|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Bicycle\Tesseract\Bridge; |
|
4
|
|
|
|
|
5
|
|
|
use Bicycle\Tesseract\BridgeInterface; |
|
6
|
|
|
|
|
7
|
|
|
/** |
|
8
|
|
|
* Please note that here we have \FFI class instance instead of FFI\TesseractInterface. |
|
9
|
|
|
*/ |
|
10
|
|
|
class FFI implements BridgeInterface |
|
11
|
|
|
{ |
|
12
|
|
|
/** @var Configuration */ |
|
13
|
|
|
private Configuration $configuration; |
|
14
|
|
|
|
|
15
|
|
|
/** |
|
16
|
|
|
* I have to use interface here, but actually we have here \FFI class instance. Do not set type for this property! |
|
17
|
|
|
* |
|
18
|
|
|
* @var FFI\TesseractInterface |
|
19
|
|
|
*/ |
|
20
|
|
|
private $ffiInstance; |
|
21
|
|
|
|
|
22
|
|
|
/** |
|
23
|
|
|
* {@inheritDoc} |
|
24
|
|
|
*/ |
|
25
|
|
|
public function __construct(Configuration $configuration) |
|
26
|
|
|
{ |
|
27
|
|
|
if (!extension_loaded('ffi')) { |
|
28
|
|
|
throw new Exception\ExtensionRequiredException('FFI extension is required for this functionality'); |
|
29
|
|
|
} |
|
30
|
|
|
// Prevent configuration change in runtime |
|
31
|
|
|
$this->configuration = clone $configuration; |
|
32
|
|
|
$headerPath = $this->configuration->getCApiHeaderpath(); |
|
33
|
|
|
if ( |
|
34
|
|
|
empty($headerPath) || |
|
35
|
|
|
empty($definitions = file_get_contents($headerPath)) |
|
36
|
|
|
) { |
|
37
|
|
|
throw new Exception\Exception('Cannot use FFI without valid header file'); |
|
38
|
|
|
} |
|
39
|
|
|
try { |
|
40
|
|
|
$libaryPath = $this->configuration->getSharedLibraryPath(); |
|
41
|
|
|
if (empty($libaryPath)) { |
|
42
|
|
|
throw new Exception\Exception('Problem with connecting library via FFI: empty library path'); |
|
43
|
|
|
} |
|
44
|
|
|
/** @var FFI\TesseractInterface ffiInstance */ |
|
45
|
|
|
$this->ffiInstance = \FFI::cdef( |
|
46
|
|
|
$definitions, |
|
47
|
|
|
$libaryPath |
|
48
|
|
|
); |
|
49
|
|
|
} catch (\FFI\Exception $e) { |
|
50
|
|
|
throw new Exception\Exception(sprintf('Problem with connecting library via FFI: %s', $e->getMessage())); |
|
51
|
|
|
} |
|
52
|
|
|
} |
|
53
|
|
|
|
|
54
|
|
|
/** |
|
55
|
|
|
* {@inheritDoc} |
|
56
|
|
|
*/ |
|
57
|
|
|
public function getVersion(): string |
|
58
|
|
|
{ |
|
59
|
|
|
return $this->ffiInstance->TessVersion(); |
|
60
|
|
|
} |
|
61
|
|
|
|
|
62
|
|
|
/** |
|
63
|
|
|
* {@inheritDoc} |
|
64
|
|
|
*/ |
|
65
|
|
|
public function getAvailableLanguages(): array |
|
66
|
|
|
{ |
|
67
|
|
|
$result = []; |
|
68
|
|
|
/** |
|
69
|
|
|
* @psalm-suppress MixedAssignment |
|
70
|
|
|
*/ |
|
71
|
|
|
$baseApiHandle = $this->ffiInstance->TessBaseAPICreate(); |
|
72
|
|
|
$initFailed = $this->ffiInstance->TessBaseAPIInit3($baseApiHandle, null, null); // Tesseract initialization |
|
73
|
|
|
if ($initFailed) { |
|
74
|
|
|
$this->ffiInstance->TessBaseAPIDelete($baseApiHandle); |
|
75
|
|
|
throw new Exception\Exception('Cannot initialize tesseract'); |
|
76
|
|
|
} |
|
77
|
|
|
$languages = $this->ffiInstance->TessBaseAPIGetAvailableLanguagesAsVector($baseApiHandle); |
|
78
|
|
|
$counter = 0; |
|
79
|
|
|
// According to body of TessBaseAPIGetAvailableLanguagesAsVector method, last element will be nullptr |
|
80
|
|
|
while (!\is_null($languages[$counter])) { |
|
81
|
|
|
/** @psalm-suppress MixedAssignment */ |
|
82
|
|
|
$result[] = \FFI::string($languages[$counter++]); |
|
83
|
|
|
} |
|
84
|
|
|
$this->ffiInstance->TessBaseAPIEnd($baseApiHandle); |
|
85
|
|
|
$this->ffiInstance->TessBaseAPIDelete($baseApiHandle); |
|
86
|
|
|
|
|
87
|
|
|
return $result; |
|
88
|
|
|
} |
|
89
|
|
|
|
|
90
|
|
|
/** |
|
91
|
|
|
* {@inheritDoc} |
|
92
|
|
|
*/ |
|
93
|
|
|
public function recognizeFromFile(string $filename, array $languages = []): string |
|
94
|
|
|
{ |
|
95
|
|
|
if (!\is_readable($filename)) { |
|
96
|
|
|
throw new Exception\InputProblemException('Cannot read input file'); |
|
97
|
|
|
} |
|
98
|
|
|
if (empty($languages)) { |
|
99
|
|
|
$languages[] = 'eng'; |
|
100
|
|
|
} elseif ( |
|
101
|
|
|
count($intersection = array_intersect($languages, $this->getAvailableLanguages())) !== count($languages) |
|
102
|
|
|
) { |
|
103
|
|
|
$exceptionMessage = sprintf( |
|
104
|
|
|
'Unknown language(s) %s for recognition.', |
|
105
|
|
|
implode(', ', array_diff($languages, $intersection)) |
|
106
|
|
|
); |
|
107
|
|
|
throw new Exception\UnavailableLanguageException($exceptionMessage); |
|
108
|
|
|
} |
|
109
|
|
|
|
|
110
|
|
|
$resultText = ''; |
|
111
|
|
|
|
|
112
|
|
|
/** |
|
113
|
|
|
* @psalm-suppress MixedAssignment |
|
114
|
|
|
*/ |
|
115
|
|
|
$baseApiHandle = $this->ffiInstance->TessBaseAPICreate(); |
|
116
|
|
|
$initFailed = $this->ffiInstance->TessBaseAPIInit3( |
|
117
|
|
|
$baseApiHandle, |
|
118
|
|
|
null, |
|
119
|
|
|
implode('+', $languages) |
|
120
|
|
|
); // Tesseract initialization |
|
121
|
|
|
if ($initFailed) { |
|
122
|
|
|
$this->ffiInstance->TessBaseAPIDelete($baseApiHandle); |
|
123
|
|
|
throw new Exception\Exception('Cannot initialize tesseract'); |
|
124
|
|
|
} |
|
125
|
|
|
|
|
126
|
|
|
if ($this->ffiInstance->TessBaseAPIProcessPages($baseApiHandle, $filename, null, 0, null)) { |
|
127
|
|
|
$resultText = $this->ffiInstance->TessBaseAPIGetUTF8Text($baseApiHandle); |
|
128
|
|
|
/** @var string $resultText */ |
|
129
|
|
|
$resultText = \FFI::string($resultText); |
|
130
|
|
|
} |
|
131
|
|
|
|
|
132
|
|
|
$this->ffiInstance->TessBaseAPIEnd($baseApiHandle); |
|
133
|
|
|
$this->ffiInstance->TessBaseAPIDelete($baseApiHandle); |
|
134
|
|
|
|
|
135
|
|
|
return $resultText; |
|
136
|
|
|
} |
|
137
|
|
|
} |
|
138
|
|
|
|
vkhramtsov /
tesseract-bridge
