Parser.php 19 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483
  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 $Id$
  20. */
  21. /** Zend_Pdf_Element */
  22. require_once 'Zend/Pdf/Element.php';
  23. /** Zend_Pdf_Element_Array */
  24. require_once 'Zend/Pdf/Element/Array.php';
  25. /** Zend_Pdf_Element_String_Binary */
  26. require_once 'Zend/Pdf/Element/String/Binary.php';
  27. /** Zend_Pdf_Element_Boolean */
  28. require_once 'Zend/Pdf/Element/Boolean.php';
  29. /** Zend_Pdf_Element_Dictionary */
  30. require_once 'Zend/Pdf/Element/Dictionary.php';
  31. /** Zend_Pdf_Element_Name */
  32. require_once 'Zend/Pdf/Element/Name.php';
  33. /** Zend_Pdf_Element_Numeric */
  34. require_once 'Zend/Pdf/Element/Numeric.php';
  35. /** Zend_Pdf_Element_Object */
  36. require_once 'Zend/Pdf/Element/Object.php';
  37. /** Zend_Pdf_Element_Reference */
  38. require_once 'Zend/Pdf/Element/Reference.php';
  39. /** Zend_Pdf_Element_Object_Stream */
  40. require_once 'Zend/Pdf/Element/Object/Stream.php';
  41. /** Zend_Pdf_Element_String */
  42. require_once 'Zend/Pdf/Element/String.php';
  43. /** Zend_Pdf_Element_Null */
  44. require_once 'Zend/Pdf/Element/Null.php';
  45. /** Zend_Pdf_Element_Reference_Context */
  46. require_once 'Zend/Pdf/Element/Reference/Context.php';
  47. /** Zend_Pdf_Element_Reference_Table */
  48. require_once 'Zend/Pdf/Element/Reference/Table.php';
  49. /** Zend_Pdf_Trailer_Keeper */
  50. require_once 'Zend/Pdf/Trailer/Keeper.php';
  51. /** Zend_Pdf_ElementFactory_Interface */
  52. require_once 'Zend/Pdf/ElementFactory/Interface.php';
  53. /** Zend_Pdf_StringParser */
  54. require_once 'Zend/Pdf/StringParser.php';
  55. /**
  56. * PDF file parser
  57. *
  58. * @package Zend_Pdf
  59. * @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
  60. * @license http://framework.zend.com/license/new-bsd New BSD License
  61. */
  62. class Zend_Pdf_Parser
  63. {
  64. /**
  65. * String parser
  66. *
  67. * @var Zend_Pdf_StringParser
  68. */
  69. private $_stringParser;
  70. /**
  71. * Last PDF file trailer
  72. *
  73. * @var Zend_Pdf_Trailer_Keeper
  74. */
  75. private $_trailer;
  76. /**
  77. * PDF version specified in the file header
  78. *
  79. * @var string
  80. */
  81. private $_pdfVersion;
  82. /**
  83. * Get length of source PDF
  84. *
  85. * @return integer
  86. */
  87. public function getPDFLength()
  88. {
  89. return strlen($this->_stringParser->data);
  90. }
  91. /**
  92. * Get PDF String
  93. *
  94. * @return string
  95. */
  96. public function getPDFString()
  97. {
  98. return $this->_stringParser->data;
  99. }
  100. /**
  101. * PDF version specified in the file header
  102. *
  103. * @return string
  104. */
  105. public function getPDFVersion()
  106. {
  107. return $this->_pdfVersion;
  108. }
  109. /**
  110. * Load XReference table and referenced objects
  111. *
  112. * @param integer $offset
  113. * @throws Zend_Pdf_Exception
  114. * @return Zend_Pdf_Trailer_Keeper
  115. */
  116. private function _loadXRefTable($offset)
  117. {
  118. $this->_stringParser->offset = $offset;
  119. $refTable = new Zend_Pdf_Element_Reference_Table();
  120. $context = new Zend_Pdf_Element_Reference_Context($this->_stringParser, $refTable);
  121. $this->_stringParser->setContext($context);
  122. $nextLexeme = $this->_stringParser->readLexeme();
  123. if ($nextLexeme == 'xref') {
  124. /**
  125. * Common cross-reference table
  126. */
  127. $this->_stringParser->skipWhiteSpace();
  128. while ( ($nextLexeme = $this->_stringParser->readLexeme()) != 'trailer' ) {
  129. if (!ctype_digit($nextLexeme)) {
  130. 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)));
  131. }
  132. $objNum = (int)$nextLexeme;
  133. $refCount = $this->_stringParser->readLexeme();
  134. if (!ctype_digit($refCount)) {
  135. 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)));
  136. }
  137. $this->_stringParser->skipWhiteSpace();
  138. while ($refCount > 0) {
  139. $objectOffset = substr($this->_stringParser->data, $this->_stringParser->offset, 10);
  140. if (!ctype_digit($objectOffset)) {
  141. throw new Zend_Pdf_Exception(sprintf('PDF file cross-reference table syntax error. Offset - 0x%X. Offset must contain only digits.', $this->_stringParser->offset));
  142. }
  143. // Force $objectOffset to be treated as decimal instead of octal number
  144. for ($numStart = 0; $numStart < strlen($objectOffset)-1; $numStart++) {
  145. if ($objectOffset[$numStart] != '0') {
  146. break;
  147. }
  148. }
  149. $objectOffset = substr($objectOffset, $numStart);
  150. $this->_stringParser->offset += 10;
  151. if ( !Zend_Pdf_StringParser::isWhiteSpace(ord( $this->_stringParser->data[$this->_stringParser->offset] )) ) {
  152. 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));
  153. }
  154. $this->_stringParser->offset++;
  155. $genNumber = substr($this->_stringParser->data, $this->_stringParser->offset, 5);
  156. if (!ctype_digit($objectOffset)) {
  157. throw new Zend_Pdf_Exception(sprintf('PDF file cross-reference table syntax error. Offset - 0x%X. Offset must contain only digits.', $this->_stringParser->offset));
  158. }
  159. // Force $objectOffset to be treated as decimal instead of octal number
  160. for ($numStart = 0; $numStart < strlen($genNumber)-1; $numStart++) {
  161. if ($genNumber[$numStart] != '0') {
  162. break;
  163. }
  164. }
  165. $genNumber = substr($genNumber, $numStart);
  166. $this->_stringParser->offset += 5;
  167. if ( !Zend_Pdf_StringParser::isWhiteSpace(ord( $this->_stringParser->data[$this->_stringParser->offset] )) ) {
  168. 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));
  169. }
  170. $this->_stringParser->offset++;
  171. $inUseKey = $this->_stringParser->data[$this->_stringParser->offset];
  172. $this->_stringParser->offset++;
  173. switch ($inUseKey) {
  174. case 'f':
  175. // free entry
  176. unset( $this->_refTable[$objNum . ' ' . $genNumber . ' R'] );
  177. $refTable->addReference($objNum . ' ' . $genNumber . ' R',
  178. $objectOffset,
  179. false);
  180. break;
  181. case 'n':
  182. // in-use entry
  183. $refTable->addReference($objNum . ' ' . $genNumber . ' R',
  184. $objectOffset,
  185. true);
  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. if ( !Zend_Pdf_StringParser::isWhiteSpace(ord( $this->_stringParser->data[$this->_stringParser->offset] )) ) {
  192. 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));
  193. }
  194. $this->_stringParser->offset++;
  195. $refCount--;
  196. $objNum++;
  197. }
  198. }
  199. $trailerDictOffset = $this->_stringParser->offset;
  200. $trailerDict = $this->_stringParser->readElement();
  201. if (!$trailerDict instanceof Zend_Pdf_Element_Dictionary) {
  202. throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. Dictionary expected after \'trailer\' keyword.', $trailerDictOffset));
  203. }
  204. } else {
  205. $xrefStream = $this->_stringParser->getObject($offset, $context);
  206. if (!$xrefStream instanceof Zend_Pdf_Element_Object_Stream) {
  207. throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. Cross-reference stream expected.', $offset));
  208. }
  209. $trailerDict = $xrefStream->dictionary;
  210. if ($trailerDict->Type->value != 'XRef') {
  211. throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. Cross-reference stream object must have /Type property assigned to /XRef.', $offset));
  212. }
  213. if ($trailerDict->W === null || $trailerDict->W->getType() != Zend_Pdf_Element::TYPE_ARRAY) {
  214. 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));
  215. }
  216. $entryField1Size = $trailerDict->W->items[0]->value;
  217. $entryField2Size = $trailerDict->W->items[1]->value;
  218. $entryField3Size = $trailerDict->W->items[2]->value;
  219. if ($entryField2Size == 0 || $entryField3Size == 0) {
  220. 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));
  221. }
  222. $xrefStreamData = &$xrefStream->value;
  223. if ($trailerDict->Index !== null) {
  224. if ($trailerDict->Index->getType() != Zend_Pdf_Element::TYPE_ARRAY) {
  225. throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. Cross reference stream dictionary Index entry must be an array.', $offset));
  226. }
  227. $sections = count($trailerDict->Index->items)/2;
  228. } else {
  229. $sections = 1;
  230. }
  231. $streamOffset = 0;
  232. $size = $entryField1Size + $entryField2Size + $entryField3Size;
  233. $entries = strlen($xrefStreamData)/$size;
  234. for ($count = 0; $count < $sections; $count++) {
  235. if ($trailerDict->Index !== null) {
  236. $objNum = $trailerDict->Index->items[$count*2 ]->value;
  237. $entries = $trailerDict->Index->items[$count*2 + 1]->value;
  238. } else {
  239. $objNum = 0;
  240. $entries = $trailerDict->Size->value;
  241. }
  242. for ($count2 = 0; $count2 < $entries; $count2++) {
  243. if ($entryField1Size == 0) {
  244. $type = 1;
  245. } else if ($entryField1Size == 1) { // Optimyze one-byte field case
  246. $type = ord($xrefStreamData[$streamOffset++]);
  247. } else {
  248. $type = Zend_Pdf_StringParser::parseIntFromStream($xrefStreamData, $streamOffset, $entryField1Size);
  249. $streamOffset += $entryField1Size;
  250. }
  251. if ($entryField2Size == 1) { // Optimyze one-byte field case
  252. $field2 = ord($xrefStreamData[$streamOffset++]);
  253. } else {
  254. $field2 = Zend_Pdf_StringParser::parseIntFromStream($xrefStreamData, $streamOffset, $entryField2Size);
  255. $streamOffset += $entryField2Size;
  256. }
  257. if ($entryField3Size == 1) { // Optimyze one-byte field case
  258. $field3 = ord($xrefStreamData[$streamOffset++]);
  259. } else {
  260. $field3 = Zend_Pdf_StringParser::parseIntFromStream($xrefStreamData, $streamOffset, $entryField3Size);
  261. $streamOffset += $entryField3Size;
  262. }
  263. switch ($type) {
  264. case 0:
  265. // Free object
  266. $refTable->addReference($objNum . ' ' . $field3 . ' R', $field2, false);
  267. // Debug output:
  268. // echo "Free object - $objNum $field3 R, next free - $field2\n";
  269. break;
  270. case 1:
  271. // In use object
  272. $refTable->addReference($objNum . ' ' . $field3 . ' R', $field2, true);
  273. // Debug output:
  274. // echo "In-use object - $objNum $field3 R, offset - $field2\n";
  275. break;
  276. case 2:
  277. // Object in an object stream
  278. // Debug output:
  279. // echo "Compressed object - $objNum 0 R, object stream - $field2 0 R, offset - $field3\n";
  280. break;
  281. }
  282. $objNum++;
  283. }
  284. }
  285. // $streamOffset . ' ' . strlen($xrefStreamData) . "\n";
  286. // "$entries\n";
  287. throw new Zend_Pdf_Exception('Cross-reference streams are not supported yet.');
  288. }
  289. $trailerObj = new Zend_Pdf_Trailer_Keeper($trailerDict, $context);
  290. if ($trailerDict->Prev instanceof Zend_Pdf_Element_Numeric ||
  291. $trailerDict->Prev instanceof Zend_Pdf_Element_Reference ) {
  292. $trailerObj->setPrev($this->_loadXRefTable($trailerDict->Prev->value));
  293. $context->getRefTable()->setParent($trailerObj->getPrev()->getRefTable());
  294. }
  295. /**
  296. * We set '/Prev' dictionary property to the current cross-reference section offset.
  297. * It doesn't correspond to the actual data, but is true when trailer will be used
  298. * as a trailer for next generated PDF section.
  299. */
  300. $trailerObj->Prev = new Zend_Pdf_Element_Numeric($offset);
  301. return $trailerObj;
  302. }
  303. /**
  304. * Get Trailer object
  305. *
  306. * @return Zend_Pdf_Trailer_Keeper
  307. */
  308. public function getTrailer()
  309. {
  310. return $this->_trailer;
  311. }
  312. /**
  313. * Object constructor
  314. *
  315. * Note: PHP duplicates string, which is sent by value, only of it's updated.
  316. * Thus we don't need to care about overhead
  317. *
  318. * @param mixed $source
  319. * @param Zend_Pdf_ElementFactory_Interface $factory
  320. * @param boolean $load
  321. * @throws Zend_Exception
  322. */
  323. public function __construct($source, Zend_Pdf_ElementFactory_Interface $factory, $load)
  324. {
  325. if ($load) {
  326. if (($pdfFile = @fopen($source, 'rb')) === false ) {
  327. throw new Zend_Pdf_Exception( "Can not open '$source' file for reading." );
  328. }
  329. $byteCount = filesize($source);
  330. $data = fread($pdfFile, $byteCount);
  331. $byteCount -= strlen($data);
  332. while ( $byteCount > 0 && ($nextBlock = fread($pdfFile, $byteCount)) != false ) {
  333. $data .= $nextBlock;
  334. $byteCount -= strlen($nextBlock);
  335. }
  336. fclose($pdfFile);
  337. $this->_stringParser = new Zend_Pdf_StringParser($data, $factory);
  338. } else {
  339. $this->_stringParser = new Zend_Pdf_StringParser($source, $factory);
  340. }
  341. $pdfVersionComment = $this->_stringParser->readComment();
  342. if (substr($pdfVersionComment, 0, 5) != '%PDF-') {
  343. throw new Zend_Pdf_Exception('File is not a PDF.');
  344. }
  345. $pdfVersion = substr($pdfVersionComment, 5);
  346. if (version_compare($pdfVersion, '0.9', '<') ||
  347. version_compare($pdfVersion, '1.61', '>=')
  348. ) {
  349. /**
  350. * @todo
  351. * To support PDF versions 1.5 (Acrobat 6) and PDF version 1.7 (Acrobat 7)
  352. * Stream compression filter must be implemented (for compressed object streams).
  353. * Cross reference streams must be implemented
  354. */
  355. throw new Zend_Pdf_Exception(sprintf('Unsupported PDF version. Zend_Pdf supports PDF 1.0-1.4. Current version - \'%f\'', $pdfVersion));
  356. }
  357. $this->_pdfVersion = $pdfVersion;
  358. $this->_stringParser->offset = strrpos($this->_stringParser->data, '%%EOF');
  359. if ($this->_stringParser->offset === false ||
  360. strlen($this->_stringParser->data) - $this->_stringParser->offset > 7) {
  361. throw new Zend_Pdf_Exception('Pdf file syntax error. End-of-fle marker expected at the end of file.');
  362. }
  363. $this->_stringParser->offset--;
  364. /**
  365. * Go to end of cross-reference table offset
  366. */
  367. while (Zend_Pdf_StringParser::isWhiteSpace( ord($this->_stringParser->data[$this->_stringParser->offset]) )&&
  368. ($this->_stringParser->offset > 0)) {
  369. $this->_stringParser->offset--;
  370. }
  371. /**
  372. * Go to the start of cross-reference table offset
  373. */
  374. while ( (!Zend_Pdf_StringParser::isWhiteSpace( ord($this->_stringParser->data[$this->_stringParser->offset]) ))&&
  375. ($this->_stringParser->offset > 0)) {
  376. $this->_stringParser->offset--;
  377. }
  378. /**
  379. * Go to the end of 'startxref' keyword
  380. */
  381. while (Zend_Pdf_StringParser::isWhiteSpace( ord($this->_stringParser->data[$this->_stringParser->offset]) )&&
  382. ($this->_stringParser->offset > 0)) {
  383. $this->_stringParser->offset--;
  384. }
  385. /**
  386. * Go to the white space (eol marker) before 'startxref' keyword
  387. */
  388. $this->_stringParser->offset -= 9;
  389. $nextLexeme = $this->_stringParser->readLexeme();
  390. if ($nextLexeme != 'startxref') {
  391. throw new Zend_Pdf_Exception(sprintf('Pdf file syntax error. \'startxref\' keyword expected. Offset - 0x%X.', $this->_stringParser->offset-strlen($nextLexeme)));
  392. }
  393. $startXref = $this->_stringParser->readLexeme();
  394. if (!ctype_digit($startXref)) {
  395. 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)));
  396. }
  397. $this->_trailer = $this->_loadXRefTable($startXref);
  398. $factory->setObjectCount($this->_trailer->Size->value);
  399. }
  400. /**
  401. * Object destructor
  402. */
  403. public function __destruct()
  404. {
  405. $this->_stringParser->cleanUp();
  406. }
  407. }