photo / frontend

The official @github repository of the Trovebox frontend software. A photo sharing and photo management web interface for data stored "in the cloud" (i.e. Amazon S3, Rackspace CloudFiles, Google Storage).
https://trovebox.com
Apache License 2.0
1.38k stars 244 forks source link

tags extracted from exif metadata are corrupted if utf8 #1338

Open ajft opened 11 years ago

ajft commented 11 years ago

I thought this had been logged (and fixed) ages ago but found out today that the problem is still present.

Importing photos with non-ascii characters in the EXIF "keywords" field gets them imported into the "tags" of the photos, but only up until the first non-ascii character, after that all following tags are ignored. You can still enter non-ascii characters as tags through the web interface, but it is very laborious to visit every photo and fix up the tags broken on import.

eg $ exiftool -Keywords 837-630_02.JPG Keywords : cycling, flag, FRA, France, geo:lat=45.08716112, geo:lon=6.05768612, wor2001b2a, geotagged, Huez, l'Alpe d'Huez, Rhône-Alpes, TDF, 837-630, 837-630_02, APS-film, scanned

Then see http://openphoto.ajft.org/p/5gu/view, the tags are:

837-630, 837-630_02 2001 APS-film cycling flag FRA France geo:lat=45.08716112 geo:lon=6.05768612 geotagged Huez July l'Alpe d'Huez Rh

Note that the last tag is "Rh" which is the first two characters from "Rhône-Alpes"

hfiguiere commented 11 years ago

Exiftool might just be doing the right thing. Do you a link to the original file?

ajft commented 11 years ago

I've uploaded the original file to http://ajft.org/photos/837-630/837-630_02.JPG

Exiftool reports the full set of tags, but when I imported it into openphoto.ajft.org it didn't import all tags. Import was done with Google chrome on a Mac .

Adrian

On Wed, Jul 24, 2013 at 2:22 AM, Hubert Figuière notifications@github.comwrote:

Exiftool might just be doing the right thing. Do you a link to the original file?

— Reply to this email directly or view it on GitHubhttps://github.com/photo/frontend/issues/1338#issuecomment-21426384 .

Adrian Tritschler Melbourne, Australia

hfiguiere commented 11 years ago

The browser shouldn't matter, this is done on the backend in PHP.

ajft commented 11 years ago

On Wed, Jul 24, 2013 at 11:33 AM, Hubert Figuière notifications@github.comwrote:

The browser shouldn't matter, this is done on the backend in PHP.

True, I was just trying to get my thoughts in order of how that photo got uploaded.

Especially now that I have a counter example where similar tags did get imported.

$ exiftool -Keywords -s 837-630_01.JPG Keywords : France, 837-630, 837-630_01, APS-film, FRA, geo:lat=45.14167015, geo:lon=6.11672402, geotagged, La Villette, l'Alpe d'Huez, Oz, Rhône-Alpes, scanned, wideopenroad, wor2001b2a

My workflow is:

Photo 837-630_02 .. 09 which ended up with tags that didn't import were all tagged under XP on the Mac Photo 837-630_01 with the tags that did work was tagged on Windows 7

Running a command line exiftool on any of the photos afterwards shows all the tags.

Adrian

Adrian Tritschler Melbourne, Australia

hfiguiere commented 11 years ago

It seems that the -01 file has a charset indicated in the IPTC. Not the -02 file. - not sure if that's the problem, I haven't dug into the PHP code for that.

What if you don't use geosetter and it is just about Picasa on Windows 7 vs Picasa on Mac?

ajft commented 11 years ago

Hubert,

I'll test that, but won't be able to for a few more hours.

Adrian

On Wed, Jul 24, 2013 at 11:57 AM, Hubert Figuière notifications@github.comwrote:

It seems that the -01 file has a charset indicated in the IPTC. Not the -02 file. - not sure if that's the problem, I haven't dug into the PHP code for that.

What if you don't use geosetter and it is just about Picasa on Windows 7 vs Picasa on Mac?

Adrian Tritschler Melbourne, Australia