chore: stash

pkg/vector/mgr.py

@@ -1,13 +1,18 @@
 from __future__ import annotations
 
 from ..core import app
+from .vdb import VectorDatabase
+from .vdbs.chroma import ChromaVectorDatabase
 
 
 class VectorDBManager:
     ap: app.Application
+    vector_db: VectorDatabase = None
 
     def __init__(self, ap: app.Application):
         self.ap = ap
 
     async def initialize(self):
-        pass
+        # 初始化 Chroma 向量数据库（可扩展为多种实现）
+        if self.vector_db is None:
+            self.vector_db = ChromaVectorDatabase(self.ap)

pkg/vector/vdb.py

@@ -1,7 +1,33 @@
 from __future__ import annotations
-
 import abc
+from typing import Any, List, Dict
+import numpy as np
 
 
 class VectorDatabase(abc.ABC):
-    pass
+    @abc.abstractmethod
+    def add_embeddings(
+        self,
+        collection: str,
+        ids: List[str],
+        embeddings: np.ndarray,
+        metadatas: List[Dict[str, Any]],
+        documents: List[str],
+    ) -> None:
+        """向指定 collection 添加向量数据。"""
+        pass
+
+    @abc.abstractmethod
+    def search(self, collection: str, query_embedding: np.ndarray, k: int = 5) -> Dict[str, Any]:
+        """在指定 collection 中检索最相似的向量。"""
+        pass
+
+    @abc.abstractmethod
+    def delete_by_metadata(self, collection: str, where: Dict[str, Any]) -> None:
+        """根据元数据删除指定 collection 中的向量。"""
+        pass
+
+    @abc.abstractmethod
+    def get_or_create_collection(self, collection: str):
+        """获取或创建 collection。"""
+        pass

pkg/vector/vdbs/chroma.py

@@ -0,0 +1,46 @@
+from __future__ import annotations
+import numpy as np
+from typing import Any, List, Dict
+from chromadb import PersistentClient
+from pkg.vector.vdb import VectorDatabase
+from pkg.core import app
+
+
+class ChromaVectorDatabase(VectorDatabase):
+    def __init__(self, ap: app.Application, base_path: str = './data/chroma'):
+        self.ap = ap
+        self.client = PersistentClient(path=base_path)
+        self._collections = {}
+
+    def get_or_create_collection(self, collection: str):
+        if collection not in self._collections:
+            self._collections[collection] = self.client.get_or_create_collection(name=collection)
+            self.ap.logger.info(f"Chroma collection '{collection}' accessed/created.")
+        return self._collections[collection]
+
+    def add_embeddings(
+        self,
+        collection: str,
+        ids: List[str],
+        embeddings: np.ndarray,
+        metadatas: List[Dict[str, Any]],
+        documents: List[str],
+    ) -> None:
+        col = self.get_or_create_collection(collection)
+        col.add(embeddings=embeddings.tolist(), ids=ids, metadatas=metadatas, documents=documents)
+        self.ap.logger.info(f"Added {len(ids)} embeddings to Chroma collection '{collection}'.")
+
+    def search(self, collection: str, query_embedding: np.ndarray, k: int = 5) -> Dict[str, Any]:
+        col = self.get_or_create_collection(collection)
+        results = col.query(
+            query_embeddings=query_embedding.tolist(),
+            n_results=k,
+            include=['metadatas', 'distances', 'documents'],
+        )
+        self.ap.logger.debug(f"Chroma search in '{collection}' returned {len(results.get('ids', [[]])[0])} results.")
+        return results
+
+    def delete_by_metadata(self, collection: str, where: Dict[str, Any]) -> None:
+        col = self.get_or_create_collection(collection)
+        col.delete(where=where)
+        self.ap.logger.info(f"Deleted embeddings from Chroma collection '{collection}' with filter: {where}")