|
| 1 | +import logging |
| 2 | +import os |
| 3 | +from collections.abc import AsyncIterable |
| 4 | +from datetime import datetime |
| 5 | + |
| 6 | +from pynumaflow import setup_logging |
| 7 | +from pynumaflow.accumulator import Accumulator, AccumulatorAsyncServer |
| 8 | +from pynumaflow.accumulator import ( |
| 9 | + Message, |
| 10 | + Datum, |
| 11 | +) |
| 12 | +from pynumaflow.shared.asynciter import NonBlockingIterator |
| 13 | + |
| 14 | +_LOGGER = setup_logging(__name__) |
| 15 | +if os.getenv("PYTHONDEBUG"): |
| 16 | + _LOGGER.setLevel(logging.DEBUG) |
| 17 | + |
| 18 | + |
| 19 | +class StreamSorter(Accumulator): |
| 20 | + def __init__(self): |
| 21 | + _LOGGER.info("StreamSorter initialized") |
| 22 | + self.latest_wm = datetime.fromtimestamp(-1) |
| 23 | + self.sorted_buffer: list[Datum] = [] |
| 24 | + |
| 25 | + async def handler( |
| 26 | + self, |
| 27 | + datums: AsyncIterable[Datum], |
| 28 | + output: NonBlockingIterator, |
| 29 | + ): |
| 30 | + _LOGGER.info("StreamSorter handler started") |
| 31 | + async for datum in datums: |
| 32 | + _LOGGER.info( |
| 33 | + f"Received datum with event time: {datum.event_time}, " |
| 34 | + f"Current latest watermark: {self.latest_wm}, " |
| 35 | + f"Datum watermark: {datum.watermark}" |
| 36 | + ) |
| 37 | + |
| 38 | + # If watermark has moved forward |
| 39 | + if datum.watermark and datum.watermark > self.latest_wm: |
| 40 | + self.latest_wm = datum.watermark |
| 41 | + await self.flush_buffer(output) |
| 42 | + |
| 43 | + self.insert_sorted(datum) |
| 44 | + |
| 45 | + def insert_sorted(self, datum: Datum): |
| 46 | + # Binary insert to keep sorted buffer in order |
| 47 | + left, right = 0, len(self.sorted_buffer) |
| 48 | + while left < right: |
| 49 | + mid = (left + right) // 2 |
| 50 | + if self.sorted_buffer[mid].event_time > datum.event_time: |
| 51 | + right = mid |
| 52 | + else: |
| 53 | + left = mid + 1 |
| 54 | + self.sorted_buffer.insert(left, datum) |
| 55 | + |
| 56 | + async def flush_buffer(self, output: NonBlockingIterator): |
| 57 | + _LOGGER.info(f"Watermark updated, flushing sortedBuffer: {self.latest_wm}") |
| 58 | + i = 0 |
| 59 | + for datum in self.sorted_buffer: |
| 60 | + if datum.event_time > self.latest_wm: |
| 61 | + break |
| 62 | + await output.put(Message.from_datum(datum)) |
| 63 | + _LOGGER.info(f"Sent datum with event time: {datum.event_time}") |
| 64 | + i += 1 |
| 65 | + # Remove flushed items |
| 66 | + self.sorted_buffer = self.sorted_buffer[i:] |
| 67 | + |
| 68 | + |
| 69 | +if __name__ == "__main__": |
| 70 | + grpc_server = None |
| 71 | + grpc_server = AccumulatorAsyncServer(StreamSorter) |
| 72 | + grpc_server.start() |
0 commit comments