Compare commits

..
138 changed files with 3842 additions and 8656 deletions

No files matched your search

-14
View File
@@ -37,17 +37,6 @@ jobs:
- blockchain
- other
steps:
- name: Configure sysctl limits
run: |
sudo swapoff -a
sudo sysctl -w vm.swappiness=1
sudo sysctl -w fs.file-max=262144
sudo sysctl -w vm.max_map_count=262144
- name: Runs Elasticsearch
uses: elastic/elastic-github-actions/elasticsearch@master
with:
stack-version: 7.6.0
- uses: actions/checkout@v2
- uses: actions/setup-python@v1
with:
@@ -78,9 +67,6 @@ jobs:
run: |
pip install pyinstaller
pip install -e .
# https://stackoverflow.com/a/61693590
# https://github.com/pypa/setuptools/issues/1963
pip install --upgrade 'setuptools<45.0.0'
- if: startsWith(matrix.os, 'windows') == false
name: Build & Run (Unix)
run: |
+1 -6
View File
@@ -6,17 +6,12 @@
/.coverage*
/lbry-venv
/venv
/lbry/blockchain
lbry.egg-info
__pycache__
_trial_temp/
trending*.log
/tests/integration/blockchain/files
/tests/.coverage.*
/lbry/wallet/bin
/.vscode
/.gitignore
/lbry/wallet/bin
+2 -10
View File
@@ -1,10 +1,5 @@
default:
image: python:3.7
services:
- name: docker.elastic.co/elasticsearch/elasticsearch:7.11.2
alias: elasticsearch
variables:
ELASTIC_HOST: elasticsearch
#cache:
@@ -46,7 +41,7 @@ test:datanetwork-integration:
stage: test
script:
- pip install tox-travis
- tox -e datanetwork --recreate
- tox -e datanetwork
test:blockchain-integration:
stage: test
@@ -99,7 +94,6 @@ build:linux:
- apt-get update
- apt-get install -y --no-install-recommends python3.7-dev
- python3.7 <(curl -q https://bootstrap.pypa.io/get-pip.py) # make sure we get pip with python3.7
- pip install lbry-libtorrent
build:mac:
extends: .build
@@ -126,11 +120,9 @@ build:windows:
GIT_DEPTH: 5
before_script:
- ./docker/install_choco.ps1
- choco install -y --x64 python --version=3.7.9
- choco install -y 7zip checksum
- choco install -y --x86 python3 7zip checksum
- python --version # | findstr /B "Python 3\.7\." # dont upgrade python on every run. just make sure we're on the right Python
- pip --version # | findstr /E '\(python 3\.7\)'
- python -c "import sys;print(f'{str(64 if sys.maxsize > 2**32 else 32)} bit python\n{sys.platform}')"
- pip install virtualenv pywin32
- virtualenv venv
- venv/Scripts/activate.ps1
+1 -2
View File
@@ -1,7 +1,6 @@
.PHONY: install tools lint test idea
install:
pip install https://s3.amazonaws.com/files.lbry.io/python_libtorrent-1.2.4-py3-none-any.whl
CFLAGS="-DSQLITE_MAX_VARIABLE_NUMBER=2500000" pip install -U https://github.com/rogerbinns/apsw/releases/download/3.30.1-r1/apsw-3.30.1-r1.zip \
--global-option=fetch \
--global-option=--version --global-option=3.30.1 --global-option=--all \
@@ -9,7 +8,7 @@ install:
pip install -e .
tools:
pip install mypy==0.701 pylint==2.4.4
pip install mypy==0.701
pip install coverage astroid pylint
lint:
+2 -4
View File
@@ -1,4 +1,4 @@
FROM debian:10-slim
FROM ubuntu:19.10
ARG user=lbry
ARG db_dir=/database
@@ -13,9 +13,7 @@ RUN apt-get update && \
wget \
tar unzip \
build-essential \
pkg-config \
libleveldb-dev \
python3.7 \
python3 \
python3-dev \
python3-pip \
python3-wheel \
+20 -26
View File
@@ -1,40 +1,34 @@
version: "3"
volumes:
lbrycrd:
wallet_server:
es01:
services:
lbrycrd:
image: lbry/lbrycrd:${LBRYCRD_TAG:-latest-release}
restart: always
ports: # accessible from host
- "9246:9246" # rpc port
expose: # internal to docker network. also this doesn't do anything. its for documentation only.
- "9245" # node-to-node comms port
volumes:
- "lbrycrd:/data/.lbrycrd"
environment:
- RUN_MODE=default
- SNAPSHOT_URL=${LBRYCRD_SNAPSHOT_URL-https://lbry.com/snapshot/blockchain}
- RPC_ALLOW_IP=0.0.0.0/0
wallet_server:
depends_on:
- es01
image: lbry/wallet-server:${WALLET_SERVER_TAG:-latest-release}
depends_on:
- lbrycrd
restart: always
network_mode: host
ports:
- "50001:50001" # rpc port
- "2112:2112" # uncomment to enable prometheus
- "50005:50005" # websocket port
#- "2112:2112" # uncomment to enable prometheus
volumes:
- "wallet_server:/database"
env_file: [/home/lbry/wallet-server-env]
environment:
- DAEMON_URL=http://lbry:lbry@127.0.0.1:9245
- TCP_PORT=50001
- PROMETHEUS_PORT=2112
es01:
image: docker.elastic.co/elasticsearch/elasticsearch:7.11.0
container_name: es01
environment:
- node.name=es01
- discovery.type=single-node
- indices.query.bool.max_clause_count=4096
- bootstrap.memory_lock=true
- "ES_JAVA_OPTS=-Xms8g -Xmx8g" # no more than 32, remember to disable swap
ulimits:
memlock:
soft: -1
hard: -1
volumes:
- es01:/usr/share/elasticsearch/data
ports:
- 127.0.0.1:9200:9200
- SNAPSHOT_URL=${WALLET_SERVER_SNAPSHOT_URL-https://lbry.com/snapshot/wallet}
- DAEMON_URL=http://lbry:lbry@lbrycrd:9245
+1 -3
View File
@@ -20,6 +20,4 @@ if [[ -n "$SNAPSHOT_URL" ]] && [[ ! -f /database/claims.db ]]; then
rm "$filename"
fi
/home/lbry/.local/bin/lbry-hub-elastic-sync /database/claims.db
echo 'starting server'
/home/lbry/.local/bin/lbry-hub "$@"
/home/lbry/.local/bin/torba-server "$@"
+187 -615
View File
File diff suppressed because it is too large. Load diff
+1 -1
View File
@@ -1,2 +1,2 @@
__version__ = "0.92.0"
__version__ = "0.70.0"
version = tuple(map(int, __version__.split('.'))) # pylint: disable=invalid-name
+10 -24
View File
@@ -110,7 +110,7 @@ class AbstractBlob:
if reader in self.readers:
self.readers.remove(reader)
def _write_blob(self, blob_bytes: bytes) -> asyncio.Task:
def _write_blob(self, blob_bytes: bytes):
raise NotImplementedError()
def set_length(self, length) -> None:
@@ -198,17 +198,11 @@ class AbstractBlob:
def save_verified_blob(self, verified_bytes: bytes):
if self.verified.is_set():
return
def update_events(_):
self.verified.set()
self.writing.clear()
if self.is_writeable():
self.writing.set()
task = self._write_blob(verified_bytes)
task.add_done_callback(update_events)
self._write_blob(verified_bytes)
self.verified.set()
if self.blob_completed_callback:
task.add_done_callback(lambda _: self.blob_completed_callback(self))
self.blob_completed_callback(self)
def get_blob_writer(self, peer_address: typing.Optional[str] = None,
peer_port: typing.Optional[int] = None) -> HashBlobWriter:
@@ -267,11 +261,9 @@ class BlobBuffer(AbstractBlob):
self.verified.clear()
def _write_blob(self, blob_bytes: bytes):
async def write():
if self._verified_bytes:
raise OSError("already have bytes for blob")
self._verified_bytes = BytesIO(blob_bytes)
return self.loop.create_task(write())
if self._verified_bytes:
raise OSError("already have bytes for blob")
self._verified_bytes = BytesIO(blob_bytes)
def delete(self):
if self._verified_bytes:
@@ -327,19 +319,13 @@ class BlobFile(AbstractBlob):
handle.close()
def _write_blob(self, blob_bytes: bytes):
def _write_blob():
with open(self.file_path, 'wb') as f:
f.write(blob_bytes)
async def write_blob():
await self.loop.run_in_executor(None, _write_blob)
return self.loop.create_task(write_blob())
with open(self.file_path, 'wb') as f:
f.write(blob_bytes)
def delete(self):
super().delete()
if os.path.isfile(self.file_path):
os.remove(self.file_path)
return super().delete()
@classmethod
async def create_from_unencrypted(
+2 -2
View File
@@ -2,7 +2,7 @@ import os
import typing
import asyncio
import logging
from lbry.utils import LRUCacheWithMetrics
from lbry.utils import LRUCache
from lbry.blob.blob_file import is_valid_blobhash, BlobFile, BlobBuffer, AbstractBlob
from lbry.stream.descriptor import StreamDescriptor
from lbry.connection_manager import ConnectionManager
@@ -32,7 +32,7 @@ class BlobManager:
else self._node_data_store.completed_blobs
self.blobs: typing.Dict[str, AbstractBlob] = {}
self.config = config
self.decrypted_blob_lru_cache = None if not self.config.blob_lru_cache_size else LRUCacheWithMetrics(
self.decrypted_blob_lru_cache = None if not self.config.blob_lru_cache_size else LRUCache(
self.config.blob_lru_cache_size)
self.connection_manager = ConnectionManager(loop)
-2
View File
@@ -152,8 +152,6 @@ class BlobExchangeClientProtocol(asyncio.Protocol):
log.debug(msg)
msg = f"downloaded {self.blob.blob_hash[:8]} from {self.peer_address}:{self.peer_port}"
await asyncio.wait_for(self.writer.finished, self.peer_timeout, loop=self.loop)
# wait for the io to finish
await self.blob.verified.wait()
log.info("%s at %fMB/s", msg,
round((float(self._blob_bytes_received) /
float(time.perf_counter() - start_time)) / 1000000.0, 2))
+17 -35
View File
@@ -5,8 +5,9 @@ import typing
import logging
from argparse import ArgumentParser
from contextlib import contextmanager
from appdirs import user_data_dir, user_config_dir
import yaml
from appdirs import user_data_dir, user_config_dir
from lbry.error import InvalidCurrencyError
from lbry.dht import constants
from lbry.wallet.coinselection import STRATEGIES
@@ -276,23 +277,14 @@ class Strings(ListSetting):
class EnvironmentAccess:
PREFIX = 'LBRY_'
def __init__(self, config: 'BaseConfig', environ: dict):
self.configuration = config
self.data = {}
if environ:
self.load(environ)
def load(self, environ):
for setting in self.configuration.get_settings():
value = environ.get(f'{self.PREFIX}{setting.name.upper()}', NOT_SET)
if value != NOT_SET and not (isinstance(setting, ListSetting) and value is None):
self.data[setting.name] = setting.deserialize(value)
def __init__(self, environ: dict):
self.environ = environ
def __contains__(self, item: str):
return item in self.data
return f'{self.PREFIX}{item.upper()}' in self.environ
def __getitem__(self, item: str):
return self.data[item]
return self.environ[f'{self.PREFIX}{item.upper()}']
class ArgumentAccess:
@@ -333,7 +325,7 @@ class ConfigFileAccess:
cls = type(self.configuration)
with open(self.path, 'r') as config_file:
raw = config_file.read()
serialized = yaml.safe_load(raw) or {}
serialized = yaml.load(raw) or {}
for key, value in serialized.items():
attr = getattr(cls, key, None)
if attr is None:
@@ -451,7 +443,7 @@ class BaseConfig:
self.arguments = ArgumentAccess(self, args)
def set_environment(self, environ=None):
self.environment = EnvironmentAccess(self, environ or os.environ)
self.environment = EnvironmentAccess(environ or os.environ)
def set_persisted(self, config_file_path=None):
if config_file_path is None:
@@ -477,16 +469,16 @@ class TranscodeConfig(BaseConfig):
'', previous_names=['ffmpeg_folder'])
video_encoder = String('FFmpeg codec and parameters for the video encoding. '
'Example: libaom-av1 -crf 25 -b:v 0 -strict experimental',
'libx264 -crf 24 -preset faster -pix_fmt yuv420p')
video_bitrate_maximum = Integer('Maximum bits per second allowed for video streams (0 to disable).', 5_000_000)
'libx264 -crf 21 -preset faster -pix_fmt yuv420p')
video_bitrate_maximum = Integer('Maximum bits per second allowed for video streams (0 to disable).', 8400000)
video_scaler = String('FFmpeg scaling parameters for reducing bitrate. '
'Example: -vf "scale=-2:720,fps=24" -maxrate 5M -bufsize 3M',
r'-vf "scale=if(gte(iw\,ih)\,min(1920\,iw)\,-2):if(lt(iw\,ih)\,min(1920\,ih)\,-2)" '
r'-maxrate 5500K -bufsize 5000K')
r'-maxrate 8400K -bufsize 5000K')
audio_encoder = String('FFmpeg codec and parameters for the audio encoding. '
'Example: libopus -b:a 128k',
'aac -b:a 160k')
volume_filter = String('FFmpeg filter for audio normalization. Exmple: -af loudnorm', '')
volume_filter = String('FFmpeg filter for audio normalization.', '-af loudnorm')
volume_analysis_time = Integer('Maximum seconds into the file that we examine audio volume (0 to disable).', 240)
@@ -585,14 +577,9 @@ class Config(CLIConfig):
)
# servers
reflector_servers = Servers("Reflector re-hosting servers for mirroring publishes", [
reflector_servers = Servers("Reflector re-hosting servers", [
('reflector.lbry.com', 5566)
])
fixed_peers = Servers("Fixed peers to fall back to if none are found on P2P for a blob", [
('cdn.reflector.lbry.com', 5567)
])
lbryum_servers = Servers("SPV wallet servers", [
('spv11.lbry.com', 50001),
('spv12.lbry.com', 50001),
@@ -611,7 +598,7 @@ class Config(CLIConfig):
('lbrynet4.lbry.com', 4444) # ASIA
])
comment_server = String("Comment server API URL", "https://comments.lbry.com/api/v2")
comment_server = String("Comment server API URL", "https://comments.lbry.com/api")
# blockchain
blockchain_name = String("Blockchain name - lbrycrd_main, lbrycrd_regtest, or lbrycrd_testnet", 'lbrycrd_main')
@@ -624,9 +611,6 @@ class Config(CLIConfig):
previous_names=['upload_log', 'upload_log', 'share_debug_info']
)
track_bandwidth = Toggle("Track bandwidth usage", True)
allowed_origin = String(
"Allowed `Origin` header value for API request (sent by browser), use * to allow "
"all hosts; default is to only allow API requests with no `Origin` value.", "")
# media server
streaming_server = String('Host name and port to serve streaming media over range requests',
@@ -638,7 +622,6 @@ class Config(CLIConfig):
"Strategy to use when selecting UTXOs for a transaction",
STRATEGIES, "standard")
transaction_cache_size = Integer("Transaction cache size", 2 ** 17)
save_resolved_claims = Toggle(
"Save content claims to the database when they are resolved to keep file_list up to date, "
"only disable this if file_x commands are not needed", True
@@ -708,10 +691,9 @@ def get_darwin_directories() -> typing.Tuple[str, str, str]:
def get_linux_directories() -> typing.Tuple[str, str, str]:
try:
with open(os.path.join(user_config_dir(), 'user-dirs.dirs'), 'r') as xdg:
down_dir = re.search(r'XDG_DOWNLOAD_DIR=(.+)', xdg.read())
if down_dir:
down_dir = re.sub(r'\$HOME', os.getenv('HOME') or os.path.expanduser("~/"), down_dir.group(1))
download_dir = re.sub('\"', '', down_dir)
down_dir = re.search(r'XDG_DOWNLOAD_DIR=(.+)', xdg.read()).group(1)
down_dir = re.sub(r'\$HOME', os.getenv('HOME') or os.path.expanduser("~/"), down_dir)
download_dir = re.sub('\"', '', down_dir)
except OSError:
download_dir = os.getenv('XDG_DOWNLOAD_DIR')
if not download_dir:
+6 -3
View File
@@ -42,8 +42,7 @@ class ConnectionManager:
self.outgoing_connected.add(host_and_port)
def connection_received(self, host_and_port: str):
# self.incoming_connected.add(host_and_port)
pass
self.incoming_connected.add(host_and_port)
def outgoing_connection_lost(self, host_and_port: str):
if self._running and host_and_port in self.outgoing_connected:
@@ -62,7 +61,9 @@ class ConnectionManager:
'total_sent': 0,
'total_received': 0,
'max_incoming_mbs': 0.0,
'max_outgoing_mbs': 0.0
'max_outgoing_mbs': 0.0,
'outbound_connections': 0,
'incoming_connections': 0
}
while True:
@@ -87,6 +88,8 @@ class ConnectionManager:
self._max_outgoing_mbs = max(self._max_outgoing_mbs, self._status['total_outgoing_mbs'])
self._status['max_incoming_mbs'] = self._max_incoming_mbs
self._status['max_outgoing_mbs'] = self._max_outgoing_mbs
self._status['outbound_connections'] = len(self.outgoing_connected)
self._status['incoming_connections'] = len(self.incoming_connected)
def stop(self):
if self._task:
+20 -3
View File
@@ -1,14 +1,14 @@
import typing
import asyncio
import logging
import ipaddress
from binascii import hexlify
from dataclasses import dataclass, field
from functools import lru_cache
from lbry.utils import is_valid_public_ipv4 as _is_valid_public_ipv4
from lbry.dht import constants
from lbry.dht.serialization.datagram import make_compact_address, make_compact_ip, decode_compact_address
ALLOW_LOCALHOST = False
log = logging.getLogger(__name__)
@@ -20,9 +20,26 @@ def make_kademlia_peer(node_id: typing.Optional[bytes], address: typing.Optional
return KademliaPeer(address, node_id, udp_port, tcp_port=tcp_port, allow_localhost=allow_localhost)
# the ipaddress module does not show these subnets as reserved
CARRIER_GRADE_NAT_SUBNET = ipaddress.ip_network('100.64.0.0/10')
IPV4_TO_6_RELAY_SUBNET = ipaddress.ip_network('192.88.99.0/24')
ALLOW_LOCALHOST = False
def is_valid_public_ipv4(address, allow_localhost: bool = False):
allow_localhost = bool(allow_localhost or ALLOW_LOCALHOST)
return _is_valid_public_ipv4(address, allow_localhost)
try:
parsed_ip = ipaddress.ip_address(address)
if parsed_ip.is_loopback and allow_localhost:
return True
return not any((parsed_ip.version != 4, parsed_ip.is_unspecified, parsed_ip.is_link_local,
parsed_ip.is_loopback, parsed_ip.is_multicast, parsed_ip.is_reserved, parsed_ip.is_private,
parsed_ip.is_reserved,
CARRIER_GRADE_NAT_SUBNET.supernet_of(ipaddress.ip_network(f"{address}/32")),
IPV4_TO_6_RELAY_SUBNET.supernet_of(ipaddress.ip_network(f"{address}/32"))))
except ipaddress.AddressValueError:
return False
class PeerManager:
+1 -2
View File
@@ -10,7 +10,6 @@ from asyncio.protocols import DatagramProtocol
from asyncio.transports import DatagramTransport
from lbry.dht import constants
from lbry.dht.serialization.bencoding import DecodeError
from lbry.dht.serialization.datagram import decode_datagram, ErrorDatagram, ResponseDatagram, RequestDatagram
from lbry.dht.serialization.datagram import RESPONSE_TYPE, ERROR_TYPE, PAGE_KEY
from lbry.dht.error import RemoteException, TransportNotConnected
@@ -555,7 +554,7 @@ class KademliaProtocol(DatagramProtocol):
def datagram_received(self, datagram: bytes, address: typing.Tuple[str, int]) -> None: # pylint: disable=arguments-differ
try:
message = decode_datagram(datagram)
except (ValueError, TypeError, DecodeError):
except (ValueError, TypeError):
self.peer_manager.report_failure(address[0], address[1])
log.warning("Couldn't decode dht datagram from %s: %s", address, binascii.hexlify(datagram).decode())
return
+7 -17
View File
@@ -144,7 +144,7 @@ class ErrorDatagram(KademliaDatagramBase):
self.response = response.decode()
def _decode_datagram(datagram: bytes):
def decode_datagram(datagram: bytes) -> typing.Union[RequestDatagram, ResponseDatagram, ErrorDatagram]:
msg_types = {
REQUEST_TYPE: RequestDatagram,
RESPONSE_TYPE: ResponseDatagram,
@@ -152,29 +152,19 @@ def _decode_datagram(datagram: bytes):
}
primitive: typing.Dict = bdecode(datagram)
converted = {
str(k).encode() if not isinstance(k, bytes) else k: v for k, v in primitive.items()
}
if converted[b'0'] in [REQUEST_TYPE, ERROR_TYPE, RESPONSE_TYPE]: # pylint: disable=unsubscriptable-object
datagram_type = converted[b'0'] # pylint: disable=unsubscriptable-object
if primitive[0] in [REQUEST_TYPE, ERROR_TYPE, RESPONSE_TYPE]: # pylint: disable=unsubscriptable-object
datagram_type = primitive[0] # pylint: disable=unsubscriptable-object
else:
raise ValueError("invalid datagram type")
datagram_class = msg_types[datagram_type]
decoded = {
k: converted[str(i).encode()] # pylint: disable=unsubscriptable-object
k: primitive[i] # pylint: disable=unsubscriptable-object
for i, k in enumerate(datagram_class.required_fields)
if str(i).encode() in converted # pylint: disable=unsupported-membership-test
if i in primitive # pylint: disable=unsupported-membership-test
}
for i, _ in enumerate(OPTIONAL_FIELDS):
if str(i + OPTIONAL_ARG_OFFSET).encode() in converted:
decoded[i + OPTIONAL_ARG_OFFSET] = converted[str(i + OPTIONAL_ARG_OFFSET).encode()]
return decoded, datagram_class
def decode_datagram(datagram: bytes) -> typing.Union[RequestDatagram, ResponseDatagram, ErrorDatagram]:
decoded, datagram_class = _decode_datagram(datagram)
if i + OPTIONAL_ARG_OFFSET in primitive:
decoded[i + OPTIONAL_ARG_OFFSET] = primitive[i + OPTIONAL_ARG_OFFSET]
return datagram_class(**decoded)
+5
View File
@@ -14,6 +14,7 @@ from aiohttp.web import GracefulExit
from docopt import docopt
from lbry import __version__ as lbrynet_version
from lbry.extras.daemon.loggly_handler import get_loggly_handler
from lbry.extras.daemon.daemon import Daemon
from lbry.conf import Config, CLIConfig
@@ -254,6 +255,10 @@ def setup_logging(logger: logging.Logger, args: argparse.Namespace, conf: Config
else:
logger.getChild('lbry').setLevel(logging.DEBUG)
loggly_handler = get_loggly_handler(conf)
loggly_handler.setLevel(logging.ERROR)
logger.getChild('lbry').addHandler(loggly_handler)
def run_daemon(args: argparse.Namespace, conf: Config):
loop = asyncio.get_event_loop()
+2 -2
View File
@@ -66,7 +66,7 @@ def _download_properties(conf: Config, external_ip: str, resolve_duration: float
"node_rpc_timeout": conf.node_rpc_timeout,
"peer_connect_timeout": conf.peer_connect_timeout,
"blob_download_timeout": conf.blob_download_timeout,
"use_fixed_peers": len(conf.fixed_peers) > 0,
"use_fixed_peers": len(conf.reflector_servers) > 0,
"fixed_peer_delay": fixed_peer_delay,
"added_fixed_peers": added_fixed_peers,
"active_peer_count": active_peer_count,
@@ -132,7 +132,7 @@ class AnalyticsManager:
async def run(self):
while True:
if self.enabled:
self.external_ip, _ = await utils.get_external_ip(self.conf.lbryum_servers)
self.external_ip = await utils.get_external_ip()
await self._send_heartbeat()
await asyncio.sleep(1800)
+18 -31
View File
@@ -22,54 +22,41 @@ def cid2hash(claim_id: str) -> bytes:
return binascii.unhexlify(claim_id.encode())[::-1]
def is_comment_signed_by_channel(comment: dict, channel: Output, sign_comment_id=False):
def is_comment_signed_by_channel(comment: dict, channel: Output, abandon=False):
if isinstance(channel, Output):
try:
signing_field = comment['comment_id'] if sign_comment_id else comment['comment']
return verify(channel, signing_field.encode(), comment, cid2hash(comment['channel_id']))
signing_field = comment['comment_id'] if abandon else comment['comment']
pieces = [
comment['signing_ts'].encode(),
cid2hash(comment['channel_id']),
signing_field.encode()
]
return Output.is_signature_valid(
get_encoded_signature(comment['signature']),
sha256(b''.join(pieces)),
channel.claim.channel.public_key_bytes
)
except KeyError:
pass
return False
def verify(channel, data, signature, channel_hash=None):
pieces = [
signature['signing_ts'].encode(),
channel_hash or channel.claim_hash,
data
]
return Output.is_signature_valid(
get_encoded_signature(signature['signature']),
sha256(b''.join(pieces)),
channel.claim.channel.public_key_bytes
)
def sign_comment(comment: dict, channel: Output, sign_comment_id=False):
signing_field = comment['comment_id'] if sign_comment_id else comment['comment']
comment.update(sign(channel, signing_field.encode()))
def sign(channel, data):
def sign_comment(comment: dict, channel: Output, abandon=False):
timestamp = str(int(time.time()))
pieces = [timestamp.encode(), channel.claim_hash, data]
signing_field = comment['comment_id'] if abandon else comment['comment']
pieces = [timestamp.encode(), channel.claim_hash, signing_field.encode()]
digest = sha256(b''.join(pieces))
signature = channel.private_key.sign_digest_deterministic(digest, hashfunc=hashlib.sha256)
return {
comment.update({
'signature': binascii.hexlify(signature).decode(),
'signing_ts': timestamp
}
def sign_reaction(reaction: dict, channel: Output):
signing_field = reaction['channel_name']
reaction.update(sign(channel, signing_field.encode()))
})
async def jsonrpc_post(url: str, method: str, params: dict = None, **kwargs) -> any:
params = params or {}
params.update(kwargs)
json_body = {'jsonrpc': '2.0', 'id': 1, 'method': method, 'params': params}
json_body = {'jsonrpc': '2.0', 'id': None, 'method': method, 'params': params}
async with utils.aiohttp_request('POST', url, json=json_body) as response:
try:
result = await response.json()
+25 -72
View File
@@ -17,17 +17,11 @@ from lbry.dht.blob_announcer import BlobAnnouncer
from lbry.blob.blob_manager import BlobManager
from lbry.blob_exchange.server import BlobServer
from lbry.stream.stream_manager import StreamManager
from lbry.file.file_manager import FileManager
from lbry.extras.daemon.component import Component
from lbry.extras.daemon.exchange_rate_manager import ExchangeRateManager
from lbry.extras.daemon.storage import SQLiteStorage
from lbry.torrent.torrent_manager import TorrentManager
from lbry.wallet import WalletManager
from lbry.wallet.usage_payment import WalletServerPayer
try:
from lbry.torrent.session import TorrentSession
except ImportError:
TorrentSession = None
log = logging.getLogger(__name__)
@@ -39,11 +33,10 @@ WALLET_COMPONENT = "wallet"
WALLET_SERVER_PAYMENTS_COMPONENT = "wallet_server_payments"
DHT_COMPONENT = "dht"
HASH_ANNOUNCER_COMPONENT = "hash_announcer"
FILE_MANAGER_COMPONENT = "file_manager"
STREAM_MANAGER_COMPONENT = "stream_manager"
PEER_PROTOCOL_SERVER_COMPONENT = "peer_protocol_server"
UPNP_COMPONENT = "upnp"
EXCHANGE_RATE_MANAGER_COMPONENT = "exchange_rate_manager"
LIBTORRENT_COMPONENT = "libtorrent_component"
class DatabaseComponent(Component):
@@ -119,14 +112,13 @@ class WalletComponent(Component):
async def get_status(self):
if self.wallet_manager is None:
return
is_connected = self.wallet_manager.ledger.network.is_connected
sessions = []
session_pool = self.wallet_manager.ledger.network.session_pool
sessions = session_pool.sessions
connected = None
if is_connected:
addr, port = self.wallet_manager.ledger.network.client.server
connected = f"{addr}:{port}"
sessions.append(self.wallet_manager.ledger.network.client)
if self.wallet_manager.ledger.network.client:
addr_and_port = self.wallet_manager.ledger.network.client.server_address_and_port
if addr_and_port:
connected = f"{addr_and_port[0]}:{addr_and_port[1]}"
result = {
'connected': connected,
'connected_features': self.wallet_manager.ledger.network.server_features,
@@ -138,8 +130,8 @@ class WalletComponent(Component):
'availability': session.available,
} for session in sessions
],
'known_servers': len(self.wallet_manager.ledger.network.config['default_servers']),
'available_servers': 1 if is_connected else 0
'known_servers': len(sessions),
'available_servers': len(list(session_pool.available_sessions))
}
if self.wallet_manager.ledger.network.remote_height:
@@ -275,7 +267,7 @@ class DHTComponent(Component):
external_ip = upnp_component.external_ip
storage = self.component_manager.get_component(DATABASE_COMPONENT)
if not external_ip:
external_ip, _ = await utils.get_external_ip(self.conf.lbryum_servers)
external_ip = await utils.get_external_ip()
if not external_ip:
log.warning("failed to get external ip")
@@ -327,23 +319,23 @@ class HashAnnouncerComponent(Component):
}
class FileManagerComponent(Component):
component_name = FILE_MANAGER_COMPONENT
class StreamManagerComponent(Component):
component_name = STREAM_MANAGER_COMPONENT
depends_on = [BLOB_COMPONENT, DATABASE_COMPONENT, WALLET_COMPONENT]
def __init__(self, component_manager):
super().__init__(component_manager)
self.file_manager: typing.Optional[FileManager] = None
self.stream_manager: typing.Optional[StreamManager] = None
@property
def component(self) -> typing.Optional[FileManager]:
return self.file_manager
def component(self) -> typing.Optional[StreamManager]:
return self.stream_manager
async def get_status(self):
if not self.file_manager:
if not self.stream_manager:
return
return {
'managed_files': len(self.file_manager.get_filtered()),
'managed_files': len(self.stream_manager.streams),
}
async def start(self):
@@ -352,55 +344,16 @@ class FileManagerComponent(Component):
wallet = self.component_manager.get_component(WALLET_COMPONENT)
node = self.component_manager.get_component(DHT_COMPONENT) \
if self.component_manager.has_component(DHT_COMPONENT) else None
try:
torrent = self.component_manager.get_component(LIBTORRENT_COMPONENT) if TorrentSession else None
except NameError:
torrent = None
log.info('Starting the file manager')
loop = asyncio.get_event_loop()
self.file_manager = FileManager(
loop, self.conf, wallet, storage, self.component_manager.analytics_manager
self.stream_manager = StreamManager(
loop, self.conf, blob_manager, wallet, storage, node, self.component_manager.analytics_manager
)
self.file_manager.source_managers['stream'] = StreamManager(
loop, self.conf, blob_manager, wallet, storage, node,
)
if TorrentSession and LIBTORRENT_COMPONENT not in self.conf.components_to_skip:
self.file_manager.source_managers['torrent'] = TorrentManager(
loop, self.conf, torrent, storage, self.component_manager.analytics_manager
)
await self.file_manager.start()
await self.stream_manager.start()
log.info('Done setting up file manager')
async def stop(self):
self.file_manager.stop()
class TorrentComponent(Component):
component_name = LIBTORRENT_COMPONENT
def __init__(self, component_manager):
super().__init__(component_manager)
self.torrent_session = None
@property
def component(self) -> typing.Optional[TorrentSession]:
return self.torrent_session
async def get_status(self):
if not self.torrent_session:
return
return {
'running': True, # TODO: what to return here?
}
async def start(self):
if TorrentSession:
self.torrent_session = TorrentSession(asyncio.get_event_loop(), None)
await self.torrent_session.bind() # TODO: specify host/port
async def stop(self):
if self.torrent_session:
await self.torrent_session.pause()
self.stream_manager.stop()
class PeerProtocolServerComponent(Component):
@@ -476,7 +429,7 @@ class UPnPComponent(Component):
pass
if external_ip and not is_valid_public_ipv4(external_ip):
log.warning("UPnP returned a private/reserved ip - %s, checking lbry.com fallback", external_ip)
external_ip, _ = await utils.get_external_ip(self.conf.lbryum_servers)
external_ip = await utils.get_external_ip()
if self.external_ip and self.external_ip != external_ip:
log.info("external ip changed from %s to %s", self.external_ip, external_ip)
if external_ip:
@@ -534,7 +487,7 @@ class UPnPComponent(Component):
async def start(self):
log.info("detecting external ip")
if not self.use_upnp:
self.external_ip, _ = await utils.get_external_ip(self.conf.lbryum_servers)
self.external_ip = await utils.get_external_ip()
return
success = False
await self._maintain_redirects()
@@ -549,9 +502,9 @@ class UPnPComponent(Component):
else:
log.error("failed to setup upnp")
if not self.external_ip:
self.external_ip, probed_url = await utils.get_external_ip(self.conf.lbryum_servers)
self.external_ip = await utils.get_external_ip()
if self.external_ip:
log.info("detected external ip using %s fallback", probed_url)
log.info("detected external ip using lbry.com fallback")
if self.component_manager.analytics_manager:
self.component_manager.loop.create_task(
self.component_manager.analytics_manager.send_upnp_setup_success_fail(
File diff suppressed because it is too large. Load diff
+59 -111
View File
@@ -2,7 +2,6 @@ import json
import time
import asyncio
import logging
from statistics import median
from decimal import Decimal
from typing import Optional, Iterable, Type
from aiohttp.client_exceptions import ContentTypeError
@@ -59,12 +58,9 @@ class MarketFeed:
raise NotImplementedError()
async def get_response(self):
async with aiohttp_request(
'get', self.url, params=self.params,
timeout=self.request_timeout, headers={"User-Agent": "lbrynet"}
) as response:
async with aiohttp_request('get', self.url, params=self.params, timeout=self.request_timeout) as response:
try:
self._last_response = await response.json(content_type=None)
self._last_response = await response.json()
except ContentTypeError as e:
self._last_response = {}
log.warning("Could not parse exchange rate response from %s: %s", self.name, e.message)
@@ -108,115 +104,70 @@ class MarketFeed:
self.event.clear()
class BaseBittrexFeed(MarketFeed):
class BittrexFeed(MarketFeed):
name = "Bittrex"
market = None
url = None
market = "BTCLBC"
url = "https://bittrex.com/api/v1.1/public/getmarkethistory"
params = {'market': 'BTC-LBC', 'count': 50}
fee = 0.0025
def get_rate_from_response(self, json_response):
if 'lastTradeRate' not in json_response:
raise InvalidExchangeRateResponseError(self.name, 'result not found')
return 1.0 / float(json_response['lastTradeRate'])
class BittrexBTCFeed(BaseBittrexFeed):
market = "BTCLBC"
url = "https://api.bittrex.com/v3/markets/LBC-BTC/ticker"
class BittrexUSDFeed(BaseBittrexFeed):
market = "USDLBC"
url = "https://api.bittrex.com/v3/markets/LBC-USD/ticker"
class BaseCryptonatorFeed(MarketFeed):
name = "Cryptonator"
market = None
url = None
def get_rate_from_response(self, json_response):
if 'ticker' not in json_response or 'price' not in json_response['ticker']:
raise InvalidExchangeRateResponseError(self.name, 'result not found')
return float(json_response['ticker']['price'])
class CryptonatorBTCFeed(BaseCryptonatorFeed):
market = "BTCLBC"
url = "https://api.cryptonator.com/api/ticker/btc-lbc"
class CryptonatorUSDFeed(BaseCryptonatorFeed):
market = "USDLBC"
url = "https://api.cryptonator.com/api/ticker/usd-lbc"
class BaseCoinExFeed(MarketFeed):
name = "CoinEx"
market = None
url = None
def get_rate_from_response(self, json_response):
if 'data' not in json_response or \
'ticker' not in json_response['data'] or \
'last' not in json_response['data']['ticker']:
raise InvalidExchangeRateResponseError(self.name, 'result not found')
return 1.0 / float(json_response['data']['ticker']['last'])
class CoinExBTCFeed(BaseCoinExFeed):
market = "BTCLBC"
url = "https://api.coinex.com/v1/market/ticker?market=LBCBTC"
class CoinExUSDFeed(BaseCoinExFeed):
market = "USDLBC"
url = "https://api.coinex.com/v1/market/ticker?market=LBCUSDT"
class BaseHotbitFeed(MarketFeed):
name = "hotbit"
market = None
url = "https://api.hotbit.io/api/v1/market.last"
def get_rate_from_response(self, json_response):
if 'result' not in json_response:
raise InvalidExchangeRateResponseError(self.name, 'result not found')
return 1.0 / float(json_response['result'])
trades = json_response['result']
if len(trades) == 0:
raise InvalidExchangeRateResponseError(self.name, 'trades not found')
totals = sum([i['Total'] for i in trades])
qtys = sum([i['Quantity'] for i in trades])
if totals <= 0 or qtys <= 0:
raise InvalidExchangeRateResponseError(self.name, 'quantities were not positive')
vwap = totals / qtys
return float(1.0 / vwap)
class HotbitBTCFeed(BaseHotbitFeed):
class LBRYFeed(MarketFeed):
name = "lbry.com"
market = "BTCLBC"
params = {"market": "LBC/BTC"}
class HotbitUSDFeed(BaseHotbitFeed):
market = "USDLBC"
params = {"market": "LBC/USDT"}
class UPbitBTCFeed(MarketFeed):
name = "UPbit"
market = "BTCLBC"
url = "https://api.upbit.com/v1/ticker"
params = {"markets": "BTC-LBC"}
url = "https://api.lbry.com/lbc/exchange_rate"
def get_rate_from_response(self, json_response):
if len(json_response) != 1 or 'trade_price' not in json_response[0]:
if 'data' not in json_response:
raise InvalidExchangeRateResponseError(self.name, 'result not found')
return 1.0 / float(json_response[0]['trade_price'])
return 1.0 / json_response['data']['lbc_btc']
class LBRYBTCFeed(LBRYFeed):
market = "USDBTC"
def get_rate_from_response(self, json_response):
if 'data' not in json_response:
raise InvalidExchangeRateResponseError(self.name, 'result not found')
return 1.0 / json_response['data']['btc_usd']
class CryptonatorFeed(MarketFeed):
name = "cryptonator.com"
market = "BTCLBC"
url = "https://api.cryptonator.com/api/ticker/btc-lbc"
def get_rate_from_response(self, json_response):
if 'ticker' not in json_response or len(json_response['ticker']) == 0 or \
'success' not in json_response or json_response['success'] is not True:
raise InvalidExchangeRateResponseError(self.name, 'result not found')
return float(json_response['ticker']['price'])
class CryptonatorBTCFeed(CryptonatorFeed):
market = "USDBTC"
url = "https://api.cryptonator.com/api/ticker/usd-btc"
FEEDS: Iterable[Type[MarketFeed]] = (
BittrexBTCFeed,
BittrexUSDFeed,
CryptonatorBTCFeed,
CryptonatorUSDFeed,
CoinExBTCFeed,
CoinExUSDFeed,
HotbitBTCFeed,
HotbitUSDFeed,
UPbitBTCFeed,
LBRYFeed,
LBRYBTCFeed,
BittrexFeed,
# CryptonatorFeed,
# CryptonatorBTCFeed,
)
@@ -240,23 +191,20 @@ class ExchangeRateManager:
source.stop()
def convert_currency(self, from_currency, to_currency, amount):
log.debug(
"Converting %f %s to %s, rates: %s",
amount, from_currency, to_currency,
[market.rate for market in self.market_feeds]
)
rates = [market.rate for market in self.market_feeds]
log.debug("Converting %f %s to %s, rates: %s", amount, from_currency, to_currency, rates)
if from_currency == to_currency:
return round(amount, 8)
rates = []
for market in self.market_feeds:
if (market.has_rate and market.is_online and
market.rate.currency_pair == (from_currency, to_currency)):
rates.append(market.rate.spot)
if rates:
return round(amount * Decimal(median(rates)), 8)
return round(amount * Decimal(market.rate.spot), 8)
for market in self.market_feeds:
if (market.has_rate and market.is_online and
market.rate.currency_pair[0] == from_currency):
return round(self.convert_currency(
market.rate.currency_pair[1], to_currency, amount * Decimal(market.rate.spot)), 8)
raise CurrencyConversionError(
f'Unable to convert {amount} from {from_currency} to {to_currency}')
+36 -76
View File
@@ -7,8 +7,6 @@ from json import JSONEncoder
from google.protobuf.message import DecodeError
from lbry.schema.claim import Claim
from lbry.schema.support import Support
from lbry.torrent.torrent_manager import TorrentSource
from lbry.wallet import Wallet, Ledger, Account, Transaction, Output
from lbry.wallet.bip32 import PubKey
from lbry.wallet.dewies import dewies_to_lbc
@@ -128,7 +126,7 @@ class JSONResponseEncoder(JSONEncoder):
return self.encode_account(obj)
if isinstance(obj, Wallet):
return self.encode_wallet(obj)
if isinstance(obj, (ManagedStream, TorrentSource)):
if isinstance(obj, ManagedStream):
return self.encode_file(obj)
if isinstance(obj, Transaction):
return self.encode_transaction(obj)
@@ -136,8 +134,6 @@ class JSONResponseEncoder(JSONEncoder):
return self.encode_output(obj)
if isinstance(obj, Claim):
return self.encode_claim(obj)
if isinstance(obj, Support):
return obj.to_dict()
if isinstance(obj, PubKey):
return obj.extended_key_string()
if isinstance(obj, datetime):
@@ -223,28 +219,25 @@ class JSONResponseEncoder(JSONEncoder):
output['claims'] = [self.encode_output(o) for o in txo.claims]
if txo.reposted_claim is not None:
output['reposted_claim'] = self.encode_output(txo.reposted_claim)
if txo.script.is_claim_name or txo.script.is_update_claim or txo.script.is_support_claim_data:
try:
output['value'] = txo.signable
if self.include_protobuf:
output['protobuf'] = hexlify(txo.signable.to_bytes())
if txo.purchase_receipt is not None:
output['purchase_receipt'] = self.encode_output(txo.purchase_receipt)
if txo.script.is_claim_name or txo.script.is_update_claim:
if txo.script.is_claim_name or txo.script.is_update_claim:
try:
output['value'] = txo.claim
output['value_type'] = txo.claim.claim_type
if self.include_protobuf:
output['protobuf'] = hexlify(txo.claim.to_bytes())
if txo.purchase_receipt is not None:
output['purchase_receipt'] = self.encode_output(txo.purchase_receipt)
if txo.claim.is_channel:
output['has_signing_key'] = txo.has_private_key
elif txo.script.is_support_claim_data:
output['value_type'] = 'emoji'
if check_signature and txo.signable.is_signed:
if txo.channel is not None:
output['signing_channel'] = self.encode_output(txo.channel)
output['is_channel_signature_valid'] = txo.is_signed_by(txo.channel, self.ledger)
else:
output['signing_channel'] = {'channel_id': txo.signable.signing_channel_id}
output['is_channel_signature_valid'] = False
except DecodeError:
pass
if check_signature and txo.claim.is_signed:
if txo.channel is not None:
output['signing_channel'] = self.encode_output(txo.channel)
output['is_channel_signature_valid'] = txo.is_signed_by(txo.channel, self.ledger)
else:
output['signing_channel'] = {'channel_id': txo.claim.signing_channel_id}
output['is_channel_signature_valid'] = False
except DecodeError:
pass
return output
def encode_claim_meta(self, meta):
@@ -280,32 +273,26 @@ class JSONResponseEncoder(JSONEncoder):
output_exists = managed_stream.output_file_exists
tx_height = managed_stream.stream_claim_info.height
best_height = self.ledger.headers.height
is_stream = hasattr(managed_stream, 'stream_hash')
if is_stream:
total_bytes_lower_bound = managed_stream.descriptor.lower_bound_decrypted_length()
total_bytes = managed_stream.descriptor.upper_bound_decrypted_length()
else:
total_bytes_lower_bound = total_bytes = managed_stream.torrent_length
result = {
'streaming_url': None,
return {
'streaming_url': managed_stream.stream_url,
'completed': managed_stream.completed,
'file_name': None,
'download_directory': None,
'download_path': None,
'file_name': managed_stream.file_name if output_exists else None,
'download_directory': managed_stream.download_directory if output_exists else None,
'download_path': managed_stream.full_path if output_exists else None,
'points_paid': 0.0,
'stopped': not managed_stream.running,
'stream_hash': None,
'stream_name': None,
'suggested_file_name': None,
'sd_hash': None,
'mime_type': None,
'key': None,
'total_bytes_lower_bound': total_bytes_lower_bound,
'total_bytes': total_bytes,
'stream_hash': managed_stream.stream_hash,
'stream_name': managed_stream.descriptor.stream_name,
'suggested_file_name': managed_stream.descriptor.suggested_file_name,
'sd_hash': managed_stream.descriptor.sd_hash,
'mime_type': managed_stream.mime_type,
'key': managed_stream.descriptor.key,
'total_bytes_lower_bound': managed_stream.descriptor.lower_bound_decrypted_length(),
'total_bytes': managed_stream.descriptor.upper_bound_decrypted_length(),
'written_bytes': managed_stream.written_bytes,
'blobs_completed': None,
'blobs_in_stream': None,
'blobs_remaining': None,
'blobs_completed': managed_stream.blobs_completed,
'blobs_in_stream': managed_stream.blobs_in_stream,
'blobs_remaining': managed_stream.blobs_remaining,
'status': managed_stream.status,
'claim_id': managed_stream.claim_id,
'txid': managed_stream.txid,
@@ -322,37 +309,10 @@ class JSONResponseEncoder(JSONEncoder):
'height': tx_height,
'confirmations': (best_height + 1) - tx_height if tx_height > 0 else tx_height,
'timestamp': self.ledger.headers.estimated_timestamp(tx_height),
'is_fully_reflected': False,
'reflector_progress': False,
'uploading_to_reflector': False
'is_fully_reflected': managed_stream.is_fully_reflected,
'reflector_progress': managed_stream.reflector_progress,
'uploading_to_reflector': managed_stream.uploading_to_reflector
}
if is_stream:
result.update({
'streaming_url': managed_stream.stream_url,
'stream_hash': managed_stream.stream_hash,
'stream_name': managed_stream.descriptor.stream_name,
'suggested_file_name': managed_stream.descriptor.suggested_file_name,
'sd_hash': managed_stream.descriptor.sd_hash,
'mime_type': managed_stream.mime_type,
'key': managed_stream.descriptor.key,
'blobs_completed': managed_stream.blobs_completed,
'blobs_in_stream': managed_stream.blobs_in_stream,
'blobs_remaining': managed_stream.blobs_remaining,
'is_fully_reflected': managed_stream.is_fully_reflected,
'reflector_progress': managed_stream.reflector_progress,
'uploading_to_reflector': managed_stream.uploading_to_reflector
})
else:
result.update({
'streaming_url': f'file://{managed_stream.full_path}',
})
if output_exists:
result.update({
'file_name': managed_stream.file_name,
'download_directory': managed_stream.download_directory,
'download_path': managed_stream.full_path,
})
return result
def encode_claim(self, claim):
encoded = getattr(claim, claim.claim_type).to_dict()
+95
View File
@@ -0,0 +1,95 @@
import asyncio
import json
import logging.handlers
import traceback
import typing
from aiohttp.client_exceptions import ClientError
import aiohttp
from lbry import utils, __version__
if typing.TYPE_CHECKING:
from lbry.conf import Config
LOGGLY_TOKEN = 'BQEzZmMzLJHgAGxkBF00LGD0YGuyATVgAmqxAQEuAQZ2BQH4'
class JsonFormatter(logging.Formatter):
"""Format log records using json serialization"""
def __init__(self, **kwargs):
super().__init__()
self.attributes = kwargs
def format(self, record):
data = {
'loggerName': record.name,
'asciTime': self.formatTime(record),
'fileName': record.filename,
'functionName': record.funcName,
'levelNo': record.levelno,
'lineNo': record.lineno,
'levelName': record.levelname,
'message': record.getMessage(),
}
data.update(self.attributes)
if record.exc_info:
data['exc_info'] = self.formatException(record.exc_info)
return json.dumps(data)
class HTTPSLogglyHandler(logging.Handler):
def __init__(self, loggly_token: str, config: 'Config'):
super().__init__()
self.cookies = {}
self.url = "https://logs-01.loggly.com/inputs/{token}/tag/{tag}".format(
token=utils.deobfuscate(loggly_token), tag='lbrynet-' + __version__
)
self._loop = asyncio.get_event_loop()
self._session = aiohttp.ClientSession()
self._config = config
@property
def enabled(self):
return self._config.share_usage_data
@staticmethod
def get_full_message(record):
if record.exc_info:
return '\n'.join(traceback.format_exception(*record.exc_info))
else:
return record.getMessage()
async def _emit(self, record, retry=True):
data = self.format(record).encode()
try:
async with self._session.post(self.url, data=data,
cookies=self.cookies) as response:
self.cookies.update(response.cookies)
except ClientError:
if self._loop.is_running() and retry and self.enabled:
await self._session.close()
self._session = aiohttp.ClientSession()
return await self._emit(record, retry=False)
def emit(self, record):
if not self.enabled:
return
try:
asyncio.ensure_future(self._emit(record), loop=self._loop)
except RuntimeError: # TODO: use a second loop
print(f"\nfailed to send traceback to loggly, please file an issue with the following traceback:\n"
f"{self.format(record)}")
def close(self):
super().close()
try:
loop = asyncio.get_event_loop()
loop.run_until_complete(self._session.close())
except RuntimeError:
pass
def get_loggly_handler(config):
handler = HTTPSLogglyHandler(LOGGLY_TOKEN, config=config)
handler.setFormatter(JsonFormatter())
return handler
-31
View File
@@ -1,31 +0,0 @@
import logging
from aiohttp import web
log = logging.getLogger(__name__)
def ensure_request_allowed(request, conf):
if is_request_allowed(request, conf):
return
if conf.allowed_origin:
log.warning(
"API requests with Origin '%s' are not allowed, "
"configuration 'allowed_origin' limits requests to: '%s'",
request.headers.get('Origin'), conf.allowed_origin
)
else:
log.warning(
"API requests with Origin '%s' are not allowed, "
"update configuration 'allowed_origin' to enable this origin.",
request.headers.get('Origin')
)
raise web.HTTPForbidden()
def is_request_allowed(request, conf) -> bool:
origin = request.headers.get('Origin')
return (
origin is None or
origin == conf.allowed_origin or
conf.allowed_origin == '*'
)
+2 -33
View File
@@ -9,7 +9,7 @@ from typing import Optional
from lbry.wallet import SQLiteMixin
from lbry.conf import Config
from lbry.wallet.dewies import dewies_to_lbc, lbc_to_dewies
from lbry.wallet.transaction import Transaction, Output
from lbry.wallet.transaction import Transaction
from lbry.schema.claim import Claim
from lbry.dht.constants import DATA_EXPIRATION
from lbry.blob.blob_info import BlobInfo
@@ -727,19 +727,6 @@ class SQLiteStorage(SQLiteMixin):
if claim_id_to_supports:
await self.save_supports(claim_id_to_supports)
def save_claim_from_output(self, ledger, *outputs: Output):
return self.save_claims([{
"claim_id": output.claim_id,
"name": output.claim_name,
"amount": dewies_to_lbc(output.amount),
"address": output.get_address(ledger),
"txid": output.tx_ref.id,
"nout": output.position,
"value": output.claim,
"height": output.tx_ref.height,
"claim_sequence": -1,
} for output in outputs])
def save_claims_for_resolve(self, claim_infos):
to_save = {}
for info in claim_infos:
@@ -753,8 +740,7 @@ class SQLiteStorage(SQLiteMixin):
return self.save_claims(to_save.values())
@staticmethod
def _save_content_claim(transaction, claim_outpoint, stream_hash=None, bt_infohash=None):
assert stream_hash or bt_infohash
def _save_content_claim(transaction, claim_outpoint, stream_hash):
# get the claim id and serialized metadata
claim_info = transaction.execute(
"select claim_id, serialized_metadata from claim where claim_outpoint=?", (claim_outpoint,)
@@ -802,19 +788,6 @@ class SQLiteStorage(SQLiteMixin):
if stream_hash in self.content_claim_callbacks:
await self.content_claim_callbacks[stream_hash]()
async def save_torrent_content_claim(self, bt_infohash, claim_outpoint, length, name):
def _save_torrent(transaction):
transaction.execute(
"insert or replace into torrent values (?, NULL, ?, ?)", (bt_infohash, length, name)
).fetchall()
transaction.execute(
"insert or replace into content_claim values (NULL, ?, ?)", (bt_infohash, claim_outpoint)
).fetchall()
await self.db.run(_save_torrent)
# update corresponding ManagedEncryptedFileDownloader object
if bt_infohash in self.content_claim_callbacks:
await self.content_claim_callbacks[bt_infohash]()
async def get_content_claim(self, stream_hash: str, include_supports: typing.Optional[bool] = True) -> typing.Dict:
claims = await self.db.run(get_claims_from_stream_hashes, [stream_hash])
claim = None
@@ -826,10 +799,6 @@ class SQLiteStorage(SQLiteMixin):
claim['effective_amount'] = calculate_effective_amount(claim['amount'], supports)
return claim
async def get_content_claim_for_torrent(self, bt_infohash):
claims = await self.db.run(get_claims_from_torrent_info_hashes, [bt_infohash])
return claims[bt_infohash].as_dict() if claims else None
# # # # # # # # # reflector functions # # # # # # # # #
def update_reflected_stream(self, sd_hash, reflector_address, success=True):
View File
Whitespace-only changes.
-302
View File
@@ -1,302 +0,0 @@
import asyncio
import logging
import typing
from typing import Optional
from aiohttp.web import Request
from lbry.error import ResolveError, DownloadSDTimeoutError, InsufficientFundsError
from lbry.error import ResolveTimeoutError, DownloadDataTimeoutError, KeyFeeAboveMaxAllowedError
from lbry.stream.managed_stream import ManagedStream
from lbry.torrent.torrent_manager import TorrentSource
from lbry.utils import cache_concurrent
from lbry.schema.url import URL
from lbry.wallet.dewies import dewies_to_lbc
from lbry.file.source_manager import SourceManager
from lbry.file.source import ManagedDownloadSource
if typing.TYPE_CHECKING:
from lbry.conf import Config
from lbry.extras.daemon.analytics import AnalyticsManager
from lbry.extras.daemon.storage import SQLiteStorage
from lbry.wallet import WalletManager, Output
from lbry.extras.daemon.exchange_rate_manager import ExchangeRateManager
log = logging.getLogger(__name__)
class FileManager:
def __init__(self, loop: asyncio.AbstractEventLoop, config: 'Config', wallet_manager: 'WalletManager',
storage: 'SQLiteStorage', analytics_manager: Optional['AnalyticsManager'] = None):
self.loop = loop
self.config = config
self.wallet_manager = wallet_manager
self.storage = storage
self.analytics_manager = analytics_manager
self.source_managers: typing.Dict[str, SourceManager] = {}
self.started = asyncio.Event()
@property
def streams(self):
return self.source_managers['stream']._sources
async def create_stream(self, file_path: str, key: Optional[bytes] = None, **kwargs) -> ManagedDownloadSource:
if 'stream' in self.source_managers:
return await self.source_managers['stream'].create(file_path, key, **kwargs)
raise NotImplementedError
async def start(self):
await asyncio.gather(*(source_manager.start() for source_manager in self.source_managers.values()))
for manager in self.source_managers.values():
await manager.started.wait()
self.started.set()
def stop(self):
for manager in self.source_managers.values():
# fixme: pop or not?
manager.stop()
self.started.clear()
@cache_concurrent
async def download_from_uri(self, uri, exchange_rate_manager: 'ExchangeRateManager',
timeout: Optional[float] = None, file_name: Optional[str] = None,
download_directory: Optional[str] = None,
save_file: Optional[bool] = None, resolve_timeout: float = 3.0,
wallet: Optional['Wallet'] = None) -> ManagedDownloadSource:
wallet = wallet or self.wallet_manager.default_wallet
timeout = timeout or self.config.download_timeout
start_time = self.loop.time()
resolved_time = None
stream = None
claim = None
error = None
outpoint = None
if save_file is None:
save_file = self.config.save_files
if file_name and not save_file:
save_file = True
if save_file:
download_directory = download_directory or self.config.download_dir
else:
download_directory = None
payment = None
try:
# resolve the claim
if not URL.parse(uri).has_stream:
raise ResolveError("cannot download a channel claim, specify a /path")
try:
resolved_result = await asyncio.wait_for(
self.wallet_manager.ledger.resolve(
wallet.accounts, [uri],
include_purchase_receipt=True,
include_is_my_output=True
), resolve_timeout
)
except asyncio.TimeoutError:
raise ResolveTimeoutError(uri)
except Exception as err:
if isinstance(err, asyncio.CancelledError):
raise
log.exception("Unexpected error resolving stream:")
raise ResolveError(f"Unexpected error resolving stream: {str(err)}")
if 'error' in resolved_result:
raise ResolveError(f"Unexpected error resolving uri for download: {resolved_result['error']}")
if not resolved_result or uri not in resolved_result:
raise ResolveError(f"Failed to resolve stream at '{uri}'")
txo = resolved_result[uri]
if isinstance(txo, dict):
raise ResolveError(f"Failed to resolve stream at '{uri}': {txo}")
claim = txo.claim
outpoint = f"{txo.tx_ref.id}:{txo.position}"
resolved_time = self.loop.time() - start_time
await self.storage.save_claim_from_output(self.wallet_manager.ledger, txo)
####################
# update or replace
####################
if claim.stream.source.bt_infohash:
source_manager = self.source_managers['torrent']
existing = source_manager.get_filtered(bt_infohash=claim.stream.source.bt_infohash)
elif claim.stream.source.sd_hash:
source_manager = self.source_managers['stream']
existing = source_manager.get_filtered(sd_hash=claim.stream.source.sd_hash)
else:
raise ResolveError(f"There is nothing to download at {uri} - Source is unknown or unset")
# resume or update an existing stream, if the stream changed: download it and delete the old one after
to_replace, updated_stream = None, None
if existing and existing[0].claim_id != txo.claim_id:
raise ResolveError(f"stream for {existing[0].claim_id} collides with existing download {txo.claim_id}")
if existing:
log.info("claim contains a metadata only update to a stream we have")
if claim.stream.source.bt_infohash:
await self.storage.save_torrent_content_claim(
existing[0].identifier, outpoint, existing[0].torrent_length, existing[0].torrent_name
)
claim_info = await self.storage.get_content_claim_for_torrent(existing[0].identifier)
existing[0].set_claim(claim_info, claim)
else:
await self.storage.save_content_claim(
existing[0].stream_hash, outpoint
)
await source_manager._update_content_claim(existing[0])
updated_stream = existing[0]
else:
existing_for_claim_id = self.get_filtered(claim_id=txo.claim_id)
if existing_for_claim_id:
log.info("claim contains an update to a stream we have, downloading it")
if save_file and existing_for_claim_id[0].output_file_exists:
save_file = False
if not claim.stream.source.bt_infohash:
existing_for_claim_id[0].downloader.node = source_manager.node
await existing_for_claim_id[0].start(timeout=timeout, save_now=save_file)
if not existing_for_claim_id[0].output_file_exists and (
save_file or file_name or download_directory):
await existing_for_claim_id[0].save_file(
file_name=file_name, download_directory=download_directory
)
to_replace = existing_for_claim_id[0]
# resume or update an existing stream, if the stream changed: download it and delete the old one after
if updated_stream:
log.info("already have stream for %s", uri)
if save_file and updated_stream.output_file_exists:
save_file = False
if not claim.stream.source.bt_infohash:
updated_stream.downloader.node = source_manager.node
await updated_stream.start(timeout=timeout, save_now=save_file)
if not updated_stream.output_file_exists and (save_file or file_name or download_directory):
await updated_stream.save_file(
file_name=file_name, download_directory=download_directory
)
return updated_stream
####################
# pay fee
####################
needs_purchasing = (
not to_replace and
not txo.is_my_output and
txo.has_price and
not txo.purchase_receipt
)
if needs_purchasing:
payment = await self.wallet_manager.create_purchase_transaction(
wallet.accounts, txo, exchange_rate_manager
)
####################
# make downloader and wait for start
####################
if not claim.stream.source.bt_infohash:
# fixme: this shouldnt be here
stream = ManagedStream(
self.loop, self.config, source_manager.blob_manager, claim.stream.source.sd_hash,
download_directory, file_name, ManagedStream.STATUS_RUNNING, content_fee=payment,
analytics_manager=self.analytics_manager
)
stream.downloader.node = source_manager.node
else:
stream = TorrentSource(
self.loop, self.config, self.storage, identifier=claim.stream.source.bt_infohash,
file_name=file_name, download_directory=download_directory or self.config.download_dir,
status=ManagedStream.STATUS_RUNNING,
analytics_manager=self.analytics_manager,
torrent_session=source_manager.torrent_session
)
log.info("starting download for %s", uri)
before_download = self.loop.time()
await stream.start(timeout, save_file)
####################
# success case: delete to_replace if applicable, broadcast fee payment
####################
if to_replace: # delete old stream now that the replacement has started downloading
await source_manager.delete(to_replace)
if payment is not None:
await self.wallet_manager.broadcast_or_release(payment)
payment = None # to avoid releasing in `finally` later
log.info("paid fee of %s for %s", dewies_to_lbc(stream.content_fee.outputs[0].amount), uri)
await self.storage.save_content_fee(stream.stream_hash, stream.content_fee)
source_manager.add(stream)
if not claim.stream.source.bt_infohash:
await self.storage.save_content_claim(stream.stream_hash, outpoint)
else:
await self.storage.save_torrent_content_claim(
stream.identifier, outpoint, stream.torrent_length, stream.torrent_name
)
claim_info = await self.storage.get_content_claim_for_torrent(stream.identifier)
stream.set_claim(claim_info, claim)
if save_file:
await asyncio.wait_for(stream.save_file(), timeout - (self.loop.time() - before_download),
loop=self.loop)
return stream
except asyncio.TimeoutError:
error = DownloadDataTimeoutError(stream.sd_hash)
raise error
except Exception as err: # forgive data timeout, don't delete stream
expected = (DownloadSDTimeoutError, DownloadDataTimeoutError, InsufficientFundsError,
KeyFeeAboveMaxAllowedError)
if isinstance(err, expected):
log.warning("Failed to download %s: %s", uri, str(err))
elif isinstance(err, asyncio.CancelledError):
pass
else:
log.exception("Unexpected error downloading stream:")
error = err
raise
finally:
if payment is not None:
# payment is set to None after broadcasting, if we're here an exception probably happened
await self.wallet_manager.ledger.release_tx(payment)
if self.analytics_manager and claim and claim.stream.source.bt_infohash:
# TODO: analytics for torrents
pass
elif self.analytics_manager and (error or (stream and (stream.downloader.time_to_descriptor or
stream.downloader.time_to_first_bytes))):
server = self.wallet_manager.ledger.network.client.server
self.loop.create_task(
self.analytics_manager.send_time_to_first_bytes(
resolved_time, self.loop.time() - start_time, None if not stream else stream.download_id,
uri, outpoint,
None if not stream else len(stream.downloader.blob_downloader.active_connections),
None if not stream else len(stream.downloader.blob_downloader.scores),
None if not stream else len(stream.downloader.blob_downloader.connection_failures),
False if not stream else stream.downloader.added_fixed_peers,
self.config.fixed_peer_delay if not stream else stream.downloader.fixed_peers_delay,
None if not stream else stream.sd_hash,
None if not stream else stream.downloader.time_to_descriptor,
None if not (stream and stream.descriptor) else stream.descriptor.blobs[0].blob_hash,
None if not (stream and stream.descriptor) else stream.descriptor.blobs[0].length,
None if not stream else stream.downloader.time_to_first_bytes,
None if not error else error.__class__.__name__,
None if not error else str(error),
None if not server else f"{server[0]}:{server[1]}"
)
)
async def stream_partial_content(self, request: Request, sd_hash: str):
return await self.source_managers['stream'].stream_partial_content(request, sd_hash)
def get_filtered(self, *args, **kwargs) -> typing.List[ManagedDownloadSource]:
"""
Get a list of filtered and sorted ManagedStream objects
:param sort_by: field to sort by
:param reverse: reverse sorting
:param comparison: comparison operator used for filtering
:param search_by: fields and values to filter by
"""
return sum((manager.get_filtered(*args, **kwargs) for manager in self.source_managers.values()), [])
async def delete(self, source: ManagedDownloadSource, delete_file=False):
for manager in self.source_managers.values():
await manager.delete(source, delete_file)
-161
View File
@@ -1,161 +0,0 @@
import os
import asyncio
import typing
import logging
import binascii
from typing import Optional
from lbry.utils import generate_id
from lbry.extras.daemon.storage import StoredContentClaim
if typing.TYPE_CHECKING:
from lbry.conf import Config
from lbry.extras.daemon.analytics import AnalyticsManager
from lbry.wallet.transaction import Transaction
from lbry.extras.daemon.storage import SQLiteStorage
log = logging.getLogger(__name__)
class ManagedDownloadSource:
STATUS_RUNNING = "running"
STATUS_STOPPED = "stopped"
STATUS_FINISHED = "finished"
SAVING_ID = 1
STREAMING_ID = 2
def __init__(self, loop: asyncio.AbstractEventLoop, config: 'Config', storage: 'SQLiteStorage', identifier: str,
file_name: Optional[str] = None, download_directory: Optional[str] = None,
status: Optional[str] = STATUS_STOPPED, claim: Optional[StoredContentClaim] = None,
download_id: Optional[str] = None, rowid: Optional[int] = None,
content_fee: Optional['Transaction'] = None,
analytics_manager: Optional['AnalyticsManager'] = None,
added_on: Optional[int] = None):
self.loop = loop
self.storage = storage
self.config = config
self.identifier = identifier
self.download_directory = download_directory
self._file_name = file_name
self._status = status
self.stream_claim_info = claim
self.download_id = download_id or binascii.hexlify(generate_id()).decode()
self.rowid = rowid
self.content_fee = content_fee
self.purchase_receipt = None
self._added_on = added_on
self.analytics_manager = analytics_manager
self.saving = asyncio.Event(loop=self.loop)
self.finished_writing = asyncio.Event(loop=self.loop)
self.started_writing = asyncio.Event(loop=self.loop)
self.finished_write_attempt = asyncio.Event(loop=self.loop)
# @classmethod
# async def create(cls, loop: asyncio.AbstractEventLoop, config: 'Config', file_path: str,
# key: Optional[bytes] = None,
# iv_generator: Optional[typing.Generator[bytes, None, None]] = None) -> 'ManagedDownloadSource':
# raise NotImplementedError()
async def start(self, timeout: Optional[float] = None, save_now: Optional[bool] = False):
raise NotImplementedError()
async def stop(self, finished: bool = False):
raise NotImplementedError()
async def save_file(self, file_name: Optional[str] = None, download_directory: Optional[str] = None):
raise NotImplementedError()
def stop_tasks(self):
raise NotImplementedError()
def set_claim(self, claim_info: typing.Dict, claim: 'Claim'):
self.stream_claim_info = StoredContentClaim(
f"{claim_info['txid']}:{claim_info['nout']}", claim_info['claim_id'],
claim_info['name'], claim_info['amount'], claim_info['height'],
binascii.hexlify(claim.to_bytes()).decode(), claim.signing_channel_id, claim_info['address'],
claim_info['claim_sequence'], claim_info.get('channel_name')
)
# async def update_content_claim(self, claim_info: Optional[typing.Dict] = None):
# if not claim_info:
# claim_info = await self.blob_manager.storage.get_content_claim(self.stream_hash)
# self.set_claim(claim_info, claim_info['value'])
@property
def file_name(self) -> Optional[str]:
return self._file_name
@property
def added_on(self) -> Optional[int]:
return self._added_on
@property
def status(self) -> str:
return self._status
@property
def completed(self):
raise NotImplementedError()
# @property
# def stream_url(self):
# return f"http://{self.config.streaming_host}:{self.config.streaming_port}/stream/{self.sd_hash}
@property
def finished(self) -> bool:
return self.status == self.STATUS_FINISHED
@property
def running(self) -> bool:
return self.status == self.STATUS_RUNNING
@property
def claim_id(self) -> Optional[str]:
return None if not self.stream_claim_info else self.stream_claim_info.claim_id
@property
def txid(self) -> Optional[str]:
return None if not self.stream_claim_info else self.stream_claim_info.txid
@property
def nout(self) -> Optional[int]:
return None if not self.stream_claim_info else self.stream_claim_info.nout
@property
def outpoint(self) -> Optional[str]:
return None if not self.stream_claim_info else self.stream_claim_info.outpoint
@property
def claim_height(self) -> Optional[int]:
return None if not self.stream_claim_info else self.stream_claim_info.height
@property
def channel_claim_id(self) -> Optional[str]:
return None if not self.stream_claim_info else self.stream_claim_info.channel_claim_id
@property
def channel_name(self) -> Optional[str]:
return None if not self.stream_claim_info else self.stream_claim_info.channel_name
@property
def claim_name(self) -> Optional[str]:
return None if not self.stream_claim_info else self.stream_claim_info.claim_name
@property
def metadata(self) -> Optional[typing.Dict]:
return None if not self.stream_claim_info else self.stream_claim_info.claim.stream.to_dict()
@property
def metadata_protobuf(self) -> bytes:
if self.stream_claim_info:
return binascii.hexlify(self.stream_claim_info.claim.to_bytes())
@property
def full_path(self) -> Optional[str]:
return os.path.join(self.download_directory, os.path.basename(self.file_name)) \
if self.file_name and self.download_directory else None
@property
def output_file_exists(self):
return os.path.isfile(self.full_path) if self.full_path else False
-138
View File
@@ -1,138 +0,0 @@
import os
import asyncio
import logging
import typing
from typing import Optional
from lbry.file.source import ManagedDownloadSource
if typing.TYPE_CHECKING:
from lbry.conf import Config
from lbry.extras.daemon.analytics import AnalyticsManager
from lbry.extras.daemon.storage import SQLiteStorage
log = logging.getLogger(__name__)
COMPARISON_OPERATORS = {
'eq': lambda a, b: a == b,
'ne': lambda a, b: a != b,
'g': lambda a, b: a > b,
'l': lambda a, b: a < b,
'ge': lambda a, b: a >= b,
'le': lambda a, b: a <= b,
}
class SourceManager:
filter_fields = {
'rowid',
'status',
'file_name',
'added_on',
'download_path',
'claim_name',
'claim_height',
'claim_id',
'outpoint',
'txid',
'nout',
'channel_claim_id',
'channel_name',
'completed'
}
set_filter_fields = {
"claim_ids": "claim_id",
"channel_claim_ids": "channel_claim_id",
"outpoints": "outpoint"
}
source_class = ManagedDownloadSource
def __init__(self, loop: asyncio.AbstractEventLoop, config: 'Config', storage: 'SQLiteStorage',
analytics_manager: Optional['AnalyticsManager'] = None):
self.loop = loop
self.config = config
self.storage = storage
self.analytics_manager = analytics_manager
self._sources: typing.Dict[str, ManagedDownloadSource] = {}
self.started = asyncio.Event(loop=self.loop)
def add(self, source: ManagedDownloadSource):
self._sources[source.identifier] = source
def remove(self, source: ManagedDownloadSource):
if source.identifier not in self._sources:
return
self._sources.pop(source.identifier)
source.stop_tasks()
async def initialize_from_database(self):
raise NotImplementedError()
async def start(self):
await self.initialize_from_database()
self.started.set()
def stop(self):
while self._sources:
_, source = self._sources.popitem()
source.stop_tasks()
self.started.clear()
async def create(self, file_path: str, key: Optional[bytes] = None,
iv_generator: Optional[typing.Generator[bytes, None, None]] = None) -> ManagedDownloadSource:
raise NotImplementedError()
async def delete(self, source: ManagedDownloadSource, delete_file: Optional[bool] = False):
self.remove(source)
if delete_file and source.output_file_exists:
os.remove(source.full_path)
def get_filtered(self, sort_by: Optional[str] = None, reverse: Optional[bool] = False,
comparison: Optional[str] = None, **search_by) -> typing.List[ManagedDownloadSource]:
"""
Get a list of filtered and sorted ManagedStream objects
:param sort_by: field to sort by
:param reverse: reverse sorting
:param comparison: comparison operator used for filtering
:param search_by: fields and values to filter by
"""
if sort_by and sort_by not in self.filter_fields:
raise ValueError(f"'{sort_by}' is not a valid field to sort by")
if comparison and comparison not in COMPARISON_OPERATORS:
raise ValueError(f"'{comparison}' is not a valid comparison")
if 'full_status' in search_by:
del search_by['full_status']
for search in search_by:
if search not in self.filter_fields:
raise ValueError(f"'{search}' is not a valid search operation")
compare_sets = {}
if isinstance(search_by.get('claim_id'), list):
compare_sets['claim_ids'] = search_by.pop('claim_id')
if isinstance(search_by.get('outpoint'), list):
compare_sets['outpoints'] = search_by.pop('outpoint')
if isinstance(search_by.get('channel_claim_id'), list):
compare_sets['channel_claim_ids'] = search_by.pop('channel_claim_id')
if search_by or compare_sets:
comparison = comparison or 'eq'
streams = []
for stream in self._sources.values():
if compare_sets and not all(
getattr(stream, self.set_filter_fields[set_search]) in val
for set_search, val in compare_sets.items()):
continue
if search_by and not all(
COMPARISON_OPERATORS[comparison](getattr(stream, search), val)
for search, val in search_by.items()):
continue
streams.append(stream)
else:
streams = list(self._sources.values())
if sort_by:
streams.sort(key=lambda s: getattr(s, sort_by))
if reverse:
streams.reverse()
return streams
+7 -7
View File
@@ -199,8 +199,6 @@ class VideoFileAnalyzer:
if not {"aac", "mp3", "flac", "vorbis", "opus"}.intersection(codec.split(",")):
return "Audio codec is not in the approved list of AAC, FLAC, MP3, Vorbis, and Opus. " \
f"Actual: {codec} [{stream['codec_long_name']}]"
if int(stream['sample_rate']) > 48000:
return "Sample rate out of range"
return ""
@@ -306,6 +304,9 @@ class VideoFileAnalyzer:
raise Exception(f"The audio encoder is not available. Requested: {encoder or 'aac'}")
async def _get_volume_filter(self):
return self._conf.volume_filter if self._conf.volume_filter else "-af loudnorm"
@staticmethod
def _get_best_container_extension(scan_data, video_encoder):
# the container is chosen by the video format
@@ -409,7 +410,7 @@ class VideoFileAnalyzer:
# the plan for transcoding:
# we have to re-encode the video if it is in a nonstandard format
# we also re-encode if we are h264 but not yuv420p (both errors caught in video_msg)
# we also re-encode if our bitrate or sample rate is too high
# we also re-encode if our bitrate is too high
try:
transcode_command = [f'-i "{file_path}" -y -c:s copy -c:d copy -c:v']
@@ -429,10 +430,9 @@ class VideoFileAnalyzer:
if audio_msg or volume_msg:
audio_encoder = await self._get_audio_encoder(extension)
transcode_command.append(audio_encoder)
if volume_msg and self._conf.volume_filter:
transcode_command.append(self._conf.volume_filter)
if audio_msg == "Sample rate out of range":
transcode_command.append(" -ar 48000 ")
if volume_msg:
volume_filter = await self._get_volume_filter()
transcode_command.append(volume_filter)
else:
transcode_command.append("copy")
-68
View File
@@ -1,68 +0,0 @@
import time
import logging
import asyncio
import asyncio.tasks
from aiohttp import web
from prometheus_client import generate_latest as prom_generate_latest
from prometheus_client import Counter, Histogram, Gauge
PROBES_IN_FLIGHT = Counter("probes_in_flight", "Number of loop probes in flight", namespace='asyncio')
PROBES_FINISHED = Counter("probes_finished", "Number of finished loop probes", namespace='asyncio')
PROBE_TIMES = Histogram("probe_times", "Loop probe times", namespace='asyncio')
TASK_COUNT = Gauge("running_tasks", "Number of running tasks", namespace='asyncio')
def get_loop_metrics(delay=1):
loop = asyncio.get_event_loop()
def callback(started):
PROBE_TIMES.observe(time.perf_counter() - started - delay)
PROBES_FINISHED.inc()
async def monitor_loop_responsiveness():
while True:
now = time.perf_counter()
loop.call_later(delay, callback, now)
PROBES_IN_FLIGHT.inc()
TASK_COUNT.set(len(asyncio.tasks._all_tasks))
await asyncio.sleep(delay)
return loop.create_task(monitor_loop_responsiveness())
class PrometheusServer:
def __init__(self, logger=None):
self.runner = None
self.logger = logger or logging.getLogger(__name__)
self._monitor_loop_task = None
async def start(self, interface: str, port: int):
self.logger.info("start prometheus metrics")
prom_app = web.Application()
prom_app.router.add_get('/metrics', self.handle_metrics_get_request)
self.runner = web.AppRunner(prom_app)
await self.runner.setup()
metrics_site = web.TCPSite(self.runner, interface, port, shutdown_timeout=.5)
await metrics_site.start()
self.logger.info(
'prometheus metrics server listening on %s:%i', *metrics_site._server.sockets[0].getsockname()[:2]
)
self._monitor_loop_task = get_loop_metrics()
async def handle_metrics_get_request(self, request: web.Request):
try:
return web.Response(
text=prom_generate_latest().decode(),
content_type='text/plain; version=0.0.4'
)
except Exception:
self.logger.exception('could not generate prometheus data')
raise
async def stop(self):
if self._monitor_loop_task and not self._monitor_loop_task.done():
self._monitor_loop_task.cancel()
self._monitor_loop_task = None
await self.runner.cleanup()
+5 -5
View File
@@ -30,10 +30,14 @@ class Claim(Signable):
COLLECTION = 'collection'
REPOST = 'repost'
__slots__ = ()
__slots__ = 'version',
message_class = ClaimMessage
def __init__(self, message=None):
super().__init__(message)
self.version = 2
@property
def claim_type(self) -> str:
return self.message.WhichOneof('type')
@@ -303,10 +307,6 @@ class Stream(BaseClaim):
def has_fee(self) -> bool:
return self.message.HasField('fee')
@property
def has_source(self) -> bool:
return self.message.HasField('source')
@property
def source(self) -> Source:
return Source(self.message.source)
+2 -6
View File
@@ -46,8 +46,8 @@ types_map = {
'.ksh': ('text/plain', 'document'),
'.latex': ('application/x-latex', 'binary'),
'.m1v': ('video/mpeg', 'video'),
'.m3u': ('application/x-mpegurl', 'audio'),
'.m3u8': ('application/x-mpegurl', 'video'),
'.m3u': ('application/vnd.apple.mpegurl', 'audio'),
'.m3u8': ('application/vnd.apple.mpegurl', 'audio'),
'.man': ('application/x-troff-man', 'document'),
'.markdown': ('text/markdown', 'document'),
'.md': ('text/markdown', 'document'),
@@ -61,12 +61,10 @@ types_map = {
'.mp3': ('audio/mpeg', 'audio'),
'.mp4': ('video/mp4', 'video'),
'.mpa': ('video/mpeg', 'video'),
'.mpd': ('application/dash+xml', 'video'),
'.mpe': ('video/mpeg', 'video'),
'.mpeg': ('video/mpeg', 'video'),
'.mpg': ('video/mpeg', 'video'),
'.ms': ('application/x-troff-ms', 'binary'),
'.m4s': ('video/iso.segment', 'binary'),
'.nc': ('application/x-netcdf', 'binary'),
'.nws': ('message/rfc822', 'document'),
'.o': ('application/octet-stream', 'binary'),
@@ -120,12 +118,10 @@ types_map = {
'.tif': ('image/tiff', 'image'),
'.tiff': ('image/tiff', 'image'),
'.tr': ('application/x-troff', 'binary'),
'.ts': ('video/mp2t', 'video'),
'.tsv': ('text/tab-separated-values', 'document'),
'.txt': ('text/plain', 'document'),
'.ustar': ('application/x-ustar', 'binary'),
'.vcf': ('text/x-vcard', 'document'),
'.vtt': ('text/vtt', 'document'),
'.wav': ('audio/x-wav', 'audio'),
'.webm': ('video/webm', 'video'),
'.wiz': ('application/msword', 'document'),
+38 -34
View File
@@ -13,45 +13,50 @@ NOT_FOUND = ErrorMessage.Code.Name(ErrorMessage.NOT_FOUND)
BLOCKED = ErrorMessage.Code.Name(ErrorMessage.BLOCKED)
def set_reference(reference, txo_row):
if txo_row:
reference.tx_hash = txo_row['txo_hash'][:32]
reference.nout = struct.unpack('<I', txo_row['txo_hash'][32:])[0]
reference.height = txo_row['height']
def set_reference(reference, claim_hash, rows):
if claim_hash:
for txo in rows:
if claim_hash == txo['claim_hash']:
reference.tx_hash = txo['txo_hash'][:32]
reference.nout = struct.unpack('<I', txo['txo_hash'][32:])[0]
reference.height = txo['height']
return
class Censor:
NOT_CENSORED = 0
SEARCH = 1
RESOLVE = 2
__slots__ = 'streams', 'channels', 'censored', 'total'
__slots__ = 'censor_type', 'censored'
def __init__(self, censor_type):
self.censor_type = censor_type
def __init__(self, streams: dict = None, channels: dict = None):
self.streams = streams or {}
self.channels = channels or {}
self.censored = {}
def is_censored(self, row):
return (row.get('censor_type') or self.NOT_CENSORED) >= self.censor_type
def apply(self, rows):
return [row for row in rows if not self.censor(row)]
self.total = 0
def censor(self, row) -> bool:
if self.is_censored(row):
censoring_channel_hash = row['censoring_channel_hash']
self.censored.setdefault(censoring_channel_hash, set())
self.censored[censoring_channel_hash].add(row['tx_hash'])
return True
return False
was_censored = False
for claim_hash, lookup in (
(row['claim_hash'], self.streams),
(row['claim_hash'], self.channels),
(row['channel_hash'], self.channels),
(row['reposted_claim_hash'], self.streams),
(row['reposted_claim_hash'], self.channels)):
censoring_channel_hash = lookup.get(claim_hash)
if censoring_channel_hash:
was_censored = True
self.censored.setdefault(censoring_channel_hash, 0)
self.censored[censoring_channel_hash] += 1
break
if was_censored:
self.total += 1
return was_censored
def to_message(self, outputs: OutputsMessage, extra_txo_rows: dict):
def to_message(self, outputs: OutputsMessage, extra_txo_rows):
outputs.blocked_total = self.total
for censoring_channel_hash, count in self.censored.items():
blocked = outputs.blocked.add()
blocked.count = len(count)
set_reference(blocked.channel, extra_txo_rows.get(censoring_channel_hash))
outputs.blocked_total += len(count)
blocked.count = count
set_reference(blocked.channel, censoring_channel_hash, extra_txo_rows)
class Outputs:
@@ -156,7 +161,6 @@ class Outputs:
@classmethod
def to_bytes(cls, txo_rows, extra_txo_rows, offset=0, total=None, blocked: Censor = None) -> bytes:
extra_txo_rows = {row['claim_hash']: row for row in extra_txo_rows}
page = OutputsMessage()
page.offset = offset
if total is not None:
@@ -165,12 +169,12 @@ class Outputs:
blocked.to_message(page, extra_txo_rows)
for row in txo_rows:
cls.row_to_message(row, page.txos.add(), extra_txo_rows)
for row in extra_txo_rows.values():
for row in extra_txo_rows:
cls.row_to_message(row, page.extra_txos.add(), extra_txo_rows)
return page.SerializeToString()
@classmethod
def row_to_message(cls, txo, txo_message, extra_row_dict: dict):
def row_to_message(cls, txo, txo_message, extra_txo_rows):
if isinstance(txo, Exception):
txo_message.error.text = txo.args[0]
if isinstance(txo, ValueError):
@@ -179,7 +183,7 @@ class Outputs:
txo_message.error.code = ErrorMessage.NOT_FOUND
elif isinstance(txo, ResolveCensoredError):
txo_message.error.code = ErrorMessage.BLOCKED
set_reference(txo_message.error.blocked.channel, extra_row_dict.get(txo.censor_hash))
set_reference(txo_message.error.blocked.channel, txo.censor_hash, extra_txo_rows)
return
txo_message.tx_hash = txo['txo_hash'][:32]
txo_message.nout, = struct.unpack('<I', txo['txo_hash'][32:])
@@ -202,5 +206,5 @@ class Outputs:
txo_message.claim.trending_mixed = txo['trending_mixed']
txo_message.claim.trending_local = txo['trending_local']
txo_message.claim.trending_global = txo['trending_global']
set_reference(txo_message.claim.channel, extra_row_dict.get(txo['channel_hash']))
set_reference(txo_message.claim.repost, extra_row_dict.get(txo['reposted_claim_hash']))
set_reference(txo_message.claim.channel, txo['channel_hash'], extra_txo_rows)
set_reference(txo_message.claim.repost, txo['reposted_claim_hash'], extra_txo_rows)
+1 -10
View File
@@ -1,15 +1,6 @@
from lbry.schema.base import Signable
from lbry.schema.types.v2.support_pb2 import Support as SupportMessage
class Support(Signable):
__slots__ = ()
message_class = SupportMessage
@property
def emoji(self) -> str:
return self.message.emoji
@emoji.setter
def emoji(self, emoji: str):
self.message.emoji = emoji
message_class = None # TODO: add support protobufs
-69
View File
@@ -1,69 +0,0 @@
# Generated by the protocol buffer compiler. DO NOT EDIT!
# source: support.proto
import sys
_b=sys.version_info[0]<3 and (lambda x:x) or (lambda x:x.encode('latin1'))
from google.protobuf import descriptor as _descriptor
from google.protobuf import message as _message
from google.protobuf import reflection as _reflection
from google.protobuf import symbol_database as _symbol_database
from google.protobuf import descriptor_pb2
# @@protoc_insertion_point(imports)
_sym_db = _symbol_database.Default()
DESCRIPTOR = _descriptor.FileDescriptor(
name='support.proto',
package='pb',
syntax='proto3',
serialized_pb=_b('\n\rsupport.proto\x12\x02pb\"\x18\n\x07Support\x12\r\n\x05\x65moji\x18\x01 \x01(\tb\x06proto3')
)
_sym_db.RegisterFileDescriptor(DESCRIPTOR)
_SUPPORT = _descriptor.Descriptor(
name='Support',
full_name='pb.Support',
filename=None,
file=DESCRIPTOR,
containing_type=None,
fields=[
_descriptor.FieldDescriptor(
name='emoji', full_name='pb.Support.emoji', index=0,
number=1, type=9, cpp_type=9, label=1,
has_default_value=False, default_value=_b("").decode('utf-8'),
message_type=None, enum_type=None, containing_type=None,
is_extension=False, extension_scope=None,
options=None),
],
extensions=[
],
nested_types=[],
enum_types=[
],
options=None,
is_extendable=False,
syntax='proto3',
extension_ranges=[],
oneofs=[
],
serialized_start=21,
serialized_end=45,
)
DESCRIPTOR.message_types_by_name['Support'] = _SUPPORT
Support = _reflection.GeneratedProtocolMessageType('Support', (_message.Message,), dict(
DESCRIPTOR = _SUPPORT,
__module__ = 'support_pb2'
# @@protoc_insertion_point(class_scope:pb.Support)
))
_sym_db.RegisterMessage(Support)
# @@protoc_insertion_point(module_scope)
+9 -10
View File
@@ -22,7 +22,8 @@ def _create_url_regex():
return _group(
_named(name+"_name", prefix + invalid_names_regex) +
_oneof(
_group('[:#]' + _named(name+"_claim_id", "[0-9a-f]{1,40}")),
_group('#' + _named(name+"_claim_id", "[0-9a-f]{1,40}")),
_group(':' + _named(name+"_sequence", '[1-9][0-9]*')),
_group(r'\$' + _named(name+"_amount_order", '[1-9][0-9]*'))
) + '?'
)
@@ -49,31 +50,28 @@ def normalize_name(name):
class PathSegment(NamedTuple):
name: str
claim_id: str = None
sequence: int = None
amount_order: int = None
@property
def normalized(self):
return normalize_name(self.name)
@property
def is_shortid(self):
return self.claim_id is not None and len(self.claim_id) < 40
@property
def is_fullid(self):
return self.claim_id is not None and len(self.claim_id) == 40
def to_dict(self):
q = {'name': self.name}
if self.claim_id is not None:
q['claim_id'] = self.claim_id
if self.sequence is not None:
q['sequence'] = self.sequence
if self.amount_order is not None:
q['amount_order'] = self.amount_order
return q
def __str__(self):
if self.claim_id is not None:
return f"{self.name}:{self.claim_id}"
return f"{self.name}#{self.claim_id}"
elif self.sequence is not None:
return f"{self.name}:{self.sequence}"
elif self.amount_order is not None:
return f"{self.name}${self.amount_order}"
return self.name
@@ -120,6 +118,7 @@ class URL(NamedTuple):
segments[segment] = PathSegment(
parts[f'{segment}_name'],
parts[f'{segment}_claim_id'],
parts[f'{segment}_sequence'],
parts[f'{segment}_amount_order']
)
+10 -17
View File
@@ -44,25 +44,18 @@ def random_iv_generator() -> typing.Generator[bytes, None, None]:
yield os.urandom(AES.block_size // 8)
def read_bytes(file_path: str, offset: int, to_read: int):
with open(file_path, 'rb') as f:
f.seek(offset)
return f.read(to_read)
async def file_reader(file_path: str):
def file_reader(file_path: str):
length = int(os.stat(file_path).st_size)
offset = 0
while offset < length:
bytes_to_read = min((length - offset), MAX_BLOB_SIZE - 1)
if not bytes_to_read:
break
blob_bytes = await asyncio.get_event_loop().run_in_executor(
None, read_bytes, file_path, offset, bytes_to_read
)
yield blob_bytes
offset += bytes_to_read
with open(file_path, 'rb') as stream_file:
while offset < length:
bytes_to_read = min((length - offset), MAX_BLOB_SIZE - 1)
if not bytes_to_read:
break
blob_bytes = stream_file.read(bytes_to_read)
yield blob_bytes
offset += bytes_to_read
def sanitize_file_name(dirty_name: str, default_file_name: str = 'lbry_download'):
@@ -252,7 +245,7 @@ class StreamDescriptor:
iv_generator = iv_generator or random_iv_generator()
key = key or os.urandom(AES.block_size // 8)
blob_num = -1
async for blob_bytes in file_reader(file_path):
for blob_bytes in file_reader(file_path):
blob_num += 1
blob_info = await BlobFile.create_from_unencrypted(
loop, blob_dir, key, next(iv_generator), blob_bytes, blob_num, blob_completed_callback
+7 -7
View File
@@ -40,7 +40,7 @@ class StreamDownloader:
async def cached_read_blob(blob_info: 'BlobInfo') -> bytes:
return await self.read_blob(blob_info, 2)
if self.blob_manager.decrypted_blob_lru_cache is not None:
if self.blob_manager.decrypted_blob_lru_cache:
cached_read_blob = lru_cache_concurrent(override_lru_cache=self.blob_manager.decrypted_blob_lru_cache)(
cached_read_blob
)
@@ -51,15 +51,15 @@ class StreamDownloader:
def _delayed_add_fixed_peers():
self.added_fixed_peers = True
self.peer_queue.put_nowait([
make_kademlia_peer(None, address, None, tcp_port=port, allow_localhost=True)
make_kademlia_peer(None, address, None, tcp_port=port + 1, allow_localhost=True)
for address, port in addresses
])
if not self.config.fixed_peers:
if not self.config.reflector_servers:
return
addresses = [
(await resolve_host(url, port, proto='tcp'), port)
for url, port in self.config.fixed_peers
(await resolve_host(url, port + 1, proto='tcp'), port)
for url, port in self.config.reflector_servers
]
if 'dht' in self.config.components_to_skip or not self.node or not \
len(self.node.protocol.routing_table.get_peers()) > 0:
@@ -92,8 +92,8 @@ class StreamDownloader:
async def start(self, node: typing.Optional['Node'] = None, connection_id: int = 0):
# set up peer accumulation
self.node = node or self.node # fixme: this shouldnt be set here!
if self.node:
if node:
self.node = node
if self.accumulate_task and not self.accumulate_task.done():
self.accumulate_task.cancel()
_, self.accumulate_task = self.node.accumulate_peers(self.search_queue, self.peer_queue)
+175 -77
View File
@@ -3,8 +3,9 @@ import asyncio
import time
import typing
import logging
from typing import Optional
import binascii
from aiohttp.web import Request, StreamResponse, HTTPRequestRangeNotSatisfiable
from lbry.utils import generate_id
from lbry.error import DownloadSDTimeoutError
from lbry.schema.mime_types import guess_media_type
from lbry.stream.downloader import StreamDownloader
@@ -12,7 +13,6 @@ from lbry.stream.descriptor import StreamDescriptor, sanitize_file_name
from lbry.stream.reflector.client import StreamReflectorClient
from lbry.extras.daemon.storage import StoredContentClaim
from lbry.blob import MAX_BLOB_SIZE
from lbry.file.source import ManagedDownloadSource
if typing.TYPE_CHECKING:
from lbry.conf import Config
@@ -40,35 +40,80 @@ async def get_next_available_file_name(loop: asyncio.AbstractEventLoop, download
return await loop.run_in_executor(None, _get_next_available_file_name, download_directory, file_name)
class ManagedStream(ManagedDownloadSource):
class ManagedStream:
STATUS_RUNNING = "running"
STATUS_STOPPED = "stopped"
STATUS_FINISHED = "finished"
SAVING_ID = 1
STREAMING_ID = 2
__slots__ = [
'loop',
'config',
'blob_manager',
'sd_hash',
'download_directory',
'_file_name',
'_added_on',
'_status',
'stream_claim_info',
'download_id',
'rowid',
'content_fee',
'purchase_receipt',
'downloader',
'analytics_manager',
'fully_reflected',
'reflector_progress',
'file_output_task',
'delayed_stop_task',
'streaming_responses',
'streaming',
'_running',
'saving',
'finished_writing',
'started_writing',
'finished_write_attempt',
'uploading_to_reflector'
]
def __init__(self, loop: asyncio.AbstractEventLoop, config: 'Config', blob_manager: 'BlobManager',
sd_hash: str, download_directory: Optional[str] = None, file_name: Optional[str] = None,
status: Optional[str] = ManagedDownloadSource.STATUS_STOPPED,
claim: Optional[StoredContentClaim] = None,
download_id: Optional[str] = None, rowid: Optional[int] = None,
descriptor: Optional[StreamDescriptor] = None,
content_fee: Optional['Transaction'] = None,
analytics_manager: Optional['AnalyticsManager'] = None,
added_on: Optional[int] = None):
super().__init__(loop, config, blob_manager.storage, sd_hash, file_name, download_directory, status, claim,
download_id, rowid, content_fee, analytics_manager, added_on)
sd_hash: str, download_directory: typing.Optional[str] = None, file_name: typing.Optional[str] = None,
status: typing.Optional[str] = STATUS_STOPPED, claim: typing.Optional[StoredContentClaim] = None,
download_id: typing.Optional[str] = None, rowid: typing.Optional[int] = None,
descriptor: typing.Optional[StreamDescriptor] = None,
content_fee: typing.Optional['Transaction'] = None,
analytics_manager: typing.Optional['AnalyticsManager'] = None,
added_on: typing.Optional[int] = None):
self.loop = loop
self.config = config
self.blob_manager = blob_manager
self.sd_hash = sd_hash
self.download_directory = download_directory
self._file_name = file_name
self._status = status
self.stream_claim_info = claim
self.download_id = download_id or binascii.hexlify(generate_id()).decode()
self.rowid = rowid
self.content_fee = content_fee
self.purchase_receipt = None
self._added_on = added_on
self.downloader = StreamDownloader(self.loop, self.config, self.blob_manager, sd_hash, descriptor)
self.analytics_manager = analytics_manager
self.fully_reflected = asyncio.Event(loop=self.loop)
self.reflector_progress = 0
self.uploading_to_reflector = False
self.file_output_task: typing.Optional[asyncio.Task] = None
self.delayed_stop_task: typing.Optional[asyncio.Task] = None
self.streaming_responses: typing.List[typing.Tuple[Request, StreamResponse]] = []
self.fully_reflected = asyncio.Event(loop=self.loop)
self.streaming = asyncio.Event(loop=self.loop)
self._running = asyncio.Event(loop=self.loop)
@property
def sd_hash(self) -> str:
return self.identifier
self.saving = asyncio.Event(loop=self.loop)
self.finished_writing = asyncio.Event(loop=self.loop)
self.started_writing = asyncio.Event(loop=self.loop)
self.finished_write_attempt = asyncio.Event(loop=self.loop)
@property
def is_fully_reflected(self) -> bool:
@@ -83,9 +128,17 @@ class ManagedStream(ManagedDownloadSource):
return self.descriptor.stream_hash
@property
def file_name(self) -> Optional[str]:
def file_name(self) -> typing.Optional[str]:
return self._file_name or (self.descriptor.suggested_file_name if self.descriptor else None)
@property
def added_on(self) -> typing.Optional[int]:
return self._added_on
@property
def status(self) -> str:
return self._status
@property
def written_bytes(self) -> int:
return 0 if not self.output_file_exists else os.stat(self.full_path).st_size
@@ -103,6 +156,55 @@ class ManagedStream(ManagedDownloadSource):
self._status = status
await self.blob_manager.storage.change_file_status(self.stream_hash, status)
@property
def finished(self) -> bool:
return self.status == self.STATUS_FINISHED
@property
def running(self) -> bool:
return self.status == self.STATUS_RUNNING
@property
def claim_id(self) -> typing.Optional[str]:
return None if not self.stream_claim_info else self.stream_claim_info.claim_id
@property
def txid(self) -> typing.Optional[str]:
return None if not self.stream_claim_info else self.stream_claim_info.txid
@property
def nout(self) -> typing.Optional[int]:
return None if not self.stream_claim_info else self.stream_claim_info.nout
@property
def outpoint(self) -> typing.Optional[str]:
return None if not self.stream_claim_info else self.stream_claim_info.outpoint
@property
def claim_height(self) -> typing.Optional[int]:
return None if not self.stream_claim_info else self.stream_claim_info.height
@property
def channel_claim_id(self) -> typing.Optional[str]:
return None if not self.stream_claim_info else self.stream_claim_info.channel_claim_id
@property
def channel_name(self) -> typing.Optional[str]:
return None if not self.stream_claim_info else self.stream_claim_info.channel_name
@property
def claim_name(self) -> typing.Optional[str]:
return None if not self.stream_claim_info else self.stream_claim_info.claim_name
@property
def metadata(self) -> typing.Optional[typing.Dict]:
return None if not self.stream_claim_info else self.stream_claim_info.claim.stream.to_dict()
@property
def metadata_protobuf(self) -> bytes:
if self.stream_claim_info:
return binascii.hexlify(self.stream_claim_info.claim.to_bytes())
@property
def blobs_completed(self) -> int:
return sum([1 if b.blob_hash in self.blob_manager.completed_blob_hashes else 0
@@ -116,34 +218,39 @@ class ManagedStream(ManagedDownloadSource):
def blobs_remaining(self) -> int:
return self.blobs_in_stream - self.blobs_completed
@property
def full_path(self) -> typing.Optional[str]:
return os.path.join(self.download_directory, os.path.basename(self.file_name)) \
if self.file_name and self.download_directory else None
@property
def output_file_exists(self):
return os.path.isfile(self.full_path) if self.full_path else False
@property
def mime_type(self):
return guess_media_type(os.path.basename(self.descriptor.suggested_file_name))[0]
@property
def download_path(self):
return f"{self.download_directory}/{self._file_name}" if self.download_directory and self._file_name else None
@classmethod
async def create(cls, loop: asyncio.AbstractEventLoop, config: 'Config', blob_manager: 'BlobManager',
file_path: str, key: typing.Optional[bytes] = None,
iv_generator: typing.Optional[typing.Generator[bytes, None, None]] = None) -> 'ManagedStream':
"""
Generate a stream from a file and save it to the db
"""
descriptor = await StreamDescriptor.create_stream(
loop, blob_manager.blob_dir, file_path, key=key, iv_generator=iv_generator,
blob_completed_callback=blob_manager.blob_completed
)
await blob_manager.storage.store_stream(
blob_manager.get_blob(descriptor.sd_hash), descriptor
)
row_id = await blob_manager.storage.save_published_file(descriptor.stream_hash, os.path.basename(file_path),
os.path.dirname(file_path), 0)
return cls(loop, config, blob_manager, descriptor.sd_hash, os.path.dirname(file_path),
os.path.basename(file_path), status=cls.STATUS_FINISHED, rowid=row_id, descriptor=descriptor)
# @classmethod
# async def create(cls, loop: asyncio.AbstractEventLoop, config: 'Config',
# file_path: str, key: Optional[bytes] = None,
# iv_generator: Optional[typing.Generator[bytes, None, None]] = None) -> 'ManagedDownloadSource':
# """
# Generate a stream from a file and save it to the db
# """
# descriptor = await StreamDescriptor.create_stream(
# loop, blob_manager.blob_dir, file_path, key=key, iv_generator=iv_generator,
# blob_completed_callback=blob_manager.blob_completed
# )
# await blob_manager.storage.store_stream(
# blob_manager.get_blob(descriptor.sd_hash), descriptor
# )
# row_id = await blob_manager.storage.save_published_file(descriptor.stream_hash, os.path.basename(file_path),
# os.path.dirname(file_path), 0)
# return cls(loop, config, blob_manager, descriptor.sd_hash, os.path.dirname(file_path),
# os.path.basename(file_path), status=cls.STATUS_FINISHED, rowid=row_id, descriptor=descriptor)
async def start(self, timeout: Optional[float] = None,
async def start(self, node: typing.Optional['Node'] = None, timeout: typing.Optional[float] = None,
save_now: bool = False):
timeout = timeout or self.config.download_timeout
if self._running.is_set():
@@ -151,7 +258,7 @@ class ManagedStream(ManagedDownloadSource):
log.info("start downloader for stream (sd hash: %s)", self.sd_hash)
self._running.set()
try:
await asyncio.wait_for(self.downloader.start(), timeout, loop=self.loop)
await asyncio.wait_for(self.downloader.start(node), timeout, loop=self.loop)
except asyncio.TimeoutError:
self._running.clear()
raise DownloadSDTimeoutError(self.sd_hash)
@@ -161,11 +268,6 @@ class ManagedStream(ManagedDownloadSource):
self.delayed_stop_task = self.loop.create_task(self._delayed_stop())
if not await self.blob_manager.storage.file_exists(self.sd_hash):
if save_now:
if not self._file_name:
self._file_name = await get_next_available_file_name(
self.loop, self.download_directory,
self._file_name or sanitize_file_name(self.descriptor.suggested_file_name)
)
file_name, download_dir = self._file_name, self.download_directory
else:
file_name, download_dir = None, None
@@ -185,7 +287,7 @@ class ManagedStream(ManagedDownloadSource):
if (finished and self.status != self.STATUS_FINISHED) or self.status == self.STATUS_RUNNING:
await self.update_status(self.STATUS_FINISHED if finished else self.STATUS_STOPPED)
async def _aiter_read_stream(self, start_blob_num: Optional[int] = 0, connection_id: int = 0)\
async def _aiter_read_stream(self, start_blob_num: typing.Optional[int] = 0, connection_id: int = 0)\
-> typing.AsyncIterator[typing.Tuple['BlobInfo', bytes]]:
if start_blob_num >= len(self.descriptor.blobs[:-1]):
raise IndexError(start_blob_num)
@@ -197,13 +299,13 @@ class ManagedStream(ManagedDownloadSource):
decrypted = await self.downloader.read_blob(blob_info, connection_id)
yield (blob_info, decrypted)
async def stream_file(self, request: Request) -> StreamResponse:
async def stream_file(self, request: Request, node: typing.Optional['Node'] = None) -> StreamResponse:
log.info("stream file to browser for lbry://%s#%s (sd hash %s...)", self.claim_name, self.claim_id,
self.sd_hash[:6])
headers, size, skip_blobs, first_blob_start_offset = self._prepare_range_response_headers(
request.headers.get('range', 'bytes=0-')
)
await self.start()
await self.start(node)
response = StreamResponse(
status=206,
headers=headers
@@ -241,10 +343,9 @@ class ManagedStream(ManagedDownloadSource):
self.streaming.clear()
@staticmethod
def _write_decrypted_blob(output_path: str, data: bytes):
with open(output_path, 'ab') as handle:
handle.write(data)
handle.flush()
def _write_decrypted_blob(handle: typing.IO, data: bytes):
handle.write(data)
handle.flush()
async def _save_file(self, output_path: str):
log.info("save file for lbry://%s#%s (sd hash %s...) -> %s", self.claim_name, self.claim_id, self.sd_hash[:6],
@@ -254,12 +355,12 @@ class ManagedStream(ManagedDownloadSource):
self.finished_writing.clear()
self.started_writing.clear()
try:
open(output_path, 'wb').close()
async for blob_info, decrypted in self._aiter_read_stream(connection_id=self.SAVING_ID):
log.info("write blob %i/%i", blob_info.blob_num + 1, len(self.descriptor.blobs) - 1)
await self.loop.run_in_executor(None, self._write_decrypted_blob, output_path, decrypted)
if not self.started_writing.is_set():
self.started_writing.set()
with open(output_path, 'wb') as file_write_handle:
async for blob_info, decrypted in self._aiter_read_stream(connection_id=self.SAVING_ID):
log.info("write blob %i/%i", blob_info.blob_num + 1, len(self.descriptor.blobs) - 1)
await self.loop.run_in_executor(None, self._write_decrypted_blob, file_write_handle, decrypted)
if not self.started_writing.is_set():
self.started_writing.set()
await self.update_status(ManagedStream.STATUS_FINISHED)
if self.analytics_manager:
self.loop.create_task(self.analytics_manager.send_download_finished(
@@ -289,8 +390,9 @@ class ManagedStream(ManagedDownloadSource):
self.saving.clear()
self.finished_write_attempt.set()
async def save_file(self, file_name: Optional[str] = None, download_directory: Optional[str] = None):
await self.start()
async def save_file(self, file_name: typing.Optional[str] = None, download_directory: typing.Optional[str] = None,
node: typing.Optional['Node'] = None):
await self.start(node)
if self.file_output_task and not self.file_output_task.done(): # cancel an already running save task
self.file_output_task.cancel()
self.download_directory = download_directory or self.download_directory or self.config.download_dir
@@ -354,30 +456,26 @@ class ManagedStream(ManagedDownloadSource):
self.reflector_progress = int((i + 1) / len(we_have) * 100)
except (asyncio.TimeoutError, ValueError):
return sent
except ConnectionError:
return sent
except (OSError, Exception) as err:
if isinstance(err, asyncio.CancelledError):
log.warning("stopped uploading %s#%s to reflector", self.claim_name, self.claim_id)
elif isinstance(err, OSError):
log.warning(
"stopped uploading %s#%s to reflector because blobs were deleted or moved", self.claim_name,
self.claim_id
)
else:
log.exception("unexpected error reflecting %s#%s", self.claim_name, self.claim_id)
except ConnectionRefusedError:
return sent
finally:
if protocol.transport:
protocol.transport.close()
self.uploading_to_reflector = False
if not self.fully_reflected.is_set():
self.fully_reflected.set()
await self.blob_manager.storage.update_reflected_stream(self.sd_hash, f"{host}:{port}")
return sent
async def update_content_claim(self, claim_info: Optional[typing.Dict] = None):
def set_claim(self, claim_info: typing.Dict, claim: 'Claim'):
self.stream_claim_info = StoredContentClaim(
f"{claim_info['txid']}:{claim_info['nout']}", claim_info['claim_id'],
claim_info['name'], claim_info['amount'], claim_info['height'],
binascii.hexlify(claim.to_bytes()).decode(), claim.signing_channel_id, claim_info['address'],
claim_info['claim_sequence'], claim_info.get('channel_name')
)
async def update_content_claim(self, claim_info: typing.Optional[typing.Dict] = None):
if not claim_info:
claim_info = await self.blob_manager.storage.get_content_claim(self.stream_hash)
self.set_claim(claim_info, claim_info['value'])
+7 -31
View File
@@ -35,8 +35,6 @@ class StreamReflectorClient(asyncio.Protocol):
def connection_lost(self, exc: typing.Optional[Exception]):
self.transport = None
self.connected.clear()
if self.pending_request:
self.pending_request.cancel()
if self.reflected_blobs:
log.info("Finished sending reflector %i blobs", len(self.reflected_blobs))
@@ -58,18 +56,12 @@ class StreamReflectorClient(asyncio.Protocol):
self.response_buff = b''
return
async def send_request(self, request_dict: typing.Dict, timeout: int = 180):
async def send_request(self, request_dict: typing.Dict):
msg = json.dumps(request_dict)
self.transport.write(msg.encode())
try:
self.transport.write(msg.encode())
self.pending_request = self.loop.create_task(asyncio.wait_for(self.response_queue.get(), timeout))
self.pending_request = self.loop.create_task(self.response_queue.get())
return await self.pending_request
except (AttributeError, asyncio.CancelledError) as err:
# attribute error happens when we transport.write after disconnect
# cancelled error happens when the pending_request task is cancelled by a disconnect
if self.transport:
self.transport.close()
raise err if isinstance(err, asyncio.CancelledError) else asyncio.CancelledError()
finally:
self.pending_request = None
@@ -94,16 +86,8 @@ class StreamReflectorClient(asyncio.Protocol):
needed = response.get('needed_blobs', [])
sent_sd = False
if response['send_sd_blob']:
try:
sent = await sd_blob.sendfile(self)
if sent == -1:
log.warning("failed to send sd blob")
raise asyncio.CancelledError()
received = await asyncio.wait_for(self.response_queue.get(), 30)
except asyncio.CancelledError as err:
if self.transport:
self.transport.close()
raise err
await sd_blob.sendfile(self)
received = await self.response_queue.get()
if received.get('received_sd_blob'):
sent_sd = True
if not needed:
@@ -126,16 +110,8 @@ class StreamReflectorClient(asyncio.Protocol):
if 'send_blob' not in response:
raise ValueError("I don't know whether to send the blob or not!")
if response['send_blob']:
try:
sent = await blob.sendfile(self)
if sent == -1:
log.warning("failed to send blob")
raise asyncio.CancelledError()
received = await asyncio.wait_for(self.response_queue.get(), 30)
except asyncio.CancelledError as err:
if self.transport:
self.transport.close()
raise err
await blob.sendfile(self)
received = await self.response_queue.get()
if received.get('received_blob'):
self.reflected_blobs.append(blob.blob_hash)
log.info("Sent reflector blob %s", blob.blob_hash[:8])
+6 -40
View File
@@ -15,9 +15,7 @@ log = logging.getLogger(__name__)
class ReflectorServerProtocol(asyncio.Protocol):
def __init__(self, blob_manager: 'BlobManager', response_chunk_size: int = 10000,
stop_event: asyncio.Event = None, incoming_event: asyncio.Event = None,
not_incoming_event: asyncio.Event = None):
def __init__(self, blob_manager: 'BlobManager', response_chunk_size: int = 10000):
self.loop = asyncio.get_event_loop()
self.blob_manager = blob_manager
self.server_task: asyncio.Task = None
@@ -29,25 +27,11 @@ class ReflectorServerProtocol(asyncio.Protocol):
self.descriptor: typing.Optional['StreamDescriptor'] = None
self.sd_blob: typing.Optional['BlobFile'] = None
self.received = []
self.incoming = incoming_event or asyncio.Event(loop=self.loop)
self.not_incoming = not_incoming_event or asyncio.Event(loop=self.loop)
self.stop_event = stop_event or asyncio.Event(loop=self.loop)
self.incoming = asyncio.Event(loop=self.loop)
self.chunk_size = response_chunk_size
self.wait_for_stop_task: typing.Optional[asyncio.Task] = None
async def wait_for_stop(self):
await self.stop_event.wait()
if self.transport:
self.transport.close()
def connection_made(self, transport):
self.transport = transport
self.wait_for_stop_task = self.loop.create_task(self.wait_for_stop())
def connection_lost(self, exc):
if self.wait_for_stop_task:
self.wait_for_stop_task.cancel()
self.wait_for_stop_task = None
def data_received(self, data: bytes):
if self.incoming.is_set():
@@ -89,7 +73,6 @@ class ReflectorServerProtocol(asyncio.Protocol):
self.sd_blob = self.blob_manager.get_blob(request['sd_blob_hash'], request['sd_blob_size'])
if not self.sd_blob.get_is_verified():
self.writer = self.sd_blob.get_blob_writer(self.transport.get_extra_info('peername'))
self.not_incoming.clear()
self.incoming.set()
self.send_response({"send_sd_blob": True})
try:
@@ -103,7 +86,6 @@ class ReflectorServerProtocol(asyncio.Protocol):
self.transport.close()
finally:
self.incoming.clear()
self.not_incoming.set()
self.writer.close_handle()
self.writer = None
else:
@@ -111,7 +93,6 @@ class ReflectorServerProtocol(asyncio.Protocol):
self.loop, self.blob_manager.blob_dir, self.sd_blob
)
self.incoming.clear()
self.not_incoming.set()
if self.writer:
self.writer.close_handle()
self.writer = None
@@ -131,7 +112,6 @@ class ReflectorServerProtocol(asyncio.Protocol):
blob = self.blob_manager.get_blob(request['blob_hash'], request['blob_size'])
if not blob.get_is_verified():
self.writer = blob.get_blob_writer(self.transport.get_extra_info('peername'))
self.not_incoming.clear()
self.incoming.set()
self.send_response({"send_blob": True})
try:
@@ -140,7 +120,6 @@ class ReflectorServerProtocol(asyncio.Protocol):
except asyncio.TimeoutError:
self.send_response({"received_blob": False})
self.incoming.clear()
self.not_incoming.set()
self.writer.close_handle()
self.writer = None
else:
@@ -151,18 +130,12 @@ class ReflectorServerProtocol(asyncio.Protocol):
class ReflectorServer:
def __init__(self, blob_manager: 'BlobManager', response_chunk_size: int = 10000,
stop_event: asyncio.Event = None, incoming_event: asyncio.Event = None,
not_incoming_event: asyncio.Event = None):
def __init__(self, blob_manager: 'BlobManager', response_chunk_size: int = 10000):
self.loop = asyncio.get_event_loop()
self.blob_manager = blob_manager
self.server_task: typing.Optional[asyncio.Task] = None
self.started_listening = asyncio.Event(loop=self.loop)
self.stopped_listening = asyncio.Event(loop=self.loop)
self.incoming_event = incoming_event or asyncio.Event(loop=self.loop)
self.not_incoming_event = not_incoming_event or asyncio.Event(loop=self.loop)
self.response_chunk_size = response_chunk_size
self.stop_event = stop_event
def start_server(self, port: int, interface: typing.Optional[str] = '0.0.0.0'):
if self.server_task is not None:
@@ -170,20 +143,13 @@ class ReflectorServer:
async def _start_server():
server = await self.loop.create_server(
lambda: ReflectorServerProtocol(
self.blob_manager, self.response_chunk_size, self.stop_event, self.incoming_event,
self.not_incoming_event
),
lambda: ReflectorServerProtocol(self.blob_manager, self.response_chunk_size),
interface, port
)
self.started_listening.set()
self.stopped_listening.clear()
log.info("Reflector server listening on TCP %s:%i", interface, port)
try:
async with server:
await server.serve_forever()
finally:
self.stopped_listening.set()
async with server:
await server.serve_forever()
self.server_task = self.loop.create_task(_start_server())
+351 -96
View File
@@ -6,67 +6,93 @@ import random
import typing
from typing import Optional
from aiohttp.web import Request
from lbry.error import InvalidStreamDescriptorError
from lbry.file.source_manager import SourceManager
from lbry.error import ResolveError, InvalidStreamDescriptorError, DownloadSDTimeoutError, InsufficientFundsError
from lbry.error import ResolveTimeoutError, DownloadDataTimeoutError, KeyFeeAboveMaxAllowedError
from lbry.utils import cache_concurrent
from lbry.stream.descriptor import StreamDescriptor
from lbry.stream.managed_stream import ManagedStream
from lbry.file.source import ManagedDownloadSource
from lbry.schema.claim import Claim
from lbry.schema.url import URL
from lbry.wallet.dewies import dewies_to_lbc
from lbry.wallet import Output
if typing.TYPE_CHECKING:
from lbry.conf import Config
from lbry.blob.blob_manager import BlobManager
from lbry.dht.node import Node
from lbry.wallet.wallet import WalletManager
from lbry.wallet.transaction import Transaction
from lbry.extras.daemon.analytics import AnalyticsManager
from lbry.extras.daemon.storage import SQLiteStorage, StoredContentClaim
from lbry.extras.daemon.exchange_rate_manager import ExchangeRateManager
from lbry.wallet.transaction import Transaction
from lbry.wallet.manager import WalletManager
from lbry.wallet.wallet import Wallet
log = logging.getLogger(__name__)
FILTER_FIELDS = [
'rowid',
'status',
'file_name',
'added_on',
'sd_hash',
'stream_hash',
'claim_name',
'claim_height',
'claim_id',
'outpoint',
'txid',
'nout',
'channel_claim_id',
'channel_name',
'full_status', # TODO: remove
'blobs_remaining',
'blobs_in_stream'
]
def path_or_none(encoded_path) -> Optional[str]:
if not encoded_path:
SET_FILTER_FIELDS = {
"claim_ids": "claim_id",
"channel_claim_ids": "channel_claim_id",
"outpoints": "outpoint"
}
COMPARISON_OPERATORS = {
'eq': lambda a, b: a == b,
'ne': lambda a, b: a != b,
'g': lambda a, b: a > b,
'l': lambda a, b: a < b,
'ge': lambda a, b: a >= b,
'le': lambda a, b: a <= b,
'in': lambda a, b: a in b
}
def path_or_none(path) -> Optional[str]:
if not path:
return
return binascii.unhexlify(encoded_path).decode()
return binascii.unhexlify(path).decode()
class StreamManager(SourceManager):
_sources: typing.Dict[str, ManagedStream]
filter_fields = SourceManager.filter_fields
filter_fields.update({
'sd_hash',
'stream_hash',
'full_status', # TODO: remove
'blobs_remaining',
'blobs_in_stream',
'uploading_to_reflector',
'is_fully_reflected'
})
class StreamManager:
def __init__(self, loop: asyncio.AbstractEventLoop, config: 'Config', blob_manager: 'BlobManager',
wallet_manager: 'WalletManager', storage: 'SQLiteStorage', node: Optional['Node'],
analytics_manager: Optional['AnalyticsManager'] = None):
super().__init__(loop, config, storage, analytics_manager)
self.loop = loop
self.config = config
self.blob_manager = blob_manager
self.wallet_manager = wallet_manager
self.storage = storage
self.node = node
self.analytics_manager = analytics_manager
self.streams: typing.Dict[str, ManagedStream] = {}
self.resume_saving_task: Optional[asyncio.Task] = None
self.re_reflect_task: Optional[asyncio.Task] = None
self.update_stream_finished_futs: typing.List[asyncio.Future] = []
self.running_reflector_uploads: typing.Dict[str, asyncio.Task] = {}
self.started = asyncio.Event(loop=self.loop)
@property
def streams(self):
return self._sources
def add(self, source: ManagedStream):
super().add(source)
self.storage.content_claim_callbacks[source.stream_hash] = lambda: self._update_content_claim(source)
async def _update_content_claim(self, stream: ManagedStream):
claim_info = await self.storage.get_content_claim(stream.stream_hash)
self._sources.setdefault(stream.sd_hash, stream).set_claim(claim_info, claim_info['value'])
self.streams.setdefault(stream.sd_hash, stream).set_claim(claim_info, claim_info['value'])
async def recover_streams(self, file_infos: typing.List[typing.Dict]):
to_restore = []
@@ -97,10 +123,10 @@ class StreamManager(SourceManager):
# if self.blob_manager._save_blobs:
# log.info("Recovered %i/%i attempted streams", len(to_restore), len(file_infos))
async def _load_stream(self, rowid: int, sd_hash: str, file_name: Optional[str],
download_directory: Optional[str], status: str,
claim: Optional['StoredContentClaim'], content_fee: Optional['Transaction'],
added_on: Optional[int], fully_reflected: Optional[bool]):
async def add_stream(self, rowid: int, sd_hash: str, file_name: Optional[str],
download_directory: Optional[str], status: str,
claim: Optional['StoredContentClaim'], content_fee: Optional['Transaction'],
added_on: Optional[int], fully_reflected: bool):
try:
descriptor = await self.blob_manager.get_stream_descriptor(sd_hash)
except InvalidStreamDescriptorError as err:
@@ -113,9 +139,10 @@ class StreamManager(SourceManager):
)
if fully_reflected:
stream.fully_reflected.set()
self.add(stream)
self.streams[sd_hash] = stream
self.storage.content_claim_callbacks[stream.stream_hash] = lambda: self._update_content_claim(stream)
async def initialize_from_database(self):
async def load_and_resume_streams_from_database(self):
to_recover = []
to_start = []
@@ -129,6 +156,7 @@ class StreamManager(SourceManager):
to_recover.append(file_info)
to_start.append(file_info)
if to_recover:
log.info("Recover %i files", len(to_recover))
await self.recover_streams(to_recover)
log.info("Initializing %i files", len(to_start))
@@ -139,7 +167,7 @@ class StreamManager(SourceManager):
download_directory = path_or_none(file_info['download_directory'])
if file_name and download_directory and not file_info['saved_file'] and file_info['status'] == 'running':
to_resume_saving.append((file_name, download_directory, file_info['sd_hash']))
add_stream_tasks.append(self.loop.create_task(self._load_stream(
add_stream_tasks.append(self.loop.create_task(self.add_stream(
file_info['rowid'], file_info['sd_hash'], file_name,
download_directory, file_info['status'],
file_info['claim'], file_info['content_fee'],
@@ -147,32 +175,25 @@ class StreamManager(SourceManager):
)))
if add_stream_tasks:
await asyncio.gather(*add_stream_tasks, loop=self.loop)
log.info("Started stream manager with %i files", len(self._sources))
log.info("Started stream manager with %i files", len(self.streams))
if not self.node:
log.info("no DHT node given, resuming downloads trusting that we can contact reflector")
if to_resume_saving:
log.info("Resuming saving %i files", len(to_resume_saving))
self.resume_saving_task = asyncio.ensure_future(asyncio.gather(
*(self._sources[sd_hash].save_file(file_name, download_directory)
for (file_name, download_directory, sd_hash) in to_resume_saving),
loop=self.loop
))
self.resume_saving_task = self.loop.create_task(self.resume(to_resume_saving))
async def resume(self, to_resume_saving):
log.info("Resuming saving %i files", len(to_resume_saving))
await asyncio.gather(
*(self.streams[sd_hash].save_file(file_name, download_directory, node=self.node)
for (file_name, download_directory, sd_hash) in to_resume_saving),
loop=self.loop
)
async def reflect_streams(self):
try:
return await self._reflect_streams()
except asyncio.CancelledError:
raise
except Exception:
log.exception("reflector task encountered an unexpected error!")
async def _reflect_streams(self):
# todo: those debug statements are temporary for #2987 - remove them if its closed
while True:
if self.config.reflect_streams and self.config.reflector_servers:
log.debug("collecting streams to reflect")
sd_hashes = await self.storage.get_streams_to_re_reflect()
sd_hashes = [sd for sd in sd_hashes if sd in self._sources]
sd_hashes = [sd for sd in sd_hashes if sd in self.streams]
batch = []
while sd_hashes:
stream = self.streams[sd_hashes.pop()]
@@ -181,26 +202,25 @@ class StreamManager(SourceManager):
stream.fully_reflected.is_set():
batch.append(self.reflect_stream(stream))
if len(batch) >= self.config.concurrent_reflector_uploads:
log.debug("waiting for batch of %s reflecting streams", len(batch))
await asyncio.gather(*batch, loop=self.loop)
log.debug("done processing %s streams", len(batch))
batch = []
if batch:
log.debug("waiting for batch of %s reflecting streams", len(batch))
await asyncio.gather(*batch, loop=self.loop)
log.debug("done processing %s streams", len(batch))
await asyncio.sleep(300, loop=self.loop)
async def start(self):
await super().start()
await self.load_and_resume_streams_from_database()
self.re_reflect_task = self.loop.create_task(self.reflect_streams())
self.started.set()
def stop(self):
super().stop()
if self.resume_saving_task and not self.resume_saving_task.done():
self.resume_saving_task.cancel()
if self.re_reflect_task and not self.re_reflect_task.done():
self.re_reflect_task.cancel()
while self.streams:
_, stream = self.streams.popitem()
stream.stop_tasks()
while self.update_stream_finished_futs:
self.update_stream_finished_futs.pop().cancel()
while self.running_reflector_uploads:
@@ -223,45 +243,280 @@ class StreamManager(SourceManager):
)
return task
async def create(self, file_path: str, key: Optional[bytes] = None,
iv_generator: Optional[typing.Generator[bytes, None, None]] = None) -> ManagedStream:
descriptor = await StreamDescriptor.create_stream(
self.loop, self.blob_manager.blob_dir, file_path, key=key, iv_generator=iv_generator,
blob_completed_callback=self.blob_manager.blob_completed
)
await self.storage.store_stream(
self.blob_manager.get_blob(descriptor.sd_hash), descriptor
)
row_id = await self.storage.save_published_file(
descriptor.stream_hash, os.path.basename(file_path), os.path.dirname(file_path), 0
)
stream = ManagedStream(
self.loop, self.config, self.blob_manager, descriptor.sd_hash, os.path.dirname(file_path),
os.path.basename(file_path), status=ManagedDownloadSource.STATUS_FINISHED,
rowid=row_id, descriptor=descriptor
)
async def create_stream(self, file_path: str, key: Optional[bytes] = None,
iv_generator: Optional[typing.Generator[bytes, None, None]] = None) -> ManagedStream:
stream = await ManagedStream.create(self.loop, self.config, self.blob_manager, file_path, key, iv_generator)
self.streams[stream.sd_hash] = stream
self.storage.content_claim_callbacks[stream.stream_hash] = lambda: self._update_content_claim(stream)
if self.config.reflect_streams and self.config.reflector_servers:
self.reflect_stream(stream)
return stream
async def delete(self, source: ManagedDownloadSource, delete_file: Optional[bool] = False):
if not isinstance(source, ManagedStream):
return
if source.identifier in self.running_reflector_uploads:
self.running_reflector_uploads[source.identifier].cancel()
source.stop_tasks()
if source.identifier in self.streams:
del self.streams[source.identifier]
blob_hashes = [source.identifier] + [b.blob_hash for b in source.descriptor.blobs[:-1]]
async def delete_stream(self, stream: ManagedStream, delete_file: Optional[bool] = False):
if stream.sd_hash in self.running_reflector_uploads:
self.running_reflector_uploads[stream.sd_hash].cancel()
stream.stop_tasks()
if stream.sd_hash in self.streams:
del self.streams[stream.sd_hash]
blob_hashes = [stream.sd_hash] + [b.blob_hash for b in stream.descriptor.blobs[:-1]]
await self.blob_manager.delete_blobs(blob_hashes, delete_from_db=False)
await self.storage.delete_stream(source.descriptor)
if delete_file and source.output_file_exists:
os.remove(source.full_path)
await self.storage.delete_stream(stream.descriptor)
if delete_file and stream.output_file_exists:
os.remove(stream.full_path)
def get_stream_by_stream_hash(self, stream_hash: str) -> Optional[ManagedStream]:
streams = tuple(filter(lambda stream: stream.stream_hash == stream_hash, self.streams.values()))
if streams:
return streams[0]
def get_filtered_streams(self, sort_by: Optional[str] = None, reverse: Optional[bool] = False,
comparison: Optional[str] = None,
**search_by) -> typing.List[ManagedStream]:
"""
Get a list of filtered and sorted ManagedStream objects
:param sort_by: field to sort by
:param reverse: reverse sorting
:param comparison: comparison operator used for filtering
:param search_by: fields and values to filter by
"""
if sort_by and sort_by not in FILTER_FIELDS:
raise ValueError(f"'{sort_by}' is not a valid field to sort by")
if comparison and comparison not in COMPARISON_OPERATORS:
raise ValueError(f"'{comparison}' is not a valid comparison")
if 'full_status' in search_by:
del search_by['full_status']
for search in search_by:
if search not in FILTER_FIELDS:
raise ValueError(f"'{search}' is not a valid search operation")
compare_sets = {}
if isinstance(search_by.get('claim_id'), list):
compare_sets['claim_ids'] = search_by.pop('claim_id')
if isinstance(search_by.get('outpoint'), list):
compare_sets['outpoints'] = search_by.pop('outpoint')
if isinstance(search_by.get('channel_claim_id'), list):
compare_sets['channel_claim_ids'] = search_by.pop('channel_claim_id')
if search_by:
comparison = comparison or 'eq'
streams = []
for stream in self.streams.values():
matched = False
for set_search, val in compare_sets.items():
if COMPARISON_OPERATORS[comparison](getattr(stream, SET_FILTER_FIELDS[set_search]), val):
streams.append(stream)
matched = True
break
if matched:
continue
for search, val in search_by.items():
this_stream = getattr(stream, search)
if COMPARISON_OPERATORS[comparison](this_stream, val):
streams.append(stream)
break
else:
streams = list(self.streams.values())
if sort_by:
streams.sort(key=lambda s: getattr(s, sort_by))
if reverse:
streams.reverse()
return streams
async def _check_update_or_replace(self, outpoint: str, claim_id: str, claim: Claim
) -> typing.Tuple[Optional[ManagedStream], Optional[ManagedStream]]:
existing = self.get_filtered_streams(outpoint=outpoint)
if existing:
return existing[0], None
existing = self.get_filtered_streams(sd_hash=claim.stream.source.sd_hash)
if existing and existing[0].claim_id != claim_id:
raise ResolveError(f"stream for {existing[0].claim_id} collides with existing download {claim_id}")
if existing:
log.info("claim contains a metadata only update to a stream we have")
await self.storage.save_content_claim(
existing[0].stream_hash, outpoint
)
await self._update_content_claim(existing[0])
return existing[0], None
else:
existing_for_claim_id = self.get_filtered_streams(claim_id=claim_id)
if existing_for_claim_id:
log.info("claim contains an update to a stream we have, downloading it")
return None, existing_for_claim_id[0]
return None, None
@staticmethod
def _convert_to_old_resolve_output(wallet_manager, resolves):
result = {}
for url, txo in resolves.items():
if isinstance(txo, Output):
tx_height = txo.tx_ref.height
best_height = wallet_manager.ledger.headers.height
result[url] = {
'name': txo.claim_name,
'value': txo.claim,
'protobuf': binascii.hexlify(txo.claim.to_bytes()),
'claim_id': txo.claim_id,
'txid': txo.tx_ref.id,
'nout': txo.position,
'amount': dewies_to_lbc(txo.amount),
'effective_amount': txo.meta.get('effective_amount', 0),
'height': tx_height,
'confirmations': (best_height+1) - tx_height if tx_height > 0 else tx_height,
'claim_sequence': -1,
'address': txo.get_address(wallet_manager.ledger),
'valid_at_height': txo.meta.get('activation_height', None),
'timestamp': wallet_manager.ledger.headers.estimated_timestamp(tx_height),
'supports': []
}
else:
result[url] = txo
return result
@cache_concurrent
async def download_stream_from_uri(self, uri, exchange_rate_manager: 'ExchangeRateManager',
timeout: Optional[float] = None,
file_name: Optional[str] = None,
download_directory: Optional[str] = None,
save_file: Optional[bool] = None,
resolve_timeout: float = 3.0,
wallet: Optional['Wallet'] = None) -> ManagedStream:
manager = self.wallet_manager
wallet = wallet or manager.default_wallet
timeout = timeout or self.config.download_timeout
start_time = self.loop.time()
resolved_time = None
stream = None
txo: Optional[Output] = None
error = None
outpoint = None
if save_file is None:
save_file = self.config.save_files
if file_name and not save_file:
save_file = True
if save_file:
download_directory = download_directory or self.config.download_dir
else:
download_directory = None
payment = None
try:
# resolve the claim
if not URL.parse(uri).has_stream:
raise ResolveError("cannot download a channel claim, specify a /path")
try:
response = await asyncio.wait_for(
manager.ledger.resolve(wallet.accounts, [uri], include_purchase_receipt=True),
resolve_timeout
)
resolved_result = self._convert_to_old_resolve_output(manager, response)
except asyncio.TimeoutError:
raise ResolveTimeoutError(uri)
except Exception as err:
if isinstance(err, asyncio.CancelledError): # TODO: remove when updated to 3.8
raise
log.exception("Unexpected error resolving stream:")
raise ResolveError(f"Unexpected error resolving stream: {str(err)}")
await self.storage.save_claims_for_resolve([
value for value in resolved_result.values() if 'error' not in value
])
resolved = resolved_result.get(uri, {})
resolved = resolved if 'value' in resolved else resolved.get('claim')
if not resolved:
raise ResolveError(f"Failed to resolve stream at '{uri}'")
if 'error' in resolved:
raise ResolveError(f"error resolving stream: {resolved['error']}")
txo = response[uri]
claim = Claim.from_bytes(binascii.unhexlify(resolved['protobuf']))
outpoint = f"{resolved['txid']}:{resolved['nout']}"
resolved_time = self.loop.time() - start_time
# resume or update an existing stream, if the stream changed: download it and delete the old one after
updated_stream, to_replace = await self._check_update_or_replace(outpoint, resolved['claim_id'], claim)
if updated_stream:
log.info("already have stream for %s", uri)
if save_file and updated_stream.output_file_exists:
save_file = False
await updated_stream.start(node=self.node, timeout=timeout, save_now=save_file)
if not updated_stream.output_file_exists and (save_file or file_name or download_directory):
await updated_stream.save_file(
file_name=file_name, download_directory=download_directory, node=self.node
)
return updated_stream
if not to_replace and txo.has_price and not txo.purchase_receipt:
payment = await manager.create_purchase_transaction(
wallet.accounts, txo, exchange_rate_manager
)
stream = ManagedStream(
self.loop, self.config, self.blob_manager, claim.stream.source.sd_hash, download_directory,
file_name, ManagedStream.STATUS_RUNNING, content_fee=payment,
analytics_manager=self.analytics_manager
)
log.info("starting download for %s", uri)
before_download = self.loop.time()
await stream.start(self.node, timeout)
stream.set_claim(resolved, claim)
if to_replace: # delete old stream now that the replacement has started downloading
await self.delete_stream(to_replace)
if payment is not None:
await manager.broadcast_or_release(payment)
payment = None # to avoid releasing in `finally` later
log.info("paid fee of %s for %s", dewies_to_lbc(stream.content_fee.outputs[0].amount), uri)
await self.storage.save_content_fee(stream.stream_hash, stream.content_fee)
self.streams[stream.sd_hash] = stream
self.storage.content_claim_callbacks[stream.stream_hash] = lambda: self._update_content_claim(stream)
await self.storage.save_content_claim(stream.stream_hash, outpoint)
if save_file:
await asyncio.wait_for(stream.save_file(node=self.node), timeout - (self.loop.time() - before_download),
loop=self.loop)
return stream
except asyncio.TimeoutError:
error = DownloadDataTimeoutError(stream.sd_hash)
raise error
except Exception as err: # forgive data timeout, don't delete stream
expected = (DownloadSDTimeoutError, DownloadDataTimeoutError, InsufficientFundsError,
KeyFeeAboveMaxAllowedError)
if isinstance(err, expected):
log.warning("Failed to download %s: %s", uri, str(err))
elif isinstance(err, asyncio.CancelledError):
pass
else:
log.exception("Unexpected error downloading stream:")
error = err
raise
finally:
if payment is not None:
# payment is set to None after broadcasting, if we're here an exception probably happened
await manager.ledger.release_tx(payment)
if self.analytics_manager and (error or (stream and (stream.downloader.time_to_descriptor or
stream.downloader.time_to_first_bytes))):
server = self.wallet_manager.ledger.network.client.server
self.loop.create_task(
self.analytics_manager.send_time_to_first_bytes(
resolved_time, self.loop.time() - start_time, None if not stream else stream.download_id,
uri, outpoint,
None if not stream else len(stream.downloader.blob_downloader.active_connections),
None if not stream else len(stream.downloader.blob_downloader.scores),
None if not stream else len(stream.downloader.blob_downloader.connection_failures),
False if not stream else stream.downloader.added_fixed_peers,
self.config.fixed_peer_delay if not stream else stream.downloader.fixed_peers_delay,
None if not stream else stream.sd_hash,
None if not stream else stream.downloader.time_to_descriptor,
None if not (stream and stream.descriptor) else stream.descriptor.blobs[0].blob_hash,
None if not (stream and stream.descriptor) else stream.descriptor.blobs[0].length,
None if not stream else stream.downloader.time_to_first_bytes,
None if not error else error.__class__.__name__,
None if not error else str(error),
None if not server else f"{server[0]}:{server[1]}"
)
)
async def stream_partial_content(self, request: Request, sd_hash: str):
stream = self._sources[sd_hash]
if not stream.downloader.node:
stream.downloader.node = self.node
return await stream.stream_file(request)
return await self.streams[sd_hash].stream_file(request, self.node)
+8 -25
View File
@@ -23,13 +23,12 @@ from lbry.wallet.orchstr8.node import BlockchainNode, WalletNode
from lbry.extras.daemon.daemon import Daemon, jsonrpc_dumps_pretty
from lbry.extras.daemon.components import Component, WalletComponent
from lbry.extras.daemon.components import (
DHT_COMPONENT,
HASH_ANNOUNCER_COMPONENT, PEER_PROTOCOL_SERVER_COMPONENT,
UPNP_COMPONENT, EXCHANGE_RATE_MANAGER_COMPONENT, LIBTORRENT_COMPONENT
DHT_COMPONENT, HASH_ANNOUNCER_COMPONENT, PEER_PROTOCOL_SERVER_COMPONENT,
UPNP_COMPONENT, EXCHANGE_RATE_MANAGER_COMPONENT
)
from lbry.extras.daemon.componentmanager import ComponentManager
from lbry.extras.daemon.exchange_rate_manager import (
ExchangeRateManager, ExchangeRate, BittrexBTCFeed, BittrexUSDFeed
ExchangeRateManager, ExchangeRate, LBRYFeed, LBRYBTCFeed
)
from lbry.extras.daemon.storage import SQLiteStorage
from lbry.blob.blob_manager import BlobManager
@@ -84,7 +83,6 @@ class AsyncioTestCase(unittest.TestCase):
# https://bugs.python.org/issue32972
LOOP_SLOW_CALLBACK_DURATION = 0.2
TIMEOUT = 120.0
maxDiff = None
@@ -138,8 +136,6 @@ class AsyncioTestCase(unittest.TestCase):
with outcome.testPartExecutor(self, isTest=True):
maybe_coroutine = testMethod()
if asyncio.iscoroutine(maybe_coroutine):
if self.TIMEOUT:
self.loop.call_later(self.TIMEOUT, self.cancel)
self.loop.run_until_complete(maybe_coroutine)
outcome.expecting_failure = False
with outcome.testPartExecutor(self):
@@ -192,12 +188,6 @@ class AsyncioTestCase(unittest.TestCase):
if asyncio.iscoroutine(maybe_coroutine):
self.loop.run_until_complete(maybe_coroutine)
def cancel(self):
for task in asyncio.all_tasks(self.loop):
if not task.done():
task.print_stack()
task.cancel()
class AdvanceTimeTestCase(AsyncioTestCase):
@@ -291,8 +281,8 @@ class FakeExchangeRateManager(ExchangeRateManager):
def get_fake_exchange_rate_manager(rates=None):
return FakeExchangeRateManager(
[BittrexBTCFeed(), BittrexUSDFeed()],
rates or {'BTCLBC': 3.0, 'USDLBC': 2.0}
[LBRYFeed(), LBRYBTCFeed()],
rates or {'BTCLBC': 3.0, 'USDBTC': 2.0}
)
@@ -330,7 +320,6 @@ class CommandTestCase(IntegrationTestCase):
self.server_blob_manager = None
self.server = None
self.reflector = None
self.skip_libtorrent = True
async def asyncSetUp(self):
await super().asyncSetUp()
@@ -351,7 +340,6 @@ class CommandTestCase(IntegrationTestCase):
server_tmp_dir = tempfile.mkdtemp()
self.addCleanup(shutil.rmtree, server_tmp_dir)
self.server_config = Config()
self.server_config.transaction_cache_size = 10000
self.server_storage = SQLiteStorage(self.server_config, ':memory:')
await self.server_storage.open()
@@ -398,16 +386,12 @@ class CommandTestCase(IntegrationTestCase):
conf.blockchain_name = 'lbrycrd_regtest'
conf.lbryum_servers = [('127.0.0.1', 50001)]
conf.reflector_servers = [('127.0.0.1', 5566)]
conf.fixed_peers = [('127.0.0.1', 5567)]
conf.known_dht_nodes = []
conf.blob_lru_cache_size = self.blob_lru_cache_size
conf.transaction_cache_size = 10000
conf.components_to_skip = [
DHT_COMPONENT, UPNP_COMPONENT, HASH_ANNOUNCER_COMPONENT,
PEER_PROTOCOL_SERVER_COMPONENT
]
if self.skip_libtorrent:
conf.components_to_skip.append(LIBTORRENT_COMPONENT)
wallet_node.manager.config = conf
def wallet_maker(component_manager):
@@ -419,7 +403,7 @@ class CommandTestCase(IntegrationTestCase):
daemon = Daemon(conf, ComponentManager(
conf, skip_components=conf.components_to_skip, wallet=wallet_maker,
exchange_rate_manager=partial(ExchangeRateManagerComponent, rates={
'BTCLBC': 1.0, 'USDLBC': 2.0
'BTCLBC': 1.0, 'USDBTC': 2.0
})
))
await daemon.initialize()
@@ -448,9 +432,8 @@ class CommandTestCase(IntegrationTestCase):
async def generate(self, blocks):
""" Ask lbrycrd to generate some blocks and wait until ledger has them. """
prepare = self.ledger.on_header.where(self.blockchain.is_expected_block)
await self.blockchain.generate(blocks)
await prepare # no guarantee that it didn't happen already, so start waiting from before calling generate
await self.ledger.on_header.where(self.blockchain.is_expected_block)
async def blockchain_claim_name(self, name: str, value: str, amount: str, confirm=True):
txid = await self.blockchain._cli_cmnd('claimname', name, value, amount)
@@ -490,7 +473,7 @@ class CommandTestCase(IntegrationTestCase):
async def stream_create(
self, name='hovercraft', bid='1.0', file_path=None,
data=b'hi!', confirm=True, prefix=None, suffix=None, **kwargs):
if file_path is None and data is not None:
if file_path is None:
file_path = self.create_upload_file(data=data, prefix=prefix, suffix=suffix)
return await self.confirm_and_render(
self.daemon.jsonrpc_stream_create(name, bid, file_path=file_path, **kwargs), confirm
View File
Whitespace-only changes.
-290
View File
@@ -1,290 +0,0 @@
import asyncio
import binascii
import os
import logging
import random
from hashlib import sha1
from tempfile import mkdtemp
from typing import Optional
import libtorrent
NOTIFICATION_MASKS = [
"error",
"peer",
"port_mapping",
"storage",
"tracker",
"debug",
"status",
"progress",
"ip_block",
"dht",
"stats",
"session_log",
"torrent_log",
"peer_log",
"incoming_request",
"dht_log",
"dht_operation",
"port_mapping_log",
"picker_log",
"file_progress",
"piece_progress",
"upload",
"block_progress"
]
log = logging.getLogger(__name__)
DEFAULT_FLAGS = ( # fixme: somehow the logic here is inverted?
libtorrent.add_torrent_params_flags_t.flag_auto_managed
| libtorrent.add_torrent_params_flags_t.flag_update_subscribe
)
def get_notification_type(notification) -> str:
for i, notification_type in enumerate(NOTIFICATION_MASKS):
if (1 << i) & notification:
return notification_type
raise ValueError("unrecognized notification type")
class TorrentHandle:
def __init__(self, loop, executor, handle):
self._loop = loop
self._executor = executor
self._handle: libtorrent.torrent_handle = handle
self.started = asyncio.Event(loop=loop)
self.finished = asyncio.Event(loop=loop)
self.metadata_completed = asyncio.Event(loop=loop)
self.size = 0
self.total_wanted_done = 0
self.name = ''
self.tasks = []
self.torrent_file: Optional[libtorrent.file_storage] = None
self._base_path = None
self._handle.set_sequential_download(1)
@property
def largest_file(self) -> Optional[str]:
if not self.torrent_file:
return None
index = self.largest_file_index
return os.path.join(self._base_path, self.torrent_file.at(index).path)
@property
def largest_file_index(self):
largest_size, index = 0, 0
for file_num in range(self.torrent_file.num_files()):
if self.torrent_file.file_size(file_num) > largest_size:
largest_size = self.torrent_file.file_size(file_num)
index = file_num
return index
def stop_tasks(self):
while self.tasks:
self.tasks.pop().cancel()
def _show_status(self):
# fixme: cleanup
if not self._handle.is_valid():
return
status = self._handle.status()
if status.has_metadata:
self.size = status.total_wanted
self.total_wanted_done = status.total_wanted_done
self.name = status.name
if not self.metadata_completed.is_set():
self.metadata_completed.set()
log.info("Metadata completed for btih:%s - %s", status.info_hash, self.name)
self.torrent_file = self._handle.get_torrent_info().files()
self._base_path = status.save_path
first_piece = self.torrent_file.at(self.largest_file_index).offset
if not self.started.is_set():
if self._handle.have_piece(first_piece):
self.started.set()
else:
# prioritize it
self._handle.set_piece_deadline(first_piece, 100)
if not status.is_seeding:
log.debug('%.2f%% complete (down: %.1f kB/s up: %.1f kB/s peers: %d seeds: %d) %s - %s',
status.progress * 100, status.download_rate / 1000, status.upload_rate / 1000,
status.num_peers, status.num_seeds, status.state, status.save_path)
elif not self.finished.is_set():
self.finished.set()
log.info("Torrent finished: %s", self.name)
async def status_loop(self):
while True:
self._show_status()
if self.finished.is_set():
break
await asyncio.sleep(0.1, loop=self._loop)
async def pause(self):
await self._loop.run_in_executor(
self._executor, self._handle.pause
)
async def resume(self):
await self._loop.run_in_executor(
self._executor, lambda: self._handle.resume() # pylint: disable=unnecessary-lambda
)
class TorrentSession:
def __init__(self, loop, executor):
self._loop = loop
self._executor = executor
self._session: Optional[libtorrent.session] = None
self._handles = {}
self.tasks = []
self.wait_start = True
async def add_fake_torrent(self):
tmpdir = mkdtemp()
info, btih = _create_fake_torrent(tmpdir)
flags = libtorrent.add_torrent_params_flags_t.flag_seed_mode
handle = self._session.add_torrent({
'ti': info, 'save_path': tmpdir, 'flags': flags
})
self._handles[btih] = TorrentHandle(self._loop, self._executor, handle)
return btih
async def bind(self, interface: str = '0.0.0.0', port: int = 10889):
settings = {
'listen_interfaces': f"{interface}:{port}",
'enable_outgoing_utp': True,
'enable_incoming_utp': True,
'enable_outgoing_tcp': False,
'enable_incoming_tcp': False
}
self._session = await self._loop.run_in_executor(
self._executor, libtorrent.session, settings # pylint: disable=c-extension-no-member
)
self.tasks.append(self._loop.create_task(self.process_alerts()))
def stop(self):
while self.tasks:
self.tasks.pop().cancel()
self._session.save_state()
self._session.pause()
self._session.stop_dht()
self._session.stop_lsd()
self._session.stop_natpmp()
self._session.stop_upnp()
self._session = None
def _pop_alerts(self):
for alert in self._session.pop_alerts():
log.info("torrent alert: %s", alert)
async def process_alerts(self):
while True:
await self._loop.run_in_executor(
self._executor, self._pop_alerts
)
await asyncio.sleep(1, loop=self._loop)
async def pause(self):
await self._loop.run_in_executor(
self._executor, lambda: self._session.save_state() # pylint: disable=unnecessary-lambda
)
await self._loop.run_in_executor(
self._executor, lambda: self._session.pause() # pylint: disable=unnecessary-lambda
)
async def resume(self):
await self._loop.run_in_executor(
self._executor, self._session.resume
)
def _add_torrent(self, btih: str, download_directory: Optional[str]):
params = {'info_hash': binascii.unhexlify(btih.encode()), 'flags': DEFAULT_FLAGS}
if download_directory:
params['save_path'] = download_directory
handle = self._session.add_torrent(params)
handle.force_dht_announce()
self._handles[btih] = TorrentHandle(self._loop, self._executor, handle)
def full_path(self, btih):
return self._handles[btih].largest_file
async def add_torrent(self, btih, download_path):
await self._loop.run_in_executor(
self._executor, self._add_torrent, btih, download_path
)
self._handles[btih].tasks.append(self._loop.create_task(self._handles[btih].status_loop()))
await self._handles[btih].metadata_completed.wait()
if self.wait_start:
# fixme: temporary until we add streaming support, otherwise playback fails!
await self._handles[btih].started.wait()
def remove_torrent(self, btih, remove_files=False):
if btih in self._handles:
handle = self._handles[btih]
handle.stop_tasks()
self._session.remove_torrent(handle._handle, 1 if remove_files else 0)
self._handles.pop(btih)
async def save_file(self, btih, download_directory):
handle = self._handles[btih]
await handle.resume()
def get_size(self, btih):
return self._handles[btih].size
def get_name(self, btih):
return self._handles[btih].name
def get_downloaded(self, btih):
return self._handles[btih].total_wanted_done
def is_completed(self, btih):
return self._handles[btih].finished.is_set()
def get_magnet_uri(btih):
return f"magnet:?xt=urn:btih:{btih}"
def _create_fake_torrent(tmpdir):
# beware, that's just for testing
path = os.path.join(tmpdir, 'tmp')
with open(path, 'wb') as myfile:
size = myfile.write(bytes([random.randint(0, 255) for _ in range(40)]) * 1024)
file_storage = libtorrent.file_storage()
file_storage.add_file('tmp', size)
t = libtorrent.create_torrent(file_storage, 0, 4 * 1024 * 1024)
libtorrent.set_piece_hashes(t, tmpdir)
info = libtorrent.torrent_info(t.generate())
btih = sha1(info.metadata()).hexdigest()
return info, btih
async def main():
if os.path.exists("~/Downloads/ubuntu-18.04.3-live-server-amd64.torrent"):
os.remove("~/Downloads/ubuntu-18.04.3-live-server-amd64.torrent")
if os.path.exists("~/Downloads/ubuntu-18.04.3-live-server-amd64.iso"):
os.remove("~/Downloads/ubuntu-18.04.3-live-server-amd64.iso")
btih = "dd8255ecdc7ca55fb0bbf81323d87062db1f6d1c"
executor = None
session = TorrentSession(asyncio.get_event_loop(), executor)
session2 = TorrentSession(asyncio.get_event_loop(), executor)
await session.bind('localhost', port=4040)
await session2.bind('localhost', port=4041)
btih = await session.add_fake_torrent()
session2._session.add_dht_node(('localhost', 4040))
await session2.add_torrent(btih, "/tmp/down")
while True:
await asyncio.sleep(100)
await session.pause()
executor.shutdown()
if __name__ == "__main__":
asyncio.run(main())
-72
View File
@@ -1,72 +0,0 @@
import asyncio
import logging
import typing
log = logging.getLogger(__name__)
class TorrentInfo:
__slots__ = ('dht_seeds', 'http_seeds', 'trackers', 'total_size')
def __init__(self, dht_seeds: typing.Tuple[typing.Tuple[str, int]],
http_seeds: typing.Tuple[typing.Dict[str, typing.Any]],
trackers: typing.Tuple[typing.Tuple[str, int]], total_size: int):
self.dht_seeds = dht_seeds
self.http_seeds = http_seeds
self.trackers = trackers
self.total_size = total_size
@classmethod
def from_libtorrent_info(cls, torrent_info):
return cls(
torrent_info.nodes(), tuple(
{
'url': web_seed['url'],
'type': web_seed['type'],
'auth': web_seed['auth']
} for web_seed in torrent_info.web_seeds()
), tuple(
(tracker.url, tracker.tier) for tracker in torrent_info.trackers()
), torrent_info.total_size()
)
class Torrent:
def __init__(self, loop, handle):
self._loop = loop
self._handle = handle
self.finished = asyncio.Event(loop=loop)
def _threaded_update_status(self):
status = self._handle.status()
if not status.is_seeding:
log.info(
'%.2f%% complete (down: %.1f kB/s up: %.1f kB/s peers: %d) %s',
status.progress * 100, status.download_rate / 1000, status.upload_rate / 1000,
status.num_peers, status.state
)
elif not self.finished.is_set():
self.finished.set()
async def wait_for_finished(self):
while True:
await self._loop.run_in_executor(
None, self._threaded_update_status
)
if self.finished.is_set():
log.info("finished downloading torrent!")
await self.pause()
break
await asyncio.sleep(1, loop=self._loop)
async def pause(self):
log.info("pause torrent")
await self._loop.run_in_executor(
None, self._handle.pause
)
async def resume(self):
await self._loop.run_in_executor(
None, self._handle.resume
)
-140
View File
@@ -1,140 +0,0 @@
import asyncio
import binascii
import logging
import os
import typing
from typing import Optional
from aiohttp.web import Request
from lbry.file.source_manager import SourceManager
from lbry.file.source import ManagedDownloadSource
if typing.TYPE_CHECKING:
from lbry.torrent.session import TorrentSession
from lbry.conf import Config
from lbry.wallet.transaction import Transaction
from lbry.extras.daemon.analytics import AnalyticsManager
from lbry.extras.daemon.storage import SQLiteStorage, StoredContentClaim
from lbry.extras.daemon.storage import StoredContentClaim
log = logging.getLogger(__name__)
def path_or_none(encoded_path) -> Optional[str]:
if not encoded_path:
return
return binascii.unhexlify(encoded_path).decode()
class TorrentSource(ManagedDownloadSource):
STATUS_STOPPED = "stopped"
filter_fields = SourceManager.filter_fields
filter_fields.update({
'bt_infohash'
})
def __init__(self, loop: asyncio.AbstractEventLoop, config: 'Config', storage: 'SQLiteStorage', identifier: str,
file_name: Optional[str] = None, download_directory: Optional[str] = None,
status: Optional[str] = STATUS_STOPPED, claim: Optional['StoredContentClaim'] = None,
download_id: Optional[str] = None, rowid: Optional[int] = None,
content_fee: Optional['Transaction'] = None,
analytics_manager: Optional['AnalyticsManager'] = None,
added_on: Optional[int] = None, torrent_session: Optional['TorrentSession'] = None):
super().__init__(loop, config, storage, identifier, file_name, download_directory, status, claim, download_id,
rowid, content_fee, analytics_manager, added_on)
self.torrent_session = torrent_session
@property
def full_path(self) -> Optional[str]:
full_path = self.torrent_session.full_path(self.identifier)
self.download_directory = os.path.dirname(full_path)
return full_path
async def start(self, timeout: Optional[float] = None, save_now: Optional[bool] = False):
await self.torrent_session.add_torrent(self.identifier, self.download_directory)
async def stop(self, finished: bool = False):
await self.torrent_session.remove_torrent(self.identifier)
async def save_file(self, file_name: Optional[str] = None, download_directory: Optional[str] = None):
await self.torrent_session.save_file(self.identifier, download_directory)
@property
def torrent_length(self):
return self.torrent_session.get_size(self.identifier)
@property
def written_bytes(self):
return self.torrent_session.get_downloaded(self.identifier)
@property
def torrent_name(self):
return self.torrent_session.get_name(self.identifier)
@property
def bt_infohash(self):
return self.identifier
def stop_tasks(self):
pass
@property
def completed(self):
return self.torrent_session.is_completed(self.identifier)
class TorrentManager(SourceManager):
_sources: typing.Dict[str, ManagedDownloadSource]
filter_fields = set(SourceManager.filter_fields)
filter_fields.update({
'bt_infohash',
'blobs_remaining', # TODO: here they call them "parts", but its pretty much the same concept
'blobs_in_stream'
})
def __init__(self, loop: asyncio.AbstractEventLoop, config: 'Config', torrent_session: 'TorrentSession',
storage: 'SQLiteStorage', analytics_manager: Optional['AnalyticsManager'] = None):
super().__init__(loop, config, storage, analytics_manager)
self.torrent_session: 'TorrentSession' = torrent_session
async def recover_streams(self, file_infos: typing.List[typing.Dict]):
raise NotImplementedError
async def _load_stream(self, rowid: int, bt_infohash: str, file_name: Optional[str],
download_directory: Optional[str], status: str,
claim: Optional['StoredContentClaim'], content_fee: Optional['Transaction'],
added_on: Optional[int]):
stream = TorrentSource(
self.loop, self.config, self.storage, identifier=bt_infohash, file_name=file_name,
download_directory=download_directory, status=status, claim=claim, rowid=rowid,
content_fee=content_fee, analytics_manager=self.analytics_manager, added_on=added_on,
torrent_session=self.torrent_session
)
self.add(stream)
async def initialize_from_database(self):
pass
async def start(self):
await super().start()
def stop(self):
super().stop()
log.info("finished stopping the torrent manager")
async def delete(self, source: ManagedDownloadSource, delete_file: Optional[bool] = False):
await super().delete(source, delete_file)
self.torrent_session.remove_torrent(source.identifier, delete_file)
async def create(self, file_path: str, key: Optional[bytes] = None,
iv_generator: Optional[typing.Generator[bytes, None, None]] = None):
raise NotImplementedError
async def _delete(self, source: ManagedDownloadSource, delete_file: Optional[bool] = False):
raise NotImplementedError
# blob_hashes = [source.sd_hash] + [b.blob_hash for b in source.descriptor.blobs[:-1]]
# await self.blob_manager.delete_blobs(blob_hashes, delete_from_db=False)
# await self.storage.delete_stream(source.descriptor)
async def stream_partial_content(self, request: Request, sd_hash: str):
raise NotImplementedError
+13 -207
View File
@@ -3,7 +3,6 @@ import codecs
import datetime
import random
import socket
import time
import string
import sys
import json
@@ -20,11 +19,8 @@ import pkg_resources
import certifi
import aiohttp
from prometheus_client import Counter
from prometheus_client.registry import REGISTRY
from lbry.schema.claim import Claim
log = logging.getLogger(__name__)
@@ -105,6 +101,10 @@ def check_connection(server="lbry.com", port=80, timeout=5) -> bool:
return False
async def async_check_connection(server="lbry.com", port=80, timeout=1) -> bool:
return await asyncio.get_event_loop().run_in_executor(None, check_connection, server, port, timeout)
def random_string(length=10, chars=string.ascii_lowercase):
return ''.join([random.choice(chars) for _ in range(length)])
@@ -188,8 +188,6 @@ def cache_concurrent(async_fn):
async def resolve_host(url: str, port: int, proto: str) -> str:
if proto not in ['udp', 'tcp']:
raise Exception("invalid protocol")
if url.lower() == 'localhost':
return '127.0.0.1'
try:
if ipaddress.ip_address(url):
return url
@@ -204,101 +202,18 @@ async def resolve_host(url: str, port: int, proto: str) -> str:
))[0][4][0]
class LRUCacheWithMetrics:
__slots__ = [
'capacity',
'cache',
'_track_metrics',
'hits',
'misses'
]
def __init__(self, capacity: int, metric_name: typing.Optional[str] = None, namespace: str = "daemon_cache"):
self.capacity = capacity
self.cache = collections.OrderedDict()
if metric_name is None:
self._track_metrics = False
self.hits = self.misses = None
else:
self._track_metrics = True
try:
self.hits = Counter(
f"{metric_name}_cache_hit_count", "Number of cache hits", namespace=namespace
)
self.misses = Counter(
f"{metric_name}_cache_miss_count", "Number of cache misses", namespace=namespace
)
except ValueError as err:
log.debug("failed to set up prometheus %s_cache_miss_count metric: %s", metric_name, err)
self._track_metrics = False
self.hits = self.misses = None
def get(self, key, default=None):
try:
value = self.cache.pop(key)
if self._track_metrics:
self.hits.inc()
except KeyError:
if self._track_metrics:
self.misses.inc()
return default
self.cache[key] = value
return value
def set(self, key, value):
try:
self.cache.pop(key)
except KeyError:
if len(self.cache) >= self.capacity:
self.cache.popitem(last=False)
self.cache[key] = value
def clear(self):
self.cache.clear()
def pop(self, key):
return self.cache.pop(key)
def __setitem__(self, key, value):
return self.set(key, value)
def __getitem__(self, item):
return self.get(item)
def __contains__(self, item) -> bool:
return item in self.cache
def __len__(self):
return len(self.cache)
def __delitem__(self, key):
self.cache.pop(key)
def __del__(self):
self.clear()
if self._track_metrics: # needed for tests
try:
REGISTRY.unregister(self.hits)
REGISTRY.unregister(self.misses)
except AttributeError:
pass
class LRUCache:
__slots__ = [
'capacity',
'cache'
]
def __init__(self, capacity: int):
def __init__(self, capacity):
self.capacity = capacity
self.cache = collections.OrderedDict()
def get(self, key, default=None):
try:
value = self.cache.pop(key)
except KeyError:
return default
def get(self, key):
value = self.cache.pop(key)
self.cache[key] = value
return value
@@ -310,37 +225,16 @@ class LRUCache:
self.cache.popitem(last=False)
self.cache[key] = value
def clear(self):
self.cache.clear()
def pop(self, key):
return self.cache.pop(key)
def __setitem__(self, key, value):
return self.set(key, value)
def __getitem__(self, item):
return self.get(item)
def __contains__(self, item) -> bool:
return item in self.cache
def __len__(self):
return len(self.cache)
def __delitem__(self, key):
self.cache.pop(key)
def __del__(self):
self.clear()
def lru_cache_concurrent(cache_size: typing.Optional[int] = None,
override_lru_cache: typing.Optional[LRUCacheWithMetrics] = None):
override_lru_cache: typing.Optional[LRUCache] = None):
if not cache_size and override_lru_cache is None:
raise ValueError("invalid cache size")
concurrent_cache = {}
lru_cache = override_lru_cache if override_lru_cache is not None else LRUCacheWithMetrics(cache_size)
lru_cache = override_lru_cache or LRUCache(cache_size)
def wrapper(async_fn):
@@ -371,108 +265,20 @@ def get_ssl_context() -> ssl.SSLContext:
@contextlib.asynccontextmanager
async def aiohttp_request(method, url, **kwargs) -> typing.AsyncContextManager[aiohttp.ClientResponse]:
async with aiohttp.ClientSession() as session:
async with session.request(method, url, **kwargs) as response:
async with session.request(method, url, ssl=get_ssl_context(), **kwargs) as response:
yield response
# the ipaddress module does not show these subnets as reserved
CARRIER_GRADE_NAT_SUBNET = ipaddress.ip_network('100.64.0.0/10')
IPV4_TO_6_RELAY_SUBNET = ipaddress.ip_network('192.88.99.0/24')
def is_valid_public_ipv4(address, allow_localhost: bool = False):
try:
parsed_ip = ipaddress.ip_address(address)
if parsed_ip.is_loopback and allow_localhost:
return True
if any((parsed_ip.version != 4, parsed_ip.is_unspecified, parsed_ip.is_link_local, parsed_ip.is_loopback,
parsed_ip.is_multicast, parsed_ip.is_reserved, parsed_ip.is_private, parsed_ip.is_reserved)):
return False
else:
return not any((CARRIER_GRADE_NAT_SUBNET.supernet_of(ipaddress.ip_network(f"{address}/32")),
IPV4_TO_6_RELAY_SUBNET.supernet_of(ipaddress.ip_network(f"{address}/32"))))
except (ipaddress.AddressValueError, ValueError):
return False
async def fallback_get_external_ip(): # used if spv servers can't be used for ip detection
async def get_external_ip() -> typing.Optional[str]: # used if upnp is disabled or non-functioning
try:
async with aiohttp_request("get", "https://api.lbry.com/ip") as resp:
response = await resp.json()
if response['success']:
return response['data']['ip'], None
return response['data']['ip']
except Exception:
return None, None
async def _get_external_ip(default_servers) -> typing.Tuple[typing.Optional[str], typing.Optional[str]]:
# used if upnp is disabled or non-functioning
from lbry.wallet.server.udp import SPVStatusClientProtocol # pylint: disable=C0415
hostname_to_ip = {}
ip_to_hostnames = collections.defaultdict(list)
async def resolve_spv(server, port):
try:
server_addr = await resolve_host(server, port, 'udp')
hostname_to_ip[server] = (server_addr, port)
ip_to_hostnames[(server_addr, port)].append(server)
except Exception:
log.exception("error looking up dns for spv servers")
# accumulate the dns results
await asyncio.gather(*(resolve_spv(server, port) for (server, port) in default_servers))
loop = asyncio.get_event_loop()
pong_responses = asyncio.Queue()
connection = SPVStatusClientProtocol(pong_responses)
try:
await loop.create_datagram_endpoint(lambda: connection, ('0.0.0.0', 0))
# could raise OSError if it cant bind
randomized_servers = list(ip_to_hostnames.keys())
random.shuffle(randomized_servers)
for server in randomized_servers:
connection.ping(server)
try:
_, pong = await asyncio.wait_for(pong_responses.get(), 1)
if is_valid_public_ipv4(pong.ip_address):
return pong.ip_address, ip_to_hostnames[server][0]
except asyncio.TimeoutError:
pass
return None, None
finally:
connection.close()
async def get_external_ip(default_servers) -> typing.Tuple[typing.Optional[str], typing.Optional[str]]:
ip_from_spv_servers = await _get_external_ip(default_servers)
if not ip_from_spv_servers[1]:
return await fallback_get_external_ip()
return ip_from_spv_servers
return
def is_running_from_bundle():
# see https://pyinstaller.readthedocs.io/en/stable/runtime-information.html
return getattr(sys, 'frozen', False) and hasattr(sys, '_MEIPASS')
class LockWithMetrics(asyncio.Lock):
def __init__(self, acquire_metric, held_time_metric, loop=None):
super().__init__(loop=loop)
self._acquire_metric = acquire_metric
self._lock_held_time_metric = held_time_metric
self._lock_acquired_time = None
async def acquire(self):
start = time.perf_counter()
try:
return await super().acquire()
finally:
self._lock_acquired_time = time.perf_counter()
self._acquire_metric.observe(self._lock_acquired_time - start)
def release(self):
try:
return super().release()
finally:
self._lock_held_time_metric.observe(time.perf_counter() - self._lock_acquired_time)
+1 -1
View File
@@ -2,7 +2,7 @@ __node_daemon__ = 'lbrycrdd'
__node_cli__ = 'lbrycrd-cli'
__node_bin__ = ''
__node_url__ = (
'https://github.com/lbryio/lbrycrd/releases/download/v0.17.4.6/lbrycrd-linux-1746.zip'
'https://github.com/lbryio/lbrycrd/releases/download/v0.17.4.5/lbrycrd-linux-1745.zip'
)
__spvserver__ = 'lbry.wallet.server.coin.LBCRegTest'
+4 -6
View File
@@ -311,7 +311,7 @@ class Account:
private_key=private_key,
public_key=public_key,
address_generator=d.get('address_generator', {}),
modified_on=int(d.get('modified_on', time.time())),
modified_on=d.get('modified_on', time.time()),
channel_keys=d.get('certificates', {})
)
@@ -343,7 +343,7 @@ class Account:
def merge(self, d: dict):
if d.get('modified_on', 0) > self.modified_on:
self.name = d['name']
self.modified_on = int(d.get('modified_on', time.time()))
self.modified_on = d.get('modified_on', time.time())
assert self.address_generator.name == d['address_generator']['name']
for chain_name in ('change', 'receiving'):
if chain_name in d['address_generator']:
@@ -525,13 +525,11 @@ class Account:
channel_pubkey_hash = self.ledger.public_key_to_address(public_key_bytes)
self.channel_keys[channel_pubkey_hash] = private_key.to_pem().decode()
async def get_channel_private_key(self, public_key_bytes):
def get_channel_private_key(self, public_key_bytes):
channel_pubkey_hash = self.ledger.public_key_to_address(public_key_bytes)
private_key_pem = self.channel_keys.get(channel_pubkey_hash)
if private_key_pem:
return await asyncio.get_event_loop().run_in_executor(
None, ecdsa.SigningKey.from_pem, private_key_pem, sha256
)
return ecdsa.SigningKey.from_pem(private_key_pem, hashfunc=sha256)
async def maybe_migrate_certificates(self):
def to_der(private_key_pem):
-141
View File
@@ -740,145 +740,4 @@ HASHES = {
738000: 'aebdf15b23eb7a37600f67d45bf6586b1d5bff3d5f3459adc2f6211ab3dd0bcb',
739000: '3f5a894ac42f95f7d54ce25c42ea0baf1a05b2da0e9406978de0dc53484d8b04',
740000: '55debc22f995d844eafa0a90296c9f4f433e2b7f38456fff45dd3c66cef04e37',
741000: '927b47fc909b4b55c067bbd75d8638af1400fac076cb642e9500a747d849e458',
742000: '97fa3d83eb94114496e418c118f549ebfb8f6d123d0b40a12ecb093239557646',
743000: '482b66d8d5084703079c28e3ae69e5dee735f762d6fcf9743e75f04e139fd181',
744000: 'f406890d5c70808a58fb14429bad812a3185bdb9dace1aa57de76663f92b5013',
745000: '2bd0802cbb8aa4441a159104d39515a4ff6fc8dfe616bc83e88197847c78bcff',
746000: '24d090a7b6359db3d5d714a69ddc9a6f2e8ff8f044b723220a8ba32df785fd54',
747000: '07c4ce9ce5310ee472cf753ddb03c39c5fee6c910d491daffd38615205411633',
748000: 'ea913798c0f09d0a27eae7c852954c2c88b8c3b7f23f8fba26b68a3952d0ffde',
749000: '23f256adebfe35d49ba84ad49f3f71fc67f7745091c91f22e65f1cc2e23b8f2c',
750000: '96db12ee3a295f3d5c56d244e6e7493f58c08d3427e379940e5d4f891a41ec26',
751000: 'cedaf12415dac1314942e58ced80830b92fbfabc41f42a0b0f054f0672ef9822',
752000: '293606bcd9fbbee5584724301b2cf86bb69204820023e1fb46c238ddfbc660ab',
753000: 'f4d43cbb38b7d97919dedc0f5a6dc8007896c4f443b76f3e5693e25bc46760cf',
754000: 'fcaad22fd815311280fe451086516375d1d9d92b2990c7c351407df5aa19011e',
755000: 'b9276f10d1844cb5b0308766c8db960490ac34a73c4653d0a91202789a6ccb9b',
756000: '2fe5581f1110c1c8dcea46cad647551bd6bd640cb37738d863e189bd8f368347',
757000: 'b9d915f366f0b010429a52245b0fb02774157eb9fd8f66bce32dcd3acc71c2a1',
758000: '62d1854fc15db56b5d0e05ceeb54c1297966bf9dc7f7a0a14b42c059fc485d1b',
759000: 'f4ca9f69d16d092f4a0ea5102e6343b21204c4ea9cd9b22cddd77dbb5d68ade3',
760000: 'df3bb86641330d8cc7f55a2fd0da28251219e95babe960a308b18e08a7d88fc8',
761000: 'a93029475de4bc7569b6ae802d658cd91c84cc253772712a279f140a6c3b91b1',
762000: '307e289dc6ec8bcd62ca8831e4159d5edd780f2fae55ba55dd446225450f46f8',
763000: '293f73514abca24f374473bd0394179812952a04ea13dc60ef5ada5331fa274f',
764000: 'dd8b082db9281e3d9bacf15d6b352fda186d2d2923c7731844d0d4764dd71db8',
765000: '201239e562d2571bf47347b3522fff89632aecea3b2d8cef05151f88b2b0bcdb',
766000: '4a55a538b51b5650979e64521998cd5c5ad055ba9f3ac0e3e2a28febc6cc2798',
767000: '3916666f2adbb05ea98ec1961f9546b9afa0f6910ec95e42ce37267f2ae4f79c',
768000: 'dc0ad881eedcb5fd4954238f462080d6e7636b058d481698ed1c077e0ce2207e',
769000: 'eaf10a1e1ec6e129289b8479a05df03e0808f1f0946f1995de6524e9ebe7a461',
770000: '7200c64f22e32de7f999583361c933680fc9a2ffcb9a5ab73d3076fd49ec7537',
771000: 'd883111a2eeacff80ce31df35ab6c943805b9e48877b413fccf371e5dbfa7fb2',
772000: '3977d3c60edb9c80c97bb2b759b1659cbb650ad2d3a6f61d2caec83f1b2ae84c',
773000: '9c7175fb8646a1a82383b4c534fd01bcf92d65c43d87ae854d51a784b04dc77e',
774000: 'e0e92485f86e5fffa87b3497424e43b02a37710517d9d3f272392e8cdc56e5e9',
775000: '6395229113d3aa2105afbaeb8b59621a536fc61fe272314b2fc3bdda98dd66cc',
776000: 'b4b00207328b5f032bd4f0b634f91323ff520ada8c8bfec241b23c8e4bfd5a4e',
777000: '14cdc6f5f7b4bd5bad745dfe6fcd114e9194026412a2e1b3f345be2eef433d16',
778000: 'd3cd7b68be504c32117b670d38d59d44b02dcf3d65811efc2ca5531d902623cc',
779000: 'afcd220e4040cb5f92d4b38fc204e59822df2218f767f2c4b33597b238a35f77',
780000: '78252a9cfc289a70192ed8dd3dddeb1b9a4f9b8eff9a5d0ac259b3254472cf68',
781000: '02ebc3f17d947481a311b4771c254f1e002b6a9198d4a5258ce6c13165aadddc',
782000: '8dd9f1f372ee6d688a0bcdc3b342c77804ba5a646a218be4bc2aa02d846206c0',
783000: 'e46b0d02ec2ef488fae455665e107520e1bd2b4f35ca52af7ad8addd2f72fa73',
784000: '9ee8a8de94231e3ae3a610b82fdbca48dc14d9b80791d20af6c365a31822df6f',
785000: '21e1cc12def8173a50158b2833bd91a62140c61646f5e08aecaee3e6da20735e',
786000: 'b3e659f84d73de42888cc0f2b69bae71dd5fa6756a437a4b21958b182faa316e',
787000: 'a9be7ba00ea6a9ea6bd03d8412ec014ca7e8cda6bdc33382f165e702811b8836',
788000: 'a4c14729f8a68c03f5a0ccd890ac6a92b39c143f1f752fe81ad051eb52d8dce0',
789000: '5cf66d224e5645097efc9c3c0392b51c8ca8ea1295151921a7912a2f04ee1274',
790000: '676769ade71c33bc102bce416e66eb2c6794b03d7b8f5a590c87c380da463775',
791000: '0228e074451797bf6bfbc941bcafcbadc972d32e4e1e0c5da015513f65714217',
792000: '0fa3d00a1f19c5ac060e10a410cf7cea18eac5f89018d79ce51ac3fc66bbb365',
793000: '5f68d0868b424e32f5ce3d8e7d9f18979da7b831b8ef4e3974d62fb20ff53a97',
794000: '34508c56423739c00a837801b654b07decb274d02b383eff396d23c4d64bc0e9',
795000: '7f70910c855d1fd88cd7f9be8a3b94314ee408a31a2da6301404bf8deb07c12c',
796000: 'b74ab8813b1d2a0967fea0e66597572e5f0b5a285e21f5150fcc9d5f757de130',
797000: 'bba27b1491d907ab1baa456cb651dc5b071231b1b6ad27b62d351ca12c25dbfd',
798000: 'e75dcb15b2fc91f02e75e600dde9f6f46c09672533bc82a5d6916c4a2cd8613a',
799000: 'adf62c826a3e0b33af439a7881918ae4ce19c5fb2ca37d21243415f7d716aa65',
800000: 'd8f0ca13a8c8a19c254a3a6ba15150a34711dca96f2d877162cc44aa2acfb268',
801000: '2a8c7104c4040a2bc31913ae25e9361df5bac9477368c708f86c1ca640480887',
802000: '1f3b09d3561c4a8a056b263289bd492dc6c0d604c3fa195935e735d1c0ddc40e',
803000: '037769628c40a701fdb4b16d79084b8fbb319fde79770a7ac842f3cdc813099e',
804000: 'a0c6a089e5fa1e3589ca282085fe7201a5705776d81b257ffd252b2947fa6428',
805000: 'b2ac99bfc4a488e7b7624b31ee061991a6dd0881bb005cd13f3dd2e66a08fe19',
806000: 'ffe63cb999a278280b80a667d2dcb60c40e43a53f733914d8bec808b694ebf83',
807000: 'eddb09fc6c4869a59b520d0befb1fb6ac952333f3cc5de086539c85ea8558778',
808000: '0f4fb3f9172e52897ea992d9f3a2024126c4d2e63e9888739f11fb1f5e4c1f46',
809000: '9641dd720d23ced2f1cb6e5cf46ac4e547afb9f56263c4cf58e3b19d407cf401',
810000: 'de6dc953acd7e5ef213b3aaf1c4a9ee1d5b756bfce5525ee105214647e243a85',
811000: 'c52c83712ca12b24b2db1b4a575e7f352b1d560cbf702e121a03bdca9e8be23d',
812000: '83143734bb965318a53a38a7e403dcdb3e3fadedb01ab12c370417fc2a0655c0',
813000: 'e480deff10c5a84fc957e3aed936690e24b74dd08fa8858a8a953c2f7383b914',
814000: '810d33afcee07b9abe16c6cdc3a041038daa131c476b0daf48a080007f08b490',
815000: 'b4aeb9e16fddd27844b2d56bc2b221134039bb5642c9e9ba88372afbdeac3972',
816000: '86e73b67aae3d248011b8f66ed414cb8a9ba4b2a3cf7e32773cfbff055d719b7',
817000: '3ebb8b83752b48242016cb682f0f6bd14e15371bf1163a5933193eaa0edeb351',
818000: '4d925e17f642f220bbf317d3d5355d2f41fbce325f190f8c3b32dc0b337d24d6',
819000: 'b9cc126d620f6b99d90a00d35957b0e428aaaa7c986bc9e816a60e4334572961',
820000: '9c2f8c142bed1f94dca29276f7c83958be8cfe11773bb9b56c808fbcf7d3b1f8',
821000: 'e5509eb98895cfa12a8da5d54c1df3f52472ffcbdf707adbf84a4a9c5d356203',
822000: '764aada4802ebfe4ef935ab50af06a4f83aa556c49fdde3d9e12e1abd230c16b',
823000: '1dbd745c2e96a365d865f990d109137d32d42977f503af55d8c00b109d31d3c3',
824000: '954304a0b0c8f549c3bffd5ff46b5b8f05b0f0fde2a36f24fd5af9d774fb3079',
825000: '17808b14f2056c1a5d46cb7617e9de9be6a1a6084edbc1bdb778586467a72297',
826000: '3ca1167d4cac8b187829b23001b438617c43704b42462c4eb001b0d434cb9651',
827000: '246d1607245e4a202f420393ac2e30e9cbf5eb5570dc997073b897f6d8643023',
828000: '1764730a8dc3e89d02d168ff6bb54e8c903820b74711af6ff27bd0c8545577e7',
829000: 'd9f3ab0cd823c6305bd8b95a96188bb4f2ca90b4d66c5d12293e8b6192bac0f2',
830000: 'd4ff51f0092b04aedf8d39937680d8e8309b1be21d36e7833ed36f8e30aad6ea',
831000: '3e92e76721b962396dce52993fa7606552f0907b38f7b2bd7b21ada98c145f47',
832000: 'df12fcdb4cbe53ba627ace6de898298de175f8671d3d90170732d110fcdc34b8',
833000: '25167ff38ae4a5964b618cabe0a12d4de62ac7a4c47448cdb4499e09e108d5b9',
834000: 'd31f5309ea179a1e386e835fc372e47dcda6871a3a239abfba50c4f368994f13',
835000: 'aff7e8dd3e55ea807fcbe284014075f420b3a23f1b0eb47bacdc1c91d2899813',
836000: '3b5ac6d64c470739bb17d1544a285affb40f2d33e92687e5ba7c5ac602e0d72a',
837000: 'd5619cbfe4f27c55f2bf9351b4891636cf64fef88212a5eeeae7bd3de47fe0bd',
838000: '1f9102a49c6ac470cb5d0050e5300b1443840d6d65719b835e3bea484aafb2ec',
839000: '3f63e391f0fbc5787fbe4ace3bada3816261294ea1c6ee435001801023682f90',
840000: '777894fd12bd0d6dee7bcde2995c68e55e7094e3122da38571e4b6c4304b75e0',
841000: 'ceb0c598c788e25e43e25aa4beff5c7377035824844cf1675eaea537074df028',
842000: '8661cf2065dc713d2ba043f0b81f0effcc940eeb3e91906a21ff22c210561dcd',
843000: '0dc2766f90415009d0c86bedffee6ebcf58042eb08262c0c67c4e9ed86b2aec8',
844000: '26d072da864cab268a12794977b04ec44fb69ef3978e2342e82225974dac54dd',
845000: '95e93bb60be8d5f07a1f4d26290c914957a82fc9d26ae8a3f20082eda27406ff',
846000: 'f1bdc39af7705e58ab8b6c31dc70dce1e115db1cfd8cc9b037949dfbec82a59a',
847000: 'f5f10f06396ecf2765d8a081141d489737c1d8d57c281f28f57c4cb2f90db883',
848000: '331b8ef08605bae8d749893af9ed54f0df4f07a5a002108a2a0aea82d0360979',
849000: '75b5f6233ab9a1bbc3c8b2893e5b22a0aa98e7ea635261255dc3c281f67d2260',
850000: '5d7e6fe83e0ea1910a54a00090704737671d6f44df4228e21440ad1fc15e595f',
851000: '7822db25d3ff0f6695ee38bad91edf317b5c6611673d28f1d22053110bb558be',
852000: '2f0effad83a3561fc1a2806a562786a641d9ddb18d16bb9308006e7d324a21e9',
853000: 'f603b2eaff11d5296377d990651317d40a1b2599ad2c5250eab131090f4b9458',
854000: '34d59b26a50f18a9f250736d0f2e69d28b7e196fbef9b8a26c6b0b75c16aa194',
855000: '76dd1ffff3946c0878969886fcf177ce5ab5560df19ddf006f9bcb02ae3e4e4f',
856000: '74ff0b6f64e9dd5802fec2aac1d3ae194d28b9264114adaf0a882b46c8c918fe',
857000: '7b5badfa2e4f40aa597a504d7ebe83c3705a2c6169a8c168ce293db223bc2d32',
858000: '2bb0767a0f72b20d45ecfc3e34517dbda16d85758e040cf0e147f4cbd0cc57ac',
859000: '3d741b9c365a91ed76f85824b94d19ec19b608d232660840ba59c7aa4b2cb67f',
860000: 'd481a5a117878c0e3acd1f5844e150fb30e617577947d9846b1d214d703b71b0',
861000: '54033424e488a3f1ad6946d4a6d9acb48465d6b1dbe8e1c2504a54cc84d7cad4',
862000: '464bc3820a8cc8844dc9e26c388009e9982c656d46ef4b4fd0a2cb0e4eea0aaa',
863000: 'd1aa94be2174f66780c4f226b9da3f6712b0f37af8dec33360bea83ca261b342',
864000: '8c16008f11de5bc395d88cd802514ff647450f1bc136724b9aaf2ccce10a494f',
865000: '3dae86012e97a201e2e1a47c899001ac00f78dc108026ed7c4194858c6c6dd5a',
866000: 'afe5b0ccab995e1a1fa25fbc24c1d4b1a92c43042d03395f8743dcd806e72fd8',
867000: 'c83716ac171aa9ab0d414833db340fa30e82bfda6cc616d3038529caab9b5600',
868000: '8c409fe03cd35ef2d8e366818788b40eaeb4c8f6ae91450d75f4a66ca5f69cad',
869000: '1d47909ceba790b8e1ce2e9902ee2775ea99e58efdb95668f9803a8ccf95f286',
870000: '9adf5da1476388f053aa42de636da169d1cf1c9652cdf7cd9ad4fb18a0eb3388',
871000: '8ad57fb1e74bcba0b5614fbac003be2bb32275dd85b38f2d28a0585005a99cfc',
872000: '84a32e92012a356106e9657da8dab1a5491ea588fc29d411c69b20680c666420',
873000: 'adf5921bbbfaa43929f67e6a070975313b77b456e262c700a27be611fceb17ae',
874000: '09eaa7c4b18c79a46a2895190333f72336826223d5c986849a06f5153f49f2a5',
875000: '235d7e4f31966507312149ea4c5e294aa84c695cf840117f0ef5963be7a0bda1',
876000: '9aa9cb806ccbec0475ac330b496c5b2edeba38ba3f1e13ddd54a01457634a288',
877000: 'c1e7f9b2b20bb1c4c0deadbc786d31fdf36f262325342aa23d1a66e2846b22bc',
878000: 'ee0d2b20ac28ce23ab38698a57c6beff14f12b7af9d027c05cc92f652695f46b',
879000: '0eb0810f4b81d1845b0a88f05449408df2e45715c9210a656f45278c5fdf7956',
880000: 'e7d613027e3b4ca38d09bbef07998b57db237c6d67f1e8ea50024d2e0d9a1a72',
881000: '21af4d355d8756b8bf0369b2d79b5c824148ae069026ba5c14f9dd6b7555e1db',
}
+1 -1
View File
@@ -5,7 +5,7 @@ from lbry.wallet.transaction import OutputEffectiveAmountEstimator
MAXIMUM_TRIES = 100000
STRATEGIES = ['sqlite'] # sqlite coin chooser is in database.py
STRATEGIES = []
def strategy(method):
+65 -287
View File
@@ -4,33 +4,22 @@ import asyncio
import sqlite3
import platform
from binascii import hexlify
from collections import defaultdict
from dataclasses import dataclass
from contextvars import ContextVar
from concurrent.futures.thread import ThreadPoolExecutor
from concurrent.futures.process import ProcessPoolExecutor
from typing import Tuple, List, Union, Callable, Any, Awaitable, Iterable, Dict, Optional
from datetime import date
from prometheus_client import Gauge, Counter, Histogram
from lbry.utils import LockWithMetrics
from .bip32 import PubKey
from .transaction import Transaction, Output, OutputScript, TXRefImmutable, Input
from .transaction import Transaction, Output, OutputScript, TXRefImmutable
from .constants import TXO_TYPES, CLAIM_TYPES
from .util import date_to_julian_day
from concurrent.futures.thread import ThreadPoolExecutor # pylint: disable=wrong-import-order
if platform.system() == 'Windows' or 'ANDROID_ARGUMENT' or 'KIVY_BUILD' in os.environ:
from concurrent.futures.thread import ThreadPoolExecutor as ReaderExecutorClass # pylint: disable=reimported
else:
from concurrent.futures.process import ProcessPoolExecutor as ReaderExecutorClass
log = logging.getLogger(__name__)
sqlite3.enable_callback_tracebacks(True)
HISTOGRAM_BUCKETS = (
.005, .01, .025, .05, .075, .1, .25, .5, .75, 1.0, 2.5, 5.0, 7.5, 10.0, 15.0, 20.0, 30.0, 60.0, float('inf')
)
@dataclass
class ReaderProcessState:
@@ -66,39 +55,24 @@ def run_read_only_fetchone(sql, params):
raise
if platform.system() == 'Windows' or 'ANDROID_ARGUMENT' in os.environ:
ReaderExecutorClass = ThreadPoolExecutor
else:
ReaderExecutorClass = ProcessPoolExecutor
class AIOSQLite:
reader_executor: ReaderExecutorClass
waiting_writes_metric = Gauge(
"waiting_writes_count", "Number of waiting db writes", namespace="daemon_database"
)
waiting_reads_metric = Gauge(
"waiting_reads_count", "Number of waiting db writes", namespace="daemon_database"
)
write_count_metric = Counter(
"write_count", "Number of database writes", namespace="daemon_database"
)
read_count_metric = Counter(
"read_count", "Number of database reads", namespace="daemon_database"
)
acquire_write_lock_metric = Histogram(
f'write_lock_acquired', 'Time to acquire the write lock', namespace="daemon_database", buckets=HISTOGRAM_BUCKETS
)
held_write_lock_metric = Histogram(
f'write_lock_held', 'Length of time the write lock is held for', namespace="daemon_database",
buckets=HISTOGRAM_BUCKETS
)
def __init__(self):
# has to be single threaded as there is no mapping of thread:connection
self.writer_executor = ThreadPoolExecutor(max_workers=1)
self.writer_connection: Optional[sqlite3.Connection] = None
self._closing = False
self.query_count = 0
self.write_lock = LockWithMetrics(self.acquire_write_lock_metric, self.held_write_lock_metric)
self.write_lock = asyncio.Lock()
self.writers = 0
self.read_ready = asyncio.Event()
self.urgent_read_done = asyncio.Event()
@classmethod
async def connect(cls, path: Union[bytes, str], *args, **kwargs):
@@ -114,20 +88,13 @@ class AIOSQLite:
)
await asyncio.get_event_loop().run_in_executor(db.writer_executor, _connect_writer)
db.read_ready.set()
db.urgent_read_done.set()
return db
async def close(self):
if self._closing:
return
self._closing = True
def __checkpoint_and_close(conn: sqlite3.Connection):
conn.execute("PRAGMA WAL_CHECKPOINT(FULL);")
log.info("DB checkpoint finished.")
conn.close()
await asyncio.get_event_loop().run_in_executor(
self.writer_executor, __checkpoint_and_close, self.writer_connection)
await asyncio.get_event_loop().run_in_executor(self.writer_executor, self.writer_connection.close)
self.writer_executor.shutdown(wait=True)
self.reader_executor.shutdown(wait=True)
self.read_ready.clear()
@@ -145,30 +112,12 @@ class AIOSQLite:
read_only=False, fetch_all: bool = False) -> List[dict]:
read_only_fn = run_read_only_fetchall if fetch_all else run_read_only_fetchone
parameters = parameters if parameters is not None else []
still_waiting = False
urgent_read = False
if read_only:
self.waiting_reads_metric.inc()
self.read_count_metric.inc()
try:
while self.writers and not self._closing: # more writes can come in while we are waiting for the first
if not urgent_read and still_waiting and self.urgent_read_done.is_set():
# throttle the writes if they pile up
self.urgent_read_done.clear()
urgent_read = True
# wait until the running writes have finished
await self.read_ready.wait()
still_waiting = True
if self._closing:
raise asyncio.CancelledError()
return await asyncio.get_event_loop().run_in_executor(
self.reader_executor, read_only_fn, sql, parameters
)
finally:
if urgent_read:
# unthrottle the writers if they had to be throttled
self.urgent_read_done.set()
self.waiting_reads_metric.dec()
while self.writers:
await self.read_ready.wait()
return await asyncio.get_event_loop().run_in_executor(
self.reader_executor, read_only_fn, sql, parameters
)
if fetch_all:
return await self.run(lambda conn: conn.execute(sql, parameters).fetchall())
return await self.run(lambda conn: conn.execute(sql, parameters).fetchone())
@@ -186,34 +135,17 @@ class AIOSQLite:
return self.run(lambda conn: conn.execute(sql, parameters))
async def run(self, fun, *args, **kwargs):
self.write_count_metric.inc()
self.waiting_writes_metric.inc()
# it's possible many writes are coming in one after the other, these can
# block reader calls for a long time
# if the reader waits for the writers to finish and then has to wait for
# yet more, it will clear the urgent_read_done event to block more writers
# piling on
try:
await self.urgent_read_done.wait()
except Exception as e:
self.waiting_writes_metric.dec()
raise e
self.writers += 1
# block readers
self.read_ready.clear()
try:
async with self.write_lock:
if self._closing:
raise asyncio.CancelledError()
async with self.write_lock:
try:
return await asyncio.get_event_loop().run_in_executor(
self.writer_executor, lambda: self.__run_transaction(fun, *args, **kwargs)
)
finally:
self.writers -= 1
self.waiting_writes_metric.dec()
if not self.writers:
# unblock the readers once the last enqueued writer finishes
self.read_ready.set()
finally:
self.writers -= 1
if not self.writers:
self.read_ready.set()
def __run_transaction(self, fun: Callable[[sqlite3.Connection, Any, Any], Any], *args, **kwargs):
self.writer_connection.execute('begin')
@@ -228,28 +160,10 @@ class AIOSQLite:
log.warning("rolled back")
raise
async def run_with_foreign_keys_disabled(self, fun, *args, **kwargs):
self.write_count_metric.inc()
self.waiting_writes_metric.inc()
try:
await self.urgent_read_done.wait()
except Exception as e:
self.waiting_writes_metric.dec()
raise e
self.writers += 1
self.read_ready.clear()
try:
async with self.write_lock:
if self._closing:
raise asyncio.CancelledError()
return await asyncio.get_event_loop().run_in_executor(
self.writer_executor, self.__run_transaction_with_foreign_keys_disabled, fun, args, kwargs
)
finally:
self.writers -= 1
self.waiting_writes_metric.dec()
if not self.writers:
self.read_ready.set()
def run_with_foreign_keys_disabled(self, fun, *args, **kwargs) -> Awaitable:
return asyncio.get_event_loop().run_in_executor(
self.writer_executor, self.__run_transaction_with_foreign_keys_disabled, fun, args, kwargs
)
def __run_transaction_with_foreign_keys_disabled(self,
fun: Callable[[sqlite3.Connection, Any, Any], Any],
@@ -387,31 +301,14 @@ def interpolate(sql, values):
return sql
def constrain_single_or_list(constraints, column, value, convert=lambda x: x, negate=False):
def constrain_single_or_list(constraints, column, value, convert=lambda x: x):
if value is not None:
if isinstance(value, list):
value = [convert(v) for v in value]
if len(value) == 1:
if negate:
constraints[f"{column}__or"] = {
f"{column}__is_null": True,
f"{column}__not": value[0]
}
else:
constraints[column] = value[0]
constraints[column] = value[0]
elif len(value) > 1:
if negate:
constraints[f"{column}__or"] = {
f"{column}__is_null": True,
f"{column}__not_in": value
}
else:
constraints[f"{column}__in"] = value
elif negate:
constraints[f"{column}__or"] = {
f"{column}__is_null": True,
f"{column}__not": convert(value)
}
constraints[f"{column}__in"] = value
else:
constraints[column] = convert(value)
return constraints
@@ -448,7 +345,7 @@ class SQLiteMixin:
return
await self.db.executescript('\n'.join(
f"DROP TABLE {table};" for table in tables
) + '\n' + 'PRAGMA WAL_CHECKPOINT(FULL);' + '\n' + 'VACUUM;')
))
await self.db.execute(self.CREATE_VERSION_TABLE)
await self.db.execute("INSERT INTO version VALUES (?)", (self.SCHEMA_VERSION,))
await self.db.executescript(self.CREATE_TABLES_QUERY)
@@ -494,105 +391,9 @@ def dict_row_factory(cursor, row):
return d
SQLITE_MAX_INTEGER = 9223372036854775807
def _get_spendable_utxos(transaction: sqlite3.Connection, accounts: List, decoded_transactions: Dict[str, Transaction],
result: Dict[Tuple[bytes, int, bool], List[int]], reserved: List[Transaction],
amount_to_reserve: int, reserved_amount: int, floor: int, ceiling: int,
fee_per_byte: int) -> int:
accounts_fmt = ",".join(["?"] * len(accounts))
txo_query = f"""
SELECT tx.txid, txo.txoid, tx.raw, tx.height, txo.position as nout, tx.is_verified, txo.amount FROM txo
INNER JOIN account_address USING (address)
LEFT JOIN txi USING (txoid)
INNER JOIN tx USING (txid)
WHERE txo.txo_type=0 AND txi.txoid IS NULL AND tx.txid IS NOT NULL AND NOT txo.is_reserved
AND txo.amount >= ? AND txo.amount < ?
"""
if accounts:
txo_query += f"""
AND account_address.account {'= ?' if len(accounts_fmt) == 1 else 'IN (' + accounts_fmt + ')'}
"""
txo_query += """
ORDER BY txo.amount ASC, tx.height DESC
"""
# prefer confirmed, but save unconfirmed utxos from this selection in case they are needed
unconfirmed = []
for row in transaction.execute(txo_query, (floor, ceiling, *accounts)):
(txid, txoid, raw, height, nout, verified, amount) = row.values()
# verified or non verified transactions were found- reset the gap count
# multiple txos can come from the same tx, only decode it once and cache
if txid not in decoded_transactions:
# cache the decoded transaction
decoded_transactions[txid] = Transaction(raw)
decoded_tx = decoded_transactions[txid]
# save the unconfirmed txo for possible use later, if still needed
if verified:
# add the txo to the reservation, minus the fee for including it
reserved_amount += amount
reserved_amount -= Input.spend(decoded_tx.outputs[nout]).size * fee_per_byte
# mark it as reserved
result[(raw, height, verified)].append(nout)
reserved.append(txoid)
# if we've reserved enough, return
if reserved_amount >= amount_to_reserve:
return reserved_amount
else:
unconfirmed.append((txid, txoid, raw, height, nout, verified, amount))
# we're popping the items, so to get them in the order they were seen they are reversed
unconfirmed.reverse()
# add available unconfirmed txos if any were previously found
while unconfirmed and reserved_amount < amount_to_reserve:
(txid, txoid, raw, height, nout, verified, amount) = unconfirmed.pop()
# it's already decoded
decoded_tx = decoded_transactions[txid]
# add to the reserved amount
reserved_amount += amount
reserved_amount -= Input.spend(decoded_tx.outputs[nout]).size * fee_per_byte
result[(raw, height, verified)].append(nout)
reserved.append(txoid)
return reserved_amount
def get_and_reserve_spendable_utxos(transaction: sqlite3.Connection, accounts: List, amount_to_reserve: int, floor: int,
fee_per_byte: int, set_reserved: bool, return_insufficient_funds: bool,
base_multiplier: int = 100):
txs = defaultdict(list)
decoded_transactions = {}
reserved = []
reserved_dewies = 0
multiplier = base_multiplier
gap_count = 0
while reserved_dewies < amount_to_reserve and gap_count < 5 and floor * multiplier < SQLITE_MAX_INTEGER:
previous_reserved_dewies = reserved_dewies
reserved_dewies = _get_spendable_utxos(
transaction, accounts, decoded_transactions, txs, reserved, amount_to_reserve, reserved_dewies,
floor, floor * multiplier, fee_per_byte
)
floor *= multiplier
if previous_reserved_dewies == reserved_dewies:
gap_count += 1
multiplier **= 2
else:
gap_count = 0
multiplier = base_multiplier
# reserve the accumulated txos if enough were found
if reserved_dewies >= amount_to_reserve:
if set_reserved:
transaction.executemany("UPDATE txo SET is_reserved = ? WHERE txoid = ?",
[(True, txoid) for txoid in reserved]).fetchall()
return txs
# return_insufficient_funds and set_reserved are used for testing
return txs if return_insufficient_funds else {}
class Database(SQLiteMixin):
SCHEMA_VERSION = "1.5"
SCHEMA_VERSION = "1.3"
PRAGMAS = """
pragma journal_mode=WAL;
@@ -704,9 +505,6 @@ class Database(SQLiteMixin):
row['txo_type'] = TXO_TYPES['stream']
elif txo.is_support:
row['txo_type'] = TXO_TYPES['support']
support = txo.can_decode_support
if support and support.is_signed:
row['channel_id'] = support.signing_channel_id
elif txo.purchase is not None:
row['txo_type'] = TXO_TYPES['purchase']
row['claim_id'] = txo.purchased_claim_id
@@ -781,7 +579,7 @@ class Database(SQLiteMixin):
return self.db.run(__many)
async def reserve_outputs(self, txos, is_reserved=True):
txoids = [(is_reserved, txo.id) for txo in txos]
txoids = ((is_reserved, txo.id) for txo in txos)
await self.db.executemany("UPDATE txo SET is_reserved = ? WHERE txoid = ?", txoids)
async def release_outputs(self, txos):
@@ -793,20 +591,6 @@ class Database(SQLiteMixin):
# 2. update address histories removing deleted TXs
return True
async def get_spendable_utxos(self, ledger, reserve_amount, accounts: Optional[Iterable], min_amount: int = 1,
fee_per_byte: int = 50, set_reserved: bool = True,
return_insufficient_funds: bool = False) -> List:
to_spend = await self.db.run(
get_and_reserve_spendable_utxos, tuple(account.id for account in accounts), reserve_amount, min_amount,
fee_per_byte, set_reserved, return_insufficient_funds
)
txos = []
for (raw, height, verified), positions in to_spend.items():
tx = Transaction(raw, height=height, is_verified=verified)
for nout in positions:
txos.append(tx.outputs[nout].get_estimator(ledger))
return txos
async def select_transactions(self, cols, accounts=None, read_only=False, **constraints):
if not {'txid', 'txid__in'}.intersection(constraints):
assert accounts, "'accounts' argument required when no 'txid' constraint is present"
@@ -965,18 +749,16 @@ class Database(SQLiteMixin):
sql.append("LEFT JOIN txi ON (txi.position=0 AND txi.txid=txo.txid)")
return await self.db.execute_fetchall(*query(' '.join(sql), **constraints), read_only=read_only)
async def get_txos(self, wallet=None, no_tx=False, no_channel_info=False, read_only=False, **constraints):
async def get_txos(self, wallet=None, no_tx=False, read_only=False, **constraints):
include_is_spent = constraints.get('include_is_spent', False)
include_is_my_input = constraints.get('include_is_my_input', False)
include_is_my_output = constraints.pop('include_is_my_output', False)
include_received_tips = constraints.pop('include_received_tips', False)
select_columns = [
"tx.txid, tx.height, tx.position as tx_position, tx.is_verified, "
"tx.txid, raw, tx.height, tx.position as tx_position, tx.is_verified, "
"txo_type, txo.position as txo_position, amount, script"
]
if not no_tx:
select_columns.append("raw")
my_accounts = {a.public_key.address for a in wallet.accounts} if wallet else set()
my_accounts_sql = ""
@@ -1054,33 +836,32 @@ class Database(SQLiteMixin):
txo.received_tips = row['received_tips']
txos.append(txo)
if not no_channel_info:
channel_ids = set()
channel_ids = set()
for txo in txos:
if txo.is_claim and txo.can_decode_claim:
if txo.claim.is_signed:
channel_ids.add(txo.claim.signing_channel_id)
if txo.claim.is_channel and wallet:
for account in wallet.accounts:
private_key = account.get_channel_private_key(
txo.claim.channel.public_key_bytes
)
if private_key:
txo.private_key = private_key
break
if channel_ids:
channels = {
txo.claim_id: txo for txo in
(await self.get_channels(
wallet=wallet,
claim_id__in=channel_ids,
read_only=read_only
))
}
for txo in txos:
if txo.is_claim and txo.can_decode_claim:
if txo.claim.is_signed:
channel_ids.add(txo.claim.signing_channel_id)
if txo.claim.is_channel and wallet:
for account in wallet.accounts:
private_key = await account.get_channel_private_key(
txo.claim.channel.public_key_bytes
)
if private_key:
txo.private_key = private_key
break
if channel_ids:
channels = {
txo.claim_id: txo for txo in
(await self.get_channels(
wallet=wallet,
claim_id__in=channel_ids,
read_only=read_only
))
}
for txo in txos:
if txo.is_claim and txo.can_decode_claim:
txo.channel = channels.get(txo.claim.signing_channel_id, None)
txo.channel = channels.get(txo.claim.signing_channel_id, None)
return txos
@@ -1287,16 +1068,13 @@ class Database(SQLiteMixin):
self.constrain_collections(constraints)
return self.get_utxo_count(**constraints)
async def release_all_outputs(self, account=None):
if account is None:
await self.db.execute_fetchall("UPDATE txo SET is_reserved = 0 WHERE is_reserved = 1")
else:
await self.db.execute_fetchall(
"UPDATE txo SET is_reserved = 0 WHERE"
" is_reserved = 1 AND txo.address IN ("
" SELECT address from account_address WHERE account = ?"
" )", (account.public_key.address, )
)
async def release_all_outputs(self, account):
await self.db.execute_fetchall(
"UPDATE txo SET is_reserved = 0 WHERE"
" is_reserved = 1 AND txo.address IN ("
" SELECT address from account_address WHERE account = ?"
" )", (account.public_key.address, )
)
def get_supports_summary(self, read_only=False, **constraints):
return self.get_txos(
+47 -44
View File
@@ -5,6 +5,7 @@ import asyncio
import logging
import zlib
from datetime import date
from concurrent.futures.thread import ThreadPoolExecutor
from io import BytesIO
from typing import Optional, Iterator, Tuple, Callable
@@ -41,22 +42,23 @@ class Headers:
validate_difficulty: bool = True
def __init__(self, path) -> None:
self.io = None
if path == ':memory:':
self.io = BytesIO()
self.path = path
self._size: Optional[int] = None
self.chunk_getter: Optional[Callable] = None
self.executor = ThreadPoolExecutor(1)
self.known_missing_checkpointed_chunks = set()
self.check_chunk_lock = asyncio.Lock()
async def open(self):
self.io = BytesIO()
if not self.executor:
self.executor = ThreadPoolExecutor(1)
if self.path != ':memory:':
def _readit():
if os.path.exists(self.path):
with open(self.path, 'r+b') as header_file:
self.io.seek(0)
self.io.write(header_file.read())
await asyncio.get_event_loop().run_in_executor(None, _readit)
if not os.path.exists(self.path):
self.io = open(self.path, 'w+b')
else:
self.io = open(self.path, 'r+b')
bytes_size = self.io.seek(0, os.SEEK_END)
self._size = bytes_size // self.header_size
max_checkpointed_height = max(self.checkpoints.keys() or [-1]) + 1000
@@ -70,14 +72,10 @@ class Headers:
await self.get_all_missing_headers()
async def close(self):
if self.io is not None:
def _close():
flags = 'r+b' if os.path.exists(self.path) else 'w+b'
with open(self.path, flags) as header_file:
header_file.write(self.io.getbuffer())
await asyncio.get_event_loop().run_in_executor(None, _close)
self.io.close()
self.io = None
if self.executor:
self.executor.shutdown()
self.executor = None
self.io.close()
@staticmethod
def serialize(header):
@@ -137,30 +135,28 @@ class Headers:
except struct.error:
raise IndexError(f"failed to get {height}, at {len(self)}")
def estimated_timestamp(self, height, try_real_headers=True):
def estimated_timestamp(self, height):
if height <= 0:
return
if try_real_headers and self.has_header(height):
offset = height * self.header_size
return struct.unpack('<I', self.io.getbuffer()[offset + 100: offset + 104])[0]
return int(self.first_block_timestamp + (height * self.timestamp_average_offset))
def estimated_julian_day(self, height):
return date_to_julian_day(date.fromtimestamp(self.estimated_timestamp(height, False)))
return date_to_julian_day(date.fromtimestamp(self.estimated_timestamp(height)))
async def get_raw_header(self, height) -> bytes:
if self.chunk_getter:
await self.ensure_chunk_at(height)
if not 0 <= height <= self.height:
raise IndexError(f"{height} is out of bounds, current height: {self.height}")
return self._read(height)
return await asyncio.get_running_loop().run_in_executor(self.executor, self._read, height)
def _read(self, height, count=1):
offset = height * self.header_size
return bytes(self.io.getbuffer()[offset: offset + self.header_size * count])
self.io.seek(height * self.header_size, os.SEEK_SET)
return self.io.read(self.header_size * count)
def chunk_hash(self, start, count):
return self.hash_header(self._read(start, count)).decode()
self.io.seek(start * self.header_size, os.SEEK_SET)
return self.hash_header(self.io.read(count * self.header_size)).decode()
async def ensure_checkpointed_size(self):
max_checkpointed_height = max(self.checkpoints.keys() or [-1])
@@ -169,7 +165,7 @@ class Headers:
async def ensure_chunk_at(self, height):
async with self.check_chunk_lock:
if self.has_header(height):
if await self.has_header(height):
log.debug("has header %s", height)
return
return await self.fetch_chunk(height)
@@ -183,7 +179,7 @@ class Headers:
)
chunk_hash = self.hash_header(chunk).decode()
if self.checkpoints.get(start) == chunk_hash:
self._write(start, chunk)
await asyncio.get_running_loop().run_in_executor(self.executor, self._write, start, chunk)
if start in self.known_missing_checkpointed_chunks:
self.known_missing_checkpointed_chunks.remove(start)
return
@@ -193,23 +189,27 @@ class Headers:
f"Checkpoint mismatch at height {start}. Expected {self.checkpoints[start]}, but got {chunk_hash} instead."
)
def has_header(self, height):
async def has_header(self, height):
normalized_height = (height // 1000) * 1000
if normalized_height in self.checkpoints:
return normalized_height not in self.known_missing_checkpointed_chunks
empty = '56944c5d3f98413ef45cf54545538103cc9f298e0575820ad3591376e2e0f65d'
all_zeroes = '789d737d4f448e554b318c94063bbfa63e9ccda6e208f5648ca76ee68896557b'
return self.chunk_hash(height, 1) not in (empty, all_zeroes)
def _has_header(height):
empty = '56944c5d3f98413ef45cf54545538103cc9f298e0575820ad3591376e2e0f65d'
all_zeroes = '789d737d4f448e554b318c94063bbfa63e9ccda6e208f5648ca76ee68896557b'
return self.chunk_hash(height, 1) not in (empty, all_zeroes)
return await asyncio.get_running_loop().run_in_executor(self.executor, _has_header, height)
async def get_all_missing_headers(self):
# Heavy operation done in one optimized shot
for chunk_height, expected_hash in reversed(list(self.checkpoints.items())):
if chunk_height in self.known_missing_checkpointed_chunks:
continue
if self.chunk_hash(chunk_height, 1000) != expected_hash:
self.known_missing_checkpointed_chunks.add(chunk_height)
return self.known_missing_checkpointed_chunks
def _io_checkall():
for chunk_height, expected_hash in reversed(list(self.checkpoints.items())):
if chunk_height in self.known_missing_checkpointed_chunks:
continue
if self.chunk_hash(chunk_height, 1000) != expected_hash:
self.known_missing_checkpointed_chunks.add(chunk_height)
return self.known_missing_checkpointed_chunks
return await asyncio.get_running_loop().run_in_executor(self.executor, _io_checkall)
@property
def height(self) -> int:
@@ -241,7 +241,7 @@ class Headers:
bail = True
chunk = chunk[:(height-e.height)*self.header_size]
if chunk:
added += self._write(height, chunk)
added += await asyncio.get_running_loop().run_in_executor(self.executor, self._write, height, chunk)
if bail:
break
return added
@@ -306,7 +306,9 @@ class Headers:
previous_header_hash = fail = None
batch_size = 36
for height in range(start_height, self.height, batch_size):
headers = self._read(height, batch_size)
headers = await asyncio.get_running_loop().run_in_executor(
self.executor, self._read, height, batch_size
)
if len(headers) % self.header_size != 0:
headers = headers[:(len(headers) // self.header_size) * self.header_size]
for header_hash, header in self._iterate_headers(height, headers):
@@ -322,11 +324,12 @@ class Headers:
assert start_height > 0 and height == start_height
if fail:
log.warning("Header file corrupted at height %s, truncating it.", height - 1)
self.io.seek(max(0, (height - 1)) * self.header_size, os.SEEK_SET)
self.io.truncate()
self.io.flush()
self._size = self.io.seek(0, os.SEEK_END) // self.header_size
return
def __truncate(at_height):
self.io.seek(max(0, (at_height - 1)) * self.header_size, os.SEEK_SET)
self.io.truncate()
self.io.flush()
self._size = self.io.seek(0, os.SEEK_END) // self.header_size
return await asyncio.get_running_loop().run_in_executor(self.executor, __truncate, height)
previous_header_hash = header_hash
@classmethod
+164 -285
View File
@@ -1,8 +1,8 @@
import os
import copy
import time
import asyncio
import logging
from io import StringIO
from datetime import datetime
from functools import partial
from operator import itemgetter
@@ -10,11 +10,11 @@ from collections import defaultdict
from binascii import hexlify, unhexlify
from typing import Dict, Tuple, Type, Iterable, List, Optional, DefaultDict, NamedTuple
import pylru
from lbry.schema.result import Outputs, INVALID, NOT_FOUND
from lbry.schema.url import URL
from lbry.crypto.hash import hash160, double_sha256, sha256
from lbry.crypto.base58 import Base58
from lbry.utils import LRUCacheWithMetrics
from .tasks import TaskGroup
from .database import Database
@@ -29,6 +29,7 @@ from .constants import TXO_TYPES, CLAIM_TYPES, COIN, NULL_HASH32
from .bip32 import PubKey, PrivateKey
from .coinselection import CoinSelector
log = logging.getLogger(__name__)
LedgerType = Type['BaseLedger']
@@ -123,6 +124,7 @@ class Ledger(metaclass=LedgerRegistry):
self.network: Network = self.config.get('network') or Network(self)
self.network.on_header.listen(self.receive_header)
self.network.on_status.listen(self.process_status_update)
self.network.on_connected.listen(self.join_network)
self.accounts = []
self.fee_per_byte: int = self.config.get('fee_per_byte', self.default_fee_per_byte)
@@ -155,19 +157,18 @@ class Ledger(metaclass=LedgerRegistry):
self._on_ready_controller = StreamController()
self.on_ready = self._on_ready_controller.stream
self._tx_cache = LRUCacheWithMetrics(self.config.get("tx_cache_size", 1024), metric_name='tx')
self._tx_cache = pylru.lrucache(100000)
self._update_tasks = TaskGroup()
self._other_tasks = TaskGroup() # that we dont need to start
self._utxo_reservation_lock = asyncio.Lock()
self._header_processing_lock = asyncio.Lock()
self._address_update_locks: DefaultDict[str, asyncio.Lock] = defaultdict(asyncio.Lock)
self._history_lock = asyncio.Lock()
self.coin_selection_strategy = None
self._known_addresses_out_of_sync = set()
self.fee_per_name_char = self.config.get('fee_per_name_char', self.default_fee_per_name_char)
self._balance_cache = LRUCacheWithMetrics(2 ** 15)
self._balance_cache = pylru.lrucache(100000)
@classmethod
def get_id(cls):
@@ -231,7 +232,7 @@ class Ledger(metaclass=LedgerRegistry):
async def get_effective_amount_estimators(self, funding_accounts: Iterable[Account]):
estimators = []
for account in funding_accounts:
utxos = await account.get_utxos(no_tx=True, no_channel_info=True)
utxos = await account.get_utxos()
for utxo in utxos:
estimators.append(utxo.get_estimator(self))
return estimators
@@ -242,15 +243,11 @@ class Ledger(metaclass=LedgerRegistry):
def get_address_count(self, **constraints):
return self.db.get_address_count(**constraints)
async def get_spendable_utxos(self, amount: int, funding_accounts: Optional[Iterable['Account']], min_amount=1):
min_amount = min(amount // 10, min_amount)
fee = Output.pay_pubkey_hash(COIN, NULL_HASH32).get_fee(self)
selector = CoinSelector(amount, fee)
async def get_spendable_utxos(self, amount: int, funding_accounts):
async with self._utxo_reservation_lock:
if self.coin_selection_strategy == 'sqlite':
return await self.db.get_spendable_utxos(self, amount + fee, funding_accounts, min_amount=min_amount,
fee_per_byte=self.fee_per_byte)
txos = await self.get_effective_amount_estimators(funding_accounts)
fee = Output.pay_pubkey_hash(COIN, NULL_HASH32).get_fee(self)
selector = CoinSelector(amount, fee)
spendables = selector.select(txos, self.coin_selection_strategy)
if spendables:
await self.reserve_outputs(s.txo for s in spendables)
@@ -328,10 +325,7 @@ class Ledger(metaclass=LedgerRegistry):
await self.network.on_connected.first
async with self._header_processing_lock:
await self._update_tasks.add(self.initial_headers_sync())
self.network.on_connected.listen(self.join_network)
asyncio.ensure_future(self.join_network())
await fully_synced
await self.db.release_all_outputs()
await asyncio.gather(*(a.maybe_migrate_certificates() for a in self.accounts))
await asyncio.gather(*(a.save_max_gap() for a in self.accounts))
if len(self.accounts) > 10:
@@ -415,7 +409,6 @@ class Ledger(metaclass=LedgerRegistry):
"Blockchain Reorganization: attempting rewind to height %s from starting height %s",
height, height+rewound
)
self._tx_cache.clear()
else:
raise IndexError(f"headers.connect() returned negative number ({added})")
@@ -480,23 +473,21 @@ class Ledger(metaclass=LedgerRegistry):
for address, remote_status in zip(batch, results):
self._update_tasks.add(self.update_history(address, remote_status, address_manager))
addresses_remaining = addresses_remaining[batch_size:]
if self.network.client and self.network.client.server_address_and_port:
log.info("subscribed to %i/%i addresses on %s:%i", len(addresses) - len(addresses_remaining),
len(addresses), *self.network.client.server_address_and_port)
if self.network.client and self.network.client.server_address_and_port:
log.info(
"finished subscribing to %i addresses on %s:%i", len(addresses),
*self.network.client.server_address_and_port
)
log.info("subscribed to %i/%i addresses on %s:%i", len(addresses) - len(addresses_remaining),
len(addresses), *self.network.client.server_address_and_port)
log.info(
"finished subscribing to %i addresses on %s:%i", len(addresses),
*self.network.client.server_address_and_port
)
def process_status_update(self, update):
address, remote_status = update
self._update_tasks.add(self.update_history(address, remote_status))
async def update_history(self, address, remote_status, address_manager: AddressManager = None,
reattempt_update: bool = True):
async def update_history(self, address, remote_status, address_manager: AddressManager = None):
async with self._address_update_locks[address]:
self._known_addresses_out_of_sync.discard(address)
local_status, local_history = await self.get_local_status_and_history(address)
if local_status == remote_status:
@@ -506,55 +497,60 @@ class Ledger(metaclass=LedgerRegistry):
remote_history = list(map(itemgetter('tx_hash', 'height'), remote_history))
we_need = set(remote_history) - set(local_history)
if not we_need:
remote_missing = set(local_history) - set(remote_history)
if remote_missing:
log.warning(
"%i transactions we have for %s are not in the remote address history",
len(remote_missing), address
)
return True
to_request = {}
pending_synced_history = {}
already_synced = set()
already_synced_offset = 0
cache_tasks: List[asyncio.Task[Transaction]] = []
synced_history = StringIO()
loop = asyncio.get_running_loop()
for i, (txid, remote_height) in enumerate(remote_history):
if i == already_synced_offset and i < len(local_history) and local_history[i] == (txid, remote_height):
pending_synced_history[i] = f'{txid}:{remote_height}:'
already_synced.add((txid, remote_height))
already_synced_offset += 1
continue
if i < len(local_history) and local_history[i] == (txid, remote_height) and not cache_tasks:
synced_history.write(f'{txid}:{remote_height}:')
else:
check_local = (txid, remote_height) not in we_need
cache_tasks.append(loop.create_task(
self.cache_transaction(txid, remote_height, check_local=check_local)
))
tx_indexes = {}
synced_txs = []
for task in cache_tasks:
tx = await task
for i, (txid, remote_height) in enumerate(remote_history):
tx_indexes[txid] = i
if (txid, remote_height) in already_synced:
continue
to_request[i] = (txid, remote_height)
check_db_for_txos = []
for txi in tx.inputs:
if txi.txo_ref.txo is not None:
continue
cache_item = self._tx_cache.get(txi.txo_ref.tx_ref.id)
if cache_item is not None:
if cache_item.tx is None:
await cache_item.has_tx.wait()
assert cache_item.tx is not None
txi.txo_ref = cache_item.tx.outputs[txi.txo_ref.position].ref
else:
check_db_for_txos.append(txi.txo_ref.id)
log.debug(
"request %i transactions, %i/%i for %s are already synced", len(to_request), len(already_synced),
len(remote_history), address
referenced_txos = {} if not check_db_for_txos else {
txo.id: txo for txo in await self.db.get_txos(
txoid__in=check_db_for_txos, order_by='txo.txoid', no_tx=True
)
}
for txi in tx.inputs:
if txi.txo_ref.txo is not None:
continue
referenced_txo = referenced_txos.get(txi.txo_ref.id)
if referenced_txo is not None:
txi.txo_ref = referenced_txo.ref
synced_history.write(f'{tx.id}:{tx.height}:')
synced_txs.append(tx)
await self.db.save_transaction_io_batch(
synced_txs, address, self.address_to_hash160(address), synced_history.getvalue()
)
remote_history_txids = set(txid for txid, _ in remote_history)
async for tx in self.request_synced_transactions(to_request, remote_history_txids, address):
pending_synced_history[tx_indexes[tx.id]] = f"{tx.id}:{tx.height}:"
if len(pending_synced_history) % 100 == 0:
log.info("Syncing address %s: %d/%d", address, len(pending_synced_history), len(to_request))
log.info("Sync finished for address %s: %d/%d", address, len(pending_synced_history), len(to_request))
assert len(pending_synced_history) == len(remote_history), \
f"{len(pending_synced_history)} vs {len(remote_history)}"
synced_history = ""
for remote_i, i in zip(range(len(remote_history)), sorted(pending_synced_history.keys())):
assert i == remote_i, f"{i} vs {remote_i}"
txid, height = remote_history[remote_i]
if f"{txid}:{height}:" != pending_synced_history[i]:
log.warning("history mismatch: %s vs %s", remote_history[remote_i], pending_synced_history[i])
synced_history += pending_synced_history[i]
await self.db.set_address_history(address, synced_history)
await asyncio.wait([
self._on_transaction_controller.add(TransactionEvent(address, tx))
for tx in synced_txs
])
if address_manager is None:
address_manager = await self.get_address_manager_for_address(address)
@@ -563,135 +559,73 @@ class Ledger(metaclass=LedgerRegistry):
await address_manager.ensure_address_gap()
local_status, local_history = \
await self.get_local_status_and_history(address, synced_history)
await self.get_local_status_and_history(address, synced_history.getvalue())
if local_status != remote_status:
if local_history == remote_history:
log.warning(
"%s has a synced history but a mismatched status", address
)
return True
remote_set = set(remote_history)
local_set = set(local_history)
log.warning(
"%s is out of sync after syncing.\n"
"Remote: %s with %d items (%i unique), local: %s with %d items (%i unique).\n"
"Histories are mismatched on %i items.\n"
"Local is missing\n"
"%s\n"
"Remote is missing\n"
"%s\n"
"******",
address, remote_status, len(remote_history), len(remote_set),
local_status, len(local_history), len(local_set), len(remote_set.symmetric_difference(local_set)),
"\n".join([f"{txid} - {height}" for txid, height in local_set.difference(remote_set)]),
"\n".join([f"{txid} - {height}" for txid, height in remote_set.difference(local_set)])
"Wallet is out of sync after syncing. Remote: %s with %d items, local: %s with %d items",
remote_status, len(remote_history), local_status, len(local_history)
)
log.warning("local: %s", local_history)
log.warning("remote: %s", remote_history)
self._known_addresses_out_of_sync.add(address)
return False
else:
log.debug("finished syncing transaction history for %s, %i known txs", address, len(local_history))
return True
async def cache_transaction(self, txid, remote_height, check_local=True):
cache_item = self._tx_cache.get(txid)
if cache_item is None:
cache_item = self._tx_cache[txid] = TransactionCacheItem()
elif cache_item.tx is not None and \
cache_item.tx.height >= remote_height and \
(cache_item.tx.is_verified or remote_height < 1):
return cache_item.tx # cached tx is already up-to-date
try:
cache_item.pending_verifications += 1
return await self._update_cache_item(cache_item, txid, remote_height, check_local)
finally:
cache_item.pending_verifications -= 1
async def _update_cache_item(self, cache_item, txid, remote_height, check_local=True):
async with cache_item.lock:
tx = cache_item.tx
if tx is None and check_local:
# check local db
tx = cache_item.tx = await self.db.get_transaction(txid=txid)
merkle = None
if tx is None:
# fetch from network
_raw, merkle = await self.network.retriable_call(
self.network.get_transaction_and_merkle, txid, remote_height
)
tx = Transaction(unhexlify(_raw), height=merkle.get('block_height'))
cache_item.tx = tx # make sure it's saved before caching it
await self.maybe_verify_transaction(tx, remote_height, merkle)
return tx
async def maybe_verify_transaction(self, tx, remote_height, merkle=None):
tx.height = remote_height
if 0 < remote_height < len(self.headers):
cached = self._tx_cache.get(tx.id)
if not cached:
# cache txs looked up by transaction_show too
cached = TransactionCacheItem()
cached.tx = tx
self._tx_cache[tx.id] = cached
if 0 < remote_height < len(self.headers) and cached.pending_verifications <= 1:
# can't be tx.pending_verifications == 1 because we have to handle the transaction_show case
if not merkle:
merkle = await self.network.retriable_call(self.network.get_merkle, tx.id, remote_height)
if 'merkle' not in merkle:
return
merkle_root = self.get_root_of_merkle_tree(merkle['merkle'], merkle['pos'], tx.hash)
header = await self.headers.get(remote_height)
tx.position = merkle['pos']
tx.is_verified = merkle_root == header['merkle_root']
return tx
async def request_transactions(self, to_request: Tuple[Tuple[str, int], ...], cached=False):
batches = [[]]
remote_heights = {}
cache_hits = set()
for txid, height in sorted(to_request, key=lambda x: x[1]):
if cached:
cached_tx = self._tx_cache.get(txid)
if cached_tx is not None:
if cached_tx.tx is not None and cached_tx.tx.is_verified:
cache_hits.add(txid)
continue
else:
self._tx_cache[txid] = TransactionCacheItem()
remote_heights[txid] = height
if len(batches[-1]) == 100:
batches.append([])
batches[-1].append(txid)
if not batches[-1]:
batches.pop()
if cached and cache_hits:
yield {txid: self._tx_cache[txid].tx for txid in cache_hits}
for batch in batches:
txs = await self._single_batch(batch, remote_heights)
if cached:
for txid, tx in txs.items():
self._tx_cache[txid].tx = tx
yield txs
async def request_synced_transactions(self, to_request, remote_history, address):
async for txs in self.request_transactions(((txid, height) for txid, height in to_request.values())):
for tx in txs.values():
yield tx
await self._sync_and_save_batch(address, remote_history, txs)
async def _single_batch(self, batch, remote_heights):
heights = {remote_heights[txid] for txid in batch}
unrestriced = 0 < min(heights) < max(heights) < max(self.headers.checkpoints or [0])
batch_result = await self.network.retriable_call(self.network.get_transaction_batch, batch, not unrestriced)
txs = {}
for txid, (raw, merkle) in batch_result.items():
remote_height = remote_heights[txid]
tx = Transaction(unhexlify(raw), height=remote_height)
txs[tx.id] = tx
await self.maybe_verify_transaction(tx, remote_height, merkle)
return txs
async def _sync_and_save_batch(self, address, remote_history, pending_txs):
await asyncio.gather(*(self._sync(tx, remote_history, pending_txs) for tx in pending_txs.values()))
await self.db.save_transaction_io_batch(
pending_txs.values(), address, self.address_to_hash160(address), ""
)
while pending_txs:
self._on_transaction_controller.add(TransactionEvent(address, pending_txs.popitem()[1]))
async def _sync(self, tx, remote_history, pending_txs):
check_db_for_txos = {}
for txi in tx.inputs:
if txi.txo_ref.txo is not None:
continue
wanted_txid = txi.txo_ref.tx_ref.id
if wanted_txid not in remote_history:
continue
if wanted_txid in pending_txs:
txi.txo_ref = pending_txs[wanted_txid].outputs[txi.txo_ref.position].ref
else:
check_db_for_txos[txi] = txi.txo_ref.id
referenced_txos = {} if not check_db_for_txos else {
txo.id: txo for txo in await self.db.get_txos(
txoid__in=list(check_db_for_txos.values()), order_by='txo.txoid', no_tx=True
)
}
for txi in check_db_for_txos:
if txi.txo_ref.id in referenced_txos:
txi.txo_ref = referenced_txos[txi.txo_ref.id].ref
else:
tx_from_db = await self.db.get_transaction(txid=txi.txo_ref.tx_ref.id)
if tx_from_db is None:
log.warning("%s not on db, not on cache, but on remote history!", txi.txo_ref.id)
else:
txi.txo_ref = tx_from_db.outputs[txi.txo_ref.position].ref
return tx
async def get_address_manager_for_address(self, address) -> Optional[AddressManager]:
details = await self.db.get_address(address=address)
@@ -705,7 +639,6 @@ class Ledger(metaclass=LedgerRegistry):
return self.network.broadcast(hexlify(tx.raw).decode())
async def wait(self, tx: Transaction, height=-1, timeout=1):
timeout = timeout or 600 # after 10 minutes there is almost 0 hope
addresses = set()
for txi in tx.inputs:
if txi.txo_ref.txo is not None:
@@ -715,40 +648,26 @@ class Ledger(metaclass=LedgerRegistry):
for txo in tx.outputs:
if txo.has_address:
addresses.add(self.hash160_to_address(txo.pubkey_hash))
start = int(time.perf_counter())
while timeout and (int(time.perf_counter()) - start) <= timeout:
if await self._wait_round(tx, height, addresses):
return
raise asyncio.TimeoutError('Timed out waiting for transaction.')
async def _wait_round(self, tx: Transaction, height: int, addresses: Iterable[str]):
records = await self.db.get_addresses(address__in=addresses)
_, pending = await asyncio.wait([
self.on_transaction.where(partial(
lambda a, e: a == e.address and e.tx.height >= height and e.tx.id == tx.id,
address_record['address']
)) for address_record in records
], timeout=1)
if not pending:
return True
records = await self.db.get_addresses(address__in=addresses)
for record in records:
local_history = (await self.get_local_status_and_history(
record['address'], history=record['history']
))[1] if record['history'] else []
for txid, local_height in local_history:
if txid == tx.id:
if local_height >= height:
return True
log.warning(
"local history has higher height than remote for %s (%i vs %i)", txid,
local_height, height
)
return False
log.warning(
"local history does not contain %s, requested height %i", tx.id, height
)
return False
], timeout=timeout)
if pending:
records = await self.db.get_addresses(address__in=addresses)
for record in records:
found = False
local_history = (await self.get_local_status_and_history(
record['address'], history=record['history']
))[1] if record['history'] else []
for txid, local_height in local_history:
if txid == tx.id and local_height >= height:
found = True
if not found:
print(record['history'], addresses, tx.id)
raise asyncio.TimeoutError('Timed out waiting for transaction.')
async def _inflate_outputs(
self, query, accounts,
@@ -759,31 +678,20 @@ class Ledger(metaclass=LedgerRegistry):
include_received_tips=False) -> Tuple[List[Output], dict, int, int]:
encoded_outputs = await query
outputs = Outputs.from_base64(encoded_outputs or b'') # TODO: why is the server returning None?
txs: List[Transaction] = []
txs = []
if len(outputs.txs) > 0:
async for tx in self.request_transactions(tuple(outputs.txs), cached=True):
txs.extend(tx.values())
txs: List[Transaction] = await asyncio.gather(*(
self.cache_transaction(*tx) for tx in outputs.txs
))
_txos, blocked = outputs.inflate(txs)
txos = []
for txo in _txos:
if isinstance(txo, Output):
# transactions and outputs are cached and shared between wallets
# we don't want to leak informaion between wallet so we add the
# wallet specific metadata on throw away copies of the txos
txo = copy.copy(txo)
channel = txo.channel
txo.purchase_receipt = None
txo.update_annotations(None)
txo.channel = channel
txos.append(txo)
txos, blocked = outputs.inflate(txs)
includes = (
include_purchase_receipt, include_is_my_output,
include_sent_supports, include_sent_tips
)
if accounts and any(includes):
copies = []
receipts = {}
if include_purchase_receipt:
priced_claims = []
@@ -800,55 +708,55 @@ class Ledger(metaclass=LedgerRegistry):
}
for txo in txos:
if isinstance(txo, Output) and txo.can_decode_claim:
# transactions and outputs are cached and shared between wallets
# we don't want to leak informaion between wallet so we add the
# wallet specific metadata on throw away copies of the txos
txo_copy = copy.copy(txo)
copies.append(txo_copy)
if include_purchase_receipt:
txo.purchase_receipt = receipts.get(txo.claim_id)
txo_copy.purchase_receipt = receipts.get(txo.claim_id)
if include_is_my_output:
mine = await self.db.get_txo_count(
claim_id=txo.claim_id, txo_type__in=CLAIM_TYPES, is_my_output=True,
is_spent=False, accounts=accounts
)
if mine:
txo.is_my_output = True
txo_copy.is_my_output = True
else:
txo.is_my_output = False
txo_copy.is_my_output = False
if include_sent_supports:
supports = await self.db.get_txo_sum(
claim_id=txo.claim_id, txo_type=TXO_TYPES['support'],
is_my_input=True, is_my_output=True,
is_spent=False, accounts=accounts
)
txo.sent_supports = supports
txo_copy.sent_supports = supports
if include_sent_tips:
tips = await self.db.get_txo_sum(
claim_id=txo.claim_id, txo_type=TXO_TYPES['support'],
is_my_input=True, is_my_output=False,
accounts=accounts
)
txo.sent_tips = tips
txo_copy.sent_tips = tips
if include_received_tips:
tips = await self.db.get_txo_sum(
claim_id=txo.claim_id, txo_type=TXO_TYPES['support'],
is_my_input=False, is_my_output=True,
accounts=accounts
)
txo.received_tips = tips
txo_copy.received_tips = tips
else:
copies.append(txo)
txos = copies
return txos, blocked, outputs.offset, outputs.total
async def resolve(self, accounts, urls, new_sdk_server=None, **kwargs):
txos = []
async def resolve(self, accounts, urls, **kwargs):
resolve = partial(self.network.retriable_call, self.network.resolve)
urls_copy = list(urls)
if new_sdk_server:
resolve = partial(self.network.new_resolve, new_sdk_server)
else:
resolve = partial(self.network.retriable_call, self.network.resolve)
txos = []
while urls_copy:
batch, urls_copy = urls_copy[:100], urls_copy[100:]
txos.extend(
(await self._inflate_outputs(
resolve(batch), accounts, **kwargs
))[0]
)
batch, urls_copy = urls_copy[:500], urls_copy[500:]
txos.extend((await self._inflate_outputs(resolve(batch), accounts, **kwargs))[0])
assert len(urls) == len(txos), "Mismatch between urls requested for resolve and responses received."
result = {}
for url, txo in zip(urls, txos):
@@ -861,20 +769,13 @@ class Ledger(metaclass=LedgerRegistry):
result[url] = txo
return result
async def sum_supports(self, new_sdk_server, **kwargs) -> List[Dict]:
return await self.network.sum_supports(new_sdk_server, **kwargs)
async def claim_search(
self, accounts, include_purchase_receipt=False, include_is_my_output=False,
new_sdk_server=None, **kwargs) -> Tuple[List[Output], dict, int, int]:
if new_sdk_server:
claim_search = partial(self.network.new_claim_search, new_sdk_server)
else:
claim_search = self.network.claim_search
**kwargs) -> Tuple[List[Output], dict, int, int]:
return await self._inflate_outputs(
claim_search(**kwargs), accounts,
self.network.claim_search(**kwargs), accounts,
include_purchase_receipt=include_purchase_receipt,
include_is_my_output=include_is_my_output,
include_is_my_output=include_is_my_output
)
async def get_claim_by_claim_id(self, accounts, claim_id, **kwargs) -> Output:
@@ -937,11 +838,6 @@ class Ledger(metaclass=LedgerRegistry):
return self.db.get_purchase_count(**constraints)
async def _resolve_for_local_results(self, accounts, txos):
txos = await self._resolve_for_local_claim_results(accounts, txos)
txos = await self._resolve_for_local_support_results(accounts, txos)
return txos
async def _resolve_for_local_claim_results(self, accounts, txos):
results = []
response = await self.resolve(
accounts, [txo.permanent_url for txo in txos if txo.can_decode_claim]
@@ -957,23 +853,6 @@ class Ledger(metaclass=LedgerRegistry):
results.append(txo)
return results
async def _resolve_for_local_support_results(self, accounts, txos):
channel_ids = set()
signed_support_txos = []
for txo in txos:
support = txo.can_decode_support
if support and support.signing_channel_id:
channel_ids.add(support.signing_channel_id)
signed_support_txos.append(txo)
if channel_ids:
channels = {
channel.claim_id: channel for channel in
(await self.claim_search(accounts, claim_ids=list(channel_ids)))[0]
}
for txo in signed_support_txos:
txo.channel = channels.get(txo.support.signing_channel_id)
return txos
async def get_claims(self, resolve=False, **constraints):
claims = await self.db.get_claims(**constraints)
if resolve:
@@ -1002,7 +881,7 @@ class Ledger(metaclass=LedgerRegistry):
return self.db.get_channel_count(**constraints)
async def resolve_collection(self, collection, offset=0, page_size=1):
claim_ids = collection.claim.collection.claims.ids[offset:page_size + offset]
claim_ids = collection.claim.collection.claims.ids[offset:page_size+offset]
try:
resolve_results, _, _, _ = await self.claim_search([], claim_ids=claim_ids)
except Exception as err:
@@ -1022,10 +901,8 @@ class Ledger(metaclass=LedgerRegistry):
claims.append(None)
return claims
async def get_collections(self, resolve_claims=0, resolve=False, **constraints):
async def get_collections(self, resolve_claims=0, **constraints):
collections = await self.db.get_collections(**constraints)
if resolve:
collections = await self._resolve_for_local_results(constraints.get('accounts', []), collections)
if resolve_claims > 0:
for collection in collections:
collection.claims = await self.resolve_collection(collection, page_size=resolve_claims)
@@ -1053,7 +930,7 @@ class Ledger(metaclass=LedgerRegistry):
'txid': tx.id,
'timestamp': ts,
'date': datetime.fromtimestamp(ts).isoformat(' ')[:-3] if tx.height > 0 else None,
'confirmations': (headers.height + 1) - tx.height if tx.height > 0 else 0,
'confirmations': (headers.height+1) - tx.height if tx.height > 0 else 0,
'claim_info': [],
'update_info': [],
'support_info': [],
@@ -1063,7 +940,7 @@ class Ledger(metaclass=LedgerRegistry):
is_my_inputs = all([txi.is_my_input for txi in tx.inputs])
if is_my_inputs:
# fees only matter if we are the ones paying them
item['value'] = dewies_to_lbc(tx.net_account_balance + tx.fee)
item['value'] = dewies_to_lbc(tx.net_account_balance+tx.fee)
item['fee'] = dewies_to_lbc(-tx.fee)
else:
# someone else paid the fees
@@ -1086,13 +963,13 @@ class Ledger(metaclass=LedgerRegistry):
if txi.txo_ref.txo is not None:
other_txo = txi.txo_ref.txo
if (other_txo.is_claim or other_txo.script.is_support_claim) \
and other_txo.claim_id == txo.claim_id:
and other_txo.claim_id == txo.claim_id:
previous = other_txo
break
if previous is not None:
item['update_info'].append({
'address': txo.get_address(self),
'balance_delta': dewies_to_lbc(previous.amount - txo.amount),
'balance_delta': dewies_to_lbc(previous.amount-txo.amount),
'amount': dewies_to_lbc(txo.amount),
'claim_id': txo.claim_id,
'claim_name': txo.claim_name,
@@ -1170,7 +1047,7 @@ class Ledger(metaclass=LedgerRegistry):
for account in accounts:
balance = self._balance_cache.get(account.id)
if not balance:
balance = self._balance_cache[account.id] = \
balance = self._balance_cache[account.id] =\
await account.get_detailed_balance(confirmations, reserved_subtotals=True)
for key, value in balance.items():
if key == 'reserved_subtotals':
@@ -1180,6 +1057,7 @@ class Ledger(metaclass=LedgerRegistry):
result[key] += value
return result
class TestNetLedger(Ledger):
network_name = 'testnet'
pubkey_address_prefix = bytes((111,))
@@ -1188,6 +1066,7 @@ class TestNetLedger(Ledger):
extended_private_key_prefix = unhexlify('04358394')
checkpoints = {}
class RegTestLedger(Ledger):
network_name = 'regtest'
headers_class = UnvalidatedHeaders
+2 -3
View File
@@ -184,7 +184,6 @@ class WalletManager:
'auto_connect': True,
'default_servers': config.lbryum_servers,
'data_path': config.wallet_dir,
'tx_cache_size': config.transaction_cache_size
}
wallets_directory = os.path.join(config.wallet_dir, 'wallets')
@@ -300,8 +299,8 @@ class WalletManager:
async def broadcast_or_release(self, tx, blocking=False):
try:
await self.ledger.broadcast(tx)
if blocking:
await self.ledger.wait(tx, timeout=None)
except:
await self.ledger.release_tx(tx)
raise
if blocking:
await self.ledger.wait(tx, timeout=None)
+179 -217
View File
@@ -1,28 +1,25 @@
import logging
import asyncio
import json
import socket
import random
from time import perf_counter
from collections import defaultdict
from operator import itemgetter
from typing import Dict, Optional, Tuple
import aiohttp
from lbry import __version__
from lbry.utils import resolve_host
from lbry.error import IncompatibleWalletServerError
from lbry.wallet.rpc import RPCSession as BaseClientSession, Connector, RPCError, ProtocolError
from lbry.wallet.stream import StreamController
from lbry.wallet.server.udp import SPVStatusClientProtocol, SPVPong
log = logging.getLogger(__name__)
class ClientSession(BaseClientSession):
def __init__(self, *args, network: 'Network', server, timeout=30, **kwargs):
def __init__(self, *args, network, server, timeout=30, on_connect_callback=None, **kwargs):
self.network = network
self.server = server
super().__init__(*args, **kwargs)
self._on_disconnect_controller = StreamController()
self.on_disconnected = self._on_disconnect_controller.stream
self.framer.max_size = self.max_errors = 1 << 32
self.timeout = timeout
self.max_seconds_idle = timeout * 2
@@ -30,6 +27,8 @@ class ClientSession(BaseClientSession):
self.connection_latency: Optional[float] = None
self._response_samples = 0
self.pending_amount = 0
self._on_connect_cb = on_connect_callback or (lambda: None)
self.trigger_urgent_reconnect = asyncio.Event()
@property
def available(self):
@@ -56,7 +55,7 @@ class ClientSession(BaseClientSession):
async def send_request(self, method, args=()):
self.pending_amount += 1
log.debug("send %s%s to %s:%i (%i timeout)", method, tuple(args), self.server[0], self.server[1], self.timeout)
log.debug("send %s%s to %s:%i", method, tuple(args), *self.server)
try:
if method == 'server.version':
return await self.send_timed_server_version_request(args, self.timeout)
@@ -67,7 +66,7 @@ class ClientSession(BaseClientSession):
log.debug("Time since last packet: %s", perf_counter() - self.last_packet_received)
if (perf_counter() - self.last_packet_received) < self.timeout:
continue
log.warning("timeout sending %s to %s:%i", method, *self.server)
log.info("timeout sending %s to %s:%i", method, *self.server)
raise asyncio.TimeoutError
if done:
try:
@@ -87,12 +86,44 @@ class ClientSession(BaseClientSession):
self.synchronous_close()
raise
except asyncio.CancelledError:
log.warning("cancelled sending %s to %s:%i", method, *self.server)
# self.synchronous_close()
log.info("cancelled sending %s to %s:%i", method, *self.server)
self.synchronous_close()
raise
finally:
self.pending_amount -= 1
async def ensure_session(self):
# Handles reconnecting and maintaining a session alive
# TODO: change to 'ping' on newer protocol (above 1.2)
retry_delay = default_delay = 1.0
while True:
try:
if self.is_closing():
await self.create_connection(self.timeout)
await self.ensure_server_version()
self._on_connect_cb()
if (perf_counter() - self.last_send) > self.max_seconds_idle or self.response_time is None:
await self.ensure_server_version()
retry_delay = default_delay
except RPCError as e:
await self.close()
log.debug("Server error, ignoring for 1h: %s:%d -- %s", *self.server, e.message)
retry_delay = 60 * 60
except IncompatibleWalletServerError:
await self.close()
retry_delay = 60 * 60
log.debug("Wallet server has an incompatible version, retrying in 1h: %s:%d", *self.server)
except (asyncio.TimeoutError, OSError):
await self.close()
retry_delay = min(60, retry_delay * 2)
log.debug("Wallet server timeout (retry in %s seconds): %s:%d", retry_delay, *self.server)
try:
await asyncio.wait_for(self.trigger_urgent_reconnect.wait(), timeout=retry_delay)
except asyncio.TimeoutError:
pass
finally:
self.trigger_urgent_reconnect.clear()
async def ensure_server_version(self, required=None, timeout=3):
required = required or self.network.PROTOCOL_VERSION
response = await asyncio.wait_for(
@@ -102,25 +133,6 @@ class ClientSession(BaseClientSession):
raise IncompatibleWalletServerError(*self.server)
return response
async def keepalive_loop(self, timeout=3, max_idle=60):
try:
while True:
now = perf_counter()
if min(self.last_send, self.last_packet_received) + max_idle < now:
await asyncio.wait_for(
self.send_request('server.ping', []), timeout=timeout
)
else:
await asyncio.sleep(max(0, max_idle - (now - self.last_send)))
except Exception as err:
if isinstance(err, asyncio.CancelledError):
log.warning("closing connection to %s:%i", *self.server)
else:
log.exception("lost connection to spv")
finally:
if not self.is_closing():
self._close()
async def create_connection(self, timeout=6):
connector = Connector(lambda: self, *self.server)
start = perf_counter()
@@ -132,14 +144,13 @@ class ClientSession(BaseClientSession):
controller.add(request.args)
def connection_lost(self, exc):
log.warning("Connection lost: %s:%d", *self.server)
log.debug("Connection lost: %s:%d", *self.server)
super().connection_lost(exc)
self.response_time = None
self.connection_latency = None
self._response_samples = 0
# self._on_disconnect_controller.add(True)
if self.network:
self.network.disconnect()
self.pending_amount = 0
self._on_disconnect_controller.add(True)
class Network:
@@ -149,9 +160,10 @@ class Network:
def __init__(self, ledger):
self.ledger = ledger
self.session_pool = SessionPool(network=self, timeout=self.config.get('connect_timeout', 6))
self.client: Optional[ClientSession] = None
self.server_features = None
# self._switch_task: Optional[asyncio.Task] = None
self._switch_task: Optional[asyncio.Task] = None
self.running = False
self.remote_height: int = 0
self._concurrency = asyncio.Semaphore(16)
@@ -170,175 +182,55 @@ class Network:
'blockchain.address.subscribe': self._on_status_controller,
}
self.aiohttp_session: Optional[aiohttp.ClientSession] = None
self._urgent_need_reconnect = asyncio.Event()
self._loop_task: Optional[asyncio.Task] = None
self._keepalive_task: Optional[asyncio.Task] = None
@property
def config(self):
return self.ledger.config
def disconnect(self):
if self._keepalive_task and not self._keepalive_task.done():
self._keepalive_task.cancel()
self._keepalive_task = None
async def start(self):
if not self.running:
self.running = True
self.aiohttp_session = aiohttp.ClientSession()
self.on_header.listen(self._update_remote_height)
self._loop_task = asyncio.create_task(self.network_loop())
self._urgent_need_reconnect.set()
def loop_task_done_callback(f):
try:
f.result()
except Exception:
if self.running:
log.exception("wallet server connection loop crashed")
self._loop_task.add_done_callback(loop_task_done_callback)
async def resolve_spv_dns(self):
hostname_to_ip = {}
ip_to_hostnames = defaultdict(list)
async def resolve_spv(server, port):
try:
server_addr = await resolve_host(server, port, 'udp')
hostname_to_ip[server] = (server_addr, port)
ip_to_hostnames[(server_addr, port)].append(server)
except socket.error:
log.warning("error looking up dns for spv server %s:%i", server, port)
except Exception:
log.exception("error looking up dns for spv server %s:%i", server, port)
# accumulate the dns results
await asyncio.gather(*(resolve_spv(server, port) for (server, port) in self.config['default_servers']))
return hostname_to_ip, ip_to_hostnames
async def get_n_fastest_spvs(self, timeout=3.0) -> Dict[Tuple[str, int], SPVPong]:
loop = asyncio.get_event_loop()
pong_responses = asyncio.Queue()
connection = SPVStatusClientProtocol(pong_responses)
sent_ping_timestamps = {}
_, ip_to_hostnames = await self.resolve_spv_dns()
n = len(ip_to_hostnames)
log.info("%i possible spv servers to try (%i urls in config)", n, len(self.config['default_servers']))
pongs = {}
try:
await loop.create_datagram_endpoint(lambda: connection, ('0.0.0.0', 0))
# could raise OSError if it cant bind
start = perf_counter()
for server in ip_to_hostnames:
connection.ping(server)
sent_ping_timestamps[server] = perf_counter()
while len(pongs) < n:
(remote, ts), pong = await asyncio.wait_for(pong_responses.get(), timeout - (perf_counter() - start))
latency = ts - start
log.info("%s:%i has latency of %sms (available: %s, height: %i)",
'/'.join(ip_to_hostnames[remote]), remote[1], round(latency * 1000, 2),
pong.available, pong.height)
if pong.available:
pongs[remote] = pong
return pongs
except asyncio.TimeoutError:
if pongs:
log.info("%i/%i probed spv servers are accepting connections", len(pongs), len(ip_to_hostnames))
return pongs
else:
log.warning("%i spv status probes failed, retrying later. servers tried: %s",
len(sent_ping_timestamps),
', '.join('/'.join(hosts) + f' ({ip})' for ip, hosts in ip_to_hostnames.items()))
return {random.choice(list(ip_to_hostnames)): None}
finally:
connection.close()
async def connect_to_fastest(self) -> Optional[ClientSession]:
fastest_spvs = await self.get_n_fastest_spvs()
for (host, port) in fastest_spvs:
client = ClientSession(network=self, server=(host, port))
try:
await client.create_connection()
log.warning("Connected to spv server %s:%i", host, port)
await client.ensure_server_version()
return client
except (asyncio.TimeoutError, ConnectionError, OSError, IncompatibleWalletServerError, RPCError):
log.warning("Connecting to %s:%d failed", host, port)
client._close()
return
async def network_loop(self):
sleep_delay = 30
async def switch_forever(self):
while self.running:
await asyncio.wait(
[asyncio.sleep(30), self._urgent_need_reconnect.wait()], return_when=asyncio.FIRST_COMPLETED
)
if self._urgent_need_reconnect.is_set():
sleep_delay = 30
self._urgent_need_reconnect.clear()
if not self.is_connected:
client = await self.connect_to_fastest()
if not client:
log.warning("failed to connect to any spv servers, retrying later")
sleep_delay *= 2
sleep_delay = min(sleep_delay, 300)
continue
log.debug("get spv server features %s:%i", *client.server)
features = await client.send_request('server.features', [])
self.client, self.server_features = client, features
log.info("subscribe to headers %s:%i", *client.server)
if self.is_connected:
await self.client.on_disconnected.first
self.server_features = None
self.client = None
continue
self.client = await self.session_pool.wait_for_fastest_session()
log.info("Switching to SPV wallet server: %s:%d", *self.client.server)
try:
self.server_features = await self.get_server_features()
self._update_remote_height((await self.subscribe_headers(),))
self._on_connected_controller.add(True)
server_str = "%s:%i" % client.server
log.info("maintaining connection to spv server %s", server_str)
self._keepalive_task = asyncio.create_task(self.client.keepalive_loop())
try:
if not self._urgent_need_reconnect.is_set():
await asyncio.wait(
[self._keepalive_task, self._urgent_need_reconnect.wait()],
return_when=asyncio.FIRST_COMPLETED
)
else:
await self._keepalive_task
if self._urgent_need_reconnect.is_set():
log.warning("urgent reconnect needed")
self._urgent_need_reconnect.clear()
if self._keepalive_task and not self._keepalive_task.done():
self._keepalive_task.cancel()
except asyncio.CancelledError:
pass
finally:
self._keepalive_task = None
self.client = None
self.server_features = None
log.warning("connection lost to %s", server_str)
log.info("network loop finished")
log.info("Subscribed to headers: %s:%d", *self.client.server)
except (asyncio.TimeoutError, ConnectionError):
log.info("Switching to %s:%d timed out, closing and retrying.", *self.client.server)
self.client.synchronous_close()
self.server_features = None
self.client = None
async def start(self):
self.running = True
self._switch_task = asyncio.ensure_future(self.switch_forever())
# this may become unnecessary when there are no more bugs found,
# but for now it helps understanding log reports
self._switch_task.add_done_callback(lambda _: log.info("Wallet client switching task stopped."))
self.session_pool.start(self.config['default_servers'])
self.on_header.listen(self._update_remote_height)
async def stop(self):
self.running = False
self.disconnect()
if self._loop_task and not self._loop_task.done():
self._loop_task.cancel()
self._loop_task = None
if self.aiohttp_session:
await self.aiohttp_session.close()
self.aiohttp_session = None
if self.running:
self.running = False
self._switch_task.cancel()
self.session_pool.stop()
@property
def is_connected(self):
return self.client and not self.client.is_closing()
def rpc(self, list_or_method, args, restricted=True, session: Optional[ClientSession] = None):
if session or self.is_connected:
session = session or self.client
def rpc(self, list_or_method, args, restricted=True):
session = self.client if restricted else self.session_pool.fastest_session
if session and not session.is_closing():
return session.send_request(list_or_method, args)
else:
self._urgent_need_reconnect.set()
self.session_pool.trigger_nodelay_connect()
raise ConnectionError("Attempting to send rpc request when connection is not available.")
async def retriable_call(self, function, *args, **kwargs):
@@ -346,15 +238,14 @@ class Network:
while self.running:
if not self.is_connected:
log.warning("Wallet server unavailable, waiting for it to come back and retry.")
self._urgent_need_reconnect.set()
await self.on_connected.first
await self.session_pool.wait_for_fastest_session()
try:
return await function(*args, **kwargs)
except asyncio.TimeoutError:
log.warning("Wallet server call timed out, retrying.")
except ConnectionError:
log.warning("connection error")
pass
raise asyncio.CancelledError() # if we got here, we are shutting down
def _update_remote_height(self, header_args):
@@ -365,10 +256,6 @@ class Network:
restricted = known_height in (None, -1, 0) or 0 > known_height > self.remote_height - 10
return self.rpc('blockchain.transaction.get', [tx_hash], restricted)
def get_transaction_batch(self, txids, restricted=True):
# use any server if its old, otherwise restrict to who gave us the history
return self.rpc('blockchain.transaction.get_batch', txids, restricted)
def get_transaction_and_merkle(self, tx_hash, known_height=None):
# use any server if its old, otherwise restrict to who gave us the history
restricted = known_height in (None, -1, 0) or 0 > known_height > self.remote_height - 10
@@ -398,13 +285,12 @@ class Network:
async def subscribe_address(self, address, *addresses):
addresses = list((address, ) + addresses)
server_addr_and_port = self.client.server_address_and_port # on disconnect client will be None
try:
return await self.rpc('blockchain.address.subscribe', addresses, True)
except asyncio.TimeoutError:
log.warning(
"timed out subscribing to addresses from %s:%i",
*server_addr_and_port
*self.client.server_address_and_port
)
# abort and cancel, we can't lose a subscription, it will happen again on reconnect
if self.client:
@@ -417,27 +303,103 @@ class Network:
def get_server_features(self):
return self.rpc('server.features', (), restricted=True)
def resolve(self, urls, session_override=None):
return self.rpc('blockchain.claimtrie.resolve', urls, False, session_override)
def get_claims_by_ids(self, claim_ids):
return self.rpc('blockchain.claimtrie.getclaimsbyids', claim_ids)
def claim_search(self, session_override=None, **kwargs):
return self.rpc('blockchain.claimtrie.search', kwargs, False, session_override)
def resolve(self, urls):
return self.rpc('blockchain.claimtrie.resolve', urls)
async def new_resolve(self, server, urls):
message = {"method": "resolve", "params": {"urls": urls, "protobuf": True}}
async with self.aiohttp_session.post(server, json=message) as r:
result = await r.json()
return result['result']
def claim_search(self, **kwargs):
return self.rpc('blockchain.claimtrie.search', kwargs)
async def new_claim_search(self, server, **kwargs):
kwargs['protobuf'] = True
message = {"method": "claim_search", "params": kwargs}
async with self.aiohttp_session.post(server, json=message) as r:
result = await r.json()
return result['result']
async def sum_supports(self, server, **kwargs):
message = {"method": "support_sum", "params": kwargs}
async with self.aiohttp_session.post(server, json=message) as r:
result = await r.json()
return result['result']
class SessionPool:
def __init__(self, network: Network, timeout: float):
self.network = network
self.sessions: Dict[ClientSession, Optional[asyncio.Task]] = dict()
self.timeout = timeout
self.new_connection_event = asyncio.Event()
@property
def online(self):
return any(not session.is_closing() for session in self.sessions)
@property
def available_sessions(self):
return (session for session in self.sessions if session.available)
@property
def fastest_session(self):
if not self.online:
return None
return min(
[((session.response_time + session.connection_latency) * (session.pending_amount + 1), session)
for session in self.available_sessions] or [(0, None)],
key=itemgetter(0)
)[1]
def _get_session_connect_callback(self, session: ClientSession):
loop = asyncio.get_event_loop()
def callback():
duplicate_connections = [
s for s in self.sessions
if s is not session and s.server_address_and_port == session.server_address_and_port
]
already_connected = None if not duplicate_connections else duplicate_connections[0]
if already_connected:
self.sessions.pop(session).cancel()
session.synchronous_close()
log.debug("wallet server %s resolves to the same server as %s, rechecking in an hour",
session.server[0], already_connected.server[0])
loop.call_later(3600, self._connect_session, session.server)
return
self.new_connection_event.set()
log.info("connected to %s:%i", *session.server)
return callback
def _connect_session(self, server: Tuple[str, int]):
session = None
for s in self.sessions:
if s.server == server:
session = s
break
if not session:
session = ClientSession(
network=self.network, server=server
)
session._on_connect_cb = self._get_session_connect_callback(session)
task = self.sessions.get(session, None)
if not task or task.done():
task = asyncio.create_task(session.ensure_session())
task.add_done_callback(lambda _: self.ensure_connections())
self.sessions[session] = task
def start(self, default_servers):
for server in default_servers:
self._connect_session(server)
def stop(self):
for session, task in self.sessions.items():
task.cancel()
session.synchronous_close()
self.sessions.clear()
def ensure_connections(self):
for session in self.sessions:
self._connect_session(session.server)
def trigger_nodelay_connect(self):
# used when other parts of the system sees we might have internet back
# bypasses the retry interval
for session in self.sessions:
session.trigger_urgent_reconnect.set()
async def wait_for_fastest_session(self):
while not self.fastest_session:
self.trigger_nodelay_connect()
self.new_connection_event.clear()
await self.new_connection_event.wait()
return self.fastest_session
+1 -8
View File
@@ -11,7 +11,6 @@ import importlib
from binascii import hexlify
from typing import Type, Optional
import urllib.request
from uuid import uuid4
import lbry
from lbry.wallet.server.server import Server
@@ -169,7 +168,6 @@ class SPVNode:
self.server = None
self.hostname = 'localhost'
self.port = 50001 + node_number # avoid conflict with default daemon
self.udp_port = self.port
self.session_timeout = 600
self.rpc_port = '0' # disabled by default
@@ -184,13 +182,10 @@ class SPVNode:
'REORG_LIMIT': '100',
'HOST': self.hostname,
'TCP_PORT': str(self.port),
'UDP_PORT': str(self.udp_port),
'SESSION_TIMEOUT': str(self.session_timeout),
'MAX_QUERY_WORKERS': '0',
'INDIVIDUAL_TAG_INDEXES': '',
'RPC_PORT': self.rpc_port,
'ES_INDEX_PREFIX': uuid4().hex,
'ES_MODE': 'writer',
'RPC_PORT': self.rpc_port
}
if extraconf:
conf.update(extraconf)
@@ -202,8 +197,6 @@ class SPVNode:
async def stop(self, cleanup=True):
try:
await self.server.db.search_index.delete_index()
await self.server.db.search_index.stop()
await self.server.stop()
finally:
cleanup and self.cleanup()
+9 -34
View File
@@ -33,16 +33,13 @@ from asyncio import Event, CancelledError
import logging
import time
from contextlib import suppress
from prometheus_client import Counter, Histogram
from lbry.wallet.tasks import TaskGroup
from .jsonrpc import Request, JSONRPCConnection, JSONRPCv2, JSONRPC, Batch, Notification
from .jsonrpc import RPCError, ProtocolError
from .framing import BadMagicError, BadChecksumError, OversizedPayloadError, BitcoinFramer, NewlineFramer
HISTOGRAM_BUCKETS = (
.005, .01, .025, .05, .075, .1, .25, .5, .75, 1.0, 2.5, 5.0, 7.5, 10.0, 15.0, 20.0, 30.0, 60.0, float('inf')
)
from lbry.wallet.server.prometheus import NOTIFICATION_COUNT, RESPONSE_TIMES, REQUEST_ERRORS_COUNT, RESET_CONNECTIONS
class Connector:
@@ -206,7 +203,7 @@ class SessionBase(asyncio.Protocol):
"""
return self._address
def peer_address_str(self, for_log=True):
def peer_address_str(self):
"""Returns the peer's IP address and port as a human-readable
string."""
if not self._address:
@@ -375,26 +372,10 @@ class BatchRequest:
raise BatchError(self)
NAMESPACE = "wallet_server"
class RPCSession(SessionBase):
"""Base class for protocols where a message can lead to a response,
for example JSON RPC."""
RESPONSE_TIMES = Histogram("response_time", "Response times", namespace=NAMESPACE,
labelnames=("method", "version"), buckets=HISTOGRAM_BUCKETS)
NOTIFICATION_COUNT = Counter("notification", "Number of notifications sent (for subscriptions)",
namespace=NAMESPACE, labelnames=("method", "version"))
REQUEST_ERRORS_COUNT = Counter(
"request_error", "Number of requests that returned errors", namespace=NAMESPACE,
labelnames=("method", "version")
)
RESET_CONNECTIONS = Counter(
"reset_clients", "Number of reset connections by client version",
namespace=NAMESPACE, labelnames=("version",)
)
def __init__(self, *, framer=None, loop=None, connection=None):
super().__init__(framer=framer, loop=loop)
self.connection = connection or self.default_connection()
@@ -407,7 +388,7 @@ class RPCSession(SessionBase):
except MemoryError:
self.logger.warning('received oversized message from %s:%s, dropping connection',
self._address[0], self._address[1])
self.RESET_CONNECTIONS.labels(version=self.client_version).inc()
RESET_CONNECTIONS.labels(version=self.client_version).inc()
self._close()
return
@@ -441,7 +422,7 @@ class RPCSession(SessionBase):
'internal server error')
if isinstance(request, Request):
message = request.send_result(result)
self.RESPONSE_TIMES.labels(
RESPONSE_TIMES.labels(
method=request.method,
version=self.client_version
).observe(time.perf_counter() - start)
@@ -449,7 +430,7 @@ class RPCSession(SessionBase):
await self._send_message(message)
if isinstance(result, Exception):
self._bump_errors()
self.REQUEST_ERRORS_COUNT.labels(
REQUEST_ERRORS_COUNT.labels(
method=request.method,
version=self.client_version
).inc()
@@ -483,17 +464,11 @@ class RPCSession(SessionBase):
raise result
return result
async def send_notification(self, method, args=()) -> bool:
async def send_notification(self, method, args=()):
"""Send an RPC notification over the network."""
message = self.connection.send_notification(Notification(method, args))
self.NOTIFICATION_COUNT.labels(method=method, version=self.client_version).inc()
try:
await self._send_message(message)
return True
except asyncio.TimeoutError:
self.logger.info("timeout sending address notification to %s", self.peer_address_str(for_log=True))
self.abort()
return False
NOTIFICATION_COUNT.labels(method=method, version=self.client_version).inc()
await self._send_message(message)
def send_batch(self, raise_errors=False):
"""Return a BatchRequest. Intended to be used like so:
-3
View File
@@ -32,13 +32,10 @@ import inspect
# other_params: None means cannot be called with keyword arguments only
# any means any name is good
from functools import lru_cache
SignatureInfo = namedtuple('SignatureInfo', 'min_args max_args '
'required_names other_names')
@lru_cache(256)
def signature_info(func):
params = inspect.signature(func).parameters
min_args = max_args = 0
-27
View File
@@ -438,17 +438,6 @@ class OutputScript(Script):
SUPPORT_CLAIM_OPCODES + PAY_SCRIPT_HASH.opcodes
))
SUPPORT_CLAIM_DATA_OPCODES = (
OP_SUPPORT_CLAIM, PUSH_SINGLE('claim_name'), PUSH_SINGLE('claim_id'), PUSH_SINGLE('support'),
OP_2DROP, OP_2DROP
)
SUPPORT_CLAIM_DATA_PUBKEY = Template('support_claim+data+pay_pubkey_hash', (
SUPPORT_CLAIM_DATA_OPCODES + PAY_PUBKEY_HASH.opcodes
))
SUPPORT_CLAIM_DATA_SCRIPT = Template('support_claim+data+pay_script_hash', (
SUPPORT_CLAIM_DATA_OPCODES + PAY_SCRIPT_HASH.opcodes
))
UPDATE_CLAIM_OPCODES = (
OP_UPDATE_CLAIM, PUSH_SINGLE('claim_name'), PUSH_SINGLE('claim_id'), PUSH_SINGLE('claim'),
OP_2DROP, OP_2DROP
@@ -485,8 +474,6 @@ class OutputScript(Script):
CLAIM_NAME_SCRIPT,
SUPPORT_CLAIM_PUBKEY,
SUPPORT_CLAIM_SCRIPT,
SUPPORT_CLAIM_DATA_PUBKEY,
SUPPORT_CLAIM_DATA_SCRIPT,
UPDATE_CLAIM_PUBKEY,
UPDATE_CLAIM_SCRIPT,
SELL_CLAIM, SELL_SCRIPT,
@@ -540,16 +527,6 @@ class OutputScript(Script):
'pubkey_hash': pubkey_hash
})
@classmethod
def pay_support_data_pubkey_hash(
cls, claim_name: bytes, claim_id: bytes, support, pubkey_hash: bytes):
return cls(template=cls.SUPPORT_CLAIM_DATA_PUBKEY, values={
'claim_name': claim_name,
'claim_id': claim_id,
'support': support,
'pubkey_hash': pubkey_hash
})
@classmethod
def sell_script(cls, price):
return cls(template=cls.SELL_SCRIPT, values={
@@ -598,10 +575,6 @@ class OutputScript(Script):
def is_support_claim(self):
return self.template.name.startswith('support_claim+')
@property
def is_support_claim_data(self):
return self.template.name.startswith('support_claim+data+')
@property
def is_sell_claim(self):
return self.template.name.startswith('sell_claim+')
+82 -120
View File
@@ -3,14 +3,14 @@ import asyncio
from struct import pack, unpack
from concurrent.futures.thread import ThreadPoolExecutor
from typing import Optional
from prometheus_client import Gauge, Histogram
import lbry
from lbry.schema.claim import Claim
from lbry.wallet.server.db.writer import SQLDB
from lbry.wallet.server.daemon import DaemonError
from lbry.wallet.server.hash import hash_to_hex_str, HASHX_LEN
from lbry.wallet.server.util import chunks, class_logger
from lbry.wallet.server.leveldb import FlushData
from lbry.wallet.server.udp import StatusServer
from lbry.wallet.server.prometheus import BLOCK_COUNT, BLOCK_UPDATE_TIMES, REORG_COUNT
class Prefetcher:
@@ -129,12 +129,6 @@ class ChainError(Exception):
"""Raised on error processing blocks."""
NAMESPACE = "wallet_server"
HISTOGRAM_BUCKETS = (
.005, .01, .025, .05, .075, .1, .25, .5, .75, 1.0, 2.5, 5.0, 7.5, 10.0, 15.0, 20.0, 30.0, 60.0, float('inf')
)
class BlockProcessor:
"""Process blocks and update the DB state to match.
@@ -142,16 +136,6 @@ class BlockProcessor:
Coordinate backing up in case of chain reorganisations.
"""
block_count_metric = Gauge(
"block_count", "Number of processed blocks", namespace=NAMESPACE
)
block_update_time_metric = Histogram(
"block_time", "Block update times", namespace=NAMESPACE, buckets=HISTOGRAM_BUCKETS
)
reorg_count_metric = Gauge(
"reorg_count", "Number of reorgs", namespace=NAMESPACE
)
def __init__(self, env, db, daemon, notifications):
self.env = env
self.db = db
@@ -171,8 +155,7 @@ class BlockProcessor:
# Caches of unflushed items.
self.headers = []
self.block_hashes = []
self.block_txs = []
self.tx_hashes = []
self.undo_infos = []
# UTXO cache
@@ -184,8 +167,6 @@ class BlockProcessor:
self.state_lock = asyncio.Lock()
self.search_cache = {}
self.history_cache = {}
self.status_server = StatusServer()
async def run_in_thread_with_lock(self, func, *args):
# Run in a thread to prevent blocking. Shielded so that
@@ -214,24 +195,16 @@ class BlockProcessor:
if hprevs == chain:
start = time.perf_counter()
await self.run_in_thread_with_lock(self.advance_blocks, blocks)
if self.sql:
await self.db.search_index.claim_consumer(self.sql.claim_producer())
for cache in self.search_cache.values():
cache.clear()
self.history_cache.clear()
self.notifications.notified_mempool_txs.clear()
await self._maybe_flush()
processed_time = time.perf_counter() - start
self.block_count_metric.set(self.height)
self.block_update_time_metric.observe(processed_time)
self.status_server.set_height(self.db.fs_height, self.db.db_tip)
BLOCK_COUNT.set(self.height)
BLOCK_UPDATE_TIMES.observe(processed_time)
if not self.db.first_sync:
s = '' if len(blocks) == 1 else 's'
self.logger.info('processed {:,d} block{} in {:.1f}s'.format(len(blocks), s, processed_time))
if self._caught_up_event.is_set():
if self.sql:
await self.db.search_index.apply_filters(self.sql.blocked_streams, self.sql.blocked_channels,
self.sql.filtered_streams, self.sql.filtered_channels)
await self.notifications.on_block(self.touched, self.height)
self.touched = set()
elif hprevs[0] != chain[0]:
@@ -255,12 +228,12 @@ class BlockProcessor:
self.logger.info('chain reorg detected')
else:
self.logger.info(f'faking a reorg of {count:,d} blocks')
await self.flush(True)
async def get_raw_blocks(last_height, hex_hashes):
heights = range(last_height, last_height - len(hex_hashes), -1)
try:
blocks = [await self.db.read_raw_block(height) for height in heights]
blocks = [self.db.read_raw_block(height) for height in heights]
self.logger.info(f'read {len(blocks)} blocks from disk')
return blocks
except FileNotFoundError:
@@ -272,27 +245,17 @@ class BlockProcessor:
self.touched.discard(None)
self.db.flush_backup(self.flush_data(), self.touched)
try:
await self.flush(True)
start, last, hashes = await self.reorg_hashes(count)
# Reverse and convert to hex strings.
hashes = [hash_to_hex_str(hash) for hash in reversed(hashes)]
self.logger.info("reorg %i block hashes", len(hashes))
for hex_hashes in chunks(hashes, 50):
raw_blocks = await get_raw_blocks(last, hex_hashes)
self.logger.info("got %i raw blocks", len(raw_blocks))
await self.run_in_thread_with_lock(self.backup_blocks, raw_blocks)
await self.run_in_thread_with_lock(flush_backup)
last -= len(raw_blocks)
await self.prefetcher.reset_height(self.height)
self.reorg_count_metric.inc()
except:
self.logger.exception("boom")
raise
finally:
self.logger.info("done with reorg")
start, last, hashes = await self.reorg_hashes(count)
# Reverse and convert to hex strings.
hashes = [hash_to_hex_str(hash) for hash in reversed(hashes)]
for hex_hashes in chunks(hashes, 50):
raw_blocks = await get_raw_blocks(last, hex_hashes)
await self.run_in_thread_with_lock(self.backup_blocks, raw_blocks)
await self.run_in_thread_with_lock(flush_backup)
last -= len(raw_blocks)
await self.run_in_thread_with_lock(self.db.sql.delete_claims_above_height, self.height)
await self.prefetcher.reset_height(self.height)
REORG_COUNT.inc()
async def reorg_hashes(self, count):
"""Return a pair (start, last, hashes) of blocks to back up during a
@@ -355,8 +318,8 @@ class BlockProcessor:
def flush_data(self):
"""The data for a flush. The lock must be taken."""
assert self.state_lock.locked()
return FlushData(self.height, self.tx_count, self.headers, self.block_hashes,
self.block_txs, self.undo_infos, self.utxo_cache,
return FlushData(self.height, self.tx_count, self.headers,
self.tx_hashes, self.undo_infos, self.utxo_cache,
self.db_deletes, self.tip)
async def flush(self, flush_utxos):
@@ -370,9 +333,11 @@ class BlockProcessor:
# performed on the DB.
if self._caught_up_event.is_set():
await self.flush(True)
elif time.perf_counter() > self.next_cache_check:
await self.flush(True)
self.next_cache_check = time.perf_counter() + 30
elif time.time() > self.next_cache_check:
flush_arg = self.check_cache_size()
if flush_arg is not None:
await self.flush(flush_arg)
self.next_cache_check = time.time() + 30
def check_cache_size(self):
"""Flush a cache if it gets too big."""
@@ -411,8 +376,7 @@ class BlockProcessor:
for block in blocks:
height += 1
undo_info = self.advance_txs(
height, block.transactions, self.coin.electrum_header(block.header, height),
self.coin.header_hash(block.header)
height, block.transactions, self.coin.electrum_header(block.header, height)
)
if height >= min_height:
self.undo_infos.append((undo_info, height))
@@ -423,26 +387,25 @@ class BlockProcessor:
self.headers.extend(headers)
self.tip = self.coin.header_hash(headers[-1])
def advance_txs(self, height, txs, header, block_hash):
self.block_hashes.append(block_hash)
self.block_txs.append((b''.join(tx_hash for tx, tx_hash in txs), [tx.raw for tx, _ in txs]))
undo_info = []
tx_num = self.tx_count
hashXs_by_tx = []
def advance_txs(self, height, txs, header):
self.tx_hashes.append(b''.join(tx_hash for tx, tx_hash in txs))
# Use local vars for speed in the loops
undo_info = []
tx_num = self.tx_count
script_hashX = self.coin.hashX_from_script
s_pack = pack
put_utxo = self.utxo_cache.__setitem__
spend_utxo = self.spend_utxo
undo_info_append = undo_info.append
update_touched = self.touched.update
append_hashX_by_tx = hashXs_by_tx.append
hashX_from_script = self.coin.hashX_from_script
hashXs_by_tx = []
append_hashXs = hashXs_by_tx.append
for tx, tx_hash in txs:
hashXs = []
append_hashX = hashXs.append
tx_numb = pack('<I', tx_num)
tx_numb = s_pack('<I', tx_num)
# Spend the inputs
for txin in tx.inputs:
@@ -455,17 +418,18 @@ class BlockProcessor:
# Add the new UTXOs
for idx, txout in enumerate(tx.outputs):
# Get the hashX. Ignore unspendable outputs
hashX = hashX_from_script(txout.pk_script)
hashX = script_hashX(txout.pk_script)
if hashX:
append_hashX(hashX)
put_utxo(tx_hash + pack('<H', idx), hashX + tx_numb + pack('<Q', txout.value))
put_utxo(tx_hash + s_pack('<H', idx),
hashX + tx_numb + s_pack('<Q', txout.value))
append_hashX_by_tx(hashXs)
append_hashXs(hashXs)
update_touched(hashXs)
self.db.total_transactions.append(tx_hash)
tx_num += 1
self.db.history.add_unflushed(hashXs_by_tx, self.tx_count)
self.tx_count = tx_num
self.db.tx_counts.append(tx_num)
@@ -482,7 +446,6 @@ class BlockProcessor:
coin = self.coin
for raw_block in raw_blocks:
self.logger.info("backup block %i", self.height)
# Check and update self.tip
block = coin.block(raw_block, self.height)
header_hash = coin.header_hash(block.header)
@@ -508,16 +471,20 @@ class BlockProcessor:
# Use local vars for speed in the loops
s_pack = pack
put_utxo = self.utxo_cache.__setitem__
spend_utxo = self.spend_utxo
script_hashX = self.coin.hashX_from_script
touched = self.touched
undo_entry_len = 12 + HASHX_LEN
for tx, tx_hash in reversed(txs):
for idx, txout in enumerate(tx.outputs):
# Spend the TX outputs. Be careful with unspendable
# outputs - we didn't save those in the first place.
hashX = self.coin.hashX_from_script(txout.pk_script)
hashX = script_hashX(txout.pk_script)
if hashX:
cache_value = self.spend_utxo(tx_hash, idx)
self.touched.add(cache_value[:-12])
cache_value = spend_utxo(tx_hash, idx)
touched.add(cache_value[:-12])
# Restore the inputs
for txin in reversed(tx.inputs):
@@ -525,10 +492,9 @@ class BlockProcessor:
continue
n -= undo_entry_len
undo_item = undo_info[n:n + undo_entry_len]
self.utxo_cache[txin.prev_hash + s_pack('<H', txin.prev_idx)] = undo_item
self.touched.add(undo_item[:-12])
self.db.total_transactions.pop()
put_utxo(txin.prev_hash + s_pack('<H', txin.prev_idx),
undo_item)
touched.add(undo_item[:-12])
assert n == 0
self.tx_count -= len(txs)
@@ -607,17 +573,13 @@ class BlockProcessor:
prefix = b'h' + tx_hash[:4] + idx_packed
candidates = {db_key: hashX for db_key, hashX
in self.db.utxo_db.iterator(prefix=prefix)}
for hdb_key, hashX in candidates.items():
tx_num_packed = hdb_key[-4:]
if len(candidates) > 1:
if len(candidates) > 1:
tx_num, = unpack('<I', tx_num_packed)
try:
hash, height = self.db.fs_tx_hash(tx_num)
except IndexError:
self.logger.error("data integrity error for hashx history: %s missing tx #%s (%s:%s)",
hashX.hex(), tx_num, hash_to_hex_str(tx_hash), tx_idx)
continue
hash, height = self.db.fs_tx_hash(tx_num)
if hash != tx_hash:
assert hash is not None # Should always be found
continue
@@ -626,17 +588,12 @@ class BlockProcessor:
# Value: the UTXO value as a 64-bit unsigned integer
udb_key = b'u' + hashX + hdb_key[-6:]
utxo_value_packed = self.db.utxo_db.get(udb_key)
if utxo_value_packed is None:
self.logger.warning(
"%s:%s is not found in UTXO db for %s", hash_to_hex_str(tx_hash), tx_idx, hash_to_hex_str(hashX)
)
raise ChainError(f"{hash_to_hex_str(tx_hash)}:{tx_idx} is not found in UTXO db for {hash_to_hex_str(hashX)}")
# Remove both entries for this UTXO
self.db_deletes.append(hdb_key)
self.db_deletes.append(udb_key)
return hashX + tx_num_packed + utxo_value_packed
if utxo_value_packed:
# Remove both entries for this UTXO
self.db_deletes.append(hdb_key)
self.db_deletes.append(udb_key)
return hashX + tx_num_packed + utxo_value_packed
self.logger.error('UTXO {hash_to_hex_str(tx_hash)} / {tx_idx} not found in "h" table')
raise ChainError('UTXO {} / {:,d} not found in "h" table'
.format(hash_to_hex_str(tx_hash), tx_idx))
@@ -649,16 +606,12 @@ class BlockProcessor:
self._caught_up_event.set()
await self.blocks_event.wait()
self.blocks_event.clear()
if self.reorg_count: # this could only happen by calling the reorg rpc
if self.reorg_count:
await self.reorg_chain(self.reorg_count)
self.reorg_count = 0
else:
blocks = self.prefetcher.get_prefetched_blocks()
try:
await self.check_and_advance_blocks(blocks)
except Exception:
self.logger.exception("error while processing txs")
raise
await self.check_and_advance_blocks(blocks)
async def _first_caught_up(self):
self.logger.info(f'caught up to height {self.height}')
@@ -692,11 +645,9 @@ class BlockProcessor:
disk before exiting, as otherwise a significant amount of work
could be lost.
"""
self._caught_up_event = caught_up_event
try:
await self._first_open_dbs()
self.status_server.set_height(self.db.fs_height, self.db.db_tip)
await asyncio.wait([
self.prefetcher.main_loop(self.height),
self._process_prefetched_blocks()
@@ -707,7 +658,6 @@ class BlockProcessor:
self.logger.exception("Block processing failed!")
raise
finally:
self.status_server.stop()
# Shut down block processing
self.logger.info('flushing to DB for a clean shutdown...')
await self.flush(True)
@@ -727,6 +677,7 @@ class BlockProcessor:
class Timer:
def __init__(self, name):
self.name = name
self.total = 0
@@ -789,27 +740,38 @@ class LBRYBlockProcessor(BlockProcessor):
self.timer = Timer('BlockProcessor')
def advance_blocks(self, blocks):
if self.sql:
self.sql.begin()
self.sql.begin()
try:
self.timer.run(super().advance_blocks, blocks)
except:
self.logger.exception(f'Error while advancing transaction in new block.')
raise
finally:
if self.sql:
self.sql.commit()
if self.sql and self.db.first_sync and self.height == self.daemon.cached_height():
self.sql.commit()
if self.db.first_sync and self.height == self.daemon.cached_height():
self.timer.run(self.sql.execute, self.sql.SEARCH_INDEXES, timer_name='executing SEARCH_INDEXES')
if self.env.individual_tag_indexes:
self.timer.run(self.sql.execute, self.sql.TAG_INDEXES, timer_name='executing TAG_INDEXES')
self.timer.run(self.sql.execute, self.sql.LANGUAGE_INDEXES, timer_name='executing LANGUAGE_INDEXES')
def advance_txs(self, height, txs, header, block_hash):
def advance_txs(self, height, txs, header):
timer = self.timer.sub_timers['advance_blocks']
undo = timer.run(super().advance_txs, height, txs, header, block_hash, timer_name='super().advance_txs')
if self.sql:
timer.run(self.sql.advance_txs, height, txs, header, self.daemon.cached_height(), forward_timer=True)
undo = timer.run(super().advance_txs, height, txs, header, timer_name='super().advance_txs')
timer.run(self.sql.advance_txs, height, txs, header, self.daemon.cached_height(), forward_timer=True)
if (height % 10000 == 0 or not self.db.first_sync) and self.logger.isEnabledFor(10):
self.timer.show(height=height)
return undo
def _checksig(self, value, address):
try:
claim_dict = Claim.from_bytes(value)
cert_id = claim_dict.signing_channel_hash
if not self.should_validate_signatures:
return cert_id
if cert_id:
cert_claim = self.db.get_claim_info(cert_id)
if cert_claim:
certificate = Claim.from_bytes(cert_claim.value)
claim_dict.validate_signature(address, certificate)
return cert_id
except Exception:
pass
+1 -3
View File
@@ -8,7 +8,7 @@ from lbry.wallet.server.server import Server
def get_argument_parser():
parser = argparse.ArgumentParser(
prog="lbry-hub"
prog="torba-server"
)
parser.add_argument("spvserver", type=str, help="Python class path to SPV server implementation.",
nargs="?", default="lbry.wallet.server.coin.LBC")
@@ -27,8 +27,6 @@ def main():
coin_class = get_coin_class(args.spvserver)
logging.basicConfig(level=logging.INFO)
logging.info('lbry.server starting')
logging.getLogger('aiohttp').setLevel(logging.WARNING)
logging.getLogger('elasticsearch').setLevel(logging.WARNING)
try:
server = Server(Env(coin_class))
server.run()
+11 -24
View File
@@ -3,14 +3,14 @@ import itertools
import json
import time
from functools import wraps
from pylru import lrucache
import aiohttp
from prometheus_client import Gauge, Histogram
from lbry.utils import LRUCacheWithMetrics
from lbry.wallet.rpc.jsonrpc import RPCError
from lbry.wallet.server.util import hex_to_bytes, class_logger
from lbry.wallet.rpc import JSONRPC
from lbry.wallet.server.prometheus import LBRYCRD_REQUEST_TIMES, LBRYCRD_PENDING_COUNT
class DaemonError(Exception):
"""Raised when the daemon returns an error in its results."""
@@ -24,23 +24,12 @@ class WorkQueueFullError(Exception):
"""Internal - when the daemon's work queue is full."""
NAMESPACE = "wallet_server"
class Daemon:
"""Handles connections to a daemon at the given URL."""
WARMING_UP = -28
id_counter = itertools.count()
lbrycrd_request_time_metric = Histogram(
"lbrycrd_request", "lbrycrd requests count", namespace=NAMESPACE, labelnames=("method",)
)
lbrycrd_pending_count_metric = Gauge(
"lbrycrd_pending_count", "Number of lbrycrd rpcs that are in flight", namespace=NAMESPACE,
labelnames=("method",)
)
def __init__(self, coin, url, max_workqueue=10, init_retry=0.25,
max_retry=4.0):
self.coin = coin
@@ -54,8 +43,8 @@ class Daemon:
self._height = None
self.available_rpcs = {}
self.connector = aiohttp.TCPConnector()
self._block_hash_cache = LRUCacheWithMetrics(100000)
self._block_cache = LRUCacheWithMetrics(2 ** 16, metric_name='block', namespace=NAMESPACE)
self._block_hash_cache = lrucache(100000)
self._block_cache = lrucache(10000)
async def close(self):
if self.connector:
@@ -140,7 +129,7 @@ class Daemon:
while True:
try:
for method in methods:
self.lbrycrd_pending_count_metric.labels(method=method).inc()
LBRYCRD_PENDING_COUNT.labels(method=method).inc()
result = await self._send_data(data)
result = processor(result)
if on_good_message:
@@ -165,7 +154,7 @@ class Daemon:
on_good_message = 'running normally'
finally:
for method in methods:
self.lbrycrd_pending_count_metric.labels(method=method).dec()
LBRYCRD_PENDING_COUNT.labels(method=method).dec()
await asyncio.sleep(retry)
retry = max(min(self.max_retry, retry * 2), self.init_retry)
@@ -186,7 +175,7 @@ class Daemon:
if params:
payload['params'] = params
result = await self._send(payload, processor)
self.lbrycrd_request_time_metric.labels(method=method).observe(time.perf_counter() - start)
LBRYCRD_REQUEST_TIMES.labels(method=method).observe(time.perf_counter() - start)
return result
async def _send_vector(self, method, params_iterable, replace_errs=False):
@@ -211,7 +200,7 @@ class Daemon:
result = []
if payload:
result = await self._send(payload, processor)
self.lbrycrd_request_time_metric.labels(method=method).observe(time.perf_counter() - start)
LBRYCRD_REQUEST_TIMES.labels(method=method).observe(time.perf_counter()-start)
return result
async def _is_rpc_available(self, method):
@@ -249,10 +238,8 @@ class Daemon:
async def deserialised_block(self, hex_hash):
"""Return the deserialised block with the given hex hash."""
if hex_hash not in self._block_cache:
block = await self._send_single('getblock', (hex_hash, True))
self._block_cache[hex_hash] = block
return block
if not self._block_cache.get(hex_hash):
self._block_cache[hex_hash] = await self._send_single('getblock', (hex_hash, True))
return self._block_cache[hex_hash]
async def raw_blocks(self, hex_hashes):
+202 -401
View File
@@ -1,8 +1,7 @@
CLAIM_TYPES = {
'stream': 1,
'channel': 2,
'repost': 3,
'collection': 4,
'repost': 3
}
STREAM_TYPES = {
@@ -11,410 +10,212 @@ STREAM_TYPES = {
'image': 3,
'document': 4,
'binary': 5,
'model': 6,
}
# 9/21/2020
MOST_USED_TAGS = {
"gaming",
"people & blogs",
"entertainment",
"music",
"pop culture",
"education",
"technology",
"blockchain",
"news",
"funny",
"science & technology",
"learning",
"gameplay",
"news & politics",
"comedy",
"bitcoin",
"beliefs",
"nature",
"art",
"economics",
"film & animation",
"lets play",
"games",
"sports",
"howto & style",
"game",
"cryptocurrency",
"playstation 4",
"automotive",
"crypto",
"mature",
"sony interactive entertainment",
"walkthrough",
"tutorial",
"video game",
"weapons",
"playthrough",
"pc",
"anime",
"how to",
"btc",
"fun",
"ethereum",
"food",
"travel & events",
"minecraft",
"science",
"autos & vehicles",
"play",
"politics",
"commentary",
"twitch",
"ps4live",
"love",
"ps4",
"nonprofits & activism",
"ps4share",
"fortnite",
"xbox",
"porn",
"video games",
"trump",
"español",
"money",
"music video",
"nintendo",
"movie",
"coronavirus",
"donald trump",
"steam",
"trailer",
"android",
"podcast",
"xbox one",
"survival",
"audio",
"linux",
"travel",
"funny moments",
"litecoin",
"animation",
"gamer",
"lets",
"playstation",
"bitcoin news",
"history",
"xxx",
"fox news",
"dance",
"god",
"adventure",
"liberal",
"2020",
"horror",
"government",
"freedom",
"reaction",
"meme",
"photography",
"truth",
"health",
"lbry",
"family",
"online",
"eth",
"crypto news",
"diy",
"trading",
"gold",
"memes",
"world",
"space",
"lol",
"covid-19",
"rpg",
"humor",
"democrat",
"film",
"call of duty",
"tech",
"religion",
"conspiracy",
"rap",
"cnn",
"hangoutsonair",
"unboxing",
"fiction",
"conservative",
"cars",
"hoa",
"epic",
"programming",
"progressive",
"cryptocurrency news",
"classical",
"jesus",
"movies",
"book",
"ps3",
"republican",
"fitness",
"books",
"multiplayer",
"animals",
"pokemon",
"bitcoin price",
"facebook",
"sharefactory",
"criptomonedas",
"cod",
"bible",
"business",
"stream",
"comics",
"how",
"fail",
"nsfw",
"new music",
"satire",
"pets & animals",
"computer",
"classical music",
"indie",
"musica",
"msnbc",
"fps",
"mod",
"sport",
"sony",
"ripple",
"auto",
"rock",
"marvel",
"complete",
"mining",
"political",
"mobile",
"pubg",
"hip hop",
"flat earth",
"xbox 360",
"reviews",
"vlogging",
"latest news",
"hack",
"tarot",
"iphone",
"media",
"cute",
"christian",
"free speech",
"trap",
"war",
"remix",
"ios",
"xrp",
"spirituality",
"song",
"league of legends",
"cat"
'model': 6
}
MATURE_TAGS = [
'nsfw', 'porn', 'xxx', 'mature', 'adult', 'sex'
]
def normalize_tag(tag):
return tag.replace(" ", "_").replace("&", "and").replace("-", "_")
COMMON_TAGS = {
tag: normalize_tag(tag) for tag in list(MOST_USED_TAGS)
"gaming": "gaming",
"people & blogs": "people_and_blogs",
"pop culture": "pop_culture",
"entertainment": "entertainment",
"technology": "technology",
"music": "music",
"funny": "funny",
"education": "education",
"learning": "learning",
"news": "news",
"gameplay": "gameplay",
"science & technology": "science_and_technology",
"playstation 4": "playstation_4",
"beliefs": "beliefs",
"nature": "nature",
"news & politics": "news_and_politics",
"comedy": "comedy",
"games": "games",
"sony interactive entertainment": "sony_interactive_entertainment",
"film & animation": "film_and_animation",
"game": "game",
"howto & style": "howto_and_style",
"weapons": "weapons",
"blockchain": "blockchain",
"video game": "video_game",
"sports": "sports",
"walkthrough": "walkthrough",
"ps4live": "ps4live",
"art": "art",
"pc": "pc",
"economics": "economics",
"automotive": "automotive",
"minecraft": "minecraft",
"playthrough": "playthrough",
"ps4share": "ps4share",
"tutorial": "tutorial",
"play": "play",
"twitch": "twitch",
"how to": "how_to",
"ps4": "ps4",
"bitcoin": "bitcoin",
"fortnite": "fortnite",
"commentary": "commentary",
"lets play": "lets_play",
"fun": "fun",
"politics": "politics",
"xbox": "xbox",
"autos & vehicles": "autos_and_vehicles",
"travel & events": "travel_and_events",
"food": "food",
"science": "science",
"mature": "mature",
"xbox one": "xbox_one",
"liberal": "liberal",
"democrat": "democrat",
"progressive": "progressive",
"survival": "survival",
"nonprofits & activism": "nonprofits_and_activism",
"cryptocurrency": "cryptocurrency",
"playstation": "playstation",
"nintendo": "nintendo",
"government": "government",
"steam": "steam",
"podcast": "podcast",
"horror": "horror",
"conservative": "conservative",
"reaction": "reaction",
"trailer": "trailer",
"love": "love",
"cnn": "cnn",
"republican": "republican",
"gamer": "gamer",
"political": "political",
"hangoutsonair": "hangoutsonair",
"hoa": "hoa",
"msnbc": "msnbc",
"cbs": "cbs",
"donald trump": "donald_trump",
"fiction": "fiction",
"fox news": "fox_news",
"anime": "anime",
"crypto": "crypto",
"ethereum": "ethereum",
"call of duty": "call_of_duty",
"multiplayer": "multiplayer",
"android": "android",
"epic": "epic",
"rpg": "rpg",
"adventure": "adventure",
"secular talk": "secular_talk",
"btc": "btc",
"atheist": "atheist",
"atheism": "atheism",
"ps3": "ps3",
"video games": "video_games",
"cod": "cod",
"agnostic": "agnostic",
"movie": "movie",
"online": "online",
"fps": "fps",
"mod": "mod",
"reviews": "reviews",
"sharefactory": "sharefactory",
"world": "world",
"space": "space",
"hilarious": "hilarious",
"stream": "stream",
"lol": "lol",
"sony": "sony",
"god": "god",
"lets": "lets",
"dance": "dance",
"pvp": "pvp",
"tech": "tech",
"zombies": "zombies",
"pokemon": "pokemon",
"fail": "fail",
"xbox 360": "xbox_360",
"film": "film",
"unboxing": "unboxing",
"animation": "animation",
"travel": "travel",
"money": "money",
"wwe": "wwe",
"how": "how",
"mods": "mods",
"pubg": "pubg",
"indie": "indie",
"strategy": "strategy",
"history": "history",
"rap": "rap",
"ios": "ios",
"sony computer entertainment": "sony_computer_entertainment",
"mobile": "mobile",
"trump": "trump",
"flat earth": "flat_earth",
"hack": "hack",
"trap": "trap",
"fox": "fox",
"vlogging": "vlogging",
"news radio": "news_radio",
"humor": "humor",
"facebook": "facebook",
"edm": "edm",
"fitness": "fitness",
"vaping": "vaping",
"hip hop": "hip_hop",
"secular": "secular",
"jesus": "jesus",
"vape": "vape",
"song": "song",
"remix": "remix",
"guitar": "guitar",
"daily": "daily",
"mining": "mining",
"diy": "diy",
"videogame": "videogame",
"pets & animals": "pets_and_animals",
"funny moments": "funny_moments",
"religion": "religion",
"death": "death",
"media": "media",
"nbc": "nbc",
"war": "war",
"freedom": "freedom",
"viral": "viral",
"meme": "meme",
"family": "family",
"gold": "gold",
"photography": "photography",
"chill": "chill",
"zombie": "zombie",
"computer": "computer",
"sniper": "sniper",
"bible": "bible",
"linux": "linux",
"overwatch": "overwatch",
"pro": "pro",
"dragon": "dragon",
"litecoin": "litecoin",
"gta": "gta",
"iphone": "iphone",
"house": "house",
"bass": "bass",
"bitcoin news": "bitcoin_news",
"wii": "wii",
"crash": "crash",
"league of legends": "league_of_legends",
"grand theft auto v": "grand_theft_auto_v",
"mario": "mario",
"mmorpg": "mmorpg",
"satire": "satire",
"fire": "fire",
"racing": "racing",
"apple": "apple",
"health": "health",
"instrumental": "instrumental",
"destiny": "destiny",
"truth": "truth",
"race": "race"
}
INDEXED_LANGUAGES = [
'none',
'en',
'aa',
'ab',
'ae',
'af',
'ak',
'am',
'an',
'ar',
'as',
'av',
'ay',
'az',
'ba',
'be',
'bg',
'bh',
'bi',
'bm',
'bn',
'bo',
'br',
'bs',
'ca',
'ce',
'ch',
'co',
'cr',
'cs',
'cu',
'cv',
'cy',
'da',
'de',
'dv',
'dz',
'ee',
'el',
'eo',
'es',
'et',
'eu',
'fa',
'ff',
'fi',
'fj',
'fo',
'fr',
'fy',
'ga',
'gd',
'gl',
'gn',
'gu',
'gv',
'ha',
'he',
'hi',
'ho',
'hr',
'ht',
'hu',
'hy',
'hz',
'ia',
'id',
'ie',
'ig',
'ii',
'ik',
'io',
'is',
'it',
'iu',
'ja',
'jv',
'ka',
'kg',
'ki',
'kj',
'kk',
'kl',
'km',
'kn',
'ko',
'kr',
'ks',
'ku',
'kv',
'kw',
'ky',
'la',
'lb',
'lg',
'li',
'ln',
'lo',
'lt',
'lu',
'lv',
'mg',
'mh',
'mi',
'mk',
'ml',
'mn',
'mr',
'ms',
'mt',
'my',
'na',
'nb',
'nd',
'ne',
'ng',
'nl',
'nn',
'no',
'nr',
'nv',
'ny',
'oc',
'oj',
'om',
'or',
'os',
'pa',
'pi',
'pl',
'ps',
'pt',
'qu',
'rm',
'rn',
'ro',
'ru',
'rw',
'sa',
'sc',
'sd',
'se',
'sg',
'si',
'sk',
'sl',
'sm',
'sn',
'so',
'sq',
'sr',
'ss',
'st',
'su',
'sv',
'sw',
'ta',
'te',
'tg',
'th',
'ti',
'tk',
'tl',
'tn',
'to',
'tr',
'ts',
'tt',
'tw',
'ty',
'ug',
'uk',
'ur',
'uz',
've',
'vi',
'vo',
'wa',
'wo',
'xh',
'yi',
'yo',
'za',
'zh',
'zu'
]
@@ -1 +0,0 @@
from .search import SearchIndex
@@ -1,61 +0,0 @@
INDEX_DEFAULT_SETTINGS = {
"settings":
{"analysis":
{"analyzer": {
"default": {"tokenizer": "whitespace", "filter": ["lowercase", "porter_stem"]}}},
"index":
{"refresh_interval": -1,
"number_of_shards": 1,
"number_of_replicas": 0,
"sort": {
"field": ["trending_mixed", "release_time"],
"order": ["desc", "desc"]
}}
},
"mappings": {
"properties": {
"claim_id": {
"fields": {
"keyword": {
"ignore_above": 256,
"type": "keyword"
}
},
"type": "text",
"index_prefixes": {
"min_chars": 1,
"max_chars": 10
}
},
"height": {"type": "integer"},
"claim_type": {"type": "byte"},
"censor_type": {"type": "byte"},
"trending_mixed": {"type": "float"},
"release_time": {"type": "long"},
}
}
}
FIELDS = {'is_controlling', 'last_take_over_height', 'claim_id', 'claim_name', 'normalized', 'tx_position', 'amount',
'timestamp', 'creation_timestamp', 'height', 'creation_height', 'activation_height', 'expiration_height',
'release_time', 'short_url', 'canonical_url', 'title', 'author', 'description', 'claim_type', 'reposted',
'stream_type', 'media_type', 'fee_amount', 'fee_currency', 'duration', 'reposted_claim_hash', 'censor_type',
'claims_in_channel', 'channel_join', 'signature_valid', 'effective_amount', 'support_amount',
'trending_group', 'trending_mixed', 'trending_local', 'trending_global', 'channel_id', 'tx_id', 'tx_nout',
'signature', 'signature_digest', 'public_key_bytes', 'public_key_hash', 'public_key_id', '_id', 'tags',
'reposted_claim_id'}
TEXT_FIELDS = {'author', 'canonical_url', 'channel_id', 'claim_name', 'description', 'claim_id',
'media_type', 'normalized', 'public_key_bytes', 'public_key_hash', 'short_url', 'signature',
'signature_digest', 'stream_type', 'title', 'tx_id', 'fee_currency', 'reposted_claim_id', 'tags'}
RANGE_FIELDS = {
'height', 'creation_height', 'activation_height', 'expiration_height',
'timestamp', 'creation_timestamp', 'duration', 'release_time', 'fee_amount',
'tx_position', 'channel_join', 'reposted', 'limit_claims_per_channel',
'amount', 'effective_amount', 'support_amount',
'trending_group', 'trending_mixed', 'censor_type',
'trending_local', 'trending_global',
}
REPLACEMENTS = {
'name': 'normalized',
'txid': 'tx_id',
'claim_hash': '_id'
}
@@ -1,531 +0,0 @@
import asyncio
import struct
from binascii import unhexlify
from decimal import Decimal
from operator import itemgetter
from typing import Optional, List, Iterable, Union
from elasticsearch import AsyncElasticsearch, NotFoundError, ConnectionError
from elasticsearch.helpers import async_streaming_bulk
from lbry.crypto.base58 import Base58
from lbry.error import ResolveCensoredError, claim_id as parse_claim_id
from lbry.schema.result import Outputs, Censor
from lbry.schema.tags import clean_tags
from lbry.schema.url import URL, normalize_name
from lbry.utils import LRUCache
from lbry.wallet.server.db.common import CLAIM_TYPES, STREAM_TYPES
from lbry.wallet.server.db.elasticsearch.constants import INDEX_DEFAULT_SETTINGS, REPLACEMENTS, FIELDS, TEXT_FIELDS, \
RANGE_FIELDS
from lbry.wallet.server.util import class_logger
class ChannelResolution(str):
@classmethod
def lookup_error(cls, url):
return LookupError(f'Could not find channel in "{url}".')
class StreamResolution(str):
@classmethod
def lookup_error(cls, url):
return LookupError(f'Could not find claim at "{url}".')
class SearchIndex:
def __init__(self, index_prefix: str, search_timeout=3.0, elastic_host='localhost', elastic_port=9200):
self.search_timeout = search_timeout
self.sync_timeout = 600 # wont hit that 99% of the time, but can hit on a fresh import
self.search_client: Optional[AsyncElasticsearch] = None
self.sync_client: Optional[AsyncElasticsearch] = None
self.index = index_prefix + 'claims'
self.logger = class_logger(__name__, self.__class__.__name__)
self.claim_cache = LRUCache(2 ** 15)
self.short_id_cache = LRUCache(2 ** 17) # never invalidated, since short ids are forever
self.search_cache = LRUCache(2 ** 17)
self.resolution_cache = LRUCache(2 ** 17)
self._elastic_host = elastic_host
self._elastic_port = elastic_port
async def start(self):
if self.sync_client:
return
hosts = [{'host': self._elastic_host, 'port': self._elastic_port}]
self.sync_client = AsyncElasticsearch(hosts, timeout=self.sync_timeout)
self.search_client = AsyncElasticsearch(hosts, timeout=self.search_timeout)
while True:
self.logger.info("DOITDOITDOIT\n\n\n\n%s:%i\n\n\n", self._elastic_host, self._elastic_port)
try:
await self.sync_client.cluster.health(wait_for_status='yellow')
break
except ConnectionError:
self.logger.warning("Failed to connect to Elasticsearch. Waiting for it!")
await asyncio.sleep(1)
res = await self.sync_client.indices.create(self.index, INDEX_DEFAULT_SETTINGS, ignore=400)
return res.get('acknowledged', False)
def stop(self):
clients = [self.sync_client, self.search_client]
self.sync_client, self.search_client = None, None
return asyncio.ensure_future(asyncio.gather(*(client.close() for client in clients)))
def delete_index(self):
return self.sync_client.indices.delete(self.index, ignore_unavailable=True)
async def _consume_claim_producer(self, claim_producer):
count = 0
for op, doc in claim_producer:
if op == 'delete':
yield {'_index': self.index, '_op_type': 'delete', '_id': doc}
else:
yield extract_doc(doc, self.index)
count += 1
if count % 100 == 0:
self.logger.info("Indexing in progress, %d claims.", count)
self.logger.info("Indexing done for %d claims.", count)
async def claim_consumer(self, claim_producer):
touched = set()
async for ok, item in async_streaming_bulk(self.sync_client, self._consume_claim_producer(claim_producer),
raise_on_error=False):
if not ok:
self.logger.warning("indexing failed for an item: %s", item)
else:
item = item.popitem()[1]
touched.add(item['_id'])
await self.sync_client.indices.refresh(self.index)
self.logger.info("Indexing done.")
def update_filter_query(self, censor_type, blockdict, channels=False):
blockdict = {key[::-1].hex(): value[::-1].hex() for key, value in blockdict.items()}
if channels:
update = expand_query(channel_id__in=list(blockdict.keys()), censor_type=f"<{censor_type}")
else:
update = expand_query(claim_id__in=list(blockdict.keys()), censor_type=f"<{censor_type}")
key = 'channel_id' if channels else 'claim_id'
update['script'] = {
"source": f"ctx._source.censor_type={censor_type}; ctx._source.censoring_channel_hash=params[ctx._source.{key}]",
"lang": "painless",
"params": blockdict
}
return update
async def apply_filters(self, blocked_streams, blocked_channels, filtered_streams, filtered_channels):
if filtered_streams:
await self.sync_client.update_by_query(
self.index, body=self.update_filter_query(Censor.SEARCH, filtered_streams), slices=4)
await self.sync_client.indices.refresh(self.index)
if filtered_channels:
await self.sync_client.update_by_query(
self.index, body=self.update_filter_query(Censor.SEARCH, filtered_channels), slices=4)
await self.sync_client.indices.refresh(self.index)
await self.sync_client.update_by_query(
self.index, body=self.update_filter_query(Censor.SEARCH, filtered_channels, True), slices=4)
await self.sync_client.indices.refresh(self.index)
if blocked_streams:
await self.sync_client.update_by_query(
self.index, body=self.update_filter_query(Censor.RESOLVE, blocked_streams), slices=4)
await self.sync_client.indices.refresh(self.index)
if blocked_channels:
await self.sync_client.update_by_query(
self.index, body=self.update_filter_query(Censor.RESOLVE, blocked_channels), slices=4)
await self.sync_client.indices.refresh(self.index)
await self.sync_client.update_by_query(
self.index, body=self.update_filter_query(Censor.RESOLVE, blocked_channels, True), slices=4)
await self.sync_client.indices.refresh(self.index)
self.search_cache.clear()
self.claim_cache.clear()
self.resolution_cache.clear()
async def session_query(self, query_name, kwargs):
offset, total = kwargs.get('offset', 0) if isinstance(kwargs, dict) else 0, 0
total_referenced = []
if query_name == 'resolve':
total_referenced, response, censor = await self.resolve(*kwargs)
else:
cache_item = ResultCacheItem.from_cache(str(kwargs), self.search_cache)
if cache_item.result is not None:
return cache_item.result
async with cache_item.lock:
if cache_item.result:
return cache_item.result
censor = Censor(Censor.SEARCH)
if kwargs.get('no_totals'):
response, offset, total = await self.search(**kwargs, censor_type=Censor.NOT_CENSORED)
else:
response, offset, total = await self.search(**kwargs)
censor.apply(response)
total_referenced.extend(response)
if censor.censored:
response, _, _ = await self.search(**kwargs, censor_type=Censor.NOT_CENSORED)
total_referenced.extend(response)
result = Outputs.to_base64(
response, await self._get_referenced_rows(total_referenced), offset, total, censor
)
cache_item.result = result
return result
return Outputs.to_base64(response, await self._get_referenced_rows(total_referenced), offset, total, censor)
async def resolve(self, *urls):
censor = Censor(Censor.RESOLVE)
results = [await self.resolve_url(url) for url in urls]
# just heat the cache
await self.populate_claim_cache(*filter(lambda x: isinstance(x, str), results))
results = [self._get_from_cache_or_error(url, result) for url, result in zip(urls, results)]
censored = [
result if not isinstance(result, dict) or not censor.censor(result)
else ResolveCensoredError(url, result['censoring_channel_hash'])
for url, result in zip(urls, results)
]
return results, censored, censor
def _get_from_cache_or_error(self, url: str, resolution: Union[LookupError, StreamResolution, ChannelResolution]):
cached = self.claim_cache.get(resolution)
return cached or (resolution if isinstance(resolution, LookupError) else resolution.lookup_error(url))
async def get_many(self, *claim_ids):
await self.populate_claim_cache(*claim_ids)
return filter(None, map(self.claim_cache.get, claim_ids))
async def populate_claim_cache(self, *claim_ids):
missing = [claim_id for claim_id in claim_ids if self.claim_cache.get(claim_id) is None]
if missing:
results = await self.search_client.mget(
index=self.index, body={"ids": missing}
)
for result in expand_result(filter(lambda doc: doc['found'], results["docs"])):
self.claim_cache.set(result['claim_id'], result)
async def full_id_from_short_id(self, name, short_id, channel_id=None):
key = (channel_id or '') + name + short_id
if key not in self.short_id_cache:
query = {'name': name, 'claim_id': short_id}
if channel_id:
query['channel_id'] = channel_id
query['order_by'] = ['^channel_join']
query['signature_valid'] = True
else:
query['order_by'] = '^creation_height'
result, _, _ = await self.search(**query, limit=1)
if len(result) == 1:
result = result[0]['claim_id']
self.short_id_cache[key] = result
return self.short_id_cache.get(key, None)
async def search(self, **kwargs):
if 'channel' in kwargs:
kwargs['channel_id'] = await self.resolve_url(kwargs.pop('channel'))
if not kwargs['channel_id'] or not isinstance(kwargs['channel_id'], str):
return [], 0, 0
try:
result = (await self.search_client.search(
expand_query(**kwargs), index=self.index, track_total_hits=False if kwargs.get('no_totals') else 10_000
))['hits']
except NotFoundError:
return [], 0, 0
return expand_result(result['hits']), 0, result.get('total', {}).get('value', 0)
async def resolve_url(self, raw_url):
if raw_url not in self.resolution_cache:
self.resolution_cache[raw_url] = await self._resolve_url(raw_url)
return self.resolution_cache[raw_url]
async def _resolve_url(self, raw_url):
try:
url = URL.parse(raw_url)
except ValueError as e:
return e
stream = LookupError(f'Could not find claim at "{raw_url}".')
channel_id = await self.resolve_channel_id(url)
if isinstance(channel_id, LookupError):
return channel_id
stream = (await self.resolve_stream(url, channel_id if isinstance(channel_id, str) else None)) or stream
if url.has_stream:
return StreamResolution(stream)
else:
return ChannelResolution(channel_id)
async def resolve_channel_id(self, url: URL):
if not url.has_channel:
return
if url.channel.is_fullid:
return url.channel.claim_id
if url.channel.is_shortid:
channel_id = await self.full_id_from_short_id(url.channel.name, url.channel.claim_id)
if not channel_id:
return LookupError(f'Could not find channel in "{url}".')
return channel_id
query = url.channel.to_dict()
if set(query) == {'name'}:
query['is_controlling'] = True
else:
query['order_by'] = ['^creation_height']
matches, _, _ = await self.search(**query, limit=1)
if matches:
channel_id = matches[0]['claim_id']
else:
return LookupError(f'Could not find channel in "{url}".')
return channel_id
async def resolve_stream(self, url: URL, channel_id: str = None):
if not url.has_stream:
return None
if url.has_channel and channel_id is None:
return None
query = url.stream.to_dict()
if url.stream.claim_id is not None:
if url.stream.is_fullid:
claim_id = url.stream.claim_id
else:
claim_id = await self.full_id_from_short_id(query['name'], query['claim_id'], channel_id)
return claim_id
if channel_id is not None:
if set(query) == {'name'}:
# temporarily emulate is_controlling for claims in channel
query['order_by'] = ['effective_amount', '^height']
else:
query['order_by'] = ['^channel_join']
query['channel_id'] = channel_id
query['signature_valid'] = True
elif set(query) == {'name'}:
query['is_controlling'] = True
matches, _, _ = await self.search(**query, limit=1)
if matches:
return matches[0]['claim_id']
async def _get_referenced_rows(self, txo_rows: List[dict]):
txo_rows = [row for row in txo_rows if isinstance(row, dict)]
referenced_ids = set(filter(None, map(itemgetter('reposted_claim_id'), txo_rows)))
referenced_ids |= set(filter(None, (row['channel_id'] for row in txo_rows)))
referenced_ids |= set(map(parse_claim_id, filter(None, (row['censoring_channel_hash'] for row in txo_rows))))
referenced_txos = []
if referenced_ids:
referenced_txos.extend(await self.get_many(*referenced_ids))
referenced_ids = set(filter(None, (row['channel_id'] for row in referenced_txos)))
if referenced_ids:
referenced_txos.extend(await self.get_many(*referenced_ids))
return referenced_txos
def extract_doc(doc, index):
doc['claim_id'] = doc.pop('claim_hash')[::-1].hex()
if doc['reposted_claim_hash'] is not None:
doc['reposted_claim_id'] = doc.pop('reposted_claim_hash')[::-1].hex()
else:
doc['reposted_claim_id'] = None
channel_hash = doc.pop('channel_hash')
doc['channel_id'] = channel_hash[::-1].hex() if channel_hash else channel_hash
channel_hash = doc.pop('censoring_channel_hash')
doc['censoring_channel_hash'] = channel_hash[::-1].hex() if channel_hash else channel_hash
txo_hash = doc.pop('txo_hash')
doc['tx_id'] = txo_hash[:32][::-1].hex()
doc['tx_nout'] = struct.unpack('<I', txo_hash[32:])[0]
doc['is_controlling'] = bool(doc['is_controlling'])
doc['signature'] = (doc.pop('signature') or b'').hex() or None
doc['signature_digest'] = (doc.pop('signature_digest') or b'').hex() or None
doc['public_key_bytes'] = (doc.pop('public_key_bytes') or b'').hex() or None
doc['public_key_hash'] = (doc.pop('public_key_hash') or b'').hex() or None
doc['signature_valid'] = bool(doc['signature_valid'])
doc['claim_type'] = doc.get('claim_type', 0) or 0
doc['stream_type'] = int(doc.get('stream_type', 0) or 0)
doc['has_source'] = bool(doc['has_source'])
return {'doc': doc, '_id': doc['claim_id'], '_index': index, '_op_type': 'update', 'doc_as_upsert': True}
def expand_query(**kwargs):
if "amount_order" in kwargs:
kwargs["limit"] = 1
kwargs["order_by"] = "effective_amount"
kwargs["offset"] = int(kwargs["amount_order"]) - 1
if 'name' in kwargs:
kwargs['name'] = normalize_name(kwargs.pop('name'))
if kwargs.get('is_controlling') is False:
kwargs.pop('is_controlling')
query = {'must': [], 'must_not': []}
collapse = None
for key, value in kwargs.items():
key = key.replace('claim.', '')
many = key.endswith('__in') or isinstance(value, list)
if many:
key = key.replace('__in', '')
value = list(filter(None, value))
if value is None or isinstance(value, list) and len(value) == 0:
continue
key = REPLACEMENTS.get(key, key)
if key in FIELDS:
partial_id = False
if key == 'claim_type':
if isinstance(value, str):
value = CLAIM_TYPES[value]
else:
value = [CLAIM_TYPES[claim_type] for claim_type in value]
if key == '_id':
if isinstance(value, Iterable):
value = [item[::-1].hex() for item in value]
else:
value = value[::-1].hex()
if not many and key in ('_id', 'claim_id') and len(value) < 20:
partial_id = True
if key == 'public_key_id':
key = 'public_key_hash'
value = Base58.decode(value)[1:21].hex()
if key == 'signature_valid':
continue # handled later
if key in TEXT_FIELDS:
key += '.keyword'
ops = {'<=': 'lte', '>=': 'gte', '<': 'lt', '>': 'gt'}
if partial_id:
query['must'].append({"prefix": {"claim_id": value}})
elif key in RANGE_FIELDS and isinstance(value, str) and value[0] in ops:
operator_length = 2 if value[:2] in ops else 1
operator, value = value[:operator_length], value[operator_length:]
if key == 'fee_amount':
value = str(Decimal(value)*1000)
query['must'].append({"range": {key: {ops[operator]: value}}})
elif many:
query['must'].append({"terms": {key: value}})
else:
if key == 'fee_amount':
value = str(Decimal(value)*1000)
query['must'].append({"term": {key: {"value": value}}})
elif key == 'not_channel_ids':
for channel_id in value:
query['must_not'].append({"term": {'channel_id.keyword': channel_id}})
query['must_not'].append({"term": {'_id': channel_id}})
elif key == 'channel_ids':
query['must'].append({"terms": {'channel_id.keyword': value}})
elif key == 'claim_ids':
query['must'].append({"terms": {'claim_id.keyword': value}})
elif key == 'media_types':
query['must'].append({"terms": {'media_type.keyword': value}})
elif key == 'stream_types':
query['must'].append({"terms": {'stream_type': [STREAM_TYPES[stype] for stype in value]}})
elif key == 'any_languages':
query['must'].append({"terms": {'languages': clean_tags(value)}})
elif key == 'any_languages':
query['must'].append({"terms": {'languages': value}})
elif key == 'all_languages':
query['must'].extend([{"term": {'languages': tag}} for tag in value])
elif key == 'any_tags':
query['must'].append({"terms": {'tags.keyword': clean_tags(value)}})
elif key == 'all_tags':
query['must'].extend([{"term": {'tags.keyword': tag}} for tag in clean_tags(value)])
elif key == 'not_tags':
query['must_not'].extend([{"term": {'tags.keyword': tag}} for tag in clean_tags(value)])
elif key == 'not_claim_id':
query['must_not'].extend([{"term": {'claim_id.keyword': cid}} for cid in value])
elif key == 'limit_claims_per_channel':
collapse = ('channel_id.keyword', value)
if kwargs.get('has_channel_signature'):
query['must'].append({"exists": {"field": "signature_digest"}})
if 'signature_valid' in kwargs:
query['must'].append({"term": {"signature_valid": bool(kwargs["signature_valid"])}})
elif 'signature_valid' in kwargs:
query.setdefault('should', [])
query["minimum_should_match"] = 1
query['should'].append({"bool": {"must_not": {"exists": {"field": "signature_digest"}}}})
query['should'].append({"term": {"signature_valid": bool(kwargs["signature_valid"])}})
if 'has_source' in kwargs:
query.setdefault('should', [])
query["minimum_should_match"] = 1
is_stream_or_repost = {"terms": {"claim_type": [CLAIM_TYPES['stream'], CLAIM_TYPES['repost']]}}
query['should'].append(
{"bool": {"must": [{"match": {"has_source": kwargs['has_source']}}, is_stream_or_repost]}})
query['should'].append({"bool": {"must_not": [is_stream_or_repost]}})
if kwargs.get('text'):
query['must'].append(
{"simple_query_string":
{"query": kwargs["text"], "fields": [
"claim_name^4", "channel_name^8", "title^1", "description^.5", "author^1", "tags^.5"
]}})
query = {
"_source": {"excludes": ["description", "title"]},
'query': {'bool': query},
"sort": [],
}
if "limit" in kwargs:
query["size"] = kwargs["limit"]
if 'offset' in kwargs:
query["from"] = kwargs["offset"]
if 'order_by' in kwargs:
if isinstance(kwargs["order_by"], str):
kwargs["order_by"] = [kwargs["order_by"]]
for value in kwargs['order_by']:
if 'trending_group' in value:
# fixme: trending_mixed is 0 for all records on variable decay, making sort slow.
continue
is_asc = value.startswith('^')
value = value[1:] if is_asc else value
value = REPLACEMENTS.get(value, value)
if value in TEXT_FIELDS:
value += '.keyword'
query['sort'].append({value: "asc" if is_asc else "desc"})
if collapse:
query["collapse"] = {
"field": collapse[0],
"inner_hits": {
"name": collapse[0],
"size": collapse[1],
"sort": query["sort"]
}
}
return query
def expand_result(results):
inner_hits = []
expanded = []
for result in results:
if result.get("inner_hits"):
for _, inner_hit in result["inner_hits"].items():
inner_hits.extend(inner_hit["hits"]["hits"])
continue
result = result['_source']
result['claim_hash'] = unhexlify(result['claim_id'])[::-1]
if result['reposted_claim_id']:
result['reposted_claim_hash'] = unhexlify(result['reposted_claim_id'])[::-1]
else:
result['reposted_claim_hash'] = None
result['channel_hash'] = unhexlify(result['channel_id'])[::-1] if result['channel_id'] else None
result['txo_hash'] = unhexlify(result['tx_id'])[::-1] + struct.pack('<I', result['tx_nout'])
result['tx_hash'] = unhexlify(result['tx_id'])[::-1]
if result['censoring_channel_hash']:
result['censoring_channel_hash'] = unhexlify(result['censoring_channel_hash'])[::-1]
expanded.append(result)
if inner_hits:
return expand_result(inner_hits)
return expanded
class ResultCacheItem:
__slots__ = '_result', 'lock', 'has_result'
def __init__(self):
self.has_result = asyncio.Event()
self.lock = asyncio.Lock()
self._result = None
@property
def result(self) -> str:
return self._result
@result.setter
def result(self, result: str):
self._result = result
if result is not None:
self.has_result.set()
@classmethod
def from_cache(cls, cache_key, cache):
cache_item = cache.get(cache_key)
if cache_item is None:
cache_item = cache[cache_key] = ResultCacheItem()
return cache_item
-114
View File
@@ -1,114 +0,0 @@
import argparse
import asyncio
import logging
import os
from collections import namedtuple
from multiprocessing import Process
import apsw
from elasticsearch import AsyncElasticsearch
from elasticsearch.helpers import async_bulk
from lbry.wallet.server.env import Env
from lbry.wallet.server.coin import LBC
from lbry.wallet.server.db.elasticsearch.search import extract_doc, SearchIndex
INDEX = 'claims'
async def get_all(db, shard_num, shards_total, limit=0):
logging.info("shard %d starting", shard_num)
def exec_factory(cursor, statement, bindings):
tpl = namedtuple('row', (d[0] for d in cursor.getdescription()))
cursor.setrowtrace(lambda cursor, row: tpl(*row))
return True
db.setexectrace(exec_factory)
total = db.execute(f"select count(*) as total from claim where height % {shards_total} = {shard_num};").fetchone()[0]
for num, claim in enumerate(db.execute(f"""
SELECT claimtrie.claim_hash as is_controlling,
claimtrie.last_take_over_height,
(select group_concat(tag, ',,') from tag where tag.claim_hash in (claim.claim_hash, claim.reposted_claim_hash)) as tags,
(select group_concat(language, ' ') from language where language.claim_hash in (claim.claim_hash, claim.reposted_claim_hash)) as languages,
(select cr.has_source from claim cr where cr.claim_hash = claim.reposted_claim_hash) as reposted_has_source,
claim.*
FROM claim LEFT JOIN claimtrie USING (claim_hash)
WHERE claim.height % {shards_total} = {shard_num}
ORDER BY claim.height desc
""")):
claim = dict(claim._asdict())
claim['has_source'] = bool(claim.pop('reposted_has_source') or claim['has_source'])
claim['censor_type'] = 0
claim['censoring_channel_hash'] = None
claim['tags'] = claim['tags'].split(',,') if claim['tags'] else []
claim['languages'] = claim['languages'].split(' ') if claim['languages'] else []
if num % 10_000 == 0:
logging.info("%d/%d", num, total)
yield extract_doc(claim, INDEX)
if 0 < limit <= num:
break
async def consume(producer):
env = Env(LBC)
logging.info("ES sync host: %s:%i", env.elastic_host, env.elastic_port)
es = AsyncElasticsearch([{'host': env.elastic_host, 'port': env.elastic_port}])
try:
await async_bulk(es, producer, request_timeout=120)
await es.indices.refresh(index=INDEX)
finally:
await es.close()
async def make_es_index():
env = Env(LBC)
index = SearchIndex('', elastic_host=env.elastic_host, elastic_port=env.elastic_port)
try:
return await index.start()
finally:
index.stop()
async def run(args, shard):
def itsbusy(*_):
logging.info("shard %d: db is busy, retry", shard)
return True
db = apsw.Connection(args.db_path, flags=apsw.SQLITE_OPEN_READONLY | apsw.SQLITE_OPEN_URI)
db.setbusyhandler(itsbusy)
db.cursor().execute('pragma journal_mode=wal;')
db.cursor().execute('pragma temp_store=memory;')
producer = get_all(db.cursor(), shard, args.clients, limit=args.blocks)
await asyncio.gather(*(consume(producer) for _ in range(min(8, args.clients))))
def __run(args, shard):
asyncio.run(run(args, shard))
def run_elastic_sync():
logging.basicConfig(level=logging.INFO)
logging.getLogger('aiohttp').setLevel(logging.WARNING)
logging.getLogger('elasticsearch').setLevel(logging.WARNING)
logging.info('lbry.server starting')
parser = argparse.ArgumentParser(prog="lbry-hub-elastic-sync")
parser.add_argument("db_path", type=str)
parser.add_argument("-c", "--clients", type=int, default=16)
parser.add_argument("-b", "--blocks", type=int, default=0)
parser.add_argument("-f", "--force", default=False, action='store_true')
args = parser.parse_args()
processes = []
if not args.force and not os.path.exists(args.db_path):
logging.info("DB path doesnt exist")
return
if not args.force and not asyncio.run(make_es_index()):
logging.info("ES is already initialized")
return
for i in range(args.clients):
processes.append(Process(target=__run, args=(args, i)))
processes[-1].start()
for process in processes:
process.join()
process.close()
+52
View File
@@ -0,0 +1,52 @@
from lbry.wallet.database import constraints_to_sql
CREATE_FULL_TEXT_SEARCH = """
create virtual table if not exists search using fts5(
claim_name, channel_name, title, description, author, tags,
content=claim, tokenize=porter
);
"""
FTS_ORDER_BY = "bm25(search, 4.0, 8.0, 1.0, 0.5, 1.0, 0.5)"
def fts_action_sql(claims=None, action='insert'):
select = {
'rowid': "claim.rowid",
'claim_name': "claim.normalized",
'channel_name': "channel.normalized",
'title': "claim.title",
'description': "claim.description",
'author': "claim.author",
'tags': "(select group_concat(tag, ' ') from tag where tag.claim_hash=claim.claim_hash)"
}
if action == 'delete':
select['search'] = '"delete"'
where, values = "", {}
if claims:
where, values = constraints_to_sql({'claim.claim_hash__in': claims})
where = 'WHERE '+where
return f"""
INSERT INTO search ({','.join(select.keys())})
SELECT {','.join(select.values())} FROM claim
LEFT JOIN claim as channel ON (claim.channel_hash=channel.claim_hash) {where}
""", values
def update_full_text_search(action, outputs, db, is_first_sync):
if is_first_sync:
return
if not outputs:
return
if action in ("before-delete", "before-update"):
db.execute(*fts_action_sql(outputs, 'delete'))
elif action in ("after-insert", "after-update"):
db.execute(*fts_action_sql(outputs, 'insert'))
else:
raise ValueError(f"Invalid action for updating full text search: '{action}'")
def first_sync_finished(db):
db.execute(*fts_action_sql())
@@ -18,7 +18,8 @@ from lbry.schema.tags import clean_tags
from lbry.schema.result import Outputs, Censor
from lbry.wallet import Ledger, RegTestLedger
from lbry.wallet.server.db.common import CLAIM_TYPES, STREAM_TYPES, COMMON_TAGS, INDEXED_LANGUAGES
from .common import CLAIM_TYPES, STREAM_TYPES, COMMON_TAGS
from .full_text_search import FTS_ORDER_BY
class SQLiteOperationalError(apsw.Error):
@@ -38,7 +39,7 @@ ATTRIBUTE_ARRAY_MAX_LENGTH = 100
INTEGER_PARAMS = {
'height', 'creation_height', 'activation_height', 'expiration_height',
'timestamp', 'creation_timestamp', 'duration', 'release_time', 'fee_amount',
'tx_position', 'channel_join', 'reposted', 'limit_claims_per_channel',
'tx_position', 'channel_join', 'reposted',
'amount', 'effective_amount', 'support_amount',
'trending_group', 'trending_mixed',
'trending_local', 'trending_global',
@@ -52,7 +53,7 @@ SEARCH_PARAMS = {
'any_locations', 'all_locations', 'not_locations',
'any_languages', 'all_languages', 'not_languages',
'is_controlling', 'limit', 'offset', 'order_by',
'no_totals', 'has_source'
'no_totals',
} | INTEGER_PARAMS
@@ -93,10 +94,10 @@ class ReaderState:
self.db.setprogresshandler(interruptor, 100)
def get_resolve_censor(self) -> Censor:
return Censor(Censor.RESOLVE)
return Censor(self.blocked_streams, self.blocked_channels)
def get_search_censor(self, limit_claims_per_channel: int) -> Censor:
return Censor(Censor.SEARCH)
def get_search_censor(self) -> Censor:
return Censor(self.filtered_streams, self.filtered_channels)
ctx: ContextVar[Optional[ReaderState]] = ContextVar('ctx')
@@ -282,7 +283,7 @@ def claims_query(cols, for_count=False, **constraints) -> Tuple[str, Dict]:
channel_ids = constraints.pop('channel_ids')
if channel_ids:
constraints['claim.channel_hash__in'] = {
unhexlify(cid)[::-1] for cid in channel_ids if cid
unhexlify(cid)[::-1] for cid in channel_ids
}
if 'not_channel_ids' in constraints:
not_channel_ids = constraints.pop('not_channel_ids')
@@ -341,7 +342,12 @@ def claims_query(cols, for_count=False, **constraints) -> Tuple[str, Dict]:
_apply_constraints_for_array_attributes(constraints, 'language', lambda _: _, for_count)
_apply_constraints_for_array_attributes(constraints, 'location', lambda _: _, for_count)
select = f"SELECT {cols} FROM claim"
if 'text' in constraints:
constraints["search"] = constraints.pop("text")
constraints["order_by"] = FTS_ORDER_BY
select = f"SELECT {cols} FROM search JOIN claim ON (search.rowid=claim.rowid)"
else:
select = f"SELECT {cols} FROM claim"
if not for_count:
select += " LEFT JOIN claimtrie USING (claim_hash)"
return query(select, **constraints)
@@ -366,7 +372,7 @@ def count_claims(**constraints) -> int:
constraints.pop('offset', None)
constraints.pop('limit', None)
constraints.pop('order_by', None)
count = select_claims(Censor(Censor.SEARCH), 'count(*) as row_count', for_count=True, **constraints)
count = select_claims(Censor(), 'count(*) as row_count', for_count=True, **constraints)
return count[0]['row_count']
@@ -415,13 +421,12 @@ def search(constraints) -> Tuple[List, List, int, int, Censor]:
assert set(constraints).issubset(SEARCH_PARAMS), \
f"Search query contains invalid arguments: {set(constraints).difference(SEARCH_PARAMS)}"
total = None
limit_claims_per_channel = constraints.pop('limit_claims_per_channel', None)
if not constraints.pop('no_totals', False):
total = count_claims(**constraints)
constraints['offset'] = abs(constraints.get('offset', 0))
constraints['limit'] = min(abs(constraints.get('limit', 10)), 50)
context = ctx.get()
search_censor = context.get_search_censor(limit_claims_per_channel)
search_censor = context.get_search_censor()
txo_rows = search_claims(search_censor, **constraints)
extra_txo_rows = _get_referenced_rows(txo_rows, search_censor.censored.keys())
return txo_rows, extra_txo_rows, constraints['offset'], total, search_censor
@@ -530,18 +535,6 @@ def _apply_constraints_for_array_attributes(constraints, attr, cleaner, for_coun
AND tag IN ({values})
)
"""
elif attr == 'language':
indexed_languages = any_items & set(INDEXED_LANGUAGES)
if indexed_languages:
any_items -= indexed_languages
for language in indexed_languages:
any_queries[f'#_any_common_languages_{language}'] = f"""
EXISTS(
SELECT 1 FROM language INDEXED BY language_{language}_idx
WHERE {CLAIM_HASH_OR_REPOST_HASH_SQL}=language.claim_hash
AND language = '{language}'
)
"""
if any_items:
@@ -554,10 +547,10 @@ def _apply_constraints_for_array_attributes(constraints, attr, cleaner, for_coun
if for_count or attr == 'tag':
if attr == 'tag':
any_queries[f'#_any_{attr}'] = f"""
((claim.claim_type != {CLAIM_TYPES['repost']}
(claim.claim_type != {CLAIM_TYPES['repost']}
AND claim.claim_hash IN (SELECT claim_hash FROM tag WHERE tag IN ({values}))) OR
(claim.claim_type == {CLAIM_TYPES['repost']} AND
claim.reposted_claim_hash IN (SELECT claim_hash FROM tag WHERE tag IN ({values}))))
claim.reposted_claim_hash IN (SELECT claim_hash FROM tag WHERE tag IN ({values})))
"""
else:
any_queries[f'#_any_{attr}'] = f"""
@@ -613,10 +606,10 @@ def _apply_constraints_for_array_attributes(constraints, attr, cleaner, for_coun
if for_count:
if attr == 'tag':
constraints[f'#_not_{attr}'] = f"""
((claim.claim_type != {CLAIM_TYPES['repost']}
AND claim.claim_hash NOT IN (SELECT claim_hash FROM tag WHERE tag IN ({values}))) OR
(claim.claim_type != {CLAIM_TYPES['repost']}
AND claim.claim_hash NOT IN (SELECT claim_hash FROM tag WHERE tag IN ({values}))) AND
(claim.claim_type == {CLAIM_TYPES['repost']} AND
claim.reposted_claim_hash NOT IN (SELECT claim_hash FROM tag WHERE tag IN ({values}))))
claim.reposted_claim_hash NOT IN (SELECT claim_hash FROM tag WHERE tag IN ({values})))
"""
else:
constraints[f'#_not_{attr}'] = f"""
+269 -323
View File
@@ -1,17 +1,18 @@
"""
AR-like trending with a delayed effect and a faster
decay rate for high valued claims.
Delayed AR with variable decay rate.
The spike height function is also simpler.
"""
import math
import copy
import time
import apsw
# Half life in blocks *for lower LBC claims* (it's shorter for whale claims)
HALF_LIFE = 200
# Whale threshold, in LBC (higher -> less DB writing)
WHALE_THRESHOLD = 10000.0
# Whale threshold (higher -> less DB writing)
WHALE_THRESHOLD = 3.0
# Decay coefficient per block
DECAY = 0.5**(1.0/HALF_LIFE)
@@ -37,7 +38,6 @@ def install(connection):
Install the trending algorithm.
"""
check_trending_values(connection)
trending_data.initialise(connection.cursor())
if TRENDING_LOG:
f = open("trending_variable_decay.log", "a")
@@ -46,6 +46,7 @@ def install(connection):
# Stub
CREATE_TREND_TABLE = ""
def check_trending_values(connection):
"""
If the trending values appear to be based on the zscore algorithm,
@@ -59,304 +60,19 @@ def check_trending_values(connection):
break
if needs_reset:
print("Resetting some columns. This might take a while...", flush=True,
end="")
print("Resetting some columns. This might take a while...", flush=True, end="")
c.execute(""" BEGIN;
UPDATE claim SET trending_group = 0;
UPDATE claim SET trending_mixed = 0;
UPDATE claim SET trending_global = 0;
UPDATE claim SET trending_local = 0;
COMMIT;""")
print("done.")
def trending_log(s):
def spike_height(x, x_old):
"""
Log a string to the log file
"""
if TRENDING_LOG:
fout = open("trending_variable_decay.log", "a")
fout.write(s)
fout.flush()
fout.close()
def trending_unit(height):
"""
Return the trending score unit at a given height.
"""
# Round to the beginning of a SAVE_INTERVAL batch of blocks.
_height = height - (height % SAVE_INTERVAL)
return 1.0/DECAY**(height % RENORM_INTERVAL)
class TrendingDB:
"""
An in-memory database of trending scores
"""
def __init__(self):
self.conn = apsw.Connection(":memory:")
self.cursor = self.conn.cursor()
self.initialised = False
self.write_needed = set()
def execute(self, query, *args, **kwargs):
return self.cursor.execute(query, *args, **kwargs)
def executemany(self, query, *args, **kwargs):
return self.cursor.executemany(query, *args, **kwargs)
def begin(self):
self.execute("BEGIN;")
def commit(self):
self.execute("COMMIT;")
def initialise(self, db):
"""
Pass in claims.db
"""
if self.initialised:
return
trending_log("Initialising trending database...")
# The need for speed
self.execute("PRAGMA JOURNAL_MODE=OFF;")
self.execute("PRAGMA SYNCHRONOUS=0;")
self.begin()
# Create the tables
self.execute("""
CREATE TABLE IF NOT EXISTS claims
(claim_hash BYTES PRIMARY KEY,
lbc REAL NOT NULL DEFAULT 0.0,
trending_score REAL NOT NULL DEFAULT 0.0)
WITHOUT ROWID;""")
self.execute("""
CREATE TABLE IF NOT EXISTS spikes
(id INTEGER PRIMARY KEY,
claim_hash BYTES NOT NULL,
height INTEGER NOT NULL,
mass REAL NOT NULL,
FOREIGN KEY (claim_hash)
REFERENCES claims (claim_hash));""")
# Clear out any existing data
self.execute("DELETE FROM claims;")
self.execute("DELETE FROM spikes;")
# Create indexes
self.execute("CREATE INDEX idx1 ON spikes (claim_hash, height, mass);")
self.execute("CREATE INDEX idx2 ON spikes (claim_hash, height, mass DESC);")
self.execute("CREATE INDEX idx3 on claims (lbc DESC, claim_hash, trending_score);")
# Import data from claims.db
for row in db.execute("""
SELECT claim_hash,
1E-8*(amount + support_amount) AS lbc,
trending_mixed
FROM claim;
"""):
self.execute("INSERT INTO claims VALUES (?, ?, ?);", row)
self.commit()
self.initialised = True
trending_log("done.\n")
def apply_spikes(self, height):
"""
Apply spikes that are due. This occurs inside a transaction.
"""
spikes = []
unit = trending_unit(height)
for row in self.execute("""
SELECT SUM(mass), claim_hash FROM spikes
WHERE height = ?
GROUP BY claim_hash;
""", (height, )):
spikes.append((row[0]*unit, row[1]))
self.write_needed.add(row[1])
self.executemany("""
UPDATE claims
SET trending_score = (trending_score + ?)
WHERE claim_hash = ?;
""", spikes)
self.execute("DELETE FROM spikes WHERE height = ?;", (height, ))
def decay_whales(self, height):
"""
Occurs inside transaction.
"""
if height % SAVE_INTERVAL != 0:
return
whales = self.execute("""
SELECT trending_score, lbc, claim_hash
FROM claims
WHERE lbc >= ?;
""", (WHALE_THRESHOLD, )).fetchall()
whales2 = []
for whale in whales:
trending, lbc, claim_hash = whale
# Overall multiplication factor for decay rate
# At WHALE_THRESHOLD, this is 1
# At 10*WHALE_THRESHOLD, it is 3
decay_rate_factor = 1.0 + 2.0*math.log10(lbc/WHALE_THRESHOLD)
# The -1 is because this is just the *extra* part being applied
factor = (DECAY**SAVE_INTERVAL)**(decay_rate_factor - 1.0)
# Decay
trending *= factor
whales2.append((trending, claim_hash))
self.write_needed.add(claim_hash)
self.executemany("UPDATE claims SET trending_score=? WHERE claim_hash=?;",
whales2)
def renorm(self, height):
"""
Renormalise trending scores. Occurs inside a transaction.
"""
if height % RENORM_INTERVAL == 0:
threshold = 1.0E-3/DECAY_PER_RENORM
for row in self.execute("""SELECT claim_hash FROM claims
WHERE ABS(trending_score) >= ?;""",
(threshold, )):
self.write_needed.add(row[0])
self.execute("""UPDATE claims SET trending_score = ?*trending_score
WHERE ABS(trending_score) >= ?;""",
(DECAY_PER_RENORM, threshold))
def write_to_claims_db(self, db, height):
"""
Write changed trending scores to claims.db.
"""
if height % SAVE_INTERVAL != 0:
return
rows = self.execute(f"""
SELECT trending_score, claim_hash
FROM claims
WHERE claim_hash IN
({','.join('?' for _ in self.write_needed)});
""", self.write_needed).fetchall()
db.executemany("""UPDATE claim SET trending_mixed = ?
WHERE claim_hash = ?;""", rows)
# Clear list of claims needing to be written to claims.db
self.write_needed = set()
def update(self, db, height, recalculate_claim_hashes):
"""
Update trending scores.
Input is a cursor to claims.db, the block height, and the list of
claims that changed.
"""
assert self.initialised
self.begin()
self.renorm(height)
# Fetch changed/new claims from claims.db
for row in db.execute(f"""
SELECT claim_hash,
1E-8*(amount + support_amount) AS lbc
FROM claim
WHERE claim_hash IN
({','.join('?' for _ in recalculate_claim_hashes)});
""", recalculate_claim_hashes):
claim_hash, lbc = row
# Insert into trending db if it does not exist
self.execute("""
INSERT INTO claims (claim_hash)
VALUES (?)
ON CONFLICT (claim_hash) DO NOTHING;""",
(claim_hash, ))
# See if it was an LBC change
old = self.execute("SELECT * FROM claims WHERE claim_hash=?;",
(claim_hash, )).fetchone()
lbc_old = old[1]
# Save new LBC value into trending db
self.execute("UPDATE claims SET lbc = ? WHERE claim_hash = ?;",
(lbc, claim_hash))
if lbc > lbc_old:
# Schedule a future spike
delay = min(int((lbc + 1E-8)**0.4), HALF_LIFE)
spike = (claim_hash, height + delay, spike_mass(lbc, lbc_old))
self.execute("""INSERT INTO spikes
(claim_hash, height, mass)
VALUES (?, ?, ?);""", spike)
elif lbc < lbc_old:
# Subtract from future spikes
penalty = spike_mass(lbc_old, lbc)
spikes = self.execute("""
SELECT * FROM spikes
WHERE claim_hash = ?
ORDER BY height ASC, mass DESC;
""", (claim_hash, )).fetchall()
for spike in spikes:
spike_id, mass = spike[0], spike[3]
if mass > penalty:
# The entire penalty merely reduces this spike
self.execute("UPDATE spikes SET mass=? WHERE id=?;",
(mass - penalty, spike_id))
penalty = 0.0
else:
# Removing this spike entirely accounts for some (or
# all) of the penalty, then move on to other spikes
self.execute("DELETE FROM spikes WHERE id=?;",
(spike_id, ))
penalty -= mass
# If penalty remains, that's a negative spike to be applied
# immediately.
if penalty > 0.0:
self.execute("""
INSERT INTO spikes (claim_hash, height, mass)
VALUES (?, ?, ?);""",
(claim_hash, height, -penalty))
self.apply_spikes(height)
self.decay_whales(height)
self.commit()
self.write_to_claims_db(db, height)
# The "global" instance to work with
# pylint: disable=C0103
trending_data = TrendingDB()
def spike_mass(x, x_old):
"""
Compute the mass of a trending spike (normed - constant units).
x_old = old LBC value
x = new LBC value
Compute the size of a trending spike (normed - constant units).
"""
# Sign of trending spike
@@ -373,16 +89,155 @@ def spike_mass(x, x_old):
return sign*mag
def run(db, height, final_height, recalculate_claim_hashes):
if height < final_height - 5*HALF_LIFE:
trending_log(f"Skipping trending calculations at block {height}.\n")
return
start = time.time()
trending_log(f"Calculating variable_decay trending at block {height}.\n")
trending_data.update(db, height, recalculate_claim_hashes)
end = time.time()
trending_log(f"Trending operations took {end - start} seconds.\n\n")
def get_time_boost(height):
"""
Return the time boost at a given height.
"""
return 1.0/DECAY**(height % RENORM_INTERVAL)
def trending_log(s):
"""
Log a string.
"""
if TRENDING_LOG:
fout = open("trending_variable_decay.log", "a")
fout.write(s)
fout.flush()
fout.close()
class TrendingData:
"""
An object of this class holds trending data
"""
def __init__(self):
# Dict from claim id to some trending info.
# Units are TIME VARIABLE in here
self.claims = {}
# Claims with >= WHALE_THRESHOLD LBC total amount
self.whales = set([])
# Have all claims been read from db yet?
self.initialised = False
# List of pending spikes.
# Units are CONSTANT in here
self.pending_spikes = []
def insert_claim_from_load(self, height, claim_hash, trending_score, total_amount):
assert not self.initialised
self.claims[claim_hash] = {"trending_score": trending_score,
"total_amount": total_amount,
"changed": False}
if trending_score >= WHALE_THRESHOLD*get_time_boost(height):
self.add_whale(claim_hash)
def add_whale(self, claim_hash):
self.whales.add(claim_hash)
def apply_spikes(self, height):
"""
Apply all pending spikes that are due at this height.
Apply with time boost ON.
"""
time_boost = get_time_boost(height)
for spike in self.pending_spikes:
if spike["height"] > height:
# Ignore
pass
if spike["height"] == height:
# Apply
self.claims[spike["claim_hash"]]["trending_score"] += time_boost*spike["size"]
self.claims[spike["claim_hash"]]["changed"] = True
if self.claims[spike["claim_hash"]]["trending_score"] >= WHALE_THRESHOLD*time_boost:
self.add_whale(spike["claim_hash"])
if spike["claim_hash"] in self.whales and \
self.claims[spike["claim_hash"]]["trending_score"] < WHALE_THRESHOLD*time_boost:
self.whales.remove(spike["claim_hash"])
# Keep only future spikes
self.pending_spikes = [s for s in self.pending_spikes \
if s["height"] > height]
def update_claim(self, height, claim_hash, total_amount):
"""
Update trending data for a claim, given its new total amount.
"""
assert self.initialised
# Extract existing total amount and trending score
# or use starting values if the claim is new
if claim_hash in self.claims:
old_state = copy.deepcopy(self.claims[claim_hash])
else:
old_state = {"trending_score": 0.0,
"total_amount": 0.0,
"changed": False}
# Calculate LBC change
change = total_amount - old_state["total_amount"]
# Modify data if there was an LBC change
if change != 0.0:
spike = spike_height(total_amount,
old_state["total_amount"])
delay = min(int((total_amount + 1E-8)**0.4), HALF_LIFE)
if change < 0.0:
# How big would the spike be for the inverse movement?
reverse_spike = spike_height(old_state["total_amount"], total_amount)
# Remove that much spike from future pending ones
for future_spike in self.pending_spikes:
if future_spike["claim_hash"] == claim_hash:
if reverse_spike >= future_spike["size"]:
reverse_spike -= future_spike["size"]
future_spike["size"] = 0.0
elif reverse_spike > 0.0:
future_spike["size"] -= reverse_spike
reverse_spike = 0.0
delay = 0
spike = -reverse_spike
self.pending_spikes.append({"height": height + delay,
"claim_hash": claim_hash,
"size": spike})
self.claims[claim_hash] = {"total_amount": total_amount,
"trending_score": old_state["trending_score"],
"changed": False}
def process_whales(self, height):
"""
Whale claims decay faster.
"""
if height % SAVE_INTERVAL != 0:
return
for claim_hash in self.whales:
trending_normed = self.claims[claim_hash]["trending_score"]/get_time_boost(height)
# Overall multiplication factor for decay rate
decay_rate_factor = trending_normed/WHALE_THRESHOLD
# The -1 is because this is just the *extra* part being applied
factor = (DECAY**SAVE_INTERVAL)**(decay_rate_factor - 1.0)
# print(claim_hash, trending_normed, decay_rate_factor)
self.claims[claim_hash]["trending_score"] *= factor
self.claims[claim_hash]["changed"] = True
def test_trending():
"""
@@ -405,12 +260,12 @@ def test_trending():
COMMIT;
""")
# Initialise trending data before anything happens with the claims
trending_data.initialise(db)
# Insert initial states of claims
everything = {"huge_whale": 0.01, "medium_whale": 0.01, "small_whale": 0.01,
"huge_whale_botted": 0.01, "minnow": 0.01}
everything = {"huge_whale": 0.01,
"huge_whale_botted": 0.01,
"medium_whale": 0.01,
"small_whale": 0.01,
"minnow": 0.01}
def to_list_of_tuples(stuff):
l = []
@@ -422,17 +277,13 @@ def test_trending():
INSERT INTO claim (claim_hash, amount) VALUES (?, 1E8*?);
""", to_list_of_tuples(everything))
# Process block zero
height = 0
run(db, height, height, everything.keys())
# Save trajectories for plotting
trajectories = {}
for row in trending_data.execute("""
SELECT claim_hash, trending_score
FROM claims;
"""):
trajectories[row[0]] = [row[1]/trending_unit(height)]
for key in trending_data.claims:
trajectories[key] = [trending_data.claims[key]["trending_score"]]
# Main loop
for height in range(1, 1000):
@@ -461,24 +312,119 @@ def test_trending():
# Call run()
run(db, height, height, everything.keys())
# Append current trending scores to trajectories
for row in db.execute("""
SELECT claim_hash, trending_mixed
FROM claim;
"""):
trajectories[row[0]].append(row[1]/trending_unit(height))
for key in trending_data.claims:
trajectories[key].append(trending_data.claims[key]["trending_score"]\
/get_time_boost(height))
dbc.close()
# pylint: disable=C0415
import matplotlib.pyplot as plt
for key in trajectories:
for key in trending_data.claims:
plt.plot(trajectories[key], label=key)
plt.legend()
plt.show()
# One global instance
# pylint: disable=C0103
trending_data = TrendingData()
def run(db, height, final_height, recalculate_claim_hashes):
if height < final_height - 5*HALF_LIFE:
trending_log("Skipping variable_decay trending at block {h}.\n".format(h=height))
return
start = time.time()
trending_log("Calculating variable_decay trending at block {h}.\n".format(h=height))
trending_log(" Length of trending data = {l}.\n"\
.format(l=len(trending_data.claims)))
# Renormalise trending scores and mark all as having changed
if height % RENORM_INTERVAL == 0:
trending_log(" Renormalising trending scores...")
keys = trending_data.claims.keys()
trending_data.whales = set([])
for key in keys:
if trending_data.claims[key]["trending_score"] != 0.0:
trending_data.claims[key]["trending_score"] *= DECAY_PER_RENORM
trending_data.claims[key]["changed"] = True
# Tiny becomes zero
if abs(trending_data.claims[key]["trending_score"]) < 1E-3:
trending_data.claims[key]["trending_score"] = 0.0
# Re-mark whales
if trending_data.claims[key]["trending_score"] >= WHALE_THRESHOLD*get_time_boost(height):
trending_data.add_whale(key)
trending_log("done.\n")
# Regular message.
trending_log(" Reading total_amounts from db and updating"\
+ " trending scores in RAM...")
# Update claims from db
if not trending_data.initialised:
trending_log("initial load...")
# On fresh launch
for row in db.execute("""
SELECT claim_hash, trending_mixed,
(amount + support_amount)
AS total_amount
FROM claim;
"""):
trending_data.insert_claim_from_load(height, row[0], row[1], 1E-8*row[2])
trending_data.initialised = True
else:
for row in db.execute(f"""
SELECT claim_hash,
(amount + support_amount)
AS total_amount
FROM claim
WHERE claim_hash IN
({','.join('?' for _ in recalculate_claim_hashes)});
""", recalculate_claim_hashes):
trending_data.update_claim(height, row[0], 1E-8*row[1])
# Apply pending spikes
trending_data.apply_spikes(height)
trending_log("done.\n")
# Write trending scores to DB
if height % SAVE_INTERVAL == 0:
trending_log(" Finding and processing whales...")
trending_log(str(len(trending_data.whales)) + " whales found...")
trending_data.process_whales(height)
trending_log("done.\n")
trending_log(" Writing trending scores to db...")
the_list = []
keys = trending_data.claims.keys()
for key in keys:
if trending_data.claims[key]["changed"]:
the_list.append((trending_data.claims[key]["trending_score"], key))
trending_data.claims[key]["changed"] = False
trending_log("{n} scores to write...".format(n=len(the_list)))
db.executemany("UPDATE claim SET trending_mixed=? WHERE claim_hash=?;",
the_list)
trending_log("done.\n")
trending_log("Trending operations took {time} seconds.\n\n"\
.format(time=time.time() - start))
if __name__ == "__main__":
+32 -128
View File
@@ -1,12 +1,12 @@
import os
import apsw
from typing import Union, Tuple, Set, List
from itertools import chain
from decimal import Decimal
from collections import namedtuple
from multiprocessing import Manager
from binascii import unhexlify, hexlify
from binascii import unhexlify
from lbry.wallet.server.leveldb import LevelDB
from lbry.wallet.server.util import class_logger
from lbry.wallet.database import query, constraints_to_sql
@@ -16,10 +16,11 @@ from lbry.schema.mime_types import guess_stream_type
from lbry.wallet import Ledger, RegTestLedger
from lbry.wallet.transaction import Transaction, Output
from lbry.wallet.server.db.canonical import register_canonical_functions
from lbry.wallet.server.db.full_text_search import update_full_text_search, CREATE_FULL_TEXT_SEARCH, first_sync_finished
from lbry.wallet.server.db.trending import TRENDING_ALGORITHMS
from .common import CLAIM_TYPES, STREAM_TYPES, COMMON_TAGS, INDEXED_LANGUAGES
from lbry.wallet.server.db.elasticsearch import SearchIndex
from .common import CLAIM_TYPES, STREAM_TYPES, COMMON_TAGS
ATTRIBUTE_ARRAY_MAX_LENGTH = 100
@@ -55,7 +56,6 @@ class SQLDB:
description text,
claim_type integer,
has_source bool,
reposted integer default 0,
-- streams
@@ -117,15 +117,6 @@ class SQLDB:
create unique index if not exists tag_claim_hash_tag_idx on tag (claim_hash, tag);
"""
CREATE_LANGUAGE_TABLE = """
create table if not exists language (
language text not null,
claim_hash bytes not null,
height integer not null
);
create unique index if not exists language_claim_hash_language_idx on language (claim_hash, language);
"""
CREATE_CLAIMTRIE_TABLE = """
create table if not exists claimtrie (
normalized text primary key,
@@ -135,22 +126,6 @@ class SQLDB:
create index if not exists claimtrie_claim_hash_idx on claimtrie (claim_hash);
"""
CREATE_CHANGELOG_TRIGGER = """
create table if not exists changelog (
claim_hash bytes primary key
);
create index if not exists claimtrie_claim_hash_idx on claimtrie (claim_hash);
create trigger if not exists claim_changelog after update on claim
begin
insert or ignore into changelog (claim_hash) values (new.claim_hash);
end;
create trigger if not exists claimtrie_changelog after update on claimtrie
begin
insert or ignore into changelog (claim_hash) values (new.claim_hash);
insert or ignore into changelog (claim_hash) values (old.claim_hash);
end;
"""
SEARCH_INDEXES = """
-- used by any tag clouds
create index if not exists tag_tag_idx on tag (tag, claim_hash);
@@ -199,18 +174,12 @@ class SQLDB:
for tag_value, tag_key in COMMON_TAGS.items()
)
LANGUAGE_INDEXES = '\n'.join(
f"create unique index if not exists language_{language}_idx on language (language, claim_hash) WHERE language='{language}';"
for language in INDEXED_LANGUAGES
)
CREATE_TABLES_QUERY = (
CREATE_CLAIM_TABLE +
CREATE_FULL_TEXT_SEARCH +
CREATE_SUPPORT_TABLE +
CREATE_CLAIMTRIE_TABLE +
CREATE_TAG_TABLE +
CREATE_CHANGELOG_TRIGGER +
CREATE_LANGUAGE_TABLE
CREATE_TAG_TABLE
)
def __init__(
@@ -220,6 +189,7 @@ class SQLDB:
self.db = None
self.logger = class_logger(__name__, self.__class__.__name__)
self.ledger = Ledger if main.coin.NET == 'mainnet' else RegTestLedger
self._fts_synced = False
self.state_manager = None
self.blocked_streams = None
self.blocked_channels = None
@@ -232,7 +202,6 @@ class SQLDB:
unhexlify(channel_id)[::-1] for channel_id in filtering_channels if channel_id
}
self.trending = trending
self.pending_deletes = set()
def open(self):
self.db = apsw.Connection(
@@ -336,7 +305,7 @@ class SQLDB:
self.execute('commit;')
def _upsertable_claims(self, txos: List[Output], header, clear_first=False):
claim_hashes, claims, tags, languages = set(), [], {}, {}
claim_hashes, claims, tags = set(), [], {}
for txo in txos:
tx = txo.tx_ref.tx
@@ -347,13 +316,6 @@ class SQLDB:
#self.logger.exception(f"Could not decode claim name for {tx.id}:{txo.position}.")
continue
language = 'none'
try:
if txo.claim.is_stream and txo.claim.stream.languages:
language = txo.claim.stream.languages[0].language
except:
pass
claim_hash = txo.claim_hash
claim_hashes.add(claim_hash)
claim_record = {
@@ -371,7 +333,6 @@ class SQLDB:
'author': None,
'duration': None,
'claim_type': None,
'has_source': False,
'stream_type': None,
'media_type': None,
'release_time': None,
@@ -389,7 +350,6 @@ class SQLDB:
if claim.is_stream:
claim_record['claim_type'] = CLAIM_TYPES['stream']
claim_record['has_source'] = claim.stream.has_source
claim_record['media_type'] = claim.stream.source.media_type
claim_record['stream_type'] = STREAM_TYPES[guess_stream_type(claim_record['media_type'])]
claim_record['title'] = claim.stream.title
@@ -412,10 +372,6 @@ class SQLDB:
claim_record['reposted_claim_hash'] = claim.repost.reference.claim_hash
elif claim.is_channel:
claim_record['claim_type'] = CLAIM_TYPES['channel']
elif claim.is_collection:
claim_record['claim_type'] = CLAIM_TYPES['collection']
languages[(language, claim_hash)] = (language, claim_hash, tx.height)
for tag in clean_tags(claim.message.tags):
tags[(tag, claim_hash)] = (tag, claim_hash, tx.height)
@@ -427,10 +383,6 @@ class SQLDB:
self.executemany(
"INSERT OR IGNORE INTO tag (tag, claim_hash, height) VALUES (?, ?, ?)", tags.values()
)
if languages:
self.executemany(
"INSERT OR IGNORE INTO language (language, claim_hash, height) VALUES (?, ?, ?)", languages.values()
)
return claims
@@ -438,14 +390,14 @@ class SQLDB:
claims = self._upsertable_claims(txos, header)
if claims:
self.executemany("""
INSERT OR REPLACE INTO claim (
INSERT OR IGNORE INTO claim (
claim_hash, claim_id, claim_name, normalized, txo_hash, tx_position, amount,
claim_type, media_type, stream_type, timestamp, creation_timestamp, has_source,
claim_type, media_type, stream_type, timestamp, creation_timestamp,
fee_currency, fee_amount, title, description, author, duration, height, reposted_claim_hash,
creation_height, release_time, activation_height, expiration_height, short_url)
VALUES (
:claim_hash, :claim_id, :claim_name, :normalized, :txo_hash, :tx_position, :amount,
:claim_type, :media_type, :stream_type, :timestamp, :timestamp, :has_source,
:claim_type, :media_type, :stream_type, :timestamp, :timestamp,
:fee_currency, :fee_amount, :title, :description, :author, :duration, :height, :reposted_claim_hash, :height,
CASE WHEN :release_time IS NOT NULL THEN :release_time ELSE :timestamp END,
CASE WHEN :normalized NOT IN (SELECT normalized FROM claimtrie) THEN :height END,
@@ -463,7 +415,7 @@ class SQLDB:
UPDATE claim SET
txo_hash=:txo_hash, tx_position=:tx_position, amount=:amount, height=:height,
claim_type=:claim_type, media_type=:media_type, stream_type=:stream_type,
timestamp=:timestamp, fee_amount=:fee_amount, fee_currency=:fee_currency, has_source=:has_source,
timestamp=:timestamp, fee_amount=:fee_amount, fee_currency=:fee_currency,
title=:title, duration=:duration, description=:description, author=:author, reposted_claim_hash=:reposted_claim_hash,
release_time=CASE WHEN :release_time IS NOT NULL THEN :release_time ELSE release_time END
WHERE claim_hash=:claim_hash;
@@ -547,7 +499,6 @@ class SQLDB:
WHERE claim_hash = ?
""", targets
)
return set(target[0] for target in targets)
def validate_channel_signatures(self, height, new_claims, updated_claims, spent_claims, affected_channels, timer):
if not new_claims and not updated_claims and not spent_claims:
@@ -821,56 +772,11 @@ class SQLDB:
f"SELECT claim_hash, normalized FROM claim WHERE expiration_height = {height}"
)
def enqueue_changes(self):
for claim in self.execute(f"""
SELECT claimtrie.claim_hash as is_controlling,
claimtrie.last_take_over_height,
(select group_concat(tag, ',,') from tag where tag.claim_hash in (claim.claim_hash, claim.reposted_claim_hash)) as tags,
(select group_concat(language, ' ') from language where language.claim_hash in (claim.claim_hash, claim.reposted_claim_hash)) as languages,
(select cr.has_source from claim cr where cr.claim_hash = claim.reposted_claim_hash) as reposted_has_source,
claim.*
FROM claim LEFT JOIN claimtrie USING (claim_hash)
WHERE claim.claim_hash in (SELECT claim_hash FROM changelog)
"""):
claim = claim._asdict()
id_set = set(filter(None, (claim['claim_hash'], claim['channel_hash'], claim['reposted_claim_hash'])))
claim['censor_type'] = 0
claim['censoring_channel_hash'] = None
claim['has_source'] = bool(claim.pop('reposted_has_source') or claim['has_source'])
for reason_id in id_set:
if reason_id in self.blocked_streams:
claim['censor_type'] = 2
claim['censoring_channel_hash'] = self.blocked_streams.get(reason_id)
elif reason_id in self.blocked_channels:
claim['censor_type'] = 2
claim['censoring_channel_hash'] = self.blocked_channels.get(reason_id)
elif reason_id in self.filtered_streams:
claim['censor_type'] = 1
claim['censoring_channel_hash'] = self.filtered_streams.get(reason_id)
elif reason_id in self.filtered_channels:
claim['censor_type'] = 1
claim['censoring_channel_hash'] = self.filtered_channels.get(reason_id)
claim['tags'] = claim['tags'].split(',,') if claim['tags'] else []
claim['languages'] = claim['languages'].split(' ') if claim['languages'] else []
yield 'update', claim
def clear_changelog(self):
self.execute("delete from changelog;")
def claim_producer(self):
while self.pending_deletes:
claim_hash = self.pending_deletes.pop()
yield 'delete', hexlify(claim_hash[::-1]).decode()
for claim in self.enqueue_changes():
yield claim
self.clear_changelog()
def advance_txs(self, height, all_txs, header, daemon_height, timer):
insert_claims = []
update_claims = []
update_claim_hashes = set()
delete_claim_hashes = self.pending_deletes
delete_claim_hashes = set()
insert_supports = []
delete_support_txo_hashes = set()
recalculate_claim_hashes = set() # added/deleted supports, added/updated claim
@@ -939,17 +845,28 @@ class SQLDB:
expire_timer.stop()
r = timer.run
r(update_full_text_search, 'before-delete',
delete_claim_hashes, self.db.cursor(), self.main.first_sync)
affected_channels = r(self.delete_claims, delete_claim_hashes)
r(self.delete_supports, delete_support_txo_hashes)
r(self.insert_claims, insert_claims, header)
r(self.calculate_reposts, insert_claims)
r(update_full_text_search, 'after-insert',
[txo.claim_hash for txo in insert_claims], self.db.cursor(), self.main.first_sync)
r(update_full_text_search, 'before-update',
[txo.claim_hash for txo in update_claims], self.db.cursor(), self.main.first_sync)
r(self.update_claims, update_claims, header)
r(update_full_text_search, 'after-update',
[txo.claim_hash for txo in update_claims], self.db.cursor(), self.main.first_sync)
r(self.validate_channel_signatures, height, insert_claims,
update_claims, delete_claim_hashes, affected_channels, forward_timer=True)
r(self.insert_supports, insert_supports)
r(self.update_claimtrie, height, recalculate_claim_hashes, deleted_claim_names, forward_timer=True)
for algorithm in self.trending:
r(algorithm.run, self.db.cursor(), height, daemon_height, recalculate_claim_hashes)
if not self._fts_synced and self.main.first_sync and height == daemon_height:
r(first_sync_finished, self.db.cursor())
self._fts_synced = True
class LBRYLevelDB(LevelDB):
@@ -961,30 +878,17 @@ class LBRYLevelDB(LevelDB):
for algorithm_name in self.env.trending_algorithms:
if algorithm_name in TRENDING_ALGORITHMS:
trending.append(TRENDING_ALGORITHMS[algorithm_name])
if self.env.es_mode == 'reader':
self.logger.info('Index mode: reader')
self.sql = None
else:
self.logger.info('Index mode: writer. Using SQLite db to sync ES')
self.sql = SQLDB(
self, path,
self.env.default('BLOCKING_CHANNEL_IDS', '').split(' '),
self.env.default('FILTERING_CHANNEL_IDS', '').split(' '),
trending
)
# Search index
self.search_index = SearchIndex(
self.env.es_index_prefix, self.env.database_query_timeout, self.env.elastic_host, self.env.elastic_port
self.sql = SQLDB(
self, path,
self.env.default('BLOCKING_CHANNEL_IDS', '').split(' '),
self.env.default('FILTERING_CHANNEL_IDS', '').split(' '),
trending
)
def close(self):
super().close()
if self.sql:
self.sql.close()
self.sql.close()
async def _open_dbs(self, *args, **kwargs):
await self.search_index.start()
await super()._open_dbs(*args, **kwargs)
if self.sql:
self.sql.open()
self.sql.open()
+2 -9
View File
@@ -33,15 +33,11 @@ class Env:
self.allow_root = self.boolean('ALLOW_ROOT', False)
self.host = self.default('HOST', 'localhost')
self.rpc_host = self.default('RPC_HOST', 'localhost')
self.elastic_host = self.default('ELASTIC_HOST', 'localhost')
self.elastic_port = self.integer('ELASTIC_PORT', 9200)
self.loop_policy = self.set_event_loop_policy()
self.obsolete(['UTXO_MB', 'HIST_MB', 'NETWORK'])
self.db_dir = self.required('DB_DIRECTORY')
self.db_engine = self.default('DB_ENGINE', 'leveldb')
self.trending_algorithms = [
trending for trending in set(self.default('TRENDING_ALGORITHMS', 'zscore').split(' ')) if trending
]
self.trending_algorithms = list(set(self.default('TRENDING_ALGORITHMS', 'zscore').split(' ')))
self.max_query_workers = self.integer('MAX_QUERY_WORKERS', None)
self.individual_tag_indexes = self.boolean('INDIVIDUAL_TAG_INDEXES', True)
self.track_metrics = self.boolean('TRACK_METRICS', False)
@@ -55,13 +51,10 @@ class Env:
coin_name = self.required('COIN').strip()
network = self.default('NET', 'mainnet').strip()
self.coin = Coin.lookup_coin_class(coin_name, network)
self.es_index_prefix = self.default('ES_INDEX_PREFIX', '')
self.es_mode = self.default('ES_MODE', 'writer')
self.cache_MB = self.integer('CACHE_MB', 1200)
self.reorg_limit = self.integer('REORG_LIMIT', self.coin.REORG_LIMIT)
# Server stuff
self.tcp_port = self.integer('TCP_PORT', None)
self.udp_port = self.integer('UDP_PORT', self.tcp_port)
self.ssl_port = self.integer('SSL_PORT', None)
if self.ssl_port:
self.ssl_certfile = self.required('SSL_CERTFILE')
@@ -99,7 +92,7 @@ class Env:
self.identities = [identity
for identity in (clearnet_identity, tor_identity)
if identity is not None]
self.database_query_timeout = float(self.integer('QUERY_TIMEOUT_MS', 3000)) / 1000.0
self.database_query_timeout = float(self.integer('QUERY_TIMEOUT_MS', 250)) / 1000.0
@classmethod
def default(cls, envvar, default):
+4 -4
View File
@@ -67,17 +67,17 @@ def hash160(x):
return ripemd160(sha256(x))
def hash_to_hex_str(x: bytes) -> str:
def hash_to_hex_str(x):
"""Convert a big-endian binary hash to displayed hex string.
Display form of a binary hash is reversed and converted to hex.
"""
return x[::-1].hex()
return bytes(reversed(x)).hex()
def hex_str_to_hash(x: str) -> bytes:
def hex_str_to_hash(x):
"""Convert a displayed hex string to a binary hash."""
return hex_to_bytes(x)[::-1]
return bytes(reversed(hex_to_bytes(x)))
class Base58Error(Exception):
+14 -14
View File
@@ -173,20 +173,20 @@ class History:
self.logger.info(f'backing up removed {nremoves:,d} history entries')
# def get_txnums(self, hashX, limit=1000):
# """Generator that returns an unpruned, sorted list of tx_nums in the
# history of a hashX. Includes both spending and receiving
# transactions. By default yields at most 1000 entries. Set
# limit to None to get them all. """
# limit = util.resolve_limit(limit)
# for key, hist in self.db.iterator(prefix=hashX):
# a = array.array('I')
# a.frombytes(hist)
# for tx_num in a:
# if limit == 0:
# return
# yield tx_num
# limit -= 1
def get_txnums(self, hashX, limit=1000):
"""Generator that returns an unpruned, sorted list of tx_nums in the
history of a hashX. Includes both spending and receiving
transactions. By default yields at most 1000 entries. Set
limit to None to get them all. """
limit = util.resolve_limit(limit)
for key, hist in self.db.iterator(prefix=hashX):
a = array.array('I')
a.frombytes(hist)
for tx_num in a:
if limit == 0:
return
yield tx_num
limit -= 1
#
# History compaction
+121 -249
View File
@@ -12,20 +12,17 @@
import asyncio
import array
import ast
import base64
import os
import time
import zlib
import typing
from typing import Optional, List, Tuple, Iterable
from asyncio import sleep
from bisect import bisect_right
from collections import namedtuple
from functools import partial
from glob import glob
from struct import pack, unpack
from concurrent.futures.thread import ThreadPoolExecutor
import attr
from lbry.utils import LRUCacheWithMetrics
from lbry.wallet.server import util
from lbry.wallet.server.hash import hash_to_hex_str, HASHX_LEN
from lbry.wallet.server.merkle import Merkle, MerkleCache
@@ -35,13 +32,6 @@ from lbry.wallet.server.history import History
UTXO = namedtuple("UTXO", "tx_num tx_pos tx_hash height value")
HEADER_PREFIX = b'H'
TX_COUNT_PREFIX = b'T'
TX_HASH_PREFIX = b'X'
TX_PREFIX = b'B'
TX_NUM_PREFIX = b'N'
BLOCK_HASH_PREFIX = b'C'
@attr.s(slots=True)
@@ -49,8 +39,7 @@ class FlushData:
height = attr.ib()
tx_count = attr.ib()
headers = attr.ib()
block_hashes = attr.ib()
block_txs = attr.ib()
block_tx_hashes = attr.ib()
# The following are flushed to the UTXO DB if undo_infos is not None
undo_infos = attr.ib()
adds = attr.ib()
@@ -74,7 +63,15 @@ class LevelDB:
self.logger = util.class_logger(__name__, self.__class__.__name__)
self.env = env
self.coin = env.coin
self.executor = None
self.executor = ThreadPoolExecutor(max(1, os.cpu_count() - 1))
# Setup block header size handlers
if self.coin.STATIC_BLOCK_HEADERS:
self.header_offset = self.coin.static_header_offset
self.header_len = self.coin.static_header_len
else:
self.header_offset = self.dynamic_header_offset
self.header_len = self.dynamic_header_len
self.logger.info(f'switching current directory to {env.db_dir}')
@@ -82,8 +79,6 @@ class LevelDB:
self.history = History()
self.utxo_db = None
self.tx_counts = None
self.headers = None
self.encoded_headers = LRUCacheWithMetrics(1 << 21, metric_name='encoded_headers', namespace='wallet_server')
self.last_flush = time.time()
self.logger.info(f'using {self.env.db_engine} for DB backend')
@@ -92,107 +87,60 @@ class LevelDB:
self.merkle = Merkle()
self.header_mc = MerkleCache(self.merkle, self.fs_block_hashes)
self.headers_db = None
self.tx_db = None
self._tx_and_merkle_cache = LRUCacheWithMetrics(2 ** 17, metric_name='tx_and_merkle', namespace="wallet_server")
self.total_transactions = None
path = partial(os.path.join, self.env.db_dir)
self.headers_file = util.LogicalFile(path('meta/headers'), 2, 16000000)
self.tx_counts_file = util.LogicalFile(path('meta/txcounts'), 2, 2000000)
self.hashes_file = util.LogicalFile(path('meta/hashes'), 4, 16000000)
if not self.coin.STATIC_BLOCK_HEADERS:
self.headers_offsets_file = util.LogicalFile(
path('meta/headers_offsets'), 2, 16000000)
async def _read_tx_counts(self):
if self.tx_counts is not None:
return
# tx_counts[N] has the cumulative number of txs at the end of
# height N. So tx_counts[0] is 1 - the genesis coinbase
def get_counts():
return tuple(
util.unpack_be_uint64(tx_count)
for tx_count in self.tx_db.iterator(prefix=TX_COUNT_PREFIX, include_key=False)
)
tx_counts = await asyncio.get_event_loop().run_in_executor(self.executor, get_counts)
assert len(tx_counts) == self.db_height + 1, f"{len(tx_counts)} vs {self.db_height + 1}"
size = (self.db_height + 1) * 4
tx_counts = self.tx_counts_file.read(0, size)
assert len(tx_counts) == size
self.tx_counts = array.array('I', tx_counts)
if self.tx_counts:
assert self.db_tx_count == self.tx_counts[-1], \
f"{self.db_tx_count} vs {self.tx_counts[-1]} ({len(self.tx_counts)} counts)"
assert self.db_tx_count == self.tx_counts[-1]
else:
assert self.db_tx_count == 0
async def _read_txids(self):
def get_txids():
return list(self.tx_db.iterator(prefix=TX_HASH_PREFIX, include_key=False))
start = time.perf_counter()
self.logger.info("loading txids")
txids = await asyncio.get_event_loop().run_in_executor(self.executor, get_txids)
assert len(txids) == len(self.tx_counts) == 0 or len(txids) == self.tx_counts[-1]
self.total_transactions = txids
ts = time.perf_counter() - start
self.logger.info("loaded %i txids in %ss", len(self.total_transactions), round(ts, 4))
async def _read_headers(self):
if self.headers is not None:
return
def get_headers():
return [
header for header in self.headers_db.iterator(prefix=HEADER_PREFIX, include_key=False)
]
headers = await asyncio.get_event_loop().run_in_executor(self.executor, get_headers)
assert len(headers) - 1 == self.db_height, f"{len(headers)} vs {self.db_height}"
self.headers = headers
async def _open_dbs(self, for_sync, compacting):
if self.executor is None:
self.executor = ThreadPoolExecutor(1)
coin_path = os.path.join(self.env.db_dir, 'COIN')
if not os.path.isfile(coin_path):
with util.open_file(coin_path, create=True) as f:
f.write(f'ElectrumX databases and metadata for '
f'{self.coin.NAME} {self.coin.NET}'.encode())
assert self.headers_db is None
self.headers_db = self.db_class('headers', for_sync)
if self.headers_db.is_new:
self.logger.info('created new headers db')
self.logger.info(f'opened headers DB (for sync: {for_sync})')
assert self.tx_db is None
self.tx_db = self.db_class('tx', for_sync)
if self.tx_db.is_new:
self.logger.info('created new tx db')
self.logger.info(f'opened tx DB (for sync: {for_sync})')
assert self.utxo_db is None
# First UTXO DB
self.utxo_db = self.db_class('utxo', for_sync)
if self.utxo_db.is_new:
self.logger.info('created new utxo db')
self.logger.info(f'opened utxo db (for sync: {for_sync})')
self.logger.info('created new database')
self.logger.info('creating metadata directory')
os.mkdir(os.path.join(self.env.db_dir, 'meta'))
coin_path = os.path.join(self.env.db_dir, 'meta', 'COIN')
with util.open_file(coin_path, create=True) as f:
f.write(f'ElectrumX databases and metadata for '
f'{self.coin.NAME} {self.coin.NET}'.encode())
if not self.coin.STATIC_BLOCK_HEADERS:
self.headers_offsets_file.write(0, bytes(8))
else:
self.logger.info(f'opened UTXO DB (for sync: {for_sync})')
self.read_utxo_state()
# Then history DB
self.utxo_flush_count = self.history.open_db(
self.db_class, for_sync, self.utxo_flush_count, compacting
)
self.utxo_flush_count = self.history.open_db(self.db_class, for_sync,
self.utxo_flush_count,
compacting)
self.clear_excess_undo_info()
# Read TX counts (requires meta directory)
await self._read_tx_counts()
if self.total_transactions is None:
await self._read_txids()
await self._read_headers()
def close(self):
self.utxo_db.close()
self.history.close_db()
self.headers_db.close()
self.tx_db.close()
self.executor.shutdown(wait=True)
self.executor = None
async def open_for_compacting(self):
await self._open_dbs(True, True)
@@ -204,28 +152,18 @@ class LevelDB:
synchronization. When serving clients we want the open files for
serving network connections.
"""
self.logger.info("opened for sync")
await self._open_dbs(True, False)
async def open_for_serving(self):
"""Open the databases for serving. If they are already open they are
closed first.
"""
self.logger.info('closing DBs to re-open for serving')
if self.utxo_db:
self.logger.info('closing DBs to re-open for serving')
self.utxo_db.close()
self.history.close_db()
self.utxo_db = None
if self.headers_db:
self.headers_db.close()
self.headers_db = None
if self.tx_db:
self.tx_db.close()
self.tx_db = None
await self._open_dbs(False, False)
self.logger.info("opened for serving")
# Header merkle cache
@@ -247,7 +185,7 @@ class LevelDB:
assert flush_data.height == self.fs_height == self.db_height
assert flush_data.tip == self.db_tip
assert not flush_data.headers
assert not flush_data.block_txs
assert not flush_data.block_tx_hashes
assert not flush_data.adds
assert not flush_data.deletes
assert not flush_data.undo_infos
@@ -305,50 +243,36 @@ class LevelDB:
"""
prior_tx_count = (self.tx_counts[self.fs_height]
if self.fs_height >= 0 else 0)
assert len(flush_data.block_txs) == len(flush_data.headers)
assert len(flush_data.block_tx_hashes) == len(flush_data.headers)
assert flush_data.height == self.fs_height + len(flush_data.headers)
assert flush_data.tx_count == (self.tx_counts[-1] if self.tx_counts
else 0)
assert len(self.tx_counts) == flush_data.height + 1
assert len(
b''.join(hashes for hashes, _ in flush_data.block_txs)
) // 32 == flush_data.tx_count - prior_tx_count
hashes = b''.join(flush_data.block_tx_hashes)
flush_data.block_tx_hashes.clear()
assert len(hashes) % 32 == 0
assert len(hashes) // 32 == flush_data.tx_count - prior_tx_count
# Write the headers
start_time = time.perf_counter()
with self.headers_db.write_batch() as batch:
batch_put = batch.put
for i, header in enumerate(flush_data.headers):
batch_put(HEADER_PREFIX + util.pack_be_uint64(self.fs_height + i + 1), header)
self.headers.append(header)
# Write the headers, tx counts, and tx hashes
start_time = time.time()
height_start = self.fs_height + 1
offset = self.header_offset(height_start)
self.headers_file.write(offset, b''.join(flush_data.headers))
self.fs_update_header_offsets(offset, height_start, flush_data.headers)
flush_data.headers.clear()
height_start = self.fs_height + 1
tx_num = prior_tx_count
with self.tx_db.write_batch() as batch:
batch_put = batch.put
for block_hash, (tx_hashes, txs) in zip(flush_data.block_hashes, flush_data.block_txs):
tx_count = self.tx_counts[height_start]
batch_put(BLOCK_HASH_PREFIX + util.pack_be_uint64(height_start), block_hash[::-1])
batch_put(TX_COUNT_PREFIX + util.pack_be_uint64(height_start), util.pack_be_uint64(tx_count))
height_start += 1
offset = 0
while offset < len(tx_hashes):
batch_put(TX_HASH_PREFIX + util.pack_be_uint64(tx_num), tx_hashes[offset:offset+32])
batch_put(TX_NUM_PREFIX + tx_hashes[offset:offset+32], util.pack_be_uint64(tx_num))
batch_put(TX_PREFIX + tx_hashes[offset:offset+32], txs[offset // 32])
tx_num += 1
offset += 32
flush_data.block_txs.clear()
flush_data.block_hashes.clear()
offset = height_start * self.tx_counts.itemsize
self.tx_counts_file.write(offset,
self.tx_counts[height_start:].tobytes())
offset = prior_tx_count * 32
self.hashes_file.write(offset, hashes)
self.fs_height = flush_data.height
self.fs_tx_count = flush_data.tx_count
elapsed = time.perf_counter() - start_time
self.logger.info(f'flushed filesystem data in {elapsed:.2f}s')
if self.utxo_db.for_sync:
elapsed = time.time() - start_time
self.logger.info(f'flushed filesystem data in {elapsed:.2f}s')
def flush_history(self):
self.history.flush()
@@ -407,7 +331,7 @@ class LevelDB:
def flush_backup(self, flush_data, touched):
"""Like flush_dbs() but when backing up. All UTXOs are flushed."""
assert not flush_data.headers
assert not flush_data.block_txs
assert not flush_data.block_tx_hashes
assert flush_data.height < self.db_height
self.history.assert_flushed()
@@ -426,33 +350,43 @@ class LevelDB:
f'{elapsed:.1f}s. Height {flush_data.height:,d} '
f'txs: {flush_data.tx_count:,d} ({tx_delta:+,d})')
def fs_update_header_offsets(self, offset_start, height_start, headers):
if self.coin.STATIC_BLOCK_HEADERS:
return
offset = offset_start
offsets = []
for h in headers:
offset += len(h)
offsets.append(pack("<Q", offset))
# For each header we get the offset of the next header, hence we
# start writing from the next height
pos = (height_start + 1) * 8
self.headers_offsets_file.write(pos, b''.join(offsets))
def dynamic_header_offset(self, height):
assert not self.coin.STATIC_BLOCK_HEADERS
offset, = unpack('<Q', self.headers_offsets_file.read(height * 8, 8))
return offset
def dynamic_header_len(self, height):
return self.dynamic_header_offset(height + 1)\
- self.dynamic_header_offset(height)
def backup_fs(self, height, tx_count):
"""Back up during a reorg. This just updates our pointers."""
while self.fs_height > height:
self.fs_height -= 1
self.headers.pop()
self.fs_height = height
self.fs_tx_count = tx_count
# Truncate header_mc: header count is 1 more than the height.
self.header_mc.truncate(height + 1)
def raw_header(self, height):
async def raw_header(self, height):
"""Return the binary header at the given height."""
header, n = self.read_headers(height, 1)
header, n = await self.read_headers(height, 1)
if n != 1:
raise IndexError(f'height {height:,d} out of range')
return header
def encode_headers(self, start_height, count, headers):
key = (start_height, count)
if not self.encoded_headers.get(key):
compressobj = zlib.compressobj(wbits=-15, level=1, memLevel=9)
headers = base64.b64encode(compressobj.compress(headers) + compressobj.flush()).decode()
if start_height % 1000 != 0:
return headers
self.encoded_headers[key] = headers
return self.encoded_headers.get(key)
def read_headers(self, start_height, count) -> typing.Tuple[bytes, int]:
async def read_headers(self, start_height, count):
"""Requires start_height >= 0, count >= 0. Reads as many headers as
are available starting at start_height up to count. This
would be zero if start_height is beyond self.db_height, for
@@ -461,15 +395,20 @@ class LevelDB:
Returns a (binary, n) pair where binary is the concatenated
binary headers, and n is the count of headers returned.
"""
if start_height < 0 or count < 0:
raise self.DBError(f'{count:,d} headers starting at '
f'{start_height:,d} not on disk')
disk_count = max(0, min(count, self.db_height + 1 - start_height))
if disk_count:
return b''.join(self.headers[start_height:start_height + disk_count]), disk_count
return b'', 0
def read_headers():
# Read some from disk
disk_count = max(0, min(count, self.db_height + 1 - start_height))
if disk_count:
offset = self.header_offset(start_height)
size = self.header_offset(start_height + disk_count) - offset
return self.headers_file.read(offset, size), disk_count
return b'', 0
return await asyncio.get_event_loop().run_in_executor(self.executor, read_headers)
def fs_tx_hash(self, tx_num):
"""Return a par (tx_hash, tx_height) for the given tx number.
@@ -477,67 +416,23 @@ class LevelDB:
If the tx_height is not on disk, returns (None, tx_height)."""
tx_height = bisect_right(self.tx_counts, tx_num)
if tx_height > self.db_height:
return None, tx_height
try:
return self.total_transactions[tx_num], tx_height
except IndexError:
self.logger.exception(
"Failed to access a cached transaction, known bug #3142 "
"should be fixed in #3205"
)
return None, tx_height
def _fs_transactions(self, txids: Iterable[str]):
unpack_be_uint64 = util.unpack_be_uint64
tx_counts = self.tx_counts
tx_db_get = self.tx_db.get
tx_cache = self._tx_and_merkle_cache
tx_infos = {}
for tx_hash in txids:
cached_tx = tx_cache.get(tx_hash)
if cached_tx:
tx, merkle = cached_tx
else:
tx_hash_bytes = bytes.fromhex(tx_hash)[::-1]
tx_num = tx_db_get(TX_NUM_PREFIX + tx_hash_bytes)
tx = None
tx_height = -1
if tx_num is not None:
tx_num = unpack_be_uint64(tx_num)
tx_height = bisect_right(tx_counts, tx_num)
if tx_height < self.db_height:
tx = tx_db_get(TX_PREFIX + tx_hash_bytes)
if tx_height == -1:
merkle = {
'block_height': -1
}
else:
tx_pos = tx_num - tx_counts[tx_height - 1]
branch, root = self.merkle.branch_and_root(
self.total_transactions[tx_counts[tx_height - 1]:tx_counts[tx_height]], tx_pos
)
merkle = {
'block_height': tx_height,
'merkle': [
hash_to_hex_str(hash)
for hash in branch
],
'pos': tx_pos
}
if tx_height + 10 < self.db_height:
tx_cache[tx_hash] = tx, merkle
tx_infos[tx_hash] = (None if not tx else tx.hex(), merkle)
return tx_infos
async def fs_transactions(self, txids):
return await asyncio.get_event_loop().run_in_executor(self.executor, self._fs_transactions, txids)
tx_hash = None
else:
tx_hash = self.hashes_file.read(tx_num * 32, 32)
return tx_hash, tx_height
async def fs_block_hashes(self, height, count):
if height + count > len(self.headers):
raise self.DBError(f'only got {len(self.headers) - height:,d} headers starting at {height:,d}, not {count:,d}')
return [self.coin.header_hash(header) for header in self.headers[height:height + count]]
headers_concat, headers_count = await self.read_headers(height, count)
if headers_count != count:
raise self.DBError(f'only got {headers_count:,d} headers starting at {height:,d}, not {count:,d}')
offset = 0
headers = []
for n in range(count):
hlen = self.header_len(height + n)
headers.append(headers_concat[offset:offset + hlen])
offset += hlen
return [self.coin.header_hash(header) for header in headers]
async def limited_history(self, hashX, *, limit=1000):
"""Return an unpruned, sorted list of (tx_hash, height) tuples of
@@ -546,33 +441,15 @@ class LevelDB:
transactions. By default returns at most 1000 entries. Set
limit to None to get them all.
"""
def read_history():
db_height = self.db_height
tx_counts = self.tx_counts
cnt = 0
txs = []
for hist in self.history.db.iterator(prefix=hashX, include_key=False):
a = array.array('I')
a.frombytes(hist)
for tx_num in a:
tx_height = bisect_right(tx_counts, tx_num)
if tx_height > db_height:
return
txs.append((tx_num, tx_height))
cnt += 1
if limit and cnt >= limit:
break
if limit and cnt >= limit:
break
return txs
tx_nums = list(self.history.get_txnums(hashX, limit))
fs_tx_hash = self.fs_tx_hash
return [fs_tx_hash(tx_num) for tx_num in tx_nums]
while True:
history = await asyncio.get_event_loop().run_in_executor(self.executor, read_history)
if history is not None:
return [(self.total_transactions[tx_num], tx_height) for (tx_num, tx_height) in history]
if all(hash is not None for hash, height in history):
return history
self.logger.warning(f'limited_history: tx hash '
f'not found (reorg?), retrying...')
await sleep(0.25)
@@ -597,20 +474,16 @@ class LevelDB:
batch_put(self.undo_key(height), b''.join(undo_info))
def raw_block_prefix(self):
return 'block'
return 'meta/block'
def raw_block_path(self, height):
return os.path.join(self.env.db_dir, f'{self.raw_block_prefix()}{height:d}')
async def read_raw_block(self, height):
def read_raw_block(self, height):
"""Returns a raw block read from disk. Raises FileNotFoundError
if the block isn't on-disk."""
def read():
with util.open_file(self.raw_block_path(height)) as f:
return f.read(-1)
return await asyncio.get_event_loop().run_in_executor(self.executor, read)
with util.open_file(self.raw_block_path(height)) as f:
return f.read(-1)
def write_raw_block(self, block, height):
"""Write a raw block to disk."""
@@ -729,14 +602,13 @@ class LevelDB:
utxos = []
utxos_append = utxos.append
s_unpack = unpack
fs_tx_hash = self.fs_tx_hash
# Key: b'u' + address_hashX + tx_idx + tx_num
# Value: the UTXO value as a 64-bit unsigned integer
prefix = b'u' + hashX
for db_key, db_value in self.utxo_db.iterator(prefix=prefix):
tx_pos, tx_num = s_unpack('<HI', db_key[-6:])
value, = unpack('<Q', db_value)
tx_hash, height = fs_tx_hash(tx_num)
tx_hash, height = self.fs_tx_hash(tx_num)
utxos_append(UTXO(tx_num, tx_pos, tx_hash, height, value))
return utxos
+45 -78
View File
@@ -6,12 +6,13 @@
# and warranty status of this software.
"""Mempool handling."""
import os
import asyncio
import itertools
import time
from abc import ABC, abstractmethod
from collections import defaultdict
from prometheus_client import Histogram
from concurrent.futures.thread import ThreadPoolExecutor
import attr
@@ -72,22 +73,12 @@ class MemPoolAPI(ABC):
"""
@abstractmethod
async def on_mempool(self, touched, new_touched, height):
async def on_mempool(self, touched, height):
"""Called each time the mempool is synchronized. touched is a set of
hashXs touched since the previous call. height is the
daemon's height at the time the mempool was obtained."""
NAMESPACE = "wallet_server"
HISTOGRAM_BUCKETS = (
.005, .01, .025, .05, .075, .1, .25, .5, .75, 1.0, 2.5, 5.0, 7.5, 10.0, 15.0, 20.0, 30.0, 60.0, float('inf')
)
mempool_process_time_metric = Histogram(
"processed_mempool", "Time to process mempool and notify touched addresses",
namespace=NAMESPACE, buckets=HISTOGRAM_BUCKETS
)
class MemPool:
"""Representation of the daemon's mempool.
@@ -102,7 +93,7 @@ class MemPool:
hashXs: hashX -> set of all hashes of txs touching the hashX
"""
def __init__(self, coin, api, refresh_secs=1.0, log_status_secs=120.0):
def __init__(self, coin, api, refresh_secs=5.0, log_status_secs=120.0):
assert isinstance(api, MemPoolAPI)
self.coin = coin
self.api = api
@@ -115,8 +106,7 @@ class MemPool:
# Prevents mempool refreshes during fee histogram calculation
self.lock = asyncio.Lock()
self.wakeup = asyncio.Event()
self.mempool_process_time_metric = mempool_process_time_metric
self.notified_mempool_txs = set()
self.executor = ThreadPoolExecutor(max(os.cpu_count() - 1, 1))
async def _logging(self, synchronized_event):
"""Print regular logs of mempool stats."""
@@ -133,11 +123,15 @@ class MemPool:
await synchronized_event.wait()
async def _refresh_histogram(self, synchronized_event):
while True:
await synchronized_event.wait()
async with self.lock:
self._update_histogram(100_000)
await asyncio.sleep(self.coin.MEMPOOL_HISTOGRAM_REFRESH_SECS)
try:
while True:
await synchronized_event.wait()
async with self.lock:
# Threaded as can be expensive
await asyncio.get_event_loop().run_in_executor(self.executor, self._update_histogram, 100_000)
await asyncio.sleep(self.coin.MEMPOOL_HISTOGRAM_REFRESH_SECS)
finally:
self.executor.shutdown(wait=True)
def _update_histogram(self, bin_size):
# Build a histogram by fee rate
@@ -210,36 +204,19 @@ class MemPool:
return deferred, {prevout: utxo_map[prevout] for prevout in unspent}
async def _mempool_loop(self, synchronized_event):
try:
return await self._refresh_hashes(synchronized_event)
except asyncio.CancelledError:
raise
except Exception as e:
self.logger.exception("MEMPOOL DIED")
raise e
async def _refresh_hashes(self, synchronized_event):
"""Refresh our view of the daemon's mempool."""
while True:
start = time.perf_counter()
height = self.api.cached_height()
hex_hashes = await self.api.mempool_hashes()
if height != await self.api.height():
continue
hashes = {hex_str_to_hash(hh) for hh in hex_hashes}
async with self.lock:
new_hashes = hashes.difference(self.notified_mempool_txs)
touched = await self._process_mempool(hashes)
self.notified_mempool_txs.update(new_hashes)
new_touched = {
touched_hashx for touched_hashx, txs in self.hashXs.items() if txs.intersection(new_hashes)
}
synchronized_event.set()
synchronized_event.clear()
await self.api.on_mempool(touched, new_touched, height)
duration = time.perf_counter() - start
self.mempool_process_time_metric.observe(duration)
await self.api.on_mempool(touched, height)
try:
# we wait up to `refresh_secs` but go early if a broadcast happens (which triggers wakeup event)
await asyncio.wait_for(self.wakeup.wait(), timeout=self.refresh_secs)
@@ -251,8 +228,7 @@ class MemPool:
async def _process_mempool(self, all_hashes):
# Re-sync with the new set of hashes
txs = self.txs
hashXs = self.hashXs # hashX: [tx_hash, ...]
hashXs = self.hashXs
touched = set()
# First handle txs that have disappeared
@@ -283,8 +259,8 @@ class MemPool:
# FIXME: this is not particularly efficient
while tx_map and len(tx_map) != prior_count:
prior_count = len(tx_map)
tx_map, utxo_map = self._accept_transactions(tx_map, utxo_map, touched)
tx_map, utxo_map = self._accept_transactions(tx_map, utxo_map,
touched)
if tx_map:
self.logger.info(f'{len(tx_map)} txs dropped')
@@ -295,25 +271,30 @@ class MemPool:
hex_hashes_iter = (hash_to_hex_str(hash) for hash in hashes)
raw_txs = await self.api.raw_transactions(hex_hashes_iter)
to_hashX = self.coin.hashX_from_script
deserializer = self.coin.DESERIALIZER
def deserialize_txs(): # This function is pure
to_hashX = self.coin.hashX_from_script
deserializer = self.coin.DESERIALIZER
tx_map = {}
for hash, raw_tx in zip(hashes, raw_txs):
# The daemon may have evicted the tx from its
# mempool or it may have gotten in a block
if not raw_tx:
continue
tx, tx_size = deserializer(raw_tx).read_tx_and_vsize()
# Convert the inputs and outputs into (hashX, value) pairs
# Drop generation-like inputs from MemPoolTx.prevouts
txin_pairs = tuple((txin.prev_hash, txin.prev_idx)
for txin in tx.inputs
if not txin.is_generation())
txout_pairs = tuple((to_hashX(txout.pk_script), txout.value)
for txout in tx.outputs)
tx_map[hash] = MemPoolTx(txin_pairs, None, txout_pairs,
0, tx_size)
txs = {}
for hash, raw_tx in zip(hashes, raw_txs):
# The daemon may have evicted the tx from its
# mempool or it may have gotten in a block
if not raw_tx:
continue
tx, tx_size = deserializer(raw_tx).read_tx_and_vsize()
# Convert the inputs and outputs into (hashX, value) pairs
# Drop generation-like inputs from MemPoolTx.prevouts
txin_pairs = tuple((txin.prev_hash, txin.prev_idx)
for txin in tx.inputs
if not txin.is_generation())
txout_pairs = tuple((to_hashX(txout.pk_script), txout.value)
for txout in tx.outputs)
txs[hash] = MemPoolTx(txin_pairs, None, txout_pairs,
0, tx_size)
return txs
# Thread this potentially slow operation so as not to block
tx_map = await asyncio.get_event_loop().run_in_executor(self.executor, deserialize_txs)
# Determine all prevouts not in the mempool, and fetch the
# UTXO information from the database. Failed prevout lookups
@@ -335,8 +316,8 @@ class MemPool:
async def keep_synchronized(self, synchronized_event):
"""Keep the mempool synchronized with the daemon."""
await asyncio.wait([
self._mempool_loop(synchronized_event),
# self._refresh_histogram(synchronized_event),
self._refresh_hashes(synchronized_event),
self._refresh_histogram(synchronized_event),
self._logging(synchronized_event)
])
@@ -353,7 +334,7 @@ class MemPool:
value += sum(v for h168, v in tx.out_pairs if h168 == hashX)
return value
def compact_fee_histogram(self):
async def compact_fee_histogram(self):
"""Return a compact fee histogram of the current mempool."""
return self.cached_compact_histogram
@@ -370,7 +351,7 @@ class MemPool:
result.update(tx.prevouts)
return result
def transaction_summaries(self, hashX):
async def transaction_summaries(self, hashX):
"""Return a list of MemPoolTxSummary objects for the hashX."""
result = []
for tx_hash in self.hashXs.get(hashX, ()):
@@ -393,17 +374,3 @@ class MemPool:
if hX == hashX:
utxos.append(UTXO(-1, pos, tx_hash, 0, value))
return utxos
def get_mempool_height(self, tx_hash):
# Height Progression
# -2: not broadcast
# -1: in mempool but has unconfirmed inputs
# 0: in mempool and all inputs confirmed
# +num: confirmed in a specific block (height)
if tx_hash not in self.txs:
return -2
tx = self.txs[tx_hash]
unspent_inputs = sum(1 if hash in self.txs else 0 for hash, idx in tx.prevouts)
if unspent_inputs:
return -1
return 0
+1 -1
View File
@@ -63,7 +63,7 @@ class Merkle:
raise TypeError('index must be an integer')
# This also asserts hashes is not empty
if not 0 <= index < len(hashes):
raise ValueError(f"index '{index}/{len(hashes)}' out of range")
raise ValueError('index out of range')
natural_length = self.branch_length(len(hashes))
if length is None:
length = natural_length
+1 -1
View File
@@ -319,7 +319,7 @@ class PeerManager:
# Check prior header too in case of hard fork.
check_height = min(our_height, their_height)
raw_header = self.db.raw_header(check_height)
raw_header = await self.db.raw_header(check_height)
if ptuple >= (1, 4):
ours = raw_header.hex()
message = 'blockchain.block.header'
+89
View File
@@ -0,0 +1,89 @@
import os
from aiohttp import web
from prometheus_client import Counter, Info, generate_latest as prom_generate_latest, Histogram, Gauge
from lbry import __version__ as version
from lbry.build_info import BUILD, COMMIT_HASH, DOCKER_TAG
from lbry.wallet.server import util
import lbry.wallet.server.version as wallet_server_version
NAMESPACE = "wallet_server"
CPU_COUNT = f"{os.cpu_count()}"
VERSION_INFO = Info('build', 'Wallet server build info (e.g. version, commit hash)', namespace=NAMESPACE)
VERSION_INFO.info({
'build': BUILD,
"commit": COMMIT_HASH,
"docker_tag": DOCKER_TAG,
'version': version,
"min_version": util.version_string(wallet_server_version.PROTOCOL_MIN),
"cpu_count": CPU_COUNT
})
SESSIONS_COUNT = Gauge("session_count", "Number of connected client sessions", namespace=NAMESPACE,
labelnames=("version", ))
REQUESTS_COUNT = Counter("requests_count", "Number of requests received", namespace=NAMESPACE,
labelnames=("method", "version"))
RESPONSE_TIMES = Histogram("response_time", "Response times", namespace=NAMESPACE, labelnames=("method", "version"))
NOTIFICATION_COUNT = Counter("notification", "Number of notifications sent (for subscriptions)",
namespace=NAMESPACE, labelnames=("method", "version"))
REQUEST_ERRORS_COUNT = Counter("request_error", "Number of requests that returned errors", namespace=NAMESPACE,
labelnames=("method", "version"))
SQLITE_INTERRUPT_COUNT = Counter("interrupt", "Number of interrupted queries", namespace=NAMESPACE)
SQLITE_OPERATIONAL_ERROR_COUNT = Counter(
"operational_error", "Number of queries that raised operational errors", namespace=NAMESPACE
)
SQLITE_INTERNAL_ERROR_COUNT = Counter(
"internal_error", "Number of queries raising unexpected errors", namespace=NAMESPACE
)
SQLITE_EXECUTOR_TIMES = Histogram("executor_time", "SQLite executor times", namespace=NAMESPACE)
SQLITE_PENDING_COUNT = Gauge(
"pending_queries_count", "Number of pending and running sqlite queries", namespace=NAMESPACE
)
LBRYCRD_REQUEST_TIMES = Histogram(
"lbrycrd_request", "lbrycrd requests count", namespace=NAMESPACE, labelnames=("method",)
)
LBRYCRD_PENDING_COUNT = Gauge(
"lbrycrd_pending_count", "Number of lbrycrd rpcs that are in flight", namespace=NAMESPACE, labelnames=("method",)
)
CLIENT_VERSIONS = Counter(
"clients", "Number of connections received per client version",
namespace=NAMESPACE, labelnames=("version",)
)
BLOCK_COUNT = Gauge(
"block_count", "Number of processed blocks", namespace=NAMESPACE
)
BLOCK_UPDATE_TIMES = Histogram("block_time", "Block update times", namespace=NAMESPACE)
REORG_COUNT = Gauge(
"reorg_count", "Number of reorgs", namespace=NAMESPACE
)
RESET_CONNECTIONS = Counter(
"reset_clients", "Number of reset connections by client version",
namespace=NAMESPACE, labelnames=("version",)
)
class PrometheusServer:
def __init__(self):
self.logger = util.class_logger(__name__, self.__class__.__name__)
self.runner = None
async def start(self, port: int):
prom_app = web.Application()
prom_app.router.add_get('/metrics', self.handle_metrics_get_request)
self.runner = web.AppRunner(prom_app)
await self.runner.setup()
metrics_site = web.TCPSite(self.runner, "0.0.0.0", port, shutdown_timeout=.5)
await metrics_site.start()
self.logger.info('metrics server listening on %s:%i', *metrics_site._server.sockets[0].getsockname()[:2])
async def handle_metrics_get_request(self, request: web.Request):
try:
return web.Response(
text=prom_generate_latest().decode(),
content_type='text/plain; version=0.0.4'
)
except Exception:
self.logger.exception('could not generate prometheus data')
raise
async def stop(self):
await self.runner.cleanup()
+10 -21
View File
@@ -6,7 +6,7 @@ import typing
import lbry
from lbry.wallet.server.mempool import MemPool, MemPoolAPI
from lbry.prometheus import PrometheusServer
from lbry.wallet.server.prometheus import PrometheusServer
class Notifications:
@@ -25,10 +25,9 @@ class Notifications:
def __init__(self):
self._touched_mp = {}
self._touched_bp = {}
self.notified_mempool_txs = set()
self._highest_block = -1
async def _maybe_notify(self, new_touched):
async def _maybe_notify(self):
tmp, tbp = self._touched_mp, self._touched_bp
common = set(tmp).intersection(tbp)
if common:
@@ -45,24 +44,24 @@ class Notifications:
del tmp[old]
for old in [h for h in tbp if h <= height]:
touched.update(tbp.pop(old))
await self.notify(height, touched, new_touched)
await self.notify(height, touched)
async def notify(self, height, touched, new_touched):
async def notify(self, height, touched):
pass
async def start(self, height, notify_func):
self._highest_block = height
self.notify = notify_func
await self.notify(height, set(), set())
await self.notify(height, set())
async def on_mempool(self, touched, new_touched, height):
async def on_mempool(self, touched, height):
self._touched_mp[height] = touched
await self._maybe_notify(new_touched)
await self._maybe_notify()
async def on_block(self, touched, height):
self._touched_bp[height] = touched
self._highest_block = height
await self._maybe_notify(set())
await self._maybe_notify()
class Server:
@@ -85,16 +84,12 @@ class Server:
notifications.mempool_hashes = daemon.mempool_hashes
notifications.raw_transactions = daemon.getrawtransactions
notifications.lookup_utxos = db.lookup_utxos
MemPoolAPI.register(Notifications)
self.mempool = mempool = MemPool(env.coin, notifications)
notifications.notified_mempool_txs = self.mempool.notified_mempool_txs
self.session_mgr = env.coin.SESSION_MANAGER(
env, db, bp, daemon, mempool, self.shutdown_event
)
self._indexer_task = None
async def start(self):
env = self.env
@@ -111,17 +106,11 @@ class Server:
self.cancellable_tasks.append(asyncio.ensure_future(run(*args, _flag)))
return _flag.wait()
await self.start_prometheus()
if self.env.udp_port:
await self.bp.status_server.start(
0, bytes.fromhex(self.bp.coin.GENESIS_HASH)[::-1],
self.env.host, self.env.udp_port
)
await _start_cancellable(self.bp.fetch_and_process_blocks)
await self.db.populate_header_merkle_cache()
await _start_cancellable(self.mempool.keep_synchronized)
await _start_cancellable(self.session_mgr.serve, self.notifications)
await self.start_prometheus()
async def stop(self):
for task in reversed(self.cancellable_tasks):
@@ -154,4 +143,4 @@ class Server:
async def start_prometheus(self):
if not self.prometheus_server and self.env.prometheus_port:
self.prometheus_server = PrometheusServer()
await self.prometheus_server.start("0.0.0.0", self.env.prometheus_port)
await self.prometheus_server.start(self.env.prometheus_port)
File diff suppressed because it is too large. Load diff
+6 -4
View File
@@ -77,16 +77,18 @@ class LevelDB(Storage):
import plyvel
cls.module = plyvel
def open(self, name, create, lru_cache_size=None):
mof = 10000
def open(self, name, create):
mof = 512 if self.for_sync else 128
path = os.path.join(self.db_dir, name)
# Use snappy compression (the default)
self.db = self.module.DB(path, create_if_missing=create, max_open_files=mof)
self.db = self.module.DB(path, create_if_missing=create,
max_open_files=mof)
self.close = self.db.close
self.get = self.db.get
self.put = self.db.put
self.iterator = self.db.iterator
self.write_batch = partial(self.db.write_batch, transaction=True, sync=True)
self.write_batch = partial(self.db.write_batch, transaction=True,
sync=True)
class RocksDB(Storage):
-199
View File
@@ -1,199 +0,0 @@
import asyncio
import struct
from time import perf_counter
import logging
from typing import Optional, Tuple, NamedTuple
from lbry.utils import LRUCache
# from prometheus_client import Counter
log = logging.getLogger(__name__)
_MAGIC = 1446058291 # genesis blocktime (which is actually wrong)
# ping_count_metric = Counter("ping_count", "Number of pings received", namespace='wallet_server_status')
_PAD_BYTES = b'\x00' * 64
class SPVPing(NamedTuple):
magic: int
protocol_version: int
pad_bytes: bytes
def encode(self):
return struct.pack(b'!lB64s', *self)
@staticmethod
def make(protocol_version=1) -> bytes:
return SPVPing(_MAGIC, protocol_version, _PAD_BYTES).encode()
@classmethod
def decode(cls, packet: bytes):
decoded = cls(*struct.unpack(b'!lB64s', packet[:69]))
if decoded.magic != _MAGIC:
raise ValueError("invalid magic bytes")
return decoded
class SPVPong(NamedTuple):
protocol_version: int
flags: int
height: int
tip: bytes
source_address_raw: bytes
def encode(self):
return struct.pack(b'!BBl32s4s', *self)
@staticmethod
def make(height: int, tip: bytes, flags: int, protocol_version: int = 1) -> bytes:
# note: drops the last 4 bytes so the result can be cached and have addresses added to it as needed
return SPVPong(protocol_version, flags, height, tip, b'\x00\x00\x00\x00').encode()[:38]
@classmethod
def decode(cls, packet: bytes):
return cls(*struct.unpack(b'!BBl32s4s', packet[:42]))
@property
def available(self) -> bool:
return (self.flags & 0b00000001) > 0
@property
def ip_address(self) -> str:
return ".".join(map(str, self.source_address_raw))
def __repr__(self) -> str:
return f"SPVPong(external_ip={self.ip_address}, version={self.protocol_version}, " \
f"available={'True' if self.flags & 1 > 0 else 'False'}," \
f" height={self.height}, tip={self.tip[::-1].hex()})"
class SPVServerStatusProtocol(asyncio.DatagramProtocol):
PROTOCOL_VERSION = 1
def __init__(self, height: int, tip: bytes, throttle_cache_size: int = 1024, throttle_reqs_per_sec: int = 10):
super().__init__()
self.transport: Optional[asyncio.transports.DatagramTransport] = None
self._height = height
self._tip = tip
self._flags = 0
self._cached_response = None
self.update_cached_response()
self._throttle = LRUCache(throttle_cache_size)
self._should_log = LRUCache(throttle_cache_size)
self._min_delay = 1 / throttle_reqs_per_sec
def update_cached_response(self):
self._cached_response = SPVPong.make(self._height, self._tip, self._flags, self.PROTOCOL_VERSION)
def set_unavailable(self):
self._flags &= 0b11111110
self.update_cached_response()
def set_available(self):
self._flags |= 0b00000001
self.update_cached_response()
def set_height(self, height: int, tip: bytes):
self._height, self._tip = height, tip
self.update_cached_response()
def should_throttle(self, host: str):
now = perf_counter()
last_requested = self._throttle.get(host, default=0)
self._throttle[host] = now
if now - last_requested < self._min_delay:
log_cnt = self._should_log.get(host, default=0) + 1
if log_cnt % 100 == 0:
log.warning("throttle spv status to %s", host)
self._should_log[host] = log_cnt
return True
return False
def make_pong(self, host):
return self._cached_response + bytes(int(b) for b in host.split("."))
def datagram_received(self, data: bytes, addr: Tuple[str, int]):
if self.should_throttle(addr[0]):
return
try:
SPVPing.decode(data)
except (ValueError, struct.error, AttributeError, TypeError):
# log.exception("derp")
return
self.transport.sendto(self.make_pong(addr[0]), addr)
# ping_count_metric.inc()
def connection_made(self, transport) -> None:
self.transport = transport
def connection_lost(self, exc: Optional[Exception]) -> None:
self.transport = None
def close(self):
if self.transport:
self.transport.close()
class StatusServer:
def __init__(self):
self._protocol: Optional[SPVServerStatusProtocol] = None
async def start(self, height: int, tip: bytes, interface: str, port: int):
if self.is_running:
return
loop = asyncio.get_event_loop()
self._protocol = SPVServerStatusProtocol(height, tip)
interface = interface if interface.lower() != 'localhost' else '127.0.0.1'
await loop.create_datagram_endpoint(lambda: self._protocol, (interface, port))
log.info("started udp status server on %s:%i", interface, port)
def stop(self):
if self.is_running:
self._protocol.close()
self._protocol = None
@property
def is_running(self):
return self._protocol is not None
def set_unavailable(self):
if self.is_running:
self._protocol.set_unavailable()
def set_available(self):
if self.is_running:
self._protocol.set_available()
def set_height(self, height: int, tip: bytes):
if self.is_running:
self._protocol.set_height(height, tip)
class SPVStatusClientProtocol(asyncio.DatagramProtocol):
PROTOCOL_VERSION = 1
def __init__(self, responses: asyncio.Queue):
super().__init__()
self.transport: Optional[asyncio.transports.DatagramTransport] = None
self.responses = responses
self._ping_packet = SPVPing.make(self.PROTOCOL_VERSION)
def datagram_received(self, data: bytes, addr: Tuple[str, int]):
try:
self.responses.put_nowait(((addr, perf_counter()), SPVPong.decode(data)))
except (ValueError, struct.error, AttributeError, TypeError, RuntimeError):
return
def connection_made(self, transport) -> None:
self.transport = transport
def connection_lost(self, exc: Optional[Exception]) -> None:
self.transport = None
log.info("closed udp spv server selection client")
def ping(self, server: Tuple[str, int]):
self.transport.sendto(self._ping_packet, server)
def close(self):
# log.info("close udp client")
if self.transport:
self.transport.close()
+1 -3
View File
@@ -333,13 +333,11 @@ unpack_le_uint64_from = struct_le_Q.unpack_from
unpack_be_uint16_from = struct_be_H.unpack_from
unpack_be_uint32_from = struct_be_I.unpack_from
unpack_be_uint64 = lambda x: int.from_bytes(x, byteorder='big')
pack_le_int32 = struct_le_i.pack
pack_le_int64 = struct_le_q.pack
pack_le_uint16 = struct_le_H.pack
pack_le_uint32 = struct_le_I.pack
pack_be_uint64 = lambda x: x.to_bytes(8, byteorder='big')
pack_le_uint64 = struct_le_Q.pack
pack_be_uint16 = struct_be_H.pack
pack_be_uint32 = struct_be_I.pack
pack_byte = structB.pack
+26 -83
View File
@@ -2,7 +2,6 @@ import struct
import hashlib
import logging
import typing
import asyncio
from binascii import hexlify, unhexlify
from typing import List, Iterable, Optional, Tuple
@@ -19,9 +18,7 @@ from lbry.crypto.hash import hash160, sha256
from lbry.crypto.base58 import Base58
from lbry.schema.url import normalize_name
from lbry.schema.claim import Claim
from lbry.schema.base import Signable
from lbry.schema.purchase import Purchase
from lbry.schema.support import Support
from .script import InputScript, OutputScript
from .constants import COIN, NULL_HASH32
@@ -213,7 +210,7 @@ class Output(InputOutput):
'amount', 'script', 'is_internal_transfer', 'is_spent', 'is_my_output', 'is_my_input',
'channel', 'private_key', 'meta', 'sent_supports', 'sent_tips', 'received_tips',
'purchase', 'purchased_claim', 'purchase_receipt',
'reposted_claim', 'claims', '_signable'
'reposted_claim', 'claims',
)
def __init__(self, amount: int, script: OutputScript,
@@ -241,7 +238,6 @@ class Output(InputOutput):
self.purchase_receipt: 'Output' = None # txo representing purchase receipt for this claim
self.reposted_claim: 'Output' = None # txo representing claim being reposted
self.claims: List['Output'] = None # resolved claims for collection
self._signable: Optional[Signable] = None
self.meta = {}
def update_annotations(self, annotated: 'Output'):
@@ -315,10 +311,6 @@ class Output(InputOutput):
def is_support(self) -> bool:
return self.script.is_support_claim
@property
def is_support_data(self) -> bool:
return self.script.is_support_claim_data
@property
def claim_hash(self) -> bytes:
if self.script.is_claim_name:
@@ -354,33 +346,9 @@ class Output(InputOutput):
def can_decode_claim(self):
try:
return self.claim
except Exception:
except: # pylint: disable=bare-except
return False
@property
def support(self) -> Support:
if self.is_support_data:
if not isinstance(self.script.values['support'], Support):
self.script.values['support'] = Support.from_bytes(self.script.values['support'])
return self.script.values['support']
raise ValueError('Only supports with data can be represented as Supports.')
@property
def can_decode_support(self):
try:
return self.support
except Exception:
return False
@property
def signable(self) -> Signable:
if self._signable is None:
if self.is_claim:
self._signable = self.claim
elif self.is_support_data:
self._signable = self.support
return self._signable
@property
def permanent_url(self) -> str:
if self.script.is_claim_involved:
@@ -392,22 +360,22 @@ class Output(InputOutput):
return self.private_key is not None
def get_signature_digest(self, ledger):
if self.signable.unsigned_payload:
if self.claim.unsigned_payload:
pieces = [
Base58.decode(self.get_address(ledger)),
self.signable.unsigned_payload,
self.signable.signing_channel_hash[::-1]
self.claim.unsigned_payload,
self.claim.signing_channel_hash[::-1]
]
else:
pieces = [
self.tx_ref.tx.inputs[0].txo_ref.hash,
self.signable.signing_channel_hash,
self.signable.to_message_bytes()
self.claim.signing_channel_hash,
self.claim.to_message_bytes()
]
return sha256(b''.join(pieces))
def get_encoded_signature(self):
signature = hexlify(self.signable.signature)
signature = hexlify(self.claim.signature)
r = int(signature[:int(len(signature)/2)], 16)
s = int(signature[int(len(signature)/2):], 16)
return ecdsa.util.sigencode_der(r, s, len(signature)*4)
@@ -431,23 +399,21 @@ class Output(InputOutput):
def sign(self, channel: 'Output', first_input_id=None):
self.channel = channel
self.signable.signing_channel_hash = channel.claim_hash
self.claim.signing_channel_hash = channel.claim_hash
digest = sha256(b''.join([
first_input_id or self.tx_ref.tx.inputs[0].txo_ref.hash,
self.signable.signing_channel_hash,
self.signable.to_message_bytes()
self.claim.signing_channel_hash,
self.claim.to_message_bytes()
]))
self.signable.signature = channel.private_key.sign_digest_deterministic(digest, hashfunc=hashlib.sha256)
self.claim.signature = channel.private_key.sign_digest_deterministic(digest, hashfunc=hashlib.sha256)
self.script.generate()
def clear_signature(self):
self.channel = None
self.signable.clear_signature()
self.claim.clear_signature()
async def generate_channel_private_key(self):
self.private_key = await asyncio.get_event_loop().run_in_executor(
None, ecdsa.SigningKey.generate, ecdsa.SECP256k1, None, hashlib.sha256
)
def generate_channel_private_key(self):
self.private_key = ecdsa.SigningKey.generate(curve=ecdsa.SECP256k1, hashfunc=hashlib.sha256)
self.claim.channel.public_key_bytes = self.private_key.get_verifying_key().to_der()
self.script.generate()
return self.private_key
@@ -477,14 +443,6 @@ class Output(InputOutput):
)
return cls(amount, script)
@classmethod
def pay_support_data_pubkey_hash(
cls, amount: int, claim_name: str, claim_id: str, support: Support, pubkey_hash: bytes) -> 'Output':
script = OutputScript.pay_support_data_pubkey_hash(
claim_name.encode(), unhexlify(claim_id)[::-1], support, pubkey_hash
)
return cls(amount, script)
@classmethod
def add_purchase_data(cls, purchase: Purchase) -> 'Output':
script = OutputScript.return_data(purchase)
@@ -539,7 +497,6 @@ class Transaction:
height: int = -2, position: int = -1, julian_day: int = None) -> None:
self._raw = raw
self._raw_sans_segwit = None
self._raw_outputs = None
self.is_segwit_flag = 0
self.witnesses: List[bytes] = []
self.ref = TXRefMutable(self)
@@ -601,7 +558,6 @@ class Transaction:
def _reset(self):
self._raw = None
self._raw_sans_segwit = None
self._raw_outputs = None
self.ref.reset()
@property
@@ -695,7 +651,9 @@ class Transaction:
stream.write_compact_size(len(self._inputs))
for txin in self._inputs:
txin.serialize_to(stream)
self._serialize_outputs(stream)
stream.write_compact_size(len(self._outputs))
for txout in self._outputs:
txout.serialize_to(stream)
stream.write_uint32(self.locktime)
return stream.get_bytes()
@@ -709,19 +667,13 @@ class Transaction:
txin.serialize_to(stream, txin.txo_ref.txo.script.source)
else:
txin.serialize_to(stream, b'')
self._serialize_outputs(stream)
stream.write_compact_size(len(self._outputs))
for txout in self._outputs:
txout.serialize_to(stream)
stream.write_uint32(self.locktime)
stream.write_uint32(self.signature_hash_type(1)) # signature hash type: SIGHASH_ALL
return stream.get_bytes()
def _serialize_outputs(self, stream):
if self._raw_outputs is None:
self._raw_outputs = BCDataStream()
self._raw_outputs.write_compact_size(len(self._outputs))
for txout in self._outputs:
txout.serialize_to(self._raw_outputs)
stream.write(self._raw_outputs.get_bytes())
def _deserialize(self):
if self._raw is not None:
stream = BCDataStream(self._raw)
@@ -850,7 +802,6 @@ class Transaction:
return hash_type
async def sign(self, funding_accounts: Iterable['Account']):
self._reset()
ledger, wallet = self.ensure_all_have_same_ledger_and_wallet(funding_accounts)
for i, txi in enumerate(self._inputs):
assert txi.script is not None
@@ -906,20 +857,12 @@ class Transaction:
@classmethod
def support(cls, claim_name: str, claim_id: str, amount: int, holding_address: str,
funding_accounts: List['Account'], change_account: 'Account', signing_channel: Output = None):
funding_accounts: List['Account'], change_account: 'Account'):
ledger, _ = cls.ensure_all_have_same_ledger_and_wallet(funding_accounts, change_account)
if signing_channel is not None:
support = Support()
support.emoji = '👍'
support_output = Output.pay_support_data_pubkey_hash(
amount, claim_name, claim_id, support, ledger.address_to_hash160(holding_address)
)
support_output.sign(signing_channel, b'placeholder txid:nout')
else:
support_output = Output.pay_support_pubkey_hash(
amount, claim_name, claim_id, ledger.address_to_hash160(holding_address)
)
return cls.create([], [support_output], funding_accounts, change_account, sign=False)
support_output = Output.pay_support_pubkey_hash(
amount, claim_name, claim_id, ledger.address_to_hash160(holding_address)
)
return cls.create([], [support_output], funding_accounts, change_account)
@classmethod
def purchase(cls, claim_id: str, amount: int, merchant_address: bytes,
+4 -9
View File
@@ -35,7 +35,7 @@ class TimestampedPreferences(UserDict):
def __setitem__(self, key, value):
self.data[key] = {
'value': value,
'ts': int(time.time())
'ts': time.time()
}
def __repr__(self):
@@ -145,15 +145,14 @@ class Wallet:
elif not self.is_locked:
log.warning(
"Disk encryption requested but no password available for encryption. "
"Resetting encryption preferences and saving wallet in an unencrypted state."
"Saving wallet in an unencrypted state."
)
self.preferences[ENCRYPT_ON_DISK] = False
return self.storage.write(self.to_dict())
@property
def hash(self) -> bytes:
h = sha256()
if self.is_encrypted:
if self.preferences.get(ENCRYPT_ON_DISK, False):
assert self.encryption_password is not None, \
"Encryption is enabled but no password is available, cannot generate hash."
h.update(self.encryption_password.encode())
@@ -220,11 +219,7 @@ class Wallet:
@property
def is_encrypted(self) -> bool:
# either its locked or it was unlocked using a password.
# if its set to encrypt on preferences but isnt encrypted and no password was given so far,
# then its not encrypted
return self.is_locked or (
self.preferences.get(ENCRYPT_ON_DISK, False) and self.encryption_password is not None)
return self.is_locked or self.preferences.get(ENCRYPT_ON_DISK, False)
def decrypt(self):
assert not self.is_locked, "Cannot decrypt a locked wallet, unlock first."
+177
View File
@@ -0,0 +1,177 @@
import os
import time
import textwrap
import argparse
import asyncio
import logging
from concurrent.futures.process import ProcessPoolExecutor
from lbry.wallet.server.db.reader import search_to_bytes, initializer, _get_claims, interpolate
from lbry.wallet.ledger import MainNetLedger
log = logging.getLogger(__name__)
log.addHandler(logging.StreamHandler())
log.setLevel(logging.CRITICAL)
DEFAULT_ANY_TAGS = [
'blockchain',
'news',
'learning',
'technology',
'automotive',
'economics',
'food',
'science',
'art',
'nature'
]
COMMON_AND_RARE = [
'gaming',
'ufos'
]
COMMON_AND_RARE2 = [
'city fix',
'gaming'
]
RARE_ANY_TAGS = [
'city fix',
'ufos',
]
CITY_FIX = [
'city fix'
]
MATURE_TAGS = [
'porn',
'nsfw',
'mature',
'xxx'
]
ORDER_BY = [
[
"trending_global",
"trending_mixed",
],
[
"release_time"
],
[
"effective_amount"
]
]
def get_args(limit=20):
args = []
any_tags_combinations = [DEFAULT_ANY_TAGS, COMMON_AND_RARE, RARE_ANY_TAGS, COMMON_AND_RARE2, CITY_FIX, []]
not_tags_combinations = [MATURE_TAGS, []]
for no_fee in [False, True]:
for claim_type in [None, 'stream', 'channel']:
for no_totals in [True]:
for offset in [0, 100]:
for any_tags in any_tags_combinations:
for not_tags in not_tags_combinations:
for order_by in ORDER_BY:
kw = {
'order_by': order_by,
'offset': offset,
'limit': limit,
'no_totals': no_totals
}
if not_tags:
kw['not_tags'] = not_tags
if any_tags:
kw['any_tags'] = any_tags
if claim_type:
kw['claim_type'] = claim_type
if no_fee:
kw['fee_amount'] = 0
args.append(kw)
print(f"-- Trying {len(args)} argument combinations")
return args
def _search(kwargs):
start = time.perf_counter()
error = None
try:
search_to_bytes(kwargs)
except Exception as err:
error = str(err)
return time.perf_counter() - start, kwargs, error
async def search(executor, kwargs):
return await asyncio.get_running_loop().run_in_executor(
executor, _search, kwargs
)
async def main(db_path, max_query_time):
args = dict(initializer=initializer, initargs=(log, db_path, MainNetLedger, 0.25))
workers = max(os.cpu_count(), 4)
log.info(f"using {workers} reader processes")
query_executor = ProcessPoolExecutor(workers, **args)
tasks = [search(query_executor, constraints) for constraints in get_args()]
try:
results = await asyncio.gather(*tasks)
query_times = [
{
'sql': interpolate(*_get_claims("""
claimtrie.claim_hash as is_controlling,
claimtrie.last_take_over_height,
claim.claim_hash, claim.txo_hash,
claim.claims_in_channel,
claim.height, claim.creation_height,
claim.activation_height, claim.expiration_height,
claim.effective_amount, claim.support_amount,
claim.trending_group, claim.trending_mixed,
claim.trending_local, claim.trending_global,
claim.short_url, claim.canonical_url,
claim.channel_hash, channel.txo_hash AS channel_txo_hash,
channel.height AS channel_height, claim.signature_valid
""", **constraints)),
'duration': ts,
'error': error
}
for ts, constraints, error in results
]
errored = [query_info for query_info in query_times if query_info['error']]
errors = {str(query_info['error']): [] for query_info in errored}
for error in errored:
errors[str(error['error'])].append(error['sql'])
slow = [
query_info for query_info in query_times
if not query_info['error'] and query_info['duration'] > (max_query_time / 2.0)
]
fast = [
query_info for query_info in query_times
if not query_info['error'] and query_info['duration'] <= (max_query_time / 2.0)
]
print(f"-- {len(fast)} queries were fast")
slow.sort(key=lambda query_info: query_info['duration'], reverse=True)
print(f"-- Failing queries:")
for error in errors:
print(f"-- Failure: \"{error}\"")
for failing_query in errors[error]:
print(f"{textwrap.dedent(failing_query)};\n")
print()
print(f"-- Slow queries:")
for slow_query in slow:
print(f"-- Query took {slow_query['duration']}\n{textwrap.dedent(slow_query['sql'])};\n")
finally:
query_executor.shutdown()
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument('--db_path', dest='db_path', default=os.path.expanduser('~/claims.db'), type=str)
parser.add_argument('--max_time', dest='max_time', default=0.25, type=float)
args = parser.parse_args()
db_path = args.db_path
max_query_time = args.max_time
asyncio.run(main(db_path, max_query_time))
-1
View File
@@ -1,7 +1,6 @@
import curses
import time
import asyncio
import lbry.wallet
from lbry.conf import Config
from lbry.extras.daemon.client import daemon_rpc
-1
View File
@@ -3,7 +3,6 @@ import os
import asyncio
import socket
import ipaddress
import lbry.wallet
from lbry.conf import Config
from lbry.extras.daemon.storage import SQLiteStorage
from lbry.blob.blob_manager import BlobManager
+2 -29
View File
@@ -13,7 +13,6 @@ from lbry.extras.cli import set_kwargs, get_argument_parser
from lbry.extras.daemon.daemon import (
Daemon, jsonrpc_dumps_pretty, encode_pagination_doc
)
from tests.integration.other.test_comment_commands import MockedCommentServer
from lbry.extras.daemon.json_response_encoder import (
encode_tx_doc, encode_txo_doc, encode_account_doc, encode_file_doc,
encode_wallet_doc
@@ -26,8 +25,7 @@ RETURN_DOCS = {
'File': encode_file_doc(),
'Transaction': encode_tx_doc(),
'Output': encode_txo_doc(),
'Address': 'an address in base58',
'Dict': 'glorious data in dictionary',
'Address': 'an address in base58'
}
@@ -67,10 +65,6 @@ class Examples(CommandTestCase):
async def asyncSetUp(self):
await super().asyncSetUp()
self.daemon.conf.comment_server = 'http://localhost:2903/api'
self.comment_server = MockedCommentServer(2903)
await self.comment_server.start()
self.addCleanup(self.comment_server.stop)
self.recorder = ExampleRecorder(self)
async def play(self):
@@ -342,8 +336,7 @@ class Examples(CommandTestCase):
'Use the parent_id param to make replies',
'comment', 'create',
'--comment="I have photographic evidence confirming Sasquatch exists"',
f'--channel_name=@channel', f'--parent_id={comment["comment_id"]}',
f'--claim_id={stream_id}'
f'--channel_name=@channel', f'--parent_id={comment["comment_id"]}'
)
await r(
@@ -365,26 +358,6 @@ class Examples(CommandTestCase):
await self.daemon.jsonrpc_comment_abandon(reply['comment_id'])
# collections
collection = await r(
'Create a collection of one stream',
'collection', 'create',
'--name=tom', '--bid=1.0',
f'--channel_id={channel_id}',
f'--claims={stream_id}'
)
await self.on_transaction_dict(collection)
await self.generate(1)
await self.on_transaction_dict(collection)
await r(
'List collections',
'collection', 'list',
'--resolve', '--resolve_claims=1',
)
# files
file_list_result = (await r(
+62
View File
@@ -0,0 +1,62 @@
import uvloop, asyncio, time, sys, logging
from concurrent.futures import ProcessPoolExecutor
from lbry.wallet.server.db import reader
from lbry.wallet.server.metrics import calculate_avg_percentiles
db_path = '../../../lbryconf/wallet-server/claims.db'
default_query_timout = 0.25
log = logging.getLogger(__name__)
log.addHandler(logging.StreamHandler())
async def run_times(executor, iterations, show=True):
start = time.perf_counter()
timings = await asyncio.gather(*(asyncio.get_running_loop().run_in_executor(
executor, reader.search_to_bytes, {
'no_totals': True,
'offset': 0,
'limit': 20,
'any_tags': [
'ufos', 'city fix'
],
'not_tags': [
'porn', 'mature', 'xxx', 'nsfw'
],
'order_by': [
'release_time'
]
}
) for _ in range(iterations)))
timings = [r[1]['execute_query'][0]['total'] for r in timings]
total = int((time.perf_counter() - start) * 100)
if show:
avg = sum(timings)/len(timings)
print(f"{iterations:4}: {total}ms total concurrent, {len(timings)*avg*1000:.3f}s total sequential (avg*runs)")
print(f" {total/len(timings):.1f}ms/query concurrent (total/runs)")
print(f" {avg:.1f}ms/query actual average (sum(queries)/runs)")
stats = calculate_avg_percentiles(timings)
print(f" min: {stats[1]}, 5%: {stats[2]}, 25%: {stats[3]}, 50%: {stats[4]}, 75%: {stats[5]}, 95%: {stats[6]}, max: {stats[7]}")
sys.stdout.write(' sample:')
for i, t in zip(range(10), timings[::-1]):
sys.stdout.write(f' {t}ms')
print(' ...\n' if len(timings) > 10 else '\n')
async def main():
executor = ProcessPoolExecutor(
4, initializer=reader.initializer, initargs=(log, db_path, 'mainnet', 1.0, True)
)
#await run_times(executor, 4, show=False)
#await run_times(executor, 1)
await run_times(executor, 2**3)
await run_times(executor, 2**5)
await run_times(executor, 2**7)
#await run_times(executor, 2**9)
#await run_times(executor, 2**11)
#await run_times(executor, 2**13)
executor.shutdown(True)
if __name__ == '__main__':
uvloop.install()
asyncio.run(main())
+1 -3
View File
@@ -6,7 +6,7 @@ source =
lbry
.tox/*/lib/python*/site-packages/lbry
[cryptography.*,coincurve.*,pbkdf2, libtorrent]
[cryptography.*,coincurve.*,pbkdf2]
ignore_missing_imports = True
[pylint]
@@ -18,7 +18,6 @@ max-line-length=120
good-names=T,t,n,i,j,k,x,y,s,f,d,h,c,e,op,db,tx,io,cachedproperty,log,id,r,iv,ts,l
valid-metaclass-classmethod-first-arg=mcs
disable=
c-extension-no-member,
fixme,
broad-except,
no-else-return,
@@ -35,7 +34,6 @@ disable=
too-many-statements,
too-many-nested-blocks,
too-many-public-methods,
too-many-return-statements,
too-many-instance-attributes,
protected-access,
unused-argument
+4 -6
View File
@@ -28,14 +28,13 @@ setup(
entry_points={
'console_scripts': [
'lbrynet=lbry.extras.cli:main',
'lbry-hub=lbry.wallet.server.cli:main',
'torba-server=lbry.wallet.server.cli:main',
'orchstr8=lbry.wallet.orchstr8.cli:main',
'lbry-hub-elastic-sync=lbry.wallet.server.db.elasticsearch.sync:run_elastic_sync'
],
},
install_requires=[
'aiohttp==3.5.4',
'aioupnp==0.0.18',
'aioupnp==0.0.17',
'appdirs==1.4.3',
'certifi>=2018.11.29',
'colorama==0.3.7',
@@ -47,15 +46,14 @@ setup(
'msgpack==0.6.1',
'prometheus_client==0.7.1',
'ecdsa==0.13.3',
'pyyaml==5.3.1',
'pyyaml==4.2b1',
'docopt==0.6.2',
'hachoir',
'multidict==4.6.1',
'coincurve==11.0.0',
'pbkdf2==1.3',
'attrs==18.2.0',
'pylru==1.1.0',
'elasticsearch==7.10.1'
'pylru==1.1.0'
] + PLYVEL,
classifiers=[
'Framework :: AsyncIO',
@@ -2,6 +2,7 @@ import logging
import asyncio
from binascii import hexlify
from lbry.testcase import CommandTestCase
from lbry.wallet.server.prometheus import REORG_COUNT
class BlockchainReorganizationTests(CommandTestCase):
@@ -9,39 +10,23 @@ class BlockchainReorganizationTests(CommandTestCase):
VERBOSITY = logging.WARN
async def assertBlockHash(self, height):
bp = self.conductor.spv_node.server.bp
def get_txids():
return [
bp.db.fs_tx_hash(tx_num)[0][::-1].hex()
for tx_num in range(bp.db.tx_counts[height - 1], bp.db.tx_counts[height])
]
block_hash = await self.blockchain.get_block_hash(height)
self.assertEqual(block_hash, (await self.ledger.headers.hash(height)).decode())
self.assertEqual(block_hash, (await bp.db.fs_block_hashes(height, 1))[0][::-1].hex())
txids = await asyncio.get_event_loop().run_in_executor(bp.db.executor, get_txids)
txs = await bp.db.fs_transactions(txids)
block_txs = (await bp.daemon.deserialised_block(block_hash))['tx']
self.assertSetEqual(set(block_txs), set(txs.keys()), msg='leveldb/lbrycrd is missing transactions')
self.assertListEqual(block_txs, list(txs.keys()), msg='leveldb/lbrycrd transactions are of order')
self.assertEqual(
(await self.ledger.headers.hash(height)).decode(),
await self.blockchain.get_block_hash(height)
)
async def test_reorg(self):
bp = self.conductor.spv_node.server.bp
bp.reorg_count_metric.set(0)
REORG_COUNT.set(0)
# invalidate current block, move forward 2
height = 206
self.assertEqual(self.ledger.headers.height, height)
await self.assertBlockHash(height)
self.assertEqual(self.ledger.headers.height, 206)
await self.assertBlockHash(206)
await self.blockchain.invalidate_block((await self.ledger.headers.hash(206)).decode())
await self.blockchain.generate(2)
await self.ledger.on_header.where(lambda e: e.height == 207)
self.assertEqual(self.ledger.headers.height, 207)
await self.assertBlockHash(206)
await self.assertBlockHash(207)
self.assertEqual(1, bp.reorg_count_metric._samples()[0][2])
self.assertEqual(1, REORG_COUNT._samples()[0][2])
# invalidate current block, move forward 3
await self.blockchain.invalidate_block((await self.ledger.headers.hash(206)).decode())
@@ -51,12 +36,7 @@ class BlockchainReorganizationTests(CommandTestCase):
await self.assertBlockHash(206)
await self.assertBlockHash(207)
await self.assertBlockHash(208)
self.assertEqual(2, bp.reorg_count_metric._samples()[0][2])
await self.blockchain.generate(3)
await self.ledger.on_header.where(lambda e: e.height == 211)
await self.assertBlockHash(209)
await self.assertBlockHash(210)
await self.assertBlockHash(211)
self.assertEqual(2, REORG_COUNT._samples()[0][2])
async def test_reorg_change_claim_height(self):
# sanity check
@@ -71,8 +51,6 @@ class BlockchainReorganizationTests(CommandTestCase):
# create a claim and verify it's returned by claim_search
self.assertEqual(self.ledger.headers.height, 207)
await self.assertBlockHash(207)
broadcast_tx = await self.daemon.jsonrpc_stream_create(
'hovercraft', '1.0', file_path=self.create_upload_file(data=b'hi!')
)
@@ -80,8 +58,6 @@ class BlockchainReorganizationTests(CommandTestCase):
await self.generate(1)
await self.ledger.wait(broadcast_tx, self.blockchain.block_expected)
self.assertEqual(self.ledger.headers.height, 208)
await self.assertBlockHash(208)
txos, _, _, _ = await self.ledger.claim_search([], name='hovercraft')
self.assertEqual(1, len(txos))
txo = txos[0]
@@ -99,21 +75,29 @@ class BlockchainReorganizationTests(CommandTestCase):
await self.blockchain.clear_mempool()
await self.blockchain.generate(2)
# wait for the client to catch up and verify the reorg
await asyncio.wait_for(self.on_header(209), 3.0)
await self.assertBlockHash(207)
await self.assertBlockHash(208)
await self.assertBlockHash(209)
# verify the claim was dropped from block 208 as returned by lbrycrdd
reorg_block_hash = await self.blockchain.get_block_hash(208)
self.assertNotEqual(invalidated_block_hash, reorg_block_hash)
block_207 = await self.blockchain.get_block(reorg_block_hash)
self.assertNotIn(txo.tx_ref.id, block_207['tx'])
# wait for the client to catch up and verify the reorg
await asyncio.wait_for(self.on_header(209), 3.0)
await self.assertBlockHash(207)
await self.assertBlockHash(208)
await self.assertBlockHash(209)
client_reorg_block_hash = (await self.ledger.headers.hash(208)).decode()
self.assertEqual(client_reorg_block_hash, reorg_block_hash)
# verify the dropped claim is no longer returned by claim search
txos, _, _, _ = await self.ledger.claim_search([], name='hovercraft')
self.assertListEqual(txos, [])
# verify the claim published a block earlier wasn't also reverted
txos, _, _, _ = await self.ledger.claim_search([], name='still-valid')
self.assertEqual(1, len(txos))
self.assertEqual(207, txos[0].tx_ref.height)
# broadcast the claim in a different block
new_txid = await self.blockchain.sendrawtransaction(hexlify(broadcast_tx.raw).decode())
self.assertEqual(broadcast_tx.id, new_txid)
@@ -3,15 +3,12 @@ import tempfile
import logging
import asyncio
from binascii import unhexlify
from unittest import skip
from urllib.request import urlopen
from lbry.error import InsufficientFundsError
from lbry.extras.daemon.comment_client import verify
from lbry.extras.daemon.daemon import DEFAULT_PAGE_SIZE
from lbry.testcase import CommandTestCase
from lbry.wallet.orchstr8.node import SPVNode
from lbry.wallet.transaction import Transaction
from lbry.wallet.util import satoshis_to_coins as lbc
@@ -74,11 +71,9 @@ class ClaimSearchCommand(ClaimTestCase):
for claim, result in zip(claims, results):
self.assertEqual(
(claim['txid'], self.get_claim_id(claim)),
(result['txid'], result['claim_id']),
f"{claim['outputs'][0]['name']} != {result['name']}"
(result['txid'], result['claim_id'])
)
@skip("doesnt happen on ES...?")
async def test_disconnect_on_memory_error(self):
claim_ids = [
'0000000000000000000000000000000000000000',
@@ -94,22 +89,10 @@ class ClaimSearchCommand(ClaimTestCase):
# 23829 claim ids makes the request just large enough
claim_ids = [
'0000000000000000000000000000000000000000',
] * 33829
] * 23829
with self.assertRaises(ConnectionResetError):
await self.claim_search(claim_ids=claim_ids)
async def test_claim_search_as_reader_server(self):
node2 = SPVNode(self.conductor.spv_module, node_number=2)
current_prefix = self.conductor.spv_node.server.bp.env.es_index_prefix
await node2.start(self.blockchain, extraconf={'ES_MODE': 'reader', 'ES_INDEX_PREFIX': current_prefix})
self.addCleanup(node2.stop)
self.ledger.network.config['default_servers'] = [(node2.hostname, node2.port)]
await self.ledger.stop()
await self.ledger.start()
channel2 = await self.channel_create('@abc', '0.1', allow_duplicate_name=True)
await asyncio.sleep(1) # fixme: find a way to block on the writer
await self.assertFindsClaims([channel2], name='@abc')
async def test_basic_claim_search(self):
await self.create_channel()
channel_txo = self.channel['outputs'][0]
@@ -150,7 +133,6 @@ class ClaimSearchCommand(ClaimTestCase):
claims = [three, two, signed]
await self.assertFindsClaims(claims, channel_ids=[self.channel_id])
await self.assertFindsClaims(claims, channel=f"@abc#{self.channel_id}")
await self.assertFindsClaims([], channel=f"@inexistent")
await self.assertFindsClaims([three, two, signed2, signed], channel_ids=[channel_id2, self.channel_id])
await self.channel_abandon(claim_id=self.channel_id)
await self.assertFindsClaims([], channel=f"@abc#{self.channel_id}", valid_channel_signature=True)
@@ -174,20 +156,6 @@ class ClaimSearchCommand(ClaimTestCase):
# abandoned stream won't show up for streams in channel search
await self.stream_abandon(txid=signed2['txid'], nout=0)
await self.assertFindsClaims([], channel_ids=[channel_id2])
# resolve by claim ids
await self.assertFindsClaims([three, two], claim_ids=[self.get_claim_id(three), self.get_claim_id(two)])
await self.assertFindsClaims([three], claim_id=self.get_claim_id(three))
await self.assertFindsClaims([three], claim_id=self.get_claim_id(three), text='*')
async def test_source_filter(self):
channel = await self.channel_create('@abc')
no_source = await self.stream_create('no-source', data=None)
normal = await self.stream_create('normal', data=b'normal')
normal_repost = await self.stream_repost(self.get_claim_id(normal), 'normal-repost')
no_source_repost = await self.stream_repost(self.get_claim_id(no_source), 'no-source-repost')
await self.assertFindsClaims([no_source_repost, no_source, channel], has_no_source=True)
await self.assertFindsClaims([normal_repost, normal, channel], has_source=True)
await self.assertFindsClaims([no_source_repost, normal_repost, normal, no_source, channel])
async def test_pagination(self):
await self.create_channel()
@@ -294,19 +262,6 @@ class ClaimSearchCommand(ClaimTestCase):
await self.assertFindsClaims([claim4, claim3, claim2], fee_amount='<1.0', fee_currency='lbc')
await self.assertFindsClaims([claim3], fee_amount='0.5', fee_currency='lbc')
await self.assertFindsClaims([claim5], fee_currency='usd')
async def test_search_by_language(self):
claim1 = await self.stream_create('claim1', fee_amount='1.0', fee_currency='lbc')
claim2 = await self.stream_create('claim2', fee_amount='0.9', fee_currency='lbc')
claim3 = await self.stream_create('claim3', fee_amount='0.5', fee_currency='lbc', languages='en')
claim4 = await self.stream_create('claim4', fee_amount='0.1', fee_currency='lbc', languages='en')
claim5 = await self.stream_create('claim5', fee_amount='1.0', fee_currency='usd', languages='es')
await self.assertFindsClaims([claim4, claim3], any_languages=['en'])
await self.assertFindsClaims([claim2, claim1], any_languages=['none'])
await self.assertFindsClaims([claim4, claim3, claim2, claim1], any_languages=['none', 'en'])
await self.assertFindsClaims([claim5], any_languages=['es'])
await self.assertFindsClaims([claim5, claim4, claim3], any_languages=['en', 'es'])
await self.assertFindsClaims([], fee_currency='foo')
async def test_search_by_channel(self):
@@ -364,25 +319,6 @@ class ClaimSearchCommand(ClaimTestCase):
not_channel_ids=[chan2_id], has_channel_signature=True, valid_channel_signature=True)
await match([], not_channel_ids=[chan1_id, chan2_id], has_channel_signature=True, valid_channel_signature=True)
async def test_limit_claims_per_channel(self):
match = self.assertFindsClaims
chan1_id = self.get_claim_id(await self.channel_create('@chan1'))
chan2_id = self.get_claim_id(await self.channel_create('@chan2'))
claim1 = await self.stream_create('claim1')
claim2 = await self.stream_create('claim2', channel_id=chan1_id)
claim3 = await self.stream_create('claim3', channel_id=chan1_id)
claim4 = await self.stream_create('claim4', channel_id=chan1_id)
claim5 = await self.stream_create('claim5', channel_id=chan2_id)
claim6 = await self.stream_create('claim6', channel_id=chan2_id)
await match(
[claim6, claim5, claim4, claim3, claim1],
limit_claims_per_channel=2, claim_type='stream'
)
await match(
[claim6, claim5, claim4, claim3, claim2, claim1],
limit_claims_per_channel=3, claim_type='stream'
)
async def test_claim_type_and_media_type_search(self):
# create an invalid/unknown claim
address = await self.account.receiving.get_or_create_usable_address()
@@ -396,7 +332,6 @@ class ClaimSearchCommand(ClaimTestCase):
video = await self.stream_create('chrome', file_path=self.video_file_name)
image = await self.stream_create('blank-image', data=self.image_data, suffix='.png')
repost = await self.stream_repost(self.get_claim_id(image))
collection = await self.collection_create('a-collection', claims=[self.get_claim_id(video)])
channel = await self.channel_create()
unknown = self.sout(tx)
@@ -404,7 +339,6 @@ class ClaimSearchCommand(ClaimTestCase):
await self.assertFindsClaims([image, video, octet, unknown], claim_type='stream')
await self.assertFindsClaims([channel], claim_type='channel')
await self.assertFindsClaims([repost], claim_type='repost')
await self.assertFindsClaims([collection], claim_type='collection')
# stream_type
await self.assertFindsClaims([octet, unknown], stream_types=['binary'])
@@ -455,11 +389,10 @@ class ClaimSearchCommand(ClaimTestCase):
await self.assertFindsClaims([claim2], text='autobiography')
await self.assertFindsClaims([claim3], text='history')
await self.assertFindsClaims([claim4], text='conspiracy')
await self.assertFindsClaims([], text='conspiracy+history')
await self.assertFindsClaims([claim4, claim3], text='conspiracy|history')
await self.assertFindsClaims([claim1, claim4, claim2, claim3], text='documentary', order_by=[])
# todo: check why claim1 and claim2 order changed. used to be ...claim1, claim2...
await self.assertFindsClaims([claim4, claim2, claim1, claim3], text='satoshi', order_by=[])
await self.assertFindsClaims([], text='conspiracy AND history')
await self.assertFindsClaims([claim4, claim3], text='conspiracy OR history')
await self.assertFindsClaims([claim1, claim4, claim2, claim3], text='documentary')
await self.assertFindsClaims([claim4, claim1, claim2, claim3], text='satoshi')
claim2 = await self.stream_update(
self.get_claim_id(claim2), clear_tags=True, tags=['cloud'],
@@ -495,47 +428,6 @@ class TransactionCommands(ClaimTestCase):
class TransactionOutputCommands(ClaimTestCase):
async def test_txo_list_resolve_supports(self):
channel = self.get_claim_id(await self.channel_create('@identity'))
stream = self.get_claim_id(await self.stream_create())
support = await self.support_create(stream, channel_id=channel)
r, = await self.txo_list(type='support')
self.assertEqual(r['txid'], support['txid'])
self.assertNotIn('name', r['signing_channel'])
r, = await self.txo_list(type='support', resolve=True)
self.assertIn('name', r['signing_channel'])
self.assertEqual(r['signing_channel']['name'], '@identity')
async def test_txo_list_by_channel_filtering(self):
channel_foo = self.get_claim_id(await self.channel_create('@foo'))
channel_bar = self.get_claim_id(await self.channel_create('@bar'))
stream_a = self.get_claim_id(await self.stream_create('a', channel_id=channel_foo))
stream_b = self.get_claim_id(await self.stream_create('b', channel_id=channel_foo))
stream_c = self.get_claim_id(await self.stream_create('c', channel_id=channel_bar))
stream_d = self.get_claim_id(await self.stream_create('d'))
support_c = await self.support_create(stream_c, '0.3', channel_id=channel_foo)
support_d = await self.support_create(stream_d, '0.3', channel_id=channel_bar)
r = await self.txo_list(type='stream')
self.assertEqual({stream_a, stream_b, stream_c, stream_d}, {c['claim_id'] for c in r})
r = await self.txo_list(type='stream', channel_id=channel_foo)
self.assertEqual({stream_a, stream_b}, {c['claim_id'] for c in r})
r = await self.txo_list(type='support', channel_id=channel_foo)
self.assertEqual({support_c['txid']}, {s['txid'] for s in r})
r = await self.txo_list(type='support', channel_id=channel_bar)
self.assertEqual({support_d['txid']}, {s['txid'] for s in r})
r = await self.txo_list(type='stream', channel_id=[channel_foo, channel_bar])
self.assertEqual({stream_a, stream_b, stream_c}, {c['claim_id'] for c in r})
r = await self.txo_list(type='stream', not_channel_id=channel_foo)
self.assertEqual({stream_c, stream_d}, {c['claim_id'] for c in r})
r = await self.txo_list(type='stream', not_channel_id=[channel_foo, channel_bar])
self.assertEqual({stream_d}, {c['claim_id'] for c in r})
async def test_txo_list_and_sum_filtering(self):
channel_id = self.get_claim_id(await self.channel_create())
self.assertEqual('1.0', lbc(await self.txo_sum(type='channel', is_not_spent=True)))
@@ -769,15 +661,12 @@ class ClaimCommands(ClaimTestCase):
channel_id = self.get_claim_id(await self.channel_create())
stream_id = self.get_claim_id(await self.stream_create())
await self.stream_update(stream_id, title='foo')
# type filtering
r = await self.claim_list(claim_type='channel')
self.assertEqual(1, len(r))
self.assertEqual('channel', r[0]['value_type'])
# catch a bug where cli sends is_spent=False by default
r = await self.claim_list(claim_type='stream', is_spent=False)
r = await self.claim_list(claim_type='stream')
self.assertEqual(1, len(r))
self.assertEqual('stream', r[0]['value_type'])
@@ -1080,18 +969,6 @@ class ChannelCommands(CommandTestCase):
self.assertItemCount(await self.daemon.jsonrpc_channel_list(account_id=self.account.id), 2)
self.assertItemCount(await self.daemon.jsonrpc_channel_list(account_id=account2_id), 1)
async def test_sign_hex_encoded_data(self):
data_to_sign = "CAFEBABE"
# claim new name
await self.channel_create('@someotherchan')
channel_tx = await self.daemon.jsonrpc_channel_create('@signer', '0.1')
await self.confirm_tx(channel_tx.id)
channel = channel_tx.outputs[0]
signature1 = await self.out(self.daemon.jsonrpc_channel_sign(channel_name='@signer', hexdata=data_to_sign))
signature2 = await self.out(self.daemon.jsonrpc_channel_sign(channel_id=channel.claim_id, hexdata=data_to_sign))
self.assertTrue(verify(channel, unhexlify(data_to_sign), signature1))
self.assertTrue(verify(channel, unhexlify(data_to_sign), signature2))
async def test_channel_export_import_before_sending_channel(self):
# export
tx = await self.channel_create('@foo', '1.0')
@@ -1370,11 +1247,6 @@ class StreamCommands(ClaimTestCase):
self.assertEqual(1, blocked['channels'][0]['blocked'])
self.assertTrue(blocked['channels'][0]['channel']['short_url'].startswith('lbry://@filtering#'))
# same search, but details omitted by 'no_totals'
last_result = result
result = await self.out(self.daemon.jsonrpc_claim_search(name='bad_content', no_totals=True))
self.assertEqual(result['items'], last_result['items'])
# search inside channel containing filtered content
result = await self.out(self.daemon.jsonrpc_claim_search(channel='@some_channel'))
filtered = result['blocked']
@@ -1384,11 +1256,6 @@ class StreamCommands(ClaimTestCase):
self.assertEqual(1, filtered['channels'][0]['blocked'])
self.assertTrue(filtered['channels'][0]['channel']['short_url'].startswith('lbry://@filtering#'))
# same search, but details omitted by 'no_totals'
last_result = result
result = await self.out(self.daemon.jsonrpc_claim_search(channel='@some_channel', no_totals=True))
self.assertEqual(result['items'], last_result['items'])
# content was filtered by not_tag before censoring
result = await self.out(self.daemon.jsonrpc_claim_search(channel='@some_channel', not_tags=["good", "bad"]))
self.assertEqual(0, len(result['items']))
@@ -1442,13 +1309,6 @@ class StreamCommands(ClaimTestCase):
self.assertEqual(3, filtered['channels'][0]['blocked'])
self.assertTrue(filtered['channels'][0]['channel']['short_url'].startswith('lbry://@filtering#'))
# same search, but details omitted by 'no_totals'
last_result = result
result = await self.out(
self.daemon.jsonrpc_claim_search(any_tags=['bad-stuff'], order_by=['height'], no_totals=True)
)
self.assertEqual(result['items'], last_result['items'])
# filtered channel should still resolve
result = await self.resolve('lbry://@bad_channel')
self.assertEqual(bad_channel_id, result['claim_id'])
@@ -1893,6 +1753,9 @@ class StreamCommands(ClaimTestCase):
with self.assertRaisesRegex(Exception, "'bid' is a required argument for new publishes."):
await self.daemon.jsonrpc_publish('foo')
with self.assertRaisesRegex(Exception, "'file_path' is a required argument for new publishes."):
await self.daemon.jsonrpc_publish('foo', bid='1.0')
# successfully create stream
with tempfile.NamedTemporaryFile() as file:
file.write(b'hi')
@@ -1935,38 +1798,6 @@ class StreamCommands(ClaimTestCase):
self.assertEqual(claim['value']['languages'], ['uk-UA'])
self.assertEqual(claim['value']['tags'], ['anime'])
# publish a stream with no source
tx5 = await self.publish(
'future-release', bid='0.1', languages='uk-UA', tags=['Anime', 'anime ']
)
self.assertItemCount(await self.daemon.jsonrpc_file_list(), 2)
claim = await self.resolve('lbry://future-release')
self.assertEqual(claim['txid'], tx5['outputs'][0]['txid'])
self.assertNotIn('signing_channel', claim)
self.assertEqual(claim['value']['languages'], ['uk-UA'])
self.assertEqual(claim['value']['tags'], ['anime'])
self.assertNotIn('source', claim['value'])
# change metadata before the release
await self.publish(
'future-release', bid='0.1', tags=['Anime', 'anime ', 'psy-trance'], title='Psy will be over 9000!!!'
)
self.assertItemCount(await self.daemon.jsonrpc_file_list(), 2)
claim = await self.resolve('lbry://future-release')
self.assertEqual(claim['value']['tags'], ['anime', 'psy-trance'])
self.assertEqual(claim['value']['title'], 'Psy will be over 9000!!!')
self.assertNotIn('source', claim['value'])
# update the stream to have a source
with tempfile.NamedTemporaryFile() as file:
file.write(b'hi')
file.flush()
tx6 = await self.publish('future-release', file_path=file.name, tags=['something-else'])
claim = await self.resolve('lbry://future-release')
self.assertEqual(claim['txid'], tx6['outputs'][0]['txid'])
self.assertEqual(claim['value']['tags'], ['something-else'])
self.assertIn('source', claim['value'])
class SupportCommands(CommandTestCase):
@@ -1994,8 +1825,7 @@ class SupportCommands(CommandTestCase):
# send a tip to the claim using account2
tip = await self.out(
self.daemon.jsonrpc_support_create(
claim_id, '1.0', True, account_id=account2.id, wallet_id='wallet2',
funding_account_ids=[account2.id])
claim_id, '1.0', True, account2.id, 'wallet2', funding_account_ids=[account2.id])
)
await self.confirm_tx(tip['txid'])
@@ -2026,8 +1856,7 @@ class SupportCommands(CommandTestCase):
# send a support to the claim using account2
support = await self.out(
self.daemon.jsonrpc_support_create(
claim_id, '2.0', False, account_id=account2.id, wallet_id='wallet2',
funding_account_ids=[account2.id])
claim_id, '2.0', False, account2.id, 'wallet2', funding_account_ids=[account2.id])
)
await self.confirm_tx(support['txid'])
@@ -2054,16 +1883,6 @@ class SupportCommands(CommandTestCase):
self.assertTrue(txs[1]['support_info'][0]['is_tip'])
self.assertTrue(txs[1]['support_info'][0]['is_spent'])
async def test_signed_supports_with_no_change_txo_regression(self):
# reproduces a bug where transactions did not get properly signed
# if there was no change and just a single output
# lbrycrd returned 'the transaction was rejected by network rules.'
channel_id = self.get_claim_id(await self.channel_create())
stream_id = self.get_claim_id(await self.stream_create())
tx = await self.support_create(stream_id, '7.967598', channel_id=channel_id)
self.assertEqual(len(tx['outputs']), 1) # must be one to reproduce bug
self.assertTrue(tx['outputs'][0]['is_channel_signature_valid'])
class CollectionCommands(CommandTestCase):
@@ -2107,10 +1926,6 @@ class CollectionCommands(CommandTestCase):
await self.collection_update(claim_id, clear_claims=True, claims=claim_ids[:2])
collections = await self.out(self.daemon.jsonrpc_collection_list())
self.assertEquals(len(collections['items']), 2)
self.assertNotIn('canonical_url', collections['items'][0])
resolved_collections = await self.out(self.daemon.jsonrpc_collection_list(resolve=True))
self.assertIn('canonical_url', resolved_collections['items'][0])
await self.collection_abandon(claim_id)
self.assertItemCount(await self.daemon.jsonrpc_collection_list(), 1)
Loaded 100 of 138 files, more files were not shown because too many files have changed in this diff. Show more