/framework/vendor/zend/Zend/Pdf/StringParser.php
PHP | 724 lines | 446 code | 91 blank | 187 comment | 108 complexity | eef744b1653f0550ce9ee9711d21d452 MD5 | raw file
1<?php 2/** 3 * Zend Framework 4 * 5 * LICENSE 6 * 7 * This source file is subject to the new BSD license that is bundled 8 * with this package in the file LICENSE.txt. 9 * It is also available through the world-wide-web at this URL: 10 * http://framework.zend.com/license/new-bsd 11 * If you did not receive a copy of the license and are unable to 12 * obtain it through the world-wide-web, please send an email 13 * to license@zend.com so we can send you a copy immediately. 14 * 15 * @category Zend 16 * @package Zend_Pdf 17 * @copyright Copyright (c) 2005-2010 Zend Technologies USA Inc. (http://www.zend.com) 18 * @license http://framework.zend.com/license/new-bsd New BSD License 19 * @version $Id: StringParser.php 20096 2010-01-06 02:05:09Z bkarwin $ 20 */ 21 22 23/** Internally used classes */ 24require_once 'Zend/Pdf/Element/Array.php'; 25require_once 'Zend/Pdf/Element/String/Binary.php'; 26require_once 'Zend/Pdf/Element/Boolean.php'; 27require_once 'Zend/Pdf/Element/Dictionary.php'; 28require_once 'Zend/Pdf/Element/Name.php'; 29require_once 'Zend/Pdf/Element/Null.php'; 30require_once 'Zend/Pdf/Element/Numeric.php'; 31require_once 'Zend/Pdf/Element/Object.php'; 32require_once 'Zend/Pdf/Element/Object/Stream.php'; 33require_once 'Zend/Pdf/Element/Reference.php'; 34require_once 'Zend/Pdf/Element/String.php'; 35 36 37/** 38 * PDF string parser 39 * 40 * @package Zend_Pdf 41 * @copyright Copyright (c) 2005-2010 Zend Technologies USA Inc. (http://www.zend.com) 42 * @license http://framework.zend.com/license/new-bsd New BSD License 43 */ 44class Zend_Pdf_StringParser 45{ 46 /** 47 * Source PDF 48 * 49 * @var string 50 */ 51 public $data = ''; 52 53 /** 54 * Current position in a data 55 * 56 * @var integer 57 */ 58 public $offset = 0; 59 60 /** 61 * Current reference context 62 * 63 * @var Zend_Pdf_Element_Reference_Context 64 */ 65 private $_context = null; 66 67 /** 68 * Array of elements of the currently parsed object/trailer 69 * 70 * @var array 71 */ 72 private $_elements = array(); 73 74 /** 75 * PDF objects factory. 76 * 77 * @var Zend_Pdf_ElementFactory_Interface 78 */ 79 private $_objFactory = null; 80 81 82 /** 83 * Clean up resources. 84 * 85 * Clear current state to remove cyclic object references 86 */ 87 public function cleanUp() 88 { 89 $this->_context = null; 90 $this->_elements = array(); 91 $this->_objFactory = null; 92 } 93 94 /** 95 * Character with code $chCode is white space 96 * 97 * @param integer $chCode 98 * @return boolean 99 */ 100 public static function isWhiteSpace($chCode) 101 { 102 if ($chCode == 0x00 || // null character 103 $chCode == 0x09 || // Tab 104 $chCode == 0x0A || // Line feed 105 $chCode == 0x0C || // Form Feed 106 $chCode == 0x0D || // Carriage return 107 $chCode == 0x20 // Space 108 ) { 109 return true; 110 } else { 111 return false; 112 } 113 } 114 115 116 /** 117 * Character with code $chCode is a delimiter character 118 * 119 * @param integer $chCode 120 * @return boolean 121 */ 122 public static function isDelimiter($chCode ) 123 { 124 if ($chCode == 0x28 || // '(' 125 $chCode == 0x29 || // ')' 126 $chCode == 0x3C || // '<' 127 $chCode == 0x3E || // '>' 128 $chCode == 0x5B || // '[' 129 $chCode == 0x5D || // ']' 130 $chCode == 0x7B || // '{' 131 $chCode == 0x7D || // '}' 132 $chCode == 0x2F || // '/' 133 $chCode == 0x25 // '%' 134 ) { 135 return true; 136 } else { 137 return false; 138 } 139 } 140 141 142 /** 143 * Skip white space 144 * 145 * @param boolean $skipComment 146 */ 147 public function skipWhiteSpace($skipComment = true) 148 { 149 if ($skipComment) { 150 while (true) { 151 $this->offset += strspn($this->data, "\x00\t\n\f\r ", $this->offset); 152 153 if ($this->offset < strlen($this->data) && $this->data[$this->offset] == '%') { 154 // Skip comment 155 $this->offset += strcspn($this->data, "\r\n", $this->offset); 156 } else { 157 // Non white space character not equal to '%' is found 158 return; 159 } 160 } 161 } else { 162 $this->offset += strspn($this->data, "\x00\t\n\f\r ", $this->offset); 163 } 164 165// /** Original (non-optimized) implementation. */ 166// 167// while ($this->offset < strlen($this->data)) { 168// if (strpos("\x00\t\n\f\r ", $this->data[$this->offset]) !== false) { 169// $this->offset++; 170// } else if (ord($this->data[$this->offset]) == 0x25 && $skipComment) { // '%' 171// $this->skipComment(); 172// } else { 173// return; 174// } 175// } 176 } 177 178 179 /** 180 * Skip comment 181 */ 182 public function skipComment() 183 { 184 while ($this->offset < strlen($this->data)) 185 { 186 if (ord($this->data[$this->offset]) != 0x0A || // Line feed 187 ord($this->data[$this->offset]) != 0x0d // Carriage return 188 ) { 189 $this->offset++; 190 } else { 191 return; 192 } 193 } 194 } 195 196 197 /** 198 * Read comment line 199 * 200 * @return string 201 */ 202 public function readComment() 203 { 204 $this->skipWhiteSpace(false); 205 206 /** Check if it's a comment line */ 207 if ($this->data[$this->offset] != '%') { 208 return ''; 209 } 210 211 for ($start = $this->offset; 212 $this->offset < strlen($this->data); 213 $this->offset++) { 214 if (ord($this->data[$this->offset]) == 0x0A || // Line feed 215 ord($this->data[$this->offset]) == 0x0d // Carriage return 216 ) { 217 break; 218 } 219 } 220 221 return substr($this->data, $start, $this->offset-$start); 222 } 223 224 225 /** 226 * Returns next lexeme from a pdf stream 227 * 228 * @return string 229 */ 230 public function readLexeme() 231 { 232 // $this->skipWhiteSpace(); 233 while (true) { 234 $this->offset += strspn($this->data, "\x00\t\n\f\r ", $this->offset); 235 236 if ($this->offset < strlen($this->data) && $this->data[$this->offset] == '%') { 237 $this->offset += strcspn($this->data, "\r\n", $this->offset); 238 } else { 239 break; 240 } 241 } 242 243 if ($this->offset >= strlen($this->data)) { 244 return ''; 245 } 246 247 if ( /* self::isDelimiter( ord($this->data[$start]) ) */ 248 strpos('()<>[]{}/%', $this->data[$this->offset]) !== false ) { 249 250 switch (substr($this->data, $this->offset, 2)) { 251 case '<<': 252 $this->offset += 2; 253 return '<<'; 254 break; 255 256 case '>>': 257 $this->offset += 2; 258 return '>>'; 259 break; 260 261 default: 262 return $this->data[$this->offset++]; 263 break; 264 } 265 } else { 266 $start = $this->offset; 267 $this->offset += strcspn($this->data, "()<>[]{}/%\x00\t\n\f\r ", $this->offset); 268 269 return substr($this->data, $start, $this->offset - $start); 270 } 271 } 272 273 274 /** 275 * Read elemental object from a PDF stream 276 * 277 * @return Zend_Pdf_Element 278 * @throws Zend_Pdf_Exception 279 */ 280 public function readElement($nextLexeme = null) 281 { 282 if ($nextLexeme === null) { 283 $nextLexeme = $this->readLexeme(); 284 } 285 286 /** 287 * Note: readElement() method is a public method and could be invoked from other classes. 288 * If readElement() is used not by Zend_Pdf_StringParser::getObject() method, then we should not care 289 * about _elements member management. 290 */ 291 switch ($nextLexeme) { 292 case '(': 293 return ($this->_elements[] = $this->_readString()); 294 295 case '<': 296 return ($this->_elements[] = $this->_readBinaryString()); 297 298 case '/': 299 return ($this->_elements[] = new Zend_Pdf_Element_Name( 300 Zend_Pdf_Element_Name::unescape( $this->readLexeme() ) 301 )); 302 303 case '[': 304 return ($this->_elements[] = $this->_readArray()); 305 306 case '<<': 307 return ($this->_elements[] = $this->_readDictionary()); 308 309 case ')': 310 // fall through to next case 311 case '>': 312 // fall through to next case 313 case ']': 314 // fall through to next case 315 case '>>': 316 // fall through to next case 317 case '{': 318 // fall through to next case 319 case '}': 320 require_once 'Zend/Pdf/Exception.php'; 321 throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X.', 322 $this->offset)); 323 324 default: 325 if (strcasecmp($nextLexeme, 'true') == 0) { 326 return ($this->_elements[] = new Zend_Pdf_Element_Boolean(true)); 327 } else if (strcasecmp($nextLexeme, 'false') == 0) { 328 return ($this->_elements[] = new Zend_Pdf_Element_Boolean(false)); 329 } else if (strcasecmp($nextLexeme, 'null') == 0) { 330 return ($this->_elements[] = new Zend_Pdf_Element_Null()); 331 } 332 333 $ref = $this->_readReference($nextLexeme); 334 if ($ref !== null) { 335 return ($this->_elements[] = $ref); 336 } 337 338 return ($this->_elements[] = $this->_readNumeric($nextLexeme)); 339 } 340 } 341 342 343 /** 344 * Read string PDF object 345 * Also reads trailing ')' from a pdf stream 346 * 347 * @return Zend_Pdf_Element_String 348 * @throws Zend_Pdf_Exception 349 */ 350 private function _readString() 351 { 352 $start = $this->offset; 353 $openedBrackets = 1; 354 355 $this->offset += strcspn($this->data, '()\\', $this->offset); 356 357 while ($this->offset < strlen($this->data)) { 358 switch (ord( $this->data[$this->offset] )) { 359 case 0x28: // '(' - opened bracket in the string, needs balanced pair. 360 $this->offset++; 361 $openedBrackets++; 362 break; 363 364 case 0x29: // ')' - pair to the opened bracket 365 $this->offset++; 366 $openedBrackets--; 367 break; 368 369 case 0x5C: // '\\' - escape sequence, skip next char from a check 370 $this->offset += 2; 371 } 372 373 if ($openedBrackets == 0) { 374 break; // end of string 375 } 376 377 $this->offset += strcspn($this->data, '()\\', $this->offset); 378 } 379 if ($openedBrackets != 0) { 380 require_once 'Zend/Pdf/Exception.php'; 381 throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Unexpected end of file while string reading. Offset - 0x%X. \')\' expected.', $start)); 382 } 383 384 return new Zend_Pdf_Element_String(Zend_Pdf_Element_String::unescape( substr($this->data, 385 $start, 386 $this->offset - $start - 1) )); 387 } 388 389 390 /** 391 * Read binary string PDF object 392 * Also reads trailing '>' from a pdf stream 393 * 394 * @return Zend_Pdf_Element_String_Binary 395 * @throws Zend_Pdf_Exception 396 */ 397 private function _readBinaryString() 398 { 399 $start = $this->offset; 400 401 $this->offset += strspn($this->data, "\x00\t\n\f\r 0123456789abcdefABCDEF", $this->offset); 402 403 if ($this->offset >= strlen($this->data) - 1) { 404 require_once 'Zend/Pdf/Exception.php'; 405 throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Unexpected end of file while reading binary string. Offset - 0x%X. \'>\' expected.', $start)); 406 } 407 408 if ($this->data[$this->offset++] != '>') { 409 require_once 'Zend/Pdf/Exception.php'; 410 throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Unexpected character while binary string reading. Offset - 0x%X.', $this->offset)); 411 } 412 413 return new Zend_Pdf_Element_String_Binary( 414 Zend_Pdf_Element_String_Binary::unescape( substr($this->data, 415 $start, 416 $this->offset - $start - 1) )); 417 } 418 419 420 /** 421 * Read array PDF object 422 * Also reads trailing ']' from a pdf stream 423 * 424 * @return Zend_Pdf_Element_Array 425 * @throws Zend_Pdf_Exception 426 */ 427 private function _readArray() 428 { 429 $elements = array(); 430 431 while ( strlen($nextLexeme = $this->readLexeme()) != 0 ) { 432 if ($nextLexeme != ']') { 433 $elements[] = $this->readElement($nextLexeme); 434 } else { 435 return new Zend_Pdf_Element_Array($elements); 436 } 437 } 438 439 require_once 'Zend/Pdf/Exception.php'; 440 throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Unexpected end of file while array reading. Offset - 0x%X. \']\' expected.', $this->offset)); 441 } 442 443 444 /** 445 * Read dictionary PDF object 446 * Also reads trailing '>>' from a pdf stream 447 * 448 * @return Zend_Pdf_Element_Dictionary 449 * @throws Zend_Pdf_Exception 450 */ 451 private function _readDictionary() 452 { 453 $dictionary = new Zend_Pdf_Element_Dictionary(); 454 455 while ( strlen($nextLexeme = $this->readLexeme()) != 0 ) { 456 if ($nextLexeme != '>>') { 457 $nameStart = $this->offset - strlen($nextLexeme); 458 459 $name = $this->readElement($nextLexeme); 460 $value = $this->readElement(); 461 462 if (!$name instanceof Zend_Pdf_Element_Name) { 463 require_once 'Zend/Pdf/Exception.php'; 464 throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Name object expected while dictionary reading. Offset - 0x%X.', $nameStart)); 465 } 466 467 $dictionary->add($name, $value); 468 } else { 469 return $dictionary; 470 } 471 } 472 473 require_once 'Zend/Pdf/Exception.php'; 474 throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Unexpected end of file while dictionary reading. Offset - 0x%X. \'>>\' expected.', $this->offset)); 475 } 476 477 478 /** 479 * Read reference PDF object 480 * 481 * @param string $nextLexeme 482 * @return Zend_Pdf_Element_Reference 483 */ 484 private function _readReference($nextLexeme = null) 485 { 486 $start = $this->offset; 487 488 if ($nextLexeme === null) { 489 $objNum = $this->readLexeme(); 490 } else { 491 $objNum = $nextLexeme; 492 } 493 if (!ctype_digit($objNum)) { // it's not a reference 494 $this->offset = $start; 495 return null; 496 } 497 498 $genNum = $this->readLexeme(); 499 if (!ctype_digit($genNum)) { // it's not a reference 500 $this->offset = $start; 501 return null; 502 } 503 504 $rMark = $this->readLexeme(); 505 if ($rMark != 'R') { // it's not a reference 506 $this->offset = $start; 507 return null; 508 } 509 510 $ref = new Zend_Pdf_Element_Reference((int)$objNum, (int)$genNum, $this->_context, $this->_objFactory->resolve()); 511 512 return $ref; 513 } 514 515 516 /** 517 * Read numeric PDF object 518 * 519 * @param string $nextLexeme 520 * @return Zend_Pdf_Element_Numeric 521 */ 522 private function _readNumeric($nextLexeme = null) 523 { 524 if ($nextLexeme === null) { 525 $nextLexeme = $this->readLexeme(); 526 } 527 528 return new Zend_Pdf_Element_Numeric($nextLexeme); 529 } 530 531 532 /** 533 * Read inderect object from a PDF stream 534 * 535 * @param integer $offset 536 * @param Zend_Pdf_Element_Reference_Context $context 537 * @return Zend_Pdf_Element_Object 538 */ 539 public function getObject($offset, Zend_Pdf_Element_Reference_Context $context) 540 { 541 if ($offset === null ) { 542 return new Zend_Pdf_Element_Null(); 543 } 544 545 // Save current offset to make getObject() reentrant 546 $offsetSave = $this->offset; 547 548 $this->offset = $offset; 549 $this->_context = $context; 550 $this->_elements = array(); 551 552 $objNum = $this->readLexeme(); 553 if (!ctype_digit($objNum)) { 554 require_once 'Zend/Pdf/Exception.php'; 555 throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. Object number expected.', $this->offset - strlen($objNum))); 556 } 557 558 $genNum = $this->readLexeme(); 559 if (!ctype_digit($genNum)) { 560 require_once 'Zend/Pdf/Exception.php'; 561 throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. Object generation number expected.', $this->offset - strlen($genNum))); 562 } 563 564 $objKeyword = $this->readLexeme(); 565 if ($objKeyword != 'obj') { 566 require_once 'Zend/Pdf/Exception.php'; 567 throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. \'obj\' keyword expected.', $this->offset - strlen($objKeyword))); 568 } 569 570 $objValue = $this->readElement(); 571 572 $nextLexeme = $this->readLexeme(); 573 574 if( $nextLexeme == 'endobj' ) { 575 /** 576 * Object is not generated by factory (thus it's not marked as modified object). 577 * But factory is assigned to the obect. 578 */ 579 $obj = new Zend_Pdf_Element_Object($objValue, (int)$objNum, (int)$genNum, $this->_objFactory->resolve()); 580 581 foreach ($this->_elements as $element) { 582 $element->setParentObject($obj); 583 } 584 585 // Restore offset value 586 $this->offset = $offsetSave; 587 588 return $obj; 589 } 590 591 /** 592 * It's a stream object 593 */ 594 if ($nextLexeme != 'stream') { 595 require_once 'Zend/Pdf/Exception.php'; 596 throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. \'endobj\' or \'stream\' keywords expected.', $this->offset - strlen($nextLexeme))); 597 } 598 599 if (!$objValue instanceof Zend_Pdf_Element_Dictionary) { 600 require_once 'Zend/Pdf/Exception.php'; 601 throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. Stream extent must be preceded by stream dictionary.', $this->offset - strlen($nextLexeme))); 602 } 603 604 /** 605 * References are automatically dereferenced at this moment. 606 */ 607 $streamLength = $objValue->Length->value; 608 609 /** 610 * 'stream' keyword must be followed by either cr-lf sequence or lf character only. 611 * This restriction gives the possibility to recognize all cases exactly 612 */ 613 if ($this->data[$this->offset] == "\r" && 614 $this->data[$this->offset + 1] == "\n" ) { 615 $this->offset += 2; 616 } else if ($this->data[$this->offset] == "\n" ) { 617 $this->offset++; 618 } else { 619 require_once 'Zend/Pdf/Exception.php'; 620 throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. \'stream\' must be followed by either cr-lf sequence or lf character only.', $this->offset - strlen($nextLexeme))); 621 } 622 623 $dataOffset = $this->offset; 624 625 $this->offset += $streamLength; 626 627 $nextLexeme = $this->readLexeme(); 628 if ($nextLexeme != 'endstream') { 629 require_once 'Zend/Pdf/Exception.php'; 630 throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. \'endstream\' keyword expected.', $this->offset - strlen($nextLexeme))); 631 } 632 633 $nextLexeme = $this->readLexeme(); 634 if ($nextLexeme != 'endobj') { 635 require_once 'Zend/Pdf/Exception.php'; 636 throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. \'endobj\' keyword expected.', $this->offset - strlen($nextLexeme))); 637 } 638 639 $obj = new Zend_Pdf_Element_Object_Stream(substr($this->data, 640 $dataOffset, 641 $streamLength), 642 (int)$objNum, 643 (int)$genNum, 644 $this->_objFactory->resolve(), 645 $objValue); 646 647 foreach ($this->_elements as $element) { 648 $element->setParentObject($obj); 649 } 650 651 // Restore offset value 652 $this->offset = $offsetSave; 653 654 return $obj; 655 } 656 657 658 /** 659 * Get length of source string 660 * 661 * @return integer 662 */ 663 public function getLength() 664 { 665 return strlen($this->data); 666 } 667 668 /** 669 * Get source string 670 * 671 * @return string 672 */ 673 public function getString() 674 { 675 return $this->data; 676 } 677 678 679 /** 680 * Parse integer value from a binary stream 681 * 682 * @param string $stream 683 * @param integer $offset 684 * @param integer $size 685 * @return integer 686 */ 687 public static function parseIntFromStream($stream, $offset, $size) 688 { 689 $value = 0; 690 for ($count = 0; $count < $size; $count++) { 691 $value *= 256; 692 $value += ord($stream[$offset + $count]); 693 } 694 695 return $value; 696 } 697 698 699 700 /** 701 * Set current context 702 * 703 * @param Zend_Pdf_Element_Reference_Context $context 704 */ 705 public function setContext(Zend_Pdf_Element_Reference_Context $context) 706 { 707 $this->_context = $context; 708 } 709 710 /** 711 * Object constructor 712 * 713 * Note: PHP duplicates string, which is sent by value, only of it's updated. 714 * Thus we don't need to care about overhead 715 * 716 * @param string $pdfString 717 * @param Zend_Pdf_ElementFactory_Interface $factory 718 */ 719 public function __construct($source, Zend_Pdf_ElementFactory_Interface $factory) 720 { 721 $this->data = $source; 722 $this->_objFactory = $factory; 723 } 724}