identify does not correctly detect exif encoding

[suweller]

I have the following version of identify:

Code: Select all

$ identify --version
Version: ImageMagick 6.5.1-0 2009-08-27 Q16 OpenMP http://www.imagemagick.org
Copyright: Copyright (C) 1999-2009 ImageMagick Studio LLC

when I execute:

Code: Select all

$ identify -format "%[IPTC:2:25]" ~/Desktop/large_2010_4_430069-full.jpeg
Blume;Bl�te;Fr�hling;Sommer;Bl�tter

Just in case it will correctly display this in IE or windows browsers, special characters are replaced with a black "<?>" character

If someone wants to reproduce this, just download my avatar and run the identify command as mentioned above on it.

[magick]

Post a URL to your image. We tried your avatar but it has an EXIF profile but does not include an IPTC profile.

[suweller]

You are quite correct. I have uploaded the original image here.

Also, I should have named this topic 'identify does not correctly detect IPTC character encoding'

Thanks for the quick response.

[magick]

Can you be specific on when it is failing to display properly? We tried your command on on Linux box and the accent characters displayed properly.

[suweller]

That is strange... I ran the command on linux too...
I opened a gnome terminal on ubuntu 9.10, ran the command and it failed to display special characters.

What kind of linux box did you run it on?

[magick]

We're running CentOS 5.4.

[snibgo]

IM is emitting the strings as character set ISO-8859-1. They appear to be in the file as both ISO-8859-1 and Unicode. (I don't know where IM gets the data from.)

When I set a Ubuntu 9.04 terminal as character set ISO-8859-1, I get characters with umlauts.

[suweller]

Hi,

Thank you for your help. It does seem to be terminal/system settings.

ImageMagick returns IPTC data as latin-1 but my system is set to utf8.
The code below displays the image IPTC data correctly.

Code: Select all

identify -format "%[IPTC:2:25]" ~/Desktop/large_2010_4_430069-full.jpeg | iconv -f latin1 -t utf8 -

Is this a problem of my system which should have a setting somewhere stating which encoding it uses or does ImageMagick simply ignore my terminal character encoding?