<?php

/**
 * TrustedLinks MediaWiki extension.
 *
 * @author Laird Shaw based on code by James Paige
 *
 * Description:
 *
 * Prevents users lacking the 'postlink' right from inserting URLs or URL-like
 * text into articles on your wiki. Checks for these anywhere in the wikitext
 * of the page, not only when enclosed in square brackets. The regular
 * expression that detects URLs and URL-like text can be customised; by default
 * it detects URLs starting with http(s):// and (s)ftp://, as well as bare
 * domain names starting with a variant of www, such as www2.spam.com, and most
 * bare URLs that extend beyond the domain name, such as spam.com/spammy-page.
 *
 * Best used to block spam by setting this right for a group of users trusted not
 * to spam links (e.g. as for the 'trusted' group below), and then adding users
 * to that group on demand using [[Special:UserRights]].
 *
 * Installation:
 *
 * Create a "TrustedLinks" directory under the "extensions" directory of your
 * MediaWiki root and copy this file and the accompanying TrustedLinks.i18n.php
 * file into that directory. Copy the following text verbatim into your LocalSettings.php
 * file and edit it per the comments:

require_once("$IP/extensions/TrustedLinks/TrustedLinks.php");

# Set permissions as desired - the following are typical use cases.
$wgGroupPermissions['bureaucrat']['postlink'] = true;
$wgGroupPermissions['sysop'     ]['postlink'] = true;
$wgGroupPermissions['trusted'   ]['postlink'] = true;

# Uncomment this to customise the regular expression that is used to extract
# URLs and URL-like text from the raw wikitext.
#$wgTrustedLinksRegex = '/yourregexhere/';

## End TrustedLinks entries in LocalSettings.php

 *
 * Note that if you wish to log or be notified by email of failed link insertion
 * attempts, you can install the UntrustedLinksLogger extension.
 *
 */

$wgHooks['EditFilter'][] = 'TrustedLinks::OnEditFilter';
$wgExtensionMessagesFiles['TrustedLinks'] = dirname(__FILE__).'/TrustedLinks.i18n.php';
$wgExtensionCredits['other']['TrustedLinks'] = array(
	'path' => __FILE__,
	'name' => 'TrustedLinks',
	'author' => 'James Paige and Laird Shaw',
	'url' => 'https://www.mediawiki.org/wiki/Extension:NotEvil',
	'descriptionmsg' => 'trustedlinks-desc',
	'version' => '1.0',
);

class TrustedLinks {
	// From http://daringfireball.net/2010/07/improved_regex_for_matching_urls, with modifications
	// to avoid matching internal wiki links: we limit protocols to http(s) and (s)ftp.
	static $defaultUrlRegex = '#(?i)\\b((?:(?:http(s)?|(s)?ftp)://|www\\d{0,3}[.]|[a-z0-9.\\-]+[.][a-z]{2,4}/)(?:[^\\s()<>]+|\\(([^\\s()<>]+|(\\([^\\s()<>]+\\)))*\\))+(?:\\(([^\\s()<>]+|(\\([^\\s()<>]+\\)))*\\)|[^\\s`!()\\[\\]{};:\'".,<>?«»“”‘’]))#';

	/**
	 * The entry point to the extension. Implements the EditFilter hook.
	 */
	static function OnEditFilter($EditPage, $newText, $section, &$hookError, $summary) {
		global $wgUser;

		$article = $EditPage->getArticle();
		$oldText = $article->getRawText();
		$addedUrls = self::getNewUrls($oldText, $newText);
		if ($addedUrls) {
			$mayPostUrls = $wgUser->isAllowed('postlink');
			if (!$mayPostUrls) {
				$errorMessage = '';
				// A hook function return of false means to abort our planned cancellation of
				// the page save. A return of true means to continue our planned cancellation of
				// the page save, optionally with an error message generated by the hook
				// function.
				if (wfRunHooks('UntrustedLinkAttempt', array($EditPage, $addedUrls, $newText, &$errorMessage))) {
					if (!$errorMessage) {
						if (function_exists('wfLoadExtensionMessages')) {
							wfLoadExtensionMessages('TrustedLinks');
						}
						$wfMsgFunc = function_exists('wfMessage') ? 'wfMessage' : 'wfMsg';
						$errorMessage = $wfMsgFunc('trustedlinks_noright');
					}
 					// Setting $hookError indicates failure even though we're returning
					// true, and for the core code to display $hookError as an error message.
					$hookError = $errorMessage;
				}
			}
		}

		return true;
	}

	/**
	 * Returns true if a URL or URL-like text is present in the string $newText
	 * that was not present in the string $oldText, and false otherwise.
	 */
	static function wasLinkAdded($oldText, $newText) {
		return self::getNewUrls($oldText, $newText) ? true : false;
	}

	/**
	 * Returns a complete list of unique URLs or URL-like texts present in
	 * the string $newText that were not present in the string $oldText.
	 */
	static function getNewUrls($oldText, $newText) {
		$oldTextArr = explode("\n", $oldText);
		$newTextArr = explode("\n", $newText);
		$newOrChangedLines = self::getNewOrChangedLines($oldTextArr, $newTextArr);
		$urlsInNewOrChangedLines = self::extractUrls($newOrChangedLines);
		if (!$urlsInNewOrChangedLines) {
			return array();
		} else {
			$preExistingUrls = self::extractUrls($oldTextArr);
			return array_diff($urlsInNewOrChangedLines, $preExistingUrls);
		}
	}

	/**
	 * Returns a complete list of unique URLs or URL-like texts present in
	 * the array of strings that is $lines. Matches are performed against
	 * $wgTrustedLinksRegex if set, otherwise a default regex.
	 */
	static function extractUrls($lines) {
		global $wgTrustedLinksRegex;

		$urlRegex = isset($wgTrustedLinksRegex) ? $wgTrustedLinksRegex : self::$defaultUrlRegex;

		$urls = array();
		foreach ($lines as $line) {
			if (preg_match_all($urlRegex, $line, $matches, PREG_PATTERN_ORDER)) {
				foreach ($matches[0] as $url) {
					if (!in_array($url, $urls)) $urls[] = $url;
				}
			}
		}

		return $urls;
	}

	/**
	 * Returns an array of lines (strings) which are either new or changed
	 * in the diff between the arrays of strings $oldTextArr and $newTextArr.
	 */
	static function getNewOrChangedLines($oldTextArr, $newTextArr) {
		$WikiDiff3 = new WikiDiff3();
		$diffRanges = $WikiDiff3->diff_range($oldTextArr, $newTextArr);
		$newOrChangedLines = array();
		foreach ($diffRanges as $diffRange) {
			for ($i = $diffRange->rightstart; $i < $diffRange->rightend; $i++) {
				$newOrChangedLines[] = $newTextArr[$i];
			}
		}

		return $newOrChangedLines;
	}
}

?>