1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
/** |
4
|
|
|
* This file is based on code of tecnickcom/TCPDF PDF library. |
5
|
|
|
* |
6
|
|
|
* Original author Nicola Asuni ([email protected]) and |
7
|
|
|
* contributors (https://github.com/tecnickcom/TCPDF/graphs/contributors). |
8
|
|
|
* |
9
|
|
|
* @see https://github.com/tecnickcom/TCPDF |
10
|
|
|
* |
11
|
|
|
* Original code was licensed on the terms of the LGPL v3. |
12
|
|
|
* |
13
|
|
|
* ------------------------------------------------------------------------------ |
14
|
|
|
* |
15
|
|
|
* @file This file is part of the PdfParser library. |
16
|
|
|
* |
17
|
|
|
* @author Konrad Abicht <[email protected]> |
18
|
|
|
* @date 2020-01-06 |
19
|
|
|
* |
20
|
|
|
* @license LGPLv3 |
21
|
|
|
* @url <https://github.com/smalot/pdfparser> |
22
|
|
|
* |
23
|
|
|
* PdfParser is a pdf library written in PHP, extraction oriented. |
24
|
|
|
* Copyright (C) 2017 - Sébastien MALOT <[email protected]> |
25
|
|
|
* |
26
|
|
|
* This program is free software: you can redistribute it and/or modify |
27
|
|
|
* it under the terms of the GNU Lesser General Public License as published by |
28
|
|
|
* the Free Software Foundation, either version 3 of the License, or |
29
|
|
|
* (at your option) any later version. |
30
|
|
|
* |
31
|
|
|
* This program is distributed in the hope that it will be useful, |
32
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of |
33
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
34
|
|
|
* GNU Lesser General Public License for more details. |
35
|
|
|
* |
36
|
|
|
* You should have received a copy of the GNU Lesser General Public License |
37
|
|
|
* along with this program. |
38
|
|
|
* If not, see <http://www.pdfparser.org/sites/default/LICENSE.txt>. |
39
|
|
|
*/ |
40
|
|
|
|
41
|
|
|
namespace Smalot\PdfParser\RawData; |
42
|
|
|
|
43
|
|
|
use Exception; |
44
|
|
|
|
45
|
|
|
class RawDataParser |
46
|
|
|
{ |
47
|
|
|
/** |
48
|
|
|
* Configuration array. |
49
|
|
|
*/ |
50
|
|
|
protected $cfg = [ |
51
|
|
|
// if `true` ignore filter decoding errors |
52
|
|
|
'ignore_filter_decoding_errors' => true, |
53
|
|
|
// if `true` ignore missing filter decoding errors |
54
|
|
|
'ignore_missing_filter_decoders' => true, |
55
|
|
|
]; |
56
|
|
|
|
57
|
|
|
protected $filterHelper; |
58
|
|
|
|
59
|
|
|
protected $xrefCache; |
60
|
|
|
|
61
|
|
|
/** |
62
|
|
|
* @param array $cfg Configuration array, default is [] |
63
|
|
|
*/ |
64
|
16 |
|
public function __construct($cfg = []) |
65
|
|
|
{ |
66
|
|
|
// merge given array with default values |
67
|
16 |
|
$this->cfg = array_merge($this->cfg, $cfg); |
68
|
|
|
|
69
|
16 |
|
$this->filterHelper = new FilterHelper(); |
70
|
16 |
|
} |
71
|
|
|
|
72
|
|
|
/** |
73
|
|
|
* Decode the specified stream. |
74
|
|
|
* |
75
|
|
|
* @param string $pdfData PDF data |
76
|
|
|
* @param array $xref |
77
|
|
|
* @param array $sdic Stream's dictionary array |
78
|
|
|
* @param string $stream Stream to decode |
79
|
|
|
* |
80
|
|
|
* @return array containing decoded stream data and remaining filters |
81
|
|
|
*/ |
82
|
16 |
|
protected function decodeStream($pdfData, $xref, $sdic, $stream) |
83
|
|
|
{ |
84
|
|
|
// get stream length and filters |
85
|
16 |
|
$slength = \strlen($stream); |
86
|
16 |
|
if ($slength <= 0) { |
87
|
|
|
return ['', []]; |
88
|
|
|
} |
89
|
16 |
|
$filters = []; |
90
|
16 |
|
foreach ($sdic as $k => $v) { |
91
|
16 |
|
if ('/' == $v[0]) { |
92
|
16 |
|
if (('Length' == $v[1]) and (isset($sdic[($k + 1)])) and ('numeric' == $sdic[($k + 1)][0])) { |
93
|
|
|
// get declared stream length |
94
|
16 |
|
$declength = (int) ($sdic[($k + 1)][1]); |
95
|
16 |
|
if ($declength < $slength) { |
96
|
16 |
|
$stream = substr($stream, 0, $declength); |
97
|
16 |
|
$slength = $declength; |
98
|
|
|
} |
99
|
16 |
|
} elseif (('Filter' == $v[1]) and (isset($sdic[($k + 1)]))) { |
100
|
|
|
// resolve indirect object |
101
|
16 |
|
$objval = $this->getObjectVal($pdfData, $xref, $sdic[($k + 1)]); |
102
|
16 |
|
if ('/' == $objval[0]) { |
103
|
|
|
// single filter |
104
|
16 |
|
$filters[] = $objval[1]; |
105
|
1 |
|
} elseif ('[' == $objval[0]) { |
106
|
|
|
// array of filters |
107
|
1 |
|
foreach ($objval[1] as $flt) { |
108
|
1 |
|
if ('/' == $flt[0]) { |
109
|
16 |
|
$filters[] = $flt[1]; |
110
|
|
|
} |
111
|
|
|
} |
112
|
|
|
} |
113
|
|
|
} |
114
|
|
|
} |
115
|
|
|
} |
116
|
|
|
|
117
|
|
|
// decode the stream |
118
|
16 |
|
$remaining_filters = []; |
119
|
16 |
|
foreach ($filters as $filter) { |
120
|
16 |
|
if (\in_array($filter, $this->filterHelper->getAvailableFilters())) { |
121
|
|
|
try { |
122
|
16 |
|
$stream = $this->filterHelper->decodeFilter($filter, $stream); |
123
|
1 |
|
} catch (Exception $e) { |
124
|
1 |
|
$emsg = $e->getMessage(); |
125
|
1 |
|
if ((('~' == $emsg[0]) && !$this->cfg['ignore_missing_filter_decoders']) |
126
|
1 |
|
|| (('~' != $emsg[0]) && !$this->cfg['ignore_filter_decoding_errors']) |
127
|
|
|
) { |
128
|
16 |
|
throw new Exception($e->getMessage()); |
129
|
|
|
} |
130
|
|
|
} |
131
|
|
|
} else { |
132
|
|
|
// add missing filter to array |
133
|
16 |
|
$remaining_filters[] = $filter; |
134
|
|
|
} |
135
|
|
|
} |
136
|
|
|
|
137
|
16 |
|
return [$stream, $remaining_filters]; |
138
|
|
|
} |
139
|
|
|
|
140
|
|
|
/** |
141
|
|
|
* Decode the Cross-Reference section |
142
|
|
|
* |
143
|
|
|
* @param string $pdfData PDF data |
144
|
|
|
* @param int $startxref Offset at which the xref section starts (position of the 'xref' keyword) |
145
|
|
|
* @param array $xref Previous xref array (if any) |
146
|
|
|
* |
147
|
|
|
* @return array containing xref and trailer data |
148
|
|
|
*/ |
149
|
15 |
|
protected function decodeXref($pdfData, $startxref, $xref = []) |
150
|
|
|
{ |
151
|
15 |
|
$startxref += 4; // 4 is the length of the word 'xref' |
152
|
|
|
// 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) |
153
|
15 |
|
$offset = $startxref + strspn($pdfData, "\x00\x09\x0a\x0c\x0d\x20", $startxref); |
154
|
|
|
// initialize object number |
155
|
15 |
|
$obj_num = 0; |
156
|
|
|
// search for cross-reference entries or subsection |
157
|
15 |
|
while (preg_match('/([0-9]+)[\x20]([0-9]+)[\x20]?([nf]?)(\r\n|[\x20]?[\r\n])/', $pdfData, $matches, PREG_OFFSET_CAPTURE, $offset) > 0) { |
158
|
15 |
|
if ($matches[0][1] != $offset) { |
159
|
|
|
// we are on another section |
160
|
2 |
|
break; |
161
|
|
|
} |
162
|
15 |
|
$offset += \strlen($matches[0][0]); |
163
|
15 |
|
if ('n' == $matches[3][0]) { |
164
|
|
|
// create unique object index: [object number]_[generation number] |
165
|
15 |
|
$index = $obj_num.'_'.(int) ($matches[2][0]); |
166
|
|
|
// check if object already exist |
167
|
15 |
|
if (!isset($xref['xref'][$index])) { |
168
|
|
|
// store object offset position |
169
|
15 |
|
$xref['xref'][$index] = (int) ($matches[1][0]); |
170
|
|
|
} |
171
|
15 |
|
++$obj_num; |
172
|
15 |
|
} elseif ('f' == $matches[3][0]) { |
173
|
15 |
|
++$obj_num; |
174
|
|
|
} else { |
175
|
|
|
// object number (index) |
176
|
15 |
|
$obj_num = (int) ($matches[1][0]); |
177
|
|
|
} |
178
|
|
|
} |
179
|
|
|
// get trailer data |
180
|
15 |
|
if (preg_match('/trailer[\s]*<<(.*)>>/isU', $pdfData, $matches, PREG_OFFSET_CAPTURE, $offset) > 0) { |
181
|
15 |
|
$trailer_data = $matches[1][0]; |
182
|
15 |
|
if (!isset($xref['trailer']) or empty($xref['trailer'])) { |
183
|
|
|
// get only the last updated version |
184
|
15 |
|
$xref['trailer'] = []; |
185
|
|
|
// parse trailer_data |
186
|
15 |
|
if (preg_match('/Size[\s]+([0-9]+)/i', $trailer_data, $matches) > 0) { |
187
|
15 |
|
$xref['trailer']['size'] = (int) ($matches[1]); |
188
|
|
|
} |
189
|
15 |
|
if (preg_match('/Root[\s]+([0-9]+)[\s]+([0-9]+)[\s]+R/i', $trailer_data, $matches) > 0) { |
190
|
15 |
|
$xref['trailer']['root'] = (int) ($matches[1]).'_'.(int) ($matches[2]); |
191
|
|
|
} |
192
|
15 |
|
if (preg_match('/Encrypt[\s]+([0-9]+)[\s]+([0-9]+)[\s]+R/i', $trailer_data, $matches) > 0) { |
193
|
|
|
$xref['trailer']['encrypt'] = (int) ($matches[1]).'_'.(int) ($matches[2]); |
194
|
|
|
} |
195
|
15 |
|
if (preg_match('/Info[\s]+([0-9]+)[\s]+([0-9]+)[\s]+R/i', $trailer_data, $matches) > 0) { |
196
|
14 |
|
$xref['trailer']['info'] = (int) ($matches[1]).'_'.(int) ($matches[2]); |
197
|
|
|
} |
198
|
15 |
|
if (preg_match('/ID[\s]*[\[][\s]*[<]([^>]*)[>][\s]*[<]([^>]*)[>]/i', $trailer_data, $matches) > 0) { |
199
|
14 |
|
$xref['trailer']['id'] = []; |
200
|
14 |
|
$xref['trailer']['id'][0] = $matches[1]; |
201
|
14 |
|
$xref['trailer']['id'][1] = $matches[2]; |
202
|
|
|
} |
203
|
|
|
} |
204
|
15 |
|
if (preg_match('/Prev[\s]+([0-9]+)/i', $trailer_data, $matches) > 0) { |
205
|
|
|
// get previous xref |
206
|
15 |
|
$xref = $this->getXrefData($pdfData, (int) ($matches[1]), $xref); |
207
|
|
|
} |
208
|
|
|
} else { |
209
|
|
|
throw new Exception('Unable to find trailer'); |
210
|
|
|
} |
211
|
|
|
|
212
|
15 |
|
return $xref; |
213
|
|
|
} |
214
|
|
|
|
215
|
|
|
/** |
216
|
|
|
* Decode the Cross-Reference Stream section |
217
|
|
|
* |
218
|
|
|
* @param string $pdfData PDF data |
219
|
|
|
* @param int $startxref Offset at which the xref section starts |
220
|
|
|
* @param array $xref Previous xref array (if any) |
221
|
|
|
* |
222
|
|
|
* @return array containing xref and trailer data |
223
|
|
|
* |
224
|
|
|
* @throws Exception if unknown PNG predictor detected |
225
|
|
|
*/ |
226
|
2 |
|
protected function decodeXrefStream($pdfData, $startxref, $xref = []) |
227
|
|
|
{ |
228
|
|
|
// try to read Cross-Reference Stream |
229
|
2 |
|
$xrefobj = $this->getRawObject($pdfData, $startxref); |
230
|
2 |
|
$xrefcrs = $this->getIndirectObject($pdfData, $xref, $xrefobj[1], $startxref, true); |
231
|
2 |
|
if (!isset($xref['trailer']) or empty($xref['trailer'])) { |
232
|
|
|
// get only the last updated version |
233
|
2 |
|
$xref['trailer'] = []; |
234
|
2 |
|
$filltrailer = true; |
235
|
|
|
} else { |
236
|
2 |
|
$filltrailer = false; |
237
|
|
|
} |
238
|
2 |
|
if (!isset($xref['xref'])) { |
239
|
2 |
|
$xref['xref'] = []; |
240
|
|
|
} |
241
|
2 |
|
$valid_crs = false; |
242
|
2 |
|
$columns = 0; |
243
|
2 |
|
$sarr = $xrefcrs[0][1]; |
244
|
2 |
|
if (!\is_array($sarr)) { |
245
|
|
|
$sarr = []; |
246
|
|
|
} |
247
|
|
|
|
248
|
2 |
|
$wb = []; |
249
|
|
|
|
250
|
2 |
|
foreach ($sarr as $k => $v) { |
251
|
|
|
if ( |
252
|
2 |
|
('/' == $v[0]) |
253
|
2 |
|
&& ('Type' == $v[1]) |
254
|
|
|
&& ( |
255
|
2 |
|
isset($sarr[($k + 1)]) |
256
|
2 |
|
&& '/' == $sarr[($k + 1)][0] |
257
|
2 |
|
&& 'XRef' == $sarr[($k + 1)][1] |
258
|
|
|
) |
259
|
|
|
) { |
260
|
2 |
|
$valid_crs = true; |
261
|
2 |
|
} elseif (('/' == $v[0]) and ('Index' == $v[1]) and (isset($sarr[($k + 1)]))) { |
262
|
|
|
// first object number in the subsection |
263
|
2 |
|
$index_first = (int) ($sarr[($k + 1)][1][0][1]); |
264
|
2 |
|
} elseif (('/' == $v[0]) and ('Prev' == $v[1]) and (isset($sarr[($k + 1)]) and ('numeric' == $sarr[($k + 1)][0]))) { |
265
|
|
|
// get previous xref offset |
266
|
2 |
|
$prevxref = (int) ($sarr[($k + 1)][1]); |
267
|
2 |
|
} elseif (('/' == $v[0]) and ('W' == $v[1]) and (isset($sarr[($k + 1)]))) { |
268
|
|
|
// number of bytes (in the decoded stream) of the corresponding field |
269
|
2 |
|
$wb[0] = (int) ($sarr[($k + 1)][1][0][1]); |
270
|
2 |
|
$wb[1] = (int) ($sarr[($k + 1)][1][1][1]); |
271
|
2 |
|
$wb[2] = (int) ($sarr[($k + 1)][1][2][1]); |
272
|
2 |
|
} elseif (('/' == $v[0]) and ('DecodeParms' == $v[1]) and (isset($sarr[($k + 1)][1]))) { |
273
|
2 |
|
$decpar = $sarr[($k + 1)][1]; |
274
|
2 |
|
foreach ($decpar as $kdc => $vdc) { |
275
|
|
|
if ( |
276
|
2 |
|
'/' == $vdc[0] |
277
|
2 |
|
&& 'Columns' == $vdc[1] |
278
|
|
|
&& ( |
279
|
2 |
|
isset($decpar[($kdc + 1)]) |
280
|
2 |
|
&& 'numeric' == $decpar[($kdc + 1)][0] |
281
|
|
|
) |
282
|
|
|
) { |
283
|
2 |
|
$columns = (int) ($decpar[($kdc + 1)][1]); |
284
|
|
|
} elseif ( |
285
|
2 |
|
'/' == $vdc[0] |
286
|
2 |
|
&& 'Predictor' == $vdc[1] |
287
|
|
|
&& ( |
288
|
2 |
|
isset($decpar[($kdc + 1)]) |
289
|
2 |
|
&& 'numeric' == $decpar[($kdc + 1)][0] |
290
|
|
|
) |
291
|
|
|
) { |
292
|
2 |
|
$predictor = (int) ($decpar[($kdc + 1)][1]); |
|
|
|
|
293
|
|
|
} |
294
|
|
|
} |
295
|
2 |
|
} elseif ($filltrailer) { |
296
|
2 |
|
if (('/' == $v[0]) and ('Size' == $v[1]) and (isset($sarr[($k + 1)]) and ('numeric' == $sarr[($k + 1)][0]))) { |
297
|
2 |
|
$xref['trailer']['size'] = $sarr[($k + 1)][1]; |
298
|
2 |
|
} elseif (('/' == $v[0]) and ('Root' == $v[1]) and (isset($sarr[($k + 1)]) and ('objref' == $sarr[($k + 1)][0]))) { |
299
|
2 |
|
$xref['trailer']['root'] = $sarr[($k + 1)][1]; |
300
|
2 |
|
} elseif (('/' == $v[0]) and ('Info' == $v[1]) and (isset($sarr[($k + 1)]) and ('objref' == $sarr[($k + 1)][0]))) { |
301
|
2 |
|
$xref['trailer']['info'] = $sarr[($k + 1)][1]; |
302
|
2 |
|
} elseif (('/' == $v[0]) and ('Encrypt' == $v[1]) and (isset($sarr[($k + 1)]) and ('objref' == $sarr[($k + 1)][0]))) { |
303
|
|
|
$xref['trailer']['encrypt'] = $sarr[($k + 1)][1]; |
304
|
2 |
|
} elseif (('/' == $v[0]) and ('ID' == $v[1]) and (isset($sarr[($k + 1)]))) { |
305
|
2 |
|
$xref['trailer']['id'] = []; |
306
|
2 |
|
$xref['trailer']['id'][0] = $sarr[($k + 1)][1][0][1]; |
307
|
2 |
|
$xref['trailer']['id'][1] = $sarr[($k + 1)][1][1][1]; |
308
|
|
|
} |
309
|
|
|
} |
310
|
|
|
} |
311
|
|
|
|
312
|
|
|
// decode data |
313
|
2 |
|
if ($valid_crs and isset($xrefcrs[1][3][0])) { |
314
|
|
|
// number of bytes in a row |
315
|
2 |
|
$rowlen = ($columns + 1); |
316
|
|
|
// convert the stream into an array of integers |
317
|
2 |
|
$sdata = unpack('C*', $xrefcrs[1][3][0]); |
318
|
|
|
// split the rows |
319
|
2 |
|
$sdata = array_chunk($sdata, $rowlen); |
|
|
|
|
320
|
|
|
// initialize decoded array |
321
|
2 |
|
$ddata = []; |
322
|
|
|
// initialize first row with zeros |
323
|
2 |
|
$prev_row = array_fill(0, $rowlen, 0); |
324
|
|
|
// for each row apply PNG unpredictor |
325
|
2 |
|
foreach ($sdata as $k => $row) { |
326
|
|
|
// initialize new row |
327
|
2 |
|
$ddata[$k] = []; |
328
|
|
|
// get PNG predictor value |
329
|
2 |
|
$predictor = (10 + $row[0]); |
330
|
|
|
// for each byte on the row |
331
|
2 |
|
for ($i = 1; $i <= $columns; ++$i) { |
332
|
|
|
// new index |
333
|
2 |
|
$j = ($i - 1); |
334
|
2 |
|
$row_up = $prev_row[$j]; |
335
|
2 |
|
if (1 == $i) { |
336
|
2 |
|
$row_left = 0; |
337
|
2 |
|
$row_upleft = 0; |
338
|
|
|
} else { |
339
|
2 |
|
$row_left = $row[($i - 1)]; |
340
|
2 |
|
$row_upleft = $prev_row[($j - 1)]; |
341
|
|
|
} |
342
|
|
|
switch ($predictor) { |
343
|
2 |
|
case 10: // PNG prediction (on encoding, PNG None on all rows) |
344
|
|
|
$ddata[$k][$j] = $row[$i]; |
345
|
|
|
break; |
346
|
|
|
|
347
|
2 |
|
case 11: // PNG prediction (on encoding, PNG Sub on all rows) |
348
|
|
|
$ddata[$k][$j] = (($row[$i] + $row_left) & 0xff); |
349
|
|
|
break; |
350
|
|
|
|
351
|
2 |
|
case 12: // PNG prediction (on encoding, PNG Up on all rows) |
352
|
2 |
|
$ddata[$k][$j] = (($row[$i] + $row_up) & 0xff); |
353
|
2 |
|
break; |
354
|
|
|
|
355
|
|
|
case 13: // PNG prediction (on encoding, PNG Average on all rows) |
356
|
|
|
$ddata[$k][$j] = (($row[$i] + (($row_left + $row_up) / 2)) & 0xff); |
357
|
|
|
break; |
358
|
|
|
|
359
|
|
|
case 14: // PNG prediction (on encoding, PNG Paeth on all rows) |
360
|
|
|
// initial estimate |
361
|
|
|
$p = ($row_left + $row_up - $row_upleft); |
362
|
|
|
// distances |
363
|
|
|
$pa = abs($p - $row_left); |
364
|
|
|
$pb = abs($p - $row_up); |
365
|
|
|
$pc = abs($p - $row_upleft); |
366
|
|
|
$pmin = min($pa, $pb, $pc); |
367
|
|
|
// return minimum distance |
368
|
|
|
switch ($pmin) { |
369
|
|
|
case $pa: |
370
|
|
|
$ddata[$k][$j] = (($row[$i] + $row_left) & 0xff); |
371
|
|
|
break; |
372
|
|
|
|
373
|
|
|
case $pb: |
374
|
|
|
$ddata[$k][$j] = (($row[$i] + $row_up) & 0xff); |
375
|
|
|
break; |
376
|
|
|
|
377
|
|
|
case $pc: |
378
|
|
|
$ddata[$k][$j] = (($row[$i] + $row_upleft) & 0xff); |
379
|
|
|
break; |
380
|
|
|
} |
381
|
|
|
break; |
382
|
|
|
|
383
|
|
|
default: // PNG prediction (on encoding, PNG optimum) |
384
|
|
|
throw new Exception('Unknown PNG predictor'); |
385
|
|
|
} |
386
|
|
|
} |
387
|
2 |
|
$prev_row = $ddata[$k]; |
388
|
|
|
} // end for each row |
389
|
|
|
// complete decoding |
390
|
2 |
|
$sdata = []; |
391
|
|
|
// for every row |
392
|
2 |
|
foreach ($ddata as $k => $row) { |
393
|
|
|
// initialize new row |
394
|
2 |
|
$sdata[$k] = [0, 0, 0]; |
395
|
2 |
|
if (0 == $wb[0]) { |
396
|
|
|
// default type field |
397
|
|
|
$sdata[$k][0] = 1; |
398
|
|
|
} |
399
|
2 |
|
$i = 0; // count bytes in the row |
400
|
|
|
// for every column |
401
|
2 |
|
for ($c = 0; $c < 3; ++$c) { |
402
|
|
|
// for every byte on the column |
403
|
2 |
|
for ($b = 0; $b < $wb[$c]; ++$b) { |
404
|
2 |
|
if (isset($row[$i])) { |
405
|
2 |
|
$sdata[$k][$c] += ($row[$i] << (($wb[$c] - 1 - $b) * 8)); |
406
|
|
|
} |
407
|
2 |
|
++$i; |
408
|
|
|
} |
409
|
|
|
} |
410
|
|
|
} |
411
|
2 |
|
$ddata = []; |
|
|
|
|
412
|
|
|
// fill xref |
413
|
2 |
|
if (isset($index_first)) { |
414
|
2 |
|
$obj_num = $index_first; |
415
|
|
|
} else { |
416
|
2 |
|
$obj_num = 0; |
417
|
|
|
} |
418
|
2 |
|
foreach ($sdata as $k => $row) { |
419
|
2 |
|
switch ($row[0]) { |
420
|
2 |
|
case 0: // (f) linked list of free objects |
421
|
2 |
|
break; |
422
|
|
|
|
423
|
2 |
|
case 1: // (n) objects that are in use but are not compressed |
424
|
|
|
// create unique object index: [object number]_[generation number] |
425
|
2 |
|
$index = $obj_num.'_'.$row[2]; |
426
|
|
|
// check if object already exist |
427
|
2 |
|
if (!isset($xref['xref'][$index])) { |
428
|
|
|
// store object offset position |
429
|
2 |
|
$xref['xref'][$index] = $row[1]; |
430
|
|
|
} |
431
|
2 |
|
break; |
432
|
|
|
|
433
|
2 |
|
case 2: // compressed objects |
434
|
|
|
// $row[1] = object number of the object stream in which this object is stored |
435
|
|
|
// $row[2] = index of this object within the object stream |
436
|
2 |
|
$index = $row[1].'_0_'.$row[2]; |
437
|
2 |
|
$xref['xref'][$index] = -1; |
438
|
2 |
|
break; |
439
|
|
|
|
440
|
|
|
default: // null objects |
441
|
|
|
break; |
442
|
|
|
} |
443
|
2 |
|
++$obj_num; |
444
|
|
|
} |
445
|
|
|
} // end decoding data |
446
|
2 |
|
if (isset($prevxref)) { |
447
|
|
|
// get previous xref |
448
|
2 |
|
$xref = $this->getXrefData($pdfData, $prevxref, $xref); |
449
|
|
|
} |
450
|
|
|
|
451
|
2 |
|
return $xref; |
452
|
|
|
} |
453
|
|
|
|
454
|
|
|
/** |
455
|
|
|
* Get content of indirect object. |
456
|
|
|
* |
457
|
|
|
* @param string $pdfData PDF data |
458
|
|
|
* @param array $xref |
459
|
|
|
* @param string $obj_ref Object number and generation number separated by underscore character |
460
|
|
|
* @param int $offset Object offset |
461
|
|
|
* @param bool $decoding If true decode streams |
462
|
|
|
* |
463
|
|
|
* @return array containing object data |
464
|
|
|
* |
465
|
|
|
* @throws Exception if invalid object reference found |
466
|
|
|
*/ |
467
|
16 |
|
protected function getIndirectObject($pdfData, $xref, $obj_ref, $offset = 0, $decoding = true) |
468
|
|
|
{ |
469
|
16 |
|
$obj = explode('_', $obj_ref); |
470
|
16 |
|
if ((false === $obj) or (2 != \count($obj))) { |
471
|
|
|
throw new Exception('Invalid object reference for $obj.'); |
472
|
|
|
} |
473
|
16 |
|
$objref = $obj[0].' '.$obj[1].' obj'; |
474
|
|
|
// ignore leading zeros |
475
|
16 |
|
$offset += strspn($pdfData, '0', $offset); |
476
|
16 |
|
if (strpos($pdfData, $objref, $offset) != $offset) { |
477
|
|
|
// an indirect reference to an undefined object shall be considered a reference to the null object |
478
|
|
|
return ['null', 'null', $offset]; |
479
|
|
|
} |
480
|
|
|
// starting position of object content |
481
|
16 |
|
$offset += \strlen($objref); |
482
|
|
|
// get array of object content |
483
|
16 |
|
$objdata = []; |
484
|
16 |
|
$i = 0; // object main index |
485
|
|
|
do { |
486
|
16 |
|
$oldoffset = $offset; |
487
|
|
|
// get element |
488
|
16 |
|
$element = $this->getRawObject($pdfData, $offset); |
489
|
16 |
|
$offset = $element[2]; |
490
|
|
|
// decode stream using stream's dictionary information |
491
|
16 |
|
if ($decoding and ('stream' == $element[0]) and (isset($objdata[($i - 1)][0])) and ('<<' == $objdata[($i - 1)][0])) { |
492
|
16 |
|
$element[3] = $this->decodeStream($pdfData, $xref, $objdata[($i - 1)][1], $element[1]); |
493
|
|
|
} |
494
|
16 |
|
$objdata[$i] = $element; |
495
|
16 |
|
++$i; |
496
|
16 |
|
} while (('endobj' != $element[0]) and ($offset != $oldoffset)); |
497
|
|
|
|
498
|
|
|
// remove closing delimiter |
499
|
16 |
|
array_pop($objdata); |
500
|
|
|
|
501
|
|
|
// return raw object content |
502
|
16 |
|
return $objdata; |
503
|
|
|
} |
504
|
|
|
|
505
|
|
|
/** |
506
|
|
|
* Get the content of object, resolving indect object reference if necessary. |
507
|
|
|
* |
508
|
|
|
* @param string $pdfData PDF data |
509
|
|
|
* @param string $obj Object value |
510
|
|
|
* |
511
|
|
|
* @return array containing object data |
512
|
|
|
*/ |
513
|
16 |
|
protected function getObjectVal($pdfData, $xref, $obj) |
514
|
|
|
{ |
515
|
16 |
|
if ('objref' == $obj[0]) { |
516
|
|
|
// reference to indirect object |
517
|
|
|
if (isset($this->objects[$obj[1]])) { |
518
|
|
|
// this object has been already parsed |
519
|
|
|
return $this->objects[$obj[1]]; |
520
|
|
|
} elseif (isset($xref[$obj[1]])) { |
521
|
|
|
// parse new object |
522
|
|
|
$this->objects[$obj[1]] = $this->getIndirectObject($pdfData, $xref, $obj[1], $xref[$obj[1]], false); |
|
|
|
|
523
|
|
|
|
524
|
|
|
return $this->objects[$obj[1]]; |
525
|
|
|
} |
526
|
|
|
} |
527
|
|
|
|
528
|
16 |
|
return $obj; |
|
|
|
|
529
|
|
|
} |
530
|
|
|
|
531
|
|
|
/** |
532
|
|
|
* Get object type, raw value and offset to next object |
533
|
|
|
* |
534
|
|
|
* @param int $offset Object offset |
535
|
|
|
* |
536
|
|
|
* @return array containing object type, raw value and offset to next object |
537
|
|
|
*/ |
538
|
16 |
|
protected function getRawObject($pdfData, $offset = 0) |
539
|
|
|
{ |
540
|
16 |
|
$objtype = ''; // object type to be returned |
541
|
16 |
|
$objval = ''; // object value to be returned |
542
|
|
|
|
543
|
|
|
/* |
544
|
|
|
* skip initial white space chars: |
545
|
|
|
* \x00 null (NUL) |
546
|
|
|
* \x09 horizontal tab (HT) |
547
|
|
|
* \x0A line feed (LF) |
548
|
|
|
* \x0C form feed (FF) |
549
|
|
|
* \x0D carriage return (CR) |
550
|
|
|
* \x20 space (SP) |
551
|
|
|
*/ |
552
|
16 |
|
$offset += strspn($pdfData, "\x00\x09\x0a\x0c\x0d\x20", $offset); |
553
|
|
|
|
554
|
|
|
// get first char |
555
|
16 |
|
$char = $pdfData[$offset]; |
556
|
|
|
// get object type |
557
|
|
|
switch ($char) { |
558
|
16 |
|
case '%': // \x25 PERCENT SIGN |
559
|
|
|
// skip comment and search for next token |
560
|
|
|
$next = strcspn($pdfData, "\r\n", $offset); |
561
|
|
|
if ($next > 0) { |
562
|
|
|
$offset += $next; |
563
|
|
|
|
564
|
|
|
return $this->getRawObject($pdfData, $offset); |
565
|
|
|
} |
566
|
|
|
break; |
567
|
|
|
|
568
|
16 |
|
case '/': // \x2F SOLIDUS |
569
|
|
|
// name object |
570
|
16 |
|
$objtype = $char; |
571
|
16 |
|
++$offset; |
572
|
16 |
|
$pregResult = preg_match( |
573
|
16 |
|
'/^([^\x00\x09\x0a\x0c\x0d\x20\s\x28\x29\x3c\x3e\x5b\x5d\x7b\x7d\x2f\x25]+)/', |
574
|
16 |
|
substr($pdfData, $offset, 256), |
575
|
16 |
|
$matches |
576
|
|
|
); |
577
|
16 |
|
if (1 == $pregResult) { |
578
|
16 |
|
$objval = $matches[1]; // unescaped value |
579
|
16 |
|
$offset += \strlen($objval); |
580
|
|
|
} |
581
|
16 |
|
break; |
582
|
|
|
|
583
|
16 |
|
case '(': // \x28 LEFT PARENTHESIS |
584
|
16 |
|
case ')': // \x29 RIGHT PARENTHESIS |
585
|
|
|
// literal string object |
586
|
16 |
|
$objtype = $char; |
587
|
16 |
|
++$offset; |
588
|
16 |
|
$strpos = $offset; |
589
|
16 |
|
if ('(' == $char) { |
590
|
16 |
|
$open_bracket = 1; |
591
|
16 |
|
while ($open_bracket > 0) { |
592
|
16 |
|
if (!isset($pdfData[$strpos])) { |
593
|
|
|
break; |
594
|
|
|
} |
595
|
16 |
|
$ch = $pdfData[$strpos]; |
596
|
|
|
switch ($ch) { |
597
|
16 |
|
case '\\': // REVERSE SOLIDUS (5Ch) (Backslash) |
598
|
|
|
// skip next character |
599
|
14 |
|
++$strpos; |
600
|
14 |
|
break; |
601
|
|
|
|
602
|
16 |
|
case '(': // LEFT PARENHESIS (28h) |
603
|
|
|
++$open_bracket; |
604
|
|
|
break; |
605
|
|
|
|
606
|
16 |
|
case ')': // RIGHT PARENTHESIS (29h) |
607
|
16 |
|
--$open_bracket; |
608
|
16 |
|
break; |
609
|
|
|
} |
610
|
16 |
|
++$strpos; |
611
|
|
|
} |
612
|
16 |
|
$objval = substr($pdfData, $offset, ($strpos - $offset - 1)); |
613
|
16 |
|
$offset = $strpos; |
614
|
|
|
} |
615
|
16 |
|
break; |
616
|
|
|
|
617
|
16 |
|
case '[': // \x5B LEFT SQUARE BRACKET |
618
|
16 |
|
case ']': // \x5D RIGHT SQUARE BRACKET |
619
|
|
|
// array object |
620
|
16 |
|
$objtype = $char; |
621
|
16 |
|
++$offset; |
622
|
16 |
|
if ('[' == $char) { |
623
|
|
|
// get array content |
624
|
16 |
|
$objval = []; |
625
|
|
|
do { |
626
|
|
|
// get element |
627
|
16 |
|
$element = $this->getRawObject($pdfData, $offset); |
628
|
16 |
|
$offset = $element[2]; |
629
|
16 |
|
$objval[] = $element; |
630
|
16 |
|
} while (']' != $element[0]); |
631
|
|
|
// remove closing delimiter |
632
|
16 |
|
array_pop($objval); |
633
|
|
|
} |
634
|
16 |
|
break; |
635
|
|
|
|
636
|
16 |
|
case '<': // \x3C LESS-THAN SIGN |
637
|
16 |
|
case '>': // \x3E GREATER-THAN SIGN |
638
|
16 |
|
if (isset($pdfData[($offset + 1)]) and ($pdfData[($offset + 1)] == $char)) { |
639
|
|
|
// dictionary object |
640
|
16 |
|
$objtype = $char.$char; |
641
|
16 |
|
$offset += 2; |
642
|
16 |
|
if ('<' == $char) { |
643
|
|
|
// get array content |
644
|
16 |
|
$objval = []; |
645
|
|
|
do { |
646
|
|
|
// get element |
647
|
16 |
|
$element = $this->getRawObject($pdfData, $offset); |
648
|
16 |
|
$offset = $element[2]; |
649
|
16 |
|
$objval[] = $element; |
650
|
16 |
|
} while ('>>' != $element[0]); |
651
|
|
|
// remove closing delimiter |
652
|
16 |
|
array_pop($objval); |
653
|
|
|
} |
654
|
|
|
} else { |
655
|
|
|
// hexadecimal string object |
656
|
4 |
|
$objtype = $char; |
657
|
4 |
|
++$offset; |
658
|
4 |
|
$pregResult = preg_match( |
659
|
4 |
|
'/^([0-9A-Fa-f\x09\x0a\x0c\x0d\x20]+)>/iU', |
660
|
4 |
|
substr($pdfData, $offset), |
661
|
4 |
|
$matches |
662
|
|
|
); |
663
|
4 |
|
if (('<' == $char) && 1 == $pregResult) { |
664
|
|
|
// remove white space characters |
665
|
4 |
|
$objval = strtr($matches[1], "\x09\x0a\x0c\x0d\x20", ''); |
666
|
4 |
|
$offset += \strlen($matches[0]); |
667
|
|
|
} elseif (false !== ($endpos = strpos($pdfData, '>', $offset))) { |
668
|
|
|
$offset = $endpos + 1; |
669
|
|
|
} |
670
|
|
|
} |
671
|
16 |
|
break; |
672
|
|
|
|
673
|
|
|
default: |
674
|
16 |
|
if ('endobj' == substr($pdfData, $offset, 6)) { |
675
|
|
|
// indirect object |
676
|
16 |
|
$objtype = 'endobj'; |
677
|
16 |
|
$offset += 6; |
678
|
16 |
|
} elseif ('null' == substr($pdfData, $offset, 4)) { |
679
|
|
|
// null object |
680
|
2 |
|
$objtype = 'null'; |
681
|
2 |
|
$offset += 4; |
682
|
2 |
|
$objval = 'null'; |
683
|
16 |
|
} elseif ('true' == substr($pdfData, $offset, 4)) { |
684
|
|
|
// boolean true object |
685
|
4 |
|
$objtype = 'boolean'; |
686
|
4 |
|
$offset += 4; |
687
|
4 |
|
$objval = 'true'; |
688
|
16 |
|
} elseif ('false' == substr($pdfData, $offset, 5)) { |
689
|
|
|
// boolean false object |
690
|
|
|
$objtype = 'boolean'; |
691
|
|
|
$offset += 5; |
692
|
|
|
$objval = 'false'; |
693
|
16 |
|
} elseif ('stream' == substr($pdfData, $offset, 6)) { |
694
|
|
|
// start stream object |
695
|
16 |
|
$objtype = 'stream'; |
696
|
16 |
|
$offset += 6; |
697
|
16 |
|
if (1 == preg_match('/^([\r]?[\n])/isU', substr($pdfData, $offset), $matches)) { |
698
|
16 |
|
$offset += \strlen($matches[0]); |
699
|
16 |
|
$pregResult = preg_match( |
700
|
16 |
|
'/(endstream)[\x09\x0a\x0c\x0d\x20]/isU', |
701
|
16 |
|
substr($pdfData, $offset), |
702
|
16 |
|
$matches, |
703
|
16 |
|
PREG_OFFSET_CAPTURE |
704
|
|
|
); |
705
|
16 |
|
if (1 == $pregResult) { |
706
|
16 |
|
$objval = substr($pdfData, $offset, $matches[0][1]); |
707
|
16 |
|
$offset += $matches[1][1]; |
708
|
|
|
} |
709
|
|
|
} |
710
|
16 |
|
} elseif ('endstream' == substr($pdfData, $offset, 9)) { |
711
|
|
|
// end stream object |
712
|
16 |
|
$objtype = 'endstream'; |
713
|
16 |
|
$offset += 9; |
714
|
16 |
|
} elseif (1 == preg_match('/^([0-9]+)[\s]+([0-9]+)[\s]+R/iU', substr($pdfData, $offset, 33), $matches)) { |
715
|
|
|
// indirect object reference |
716
|
16 |
|
$objtype = 'objref'; |
717
|
16 |
|
$offset += \strlen($matches[0]); |
718
|
16 |
|
$objval = (int) ($matches[1]).'_'.(int) ($matches[2]); |
719
|
16 |
|
} elseif (1 == preg_match('/^([0-9]+)[\s]+([0-9]+)[\s]+obj/iU', substr($pdfData, $offset, 33), $matches)) { |
720
|
|
|
// object start |
721
|
2 |
|
$objtype = 'obj'; |
722
|
2 |
|
$objval = (int) ($matches[1]).'_'.(int) ($matches[2]); |
723
|
2 |
|
$offset += \strlen($matches[0]); |
724
|
16 |
|
} elseif (($numlen = strspn($pdfData, '+-.0123456789', $offset)) > 0) { |
725
|
|
|
// numeric object |
726
|
16 |
|
$objtype = 'numeric'; |
727
|
16 |
|
$objval = substr($pdfData, $offset, $numlen); |
728
|
16 |
|
$offset += $numlen; |
729
|
|
|
} |
730
|
16 |
|
break; |
731
|
|
|
} |
732
|
|
|
|
733
|
16 |
|
return [$objtype, $objval, $offset]; |
734
|
|
|
} |
735
|
|
|
|
736
|
|
|
/** |
737
|
|
|
* Get Cross-Reference (xref) table and trailer data from PDF document data. |
738
|
|
|
* |
739
|
|
|
* @param string $pdfData |
740
|
|
|
* @param int $offset xref offset (if know) |
741
|
|
|
* @param array $xref previous xref array (if any) |
742
|
|
|
* |
743
|
|
|
* @return array containing xref and trailer data |
744
|
|
|
* |
745
|
|
|
* @throws Exception if it was unable to find startxref |
746
|
|
|
* @throws Exception if it was unable to find xref |
747
|
|
|
*/ |
748
|
16 |
|
protected function getXrefData($pdfData, $offset = 0, $xref = []) |
749
|
|
|
{ |
750
|
16 |
|
$startxrefPreg = preg_match( |
751
|
16 |
|
'/[\r\n]startxref[\s]*[\r\n]+([0-9]+)[\s]*[\r\n]+%%EOF/i', |
752
|
16 |
|
$pdfData, |
753
|
16 |
|
$matches, |
754
|
16 |
|
PREG_OFFSET_CAPTURE, |
755
|
16 |
|
$offset |
756
|
|
|
); |
757
|
|
|
|
758
|
16 |
|
if (0 == $offset) { |
759
|
|
|
// find last startxref |
760
|
16 |
|
$pregResult = preg_match_all( |
761
|
16 |
|
'/[\r\n]startxref[\s]*[\r\n]+([0-9]+)[\s]*[\r\n]+%%EOF/i', |
762
|
16 |
|
$pdfData, $matches, |
763
|
16 |
|
PREG_SET_ORDER, |
764
|
16 |
|
$offset |
765
|
|
|
); |
766
|
16 |
|
if (0 == $pregResult) { |
767
|
|
|
throw new Exception('Unable to find startxref'); |
768
|
|
|
} |
769
|
16 |
|
$matches = array_pop($matches); |
770
|
16 |
|
$startxref = $matches[1]; |
771
|
4 |
|
} elseif (strpos($pdfData, 'xref', $offset) == $offset) { |
772
|
|
|
// Already pointing at the xref table |
773
|
2 |
|
$startxref = $offset; |
774
|
2 |
|
} elseif (preg_match('/([0-9]+[\s][0-9]+[\s]obj)/i', $pdfData, $matches, PREG_OFFSET_CAPTURE, $offset)) { |
775
|
|
|
// Cross-Reference Stream object |
776
|
2 |
|
$startxref = $offset; |
777
|
|
|
} elseif ($startxrefPreg) { |
778
|
|
|
// startxref found |
779
|
|
|
$startxref = $matches[1][0]; |
780
|
|
|
} else { |
781
|
|
|
throw new Exception('Unable to find startxref'); |
782
|
|
|
} |
783
|
|
|
|
784
|
|
|
// check xref position |
785
|
16 |
|
if (strpos($pdfData, 'xref', $startxref) == $startxref) { |
786
|
|
|
// Cross-Reference |
787
|
15 |
|
$xref = $this->decodeXref($pdfData, $startxref, $xref); |
788
|
|
|
} else { |
789
|
|
|
// Cross-Reference Stream |
790
|
2 |
|
$xref = $this->decodeXrefStream($pdfData, $startxref, $xref); |
791
|
|
|
} |
792
|
16 |
|
if (empty($xref)) { |
793
|
|
|
throw new Exception('Unable to find xref'); |
794
|
|
|
} |
795
|
|
|
|
796
|
16 |
|
return $xref; |
797
|
|
|
} |
798
|
|
|
|
799
|
|
|
/** |
800
|
|
|
* Parses PDF data and returns extracted data as array. |
801
|
|
|
* |
802
|
|
|
* @param string $data PDF data to parse |
803
|
|
|
* |
804
|
|
|
* @return array array of parsed PDF document objects |
805
|
|
|
* |
806
|
|
|
* @throws Exception if empty PDF data given |
807
|
|
|
* @throws Exception if PDF data missing %PDF header |
808
|
|
|
*/ |
809
|
16 |
|
public function parseData($data) |
810
|
|
|
{ |
811
|
16 |
|
if (empty($data)) { |
812
|
|
|
throw new Exception('Empty PDF data given.'); |
813
|
|
|
} |
814
|
|
|
// find the pdf header starting position |
815
|
16 |
|
if (false === ($trimpos = strpos($data, '%PDF-'))) { |
816
|
|
|
throw new Exception('Invalid PDF data: missing %PDF header.'); |
817
|
|
|
} |
818
|
|
|
|
819
|
|
|
// get PDF content string |
820
|
16 |
|
$pdfData = substr($data, $trimpos); |
821
|
|
|
|
822
|
|
|
// get xref and trailer data |
823
|
16 |
|
$xref = $this->getXrefData($pdfData); |
824
|
|
|
|
825
|
|
|
// parse all document objects |
826
|
16 |
|
$objects = []; |
827
|
16 |
|
foreach ($xref['xref'] as $obj => $offset) { |
828
|
16 |
|
if (!isset($objects[$obj]) and ($offset > 0)) { |
829
|
|
|
// decode objects with positive offset |
830
|
16 |
|
$objects[$obj] = $this->getIndirectObject($pdfData, $xref, $obj, $offset, true); |
831
|
|
|
} |
832
|
|
|
} |
833
|
|
|
|
834
|
16 |
|
return [$xref, $objects]; |
835
|
|
|
} |
836
|
|
|
} |
837
|
|
|
|