[Summary view]

Differences Between: [Versions 310 and 311] [Versions 310 and 400] [Versions 310 and 401] [Versions 310 and 402] [Versions 310 and 403]

   1  <?php
   2  /**
   3   * This file is part of FPDI
   4   *
   5   * @package   setasign\Fpdi
   6   * @copyright Copyright (c) 2019 Setasign - Jan Slabon (https://www.setasign.com)
   7   * @license   http://opensource.org/licenses/mit-license The MIT License
   8   */
   9  
  10  namespace setasign\Fpdi\PdfParser\CrossReference;
  11  
  12  use setasign\Fpdi\PdfParser\PdfParser;
  13  use setasign\Fpdi\PdfParser\Type\PdfDictionary;
  14  use setasign\Fpdi\PdfParser\Type\PdfIndirectObject;
  15  use setasign\Fpdi\PdfParser\Type\PdfNumeric;
  16  use setasign\Fpdi\PdfParser\Type\PdfStream;
  17  use setasign\Fpdi\PdfParser\Type\PdfToken;
  18  use setasign\Fpdi\PdfParser\Type\PdfTypeException;
  19  
  20  /**
  21   * Class CrossReference
  22   *
  23   * This class processes the standard cross reference of a PDF document.
  24   *
  25   * @package setasign\Fpdi\PdfParser\CrossReference
  26   */
  27  class CrossReference
  28  {
  29      /**
  30       * The byte length in which the "startxref" keyword should be searched.
  31       *
  32       * @var int
  33       */
  34      static public $trailerSearchLength = 5500;
  35  
  36      /**
  37       * @var int
  38       */
  39      protected $fileHeaderOffset = 0;
  40  
  41      /**
  42       * @var PdfParser
  43       */
  44      protected $parser;
  45  
  46      /**
  47       * @var ReaderInterface[]
  48       */
  49      protected $readers = [];
  50  
  51      /**
  52       * CrossReference constructor.
  53       *
  54       * @param PdfParser $parser
  55       * @throws CrossReferenceException
  56       * @throws PdfTypeException
  57       */
  58      public function __construct(PdfParser $parser, $fileHeaderOffset = 0)
  59      {
  60          $this->parser = $parser;
  61          $this->fileHeaderOffset = $fileHeaderOffset;
  62  
  63          $offset = $this->findStartXref();
  64          $reader = null;
  65          /** @noinspection TypeUnsafeComparisonInspection */
  66          while ($offset != false) { // By doing an unsafe comparsion we ignore faulty references to byte offset 0
  67              try {
  68                  $reader = $this->readXref($offset + $this->fileHeaderOffset);
  69              } catch (CrossReferenceException $e) {
  70                  // sometimes the file header offset is part of the byte offsets, so let's retry by resetting it to zero.
  71                  if ($e->getCode() === CrossReferenceException::INVALID_DATA && $this->fileHeaderOffset !== 0) {
  72                      $this->fileHeaderOffset = 0;
  73                      $reader = $this->readXref($offset + $this->fileHeaderOffset);
  74                  } else {
  75                      throw $e;
  76                  }
  77              }
  78  
  79              $trailer = $reader->getTrailer();
  80              $this->checkForEncryption($trailer);
  81              $this->readers[] = $reader;
  82  
  83              if (isset($trailer->value['Prev'])) {
  84                  $offset = $trailer->value['Prev']->value;
  85              } else {
  86                  $offset = false;
  87              }
  88          }
  89  
  90          // fix faulty sub-section header
  91          if ($reader instanceof FixedReader) {
  92              /**
  93               * @var FixedReader $reader
  94               */
  95              $reader->fixFaultySubSectionShift();
  96          }
  97  
  98          if ($reader === null) {
  99              throw new CrossReferenceException('No cross-reference found.', CrossReferenceException::NO_XREF_FOUND);
 100          }
 101      }
 102  
 103      /**
 104       * Get the size of the cross reference.
 105       *
 106       * @return integer
 107       */
 108      public function getSize()
 109      {
 110          return $this->getTrailer()->value['Size']->value;
 111      }
 112  
 113      /**
 114       * Get the trailer dictionary.
 115       *
 116       * @return PdfDictionary
 117       */
 118      public function getTrailer()
 119      {
 120          return $this->readers[0]->getTrailer();
 121      }
 122  
 123      /**
 124       * Get the cross reference readser instances.
 125       *
 126       * @return ReaderInterface[]
 127       */
 128      public function getReaders()
 129      {
 130          return $this->readers;
 131      }
 132  
 133      /**
 134       * Get the offset by an object number.
 135       *
 136       * @param int $objectNumber
 137       * @return integer|bool
 138       */
 139      public function getOffsetFor($objectNumber)
 140      {
 141          foreach ($this->getReaders() as $reader) {
 142              $offset = $reader->getOffsetFor($objectNumber);
 143              if ($offset !== false) {
 144                  return $offset;
 145              }
 146          }
 147  
 148          return false;
 149      }
 150  
 151      /**
 152       * Get an indirect object by its object number.
 153       *
 154       * @param int $objectNumber
 155       * @return PdfIndirectObject
 156       * @throws CrossReferenceException
 157       */
 158      public function getIndirectObject($objectNumber)
 159      {
 160          $offset = $this->getOffsetFor($objectNumber);
 161          if ($offset === false) {
 162              throw new CrossReferenceException(
 163                  \sprintf('Object (id:%s) not found.', $objectNumber),
 164                  CrossReferenceException::OBJECT_NOT_FOUND
 165              );
 166          }
 167  
 168          $parser = $this->parser;
 169  
 170          $parser->getTokenizer()->clearStack();
 171          $parser->getStreamReader()->reset($offset + $this->fileHeaderOffset);
 172  
 173          try {
 174              /** @var PdfIndirectObject $object */
 175              $object = $parser->readValue(null, PdfIndirectObject::class);
 176          } catch (PdfTypeException $e) {
 177              throw new CrossReferenceException(
 178                  \sprintf('Object (id:%s) not found at location (%s).', $objectNumber, $offset),
 179                  CrossReferenceException::OBJECT_NOT_FOUND,
 180                  $e
 181              );
 182          }
 183  
 184          if ($object->objectNumber !== $objectNumber) {
 185              throw new CrossReferenceException(
 186                  \sprintf('Wrong object found, got %s while %s was expected.', $object->objectNumber, $objectNumber),
 187                  CrossReferenceException::OBJECT_NOT_FOUND
 188              );
 189          }
 190  
 191          return $object;
 192      }
 193  
 194      /**
 195       * Read the cross-reference table at a given offset.
 196       *
 197       * Internally the method will try to evaluate the best reader for this cross-reference.
 198       *
 199       * @param int $offset
 200       * @return ReaderInterface
 201       * @throws CrossReferenceException
 202       * @throws PdfTypeException
 203       */
 204      protected function readXref($offset)
 205      {
 206          $this->parser->getStreamReader()->reset($offset);
 207          $this->parser->getTokenizer()->clearStack();
 208          $initValue = $this->parser->readValue();
 209          
 210          return $this->initReaderInstance($initValue);
 211      }
 212  
 213      /**
 214       * Get a cross-reference reader instance.
 215       *
 216       * @param PdfToken|PdfIndirectObject $initValue
 217       * @return ReaderInterface|bool
 218       * @throws CrossReferenceException
 219       * @throws PdfTypeException
 220       */
 221      protected function initReaderInstance($initValue)
 222      {
 223          $position = $this->parser->getStreamReader()->getPosition()
 224              + $this->parser->getStreamReader()->getOffset() + $this->fileHeaderOffset;
 225  
 226          if ($initValue instanceof PdfToken && $initValue->value === 'xref') {
 227              try {
 228                  return new FixedReader($this->parser);
 229              } catch (CrossReferenceException $e) {
 230                  $this->parser->getStreamReader()->reset($position);
 231                  $this->parser->getTokenizer()->clearStack();
 232  
 233                  return new LineReader($this->parser);
 234              }
 235          }
 236  
 237          if ($initValue instanceof PdfIndirectObject) {
 238              // check for encryption
 239              $stream = PdfStream::ensure($initValue->value);
 240  
 241              $type = PdfDictionary::get($stream->value, 'Type');
 242              if ($type->value !== 'XRef') {
 243                  throw new CrossReferenceException(
 244                      'The xref position points to an incorrect object type.',
 245                      CrossReferenceException::INVALID_DATA
 246                  );
 247              }
 248  
 249              $this->checkForEncryption($stream->value);
 250  
 251              throw new CrossReferenceException(
 252                  'This PDF document probably uses a compression technique which is not supported by the ' .
 253                  'free parser shipped with FPDI. (See https://www.setasign.com/fpdi-pdf-parser for more details)',
 254                  CrossReferenceException::COMPRESSED_XREF
 255              );
 256          }
 257  
 258          throw new CrossReferenceException(
 259              'The xref position points to an incorrect object type.',
 260              CrossReferenceException::INVALID_DATA
 261          );
 262      }
 263  
 264      /**
 265       * Check for encryption.
 266       *
 267       * @param PdfDictionary $dictionary
 268       * @throws CrossReferenceException
 269       */
 270      protected function checkForEncryption(PdfDictionary $dictionary)
 271      {
 272          if (isset($dictionary->value['Encrypt'])) {
 273              throw new CrossReferenceException(
 274                  'This PDF document is encrypted and cannot be processed with FPDI.',
 275                  CrossReferenceException::ENCRYPTED
 276              );
 277          }
 278      }
 279  
 280      /**
 281       * Find the start position for the first cross-reference.
 282       *
 283       * @return int The byte-offset position of the first cross-reference.
 284       * @throws CrossReferenceException
 285       */
 286      protected function findStartXref()
 287      {
 288          $reader = $this->parser->getStreamReader();
 289          $reader->reset(-self::$trailerSearchLength, self::$trailerSearchLength);
 290  
 291          $buffer = $reader->getBuffer(false);
 292          $pos = \strrpos($buffer, 'startxref');
 293          $addOffset = 9;
 294          if ($pos === false) {
 295              // Some corrupted documents uses startref, instead of startxref
 296              $pos = \strrpos($buffer, 'startref');
 297              if ($pos === false) {
 298                  throw new CrossReferenceException(
 299                      'Unable to find pointer to xref table',
 300                      CrossReferenceException::NO_STARTXREF_FOUND
 301                  );
 302              }
 303              $addOffset = 8;
 304          }
 305  
 306          $reader->setOffset($pos + $addOffset);
 307  
 308          try {
 309              $value = $this->parser->readValue(null, PdfNumeric::class);
 310          } catch (PdfTypeException $e) {
 311              throw new CrossReferenceException(
 312                  'Invalid data after startxref keyword.',
 313                  CrossReferenceException::INVALID_DATA,
 314                  $e
 315              );
 316          }
 317  
 318          return $value->value;
 319      }
 320  }