Page MenuHomeGitPull.it

First steps with boz-mw MediaWiki API framework
Updated 384 Days AgoPublic

Version 34 of 46: You are viewing an older version of this document, as it appeared on Mar 10 2023, 09:12.

This is boz-mw, another MediaWiki API handler in PHP with batteries included! It has a tons of features that will make your head spin!

This is a library to interact with MediaWiki and Wikibase APIs. There are also some boz-mw command line tools.

Features

You may ask what we can offer to you:

  • read/write support for Wikidata
  • read/write support for Wikimedia Commons' Structured Data
  • file upload support for Wikimedia Commons
  • support for some other known wiki(s)
  • support for your custom wiki
  • lightweight project, well designed using OOP
  • it does not require Composer, but it's on Composer
  • it supports all versions from PHP 5.5 to PHP 8.2

Actually, this framework is useful for:

Download

git clone https://gitpull.it/source/boz-mw.git

Install

apt install php-cli php-curl

Command line tools

See command line tools.

Command line script replace.php

The replace.php allows you to do some sobstitutions in a wiki.

Command line script mega-export.php

The mega-export.php allows you to export the _full_ page history of whatever page.

API framework showcase

Here some usage examples.

Start: select your wiki

First of all you should init a wiki:

<?php
require 'boz-mw/autoload-with-laser-cannon.php';

// load Wikidata
$wikidata = wikidata();

// load Wikidata (this is the same)
// $wikidata = wiki( 'wikidatawiki' );

// load Wikidata (this is the same)
// $wikidata = \wm\Wikidata::instance();

// load Wikipedia in Italian (this is the same)
$itwiki = itwiki();

// load Wikipedia in Italian (this is the same)
// $itwiki = wiki( 'itwiki' );

// load Wikimedia Commons
$commons = commons();

// load Wikimedia Commons (this is the same)
// $commons = wiki( 'commonswiki' );

// load Wikimedia Commons (this is the same)
// $commons = \wm\Commons::instance();

// load Meta-Wiki
$meta = meta();

// load Meta-wiki (this is the same)
// $meta = wiki( 'metawiki' );

// load Meta-wiki this is the same
// $meta = \wm\MetaWiki::instance();

Basic MediaWiki API read query

You probably want a very basic example to get info from a very generic API of a MediaWiki.

Here an over-simplified example, with the assumption that you get just one (/ very few) results:

<?php
require 'boz-mw/autoload-with-laser-cannon.php';

// load it.wiki
$wiki = itwiki();

$response =
	$wiki->fetch( [
		'action' => 'query',
		'prop'   => 'info',
		'titles' => 'Pagina principale',
	] );

// dramatically take the very first page since you requested just one page
$page = array_pop( $response->query->pages );

// show interesting information
print_r( $page->title );
print_r( $page->pageid );

Note: See MediaWiki API action=query documentation.

This example is over-simplified. In the real life you need to get multiple pages. Also, only few pages are returned for each request, and you need to execute multiple queries to "continue" and get all remaining results. See the next example for that.

API query with continuation

To obtain a long result set from the server (with continuation support):

<?php
require 'boz-mw/autoload-with-laser-cannon.php';

$wiki = itwiki();

$request =
	$wiki->createQuery( [
		'action'  => 'query',
		'list'    => 'categorymembers',
		'cmtitle' => 'Categoria:Software con licenza GNU GPL',
	] );

// automatic query continuation
foreach( $request as $response ) {

        // loop each category member
	$pages = $response->query->categorymembers ?? [];
	foreach( $pages as $page ) {

		// do something
		var_dump( $page->pageid );
		var_dump( $page->ns );
		var_dump( $page->title );
		print_r( $page );
	}
}
NOTE: Trust me, there is a lot of stuff under the hood to make this work so easily for you. After the first request, it is actually expanding it, to ask for the other results from the server, until you get them all. Doing an "API continuation" by hand is really tiring. You are welcome.

Login and Edit API query

<?php
require 'boz-mw/autoload-with-laser-cannon.php';

// insert here your bot username and password from your wiki
// https://meta.wikimedia.org/wiki/Special:BotPasswords
$bot_user     = '';
$bot_password = '';

// pick a wiki
$wiki = itwiki();

// login
$wiki->login( $bot_user, $bot_password );

// save a page
// if something bad happen here, it will quit your program (throwing a detailed exception)
$response = $wiki->edit( [
	'title'   => 'Wikipedia:Pagina delle prove',
	'text'    => 'My test wikitext with boz-mw (sorry for this test)',
	'summary' => 'My test edit summary with boz-mw (sorry for this test)',
] );

// do something here

// a number
var_dump( $response->edit->pageid );

// a number
var_dump( $response->edit->oldrevid );

// a number
var_dump( $response->edit->newrevid );

This example works. If you run this example this page will be edited:

Please do not abuse!

NOTE: See MediaWiki action=edit documentation. Yes, the token parameter is automatically handled for you automagically.

Quick start bot wizard and serious-business error management

Do you need to configure your bot with its username and password? We have a wizard for that!

<?php
require 'boz-mw/autoload-with-laser-cannon.php';

/*
 * Check your configuration
 *
 * A nice wizard will help you if it does not exist <3
 */
config_wizard( 'config.php' );

/*
 * Login in your wiki
 *
 * It uses your bot username and password from your config.
 */
$wiki = itwiki()->login();

/*
 * Set the page you want to edit
 *
 * This is an not-writable example to try what happens! <3
 */
$page_title = 'Special:Nothing';

/**
 * Try to save the page
 *
 * The "try" block is useful to "catch" errors instead of dying.
 *
 */
try {

	// try to save a page
	$response = $wiki->edit( [
		'title'   => $page_title,
		'text'    => 'My test wikitext with boz-mw (sorry for this test)',
		'summary' => 'My test edit summary with boz-mw (sorry for this test)',
	] );

	// if you are here everything was OK

	// do something here

	// a number
	var_dump( $response->edit->pageid );

	// a number
	var_dump( $response->edit->oldrevid );

	// a number
	var_dump( $response->edit->newrevid );

/**
 * The catch block can be used to detect a specific exception
 *
 * Some of them:
 *   MaxLagException
 *   PageCannotExistException
 *   EditConflictException
 *   ProtectedPageException
 *   ReadApiDeniedException
 *   ReadOnlyException
 *   PermissionDeniedException
 *   PageCannotExistException
 *
 * and more!
 *
 * In this case we will detect the 'page cannot exist' exception.
 */
} catch( API\PageCannotExistException $e ) {

	echo "Sorry but you are editing a page that cannot exist. \n";

// generic error handler
} catch( Exception $e ) {

	// other exceptions
	echo "Something unexpected happened:  ";
	echo $e->getMessage();
	echo "\n";

}

// do something in any case
echo "End. \n";

This example works. If you run this example this page will be edited:

Please do not abuse!

NOTE: The wizard will just create a configuration file with these constants inside:
\mw\API::$DEFAULT_USERNAME = '':
\mw\API::$DEFAULT_PASSWORD = '';

Wikidata SPARQL query

What if you want to list all the cats from Wikidata?

<?php
require 'boz-mw/autoload-with-laser-cannon.php';

$wikidata = wikidata();

// you should know how to build a SPARQL query
$query  = 'SELECT ?item ?itemLabel WHERE {';
$query .= ' ?item wdt:P31 wd:Q146 ';
$query .= ' SERVICE wikibase:label { bd:serviceParam wikibase:language "en". } ';
$query .= '}';

// query Wikidata and decode the response
$rows = $wikidata::querySPARQL( $query );

// for each cat
foreach( $rows as $row ) {

	// example: 'http://www.wikidata.org/entity/Q5317221'
	$url = $row->item->value;

	// example: 'Q5317221'
	$id = basename( $url );

	// example: 'Dusty the Klepto Kitty'
	$itemLabel = $row->itemLabel->value;

	echo "Found cat ID: $id. Name: $itemLabel \n";
}

Wikidata read API

Reading a Wikidata entity is really simple.

Under the hood it uses this API entry point:

https://www.wikidata.org/w/api.php?action=help&modules=wbgetentities

But the wrapper is really simplified.

<?php
require 'boz-mw/autoload-with-laser-cannon.php';

$wikidata = wiki( 'wikidatawiki' );

// query the Wikidata Sandbox
$data = $wikidata->fetchSingleEntity( 'Q4115189', [
	'props' => [
		'descriptions',
		'labels',
		'claims',
	],
] );

// get descriptions and labels
$italian_description  = $data->getDescriptionValue( 'it' );
$italian_label        = $data->getLabelValue( 'it' );

// look all the Claims
var_dump( $data->getClaimsGrouped() );

Wikidata edit API

Editing a Wikidata item, honestly, it's stuff for advanced people, but our wrapper simplifies lot of things. Trust me.

Under the hood it uses this API entry point:

https://www.wikidata.org/w/api.php?action=help&modules=wbeditentity

In this example you:

  • overwrite a Wikidata label
  • overwrite a Wikidata description
  • add a Commons Category (← this is called a "Claim" in Wikidata)

To do all these changes, here a full example:

<?php
require 'boz-mw/autoload-with-laser-cannon.php';

// load a dedicated configuration - it will create that if missing
config_wizard( 'config.php' );

$wikidata = wiki( 'wikidatawiki' );

// this object registers your proposed changes
$newdata = $wikidata->createDataModel();

// prepare a new Wikimedia Commons category
$statement = new \wb\StatementCommonsCategory( 'P373', 'Test category name' );
$newdata->addClaim( $statement );

$newdata->setLabelValue( 'en', "New label" );

$newdata->setDescriptionValue( 'en', "New Description" );

$wikidata->login();

// this tries to save all your proposed changes in the Wikidata Sandbox
$newdata->editEntity( [
	'id' => 'Q4115189',
] );

Upload API query

Uploading a file requires to respect the RFC1341 about an HTTP multipart request.

Well, we made it easy:

<?php
require 'boz-mw/autoload.php';

// this can be an URL or a local pathname
$photo_url = 'http://.../libre-image.jpg';
$wiki->upload( [
	'comment'  => 'upload file about...',
	'text'     => 'bla bla [[bla]]',
	'filename' => 'Libre image.jpg',
	\network\ContentDisposition::createFromNameURLType( 'file', $photo_url, 'image/jpg' ),
] );

Eventually see the ContentDisposition class for some other constructors.

Where to test

Please use your own wiki to test this framework or at least use the Wikimedia Wikis' Sandboxes!

Some known pages you can destroy:

How to add another Wiki

Thank you for adding another wiki!

To create another wiki, create a file similar to this one:

https://gitpull.it/source/boz-mw/browse/master/include/web/LandscapeforWiki.php

Then, mention it in the list called MediaWikis::allClasses(). Example:

https://gitpull.it/source/boz-mw/browse/master/include/web/MediaWikis.php

Feel free to commit your wiki in the core and send upstream! So other can contribute.

How to contribute? How to pull request?

Thank you for your code contributions! Follow the instructions in the Download and Install sections. Then:

Please start creating a Task here first:

Describe your problem / goal. Then we can help you in sharing your patch with upstream.

After you created the Task, the best method is, run this command from your local repository:

arc diff

And follow the wizard.

https://we.phorge.it/book/phorge/article/arcanist_quick_start/

If you have no time for all this mess, feel free to just do a pull request on the proprietary GitHub platform:

https://github.com/valerio-bozzolan/boz-mw/

Bonus point: after your pull request, please create an account in gitpull.it:

https://gitpull.it/auth/start/

If you do a pull request in GitHub, it's OK: I will do my best to merge your patch with my weird workflow Phabricator-based, for you. You're welcome!

Other examples?

Feel free to fork and improve this documentation! Or just look inside the /include directory where there is some inline documentation for you!

Troubleshooting

You can enable debug mode with bozmw_debug(). Example:

<?php
require 'boz-mw/autoload-with-laser-cannon.php';

bozmw_debug();

// other code

Testing Unit Files

To run the Unit tests, just execute this in the main directory:

phpunit --bootstrap=phpunit/load.php --testdox phpunit

Known usages

Last Author
valerio.bozzolan
Last Edited
Mar 10 2023, 09:12

Document Hierarchy

Event Timeline

valerio.bozzolan changed the title from First Steps With Boz-mw to First steps with boz-mw MediaWiki API framework.Mar 4 2020, 00:00
valerio.bozzolan created this object.
valerio.bozzolan changed the edit policy from "Administrators" to "All Users".
valerio.bozzolan edited the content of this document. (Show Details)
valerio.bozzolan edited the content of this document. (Show Details)
valerio.bozzolan edited the content of this document. (Show Details)
valerio.bozzolan edited the content of this document. (Show Details)
valerio.bozzolan edited the content of this document. (Show Details)
valerio.bozzolan edited the content of this document. (Show Details)
valerio.bozzolan edited the content of this document. (Show Details)