数据库 - 新增迁移 0003:timeline_events.memory_source_id 外键 → memory_sources,便于按 ingest 源做时间线幂等 后端 - 记忆 - 新增 ingest 后 LLM 富化(摘要/事实/时间线),可配置开关与最大字符数 - 新增证据包组装:合并 chunk、摘要、事实、时间线、故事等检索结果;支持空 query 时是否仍带 rolling 等开关 - repo/retriever/service/router/schemas/summarizer/timeline/extractor 等扩展;文档 memory-retrieval.md 更新 后端 - 对话 WS - 增加 PING/PONG;分段 ASR 日志与空音频处理;转写失败与「无助手回复」错误提示更明确 - 助手多段回复持久化使用统一分隔符,与分段逻辑一致 后端 - Agent - reply_limits:按 [SPLIT] 与段落拆段,并保证非空 fallback,供 WS 与 TTS 多段下发 后端 - 回忆录任务 - transcript ingest 记录 source_id;任务成功结?
150 lines
5.1 KiB
Python
150 lines
5.1 KiB
Python
"""Durable conversation turn persistence + Redis cache sync (feature layer)."""
|
|
|
|
from __future__ import annotations
|
|
|
|
import uuid
|
|
from datetime import datetime, timedelta, timezone
|
|
from typing import Any
|
|
|
|
from sqlalchemy.ext.asyncio import AsyncSession
|
|
|
|
from app.core import redis as redis_core
|
|
from app.core.logging import get_logger
|
|
from app.features.conversation import repo
|
|
from app.features.conversation.models import ConversationMessage
|
|
from app.features.conversation.session_history import (
|
|
conversation_messages_to_redis_history,
|
|
)
|
|
|
|
logger = get_logger(__name__)
|
|
|
|
# 与 LLM / 客户端约定:多段助手消息用 [SPLIT] 拼接,便于拆成多条气泡与多段 TTS
|
|
AI_RESPONSE_SEGMENT_JOIN = "[SPLIT]"
|
|
|
|
|
|
def _utc_now() -> datetime:
|
|
return datetime.now(timezone.utc)
|
|
|
|
|
|
class ConversationHistoryStore:
|
|
def __init__(self, db: AsyncSession):
|
|
self._db = db
|
|
|
|
async def load_canonical_history(
|
|
self, conversation_id: str
|
|
) -> list[dict[str, Any]]:
|
|
rows = await repo.get_conversation_messages(conversation_id, self._db)
|
|
return conversation_messages_to_redis_history(rows)
|
|
|
|
async def _touch_conversation(
|
|
self, conversation_id: str, *, occurred_at: datetime
|
|
) -> None:
|
|
conversation = await repo.get_conversation(conversation_id, self._db)
|
|
if conversation is None:
|
|
return
|
|
current = getattr(conversation, "last_message_at", None)
|
|
if current is None or current < occurred_at:
|
|
conversation.last_message_at = occurred_at
|
|
|
|
async def _sync_redis_from_db(self, conversation_id: str) -> None:
|
|
hist = await self.load_canonical_history(conversation_id)
|
|
await redis_core.redis_service.set_conversation_history(conversation_id, hist)
|
|
|
|
async def _sync_redis_best_effort(self, conversation_id: str) -> None:
|
|
try:
|
|
await self._sync_redis_from_db(conversation_id)
|
|
except Exception as exc:
|
|
logger.warning("conversation history cache sync skipped: {}", exc)
|
|
|
|
async def record_ai_only_turn(
|
|
self, conversation_id: str, responses: list[str]
|
|
) -> str | None:
|
|
if not responses:
|
|
return None
|
|
combined = AI_RESPONSE_SEGMENT_JOIN.join(responses)
|
|
created_at = _utc_now()
|
|
msg = ConversationMessage(
|
|
id=str(uuid.uuid4()),
|
|
conversation_id=conversation_id,
|
|
role="ai",
|
|
content=combined,
|
|
message_type="text",
|
|
created_at=created_at,
|
|
)
|
|
repo.add_conversation_message(msg, self._db)
|
|
await self._touch_conversation(conversation_id, occurred_at=created_at)
|
|
await self._db.commit()
|
|
await self._sync_redis_best_effort(conversation_id)
|
|
return msg.id
|
|
|
|
async def record_human_ai_turn(
|
|
self,
|
|
conversation_id: str,
|
|
user_message: str,
|
|
responses: list[str],
|
|
*,
|
|
user_message_timestamp: datetime | None,
|
|
is_from_voice: bool,
|
|
voice_session_id: str | None,
|
|
audio_duration_seconds: int | None,
|
|
tts_audio_urls: list[str] | None,
|
|
segment_id: str | None,
|
|
) -> str | None:
|
|
if not responses:
|
|
return None
|
|
human_ts = user_message_timestamp or _utc_now()
|
|
if human_ts.tzinfo is None:
|
|
human_ts = human_ts.replace(tzinfo=timezone.utc)
|
|
ai_ts = human_ts + timedelta(microseconds=1)
|
|
human_type = "audio" if is_from_voice else "text"
|
|
human = ConversationMessage(
|
|
id=str(uuid.uuid4()),
|
|
conversation_id=conversation_id,
|
|
role="human",
|
|
content=user_message,
|
|
message_type=human_type,
|
|
voice_session_id=voice_session_id,
|
|
duration_seconds=audio_duration_seconds
|
|
if audio_duration_seconds is not None and audio_duration_seconds > 0
|
|
else None,
|
|
segment_id=segment_id,
|
|
created_at=human_ts,
|
|
)
|
|
combined = AI_RESPONSE_SEGMENT_JOIN.join(responses)
|
|
ai = ConversationMessage(
|
|
id=str(uuid.uuid4()),
|
|
conversation_id=conversation_id,
|
|
role="ai",
|
|
content=combined,
|
|
message_type="text",
|
|
tts_audio_urls=tts_audio_urls if tts_audio_urls else None,
|
|
segment_id=segment_id,
|
|
created_at=ai_ts,
|
|
)
|
|
repo.add_conversation_message(human, self._db)
|
|
repo.add_conversation_message(ai, self._db)
|
|
await self._touch_conversation(conversation_id, occurred_at=ai_ts)
|
|
await self._db.commit()
|
|
await self._sync_redis_best_effort(conversation_id)
|
|
return ai.id
|
|
|
|
async def attach_ai_tts_audio_urls(
|
|
self,
|
|
conversation_id: str,
|
|
*,
|
|
tts_audio_urls: list[str],
|
|
segment_id: str | None = None,
|
|
) -> None:
|
|
if not tts_audio_urls:
|
|
return
|
|
row = await repo.set_latest_ai_message_tts_audio_urls(
|
|
conversation_id,
|
|
self._db,
|
|
tts_audio_urls=tts_audio_urls,
|
|
segment_id=segment_id,
|
|
)
|
|
if row is None:
|
|
return
|
|
await self._db.commit()
|
|
await self._sync_redis_best_effort(conversation_id)
|