Update .keyword -> .raw for open search queries

Author: amywieliczka

calisphere/collection_views.py

@@ -161,10 +161,10 @@ def _parse_custom_facets(self):
                         {
                             'form_name': custom_facet['facet_field'],
                             'facet_field': (
-                                f"{custom_facet['facet_field'][:-3]}.keyword"),
+                                f"{custom_facet['facet_field'][:-3]}.raw"),
                             'display_name': custom_facet['label'],
                             'filter_field': (
-                                f"{custom_facet['facet_field'][:-3]}.keyword"),
+                                f"{custom_facet['facet_field'][:-3]}.raw"),
                             'sort_by': custom_facet['sort_by'],
                             'faceting_allowed': True
                         }
@@ -310,7 +310,7 @@ def get_mosaic(self):
                 repositories.append(repository['name'])
 
         if self.index == 'es':
-            sort = ("sort_title.keyword", "asc")
+            sort = ("sort_title.raw", "asc")
         else:
             sort = ("sort_title", "asc")
 

calisphere/constants.py

@@ -36,7 +36,32 @@
 FacetDisplay = namedtuple(
     'FacetDisplay', 'facet, display')
 
-# solr schema fields that have a `_ss` varient for facets
+# index schema fields that are of type=keyword so we can get
+# facets directly without needing an _ss or .raw suffix
+UCLDC_SCHEMA_TERM_FIELDS = [
+    'calisphere-id',
+    'id',
+    'campus_name',
+    'campus_data',
+    'campus_url',
+    'campus_id',
+    'collection_name',
+    'collection_data',
+    'collection_url',
+    'collection_id',
+    'sort_collection_data',
+    'repository_name',
+    'repository_data',
+    'repository_url',
+    'repository_id',
+    'rights_uri',
+    'url_item',
+    'fetcher_type',
+    'mapper_type'
+]
+
+# index schema fields that are of type=text and thus need a 
+# solr _ss or opensearch .raw suffix to get facets
 UCLDC_SCHEMA_FACETS = [
   FacetDisplay("title", "title"),
   FacetDisplay("alternative_title", "alternative title"),
@@ -70,7 +95,7 @@
     for fd in UCLDC_SCHEMA_FACETS
 ]
 UCLDC_ES_SCHEMA_FACETS = [
-    FacetDisplayField(fd.facet, fd.display, f"{fd.facet}.keyword")
+    FacetDisplayField(fd.facet, fd.display, f"{fd.facet}.raw")
     for fd in UCLDC_SCHEMA_FACETS
 ]
 

calisphere/es_cache_retry.py

@@ -1,6 +1,7 @@
 """ logic for cache / retry for es (opensearch) and JSON from registry
 """
 
+from calisphere.constants import UCLDC_SCHEMA_TERM_FIELDS
 from future import standard_library
 from django.core.cache import cache
 from django.conf import settings
@@ -166,9 +167,9 @@ def es_mlt(item_id):
         "query": {
             "more_like_this": {
                 "fields": [
-                    "title.keyword",
+                    "title.raw",
                     "collection_data",
-                    "subject.keyword",
+                    "subject.raw",
                 ],
                 "like": [
                     {"_id": item_id}
@@ -241,33 +242,12 @@ def query_encode(query_string: str = None,
                 es_params['query'] = es_filters[0]
 
     if facets:
-        keyword_fields = [
-            'calisphere-id',
-            'id',
-            'campus_name',
-            'campus_data',
-            'campus_url',
-            'campus_id',
-            'collection_name',
-            'collection_data',
-            'collection_url',
-            'collection_id',
-            'sort_collection_data',
-            'repository_name',
-            'repository_data',
-            'repository_url',
-            'repository_id',
-            'rights_uri',
-            'url_item',
-            'fetcher_type',
-            'mapper_type'
-        ]
         aggs = {}
         for facet in facets:
-            if facet in keyword_fields or facet[-8:] == '.keyword':
+            if facet in UCLDC_SCHEMA_TERM_FIELDS or facet[-4:] == '.raw':
                 field = facet
             else:
-                field = f'{facet}.keyword'
+                field = f'{facet}.raw'
 
             aggs[facet] = {
                 "terms": {

calisphere/facet_filter_type.py

@@ -170,7 +170,7 @@ class ESRelationFF(ESFacetFilterType):
     form_name = 'relation_ss'
     facet_field = 'relation'
     display_name = 'Relation'
-    filter_field = 'relation.keyword'
+    filter_field = 'relation.raw'
     sort_by = 'value'
     faceting_allowed = False
 
@@ -186,7 +186,7 @@ class ESTypeFF(ESFacetFilterType):
     form_name = 'type_ss'
     facet_field = 'type'
     display_name = 'Type of Item'
-    filter_field = 'type.keyword'
+    filter_field = 'type.raw'
 
 
 class DecadeFF(FacetFilterType):
@@ -201,7 +201,7 @@ class ESDecadeFF(ESFacetFilterType):
     form_name = 'facet_decade'
     facet_field = 'date'
     display_name = 'Decade'
-    filter_field = 'date.keyword'
+    filter_field = 'date.raw'
     sort_by = 'value'
 
 

calisphere/search_form.py

@@ -10,15 +10,15 @@ def solr_escape(text):
 
 class SortField(object):
     default = 'relevance'
-    no_keyword = 'a'
+    default_without_query_string = 'a'
 
     def __init__(self, request):
         if (request.get('q')
            or request.getlist('rq')
            or request.getlist('fq')):
             self.sort = request.get('sort', self.default)
         else:
-            self.sort = request.get('sort', self.no_keyword)
+            self.sort = request.get('sort', self.default_without_query_string)
 
 
 class SearchForm(object):
@@ -389,7 +389,7 @@ class ESCampusCarouselForm(CampusCarouselForm):
 
 class AltSortField(SortField):
     default = 'oldest-end'
-    no_keyword = 'oldest-end'
+    default_without_query_string = 'oldest-end'
 
 
 class CollectionFacetValueForm(CollectionForm):