Fix: support combined text/query searches (#14443)

src/documents/search/_backend.py

@@ -926,6 +926,7 @@ class TantivyBackend:
         sort_field: str | None = None,
         sort_reverse: bool = False,
         search_mode: SearchMode = SearchMode.QUERY,
+        filter_query: str | None = None,
         limit: int | None = None,
     ) -> list[int]:
         """
@@ -941,6 +942,7 @@ class TantivyBackend:
             sort_field: Field to sort by (None for relevance ranking)
             sort_reverse: Whether to reverse the sort order
             search_mode: Query parsing mode (QUERY, TEXT, or TITLE)
+            filter_query: Optional advanced query results must also match
             limit: Maximum number of IDs to return (None = all matching docs)
 
         Returns:
@@ -948,6 +950,13 @@ class TantivyBackend:
         """
         self._ensure_open()
         user_query = self._parse_query(query, search_mode)
+        if filter_query is not None:
+            user_query = user_query.and_must_match(
+                tantivy.Query.const_score_query(
+                    self._parse_query(filter_query, SearchMode.QUERY),
+                    0.0,  # zero so doesn't affect relevance
+                ),
+            )
         final_query = self._apply_permission_filter(user_query, user)
 
         searcher = self._index.searcher()

src/documents/tests/test_api_bulk_edit.py

@@ -835,6 +835,7 @@ class TestBulkEditAPI(DirectoriesMixin, APITestCase):
         for filters in (
             {"text": "new doc 2017-03-16"},
             {"title_search": "apple"},
+            {"title_search": "apple", "query": "created:[-1 year to now]"},
         ):
             with self.subTest(filters=filters):
                 get_backend.return_value.search_ids.return_value = [self.doc2.id]
@@ -886,7 +887,7 @@ class TestBulkEditAPI(DirectoriesMixin, APITestCase):
                     "all": True,
                     "filters": {
                         "text": "B",
-                        "query": "c1",
+                        "title_search": "c1",
                     },
                     "method": "set_storage_path",
                     "parameters": {"storage_path": self.sp1.id},

src/documents/tests/test_api_search.py

@@ -245,8 +245,68 @@ class TestDocumentSearchApi(DirectoriesMixin, APITestCase):
         self.assertEqual(response.data["count"], 1)
         self.assertEqual(response.data["results"][0]["id"], title_match.id)
 
+    def test_simple_search_combined_with_query_filter(self) -> None:
+        """
+        GIVEN:
+            - A document with the term in its title, created this year
+            - A document with the term only in its content, created this year
+            - A document with the term in its title, created years ago
+        WHEN:
+            - A title or text search is combined with a relative created
+              date query, as the filter editor sends it
+        THEN:
+            - The simple search keeps its own field scope and the query only
+              narrows the results (issue #14436)
+        """
+        this_year = timezone.localdate().replace(month=1, day=1)
+        title_match = Document.objects.create(
+            title="foo report",
+            content="nothing relevant",
+            checksum="F1",
+            created=this_year,
+            pk=21,
+        )
+        content_match = Document.objects.create(
+            title="something else",
+            content="foo appears here",
+            checksum="F2",
+            created=this_year,
+            pk=22,
+        )
+        old_title_match = Document.objects.create(
+            title="foo archive",
+            content="nothing relevant",
+            checksum="F3",
+            created=datetime.date(2001, 1, 1),
+            pk=23,
+        )
+        backend = get_backend()
+        for doc in (title_match, content_match, old_title_match):
+            backend.add_or_update(doc)
+
+        query = "created:[-1 year to now]"
+        response = self.client.get(
+            "/api/documents/",
+            {"title_search": "foo", "query": query},
+        )
+        self.assertEqual(response.status_code, status.HTTP_200_OK)
+        self.assertEqual(
+            [r["id"] for r in response.data["results"]],
+            [title_match.id],
+        )
+
+        response = self.client.get(
+            "/api/documents/",
+            {"text": "foo", "query": query},
+        )
+        self.assertEqual(response.status_code, status.HTTP_200_OK)
+        self.assertCountEqual(
+            [r["id"] for r in response.data["results"]],
+            [title_match.id, content_match.id],
+        )
+
     def test_search_rejects_multiple_search_modes(self) -> None:
-        response = self.client.get("/api/documents/?text=bank&query=bank")
+        response = self.client.get("/api/documents/?text=bank&title_search=bank")
         self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)
         self.assertEqual(
             response.data["detail"],

src/documents/views.py

@@ -289,6 +289,7 @@ logger = logging.getLogger("paperless.api")
 # clauses efficiently, so this threshold mainly protects SQLite users.
 _TANTIVY_INTERSECT_THRESHOLD = 5_000
 _TANTIVY_SEARCH_PARAM_NAMES = ("text", "title_search", "query", "more_like_id")
+_TANTIVY_COMBINABLE_SEARCH_PARAMS = ({"text", "query"}, {"title_search", "query"})
 
 # whoosh-compat's fieldname tagger (used only for SearchMode.QUERY, via the
 # whoosh grammar in parse_user_query) is O(n^2) in plain word characters:
@@ -308,6 +309,18 @@ _TANTIVY_SEARCH_PARAM_NAMES = ("text", "title_search", "query", "more_like_id")
 _MAX_QUERY_LENGTH: Final[int] = 4096
 
 
+def _validate_tantivy_search_params(active: list[str]) -> None:
+    # may also have a filter query alongside a simple search e.g. relative dates
+    if len(active) > 1 and set(active) not in _TANTIVY_COMBINABLE_SEARCH_PARAMS:
+        raise ValidationError(
+            {
+                "detail": _(
+                    "Specify only one of text, title_search, query, or more_like_id.",
+                ),
+            },
+        )
+
+
 def _get_tantivy_query_and_mode(params):
     from documents.search import QueryTooLongError
     from documents.search import SearchMode
@@ -321,9 +334,15 @@ def _get_tantivy_query_and_mode(params):
     else:
         return None  # pragma: no cover
 
-    if len(raw) > _MAX_QUERY_LENGTH:
-        raise QueryTooLongError(len(raw), _MAX_QUERY_LENGTH)
-    return raw, mode
+    filter_query = (
+        str(params["query"])
+        if mode is not SearchMode.QUERY and "query" in params
+        else None
+    )
+    for value in (raw, filter_query):
+        if value is not None and len(value) > _MAX_QUERY_LENGTH:
+            raise QueryTooLongError(len(value), _MAX_QUERY_LENGTH)
+    return raw, mode, filter_query
 
 
 def _get_more_like_id(query_params: dict[str, Any], user: User | None) -> int:
@@ -2566,15 +2585,9 @@ class UnifiedSearchViewSet(DocumentViewSet):
 
         def parse_search_params() -> SearchParams:
             """Extract query string, search mode, and ordering from request."""
-            active = self._get_active_search_params(request)
-            if len(active) > 1:
-                raise ValidationError(
-                    {
-                        "detail": _(
-                            "Specify only one of text, title_search, query, or more_like_id.",
-                        ),
-                    },
-                )
+            _validate_tantivy_search_params(
+                self._get_active_search_params(request),
+            )
 
             ordering_param = request.query_params.get("ordering", "")
             sort_reverse = ordering_param.startswith("-")
@@ -2636,7 +2649,9 @@ class UnifiedSearchViewSet(DocumentViewSet):
             filtered_qs: QuerySet[Document],
         ) -> SearchResultPage:
             """Handle text/title/query search: IDs, ORM intersection, page highlights."""
-            query_str, search_mode = _get_tantivy_query_and_mode(request.query_params)
+            query_str, search_mode, filter_query = _get_tantivy_query_and_mode(
+                request.query_params,
+            )
 
             # "score" is not a real Tantivy sort field — it means relevance order,
             # which is Tantivy's default when no sort field is specified.
@@ -2649,6 +2664,7 @@ class UnifiedSearchViewSet(DocumentViewSet):
                 ),
                 sort_reverse=sort_reverse,
                 search_mode=search_mode,
+                filter_query=filter_query,
             )
             ordered_ids = intersect_and_order(
                 all_ids,
@@ -2883,14 +2899,7 @@ class DocumentSelectionMixin:
         ]
         if not search_filters:
             return None
-        if len(search_filters) > 1:
-            raise ValidationError(
-                {
-                    "detail": _(
-                        "Specify only one of text, title_search, query, or more_like_id.",
-                    ),
-                },
-            )
+        _validate_tantivy_search_params(search_filters)
 
         from documents.search import SearchQueryError
         from documents.search import get_backend
@@ -2909,11 +2918,14 @@ class DocumentSelectionMixin:
                     user=search_user,
                 )
             else:
-                query_str, search_mode = _get_tantivy_query_and_mode(filters)
+                query_str, search_mode, filter_query = _get_tantivy_query_and_mode(
+                    filters,
+                )
                 search_ids = backend.search_ids(
                     query_str,
                     user=search_user,
                     search_mode=search_mode,
+                    filter_query=filter_query,
                 )
         except SearchQueryError as e:
             # Same user-fixable-query mapping as the search list endpoint: