maintenance/language/checkLanguage.php

   1 <?php
   2 /**
   3  * Check a language file.
   4  *
   5  * @package MediaWiki
   6  * @subpackage Maintenance
   7  */
   8
   9 require_once( dirname(__FILE__).'/../commandLine.inc' );
  10 require_once( 'languages.inc' );
  11
  12 /**
  13  * Check a language.
  14  *
  15  * @param $code The language code.
  16  */
  17 function checkLanguage( $code ) {
  18         global $wgLanguages, $wgGeneralMessages, $wgRequiredMessagesNumber, $wgDisplayLevel, $wgLinks, $wgWikiLanguage, $wgChecks;
  19
  20         # Get messages
  21         $messages = $wgLanguages->getMessages( $code );
  22         $messagesNumber = count( $messages['translated'] );
  23
  24         # Skip the checks if specified
  25         if ( $wgDisplayLevel == 0 ) {
  26                 return;
  27         }
  28
  29         # Untranslated messages
  30         if ( in_array( 'untranslated', $wgChecks ) ) {
  31                 $untranslatedMessages = $wgLanguages->getUntranslatedMessages( $code );
  32                 $untranslatedMessagesNumber = count( $untranslatedMessages );
  33                 $wgLanguages->outputMessagesList( $untranslatedMessages, $code, "\n$untranslatedMessagesNumber messages of $wgRequiredMessagesNumber are not translated to $code, but exist in en:", $wgDisplayLevel, $wgLinks, $wgWikiLanguage );
  34         }
  35
  36         # Duplicate messages
  37         if ( in_array( 'duplicate', $wgChecks ) ) {
  38                 $duplicateMessages = $wgLanguages->getDuplicateMessages( $code );
  39                 $duplicateMessagesNumber = count( $duplicateMessages );
  40                 $wgLanguages->outputMessagesList( $duplicateMessages, $code, "\n$duplicateMessagesNumber messages of $messagesNumber are translated the same in en and $code:", $wgDisplayLevel, $wgLinks, $wgWikiLanguage );
  41         }
  42
  43         # Obsolete messages
  44         if ( in_array( 'obsolete', $wgChecks ) ) {
  45                 $obsoleteMessages = $messages['obsolete'];
  46                 $obsoleteMessagesNumber = count( $obsoleteMessages );
  47                 $wgLanguages->outputMessagesList( $obsoleteMessages, $code, "\n$obsoleteMessagesNumber messages of $messagesNumber are not exist in en (or are in the ignored list), but still exist in $code:", $wgDisplayLevel, $wgLinks, $wgWikiLanguage );
  48         }
  49
  50         # Messages without variables
  51         if ( in_array( 'variables', $wgChecks ) ) {
  52                 $messagesWithoutVariables = $wgLanguages->getMessagesWithoutVariables( $code );
  53                 $messagesWithoutVariablesNumber = count( $messagesWithoutVariables );
  54                 $wgLanguages->outputMessagesList( $messagesWithoutVariables, $code, "\n$messagesWithoutVariablesNumber messages of $messagesNumber in $code don't use some variables while en uses them:", $wgDisplayLevel, $wgLinks, $wgWikiLanguage );
  55         }
  56
  57         # Empty messages
  58         if ( in_array( 'empty', $wgChecks ) ) {
  59                 $emptyMessages = $wgLanguages->getEmptyMessages( $code );
  60                 $emptyMessagesNumber = count( $emptyMessages );
  61                 $wgLanguages->outputMessagesList( $emptyMessages, $code, "\n$emptyMessagesNumber messages of $messagesNumber in $code are empty or -:", $wgDisplayLevel, $wgLinks, $wgWikiLanguage );
  62         }
  63
  64         # Messages with whitespace
  65         if ( in_array( 'whitespace', $wgChecks ) ) {
  66                 $messagesWithWhitespace = $wgLanguages->getMessagesWithWhitespace( $code );
  67                 $messagesWithWhitespaceNumber = count( $messagesWithWhitespace );
  68                 $wgLanguages->outputMessagesList( $messagesWithWhitespace, $code, "\n$messagesWithWhitespaceNumber messages of $messagesNumber in $code have a trailing whitespace:", $wgDisplayLevel, $wgLinks, $wgWikiLanguage );
  69         }
  70
  71         # Non-XHTML messages
  72         if ( in_array( 'xhtml', $wgChecks ) ) {
  73                 $nonXHTMLMessages = $wgLanguages->getNonXHTMLMessages( $code );
  74                 $nonXHTMLMessagesNumber = count( $nonXHTMLMessages );
  75                 $wgLanguages->outputMessagesList( $nonXHTMLMessages, $code, "\n$nonXHTMLMessagesNumber messages of $messagesNumber in $code are not well-formed XHTML:", $wgDisplayLevel, $wgLinks, $wgWikiLanguage );
  76         }
  77
  78         # Messages with wrong characters
  79         if ( in_array( 'chars', $wgChecks ) ) {
  80                 $messagesWithWrongChars = $wgLanguages->getMessagesWithWrongChars( $code );
  81                 $messagesWithWrongCharsNumber = count( $messagesWithWrongChars );
  82                 $wgLanguages->outputMessagesList( $messagesWithWrongChars, $code, "\n$messagesWithWrongCharsNumber messages of $messagesNumber in $code include hidden chars which should not be used in the messages:", $wgDisplayLevel, $wgLinks, $wgWikiLanguage );
  83         }
  84 }
  85
  86 # Show help
  87 if ( isset( $options['help'] ) ) {
  88         echo <<<END
  89 Run this script to check a specific language file, or all of them.
  90 Parameters:
  91         * lang: Language code (default: the installation default language). You can also specify "all" to check all the languages.
  92         * help: Show this help.
  93         * level: Show the following level (default: 2).
  94         * links: Link the message values (default off).
  95         * wikilang: For the links, what is the content language of the wiki to display the output in (default en).
  96         * whitelist: Make only the following checks (form: code,code).
  97         * blacklist: Don't make the following checks (form: code,code).
  98         * duplicate: Additionally check for messages which are translated the same to English (default off).
  99         * noexif: Don't check for EXIF messages (a bit hard and boring to translate), if you know that they are currently not translated and want to focus on other problems (default off).
 100 Check codes (ideally, all of them should result 0; all the checks are executed by default):
 101         * untranslated: Messages which are required to translate, but are not translated.
 102         * obsolete: Messages which are untranslatable, but translated.
 103         * variables: Messages without variables which should be used.
 104         * empty: Empty messages.
 105         * whitespace: Messages which have trailing whitespace.
 106         * xhtml: Messages which are not well-formed XHTML.
 107         * chars: Messages with hidden characters.
 108 Display levels (default: 2):
 109         * 0: Skip the checks (useful for checking syntax).
 110         * 1: Show only the stub headers and number of wrong messages, without list of messages.
 111         * 2: Show only the headers and the message keys, without the message values.
 112         * 3: Show both the headers and the complete messages, with both keys and values.
 113
 114 END;
 115         exit();
 116 }
 117
 118 # Get the language code
 119 if ( isset( $options['lang'] ) ) {
 120         $wgCode = $options['lang'];
 121 } else {
 122         $wgCode = $wgContLang->getCode();
 123 }
 124
 125 # Get the display level
 126 if ( isset( $options['level'] ) ) {
 127         $wgDisplayLevel = $options['level'];
 128 } else {
 129         $wgDisplayLevel = 2;
 130 }
 131
 132 # Get the links options
 133 $wgLinks = isset( $options['links'] );
 134 $wgWikiLanguage = isset( $options['wikilang'] ) ? $options['wikilang'] : 'en';
 135
 136 # Get the checks to do
 137 $wgChecks = array( 'untranslated', 'obsolete', 'variables', 'empty', 'whitespace', 'xhtml', 'chars' );
 138 if ( isset( $options['whitelist'] ) ) {
 139         $wgChecks = explode( ',', $options['whitelist'] );
 140 } elseif ( isset( $options['blacklist'] ) ) {
 141         $wgChecks = array_diff( $wgChecks, explode( ',', $options['blacklist'] ) );
 142 }
 143
 144 # Add duplicate option if specified
 145 if ( isset( $options['duplicate'] ) ) {
 146         $wgChecks[] = 'duplicate';
 147 }
 148
 149 # Should check for EXIF?
 150 $wgCheckEXIF = !isset( $options['noexif'] );
 151
 152 # Get language objects
 153 $wgLanguages = new languages( $wgCheckEXIF );
 154
 155 # Get the general messages
 156 $wgGeneralMessages = $wgLanguages->getGeneralMessages();
 157 $wgRequiredMessagesNumber = count( $wgGeneralMessages['required'] );
 158
 159 # Check the language
 160 if ( $wgCode == 'all' ) {
 161         foreach ( $wgLanguages->getLanguages() as $language ) {
 162                 if ( $language != 'en' && $language != 'enRTL' ) {
 163                         checkLanguage( $language );
 164                 }
 165         }
 166 } else {
 167         # Can't check English
 168         if ( $wgCode == 'en' ) {
 169                 echo "Current selected language is English, which cannot be checked.\n";
 170         } else if ( $wgCode == 'enRTL' ) {
 171                 echo "Current selected language is RTL English, which cannot be checked.\n";
 172         } else {
 173                 checkLanguage( $wgCode );
 174         }
 175 }
 176
 177 ?>