| Index: trunk/extensions/wikihiero/wikihiero.body.php |
| — | — | @@ -35,12 +35,6 @@ |
| 36 | 36 | define( "WH_MODE_IMAGE", 3 ); // picture (PNG) // not supported |
| 37 | 37 | define( "WH_MODE_RAW", 4 ); // MdC test as it |
| 38 | 38 | |
| 39 | | -define( "WH_TYPE_NONE", 0 ); |
| 40 | | -define( "WH_TYPE_GLYPH", 1 ); // rendered items |
| 41 | | -define( "WH_TYPE_CODE", 2 ); // single code as ':', '*', '!', '(' or ')' |
| 42 | | -define( "WH_TYPE_SPECIAL", 3 ); // advanced code (more than 1 caracter) |
| 43 | | -define( "WH_TYPE_END", 4 ); // end of line '!' |
| 44 | | - |
| 45 | 39 | define( "WH_SCALE_DEFAULT", -1 ); // use default scale |
| 46 | 40 | |
| 47 | 41 | global $wgExtensionAssetsPath; |
| — | — | @@ -278,79 +272,11 @@ |
| 279 | 273 | $html .= "<hr />\n"; |
| 280 | 274 | } |
| 281 | 275 | |
| 282 | | - // ------------------------------------------------------------------------ |
| 283 | | - // Split text into blocks, then split blocks into items |
| 284 | | - $blocks = array(); |
| 285 | | - $blocks[0] = array(); |
| 286 | | - $blocks[0][0] = ""; |
| 287 | | - $blocks_id = 0; |
| 288 | | - $item_id = 0; |
| 289 | | - $parenthesis = 0; |
| 290 | | - $type = WH_TYPE_NONE; |
| | 276 | + $tokenizer = new HieroTokenizer( $hiero ); |
| | 277 | + $blocks = $tokenizer->tokenize(); |
| | 278 | + $contentHtml = $tableHtml = $tableContentHtml = ""; |
| 291 | 279 | $is_cartouche = false; |
| 292 | 280 | |
| 293 | | - for ( $char = 0; $char < strlen( $hiero ); $char++ ) { |
| 294 | | - |
| 295 | | - if ( $hiero[$char] == '(' ) { |
| 296 | | - $parenthesis++; |
| 297 | | - } elseif ( $hiero[$char] == ')' ) { |
| 298 | | - $parenthesis--; |
| 299 | | - } |
| 300 | | - |
| 301 | | - if ( $parenthesis == 0 ) { |
| 302 | | - if ( $hiero[$char] == '-' || $hiero[$char] == ' ' ) { |
| 303 | | - if ( $type != WH_TYPE_NONE ) { |
| 304 | | - $blocks_id++; |
| 305 | | - $blocks[$blocks_id] = array(); |
| 306 | | - $item_id = 0; |
| 307 | | - $blocks[$blocks_id][$item_id] = ""; |
| 308 | | - $type = WH_TYPE_NONE; |
| 309 | | - } |
| 310 | | - } |
| 311 | | - } else {// don't split block if inside parenthesis |
| 312 | | - if ( $hiero[$char] == '-' ) { |
| 313 | | - $item_id++; |
| 314 | | - $blocks[$blocks_id][$item_id] = '-'; |
| 315 | | - $type = WH_TYPE_CODE; |
| 316 | | - } |
| 317 | | - } |
| 318 | | - |
| 319 | | - if ( $hiero[$char] == '!' ) { |
| 320 | | - if ( $item_id > 0 ) { |
| 321 | | - $blocks_id++; |
| 322 | | - $blocks[$blocks_id] = array(); |
| 323 | | - $item_id = 0; |
| 324 | | - } |
| 325 | | - $blocks[$blocks_id][$item_id] = $hiero[$char]; |
| 326 | | - $type = WH_TYPE_END; |
| 327 | | - |
| 328 | | - } elseif ( preg_match( "/[*:()]/", $hiero[$char] ) ) { |
| 329 | | - |
| 330 | | - if ( $type == WH_TYPE_GLYPH || $type == WH_TYPE_CODE ) { |
| 331 | | - $item_id++; |
| 332 | | - $blocks[$blocks_id][$item_id] = ""; |
| 333 | | - } |
| 334 | | - $blocks[$blocks_id][$item_id] = $hiero[$char]; |
| 335 | | - $type = WH_TYPE_CODE; |
| 336 | | - |
| 337 | | - } elseif ( ctype_alnum( $hiero[$char] ) || $hiero[$char] == '.' || $hiero[$char] == '<' |
| 338 | | - || $hiero[$char] == '>' || $hiero[$char] == '\\' ) { |
| 339 | | - if ( $type == WH_TYPE_END ) { |
| 340 | | - $blocks_id++; |
| 341 | | - $blocks[$blocks_id] = array(); |
| 342 | | - $item_id = 0; |
| 343 | | - $blocks[$blocks_id][$item_id] = ""; |
| 344 | | - } elseif ( $type == WH_TYPE_CODE ) { |
| 345 | | - $item_id++; |
| 346 | | - $blocks[$blocks_id][$item_id] = ""; |
| 347 | | - } |
| 348 | | - $blocks[$blocks_id][$item_id] .= $hiero[$char]; |
| 349 | | - $type = WH_TYPE_GLYPH; |
| 350 | | - } |
| 351 | | - } |
| 352 | | - |
| 353 | | - $contentHtml = $tableHtml = $tableContentHtml = ""; |
| 354 | | - |
| 355 | 281 | // ------------------------------------------------------------------------ |
| 356 | 282 | // Loop into all blocks |
| 357 | 283 | foreach ( $blocks as $code ) { |
| — | — | @@ -499,3 +425,106 @@ |
| 500 | 426 | return substr( $file, strlen( self::IMAGE_PREFIX ), -( 1 + strlen( self::IMAGE_EXT ) ) ); |
| 501 | 427 | } |
| 502 | 428 | } |
| | 429 | + |
| | 430 | +/** |
| | 431 | + * Hieroglyphs tokenizer class |
| | 432 | + */ |
| | 433 | +/*private*/ class HieroTokenizer { |
| | 434 | + const TYPE_NONE = 0; |
| | 435 | + const TYPE_GLYPH = 1; // rendered items |
| | 436 | + const TYPE_CODE = 2; // single code as ':', '*', '!', '(' or ')' |
| | 437 | + const TYPE_SPECIAL = 3; // advanced code (more than 1 caracter) |
| | 438 | + const TYPE_END = 4; // end of line '!' |
| | 439 | + |
| | 440 | + private $text; |
| | 441 | + private $blocks = false; |
| | 442 | + private $blocks_id = 0; |
| | 443 | + private $item_id = 0; |
| | 444 | + |
| | 445 | + /** |
| | 446 | + * Constructor |
| | 447 | + * |
| | 448 | + * @param $text string: |
| | 449 | + */ |
| | 450 | + public function __construct( $text ) { |
| | 451 | + $this->text = $text; |
| | 452 | + } |
| | 453 | + |
| | 454 | + /** |
| | 455 | + * Split text into blocks, then split blocks into items |
| | 456 | + * |
| | 457 | + * @return array: tokenized text |
| | 458 | + */ |
| | 459 | + public function tokenize() { |
| | 460 | + if ( $this->blocks !== false ) { |
| | 461 | + return $this->blocks; |
| | 462 | + } |
| | 463 | + $this->blocks = array( array( '' ) ); |
| | 464 | + $parentheses = 0; |
| | 465 | + $type = self::TYPE_NONE; |
| | 466 | + |
| | 467 | + for ( $i = 0; $i < strlen( $this->text ); $i++ ) { |
| | 468 | + $char = $this->text[$i]; |
| | 469 | + |
| | 470 | + if ( $char == '(' ) { |
| | 471 | + $parentheses++; |
| | 472 | + } elseif ( $char == ')' ) { |
| | 473 | + $parentheses--; |
| | 474 | + } |
| | 475 | + |
| | 476 | + if ( $parentheses == 0 ) { |
| | 477 | + if ( $char == '-' || $char == ' ' ) { |
| | 478 | + if ( $type != self::TYPE_NONE ) { |
| | 479 | + $this->addBlock( '' ); |
| | 480 | + $type = self::TYPE_NONE; |
| | 481 | + } |
| | 482 | + } |
| | 483 | + } else {// don't split block if inside parentheses |
| | 484 | + if ( $char == '-' ) { |
| | 485 | + $this->addItem( '-' ); |
| | 486 | + $type = self::TYPE_CODE; |
| | 487 | + } |
| | 488 | + } |
| | 489 | + |
| | 490 | + if ( $char == '!' ) { |
| | 491 | + if ( $this->item_id > 0 ) { |
| | 492 | + $this->addBlock(); |
| | 493 | + } |
| | 494 | + $this->blocks[$this->blocks_id][$this->item_id] = $char; |
| | 495 | + $type = self::TYPE_END; |
| | 496 | + |
| | 497 | + } elseif ( preg_match( '/[*:()]/', $char ) ) { |
| | 498 | + if ( $type == self::TYPE_GLYPH || $type == self::TYPE_CODE ) { |
| | 499 | + $this->addItem( '' ); |
| | 500 | + } |
| | 501 | + $this->blocks[$this->blocks_id][$this->item_id] = $char; |
| | 502 | + $type = self::TYPE_CODE; |
| | 503 | + |
| | 504 | + } elseif ( ctype_alnum( $char ) || $char == '.' || $char == '<' |
| | 505 | + || $char == '>' || $char == '\\' ) { |
| | 506 | + if ( $type == self::TYPE_END ) { |
| | 507 | + $this->addBlock( '' ); |
| | 508 | + } elseif ( $type == self::TYPE_CODE ) { |
| | 509 | + $this->addItem( '' ); |
| | 510 | + } |
| | 511 | + $this->blocks[$this->blocks_id][$this->item_id] .= $char; |
| | 512 | + $type = self::TYPE_GLYPH; |
| | 513 | + } |
| | 514 | + } |
| | 515 | + return $this->blocks; |
| | 516 | + } |
| | 517 | + |
| | 518 | + private function addBlock( $newItem = false ) { |
| | 519 | + $this->blocks_id++; |
| | 520 | + $this->blocks[$this->blocks_id] = array(); |
| | 521 | + $this->item_id = 0; |
| | 522 | + if ( $newItem !== false ) { |
| | 523 | + $this->blocks[$this->blocks_id][$this->item_id] = $newItem; |
| | 524 | + } |
| | 525 | + } |
| | 526 | + |
| | 527 | + private function addItem( $item ) { |
| | 528 | + $this->item_id++; |
| | 529 | + $this->blocks[$this->blocks_id][$this->item_id] = $item; |
| | 530 | + } |
| | 531 | +} |