diff options
Diffstat (limited to 'maintenance/language/checkLanguage.inc')
| -rw-r--r-- | maintenance/language/checkLanguage.inc | 376 |
1 files changed, 277 insertions, 99 deletions
diff --git a/maintenance/language/checkLanguage.inc b/maintenance/language/checkLanguage.inc index 2cfd1b04..52281b57 100644 --- a/maintenance/language/checkLanguage.inc +++ b/maintenance/language/checkLanguage.inc @@ -13,41 +13,36 @@ class CheckLanguageCLI { protected $checks = array(); protected $L = null; - protected $defaultChecks = array( - 'untranslated', 'obsolete', 'variables', 'empty', 'plural', - 'whitespace', 'xhtml', 'chars', 'links', 'unbalanced' - ); - protected $results = array(); private $includeExif = false; /** - * GLOBALS: $wgLanguageCode; + * Constructor. + * @param $options Options for script. */ public function __construct( Array $options ) { - if ( isset( $options['help'] ) ) { echo $this->help(); exit(); } - if ( isset($options['lang']) ) { + if ( isset( $options['lang'] ) ) { $this->code = $options['lang']; } else { global $wgLanguageCode; $this->code = $wgLanguageCode; } - if ( isset($options['level']) ) { + if ( isset( $options['level'] ) ) { $this->level = $options['level']; } - $this->doLinks = isset($options['links']); - $this->includeExif = !isset($options['noexif']); - $this->checkAll = isset($options['all']); + $this->doLinks = isset( $options['links'] ); + $this->includeExif = !isset( $options['noexif'] ); + $this->checkAll = isset( $options['all'] ); - if ( isset($options['wikilang']) ) { + if ( isset( $options['wikilang'] ) ) { $this->wikiCode = $options['wikilang']; } @@ -55,57 +50,136 @@ class CheckLanguageCLI { $this->checks = explode( ',', $options['whitelist'] ); } elseif ( isset( $options['blacklist'] ) ) { $this->checks = array_diff( - $this->defaultChecks, + isset( $options['easy'] ) ? $this->easyChecks() : $this->defaultChecks(), explode( ',', $options['blacklist'] ) ); + } elseif ( isset( $options['easy'] ) ) { + $this->checks = $this->easyChecks(); } else { - $this->checks = $this->defaultChecks; + $this->checks = $this->defaultChecks(); } - if ( isset($options['output']) ) { + if ( isset( $options['output'] ) ) { $this->output = $options['output']; } - # Some additional checks not enabled by default - if ( isset( $options['duplicate'] ) ) { - $this->checks[] = 'duplicate'; - } - $this->L = new languages( $this->includeExif ); } + /** + * Get the default checks. + * @return A list of the default checks. + */ + protected function defaultChecks() { + return array( + 'untranslated', 'duplicate', 'obsolete', 'variables', 'empty', 'plural', + 'whitespace', 'xhtml', 'chars', 'links', 'unbalanced', 'namespace', + 'projecttalk', 'magic', 'magic-old', 'magic-over', 'magic-case', + 'special', 'special-old', + ); + } + + /** + * Get the checks which check other things than messages. + * @return A list of the non-message checks. + */ + protected function nonMessageChecks() { + return array( + 'namespace', 'projecttalk', 'magic', 'magic-old', 'magic-over', + 'magic-case', 'special', 'special-old', + ); + } + + /** + * Get the checks that can easily be treated by non-speakers of the language. + * @return A list of the easy checks. + */ + protected function easyChecks() { + return array( + 'duplicate', 'obsolete', 'empty', 'whitespace', 'xhtml', 'chars', 'magic-old', + 'magic-over', 'magic-case', 'special-old', + ); + } + + /** + * Get all checks. + * @return An array of all check names mapped to their function names. + */ protected function getChecks() { - $checks = array(); - $checks['untranslated'] = 'getUntranslatedMessages'; - $checks['duplicate'] = 'getDuplicateMessages'; - $checks['obsolete'] = 'getObsoleteMessages'; - $checks['variables'] = 'getMessagesWithoutVariables'; - $checks['plural'] = 'getMessagesWithoutPlural'; - $checks['empty'] = 'getEmptyMessages'; - $checks['whitespace'] = 'getMessagesWithWhitespace'; - $checks['xhtml'] = 'getNonXHTMLMessages'; - $checks['chars'] = 'getMessagesWithWrongChars'; - $checks['links'] = 'getMessagesWithDubiousLinks'; - $checks['unbalanced'] = 'getMessagesWithUnbalanced'; - return $checks; + return array( + 'untranslated' => 'getUntranslatedMessages', + 'duplicate' => 'getDuplicateMessages', + 'obsolete' => 'getObsoleteMessages', + 'variables' => 'getMessagesWithMismatchVariables', + 'plural' => 'getMessagesWithoutPlural', + 'empty' => 'getEmptyMessages', + 'whitespace' => 'getMessagesWithWhitespace', + 'xhtml' => 'getNonXHTMLMessages', + 'chars' => 'getMessagesWithWrongChars', + 'links' => 'getMessagesWithDubiousLinks', + 'unbalanced' => 'getMessagesWithUnbalanced', + 'namespace' => 'getUntranslatedNamespaces', + 'projecttalk' => 'getProblematicProjectTalks', + 'magic' => 'getUntranslatedMagicWords', + 'magic-old' => 'getObsoleteMagicWords', + 'magic-over' => 'getOverridingMagicWords', + 'magic-case' => 'getCaseMismatchMagicWords', + 'special' => 'getUntraslatedSpecialPages', + 'special-old' => 'getObsoleteSpecialPages', + ); + } + + /** + * Get total count for each check non-messages check. + * @return An array of all check names mapped to a two-element array: + * function name to get the total count and language code or null + * for checked code. + */ + protected function getTotalCount() { + return array( + 'namespace' => array( 'getNamespaceNames', 'en' ), + 'projecttalk' => null, + 'magic' => array( 'getMagicWords', 'en' ), + 'magic-old' => array( 'getMagicWords', null ), + 'magic-over' => array( 'getMagicWords', null ), + 'magic-case' => array( 'getMagicWords', null ), + 'special' => array( 'getSpecialPageAliases', 'en' ), + 'special-old' => array( 'getSpecialPageAliases', null ), + ); } + /** + * Get all check descriptions. + * @return An array of all check names mapped to their descriptions. + */ protected function getDescriptions() { - $descriptions = array(); - $descriptions['untranslated'] = '$1 message(s) of $2 are not translated to $3, but exist in en:'; - $descriptions['duplicate'] = '$1 message(s) of $2 are translated the same in en and $3:'; - $descriptions['obsolete'] = '$1 message(s) of $2 do not exist in en or are in the ignore list, but are in $3'; - $descriptions['variables'] = '$1 message(s) of $2 in $3 don\'t use some variables that en uses:'; - $descriptions['plural'] = '$1 message(s) of $2 in $3 don\'t use {{plural}} while en uses:'; - $descriptions['empty'] = '$1 message(s) of $2 in $3 are empty or -:'; - $descriptions['whitespace'] = '$1 message(s) of $2 in $3 have trailing whitespace:'; - $descriptions['xhtml'] = '$1 message(s) of $2 in $3 contain illegal XHTML:'; - $descriptions['chars'] = '$1 message(s) of $2 in $3 include hidden chars which should not be used in the messages:'; - $descriptions['links'] = '$1 message(s) of $2 in $3 have problematic link(s):'; - $descriptions['unbalanced'] = '$1 message(s) of $2 in $3 have unbalanced {[]}:'; - return $descriptions; + return array( + 'untranslated' => '$1 message(s) of $2 are not translated to $3, but exist in en:', + 'duplicate' => '$1 message(s) of $2 are translated the same in en and $3:', + 'obsolete' => '$1 message(s) of $2 do not exist in en or are in the ignore list, but exist in $3:', + 'variables' => '$1 message(s) of $2 in $3 don\'t match the variables used in en:', + 'plural' => '$1 message(s) of $2 in $3 don\'t use {{plural}} while en uses:', + 'empty' => '$1 message(s) of $2 in $3 are empty or -:', + 'whitespace' => '$1 message(s) of $2 in $3 have trailing whitespace:', + 'xhtml' => '$1 message(s) of $2 in $3 contain illegal XHTML:', + 'chars' => '$1 message(s) of $2 in $3 include hidden chars which should not be used in the messages:', + 'links' => '$1 message(s) of $2 in $3 have problematic link(s):', + 'unbalanced' => '$1 message(s) of $2 in $3 have unbalanced {[]}:', + 'namespace' => '$1 namespace name(s) of $2 are not translated to $3, but exist in en:', + 'projecttalk' => '$1 namespace name(s) and alias(es) in $3 are project talk namespaces without the parameter:', + 'magic' => '$1 magic word(s) of $2 are not translated to $3, but exist in en:', + 'magic-old' => '$1 magic word(s) of $2 do not exist in en, but exist in $3:', + 'magic-over' => '$1 magic word(s) of $2 in $3 do not contain the original en word(s):', + 'magic-case' => '$1 magic word(s) of $2 in $3 change the case-sensitivity of the original en word:', + 'special' => '$1 special page alias(es) of $2 are not translated to $3, but exist in en:', + 'special-old' => '$1 special page alias(es) of $2 do not exist in en, but exist in $3:', + ); } + /** + * Get help. + * @return The help string. + */ protected function help() { return <<<ENDS Run this script to check a specific language file, or all of them. @@ -114,24 +188,32 @@ Parameters: * lang: Language code (default: the installation default language). * all: Check all customized languages. * help: Show this help. - * level: Show the following level (default: 2). + * level: Show the following display level (default: 2). * links: Link the message values (default off). * wikilang: For the links, what is the content language of the wiki to display the output in (default en). * whitelist: Do only the following checks (form: code,code). * blacklist: Don't do the following checks (form: code,code). - * duplicate: Additionally check for messages which are translated the same to English (default off). + * easy: Do only the easy checks, which can be treated by non-speakers of the language. * 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). -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): +Check codes (ideally, all of them should result 0; all the checks are executed by default (except language-specific check blacklists in checkLanguage.inc): * untranslated: Messages which are required to translate, but are not translated. * duplicate: Messages which translation equal to fallback - * obsolete: Messages which are untranslatable, but translated. - * variables: Messages without variables which should be used. - * empty: Empty messages. + * obsolete: Messages which are untranslatable or do not exist, but are translated. + * variables: Messages without variables which should be used, or with variables which shouldn't be used. + * empty: Empty messages and messages that contain only -. * whitespace: Messages which have trailing whitespace. * xhtml: Messages which are not well-formed XHTML (checks only few common errors). * chars: Messages with hidden characters. * links: Messages which contains broken links to pages (does not find all). * unbalanced: Messages which contains unequal numbers of opening {[ and closing ]}. + * namespace: Namespace names that were not translated. + * projecttalk: Namespace names and aliases where the project talk does not contain $1. + * magic: Magic words that were not translated. + * magic-old: Magic words which do not exist. + * magic-over: Magic words that override the original English word. + * magic-case: Magic words whose translation changes the case-sensitivity of the original English word. + * special: Special page names that were not translated. + * special-old: Special page names which do not exist. Display levels (default: 2): * 0: Skip the checks (useful for checking syntax). * 1: Show only the stub headers and number of wrong messages, without list of messages. @@ -141,10 +223,13 @@ Display levels (default: 2): ENDS; } + /** + * Execute the script. + */ public function execute() { $this->doChecks(); if ( $this->level > 0 ) { - switch ($this->output) { + switch ( $this->output ) { case 'plain': $this->outputText(); break; @@ -152,11 +237,14 @@ ENDS; $this->outputWiki(); break; default: - throw new MWException( "Invalid output type $this->output"); + throw new MWException( "Invalid output type $this->output" ); } } } + /** + * Execute the checks. + */ protected function doChecks() { $ignoredCodes = array( 'en', 'enRTL' ); @@ -164,24 +252,33 @@ ENDS; # Check the language if ( $this->checkAll ) { foreach ( $this->L->getLanguages() as $language ) { - if ( !in_array($language, $ignoredCodes) ) { + if ( !in_array( $language, $ignoredCodes ) ) { $this->results[$language] = $this->checkLanguage( $language ); } } } else { - if ( in_array($this->code, $ignoredCodes) ) { - throw new MWException("Cannot check code $this->code."); + if ( in_array( $this->code, $ignoredCodes ) ) { + throw new MWException( "Cannot check code $this->code." ); } else { $this->results[$this->code] = $this->checkLanguage( $this->code ); } } } + /** + * Get the check blacklist. + * @return The list of checks which should not be executed. + */ protected function getCheckBlacklist() { global $checkBlacklist; return $checkBlacklist; } + /** + * Check a language. + * @param $code The language code. + * @return The results. + */ protected function checkLanguage( $code ) { # Syntax check only if ( $this->level === 0 ) { @@ -193,22 +290,28 @@ ENDS; $checkFunctions = $this->getChecks(); $checkBlacklist = $this->getCheckBlacklist(); foreach ( $this->checks as $check ) { - if ( isset($checkBlacklist[$code]) && - in_array($check, $checkBlacklist[$code]) ) { + if ( isset( $checkBlacklist[$code] ) && + in_array( $check, $checkBlacklist[$code] ) ) { $result[$check] = array(); continue; } $callback = array( $this->L, $checkFunctions[$check] ); - if ( !is_callable($callback ) ) { + if ( !is_callable( $callback ) ) { throw new MWException( "Unkown check $check." ); } - $results[$check] = call_user_func( $callback , $code ); + $results[$check] = call_user_func( $callback, $code ); } return $results; } + /** + * Format a message key. + * @param $key The message key. + * @param $code The language code. + * @return The formatted message key. + */ protected function formatKey( $key, $code ) { if ( $this->doLinks ) { $displayKey = ucfirst( $key ); @@ -222,28 +325,44 @@ ENDS; } } + /** + * Output the checks results as plain text. + * @return The checks results as plain text. + */ protected function outputText() { foreach ( $this->results as $code => $results ) { $translated = $this->L->getMessages( $code ); $translated = count( $translated['translated'] ); - $translatable = $this->L->getGeneralMessages(); - $translatable = count( $translatable['translatable'] ); foreach ( $results as $check => $messages ) { $count = count( $messages ); if ( $count ) { + if ( $check == 'untranslated' ) { + $translatable = $this->L->getGeneralMessages(); + $total = count( $translatable['translatable'] ); + } elseif ( in_array( $check, $this->nonMessageChecks() ) ) { + $totalCount = $this->getTotalCount(); + $totalCount = $totalCount[$check]; + $callback = array( $this->L, $totalCount[0] ); + $callCode = $totalCount[1] ? $totalCount[1] : $code; + $total = count( call_user_func( $callback, $callCode ) ); + } else { + $total = $translated; + } $search = array( '$1', '$2', '$3' ); - $replace = array( $count, $check == 'untranslated' ? $translatable: $translated, $code ); + $replace = array( $count, $total, $code ); $descriptions = $this->getDescriptions(); echo "\n" . str_replace( $search, $replace, $descriptions[$check] ) . "\n"; if ( $this->level == 1 ) { echo "[messages are hidden]\n"; } else { foreach ( $messages as $key => $value ) { - $displayKey = $this->formatKey( $key, $code ); - if ( $this->level == 2 ) { - echo "* $displayKey\n"; + if( !in_array( $check, $this->nonMessageChecks() ) ) { + $key = $this->formatKey( $key, $code ); + } + if ( $this->level == 2 || empty( $value ) ) { + echo "* $key\n"; } else { - echo "* $displayKey: '$value'\n"; + echo "* $key: '$value'\n"; } } } @@ -253,7 +372,8 @@ ENDS; } /** - * Globals: $wgContLang, $IP + * Output the checks results as wiki text. + * @return The checks results as wiki text. */ function outputWiki() { global $wgContLang, $IP; @@ -265,6 +385,9 @@ ENDS; $problems = 0; $detailTextForLangChecks = array(); foreach ( $results as $check => $messages ) { + if( in_array( $check, $this->nonMessageChecks() ) ) { + continue; + } $count = count( $messages ); if ( $count ) { $problems += $count; @@ -273,7 +396,7 @@ ENDS; $displayKey = $this->formatKey( $key, $code ); $messageDetails[] = $displayKey; } - $detailTextForLangChecks[] = "===$code-$check===\n* " . implode( ', ', $messageDetails ); + $detailTextForLangChecks[] = "=== $code-$check ===\n* " . implode( ', ', $messageDetails ); $numbers[] = "'''[[#$code-$check|$count]]'''"; } else { $numbers[] = $count; @@ -285,7 +408,10 @@ ENDS; $detailText .= $detailTextForLang . implode( "\n", $detailTextForLangChecks ) . "\n"; } - if ( !$problems ) { continue; } // Don't list languages without problems + if ( !$problems ) { + # Don't list languages without problems + continue; + } $language = $wgContLang->getLanguageName( $code ); $rows[] = "| $language || $code || $problems || " . implode( ' || ', $numbers ); } @@ -297,7 +423,7 @@ ENDS; '''Check results are for:''' <code>$version</code> -{| class="sortable wikitable" border="2" cellpadding="4" cellspacing="0" style="background-color: #F9F9F9; border: 1px #AAAAAA solid; border-collapse: collapse; clear:both;" +{| class="sortable wikitable" border="2" cellpadding="4" cellspacing="0" style="background-color: #F9F9F9; border: 1px #AAAAAA solid; border-collapse: collapse; clear: both;" $tableRows |} @@ -306,46 +432,50 @@ $detailText EOL; } + /** + * Check if there are any results for the checks, in any language. + * @return True if there are any results, false if not. + */ protected function isEmpty() { - $empty = true; foreach( $this->results as $code => $results ) { foreach( $results as $check => $messages ) { if( !empty( $messages ) ) { - $empty = false; - break; + return false; } } - if( !$empty ) { - break; - } } - return $empty; + return true; } } class CheckExtensionsCLI extends CheckLanguageCLI { private $extensions; + /** + * Constructor. + * @param $options Options for script. + * @param $extension The extension name (or names). + */ public function __construct( Array $options, $extension ) { if ( isset( $options['help'] ) ) { echo $this->help(); exit(); } - if ( isset($options['lang']) ) { + if ( isset( $options['lang'] ) ) { $this->code = $options['lang']; } else { global $wgLanguageCode; $this->code = $wgLanguageCode; } - if ( isset($options['level']) ) { + if ( isset( $options['level'] ) ) { $this->level = $options['level']; } - $this->doLinks = isset($options['links']); + $this->doLinks = isset( $options['links'] ); - if ( isset($options['wikilang']) ) { + if ( isset( $options['wikilang'] ) ) { $this->wikiCode = $options['wikilang']; } @@ -353,14 +483,16 @@ class CheckExtensionsCLI extends CheckLanguageCLI { $this->checks = explode( ',', $options['whitelist'] ); } elseif ( isset( $options['blacklist'] ) ) { $this->checks = array_diff( - $this->defaultChecks, + isset( $options['easy'] ) ? $this->easyChecks() : $this->defaultChecks(), explode( ',', $options['blacklist'] ) ); + } elseif ( isset( $options['easy'] ) ) { + $this->checks = $this->easyChecks(); } else { - $this->checks = $this->defaultChecks; + $this->checks = $this->defaultChecks(); } - if ( isset($options['output']) ) { + if ( isset( $options['output'] ) ) { $this->output = $options['output']; } @@ -372,23 +504,29 @@ class CheckExtensionsCLI extends CheckLanguageCLI { $this->extensions = array(); $extensions = new PremadeMediawikiExtensionGroups(); $extensions->addAll(); - if( $extension == 'all' ) { - foreach( MessageGroups::singleton()->getGroups() as $group ) { - if( strpos( $group->getId(), 'ext-' ) === 0 && !$group->isMeta() ) { + if ( $extension == 'all' ) { + foreach ( MessageGroups::singleton()->getGroups() as $group ) { + if ( strpos( $group->getId(), 'ext-' ) === 0 && !$group->isMeta() ) { $this->extensions[] = new extensionLanguages( $group ); } } - } elseif( $extension == 'wikimedia' ) { + } elseif ( $extension == 'wikimedia' ) { $wikimedia = MessageGroups::getGroup( 'ext-0-wikimedia' ); - foreach( $wikimedia->wmfextensions() as $extension ) { + foreach ( $wikimedia->wmfextensions() as $extension ) { $group = MessageGroups::getGroup( $extension ); $this->extensions[] = new extensionLanguages( $group ); } + } elseif ( $extension == 'flaggedrevs' ) { + foreach ( MessageGroups::singleton()->getGroups() as $group ) { + if ( strpos( $group->getId(), 'ext-flaggedrevs-' ) === 0 && !$group->isMeta() ) { + $this->extensions[] = new extensionLanguages( $group ); + } + } } else { $extensions = explode( ',', $extension ); - foreach( $extensions as $extension ) { + foreach ( $extensions as $extension ) { $group = MessageGroups::getGroup( 'ext-' . $extension ); - if( $group ) { + if ( $group ) { $extension = new extensionLanguages( $group ); $this->extensions[] = $extension; } else { @@ -398,25 +536,58 @@ class CheckExtensionsCLI extends CheckLanguageCLI { } } + /** + * Get the default checks. + * @return A list of the default checks. + */ + protected function defaultChecks() { + return array( + 'untranslated', 'duplicate', 'obsolete', 'variables', 'empty', 'plural', + 'whitespace', 'xhtml', 'chars', 'links', 'unbalanced', + ); + } + + /** + * Get the checks which check other things than messages. + * @return A list of the non-message checks. + */ + protected function nonMessageChecks() { + return array(); + } + + /** + * Get the checks that can easily be treated by non-speakers of the language. + * @return A list of the easy checks. + */ + protected function easyChecks() { + return array( + 'duplicate', 'obsolete', 'empty', 'whitespace', 'xhtml', 'chars', + ); + } + + /** + * Get help. + * @return The help string. + */ protected function help() { return <<<ENDS Run this script to check the status of a specific language in extensions, or all of them. Command line settings are in form --parameter[=value], except for the first one. Parameters: - * First parameter (mandatory): Extension name, multiple extension names (separated by commas), "all" for all the extensions or "wikimedia" for extensions used by Wikimedia. + * 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. * lang: Language code (default: the installation default language). * help: Show this help. - * level: Show the following level (default: 2). + * level: Show the following display level (default: 2). * links: Link the message values (default off). * wikilang: For the links, what is the content language of the wiki to display the output in (default en). * whitelist: Do only the following checks (form: code,code). * blacklist: Do not perform the following checks (form: code,code). - * duplicate: Additionally check for messages which are translated the same to English (default off). -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): + * easy: Do only the easy checks, which can be treated by non-speakers of the language. +Check codes (ideally, all of them should result 0; all the checks are executed by default (except language-specific check blacklists in checkLanguage.inc): * untranslated: Messages which are required to translate, but are not translated. * duplicate: Messages which translation equal to fallback * obsolete: Messages which are untranslatable, but translated. - * variables: Messages without variables which should be used. + * variables: Messages without variables which should be used, or with variables which shouldn't be used. * empty: Empty messages. * whitespace: Messages which have trailing whitespace. * xhtml: Messages which are not well-formed XHTML (checks only few common errors). @@ -432,10 +603,17 @@ Display levels (default: 2): ENDS; } + /** + * Execute the script. + */ public function execute() { $this->doChecks(); } + /** + * Check a language and show the results. + * @param $code The language code. + */ protected function checkLanguage( $code ) { foreach( $this->extensions as $extension ) { $this->L = $extension; |