Made SearchIndex classes thread-safe. Thanks to craigds for the report & original patch.

Author: toastdriven

haystack/indexes.py

@@ -1,5 +1,6 @@
 import copy
 import sys
+import threading
 from django.db.models import signals
 from django.utils.encoding import force_unicode
 from haystack.constants import ID, DJANGO_CT, DJANGO_ID
@@ -56,7 +57,7 @@ def __new__(cls, name, bases, attrs):
         return super(DeclarativeMetaclass, cls).__new__(cls, name, bases, attrs)
 
 
-class SearchIndex(object):
+class SearchIndex(threading.local):
     """
     Base class for building indexes.
 

tests/core/tests/indexes.py

@@ -1,4 +1,7 @@
 import datetime
+import Queue
+from threading import Thread
+import time
 from django.test import TestCase
 from haystack.indexes import *
 from core.models import MockModel, AThirdMockModel
@@ -195,6 +198,54 @@ def test_custom_prepare(self):
         self.assertEqual(len(self.cmi.full_prepare(mock)), 11)
         self.assertEqual(sorted(self.cmi.full_prepare(mock).keys()), ['author', 'author_exact', 'content', 'django_ct', 'django_id', 'extra', 'hello', 'id', 'pub_date', 'pub_date_exact', 'whee'])
 
+    def test_thread_safety(self):
+        # This is a regression. ``SearchIndex`` used to write to
+        # ``self.prepared_data``, which would leak between threads if things
+        # went too fast.
+        exceptions = []
+
+        def threaded_prepare(queue, index, model):
+            try:
+                index.queue = queue
+                prepped = index.prepare(model)
+            except Exception, e:
+                exceptions.append(e)
+                raise
+
+        class ThreadedSearchIndex(GoodMockSearchIndex):
+            def prepare_author(self, obj):
+                if obj.pk == 20:
+                    time.sleep(0.1)
+                else:
+                    time.sleep(0.5)
+
+                queue.put(self.prepared_data['author'])
+                return self.prepared_data['author']
+
+        tmi = ThreadedSearchIndex(MockModel, backend=self.msb)
+        queue = Queue.Queue()
+        mock_1 = MockModel()
+        mock_1.pk = 20
+        mock_1.author = 'foo'
+        mock_1.pub_date = datetime.datetime(2009, 1, 31, 4, 19, 0)
+        mock_2 = MockModel()
+        mock_2.pk = 21
+        mock_2.author = 'daniel%s' % mock_2.id
+        mock_2.pub_date = datetime.datetime(2009, 1, 31, 4, 19, 0)
+
+        th1 = Thread(target=threaded_prepare, args=(queue, tmi, mock_1))
+        th2 = Thread(target=threaded_prepare, args=(queue, tmi, mock_2))
+
+        th1.start()
+        th2.start()
+        th1.join()
+        th2.join()
+
+        mock_1_result = queue.get()
+        mock_2_result = queue.get()
+        self.assertEqual(mock_1_result, u'foo')
+        self.assertEqual(mock_2_result, u'daniel21')
+
     def test_custom_prepare_author(self):
         mock = MockModel()
         mock.pk = 20

tests/settings.py

@@ -16,6 +16,7 @@
     'core',
 ]
 
+SITE_ID = 1
 ROOT_URLCONF = 'core.urls'
 
 HAYSTACK_SITECONF = 'core.search_sites'