Issue 26293: Embedded zipfile fields dependent on absolute position

Example:

from zipfile import ZipFile
with open('a.zipp', 'wb') as base:
    base.write(b'old\n')
    with ZipFile(base, 'a') as myzip:
        myzip.write('eggs.txt')

If the embedded zip portion of the file is extracted (first four bytes deleted), some fields will be incorrect in the resultant file - commenting out line 3 produces a file that can serve as a comparison.  These differences cause issues opening with some zip library implementations.

My best guess is that this is related to this line: https://github.com/python/cpython/blob/master/Lib/zipfile.py#L1459

I'm not familiar with iOS development, but I'd hazard a guess that "some zip library implementations" means a (the official?) iOS zip library.

Yes, this may be a bug. After appending all offsets are absolute file positions. Here is a patch that presumably fixes this issue. Unfortunately I can't write tests for this issue. ZipFile is too lenient and all written tests are passes with unpatched code. Please test the patch manually.

spoo, can you confirm that the patch fixes your issue?

On 06/03/2016 09:27 PM, Serhiy Storchaka wrote:
> 
> Serhiy Storchaka added the comment:
> 
> spoo, can you confirm that the patch fixes your issue?
> 
> ----------
> 
> _______________________________________
> Python tracker <report@bugs.python.org>
> <http://bugs.python.org/issue26293>
> _______________________________________
> 

Yes, sorry, and thank you for looking into this!  It may take me a few
days to be able to test it.

Thank you for looking into this!  I don't have my tablet to test with at
the moment, but testing locally with zipinfo there used to be a warning
about missing bytes and now there is none.  I will test with the tablet
in a week or two when I get it back.

I confirmed this fixes the issue loading zips on an iPad.

New changeset 005704f5634f by Serhiy Storchaka in branch '3.5':
Issue #26293: Fixed writing ZIP files that starts not from the start of the
https://hg.python.org/cpython/rev/005704f5634f

New changeset b06e75ae1981 by Serhiy Storchaka in branch '3.6':
Issue #26293: Fixed writing ZIP files that starts not from the start of the
https://hg.python.org/cpython/rev/b06e75ae1981

New changeset 64a19fe3a16a by Serhiy Storchaka in branch 'default':
Issue #26293: Fixed writing ZIP files that starts not from the start of the
https://hg.python.org/cpython/rev/64a19fe3a16a

New changeset 9a99a88301ef by Serhiy Storchaka in branch '2.7':
Issue #26293: Fixed writing ZIP files that starts not from the start of the
https://hg.python.org/cpython/rev/9a99a88301ef

Thank you for your report and testing.

Unfortunately we should revert this change. It breaks other cases. See issue29094.

New changeset 3763ea865cee5bbabcce11cd577811135e0fc747 by Serhiy Storchaka in branch 'master':
Revert bpo-26293 for zipfile breakage. See also bpo-29094. (#1484)
https://github.com/python/cpython/commit/3763ea865cee5bbabcce11cd577811135e0fc747

New changeset 70dc6a7a0b7f104d87512556fca242c2ca96a010 by Serhiy Storchaka in branch '3.6':
[3.6] Revert bpo-26293 for zipfile breakage. See also bpo-29094. (GH-1484). (#1485)
https://github.com/python/cpython/commit/70dc6a7a0b7f104d87512556fca242c2ca96a010

New changeset c8faabce6ef318f3b425c6defd846e274d61e2ef by Serhiy Storchaka in branch '3.5':
[3.5] Revert bpo-26293 for zipfile breakage. See also bpo-29094. (GH-1484). (#1486)
https://github.com/python/cpython/commit/c8faabce6ef318f3b425c6defd846e274d61e2ef

resolution: fixed
stage: patch review -> resolved

keywords: + patch
stage: patch review

nosy: + twouters, alanmcintyre, serhiy.storchaka