Add indexed_only and quantization support to search parameters

Author: srimon12

docs/search.md

@@ -107,8 +107,7 @@ Use these when you want to debug retrieval quality or tune recall without changi
 | `WITH { hnsw_ef: 128 }` | Increase HNSW exploration at query time |
 | `WITH { exact: true }` | Force exact KNN explicitly |
 | `WITH { acorn: true }` | Enable ACORN for filtered queries |
-| `WITH { indexed_only: true }` | Restrict the query to indexed segments only |
-| `WITH { quantization: { ... } }` | Tune quantized-search behavior at query time |
+| `WITH { indexed_only: true, quantization: { rescore: true } }` | Prefer indexed vectors and apply quantization controls |
 | `WITH { mmr_diversity: 0.5, mmr_candidates: 50 }` | Apply native MMR diversification after nearest-neighbor retrieval |
 
 - `EXACT` can appear after `LIMIT` or after `RERANK`

src/qql/parser.py

@@ -26,6 +26,7 @@
     QuantizationSearchWith,
     QuantizationConfig,
     QuantizationType,
+    QuantizationSearchWith,
     RecommendStmt,
     SelectStmt,
     ScrollStmt,

tests/test_executor.py

@@ -792,7 +792,6 @@ def test_sparse_search_forwards_search_params(self, executor, mock_client, mocke
         search_params = mock_client.query_points.call_args.kwargs["search_params"]
         assert search_params.exact is True
         assert search_params.indexed_only is True
-
     def test_dense_search_against_hybrid_collection_uses_dense_vector_name(
         self, executor, mock_client, mocker
     ):

tests/test_parser.py

@@ -22,6 +22,7 @@
     NotInExpr,
     OrExpr,
     QuantizationType,
+    QuantizationSearchWith,
     RecommendStmt,
     SelectStmt,
     ScrollStmt,
@@ -916,13 +917,6 @@ def test_with_acorn(self):
         assert node.with_clause is not None
         assert node.with_clause.acorn is True
 
-    def test_with_multiple_params(self):
-        node = parse(
-            "SEARCH col SIMILAR TO 'q' LIMIT 5 WITH { hnsw_ef: 256, acorn: true }"
-        )
-        assert node.with_clause.hnsw_ef == 256
-        assert node.with_clause.acorn is True
-
     def test_with_indexed_only(self):
         node = parse("SEARCH col SIMILAR TO 'q' LIMIT 5 WITH { indexed_only: true }")
         assert node.with_clause is not None
@@ -939,6 +933,13 @@ def test_with_quantization(self):
         assert node.with_clause.quantization.rescore is False
         assert node.with_clause.quantization.oversampling == pytest.approx(2.0)
 
+    def test_with_multiple_params(self):
+        node = parse(
+            "SEARCH col SIMILAR TO 'q' LIMIT 5 WITH { hnsw_ef: 256, acorn: true }"
+        )
+        assert node.with_clause.hnsw_ef == 256
+        assert node.with_clause.acorn is True
+
     def test_with_mmr_params(self):
         node = parse(
             "SEARCH col SIMILAR TO 'q' LIMIT 5 "
@@ -986,6 +987,10 @@ def test_with_mmr_candidates_non_positive_raises(self):
         with pytest.raises(QQLSyntaxError, match="mmr_candidates must be a positive integer"):
             parse("SEARCH col SIMILAR TO 'q' LIMIT 5 WITH { mmr_candidates: 0 }")
 
+    def test_with_quantization_unknown_key_raises(self):
+        with pytest.raises(QQLSyntaxError):
+            parse("SEARCH col SIMILAR TO 'q' LIMIT 5 WITH { quantization: { unknown: true } }")
+
     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