Improve efficiency and prefetching for people related people counts (#…

…1595)

geniza/corpus/models.py

@@ -5,6 +5,7 @@
 from copy import deepcopy
 from functools import cached_property
 from itertools import chain
+from time import sleep
 
 from django.conf import settings
 from django.contrib import admin, messages

geniza/entities/models.py

@@ -3,6 +3,7 @@
 from datetime import datetime
 from math import modf
 from operator import itemgetter
+from time import sleep
 
 from django.conf import settings
 from django.contrib.admin.models import CHANGE, LogEntry
@@ -475,6 +476,22 @@ def get_absolute_url(self):
         else:
             return None
 
+    @property
+    def related_people_count(self):
+        """Get a count of related people without duplicates, taking into account
+        converse relationships"""
+        # used for indexing and display
+        people_relations = (
+            self.from_person.annotate(related_id=F("from_person"))
+            .values_list("related_id", flat=True)
+            .union(
+                self.to_person.annotate(
+                    related_id=F("to_person"),
+                ).values_list("related_id", flat=True)
+            )
+        )
+        return len(set(people_relations))
+
     def related_people(self):
         """Set of all people related to this person, with relationship type and
         any notes on the relationship, taking into account converse relations"""
@@ -805,6 +822,30 @@ def items_to_index(cls):
             "names",
             "role",
             "relationships",
+            "from_person",
+            "to_person",
+            "personplacerelation_set",
+            Prefetch(
+                "persondocumentrelation_set",
+                queryset=PersonDocumentRelation.objects.select_related("type"),
+            ),
+            Prefetch(
+                "documents",
+                queryset=Document.objects.prefetch_related("dating_set"),
+            ),
+        )
+
+    @classmethod
+    def prep_index_chunk(cls, chunk):
+        """Prefetch related information when indexing in chunks
+        (modifies queryset chunk in place)"""
+        models.prefetch_related_objects(
+            chunk,
+            "names",
+            "role",
+            "relationships",
+            "from_person",
+            "to_person",
             "personplacerelation_set",
             Prefetch(
                 "persondocumentrelation_set",
@@ -815,6 +856,7 @@ def items_to_index(cls):
                 queryset=Document.objects.prefetch_related("dating_set"),
             ),
         )
+        return chunk
 
     def index_data(self):
         """data for indexing in Solr"""
@@ -830,7 +872,7 @@ def index_data(self):
                 "url_s": self.get_absolute_url(),
                 # related object counts
                 "documents_i": self.documents.count(),
-                "people_i": self.relationships.count(),
+                "people_i": self.related_people_count,
                 "places_i": self.personplacerelation_set.count(),
                 # kinds of relationships to documents
                 "document_relation_ss": list(

geniza/entities/templates/entities/snippets/person_tabs.html

@@ -2,7 +2,7 @@
 <!-- person detail page navigation -->
 {# Translators: accessibility text label for person detail view tabs navigation #}
 <nav aria-label="{% translate "tabs" %}" id="tabs">
-    {% with n_reldocs=person.documents.count n_relpeople=person.related_people|length n_relplaces=person.personplacerelation_set.count %}
+    {% with n_reldocs=person.documents.count n_relpeople=person.related_people_count n_relplaces=person.personplacerelation_set.count %}
         <ul class="tabs">
             {% url 'entities:person' slug=person.slug as person_url %}
             {% translate "Person Details" as details_text %}

geniza/entities/views.py

@@ -363,7 +363,7 @@ class PersonPeopleView(RelatedPeopleMixin, PersonDetailView):
     def page_description(self):
         """Description of a person related people page, with count"""
         obj = self.get_object()
-        count = len(obj.related_people())
+        count = obj.related_people_count
         # Translators: description of related people page, for search engines
         return ngettext(
             "%(count)d related person",