3 * Helper class for checkLanguage.php script.
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
21 * @ingroup MaintenanceLanguage
25 * @ingroup MaintenanceLanguage
27 class CheckLanguageCLI {
28 protected $code = null;
30 protected $doLinks = false;
31 protected $linksPrefix = '';
32 protected $wikiCode = 'en';
33 protected $checkAll = false;
34 protected $output = 'plain';
35 protected $checks = array();
38 protected $results = array();
40 private $includeExif = false;
44 * @param $options Options for script.
46 public function __construct( Array $options ) {
47 if ( isset( $options['help'] ) ) {
52 if ( isset( $options['lang'] ) ) {
53 $this->code = $options['lang'];
55 global $wgLanguageCode;
56 $this->code = $wgLanguageCode;
59 if ( isset( $options['level'] ) ) {
60 $this->level = $options['level'];
63 $this->doLinks = isset( $options['links'] );
64 $this->includeExif = !isset( $options['noexif'] );
65 $this->checkAll = isset( $options['all'] );
67 if ( isset( $options['prefix'] ) ) {
68 $this->linksPrefix = $options['prefix'];
71 if ( isset( $options['wikilang'] ) ) {
72 $this->wikiCode = $options['wikilang'];
75 if ( isset( $options['whitelist'] ) ) {
76 $this->checks = explode( ',', $options['whitelist'] );
77 } elseif ( isset( $options['blacklist'] ) ) {
78 $this->checks = array_diff(
79 isset( $options['easy'] ) ? $this->easyChecks() : $this->defaultChecks(),
80 explode( ',', $options['blacklist'] )
82 } elseif ( isset( $options['easy'] ) ) {
83 $this->checks = $this->easyChecks();
85 $this->checks = $this->defaultChecks();
88 if ( isset( $options['output'] ) ) {
89 $this->output = $options['output'];
92 $this->L = new languages( $this->includeExif );
96 * Get the default checks.
97 * @return A list of the default checks.
99 protected function defaultChecks() {
101 'untranslated', 'duplicate', 'obsolete', 'variables', 'empty', 'plural',
102 'whitespace', 'xhtml', 'chars', 'links', 'unbalanced', 'namespace',
103 'projecttalk', 'magic', 'magic-old', 'magic-over', 'magic-case',
104 'special', 'special-old',
109 * Get the checks which check other things than messages.
110 * @return A list of the non-message checks.
112 protected function nonMessageChecks() {
114 'namespace', 'projecttalk', 'magic', 'magic-old', 'magic-over',
115 'magic-case', 'special', 'special-old',
120 * Get the checks that can easily be treated by non-speakers of the language.
121 * @return Array A list of the easy checks.
123 protected function easyChecks() {
125 'duplicate', 'obsolete', 'empty', 'whitespace', 'xhtml', 'chars', 'magic-old',
126 'magic-over', 'magic-case', 'special-old',
132 * @return An array of all check names mapped to their function names.
134 protected function getChecks() {
136 'untranslated' => 'getUntranslatedMessages',
137 'duplicate' => 'getDuplicateMessages',
138 'obsolete' => 'getObsoleteMessages',
139 'variables' => 'getMessagesWithMismatchVariables',
140 'plural' => 'getMessagesWithoutPlural',
141 'empty' => 'getEmptyMessages',
142 'whitespace' => 'getMessagesWithWhitespace',
143 'xhtml' => 'getNonXHTMLMessages',
144 'chars' => 'getMessagesWithWrongChars',
145 'links' => 'getMessagesWithDubiousLinks',
146 'unbalanced' => 'getMessagesWithUnbalanced',
147 'namespace' => 'getUntranslatedNamespaces',
148 'projecttalk' => 'getProblematicProjectTalks',
149 'magic' => 'getUntranslatedMagicWords',
150 'magic-old' => 'getObsoleteMagicWords',
151 'magic-over' => 'getOverridingMagicWords',
152 'magic-case' => 'getCaseMismatchMagicWords',
153 'special' => 'getUntraslatedSpecialPages',
154 'special-old' => 'getObsoleteSpecialPages',
159 * Get total count for each check non-messages check.
160 * @return An array of all check names mapped to a two-element array:
161 * function name to get the total count and language code or null
164 protected function getTotalCount() {
166 'namespace' => array( 'getNamespaceNames', 'en' ),
167 'projecttalk' => null,
168 'magic' => array( 'getMagicWords', 'en' ),
169 'magic-old' => array( 'getMagicWords', null ),
170 'magic-over' => array( 'getMagicWords', null ),
171 'magic-case' => array( 'getMagicWords', null ),
172 'special' => array( 'getSpecialPageAliases', 'en' ),
173 'special-old' => array( 'getSpecialPageAliases', null ),
178 * Get all check descriptions.
179 * @return An array of all check names mapped to their descriptions.
181 protected function getDescriptions() {
183 'untranslated' => '$1 message(s) of $2 are not translated to $3, but exist in en:',
184 'duplicate' => '$1 message(s) of $2 are translated the same in en and $3:',
185 'obsolete' => '$1 message(s) of $2 do not exist in en or are in the ignore list, but exist in $3:',
186 'variables' => '$1 message(s) of $2 in $3 don\'t match the variables used in en:',
187 'plural' => '$1 message(s) of $2 in $3 don\'t use {{plural}} while en uses:',
188 'empty' => '$1 message(s) of $2 in $3 are empty or -:',
189 'whitespace' => '$1 message(s) of $2 in $3 have trailing whitespace:',
190 'xhtml' => '$1 message(s) of $2 in $3 contain illegal XHTML:',
191 'chars' => '$1 message(s) of $2 in $3 include hidden chars which should not be used in the messages:',
192 'links' => '$1 message(s) of $2 in $3 have problematic link(s):',
193 'unbalanced' => '$1 message(s) of $2 in $3 have unbalanced {[]}:',
194 'namespace' => '$1 namespace name(s) of $2 are not translated to $3, but exist in en:',
195 'projecttalk' => '$1 namespace name(s) and alias(es) in $3 are project talk namespaces without the parameter:',
196 'magic' => '$1 magic word(s) of $2 are not translated to $3, but exist in en:',
197 'magic-old' => '$1 magic word(s) of $2 do not exist in en, but exist in $3:',
198 'magic-over' => '$1 magic word(s) of $2 in $3 do not contain the original en word(s):',
199 'magic-case' => '$1 magic word(s) of $2 in $3 change the case-sensitivity of the original en word:',
200 'special' => '$1 special page alias(es) of $2 are not translated to $3, but exist in en:',
201 'special-old' => '$1 special page alias(es) of $2 do not exist in en, but exist in $3:',
207 * @return The help string.
209 protected function help() {
211 Run this script to check a specific language file, or all of them.
212 Command line settings are in form --parameter[=value].
214 --help: Show this help.
215 --lang: Language code (default: the installation default language).
216 --all: Check all customized languages.
217 --level: Show the following display level (default: 2):
218 * 0: Skip the checks (useful for checking syntax).
219 * 1: Show only the stub headers and number of wrong messages, without list of messages.
220 * 2: Show only the headers and the message keys, without the message values.
221 * 3: Show both the headers and the complete messages, with both keys and values.
222 --links: Link the message values (default off).
223 --prefix: prefix to add to links.
224 --wikilang: For the links, what is the content language of the wiki to display the output in (default en).
225 --noexif: Don't check for EXIF messages (a bit hard and boring to translate), if you know
226 that they are currently not translated and want to focus on other problems (default off).
227 --whitelist: Do only the following checks (form: code,code).
228 --blacklist: Don't do the following checks (form: code,code).
229 --easy: Do only the easy checks, which can be treated by non-speakers of the language.
231 Check codes (ideally, all of them should result 0; all the checks are executed by default (except language-specific check blacklists in checkLanguage.inc):
232 * untranslated: Messages which are required to translate, but are not translated.
233 * duplicate: Messages which translation equal to fallback
234 * obsolete: Messages which are untranslatable or do not exist, but are translated.
235 * variables: Messages without variables which should be used, or with variables which shouldn't be used.
236 * empty: Empty messages and messages that contain only -.
237 * whitespace: Messages which have trailing whitespace.
238 * xhtml: Messages which are not well-formed XHTML (checks only few common errors).
239 * chars: Messages with hidden characters.
240 * links: Messages which contains broken links to pages (does not find all).
241 * unbalanced: Messages which contains unequal numbers of opening {[ and closing ]}.
242 * namespace: Namespace names that were not translated.
243 * projecttalk: Namespace names and aliases where the project talk does not contain $1.
244 * magic: Magic words that were not translated.
245 * magic-old: Magic words which do not exist.
246 * magic-over: Magic words that override the original English word.
247 * magic-case: Magic words whose translation changes the case-sensitivity of the original English word.
248 * special: Special page names that were not translated.
249 * special-old: Special page names which do not exist.
255 * Execute the script.
257 public function execute() {
259 if ( $this->level > 0 ) {
260 switch ( $this->output ) {
268 throw new MWException( "Invalid output type $this->output" );
274 * Execute the checks.
276 protected function doChecks() {
277 $ignoredCodes = array( 'en', 'enRTL' );
279 $this->results = array();
281 if ( $this->checkAll ) {
282 foreach ( $this->L->getLanguages() as $language ) {
283 if ( !in_array( $language, $ignoredCodes ) ) {
284 $this->results[$language] = $this->checkLanguage( $language );
288 if ( in_array( $this->code, $ignoredCodes ) ) {
289 throw new MWException( "Cannot check code $this->code." );
291 $this->results[$this->code] = $this->checkLanguage( $this->code );
297 * Get the check blacklist.
298 * @return The list of checks which should not be executed.
300 protected function getCheckBlacklist() {
301 global $checkBlacklist;
302 return $checkBlacklist;
307 * @param $code The language code.
308 * @return The results.
310 protected function checkLanguage( $code ) {
312 if ( $this->level === 0 ) {
313 $this->L->getMessages( $code );
318 $checkFunctions = $this->getChecks();
319 $checkBlacklist = $this->getCheckBlacklist();
320 foreach ( $this->checks as $check ) {
321 if ( isset( $checkBlacklist[$code] ) &&
322 in_array( $check, $checkBlacklist[$code] ) ) {
323 $results[$check] = array();
327 $callback = array( $this->L, $checkFunctions[$check] );
328 if ( !is_callable( $callback ) ) {
329 throw new MWException( "Unkown check $check." );
331 $results[$check] = call_user_func( $callback, $code );
338 * Format a message key.
339 * @param $key The message key.
340 * @param $code The language code.
341 * @return The formatted message key.
343 protected function formatKey( $key, $code ) {
344 if ( $this->doLinks ) {
345 $displayKey = ucfirst( $key );
346 if ( $code == $this->wikiCode ) {
347 return "[[{$this->linksPrefix}MediaWiki:$displayKey|$key]]";
349 return "[[{$this->linksPrefix}MediaWiki:$displayKey/$code|$key]]";
357 * Output the checks results as plain text.
358 * @return The checks results as plain text.
360 protected function outputText() {
361 foreach ( $this->results as $code => $results ) {
362 $translated = $this->L->getMessages( $code );
363 $translated = count( $translated['translated'] );
364 foreach ( $results as $check => $messages ) {
365 $count = count( $messages );
367 if ( $check == 'untranslated' ) {
368 $translatable = $this->L->getGeneralMessages();
369 $total = count( $translatable['translatable'] );
370 } elseif ( in_array( $check, $this->nonMessageChecks() ) ) {
371 $totalCount = $this->getTotalCount();
372 $totalCount = $totalCount[$check];
373 $callback = array( $this->L, $totalCount[0] );
374 $callCode = $totalCount[1] ? $totalCount[1] : $code;
375 $total = count( call_user_func( $callback, $callCode ) );
377 $total = $translated;
379 $search = array( '$1', '$2', '$3' );
380 $replace = array( $count, $total, $code );
381 $descriptions = $this->getDescriptions();
382 echo "\n" . str_replace( $search, $replace, $descriptions[$check] ) . "\n";
383 if ( $this->level == 1 ) {
384 echo "[messages are hidden]\n";
386 foreach ( $messages as $key => $value ) {
387 if( !in_array( $check, $this->nonMessageChecks() ) ) {
388 $key = $this->formatKey( $key, $code );
390 if ( $this->level == 2 || empty( $value ) ) {
393 echo "* $key: '$value'\n";
403 * Output the checks results as wiki text.
404 * @return The checks results as wiki text.
406 function outputWiki() {
409 $rows[] = '! Language !! Code !! Total !! ' . implode( ' !! ', array_diff( $this->checks, $this->nonMessageChecks() ) );
410 foreach ( $this->results as $code => $results ) {
411 $detailTextForLang = "==$code==\n";
414 $detailTextForLangChecks = array();
415 foreach ( $results as $check => $messages ) {
416 if( in_array( $check, $this->nonMessageChecks() ) ) {
419 $count = count( $messages );
422 $messageDetails = array();
423 foreach ( $messages as $key => $details ) {
424 $displayKey = $this->formatKey( $key, $code );
425 $messageDetails[] = $displayKey;
427 $detailTextForLangChecks[] = "=== $code-$check ===\n* " . implode( ', ', $messageDetails );
428 $numbers[] = "'''[[#$code-$check|$count]]'''";
435 if ( count( $detailTextForLangChecks ) ) {
436 $detailText .= $detailTextForLang . implode( "\n", $detailTextForLangChecks ) . "\n";
440 # Don't list languages without problems
443 $language = $wgContLang->getLanguageName( $code );
444 $rows[] = "| $language || $code || $problems || " . implode( ' || ', $numbers );
447 $tableRows = implode( "\n|-\n", $rows );
449 $version = SpecialVersion::getVersion( 'nodb' );
451 '''Check results are for:''' <code>$version</code>
454 {| class="sortable wikitable" border="2" cellpadding="4" cellspacing="0" style="background-color: #F9F9F9; border: 1px #AAAAAA solid; border-collapse: collapse; clear: both;"
464 * Check if there are any results for the checks, in any language.
465 * @return True if there are any results, false if not.
467 protected function isEmpty() {
468 foreach( $this->results as $results ) {
469 foreach( $results as $messages ) {
470 if( !empty( $messages ) ) {
480 * @ingroup MaintenanceLanguage
482 class CheckExtensionsCLI extends CheckLanguageCLI {
487 * @param $options Options for script.
488 * @param $extension The extension name (or names).
490 public function __construct( Array $options, $extension ) {
491 if ( isset( $options['help'] ) ) {
496 if ( isset( $options['lang'] ) ) {
497 $this->code = $options['lang'];
499 global $wgLanguageCode;
500 $this->code = $wgLanguageCode;
503 if ( isset( $options['level'] ) ) {
504 $this->level = $options['level'];
507 $this->doLinks = isset( $options['links'] );
509 if ( isset( $options['wikilang'] ) ) {
510 $this->wikiCode = $options['wikilang'];
513 if ( isset( $options['whitelist'] ) ) {
514 $this->checks = explode( ',', $options['whitelist'] );
515 } elseif ( isset( $options['blacklist'] ) ) {
516 $this->checks = array_diff(
517 isset( $options['easy'] ) ? $this->easyChecks() : $this->defaultChecks(),
518 explode( ',', $options['blacklist'] )
520 } elseif ( isset( $options['easy'] ) ) {
521 $this->checks = $this->easyChecks();
523 $this->checks = $this->defaultChecks();
526 if ( isset( $options['output'] ) ) {
527 $this->output = $options['output'];
530 # Some additional checks not enabled by default
531 if ( isset( $options['duplicate'] ) ) {
532 $this->checks[] = 'duplicate';
535 $this->extensions = array();
536 $extensions = new PremadeMediawikiExtensionGroups();
537 $extensions->addAll();
538 if ( $extension == 'all' ) {
539 foreach ( MessageGroups::singleton()->getGroups() as $group ) {
540 if ( strpos( $group->getId(), 'ext-' ) === 0 && !$group->isMeta() ) {
541 $this->extensions[] = new extensionLanguages( $group );
544 } elseif ( $extension == 'wikimedia' ) {
545 $wikimedia = MessageGroups::getGroup( 'ext-0-wikimedia' );
546 foreach ( $wikimedia->wmfextensions() as $extension ) {
547 $group = MessageGroups::getGroup( $extension );
548 $this->extensions[] = new extensionLanguages( $group );
550 } elseif ( $extension == 'flaggedrevs' ) {
551 foreach ( MessageGroups::singleton()->getGroups() as $group ) {
552 if ( strpos( $group->getId(), 'ext-flaggedrevs-' ) === 0 && !$group->isMeta() ) {
553 $this->extensions[] = new extensionLanguages( $group );
557 $extensions = explode( ',', $extension );
558 foreach ( $extensions as $extension ) {
559 $group = MessageGroups::getGroup( 'ext-' . $extension );
561 $extension = new extensionLanguages( $group );
562 $this->extensions[] = $extension;
564 print "No such extension $extension.\n";
571 * Get the default checks.
572 * @return A list of the default checks.
574 protected function defaultChecks() {
576 'untranslated', 'duplicate', 'obsolete', 'variables', 'empty', 'plural',
577 'whitespace', 'xhtml', 'chars', 'links', 'unbalanced',
582 * Get the checks which check other things than messages.
583 * @return A list of the non-message checks.
585 protected function nonMessageChecks() {
590 * Get the checks that can easily be treated by non-speakers of the language.
591 * @return A list of the easy checks.
593 protected function easyChecks() {
595 'duplicate', 'obsolete', 'empty', 'whitespace', 'xhtml', 'chars',
601 * @return The help string.
603 protected function help() {
605 Run this script to check the status of a specific language in extensions, or all of them.
606 Command line settings are in form --parameter[=value], except for the first one.
608 * First parameter (mandatory): Extension name, multiple extension names (separated by commas), "all" for all the extensions, "wikimedia" for extensions used by Wikimedia or "flaggedrevs" for all FLaggedRevs extension messages.
609 * lang: Language code (default: the installation default language).
610 * help: Show this help.
611 * level: Show the following display level (default: 2).
612 * links: Link the message values (default off).
613 * wikilang: For the links, what is the content language of the wiki to display the output in (default en).
614 * whitelist: Do only the following checks (form: code,code).
615 * blacklist: Do not perform the following checks (form: code,code).
616 * easy: Do only the easy checks, which can be treated by non-speakers of the language.
617 Check codes (ideally, all of them should result 0; all the checks are executed by default (except language-specific check blacklists in checkLanguage.inc):
618 * untranslated: Messages which are required to translate, but are not translated.
619 * duplicate: Messages which translation equal to fallback
620 * obsolete: Messages which are untranslatable, but translated.
621 * variables: Messages without variables which should be used, or with variables which should not be used.
622 * empty: Empty messages.
623 * whitespace: Messages which have trailing whitespace.
624 * xhtml: Messages which are not well-formed XHTML (checks only few common errors).
625 * chars: Messages with hidden characters.
626 * links: Messages which contains broken links to pages (does not find all).
627 * unbalanced: Messages which contains unequal numbers of opening {[ and closing ]}.
628 Display levels (default: 2):
629 * 0: Skip the checks (useful for checking syntax).
630 * 1: Show only the stub headers and number of wrong messages, without list of messages.
631 * 2: Show only the headers and the message keys, without the message values.
632 * 3: Show both the headers and the complete messages, with both keys and values.
638 * Execute the script.
640 public function execute() {
645 * Check a language and show the results.
646 * @param $code The language code.
648 protected function checkLanguage( $code ) {
649 foreach( $this->extensions as $extension ) {
650 $this->L = $extension;
651 $this->results = array();
652 $this->results[$code] = parent::checkLanguage( $code );
654 if( !$this->isEmpty() ) {
655 echo $extension->name() . ":\n";
657 if( $this->level > 0 ) {
658 switch( $this->output ) {
666 throw new MWException( "Invalid output type $this->output" );
676 # Blacklist some checks for some languages
677 $checkBlacklist = array(
678 #'code' => array( 'check1', 'check2' ... )
679 'az' => array( 'plural' ),
680 'bo' => array( 'plural' ),
681 'dz' => array( 'plural' ),
682 'id' => array( 'plural' ),
683 'fa' => array( 'plural' ),
684 'gan' => array( 'plural' ),
685 'gan-hans' => array( 'plural' ),
686 'gan-hant' => array( 'plural' ),
687 'gn' => array( 'plural' ),
688 'hak' => array( 'plural' ),
689 'hu' => array( 'plural' ),
690 'ja' => array( 'plural' ), // Does not use plural
691 'jv' => array( 'plural' ),
692 'ka' => array( 'plural' ),
693 'kk-arab' => array( 'plural' ),
694 'kk-cyrl' => array( 'plural' ),
695 'kk-latn' => array( 'plural' ),
696 'km' => array( 'plural' ),
697 'kn' => array( 'plural' ),
698 'ko' => array( 'plural' ),
699 'lzh' => array( 'plural' ),
700 'mn' => array( 'plural' ),
701 'ms' => array( 'plural' ),
702 'my' => array( 'plural', 'chars' ), // Uses a lot zwnj
703 'sah' => array( 'plural' ),
704 'sq' => array( 'plural' ),
705 'tet' => array( 'plural' ),
706 'th' => array( 'plural' ),
707 'to' => array( 'plural' ),
708 'tr' => array( 'plural' ),
709 'vi' => array( 'plural' ),
710 'wuu' => array( 'plural' ),
711 'xmf' => array( 'plural' ),
712 'yo' => array( 'plural' ),
713 'yue' => array( 'plural' ),
714 'zh' => array( 'plural' ),
715 'zh-classical' => array( 'plural' ),
716 'zh-cn' => array( 'plural' ),
717 'zh-hans' => array( 'plural' ),
718 'zh-hant' => array( 'plural' ),
719 'zh-hk' => array( 'plural' ),
720 'zh-sg' => array( 'plural' ),
721 'zh-tw' => array( 'plural' ),
722 'zh-yue' => array( 'plural' ),