lbry-sdk/lbrynet/blob_exchange/downloader.py

112 lines
5 KiB
Python
Raw Normal View History

2019-01-22 18:52:32 +01:00
import asyncio
import typing
import logging
from lbrynet.utils import drain_tasks
from lbrynet.blob_exchange.client import request_blob
2019-01-22 18:52:32 +01:00
if typing.TYPE_CHECKING:
from lbrynet.conf import Config
2019-01-22 18:52:32 +01:00
from lbrynet.dht.node import Node
from lbrynet.dht.peer import KademliaPeer
from lbrynet.blob.blob_manager import BlobFileManager
from lbrynet.blob.blob_file import BlobFile
log = logging.getLogger(__name__)
class BlobDownloader:
def __init__(self, loop: asyncio.BaseEventLoop, config: 'Config', blob_manager: 'BlobFileManager',
peer_queue: asyncio.Queue):
2019-01-22 18:52:32 +01:00
self.loop = loop
self.config = config
2019-01-22 18:52:32 +01:00
self.blob_manager = blob_manager
self.peer_queue = peer_queue
2019-01-31 19:05:00 +01:00
self.active_connections: typing.Dict['KademliaPeer', asyncio.Task] = {} # active request_blob calls
self.ignored: typing.Set['KademliaPeer'] = set()
2019-01-31 18:28:59 +01:00
self.scores: typing.Dict['KademliaPeer', int] = {}
2019-01-22 18:52:32 +01:00
2019-02-04 22:03:08 +01:00
async def request_blob_from_peer(self, blob: 'BlobFile', peer: 'KademliaPeer'):
if blob.get_is_verified():
return
self.scores[peer] = self.scores.get(peer, 0) - 1 # starts losing score, to account for cancelled ones
2019-02-04 22:03:08 +01:00
success, keep_connection = await request_blob(
self.loop, blob, peer.address, peer.tcp_port, self.config.peer_connect_timeout,
self.config.blob_download_timeout
)
if not keep_connection and peer not in self.ignored:
self.ignored.add(peer)
log.debug("drop peer %s:%i", peer.address, peer.tcp_port)
elif keep_connection:
log.debug("keep peer %s:%i", peer.address, peer.tcp_port)
self.scores[peer] = self.scores.get(peer, 0) + 2 if success else 0
2019-01-22 18:52:32 +01:00
async def new_peer_or_finished(self, blob: 'BlobFile'):
async def get_and_re_add_peers():
new_peers = await self.peer_queue.get()
self.peer_queue.put_nowait(new_peers)
tasks = [self.loop.create_task(get_and_re_add_peers()), self.loop.create_task(blob.verified.wait())]
active_tasks = list(self.active_connections.values())
2019-01-22 18:52:32 +01:00
try:
await asyncio.wait(tasks + active_tasks, loop=self.loop, return_when='FIRST_COMPLETED')
finally:
drain_tasks(tasks)
2019-01-22 18:52:32 +01:00
2019-02-08 00:11:28 +01:00
def cleanup_active(self):
to_remove = [peer for (peer, task) in self.active_connections.items() if task.done()]
for peer in to_remove:
del self.active_connections[peer]
async def download_blob(self, blob_hash: str, length: typing.Optional[int] = None) -> 'BlobFile':
blob = self.blob_manager.get_blob(blob_hash, length)
if blob.get_is_verified():
return blob
2019-01-22 18:52:32 +01:00
try:
while not blob.get_is_verified():
batch: typing.List['KademliaPeer'] = []
while not self.peer_queue.empty():
2019-02-08 00:11:28 +01:00
batch.extend(self.peer_queue.get_nowait())
batch.sort(key=lambda peer: self.scores.get(peer, 0), reverse=True)
2019-02-08 00:11:28 +01:00
log.debug(
"running, %d peers, %d ignored, %d active",
len(batch), len(self.ignored), len(self.active_connections)
)
for peer in batch:
2019-01-31 18:28:59 +01:00
if len(self.active_connections) >= self.config.max_connections_per_download:
break
2019-01-31 19:05:00 +01:00
if peer not in self.active_connections and peer not in self.ignored:
2019-01-31 18:28:59 +01:00
log.debug("request %s from %s:%i", blob_hash[:8], peer.address, peer.tcp_port)
2019-02-04 22:03:08 +01:00
t = self.loop.create_task(self.request_blob_from_peer(blob, peer))
self.active_connections[peer] = t
await self.new_peer_or_finished(blob)
self.cleanup_active()
if batch:
self.peer_queue.put_nowait(set(batch).difference(self.ignored))
while self.active_connections:
peer, task = self.active_connections.popitem()
if task and not task.done():
task.cancel()
blob.close()
2019-02-04 22:09:50 +01:00
log.debug("downloaded %s", blob_hash[:8])
return blob
2019-01-22 18:52:32 +01:00
except asyncio.CancelledError:
while self.active_connections:
peer, task = self.active_connections.popitem()
if task and not task.done():
task.cancel()
2019-01-22 18:52:32 +01:00
raise
2019-02-08 00:11:28 +01:00
except (OSError, Exception) as e:
log.exception(e)
raise e
async def download_blob(loop, config: 'Config', blob_manager: 'BlobFileManager', node: 'Node',
blob_hash: str) -> 'BlobFile':
2019-01-31 18:28:59 +01:00
search_queue = asyncio.Queue(loop=loop, maxsize=config.max_connections_per_download)
search_queue.put_nowait(blob_hash)
peer_queue, accumulate_task = node.accumulate_peers(search_queue)
downloader = BlobDownloader(loop, config, blob_manager, peer_queue)
try:
return await downloader.download_blob(blob_hash)
finally:
if accumulate_task and not accumulate_task.done():
accumulate_task.cancel()