Use triagram similarity for user search

bookwyrm/migrations/0053_auto_20201014_1700.py

@@ -0,0 +1,15 @@
+# Generated by Django 3.0.7 on 2020-10-14 17:00
+
+from django.contrib.postgres.operations import TrigramExtension
+from django.db import migrations
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('bookwyrm', '0052_auto_20201005_2145'),
+    ]
+
+    operations = [
+        TrigramExtension()
+    ]

bookwyrm/views.py

@@ -2,7 +2,7 @@
 import re
 
 from django.contrib.auth.decorators import login_required, permission_required
-from django.contrib.postgres.search import SearchRank, SearchVector
+from django.contrib.postgres.search import TrigramSimilarity
 from django.db.models import Avg, Count, Q
 from django.http import HttpResponseBadRequest, HttpResponseNotFound,\
         JsonResponse
@@ -159,15 +159,11 @@ def search(request):
         outgoing.handle_remote_webfinger(query)
 
     # do a local user search
-    vector = SearchVector('localname', weight='A') + \
-             SearchVector('username', wieght='A')
     user_results = models.User.objects.annotate(
-        search=vector
-    ).annotate(
-        rank=SearchRank(vector, query)
+        similarity=TrigramSimilarity('username', query),
     ).filter(
-        rank__gt=0
-    ).order_by('-rank')[:10]
+        similarity__gt=0.1,
+    ).order_by('-similarity')[:10]
 
     book_results = books_manager.search(query)
     data = {