Fixed #8439 -- Complex combinations of Q-objects (using both conjunctions and

disjunctions) were producing incorrect SQL when nullable relations were
involved. This fixes that.


git-svn-id: http://code.djangoproject.com/svn/django/trunk@8832 bcc190cf-cafb-0310-a4f2-bffc1f526a37

django/db/models/sql/constants.py

@@ -15,7 +15,8 @@ GET_ITERATOR_CHUNK_SIZE = 100
 LOOKUP_SEP = '__'
 
 # Constants to make looking up tuple values clearer.
-# Join lists
+# Join lists (indexes into the tuples that are values in the alias_map
+# dictionary in the Query class).
 TABLE_NAME = 0
 RHS_ALIAS = 1
 JOIN_TYPE = 2

django/db/models/sql/query.py

@@ -716,7 +716,6 @@ class Query(object):
             alias = table_name
             self.table_map[alias] = [alias]
         self.alias_refcount[alias] = 1
-        #self.alias_map[alias] = None
         self.tables.append(alias)
         return alias, True
 
@@ -1188,6 +1187,8 @@ class Query(object):
                 subtree = False
             connector = AND
             for child in q_object.children:
+                if connector == OR:
+                    refcounts_before = self.alias_refcount.copy()
                 if isinstance(child, Node):
                     self.where.start_subtree(connector)
                     self.add_q(child, used_aliases)
@@ -1195,6 +1196,27 @@ class Query(object):
                 else:
                     self.add_filter(child, connector, q_object.negated,
                             can_reuse=used_aliases)
+                if connector == OR:
+                    # Aliases that were newly added or not used at all need to
+                    # be promoted to outer joins if they are nullable relations.
+                    # (they shouldn't turn the whole conditional into the empty
+                    # set just because they don't match anything).
+                    # FIXME: There's some (a lot of!) overlap with the similar
+                    # OR promotion in add_filter(). It's not quite identical,
+                    # but is very similar. So pulling out the common bits is
+                    # something for later (code smell: too much indentation
+                    # here)
+                    considered = {}
+                    for alias in self.tables:
+                        if alias not in used_aliases:
+                            continue
+                        if (alias not in refcounts_before or
+                                self.alias_refcount[alias] ==
+                                refcounts_before[alias]):
+                            parent = self.alias_map[alias][LHS_ALIAS]
+                            must_promote = considered.get(parent, False)
+                            promoted = self.promote_alias(alias, must_promote)
+                            considered[alias] = must_promote or promoted
                 connector = q_object.connector
             if q_object.negated:
                 self.where.negate()

tests/regressiontests/queries/models.py

@@ -238,6 +238,11 @@ __test__ = {'API_TESTS':"""
 >>> n2 = Note.objects.create(note='n2', misc='bar')
 >>> n3 = Note.objects.create(note='n3', misc='foo')
 
+>>> ann1 = Annotation.objects.create(name='a1', tag=t1)
+>>> ann1.notes.add(n1)
+>>> ann2 = Annotation.objects.create(name='a2', tag=t4)
+>>> ann2.notes.add(n2, n3)
+
 Create these out of order so that sorting by 'id' will be different to sorting
 by 'info'. Helps detect some problems later.
 >>> e2 = ExtraInfo.objects.create(info='e2', note=n2)
@@ -842,8 +847,6 @@ unpickling.
 True
 
 Bug #7277
->>> ann1 = Annotation.objects.create(name='a1', tag=t1)
->>> ann1.notes.add(n1)
 >>> n1.annotation_set.filter(Q(tag=t5) | Q(tag__children=t5) | Q(tag__children__children=t5))
 [<Annotation: a1>]
 
@@ -931,10 +934,25 @@ Bug #7302: reserved names are appropriately escaped
 >>> _ = ReservedName.objects.create(name='b',order=37)
 >>> ReservedName.objects.all().order_by('order')
 [<ReservedName: b>, <ReservedName: a>]
-
 >>> ReservedName.objects.extra(select={'stuff':'name'}, order_by=('order','stuff'))
 [<ReservedName: b>, <ReservedName: a>]
 
+Bug #8439 -- complex combinations of conjunctions, disjunctions and nullable
+relations.
+>>> Author.objects.filter(Q(item__note__extrainfo=e2)|Q(report=r1, name='xyz'))
+[<Author: a2>]
+>>> Author.objects.filter(Q(report=r1, name='xyz')|Q(item__note__extrainfo=e2))
+[<Author: a2>]
+>>> Annotation.objects.filter(Q(tag__parent=t1)|Q(notes__note='n1', name='a1'))
+[<Annotation: a1>]
+>>> xx = ExtraInfo.objects.create(info='xx', note=n3)
+>>> Note.objects.filter(Q(extrainfo__author=a1)|Q(extrainfo=xx))
+[<Note: n1>, <Note: n3>]
+>>> xx.delete()
+>>> q = Note.objects.filter(Q(extrainfo__author=a1)|Q(extrainfo=xx)).query
+>>> len([x[2] for x in q.alias_map.values() if x[2] == q.LOUTER and q.alias_refcount[x[1]]])
+1
+
 """}
 
 # In Python 2.3 and the Python 2.6 beta releases, exceptions raised in __len__