Python
"""
Agent Memory V2 - 数据层 (Data Layer)
三引擎存储架构:
- VectorStore (Chroma/Qdrant/FAISS): 向量存储,L1 Raw / L2 Fact / Schema / Intention 语义检索
- GraphStore (Kuzu/Neo4j): 图存储,实体关系 / 图遍历 / 时间线 / 知识缺口
- Cache (SQLite/MySQL): 审计/观测落库(memory_operations / pipeline_logs / system_metrics)
缓存后端:
- SqliteCache: 零依赖本地(默认)
- MysqlCache: 标准 MySQL 持久化(需 aiomysql 包)
- create_cache(): 工厂函数,根据配置选择后端
"""
from .vector_store import VectorStore, create_vector_store
from .vector_store_base import VectorStoreBase
from .graph_store import GraphStore, create_graph_store
from .graph_store_base import GraphStoreBase
from .cache_base import CacheBase
from .cache_sqlite import SqliteCache
from .cache_mysql import MysqlCache
from .cache import create_cache
from .history_store import HistoryStore
# 保留旧接口兼容
from .kv_store import KVStore, KVStoreConfig, KVBackend
__all__ = [
# V2 核心
"VectorStore",
"VectorStoreBase",
"create_vector_store",
"GraphStore",
"GraphStoreBase",
"create_graph_store",
"CacheBase",
"SqliteCache",
"MysqlCache",
"create_cache",
"HistoryStore",
# V1 兼容
"KVStore",
"KVStoreConfig",
"KVBackend",
]