maintenance/language/checkLanguage.inc

   1 <?php
   2
   3 class CheckLanguageCLI {
   4         protected $code  = null;
   5         protected $level = 2;
   6         protected $doLinks = false;
   7         protected $wikiCode = 'en';
   8         protected $checkAll = false;
   9         protected $output = 'plain';
  10         protected $checks = array();
  11         protected $L = null;
  12
  13         protected $defaultChecks = array(
  14                 'untranslated', 'obsolete', 'variables', 'empty', 'plural',
  15                 'whitespace', 'xhtml', 'chars', 'links', 'unbalanced'
  16         );
  17
  18         protected $results = array();
  19
  20         private $includeExif = false;
  21
  22         /**
  23          * GLOBALS: $wgLanguageCode;
  24          */
  25         public function __construct( Array $options ) {
  26
  27                 if ( isset( $options['help'] ) ) {
  28                         echo $this->help();
  29                         exit();
  30                 }
  31
  32                 if ( isset($options['lang']) ) {
  33                         $this->code = $options['lang'];
  34                 } else {
  35                         global $wgLanguageCode;
  36                         $this->code = $wgLanguageCode;
  37                 }
  38
  39                 if ( isset($options['level']) ) {
  40                         $this->level = $options['level'];
  41                 }
  42
  43                 $this->doLinks = isset($options['links']);
  44                 $this->includeExif = !isset($options['noexif']);
  45                 $this->checkAll = isset($options['all']);
  46
  47                 if ( isset($options['wikilang']) ) {
  48                         $this->wikiCode = $options['wikilang'];
  49                 }
  50
  51                 if ( isset( $options['whitelist'] ) ) {
  52                         $this->checks = explode( ',', $options['whitelist'] );
  53                 } elseif ( isset( $options['blacklist'] ) ) {
  54                         $this->checks = array_diff(
  55                                 $this->defaultChecks,
  56                                 explode( ',', $options['blacklist'] )
  57                         );
  58                 } else {
  59                         $this->checks = $this->defaultChecks;
  60                 }
  61
  62                 if ( isset($options['output']) ) {
  63                         $this->output = $options['output'];
  64                 }
  65
  66                 # Some additional checks not enabled by default
  67                 if ( isset( $options['duplicate'] ) ) {
  68                         $this->checks[] = 'duplicate';
  69                 }
  70
  71                 $this->L = new languages( $this->includeExif );
  72         }
  73
  74         protected function getChecks() {
  75                 $checks = array();
  76                 $checks['untranslated'] = 'getUntranslatedMessages';
  77                 $checks['duplicate'] = 'getDuplicateMessages';
  78                 $checks['obsolete'] = 'getObsoleteMessages';
  79                 $checks['variables'] = 'getMessagesWithoutVariables';
  80                 $checks['plural'] = 'getMessagesWithoutPlural';
  81                 $checks['empty'] = 'getEmptyMessages';
  82                 $checks['whitespace'] = 'getMessagesWithWhitespace';
  83                 $checks['xhtml'] = 'getNonXHTMLMessages';
  84                 $checks['chars'] = 'getMessagesWithWrongChars';
  85                 $checks['links'] = 'getMessagesWithDubiousLinks';
  86                 $checks['unbalanced'] = 'getMessagesWithUnbalanced';
  87                 return $checks;
  88         }
  89
  90         protected function getDescriptions() {
  91                 $descriptions = array();
  92                 $descriptions['untranslated'] = '$1 message(s) of $2 are not translated to $3, but exist in en:';
  93                 $descriptions['duplicate'] = '$1 message(s) of $2 are translated the same in en and $3:';
  94                 $descriptions['obsolete'] = '$1 message(s) of $2 do not exist in en or are in the ignore list, but are in $3';
  95                 $descriptions['variables'] = '$1 message(s) of $2 in $3 don\'t use some variables that en uses:';
  96                 $descriptions['plural'] = '$1 message(s) of $2 in $3 don\'t use {{plural}} while en uses:';
  97                 $descriptions['empty'] = '$1 message(s) of $2 in $3 are empty or -:';
  98                 $descriptions['whitespace'] = '$1 message(s) of $2 in $3 have trailing whitespace:';
  99                 $descriptions['xhtml'] = '$1 message(s) of $2 in $3 contain illegal XHTML:';
 100                 $descriptions['chars'] = '$1 message(s) of $2 in $3 include hidden chars which should not be used in the messages:';
 101                 $descriptions['links'] = '$1 message(s) of $2 in $3 have problematic link(s):';
 102                 $descriptions['unbalanced'] = '$1 message(s) of $2 in $3 have unbalanced {[]}:';
 103                 return $descriptions;
 104         }
 105
 106         protected function help() {
 107                 return <<<ENDS
 108 Run this script to check a specific language file, or all of them.
 109 Command line settings are in form --parameter[=value].
 110 Parameters:
 111         * lang: Language code (default: the installation default language).
 112         * all: Check all customized languages.
 113         * help: Show this help.
 114         * level: Show the following level (default: 2).
 115         * links: Link the message values (default off).
 116         * wikilang: For the links, what is the content language of the wiki to display the output in (default en).
 117         * whitelist: Do only the following checks (form: code,code).
 118         * blacklist: Don't do the following checks (form: code,code).
 119         * duplicate: Additionally check for messages which are translated the same to English (default off).
 120         * 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).
 121 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):
 122         * untranslated: Messages which are required to translate, but are not translated.
 123         * duplicate: Messages which translation equal to fallback
 124         * obsolete: Messages which are untranslatable, but translated.
 125         * variables: Messages without variables which should be used.
 126         * empty: Empty messages.
 127         * whitespace: Messages which have trailing whitespace.
 128         * xhtml: Messages which are not well-formed XHTML (checks only few common errors).
 129         * chars: Messages with hidden characters.
 130         * links: Messages which contains broken links to pages (does not find all).
 131         * unbalanced: Messages which contains unequal numbers of opening {[ and closing ]}.
 132 Display levels (default: 2):
 133         * 0: Skip the checks (useful for checking syntax).
 134         * 1: Show only the stub headers and number of wrong messages, without list of messages.
 135         * 2: Show only the headers and the message keys, without the message values.
 136         * 3: Show both the headers and the complete messages, with both keys and values.
 137
 138 ENDS;
 139         }
 140
 141         public function execute() {
 142                 $this->doChecks();
 143                 if ( $this->level > 0 ) {
 144                         switch ($this->output) {
 145                                 case 'plain':
 146                                         $this->outputText();
 147                                         break;
 148                                 case 'wiki':
 149                                         $this->outputWiki();
 150                                         break;
 151                                 default:
 152                                         throw new MWException( "Invalid output type $this->output");
 153                         }
 154                 }
 155         }
 156
 157         protected function doChecks() {
 158                 $ignoredCodes = array( 'en', 'enRTL' );
 159
 160                 $this->results = array();
 161                 # Check the language
 162                 if ( $this->checkAll ) {
 163                         foreach ( $this->L->getLanguages() as $language ) {
 164                                 if ( !in_array($language, $ignoredCodes) ) {
 165                                         $this->results[$language] = $this->checkLanguage( $language );
 166                                 }
 167                         }
 168                 } else {
 169                         if ( in_array($this->code, $ignoredCodes) ) {
 170                                 throw new MWException("Cannot check code $this->code.");
 171                         } else {
 172                                 $this->results[$this->code] = $this->checkLanguage( $this->code );
 173                         }
 174                 }
 175         }
 176
 177         protected function getCheckBlacklist() {
 178                 global $checkBlacklist;
 179                 return $checkBlacklist;
 180         }
 181
 182         protected function checkLanguage( $code ) {
 183                 # Syntax check only
 184                 if ( $this->level === 0 ) {
 185                         $this->L->getMessages( $code );
 186                         return;
 187                 }
 188
 189                 $results = array();
 190                 $checkFunctions = $this->getChecks();
 191                 $checkBlacklist = $this->getCheckBlacklist();
 192                 foreach ( $this->checks as $check ) {
 193                         if ( isset($checkBlacklist[$code]) &&
 194                                 in_array($check, $checkBlacklist[$code]) ) {
 195                                 $result[$check] = array();
 196                                 continue;
 197                         }
 198
 199                         $callback = array( $this->L, $checkFunctions[$check] );
 200                         if ( !is_callable($callback ) ) {
 201                                 throw new MWException( "Unkown check $check." );
 202                         }
 203                         $results[$check] = call_user_func( $callback , $code );
 204                 }
 205
 206                 return $results;
 207         }
 208
 209         protected function formatKey( $key, $code ) {
 210                 if ( $this->doLinks ) {
 211                         $displayKey = ucfirst( $key );
 212                         if ( $code == $this->wikiCode ) {
 213                                 return "[[MediaWiki:$displayKey|$key]]";
 214                         } else {
 215                                 return "[[MediaWiki:$displayKey/$code|$key]]";
 216                         }
 217                 } else {
 218                         return $key;
 219                 }
 220         }
 221
 222         protected function outputText() {
 223                 foreach ( $this->results as $code => $results ) {
 224                         $translated = $this->L->getMessages( $code );
 225                         $translated = count( $translated['translated'] );
 226                         $translatable = $this->L->getGeneralMessages();
 227                         $translatable = count( $translatable['translatable'] );
 228                         foreach ( $results as $check => $messages ) {
 229                                 $count = count( $messages );
 230                                 if ( $count ) {
 231                                         $search = array( '$1', '$2', '$3' );
 232                                         $replace = array( $count, $check == 'untranslated' ? $translatable: $translated, $code );
 233                                         $descriptions = $this->getDescriptions();
 234                                         echo "\n" . str_replace( $search, $replace, $descriptions[$check] ) . "\n";
 235                                         if ( $this->level == 1 ) {
 236                                                 echo "[messages are hidden]\n";
 237                                         } else {
 238                                                 foreach ( $messages as $key => $value ) {
 239                                                         $displayKey = $this->formatKey( $key, $code );
 240                                                         if ( $this->level == 2 ) {
 241                                                                 echo "* $displayKey\n";
 242                                                         } else {
 243                                                                 echo "* $displayKey:            '$value'\n";
 244                                                         }
 245                                                 }
 246                                         }
 247                                 }
 248                         }
 249                 }
 250         }
 251
 252         /**
 253          * Globals: $wgContLang, $IP
 254          */
 255         function outputWiki() {
 256                 global $wgContLang, $IP;
 257                 $detailText = '';
 258                 $rows[] = '! Language !! Code !! Total !! ' . implode( ' !! ', $this->checks );
 259                 foreach ( $this->results as $code => $results ) {
 260                         $detailTextForLang = "==$code==\n";
 261                         $numbers = array();
 262                         $problems = 0;
 263                         $detailTextForLangChecks = array();
 264                         foreach ( $results as $check => $messages ) {
 265                                 $count = count( $messages );
 266                                 if ( $count ) {
 267                                         $problems += $count;
 268                                         $messageDetails = array();
 269                                         foreach ( $messages as $key => $details ) {
 270                                                 $displayKey = $this->formatKey( $key, $code );
 271                                                 $messageDetails[] = $displayKey;
 272                                         }
 273                                         $detailTextForLangChecks[] = "===$code-$check===\n* " . implode( ', ', $messageDetails );
 274                                         $numbers[] = "'''[[#$code-$check|$count]]'''";
 275                                 } else {
 276                                         $numbers[] = $count;
 277                                 }
 278
 279                         }
 280
 281                         if ( count( $detailTextForLangChecks ) ) {
 282                                 $detailText .= $detailTextForLang . implode( "\n", $detailTextForLangChecks ) . "\n";
 283                         }
 284
 285                         if ( !$problems ) { continue; } // Don't list languages without problems
 286                         $language = $wgContLang->getLanguageName( $code );
 287                         $rows[] = "| $language || $code || $problems || " . implode( ' || ', $numbers );
 288                 }
 289
 290                 $tableRows = implode( "\n|-\n", $rows );
 291
 292                 $version = SpecialVersion::getVersion( $IP );
 293                 echo <<<EOL
 294 '''Check results are for:''' <code>$version</code>
 295
 296
 297 {| class="sortable wikitable" border="2" cellpadding="4" cellspacing="0" style="background-color: #F9F9F9; border: 1px #AAAAAA solid; border-collapse: collapse; clear:both;"
 298 $tableRows
 299 |}
 300
 301 $detailText
 302
 303 EOL;
 304         }
 305 }
 306
 307 class CheckExtensionsCLI extends CheckLanguageCLI {
 308         private $extensions;
 309
 310         public function __construct( Array $options, $extension ) {
 311                 if ( isset( $options['help'] ) ) {
 312                         echo $this->help();
 313                         exit();
 314                 }
 315
 316                 if ( isset($options['lang']) ) {
 317                         $this->code = $options['lang'];
 318                 } else {
 319                         global $wgLanguageCode;
 320                         $this->code = $wgLanguageCode;
 321                 }
 322
 323                 if ( isset($options['level']) ) {
 324                         $this->level = $options['level'];
 325                 }
 326
 327                 $this->doLinks = isset($options['links']);
 328
 329                 if ( isset($options['wikilang']) ) {
 330                         $this->wikiCode = $options['wikilang'];
 331                 }
 332
 333                 if ( isset( $options['whitelist'] ) ) {
 334                         $this->checks = explode( ',', $options['whitelist'] );
 335                 } elseif ( isset( $options['blacklist'] ) ) {
 336                         $this->checks = array_diff(
 337                                 $this->defaultChecks,
 338                                 explode( ',', $options['blacklist'] )
 339                         );
 340                 } else {
 341                         $this->checks = $this->defaultChecks;
 342                 }
 343
 344                 if ( isset($options['output']) ) {
 345                         $this->output = $options['output'];
 346                 }
 347
 348                 # Some additional checks not enabled by default
 349                 if ( isset( $options['duplicate'] ) ) {
 350                         $this->checks[] = 'duplicate';
 351                 }
 352
 353                 if( $extension == 'all' ) {
 354                         $this->extensions = array();
 355                         foreach( MessageGroups::singleton()->getGroups() as $group ) {
 356                                 if( strpos( $group->getId(), 'ext-' ) === 0 && !$group->isMeta() ) {
 357                                         $this->extensions[] = new extensionLanguages( $group );
 358                                 }
 359                         }
 360                 } else {
 361                         $group = MessageGroups::getGroup( 'ext-' . $extension );
 362                         if( $group ) {
 363                                 $extension = new extensionLanguages( $group );
 364                                 $this->extensions = array( $extension );
 365                         } else {
 366                                 print "No such extension $extension.\n";
 367                                 $this->extensions = array();
 368                         }
 369                 }
 370         }
 371
 372         protected function help() {
 373                 return <<<ENDS
 374 Run this script to check the status of a specific language in extensions, or all of them.
 375 Command line settings are in form --parameter[=value], except for the first one.
 376 Parameters:
 377         * First parameter (mandatory): Extension name, or "all" for all the extensions.
 378         * lang: Language code (default: the installation default language).
 379         * help: Show this help.
 380         * level: Show the following level (default: 2).
 381         * links: Link the message values (default off).
 382         * wikilang: For the links, what is the content language of the wiki to display the output in (default en).
 383         * whitelist: Do only the following checks (form: code,code).
 384         * blacklist: Don't do the following checks (form: code,code).
 385         * duplicate: Additionally check for messages which are translated the same to English (default off).
 386 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):
 387         * untranslated: Messages which are required to translate, but are not translated.
 388         * duplicate: Messages which translation equal to fallback
 389         * obsolete: Messages which are untranslatable, but translated.
 390         * variables: Messages without variables which should be used.
 391         * empty: Empty messages.
 392         * whitespace: Messages which have trailing whitespace.
 393         * xhtml: Messages which are not well-formed XHTML (checks only few common errors).
 394         * chars: Messages with hidden characters.
 395         * links: Messages which contains broken links to pages (does not find all).
 396         * unbalanced: Messages which contains unequal numbers of opening {[ and closing ]}.
 397 Display levels (default: 2):
 398         * 0: Skip the checks (useful for checking syntax).
 399         * 1: Show only the stub headers and number of wrong messages, without list of messages.
 400         * 2: Show only the headers and the message keys, without the message values.
 401         * 3: Show both the headers and the complete messages, with both keys and values.
 402
 403 ENDS;
 404         }
 405
 406         public function execute() {
 407                 $this->doChecks();
 408         }
 409
 410         protected function checkLanguage( $code ) {
 411                 foreach( $this->extensions as $extension ) {
 412                         echo $extension->name() . ":\n";
 413
 414                         $this->L = $extension;
 415                         $this->results = array();
 416                         $this->results[$code] = parent::checkLanguage( $code );
 417
 418                         if( $this->level > 0 ) {
 419                                 switch( $this->output ) {
 420                                         case 'plain':
 421                                                 $this->outputText();
 422                                                 break;
 423                                         case 'wiki':
 424                                                 $this->outputWiki();
 425                                                 break;
 426                                         default:
 427                                                 throw new MWException( "Invalid output type $this->output" );
 428                                 }
 429                         }
 430
 431                         echo "\n";
 432                 }
 433         }
 434 }
 435
 436 # Blacklist some checks for some languages
 437 $checkBlacklist = array(
 438 #'code'        => array( 'check1', 'check2' ... )
 439 'gan'          => array( 'plural' ),
 440 'hak'          => array( 'plural' ),
 441 'ja'           => array( 'plural' ), // Does not use plural
 442 'ka'           => array( 'plural' ),
 443 'kk-arab'      => array( 'plural' ),
 444 'kk-cyrl'      => array( 'plural' ),
 445 'kk-latn'      => array( 'plural' ),
 446 'ko'           => array( 'plural' ),
 447 'mn'           => array( 'plural' ),
 448 'ms'           => array( 'plural' ),
 449 'my'           => array( 'chars' ),  // Uses a lot zwnj
 450 'sq'           => array( 'plural' ),
 451 'tet'          => array( 'plural' ),
 452 'th'           => array( 'plural' ),
 453 'wuu'          => array( 'plural' ),
 454 'xmf'          => array( 'plural' ),
 455 'yue'          => array( 'plural' ),
 456 'zh'           => array( 'plural' ),
 457 'zh-classical' => array( 'plural' ),
 458 'zh-cn'        => array( 'plural' ),
 459 'zh-hans'      => array( 'plural' ),
 460 'zh-hant'      => array( 'plural' ),
 461 'zh-hk'        => array( 'plural' ),
 462 'zh-sg'        => array( 'plural' ),
 463 'zh-tw'        => array( 'plural' ),
 464 'zh-yue'       => array( 'plural' ),
 465 );