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 22311 2010-05-27 12:57:37Z padraic $

    20  */

    21 

    22 

    23 /** Internally used classes */

    24 require_once 'Zend/Pdf/Element/Array.php';

    25 require_once 'Zend/Pdf/Element/String/Binary.php';

    26 require_once 'Zend/Pdf/Element/Boolean.php';

    27 require_once 'Zend/Pdf/Element/Dictionary.php';

    28 require_once 'Zend/Pdf/Element/Name.php';

    29 require_once 'Zend/Pdf/Element/Null.php';

    30 require_once 'Zend/Pdf/Element/Numeric.php';

    31 require_once 'Zend/Pdf/Element/Object.php';

    32 require_once 'Zend/Pdf/Element/Object/Stream.php';

    33 require_once 'Zend/Pdf/Element/Reference.php';

    34 require_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  */

    44 class 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             $compare = '';

   268             if( version_compare( phpversion(), '5.2.5' ) >= 0) {

   269                 $compare = "()<>[]{}/%\x00\t\n\f\r ";

   270             } else {

   271                 $compare = "()<>[]{}/%\x00\t\n\r ";

   272             }

   273 

   274             $this->offset += strcspn($this->data, $compare, $this->offset);

   275  

   276             return substr($this->data, $start, $this->offset - $start);

   277         }

   278     }

   279 

   280 

   281     /**

   282      * Read elemental object from a PDF stream

   283      *

   284      * @return Zend_Pdf_Element

   285      * @throws Zend_Pdf_Exception

   286      */

   287     public function readElement($nextLexeme = null)

   288     {

   289         if ($nextLexeme === null) {

   290             $nextLexeme = $this->readLexeme();

   291         }

   292 

   293         /**

   294          * Note: readElement() method is a public method and could be invoked from other classes.

   295          * If readElement() is used not by Zend_Pdf_StringParser::getObject() method, then we should not care

   296          * about _elements member management.

   297          */

   298         switch ($nextLexeme) {

   299             case '(':

   300                 return ($this->_elements[] = $this->_readString());

   301 

   302             case '<':

   303                 return ($this->_elements[] = $this->_readBinaryString());

   304 

   305             case '/':

   306                 return ($this->_elements[] = new Zend_Pdf_Element_Name(

   307                                                     Zend_Pdf_Element_Name::unescape( $this->readLexeme() )

   308                                                                       ));

   309 

   310             case '[':

   311                 return ($this->_elements[] = $this->_readArray());

   312 

   313             case '<<':

   314                 return ($this->_elements[] = $this->_readDictionary());

   315 

   316             case ')':

   317                 // fall through to next case

   318             case '>':

   319                 // fall through to next case

   320             case ']':

   321                 // fall through to next case

   322             case '>>':

   323                 // fall through to next case

   324             case '{':

   325                 // fall through to next case

   326             case '}':

   327                 require_once 'Zend/Pdf/Exception.php';

   328                 throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X.',

   329                                                 $this->offset));

   330 

   331             default:

   332                 if (strcasecmp($nextLexeme, 'true') == 0) {

   333                     return ($this->_elements[] = new Zend_Pdf_Element_Boolean(true));

   334                 } else if (strcasecmp($nextLexeme, 'false') == 0) {

   335                     return ($this->_elements[] = new Zend_Pdf_Element_Boolean(false));

   336                 } else if (strcasecmp($nextLexeme, 'null') == 0) {

   337                     return ($this->_elements[] = new Zend_Pdf_Element_Null());

   338                 }

   339 

   340                 $ref = $this->_readReference($nextLexeme);

   341                 if ($ref !== null) {

   342                     return ($this->_elements[] = $ref);

   343                 }

   344 

   345                 return ($this->_elements[] = $this->_readNumeric($nextLexeme));

   346         }

   347     }

   348 

   349 

   350     /**

   351      * Read string PDF object

   352      * Also reads trailing ')' from a pdf stream

   353      *

   354      * @return Zend_Pdf_Element_String

   355      * @throws Zend_Pdf_Exception

   356      */

   357     private function _readString()

   358     {

   359         $start = $this->offset;

   360         $openedBrackets = 1;

   361 

   362         $this->offset += strcspn($this->data, '()\\', $this->offset);

   363 

   364         while ($this->offset < strlen($this->data)) {

   365             switch (ord( $this->data[$this->offset] )) {

   366                 case 0x28: // '(' - opened bracket in the string, needs balanced pair.

   367                     $this->offset++;

   368                     $openedBrackets++;

   369                     break;

   370 

   371                 case 0x29: // ')' - pair to the opened bracket

   372                     $this->offset++;

   373                     $openedBrackets--;

   374                     break;

   375 

   376                 case 0x5C: // '\\' - escape sequence, skip next char from a check

   377                     $this->offset += 2;

   378             }

   379 

   380             if ($openedBrackets == 0) {

   381                 break; // end of string

   382             }

   383 

   384             $this->offset += strcspn($this->data, '()\\', $this->offset);

   385         }

   386         if ($openedBrackets != 0) {

   387             require_once 'Zend/Pdf/Exception.php';

   388             throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Unexpected end of file while string reading. Offset - 0x%X. \')\' expected.', $start));

   389         }

   390 

   391         return new Zend_Pdf_Element_String(Zend_Pdf_Element_String::unescape( substr($this->data,

   392                                                                                      $start,

   393                                                                                      $this->offset - $start - 1) ));

   394     }

   395 

   396 

   397     /**

   398      * Read binary string PDF object

   399      * Also reads trailing '>' from a pdf stream

   400      *

   401      * @return Zend_Pdf_Element_String_Binary

   402      * @throws Zend_Pdf_Exception

   403      */

   404     private function _readBinaryString()

   405     {

   406         $start = $this->offset;

   407 

   408         $this->offset += strspn($this->data, "\x00\t\n\f\r 0123456789abcdefABCDEF", $this->offset);

   409 

   410         if ($this->offset >= strlen($this->data) - 1) {

   411             require_once 'Zend/Pdf/Exception.php';

   412             throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Unexpected end of file while reading binary string. Offset - 0x%X. \'>\' expected.', $start));

   413         }

   414 

   415         if ($this->data[$this->offset++] != '>') {

   416             require_once 'Zend/Pdf/Exception.php';

   417             throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Unexpected character while binary string reading. Offset - 0x%X.', $this->offset));

   418         }

   419 

   420         return new Zend_Pdf_Element_String_Binary(

   421                        Zend_Pdf_Element_String_Binary::unescape( substr($this->data,

   422                                                                         $start,

   423                                                                         $this->offset - $start - 1) ));

   424     }

   425 

   426 

   427     /**

   428      * Read array PDF object

   429      * Also reads trailing ']' from a pdf stream

   430      *

   431      * @return Zend_Pdf_Element_Array

   432      * @throws Zend_Pdf_Exception

   433      */

   434     private function _readArray()

   435     {

   436         $elements = array();

   437 

   438         while ( strlen($nextLexeme = $this->readLexeme()) != 0 ) {

   439             if ($nextLexeme != ']') {

   440                 $elements[] = $this->readElement($nextLexeme);

   441             } else {

   442                 return new Zend_Pdf_Element_Array($elements);

   443             }

   444         }

   445 

   446         require_once 'Zend/Pdf/Exception.php';

   447         throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Unexpected end of file while array reading. Offset - 0x%X. \']\' expected.', $this->offset));

   448     }

   449 

   450 

   451     /**

   452      * Read dictionary PDF object

   453      * Also reads trailing '>>' from a pdf stream

   454      *

   455      * @return Zend_Pdf_Element_Dictionary

   456      * @throws Zend_Pdf_Exception

   457      */

   458     private function _readDictionary()

   459     {

   460         $dictionary = new Zend_Pdf_Element_Dictionary();

   461 

   462         while ( strlen($nextLexeme = $this->readLexeme()) != 0 ) {

   463             if ($nextLexeme != '>>') {

   464                 $nameStart = $this->offset - strlen($nextLexeme);

   465 

   466                 $name  = $this->readElement($nextLexeme);

   467                 $value = $this->readElement();

   468 

   469                 if (!$name instanceof Zend_Pdf_Element_Name) {

   470                     require_once 'Zend/Pdf/Exception.php';

   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         require_once 'Zend/Pdf/Exception.php';

   481         throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Unexpected end of file while dictionary reading. Offset - 0x%X. \'>>\' expected.', $this->offset));

   482     }

   483 

   484 

   485     /**

   486      * Read reference PDF object

   487      *

   488      * @param string $nextLexeme

   489      * @return Zend_Pdf_Element_Reference

   490      */

   491     private function _readReference($nextLexeme = null)

   492     {

   493         $start = $this->offset;

   494 

   495         if ($nextLexeme === null) {

   496             $objNum = $this->readLexeme();

   497         } else {

   498             $objNum = $nextLexeme;

   499         }

   500         if (!ctype_digit($objNum)) { // it's not a reference

   501             $this->offset = $start;

   502             return null;

   503         }

   504 

   505         $genNum = $this->readLexeme();

   506         if (!ctype_digit($genNum)) { // it's not a reference

   507             $this->offset = $start;

   508             return null;

   509         }

   510 

   511         $rMark  = $this->readLexeme();

   512         if ($rMark != 'R') { // it's not a reference

   513             $this->offset = $start;

   514             return null;

   515         }

   516 

   517         $ref = new Zend_Pdf_Element_Reference((int)$objNum, (int)$genNum, $this->_context, $this->_objFactory->resolve());

   518 

   519         return $ref;

   520     }

   521 

   522 

   523     /**

   524      * Read numeric PDF object

   525      *

   526      * @param string $nextLexeme

   527      * @return Zend_Pdf_Element_Numeric

   528      */

   529     private function _readNumeric($nextLexeme = null)

   530     {

   531         if ($nextLexeme === null) {

   532             $nextLexeme = $this->readLexeme();

   533         }

   534 

   535         return new Zend_Pdf_Element_Numeric($nextLexeme);

   536     }

   537 

   538 

   539     /**

   540      * Read inderect object from a PDF stream

   541      *

   542      * @param integer $offset

   543      * @param Zend_Pdf_Element_Reference_Context $context

   544      * @return Zend_Pdf_Element_Object

   545      */

   546     public function getObject($offset, Zend_Pdf_Element_Reference_Context $context)

   547     {

   548         if ($offset === null ) {

   549             return new Zend_Pdf_Element_Null();

   550         }

   551 

   552         // Save current offset to make getObject() reentrant

   553         $offsetSave = $this->offset;

   554 

   555         $this->offset    = $offset;

   556         $this->_context  = $context;

   557         $this->_elements = array();

   558 

   559         $objNum = $this->readLexeme();

   560         if (!ctype_digit($objNum)) {

   561             require_once 'Zend/Pdf/Exception.php';

   562             throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. Object number expected.', $this->offset - strlen($objNum)));

   563         }

   564 

   565         $genNum = $this->readLexeme();

   566         if (!ctype_digit($genNum)) {

   567             require_once 'Zend/Pdf/Exception.php';

   568             throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. Object generation number expected.', $this->offset - strlen($genNum)));

   569         }

   570 

   571         $objKeyword = $this->readLexeme();

   572         if ($objKeyword != 'obj') {

   573             require_once 'Zend/Pdf/Exception.php';

   574             throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. \'obj\' keyword expected.', $this->offset - strlen($objKeyword)));

   575         }

   576 

   577         $objValue = $this->readElement();

   578 

   579         $nextLexeme = $this->readLexeme();

   580 

   581         if( $nextLexeme == 'endobj' ) {

   582             /**

   583              * Object is not generated by factory (thus it's not marked as modified object).

   584              * But factory is assigned to the obect.

   585              */

   586             $obj = new Zend_Pdf_Element_Object($objValue, (int)$objNum, (int)$genNum, $this->_objFactory->resolve());

   587 

   588             foreach ($this->_elements as $element) {

   589                 $element->setParentObject($obj);

   590             }

   591 

   592             // Restore offset value

   593             $this->offset = $offsetSave;

   594 

   595             return $obj;

   596         }

   597 

   598         /**

   599          * It's a stream object

   600          */

   601         if ($nextLexeme != 'stream') {

   602             require_once 'Zend/Pdf/Exception.php';

   603             throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. \'endobj\' or \'stream\' keywords expected.', $this->offset - strlen($nextLexeme)));

   604         }

   605 

   606         if (!$objValue instanceof Zend_Pdf_Element_Dictionary) {

   607             require_once 'Zend/Pdf/Exception.php';

   608             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)));

   609         }

   610 

   611         /**

   612          * References are automatically dereferenced at this moment.

   613          */

   614         $streamLength = $objValue->Length->value;

   615 

   616         /**

   617          * 'stream' keyword must be followed by either cr-lf sequence or lf character only.

   618          * This restriction gives the possibility to recognize all cases exactly

   619          */

   620         if ($this->data[$this->offset] == "\r" &&

   621             $this->data[$this->offset + 1] == "\n"    ) {

   622             $this->offset += 2;

   623         } else if ($this->data[$this->offset] == "\n"    ) {

   624             $this->offset++;

   625         } else {

   626             require_once 'Zend/Pdf/Exception.php';

   627             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)));

   628         }

   629 

   630         $dataOffset = $this->offset;

   631 

   632         $this->offset += $streamLength;

   633 

   634         $nextLexeme = $this->readLexeme();

   635         if ($nextLexeme != 'endstream') {

   636             require_once 'Zend/Pdf/Exception.php';

   637             throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. \'endstream\' keyword expected.', $this->offset - strlen($nextLexeme)));

   638         }

   639 

   640         $nextLexeme = $this->readLexeme();

   641         if ($nextLexeme != 'endobj') {

   642             require_once 'Zend/Pdf/Exception.php';

   643             throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. \'endobj\' keyword expected.', $this->offset - strlen($nextLexeme)));

   644         }

   645 

   646         $obj = new Zend_Pdf_Element_Object_Stream(substr($this->data,

   647                                                          $dataOffset,

   648                                                          $streamLength),

   649                                                   (int)$objNum,

   650                                                   (int)$genNum,

   651                                                   $this->_objFactory->resolve(),

   652                                                   $objValue);

   653 

   654         foreach ($this->_elements as $element) {

   655             $element->setParentObject($obj);

   656         }

   657 

   658         // Restore offset value

   659         $this->offset = $offsetSave;

   660 

   661         return $obj;

   662     }

   663 

   664 

   665     /**

   666      * Get length of source string

   667      *

   668      * @return integer

   669      */

   670     public function getLength()

   671     {

   672         return strlen($this->data);

   673     }

   674 

   675     /**

   676      * Get source string

   677      *

   678      * @return string

   679      */

   680     public function getString()

   681     {

   682         return $this->data;

   683     }

   684 

   685 

   686     /**

   687      * Parse integer value from a binary stream

   688      *

   689      * @param string $stream

   690      * @param integer $offset

   691      * @param integer $size

   692      * @return integer

   693      */

   694     public static function parseIntFromStream($stream, $offset, $size)

   695     {

   696         $value = 0;

   697         for ($count = 0; $count < $size; $count++) {

   698             $value *= 256;

   699             $value += ord($stream[$offset + $count]);

   700         }

   701 

   702         return $value;

   703     }

   704 

   705 

   706 

   707     /**

   708      * Set current context

   709      *

   710      * @param Zend_Pdf_Element_Reference_Context $context

   711      */

   712     public function setContext(Zend_Pdf_Element_Reference_Context $context)

   713     {

   714         $this->_context = $context;

   715     }

   716 

   717     /**

   718      * Object constructor

   719      *

   720      * Note: PHP duplicates string, which is sent by value, only of it's updated.

   721      * Thus we don't need to care about overhead

   722      *

   723      * @param string $pdfString

   724      * @param Zend_Pdf_ElementFactory_Interface $factory

   725      */

   726     public function __construct($source, Zend_Pdf_ElementFactory_Interface $factory)

   727     {

   728         $this->data         = $source;

   729         $this->_objFactory  = $factory;

   730     }

   731 }