|
| 1 | +import asyncio |
| 2 | +from typing import Union, Dict, NamedTuple |
| 3 | +import queue |
| 4 | + |
| 5 | +from google.api_core.exceptions import FailedPrecondition, GoogleAPICallError |
| 6 | +from google.cloud.pubsub_v1.subscriber.message import Message |
| 7 | +from google.pubsub_v1 import PubsubMessage |
| 8 | + |
| 9 | +from google.cloud.pubsublite.cloudpubsub.flow_control_settings import FlowControlSettings |
| 10 | +from google.cloud.pubsublite.cloudpubsub.internal.ack_set_tracker import AckSetTracker |
| 11 | +from google.cloud.pubsublite.cloudpubsub.message_transformer import MessageTransformer |
| 12 | +from google.cloud.pubsublite.cloudpubsub.nack_handler import NackHandler |
| 13 | +from google.cloud.pubsublite.cloudpubsub.subscriber import AsyncSubscriber |
| 14 | +from google.cloud.pubsublite.internal.wire.permanent_failable import PermanentFailable |
| 15 | +from google.cloud.pubsublite.internal.wire.subscriber import Subscriber |
| 16 | +from google.cloud.pubsublite_v1 import FlowControlRequest, SequencedMessage |
| 17 | +from google.cloud.pubsub_v1.subscriber._protocol import requests |
| 18 | + |
| 19 | + |
| 20 | +class _SizedMessage(NamedTuple): |
| 21 | + message: PubsubMessage |
| 22 | + size_bytes: int |
| 23 | + |
| 24 | + |
| 25 | +class SinglePartitionSubscriber(PermanentFailable, AsyncSubscriber): |
| 26 | + _underlying: Subscriber |
| 27 | + _flow_control_settings: FlowControlSettings |
| 28 | + _ack_set_tracker: AckSetTracker |
| 29 | + _nack_handler: NackHandler |
| 30 | + _transformer: MessageTransformer |
| 31 | + |
| 32 | + _queue: queue.Queue |
| 33 | + _messages_by_offset: Dict[int, _SizedMessage] |
| 34 | + _looper_future: asyncio.Future |
| 35 | + |
| 36 | + def __init__(self, underlying: Subscriber, flow_control_settings: FlowControlSettings, ack_set_tracker: AckSetTracker, |
| 37 | + nack_handler: NackHandler, transformer: MessageTransformer): |
| 38 | + super().__init__() |
| 39 | + self._underlying = underlying |
| 40 | + self._flow_control_settings = flow_control_settings |
| 41 | + self._ack_set_tracker = ack_set_tracker |
| 42 | + self._nack_handler = nack_handler |
| 43 | + self._transformer = transformer |
| 44 | + |
| 45 | + self._queue = queue.Queue() |
| 46 | + self._messages_by_offset = {} |
| 47 | + |
| 48 | + async def read(self) -> Message: |
| 49 | + message: SequencedMessage = await self.await_unless_failed(self._underlying.read()) |
| 50 | + try: |
| 51 | + cps_message = self._transformer.transform(message) |
| 52 | + offset = message.cursor.offset |
| 53 | + self._ack_set_tracker.track(offset) |
| 54 | + self._messages_by_offset[offset] = _SizedMessage(cps_message, message.size_bytes) |
| 55 | + wrapped_message = Message(cps_message._pb, ack_id=str(offset), delivery_attempt=0, request_queue=self._queue) |
| 56 | + return wrapped_message |
| 57 | + except GoogleAPICallError as e: |
| 58 | + self.fail(e) |
| 59 | + raise e |
| 60 | + |
| 61 | + async def _handle_ack(self, message: requests.AckRequest): |
| 62 | + offset = int(message.ack_id) |
| 63 | + await self._underlying.allow_flow( |
| 64 | + FlowControlRequest(allowed_messages=1, allowed_bytes=self._messages_by_offset[offset].size_bytes)) |
| 65 | + del self._messages_by_offset[offset] |
| 66 | + try: |
| 67 | + await self._ack_set_tracker.ack(offset) |
| 68 | + except GoogleAPICallError as e: |
| 69 | + self.fail(e) |
| 70 | + |
| 71 | + def _handle_nack(self, message: requests.NackRequest): |
| 72 | + offset = int(message.ack_id) |
| 73 | + sized_message = self._messages_by_offset[offset] |
| 74 | + try: |
| 75 | + # Put the ack request back into the queue since the callback may be called from another thread. |
| 76 | + self._nack_handler.on_nack(sized_message.message, |
| 77 | + lambda: self._queue.put(requests.AckRequest( |
| 78 | + ack_id=message.ack_id, |
| 79 | + byte_size=0, # Ignored |
| 80 | + time_to_ack=0, # Ignored |
| 81 | + ordering_key="" # Ignored |
| 82 | + ))) |
| 83 | + except GoogleAPICallError as e: |
| 84 | + self.fail(e) |
| 85 | + |
| 86 | + async def _handle_queue_message(self, message: Union[ |
| 87 | + requests.AckRequest, requests.DropRequest, requests.ModAckRequest, requests.NackRequest]): |
| 88 | + if isinstance(message, requests.DropRequest) or isinstance(message, requests.ModAckRequest): |
| 89 | + self.fail(FailedPrecondition("Called internal method of google.cloud.pubsub_v1.subscriber.message.Message " |
| 90 | + f"Pub/Sub Lite does not support: {message}")) |
| 91 | + elif isinstance(message, requests.AckRequest): |
| 92 | + await self._handle_ack(message) |
| 93 | + else: |
| 94 | + self._handle_nack(message) |
| 95 | + |
| 96 | + async def _looper(self): |
| 97 | + while True: |
| 98 | + try: |
| 99 | + # This is not an asyncio.Queue, and therefore we cannot do `await self._queue.get()`. |
| 100 | + # A blocking wait would block the event loop, this needs to be a queue.Queue for |
| 101 | + # compatibility with the Cloud Pub/Sub Message's requirements. |
| 102 | + queue_message = self._queue.get_nowait() |
| 103 | + await self._handle_queue_message(queue_message) |
| 104 | + except queue.Empty: |
| 105 | + await asyncio.sleep(.1) |
| 106 | + |
| 107 | + async def __aenter__(self): |
| 108 | + await self._ack_set_tracker.__aenter__() |
| 109 | + await self._underlying.__aenter__() |
| 110 | + self._looper_future = asyncio.ensure_future(self._looper()) |
| 111 | + await self._underlying.allow_flow(FlowControlRequest( |
| 112 | + allowed_messages=self._flow_control_settings.messages_outstanding, |
| 113 | + allowed_bytes=self._flow_control_settings.bytes_outstanding)) |
| 114 | + return self |
| 115 | + |
| 116 | + async def __aexit__(self, exc_type, exc_value, traceback): |
| 117 | + self._looper_future.cancel() |
| 118 | + try: |
| 119 | + await self._looper_future |
| 120 | + except asyncio.CancelledError: |
| 121 | + pass |
| 122 | + await self._underlying.__aexit__(exc_type, exc_value, traceback) |
| 123 | + await self._ack_set_tracker.__aexit__(exc_type, exc_value, traceback) |
0 commit comments