Project:Support desk

Jump to navigation Jump to search

About this board

Welcome to the MediaWiki Support desk, where you can ask MediaWiki questions!

(Read this message in a different language)

See also

Other places to ask for help:

Before you post

Post a new question

  1. To help us answer your questions, please indicate which versions you are using, as found on your wiki's Special:Version page:
    • MediaWiki version
    • PHP version
    • Database type and version
  2. Please include the web address (URL) to your wiki if possible. It's often easier for us to identify the source of the problem if we can see the error directly.
  3. To start a new thread, click "Start a new topic".

Search function in File list doesn't work properly

Summary by Ciencia Al Poder
Zioalex (talkcontribs)
Fereal (talkcontribs)

I've always thought the File List search needed some sort of option to activate since it never worked for me.

Checking the file, the query generated attempts to convert both the image names and the search query to lowercase before using a LIKE comparison. I've checked the database and apparently the image name column is set as VARBINARY, preventing the lowercase conversion of the image name thus botching the comparison.

TL;DR Go to includes\specials\SpecialListfiles.php. On Line 84, change:

$this->mQueryConds[] = 'LOWER(img_name)' .


$this->mQueryConds[] = 'CONVERT(img_name USING latin1)' .

Necrobumping since this problem still persists in 1.19.

kthxbye. (talkcontribs)

It appears that somewhere around v1.22 they introduced a new function to replace the mQueryConds member variable. When they did this, the patch above recommended stopped working. I believe this is because the new function, "protected fuction buildQueryConds" handles this task now.

TL;DR Go to includes/specials/SpecialListfiles.php, Line 138:

 $conds[] = 'LOWER(' . $prefix . '_name)' .

I'm not sure what to change this to.

For the full function, see here:

 protected function buildQueryConds( $table ) {
               $prefix = $table === 'image' ? 'img' : 'oi';
               $conds = array();

               if ( !is_null( $this->mUserName ) ) {
                       $conds[ $prefix . '_user_text' ] = $this->mUserName;

               if ( $this->mSearch !==  ) {
                       $nt = Title::newFromURL( $this->mSearch );
                       if ( $nt ) {
                               $dbr = wfGetDB( DB_SLAVE );
                               $conds[] = 'LOWER(' . $prefix . '_name)' .
                                       $dbr->buildLike( $dbr->anyString(),
                                               strtolower( $nt->getDBkey() ), $dbr->anyString() );

               if ( $table === 'oldimage' ) {
                       // Don't want to deal with revdel.
                       // Future fixme: Show partial information as appropriate.
                       // Would have to be careful about filtering by username when username is deleted.
                       $conds['oi_deleted'] = 0;

               // Add mQueryConds in case anyone was subclassing and using the old variable.
               return $conds + $this->mQueryConds;
       } (talkcontribs)

Nevermind, I figured it out.

TL;DR Go to includes/specials/SpecialListfiles.php, Line 138 change:

$conds[] = 'LOWER(' . $prefix . '_name)' .


$conds[] = 'CONVERT(' . $prefix . '_name USING latin1)' .
Ciencia Al Poder (talkcontribs) (talkcontribs)

Unless i'm reading it wrong, that bug seems to be more based on an error in handling non standard characters. They couldn't search because the db table was not accepting UTF8.

With the issue described above, its not a special character but just a capital letter. I didn't need to change the db table format, just how the query was being submitted to the db. (talkcontribs)

nevermind, both issues are related to the VARBINARY column type. In the resolution above we changed the query, in the resolution in the bug, they changed the column type. either way would resolve it I guess.

Ciencia Al Poder (talkcontribs)

The efficient way is to convert the database field, specially if it's part of an index, because searching for that column would use the index. If you need to convert the value for searching, then the index won't be used and the database engine would need to do a full table scan instead. (talkcontribs)

I am A newbie here. I accidently changed the subject of the thread. I actually wanted to start a new one. Please change it back. Sorry. (talkcontribs)
Loman87 (talkcontribs)

Hi everybody,

sorry but I am facing this same issue in 1.27.1. In Special:ListFiles I can't retrieve any file using letters (either upper and lower case) in the query; if I use only numbers I get results.

The solution proposed in this thread, seems not to be still valid for newer versions of MW. Maybe is this another bug?

Thanks for your collaboration!

Ciencia Al Poder (talkcontribs)

Can you check the datatype of the column img_name from the image table? It should be varbinary.

Loman87 (talkcontribs)


I am not sure if the following is what you asked, however this is what I find in tables.sql for the image table:

CREATE TABLE /*_*/image (

  -- Filename.

  -- This is also the title of the associated description page,

  -- which will be in namespace 6 (NS_FILE).

  img_name varchar(255) binary NOT NULL default '' PRIMARY KEY,

  -- File size in bytes.

  img_size int unsigned NOT NULL default 0,

  -- For images, size in pixels.

  img_width int NOT NULL default 0,

  img_height int NOT NULL default 0,

  -- Extracted Exif metadata stored as a serialized PHP array.

  img_metadata mediumblob NOT NULL,

  -- For images, bits per pixel if known.

  img_bits int NOT NULL default 0,

  -- Media type as defined by the MEDIATYPE_xxx constants


  -- major part of a MIME media type as defined by IANA

  -- see

  -- for "chemical" cf. by the ACS

  img_major_mime ENUM("unknown", "application", "audio", "image", "text", "video", "message", "model", "multipart", "chemical") NOT NULL default "unknown",

  -- minor part of a MIME media type as defined by IANA

  -- the minor parts are not required to adher to any standard

  -- but should be consistent throughout the database

  -- see

  img_minor_mime varbinary(100) NOT NULL default "unknown",

 -- Description field as entered by the uploader.

  -- This is displayed in image upload history and logs.

  img_description varbinary(767) NOT NULL,

  -- user_id and user_name of uploader.

  img_user int unsigned NOT NULL default 0,

  img_user_text varchar(255) binary NOT NULL,

  -- Time of the upload.

  img_timestamp varbinary(14) NOT NULL default '',

  -- SHA-1 content hash in base-36

  img_sha1 varbinary(32) NOT NULL default ''

) /*$wgDBTableOptions*/;

Loman87 (talkcontribs)

Ok, maybe I understand now. I have to change  

img_name varchar(255) binary NOT NULL default '' PRIMARY KEY,


  img_name varbinary(255) NOT NULL default '' PRIMARY KEY,

is that right?

Loman87 (talkcontribs)

I made this change but the result is the same...any other idea?

Ciencia Al Poder (talkcontribs)

tables.sql is how the tables are created, and they should already exist on your installation. Changing the script will do nothing unless you install MediaWiki on a new database. Also, you should not be editing MediaWiki files directly.

What I asked is to check the type and collation of the actual tables on your database. The tables.sql may be changed on each upgrade, and if your wiki is old it may have the wrong datatype.

In any case, varchar(X) binary is not equivalent to varbinary(X) (see this) and I didn't check the actual type of those columns in tables.sql, so this were mostly instructions for you to check rather than changes to do to your database.

Collations other than *_bin are probably wrong. This feature is probably still broken since task T34207 is still open. I wonder why using this instead of Special:Search.

Loman87 (talkcontribs)

Good morning and thanks for you patience.

I check the actual database and this is the result:


stdClass Object


    [Field] => img_name

    [Type] => varbinary(255)

    [Collation] =>

    [Null] => NO

    [Key] => PRI

    [Default] =>

    [Extra] =>

    [Privileges] => select,insert,update,references

    [Comment] =>

The datatype seems ok; the collation instead is not *_bin as you stated. Do I have to change it? How?

I use Special:ListFiles because the results presentation is cleaner and also because the results have a chronological order (for some purposes pertinence is not always the best choice). For me it is easier to work here, but now I have a lot of files and I need to use the search function.

Thanks again for your collaboration!

Ciencia Al Poder (talkcontribs)

Looks like, in order to be able to search, the field mist be varchar(255) binary, not varbinary(255)

Cascosoft (talkcontribs)


I am experiencing this issue with the following configuration:

MediaWiki 1.30.0
PHP (apache2handler)
MySQL 5.7.22-0ubuntu0.16.04.1

Search in the Special:ListFiles only returns results when using numbers in the search query (and these numbers are present in a file name).

Is this a bug or something broke in my wiki?

Most comments (up) seem to indicate is a known issue from previous versions. If so, is there a patch?

PorkCharSui79 (talkcontribs)

I'm also experiencing this problem with the following configuration:

MediaWiki 1.31.0
PHP 7.0.33 (apache2handler)
MySQL 5.6.43-84.3

If I omit the first letter of the file I can find them. It doesn't apply to numbers though, they are found without a problem.

Jamiehutber (talkcontribs)

Yep, also having the exact same problem... sickening :( Exactly as described above PorkCharSui79

Cornulio (talkcontribs)

We have the same Problem with

mediawiki 1.31.0

PHP 7.2.16

MariaDB: 5.5.60

but I found a wierd little workaround:

e.g. I search for a file named Start

I can find it if I just type S

I can find it if I type art or tart

BUT I can't find it when I type Start or even just St

Haakmak (talkcontribs)

Same Problem.

MediaWiki     1.31.0

PHP (fpm-fcgi)

MySQL     5.7.27-0ubuntu0.18.04.1

Does anyone have a solution? Thank you

Fereal (talkcontribs)

Can't believe a problem I've encountered from basically fresh MediaWiki installations has an answer from myself from 8 years ago.

Go to /includes/specials/pagers/. Open ImageListPager.php. Find the line:

$conds[] = 'LOWER(' . $prefix . '_name)' .

And replace it with:

$conds[] = 'CONVERT(' . $prefix . '_name USING utf8)' .

This fix is only for Special:Listfiles. Don't expect fixes for vanilla search from devs since Wikimedia projects use their own search functions.

Cornulio (talkcontribs)

Thank you! works fine! (talkcontribs)

@Fereal Thank you very much!!!!

Bruceillest (talkcontribs)

@Fereal Thanks this worked for my Wiki version 1.32.0. The Fix is still going strong!

S0ring (talkcontribs)

This workaround is working, why isn't patched and deployed into the next MW version?

Ciencia Al Poder (talkcontribs)
S0ring (talkcontribs)

The patch is addressed to fix the file SpecialNewimages.php, while the workaround from this topic fixes the file ImageListPager.php (!)

2001:420:C0C0:1002:0:0:0:35D (talkcontribs)

I also found this patch worked. Is there any way we can promote that patch for review again?

Gruniversal (talkcontribs)
Ernstkm (talkcontribs)

Those wishing to automate the process can apply Gruniversal's patch directly onto your 1.35.x installation with

cd /path/to/your/mediawiki/installation
curl \
  | base64 -d | patch -p1 --dry-run

Use base64 -D on macOS (and maybe BSD) and remove the --dry-run to actually do the thing if you get no complaints from patch --dry-run.

This will not apply cleanly to 1.36 and above, so you'll want to make the changes by hand, based on some variation of Fereal's suggestions above, or the patch from phab:T34207.

It might help to make backups of the core files that are about to be modified to something like .factory first, so you can more easily diff to see what changed, and go back if everything else goes disastrously, although patch has a -R ("reverse") option, if that's any consolation.

2600:1700:4BA0:8430:9836:C644:BD8:87C3 (talkcontribs)

Still works, still ridiculous that a one-line fix has persisted in a talk page for 10 years without promotion.

Reply to "Search function in File list doesn't work properly"
Thatnewman (talkcontribs)

I Already designed my mainpage and i am trying to remove main page text... Something like this below

Main Page

From Wikipedia

Reply to "Remove The Main Page Text"
Roxanegachameep (talkcontribs)

Here is some of the best games

Thatnewman (talkcontribs)

How do I install all the websites links on my ''Wikipedia sister projects''

only Mediawiki and Wiktionary that is working.

Bawolff (talkcontribs)
Reply to "Wikipedia sister projects"

"Citation needed" template missing

3 (talkcontribs)

I've just installed a fresh copy of MediaWiki, and have created a few pages. However, when I try to invoke the "Citation needed" template, either with {{Cn}} or {{Citation needed}}, it renders on the page as Template:Cn or Template:Citation needed respectively.

Is there a specific extension that needs to be enabled? Do the citation templates have to be added to an installation manually? I apologise if this is a basic issue, but I haven't been able to find any information online besides the MediaWiki article on usage of the template, not it's installation...

Jonathan3 (talkcontribs)
Bawolff (talkcontribs)
Reply to ""Citation needed" template missing"

Unable to generate Thumbnails : escapeshellarg() has been disabled for security reasons

Ironhide1975 (talkcontribs)

Okay so everything was working just fine and when I tried to upload images I received this error.

: escapeshellarg() has been disabled for security reasons

I did some research and set $wgUseImageMagick = false;

This allows the file to be uploaded but when I go to view the image on the page I get the following error.

Error creating thumbnail: File missing

What gives?

Jonathan3 (talkcontribs)
Bawolff (talkcontribs)

The mediawiki debug log might have additional info on wht gd didnt create the thumbnail. See How to debug.

Strongly reccomend getting a different host that doesnt dusable escapeshellarg.

If all else fails, try also setting $wgUseImageResize = false;

Ironhide1975 (talkcontribs)


Thank you for the info however here is the problems I'm running into. (And I've also read everything I could find on the internet concerning this.)

1) There is no disabled functions in my php file.

2) I sent a request to my provider to re-enable escapeshellarg

3) I am not a PHP programmer by heart (more of a Classic ASP guy) so I don't know what option 3 does.


I did reach out to a new hosting provider to see if I can migrate the site over as it's getting pretty big in itself.

Jonathan3 (talkcontribs)

The whole thumbnail thing is a mystery. I had a problem with PDF thumbnails which nobody could get to the bottom of. When I moved server it just started to work.

Bawolff (talkcontribs)

> There is no disabled functions in my php file.

You're probably looking at the wrong php.ini file. (Create a page containing just <?php phpinfo(); name it something ending in .php, upload to your webserver, and view it in the browser. It will tell you where the php.ini files are located).

Reply to "Unable to generate Thumbnails : escapeshellarg() has been disabled for security reasons"

My wiki page has randomly started to run really sluggish in the last few days...

B1Bayon (talkcontribs)

Wiki page of 8 months, have recently in the last few days started to run really slow

Nothing has changed in the back-end for months, no new extensions, no configurations, no new setups

The wiki has at least 5000 articles, but the system has randomly only just started to run slow the last few days, and I got notification from my web hosting that my site was using too much resources and they recommended that I upgrade my service, or I need to optimize my wiki... I don't understand what has changed and how this occurred? Does anyone have any guidance on this. is where the wiki sits.

Jonathan3 (talkcontribs)

Likely it's just the server being slow and not your website's fault... get a $5 Digital Ocean droplet (or Linode or whatever) and you'll never look back!

B1Bayon (talkcontribs)

I'll take a look at what Digital Ocean is..

B1Bayon (talkcontribs)

@Jonathan3 Hi, I don't quite understand what im supposed to do with the droplet as you suggest? Is this supposed to replace my webhosting?

Jonathan3 (talkcontribs)

It would do. It's an "unmanaged VPS". The droplet thing is a silly word they use for that because it ties in with "Ocean". The way I understand it is that a virtual private server is actually shared with others but in a different way, and a way that lets you do most of what you could do if you had a dedicated server (root access etc).

Maybe another solution would be to try out a new shared hosting server to see whether it's faster than the current one.

There are lots of pages here and on the internet on how to speed up MediaWiki, so maybe those would help too.

Bawolff (talkcontribs)

Its weird if it suddenly changed.

  • check webserver access log for changes in access patterns. Maybe a bot is loading your website a lot. Maybe it got linked somewhere popular
  • if $wgMainCacheType is set to something, make sure whatever you set it to still works
  • [advanced] use Profiling to see what is taking resources.

Last of all, although its unlikely, also check to ensure your website hasnt been hacked or anything like that.

B1Bayon (talkcontribs)

Thanks guys, it looked like a few bots were scanning it SemrushBot, MJ12bot and one other

Jonathan3 (talkcontribs)

SemrushBot rings a bell. A couple of times when my wiki went down it was the same thing. I added Crawl-delay: 20 to robots.txt and that seems to have helped (I don't know for sure).

Sorry for blaming your host unnecessarily! My wiki was about as slow as yours is now at about 15 seconds when tested on "Pingdom". I'd tried all sorts but the answer was moving server - it went down to well under a second.

B1Bayon (talkcontribs)

haha no worries, it seemed that i was spammed with 3 different bots, I've added the crawl-delay and will monitor and see how it goes.

Reply to "My wiki page has randomly started to run really sluggish in the last few days..."
Karlpajak (talkcontribs)

How do you view all the templates in All Pages?

Jonathan3 (talkcontribs)
Karlpajak (talkcontribs)

I see it! Never would have found it. Thanks for pointing out the other page types on the All Pages page. I don't know what the other pages there are for, but I'll dig in to the topics and figure it out.

Thanks again!

- Karl

Bawolff (talkcontribs)
Reply to "Seeing Templates in All Pages"
Karlpajak (talkcontribs)

How do you make the font size smaller in the Sidebox so it takes up less space. I'm trying to limit the size the sidebox takes up to the right of the text.

Jonathan3 (talkcontribs)

You could use

<div style="font-size:90%">...</div>
Bawolff (talkcontribs)

If you mean the sidebar, depends on what skin.

in Vector, something like the following in MediaWiki:Vector.css

#mw-panel { width: 9em !important; font-size: 90%; }
#content { margin-left: 10em !important; }

This is a bit untested, and might break when changing versions. Use at your own risk.

Reply to "Font size of Sidebox"

How to make wiki like Wiktionary, commons, wikidata etc?

Maniues (talkcontribs)

Wikimedia Commons, Wikisource, Wiktionary, Wikidata and others are based on MediaWiki. How can I make site like Commons and others?

Bawolff (talkcontribs)

Can you be more specific as to what is preventing you?

Maniues (talkcontribs)

@Bawolff I want to make a local website that is "copy" of Wikimedia Commons. I want to setup MediaWiki not to "Wikipedia mode" but to "File hoster" mode.

The same applies to Wiktionary and to keep datas like wikidata

Jonathan3 (talkcontribs)

Sounds like a big undertaking! What's it for?

Maniues (talkcontribs)

@Jonathan3 No, I think. I would like to have a source code of Wikimedia Commons. I'm interested in MediaWiki and I want to setup a small dictionary or hosting with my gallery.

I am not excluding future projects, but I would like to know how WMF configured MediaWiki to act as image hosting, dictionary or data center

Bawolff (talkcontribs)

Wikimedia commons is mostly just using normal MediaWiki's upload stuff + $wgForeignFileRepos so other wikis can use its images. There are some specialized extensions related to it, you can see a list at commons:Special:Version, however they are not really core to the experience. It also has some performance related modifications, but they wouldn't be very relevant to you I presume.

Wiktionary is pretty much just a generic MediaWiki site.

Wikidata uses Extension:Wikibase.

The config for all Wikimedia wikis (excluding things like the DB password, of course) is available at [git version: ] (manual:$wgConf may help in understanding it). Config files are very large, and can be hard to follow.

Some lower level server config is also available at . There is also some documentation that is specific to Wikimedia's set up at

Reply to "How to make wiki like Wiktionary, commons, wikidata etc?"