MediaWiki:Gadget-OrangeLinks.js

/**
 * General idea: for each bluelink with an anchor, the script fetches the categories for its target page
 * and checks whether it contains a part-of-speech category. If a suitable category is found, the script
 * assumes the anchor is valid. If not, the link is coloured orange.
 * 
 * Previous version by [[User:Yair rand]], based in turn on an idea by [[User:Hippietrail]].
 * This script is a complete rewrite.
 */

// <nowiki>
/* jshint esversion: 5, loopfunc: true, latedef: true, scripturl: true, undef: true, unused: true */
/* globals mw, jQuery */

(function () {
'use strict';

var api = new mw.Api();
var fresh = [], // list of titles to get categories of
	queue = {}, // map from title to list of links with that title
	catcache = {}; // map from page name to list of categories

var articlePathRx = new RegExp('^' + mw.RegExp.escape(mw.config.get('wgArticlePath')).replace('\\$1', '(.+)') + '$'); // hax

var isDebug = mw.util.getParamValue('debug') && window.console;
var debugLogger = {};
// add methods that call console method if "isDebug" is defined
[ "error", "info" ].forEach(function (key) {
	debugLogger[key] = function () {
		if (isDebug)
			window.console[key].apply(console, arguments);
	};
});

// see [[mw:Manual:index.php]]
// alternative: /^\/(?:w(?:\/(?:index\.php)?)?)?$/
var indexPhp = [ '/', '/w', '/w/', '/w/index.php' ];

function getPageName(url) {
	if (!(url instanceof mw.Uri))
		throw new TypeError("Expected mw.Uri");
	
	var match = articlePathRx.exec(url.path);
	return match ? match[1]
		// if path is index.php, page name is value of "title" parameter
		: indexPhp.indexOf(url.path) !== -1 ? url.query.title
		: undefined;
}

// header is not language header if it does not begin with a character that
// occurs at the beginning of a language name (refer to
// [[Special:PermaLink/51354043]]) and is not a recognized non-language header
var POSHeader = /^(Noun|Proper noun|Verb|Adjective|Adverb|Participle|Interjection|Particle|Suffix|Prefix|Symbol|Phrase|Prepositional phrase|Han [ct]|Hanja|Hanzi|Kanji|Romanization|Gismu|Rafsi|Brivla|Idiom|Acronym|Initialism)/;
var otherHeader = /^(Etymology|Pronunciation|Alternative forms|See also|External links|Derived terms|Related terms|Coordinate terms|Descendants|Inflection|Declension|Conjugation|Usage notes|Translations|References|Anagrams|Synonyms|Antonyms|Meronyms|Holonyms|Hyponyms|Hypernyms|Quotations|Statistics|Compounds)/i;
function maybeLanguageHeader(header) {
	return header && /^[A-Z'ÀÁÖǀǁǂǃ]/.test(header)
	&& !(POSHeader.test(header) || otherHeader.test(header));
}

// get page title if link points to another page on this wiki
function getTitle(link) {
	if (!(link instanceof HTMLAnchorElement))
		throw new TypeError("Expected HTMLAnchorElement");
	
	// XXX: is there no other way to skip TabbedLanguages tabs?
	var parentClasses = link.parentNode.classList;
	if (parentClasses.contains("unselectedTab")
	|| parentClasses.contains("selectedTab"))
		return;
	
	// skip empty hrefs, script URLs, local links
	var rawHref = link.getAttribute('href');
	if (!(rawHref && rawHref.indexOf("javascript:") !== 0 && rawHref[0] !== "#"))
		return;
	
	var url = new mw.Uri(link.href);
	// check that the domain is correct
	if (url.getAuthority() !== location.hostname)
		return;
	
	var pageName = getPageName(url);
	
	return pageName
		? new mw.Title(decodeURIComponent(pageName
			+ '#' + url.fragment.replace(/\.([0-9A-Fa-f][0-9A-Fa-f])/g, '%$1')))
		: undefined;
}

// XXX: save resources by caching something or other between calls to
// processLink on same link?
function processLink(link) {
	try {
		if (link._ORANGED === link.href) // XXX: eliminate this hack
			return;

		var title = getTitle(link);
		
		// check that title is in main namespace and that the fragment could be
		// a language name
		if (!(title && title.getNamespaceId() === 0 && maybeLanguageHeader(title.fragment)))
			return;
		
		var prefixedText = title.getPrefixedText();
		var categories = catcache[prefixedText];
		if (categories) {
			link._ORANGED = link.href;
			
			// XXX: {{senseid}} hack
			// assumes sense ids start with Basic Latin lowercase letter and
			// that language names do not contain hyphen followed by Basic Latin
			// lowercase letter, which is not always true:
			// search for ': hastemplate:"senseid" insource:/senseid\|[^\|]+\|[^a-z]/'
			// with single quotes removed and see [[Special:PermaLink/51356029]]
			var langName = title.fragment.replace(/-[a-z].*$/, '');
			
			// XXX: discount "German Low German", etc. but allow
			// "Vietnamese Han tu" and "Vietnamese Nom" to count as an
			// existing vi entry
			// XXX: use /^ (lemmas|non-lemma forms)$/ instead?
			if (!categories.some(function (category) {
				return category.indexOf(langName) === 0
				&& /^ ([a-z]|Han tu|Nom)/.test(category.substring(langName.length));
			})) {
				link.classList.add('partlynew');
				debugLogger.info('partlynew ', link, '; cats=', categories,
					' frag=', title.fragment, ' langName=', langName);
			}
		} else {
			var entry = queue[prefixedText];
			if (!entry) {
				entry = queue[prefixedText] = [];
				fresh.push(prefixedText);
			}
			entry.push(link);
		}
	} catch (e) {
		debugLogger.error(e, 'while processing', link, ' href=', link.href);
	}
}

function processQueue(queue) {
	function collect(titles) {
		return jQuery.Deferred(function (d) {
			var query = {
				'action': 'query',
				'titles': titles.join('|'),
				'redirects': 1,
				'prop': 'categories',
				'cllimit': 100,
			};
			
			function fail(code, details, xhr) {
				d.reject(code, details, xhr);
			}
			
			function pluckResults(result) {
				debugLogger.info('result', result);
				var pages = result.query.pages, redirects = result.query.redirects;
				for (var pageid in pages) {
					var title = new mw.Title(pages[pageid].title).getPrefixedText();
					var categories = pages[pageid].categories || [];
					var cache = catcache[title];
					if (!cache)
						cache = catcache[title] = [];
					
					// 'Category:'.length === 9
					for (var i = 0; i < categories.length; ++i)
						cache.push(categories[i].title.substring(9));

					debugLogger.info('cache for', title, 'is', cache);
				}
				
				if (redirects) {
					redirects.forEach(function (redirect) {
						var origin = new mw.Title(redirect.from).getPrefixedText();
						var target = new mw.Title(redirect.to).getPrefixedText();
						catcache[origin] = catcache[target];
					});
				}
				
				if (result['continue']) {
					api.get(jQuery.extend(query, result['continue'])).then(pluckResults, fail);
				} else
					d.resolve(titles);
			}

			api.get(query).then(pluckResults, fail);
		});
	}
	
	while (fresh.length > 0) {
		collect(fresh.splice(0, 25)).then(function (titles) {
			titles.forEach(function (title) {
				queue[title].forEach(processLink);
			});
		}, function (code, details, xhr) {
			console.error(code, details, xhr);
		});
	}
	fresh = [];
}

Array.prototype.forEach.call(document.getElementsByTagName('a'), processLink);
processQueue(queue);

if (mw.util.getParamValue('@orangelinks.no_live')
|| typeof window.MutationObserver !== "function")
	return;

var mo = new MutationObserver(function (events) {
	for (var i = 0; i < events.length; ++i) {
		var event = events[i];
		if (event.type === 'childList') {
			event.addedNodes.forEach(function (addedNode) {
				if (addedNode instanceof Element) {
					if (addedNode.tagName === 'A')
						processLink(addedNode);
					else
						Array.prototype.forEach.call(
							addedNode.getElementsByTagName('a'), processLink);
				}
			});
		} else if (event.type === 'attributes' && event.target.tagName === 'A'
		&& (event.attributeName === 'href' || event.attributeName === 'class'))
			processLink(event.target);
	}
	processQueue(queue);
});

mo.observe(document, {
	attributes: true,
	childList: true,
	subtree: true
});

})();

// </nowiki>