pyexiv2

Pb setting IPTC Tags

Bug #363873 reported by snowgoon88 on 2009-04-19

This bug affects 4 people

Affects		Status	Importance	Assigned to	Milestone
	Phatch	Triaged	Medium	Unassigned
	pyexiv2	Fix Released	High	Olivier Tilloy	pyexiv2 0.2 "Commuting"

Bug Description

When setting a "string" IPTC Tags with 'str', the cached value becomes a tuple. It is also the case with some multi-valued IPTC tag (like 'Iptc.Application2.Keywords') when set with one single str value.
But when the image is read again, it has its correct value.
I thought it was relevant to bug #183332 (https://bugs.launchpad.net/pyexiv2/+bug/183332) but I checked:
- the bug's fixe (fix N°95 in bzr) is in my version of pyexiv2 (debian 0.1.2-4+b1)
- even with pyexiv 0.1.3 (with exiv2 0.18) the "bug" is still there.

In [1]: import pyexiv2
In [2]: im = pyexiv2.Image("Images/p20080921_105313.jpg")
In [3]: im.readMetadata()
In [4]: im['Iptc.Application2.City']
Out[4]: 'coucou'
In [5]: im['Iptc.Application2.City'] = 'beep'
In [6]: im['Iptc.Application2.City']
Out[6]: ('b', 'e', 'e', 'p')

In [7]: im.writeMetadata()
In [8]: im.readMetadata()
In [9]: im['Iptc.Application2.City']
Out[9]: ('b', 'e', 'e', 'p')

In [10]: im = pyexiv2.Image("Images/p20080921_105313.jpg")
In [11]: im.readMetadata()
In [12]: im['Iptc.Application2.City']
Out[12]: 'beep'

Related branches

lp://staging/pyexiv2/0.2.x

Revision history for this message

Olivier Tilloy (osomon) wrote on 2009-04-20:

Thanks for the report.
This bug should be fixed in the 0.2 branch.

Changed in pyexiv2:
assignee:	nobody → Olivier Tilloy (osomon)
importance:	Undecided → High
milestone:	none → 0.2
status:	New → Confirmed

Revision history for this message

Olivier Tilloy (osomon) wrote on 2009-10-26:

Fix committed with revision 234 of the 0.2 branch.
Trying to set the values of an IPTC tag to something else than a list or a tuple now raises a TypeError.

Changed in pyexiv2:
status:	Confirmed → Fix Committed

Revision history for this message

Andrea Ambu (andreambu) wrote on 2010-02-15:

hypothetical patch for 0.1.x series of bug #363873 Edit (1.9 KiB, text/plain)

Hi Olivier,

does it require any particular effort to fix this for series 0.1.x too?
Ubuntu is actually shipping 0.1.3-3build1 and an updated of the current shipped library would be very nice to see.
I tried to play a little with and come up deleting two lines. On the bottom there is the diff, it doesn't seem to break the unittest, anyway that test must be there for a reason, what would not work in this way?

Andrea

=== modified file 'src/pyexiv2.py'
--- src/pyexiv2.py 2009-03-14 21:43:49 +0000
+++ src/pyexiv2.py 2010-02-15 18:32:50 +0000
@@ -654,8 +654,6 @@
      except KeyError:
       pass
    if len(newValues) > 0:
- if len(newValues) == 1:
- newValues = newValues[0]
     self.__iptcTagsDict[key] = tuple([ConvertToPythonType(tagFamily, typeName, str(v)) for v in newValues])
    else:
     if self.__iptcTagsDict.has_key(key):

Revision history for this message

Olivier Tilloy (osomon) wrote on 2010-02-16:

Hi Andrea,
The patch seems to make sense from what I can read, and I can't recall why this test is here in the first place.
It doesn't break the unit tests, but on the other hand unit tests don't really check thoroughly this piece of code.
I'm not against merging the patch, but I'm concerned about potential regressions we cannot spot right now.
I'm currently focusing my efforts on putting the finishing touch to the 0.2 branch to release it as soon as possible, in the hope that I can drop maintenance of the 0.1 branch which proved to be buggy in various places and difficult to maintain.
If you are willing to do some more testing, I certainly wouldn't mind merging the patch, and we can then propose it for inclusion in the debian/ubuntu packages.

Stani (stani) on 2010-02-25

Changed in phatch:
assignee:	nobody → Andrea Ambu (andreambu)
importance:	Undecided → Medium
status:	New → In Progress

Revision history for this message

Andrea Ambu (andreambu) wrote on 2010-02-25:

Well we are actually using the library in Phatch. When debugging I came up to that check. For each test I did it seems to work. If you think of some specific test just tell me and I'll try to test it.

Andrea

Olivier Tilloy (osomon) on 2010-03-25