|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Spatie\PdfToText; |
|
4
|
|
|
|
|
5
|
|
|
use Spatie\PdfToText\Exceptions\CouldNotExtractText; |
|
6
|
|
|
use Spatie\PdfToText\Exceptions\PdfNotFound; |
|
7
|
|
|
use Symfony\Component\Process\Process; |
|
8
|
|
|
|
|
9
|
|
|
class Pdf |
|
10
|
|
|
{ |
|
11
|
|
|
protected $pdf; |
|
12
|
|
|
|
|
13
|
|
|
protected $binPath; |
|
14
|
|
|
|
|
15
|
|
|
protected $options = []; |
|
16
|
|
|
|
|
17
|
|
|
public function __construct(string $binPath = null) |
|
18
|
|
|
{ |
|
19
|
|
|
$this->binPath = $binPath ?? '/usr/bin/pdftotext'; |
|
20
|
|
|
} |
|
21
|
|
|
|
|
22
|
|
|
public function setPdf(string $pdf) : self |
|
23
|
|
|
{ |
|
24
|
|
|
if (!is_readable($pdf)) { |
|
25
|
|
|
throw new PdfNotFound(sprintf('could not find or read pdf `%s`', $pdf)); |
|
26
|
|
|
} |
|
27
|
|
|
|
|
28
|
|
|
$this->pdf = $pdf; |
|
29
|
|
|
|
|
30
|
|
|
return $this; |
|
31
|
|
|
} |
|
32
|
|
|
|
|
33
|
|
|
public function setOptions(array $options) : self |
|
34
|
|
|
{ |
|
35
|
|
|
$this->options = $this->parseOptions($options); |
|
36
|
|
|
|
|
37
|
|
|
return $this; |
|
38
|
|
|
} |
|
39
|
|
|
|
|
40
|
|
|
public function addOptions(array $options) : self |
|
41
|
|
|
{ |
|
42
|
|
|
$this->options = array_merge( |
|
43
|
|
|
$this->options, |
|
44
|
|
|
$this->parseOptions($options) |
|
45
|
|
|
); |
|
46
|
|
|
|
|
47
|
|
|
return $this; |
|
48
|
|
|
} |
|
49
|
|
|
|
|
50
|
|
|
protected function parseOptions(array $options) : array |
|
51
|
|
|
{ |
|
52
|
|
|
$mapper = function (string $content) : array { |
|
53
|
|
|
$content = trim($content); |
|
54
|
|
|
if ('-' !== ($content[0] ?? '')) { |
|
55
|
|
|
$content = '-'.$content; |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
|
|
return explode(' ', $content, 2); |
|
59
|
|
|
}; |
|
60
|
|
|
|
|
61
|
|
|
$reducer = function (array $carry, array $option) : array { |
|
62
|
|
|
return array_merge($carry, $option); |
|
63
|
|
|
}; |
|
64
|
|
|
|
|
65
|
|
|
return array_reduce(array_map($mapper, $options), $reducer, []); |
|
66
|
|
|
} |
|
67
|
|
|
|
|
68
|
|
|
public function text() : string |
|
69
|
|
|
{ |
|
70
|
|
|
$process = new Process(array_merge([$this->binPath], $this->options, [$this->pdf, '-'])); |
|
71
|
|
|
$process->run(); |
|
72
|
|
|
if (!$process->isSuccessful()) { |
|
73
|
|
|
throw new CouldNotExtractText($process); |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
return trim($process->getOutput(), " \t\n\r\0\x0B\x0C"); |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
|
|
public static function getText(string $pdf, string $binPath = null, array $options = []) : string |
|
80
|
|
|
{ |
|
81
|
|
|
return (new static($binPath)) |
|
82
|
|
|
->setOptions($options) |
|
83
|
|
|
->setPdf($pdf) |
|
84
|
|
|
->text() |
|
85
|
|
|
; |
|
86
|
|
|
} |
|
87
|
|
|
} |
|
88
|
|
|
|
spatie /
pdf-to-text
