ILIAS  release_5-1 Revision 5.0.0-5477-g43f3e3fab5f
tcpdf_parser.php
Go to the documentation of this file.
1<?php
2//============================================================+
3// File name : tcpdf_parser.php
4// Version : 1.0.16
5// Begin : 2011-05-23
6// Last Update : 2015-04-28
7// Author : Nicola Asuni - Tecnick.com LTD - www.tecnick.com - info@tecnick.com
8// License : http://www.tecnick.com/pagefiles/tcpdf/LICENSE.TXT GNU-LGPLv3
9// -------------------------------------------------------------------
10// Copyright (C) 2011-2015 Nicola Asuni - Tecnick.com LTD
11//
12// This file is part of TCPDF software library.
13//
14// TCPDF is free software: you can redistribute it and/or modify it
15// under the terms of the GNU Lesser General Public License as
16// published by the Free Software Foundation, either version 3 of the
17// License, or (at your option) any later version.
18//
19// TCPDF is distributed in the hope that it will be useful, but
20// WITHOUT ANY WARRANTY; without even the implied warranty of
21// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
22// See the GNU Lesser General Public License for more details.
23//
24// You should have received a copy of the License
25// along with TCPDF. If not, see
26// <http://www.tecnick.com/pagefiles/tcpdf/LICENSE.TXT>.
27//
28// See LICENSE.TXT file for more information.
29// -------------------------------------------------------------------
30//
31// Description : This is a PHP class for parsing PDF documents.
32//
33//============================================================+
34
43// include class for decoding filters
44require_once(dirname(__FILE__).'/include/tcpdf_filters.php');
45
54class TCPDF_PARSER {
55
60 private $pdfdata = '';
61
66 protected $xref = array();
67
72 protected $objects = array();
73
78 private $FilterDecoders;
79
84 private $cfg = array(
85 'die_for_errors' => false,
86 'ignore_filter_decoding_errors' => true,
87 'ignore_missing_filter_decoders' => true,
88 );
89
90// -----------------------------------------------------------------------------
91
102 public function __construct($data, $cfg=array()) {
103 if (empty($data)) {
104 $this->Error('Empty PDF data.');
105 }
106 // find the pdf header starting position
107 if (($trimpos = strpos($data, '%PDF-')) === FALSE) {
108 $this->Error('Invalid PDF data: missing %PDF header.');
109 }
110 // get PDF content string
111 $this->pdfdata = substr($data, $trimpos);
112 // get length
113 $pdflen = strlen($this->pdfdata);
114 // set configuration parameters
115 $this->setConfig($cfg);
116 // get xref and trailer data
117 $this->xref = $this->getXrefData();
118 // parse all document objects
119 $this->objects = array();
120 foreach ($this->xref['xref'] as $obj => $offset) {
121 if (!isset($this->objects[$obj]) AND ($offset > 0)) {
122 // decode objects with positive offset
123 $this->objects[$obj] = $this->getIndirectObject($obj, $offset, true);
124 }
125 }
126 // release some memory
127 unset($this->pdfdata);
128 $this->pdfdata = '';
129 }
130
139 protected function setConfig($cfg) {
140 if (isset($cfg['die_for_errors'])) {
141 $this->cfg['die_for_errors'] = !!$cfg['die_for_errors'];
142 }
143 if (isset($cfg['ignore_filter_decoding_errors'])) {
144 $this->cfg['ignore_filter_decoding_errors'] = !!$cfg['ignore_filter_decoding_errors'];
145 }
146 if (isset($cfg['ignore_missing_filter_decoders'])) {
147 $this->cfg['ignore_missing_filter_decoders'] = !!$cfg['ignore_missing_filter_decoders'];
148 }
149 }
150
157 public function getParsedData() {
158 return array($this->xref, $this->objects);
159 }
160
169 protected function getXrefData($offset=0, $xref=array()) {
170 if ($offset == 0) {
171 // find last startxref
172 if (preg_match_all('/[\r\n]startxref[\s]*[\r\n]+([0-9]+)[\s]*[\r\n]+%%EOF/i', $this->pdfdata, $matches, PREG_SET_ORDER, $offset) == 0) {
173 $this->Error('Unable to find startxref');
174 }
175 $matches = array_pop($matches);
176 $startxref = $matches[1];
177 } elseif (strpos($this->pdfdata, 'xref', $offset) == $offset) {
178 // Already pointing at the xref table
179 $startxref = $offset;
180 } elseif (preg_match('/([0-9]+[\s][0-9]+[\s]obj)/i', $this->pdfdata, $matches, PREG_OFFSET_CAPTURE, $offset)) {
181 // Cross-Reference Stream object
182 $startxref = $offset;
183 } elseif (preg_match('/[\r\n]startxref[\s]*[\r\n]+([0-9]+)[\s]*[\r\n]+%%EOF/i', $this->pdfdata, $matches, PREG_OFFSET_CAPTURE, $offset)) {
184 // startxref found
185 $startxref = $matches[1][0];
186 } else {
187 $this->Error('Unable to find startxref');
188 }
189 // check xref position
190 if (strpos($this->pdfdata, 'xref', $startxref) == $startxref) {
191 // Cross-Reference
192 $xref = $this->decodeXref($startxref, $xref);
193 } else {
194 // Cross-Reference Stream
195 $xref = $this->decodeXrefStream($startxref, $xref);
196 }
197 if (empty($xref)) {
198 $this->Error('Unable to find xref');
199 }
200 return $xref;
201 }
202
211 protected function decodeXref($startxref, $xref=array()) {
212 $startxref += 4; // 4 is the length of the word 'xref'
213 // skip initial white space chars: \x00 null (NUL), \x09 horizontal tab (HT), \x0A line feed (LF), \x0C form feed (FF), \x0D carriage return (CR), \x20 space (SP)
214 $offset = $startxref + strspn($this->pdfdata, "\x00\x09\x0a\x0c\x0d\x20", $startxref);
215 // initialize object number
216 $obj_num = 0;
217 // search for cross-reference entries or subsection
218 while (preg_match('/([0-9]+)[\x20]([0-9]+)[\x20]?([nf]?)(\r\n|[\x20]?[\r\n])/', $this->pdfdata, $matches, PREG_OFFSET_CAPTURE, $offset) > 0) {
219 if ($matches[0][1] != $offset) {
220 // we are on another section
221 break;
222 }
223 $offset += strlen($matches[0][0]);
224 if ($matches[3][0] == 'n') {
225 // create unique object index: [object number]_[generation number]
226 $index = $obj_num.'_'.intval($matches[2][0]);
227 // check if object already exist
228 if (!isset($xref['xref'][$index])) {
229 // store object offset position
230 $xref['xref'][$index] = intval($matches[1][0]);
231 }
232 ++$obj_num;
233 } elseif ($matches[3][0] == 'f') {
234 ++$obj_num;
235 } else {
236 // object number (index)
237 $obj_num = intval($matches[1][0]);
238 }
239 }
240 // get trailer data
241 if (preg_match('/trailer[\s]*<<(.*)>>/isU', $this->pdfdata, $matches, PREG_OFFSET_CAPTURE, $offset) > 0) {
242 $trailer_data = $matches[1][0];
243 if (!isset($xref['trailer']) OR empty($xref['trailer'])) {
244 // get only the last updated version
245 $xref['trailer'] = array();
246 // parse trailer_data
247 if (preg_match('/Size[\s]+([0-9]+)/i', $trailer_data, $matches) > 0) {
248 $xref['trailer']['size'] = intval($matches[1]);
249 }
250 if (preg_match('/Root[\s]+([0-9]+)[\s]+([0-9]+)[\s]+R/i', $trailer_data, $matches) > 0) {
251 $xref['trailer']['root'] = intval($matches[1]).'_'.intval($matches[2]);
252 }
253 if (preg_match('/Encrypt[\s]+([0-9]+)[\s]+([0-9]+)[\s]+R/i', $trailer_data, $matches) > 0) {
254 $xref['trailer']['encrypt'] = intval($matches[1]).'_'.intval($matches[2]);
255 }
256 if (preg_match('/Info[\s]+([0-9]+)[\s]+([0-9]+)[\s]+R/i', $trailer_data, $matches) > 0) {
257 $xref['trailer']['info'] = intval($matches[1]).'_'.intval($matches[2]);
258 }
259 if (preg_match('/ID[\s]*[\[][\s]*[<]([^>]*)[>][\s]*[<]([^>]*)[>]/i', $trailer_data, $matches) > 0) {
260 $xref['trailer']['id'] = array();
261 $xref['trailer']['id'][0] = $matches[1];
262 $xref['trailer']['id'][1] = $matches[2];
263 }
264 }
265 if (preg_match('/Prev[\s]+([0-9]+)/i', $trailer_data, $matches) > 0) {
266 // get previous xref
267 $xref = $this->getXrefData(intval($matches[1]), $xref);
268 }
269 } else {
270 $this->Error('Unable to find trailer');
271 }
272 return $xref;
273 }
274
283 protected function decodeXrefStream($startxref, $xref=array()) {
284 // try to read Cross-Reference Stream
285 $xrefobj = $this->getRawObject($startxref);
286 $xrefcrs = $this->getIndirectObject($xrefobj[1], $startxref, true);
287 if (!isset($xref['trailer']) OR empty($xref['trailer'])) {
288 // get only the last updated version
289 $xref['trailer'] = array();
290 $filltrailer = true;
291 } else {
292 $filltrailer = false;
293 }
294 if (!isset($xref['xref'])) {
295 $xref['xref'] = array();
296 }
297 $valid_crs = false;
298 $columns = 0;
299 $sarr = $xrefcrs[0][1];
300 if (!is_array($sarr)) {
301 $sarr = array();
302 }
303 foreach ($sarr as $k => $v) {
304 if (($v[0] == '/') AND ($v[1] == 'Type') AND (isset($sarr[($k +1)]) AND ($sarr[($k +1)][0] == '/') AND ($sarr[($k +1)][1] == 'XRef'))) {
305 $valid_crs = true;
306 } elseif (($v[0] == '/') AND ($v[1] == 'Index') AND (isset($sarr[($k +1)]))) {
307 // first object number in the subsection
308 $index_first = intval($sarr[($k +1)][1][0][1]);
309 // number of entries in the subsection
310 $index_entries = intval($sarr[($k +1)][1][1][1]);
311 } elseif (($v[0] == '/') AND ($v[1] == 'Prev') AND (isset($sarr[($k +1)]) AND ($sarr[($k +1)][0] == 'numeric'))) {
312 // get previous xref offset
313 $prevxref = intval($sarr[($k +1)][1]);
314 } elseif (($v[0] == '/') AND ($v[1] == 'W') AND (isset($sarr[($k +1)]))) {
315 // number of bytes (in the decoded stream) of the corresponding field
316 $wb = array();
317 $wb[0] = intval($sarr[($k +1)][1][0][1]);
318 $wb[1] = intval($sarr[($k +1)][1][1][1]);
319 $wb[2] = intval($sarr[($k +1)][1][2][1]);
320 } elseif (($v[0] == '/') AND ($v[1] == 'DecodeParms') AND (isset($sarr[($k +1)][1]))) {
321 $decpar = $sarr[($k +1)][1];
322 foreach ($decpar as $kdc => $vdc) {
323 if (($vdc[0] == '/') AND ($vdc[1] == 'Columns') AND (isset($decpar[($kdc +1)]) AND ($decpar[($kdc +1)][0] == 'numeric'))) {
324 $columns = intval($decpar[($kdc +1)][1]);
325 } elseif (($vdc[0] == '/') AND ($vdc[1] == 'Predictor') AND (isset($decpar[($kdc +1)]) AND ($decpar[($kdc +1)][0] == 'numeric'))) {
326 $predictor = intval($decpar[($kdc +1)][1]);
327 }
328 }
329 } elseif ($filltrailer) {
330 if (($v[0] == '/') AND ($v[1] == 'Size') AND (isset($sarr[($k +1)]) AND ($sarr[($k +1)][0] == 'numeric'))) {
331 $xref['trailer']['size'] = $sarr[($k +1)][1];
332 } elseif (($v[0] == '/') AND ($v[1] == 'Root') AND (isset($sarr[($k +1)]) AND ($sarr[($k +1)][0] == 'objref'))) {
333 $xref['trailer']['root'] = $sarr[($k +1)][1];
334 } elseif (($v[0] == '/') AND ($v[1] == 'Info') AND (isset($sarr[($k +1)]) AND ($sarr[($k +1)][0] == 'objref'))) {
335 $xref['trailer']['info'] = $sarr[($k +1)][1];
336 } elseif (($v[0] == '/') AND ($v[1] == 'Encrypt') AND (isset($sarr[($k +1)]) AND ($sarr[($k +1)][0] == 'objref'))) {
337 $xref['trailer']['encrypt'] = $sarr[($k +1)][1];
338 } elseif (($v[0] == '/') AND ($v[1] == 'ID') AND (isset($sarr[($k +1)]))) {
339 $xref['trailer']['id'] = array();
340 $xref['trailer']['id'][0] = $sarr[($k +1)][1][0][1];
341 $xref['trailer']['id'][1] = $sarr[($k +1)][1][1][1];
342 }
343 }
344 }
345 // decode data
346 if ($valid_crs AND isset($xrefcrs[1][3][0])) {
347 // number of bytes in a row
348 $rowlen = ($columns + 1);
349 // convert the stream into an array of integers
350 $sdata = unpack('C*', $xrefcrs[1][3][0]);
351 // split the rows
352 $sdata = array_chunk($sdata, $rowlen);
353 // initialize decoded array
354 $ddata = array();
355 // initialize first row with zeros
356 $prev_row = array_fill (0, $rowlen, 0);
357 // for each row apply PNG unpredictor
358 foreach ($sdata as $k => $row) {
359 // initialize new row
360 $ddata[$k] = array();
361 // get PNG predictor value
362 $predictor = (10 + $row[0]);
363 // for each byte on the row
364 for ($i=1; $i<=$columns; ++$i) {
365 // new index
366 $j = ($i - 1);
367 $row_up = $prev_row[$j];
368 if ($i == 1) {
369 $row_left = 0;
370 $row_upleft = 0;
371 } else {
372 $row_left = $row[($i - 1)];
373 $row_upleft = $prev_row[($j - 1)];
374 }
375 switch ($predictor) {
376 case 10: { // PNG prediction (on encoding, PNG None on all rows)
377 $ddata[$k][$j] = $row[$i];
378 break;
379 }
380 case 11: { // PNG prediction (on encoding, PNG Sub on all rows)
381 $ddata[$k][$j] = (($row[$i] + $row_left) & 0xff);
382 break;
383 }
384 case 12: { // PNG prediction (on encoding, PNG Up on all rows)
385 $ddata[$k][$j] = (($row[$i] + $row_up) & 0xff);
386 break;
387 }
388 case 13: { // PNG prediction (on encoding, PNG Average on all rows)
389 $ddata[$k][$j] = (($row[$i] + (($row_left + $row_up) / 2)) & 0xff);
390 break;
391 }
392 case 14: { // PNG prediction (on encoding, PNG Paeth on all rows)
393 // initial estimate
394 $p = ($row_left + $row_up - $row_upleft);
395 // distances
396 $pa = abs($p - $row_left);
397 $pb = abs($p - $row_up);
398 $pc = abs($p - $row_upleft);
399 $pmin = min($pa, $pb, $pc);
400 // return minimum distance
401 switch ($pmin) {
402 case $pa: {
403 $ddata[$k][$j] = (($row[$i] + $row_left) & 0xff);
404 break;
405 }
406 case $pb: {
407 $ddata[$k][$j] = (($row[$i] + $row_up) & 0xff);
408 break;
409 }
410 case $pc: {
411 $ddata[$k][$j] = (($row[$i] + $row_upleft) & 0xff);
412 break;
413 }
414 }
415 break;
416 }
417 default: { // PNG prediction (on encoding, PNG optimum)
418 $this->Error('Unknown PNG predictor');
419 break;
420 }
421 }
422 }
423 $prev_row = $ddata[$k];
424 } // end for each row
425 // complete decoding
426 $sdata = array();
427 // for every row
428 foreach ($ddata as $k => $row) {
429 // initialize new row
430 $sdata[$k] = array(0, 0, 0);
431 if ($wb[0] == 0) {
432 // default type field
433 $sdata[$k][0] = 1;
434 }
435 $i = 0; // count bytes in the row
436 // for every column
437 for ($c = 0; $c < 3; ++$c) {
438 // for every byte on the column
439 for ($b = 0; $b < $wb[$c]; ++$b) {
440 if (isset($row[$i])) {
441 $sdata[$k][$c] += ($row[$i] << (($wb[$c] - 1 - $b) * 8));
442 }
443 ++$i;
444 }
445 }
446 }
447 $ddata = array();
448 // fill xref
449 if (isset($index_first)) {
450 $obj_num = $index_first;
451 } else {
452 $obj_num = 0;
453 }
454 foreach ($sdata as $k => $row) {
455 switch ($row[0]) {
456 case 0: { // (f) linked list of free objects
457 break;
458 }
459 case 1: { // (n) objects that are in use but are not compressed
460 // create unique object index: [object number]_[generation number]
461 $index = $obj_num.'_'.$row[2];
462 // check if object already exist
463 if (!isset($xref['xref'][$index])) {
464 // store object offset position
465 $xref['xref'][$index] = $row[1];
466 }
467 break;
468 }
469 case 2: { // compressed objects
470 // $row[1] = object number of the object stream in which this object is stored
471 // $row[2] = index of this object within the object stream
472 $index = $row[1].'_0_'.$row[2];
473 $xref['xref'][$index] = -1;
474 break;
475 }
476 default: { // null objects
477 break;
478 }
479 }
480 ++$obj_num;
481 }
482 } // end decoding data
483 if (isset($prevxref)) {
484 // get previous xref
485 $xref = $this->getXrefData($prevxref, $xref);
486 }
487 return $xref;
488 }
489
497 protected function getRawObject($offset=0) {
498 $objtype = ''; // object type to be returned
499 $objval = ''; // object value to be returned
500 // skip initial white space chars: \x00 null (NUL), \x09 horizontal tab (HT), \x0A line feed (LF), \x0C form feed (FF), \x0D carriage return (CR), \x20 space (SP)
501 $offset += strspn($this->pdfdata, "\x00\x09\x0a\x0c\x0d\x20", $offset);
502 // get first char
503 $char = $this->pdfdata[$offset];
504 // get object type
505 switch ($char) {
506 case '%': { // \x25 PERCENT SIGN
507 // skip comment and search for next token
508 $next = strcspn($this->pdfdata, "\r\n", $offset);
509 if ($next > 0) {
510 $offset += $next;
511 return $this->getRawObject($offset);
512 }
513 break;
514 }
515 case '/': { // \x2F SOLIDUS
516 // name object
517 $objtype = $char;
518 ++$offset;
519 if (preg_match('/^([^\x00\x09\x0a\x0c\x0d\x20\s\x28\x29\x3c\x3e\x5b\x5d\x7b\x7d\x2f\x25]+)/', substr($this->pdfdata, $offset, 256), $matches) == 1) {
520 $objval = $matches[1]; // unescaped value
521 $offset += strlen($objval);
522 }
523 break;
524 }
525 case '(': // \x28 LEFT PARENTHESIS
526 case ')': { // \x29 RIGHT PARENTHESIS
527 // literal string object
528 $objtype = $char;
529 ++$offset;
530 $strpos = $offset;
531 if ($char == '(') {
532 $open_bracket = 1;
533 while ($open_bracket > 0) {
534 if (!isset($this->pdfdata{$strpos})) {
535 break;
536 }
537 $ch = $this->pdfdata{$strpos};
538 switch ($ch) {
539 case '\\': { // REVERSE SOLIDUS (5Ch) (Backslash)
540 // skip next character
541 ++$strpos;
542 break;
543 }
544 case '(': { // LEFT PARENHESIS (28h)
545 ++$open_bracket;
546 break;
547 }
548 case ')': { // RIGHT PARENTHESIS (29h)
549 --$open_bracket;
550 break;
551 }
552 }
553 ++$strpos;
554 }
555 $objval = substr($this->pdfdata, $offset, ($strpos - $offset - 1));
556 $offset = $strpos;
557 }
558 break;
559 }
560 case '[': // \x5B LEFT SQUARE BRACKET
561 case ']': { // \x5D RIGHT SQUARE BRACKET
562 // array object
563 $objtype = $char;
564 ++$offset;
565 if ($char == '[') {
566 // get array content
567 $objval = array();
568 do {
569 // get element
570 $element = $this->getRawObject($offset);
571 $offset = $element[2];
572 $objval[] = $element;
573 } while ($element[0] != ']');
574 // remove closing delimiter
575 array_pop($objval);
576 }
577 break;
578 }
579 case '<': // \x3C LESS-THAN SIGN
580 case '>': { // \x3E GREATER-THAN SIGN
581 if (isset($this->pdfdata{($offset + 1)}) AND ($this->pdfdata{($offset + 1)} == $char)) {
582 // dictionary object
583 $objtype = $char.$char;
584 $offset += 2;
585 if ($char == '<') {
586 // get array content
587 $objval = array();
588 do {
589 // get element
590 $element = $this->getRawObject($offset);
591 $offset = $element[2];
592 $objval[] = $element;
593 } while ($element[0] != '>>');
594 // remove closing delimiter
595 array_pop($objval);
596 }
597 } else {
598 // hexadecimal string object
599 $objtype = $char;
600 ++$offset;
601 if (($char == '<') AND (preg_match('/^([0-9A-Fa-f\x09\x0a\x0c\x0d\x20]+)>/iU', substr($this->pdfdata, $offset), $matches) == 1)) {
602 // remove white space characters
603 $objval = strtr($matches[1], "\x09\x0a\x0c\x0d\x20", '');
604 $offset += strlen($matches[0]);
605 } elseif (($endpos = strpos($this->pdfdata, '>', $offset)) !== FALSE) {
606 $offset = $endpos + 1;
607 }
608 }
609 break;
610 }
611 default: {
612 if (substr($this->pdfdata, $offset, 6) == 'endobj') {
613 // indirect object
614 $objtype = 'endobj';
615 $offset += 6;
616 } elseif (substr($this->pdfdata, $offset, 4) == 'null') {
617 // null object
618 $objtype = 'null';
619 $offset += 4;
620 $objval = 'null';
621 } elseif (substr($this->pdfdata, $offset, 4) == 'true') {
622 // boolean true object
623 $objtype = 'boolean';
624 $offset += 4;
625 $objval = 'true';
626 } elseif (substr($this->pdfdata, $offset, 5) == 'false') {
627 // boolean false object
628 $objtype = 'boolean';
629 $offset += 5;
630 $objval = 'false';
631 } elseif (substr($this->pdfdata, $offset, 6) == 'stream') {
632 // start stream object
633 $objtype = 'stream';
634 $offset += 6;
635 if (preg_match('/^([\r]?[\n])/isU', substr($this->pdfdata, $offset), $matches) == 1) {
636 $offset += strlen($matches[0]);
637 if (preg_match('/(endstream)[\x09\x0a\x0c\x0d\x20]/isU', substr($this->pdfdata, $offset), $matches, PREG_OFFSET_CAPTURE) == 1) {
638 $objval = substr($this->pdfdata, $offset, $matches[0][1]);
639 $offset += $matches[1][1];
640 }
641 }
642 } elseif (substr($this->pdfdata, $offset, 9) == 'endstream') {
643 // end stream object
644 $objtype = 'endstream';
645 $offset += 9;
646 } elseif (preg_match('/^([0-9]+)[\s]+([0-9]+)[\s]+R/iU', substr($this->pdfdata, $offset, 33), $matches) == 1) {
647 // indirect object reference
648 $objtype = 'objref';
649 $offset += strlen($matches[0]);
650 $objval = intval($matches[1]).'_'.intval($matches[2]);
651 } elseif (preg_match('/^([0-9]+)[\s]+([0-9]+)[\s]+obj/iU', substr($this->pdfdata, $offset, 33), $matches) == 1) {
652 // object start
653 $objtype = 'obj';
654 $objval = intval($matches[1]).'_'.intval($matches[2]);
655 $offset += strlen ($matches[0]);
656 } elseif (($numlen = strspn($this->pdfdata, '+-.0123456789', $offset)) > 0) {
657 // numeric object
658 $objtype = 'numeric';
659 $objval = substr($this->pdfdata, $offset, $numlen);
660 $offset += $numlen;
661 }
662 break;
663 }
664 }
665 return array($objtype, $objval, $offset);
666 }
667
677 protected function getIndirectObject($obj_ref, $offset=0, $decoding=true) {
678 $obj = explode('_', $obj_ref);
679 if (($obj === false) OR (count($obj) != 2)) {
680 $this->Error('Invalid object reference: '.$obj);
681 return;
682 }
683 $objref = $obj[0].' '.$obj[1].' obj';
684 // ignore leading zeros
685 $offset += strspn($this->pdfdata, '0', $offset);
686 if (strpos($this->pdfdata, $objref, $offset) != $offset) {
687 // an indirect reference to an undefined object shall be considered a reference to the null object
688 return array('null', 'null', $offset);
689 }
690 // starting position of object content
691 $offset += strlen($objref);
692 // get array of object content
693 $objdata = array();
694 $i = 0; // object main index
695 do {
696 $oldoffset = $offset;
697 // get element
698 $element = $this->getRawObject($offset);
699 $offset = $element[2];
700 // decode stream using stream's dictionary information
701 if ($decoding AND ($element[0] == 'stream') AND (isset($objdata[($i - 1)][0])) AND ($objdata[($i - 1)][0] == '<<')) {
702 $element[3] = $this->decodeStream($objdata[($i - 1)][1], $element[1]);
703 }
704 $objdata[$i] = $element;
705 ++$i;
706 } while (($element[0] != 'endobj') AND ($offset != $oldoffset));
707 // remove closing delimiter
708 array_pop($objdata);
709 // return raw object content
710 return $objdata;
711 }
712
720 protected function getObjectVal($obj) {
721 if ($obj[0] == 'objref') {
722 // reference to indirect object
723 if (isset($this->objects[$obj[1]])) {
724 // this object has been already parsed
725 return $this->objects[$obj[1]];
726 } elseif (isset($this->xref[$obj[1]])) {
727 // parse new object
728 $this->objects[$obj[1]] = $this->getIndirectObject($obj[1], $this->xref[$obj[1]], false);
729 return $this->objects[$obj[1]];
730 }
731 }
732 return $obj;
733 }
734
743 protected function decodeStream($sdic, $stream) {
744 // get stream length and filters
745 $slength = strlen($stream);
746 if ($slength <= 0) {
747 return array('', array());
748 }
749 $filters = array();
750 foreach ($sdic as $k => $v) {
751 if ($v[0] == '/') {
752 if (($v[1] == 'Length') AND (isset($sdic[($k + 1)])) AND ($sdic[($k + 1)][0] == 'numeric')) {
753 // get declared stream length
754 $declength = intval($sdic[($k + 1)][1]);
755 if ($declength < $slength) {
756 $stream = substr($stream, 0, $declength);
757 $slength = $declength;
758 }
759 } elseif (($v[1] == 'Filter') AND (isset($sdic[($k + 1)]))) {
760 // resolve indirect object
761 $objval = $this->getObjectVal($sdic[($k + 1)]);
762 if ($objval[0] == '/') {
763 // single filter
764 $filters[] = $objval[1];
765 } elseif ($objval[0] == '[') {
766 // array of filters
767 foreach ($objval[1] as $flt) {
768 if ($flt[0] == '/') {
769 $filters[] = $flt[1];
770 }
771 }
772 }
773 }
774 }
775 }
776 // decode the stream
777 $remaining_filters = array();
778 foreach ($filters as $filter) {
779 if (in_array($filter, TCPDF_FILTERS::getAvailableFilters())) {
780 try {
781 $stream = TCPDF_FILTERS::decodeFilter($filter, $stream);
782 } catch (Exception $e) {
783 $emsg = $e->getMessage();
784 if ((($emsg[0] == '~') AND !$this->cfg['ignore_missing_filter_decoders'])
785 OR (($emsg[0] != '~') AND !$this->cfg['ignore_filter_decoding_errors'])) {
786 $this->Error($e->getMessage());
787 }
788 }
789 } else {
790 // add missing filter to array
791 $remaining_filters[] = $filter;
792 }
793 }
794 return array($stream, $remaining_filters);
795 }
796
803 public function Error($msg) {
804 if ($this->cfg['die_for_errors']) {
805 die('<strong>TCPDF_PARSER ERROR: </strong>'.$msg);
806 } else {
807 throw new Exception('TCPDF_PARSER ERROR: '.$msg);
808 }
809 }
810
811} // END OF TCPDF_PARSER CLASS
812
813//============================================================+
814// END OF FILE
815//============================================================+
$columns
if(! $in) $columns
Definition: Utf8Test.php:46
TCPDF_FILTERS\getAvailableFilters
static getAvailableFilters()
Get a list of available decoding filters.
Definition: tcpdf_filters.php:67
TCPDF_FILTERS\decodeFilter
static decodeFilter($filter, $data)
Decode data using the specified filter type.
Definition: tcpdf_filters.php:79
TCPDF_PARSER
This is a PHP class for parsing PDF documents.
Definition: tcpdf_parser.php:54
TCPDF_PARSER\decodeXrefStream
decodeXrefStream($startxref, $xref=array())
Decode the Cross-Reference Stream section.
Definition: tcpdf_parser.php:283
TCPDF_PARSER\$objects
$objects
Array of PDF objects.
Definition: tcpdf_parser.php:72
TCPDF_PARSER\__construct
__construct($data, $cfg=array())
Parse a PDF document an return an array of objects.
Definition: tcpdf_parser.php:102
TCPDF_PARSER\Error
Error($msg)
Throw an exception or print an error message and die if the K_TCPDF_PARSER_THROW_EXCEPTION_ERROR cons...
Definition: tcpdf_parser.php:803
TCPDF_PARSER\getParsedData
getParsedData()
Return an array of parsed PDF document objects.
Definition: tcpdf_parser.php:157
TCPDF_PARSER\decodeStream
decodeStream($sdic, $stream)
Decode the specified stream.
Definition: tcpdf_parser.php:743
TCPDF_PARSER\getObjectVal
getObjectVal($obj)
Get the content of object, resolving indect object reference if necessary.
Definition: tcpdf_parser.php:720
TCPDF_PARSER\getIndirectObject
getIndirectObject($obj_ref, $offset=0, $decoding=true)
Get content of indirect object.
Definition: tcpdf_parser.php:677
TCPDF_PARSER\$xref
$xref
XREF data.
Definition: tcpdf_parser.php:66
TCPDF_PARSER\$FilterDecoders
$FilterDecoders
Class object for decoding filters.
Definition: tcpdf_parser.php:78
TCPDF_PARSER\decodeXref
decodeXref($startxref, $xref=array())
Decode the Cross-Reference section.
Definition: tcpdf_parser.php:211
TCPDF_PARSER\setConfig
setConfig($cfg)
Set the configuration parameters.
Definition: tcpdf_parser.php:139
TCPDF_PARSER\$cfg
$cfg
Array of configuration parameters.
Definition: tcpdf_parser.php:84
TCPDF_PARSER\getXrefData
getXrefData($offset=0, $xref=array())
Get Cross-Reference (xref) table and trailer data from PDF document data.
Definition: tcpdf_parser.php:169
TCPDF_PARSER\getRawObject
getRawObject($offset=0)
Get object type, raw value and offset to next object.
Definition: tcpdf_parser.php:497
TCPDF_PARSER\$pdfdata
$pdfdata
Raw content of the PDF document.
Definition: tcpdf_parser.php:60
$data
$data
Definition: example_011.php:126