mirror of
https://github.com/dograh-hq/dograh.git
synced 2026-07-16 11:31:04 +02:00
Fix realtime feedback event correlation (#534)
This commit is contained in:
Abhishek
GitHub
parent
5da57b3fa5
commit
e08660c1fa
9 changed files with 497 additions and 266 deletions
|
|
@ -1,6 +1,7 @@
|
|||
import asyncio
|
||||
import io
|
||||
import wave
|
||||
from copy import deepcopy
|
||||
from datetime import UTC, datetime
|
||||
from typing import List, Optional
|
||||
|
||||
|
|
@ -104,15 +105,9 @@ class InMemoryLogsBuffer:
|
|||
def __init__(self, workflow_run_id: int):
|
||||
self._workflow_run_id = workflow_run_id
|
||||
self._events: List[dict] = []
|
||||
self._turn_counter = 0
|
||||
self._current_turn: Optional[int] = None
|
||||
self._current_node_id: Optional[str] = None
|
||||
self._current_node_name: Optional[str] = None
|
||||
self._user_speech_start_timestamp: Optional[str] = None
|
||||
self._user_speech_end_timestamp: Optional[str] = None
|
||||
self._user_speech_start_from_vad = False
|
||||
self._user_speech_end_from_vad = False
|
||||
self._bot_speech_start_timestamp: Optional[str] = None
|
||||
self._bot_speech_end_timestamp: Optional[str] = None
|
||||
|
||||
def set_current_node(self, node_id: str, node_name: str):
|
||||
"""Set the current node ID and name to be injected into subsequent events."""
|
||||
|
|
@ -129,153 +124,44 @@ class InMemoryLogsBuffer:
|
|||
"""Get the current node name."""
|
||||
return self._current_node_name
|
||||
|
||||
@staticmethod
|
||||
def _now_iso() -> str:
|
||||
return datetime.now(UTC).isoformat(timespec="milliseconds")
|
||||
def set_current_turn(self, turn: int) -> None:
|
||||
"""Set the fallback turn for non-transcript events."""
|
||||
self._current_turn = turn
|
||||
|
||||
def mark_user_started_speaking(
|
||||
self, timestamp: Optional[str] = None, *, from_vad: bool = False
|
||||
async def append(
|
||||
self,
|
||||
event: dict,
|
||||
*,
|
||||
timestamp: Optional[str] = None,
|
||||
turn: Optional[int] = None,
|
||||
node_id: Optional[str] = None,
|
||||
node_name: Optional[str] = None,
|
||||
use_current_node: bool = True,
|
||||
):
|
||||
"""Record when the user started speaking for the current turn."""
|
||||
vad_interval_is_open = (
|
||||
self._user_speech_start_from_vad and self._user_speech_end_timestamp is None
|
||||
)
|
||||
if vad_interval_is_open and not from_vad:
|
||||
return
|
||||
|
||||
self._user_speech_start_timestamp = timestamp or self._now_iso()
|
||||
self._user_speech_end_timestamp = None
|
||||
self._user_speech_start_from_vad = from_vad
|
||||
self._user_speech_end_from_vad = False
|
||||
self._update_latest_payload_start_timestamp(
|
||||
RealtimeFeedbackType.USER_TRANSCRIPTION.value,
|
||||
self._user_speech_start_timestamp,
|
||||
require_final=True,
|
||||
)
|
||||
|
||||
def mark_user_stopped_speaking(
|
||||
self, timestamp: Optional[str] = None, *, from_vad: bool = False
|
||||
):
|
||||
"""Record when the user stopped speaking and update the latest user event."""
|
||||
if self._user_speech_end_from_vad and not from_vad:
|
||||
return
|
||||
|
||||
self._user_speech_end_timestamp = timestamp or self._now_iso()
|
||||
self._user_speech_end_from_vad = from_vad
|
||||
self._update_latest_payload_end_timestamp(
|
||||
RealtimeFeedbackType.USER_TRANSCRIPTION.value,
|
||||
self._user_speech_end_timestamp,
|
||||
require_final=True,
|
||||
)
|
||||
|
||||
def mark_bot_started_speaking(self, timestamp: Optional[str] = None):
|
||||
"""Record when the bot started speaking for the current assistant turn."""
|
||||
self._bot_speech_start_timestamp = timestamp or self._now_iso()
|
||||
self._bot_speech_end_timestamp = None
|
||||
self._update_latest_payload_start_timestamp(
|
||||
RealtimeFeedbackType.BOT_TEXT.value,
|
||||
self._bot_speech_start_timestamp,
|
||||
)
|
||||
|
||||
def mark_bot_stopped_speaking(self, timestamp: Optional[str] = None):
|
||||
"""Record when the bot stopped speaking and update the latest bot event."""
|
||||
self._bot_speech_end_timestamp = timestamp or self._now_iso()
|
||||
self._update_latest_payload_end_timestamp(
|
||||
RealtimeFeedbackType.BOT_TEXT.value,
|
||||
self._bot_speech_end_timestamp,
|
||||
)
|
||||
|
||||
def _find_latest_open_payload(
|
||||
self, event_type: str, *, require_final: bool = False
|
||||
) -> dict | None:
|
||||
for event in reversed(self._events):
|
||||
if event.get("type") != event_type:
|
||||
continue
|
||||
payload = event.get("payload")
|
||||
if not isinstance(payload, dict):
|
||||
continue
|
||||
if require_final and payload.get("final") is not True:
|
||||
continue
|
||||
if payload.get("end_timestamp"):
|
||||
continue
|
||||
return payload
|
||||
return None
|
||||
|
||||
def _update_latest_payload_start_timestamp(
|
||||
self, event_type: str, start_timestamp: str, *, require_final: bool = False
|
||||
):
|
||||
payload = self._find_latest_open_payload(
|
||||
event_type, require_final=require_final
|
||||
)
|
||||
if payload is not None:
|
||||
payload["timestamp"] = start_timestamp
|
||||
|
||||
def _update_latest_payload_end_timestamp(
|
||||
self, event_type: str, end_timestamp: str, *, require_final: bool = False
|
||||
):
|
||||
payload = self._find_latest_open_payload(
|
||||
event_type, require_final=require_final
|
||||
)
|
||||
if payload is not None:
|
||||
payload["end_timestamp"] = end_timestamp
|
||||
|
||||
def _event_with_speech_timestamps(self, event: dict) -> dict:
|
||||
event_type = event.get("type")
|
||||
payload = event.get("payload")
|
||||
if not isinstance(payload, dict):
|
||||
return event
|
||||
|
||||
payload_with_timestamps = dict(payload)
|
||||
if (
|
||||
event_type == RealtimeFeedbackType.USER_TRANSCRIPTION.value
|
||||
and payload.get("final") is True
|
||||
):
|
||||
if self._user_speech_start_timestamp:
|
||||
payload_with_timestamps["timestamp"] = self._user_speech_start_timestamp
|
||||
if self._user_speech_end_timestamp:
|
||||
payload_with_timestamps["end_timestamp"] = (
|
||||
self._user_speech_end_timestamp
|
||||
)
|
||||
elif event_type == RealtimeFeedbackType.BOT_TEXT.value:
|
||||
bot_interval_is_active = self._bot_speech_end_timestamp is None
|
||||
if bot_interval_is_active and self._bot_speech_start_timestamp:
|
||||
payload_with_timestamps["timestamp"] = self._bot_speech_start_timestamp
|
||||
|
||||
if payload_with_timestamps == payload:
|
||||
return event
|
||||
return {**event, "payload": payload_with_timestamps}
|
||||
|
||||
async def append(self, event: dict):
|
||||
"""Append a feedback event to the buffer with timestamp and current node."""
|
||||
event = self._event_with_speech_timestamps(event)
|
||||
"""Append an immutable event with optional correlation metadata."""
|
||||
if use_current_node:
|
||||
node_id = self._current_node_id if node_id is None else node_id
|
||||
node_name = self._current_node_name if node_name is None else node_name
|
||||
timestamped_event = stamp_realtime_feedback_event(
|
||||
event,
|
||||
timestamp=self._now_iso(),
|
||||
turn=self._turn_counter,
|
||||
node_id=self._current_node_id,
|
||||
node_name=self._current_node_name,
|
||||
deepcopy(event),
|
||||
timestamp=timestamp or datetime.now(UTC).isoformat(timespec="milliseconds"),
|
||||
turn=self._current_turn if turn is None else turn,
|
||||
node_id=node_id,
|
||||
node_name=node_name,
|
||||
)
|
||||
self._events.append(timestamped_event)
|
||||
logger.trace(
|
||||
f"Appended event {event.get('type')} to logs buffer for workflow {self._workflow_run_id}"
|
||||
)
|
||||
|
||||
def increment_turn(self):
|
||||
"""Increment turn counter (called on user transcription completion)."""
|
||||
self._turn_counter += 1
|
||||
logger.trace(
|
||||
f"Incremented turn counter to {self._turn_counter} for workflow {self._workflow_run_id}"
|
||||
)
|
||||
|
||||
def _sorted_events(self) -> List[dict]:
|
||||
# Stable sort by the realtime (payload) timestamp when available, falling
|
||||
# back to the buffer-append timestamp. Python's sort is stable, so events
|
||||
# sharing a key retain their original insertion order — this keeps
|
||||
# consecutive bot-text chunks of a single turn contiguous.
|
||||
# Stable sort by the top-level event timestamp used by the persisted
|
||||
# realtime feedback schema. Legacy events without one fall back to their
|
||||
# payload timestamp. Events sharing a key retain insertion order.
|
||||
return sorted(self._events, key=realtime_feedback_event_sort_key)
|
||||
|
||||
def get_events(self) -> List[dict]:
|
||||
"""Get all events for final storage, ordered by realtime timestamp."""
|
||||
"""Get all events for final storage, ordered by event timestamp."""
|
||||
return self._sorted_events()
|
||||
|
||||
def contains_user_speech(self) -> bool:
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue