Fixed #10132 -- Corrected the interaction of extra() queries with the values() clause. Thanks to Glen Maynard for the report.

git-svn-id: http://code.djangoproject.com/svn/django/trunk@9838 bcc190cf-cafb-0310-a4f2-bffc1f526a37
This commit is contained in:
Russell Keith-Magee 2009-02-16 12:29:31 +00:00
parent 58ea6d4561
commit fb64ea7896
3 changed files with 28 additions and 14 deletions

View File

@ -698,7 +698,7 @@ class QuerySet(object):
Prepare the query for computing a result that contains aggregate annotations. Prepare the query for computing a result that contains aggregate annotations.
""" """
opts = self.model._meta opts = self.model._meta
if not self.query.group_by: if self.query.group_by is None:
field_names = [f.attname for f in opts.fields] field_names = [f.attname for f in opts.fields]
self.query.add_fields(field_names, False) self.query.add_fields(field_names, False)
self.query.set_group_by() self.query.set_group_by()
@ -736,6 +736,7 @@ class ValuesQuerySet(QuerySet):
aggregate_names = self.query.aggregate_select.keys() aggregate_names = self.query.aggregate_select.keys()
names = extra_names + field_names + aggregate_names names = extra_names + field_names + aggregate_names
for row in self.query.results_iter(): for row in self.query.results_iter():
yield dict(zip(names, row)) yield dict(zip(names, row))

View File

@ -68,7 +68,7 @@ class BaseQuery(object):
self.tables = [] # Aliases in the order they are created. self.tables = [] # Aliases in the order they are created.
self.where = where() self.where = where()
self.where_class = where self.where_class = where
self.group_by = [] self.group_by = None
self.having = where() self.having = where()
self.order_by = [] self.order_by = []
self.low_mark, self.high_mark = 0, None # Used for offset/limit self.low_mark, self.high_mark = 0, None # Used for offset/limit
@ -177,7 +177,10 @@ class BaseQuery(object):
obj.tables = self.tables[:] obj.tables = self.tables[:]
obj.where = deepcopy(self.where) obj.where = deepcopy(self.where)
obj.where_class = self.where_class obj.where_class = self.where_class
obj.group_by = self.group_by[:] if self.group_by is None:
obj.group_by = None
else:
obj.group_by = self.group_by[:]
obj.having = deepcopy(self.having) obj.having = deepcopy(self.having)
obj.order_by = self.order_by[:] obj.order_by = self.order_by[:]
obj.low_mark, obj.high_mark = self.low_mark, self.high_mark obj.low_mark, obj.high_mark = self.low_mark, self.high_mark
@ -272,7 +275,7 @@ class BaseQuery(object):
# If there is a group by clause, aggregating does not add useful # If there is a group by clause, aggregating does not add useful
# information but retrieves only the first row. Aggregate # information but retrieves only the first row. Aggregate
# over the subquery instead. # over the subquery instead.
if self.group_by: if self.group_by is not None:
from subqueries import AggregateQuery from subqueries import AggregateQuery
query = AggregateQuery(self.model, self.connection) query = AggregateQuery(self.model, self.connection)
@ -379,8 +382,8 @@ class BaseQuery(object):
result.append('AND') result.append('AND')
result.append(' AND '.join(self.extra_where)) result.append(' AND '.join(self.extra_where))
if self.group_by: grouping = self.get_grouping()
grouping = self.get_grouping() if grouping:
if ordering: if ordering:
# If the backend can't group by PK (i.e., any database # If the backend can't group by PK (i.e., any database
# other than MySQL), then any fields mentioned in the # other than MySQL), then any fields mentioned in the
@ -698,13 +701,15 @@ class BaseQuery(object):
""" """
qn = self.quote_name_unless_alias qn = self.quote_name_unless_alias
result = [] result = []
for col in self.group_by + self.related_select_cols: if self.group_by is not None:
if isinstance(col, (list, tuple)): group_by = self.group_by or []
result.append('%s.%s' % (qn(col[0]), qn(col[1]))) for col in group_by + self.related_select_cols + self.extra_select.keys():
elif hasattr(col, 'as_sql'): if isinstance(col, (list, tuple)):
result.append(col.as_sql(qn)) result.append('%s.%s' % (qn(col[0]), qn(col[1])))
else: elif hasattr(col, 'as_sql'):
result.append(str(col)) result.append(col.as_sql(qn))
else:
result.append(str(col))
return result return result
def get_ordering(self): def get_ordering(self):
@ -1224,7 +1229,7 @@ class BaseQuery(object):
# Aggregate references a normal field # Aggregate references a normal field
field_name = field_list[0] field_name = field_list[0]
source = opts.get_field(field_name) source = opts.get_field(field_name)
if not (self.group_by and is_summary): if not (self.group_by is not None and is_summary):
# Only use a column alias if this is a # Only use a column alias if this is a
# standalone aggregate, or an annotation # standalone aggregate, or an annotation
col = (opts.db_table, source.column) col = (opts.db_table, source.column)
@ -1816,6 +1821,7 @@ class BaseQuery(object):
primary key, and the query would be equivalent, the optimization primary key, and the query would be equivalent, the optimization
will be made automatically. will be made automatically.
""" """
self.group_by = []
if self.connection.features.allows_group_by_pk: if self.connection.features.allows_group_by_pk:
if len(self.select) == len(self.model._meta.fields): if len(self.select) == len(self.model._meta.fields):
self.group_by.append('.'.join([self.model._meta.db_table, self.group_by.append('.'.join([self.model._meta.db_table,

View File

@ -200,6 +200,13 @@ FieldError: Cannot resolve keyword 'foo' into field. Choices are: authors, conta
>>> sorted([(b.name, b.authors__age__avg, b.publisher.name, b.contact.name) for b in books]) >>> sorted([(b.name, b.authors__age__avg, b.publisher.name, b.contact.name) for b in books])
[(u'Artificial Intelligence: A Modern Approach', 51.5, u'Prentice Hall', u'Peter Norvig'), (u'Practical Django Projects', 29.0, u'Apress', u'James Bennett'), (u'Python Web Development with Django', 30.3..., u'Prentice Hall', u'Jeffrey Forcier'), (u'Sams Teach Yourself Django in 24 Hours', 45.0, u'Sams', u'Brad Dayley')] [(u'Artificial Intelligence: A Modern Approach', 51.5, u'Prentice Hall', u'Peter Norvig'), (u'Practical Django Projects', 29.0, u'Apress', u'James Bennett'), (u'Python Web Development with Django', 30.3..., u'Prentice Hall', u'Jeffrey Forcier'), (u'Sams Teach Yourself Django in 24 Hours', 45.0, u'Sams', u'Brad Dayley')]
# Regression for #10132 - If the values() clause only mentioned extra(select=) columns, those columns are used for grouping
>>> Book.objects.extra(select={'pub':'publisher_id'}).values('pub').annotate(Count('id')).order_by('pub')
[{'pub': 1, 'id__count': 2}, {'pub': 2, 'id__count': 1}, {'pub': 3, 'id__count': 2}, {'pub': 4, 'id__count': 1}]
>>> Book.objects.extra(select={'pub':'publisher_id','foo':'pages'}).values('pub').annotate(Count('id')).order_by('pub')
[{'pub': 1, 'id__count': 2}, {'pub': 2, 'id__count': 1}, {'pub': 3, 'id__count': 2}, {'pub': 4, 'id__count': 1}]
# Regression for #10199 - Aggregate calls clone the original query so the original query can still be used # Regression for #10199 - Aggregate calls clone the original query so the original query can still be used
>>> books = Book.objects.all() >>> books = Book.objects.all()
>>> _ = books.aggregate(Avg('authors__age')) >>> _ = books.aggregate(Avg('authors__age'))