Accept string based primary keys

[fixes notanumber#138]

tests/xapian_tests/models.py

@@ -14,7 +14,7 @@ class Document(models.Model):
     text = models.TextField()
 
 
-class BlogEntry(models.Model):
+class AbstractBlogEntry(models.Model):
     """
     Same as tests.core.MockModel with a few extra fields for testing various
     sorting and ordering criteria.
@@ -34,3 +34,18 @@ class BlogEntry(models.Model):
     number = models.IntegerField()
     float_number = models.FloatField()
     decimal_number = models.DecimalField(max_digits=4, decimal_places=2)
+
+    class Meta:
+        abstract = True
+
+
+class BlogEntry(AbstractBlogEntry):
+    pass
+
+
+class UUIDBlogEntry(AbstractBlogEntry):
+    """
+    A blog entry with string based primary key instead of an integer.
+    Covers #138
+    """
+    uuid = models.CharField(primary_key=True, max_length=20)

tests/xapian_tests/search_indexes.py

@@ -82,6 +82,35 @@ def prepare_empty(self, obj):
         return ''
 
 
+class UUIDBlogSearchIndex(BlogSearchIndex):
+
+    def get_model(self):
+        return models.UUIDBlogEntry
+
+    def prepare_sites(self, obj):
+        return ['%d' % (i * int(obj.pk.split('-')[1])) for i in range(1, 4)]
+
+    def prepare_tags(self, obj):
+        if obj.pk == 'uuid-1':
+            return ['a', 'b', 'c']
+        elif obj.pk == 'uuid-2':
+            return ['ab', 'bc', 'cd']
+        else:
+            return ['an', 'to', 'or']
+
+    def prepare_keys(self, obj):
+        return [i * int(obj.pk.split('-')[1]) for i in range(1, 4)]
+
+    def prepare_titles(self, obj):
+        if obj.pk == 'uuid-1':
+            return ['object one title one', 'object one title two']
+        elif obj.pk == 'uuid-2':
+            return ['object two title one', 'object two title two']
+        else:
+            return ['object three title one', 'object three title two']
+
+
+
 class CompleteBlogEntryIndex(indexes.SearchIndex):
     text = indexes.CharField(model_attr='text', document=True)
     author = indexes.CharField(model_attr='author')

tests/xapian_tests/tests/test_backend.py

@@ -17,9 +17,8 @@
 from haystack.utils.loading import UnifiedIndex
 
 from ..search_indexes import XapianNGramIndex, XapianEdgeNGramIndex, \
-    CompleteBlogEntryIndex, BlogSearchIndex
-from ..models import BlogEntry, AnotherMockModel, MockTag
-
+    CompleteBlogEntryIndex, BlogSearchIndex, UUIDBlogSearchIndex
+from ..models import BlogEntry, AnotherMockModel, MockTag, UUIDBlogEntry
 
 XAPIAN_VERSION = [int(x) for x in xapian.__version__.split('.')]
 
@@ -681,6 +680,74 @@ def test_more_like_this_with_unindexed_model(self):
         self.assertRaises(InvalidIndexError, self.backend.more_like_this, mock)
 
 
+class StringBasedPKModelBackendFeaturesTestCase(HaystackBackendTestCase, TestCase):
+    """
+    Covers #138, Must not assume django_id is an int
+    """
+
+    def get_index(self):
+        return UUIDBlogSearchIndex()
+
+    @staticmethod
+    def get_entry(i):
+        entry = UUIDBlogEntry()
+        entry.uuid = 'uuid-%s' % i
+        entry.author = 'david%s' % i
+        entry.url = 'http://example.com/%d/' % i
+        entry.boolean = bool(i % 2)
+        entry.number = i*5
+        entry.float_number = i*5.0
+        entry.decimal_number = Decimal('22.34')
+        entry.datetime = (
+            datetime.datetime(2009, 2, 25, 1, 1, 1) - datetime.timedelta(seconds=i)
+        )
+        entry.date = datetime.date(2009, 2, 23) + datetime.timedelta(days=i)
+        return entry
+
+    def setUp(self):
+        super(StringBasedPKModelBackendFeaturesTestCase, self).setUp()
+
+        self.sample_objs = []
+
+        for i in range(1, 4):
+            entry = self.get_entry(i)
+            self.sample_objs.append(entry)
+
+        self.sample_objs[0].float_number = 834.0
+        self.sample_objs[1].float_number = 35.5
+        self.sample_objs[2].float_number = 972.0
+        for obj in self.sample_objs:
+            obj.save()
+
+        self.backend.update(self.index, UUIDBlogEntry.objects.all())
+
+    def test_update(self):
+        self.assertEqual(pks(self.backend.search(xapian.Query(''))['results']),
+                         ['uuid-1', 'uuid-2', 'uuid-3'])
+
+    def test_order_by_django_id(self):
+        """
+        We need this test because ordering on more than
+        10 entries was not correct at some point.
+        """
+        self.sample_objs = []
+        pk_list = []
+        for i in range(101, 200):
+            entry = self.get_entry(i)
+            self.sample_objs.append(entry)
+            pk_list.append('uuid-%03d' % i)
+
+        for obj in self.sample_objs:
+            obj.save()
+
+        self.backend.clear()
+        self.backend.update(self.index, self.sample_objs)
+
+        results = self.backend.search(xapian.Query(''), sort_by=['-django_id'])
+        self.assertEqual(pks(results['results']), list(reversed(pk_list)))
+
+
+
 class IndexationNGramTestCase(HaystackBackendTestCase, TestCase):
     def get_index(self):
         return XapianNGramIndex()

tests/xapian_tests/tests/test_query.py

@@ -331,6 +331,10 @@ def test_multiple_filter_types(self):
                                  'zzzzzzzzzzzzzzzzzzzzzzzzz AND'
                                  ' (QQ000000000001 OR QQ000000000002 OR QQ000000000003))')
 
+    def test_filter_string_based_django_pk(self):
+        self.sq.add_filter(SQ(django_id='uuid-1'))
+        self.assertExpectedQuery(self.sq.build_query(), 'QQuuid-1')
+
     def test_log_query(self):
         reset_search_queries()
         self.assertEqual(len(connections['default'].queries), 0)

xapian_backend.py

@@ -436,7 +436,11 @@ def add_datetime_to_document(termpos, prefix, term, weight):
                         # `django_id` is an int and `django_ct` is text;
                         # besides, they are indexed by their (unstemmed) value.
                         if field['field_name'] == 'django_id':
-                            value = int(value)
+                            try:
+                                value = int(value)
+                            except ValueError:
+                                # Django_id is a string
+                                field['type'] = 'text'
                         value = _term_to_xapian_value(value, field['type'])
 
                         document.add_term(TERM_PREFIXES[field['field_name']] + value, weight)
@@ -1499,7 +1503,12 @@ def _term_query(self, term, field_name, field_type, stemmed=True):
         if field_name in ('id', 'django_id', 'django_ct'):
             # to ensure the value is serialized correctly.
             if field_name == 'django_id':
-                term = int(term)
+                try:
+                    term = int(term)
+                except ValueError:
+                    # Django_id is a string
+                    field_type = 'text'
+
             term = _term_to_xapian_value(term, field_type)
             return xapian.Query('%s%s' % (TERM_PREFIXES[field_name], term))