new: add collection metadata and tests to local mode (#1089)

* new: add collection metadata and tests to local mode

* fix: regen async client

Author: joein

qdrant_client/local/async_qdrant_local.py

@@ -885,14 +885,20 @@ async def update_collection(
         self,
         collection_name: str,
         sparse_vectors_config: Optional[Mapping[str, types.SparseVectorParams]] = None,
+        metadata: Optional[types.Payload] = None,
         **kwargs: Any,
     ) -> bool:
         _collection = self._get_collection(collection_name)
+        updated = False
         if sparse_vectors_config is not None:
             for vector_name, vector_params in sparse_vectors_config.items():
                 _collection.update_sparse_vectors_config(vector_name, vector_params)
-            return True
-        return False
+            updated = True
+        if metadata is not None:
+            _collection.config.metadata.update(metadata)
+            updated = True
+        self._save()
+        return updated
 
     def _collection_path(self, collection_name: str) -> Optional[str]:
         if self.persistent:
@@ -924,6 +930,7 @@ async def create_collection(
         ] = None,
         init_from: Optional[types.InitFrom] = None,
         sparse_vectors_config: Optional[Mapping[str, types.SparseVectorParams]] = None,
+        metadata: Optional[types.Payload] = None,
         **kwargs: Any,
     ) -> bool:
         if self.closed:
@@ -942,7 +949,9 @@ async def create_collection(
             os.makedirs(collection_path, exist_ok=True)
         collection = LocalCollection(
             rest_models.CreateCollection(
-                vectors=vectors_config or {}, sparse_vectors=sparse_vectors_config
+                vectors=vectors_config or {},
+                sparse_vectors=sparse_vectors_config,
+                metadata=metadata,
             ),
             location=collection_path,
             force_disable_check_same_thread=self.force_disable_check_same_thread,

qdrant_client/local/local_collection.py

@@ -2786,6 +2786,7 @@ def info(self) -> models.CollectionInfo:
                     max_optimization_threads=1,
                 ),
                 quantization_config=None,
+                metadata=self.config.metadata,
             ),
         )
 

qdrant_client/local/qdrant_local.py

@@ -954,16 +954,22 @@ def update_collection(
         self,
         collection_name: str,
         sparse_vectors_config: Optional[Mapping[str, types.SparseVectorParams]] = None,
+        metadata: Optional[types.Payload] = None,
         **kwargs: Any,
     ) -> bool:
         _collection = self._get_collection(collection_name)
-
+        updated = False
         if sparse_vectors_config is not None:
             for vector_name, vector_params in sparse_vectors_config.items():
                 _collection.update_sparse_vectors_config(vector_name, vector_params)
+            updated = True
 
-            return True
-        return False
+        if metadata is not None:
+            _collection.config.metadata.update(metadata)
+            updated = True
+
+        self._save()
+        return updated
 
     def _collection_path(self, collection_name: str) -> Optional[str]:
         if self.persistent:
@@ -996,6 +1002,7 @@ def create_collection(
         ] = None,
         init_from: Optional[types.InitFrom] = None,
         sparse_vectors_config: Optional[Mapping[str, types.SparseVectorParams]] = None,
+        metadata: Optional[types.Payload] = None,
         **kwargs: Any,
     ) -> bool:
         if self.closed:
@@ -1019,6 +1026,7 @@ def create_collection(
             rest_models.CreateCollection(
                 vectors=vectors_config or {},
                 sparse_vectors=sparse_vectors_config,
+                metadata=metadata,
             ),
             location=collection_path,
             force_disable_check_same_thread=self.force_disable_check_same_thread,

tests/test_local_persistence.py

@@ -150,3 +150,54 @@ def test_local_sparse_persistence(add_dense_to_config):
         client = qdrant_client.QdrantClient(path=tmpdir)
         assert client.count(default_collection_name).count == 10
         assert client.count("example_2").count == 10
+
+
+def test_update_persisence():
+    collection_name = "update_persisence"
+    with tempfile.TemporaryDirectory() as tmpdir:
+        client = qdrant_client.QdrantClient(path=tmpdir)
+
+        if client.collection_exists(collection_name):
+            client.delete_collection(collection_name)
+
+        client.create_collection(
+            collection_name,
+            vectors_config={"dense": rest.VectorParams(size=20, distance=rest.Distance.COSINE)},
+            sparse_vectors_config={
+                "text": rest.SparseVectorParams(),
+            },
+            metadata={"important": "meta information"},
+        )
+
+        original_collection_info = client.get_collection(collection_name)
+
+        assert original_collection_info.config.params.sparse_vectors["text"].modifier is None
+        assert original_collection_info.config.metadata == {"important": "meta information"}
+
+        client.update_collection(
+            collection_name,
+            sparse_vectors_config={"text": rest.SparseVectorParams(modifier=rest.Modifier.IDF)},
+            metadata={"not_important": "missing"},
+        )
+        updated_collection_info = client.get_collection(collection_name)
+        assert (
+            updated_collection_info.config.params.sparse_vectors["text"].modifier
+            == rest.Modifier.IDF
+        )
+        assert updated_collection_info.config.metadata == {
+            "important": "meta information",
+            "not_important": "missing",
+        }
+        client.close()
+        del client
+
+        client = qdrant_client.QdrantClient(path=tmpdir)
+        persisted_collection_info = client.get_collection(collection_name)
+        assert (
+            persisted_collection_info.config.params.sparse_vectors["text"].modifier
+            == rest.Modifier.IDF
+        )
+        assert persisted_collection_info.config.metadata == {
+            "important": "meta information",
+            "not_important": "missing",
+        }

tests/test_qdrant_client.py

@@ -1183,6 +1183,11 @@ def test_points_crud(prefer_grpc):
         collection_info = client.get_collection(COLLECTION_NAME)
         assert collection_info.config.metadata == collection_metadata
 
+        new_metadata = {"due_date": "12.12.2222"}
+        client.update_collection(COLLECTION_NAME, metadata=new_metadata)
+        updated_collection_info = client.get_collection(COLLECTION_NAME)
+        assert updated_collection_info.config.metadata == {**collection_metadata, **new_metadata}
+
     # Create a single point
     client.upsert(
         collection_name=COLLECTION_NAME,