Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 202 |
|
0.00% |
0 / 13 |
CRAP | |
0.00% |
0 / 1 |
| TokenizerUtils | |
0.00% |
0 / 202 |
|
0.00% |
0 / 13 |
13340 | |
0.00% |
0 / 1 |
| internalFlatten | |
0.00% |
0 / 15 |
|
0.00% |
0 / 1 |
72 | |||
| flattenIfArray | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| flattenString | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
12 | |||
| flattenStringlist | |
0.00% |
0 / 15 |
|
0.00% |
0 / 1 |
42 | |||
| getAttrVal | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| buildTableTokens | |
0.00% |
0 / 36 |
|
0.00% |
0 / 1 |
380 | |||
| buildXMLTag | |
0.00% |
0 / 13 |
|
0.00% |
0 / 1 |
20 | |||
| inlineBreaks | |
0.00% |
0 / 74 |
|
0.00% |
0 / 1 |
2256 | |||
| popComments | |
0.00% |
0 / 17 |
|
0.00% |
0 / 1 |
156 | |||
| getAutoUrlTerminatingChars | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| enforceParserResourceLimits | |
0.00% |
0 / 16 |
|
0.00% |
0 / 1 |
90 | |||
| protectAttrs | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
6 | |||
| resetAnnotationIncludeRegex | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * Utilities used in the tokenizer. |
| 4 | * @module wt2html/tokenizer_utils |
| 5 | */ |
| 6 | |
| 7 | declare( strict_types = 1 ); |
| 8 | |
| 9 | namespace Wikimedia\Parsoid\Wt2Html; |
| 10 | |
| 11 | use Wikimedia\Parsoid\Config\Env; |
| 12 | use Wikimedia\Parsoid\NodeData\DataParsoid; |
| 13 | use Wikimedia\Parsoid\NodeData\TempData; |
| 14 | use Wikimedia\Parsoid\Tokens\CommentTk; |
| 15 | use Wikimedia\Parsoid\Tokens\EndTagTk; |
| 16 | use Wikimedia\Parsoid\Tokens\SelfclosingTagTk; |
| 17 | use Wikimedia\Parsoid\Tokens\SourceRange; |
| 18 | use Wikimedia\Parsoid\Tokens\TagTk; |
| 19 | use Wikimedia\Parsoid\Tokens\Token; |
| 20 | use Wikimedia\Parsoid\Utils\DOMDataUtils; |
| 21 | use Wikimedia\Parsoid\Utils\PHPUtils; |
| 22 | use Wikimedia\Parsoid\Wikitext\Consts; |
| 23 | |
| 24 | class TokenizerUtils { |
| 25 | private static ?string $protectAttrsRegExp = null; |
| 26 | private static ?string $inclAnnRegExp = null; |
| 27 | |
| 28 | /** |
| 29 | * @param mixed $e |
| 30 | * @param ?array &$res |
| 31 | * @return mixed (same type as $e) |
| 32 | * @throws \Exception |
| 33 | */ |
| 34 | private static function internalFlatten( $e, ?array &$res ) { |
| 35 | // Don't bother flattening if we dont have an array |
| 36 | if ( !is_array( $e ) ) { |
| 37 | return $e; |
| 38 | } |
| 39 | |
| 40 | for ( $i = 0; $i < count( $e ); $i++ ) { |
| 41 | $v = $e[$i]; |
| 42 | if ( is_array( $v ) ) { |
| 43 | // Change in assumption from a shallow array to a nested array. |
| 44 | if ( $res === null ) { |
| 45 | $res = array_slice( $e, 0, $i ); |
| 46 | } |
| 47 | self::internalFlatten( $v, $res ); |
| 48 | } elseif ( $v !== null ) { |
| 49 | if ( $res !== null ) { |
| 50 | $res[] = $v; |
| 51 | } |
| 52 | } else { |
| 53 | throw new \RuntimeException( __METHOD__ . ": found falsy element $v @ posn $i" ); |
| 54 | } |
| 55 | } |
| 56 | |
| 57 | if ( $res !== null ) { |
| 58 | $e = $res; |
| 59 | } |
| 60 | return $e; |
| 61 | } |
| 62 | |
| 63 | /** |
| 64 | * If $a is an array, this recursively flattens all nested arrays. |
| 65 | * @param mixed $a |
| 66 | * @return mixed |
| 67 | */ |
| 68 | public static function flattenIfArray( $a ) { |
| 69 | return self::internalFlatten( $a, $res ); |
| 70 | } |
| 71 | |
| 72 | /** |
| 73 | * FIXME: document |
| 74 | * @param mixed $c |
| 75 | * @return mixed |
| 76 | */ |
| 77 | public static function flattenString( $c ) { |
| 78 | $out = self::flattenStringlist( $c ); |
| 79 | if ( count( $out ) === 1 && is_string( $out[0] ) ) { |
| 80 | return $out[0]; |
| 81 | } else { |
| 82 | return $out; |
| 83 | } |
| 84 | } |
| 85 | |
| 86 | /** |
| 87 | * FIXME: document |
| 88 | * @param array $c |
| 89 | * @return array |
| 90 | */ |
| 91 | public static function flattenStringlist( array $c ): array { |
| 92 | $out = []; |
| 93 | $text = ''; |
| 94 | $c = self::flattenIfArray( $c ); |
| 95 | for ( $i = 0, $l = count( $c ); $i < $l; $i++ ) { |
| 96 | $ci = $c[$i]; |
| 97 | if ( is_string( $ci ) ) { |
| 98 | if ( $ci !== '' ) { |
| 99 | $text .= $ci; |
| 100 | } |
| 101 | } else { |
| 102 | if ( $text !== '' ) { |
| 103 | $out[] = $text; |
| 104 | $text = ''; |
| 105 | } |
| 106 | $out[] = $ci; |
| 107 | } |
| 108 | } |
| 109 | if ( $text !== '' ) { |
| 110 | $out[] = $text; |
| 111 | } |
| 112 | return $out; |
| 113 | } |
| 114 | |
| 115 | /** |
| 116 | * @param mixed $value |
| 117 | * @param int $start start of TSR range |
| 118 | * @param int $end end of TSR range |
| 119 | * @return array |
| 120 | */ |
| 121 | public static function getAttrVal( $value, int $start, int $end ): array { |
| 122 | return [ 'value' => $value, 'srcOffsets' => new SourceRange( $start, $end ) ]; |
| 123 | } |
| 124 | |
| 125 | /** |
| 126 | * Build a token array representing <tag>$content</tag> alongwith |
| 127 | * appropriate attributes and TSR info set on the tokens. |
| 128 | * |
| 129 | * @param string $pegSource |
| 130 | * @param string $tagName |
| 131 | * @param string $wtChar |
| 132 | * @param mixed $attrInfo |
| 133 | * @param SourceRange $tsr |
| 134 | * @param int $endPos |
| 135 | * @param mixed $content |
| 136 | * @param bool $addEndTag |
| 137 | * @return array (of tokens) |
| 138 | */ |
| 139 | public static function buildTableTokens( |
| 140 | string $pegSource, string $tagName, string $wtChar, $attrInfo, |
| 141 | SourceRange $tsr, int $endPos, $content, bool $addEndTag = false |
| 142 | ): array { |
| 143 | $dp = new DataParsoid; |
| 144 | $dp->tsr = $tsr; |
| 145 | |
| 146 | if ( $tagName === 'td' ) { |
| 147 | if ( !$attrInfo ) { |
| 148 | // Add a flag that indicates that the tokenizer didn't |
| 149 | // encounter a "|...|" attribute box. This is useful when |
| 150 | // deciding which <td>/<th> cells need attribute fixups. |
| 151 | $dp->setTempFlag( TempData::NO_ATTRS ); |
| 152 | } else { |
| 153 | if ( !$attrInfo[0] && $attrInfo[1] === "" ) { |
| 154 | // FIXME: Skip comments between the two "|" chars |
| 155 | // [ [], "", "|"] => "||" syntax for first <td> on line |
| 156 | $dp->setTempFlag( TempData::NON_MERGEABLE_TABLE_CELL ); |
| 157 | $dp->setTempFlag( TempData::NO_ATTRS ); |
| 158 | } |
| 159 | } |
| 160 | } elseif ( $tagName === 'th' ) { |
| 161 | if ( !$attrInfo ) { |
| 162 | // Add a flag that indicates that the tokenizer didn't |
| 163 | // encounter a "|...|" attribute box. This is useful when |
| 164 | // deciding which <td>/<th> cells need attribute fixups. |
| 165 | $dp->setTempFlag( TempData::NO_ATTRS ); |
| 166 | |
| 167 | // FIXME: Skip comments between the two "!" chars |
| 168 | // "!!foo" in sol context parses as <th>!foo</th> |
| 169 | if ( |
| 170 | is_string( $content[0][0] ?? null ) && |
| 171 | str_starts_with( $content[0][0], "!" ) |
| 172 | ) { |
| 173 | $dp->setTempFlag( TempData::NON_MERGEABLE_TABLE_CELL ); |
| 174 | } |
| 175 | } |
| 176 | } |
| 177 | |
| 178 | $a = []; |
| 179 | if ( $attrInfo ) { |
| 180 | if ( $tagName !== 'caption' ) { |
| 181 | $dp->getTemp()->attrSrc = substr( |
| 182 | $pegSource, $tsr->start, $tsr->end - $tsr->start - strlen( $attrInfo[2] ) |
| 183 | ); |
| 184 | } |
| 185 | $a = $attrInfo[0]; |
| 186 | if ( !$a ) { |
| 187 | $dp->startTagSrc = $wtChar . $attrInfo[1]; |
| 188 | } |
| 189 | if ( ( !$a && $attrInfo[2] ) || $attrInfo[2] !== '|' ) { |
| 190 | // Variation from default |
| 191 | // 1. Separator present with an empty attribute block |
| 192 | // 2. Not "|" |
| 193 | $dp->attrSepSrc = $attrInfo[2]; |
| 194 | } |
| 195 | } elseif ( $tagName !== 'caption' ) { |
| 196 | $dp->getTemp()->attrSrc = ''; |
| 197 | } |
| 198 | |
| 199 | // We consider 1 the start because the table_data_tag and table_heading_tag |
| 200 | // rules don't include the pipe so it isn't accounted for in the tsr passed |
| 201 | // to this function. The rules making use of those rules do some extra |
| 202 | // bookkeeping to adjust for that on the start token returned from this |
| 203 | // function. Of course, table_caption_tag doesn't follow that same pattern |
| 204 | // but that isn't a concern here. |
| 205 | if ( $tagName !== 'caption' && $tsr->start === 1 ) { |
| 206 | $dp->setTempFlag( TempData::AT_SRC_START ); |
| 207 | } |
| 208 | |
| 209 | $tokens = [ new TagTk( $tagName, $a, $dp ) ]; |
| 210 | PHPUtils::pushArray( $tokens, $content ); |
| 211 | |
| 212 | if ( $addEndTag ) { |
| 213 | $dataParsoid = new DataParsoid; |
| 214 | $dataParsoid->tsr = new SourceRange( $endPos, $endPos ); |
| 215 | $tokens[] = new EndTagTk( $tagName, [], $dataParsoid ); |
| 216 | } else { |
| 217 | // We rely on our tree builder to close the table cell (td/th) as needed. |
| 218 | // We cannot close the cell here because cell content can come from |
| 219 | // multiple parsing contexts and we cannot close the tag in the same |
| 220 | // parsing context in which the td was opened: |
| 221 | // Ex: {{1x|{{!}}foo}}{{1x|bar}} has to output <td>foobar</td> |
| 222 | // |
| 223 | // Previously a meta marker was added here for DSR computation, but |
| 224 | // that's complicated now that marker meta handling has been removed |
| 225 | // from ComputeDSR. |
| 226 | } |
| 227 | |
| 228 | return $tokens; |
| 229 | } |
| 230 | |
| 231 | /** |
| 232 | * Build a token representing <tag>, <tag />, or </tag> |
| 233 | * with appropriate attributes set on the token. |
| 234 | * |
| 235 | * @param string $name |
| 236 | * @param string $lcName |
| 237 | * @param array $attribs |
| 238 | * @param mixed $endTag |
| 239 | * @param bool $selfClose |
| 240 | * @param SourceRange $tsr |
| 241 | * @return Token |
| 242 | */ |
| 243 | public static function buildXMLTag( string $name, string $lcName, array $attribs, $endTag, |
| 244 | bool $selfClose, SourceRange $tsr |
| 245 | ): Token { |
| 246 | $tok = null; |
| 247 | $da = new DataParsoid; |
| 248 | $da->tsr = $tsr; |
| 249 | $da->stx = 'html'; |
| 250 | |
| 251 | if ( $name !== $lcName ) { |
| 252 | $da->srcTagName = $name; |
| 253 | } |
| 254 | |
| 255 | if ( $endTag !== null ) { |
| 256 | $tok = new EndTagTk( $lcName, $attribs, $da ); |
| 257 | } elseif ( $selfClose ) { |
| 258 | $da->selfClose = true; |
| 259 | $tok = new SelfclosingTagTk( $lcName, $attribs, $da ); |
| 260 | } else { |
| 261 | $tok = new TagTk( $lcName, $attribs, $da ); |
| 262 | } |
| 263 | |
| 264 | return $tok; |
| 265 | } |
| 266 | |
| 267 | /** |
| 268 | * Inline breaks, flag-enabled rule which detects end positions for |
| 269 | * active higher-level rules in inline and other nested rules. |
| 270 | * Those inner rules are then exited, so that the outer rule can |
| 271 | * handle the end marker. |
| 272 | * @param string $input |
| 273 | * @param int $pos |
| 274 | * @param array $stops |
| 275 | * @param Env $env |
| 276 | * @return bool |
| 277 | * @throws \Exception |
| 278 | */ |
| 279 | public static function inlineBreaks( string $input, int $pos, array $stops, Env $env ): bool { |
| 280 | $c = $input[$pos]; |
| 281 | $c2 = $input[$pos + 1] ?? ''; |
| 282 | |
| 283 | switch ( $c ) { |
| 284 | case '=': |
| 285 | if ( $stops['arrow'] && $c2 === '>' ) { |
| 286 | return true; |
| 287 | } |
| 288 | if ( $stops['equal'] ) { |
| 289 | return true; |
| 290 | } |
| 291 | if ( $stops['h'] ) { |
| 292 | if ( self::$inclAnnRegExp === null ) { |
| 293 | $tags = array_merge( |
| 294 | [ 'noinclude', 'includeonly', 'onlyinclude' ], |
| 295 | $env->getSiteConfig()->getAnnotationTags() |
| 296 | ); |
| 297 | self::$inclAnnRegExp = '|<\/?(?:' . implode( '|', $tags ) . ')>'; |
| 298 | } |
| 299 | return ( $pos === strlen( $input ) - 1 |
| 300 | // possibly more equals followed by spaces or comments |
| 301 | || preg_match( '/^=*(?:[ \t]|<\!--(?:(?!-->).)*-->' |
| 302 | . self::$inclAnnRegExp . ')*(?:[\r\n]|$)/sD', |
| 303 | substr( $input, $pos + 1 ) ) ); |
| 304 | } |
| 305 | return false; |
| 306 | |
| 307 | case '|': |
| 308 | return !$stops['annOrExtTag'] && ( |
| 309 | $stops['templateArg'] |
| 310 | || $stops['tableCellArg'] |
| 311 | || $stops['linkdesc'] |
| 312 | || ( $stops['table'] |
| 313 | && $pos < strlen( $input ) - 1 |
| 314 | && preg_match( '/[}|]/', $input[$pos + 1] ) ) |
| 315 | ); |
| 316 | |
| 317 | case '!': |
| 318 | return $stops['th'] |
| 319 | && !$stops['intemplate'] |
| 320 | && $c2 === '!'; |
| 321 | |
| 322 | case '{': |
| 323 | // {{!}} pipe templates.. |
| 324 | // FIXME: Presumably these should mix with and match | above. |
| 325 | // phpcs:ignore Squiz.WhiteSpace.LanguageConstructSpacing.IncorrectSingle |
| 326 | return ( $stops['tableCellArg'] |
| 327 | && substr( $input, $pos, 5 ) === '{{!}}' ) |
| 328 | || ( $stops['table'] |
| 329 | && substr( $input, $pos, 10 ) === '{{!}}{{!}}' ); |
| 330 | |
| 331 | case '}': |
| 332 | $preproc = $stops['preproc']; |
| 333 | return ( $c2 === '}' && $preproc === '}}' ) |
| 334 | || ( $c2 === '-' && $preproc === '}-' ); |
| 335 | |
| 336 | case ':': |
| 337 | return $stops['colon'] |
| 338 | && !$stops['extlink'] |
| 339 | && !$stops['intemplate'] |
| 340 | && !$stops['linkdesc'] |
| 341 | && !( $stops['preproc'] === '}-' ); |
| 342 | |
| 343 | case ';': |
| 344 | return $stops['semicolon']; |
| 345 | |
| 346 | case "\r": |
| 347 | return $stops['table'] |
| 348 | && preg_match( '/\r\n?\s*[!|]/', substr( $input, $pos ) ); |
| 349 | |
| 350 | case "\n": |
| 351 | // The code below is just a manual / efficient |
| 352 | // version of this check. |
| 353 | // |
| 354 | // stops.table && /^\n\s*[!|]/.test(input.substr(pos)); |
| 355 | // |
| 356 | // It eliminates a substr on the string and eliminates |
| 357 | // a potential perf problem since "\n" and the inline_breaks |
| 358 | // test is common during tokenization. |
| 359 | if ( !$stops['table'] ) { |
| 360 | return false; |
| 361 | } |
| 362 | |
| 363 | // Allow leading whitespace in tables |
| 364 | |
| 365 | // Since we switched on 'c' which is input[pos], |
| 366 | // we know that input[pos] is "\n". |
| 367 | // So, the /^\n/ part of the regexp is already satisfied. |
| 368 | // Look for /\s*[!|]/ below. |
| 369 | $n = strlen( $input ); |
| 370 | for ( $i = $pos + 1; $i < $n; $i++ ) { |
| 371 | $d = $input[$i]; |
| 372 | if ( preg_match( '/[!|]/', $d ) ) { |
| 373 | return true; |
| 374 | } elseif ( !( preg_match( '/\s/', $d ) ) ) { |
| 375 | return false; |
| 376 | } |
| 377 | } |
| 378 | return false; |
| 379 | case '[': |
| 380 | // This is a special case in php's doTableStuff, added in |
| 381 | // response to T2553. If it encounters a `[[`, it bails on |
| 382 | // parsing attributes and interprets it all as content. |
| 383 | return $stops['tableCellArg'] && $c2 === '['; |
| 384 | |
| 385 | case '-': |
| 386 | // Same as above: a special case in doTableStuff, added |
| 387 | // as part of T153140 |
| 388 | return $stops['tableCellArg'] && $c2 === '{'; |
| 389 | |
| 390 | case ']': |
| 391 | if ( $stops['extlink'] ) { |
| 392 | return true; |
| 393 | } |
| 394 | return $stops['preproc'] === ']]' |
| 395 | && $c2 === ']'; |
| 396 | |
| 397 | default: |
| 398 | throw new \RuntimeException( 'Unhandled case!' ); |
| 399 | } |
| 400 | } |
| 401 | |
| 402 | /** |
| 403 | * Pop off the end comments, if any. |
| 404 | * @param array &$attrs |
| 405 | * @return array|null |
| 406 | */ |
| 407 | public static function popComments( array &$attrs ): ?array { |
| 408 | $buf = []; |
| 409 | for ( $i = count( $attrs ) - 1; $i > -1; $i-- ) { |
| 410 | $kv = $attrs[$i]; |
| 411 | if ( is_string( $kv->k ) && !$kv->v && preg_match( '/^\s*$/D', $kv->k ) ) { |
| 412 | // permit whitespace |
| 413 | array_unshift( $buf, $kv->k ); |
| 414 | } elseif ( is_array( $kv->k ) && !$kv->v ) { |
| 415 | // all should be comments |
| 416 | foreach ( $kv->k as $k ) { |
| 417 | if ( !( $k instanceof CommentTk ) ) { |
| 418 | break 2; |
| 419 | } |
| 420 | } |
| 421 | array_splice( $buf, 0, 0, $kv->k ); |
| 422 | } else { |
| 423 | break; |
| 424 | } |
| 425 | } |
| 426 | // ensure we found a comment |
| 427 | while ( $buf && !( $buf[0] instanceof CommentTk ) ) { |
| 428 | array_shift( $buf ); |
| 429 | } |
| 430 | if ( $buf ) { |
| 431 | array_splice( $attrs, -count( $buf ), count( $buf ) ); |
| 432 | return [ 'buf' => $buf, 'commentStartPos' => $buf[0]->dataParsoid->tsr->start ]; |
| 433 | } else { |
| 434 | return null; |
| 435 | } |
| 436 | } |
| 437 | |
| 438 | /** Get a string containing all the autourl terminating characters (as in legacy parser |
| 439 | * Parser.php::makeFreeExternalLink). This list is slightly context-dependent because the |
| 440 | * inclusion of the right parenthesis depends on whether the provided character array $arr |
| 441 | * contains a left parenthesis. |
| 442 | * @param bool $hasLeftParen should be true if the URL in question contains |
| 443 | * a left parenthesis. |
| 444 | * @return string |
| 445 | */ |
| 446 | public static function getAutoUrlTerminatingChars( bool $hasLeftParen ): string { |
| 447 | $chars = Consts::$strippedUrlCharacters; |
| 448 | if ( !$hasLeftParen ) { |
| 449 | $chars .= ')'; |
| 450 | } |
| 451 | return $chars; |
| 452 | } |
| 453 | |
| 454 | /** |
| 455 | * @param Env $env |
| 456 | * @param mixed $token |
| 457 | */ |
| 458 | public static function enforceParserResourceLimits( Env $env, $token ) { |
| 459 | if ( $token instanceof TagTk || $token instanceof SelfclosingTagTk ) { |
| 460 | $resource = null; |
| 461 | switch ( $token->getName() ) { |
| 462 | case 'listItem': |
| 463 | $resource = 'listItem'; |
| 464 | break; |
| 465 | case 'template': |
| 466 | $resource = 'transclusion'; |
| 467 | break; |
| 468 | case 'td': |
| 469 | case 'th': |
| 470 | $resource = 'tableCell'; |
| 471 | break; |
| 472 | } |
| 473 | if ( |
| 474 | $resource !== null && |
| 475 | $env->bumpWt2HtmlResourceUse( $resource ) === false |
| 476 | ) { |
| 477 | // `false` indicates that this bump pushed us over the threshold |
| 478 | // We don't want to log every token above that, which would be `null` |
| 479 | $env->log( 'warn', "wt2html: $resource limit exceeded" ); |
| 480 | } |
| 481 | } |
| 482 | } |
| 483 | |
| 484 | /** |
| 485 | * Protect Parsoid-inserted attributes by escaping them to prevent |
| 486 | * Parsoid-HTML spoofing in wikitext. |
| 487 | * |
| 488 | * @param string $name |
| 489 | * @return string |
| 490 | */ |
| 491 | public static function protectAttrs( string $name ): string { |
| 492 | if ( self::$protectAttrsRegExp === null ) { |
| 493 | self::$protectAttrsRegExp = "/^(about|data-mw.*|data-parsoid.*|data-x.*|" . |
| 494 | DOMDataUtils::DATA_OBJECT_ATTR_NAME . |
| 495 | '|property|rel|typeof)$/i'; |
| 496 | } |
| 497 | return preg_replace( self::$protectAttrsRegExp, 'data-x-$1', $name ); |
| 498 | } |
| 499 | |
| 500 | /** |
| 501 | * Resets $inclAnnRegExp to null to avoid test environment side effects |
| 502 | */ |
| 503 | public static function resetAnnotationIncludeRegex(): void { |
| 504 | self::$inclAnnRegExp = null; |
| 505 | } |
| 506 | |
| 507 | } |