|
| 1 | +""" |
| 2 | +Stream sorter accumulator example. |
| 3 | +
|
| 4 | +This accumulator buffers incoming data and sorts it by event time, |
| 5 | +flushing sorted data when the watermark advances. |
| 6 | +""" |
| 7 | +import asyncio |
| 8 | +from datetime import datetime |
| 9 | +from typing import AsyncIterator |
| 10 | + |
| 11 | +from pynumaflow_lite.accumulator import Datum, Message, AccumulatorAsyncServer, Accumulator |
| 12 | + |
| 13 | + |
| 14 | +class StreamSorter(Accumulator): |
| 15 | + """ |
| 16 | + A stream sorter that buffers and sorts data by event time, |
| 17 | + flushing when watermark advances. |
| 18 | + """ |
| 19 | + |
| 20 | + def __init__(self): |
| 21 | + from datetime import timezone |
| 22 | + # Initialize with a very old timestamp (timezone-aware) |
| 23 | + self.latest_wm = datetime.fromtimestamp(-1, tz=timezone.utc) |
| 24 | + self.sorted_buffer: list[Datum] = [] |
| 25 | + print("StreamSorter initialized") |
| 26 | + |
| 27 | + async def handler(self, datums: AsyncIterator[Datum]) -> AsyncIterator[Message]: |
| 28 | + """ |
| 29 | + Buffer and sort datums, yielding sorted messages when watermark advances. |
| 30 | + """ |
| 31 | + print("Handler started, waiting for datums...") |
| 32 | + datum_count = 0 |
| 33 | + |
| 34 | + async for datum in datums: |
| 35 | + datum_count += 1 |
| 36 | + print(f"Received datum #{datum_count}: event_time={datum.event_time}, " |
| 37 | + f"watermark={datum.watermark}, value={datum.value}") |
| 38 | + |
| 39 | + # If watermark has moved forward |
| 40 | + if datum.watermark and datum.watermark > self.latest_wm: |
| 41 | + old_wm = self.latest_wm |
| 42 | + self.latest_wm = datum.watermark |
| 43 | + print(f"Watermark advanced from {old_wm} to {self.latest_wm}") |
| 44 | + |
| 45 | + # Flush buffer |
| 46 | + flushed = 0 |
| 47 | + async for msg in self.flush_buffer(): |
| 48 | + flushed += 1 |
| 49 | + yield msg |
| 50 | + |
| 51 | + if flushed > 0: |
| 52 | + print(f"Flushed {flushed} messages from buffer") |
| 53 | + |
| 54 | + # Insert into sorted buffer |
| 55 | + self.insert_sorted(datum) |
| 56 | + print(f"Buffer size: {len(self.sorted_buffer)}") |
| 57 | + |
| 58 | + print(f"Handler finished. Total datums processed: {datum_count}") |
| 59 | + print(f"Remaining in buffer: {len(self.sorted_buffer)}") |
| 60 | + |
| 61 | + # Flush any remaining items in the buffer at the end |
| 62 | + if self.sorted_buffer: |
| 63 | + print("Flushing remaining buffer at end...") |
| 64 | + for datum in self.sorted_buffer: |
| 65 | + print(f" Flushing: event_time={datum.event_time}, value={datum.value}") |
| 66 | + # Use Message.from_datum to preserve all metadata |
| 67 | + yield Message.from_datum(datum) |
| 68 | + self.sorted_buffer = [] |
| 69 | + |
| 70 | + def insert_sorted(self, datum: Datum): |
| 71 | + """Binary insert to keep sorted buffer in order by event_time.""" |
| 72 | + left, right = 0, len(self.sorted_buffer) |
| 73 | + while left < right: |
| 74 | + mid = (left + right) // 2 |
| 75 | + if self.sorted_buffer[mid].event_time > datum.event_time: |
| 76 | + right = mid |
| 77 | + else: |
| 78 | + left = mid + 1 |
| 79 | + self.sorted_buffer.insert(left, datum) |
| 80 | + |
| 81 | + async def flush_buffer(self) -> AsyncIterator[Message]: |
| 82 | + """Flush all items from buffer that are before or at the watermark.""" |
| 83 | + i = 0 |
| 84 | + for datum in self.sorted_buffer: |
| 85 | + if datum.event_time > self.latest_wm: |
| 86 | + break |
| 87 | + print(f" Flushing: event_time={datum.event_time}, value={datum.value}") |
| 88 | + # Use Message.from_datum to preserve all metadata (id, headers, event_time, watermark) |
| 89 | + yield Message.from_datum(datum) |
| 90 | + i += 1 |
| 91 | + |
| 92 | + # Remove flushed items |
| 93 | + self.sorted_buffer = self.sorted_buffer[i:] |
| 94 | + |
| 95 | + |
| 96 | +async def main(): |
| 97 | + """ |
| 98 | + Start the accumulator server. |
| 99 | + """ |
| 100 | + import signal |
| 101 | + |
| 102 | + server = AccumulatorAsyncServer() |
| 103 | + |
| 104 | + # Set up signal handlers for graceful shutdown |
| 105 | + loop = asyncio.get_running_loop() |
| 106 | + try: |
| 107 | + loop.add_signal_handler(signal.SIGINT, lambda: server.stop()) |
| 108 | + loop.add_signal_handler(signal.SIGTERM, lambda: server.stop()) |
| 109 | + except (NotImplementedError, RuntimeError): |
| 110 | + pass |
| 111 | + |
| 112 | + try: |
| 113 | + print("Starting Stream Sorter Accumulator Server...") |
| 114 | + await server.start(StreamSorter) |
| 115 | + print("Shutting down gracefully...") |
| 116 | + except asyncio.CancelledError: |
| 117 | + try: |
| 118 | + server.stop() |
| 119 | + except Exception: |
| 120 | + pass |
| 121 | + return |
| 122 | + |
| 123 | + |
| 124 | +# Optional: ensure default signal handlers are in place so asyncio.run can handle them cleanly. |
| 125 | +import signal |
| 126 | +signal.signal(signal.SIGINT, signal.default_int_handler) |
| 127 | +try: |
| 128 | + signal.signal(signal.SIGTERM, signal.SIG_DFL) |
| 129 | +except AttributeError: |
| 130 | + pass |
| 131 | + |
| 132 | +if __name__ == "__main__": |
| 133 | + asyncio.run(main()) |
0 commit comments