maintenance/checkLanguage.php

   1 <?php
   2 /**
   3  * Check a language file.
   4  *
   5  * @package MediaWiki
   6  * @subpackage Maintenance
   7  */
   8
   9 require_once( '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, $wgDisplayLevel, $wgLinks, $wgChecks;
  19
  20         # Get messages number
  21         $translatableMessagesNumber = count( $wgLanguages->getTranslatableMessages() );
  22         $localMessagesNumber = count( $wgLanguages->getMessagesFor( $code ) );
  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 $translatableMessagesNumber are not translated to $code, but exist in en:", $wgDisplayLevel, $wgLinks );
  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 $localMessagesNumber are translated the same in en and $code:", $wgDisplayLevel, $wgLinks );
  41         }
  42
  43         # Obsolete messages
  44         if ( in_array( 'obsolete', $wgChecks ) ) {
  45                 $obsoleteMessages = $wgLanguages->getObsoleteMessages( $code );
  46                 $obsoleteMessagesNumber = count( $obsoleteMessages );
  47                 $wgLanguages->outputMessagesList( $obsoleteMessages, $code, "\n$obsoleteMessagesNumber messages of $localMessagesNumber are not exist in en (or are in the ignored list), but still exist in $code:", $wgDisplayLevel, $wgLinks );
  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 $localMessagesNumber in $code don't use some variables while en uses them:", $wgDisplayLevel, $wgLinks );
  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 $localMessagesNumber in $code are empty or -:", $wgDisplayLevel, $wgLinks );
  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 $localMessagesNumber in $code have a trailing whitespace:", $wgDisplayLevel, $wgLinks );
  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 $localMessagesNumber in $code are not well-formed XHTML:", $wgDisplayLevel, $wgLinks );
  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 $localMessagesNumber in $code include hidden chars which should not be used in the messages:", $wgDisplayLevel, $wgLinks );
  83         }
  84 }
  85
  86 # Show help
  87 if ( isset( $options['help'] ) ) {
  88         echo "Run this script to check a specific language file.\n";
  89         echo "Parameters:\n";
  90         echo "\t* lang: Language code (default: the installation default language). You can also specify \"all\" to check all the languages.\n";
  91         echo "\t* help: Show help.\n";
  92         echo "\t* level: Show the following level (default: 2).\n";
  93         echo "\t* links: Link the message values (default off).\n";
  94         echo "\t* whitelist: Make only the following checks (form: code,code).\n";
  95         echo "\t* blacklist: Don't make the following checks (form: code,code).\n";
  96         echo "\t* duplicate: Additionally check for messages which are translated the same to English (default off).\n";
  97         echo "Check codes (ideally, should be zero; all the checks are executed by default):\n";
  98         echo "\t* untranslated: Messages which are translatable, but not translated.\n";
  99         echo "\t* obsolete: Messages which are untranslatable, but translated.\n";
 100         echo "\t* variables: Messages without variables which should be used.\n";
 101         echo "\t* empty: Empty messages.\n";
 102         echo "\t* whitespace: Messages which have trailing whitespace.\n";
 103         echo "\t* xhtml: Messages which are not well-formed XHTML.\n";
 104         echo "\t* chars: Messages with hidden characters.\n";
 105         echo "Display levels (default: 2):\n";
 106         echo "\t* 0: Skip the checks (useful for checking syntax).\n";
 107         echo "\t* 1: Show only the stub headers and number of wrong messages, without list of messages.\n";
 108         echo "\t* 2: Show only the headers and the message keys, without the message values.\n";
 109         echo "\t* 3: Show both the headers and the complete messages, with both keys and values.\n";
 110         exit();
 111 }
 112
 113 # Get the language code
 114 if ( isset( $options['lang'] ) ) {
 115         $wgCode = $options['lang'];
 116 } else {
 117         $wgCode = $wgContLang->getCode();
 118 }
 119
 120 # Get the display level
 121 if ( isset( $options['level'] ) ) {
 122         $wgDisplayLevel = $options['level'];
 123 } else {
 124         $wgDisplayLevel = 2;
 125 }
 126
 127 # Get the links option
 128 $wgLinks = isset( $options['links'] );
 129
 130 # Get the checks to do
 131 $wgChecks = array( 'untranslated', 'obsolete', 'variables', 'empty', 'whitespace', 'xhtml', 'chars' );
 132 if ( isset( $options['whitelist'] ) ) {
 133         $wgChecks = explode( ',', $options['whitelist'] );
 134 } elseif ( isset( $options['blacklist'] ) ) {
 135         $wgChecks = array_diff( $wgChecks, explode( ',', $options['blacklist'] ) );
 136 }
 137
 138 # Add duplicate option if specified
 139 if ( isset( $options['duplicate'] ) ) {
 140         $wgChecks[] = 'duplicate';
 141 }
 142
 143 # Get language objects
 144 $wgLanguages = new languages();
 145
 146 # Check the language
 147 if ( $wgCode == 'all' ) {
 148         foreach ( $wgLanguages->getList() as $language ) {
 149                 if ( $language != 'en' && $language != 'enRTL' ) {
 150                         checkLanguage( $language );
 151                 }
 152         }
 153 } else {
 154         # Can't check English
 155         if ( $wgCode == 'en' ) {
 156                 echo "Current selected language is English, which cannot be checked.\n";
 157         } else if ( $wgCode == 'enRTL' ) {
 158                 echo "Current selected language is RTL English, which cannot be checked.\n";
 159         } else {
 160                 checkLanguage( $wgCode );
 161         }
 162 }
 163
 164 ?>