maintenance/language/checkLanguage.php

   1 <?php
   2 /**
   3  * Check a language file.
   4  *
   5  * @addtogroup Maintenance
   6  */
   7
   8 require_once( dirname(__FILE__).'/../commandLine.inc' );
   9 require_once( 'languages.inc' );
  10
  11 $cli = new CheckLanguageCLI( $options );
  12 $cli->execute();
  13
  14 class CheckLanguageCLI {
  15         private $code  = null;
  16         private $level = 2;
  17         private $doLinks = false;
  18         private $wikiCode = 'en';
  19         private $includeExif = false;
  20         private $checkAll = false;
  21         private $output = 'plain';
  22         private $checks = array();
  23
  24         private $defaultChecks = array(
  25                 'untranslated', 'obsolete', 'variables', 'empty', 'plural',
  26                 'whitespace', 'xhtml', 'chars', 'links', 'unbalanced'
  27         );
  28
  29         private $L = null;
  30
  31         /**
  32          * GLOBALS: $wgLanguageCode;
  33          */
  34         public function __construct( Array $options ) {
  35
  36                 if ( isset( $options['help'] ) ) {
  37                         echo $this->help();
  38                         exit();
  39                 }
  40
  41                 if ( isset($options['lang']) ) {
  42                         $this->code = $options['lang'];
  43                 } else {
  44                         global $wgLanguageCode;
  45                         $this->code = $wgLanguageCode;
  46                 }
  47
  48                 if ( isset($options['level']) ) {
  49                         $this->level = $options['level'];
  50                 }
  51
  52                 $this->doLinks = isset($options['links']);
  53                 $this->includeExif = !isset($options['noexif']);
  54                 $this->checkAll = isset($options['all']);
  55
  56                 if ( isset($options['wikilang']) ) {
  57                         $this->wikiCode = $options['wikilang'];
  58                 }
  59
  60                 if ( isset( $options['whitelist'] ) ) {
  61                         $this->checks = explode( ',', $options['whitelist'] );
  62                 } elseif ( isset( $options['blacklist'] ) ) {
  63                         $this->checks = array_diff(
  64                                 $this->defaultChecks,
  65                                 explode( ',', $options['blacklist'] )
  66                         );
  67                 } else {
  68                         $this->checks = $this->defaultChecks;
  69                 }
  70
  71                 if ( isset($options['output']) ) {
  72                         $this->output = $options['output'];
  73                 }
  74
  75                 # Some additional checks not enabled by default
  76                 if ( isset( $options['duplicate'] ) ) {
  77                         $this->checks[] = 'duplicate';
  78                 }
  79
  80                 $this->L = new languages( $this->includeExif );
  81         }
  82
  83         protected function getChecks() {
  84                 $checks = array();
  85                 $checks['untranslated'] = 'getUntranslatedMessages';
  86                 $checks['duplicate'] = 'getDuplicateMessages';
  87                 $checks['obsolete'] = 'getObsoleteMessages';
  88                 $checks['variables'] = 'getMessagesWithoutVariables';
  89                 $checks['plural'] = 'getMessagesWithoutPlural';
  90                 $checks['empty'] = 'getEmptyMessages';
  91                 $checks['whitespace'] = 'getMessagesWithWhitespace';
  92                 $checks['xhtml'] = 'getNonXHTMLMessages';
  93                 $checks['chars'] = 'getMessagesWithWrongChars';
  94                 $checks['links'] = 'getMessagesWithDubiousLinks';
  95                 $checks['unbalanced'] = 'getMessagesWithUnbalanced';
  96                 return $checks;
  97         }
  98
  99         protected function getDescriptions() {
 100                 $descriptions = array();
 101                 $descriptions['untranslated'] = '$1 message(s) of $2 are not translated to $3, but exist in en:';
 102                 $descriptions['duplicate'] = '$1 message(s) of $2 are translated the same in en and $3:';
 103                 $descriptions['obsolete'] = '$1 message(s) of $2 do not exist in en or are in the ignore list, but are in $3';
 104                 $descriptions['variables'] = '$1 message(s) of $2 in $3 don\'t use some variables that en uses:';
 105                 $descriptions['plural'] = '$1 message(s) of $2 in $3 don\'t use {{plural}} while en uses:';
 106                 $descriptions['empty'] = '$1 message(s) of $2 in $3 are empty or -:';
 107                 $descriptions['whitespace'] = '$1 message(s) of $2 in $3 have trailing whitespace:';
 108                 $descriptions['xhtml'] = '$1 message(s) of $2 in $3 contain illegal XHTML:';
 109                 $descriptions['chars'] = '$1 message(s) of $2 in $3 include hidden chars which should not be used in the messages:';
 110                 $descriptions['links'] = '$1 message(s) of $2 in $3 have problematic link(s):';
 111                 $descriptions['unbalanced'] = '$1 message(s) of $2 in $3 have unbalanced {[]}:';
 112                 return $descriptions;
 113         }
 114
 115         protected function help() {
 116                 return <<<ENDS
 117 Run this script to check a specific language file, or all of them.
 118 Command line settings are in form --parameter[=value].
 119 Parameters:
 120         * lang: Language code (default: the installation default language).
 121         * all: Check all customized languages
 122         * help: Show this help.
 123         * level: Show the following level (default: 2).
 124         * links: Link the message values (default off).
 125         * wikilang: For the links, what is the content language of the wiki to display the output in (default en).
 126         * whitelist: Do only the following checks (form: code,code).
 127         * blacklist: Don't do the following checks (form: code,code).
 128         * duplicate: Additionally check for messages which are translated the same to English (default off).
 129         * 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).
 130 Check codes (ideally, all of them should result 0; all the checks are executed by default (except duplicate and language specific check blacklists in checkLanguage.inc):
 131         * untranslated: Messages which are required to translate, but are not translated.
 132         * duplicate: Messages which translation equal to fallback
 133         * obsolete: Messages which are untranslatable, but translated.
 134         * variables: Messages without variables which should be used.
 135         * empty: Empty messages.
 136         * whitespace: Messages which have trailing whitespace.
 137         * xhtml: Messages which are not well-formed XHTML (checks only few common errors).
 138         * chars: Messages with hidden characters.
 139         * links: Messages which contains broken links to pages (does not find all).
 140         * unbalanced: Messages which contains unequal numbers of opening {[ and closing ]}.
 141 Display levels (default: 2):
 142         * 0: Skip the checks (useful for checking syntax).
 143         * 1: Show only the stub headers and number of wrong messages, without list of messages.
 144         * 2: Show only the headers and the message keys, without the message values.
 145         * 3: Show both the headers and the complete messages, with both keys and values.
 146
 147 ENDS;
 148         }
 149
 150         private $results = array();
 151
 152         public function execute() {
 153                 $this->doChecks();
 154                 if ( $this->level > 0 ) {
 155                         switch ($this->output) {
 156                                 case 'plain':
 157                                         $this->outputText();
 158                                         break;
 159                                 case 'wiki':
 160                                         $this->outputWiki();
 161                                         break;
 162                                 default:
 163                                         throw new MWException( "Invalid output type $this->output");
 164                         }
 165                 }
 166         }
 167
 168         protected function doChecks() {
 169                 $ignoredCodes = array( 'en', 'enRTL' );
 170
 171                 $this->results = array();
 172                 # Check the language
 173                 if ( $this->checkAll ) {
 174                         foreach ( $this->L->getLanguages() as $language ) {
 175                                 if ( !in_array($language, $ignoredCodes) ) {
 176                                         $this->results[$language] = $this->checkLanguage( $language );
 177                                 }
 178                         }
 179                 } else {
 180                         if ( in_array($this->code, $ignoredCodes) ) {
 181                                 throw new MWException("Cannot check code $this->code.");
 182                         } else {
 183                                 $this->results[$this->code] = $this->checkLanguage( $this->code );
 184                         }
 185                 }
 186         }
 187
 188         protected function getCheckBlacklist() {
 189                 static $checkBlacklist = null;
 190                 if ( $checkBlacklist === null ) {
 191                         $checkBlacklist = array();
 192                         require( dirname(__FILE__) . '/checkLanguage.inc' );
 193                 }
 194                 return $checkBlacklist;
 195         }
 196
 197         protected function checkLanguage( $code ) {
 198                 # Syntax check only
 199                 if ( $this->level === 0 ) {
 200                         $this->L->getMessages( $code );
 201                         return;
 202                 }
 203
 204                 $results = array();
 205                 $checkFunctions = $this->getChecks();
 206                 $checkBlacklist = $this->getCheckBlacklist();
 207                 foreach ( $this->checks as $check ) {
 208                         if ( isset($checkBlacklist[$code]) &&
 209                                 in_array($check, $checkBlacklist[$code]) ) {
 210                                 $result[$check] = array();
 211                                 continue;
 212                         }
 213
 214                         $callback = array( $this->L, $checkFunctions[$check] );
 215                         if ( !is_callable($callback ) ) {
 216                                 throw new MWException( "Unkown check $check." );
 217                         }
 218                         $results[$check] = call_user_func( $callback , $code );
 219                 }
 220
 221                 return $results;
 222         }
 223
 224         protected function formatKey( $key, $code ) {
 225                 if ( $this->doLinks ) {
 226                         $displayKey = ucfirst( $key );
 227                         if ( $code == $this->wikiCode ) {
 228                                 return "[[MediaWiki:$displayKey|$key]]";
 229                         } else {
 230                                 return "[[MediaWiki:$displayKey/$code|$key]]";
 231                         }
 232                 } else {
 233                         return $key;
 234         }                                               }
 235
 236         protected function outputText( ) {
 237                 foreach ( $this->results as $code => $results ) {
 238                         $translated = $this->L->getMessages( $code );
 239                         $translated = count( $translated['translated'] );
 240                         foreach ( $results as $check => $messages ) {
 241                                 $count = count( $messages );
 242                                 if ( $count ) {
 243                                         $search = array( '$1', '$2', '$3' );
 244                                         $replace = array( $count, $translated, $code );
 245                                         $descriptions = $this->getDescriptions();
 246                                         echo "\n" . str_replace( $search, $replace, $descriptions[$check] ) . "\n";
 247                                         if ( $this->level == 1 ) {
 248                                                 echo "[messages are hidden]\n";
 249                                         } else {
 250                                                 foreach ( $messages as $key => $value ) {
 251                                                         $displayKey = $this->formatKey( $key, $code );
 252                                                         if ( $this->level == 2 ) {
 253                                                                 echo "* $displayKey\n";
 254                                                         } else {
 255                                                                 echo "* $displayKey:            '$value'\n";
 256                                                         }
 257                                                 }
 258                                         }
 259                                 }
 260                         }
 261                 }
 262         }
 263
 264         /**
 265          * Globals: $wgContLang, $IP
 266          */
 267         function outputWiki() {
 268                 global $wgContLang, $IP;
 269                 $detailText = '';
 270                 $rows[] = '! Language !! Code !! Total !! ' . implode( ' !! ', $this->checks );
 271                 foreach ( $this->results as $code => $results ) {
 272                         $detailTextForLang = "==$code==\n";
 273                         $numbers = array();
 274                         $problems = 0;
 275                         $detailTextForLangChecks = array();
 276                         foreach ( $results as $check => $messages ) {
 277                                 $count = count( $messages );
 278                                 if ( $count ) {
 279                                         $problems += $count;
 280                                         $messageDetails = array();
 281                                         foreach ( $messages as $key => $details ) {
 282                                                 $displayKey = $this->formatKey( $key, $code );
 283                                                 $messageDetails[] = $displayKey;
 284                                         }
 285                                         $detailTextForLangChecks[] = "===$code-$check===\n* " . implode( ', ', $messageDetails );
 286                                         $numbers[] = "'''[[#$code-$check|$count]]'''";
 287                                 } else {
 288                                         $numbers[] = $count;
 289                                 }
 290
 291                         }
 292
 293                         if ( count( $detailTextForLangChecks ) ) {
 294                                 $detailText .= $detailTextForLang . implode( "\n", $detailTextForLangChecks ) . "\n";
 295                         }
 296
 297                         if ( !$problems ) { continue; } // Don't list languages without problems
 298                         $language = $wgContLang->getLanguageName( $code );
 299                         $rows[] = "| $language || $code || $problems || " . implode( ' || ', $numbers );
 300                 }
 301
 302                 $tableRows = implode( "\n|-\n", $rows );
 303
 304                 $version = SpecialVersion::getVersion( $IP );
 305                 echo <<<EOL
 306 '''Check results are for:''' <code>$version</code>
 307
 308
 309 {| class="sortable wikitable" border="2" cellpadding="4" cellspacing="0" style="background-color: #F9F9F9; border: 1px #AAAAAA solid; border-collapse: collapse; clear:both;"
 310 $tableRows
 311 |}
 312
 313 $detailText
 314
 315 EOL;
 316         }
 317
 318 }