Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 163 |
|
0.00% |
0 / 5 |
CRAP | |
0.00% |
0 / 1 |
| LanguageVariantHandler | |
0.00% |
0 / 163 |
|
0.00% |
0 / 5 |
3192 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| convertOne | |
0.00% |
0 / 21 |
|
0.00% |
0 / 1 |
2 | |||
| compressSpArray | |
0.00% |
0 / 14 |
|
0.00% |
0 / 1 |
42 | |||
| onLanguageVariant | |
0.00% |
0 / 126 |
|
0.00% |
0 / 1 |
2162 | |||
| onTag | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
6 | |||
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | |
| 4 | namespace Wikimedia\Parsoid\Wt2Html\TT; |
| 5 | |
| 6 | use Wikimedia\Parsoid\NodeData\DataParsoid; |
| 7 | use Wikimedia\Parsoid\Tokens\EndTagTk; |
| 8 | use Wikimedia\Parsoid\Tokens\EOFTk; |
| 9 | use Wikimedia\Parsoid\Tokens\KV; |
| 10 | use Wikimedia\Parsoid\Tokens\SourceRange; |
| 11 | use Wikimedia\Parsoid\Tokens\TagTk; |
| 12 | use Wikimedia\Parsoid\Tokens\Token; |
| 13 | use Wikimedia\Parsoid\Utils\ContentUtils; |
| 14 | use Wikimedia\Parsoid\Utils\DOMUtils; |
| 15 | use Wikimedia\Parsoid\Utils\PHPUtils; |
| 16 | use Wikimedia\Parsoid\Utils\PipelineUtils; |
| 17 | use Wikimedia\Parsoid\Wikitext\Consts; |
| 18 | use Wikimedia\Parsoid\Wt2Html\TokenTransformManager; |
| 19 | |
| 20 | /** |
| 21 | * Handler for language conversion markup, which looks like `-{ ... }-`. |
| 22 | */ |
| 23 | class LanguageVariantHandler extends TokenHandler { |
| 24 | /** @inheritDoc */ |
| 25 | public function __construct( TokenTransformManager $manager, array $options ) { |
| 26 | parent::__construct( $manager, $options ); |
| 27 | } |
| 28 | |
| 29 | /** |
| 30 | * convert one variant text to dom. |
| 31 | * @param TokenTransformManager $manager |
| 32 | * @param array $options |
| 33 | * @param string $t |
| 34 | * @param array $attribs |
| 35 | * @return array |
| 36 | */ |
| 37 | private function convertOne( TokenTransformManager $manager, array $options, string $t, |
| 38 | array $attribs ): array { |
| 39 | // we're going to fetch the actual token list from attribs |
| 40 | // (this ensures that it has gone through the earlier stages |
| 41 | // of the pipeline already to be expanded) |
| 42 | $t = PHPUtils::stripPrefix( $t, 'mw:lv' ); |
| 43 | $srcOffsets = $attribs[$t]->srcOffsets; |
| 44 | $domFragment = PipelineUtils::processContentInPipeline( |
| 45 | $this->env, $manager->getFrame(), array_merge( $attribs[$t]->v, [ new EOFTk() ] ), |
| 46 | [ |
| 47 | 'pipelineType' => 'tokens/x-mediawiki/expanded', |
| 48 | 'pipelineOpts' => [ |
| 49 | 'inlineContext' => true, |
| 50 | 'expandTemplates' => $options['expandTemplates'], |
| 51 | 'inTemplate' => $options['inTemplate'] |
| 52 | ], |
| 53 | 'srcOffsets' => $srcOffsets->value ?? null, |
| 54 | 'sol' => true |
| 55 | ] |
| 56 | ); |
| 57 | return [ |
| 58 | 'xmlstr' => ContentUtils::ppToXML( |
| 59 | $domFragment, [ 'innerXML' => true ] |
| 60 | ), |
| 61 | 'isBlock' => DOMUtils::hasBlockElementDescendant( $domFragment ), |
| 62 | ]; |
| 63 | } |
| 64 | |
| 65 | /** |
| 66 | * compress a whitespace sequence |
| 67 | * @param ?array $a |
| 68 | * @return ?array |
| 69 | */ |
| 70 | private function compressSpArray( ?array $a ): ?array { |
| 71 | $result = []; |
| 72 | $ctr = 0; |
| 73 | if ( $a === null ) { |
| 74 | return $a; |
| 75 | } |
| 76 | foreach ( $a as $sp ) { |
| 77 | if ( $sp === '' ) { |
| 78 | $ctr++; |
| 79 | } else { |
| 80 | if ( $ctr > 0 ) { |
| 81 | $result[] = $ctr; |
| 82 | $ctr = 0; |
| 83 | } |
| 84 | $result[] = $sp; |
| 85 | } |
| 86 | } |
| 87 | if ( $ctr > 0 ) { |
| 88 | $result[] = $ctr; |
| 89 | } |
| 90 | return $result; |
| 91 | } |
| 92 | |
| 93 | /** |
| 94 | * Main handler. |
| 95 | * See {@link TokenTransformManager#addTransform}'s transformation parameter |
| 96 | * @param Token $token |
| 97 | * @return TokenHandlerResult|null |
| 98 | */ |
| 99 | private function onLanguageVariant( Token $token ): ?TokenHandlerResult { |
| 100 | $manager = $this->manager; |
| 101 | $options = $this->options; |
| 102 | $attribs = $token->attribs; |
| 103 | $dataParsoid = $token->dataParsoid; |
| 104 | $tsr = $dataParsoid->tsr; |
| 105 | $flags = $dataParsoid->flags; |
| 106 | $flagSp = $dataParsoid->flagSp; |
| 107 | $isMeta = false; |
| 108 | $sawFlagA = false; |
| 109 | |
| 110 | // remove trailing semicolon marker, if present |
| 111 | $trailingSemi = false; |
| 112 | if ( count( $dataParsoid->texts ) && |
| 113 | ( $dataParsoid->texts[count( $dataParsoid->texts ) - 1]['semi'] ?? null ) |
| 114 | ) { |
| 115 | $trailingSemi = array_pop( $dataParsoid->texts )['sp'] ?? null; |
| 116 | } |
| 117 | // convert all variant texts to DOM |
| 118 | $isBlock = false; |
| 119 | $texts = array_map( function ( array $t ) use ( $manager, $options, $attribs, &$isBlock ) { |
| 120 | $text = null; |
| 121 | $from = null; |
| 122 | $to = null; |
| 123 | if ( isset( $t['twoway'] ) ) { |
| 124 | $text = $this->convertOne( $manager, $options, $t['text'], $attribs ); |
| 125 | $isBlock = $isBlock || !empty( $text['isBlock'] ); |
| 126 | return [ 'lang' => $t['lang'], 'text' => $text['xmlstr'], 'twoway' => true, 'sp' => $t['sp'] ]; |
| 127 | } elseif ( isset( $t['lang'] ) ) { |
| 128 | $from = $this->convertOne( $manager, $options, $t['from'], $attribs ); |
| 129 | $to = $this->convertOne( $manager, $options, $t['to'], $attribs ); |
| 130 | $isBlock = $isBlock || !empty( $from['isBlock'] ) || !empty( $to['isBlock'] ); |
| 131 | return [ 'lang' => $t['lang'], 'from' => $from['xmlstr'], 'to' => $to['xmlstr'], |
| 132 | 'sp' => $t['sp'] ]; |
| 133 | } else { |
| 134 | $text = $this->convertOne( $manager, $options, $t['text'], $attribs ); |
| 135 | $isBlock = $isBlock || !empty( $text['isBlock'] ); |
| 136 | return [ 'text' => $text['xmlstr'], 'sp' => [] ]; |
| 137 | } |
| 138 | }, $dataParsoid->texts ); |
| 139 | // collect two-way/one-way conversion rules |
| 140 | $oneway = []; |
| 141 | $twoway = []; |
| 142 | $sawTwoway = false; |
| 143 | $sawOneway = false; |
| 144 | $textSp = null; |
| 145 | $twowaySp = []; |
| 146 | $onewaySp = []; |
| 147 | foreach ( $texts as $t ) { |
| 148 | if ( isset( $t['twoway'] ) ) { |
| 149 | $twoway[] = [ 'l' => $t['lang'], 't' => $t['text'] ]; |
| 150 | array_push( $twowaySp, $t['sp'][0], $t['sp'][1], $t['sp'][2] ); |
| 151 | $sawTwoway = true; |
| 152 | } elseif ( isset( $t['lang'] ) ) { |
| 153 | $oneway[] = [ 'l' => $t['lang'], 'f' => $t['from'], 't' => $t['to'] ]; |
| 154 | array_push( $onewaySp, $t['sp'][0], $t['sp'][1], $t['sp'][2], $t['sp'][3] ); |
| 155 | $sawOneway = true; |
| 156 | } |
| 157 | } |
| 158 | |
| 159 | // To avoid too much data-mw bloat, only the top level keys in |
| 160 | // data-mw-variant are "human readable". Nested keys are single-letter: |
| 161 | // `l` for `language`, `t` for `text` or `to`, `f` for `from`. |
| 162 | $dataMWV = null; |
| 163 | if ( count( $flags ) === 0 && count( $dataParsoid->variants ) > 0 ) { |
| 164 | // "Restrict possible variants to a limited set" |
| 165 | $dataMWV = [ |
| 166 | 'filter' => [ 'l' => $dataParsoid->variants, 't' => $texts[0]['text'] ], |
| 167 | 'show' => true |
| 168 | ]; |
| 169 | } else { |
| 170 | $dataMWV = []; |
| 171 | foreach ( $flags as $f ) { |
| 172 | if ( array_key_exists( $f, Consts::$LCFlagMap ) ) { |
| 173 | if ( Consts::$LCFlagMap[$f] ) { |
| 174 | $dataMWV[Consts::$LCFlagMap[$f]] = true; |
| 175 | if ( $f === 'A' ) { |
| 176 | $sawFlagA = true; |
| 177 | } |
| 178 | } |
| 179 | } else { |
| 180 | $dataMWV['error'] = true; |
| 181 | } |
| 182 | } |
| 183 | // (this test is done at the top of ConverterRule::getRuleConvertedStr) |
| 184 | // (also partially in ConverterRule::parse) |
| 185 | if ( count( $texts ) === 1 && |
| 186 | !isset( $texts[0]['lang'] ) && !isset( $dataMWV['name'] ) |
| 187 | ) { |
| 188 | if ( isset( $dataMWV['add'] ) || isset( $dataMWV['remove'] ) ) { |
| 189 | $variants = [ '*' ]; |
| 190 | $twoway = array_map( static function ( string $code ) use ( $texts, &$sawTwoway ) { |
| 191 | return [ 'l' => $code, 't' => $texts[0]['text'] ]; |
| 192 | }, $variants ); |
| 193 | $sawTwoway = true; |
| 194 | } else { |
| 195 | $dataMWV['disabled'] = true; |
| 196 | unset( $dataMWV['describe'] ); |
| 197 | } |
| 198 | } |
| 199 | if ( isset( $dataMWV['describe'] ) ) { |
| 200 | if ( !$sawFlagA ) { |
| 201 | $dataMWV['show'] = true; |
| 202 | } |
| 203 | } |
| 204 | if ( isset( $dataMWV['disabled'] ) || isset( $dataMWV['name'] ) ) { |
| 205 | if ( isset( $dataMWV['disabled'] ) ) { |
| 206 | $dataMWV['disabled'] = [ 't' => $texts[0]['text'] ?? '' ]; |
| 207 | } else { |
| 208 | $dataMWV['name'] = [ 't' => $texts[0]['text'] ?? '' ]; |
| 209 | } |
| 210 | if ( isset( $dataMWV['title'] ) || isset( $dataMWV['add'] ) ) { |
| 211 | unset( $dataMWV['show'] ); |
| 212 | } else { |
| 213 | $dataMWV['show'] = true; |
| 214 | } |
| 215 | } elseif ( $sawTwoway ) { |
| 216 | $dataMWV['twoway'] = $twoway; |
| 217 | $textSp = $twowaySp; |
| 218 | if ( $sawOneway ) { |
| 219 | $dataMWV['error'] = true; |
| 220 | } |
| 221 | } else { |
| 222 | $dataMWV['oneway'] = $oneway; |
| 223 | $textSp = $onewaySp; |
| 224 | if ( !$sawOneway ) { |
| 225 | $dataMWV['error'] = true; |
| 226 | } |
| 227 | } |
| 228 | } |
| 229 | // Use meta/not meta instead of explicit 'show' flag. |
| 230 | $isMeta = !isset( $dataMWV['show'] ); |
| 231 | unset( $dataMWV['show'] ); |
| 232 | // Trim some data from data-parsoid if it matches the defaults |
| 233 | if ( count( $flagSp ) === 2 * count( $dataParsoid->original ) ) { |
| 234 | $result = true; |
| 235 | foreach ( $flagSp as $s ) { |
| 236 | if ( $s !== '' ) { |
| 237 | $result = false; |
| 238 | break; |
| 239 | } |
| 240 | } |
| 241 | if ( $result ) { |
| 242 | $flagSp = null; |
| 243 | } |
| 244 | } |
| 245 | if ( $trailingSemi !== false && $textSp ) { |
| 246 | $textSp[] = $trailingSemi; |
| 247 | } |
| 248 | |
| 249 | // Our markup is always the same, except for the contents of |
| 250 | // the data-mw-variant attribute and whether it's a span, div, or a |
| 251 | // meta, depending on (respectively) whether conversion output |
| 252 | // contains only inline content, could contain block content, |
| 253 | // or never contains any content. |
| 254 | |
| 255 | $das = new DataParsoid; |
| 256 | $das->fl = $dataParsoid->original; // original "fl"ags |
| 257 | $flSp = $this->compressSpArray( $flagSp ); // spaces around flags |
| 258 | if ( $flSp !== null ) { |
| 259 | $das->flSp = $flSp; |
| 260 | } |
| 261 | $das->src = $dataParsoid->src; |
| 262 | $tSp = $this->compressSpArray( $textSp ); // spaces around texts |
| 263 | if ( $tSp !== null ) { |
| 264 | $das->tSp = $tSp; |
| 265 | } |
| 266 | $das->tsr = new SourceRange( $tsr->start, $isMeta ? $tsr->end : ( $tsr->end - 2 ) ); |
| 267 | |
| 268 | PHPUtils::sortArray( $dataMWV ); |
| 269 | $tokens = [ |
| 270 | new TagTk( $isMeta ? 'meta' : ( $isBlock ? 'div' : 'span' ), [ |
| 271 | new KV( 'typeof', 'mw:LanguageVariant' ), |
| 272 | new KV( 'data-mw-variant', PHPUtils::jsonEncode( $dataMWV ) ) |
| 273 | ], $das |
| 274 | ) |
| 275 | ]; |
| 276 | if ( !$isMeta ) { |
| 277 | $metaDP = new DataParsoid; |
| 278 | $metaDP->tsr = new SourceRange( $tsr->end - 2, $tsr->end ); |
| 279 | $tokens[] = new EndTagTk( $isBlock ? 'div' : 'span', [], $metaDP ); |
| 280 | } |
| 281 | |
| 282 | return new TokenHandlerResult( $tokens ); |
| 283 | } |
| 284 | |
| 285 | /** |
| 286 | * @inheritDoc |
| 287 | */ |
| 288 | public function onTag( Token $token ): ?TokenHandlerResult { |
| 289 | return $token->getName() === 'language-variant' ? $this->onLanguageVariant( $token ) : null; |
| 290 | } |
| 291 | } |