library/Zend/Pdf/StringParser.php

   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) 2005-2009 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 17532 2009-08-10 19:04:14Z alexander $
  20  */
  21
  22
  23 /** Zend_Pdf_Element */
  24 require_once 'Zend/Pdf/Element.php';
  25
  26 /** Zend_Pdf_Element_Array */
  27 require_once 'Zend/Pdf/Element/Array.php';
  28
  29 /** Zend_Pdf_Element_String_Binary */
  30 require_once 'Zend/Pdf/Element/String/Binary.php';
  31
  32 /** Zend_Pdf_Element_Boolean */
  33 require_once 'Zend/Pdf/Element/Boolean.php';
  34
  35 /** Zend_Pdf_Element_Dictionary */
  36 require_once 'Zend/Pdf/Element/Dictionary.php';
  37
  38 /** Zend_Pdf_Element_Name */
  39 require_once 'Zend/Pdf/Element/Name.php';
  40
  41 /** Zend_Pdf_Element_Numeric */
  42 require_once 'Zend/Pdf/Element/Numeric.php';
  43
  44 /** Zend_Pdf_Element_Object */
  45 require_once 'Zend/Pdf/Element/Object.php';
  46
  47 /** Zend_Pdf_Element_Reference */
  48 require_once 'Zend/Pdf/Element/Reference.php';
  49
  50 /** Zend_Pdf_Element_Object_Stream */
  51 require_once 'Zend/Pdf/Element/Object/Stream.php';
  52
  53 /** Zend_Pdf_Element_String */
  54 require_once 'Zend/Pdf/Element/String.php';
  55
  56 /** Zend_Pdf_Element_Null */
  57 require_once 'Zend/Pdf/Element/Null.php';
  58
  59 /** Zend_Pdf_Element_Reference_Context */
  60 require_once 'Zend/Pdf/Element/Reference/Context.php';
  61
  62 /** Zend_Pdf_Element_Reference_Table */
  63 require_once 'Zend/Pdf/Element/Reference/Table.php';
  64
  65 /** Zend_Pdf_ElementFactory_Interface */
  66 require_once 'Zend/Pdf/ElementFactory/Interface.php';
  67
  68
  69 /**
  70  * PDF string parser
  71  *
  72  * @package    Zend_Pdf
  73  * @copyright  Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
  74  * @license    http://framework.zend.com/license/new-bsd     New BSD License
  75  */
  76 class Zend_Pdf_StringParser
  77 {
  78     /**
  79      * Source PDF
  80      *
  81      * @var string
  82      */
  83     public $data = '';
  84
  85     /**
  86      * Current position in a data
  87      *
  88      * @var integer
  89      */
  90     public $offset = 0;
  91
  92     /**
  93      * Current reference context
  94      *
  95      * @var Zend_Pdf_Element_Reference_Context
  96      */
  97     private $_context = null;
  98
  99     /**
 100      * Array of elements of the currently parsed object/trailer
 101      *
 102      * @var array
 103      */
 104     private $_elements = array();
 105
 106     /**
 107      * PDF objects factory.
 108      *
 109      * @var Zend_Pdf_ElementFactory_Interface
 110      */
 111     private $_objFactory = null;
 112
 113
 114     /**
 115      * Clean up resources.
 116      *
 117      * Clear current state to remove cyclic object references
 118      */
 119     public function cleanUp()
 120     {
 121         $this->_context = null;
 122         $this->_elements = array();
 123         $this->_objFactory = null;
 124     }
 125
 126     /**
 127      * Character with code $chCode is white space
 128      *
 129      * @param integer $chCode
 130      * @return boolean
 131      */
 132     public static function isWhiteSpace($chCode)
 133     {
 134         if ($chCode == 0x00 || // null character
 135             $chCode == 0x09 || // Tab
 136             $chCode == 0x0A || // Line feed
 137             $chCode == 0x0C || // Form Feed
 138             $chCode == 0x0D || // Carriage return
 139             $chCode == 0x20    // Space
 140            ) {
 141             return true;
 142         } else {
 143             return false;
 144         }
 145     }
 146
 147
 148     /**
 149      * Character with code $chCode is a delimiter character
 150      *
 151      * @param integer $chCode
 152      * @return boolean
 153      */
 154     public static function isDelimiter($chCode )
 155     {
 156         if ($chCode == 0x28 || // '('
 157             $chCode == 0x29 || // ')'
 158             $chCode == 0x3C || // '<'
 159             $chCode == 0x3E || // '>'
 160             $chCode == 0x5B || // '['
 161             $chCode == 0x5D || // ']'
 162             $chCode == 0x7B || // '{'
 163             $chCode == 0x7D || // '}'
 164             $chCode == 0x2F || // '/'
 165             $chCode == 0x25    // '%'
 166            ) {
 167             return true;
 168         } else {
 169             return false;
 170         }
 171     }
 172
 173
 174     /**
 175      * Skip white space
 176      *
 177      * @param boolean $skipComment
 178      */
 179     public function skipWhiteSpace($skipComment = true)
 180     {
 181         while ($this->offset < strlen($this->data)) {
 182             if (self::isWhiteSpace( ord($this->data[$this->offset]) )) {
 183                 $this->offset++;
 184             } else if (ord($this->data[$this->offset]) == 0x25 && $skipComment) { // '%'
 185                 $this->skipComment();
 186             } else {
 187                 return;
 188             }
 189         }
 190     }
 191
 192
 193     /**
 194      * Skip comment
 195      */
 196     public function skipComment()
 197     {
 198         while ($this->offset < strlen($this->data))
 199         {
 200             if (ord($this->data[$this->offset]) != 0x0A || // Line feed
 201                 ord($this->data[$this->offset]) != 0x0d    // Carriage return
 202                ) {
 203                 $this->offset++;
 204             } else {
 205                 return;
 206             }
 207         }
 208     }
 209
 210
 211     /**
 212      * Read comment line
 213      *
 214      * @return string
 215      */
 216     public function readComment()
 217     {
 218         $this->skipWhiteSpace(false);
 219
 220         /** Check if it's a comment line */
 221         if ($this->data[$this->offset] != '%') {
 222             return '';
 223         }
 224
 225         for ($start = $this->offset;
 226              $this->offset < strlen($this->data);
 227              $this->offset++) {
 228             if (ord($this->data[$this->offset]) == 0x0A || // Line feed
 229                 ord($this->data[$this->offset]) == 0x0d    // Carriage return
 230                ) {
 231                 break;
 232             }
 233         }
 234
 235         return substr($this->data, $start, $this->offset-$start);
 236     }
 237
 238
 239     /**
 240      * Returns next lexeme from a pdf stream
 241      *
 242      * @return string
 243      */
 244     public function readLexeme()
 245     {
 246         // $this->skipWhiteSpace();
 247         while (true) {
 248             $this->offset += strspn($this->data, "\x00\t\n\f\r ", $this->offset);
 249
 250             if ($this->data[$this->offset] == '%') {
 251                 preg_match('/[\r\n]/', $this->data, $matches, PREG_OFFSET_CAPTURE, $this->offset);
 252                 if (count($matches) > 0) {
 253                     $this->offset += strlen($matches[0][0]) + $matches[0][1];
 254                 } else {
 255                     $this->offset = strlen($this->data);
 256                 }
 257             } else {
 258                 break;
 259             }
 260         }
 261
 262         if ($this->offset >= strlen($this->data)) {
 263             return '';
 264         }
 265
 266         $start = $this->offset;
 267
 268         if (self::isDelimiter( ord($this->data[$start]) )) {
 269             if ($this->data[$start] == '<' && $this->offset + 1 < strlen($this->data) && $this->data[$start+1] == '<') {
 270                 $this->offset += 2;
 271                 return '<<';
 272             } else if ($this->data[$start] == '>' && $this->offset + 1 < strlen($this->data) && $this->data[$start+1] == '>') {
 273                 $this->offset += 2;
 274                 return '>>';
 275             } else {
 276                 $this->offset++;
 277                 return $this->data[$start];
 278             }
 279         } else {
 280             while ( ($this->offset < strlen($this->data)) &&
 281                     (!self::isDelimiter(  ord($this->data[$this->offset]) )) &&
 282                     (!self::isWhiteSpace( ord($this->data[$this->offset]) ))   ) {
 283                 $this->offset++;
 284             }
 285
 286             return substr($this->data, $start, $this->offset - $start);
 287         }
 288     }
 289
 290
 291     /**
 292      * Read elemental object from a PDF stream
 293      *
 294      * @return Zend_Pdf_Element
 295      * @throws Zend_Pdf_Exception
 296      */
 297     public function readElement($nextLexeme = null)
 298     {
 299         if ($nextLexeme === null) {
 300             $nextLexeme = $this->readLexeme();
 301         }
 302
 303         /**
 304          * Note: readElement() method is a public method and could be invoked from other classes.
 305          * If readElement() is used not by Zend_Pdf_StringParser::getObject() method, then we should not care
 306          * about _elements member management.
 307          */
 308         switch ($nextLexeme) {
 309             case '(':
 310                 return ($this->_elements[] = $this->_readString());
 311
 312             case '<':
 313                 return ($this->_elements[] = $this->_readBinaryString());
 314
 315             case '/':
 316                 return ($this->_elements[] = new Zend_Pdf_Element_Name(
 317                                                 Zend_Pdf_Element_Name::unescape( $this->readLexeme() )
 318                                                                       ));
 319
 320             case '[':
 321                 return ($this->_elements[] = $this->_readArray());
 322
 323             case '<<':
 324                 return ($this->_elements[] = $this->_readDictionary());
 325
 326             case ')':
 327                 // fall through to next case
 328             case '>':
 329                 // fall through to next case
 330             case ']':
 331                 // fall through to next case
 332             case '>>':
 333                 // fall through to next case
 334             case '{':
 335                 // fall through to next case
 336             case '}':
 337                 throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X.',
 338                                                 $this->offset));
 339
 340             default:
 341                 if (strcasecmp($nextLexeme, 'true') == 0) {
 342                     return ($this->_elements[] = new Zend_Pdf_Element_Boolean(true));
 343                 } else if (strcasecmp($nextLexeme, 'false') == 0) {
 344                     return ($this->_elements[] = new Zend_Pdf_Element_Boolean(false));
 345                 } else if (strcasecmp($nextLexeme, 'null') == 0) {
 346                     return ($this->_elements[] = new Zend_Pdf_Element_Null());
 347                 }
 348
 349                 $ref = $this->_readReference($nextLexeme);
 350                 if ($ref !== null) {
 351                     return ($this->_elements[] = $ref);
 352                 }
 353
 354                 return ($this->_elements[] = $this->_readNumeric($nextLexeme));
 355         }
 356     }
 357
 358
 359     /**
 360      * Read string PDF object
 361      * Also reads trailing ')' from a pdf stream
 362      *
 363      * @return Zend_Pdf_Element_String
 364      * @throws Zend_Pdf_Exception
 365      */
 366     private function _readString()
 367     {
 368         $start = $this->offset;
 369         $openedBrackets = 1;
 370
 371         while ($this->offset < strlen($this->data)) {
 372             switch (ord( $this->data[$this->offset] )) {
 373                 case 0x28: // '(' - opened bracket in the string, needs balanced pair.
 374                     $openedBrackets++;
 375                     break;
 376
 377                 case 0x29: // ')' - pair to the opened bracket
 378                     $openedBrackets--;
 379                     break;
 380
 381                 case 0x5C: // '\\' - escape sequence, skip next char from a check
 382                     $this->offset++;
 383             }
 384
 385             $this->offset++;
 386             if ($openedBrackets == 0) {
 387                 break; // end of string
 388             }
 389         }
 390         if ($openedBrackets != 0) {
 391             throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Unexpected end of file while string reading. Offset - 0x%X. \')\' expected.', $start));
 392         }
 393
 394         return new Zend_Pdf_Element_String(Zend_Pdf_Element_String::unescape( substr($this->data,
 395                                                                  $start,
 396                                                                  $this->offset - $start - 1) ));
 397     }
 398
 399
 400     /**
 401      * Read binary string PDF object
 402      * Also reads trailing '>' from a pdf stream
 403      *
 404      * @return Zend_Pdf_Element_String_Binary
 405      * @throws Zend_Pdf_Exception
 406      */
 407     private function _readBinaryString()
 408     {
 409         $start = $this->offset;
 410
 411         while ($this->offset < strlen($this->data)) {
 412             if (self::isWhiteSpace( ord($this->data[$this->offset]) ) ||
 413                 ctype_xdigit( $this->data[$this->offset] ) ) {
 414                 $this->offset++;
 415             } else if ($this->data[$this->offset] == '>') {
 416                 $this->offset++;
 417                 return new Zend_Pdf_Element_String_Binary(
 418                                Zend_Pdf_Element_String_Binary::unescape( substr($this->data,
 419                                                                     $start,
 420                                                                     $this->offset - $start - 1) ));
 421             } else {
 422                 throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Unexpected character while binary string reading. Offset - 0x%X.', $this->offset));
 423             }
 424         }
 425         throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Unexpected end of file while binary string reading. Offset - 0x%X. \'>\' expected.', $start));
 426     }
 427
 428
 429     /**
 430      * Read array PDF object
 431      * Also reads trailing ']' from a pdf stream
 432      *
 433      * @return Zend_Pdf_Element_Array
 434      * @throws Zend_Pdf_Exception
 435      */
 436     private function _readArray()
 437     {
 438         $elements = array();
 439
 440         while ( strlen($nextLexeme = $this->readLexeme()) != 0 ) {
 441             if ($nextLexeme != ']') {
 442                 $elements[] = $this->readElement($nextLexeme);
 443             } else {
 444                 return new Zend_Pdf_Element_Array($elements);
 445             }
 446         }
 447
 448         throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Unexpected end of file while array reading. Offset - 0x%X. \']\' expected.', $this->offset));
 449     }
 450
 451
 452     /**
 453      * Read dictionary PDF object
 454      * Also reads trailing '>>' from a pdf stream
 455      *
 456      * @return Zend_Pdf_Element_Dictionary
 457      * @throws Zend_Pdf_Exception
 458      */
 459     private function _readDictionary()
 460     {
 461         $dictionary = new Zend_Pdf_Element_Dictionary();
 462
 463         while ( strlen($nextLexeme = $this->readLexeme()) != 0 ) {
 464             if ($nextLexeme != '>>') {
 465                 $nameStart = $this->offset - strlen($nextLexeme);
 466
 467                 $name  = $this->readElement($nextLexeme);
 468                 $value = $this->readElement();
 469
 470                 if (!$name instanceof Zend_Pdf_Element_Name) {
 471                     throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Name object expected while dictionary reading. Offset - 0x%X.', $nameStart));
 472                 }
 473
 474                 $dictionary->add($name, $value);
 475             } else {
 476                 return $dictionary;
 477             }
 478         }
 479
 480         throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Unexpected end of file while dictionary reading. Offset - 0x%X. \'>>\' expected.', $this->offset));
 481     }
 482
 483
 484     /**
 485      * Read reference PDF object
 486      *
 487      * @param string $nextLexeme
 488      * @return Zend_Pdf_Element_Reference
 489      */
 490     private function _readReference($nextLexeme = null)
 491     {
 492         $start = $this->offset;
 493
 494         if ($nextLexeme === null) {
 495             $objNum = $this->readLexeme();
 496         } else {
 497             $objNum = $nextLexeme;
 498         }
 499         if (!ctype_digit($objNum)) { // it's not a reference
 500             $this->offset = $start;
 501             return null;
 502         }
 503
 504         $genNum = $this->readLexeme();
 505         if (!ctype_digit($genNum)) { // it's not a reference
 506             $this->offset = $start;
 507             return null;
 508         }
 509
 510         $rMark  = $this->readLexeme();
 511         if ($rMark != 'R') { // it's not a reference
 512             $this->offset = $start;
 513             return null;
 514         }
 515
 516         $ref = new Zend_Pdf_Element_Reference((int)$objNum, (int)$genNum, $this->_context, $this->_objFactory->resolve());
 517
 518         return $ref;
 519     }
 520
 521
 522     /**
 523      * Read numeric PDF object
 524      *
 525      * @param string $nextLexeme
 526      * @return Zend_Pdf_Element_Numeric
 527      */
 528     private function _readNumeric($nextLexeme = null)
 529     {
 530         if ($nextLexeme === null) {
 531             $nextLexeme = $this->readLexeme();
 532         }
 533
 534         return new Zend_Pdf_Element_Numeric($nextLexeme);
 535     }
 536
 537
 538     /**
 539      * Read inderect object from a PDF stream
 540      *
 541      * @param integer $offset
 542      * @param Zend_Pdf_Element_Reference_Context $context
 543      * @return Zend_Pdf_Element_Object
 544      */
 545     public function getObject($offset, Zend_Pdf_Element_Reference_Context $context)
 546     {
 547         if ($offset === null ) {
 548             return new Zend_Pdf_Element_Null();
 549         }
 550
 551         // Save current offset to make getObject() reentrant
 552         $offsetSave = $this->offset;
 553
 554         $this->offset    = $offset;
 555         $this->_context  = $context;
 556         $this->_elements = array();
 557
 558         $objNum = $this->readLexeme();
 559         if (!ctype_digit($objNum)) {
 560             throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. Object number expected.', $this->offset - strlen($objNum)));
 561         }
 562
 563         $genNum = $this->readLexeme();
 564         if (!ctype_digit($genNum)) {
 565             throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. Object generation number expected.', $this->offset - strlen($genNum)));
 566         }
 567
 568         $objKeyword = $this->readLexeme();
 569         if ($objKeyword != 'obj') {
 570             throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. \'obj\' keyword expected.', $this->offset - strlen($objKeyword)));
 571         }
 572
 573         $objValue = $this->readElement();
 574
 575         $nextLexeme = $this->readLexeme();
 576
 577         if( $nextLexeme == 'endobj' ) {
 578             /**
 579              * Object is not generated by factory (thus it's not marked as modified object).
 580              * But factory is assigned to the obect.
 581              */
 582             $obj = new Zend_Pdf_Element_Object($objValue, (int)$objNum, (int)$genNum, $this->_objFactory->resolve());
 583
 584             foreach ($this->_elements as $element) {
 585                 $element->setParentObject($obj);
 586             }
 587
 588             // Restore offset value
 589             $this->offset = $offsetSave;
 590
 591             return $obj;
 592         }
 593
 594         /**
 595          * It's a stream object
 596          */
 597         if ($nextLexeme != 'stream') {
 598             throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. \'endobj\' or \'stream\' keywords expected.', $this->offset - strlen($nextLexeme)));
 599         }
 600
 601         if (!$objValue instanceof Zend_Pdf_Element_Dictionary) {
 602             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)));
 603         }
 604
 605         /**
 606          * References are automatically dereferenced at this moment.
 607          */
 608         $streamLength = $objValue->Length->value;
 609
 610         /**
 611          * 'stream' keyword must be followed by either cr-lf sequence or lf character only.
 612          * This restriction gives the possibility to recognize all cases exactly
 613          */
 614         if ($this->data[$this->offset] == "\r" &&
 615             $this->data[$this->offset + 1] == "\n"    ) {
 616             $this->offset += 2;
 617         } else if ($this->data[$this->offset] == "\n"    ) {
 618             $this->offset++;
 619         } else {
 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             throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. \'endstream\' keyword expected.', $this->offset - strlen($nextLexeme)));
 630         }
 631
 632         $nextLexeme = $this->readLexeme();
 633         if ($nextLexeme != 'endobj') {
 634             throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. \'endobj\' keyword expected.', $this->offset - strlen($nextLexeme)));
 635         }
 636
 637         $obj = new Zend_Pdf_Element_Object_Stream(substr($this->data,
 638                                                          $dataOffset,
 639                                                          $streamLength),
 640                                                   (int)$objNum,
 641                                                   (int)$genNum,
 642                                                   $this->_objFactory->resolve(),
 643                                                   $objValue);
 644
 645         foreach ($this->_elements as $element) {
 646             $element->setParentObject($obj);
 647         }
 648
 649         // Restore offset value
 650         $this->offset = $offsetSave;
 651
 652         return $obj;
 653     }
 654
 655
 656     /**
 657      * Get length of source string
 658      *
 659      * @return integer
 660      */
 661     public function getLength()
 662     {
 663         return strlen($this->data);
 664     }
 665
 666     /**
 667      * Get source string
 668      *
 669      * @return string
 670      */
 671     public function getString()
 672     {
 673         return $this->data;
 674     }
 675
 676
 677     /**
 678      * Parse integer value from a binary stream
 679      *
 680      * @param string $stream
 681      * @param integer $offset
 682      * @param integer $size
 683      * @return integer
 684      */
 685     public static function parseIntFromStream($stream, $offset, $size)
 686     {
 687         $value = 0;
 688         for ($count = 0; $count < $size; $count++) {
 689             $value *= 256;
 690             $value += ord($stream[$offset + $count]);
 691         }
 692
 693         return $value;
 694     }
 695
 696
 697
 698     /**
 699      * Set current context
 700      *
 701      * @param Zend_Pdf_Element_Reference_Context $context
 702      */
 703     public function setContext(Zend_Pdf_Element_Reference_Context $context)
 704     {
 705         $this->_context = $context;
 706     }
 707
 708     /**
 709      * Object constructor
 710      *
 711      * Note: PHP duplicates string, which is sent by value, only of it's updated.
 712      * Thus we don't need to care about overhead
 713      *
 714      * @param string $pdfString
 715      * @param Zend_Pdf_ElementFactory_Interface $factory
 716      */
 717     public function __construct($source, Zend_Pdf_ElementFactory_Interface $factory)
 718     {
 719         $this->data         = $source;
 720         $this->_objFactory  = $factory;
 721     }
 722 }