fix: omit irrelevant search-result evidence/highlights

CenterForOpenScience · Feb 21, 2025 · 27c05fa · 27c05fa
1 parent c09e4f4
commit 27c05fa
Show file tree

Hide file tree

Showing 2 changed files with 27 additions and 8 deletions.
diff --git a/share/search/index_strategy/trovesearch_denorm.py b/share/search/index_strategy/trovesearch_denorm.py
@@ -564,7 +564,7 @@ def _cardsearch_handle(
             _results.append(CardsearchResult(
                 card_iri=_card_iri,
                 card_pk=_es8_hit['_id'],
-                text_match_evidence=list(self._gather_textmatch_evidence(_card_iri, _es8_hit)),
+                text_match_evidence=list(self._gather_textmatch_evidence(_card_iri, _es8_hit, cardsearch_params)),
             ))
         _relatedproperty_list: list[PropertypathUsage] = []
         if cardsearch_params.related_property_paths:
@@ -585,16 +585,20 @@ def _cardsearch_handle(
             search_params=cardsearch_params,
         )
 
-    def _gather_textmatch_evidence(self, card_iri, es8_hit) -> Iterator[TextMatchEvidence]:
+    def _gather_textmatch_evidence(self, card_iri, es8_hit, cardsearch_params) -> Iterator[TextMatchEvidence]:
         for _field, _snippets in es8_hit.get('highlight', {}).items():
             (_, _, _encoded_path) = _field.rpartition('.')
             _property_path = _parse_path_field_name(_encoded_path)
-            for _snippet in _snippets:
-                yield TextMatchEvidence(
-                    property_path=_property_path,
-                    matching_highlight=rdf.literal(_snippet),
-                    card_iri=card_iri,
-                )
+            if (  # skip highlights on non-requested text fields
+                _property_path in cardsearch_params.cardsearch_text_paths
+                or len(_property_path) in cardsearch_params.cardsearch_text_glob_depths
+            ):
+                for _snippet in _snippets:
+                    yield TextMatchEvidence(
+                        property_path=_property_path,
+                        matching_highlight=rdf.literal(_snippet),
+                        card_iri=card_iri,
+                    )
 
 
 ###

diff --git a/trove/trovesearch/search_params.py b/trove/trovesearch/search_params.py
@@ -584,6 +584,21 @@ def cardsearch_type_iris(self):
             if _filter.is_type_filter():
                 yield from _filter.value_set
 
+    @functools.cached_property
+    def cardsearch_text_paths(self) -> PropertypathSet:
+        return frozenset().union(*(
+            _textsegment.propertypath_set
+            for _textsegment in self.cardsearch_textsegment_set
+        ))
+
+    @functools.cached_property
+    def cardsearch_text_glob_depths(self) -> frozenset[int]:
+        return frozenset(
+            len(_path)
+            for _path in self.cardsearch_text_paths
+            if is_globpath(_path)
+        )
+
     def to_querydict(self) -> QueryDict:
         _querydict = super().to_querydict()
         for _qp_name, _qp_value in Textsegment.queryparams_from_textsegments('cardSearchText', self.cardsearch_textsegment_set):