Archive/Usage/Offline images

Note: This page is obsolete. It is preserved for historical reference only.

Note: This feature is currently not supported. The WMF Tarballs are severely out of date (2012 December). See the following for more info: https://lists.wikimedia.org/pipermail/xmldatadumps-l/2013-July/000868.html https://bugzilla.wikimedia.org/show_bug.cgi?id=51001 Once the WMF provides a more up to date tarball, I'll resume work on this feature. If you are interested in this feature, please send inquiries to the xmldatadumps mailing list or comment on bugzilla.

Overview [edit]

XOWA displays images by using online downloads as specified in Dev/File/Setup. However, XOWA can also be setup to use offline sources. If you have the time and storage space, you can download all the images for your wiki and run them from a hard drive. Note that the full tarball set for English Wikipedia is over 2 TB and may take 15 days to download and unzip. Even a small wiki like Simple Wikipedia requires 56 GB. Please prepare accordingly.

Download tarball [edit]

• Navigate to the Wikimedia image dump site at http://ftpmirror.your.org/pub/wikimedia/imagedumps/tarballs/fulls/
• Select the tarball for your wiki and download it.

Unzip tarball [edit]

• If you are on a Windows system, the current version of 7-Zip does not handle UTF-8 filenames (it default to CP-1252). You will need to download the latest 7-Zip alpha (I used 9.27).

You can try the latest alpha here: http://sourceforge.net/projects/sevenzip/forums/forum/45797/topic/6063877

• Unzip the tarball to the hard-drive's root.

For example, on a Windows 7 system with a hard drive mounted at W, unzip to W:\. This will create a directory called W:\wikipedia\common\.

Change config file [edit]

Two changes will have to be made to the config file:

1. commons source
   1. Search for the following: .set('src_http_commons', 'http://upload.wikimedia.org/wikipedia/commons/' , 'commons.wikimedia.org' ).ext_rules_('img_only').owner
   2. Replace it with the following: .set('src_http_commons', 'W:\wikipedia\commons\' , 'commons.wikimedia.org' ).fsys_('wnt').tarball_('y').ext_rules_('img_only').owner
2. wiki source
   1. Search for the following: .set('src_http_~{wiki_key}', 'http://upload.wikimedia.org/~{wiki_type_name}/~{wiki_lang}/' , '~{wiki_key}').ext_rules_('img_only').owner
   2. Replace it with the following: .set('src_http_~{wiki_key}', 'W:\~{wiki_type_name}\~{wiki_lang}\' , '~{wiki_key}').fsys_('wnt').tarball_('y').ext_rules_('img_only').owner

Note: the .fsys_('wnt') is not needed for Linux systems. It is a "hack" for Windows systems to handle File titles that have invalid Windows NT filesystem characters (such as " or :)

Known limitations [edit]

• The tarballs do not come with thumbnails for video or pdf files. XOWA will show them as blank.

These thumbnails are generated with other programs. Future versions of XOWA will try to generate them.

• imageMagick/inkscape will improperly identify sizes for certain svgs.

For example, on en.wikipedia.org/wiki/Earth, Earth-Moon.svg is 512x320

imageMagick incorrectly identifies it as 1000,1000.

inkscape incorrectly identifies it as 1991,316. It provides an X,Y of -825,-2.9 which doesn't help.

• inkscape will improperly show artifacts in certain svgs.

For example, on en.wikipedia.org/wiki/France, France_location_map-Regions_and_departements.svg will have random black lines around Paris

Offline thumbnails [edit]

Dumping all the thumbnails in a wiki is currently being implemented. See Dev/Design/Offline_files