feature: valueSearchFilter query param

allow filtering `index-value-search` by value iri and type e.g. ``` GET /api/v3/index-value-search ?valueSearchPropertyPath=rights &valueSearchFilter[sameAs]=https%3A//creativecommons.org/publicdomain/zero/1.0/legalcode ``` ``` GET /api/v3/index-value-search ?valueSearchPropertyPath=creator &valueSearchFilter[resourceType]=Person ```
aaxelb · Sep 8, 2023 · f6701b7 · f6701b7
1 parent aef9b4d
commit f6701b7
Show file tree

Hide file tree

Showing 2 changed files with 31 additions and 10 deletions.
diff --git a/share/search/index_strategy/trove_indexcard_flats.py b/share/search/index_strategy/trove_indexcard_flats.py
@@ -343,7 +343,6 @@ def pls_handle_cardsearch(self, cardsearch_params: CardsearchParams) -> Cardsear
 
         def pls_handle_valuesearch(self, valuesearch_params: ValuesearchParams) -> ValuesearchResponse:
             _cursor = _SimpleCursor.from_page_param(valuesearch_params.page)
-            _on_date_property = is_date_property(valuesearch_params.valuesearch_property_path[-1])
             try:
                 _es8_response = self.index_strategy.es8_client.search(
                     index=self.indexname,
@@ -357,7 +356,7 @@ def pls_handle_valuesearch(self, valuesearch_params: ValuesearchParams) -> Value
                     size=0,  # ignore cardsearch hits; just want the aggs
                     aggs=(
                         self._valuesearch_date_aggs(valuesearch_params)
-                        if _on_date_property
+                        if is_date_property(valuesearch_params.valuesearch_property_path[-1])
                         else self._valuesearch_iri_aggs(valuesearch_params, _cursor)
                     ),
                 )
@@ -495,7 +494,6 @@ def _cardsearch_aggs(self, cardsearch_params):
             return _aggs
 
         def _valuesearch_iri_aggs(self, valuesearch_params: ValuesearchParams, cursor: '_SimpleCursor'):
-            # TODO: valuesearch_filter_set (just rdf:type => nested_iri.value_type_iri)
             _nested_iri_bool = {
                 'filter': [{'term': {'nested_iri.suffuniq_path_from_focus': iri_path_as_keyword(
                     valuesearch_params.valuesearch_property_path,
@@ -505,6 +503,23 @@ def _valuesearch_iri_aggs(self, valuesearch_params: ValuesearchParams, cursor: '
                 'must_not': [],
                 'should': [],
             }
+            _nested_terms_agg = {
+                'field': 'nested_iri.iri_value',
+                # WARNING: terribly inefficient pagination (part one)
+                'size': cursor.start_index + cursor.page_size + 1,
+            }
+            _iris = list(valuesearch_params.valuesearch_iris())
+            if _iris:
+                _nested_iri_bool['filter'].append({'terms': {
+                    'nested_iri.iri_value': _iris,
+                }})
+                _nested_terms_agg['size'] = len(_iris)
+                _nested_terms_agg['include'] = _iris
+            _type_iris = list(valuesearch_params.valuesearch_type_iris())
+            if _type_iris:
+                _nested_iri_bool['filter'].append({'terms': {
+                    'nested_iri.value_type_iri': _type_iris,
+                }})
             _textq_builder = self._TextQueryBuilder('nested_iri.value_namelike_text')
             for _boolkey, _textquery in _textq_builder.textsegment_queries(valuesearch_params.valuesearch_textsegment_set):
                 _nested_iri_bool[_boolkey].append(_textquery)
@@ -516,11 +531,7 @@ def _valuesearch_iri_aggs(self, valuesearch_params: ValuesearchParams, cursor: '
                             'filter': {'bool': _nested_iri_bool},
                             'aggs': {
                                 'iri_values': {
-                                    'terms': {
-                                        'field': 'nested_iri.iri_value',
-                                        # WARNING: terribly inefficient pagination (part one)
-                                        'size': cursor.start_index + cursor.page_size + 1,
-                                    },
+                                    'terms': _nested_terms_agg,
                                     'aggs': {
                                         'type_iri': {'terms': {
                                             'field': 'nested_iri.value_type_iri',

diff --git a/share/search/search_params.py b/share/search/search_params.py
@@ -21,7 +21,7 @@
     OSFMAP_VOCAB,
 )
 from trove.vocab.trove import trove_labeler
-from trove.vocab.namespaces import RDF, TROVE
+from trove.vocab.namespaces import RDF, TROVE, OWL
 
 
 logger = logging.getLogger(__name__)
@@ -361,7 +361,7 @@ def from_queryparams(queryparams: QueryparamDict) -> 'ValuesearchParams':
         return ValuesearchParams(
             **CardsearchParams.parse_cardsearch_queryparams(queryparams),
             valuesearch_property_path=tuple(
-                osfmap_labeler.iri_for_label(_pathstep)
+                osfmap_labeler.iri_for_label(_pathstep, default=_pathstep)
                 for _pathstep in split_queryparam_value(_raw_property_path)
             ),
             valuesearch_text=_valuesearch_text,
@@ -379,6 +379,16 @@ def to_querydict(self):
             _querydict.appendlist(_filter.original_param_name, _filter.original_param_value),
         return _querydict
 
+    def valuesearch_iris(self):
+        for _filter in self.valuesearch_filter_set:
+            if _filter.property_path == (OWL.sameAs,) and _filter.operator == SearchFilter.FilterOperator.ANY_OF:
+                yield from _filter.value_set
+
+    def valuesearch_type_iris(self):
+        for _filter in self.valuesearch_filter_set:
+            if _filter.property_path == (RDF.type,) and _filter.operator == SearchFilter.FilterOperator.ANY_OF:
+                yield from _filter.value_set
+
 
 ###
 # local helpers