feat: enhance collection creation and update vector commands with explicit vector names

Author: srimon12

docs/scripts.md

@@ -120,7 +120,7 @@ Done. 41 point(s) written.
 --             configured model (see: qql connect).
 -- ============================================================
 
-CREATE COLLECTION medical_records HYBRID
+CREATE COLLECTION medical_records USING HYBRID DENSE VECTOR 'dense' SPARSE VECTOR 'sparse'
 
 -- Batch 1 / 1  (records 1–41)
 INSERT BULK INTO COLLECTION medical_records VALUES [
@@ -132,7 +132,7 @@ INSERT BULK INTO COLLECTION medical_records VALUES [
     'peer_reviewed': true
   },
   ...
-] USING HYBRID
+] USING HYBRID DENSE VECTOR 'dense' SPARSE VECTOR 'sparse'
 
 -- ============================================================
 -- End of dump

src/qql/__init__.py

@@ -15,6 +15,7 @@
 __all__ = [
     "__version__",
     "Connection",
+    "DEFAULT_MODEL",
     "QQLConfig",
     "QQLError",
     "QQLRuntimeError",

src/qql/cli.py

@@ -110,7 +110,8 @@
   [yellow]DELETE FROM[/yellow] <name> [yellow]WHERE id =[/yellow] '<id>'
       Delete a point by its ID.
 
-  [yellow]UPDATE[/yellow] <name> [yellow]SET VECTOR[/yellow] ['<dense_vector>'] [yellow]WHERE id =[/yellow] '<id>'|<int> [<vector>]
+  [yellow]UPDATE[/yellow] <name> [yellow]SET VECTOR[/yellow] [yellow]WHERE id =[/yellow] '<id>'|<int> [<vector>]
+  [yellow]UPDATE[/yellow] <name> [yellow]SET VECTOR[/yellow] '<dense_vector_name>' [yellow]WHERE id =[/yellow] '<id>'|<int> [<vector>]
       Replace the dense vector for a single point by ID.
       The point must already exist. Vector is a float array: [0.1, 0.2, ..., 0.N]
 

src/qql/executor.py

@@ -74,7 +74,6 @@
     AlterCollectionStmt,
     AndExpr,
     BetweenExpr,
-    CollectionParamsConfig,
     CollectionConfig,
     CompareExpr,
     CreateCollectionStmt,
@@ -94,7 +93,6 @@
     MatchTextExpr,
     NotExpr,
     NotInExpr,
-    OptimizersRuntimeConfig,
     OrExpr,
     QuantizationUpdate,
     QuantizationConfig,
@@ -108,8 +106,6 @@
     ShowCollectionsStmt,
     UpdateVectorStmt,
     UpdatePayloadStmt,
-    VectorsConfig,
-    HnswRuntimeConfig,
 )
 from .config import QQLConfig
 from .embedder import CrossEncoderEmbedder, Embedder, SparseEmbedder
@@ -384,10 +380,13 @@ def _execute_insert_bulk(self, node: InsertBulkStmt) -> ExecutionResult:
                 dense_name = topology.dense_using(node.dense_vector) or dense_name
                 sparse_name = topology.sparse_using(node.sparse_vector)
 
+            first_dense_vector: list[float] | None = None
             points: list[PointStruct] = []
             for vals in node.values_list:
                 point_id, payload = self._extract_point_id_and_payload(vals)
                 dense_vector = dense_embedder.embed(vals["text"])
+                if first_dense_vector is None:
+                    first_dense_vector = dense_vector
                 sparse_obj = sparse_embedder.embed(vals["text"])
                 sparse_vector = SparseVector(
                     indices=sparse_obj["indices"], values=sparse_obj["values"]
@@ -401,11 +400,11 @@ def _execute_insert_bulk(self, node: InsertBulkStmt) -> ExecutionResult:
                 )
 
             if not topology.exists:
-                first_dense = dense_embedder.embed(node.values_list[0]["text"])
+                assert first_dense_vector is not None
                 self._create_collection_and_wait(
                     collection_name=node.collection,
                     vectors_config={
-                        dense_name: VectorParams(size=len(first_dense), distance=Distance.COSINE)
+                        dense_name: VectorParams(size=len(first_dense_vector), distance=Distance.COSINE)
                     },
                     sparse_vectors_config={
                         sparse_name: SparseVectorParams(modifier=Modifier.IDF)
@@ -557,7 +556,7 @@ def _execute_alter_collection(self, node: AlterCollectionStmt) -> ExecutionResul
         topology = self._resolve_topology(node.collection)
 
         update_kwargs: dict[str, Any] = {"collection_name": node.collection}
-        vectors_config = self._build_vectors_config_diff(node.collection, node.config)
+        vectors_config = self._build_vectors_config_diff(topology, node.config)
         hnsw_config = self._build_hnsw_config(node.config)
         optimizers_config = self._build_optimizers_config(node.config)
         collection_params = self._build_collection_params_diff(node.config)
@@ -1328,12 +1327,11 @@ def _build_collection_params_diff(
 
     def _build_vectors_config_diff(
         self,
-        collection_name: str,
+        topology: CollectionTopology,
         config: CollectionConfig | None,
     ) -> dict[str, VectorParamsDiff] | None:
         if config is None or config.vectors is None:
             return None
-        topology = self._resolve_topology(collection_name)
         vector_name = topology.dense_config_key()
         return {
             vector_name: VectorParamsDiff(on_disk=config.vectors.on_disk),

tests/test_executor.py

@@ -1936,6 +1936,28 @@ def test_dense_search_explicit_vector_resolves_ambiguous_collection(
         executor.execute(node)
         assert mock_client.query_points.call_args.kwargs["using"] == "body"
 
+    def test_dense_search_explicit_vector_unknown_name_raises(
+        self, executor, mock_client
+    ):
+        from qdrant_client.models import Distance, VectorParams
+
+        mock_client.collection_exists.return_value = True
+        mock_client.get_collection.return_value.config.params.vectors = {
+            "title": VectorParams(size=384, distance=Distance.COSINE),
+            "body": VectorParams(size=384, distance=Distance.COSINE),
+        }
+        mock_client.get_collection.return_value.config.params.sparse_vectors = None
+
+        node = SearchStmt(
+            collection="col",
+            query_text="q",
+            limit=5,
+            model=None,
+            dense_vector="missing_name",
+        )
+        with pytest.raises(QQLRuntimeError, match="no dense vector named"):
+            executor.execute(node)
+
     def test_hybrid_search_forwards_search_params_to_prefetch(
         self, executor, mock_client, mock_sparse_embedder, mocker
     ):
@@ -2922,11 +2944,12 @@ class TestUpdateVectorVectorShape:
     """Gaps 12 & 13 — verify exact vector shape sent to Qdrant for named/unnamed collections."""
 
     def test_update_vector_unnamed_collection_sends_plain_list(self, executor, mock_client):
+        from qdrant_client.models import Distance, VectorParams
         from qql.ast_nodes import UpdateVectorStmt
+
         mock_client.collection_exists.return_value = True
-        # Unnamed collection: get_collection returns non-dict vectors
         info = mock_client.get_collection.return_value
-        info.config.params.vectors = [None]  # list → not a dict → unnamed
+        info.config.params.vectors = VectorParams(size=3, distance=Distance.COSINE)
 
         node = UpdateVectorStmt(collection="articles", point_id=1, vector=(0.1, 0.2, 0.3))
         executor.execute(node)

tests/test_parser.py

@@ -27,7 +27,6 @@
     OrExpr,
     QuantizationUpdate,
     QuantizationType,
-    QuantizationSearchWith,
     RecommendStmt,
     SelectStmt,
     ScrollStmt,
@@ -1122,14 +1121,6 @@ def test_with_trailing_comma(self):
         node = parse("SEARCH col SIMILAR TO 'q' LIMIT 5 WITH { hnsw_ef: 256, }")
         assert node.with_clause.hnsw_ef == 256
 
-    def test_with_quantization_unknown_key_raises(self):
-        with pytest.raises(QQLSyntaxError):
-            parse(
-                "SEARCH col SIMILAR TO 'q' LIMIT 5 "
-                "WITH { quantization: { unknown: true } }"
-            )
-
-
 class TestSparseOnlySearch:
     def test_using_sparse_sets_flag(self):
         node = parse("SEARCH col SIMILAR TO 'q' LIMIT 5 USING SPARSE")