classification
Title: zipfile does not write empty ZIP structure if close() called after __init__() as doc suggests
Type: behavior Stage:
Components: Documentation, Library (Lib) Versions: Python 2.6
process
Status: closed Resolution: out of date
Dependencies: Superseder:
Assigned To: docs@python Nosy List: Ian.Stevens, docs@python, georg.brandl
Priority: normal Keywords:

Created on 2010-12-21 16:30 by Ian.Stevens, last changed 2010-12-21 17:58 by georg.brandl. This issue is now closed.

Messages (4)
msg124433 - (view) Author: Ian Stevens (Ian.Stevens) Date: 2010-12-21 16:30
The zipfile documentation (http://docs.python.org/library/zipfile.html) states:

"If the file is created with mode 'a' or 'w' and then close()d without adding any files to the archive, the appropriate ZIP structures for an empty archive will be written to the file."

This is not the case, eg.::

    >>> from StringIO import StringIO
    >>> import zipfile
    >>> s = StringIO()
    >>> z = zipfile.ZipFile(s, 'w')
    >>> z.close()
    >>> s.len
    0

The code for zipfile (http://svn.python.org/projects/python/trunk/Lib/zipfile.py) does not support the documentation either. The ending records are written only if ZipFile._didModify is True, and that attribute is only set to True if writestr() or write() are called.

Either the code should be fixed to support writing the ending records on an empty zip, or the documentation should be changed to reflect the existing behaviour.

Test case (for Lib/test/test_zipfile)::

    def test_close_empty_zip_creates_valid_zip(self):
        # Test that close() called on a ZipFile without write creates a valid ZIP.
        zf = zipfile.ZipFile(TESTFN, "w")
        zf.close()
        chk = zipfile.is_zipfile(TESTFN)
        self.assertTrue(chk)
msg124435 - (view) Author: Georg Brandl (georg.brandl) * (Python committer) Date: 2010-12-21 16:48
This has been fixed in Python 2.7.1 (which the online docs refer to).  I assume that you're using 2.6 or an earlier version.

As for the code in SVN, the "trunk" is currently not in use; development happens in the "release27-maint", "release31-maint" and "py3k" branches (the latter being the actual trunk of development).  This irritation will go away soon once we have migrated to Hg.
msg124436 - (view) Author: Ian Stevens (Ian.Stevens) Date: 2010-12-21 16:59
Yes, I'm using 2.6. 

If this is not the expected behaviour in 2.6, the doc should reflect that with a "New in version 2.7" note.
msg124440 - (view) Author: Georg Brandl (georg.brandl) * (Python committer) Date: 2010-12-21 17:58
We usually don't do this for bugfixes, but here it makes sense I guess. r87414.
History
Date User Action Args
2010-12-21 17:58:17georg.brandlsetnosy: georg.brandl, docs@python, Ian.Stevens
messages: + msg124440
2010-12-21 16:59:25Ian.Stevenssetnosy: georg.brandl, docs@python, Ian.Stevens
messages: + msg124436
2010-12-21 16:48:15georg.brandlsetstatus: open -> closed

nosy: + georg.brandl
messages: + msg124435

resolution: out of date
2010-12-21 16:30:24Ian.Stevenscreate