Just to make the following intelligible without going through everything, I'm trying to "batch upload" images into a local MediaWiki site, using rebuildImages.php. Here is Rob's reply to my most recent question:

Either use the script I mentioned in a former post (which can be stolen from SVN trunk) or set $wgHashedUploadDirectory to false and try again.

Rob Church _______________________________________________

And, unfortunately, here is what I did in "rebuildImages.php":

function crawlMissing() { global $wgUploadDirectory, $wgHashedUploadDirectory; $wgHashedUploadDirectory = FALSE; if( $wgHashedUploadDirectory ) {

and the following command line execution and result:

kenneth-mcdonalds-computer:~/Sites/mediawiki-1.6.5/maintenance Ken$ php rebuildImages.php --missing wikidb 2006-05-17 20:06:18: crawling /Users/Ken/Sites/mediawiki-1.6.5/images wikidb 2006-05-17 20:06:19: Surprising mime type: unknown/unknown wikidb 2006-05-17 20:06:19: /Users/Ken/Sites/mediawiki-1.6.5/images/README kenneth-mcdonalds-computer:~/Sites/mediawiki-1.6.5/maintenance Ken$

I get the feeling that if only I could tell rebuildImages where I _really_ want it to get the new images from, I could get this to work. But I don't have the PHP or MW knowledge to try that with even the smallest likelihood of success. Can I simply set $wgUploadDirectory to something appropriate, and if so, is the path absolute in the computer, relative to the main MW directory, or specified somehow else :-)

Rob mentioned another possible script I could check out from SVN, and I'll try that if I have to. But I feel this is _so_ close, surely I must be able to get it to work...and I'm not really a development geek, so I'd prefer to avoid messing around with Subversion.

Thanks again, Ken

On 17/05/06, Ken McDonald ken@pixologic.com wrote:

...

With many thanks to responses to previous questions, I think I _almost_ at the point of being able to bulk upload images. Here's the situation:

• My images are in a file "uploads" in the "main" wiki directory, i.e.

the one containing the "maintenance" directory. They all have a .jpg suffix

• I have added and configured an "AdminSettings.php" file in the main

directory. I can now run the rebuildImages script, which I could not do before.

I get two different types of results when I try to execute the rebuildImages script while in the maintenance directory:

If I run "php rebuildImages.php", I get the result;

wikidb 2006-05-17 18:43:07: Processing image... wikidb 2006-05-17 18:43:07: Finished image... 0 of 2 rows updated wikidb 2006-05-17 18:43:07: Processing oldimage... wikidb 2006-05-17 18:43:07: Finished oldimage... 0 of 0 rows updated

I have two images already in the wiki, I assume that is what is being reflected in this output.

If I run "php rebuildImages.php --missing", The result is much longer, starting with:

wikidb 2006-05-17 18:45:00: no directory, skipping /Users/Ken/Sites/mediawiki-1.6.5/images/0/00 wikidb 2006-05-17 18:45:00: no directory, skipping /Users/Ken/Sites/mediawiki-1.6.5/images/0/01 wikidb 2006-05-17 18:45:00: no directory, skipping /Users/Ken/Sites/mediawiki-1.6.5/images/0/02 wikidb 2006-05-17 18:45:00: no directory, skipping /Users/Ken/Sites/mediawiki-1.6.5/images/0/03 wikidb 2006-05-17 18:45:00: no directory, skipping /Users/Ken/Sites/mediawiki-1.6.5/images/0/04 wikidb 2006-05-17 18:45:00: no directory, skipping /Users/Ken/Sites/mediawiki-1.6.5/images/0/05 wikidb 2006-05-17 18:45:00: no directory, skipping /Users/Ken/Sites/mediawiki-1.6.5/images/0/06 wikidb 2006-05-17 18:45:00: no directory, skipping /Users/Ken/Sites/mediawiki-1.6.5/images/0/07 wikidb 2006-05-17 18:45:00: no directory, skipping /Users/Ken/Sites/mediawiki-1.6.5/images/0/08 wikidb 2006-05-17 18:45:00: no directory, skipping /Users/Ken/Sites/mediawiki-1.6.5/images/0/09 wikidb 2006-05-17 18:45:00: no directory, skipping /Users/Ken/Sites/mediawiki-1.6.5/images/0/0a wikidb 2006-05-17 18:45:00: no directory, skipping /Users/Ken/Sites/mediawiki-1.6.5/images/0/0b wikidb 2006-05-17 18:45:00: no directory, skipping /Users/Ken/Sites/mediawiki-1.6.5/images/0/0c wikidb 2006-05-17 18:45:00: no directory, skipping /Users/Ken/Sites/mediawiki-1.6.5/images/0/0d wikidb 2006-05-17 18:45:00: no directory, skipping /Users/Ken/Sites/mediawiki-1.6.5/images/0/0e wikidb 2006-05-17 18:45:00: no directory, skipping /Users/Ken/Sites/mediawiki-1.6.5/images/0/0f

and so on...

I'm assuming this causes rebuildImages to look through the "images" dir, looking for files it doesn't know about, and then add them to the database.

Does this mean I need to add my files to that directory before running the script? If so, where? The directory apparently has a very specific structure, and I don't want to put images in incorrect places. Or can I get "rebuildImages" to look at another directory (eg. /wikimain/uploads), move them into the appropriate place in "images", and register the with the database?

Once again, so many thanks for helping with this. We have (at least) several hundred images that need to be put into the site, and don't really like the thought of doing it by hand :-).

Either use the script I mentioned in a former post (which can be stolen from SVN trunk) or set $wgHashedUploadDirectory to false and try again.

Rob Church _______________________________________________ MediaWiki-l mailing list MediaWiki-l@Wikimedia.org http://mail.wikipedia.org/mailman/listinfo/mediawiki-l

And the script sounds like it's failing because of some weird MIME error. Psst...the bulk importer doesn't do MIME checking at the moment.

Rob Church _______________________________________________

I didn't even know images could have mime types associated with them, except via their extensions which I assume are set correctly since I can do manual uploads :-)

In any case, Rob, I think you're right, it's time to try the batch importers. Sigh, install SVN, figure out how to download, figure out to configure, figure out how to use--I suppose this is the kind of thing they pay me decent money for, but surely I'm not the only one to question the assertion that computers make things more efficient :-)

Thanks, Ken

Rob and you are now officially my Gods. Sorry I didn't see that about the SVN gateway before, I'm just a little overwhelmed with trying to learn too much new stuff in too little time. In any case, just to indicate to others how easy this is:

1) Put your images in a directory, anywhere. I called mine 'pix'. 2) Download importImages.php and importImage.inc.php from the gateway mentioned at the end of this message, and install them in the "maintenance" subdirectory of the wiki installation. 3) Run an appropriate command line command. For my computer, in the "maintenance" dir, that was "php importImages.php ../pix jpg".

The first arg is the dir containing the images, second and subsequent args are extensions indicating the file types you wish to process. These are given _without_ the "." preceding the extension. I only used the above command (i.e single jpg extension), so can't say anything about other extensions.

The command will not overwrite files already in the directory; perhaps there is a way to force such a thing? Fortunately, I don't see us needing such a thing.

Once again, many, many thanks, Ken

Todd Pederzani wrote:

On 5/18/06, Ken McDonald ken@pixologic.com wrote:

...

Sigh, install SVN, figure out how to download, figure out to configure, figure out how to use

Why not just use your web browser along with the HTTP SVN gateway [1] that was mentioned earlier in the thread? There's really no need to learn SVN when you will only be using the script once.

[1] http://svn.wikimedia.org/viewvc/mediawiki/trunk/phase3/maintenance/

Ken McDonald wrote:

And, unfortunately, here is what I did in "rebuildImages.php":

function crawlMissing() {
    global $wgUploadDirectory, $wgHashedUploadDirectory;
    $wgHashedUploadDirectory = FALSE;

No, that'll just break everything. :)

1) Set that in your LocalSettings.php.

2) Put all your files into the images directory.

3) run rebuildImages.php --missing

4) cross fingers

I get the feeling that if only I could tell rebuildImages where I _really_ want it to get the new images from, I could get this to work.

rebuildImages is meant to update the image table for image files that are *already* in the proper location and directory layout.

It does not move files, as it expects them to already be in place.

-- brion vibber (brion @ pobox.com)