Added --schema-only flag to update_index command

CHANGELOG.txt

@@ -18,6 +18,7 @@ Changelog
  * Increased size of Save button on site settings (Liam Brenner)
  * Optimised Site.find_for_request to only perform one database query (Matthew Downey)
  * Notification messages on creating / editing sites now include the site name if specified (Chris Rogers)
+ * Added ``--schema-only`` option to ``update_index`` management command
  * Fix: The currently selected day is now highlighted only in the correct month in date pickers (Jonas Lergell)
  * Fix: Fixed crash when an image without a source file was resized with the "dynamic serve view"
  * Fix: Registered settings admin menu items now show active correctly (Matthew Downey)

docs/reference/management_commands.rst

@@ -76,6 +76,17 @@ For example, to update just the default backend:
     python manage.py update_index --backend default
 
 
+Indexing the schema only
+````````````````````````
+
+.. versionadded:: 1.5
+
+You can prevent the ``update_index`` command from indexing any data by using the ``--schema-only`` option:
+
+.. code-block:: sh
+
+    python manage.py update_index --schema-only
+
 
 .. _search_garbage_collect:
 

docs/releases/1.5.rst

@@ -38,6 +38,7 @@ Minor features
  * Increased size of Save button on site settings (Liam Brenner)
  * Optimised Site.find_for_request to only perform one database query (Matthew Downey)
  * Notification messages on creating / editing sites now include the site name if specified (Chris Rogers)
+ * Added ``--schema-only`` option to ``update_index`` management command
 
 Bug fixes
 ~~~~~~~~~

wagtail/wagtailsearch/management/commands/update_index.py

@@ -9,7 +9,7 @@ from wagtail.wagtailsearch.index import get_indexed_models
 
 
 class Command(BaseCommand):
-    def update_backend(self, backend_name):
+    def update_backend(self, backend_name, schema_only=False):
         # Print info
         self.stdout.write("Updating backend: " + backend_name)
 
@@ -36,13 +36,15 @@ class Command(BaseCommand):
             # Add model
             index.add_model(model)
 
-            # Add items (1000 at a time)
-            count = 0
-            for chunk in self.print_iter_progress(self.queryset_chunks(model.get_indexed_objects())):
-                index.add_items(model, chunk)
-                count += len(chunk)
+            # Index objects
+            object_count = 0
+            if not schema_only:
+                # Add items (1000 at a time)
+                for chunk in self.print_iter_progress(self.queryset_chunks(model.get_indexed_objects())):
+                    index.add_items(model, chunk)
+                    object_count += len(chunk)
 
-            self.stdout.write("(indexed %d objects)" % count)
+            self.stdout.write("(indexed %d objects)" % object_count)
             self.print_newline()
 
         # Finish rebuild
@@ -53,6 +55,9 @@ class Command(BaseCommand):
         parser.add_argument(
             '--backend', action='store', dest='backend_name', default=None,
             help="Specify a backend to update")
+        parser.add_argument(
+            '--schema-only', action='store', dest='schema_only', default=None,
+            help="Prevents loading any data into the index")
 
     def handle(self, **options):
         # Get list of backends to index
@@ -68,7 +73,7 @@ class Command(BaseCommand):
 
         # Update backends
         for backend_name in backend_names:
-            self.update_backend(backend_name)
+            self.update_backend(backend_name, schema_only=options['schema_only'])
 
     def print_newline(self):
         self.stdout.write('')

wagtail/wagtailsearch/tests/test_elasticsearch_backend.py

@@ -4,11 +4,15 @@ from __future__ import absolute_import, unicode_literals
 import datetime
 import json
 import os
+import time
 import unittest
 
 import mock
+
+from django.core import management
 from django.db.models import Q
 from django.test import TestCase
+from django.utils.six import StringIO
 from elasticsearch.serializer import JSONSerializer
 
 from wagtail.tests.search import models
@@ -224,6 +228,27 @@ class TestElasticSearchBackend(BackendTests, TestCase):
         results = self.backend.search("Hello World", models.SearchTest, operator='and', fields=['title'])
         self.assertEqual(list(results), [a])
 
+    def test_update_index_command_schema_only(self):
+        # Reset the index, this should clear out the index
+        self.backend.reset_index()
+
+        # Give Elasticsearch some time to catch up...
+        time.sleep(1)
+
+        results = self.backend.search(None, models.SearchTest)
+        self.assertEqual(set(results), set())
+
+        # Run update_index command
+        with self.ignore_deprecation_warnings():
+            # ignore any DeprecationWarnings thrown by models with old-style indexed_fields definitions
+            management.call_command(
+                'update_index', backend_name=self.backend_name, schema_only=True, interactive=False, stdout=StringIO()
+            )
+
+        # Unlike the test_update_index_command test. This should not give any results
+        results = self.backend.search(None, models.SearchTest)
+        self.assertEqual(set(results), set())
+
 
 class TestElasticSearchQuery(TestCase):
     def assertDictEqual(self, a, b):