Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
90.83% |
109 / 120 |
|
70.59% |
12 / 17 |
CRAP | |
0.00% |
0 / 1 |
| TranslationUnit | |
90.83% |
109 / 120 |
|
70.59% |
12 / 17 |
46.56 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
1 | |||
| setIsInline | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| isInline | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| setCanWrap | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| canWrap | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getText | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getTextWithVariables | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
2 | |||
| getTextForTrans | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
2 | |||
| getMarkedText | |
100.00% |
8 / 8 |
|
100.00% |
1 / 1 |
3 | |||
| getOldText | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getVariables | |
100.00% |
14 / 14 |
|
100.00% |
1 / 1 |
3 | |||
| serializeToArray | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| unserializeFromArray | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| getTextForRendering | |
100.00% |
40 / 40 |
|
100.00% |
1 / 1 |
12 | |||
| getIssues | |
100.00% |
20 / 20 |
|
100.00% |
1 / 1 |
5 | |||
| getHeading | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| shouldAddAnchor | |
88.89% |
8 / 9 |
|
0.00% |
0 / 1 |
6.05 | |||
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | |
| 4 | namespace MediaWiki\Extension\Translate\PageTranslation; |
| 5 | |
| 6 | use Language; |
| 7 | use MediaWiki\Extension\Translate\MessageLoading\Message; |
| 8 | use MediaWiki\Html\Html; |
| 9 | use Parser; |
| 10 | use const PREG_SET_ORDER; |
| 11 | |
| 12 | /** |
| 13 | * This class represents one translation unit in a translatable page. |
| 14 | * |
| 15 | * @author Niklas Laxström |
| 16 | * @license GPL-2.0-or-later |
| 17 | * @ingroup PageTranslation |
| 18 | */ |
| 19 | class TranslationUnit { |
| 20 | public const UNIT_MARKER_INVALID_CHARS = "_/\n<>"; |
| 21 | public const NEW_UNIT_ID = '-1'; |
| 22 | // Deprecated syntax. Example: <tvar|1>...</> |
| 23 | public const TVAR_OLD_SYNTAX_REGEX = '~<tvar\|([^>]+)>(.*?)</>~us'; |
| 24 | // Current syntax. Example: <tvar name=1>...</tvar> |
| 25 | public const TVAR_NEW_SYNTAX_REGEX = |
| 26 | <<<'REGEXP' |
| 27 | ~ |
| 28 | <tvar \s+ name \s* = \s* |
| 29 | ( ( ' (?<key1> [^']* ) ' ) | ( " (?<key2> [^"]* ) " ) | (?<key3> [^"'\s>]* ) ) |
| 30 | \s* > (?<value>.*?) </tvar \s* > |
| 31 | ~xusi |
| 32 | REGEXP; |
| 33 | /** |
| 34 | * Regular expression matching the `{{TRANSLATIONLANGUAGE}}` “magic word” |
| 35 | * (which is not a real magic word, but rather replaced in the source text) |
| 36 | */ |
| 37 | public const TRANSLATIONLANGUAGE_REGEX = '/{{\s*TRANSLATIONLANGUAGE\s*}}/'; |
| 38 | |
| 39 | /** @var string Unit name */ |
| 40 | public $id; |
| 41 | /** @var string Unit text. */ |
| 42 | public $text; |
| 43 | /** @var string Is this new, existing, changed or deleted unit. */ |
| 44 | public $type; |
| 45 | /** @var string|null Text of previous version of this unit. */ |
| 46 | public $oldText = null; |
| 47 | /** |
| 48 | * @var bool Whether this unit is inline unit. |
| 49 | * E.g. "Something <translate>foo</translate> bar". |
| 50 | */ |
| 51 | protected $inline = false; |
| 52 | /** @var bool Whether wrapping the unit is allowed */ |
| 53 | private $canWrap = true; |
| 54 | /** @var int Version number for the serialization. */ |
| 55 | private $version = 1; |
| 56 | /** @var string[] List of properties to serialize. */ |
| 57 | private static $properties = [ 'version', 'id', 'text', 'type', 'oldText', 'inline' ]; |
| 58 | |
| 59 | public function __construct( |
| 60 | string $text, |
| 61 | string $id = self::NEW_UNIT_ID, |
| 62 | string $type = 'new', |
| 63 | string $oldText = null |
| 64 | ) { |
| 65 | $this->text = $text; |
| 66 | $this->id = $id; |
| 67 | $this->type = $type; |
| 68 | $this->oldText = $oldText; |
| 69 | } |
| 70 | |
| 71 | public function setIsInline( bool $value ): void { |
| 72 | $this->inline = $value; |
| 73 | } |
| 74 | |
| 75 | public function isInline(): bool { |
| 76 | return $this->inline; |
| 77 | } |
| 78 | |
| 79 | public function setCanWrap( bool $value ): void { |
| 80 | $this->canWrap = $value; |
| 81 | } |
| 82 | |
| 83 | public function canWrap(): bool { |
| 84 | return $this->canWrap; |
| 85 | } |
| 86 | |
| 87 | /** Returns unit text unmodified */ |
| 88 | public function getText(): string { |
| 89 | return $this->text; |
| 90 | } |
| 91 | |
| 92 | /** Returns the text with tvars replaces with placeholders */ |
| 93 | public function getTextWithVariables(): string { |
| 94 | $variableReplacements = []; |
| 95 | foreach ( $this->getVariables() as $variable ) { |
| 96 | $variableReplacements[$variable->getDefinition()] = $variable->getName(); |
| 97 | } |
| 98 | |
| 99 | return strtr( $this->text, $variableReplacements ); |
| 100 | } |
| 101 | |
| 102 | /** Returns unit text with variables replaced. */ |
| 103 | public function getTextForTrans(): string { |
| 104 | $variableReplacements = []; |
| 105 | foreach ( $this->getVariables() as $variable ) { |
| 106 | $variableReplacements[$variable->getDefinition()] = $variable->getValue(); |
| 107 | } |
| 108 | |
| 109 | return strtr( $this->text, $variableReplacements ); |
| 110 | } |
| 111 | |
| 112 | /** Returns the unit text with updated or added unit marker */ |
| 113 | public function getMarkedText(): string { |
| 114 | $id = $this->id; |
| 115 | $header = "<!--T:$id-->"; |
| 116 | |
| 117 | if ( $this->getHeading( $this->text ) !== null ) { |
| 118 | $text = $this->text . ' ' . $header; |
| 119 | } else { |
| 120 | if ( $this->inline ) { |
| 121 | $text = $header . ' ' . $this->text; |
| 122 | } else { |
| 123 | $text = $header . "\n" . $this->text; |
| 124 | } |
| 125 | } |
| 126 | |
| 127 | return $text; |
| 128 | } |
| 129 | |
| 130 | /** Returns oldtext, or current text if not available */ |
| 131 | public function getOldText(): string { |
| 132 | return $this->oldText ?? $this->text; |
| 133 | } |
| 134 | |
| 135 | /** @return TranslationVariable[] */ |
| 136 | public function getVariables(): array { |
| 137 | $vars = []; |
| 138 | |
| 139 | $matches = []; |
| 140 | preg_match_all( self::TVAR_OLD_SYNTAX_REGEX, $this->text, $matches, PREG_SET_ORDER ); |
| 141 | foreach ( $matches as $m ) { |
| 142 | $vars[] = new TranslationVariable( $m[0], '$' . $m[1], $m[2] ); |
| 143 | } |
| 144 | |
| 145 | $matches = []; |
| 146 | preg_match_all( self::TVAR_NEW_SYNTAX_REGEX, $this->text, $matches, PREG_SET_ORDER ); |
| 147 | foreach ( $matches as $m ) { |
| 148 | $vars[] = new TranslationVariable( |
| 149 | $m[0], |
| 150 | // Maximum of one of these is non-empty string |
| 151 | '$' . ( $m['key1'] . $m['key2'] . $m['key3'] ), |
| 152 | $m['value'] |
| 153 | ); |
| 154 | } |
| 155 | |
| 156 | return $vars; |
| 157 | } |
| 158 | |
| 159 | /** Serialize this object to a PHP array */ |
| 160 | public function serializeToArray(): array { |
| 161 | $data = []; |
| 162 | foreach ( self::$properties as $index => $property ) { |
| 163 | // Because this is used for the JobQueue, use a list |
| 164 | // instead of an array to save space. |
| 165 | $data[$index] = $this->$property; |
| 166 | } |
| 167 | |
| 168 | return $data; |
| 169 | } |
| 170 | |
| 171 | public static function unserializeFromArray( array $data ): self { |
| 172 | // Give dummy default text, will be overridden |
| 173 | $unit = new self( '' ); |
| 174 | foreach ( self::$properties as $index => $property ) { |
| 175 | $unit->$property = $data[$index]; |
| 176 | } |
| 177 | |
| 178 | return $unit; |
| 179 | } |
| 180 | |
| 181 | public function getTextForRendering( |
| 182 | ?Message $msg, |
| 183 | Language $sourceLanguage, |
| 184 | Language $targetLanguage, |
| 185 | bool $wrapUntranslated, |
| 186 | ?Parser $parser = null |
| 187 | ): string { |
| 188 | $attributes = []; |
| 189 | $headingText = null; |
| 190 | |
| 191 | if ( $msg && $msg->translation() !== null ) { |
| 192 | $content = $msg->translation(); |
| 193 | $headingText = $this->getHeading( $msg->definition() ); |
| 194 | |
| 195 | if ( $msg->hasTag( 'fuzzy' ) ) { |
| 196 | // We do not ever want to show explicit fuzzy marks in the rendered pages |
| 197 | $content = str_replace( TRANSLATE_FUZZY, '', $content ); |
| 198 | $attributes['class'] = 'mw-translate-fuzzy'; |
| 199 | } |
| 200 | $translationLanguage = $targetLanguage->getCode(); |
| 201 | } else { |
| 202 | $content = $this->getTextWithVariables(); |
| 203 | if ( $wrapUntranslated ) { |
| 204 | $attributes['lang'] = $sourceLanguage->getHtmlCode(); |
| 205 | $attributes['dir'] = $sourceLanguage->getDir(); |
| 206 | $attributes['class'] = 'mw-content-' . $sourceLanguage->getDir(); |
| 207 | } |
| 208 | $translationLanguage = $sourceLanguage->getCode(); |
| 209 | } |
| 210 | |
| 211 | if ( $this->canWrap() && $attributes ) { |
| 212 | $tag = $this->isInline() ? 'span' : 'div'; |
| 213 | $content = $this->isInline() ? $content : "\n$content\n"; |
| 214 | $content = Html::rawElement( $tag, $attributes, $content ); |
| 215 | } |
| 216 | |
| 217 | $variableReplacements = []; |
| 218 | foreach ( $this->getVariables() as $variable ) { |
| 219 | $variableReplacements[$variable->getName()] = $variable->getValue(); |
| 220 | } |
| 221 | |
| 222 | if ( |
| 223 | $parser && |
| 224 | $this->shouldAddAnchor( |
| 225 | $sourceLanguage, |
| 226 | $targetLanguage, |
| 227 | $headingText, |
| 228 | $msg, |
| 229 | $this->isInline() |
| 230 | ) |
| 231 | ) { |
| 232 | $sectionName = substr( $parser->guessSectionNameFromWikiText( $headingText ), 1 ); |
| 233 | $attributes = [ 'id' => $sectionName ]; |
| 234 | $content = Html::rawElement( 'span', $attributes, '' ) . "\n$content"; |
| 235 | } |
| 236 | |
| 237 | $content = strtr( $content, $variableReplacements ); |
| 238 | |
| 239 | // Allow wrapping this inside variables |
| 240 | $content = preg_replace( |
| 241 | self::TRANSLATIONLANGUAGE_REGEX, |
| 242 | $translationLanguage, |
| 243 | $content |
| 244 | ); |
| 245 | |
| 246 | return $content; |
| 247 | } |
| 248 | |
| 249 | /** @return TranslationUnitIssue[] */ |
| 250 | public function getIssues(): array { |
| 251 | $issues = $usedNames = []; |
| 252 | foreach ( $this->getVariables() as $variable ) { |
| 253 | $name = $variable->getName(); |
| 254 | $pattern = '/^' . TranslatablePageInsertablesSuggester::NAME_PATTERN . '$/u'; |
| 255 | if ( !preg_match( $pattern, $name ) ) { |
| 256 | // Key by name to avoid multiple issues of the same name |
| 257 | $issues[$name] = new TranslationUnitIssue( |
| 258 | TranslationUnitIssue::WARNING, |
| 259 | 'tpt-validation-not-insertable', |
| 260 | [ wfEscapeWikiText( $name ) ] |
| 261 | ); |
| 262 | } |
| 263 | |
| 264 | $usedNames[ $name ][] = $variable->getValue(); |
| 265 | } |
| 266 | |
| 267 | foreach ( $usedNames as $name => $contents ) { |
| 268 | $uniqueValueCount = count( array_unique( $contents ) ); |
| 269 | if ( $uniqueValueCount > 1 ) { |
| 270 | $issues[] = new TranslationUnitIssue( |
| 271 | TranslationUnitIssue::ERROR, |
| 272 | 'tpt-validation-name-reuse', |
| 273 | [ wfEscapeWikiText( $name ) ] |
| 274 | ); |
| 275 | } |
| 276 | } |
| 277 | |
| 278 | return array_values( $issues ); |
| 279 | } |
| 280 | |
| 281 | /** Mimic the behavior of how Parser handles headings including handling of unbalanced "=" signs */ |
| 282 | private function getHeading( string $text ): ?string { |
| 283 | $match = []; |
| 284 | preg_match( '/^(={1,6})[ \t]*(.+?)[ \t]*\1\s*$/', $text, $match ); |
| 285 | return $match[2] ?? null; |
| 286 | } |
| 287 | |
| 288 | private function shouldAddAnchor( |
| 289 | Language $sourceLanguage, |
| 290 | Language $targetLanguage, |
| 291 | ?string $headingText, |
| 292 | ?Message $msg, |
| 293 | bool $isInline |
| 294 | ): bool { |
| 295 | // If it's not a heading, don't bother adding an anchor |
| 296 | if ( $headingText === null ) { |
| 297 | return false; |
| 298 | } |
| 299 | |
| 300 | // We only add an anchor for a translation. See: https://phabricator.wikimedia.org/T62544 |
| 301 | if ( $sourceLanguage->getCode() === $targetLanguage->getCode() ) { |
| 302 | return false; |
| 303 | } |
| 304 | |
| 305 | // Translation and the source text are same, avoid adding an anchor that would create |
| 306 | // an id attribute with duplicate value |
| 307 | if ( $msg && $msg->translation() === $msg->definition() ) { |
| 308 | return false; |
| 309 | } |
| 310 | |
| 311 | // If nowrap attribute is set, do not add the anchor |
| 312 | if ( !$this->canWrap() ) { |
| 313 | return false; |
| 314 | } |
| 315 | |
| 316 | // We don't add anchors for inline translate tags to avoid breaking input like this: |
| 317 | // Text here <translate>== not a heading ==</translate> |
| 318 | return !$isInline; |
| 319 | } |
| 320 | } |