__future__ imports fail when compiling from python ast

[talljosh]

BPO	14378
Nosy	@brettcannon, @birkenfeld, @ncoghlan, @benjaminp, @bitdancer
Files	futureimport.py: Python code demonstrating the issue

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = None
closed_at = <Date 2012-03-22.12:58:03.865>
created_at = <Date 2012-03-21.06:19:41.889>
labels = ['interpreter-core', 'type-bug', 'release-blocker']
title = '__future__ imports fail when compiling from python ast'
updated_at = <Date 2012-03-22.12:58:47.207>
user = 'https://bugs.python.org/talljosh'

bugs.python.org fields:

activity = <Date 2012-03-22.12:58:47.207>
actor = 'benjamin.peterson'
assignee = 'none'
closed = True
closed_date = <Date 2012-03-22.12:58:03.865>
closer = 'python-dev'
components = ['Interpreter Core']
creation = <Date 2012-03-21.06:19:41.889>
creator = 'talljosh'
dependencies = []
files = ['24981']
hgrepos = []
issue_num = 14378
keywords = []
message_count = 6.0
messages = ['156477', '156478', '156548', '156554', '156555', '156556']
nosy_count = 7.0
nosy_names = ['brett.cannon', 'georg.brandl', 'ncoghlan', 'benjamin.peterson', 'r.david.murray', 'python-dev', 'talljosh']
pr_nums = []
priority = 'release blocker'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'behavior'
url = 'https://bugs.python.org/issue14378'
versions = ['Python 2.7', 'Python 3.2']

[talljosh]

GOAL
I am trying to compile an AST which contains an ImportFrom node which performs a __future__ import. The ImportFrom node in question is the first node within the AST's body (as it should be, because __future__ imports must occur at the beginning of modules).

ISSUE
Compiling the AST fails with the error "SyntaxError: from __future__ imports must occur at the beginning of the file".

ANALYSIS
The future_parse() function in future.c looks for __future__ imports and identifies them based on the condition (ds->v.ImportFrom.module == future) where future is constructed using PyString_InternFromString("__future__"). That is, the module attribute of the ImportFrom node is compared by identity with the interned string "__future__".

The AST which I was compiling used the string "__future__" after extracting that string from a much longer string of user input, and as a result, the string was not interned.

The attached file futureimport.py is a simple script demonstrating this issue. I have confirmed that the issue occurs in Python 2.7.2 and 3.2.2.

[talljosh]

Incidentally, the workaround that I'm using for the time being is to run the following code before attempting to compile root_node.

for node in ast.walk(root_node):
    if isinstance(node, ast.ImportFrom) and node.module == '__future__':
        node.module = '__future__'

[python-dev]

New changeset f57cbcefde34 by Benjamin Peterson in branch '3.2':
check by equality for __future__ not identity (closes bpo-14378)
http://hg.python.org/cpython/rev/f57cbcefde34

New changeset 9d793be3b4eb by Benjamin Peterson in branch 'default':
merge 3.2 (bpo-14378)
http://hg.python.org/cpython/rev/9d793be3b4eb

[bitdancer]

After this commit the buildbots are dying randomly with segfaults.

[python-dev]

New changeset 1729ec440bb6 by Benjamin Peterson in branch '2.7':
check by equality for __future__ not identity (closes bpo-14378)
http://hg.python.org/cpython/rev/1729ec440bb6

[benjaminp]

1b467efb9b27