Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
85.45% |
94 / 110 |
|
33.33% |
1 / 3 |
CRAP | |
0.00% |
0 / 1 |
| LanguageData | |
85.45% |
94 / 110 |
|
33.33% |
1 / 3 |
43.68 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getLocalData | |
96.83% |
61 / 63 |
|
0.00% |
0 / 1 |
9 | |||
| convertDateFormat | |
69.57% |
32 / 46 |
|
0.00% |
0 / 1 |
52.71 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * Generates language-specific data used by DiscussionTools. |
| 4 | * |
| 5 | * @file |
| 6 | * @ingroup Extensions |
| 7 | * @license MIT |
| 8 | */ |
| 9 | |
| 10 | namespace MediaWiki\Extension\DiscussionTools; |
| 11 | |
| 12 | use DateTimeZone; |
| 13 | use MediaWiki\Config\Config; |
| 14 | use MediaWiki\Language\ILanguageConverter; |
| 15 | use MediaWiki\Language\Language; |
| 16 | use MediaWiki\Language\LanguageConverterFactory; |
| 17 | use MediaWiki\MainConfigNames; |
| 18 | use MediaWiki\SpecialPage\SpecialPageFactory; |
| 19 | |
| 20 | class LanguageData { |
| 21 | |
| 22 | public function __construct( |
| 23 | private readonly Config $config, |
| 24 | private readonly Language $language, |
| 25 | private readonly LanguageConverterFactory $languageConverterFactory, |
| 26 | private readonly SpecialPageFactory $specialPageFactory, |
| 27 | ) { |
| 28 | } |
| 29 | |
| 30 | /** |
| 31 | * Compute data we need to parse discussion threads on pages. |
| 32 | */ |
| 33 | public function getLocalData(): array { |
| 34 | $config = $this->config; |
| 35 | $lang = $this->language; |
| 36 | $langConv = $this->languageConverterFactory->getLanguageConverter( $lang ); |
| 37 | |
| 38 | $data = []; |
| 39 | |
| 40 | $data['dateFormat'] = []; |
| 41 | $dateFormat = $lang->getDateFormatString( 'both', $lang->dateFormat( false ) ); |
| 42 | foreach ( $langConv->getVariants() as $variant ) { |
| 43 | $convDateFormat = $this->convertDateFormat( $dateFormat, $langConv, $variant ); |
| 44 | $data['dateFormat'][$variant] = $convDateFormat; |
| 45 | } |
| 46 | |
| 47 | $data['digits'] = []; |
| 48 | foreach ( $langConv->getVariants() as $variant ) { |
| 49 | $data['digits'][$variant] = []; |
| 50 | foreach ( str_split( '0123456789' ) as $digit ) { |
| 51 | if ( $config->get( MainConfigNames::TranslateNumerals ) ) { |
| 52 | $localDigit = $lang->formatNumNoSeparators( $digit ); |
| 53 | } else { |
| 54 | $localDigit = $digit; |
| 55 | } |
| 56 | $convLocalDigit = $langConv->translate( $localDigit, $variant ); |
| 57 | $data['digits'][$variant][] = $convLocalDigit; |
| 58 | } |
| 59 | } |
| 60 | |
| 61 | // ApiQuerySiteinfo |
| 62 | $data['localTimezone'] = $config->get( MainConfigNames::Localtimezone ); |
| 63 | |
| 64 | // special page names compared against Title::getText, which contains space |
| 65 | // But aliases are stored with underscores (db key) in the alias files |
| 66 | $data['specialContributionsName'] = str_replace( '_', ' ', $this->specialPageFactory |
| 67 | ->getLocalNameFor( 'Contributions' ) ); |
| 68 | $data['specialNewSectionName'] = str_replace( '_', ' ', $this->specialPageFactory |
| 69 | ->getLocalNameFor( 'NewSection' ) ); |
| 70 | |
| 71 | $localTimezone = $config->get( MainConfigNames::Localtimezone ); |
| 72 | // Return all timezone abbreviations for the local timezone (there will often be two, for |
| 73 | // non-DST and DST timestamps, and sometimes more due to historical data, but that's okay). |
| 74 | // Avoid DateTimeZone::listAbbreviations(), it returns some half-baked list that is different |
| 75 | // from the timezone data used by everything else in PHP. |
| 76 | $timezoneTransitions = ( new DateTimeZone( $localTimezone ) )->getTransitions(); |
| 77 | if ( !is_array( $timezoneTransitions ) ) { |
| 78 | // Handle (arguably invalid) config where $wgLocaltimezone is an abbreviation like "CST" |
| 79 | // instead of a real IANA timezone name like "America/Chicago". (T312310) |
| 80 | // "DateTimeZone objects wrapping type 1 (UTC offsets) and type 2 (abbreviations) do not |
| 81 | // contain any transitions, and calling this method on them will return false." |
| 82 | // https://www.php.net/manual/en/datetimezone.gettransitions.php |
| 83 | $timezoneAbbrs = [ $localTimezone ]; |
| 84 | } else { |
| 85 | $timezoneAbbrs = array_values( array_unique( |
| 86 | array_map( static function ( $transition ) { |
| 87 | return $transition['abbr']; |
| 88 | }, $timezoneTransitions ) |
| 89 | ) ); |
| 90 | } |
| 91 | |
| 92 | $data['timezones'] = []; |
| 93 | foreach ( $langConv->getVariants() as $variant ) { |
| 94 | $data['timezones'][$variant] = array_combine( |
| 95 | array_map( static function ( string $tzMsg ) use ( $lang, $langConv, $variant ) { |
| 96 | // MWTimestamp::getTimezoneMessage() |
| 97 | // Parser::pstPass2() |
| 98 | // Messages used here: 'timezone-utc' and so on |
| 99 | $key = 'timezone-' . strtolower( trim( $tzMsg ) ); |
| 100 | $msg = wfMessage( $key )->inLanguage( $lang ); |
| 101 | // TODO: This probably causes a similar issue to https://phabricator.wikimedia.org/T221294, |
| 102 | // but we *must* check the message existence in the database, because the messages are not |
| 103 | // actually defined by MediaWiki core for any timezone other than UTC... |
| 104 | if ( $msg->exists() ) { |
| 105 | $text = $msg->text(); |
| 106 | } else { |
| 107 | $text = strtoupper( $tzMsg ); |
| 108 | } |
| 109 | $convText = $langConv->translate( $text, $variant ); |
| 110 | return $convText; |
| 111 | }, $timezoneAbbrs ), |
| 112 | array_map( 'strtoupper', $timezoneAbbrs ) |
| 113 | ); |
| 114 | } |
| 115 | |
| 116 | // Messages in content language |
| 117 | $messagesKeys = array_merge( |
| 118 | Language::WEEKDAY_MESSAGES, |
| 119 | Language::WEEKDAY_ABBREVIATED_MESSAGES, |
| 120 | Language::MONTH_MESSAGES, |
| 121 | Language::MONTH_GENITIVE_MESSAGES, |
| 122 | Language::MONTH_ABBREVIATED_MESSAGES |
| 123 | ); |
| 124 | $data['contLangMessages'] = []; |
| 125 | foreach ( $langConv->getVariants() as $variant ) { |
| 126 | $data['contLangMessages'][$variant] = array_combine( |
| 127 | $messagesKeys, |
| 128 | array_map( static function ( $key ) use ( $lang, $langConv, $variant ) { |
| 129 | $text = wfMessage( $key )->inLanguage( $lang )->text(); |
| 130 | return $langConv->translate( $text, $variant ); |
| 131 | }, $messagesKeys ) |
| 132 | ); |
| 133 | } |
| 134 | |
| 135 | return $data; |
| 136 | } |
| 137 | |
| 138 | /** |
| 139 | * Convert a date format string to a different language variant, leaving all special characters |
| 140 | * unchanged and applying language conversion to the plain text fragments. |
| 141 | */ |
| 142 | private function convertDateFormat( |
| 143 | string $format, |
| 144 | ILanguageConverter $langConv, |
| 145 | string $variant |
| 146 | ): string { |
| 147 | $formatLength = strlen( $format ); |
| 148 | $s = ''; |
| 149 | // The supported codes must match CommentParser::getTimestampRegexp() |
| 150 | for ( $p = 0; $p < $formatLength; $p++ ) { |
| 151 | $num = false; |
| 152 | $code = $format[ $p ]; |
| 153 | if ( $code === 'x' && $p < $formatLength - 1 ) { |
| 154 | $code .= $format[++$p]; |
| 155 | } |
| 156 | if ( $code === 'xk' && $p < $formatLength - 1 ) { |
| 157 | $code .= $format[++$p]; |
| 158 | } |
| 159 | |
| 160 | // LAZY SHORTCUTS that might cause bugs: |
| 161 | // * We assume that result of $langConv->translate() doesn't produce any special codes/characters |
| 162 | // * We assume that calling $langConv->translate() separately for each character is correct |
| 163 | switch ( $code ) { |
| 164 | case 'xx': |
| 165 | case 'xg': |
| 166 | case 'xn': |
| 167 | case 'd': |
| 168 | case 'D': |
| 169 | case 'j': |
| 170 | case 'l': |
| 171 | case 'F': |
| 172 | case 'M': |
| 173 | case 'm': |
| 174 | case 'n': |
| 175 | case 'Y': |
| 176 | case 'xkY': |
| 177 | case 'G': |
| 178 | case 'H': |
| 179 | case 'i': |
| 180 | case 's': |
| 181 | // Special code - pass through unchanged |
| 182 | $s .= $code; |
| 183 | break; |
| 184 | case '\\': |
| 185 | // Plain text (backslash escaping) - convert to language variant |
| 186 | if ( $p < $formatLength - 1 ) { |
| 187 | $s .= '\\' . $langConv->translate( $format[++$p], $variant ); |
| 188 | } else { |
| 189 | $s .= $code; |
| 190 | } |
| 191 | break; |
| 192 | case '"': |
| 193 | // Plain text (quoted literal) - convert to language variant |
| 194 | if ( $p < $formatLength - 1 ) { |
| 195 | $endQuote = strpos( $format, '"', $p + 1 ); |
| 196 | if ( $endQuote === false ) { |
| 197 | // No terminating quote, assume literal " |
| 198 | $s .= $code; |
| 199 | } else { |
| 200 | $s .= '"' . |
| 201 | $langConv->translate( substr( $format, $p + 1, $endQuote - $p - 1 ), $variant ) . |
| 202 | '"'; |
| 203 | $p = $endQuote; |
| 204 | } |
| 205 | } else { |
| 206 | // Quote at end of string, assume literal " |
| 207 | $s .= $code; |
| 208 | } |
| 209 | break; |
| 210 | default: |
| 211 | // Plain text - convert to language variant |
| 212 | $s .= $langConv->translate( $format[$p], $variant ); |
| 213 | } |
| 214 | } |
| 215 | |
| 216 | return $s; |
| 217 | } |
| 218 | } |