library/Zend/Pdf/Parser.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  * @package    Zend_Pdf
  16  * @copyright  Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
  17  * @license    http://framework.zend.com/license/new-bsd     New BSD License
  18  */
  19
  20 /** Zend_Pdf_Element */
  21 require_once 'Zend/Pdf/Element.php';
  22
  23 /** Zend_Pdf_Element_Array */
  24 require_once 'Zend/Pdf/Element/Array.php';
  25
  26 /** Zend_Pdf_Element_String_Binary */
  27 require_once 'Zend/Pdf/Element/String/Binary.php';
  28
  29 /** Zend_Pdf_Element_Boolean */
  30 require_once 'Zend/Pdf/Element/Boolean.php';
  31
  32 /** Zend_Pdf_Element_Dictionary */
  33 require_once 'Zend/Pdf/Element/Dictionary.php';
  34
  35 /** Zend_Pdf_Element_Name */
  36 require_once 'Zend/Pdf/Element/Name.php';
  37
  38 /** Zend_Pdf_Element_Numeric */
  39 require_once 'Zend/Pdf/Element/Numeric.php';
  40
  41 /** Zend_Pdf_Element_Object */
  42 require_once 'Zend/Pdf/Element/Object.php';
  43
  44 /** Zend_Pdf_Element_Reference */
  45 require_once 'Zend/Pdf/Element/Reference.php';
  46
  47 /** Zend_Pdf_Element_Object_Stream */
  48 require_once 'Zend/Pdf/Element/Object/Stream.php';
  49
  50 /** Zend_Pdf_Element_String */
  51 require_once 'Zend/Pdf/Element/String.php';
  52
  53 /** Zend_Pdf_Element_Null */
  54 require_once 'Zend/Pdf/Element/Null.php';
  55
  56 /** Zend_Pdf_Element_Reference_Context */
  57 require_once 'Zend/Pdf/Element/Reference/Context.php';
  58
  59 /** Zend_Pdf_Element_Reference_Table */
  60 require_once 'Zend/Pdf/Element/Reference/Table.php';
  61
  62 /** Zend_Pdf_Trailer_Keeper */
  63 require_once 'Zend/Pdf/Trailer/Keeper.php';
  64
  65 /** Zend_Pdf_ElementFactory_Interface */
  66 require_once 'Zend/Pdf/ElementFactory/Interface.php';
  67
  68 /** Zend_Pdf_PhpArray */
  69 require_once 'Zend/Pdf/PhpArray.php';
  70
  71 /** Zend_Pdf_StringParser */
  72 require_once 'Zend/Pdf/StringParser.php';
  73
  74 /** Zend_Pdf_Parser_Stream */
  75 require_once 'Zend/Pdf/Parser/Stream.php';
  76
  77
  78 /**
  79  * PDF file parser
  80  *
  81  * @package    Zend_Pdf
  82  * @copyright  Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
  83  * @license    http://framework.zend.com/license/new-bsd     New BSD License
  84  */
  85 class Zend_Pdf_Parser
  86 {
  87     /**
  88      * String parser
  89      *
  90      * @var Zend_Pdf_StringParser
  91      */
  92     private $_stringParser;
  93
  94     /**
  95      * Last PDF file trailer
  96      *
  97      * @var Zend_Pdf_Trailer_Keeper
  98      */
  99     private $_trailer;
 100
 101
 102     /**
 103      * Get length of source PDF
 104      *
 105      * @return integer
 106      */
 107     public function getPDFLength()
 108     {
 109         return strlen($this->_stringParser->data);
 110     }
 111
 112     /**
 113      * Get PDF String
 114      *
 115      * @return string
 116      */
 117     public function getPDFString()
 118     {
 119         return $this->_stringParser->data;
 120     }
 121
 122     /**
 123      * Load XReference table and referenced objects
 124      *
 125      * @param integer $offset
 126      * @throws Zend_Pdf_Exception
 127      * @return Zend_Pdf_Trailer_Keeper
 128      */
 129     private function _loadXRefTable($offset)
 130     {
 131         $this->_stringParser->offset = $offset;
 132
 133         $refTable = new Zend_Pdf_Element_Reference_Table();
 134         $context  = new Zend_Pdf_Element_Reference_Context($this->_stringParser, $refTable);
 135         $this->_stringParser->setContext($context);
 136
 137         $nextLexeme = $this->_stringParser->readLexeme();
 138         if ($nextLexeme == 'xref') {
 139             /**
 140              * Common cross-reference table
 141              */
 142             $this->_stringParser->skipWhiteSpace();
 143             while ( ($nextLexeme = $this->_stringParser->readLexeme()) != 'trailer' ) {
 144                 if (!ctype_digit($nextLexeme)) {
 145                     throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. Cross-reference table subheader values must contain only digits.', $this->_stringParser->offset-strlen($nextLexeme)));
 146                 }
 147                 $objNum = (int)$nextLexeme;
 148
 149                 $refCount = $this->_stringParser->readLexeme();
 150                 if (!ctype_digit($refCount)) {
 151                     throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. Cross-reference table subheader values must contain only digits.', $this->_stringParser->offset-strlen($refCount)));
 152                 }
 153
 154                 $this->_stringParser->skipWhiteSpace();
 155                 while ($refCount > 0) {
 156                     $objectOffset = substr($this->_stringParser->data, $this->_stringParser->offset, 10);
 157                     if (!ctype_digit($objectOffset)) {
 158                         throw new Zend_Pdf_Exception(sprintf('PDF file cross-reference table syntax error. Offset - 0x%X. Offset must contain only digits.', $this->_stringParser->offset));
 159                     }
 160                     // Force $objectOffset to be treated as decimal instead of octal number
 161                     for ($numStart = 0; $numStart < strlen($objectOffset)-1; $numStart++) {
 162                         if ($objectOffset[$numStart] != '0') {
 163                             break;
 164                         }
 165                     }
 166                     $objectOffset = substr($objectOffset, $numStart);
 167                     $this->_stringParser->offset += 10;
 168
 169                     if ( !Zend_Pdf_StringParser::isWhiteSpace(ord( $this->_stringParser->data[$this->_stringParser->offset] )) ) {
 170                         throw new Zend_Pdf_Exception(sprintf('PDF file cross-reference table syntax error. Offset - 0x%X. Value separator must be white space.', $this->_stringParser->offset));
 171                     }
 172                     $this->_stringParser->offset++;
 173
 174                     $genNumber = substr($this->_stringParser->data, $this->_stringParser->offset, 5);
 175                     if (!ctype_digit($objectOffset)) {
 176                         throw new Zend_Pdf_Exception(sprintf('PDF file cross-reference table syntax error. Offset - 0x%X. Offset must contain only digits.', $this->_stringParser->offset));
 177                     }
 178                     // Force $objectOffset to be treated as decimal instead of octal number
 179                     for ($numStart = 0; $numStart < strlen($genNumber)-1; $numStart++) {
 180                         if ($genNumber[$numStart] != '0') {
 181                             break;
 182                         }
 183                     }
 184                     $genNumber = substr($genNumber, $numStart);
 185                     $this->_stringParser->offset += 5;
 186
 187                     if ( !Zend_Pdf_StringParser::isWhiteSpace(ord( $this->_stringParser->data[$this->_stringParser->offset] )) ) {
 188                         throw new Zend_Pdf_Exception(sprintf('PDF file cross-reference table syntax error. Offset - 0x%X. Value separator must be white space.', $this->_stringParser->offset));
 189                     }
 190                     $this->_stringParser->offset++;
 191
 192                     $inUseKey = $this->_stringParser->data[$this->_stringParser->offset];
 193                     $this->_stringParser->offset++;
 194
 195                     switch ($inUseKey) {
 196                         case 'f':
 197                             // free entry
 198                             unset( $this->_refTable[$objNum . ' ' . $genNumber . ' R'] );
 199                             $refTable->addReference($objNum . ' ' . $genNumber . ' R',
 200                                                     $objectOffset,
 201                                                     false);
 202                             break;
 203
 204                         case 'n':
 205                             // in-use entry
 206
 207                             $refTable->addReference($objNum . ' ' . $genNumber . ' R',
 208                                                     $objectOffset,
 209                                                     true);
 210                     }
 211
 212                     if ( !Zend_Pdf_StringParser::isWhiteSpace(ord( $this->_stringParser->data[$this->_stringParser->offset] )) ) {
 213                         throw new Zend_Pdf_Exception(sprintf('PDF file cross-reference table syntax error. Offset - 0x%X. Value separator must be white space.', $this->_stringParser->offset));
 214                     }
 215                     $this->_stringParser->offset++;
 216                     if ( !Zend_Pdf_StringParser::isWhiteSpace(ord( $this->_stringParser->data[$this->_stringParser->offset] )) ) {
 217                         throw new Zend_Pdf_Exception(sprintf('PDF file cross-reference table syntax error. Offset - 0x%X. Value separator must be white space.', $this->_stringParser->offset));
 218                     }
 219                     $this->_stringParser->offset++;
 220
 221                     $refCount--;
 222                     $objNum++;
 223                 }
 224             }
 225
 226             $trailerDictOffset = $this->_stringParser->offset;
 227             $trailerDict = $this->_stringParser->readElement();
 228             if (!$trailerDict instanceof Zend_Pdf_Element_Dictionary) {
 229                 throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X.  Dictionary expected after \'trailer\' keyword.', $trailerDictOffset));
 230             }
 231         } else {
 232             $xrefStream = $this->_stringParser->getObject($offset, $context);
 233
 234             if (!$xrefStream instanceof Zend_Pdf_Element_Object_Stream) {
 235                 throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X.  Cross-reference stream expected.', $offset));
 236             }
 237
 238             $trailerDict = $xrefStream->dictionary;
 239             if ($trailerDict->Type->value != 'XRef') {
 240                 throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X.  Cross-reference stream object must have /Type property assigned to /XRef.', $offset));
 241             }
 242             if ($trailerDict->W === null  || $trailerDict->W->getType() != Zend_Pdf_Element::TYPE_ARRAY) {
 243                 throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. Cross reference stream dictionary doesn\'t have W entry or it\'s not an array.', $offset));
 244             }
 245
 246             $entryField1Size = $trailerDict->W->items[0]->value;
 247             $entryField2Size = $trailerDict->W->items[1]->value;
 248             $entryField3Size = $trailerDict->W->items[2]->value;
 249
 250             if ($entryField2Size == 0 || $entryField3Size == 0) {
 251                 throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. Wrong W dictionary entry. Only type field of stream entries has default value and could be zero length.', $offset));
 252             }
 253
 254             $xrefStreamData = &$xrefStream->value;
 255
 256             if ($trailerDict->Index !== null) {
 257                 if ($trailerDict->Index->getType() != Zend_Pdf_Element::TYPE_ARRAY) {
 258                     throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. Cross reference stream dictionary Index entry must be an array.', $offset));
 259                 }
 260                 $sections = count($trailerDict->Index->items)/2;
 261             } else {
 262                 $sections = 1;
 263             }
 264
 265             $streamOffset = 0;
 266
 267             $size    = $entryField1Size + $entryField2Size + $entryField3Size;
 268             $entries = strlen($xrefStreamData)/$size;
 269
 270             for ($count = 0; $count < $sections; $count++) {
 271                 if ($trailerDict->Index !== null) {
 272                     $objNum  = $trailerDict->Index->items[$count*2    ]->value;
 273                     $entries = $trailerDict->Index->items[$count*2 + 1]->value;
 274                 } else {
 275                     $objNum  = 0;
 276                     $entries = $trailerDict->Size->value;
 277                 }
 278
 279                 for ($count2 = 0; $count2 < $entries; $count2++) {
 280                     if ($entryField1Size == 0) {
 281                         $type = 1;
 282                     } else if ($entryField1Size == 1) { // Optimyze one-byte field case
 283                         $type = ord($xrefStreamData[$streamOffset++]);
 284                     } else {
 285                         $type = Zend_Pdf_StringParser::parseIntFromStream($xrefStreamData, $streamOffset, $entryField1Size);
 286                         $streamOffset += $entryField1Size;
 287                     }
 288
 289                     if ($entryField2Size == 1) { // Optimyze one-byte field case
 290                         $field2 = ord($xrefStreamData[$streamOffset++]);
 291                     } else {
 292                         $field2 = Zend_Pdf_StringParser::parseIntFromStream($xrefStreamData, $streamOffset, $entryField2Size);
 293                         $streamOffset += $entryField2Size;
 294                     }
 295
 296                     if ($entryField3Size == 1) { // Optimyze one-byte field case
 297                         $field3 = ord($xrefStreamData[$streamOffset++]);
 298                     } else {
 299                         $field3 = Zend_Pdf_StringParser::parseIntFromStream($xrefStreamData, $streamOffset, $entryField3Size);
 300                         $streamOffset += $entryField3Size;
 301                     }
 302
 303                     switch ($type) {
 304                         case 0:
 305                             // Free object
 306                             $refTable->addReference($objNum . ' ' . $field3 . ' R', $field2, false);
 307                             // Debug output:
 308                             // echo "Free object - $objNum $field3 R, next free - $field2\n";
 309                             break;
 310
 311                         case 1:
 312                             // In use object
 313                             $refTable->addReference($objNum . ' ' . $field3 . ' R', $field2, true);
 314                             // Debug output:
 315                             // echo "In-use object - $objNum $field3 R, offset - $field2\n";
 316                             break;
 317
 318                         case 2:
 319                             // Object in an object stream
 320                             // Debug output:
 321                             // echo "Compressed object - $objNum 0 R, object stream - $field2 0 R, offset - $field3\n";
 322                             break;
 323                     }
 324
 325                     $objNum++;
 326                 }
 327             }
 328
 329             // $streamOffset . ' ' . strlen($xrefStreamData) . "\n";
 330             // "$entries\n";
 331             throw new Zend_Pdf_Exception('Cross-reference streams are not supported yet.');
 332         }
 333
 334
 335         $trailerObj = new Zend_Pdf_Trailer_Keeper($trailerDict, $context);
 336         if ($trailerDict->Prev instanceof Zend_Pdf_Element_Numeric ||
 337             $trailerDict->Prev instanceof Zend_Pdf_Element_Reference ) {
 338             $trailerObj->setPrev($this->_loadXRefTable($trailerDict->Prev->value));
 339             $context->getRefTable()->setParent($trailerObj->getPrev()->getRefTable());
 340         }
 341
 342         /**
 343          * We set '/Prev' dictionary property to the current cross-reference section offset.
 344          * It doesn't correspond to the actual data, but is true when trailer will be used
 345          * as a trailer for next generated PDF section.
 346          */
 347         $trailerObj->Prev = new Zend_Pdf_Element_Numeric($offset);
 348
 349         return $trailerObj;
 350     }
 351
 352
 353     /**
 354      * Get Trailer object
 355      *
 356      * @return Zend_Pdf_Trailer_Keeper
 357      */
 358     public function getTrailer()
 359     {
 360         return $this->_trailer;
 361     }
 362
 363     /**
 364      * Object constructor
 365      *
 366      * Note: PHP duplicates string, which is sent by value, only of it's updated.
 367      * Thus we don't need to care about overhead
 368      *
 369      * @param mixed $source
 370      * @param Zend_Pdf_ElementFactory_Interface $factory
 371      * @param boolean $load
 372      * @throws Zend_Exception
 373      */
 374     public function __construct($source, Zend_Pdf_ElementFactory_Interface $factory, $load)
 375     {
 376         if ($load) {
 377             if (($pdfFile = @fopen($source, 'rb')) === false ) {
 378                 throw new Zend_Pdf_Exception( "Can not open '$source' file for reading." );
 379             }
 380
 381             $byteCount = filesize($source);
 382
 383             $data = fread($pdfFile, $byteCount);
 384             $byteCount -= strlen($data);
 385             while ( $byteCount > 0 && ($nextBlock = fread($pdfFile, $byteCount)) != false ) {
 386                 $data .= $nextBlock;
 387                 $byteCount -= strlen($nextBlock);
 388             }
 389             fclose($pdfFile);
 390
 391             $this->_stringParser = new Zend_Pdf_StringParser($data, $factory);
 392         } else {
 393             $this->_stringParser = new Zend_Pdf_StringParser($source, $factory);
 394         }
 395
 396         $pdfVersionComment = $this->_stringParser->readComment();
 397         if (substr($pdfVersionComment, 0, 5) != '%PDF-') {
 398             throw new Zend_Pdf_Exception('File is not a PDF.');
 399         }
 400
 401         $pdfVersion = (float)substr($pdfVersionComment, 5);
 402         if ($pdfVersion < 0.9 || $pdfVersion >= 1.61) {
 403             /**
 404              * @todo
 405              * To support PDF versions 1.5 (Acrobat 6) and PDF version 1.7 (Acrobat 7)
 406              * Stream compression filter must be implemented (for compressed object streams).
 407              * Cross reference streams must be implemented
 408              */
 409             throw new Zend_Pdf_Exception(sprintf('Unsupported PDF version. Zend_Pdf supports PDF 1.0-1.4. Current version - \'%f\'', $pdfVersion));
 410         }
 411
 412         $this->_stringParser->offset = strrpos($this->_stringParser->data, '%%EOF');
 413         if ($this->_stringParser->offset === false ||
 414             strlen($this->_stringParser->data) - $this->_stringParser->offset > 7) {
 415             throw new Zend_Pdf_Exception('Pdf file syntax error. End-of-fle marker expected at the end of file.');
 416         }
 417
 418         $this->_stringParser->offset--;
 419         /**
 420          * Go to end of cross-reference table offset
 421          */
 422         while (Zend_Pdf_StringParser::isWhiteSpace( ord($this->_stringParser->data[$this->_stringParser->offset]) )&&
 423                ($this->_stringParser->offset > 0)) {
 424             $this->_stringParser->offset--;
 425         }
 426         /**
 427          * Go to the start of cross-reference table offset
 428          */
 429         while ( (!Zend_Pdf_StringParser::isWhiteSpace( ord($this->_stringParser->data[$this->_stringParser->offset]) ))&&
 430                ($this->_stringParser->offset > 0)) {
 431             $this->_stringParser->offset--;
 432         }
 433         /**
 434          * Go to the end of 'startxref' keyword
 435          */
 436         while (Zend_Pdf_StringParser::isWhiteSpace( ord($this->_stringParser->data[$this->_stringParser->offset]) )&&
 437                ($this->_stringParser->offset > 0)) {
 438             $this->_stringParser->offset--;
 439         }
 440         /**
 441          * Go to the white space (eol marker) before 'startxref' keyword
 442          */
 443         $this->_stringParser->offset -= 9;
 444
 445         $nextLexeme = $this->_stringParser->readLexeme();
 446         if ($nextLexeme != 'startxref') {
 447             throw new Zend_Pdf_Exception(sprintf('Pdf file syntax error. \'startxref\' keyword expected. Offset - 0x%X.', $this->_stringParser->offset-strlen($nextLexeme)));
 448         }
 449
 450         $startXref = $this->_stringParser->readLexeme();
 451         if (!ctype_digit($startXref)) {
 452             throw new Zend_Pdf_Exception(sprintf('Pdf file syntax error. Cross-reference table offset must contain only digits. Offset - 0x%X.', $this->_stringParser->offset-strlen($nextLexeme)));
 453         }
 454
 455         $this->_trailer = $this->_loadXRefTable($startXref);
 456         $factory->setObjectCount($this->_trailer->Size->value);
 457     }
 458
 459
 460     /**
 461      * Object destructor
 462      */
 463     public function __destruct()
 464     {
 465         $this->_stringParser->cleanUp();
 466     }
 467 }