|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* PDFtk wrapper |
|
4
|
|
|
* |
|
5
|
|
|
* @copyright 2014-2024 Institute of Legal Medicine, Medical University of Innsbruck |
|
6
|
|
|
* @author Martin Pircher <[email protected]> |
|
7
|
|
|
* @author Andreas Erhard <[email protected]> |
|
8
|
|
|
* @license LGPL-3.0-only |
|
9
|
|
|
* @link http://www.gerichtsmedizin.at/ |
|
10
|
|
|
* |
|
11
|
|
|
* @package pdftk |
|
12
|
|
|
*/ |
|
13
|
|
|
|
|
14
|
|
|
namespace Gmi\Toolkit\Pdftk; |
|
15
|
|
|
|
|
16
|
|
|
use Symfony\Component\Process\Process; |
|
17
|
|
|
|
|
18
|
|
|
use Gmi\Toolkit\Pdftk\Exception\FileNotFoundException; |
|
19
|
|
|
use Gmi\Toolkit\Pdftk\Exception\PdfException; |
|
20
|
|
|
use Gmi\Toolkit\Pdftk\Util\Escaper; |
|
21
|
|
|
use Gmi\Toolkit\Pdftk\Util\ProcessFactory; |
|
22
|
|
|
|
|
23
|
|
|
use Exception; |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* Wrapper for PDFtk. |
|
27
|
|
|
* |
|
28
|
|
|
* @internal Only the methods exposed by the interfaces should be accessed from outside. |
|
29
|
|
|
*/ |
|
30
|
|
|
class PdftkWrapper implements WrapperInterface, BinaryPathAwareInterface |
|
31
|
|
|
{ |
|
32
|
|
|
use BinaryPathAwareTrait; |
|
33
|
|
|
|
|
34
|
|
|
/** |
|
35
|
|
|
* @var ProcessFactory |
|
36
|
|
|
*/ |
|
37
|
|
|
private $processFactory; |
|
38
|
|
|
|
|
39
|
|
|
/** |
|
40
|
|
|
* @var Escaper |
|
41
|
|
|
*/ |
|
42
|
|
|
private $escaper; |
|
43
|
|
|
|
|
44
|
|
|
/** |
|
45
|
|
|
* Constructor. |
|
46
|
|
|
* |
|
47
|
|
|
* @throws FileNotFoundException |
|
48
|
|
|
*/ |
|
49
|
|
|
public function __construct(string $pdftkBinary = null, ProcessFactory $processFactory = null) |
|
50
|
|
|
{ |
|
51
|
|
|
$this->setBinary($pdftkBinary ?: $this->guessBinary(PHP_OS)); |
|
52
|
|
|
$this->processFactory = $processFactory ?: new ProcessFactory(); |
|
53
|
|
|
$this->escaper = new Escaper(); |
|
54
|
|
|
} |
|
55
|
|
|
|
|
56
|
|
|
/** |
|
57
|
|
|
* Guesses the pdftk binary path based on the operating system. |
|
58
|
|
|
*/ |
|
59
|
|
|
public function guessBinary(string $operatingSystemString): string |
|
60
|
|
|
{ |
|
61
|
|
|
if (strtoupper(substr($operatingSystemString, 0, 3)) === 'WIN') { |
|
62
|
|
|
$binary = 'C:\\Program Files (x86)\\PDFtk Server\\bin\\pdftk.exe'; |
|
63
|
|
|
} else { |
|
64
|
|
|
$binary = '/usr/bin/pdftk'; |
|
65
|
|
|
} |
|
66
|
|
|
|
|
67
|
|
|
return $binary; |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
|
|
/** |
|
71
|
|
|
* {@inheritDoc} |
|
72
|
|
|
*/ |
|
73
|
|
|
public function join(array $filePaths, string $outfile): void |
|
74
|
|
|
{ |
|
75
|
|
|
$esc = $this->escaper; |
|
76
|
|
|
|
|
77
|
|
|
$filePathsEscaped = array_map(function (string $filePath) use ($esc) { |
|
78
|
|
|
return $esc->shellArg($filePath); |
|
79
|
|
|
}, $filePaths); |
|
80
|
|
|
|
|
81
|
|
|
$fileList = implode(' ', $filePathsEscaped); |
|
82
|
|
|
|
|
83
|
|
|
$commandLine = sprintf('%s %s cat output %s', $this->getBinary(), $fileList, $esc->shellArg($outfile)); |
|
84
|
|
|
|
|
85
|
|
|
/** |
|
86
|
|
|
* @var Process |
|
87
|
|
|
*/ |
|
88
|
|
|
$process = $this->processFactory->createProcess($commandLine); |
|
89
|
|
|
|
|
90
|
|
|
try { |
|
91
|
|
|
$process->mustRun(); |
|
92
|
|
|
} catch (Exception $e) { |
|
93
|
|
|
throw new PdfException($e->getMessage(), 0, $e, $process->getErrorOutput(), $process->getOutput()); |
|
94
|
|
|
} |
|
95
|
|
|
|
|
96
|
|
|
$process->getOutput(); |
|
97
|
|
|
} |
|
98
|
|
|
|
|
99
|
|
|
/** |
|
100
|
|
|
* {@inheritDoc} |
|
101
|
|
|
*/ |
|
102
|
|
|
public function split(string $infile, array $mapping, string $outputFolder = null): void |
|
103
|
|
|
{ |
|
104
|
|
|
$commandLines = $this->buildSplitCommandLines($infile, $mapping, $outputFolder); |
|
105
|
|
|
|
|
106
|
|
|
foreach ($commandLines as $commandLine) { |
|
107
|
|
|
$process = $this->processFactory->createProcess($commandLine); |
|
108
|
|
|
try { |
|
109
|
|
|
$process->mustRun(); |
|
110
|
|
|
} catch (Exception $e) { |
|
111
|
|
|
throw new PdfException($e->getMessage(), 0, $e, $process->getErrorOutput(), $process->getOutput()); |
|
112
|
|
|
} |
|
113
|
|
|
} |
|
114
|
|
|
} |
|
115
|
|
|
|
|
116
|
|
|
/** |
|
117
|
|
|
* {@inheritDoc} |
|
118
|
|
|
*/ |
|
119
|
|
|
public function reorder(string $infile, array $order, string $outfile = null): void |
|
120
|
|
|
{ |
|
121
|
|
|
$pageNumbers = implode(' ', $order); |
|
122
|
|
|
|
|
123
|
|
|
$temporaryOutFile = false; |
|
124
|
|
|
|
|
125
|
|
|
if ($outfile === null || $infile === $outfile) { |
|
126
|
|
|
$temporaryOutFile = true; |
|
127
|
|
|
$outfile = tempnam(sys_get_temp_dir(), 'pdf'); |
|
128
|
|
|
} |
|
129
|
|
|
|
|
130
|
|
|
$esc = $this->escaper; |
|
131
|
|
|
|
|
132
|
|
|
$commandLine = sprintf( |
|
133
|
|
|
'%s %s cat %s output %s', |
|
134
|
|
|
$this->getBinary(), |
|
135
|
|
|
$esc->shellArg($infile), |
|
136
|
|
|
$pageNumbers, |
|
137
|
|
|
$esc->shellArg($outfile) |
|
138
|
|
|
); |
|
139
|
|
|
|
|
140
|
|
|
/** |
|
141
|
|
|
* @var Process |
|
142
|
|
|
*/ |
|
143
|
|
|
$process = $this->processFactory->createProcess($commandLine); |
|
144
|
|
|
|
|
145
|
|
|
try { |
|
146
|
|
|
$process->mustRun(); |
|
147
|
|
|
} catch (Exception $e) { |
|
148
|
|
|
throw new PdfException( |
|
149
|
|
|
sprintf('Failed to reorder PDF "%s"! Error: %s', $infile, $e->getMessage()), |
|
150
|
|
|
0, |
|
151
|
|
|
$e, |
|
152
|
|
|
$process->getErrorOutput(), |
|
153
|
|
|
$process->getOutput() |
|
154
|
|
|
); |
|
155
|
|
|
} |
|
156
|
|
|
|
|
157
|
|
|
if ($temporaryOutFile) { |
|
158
|
|
|
unlink($infile); |
|
159
|
|
|
rename($outfile, $infile); |
|
160
|
|
|
} |
|
161
|
|
|
} |
|
162
|
|
|
|
|
163
|
|
|
/** |
|
164
|
|
|
* {@inheritDoc} |
|
165
|
|
|
*/ |
|
166
|
|
|
public function applyBookmarks(Bookmarks $bookmarks, string $infile, string $outfile = null): self |
|
167
|
|
|
{ |
|
168
|
|
|
$this->updatePdfDataFromDump($infile, $this->buildBookmarksBlock($bookmarks), $outfile); |
|
169
|
|
|
|
|
170
|
|
|
return $this; |
|
171
|
|
|
} |
|
172
|
|
|
|
|
173
|
|
|
/** |
|
174
|
|
|
* {@inheritDoc} |
|
175
|
|
|
*/ |
|
176
|
|
|
public function importBookmarks(Bookmarks $bookmarks, string $infile): self |
|
177
|
|
|
{ |
|
178
|
|
|
$dump = $this->getPdfDataDump($infile); |
|
179
|
|
|
$this->importBookmarksFromDump($bookmarks, $dump); |
|
180
|
|
|
|
|
181
|
|
|
return $this; |
|
182
|
|
|
} |
|
183
|
|
|
|
|
184
|
|
|
/** |
|
185
|
|
|
* {@inheritDoc} |
|
186
|
|
|
*/ |
|
187
|
|
|
public function importPages(Pages $pages, string $infile): self |
|
188
|
|
|
{ |
|
189
|
|
|
$dump = $this->getPdfDataDump($infile); |
|
190
|
|
|
$this->importPagesFromDump($pages, $dump); |
|
191
|
|
|
|
|
192
|
|
|
return $this; |
|
193
|
|
|
} |
|
194
|
|
|
|
|
195
|
|
|
/** |
|
196
|
|
|
* {@inheritDoc} |
|
197
|
|
|
*/ |
|
198
|
|
|
public function applyMetadata(Metadata $metadata, string $infile, string $outfile = null): self |
|
199
|
|
|
{ |
|
200
|
|
|
$metadataBlock = $this->buildMetadataBlock($metadata); |
|
201
|
|
|
|
|
202
|
|
|
$this->updatePdfDataFromDump($infile, $metadataBlock, $outfile); |
|
203
|
|
|
|
|
204
|
|
|
return $this; |
|
205
|
|
|
} |
|
206
|
|
|
|
|
207
|
|
|
/** |
|
208
|
|
|
* {@inheritDoc} |
|
209
|
|
|
*/ |
|
210
|
|
|
public function importMetadata(Metadata $metadata, string $infile): self |
|
211
|
|
|
{ |
|
212
|
|
|
$dump = $this->getPdfDataDump($infile); |
|
213
|
|
|
$this->importMetadataFromDump($metadata, $dump); |
|
214
|
|
|
|
|
215
|
|
|
return $this; |
|
216
|
|
|
} |
|
217
|
|
|
|
|
218
|
|
|
/** |
|
219
|
|
|
* Get data dump. |
|
220
|
|
|
* |
|
221
|
|
|
* @throws PdfException |
|
222
|
|
|
*/ |
|
223
|
|
|
public function getPdfDataDump(string $pdf): string |
|
224
|
|
|
{ |
|
225
|
|
|
if (!file_exists($pdf)) { |
|
226
|
|
|
throw new FileNotFoundException(sprintf('PDF "%s" not found!', $pdf)); |
|
227
|
|
|
} |
|
228
|
|
|
|
|
229
|
|
|
$esc = $this->escaper; |
|
230
|
|
|
|
|
231
|
|
|
$tempfile = tempnam(sys_get_temp_dir(), 'pdf'); |
|
232
|
|
|
$cmd = sprintf( |
|
233
|
|
|
'%s %s dump_data_utf8 output %s', |
|
234
|
|
|
$this->getBinary(), |
|
235
|
|
|
$esc->shellArg($pdf), |
|
236
|
|
|
$esc->shellArg($tempfile) |
|
237
|
|
|
); |
|
238
|
|
|
|
|
239
|
|
|
$process = $this->processFactory->createProcess($cmd); |
|
240
|
|
|
|
|
241
|
|
|
try { |
|
242
|
|
|
$process->mustRun(); |
|
243
|
|
|
} catch (Exception $e) { |
|
244
|
|
|
$exception = new PdfException( |
|
245
|
|
|
sprintf('Failed to read PDF data from "%s"! Error: %s', $pdf, $e->getMessage()), |
|
246
|
|
|
0, |
|
247
|
|
|
$e, |
|
248
|
|
|
$process->getErrorOutput(), |
|
249
|
|
|
$process->getOutput() |
|
250
|
|
|
); |
|
251
|
|
|
} |
|
252
|
|
|
|
|
253
|
|
|
$dump = file_get_contents($tempfile); |
|
254
|
|
|
unlink($tempfile); |
|
255
|
|
|
|
|
256
|
|
|
if (isset($exception)) { |
|
257
|
|
|
throw $exception; |
|
258
|
|
|
} |
|
259
|
|
|
|
|
260
|
|
|
return $dump; |
|
261
|
|
|
} |
|
262
|
|
|
|
|
263
|
|
|
/** |
|
264
|
|
|
* Update PDF data from dump. |
|
265
|
|
|
* |
|
266
|
|
|
* @param string $pdf input file |
|
267
|
|
|
* @param string $data dump data or filename of containing dump data |
|
268
|
|
|
* @param string $outfile output file (is input when null) |
|
269
|
|
|
* |
|
270
|
|
|
* @throws PdfException |
|
271
|
|
|
*/ |
|
272
|
|
|
public function updatePdfDataFromDump(string $pdf, string $data, string $outfile = null): void |
|
273
|
|
|
{ |
|
274
|
|
|
$temporaryOutFile = false; |
|
275
|
|
|
|
|
276
|
|
|
if (!file_exists($pdf)) { |
|
277
|
|
|
throw new FileNotFoundException(sprintf('PDF "%s" not found', $pdf)); |
|
278
|
|
|
} |
|
279
|
|
|
|
|
280
|
|
|
if ($outfile === null || $pdf === $outfile) { |
|
281
|
|
|
$temporaryOutFile = true; |
|
282
|
|
|
$outfile = tempnam(sys_get_temp_dir(), 'pdf'); |
|
283
|
|
|
} |
|
284
|
|
|
|
|
285
|
|
|
$tempfile = tempnam(sys_get_temp_dir(), 'pdf'); |
|
286
|
|
|
file_put_contents($tempfile, $data); |
|
287
|
|
|
|
|
288
|
|
|
$esc = $this->escaper; |
|
289
|
|
|
|
|
290
|
|
|
$cmd = sprintf( |
|
291
|
|
|
'%s %s update_info_utf8 %s output %s', |
|
292
|
|
|
$this->getBinary(), |
|
293
|
|
|
$esc->shellArg($pdf), |
|
294
|
|
|
$esc->shellArg($tempfile), |
|
295
|
|
|
$esc->shellArg($outfile) |
|
296
|
|
|
); |
|
297
|
|
|
|
|
298
|
|
|
$process = $this->processFactory->createProcess($cmd); |
|
299
|
|
|
|
|
300
|
|
|
try { |
|
301
|
|
|
$process->mustRun(); |
|
302
|
|
|
} catch (Exception $e) { |
|
303
|
|
|
$exception = new PdfException( |
|
304
|
|
|
sprintf('Failed to write PDF data to "%s"! Error: %s', $outfile, $e->getMessage()), |
|
305
|
|
|
0, |
|
306
|
|
|
$e, |
|
307
|
|
|
$process->getErrorOutput(), |
|
308
|
|
|
$process->getOutput() |
|
309
|
|
|
); |
|
310
|
|
|
} |
|
311
|
|
|
|
|
312
|
|
|
unlink($tempfile); |
|
313
|
|
|
|
|
314
|
|
|
if ($temporaryOutFile && !isset($exception)) { |
|
315
|
|
|
unlink($pdf); |
|
316
|
|
|
rename($outfile, $pdf); |
|
317
|
|
|
} |
|
318
|
|
|
|
|
319
|
|
|
if (isset($exception)) { |
|
320
|
|
|
throw $exception; |
|
321
|
|
|
} |
|
322
|
|
|
} |
|
323
|
|
|
|
|
324
|
|
|
/** |
|
325
|
|
|
* Builds the pdftk command lines for splitting. |
|
326
|
|
|
* |
|
327
|
|
|
* @return string[] |
|
328
|
|
|
*/ |
|
329
|
|
|
private function buildSplitCommandLines(string $inputFile, array $mapping, string $outputFolder = null): array |
|
330
|
|
|
{ |
|
331
|
|
|
$commandLines = []; |
|
332
|
|
|
$esc = $this->escaper; |
|
333
|
|
|
|
|
334
|
|
|
foreach ($mapping as $filename => $pages) { |
|
335
|
|
|
if ($outputFolder) { |
|
336
|
|
|
$target = sprintf('%s/%s', $outputFolder, $filename); |
|
337
|
|
|
} else { |
|
338
|
|
|
$target = $filename; |
|
339
|
|
|
} |
|
340
|
|
|
|
|
341
|
|
|
$commandLines[] = sprintf( |
|
342
|
|
|
'%s %s cat %s output %s', |
|
343
|
|
|
$this->getBinary(), |
|
344
|
|
|
$esc->shellArg($inputFile), |
|
345
|
|
|
implode(' ', $pages), |
|
346
|
|
|
$esc->shellArg($target) |
|
347
|
|
|
); |
|
348
|
|
|
} |
|
349
|
|
|
|
|
350
|
|
|
return $commandLines; |
|
351
|
|
|
} |
|
352
|
|
|
|
|
353
|
|
|
/** |
|
354
|
|
|
* Imports bookmarks from a pdftk dump. |
|
355
|
|
|
*/ |
|
356
|
|
|
private function importBookmarksFromDump(Bookmarks $bookmarks, string $dump): self |
|
357
|
|
|
{ |
|
358
|
|
|
$matches = []; |
|
359
|
|
|
$regex = '/BookmarkBegin\nBookmarkTitle: (?<title>.+)\n' . |
|
360
|
|
|
'BookmarkLevel: (?<level>[0-9]+)\nBookmarkPageNumber: (?<page>[0-9]+)/'; |
|
361
|
|
|
preg_match_all($regex, $dump, $matches, PREG_SET_ORDER); |
|
362
|
|
|
|
|
363
|
|
|
foreach ($matches as $bm) { |
|
364
|
|
|
$bookmark = new Bookmark(); |
|
365
|
|
|
$bookmark |
|
366
|
|
|
->setTitle($bm['title']) |
|
367
|
|
|
->setPageNumber((int) $bm['page']) |
|
368
|
|
|
->setLevel((int) $bm['level']) |
|
369
|
|
|
; |
|
370
|
|
|
|
|
371
|
|
|
$bookmarks->add($bookmark); |
|
372
|
|
|
} |
|
373
|
|
|
|
|
374
|
|
|
return $this; |
|
375
|
|
|
} |
|
376
|
|
|
|
|
377
|
|
|
/** |
|
378
|
|
|
* Builds an Bookmark string for all bookmarks. |
|
379
|
|
|
*/ |
|
380
|
|
|
private function buildBookmarksBlock(Bookmarks $bookmarks): string |
|
381
|
|
|
{ |
|
382
|
|
|
$result = ''; |
|
383
|
|
|
|
|
384
|
|
|
foreach ($bookmarks->all() as $bookmark) { |
|
385
|
|
|
$result .= $this->buildBookmarkBlock($bookmark); |
|
386
|
|
|
} |
|
387
|
|
|
|
|
388
|
|
|
return $result; |
|
389
|
|
|
} |
|
390
|
|
|
|
|
391
|
|
|
/** |
|
392
|
|
|
* Builds an Bookmark string for a single bookmark. |
|
393
|
|
|
*/ |
|
394
|
|
|
private function buildBookmarkBlock(Bookmark $bookmark): string |
|
395
|
|
|
{ |
|
396
|
|
|
return |
|
397
|
|
|
'BookmarkBegin' . PHP_EOL . |
|
398
|
|
|
'BookmarkTitle: ' . $bookmark->getTitle() . PHP_EOL . |
|
399
|
|
|
'BookmarkLevel: ' . $bookmark->getLevel() . PHP_EOL . |
|
400
|
|
|
'BookmarkPageNumber: ' . $bookmark->getPageNumber() . PHP_EOL; |
|
401
|
|
|
} |
|
402
|
|
|
|
|
403
|
|
|
/** |
|
404
|
|
|
* Imports page meta data from a pdftk dump. |
|
405
|
|
|
*/ |
|
406
|
|
|
public function importPagesFromDump(Pages $pages, string $dump): self |
|
407
|
|
|
{ |
|
408
|
|
|
$matches = []; |
|
409
|
|
|
$regex = '/PageMediaBegin\nPageMediaNumber: (?<page>.+)\nPageMediaRotation: (?<rotation>[0-9]+)\n' . |
|
410
|
|
|
'PageMediaRect: .*\n' . |
|
411
|
|
|
'PageMediaDimensions: (?<dim>(([0-9]\,)?[0-9]+(\.[0-9]+)?) (([0-9]\,)?[0-9]+(\.[0-9]+)?))/'; |
|
412
|
|
|
preg_match_all($regex, $dump, $matches, PREG_SET_ORDER); |
|
413
|
|
|
|
|
414
|
|
|
foreach ($matches as $p) { |
|
415
|
|
|
$page = new Page(); |
|
416
|
|
|
|
|
417
|
|
|
$dimensions = explode(' ', $p['dim']); |
|
418
|
|
|
|
|
419
|
|
|
$page |
|
420
|
|
|
->setPageNumber((int) $p['page']) |
|
421
|
|
|
->setRotation((int) $p['rotation']) |
|
422
|
|
|
->setWidth((float) str_replace(',', '', $dimensions[0])) |
|
423
|
|
|
->setHeight((float) str_replace(',', '', $dimensions[1])) |
|
424
|
|
|
; |
|
425
|
|
|
|
|
426
|
|
|
$pages->add($page); |
|
427
|
|
|
} |
|
428
|
|
|
|
|
429
|
|
|
return $this; |
|
430
|
|
|
} |
|
431
|
|
|
|
|
432
|
|
|
/** |
|
433
|
|
|
* Imports PDF metadata from a pdftk dump. |
|
434
|
|
|
*/ |
|
435
|
|
|
public function importMetadataFromDump(Metadata $metadata, string $dump): self |
|
436
|
|
|
{ |
|
437
|
|
|
$matches = []; |
|
438
|
|
|
$regex = '/InfoBegin??\r?\nInfoKey: (?<key>.*)??\r?\nInfoValue: (?<value>.*)??\r?\n/'; |
|
439
|
|
|
preg_match_all($regex, $dump, $matches, PREG_SET_ORDER); |
|
440
|
|
|
|
|
441
|
|
|
foreach ($matches as $meta) { |
|
442
|
|
|
$metadata->set($meta['key'], $meta['value']); |
|
443
|
|
|
} |
|
444
|
|
|
|
|
445
|
|
|
return $this; |
|
446
|
|
|
} |
|
447
|
|
|
|
|
448
|
|
|
/** |
|
449
|
|
|
* Builds an Metadata string for all metadata entries. |
|
450
|
|
|
*/ |
|
451
|
|
|
public function buildMetadataBlock(Metadata $metadata): string |
|
452
|
|
|
{ |
|
453
|
|
|
$result = ''; |
|
454
|
|
|
|
|
455
|
|
|
foreach ($metadata->all() as $key => $value) { |
|
456
|
|
|
$result .= 'InfoBegin' . PHP_EOL; |
|
457
|
|
|
$result .= 'InfoKey: ' . $key . PHP_EOL; |
|
458
|
|
|
$result .= 'InfoValue: ' . (string) $value . PHP_EOL; |
|
459
|
|
|
} |
|
460
|
|
|
|
|
461
|
|
|
return $result; |
|
462
|
|
|
} |
|
463
|
|
|
} |
|
464
|
|
|
|