Passed
Push — master ( 14adf3...a19d55 )
by Konrad
02:27
created

PDFObjectTest::testFormatContentInlineImages()   A

Complexity

Conditions 1
Paths 1

Size

Total Lines 30
Code Lines 17

Duplication

Lines 0
Ratio 0 %

Importance

Changes 1
Bugs 0 Features 0
Metric Value
cc 1
eloc 17
c 1
b 0
f 0
nc 1
nop 0
dl 0
loc 30
rs 9.7
1
<?php
2
3
/**
4
 * @file This file is part of the PdfParser library.
5
 *
6
 * @author  Konrad Abicht <[email protected]>
7
 *
8
 * @date    2020-06-01
9
 *
10
 * @author  Sébastien MALOT <[email protected]>
11
 *
12
 * @date    2017-01-03
13
 *
14
 * @license LGPLv3
15
 *
16
 * @url     <https://github.com/smalot/pdfparser>
17
 *
18
 *  PdfParser is a pdf library written in PHP, extraction oriented.
19
 *  Copyright (C) 2017 - Sébastien MALOT <[email protected]>
20
 *
21
 *  This program is free software: you can redistribute it and/or modify
22
 *  it under the terms of the GNU Lesser General Public License as published by
23
 *  the Free Software Foundation, either version 3 of the License, or
24
 *  (at your option) any later version.
25
 *
26
 *  This program is distributed in the hope that it will be useful,
27
 *  but WITHOUT ANY WARRANTY; without even the implied warranty of
28
 *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
29
 *  GNU Lesser General Public License for more details.
30
 *
31
 *  You should have received a copy of the GNU Lesser General Public License
32
 *  along with this program.
33
 *  If not, see <http://www.pdfparser.org/sites/default/LICENSE.txt>.
34
 */
35
36
namespace PHPUnitTests\Integration;
37
38
use PHPUnitTests\TestCase;
39
use Smalot\PdfParser\Document;
40
use Smalot\PdfParser\PDFObject;
41
42
class PDFObjectTest extends TestCase
43
{
44
    public const TYPE = 't';
45
46
    public const OPERATOR = 'o';
47
48
    public const COMMAND = 'c';
49
50
    protected function getPdfObjectInstance($document): PDFObject
51
    {
52
        return new PDFObject($document);
53
    }
54
55
    public function testGetCommandsText(): void
56
    {
57
        $content = "BT /R14 30 Tf 0.999016 0 0 1 137.4
58
342.561 Tm
59
[(A)-168.854( BC D)-220.905(\\(E\\))20.905<20>]
60
TJ /R14 17.16 Tf <20> Tj
61
0.999014 0 0 1 336.84 319.161 Tm T* ( \x00m)Tj
62
/R14 20.04 Tf
63
ET Q
64
q -124.774 124.127 5.64213 5.67154 930.307 4436.95 cm
65
BI";
66
67
        $sections = $this->getPdfObjectInstance(new Document())->getSectionsText($content);
68
69
        $offset = 0;
0 ignored issues
show
Unused Code introduced by
The assignment to $offset is dead and can be removed.
Loading history...
70
        $parts = [];
71
        foreach ($sections as $section) {
72
            $parts[] = $this->getPdfObjectInstance(new Document())->getCommandsText($section)[0];
73
        }
74
75
        $reference = [
76
            [
77
                self::TYPE => '',
78
                self::OPERATOR => 'BT',
79
                self::COMMAND => '',
80
            ],
81
            [
82
                self::TYPE => '/',
83
                self::OPERATOR => 'Tf',
84
                self::COMMAND => 'R14 30',
85
            ],
86
            [
87
                self::TYPE => '',
88
                self::OPERATOR => 'Tm',
89
                self::COMMAND => '0.999016 0 0 1 137.4 342.561',
90
            ],
91
            [
92
                self::TYPE => '[',
93
                self::OPERATOR => 'TJ',
94
                self::COMMAND => [
95
                    [
96
                        self::TYPE => '(',
97
                        self::OPERATOR => 'TJ',
98
                        self::COMMAND => 'A',
99
                    ],
100
                    [
101
                        self::TYPE => 'n',
102
                        self::OPERATOR => '',
103
                        self::COMMAND => '-168.854',
104
                    ],
105
                    [
106
                        self::TYPE => '(',
107
                        self::OPERATOR => 'TJ',
108
                        self::COMMAND => ' BC D',
109
                    ],
110
                    [
111
                        self::TYPE => 'n',
112
                        self::OPERATOR => '',
113
                        self::COMMAND => '-220.905',
114
                    ],
115
                    [
116
                        self::TYPE => '(',
117
                        self::OPERATOR => 'TJ',
118
                        self::COMMAND => '\\(E\\)',
119
                    ],
120
                    [
121
                        self::TYPE => 'n',
122
                        self::OPERATOR => '',
123
                        self::COMMAND => '20.905',
124
                    ],
125
                    [
126
                        self::TYPE => '<',
127
                        self::OPERATOR => 'TJ',
128
                        self::COMMAND => '20',
129
                    ],
130
                ],
131
            ],
132
            [
133
                self::TYPE => '/',
134
                self::OPERATOR => 'Tf',
135
                self::COMMAND => 'R14 17.16',
136
            ],
137
            [
138
                self::TYPE => '<',
139
                self::OPERATOR => 'Tj',
140
                self::COMMAND => '20',
141
            ],
142
            [
143
                self::TYPE => '',
144
                self::OPERATOR => 'Tm',
145
                self::COMMAND => '0.999014 0 0 1 336.84 319.161',
146
            ],
147
            [
148
                self::TYPE => '',
149
                self::OPERATOR => 'T*',
150
                self::COMMAND => '',
151
            ],
152
            [
153
                self::TYPE => '(',
154
                self::OPERATOR => 'Tj',
155
                self::COMMAND => " \x00m",
156
            ],
157
            [
158
                self::TYPE => '/',
159
                self::OPERATOR => 'Tf',
160
                self::COMMAND => 'R14 20.04',
161
            ],
162
            [
163
                self::TYPE => '',
164
                self::OPERATOR => 'ET',
165
                self::COMMAND => '',
166
            ],
167
            [
168
                self::TYPE => '',
169
                self::OPERATOR => 'Q',
170
                self::COMMAND => '',
171
            ],
172
            [
173
                self::TYPE => '',
174
                self::OPERATOR => 'q',
175
                self::COMMAND => '',
176
            ],
177
            [
178
                self::TYPE => '',
179
                self::OPERATOR => 'cm',
180
                self::COMMAND => '-124.774 124.127 5.64213 5.67154 930.307 4436.95',
181
            ],
182
        ];
183
184
        $this->assertEquals($parts, $reference);
185
    }
186
187
    public function testCleanContent(): void
188
    {
189
        $content = '/Shape <</MCID << /Font<8>>> BT >>BDC
190
Q
191
/CS0 cs 1 1 0  scn
192
1 i
193
/GS0 gs
194
BT
195
/TT0 1 Tf
196
0.0007 Tc 0.0018 Tw 0  Ts 100  Tz 0 Tr 24 0 0 24 51.3 639.26025 Tm
197
(Modificatio[ns] au \\(14\\) septembre 2009 ET 2010)Tj
198
EMC
199
(ABC) Tj
200
201
[ (a)-4.5(b)6(c)8.8 ( fsdfsdfsdf[]sd) ] TD
202
203
ET
204
/Shape <</MCID 2 >>BDC
205
q
206
0.03 841';
207
208
        $expected = '_____________________________________
209
Q
210
/CS0 cs 1 1 0  scn
211
1 i
212
/GS0 gs
213
BT
214
/TT0 1 Tf
215
0.0007 Tc 0.0018 Tw 0  Ts 100  Tz 0 Tr 24 0 0 24 51.3 639.26025 Tm
216
(________________________________________________)Tj
217
___
218
(___) Tj
219
220
[_____________________________________] TD
221
222
ET
223
______________________
224
q
225
0.03 841';
226
227
        $cleaned = $this->getPdfObjectInstance(new Document())->cleanContent($content, '_');
0 ignored issues
show
Deprecated Code introduced by
The function Smalot\PdfParser\PDFObject::cleanContent() has been deprecated: function is no longer used and will be removed in a future release ( Ignorable by Annotation )

If this is a false-positive, you can also ignore this issue in your code via the ignore-deprecated  annotation

227
        $cleaned = /** @scrutinizer ignore-deprecated */ $this->getPdfObjectInstance(new Document())->cleanContent($content, '_');

This function has been deprecated. The supplier of the function has supplied an explanatory message.

The explanatory message should give you some clue as to whether and when the function will be removed and what other function to use instead.

Loading history...
228
229
        $this->assertEquals($cleaned, $expected);
230
    }
231
232
    public function testFormatContent(): void
233
    {
234
        $content = '/Shape <</MCID << /Font<8>>> BT >>BDC Q /CS0 cs 1 1 0  scn 1 i
235
/GS0 gs BT /TT0 1 Tf 0.0007 Tc 0.0018 Tw 0  Ts 100  Tz 0 Tr 24 0 0 24 51.3 639.26025 Tm
236
(Modificatio[ns] au \\(14\\) septembre 2009 ET 2010)Tj EMC (ABC) Tj
237
[ (a)-4.5(b)6(c)8.8 ( fsdfsdfsdf[]sd) ] TJ ET /Shape <</MCID 2 >>BDC q 0.03 841';
238
239
        $expected = '/Shape <</MCID << /Font<8>>> BT >>BDC
240
Q
241
/CS0 cs
242
1 1 0 scn
243
1 i
244
/GS0 gs
245
BT
246
/TT0 1 Tf
247
0.0007 Tc
248
0.0018 Tw
249
0 Ts
250
100 Tz
251
0 Tr
252
24 0 0 24 51.3 639.26025 Tm
253
(Modificatio[ns] au \\(14\\) septembre 2009 ET 2010)Tj
254
EMC
255
(ABC) Tj
256
[ (a)-4.5(b)6(c)8.8 ( fsdfsdfsdf[]sd) ] TJ
257
ET
258
/Shape <</MCID 2 >>BDC
259
q
260
0.03 841';
261
262
        // Normalize line-endings
263
        $expected = str_replace(["\r\n", "\n"], ["\n", "\r\n"], $expected);
264
265
        $formatContent = new \ReflectionMethod('Smalot\PdfParser\PDFObject', 'formatContent');
266
        $formatContent->setAccessible(true);
267
        $cleaned = $formatContent->invoke($this->getPdfObjectInstance(new Document()), $content);
268
269
        $this->assertEquals($expected, $cleaned);
270
271
        // Check that binary data is rejected
272
        $content = hex2bin('a670c89d4a324e47');
273
274
        $cleaned = $formatContent->invoke($this->getPdfObjectInstance(new Document()), $content);
275
276
        $this->assertEquals('', $cleaned);
277
278
        // See: https://github.com/smalot/pdfparser/issues/668
279
        $filename = $this->rootDir.'/samples/bugs/Issue668.pdf';
280
281
        $parser = $this->getParserInstance();
282
        $document = $parser->parseFile($filename);
283
        $pages = $document->getPages();
284
285
        // Binary check is done before a regexp that causes an error
286
        $this->assertStringContainsString('Marko Nestorović PR', $pages[0]->getText());
287
288
        // mb_check_encoding(..., 'UTF-8') returns true here,
289
        // necessitating a test for UTF-8 that's more strict
290
        $content = hex2bin('0101010101010101');
291
        $cleaned = $formatContent->invoke($this->getPdfObjectInstance(new Document()), $content);
292
293
        $this->assertEquals('', $cleaned);
294
    }
295
296
    /**
297
     * Check that inline image data does not corrupt the stream
298
     *
299
     * @see: https://github.com/smalot/pdfparser/issues/691
300
     */
301
    public function testFormatContentInlineImages(): void
302
    {
303
        $formatContent = new \ReflectionMethod('Smalot\PdfParser\PDFObject', 'formatContent');
304
        $formatContent->setAccessible(true);
305
306
        $cleaned = $formatContent->invoke(
307
            $this->getPdfObjectInstance(new Document()),
308
            'BT (This BI /W 258 /H 51 /should not trigger /as a /PDF command) TD ET q 65.30 0 0 18.00 412 707 cm BI /W 544 /H 150
309
/BPC 1 /IM true /F [/A85 /Fl] ID Gb"0F_$L6!$j/a\$:ma&h\'JnJJ9S?O_EA-W+%D^ClCH=FP3s5M-gStQm\'5/hc`C?<Q)riWgtEe:Po0dY_-er6$jM@#?n`E+#(sa"0Gk3&K>CqL(^pV$_-er6Ik`"-1]Q ;~> EI Q /F002 10.00 Tf 0.00 Tw 0 g'
310
        );
311
312
        // PdfParser should not be fooled by Q's in inline image data;
313
        // Only one 'Q' command should be found
314
        $commandQ = preg_match_all('/Q\r\n/', $cleaned);
315
        $this->assertEquals(1, $commandQ);
316
317
        // The 'BI' inside a string should not be interpreted as the
318
        // beginning of an inline image command
319
        $this->assertStringContainsString('(This BI /W 258 /H 51 /should not trigger /as a /PDF command) TD', $cleaned);
320
321
        $cleaned = $formatContent->invoke(
322
            $this->getPdfObjectInstance(new Document()),
323
            'BT (This BI /W 258 /H 51 /should not () \) trigger /as a /PDF command) TD (There is no ID inline image in this data) TD (Nothing but text EI should be found) TD ET'
324
        );
325
326
        $this->assertEquals('BT'."\r\n".
327
'(This BI /W 258 /H 51 /should not () \) trigger /as a /PDF command) TD'."\r\n".
328
'(There is no ID inline image in this data) TD'."\r\n".
329
'(Nothing but text EI should be found) TD'."\r\n".
330
'ET', $cleaned);
331
    }
332
333
    public function testGetSectionsText(): void
334
    {
335
        $content = '/Shape <</MCID 1 >>BDC
336
Q
337
/CS0 cs 1 1 0  scn
338
1 i
339
/GS0 gs
340
BT
341
/TT0 1 Tf
342
0.0007 Tc 0.0018 Tw 0  Ts 100  Tz 0 Tr 24 0 0 24 51.3 639.26025 Tm
343
(Mod BT atio[ns] au \\(14\\) septembre 2009 ET 2010)Tj
344
EMC
345
(ABC) Tj
346
347
[ (a)-4.5(b) 6(c)8.8 ( fsdfsdfsdf[ sd) ] TD
348
349
ET
350
/Shape <</MCID [BT] >>BDC BT /TT1 1.5 Tf (BT )Tj ET
351
q
352
0.03 841';
353
354
        $sections = $this->getPdfObjectInstance(new Document())->getSectionsText($content);
355
356
        $this->assertEquals(
357
            [
358
                '/Shape <</MCID 1 >>BDC',
359
                'Q',
360
                'BT',
361
                '/TT0 1 Tf',
362
                '0.0007 Tc',
363
                '0.0018 Tw',
364
                '0 Ts',
365
                '100 Tz',
366
                '0 Tr',
367
                '24 0 0 24 51.3 639.26025 Tm',
368
                '(Mod BT atio[ns] au \\(14\\) septembre 2009 ET 2010)Tj',
369
                'EMC',
370
                '(ABC) Tj',
371
                '[ (a)-4.5(b) 6(c)8.8 ( fsdfsdfsdf[ sd) ] TD',
372
                'ET',
373
                '/Shape <</MCID [BT] >>BDC',
374
                'BT',
375
                '/TT1 1.5 Tf',
376
                '(BT )Tj',
377
                'ET',
378
                'q',
379
            ],
380
            $sections
381
        );
382
383
        // Test that a Name containing 'ET' doesn't close a 'BT' block
384
        // See: https://github.com/smalot/pdfparser/issues/474
385
        $content = 'BT
386
/FTxkPETkkj 8 Tf
387
1 0 0 1 535.55 627.4 Tm
388
(Hello World)TJ
389
ET';
390
391
        $sections = $this->getPdfObjectInstance(new Document())->getSectionsText($content);
392
393
        $this->assertNotEquals('/FTxkP', $sections[0]);
394
        $this->assertNotEquals('/FTxkP', $sections[1]);
395
    }
396
397
    public function testParseDictionary(): void
398
    {
399
        $data = '<</ActualText(text)/XObject<</F2 6 0 R /F3 [/Sub /Array]>> /Array[/Parsed /Data/Actual]/Silent<>>>';
400
401
        $dictionary = $this->getPdfObjectInstance(new Document())->parseDictionary($data);
402
403
        $this->assertArrayHasKey('ActualText', $dictionary);
404
        $this->assertArrayHasKey('XObject', $dictionary);
405
        $this->assertArrayHasKey('Array', $dictionary);
406
        $this->assertArrayHasKey('Silent', $dictionary);
407
408
        $this->assertCount(3, $dictionary['Array']);
409
410
        $this->assertEquals('<>', $dictionary['Silent']);
411
    }
412
413
    /**
414
     * Tests that graphics position (cm) is taken into account when
415
     * positioning text
416
     *
417
     * @see: https://github.com/smalot/pdfparser/issues/608
418
     */
419
    public function testGraphicsPositioning(): void
420
    {
421
        $filename = $this->rootDir.'/samples/bugs/Issue608.pdf';
422
423
        $parser = $this->getParserInstance();
424
        $document = $parser->parseFile($filename);
425
        $pages = $document->getPages();
426
427
        // The \n is not added if 'cm' commands are ignored
428
        $this->assertStringContainsString("Heading 1 \nLorem ipsum", $pages[0]->getText());
429
    }
430
431
    /**
432
     * Tests that ActualText text is printed for a block instead of the
433
     * contents of the Tj or TJ commands in the block.
434
     *
435
     * @see: https://github.com/smalot/pdfparser/issues/464
436
     */
437
    public function testActualText(): void
438
    {
439
        $filename = $this->rootDir.'/samples/bugs/Issue608.pdf';
440
441
        $parser = $this->getParserInstance();
442
        $document = $parser->parseFile($filename);
443
        $pages = $document->getPages();
444
445
        // An ActualText command subs in the three literal characters
446
        // 'ffi' for the single character ligature here
447
        // In addition, if $last_written_position isn't used to store
448
        // the position to insert, \n's would be erroniously inserted
449
        // on either side of the 'ffi'
450
        $this->assertStringContainsString('efficitur', $pages[0]->getText());
451
    }
452
453
    /**
454
     * Tests for the correct decoding of an Em-dash character in
455
     * certain font contexts
456
     *
457
     * See: https://github.com/smalot/pdfparser/issues/585
458
     */
459
    public function testDecodeEmDash(): void
460
    {
461
        $filename = $this->rootDir.'/samples/bugs/Issue585.pdf';
462
463
        $parser = $this->getParserInstance();
464
        $document = $parser->parseFile($filename);
465
        $pages = $document->getPages();
466
467
        $this->assertStringContainsString('слева по ходу — веревка', $pages[0]->getText());
468
    }
469
470
    /**
471
     * Tests behavior with reversed chars instruction.
472
     *
473
     * @see: https://github.com/smalot/pdfparser/issues/398
474
     */
475
    public function testReversedChars(): void
476
    {
477
        $filename = $this->rootDir.'/samples/bugs/Issue398.pdf';
478
479
        $parser = $this->getParserInstance();
480
        $document = $parser->parseFile($filename);
481
        $pages = $document->getPages();
482
483
        $pageText = $pages[0]->getText();
484
485
        $this->assertStringContainsString('שלומי טסט', $pageText);
486
        $this->assertStringContainsString('בנמל מספנות ישראל.', $pageText);
487
    }
488
489
    /**
490
     * Tests that a text stream with an improperly selected font code
491
     * page falls back to one that maps all characters.
492
     *
493
     * @see: https://github.com/smalot/pdfparser/issues/586
494
     */
495
    public function testImproperFontFallback(): void
496
    {
497
        $filename = $this->rootDir.'/samples/ImproperFontFallback.pdf';
498
499
        $parser = $this->getParserInstance();
500
        $document = $parser->parseFile($filename);
501
        $pages = $document->getPages();
502
503
        $this->assertStringContainsString('сделал', $pages[0]->getText());
504
    }
505
506
    /**
507
     * Tests that a font ID containing a hyphen / dash character was
508
     * correctly parsed
509
     *
510
     * @see: https://github.com/smalot/pdfparser/issues/145
511
     */
512
    public function testFontIDWithHyphen(): void
513
    {
514
        $pdfObject = $this->getPdfObjectInstance(new Document());
515
516
        $fontCommandHyphen = $pdfObject->getCommandsText('/FID-01 15.00 Tf');
517
518
        $this->assertEquals('/', $fontCommandHyphen[0]['t']);
519
        $this->assertEquals('Tf', $fontCommandHyphen[0]['o']);
520
        $this->assertEquals('FID-01 15.00', $fontCommandHyphen[0]['c']);
521
    }
522
523
    /**
524
     * Tests that an invalid command does not cause an error, but just
525
     * returns an empty array
526
     */
527
    public function testInvalidCommand(): void
528
    {
529
        $pdfObject = $this->getPdfObjectInstance(new Document());
530
531
        $validCommand = $pdfObject->getCommandsText('75 rg');
532
533
        $this->assertEquals('', $validCommand[0]['t']);
534
        $this->assertEquals('rg', $validCommand[0]['o']);
535
        $this->assertEquals('75', $validCommand[0]['c']);
536
537
        $invalidCommand = $pdfObject->getCommandsText('75');
538
539
        $this->assertEquals([], $invalidCommand);
540
    }
541
}
542