Add option to use collapse & expand query parser for solr grouping

Author: ffont

utils/search/backends/solr555pysolr.py

@@ -676,21 +676,34 @@ def search_sounds(self, textual_query='', query_fields=None, query_filter='', fi
 
         # Configure grouping
         if group_by_pack:
-            query.set_group_field(group_field="pack_grouping" if not similar_to else "pack_grouping_child")  # We name the fields differently to avoid solr conflicts with matches of both child and parent docs
-            query.set_group_options(
-                group_func=None,
-                group_query=None,
-                group_rows=10,  # TODO: if limit is lower than rows and start=0, this should probably be equal to limit
-                group_start=0,
-                group_limit=num_sounds_per_pack_group,  # This is the number of documents that will be returned for each group.
-                group_offset=0,
-                group_sort=None,
-                group_sort_ingroup=None,
-                group_format='grouped',
-                group_main=False,
-                group_num_groups=True,
-                group_cache_percent=0,
-                group_truncate=group_counts_as_one_in_facets)
+            use_collapse_expand_query_parser = True
+            if use_collapse_expand_query_parser:
+                current_filter = query.params.get('fq', '')
+                field_name = "pack_grouping" if not similar_to else "pack_grouping_child"
+                group_by_pack_filter = f'{{!collapse field={field_name}}}'
+                if current_filter:
+                    query.params['fq'] = [current_filter, group_by_pack_filter]
+                else:
+                    query.params['fq'] = [group_by_pack_filter]
+                query.params['fl'] = query.params['fl'] + f',{field_name}'
+                query.params['expand'] = True
+                query.params['expand.rows'] = 0
+            else:
+                query.set_group_field(group_field="pack_grouping" if not similar_to else "pack_grouping_child")  
+                query.set_group_options(
+                    group_func=None,
+                    group_query=None,
+                    group_rows=10,  # TODO: if limit is lower than rows and start=0, this should probably be equal to limit
+                    group_start=0,
+                    group_limit=num_sounds_per_pack_group,  # This is the number of documents that will be returned for each group.
+                    group_offset=0,
+                    group_sort=None,
+                    group_sort_ingroup=None,
+                    group_format='grouped',
+                    group_main=False,
+                    group_num_groups=True,
+                    group_cache_percent=0,
+                    group_truncate=group_counts_as_one_in_facets)
 
         # Do the query!
         # Note: we create a SearchResults with the same members as SolrResponseInterpreter (the response from .search()).

utils/search/backends/solr_common.py

@@ -294,6 +294,9 @@ def make_solr_query_url(solr_query_params, debug=False):
 
 class SolrResponseInterpreter:
     def __init__(self, response, next_page_query=None):
+        if "grouped" in response and "expanded" in response:
+            raise SearchEngineException("Response contains both grouped and expanded results, this is not supported")
+        
         if "grouped" in response:
             grouping_field = list(response["grouped"].keys())[0]            
             self.docs = [{
@@ -307,6 +310,19 @@ def __init__(self, response, next_page_query=None):
             self.num_rows = len(self.docs)
             self.num_found = response["grouped"][grouping_field]["ngroups"]
             self.non_grouped_number_of_results = response["grouped"][grouping_field]["matches"]
+        elif "expanded" in response:
+            collapse_field = 'pack_grouping_child' if 'pack_grouping_child' in response['responseHeader']['params']['fl'] else 'pack_grouping'
+            self.docs = [{ 
+                'id': doc['id'],
+                'score': doc['score'],
+                'n_more_in_group': response['expanded'][doc[collapse_field]]['numFound'] if doc[collapse_field] in response['expanded'] else 0,
+                'group_docs': doc,
+                'group_name': doc[collapse_field]
+            } for doc in response["response"]["docs"]]
+            self.start = int(response["response"]["start"])
+            self.num_rows = len(self.docs)
+            self.num_found = response["response"]["numFound"]  # Note that using "Collapse and Expand" for grouping we can't get the total number of uncollapsed matches
+            self.non_grouped_number_of_results = -1
         else:
             self.docs = response["response"]["docs"]
             self.start = int(response["response"]["start"])