Parser.php 19 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487
  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) 2005-2008 Zend Technologies USA Inc. (http://www.zend.com)
  17. * @license http://framework.zend.com/license/new-bsd New BSD License
  18. */
  19. /** Zend_Pdf_Element */
  20. require_once 'Zend/Pdf/Element.php';
  21. /** Zend_Pdf_Element_Array */
  22. require_once 'Zend/Pdf/Element/Array.php';
  23. /** Zend_Pdf_Element_String_Binary */
  24. require_once 'Zend/Pdf/Element/String/Binary.php';
  25. /** Zend_Pdf_Element_Boolean */
  26. require_once 'Zend/Pdf/Element/Boolean.php';
  27. /** Zend_Pdf_Element_Dictionary */
  28. require_once 'Zend/Pdf/Element/Dictionary.php';
  29. /** Zend_Pdf_Element_Name */
  30. require_once 'Zend/Pdf/Element/Name.php';
  31. /** Zend_Pdf_Element_Numeric */
  32. require_once 'Zend/Pdf/Element/Numeric.php';
  33. /** Zend_Pdf_Element_Object */
  34. require_once 'Zend/Pdf/Element/Object.php';
  35. /** Zend_Pdf_Element_Reference */
  36. require_once 'Zend/Pdf/Element/Reference.php';
  37. /** Zend_Pdf_Element_Object_Stream */
  38. require_once 'Zend/Pdf/Element/Object/Stream.php';
  39. /** Zend_Pdf_Element_String */
  40. require_once 'Zend/Pdf/Element/String.php';
  41. /** Zend_Pdf_Element_Null */
  42. require_once 'Zend/Pdf/Element/Null.php';
  43. /** Zend_Pdf_Element_Reference_Context */
  44. require_once 'Zend/Pdf/Element/Reference/Context.php';
  45. /** Zend_Pdf_Element_Reference_Table */
  46. require_once 'Zend/Pdf/Element/Reference/Table.php';
  47. /** Zend_Pdf_Trailer_Keeper */
  48. require_once 'Zend/Pdf/Trailer/Keeper.php';
  49. /** Zend_Pdf_ElementFactory_Interface */
  50. require_once 'Zend/Pdf/ElementFactory/Interface.php';
  51. /** Zend_Pdf_PhpArray */
  52. require_once 'Zend/Pdf/PhpArray.php';
  53. /** Zend_Pdf_StringParser */
  54. require_once 'Zend/Pdf/StringParser.php';
  55. /** Zend_Pdf_Parser_Stream */
  56. require_once 'Zend/Pdf/Parser/Stream.php';
  57. /**
  58. * PDF file parser
  59. *
  60. * @package Zend_Pdf
  61. * @copyright Copyright (c) 2005-2008 Zend Technologies USA Inc. (http://www.zend.com)
  62. * @license http://framework.zend.com/license/new-bsd New BSD License
  63. */
  64. class Zend_Pdf_Parser
  65. {
  66. /**
  67. * String parser
  68. *
  69. * @var Zend_Pdf_StringParser
  70. */
  71. private $_stringParser;
  72. /**
  73. * Last PDF file trailer
  74. *
  75. * @var Zend_Pdf_Trailer_Keeper
  76. */
  77. private $_trailer;
  78. /**
  79. * PDF version specified in the file header
  80. *
  81. * @var string
  82. */
  83. private $_pdfVersion;
  84. /**
  85. * Get length of source PDF
  86. *
  87. * @return integer
  88. */
  89. public function getPDFLength()
  90. {
  91. return strlen($this->_stringParser->data);
  92. }
  93. /**
  94. * Get PDF String
  95. *
  96. * @return string
  97. */
  98. public function getPDFString()
  99. {
  100. return $this->_stringParser->data;
  101. }
  102. /**
  103. * PDF version specified in the file header
  104. *
  105. * @return string
  106. */
  107. public function getPDFVersion()
  108. {
  109. return $this->_pdfVersion;
  110. }
  111. /**
  112. * Load XReference table and referenced objects
  113. *
  114. * @param integer $offset
  115. * @throws Zend_Pdf_Exception
  116. * @return Zend_Pdf_Trailer_Keeper
  117. */
  118. private function _loadXRefTable($offset)
  119. {
  120. $this->_stringParser->offset = $offset;
  121. $refTable = new Zend_Pdf_Element_Reference_Table();
  122. $context = new Zend_Pdf_Element_Reference_Context($this->_stringParser, $refTable);
  123. $this->_stringParser->setContext($context);
  124. $nextLexeme = $this->_stringParser->readLexeme();
  125. if ($nextLexeme == 'xref') {
  126. /**
  127. * Common cross-reference table
  128. */
  129. $this->_stringParser->skipWhiteSpace();
  130. while ( ($nextLexeme = $this->_stringParser->readLexeme()) != 'trailer' ) {
  131. if (!ctype_digit($nextLexeme)) {
  132. 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)));
  133. }
  134. $objNum = (int)$nextLexeme;
  135. $refCount = $this->_stringParser->readLexeme();
  136. if (!ctype_digit($refCount)) {
  137. 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)));
  138. }
  139. $this->_stringParser->skipWhiteSpace();
  140. while ($refCount > 0) {
  141. $objectOffset = substr($this->_stringParser->data, $this->_stringParser->offset, 10);
  142. if (!ctype_digit($objectOffset)) {
  143. throw new Zend_Pdf_Exception(sprintf('PDF file cross-reference table syntax error. Offset - 0x%X. Offset must contain only digits.', $this->_stringParser->offset));
  144. }
  145. // Force $objectOffset to be treated as decimal instead of octal number
  146. for ($numStart = 0; $numStart < strlen($objectOffset)-1; $numStart++) {
  147. if ($objectOffset[$numStart] != '0') {
  148. break;
  149. }
  150. }
  151. $objectOffset = substr($objectOffset, $numStart);
  152. $this->_stringParser->offset += 10;
  153. if ( !Zend_Pdf_StringParser::isWhiteSpace(ord( $this->_stringParser->data[$this->_stringParser->offset] )) ) {
  154. 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));
  155. }
  156. $this->_stringParser->offset++;
  157. $genNumber = substr($this->_stringParser->data, $this->_stringParser->offset, 5);
  158. if (!ctype_digit($objectOffset)) {
  159. throw new Zend_Pdf_Exception(sprintf('PDF file cross-reference table syntax error. Offset - 0x%X. Offset must contain only digits.', $this->_stringParser->offset));
  160. }
  161. // Force $objectOffset to be treated as decimal instead of octal number
  162. for ($numStart = 0; $numStart < strlen($genNumber)-1; $numStart++) {
  163. if ($genNumber[$numStart] != '0') {
  164. break;
  165. }
  166. }
  167. $genNumber = substr($genNumber, $numStart);
  168. $this->_stringParser->offset += 5;
  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. $inUseKey = $this->_stringParser->data[$this->_stringParser->offset];
  174. $this->_stringParser->offset++;
  175. switch ($inUseKey) {
  176. case 'f':
  177. // free entry
  178. unset( $this->_refTable[$objNum . ' ' . $genNumber . ' R'] );
  179. $refTable->addReference($objNum . ' ' . $genNumber . ' R',
  180. $objectOffset,
  181. false);
  182. break;
  183. case 'n':
  184. // in-use entry
  185. $refTable->addReference($objNum . ' ' . $genNumber . ' R',
  186. $objectOffset,
  187. true);
  188. }
  189. if ( !Zend_Pdf_StringParser::isWhiteSpace(ord( $this->_stringParser->data[$this->_stringParser->offset] )) ) {
  190. 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));
  191. }
  192. $this->_stringParser->offset++;
  193. if ( !Zend_Pdf_StringParser::isWhiteSpace(ord( $this->_stringParser->data[$this->_stringParser->offset] )) ) {
  194. 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));
  195. }
  196. $this->_stringParser->offset++;
  197. $refCount--;
  198. $objNum++;
  199. }
  200. }
  201. $trailerDictOffset = $this->_stringParser->offset;
  202. $trailerDict = $this->_stringParser->readElement();
  203. if (!$trailerDict instanceof Zend_Pdf_Element_Dictionary) {
  204. throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. Dictionary expected after \'trailer\' keyword.', $trailerDictOffset));
  205. }
  206. } else {
  207. $xrefStream = $this->_stringParser->getObject($offset, $context);
  208. if (!$xrefStream instanceof Zend_Pdf_Element_Object_Stream) {
  209. throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. Cross-reference stream expected.', $offset));
  210. }
  211. $trailerDict = $xrefStream->dictionary;
  212. if ($trailerDict->Type->value != 'XRef') {
  213. throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. Cross-reference stream object must have /Type property assigned to /XRef.', $offset));
  214. }
  215. if ($trailerDict->W === null || $trailerDict->W->getType() != Zend_Pdf_Element::TYPE_ARRAY) {
  216. 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));
  217. }
  218. $entryField1Size = $trailerDict->W->items[0]->value;
  219. $entryField2Size = $trailerDict->W->items[1]->value;
  220. $entryField3Size = $trailerDict->W->items[2]->value;
  221. if ($entryField2Size == 0 || $entryField3Size == 0) {
  222. 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));
  223. }
  224. $xrefStreamData = &$xrefStream->value;
  225. if ($trailerDict->Index !== null) {
  226. if ($trailerDict->Index->getType() != Zend_Pdf_Element::TYPE_ARRAY) {
  227. throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. Cross reference stream dictionary Index entry must be an array.', $offset));
  228. }
  229. $sections = count($trailerDict->Index->items)/2;
  230. } else {
  231. $sections = 1;
  232. }
  233. $streamOffset = 0;
  234. $size = $entryField1Size + $entryField2Size + $entryField3Size;
  235. $entries = strlen($xrefStreamData)/$size;
  236. for ($count = 0; $count < $sections; $count++) {
  237. if ($trailerDict->Index !== null) {
  238. $objNum = $trailerDict->Index->items[$count*2 ]->value;
  239. $entries = $trailerDict->Index->items[$count*2 + 1]->value;
  240. } else {
  241. $objNum = 0;
  242. $entries = $trailerDict->Size->value;
  243. }
  244. for ($count2 = 0; $count2 < $entries; $count2++) {
  245. if ($entryField1Size == 0) {
  246. $type = 1;
  247. } else if ($entryField1Size == 1) { // Optimyze one-byte field case
  248. $type = ord($xrefStreamData[$streamOffset++]);
  249. } else {
  250. $type = Zend_Pdf_StringParser::parseIntFromStream($xrefStreamData, $streamOffset, $entryField1Size);
  251. $streamOffset += $entryField1Size;
  252. }
  253. if ($entryField2Size == 1) { // Optimyze one-byte field case
  254. $field2 = ord($xrefStreamData[$streamOffset++]);
  255. } else {
  256. $field2 = Zend_Pdf_StringParser::parseIntFromStream($xrefStreamData, $streamOffset, $entryField2Size);
  257. $streamOffset += $entryField2Size;
  258. }
  259. if ($entryField3Size == 1) { // Optimyze one-byte field case
  260. $field3 = ord($xrefStreamData[$streamOffset++]);
  261. } else {
  262. $field3 = Zend_Pdf_StringParser::parseIntFromStream($xrefStreamData, $streamOffset, $entryField3Size);
  263. $streamOffset += $entryField3Size;
  264. }
  265. switch ($type) {
  266. case 0:
  267. // Free object
  268. $refTable->addReference($objNum . ' ' . $field3 . ' R', $field2, false);
  269. // Debug output:
  270. // echo "Free object - $objNum $field3 R, next free - $field2\n";
  271. break;
  272. case 1:
  273. // In use object
  274. $refTable->addReference($objNum . ' ' . $field3 . ' R', $field2, true);
  275. // Debug output:
  276. // echo "In-use object - $objNum $field3 R, offset - $field2\n";
  277. break;
  278. case 2:
  279. // Object in an object stream
  280. // Debug output:
  281. // echo "Compressed object - $objNum 0 R, object stream - $field2 0 R, offset - $field3\n";
  282. break;
  283. }
  284. $objNum++;
  285. }
  286. }
  287. // $streamOffset . ' ' . strlen($xrefStreamData) . "\n";
  288. // "$entries\n";
  289. throw new Zend_Pdf_Exception('Cross-reference streams are not supported yet.');
  290. }
  291. $trailerObj = new Zend_Pdf_Trailer_Keeper($trailerDict, $context);
  292. if ($trailerDict->Prev instanceof Zend_Pdf_Element_Numeric ||
  293. $trailerDict->Prev instanceof Zend_Pdf_Element_Reference ) {
  294. $trailerObj->setPrev($this->_loadXRefTable($trailerDict->Prev->value));
  295. $context->getRefTable()->setParent($trailerObj->getPrev()->getRefTable());
  296. }
  297. /**
  298. * We set '/Prev' dictionary property to the current cross-reference section offset.
  299. * It doesn't correspond to the actual data, but is true when trailer will be used
  300. * as a trailer for next generated PDF section.
  301. */
  302. $trailerObj->Prev = new Zend_Pdf_Element_Numeric($offset);
  303. return $trailerObj;
  304. }
  305. /**
  306. * Get Trailer object
  307. *
  308. * @return Zend_Pdf_Trailer_Keeper
  309. */
  310. public function getTrailer()
  311. {
  312. return $this->_trailer;
  313. }
  314. /**
  315. * Object constructor
  316. *
  317. * Note: PHP duplicates string, which is sent by value, only of it's updated.
  318. * Thus we don't need to care about overhead
  319. *
  320. * @param mixed $source
  321. * @param Zend_Pdf_ElementFactory_Interface $factory
  322. * @param boolean $load
  323. * @throws Zend_Exception
  324. */
  325. public function __construct($source, Zend_Pdf_ElementFactory_Interface $factory, $load)
  326. {
  327. if ($load) {
  328. if (($pdfFile = @fopen($source, 'rb')) === false ) {
  329. throw new Zend_Pdf_Exception( "Can not open '$source' file for reading." );
  330. }
  331. $byteCount = filesize($source);
  332. $data = fread($pdfFile, $byteCount);
  333. $byteCount -= strlen($data);
  334. while ( $byteCount > 0 && ($nextBlock = fread($pdfFile, $byteCount)) != false ) {
  335. $data .= $nextBlock;
  336. $byteCount -= strlen($nextBlock);
  337. }
  338. fclose($pdfFile);
  339. $this->_stringParser = new Zend_Pdf_StringParser($data, $factory);
  340. } else {
  341. $this->_stringParser = new Zend_Pdf_StringParser($source, $factory);
  342. }
  343. $pdfVersionComment = $this->_stringParser->readComment();
  344. if (substr($pdfVersionComment, 0, 5) != '%PDF-') {
  345. throw new Zend_Pdf_Exception('File is not a PDF.');
  346. }
  347. $pdfVersion = substr($pdfVersionComment, 5);
  348. if (version_compare($pdfVersion, '0.9', '<') ||
  349. version_compare($pdfVersion, '1.61', '>=')
  350. ) {
  351. /**
  352. * @todo
  353. * To support PDF versions 1.5 (Acrobat 6) and PDF version 1.7 (Acrobat 7)
  354. * Stream compression filter must be implemented (for compressed object streams).
  355. * Cross reference streams must be implemented
  356. */
  357. throw new Zend_Pdf_Exception(sprintf('Unsupported PDF version. Zend_Pdf supports PDF 1.0-1.4. Current version - \'%f\'', $pdfVersion));
  358. }
  359. $this->_pdfVersion = $pdfVersion;
  360. $this->_stringParser->offset = strrpos($this->_stringParser->data, '%%EOF');
  361. if ($this->_stringParser->offset === false ||
  362. strlen($this->_stringParser->data) - $this->_stringParser->offset > 7) {
  363. throw new Zend_Pdf_Exception('Pdf file syntax error. End-of-fle marker expected at the end of file.');
  364. }
  365. $this->_stringParser->offset--;
  366. /**
  367. * Go to end of cross-reference table offset
  368. */
  369. while (Zend_Pdf_StringParser::isWhiteSpace( ord($this->_stringParser->data[$this->_stringParser->offset]) )&&
  370. ($this->_stringParser->offset > 0)) {
  371. $this->_stringParser->offset--;
  372. }
  373. /**
  374. * Go to the start of cross-reference table offset
  375. */
  376. while ( (!Zend_Pdf_StringParser::isWhiteSpace( ord($this->_stringParser->data[$this->_stringParser->offset]) ))&&
  377. ($this->_stringParser->offset > 0)) {
  378. $this->_stringParser->offset--;
  379. }
  380. /**
  381. * Go to the end of 'startxref' keyword
  382. */
  383. while (Zend_Pdf_StringParser::isWhiteSpace( ord($this->_stringParser->data[$this->_stringParser->offset]) )&&
  384. ($this->_stringParser->offset > 0)) {
  385. $this->_stringParser->offset--;
  386. }
  387. /**
  388. * Go to the white space (eol marker) before 'startxref' keyword
  389. */
  390. $this->_stringParser->offset -= 9;
  391. $nextLexeme = $this->_stringParser->readLexeme();
  392. if ($nextLexeme != 'startxref') {
  393. throw new Zend_Pdf_Exception(sprintf('Pdf file syntax error. \'startxref\' keyword expected. Offset - 0x%X.', $this->_stringParser->offset-strlen($nextLexeme)));
  394. }
  395. $startXref = $this->_stringParser->readLexeme();
  396. if (!ctype_digit($startXref)) {
  397. 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)));
  398. }
  399. $this->_trailer = $this->_loadXRefTable($startXref);
  400. $factory->setObjectCount($this->_trailer->Size->value);
  401. }
  402. /**
  403. * Object destructor
  404. */
  405. public function __destruct()
  406. {
  407. $this->_stringParser->cleanUp();
  408. }
  409. }