stream.py 5.4 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155
  1. import asyncio
  2. import threading
  3. from datetime import datetime
  4. from enum import Enum
  5. from typing import Callable, Iterable
  6. from openhands.core.logger import openhands_logger as logger
  7. from openhands.core.utils import json
  8. from openhands.events.event import Event, EventSource
  9. from openhands.events.serialization.event import event_from_dict, event_to_dict
  10. from openhands.runtime.utils.shutdown_listener import should_continue
  11. from openhands.storage import FileStore
  12. class EventStreamSubscriber(str, Enum):
  13. AGENT_CONTROLLER = 'agent_controller'
  14. SECURITY_ANALYZER = 'security_analyzer'
  15. SERVER = 'server'
  16. RUNTIME = 'runtime'
  17. MAIN = 'main'
  18. TEST = 'test'
  19. class EventStream:
  20. sid: str
  21. file_store: FileStore
  22. # For each subscriber ID, there is a stack of callback functions - useful
  23. # when there are agent delegates
  24. _subscribers: dict[str, list[Callable]]
  25. _cur_id: int
  26. _lock: threading.Lock
  27. def __init__(self, sid: str, file_store: FileStore):
  28. self.sid = sid
  29. self.file_store = file_store
  30. self._subscribers = {}
  31. self._cur_id = 0
  32. self._lock = threading.Lock()
  33. self._reinitialize_from_file_store()
  34. def _reinitialize_from_file_store(self) -> None:
  35. try:
  36. events = self.file_store.list(f'sessions/{self.sid}/events')
  37. except FileNotFoundError:
  38. logger.debug(f'No events found for session {self.sid}')
  39. self._cur_id = 0
  40. return
  41. # if we have events, we need to find the highest id to prepare for new events
  42. for event_str in events:
  43. id = self._get_id_from_filename(event_str)
  44. if id >= self._cur_id:
  45. self._cur_id = id + 1
  46. def _get_filename_for_id(self, id: int) -> str:
  47. return f'sessions/{self.sid}/events/{id}.json'
  48. @staticmethod
  49. def _get_id_from_filename(filename: str) -> int:
  50. try:
  51. return int(filename.split('/')[-1].split('.')[0])
  52. except ValueError:
  53. logger.warning(f'get id from filename ({filename}) failed.')
  54. return -1
  55. def get_events(
  56. self,
  57. start_id=0,
  58. end_id=None,
  59. reverse=False,
  60. filter_out_type: tuple[type[Event], ...] | None = None,
  61. ) -> Iterable[Event]:
  62. if reverse:
  63. if end_id is None:
  64. end_id = self._cur_id - 1
  65. event_id = end_id
  66. while event_id >= start_id:
  67. try:
  68. event = self.get_event(event_id)
  69. if filter_out_type is None or not isinstance(
  70. event, filter_out_type
  71. ):
  72. yield event
  73. except FileNotFoundError:
  74. logger.debug(f'No event found for ID {event_id}')
  75. event_id -= 1
  76. else:
  77. event_id = start_id
  78. while should_continue():
  79. if end_id is not None and event_id > end_id:
  80. break
  81. try:
  82. event = self.get_event(event_id)
  83. if filter_out_type is None or not isinstance(
  84. event, filter_out_type
  85. ):
  86. yield event
  87. except FileNotFoundError:
  88. break
  89. event_id += 1
  90. def get_event(self, id: int) -> Event:
  91. filename = self._get_filename_for_id(id)
  92. content = self.file_store.read(filename)
  93. data = json.loads(content)
  94. return event_from_dict(data)
  95. def get_latest_event(self) -> Event:
  96. return self.get_event(self._cur_id - 1)
  97. def get_latest_event_id(self) -> int:
  98. return self._cur_id - 1
  99. def subscribe(self, id: EventStreamSubscriber, callback: Callable, append=False):
  100. if id in self._subscribers:
  101. if append:
  102. self._subscribers[id].append(callback)
  103. else:
  104. raise ValueError('Subscriber already exists: ' + id)
  105. else:
  106. self._subscribers[id] = [callback]
  107. def unsubscribe(self, id: EventStreamSubscriber):
  108. if id not in self._subscribers:
  109. logger.warning('Subscriber not found during unsubscribe: ' + id)
  110. else:
  111. self._subscribers[id].pop()
  112. if len(self._subscribers[id]) == 0:
  113. del self._subscribers[id]
  114. def add_event(self, event: Event, source: EventSource):
  115. with self._lock:
  116. event._id = self._cur_id # type: ignore [attr-defined]
  117. self._cur_id += 1
  118. logger.debug(f'Adding {type(event).__name__} id={event.id} from {source.name}')
  119. event._timestamp = datetime.now().isoformat()
  120. event._source = source # type: ignore [attr-defined]
  121. data = event_to_dict(event)
  122. if event.id is not None:
  123. self.file_store.write(self._get_filename_for_id(event.id), json.dumps(data))
  124. for key in sorted(self._subscribers.keys()):
  125. stack = self._subscribers[key]
  126. callback = stack[-1]
  127. asyncio.create_task(callback(event))
  128. def filtered_events_by_source(self, source: EventSource):
  129. for event in self.get_events():
  130. if event.source == source:
  131. yield event
  132. def clear(self):
  133. self.file_store.delete(f'sessions/{self.sid}')
  134. self._cur_id = 0
  135. # self._subscribers = {}
  136. self._reinitialize_from_file_store()