| Index: trunk/extensions/ArchiveLinks/ArchiveLinks.php |
| — | — | @@ -11,43 +11,12 @@ |
| 12 | 12 | |
| 13 | 13 | error_reporting ( E_ALL | E_STRICT ); |
| 14 | 14 | |
| 15 | | -//$wgJobClasses['synchroniseThreadArticleData'] = 'SynchroniseThreadArticleDataJob'; |
| | 15 | +$path = dirname( __FILE__ ); |
| 16 | 16 | |
| 17 | | -// Hooks |
| 18 | | -/*$wgHooks['EditPage::attemptSave'][] = array ( 'getExternalLinks' ); |
| | 17 | +$wgExtensionMessagesFiles['ArchiveLinks'] = "$path/ArchiveLinks.i18n.php"; |
| 19 | 18 | |
| 20 | | -function getExternalLinks ( $editpage ) { |
| 21 | | - //var_export ( $editpage , FALSE ); |
| 22 | | - //echo "hi"; |
| 23 | | - file_put_contents ( './stuff.txt', 'it works :D' . "\n" . var_export ( $editpage ) ); |
| 24 | | - return true; |
| 25 | | -}*/ |
| 26 | | - |
| 27 | | -//$wgHooks['LinkerMakeExternalLink'][] = 'findOutWhatTheHellThisHookGivesUs'; |
| | 19 | +$wgAutoloadClasses['ArchiveLinks'] = "$path/ArchiveLinks.class.php"; |
| 28 | 20 | |
| 29 | | -function findOutWhatTheHellThisHookGivesUs ( &$url, &$text, &$link, &$attributes ) { |
| 30 | | - for ( $i = 0, $go = true ; $go !== false; ++$i ) { |
| 31 | | - if ( file_exists ( './extensions/ArchiveLinks/stuff-' . $i . '.txt' ) ) { |
| 32 | | - continue; |
| 33 | | - } else { |
| 34 | | - file_put_contents ( "./extensions/ArchiveLinks/stuff-$i.txt", |
| 35 | | - var_export ( $url , TRUE ) . "\n\n" . |
| 36 | | - var_export ( $text , TRUE) . "\n\n" . |
| 37 | | - var_export ( $link , TRUE) . "\n\n" . |
| 38 | | - var_export ( $attributes , TRUE) . "\n\n" |
| 39 | | - ); |
| 40 | | - $go = false; |
| 41 | | - } |
| 42 | | - } |
| 43 | | - //echo "hi"; |
| 44 | | - return true; |
| 45 | | -} |
| 46 | | - |
| 47 | | -//$wgHooks['LinkerMakeExternalLink'][] = 'getExternalLinks'; |
| 48 | | -//$wgHooks['EditPage::attemptSave'][] = 'getExternalLinks'; |
| 49 | | - |
| 50 | | -$wgExtensionMessagesFiles['ArchiveLinks'] = dirname( __FILE__ ) . '/ArchiveLinks.i18n.php'; |
| 51 | | - |
| 52 | 21 | $wgHooks['ArticleSaveComplete'][] = 'ArchiveLinks::queueExternalLinks'; |
| 53 | 22 | $wgHooks['LinkerMakeExternalLink'][] = 'ArchiveLinks::rewriteLinks'; |
| 54 | 23 | |
| — | — | @@ -55,136 +24,4 @@ |
| 56 | 25 | 'archive_service' => 'wikiwix', |
| 57 | 26 | 'use_multiple_archives' => false, |
| 58 | 27 | 'run_spider_in_loop' => false, |
| 59 | | -); |
| 60 | | - |
| 61 | | -class ArchiveLinks { |
| 62 | | - public static function queueExternalLinks ( &$article ) { |
| 63 | | - global $wgParser; |
| 64 | | - $external_links = $wgParser->getOutput(); |
| 65 | | - $external_links = $external_links->mExternalLinks; |
| 66 | | - //echo "$stuff"; |
| 67 | | - //file_put_contents ( './extensions/ArchiveLinks/stuff0.txt', var_export( $external_links , TRUE )); |
| 68 | | - |
| 69 | | - $db_master = wfGetDB( DB_MASTER ); |
| 70 | | - $db_slave = wfGetDB( DB_SLAVE ); |
| 71 | | - $db_result = array(); |
| 72 | | - |
| 73 | | - $db_master->begin(); |
| 74 | | - |
| 75 | | - foreach ( $external_links as $link => $unused_value ) { |
| 76 | | - //$db_result['resource'] = $db_slave->select( 'el_archive_resource', '*', '`el_archive_resource`.`resource_url` = "' . $db_slave->strencode( $link ) . '"'); |
| 77 | | - $db_result['blacklist'] = $db_slave->select( 'el_archive_blacklist', '*', '`el_archive_blacklist`.`bl_url` = "' . $db_slave->strencode( $link ) . '"'); |
| 78 | | - $db_result['queue'] = $db_slave->select( 'el_archive_queue', '*', '`el_archive_queue`.`url` = "' . $db_slave->strencode( $link ) . '"' ); |
| 79 | | - |
| 80 | | - if ( $db_result['blacklist']->numRows() === 0 ) { |
| 81 | | - if ( $db_result['queue']->numRows() === 0 ) { |
| 82 | | - // this probably a first time job |
| 83 | | - // but we should check the logs and resource table |
| 84 | | - // to make sure |
| 85 | | - $db_master->insert( 'el_archive_queue', array ( |
| 86 | | - 'page_id' => $article->getID(), |
| 87 | | - 'url' => $link, |
| 88 | | - 'delay_time' => '0', |
| 89 | | - 'insertion_time' => time(), |
| 90 | | - 'in_progress' => '0', |
| 91 | | - )); |
| 92 | | - } else { |
| 93 | | - //this job is already in the queue, why? |
| 94 | | - // * most likely reason is it has already been inserted by another page |
| 95 | | - // * or we are checking it later because the site was down at last archival |
| 96 | | - // in either case we don't really need to do anything right now, so skip... |
| 97 | | - } |
| 98 | | - |
| 99 | | - } |
| 100 | | - |
| 101 | | - //file_put_contents ( './extensions/ArchiveLinks/stuff.txt', var_export( $db_result , TRUE )); |
| 102 | | - |
| 103 | | - //$db_master->insert('el_archive_queue', $array ); |
| 104 | | - } |
| 105 | | - |
| 106 | | - $db_master->commit(); |
| 107 | | - |
| 108 | | - return true; |
| 109 | | - } |
| 110 | | - |
| 111 | | - public static function rewriteLinks ( &$url, &$text, &$link, &$attributes ) { |
| 112 | | - if ( array_key_exists('rel', $attributes) && $attributes['rel'] === 'nofollow' ) { |
| 113 | | - global $wgArchiveLinksConfig; |
| 114 | | - if ( $wgArchiveLinksConfig['use_multiple_archives'] ) { |
| 115 | | - //need to add support for more than one archival service at once |
| 116 | | - // (a page where you can select one from a list of choices) |
| 117 | | - } else { |
| 118 | | - switch ( $wgArchiveLinksConfig['archive_service'] ) { |
| 119 | | - case 'local': |
| 120 | | - //We need to have something to figure out where the filestore is... |
| 121 | | - $link_to_archive = urlencode( substr_replace( $url, '', 0, 7 ) ); |
| 122 | | - break; |
| 123 | | - case 'wikiwix': |
| 124 | | - $link_to_archive = 'http://archive.wikiwix.com/cache/?url=' . $url; |
| 125 | | - break; |
| 126 | | - case 'webcitation': |
| 127 | | - $link_to_archive = 'http://webcitation.org/query?url=' . $url; |
| 128 | | - break; |
| 129 | | - case 'internet_archive': |
| 130 | | - default: |
| 131 | | - $link_to_archive = 'http://wayback.archive.org/web/*/' . $url; |
| 132 | | - break; |
| 133 | | - |
| 134 | | - } |
| 135 | | - } |
| 136 | | - //Note to self: need to fix this to use Html.php instead of direct html |
| 137 | | - $link = "<a rel=\"nofollow\" class=\"{$attributes['class']}\" href=\"{$url}\">{$text}</a> <sup><small><a href=\"" |
| 138 | | - . "{$link_to_archive}\">" . wfMsg( 'archive-links-cache-title' ) . '</a></small></sup> '; |
| 139 | | - return false; |
| 140 | | - } else { |
| 141 | | - return true; |
| 142 | | - } |
| 143 | | - } |
| 144 | | - |
| 145 | | - /*function retrieveLinks ( ) { |
| 146 | | - |
| 147 | | - }*/ |
| 148 | | - |
| 149 | | - /*function queueURL ( $url, &$db_master ) { |
| 150 | | - |
| 151 | | - }*/ |
| 152 | | -} |
| 153 | | - |
| 154 | | -//$wgHooks['ArticleSave'][] = 'test'; |
| 155 | | - |
| 156 | | -function test ( ) { |
| 157 | | - /*$db_master = wfGetDB( DB_MASTER ); |
| 158 | | - |
| 159 | | - $db_master->insert('el_archive_blacklist', array( |
| 160 | | - 'bl_type' => 0, |
| 161 | | - 'bl_url' => 'http://example.com', |
| 162 | | - 'bl_reason' => 'test' |
| 163 | | - ));*/ |
| 164 | | - |
| 165 | | - //$db_slave = wfGetDB( DB_SLAVE ); |
| 166 | | - |
| 167 | | - /*db_result = $db_slave->select( 'el_archive_blacklist', '*', |
| 168 | | - '`el_archive_blacklist`.`bl_url` = "' . $db_slave->strencode( 'http://example.com' ) . '"'); |
| 169 | | - */ |
| 170 | | - //$db_result['queue'] = $db_slave->select( 'el_archive_queue', '*', '`el_archive_queue`.`url` = "' . $db_slave->strencode( 'http://example.com' ) . '"' ); |
| 171 | | - |
| 172 | | - //file_put_contents ( './extensions/ArchiveLinks/stuff.txt', var_export( $db_result['queue']->numRows() , TRUE )); |
| 173 | | - //$add_quotes = 'http://example.com'; |
| 174 | | - //file_put_contents ( './extensions/ArchiveLinks/stuff.txt', var_export( $db_slave->addQuotes( $add_quotes ) , TRUE )); |
| 175 | | - |
| 176 | | - |
| 177 | | - |
| 178 | | - return false; |
| 179 | | -} |
| 180 | | -/* |
| 181 | | -class InsertURLsIntoQueue extends Job { |
| 182 | | - public function __construct( $title, $params ) { |
| 183 | | - // Replace synchroniseThreadArticleData with the an identifier for your job. |
| 184 | | - parent::__construct( 'insertURLsIntoQueue', $title, $params ); |
| 185 | | - } |
| 186 | | - |
| 187 | | - |
| 188 | | - public function run() { |
| 189 | | - |
| 190 | | - } |
| 191 | | -}*/ |
| \ No newline at end of file |
| | 28 | +); |
| \ No newline at end of file |
| Index: trunk/extensions/ArchiveLinks/ArchiveLinks.class.php |
| — | — | @@ -0,0 +1,87 @@ |
| | 2 | +<?php |
| | 3 | +/** |
| | 4 | + * Main Extension Class for Archive Links |
| | 5 | + */ |
| | 6 | +if ( !defined( 'MEDIAWIKI' ) ) { |
| | 7 | + echo( "This file is an extension to the MediaWiki software and cannot be used standalone.\n" ); |
| | 8 | + die( 1 ); |
| | 9 | +} |
| | 10 | + |
| | 11 | +class ArchiveLinks { |
| | 12 | + public static function queueExternalLinks ( &$article ) { |
| | 13 | + global $wgParser; |
| | 14 | + $external_links = $wgParser->getOutput(); |
| | 15 | + $external_links = $external_links->mExternalLinks; |
| | 16 | + |
| | 17 | + $db_master = wfGetDB( DB_MASTER ); |
| | 18 | + $db_slave = wfGetDB( DB_SLAVE ); |
| | 19 | + $db_result = array(); |
| | 20 | + |
| | 21 | + $db_master->begin(); |
| | 22 | + |
| | 23 | + foreach ( $external_links as $link => $unused_value ) { |
| | 24 | + //$db_result['resource'] = $db_slave->select( 'el_archive_resource', '*', '`el_archive_resource`.`resource_url` = "' . $db_slave->strencode( $link ) . '"'); |
| | 25 | + $db_result['blacklist'] = $db_slave->select( 'el_archive_blacklist', '*', '`el_archive_blacklist`.`bl_url` = "' . $db_slave->strencode( $link ) . '"'); |
| | 26 | + $db_result['queue'] = $db_slave->select( 'el_archive_queue', '*', '`el_archive_queue`.`url` = "' . $db_slave->strencode( $link ) . '"' ); |
| | 27 | + |
| | 28 | + if ( $db_result['blacklist']->numRows() === 0 ) { |
| | 29 | + if ( $db_result['queue']->numRows() === 0 ) { |
| | 30 | + // this probably a first time job |
| | 31 | + // but we should check the logs and resource table |
| | 32 | + // to make sure |
| | 33 | + $db_master->insert( 'el_archive_queue', array ( |
| | 34 | + 'page_id' => $article->getID(), |
| | 35 | + 'url' => $link, |
| | 36 | + 'delay_time' => '0', |
| | 37 | + 'insertion_time' => time(), |
| | 38 | + 'in_progress' => '0', |
| | 39 | + )); |
| | 40 | + } else { |
| | 41 | + //this job is already in the queue, why? |
| | 42 | + // * most likely reason is it has already been inserted by another page |
| | 43 | + // * or we are checking it later because the site was down at last archival |
| | 44 | + // in either case we don't really need to do anything right now, so skip... |
| | 45 | + } |
| | 46 | + |
| | 47 | + } |
| | 48 | + } |
| | 49 | + |
| | 50 | + $db_master->commit(); |
| | 51 | + |
| | 52 | + return true; |
| | 53 | + } |
| | 54 | + |
| | 55 | + public static function rewriteLinks ( &$url, &$text, &$link, &$attributes ) { |
| | 56 | + if ( array_key_exists('rel', $attributes) && $attributes['rel'] === 'nofollow' ) { |
| | 57 | + global $wgArchiveLinksConfig; |
| | 58 | + if ( $wgArchiveLinksConfig['use_multiple_archives'] ) { |
| | 59 | + //need to add support for more than one archival service at once |
| | 60 | + // (a page where you can select one from a list of choices) |
| | 61 | + } else { |
| | 62 | + switch ( $wgArchiveLinksConfig['archive_service'] ) { |
| | 63 | + case 'local': |
| | 64 | + //We need to have something to figure out where the filestore is... |
| | 65 | + $link_to_archive = urlencode( substr_replace( $url, '', 0, 7 ) ); |
| | 66 | + break; |
| | 67 | + case 'wikiwix': |
| | 68 | + $link_to_archive = 'http://archive.wikiwix.com/cache/?url=' . $url; |
| | 69 | + break; |
| | 70 | + case 'webcitation': |
| | 71 | + $link_to_archive = 'http://webcitation.org/query?url=' . $url; |
| | 72 | + break; |
| | 73 | + case 'internet_archive': |
| | 74 | + default: |
| | 75 | + $link_to_archive = 'http://wayback.archive.org/web/*/' . $url; |
| | 76 | + break; |
| | 77 | + |
| | 78 | + } |
| | 79 | + } |
| | 80 | + //Note to self: need to fix this to use Html.php instead of direct html |
| | 81 | + $link = "<a rel=\"nofollow\" class=\"{$attributes['class']}\" href=\"{$url}\">{$text}</a> <sup><small><a href=\"" |
| | 82 | + . "{$link_to_archive}\">" . wfMsg( 'archive-links-cache-title' ) . '</a></small></sup> '; |
| | 83 | + return false; |
| | 84 | + } else { |
| | 85 | + return true; |
| | 86 | + } |
| | 87 | + } |
| | 88 | +} |
| \ No newline at end of file |
| Property changes on: trunk/extensions/ArchiveLinks/ArchiveLinks.class.php |
| ___________________________________________________________________ |
| Added: svn:eol-style |
| 1 | 89 | + native |