e4bf0710c7
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?
118 lines
4.5 KiB
Python
118 lines
4.5 KiB
Python
from pgvector.sqlalchemy import Vector
|
|
from sqlalchemy import (
|
|
JSON,
|
|
Boolean,
|
|
Column,
|
|
DateTime,
|
|
Float,
|
|
ForeignKey,
|
|
Integer,
|
|
String,
|
|
Text,
|
|
)
|
|
from sqlalchemy.orm import relationship
|
|
from sqlalchemy.dialects.postgresql import TSVECTOR as TSVector
|
|
|
|
from app.core.db import Base, utc_now
|
|
|
|
pgvector_type = Vector(1536)
|
|
|
|
|
|
class MemorySource(Base):
|
|
__tablename__ = "memory_sources"
|
|
id = Column(String, primary_key=True)
|
|
user_id = Column(String, ForeignKey("users.id"), nullable=False, index=True)
|
|
source_type = Column(String, nullable=False) # transcript / note / draft
|
|
raw_text = Column(Text, nullable=True)
|
|
storage_key = Column(String, nullable=True)
|
|
speaker = Column(String, nullable=True)
|
|
captured_at = Column(DateTime(timezone=True), nullable=True)
|
|
status = Column(String, default="active")
|
|
conversation_id = Column(String, ForeignKey("conversations.id"), nullable=True)
|
|
created_at = Column(DateTime(timezone=True), default=utc_now)
|
|
chunks = relationship(
|
|
"MemoryChunk", back_populates="source", cascade="all, delete-orphan"
|
|
)
|
|
|
|
|
|
class MemoryChunk(Base):
|
|
__tablename__ = "memory_chunks"
|
|
id = Column(String, primary_key=True)
|
|
source_id = Column(
|
|
String, ForeignKey("memory_sources.id"), nullable=False, index=True
|
|
)
|
|
user_id = Column(String, ForeignKey("users.id"), nullable=False, index=True)
|
|
content = Column(Text, nullable=False)
|
|
# pgvector embedding — Alembic migration 负责 CREATE EXTENSION vector 及列类型
|
|
embedding = Column(pgvector_type, nullable=True)
|
|
# PostgreSQL FTS — Alembic migration 负责 generated tsvector 列 + GIN index
|
|
content_tsv = Column(TSVector, nullable=True)
|
|
chunk_index = Column(Integer, nullable=False)
|
|
speaker = Column(String, nullable=True)
|
|
event_year = Column(Integer, nullable=True)
|
|
metadata_json = Column(JSON, nullable=True)
|
|
is_excluded = Column(Boolean, default=False)
|
|
created_at = Column(DateTime(timezone=True), default=utc_now)
|
|
source = relationship("MemorySource", back_populates="chunks")
|
|
|
|
|
|
class MemorySummary(Base):
|
|
__tablename__ = "memory_summaries"
|
|
id = Column(String, primary_key=True)
|
|
user_id = Column(String, ForeignKey("users.id"), nullable=False, index=True)
|
|
summary_type = Column(String, nullable=False) # session / rolling / topic
|
|
content = Column(Text, nullable=False)
|
|
source_chunk_ids = Column(JSON, nullable=True)
|
|
created_at = Column(DateTime(timezone=True), default=utc_now)
|
|
updated_at = Column(DateTime(timezone=True), default=utc_now, onupdate=utc_now)
|
|
|
|
|
|
class MemoryFact(Base):
|
|
__tablename__ = "memory_facts"
|
|
id = Column(String, primary_key=True)
|
|
user_id = Column(String, ForeignKey("users.id"), nullable=False, index=True)
|
|
fact_type = Column(
|
|
String, nullable=False
|
|
) # person / event / relation / place / milestone
|
|
subject = Column(String, nullable=True)
|
|
predicate = Column(String, nullable=True)
|
|
object_json = Column(JSON, nullable=True)
|
|
confidence = Column(Float, default=0.0)
|
|
source_chunk_id = Column(String, ForeignKey("memory_chunks.id"), nullable=True)
|
|
status = Column(String, default="candidate") # candidate / confirmed / rejected
|
|
created_at = Column(DateTime(timezone=True), default=utc_now)
|
|
|
|
|
|
class TimelineEvent(Base):
|
|
__tablename__ = "timeline_events"
|
|
id = Column(String, primary_key=True)
|
|
user_id = Column(String, ForeignKey("users.id"), nullable=False, index=True)
|
|
memory_source_id = Column(
|
|
String,
|
|
ForeignKey("memory_sources.id", ondelete="SET NULL"),
|
|
nullable=True,
|
|
index=True,
|
|
)
|
|
event_year = Column(Integer, nullable=True)
|
|
event_date = Column(String, nullable=True)
|
|
title = Column(String, nullable=False)
|
|
description = Column(Text, nullable=True)
|
|
person_refs = Column(JSON, nullable=True)
|
|
source_fact_ids = Column(JSON, nullable=True)
|
|
created_at = Column(DateTime(timezone=True), default=utc_now)
|
|
|
|
|
|
class MemoryCurationAction(Base):
|
|
__tablename__ = "memory_curation_actions"
|
|
id = Column(String, primary_key=True)
|
|
user_id = Column(String, ForeignKey("users.id"), nullable=False, index=True)
|
|
action_type = Column(
|
|
String, nullable=False
|
|
) # exclude / restore / correct / merge / confirm / reject
|
|
target_type = Column(
|
|
String, nullable=False
|
|
) # chunk / fact / summary / timeline_event
|
|
target_id = Column(String, nullable=False)
|
|
details = Column(JSON, nullable=True)
|
|
created_at = Column(DateTime(timezone=True), default=utc_now)
|