[EMAIL PROTECTED] wrote:
Quoting Alexandre Rossi <[EMAIL PROTECTED]>:
Hi,
But there a problem for IMG_0870_exiftool_usercomment_unicode.JPG : the
exif
is in big endian format. And it this case it seems it is UTF-16BE that
should be used.
Thanks a lot, this should be okay for your examples now.
Not really, I wasn't clear enought. The exif data could be in little or big
endian. In order to know that there is a byte order field in jpeg or exif data
(II or MM).
If the exif data is in little endian, utf-16le should be used
If the exif data is in bug endian, utf-16le should be used.
http://castet.matthieu.free.fr/tmp/test_exif/IMG_0870_exiftool_usercomment_unicode.JPG
is in big endian,
http://castet.matthieu.free.fr/tmp/test_exif/IMG_0874_exiftool_usercomment_unicode.JPG
is in little endian.
But I don't know how to get the bit order info in python.
Waiting for a cleaner solution, a quick hack could be to use
im.app['APP1'][6:8].
I attach a patch that does that.
Matthieu
diff -rN -u old-lazygal/lazygal/metadata.py new-lazygal/lazygal/metadata.py
--- old-lazygal/lazygal/metadata.py 2008-06-18 21:23:57.000000000 +0200
+++ new-lazygal/lazygal/metadata.py 2008-06-18 21:23:57.000000000 +0200
@@ -207,7 +207,15 @@
text = ret
if cset == 'Unicode':
- encoding = 'utf-16be'
+ im = Image.open(self.image_path)
+ endian = im.app['APP1'][6:8]
+ if endian == 'MM':
+ encoding = 'utf-16be'
+ elif endian == 'II':
+ encoding = 'utf-16le'
+ else:
+ return ''
+
elif cset == 'Ascii':
encoding = 'ascii'
elif cset == 'Jis':
