Description

cydan33

opened on Oct 16, 2019

Issue body actions

I can't get the tagged date for some repository tags.

For example -

import git
r = git.Repo.clone_from("https://github.com/nodejs/node", "/tmp/nodegit")
t = r.tags[10]
t.object.tagged_date

Error -

~/.local/lib/python3.7/site-packages/gitdb/util.py in __getattr__(self, attr)
    251         to be created and set. Next time the same attribute is reqeusted, it is simply
    252         returned from our dict/slots. """
--> 253         self._set_cache_(attr)
    254         # will raise in case the cache was not created
    255         return object.__getattribute__(self, attr)

/tmp/GitPython/git/objects/tag.py in _set_cache_(self, attr)
     51         if attr in TagObject.__slots__:
     52             ostream = self.repo.odb.stream(self.binsha)
---> 53             lines = ostream.read().decode(defenc).splitlines()
     54
     55             _obj, hexsha = lines[0].split(" ")

UnicodeDecodeError: 'utf-8' codec can't decode byte 0xe1 in position 392: invalid continuation byte

It happens for the following repositories -
https://github.com/nodejs/node (tag "v0.1.0")
https://github.com/OpenVPN/openvpn (tag "v2.3.1")

It reproduces in the latest version.

Activity

1. Byron

   added

   acknowledged

   help wanted

   on Oct 18, 2019
2. Byron commented on Oct 18, 2019

   

   Byron

   on Oct 18, 2019

   Member

   More actions

   Thanks for posting. and for providing the information needed for reproduction! This helps a lot.

   Unfortunately, I don't see a way to fix it as UTF-8 is assumed as default encoding. It can't be overridden unless the value of sys.getdefaultencoding() is different on python startup.

   A preferable solution would be to not assume any encoding and instead work on bytes.
3. Harmon758 commented on Feb 18, 2020

   

   Harmon758

   on Feb 18, 2020

   Member

   More actions

   I'm unable to reproduce this now.
   Is this still an issue? If so, can you please provide details about your environment?
4. Harmon758

   added

   cannot reproduce

   waiting for feedback

   and removed

   help wanted

   on Feb 18, 2020
5. 3droj7 commented on Feb 20, 2020

   

   3droj7

   on Feb 20, 2020

   More actions

   Yeah it still happens, just not in the 10th tag but the 7th

   import git
   r = git.Repo.clone_from("https://github.com/nodejs/node", "/tmp/nodegit")
   t = r.tags[7]
   t.object.tagged_date
   

   ---------------------------------------------------------------------------
   UnicodeDecodeError                        Traceback (most recent call last)
   <ipython-input-5-c80fb16d948c> in <module>
   ----> 1 t.object.tagged_date
   
   /usr/local/lib/python3.7/dist-packages/gitdb/util.py in __getattr__(self, attr)
       251         to be created and set. Next time the same attribute is reqeusted, it is simply
       252         returned from our dict/slots. """
   --> 253         self._set_cache_(attr)
       254         # will raise in case the cache was not created
       255         return object.__getattribute__(self, attr)
   
   /usr/local/lib/python3.7/dist-packages/git/objects/tag.py in _set_cache_(self, attr)
        51         if attr in TagObject.__slots__:
        52             ostream = self.repo.odb.stream(self.binsha)
   ---> 53             lines = ostream.read().decode(defenc).splitlines()
        54
        55             _obj, hexsha = lines[0].split(" ")
   
   UnicodeDecodeError: 'utf-8' codec can't decode byte 0xe1 in position 392: invalid continuation byte
   
   
6. Harmon758

   removed

   cannot reproduce

   waiting for feedback

   on Feb 20, 2020
7. Harmon758

   added a commit that references this issue on Feb 20, 2020

   Replace invalid bytes when decoding TagObject stream

   This issue will close once commit a244bd1 is merged into the 'master' branch.

   a244bd1
8. Harmon758

   closed this as completedin a244bd1on Feb 20, 2020
9. Harmon758 commented on Feb 20, 2020

   

   Harmon758

   on Feb 20, 2020

   Member

   More actions

   I can't reproduce this with https://github.com/OpenVPN/openvpn/releases/tag/v2.3.1:

   >>> import git
   >>> r = git.Repo.clone_from("https://github.com/OpenVPN/openvpn", "/tmp/openvpn")
   >>> r.tags[40].object.tag
   'v2.3.1'
   >>> r.tags[40].object.tagged_date
   1364406784

   However, the annotation for https://github.com/nodejs/node/releases/tag/v0.1.0 seems to be encoded in Latin-1 (ISO-8859-1), without having an encoding header for the tag specifying that.

   Even git show fails to display the tag message properly, as it also assumes UTF-8:

   

   Although GitPython uses the filesystem's encoding rather than looking at the encoding header, that's a separate issue.
10. wip-sync

    added a commit that references this issue on Mar 7, 2020

    py-git: update to 3.1.0.

    cc95dab

Issue actions