stream.py 5.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154
  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.storage import FileStore
  11. class EventStreamSubscriber(str, Enum):
  12. AGENT_CONTROLLER = 'agent_controller'
  13. SECURITY_ANALYZER = 'security_analyzer'
  14. SERVER = 'server'
  15. RUNTIME = 'runtime'
  16. MAIN = 'main'
  17. TEST = 'test'
  18. class EventStream:
  19. sid: str
  20. file_store: FileStore
  21. # For each subscriber ID, there is a stack of callback functions - useful
  22. # when there are agent delegates
  23. _subscribers: dict[str, list[Callable]]
  24. _cur_id: int
  25. _lock: threading.Lock
  26. def __init__(self, sid: str, file_store: FileStore):
  27. self.sid = sid
  28. self.file_store = file_store
  29. self._subscribers = {}
  30. self._cur_id = 0
  31. self._lock = threading.Lock()
  32. self._reinitialize_from_file_store()
  33. def _reinitialize_from_file_store(self) -> None:
  34. try:
  35. events = self.file_store.list(f'sessions/{self.sid}/events')
  36. except FileNotFoundError:
  37. logger.debug(f'No events found for session {self.sid}')
  38. self._cur_id = 0
  39. return
  40. # if we have events, we need to find the highest id to prepare for new events
  41. for event_str in events:
  42. id = self._get_id_from_filename(event_str)
  43. if id >= self._cur_id:
  44. self._cur_id = id + 1
  45. def _get_filename_for_id(self, id: int) -> str:
  46. return f'sessions/{self.sid}/events/{id}.json'
  47. @staticmethod
  48. def _get_id_from_filename(filename: str) -> int:
  49. try:
  50. return int(filename.split('/')[-1].split('.')[0])
  51. except ValueError:
  52. logger.warning(f'get id from filename ({filename}) failed.')
  53. return -1
  54. def get_events(
  55. self,
  56. start_id=0,
  57. end_id=None,
  58. reverse=False,
  59. filter_out_type: tuple[type[Event], ...] | None = None,
  60. ) -> Iterable[Event]:
  61. if reverse:
  62. if end_id is None:
  63. end_id = self._cur_id - 1
  64. event_id = end_id
  65. while event_id >= start_id:
  66. try:
  67. event = self.get_event(event_id)
  68. if filter_out_type is None or not isinstance(
  69. event, filter_out_type
  70. ):
  71. yield event
  72. except FileNotFoundError:
  73. logger.debug(f'No event found for ID {event_id}')
  74. event_id -= 1
  75. else:
  76. event_id = start_id
  77. while True:
  78. if end_id is not None and event_id > end_id:
  79. break
  80. try:
  81. event = self.get_event(event_id)
  82. if filter_out_type is None or not isinstance(
  83. event, filter_out_type
  84. ):
  85. yield event
  86. except FileNotFoundError:
  87. break
  88. event_id += 1
  89. def get_event(self, id: int) -> Event:
  90. filename = self._get_filename_for_id(id)
  91. content = self.file_store.read(filename)
  92. data = json.loads(content)
  93. return event_from_dict(data)
  94. def get_latest_event(self) -> Event:
  95. return self.get_event(self._cur_id - 1)
  96. def get_latest_event_id(self) -> int:
  97. return self._cur_id - 1
  98. def subscribe(self, id: EventStreamSubscriber, callback: Callable, append=False):
  99. if id in self._subscribers:
  100. if append:
  101. self._subscribers[id].append(callback)
  102. else:
  103. raise ValueError('Subscriber already exists: ' + id)
  104. else:
  105. self._subscribers[id] = [callback]
  106. def unsubscribe(self, id: EventStreamSubscriber):
  107. if id not in self._subscribers:
  108. logger.warning('Subscriber not found during unsubscribe: ' + id)
  109. else:
  110. self._subscribers[id].pop()
  111. if len(self._subscribers[id]) == 0:
  112. del self._subscribers[id]
  113. def add_event(self, event: Event, source: EventSource):
  114. with self._lock:
  115. event._id = self._cur_id # type: ignore [attr-defined]
  116. self._cur_id += 1
  117. logger.debug(f'Adding {type(event).__name__} id={event.id} from {source.name}')
  118. event._timestamp = datetime.now() # type: ignore [attr-defined]
  119. event._source = source # type: ignore [attr-defined]
  120. data = event_to_dict(event)
  121. if event.id is not None:
  122. self.file_store.write(self._get_filename_for_id(event.id), json.dumps(data))
  123. for key in sorted(self._subscribers.keys()):
  124. stack = self._subscribers[key]
  125. callback = stack[-1]
  126. asyncio.create_task(callback(event))
  127. def filtered_events_by_source(self, source: EventSource):
  128. for event in self.get_events():
  129. if event.source == source:
  130. yield event
  131. def clear(self):
  132. self.file_store.delete(f'sessions/{self.sid}')
  133. self._cur_id = 0
  134. # self._subscribers = {}
  135. self._reinitialize_from_file_store()