|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/* |
|
4
|
|
|
* This file is part of the core-library package. |
|
5
|
|
|
* |
|
6
|
|
|
* (c) 2020 WEBEWEB |
|
7
|
|
|
* |
|
8
|
|
|
* For the full copyright and license information, please view the LICENSE |
|
9
|
|
|
* file that was distributed with this source code. |
|
10
|
|
|
*/ |
|
11
|
|
|
|
|
12
|
|
|
namespace WBW\Library\Core\ThirdParty\OcrLad\Serializer; |
|
13
|
|
|
|
|
14
|
|
|
use WBW\Library\Core\ThirdParty\OcrLad\Model\Document; |
|
15
|
|
|
use WBW\Library\Core\ThirdParty\OcrLad\Model\Page; |
|
16
|
|
|
use WBW\Library\Core\ThirdParty\OcrLad\Model\Word; |
|
17
|
|
|
|
|
18
|
|
|
/** |
|
19
|
|
|
* DER deserializer. |
|
20
|
|
|
* |
|
21
|
|
|
* @author webeweb <https://github.com/webeweb> |
|
22
|
|
|
* @package WBW\Library\Core\ThirdParty\OcrLad\Serializer |
|
23
|
|
|
*/ |
|
24
|
|
|
class DerDeserializer { |
|
25
|
|
|
|
|
26
|
|
|
/** |
|
27
|
|
|
* DER delimiter. |
|
28
|
|
|
* |
|
29
|
|
|
* @var string |
|
30
|
|
|
*/ |
|
31
|
|
|
const DER_DELIMITER = ";"; |
|
32
|
|
|
|
|
33
|
|
|
/** |
|
34
|
|
|
* Deserialize a document. |
|
35
|
|
|
* |
|
36
|
|
|
* @param string $filename The filename. |
|
37
|
|
|
* @return Document|null Returns the document in case of success, null otherwise. |
|
38
|
|
|
*/ |
|
39
|
|
|
public static function deserializeDocument($filename) { |
|
40
|
|
|
|
|
41
|
|
|
$model = new Document(); |
|
42
|
|
|
$model->setFilename($filename); |
|
43
|
|
|
|
|
44
|
|
|
$stream = fopen($filename, "r"); |
|
45
|
|
|
|
|
46
|
|
|
$headers = DerDeserializer::processHeaders(fgets($stream)); |
|
47
|
|
|
foreach ($headers as $current) { |
|
48
|
|
|
|
|
49
|
|
|
$page = DerDeserializer::deserializePage($current); |
|
50
|
|
|
if (null !== $page) { |
|
51
|
|
|
$model->addPage($page->setParent($model)); |
|
52
|
|
|
} |
|
53
|
|
|
} |
|
54
|
|
|
|
|
55
|
|
|
while (true !== feof($stream)) { |
|
56
|
|
|
|
|
57
|
|
|
$word = DerDeserializer::deserializeWord(fgets($stream)); |
|
58
|
|
|
if (null !== $word) { |
|
59
|
|
|
$model->addWord($word); |
|
60
|
|
|
} |
|
61
|
|
|
} |
|
62
|
|
|
|
|
63
|
|
|
fclose($stream); |
|
64
|
|
|
|
|
65
|
|
|
return DerDeserializer::processDocument($model); |
|
66
|
|
|
} |
|
67
|
|
|
|
|
68
|
|
|
/** |
|
69
|
|
|
* Deserialize a page. |
|
70
|
|
|
* |
|
71
|
|
|
* @param string $rawData The raw data. |
|
72
|
|
|
* @return Page|null Returns the page in case of success, null otherwise. |
|
73
|
|
|
*/ |
|
74
|
|
|
protected static function deserializePage($rawData) { |
|
75
|
|
|
|
|
76
|
|
|
$data = explode(DerDeserializer::DER_DELIMITER, $rawData); |
|
77
|
|
|
if (6 !== count($data)) { |
|
78
|
|
|
return null; |
|
79
|
|
|
} |
|
80
|
|
|
|
|
81
|
|
|
$model = new Page(); |
|
82
|
|
|
$model->setResolution(intval(preg_replace("/[^0-9]/", "", $data[0]))); |
|
83
|
|
|
$model->setWidth(intval($data[1])); |
|
84
|
|
|
$model->setHeight(intval($data[2])); |
|
85
|
|
|
$model->setRotation(intval($data[3])); |
|
86
|
|
|
$model->setTag(intval($data[4])); |
|
87
|
|
|
|
|
88
|
|
|
return $model; |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
/** |
|
92
|
|
|
* Deserialize a word. |
|
93
|
|
|
* |
|
94
|
|
|
* @param string $rawData The raw data. |
|
95
|
|
|
* @return Word|null Returns the word in case of success, null otherwise. |
|
96
|
|
|
*/ |
|
97
|
|
|
protected static function deserializeWord($rawData) { |
|
98
|
|
|
|
|
99
|
|
|
$data = explode(DerDeserializer::DER_DELIMITER, $rawData); |
|
100
|
|
|
if (7 !== count($data)) { |
|
101
|
|
|
return null; |
|
102
|
|
|
} |
|
103
|
|
|
|
|
104
|
|
|
preg_match("/(LB|NLB).?([0-9]+)/", trim($data[1]), $matches); |
|
105
|
|
|
|
|
106
|
|
|
$model = new Word(); |
|
107
|
|
|
$model->setContent(trim($data[0])); |
|
108
|
|
|
$model->setType($matches[1]); |
|
109
|
|
|
$model->setPage(intval($matches[2])); |
|
110
|
|
|
$model->setOcrConfidence(floatval(str_replace(",", ".", $data[2]))); |
|
111
|
|
|
$model->setX1(floatval($data[3])); |
|
112
|
|
|
$model->setY1(floatval($data[4])); |
|
113
|
|
|
$model->setX2(floatval($data[5])); |
|
114
|
|
|
$model->setY2(floatval($data[6])); |
|
115
|
|
|
|
|
116
|
|
|
return $model; |
|
117
|
|
|
} |
|
118
|
|
|
|
|
119
|
|
|
/** |
|
120
|
|
|
* Process the document. |
|
121
|
|
|
* |
|
122
|
|
|
* @param Document $document The document. |
|
123
|
|
|
* @return Document Returns the document. |
|
124
|
|
|
*/ |
|
125
|
|
|
protected static function processDocument(Document $document) { |
|
126
|
|
|
|
|
127
|
|
|
foreach ($document->getWords() as $current) { |
|
128
|
|
|
|
|
129
|
|
|
$document->index($current); |
|
130
|
|
|
|
|
131
|
|
|
$page = $document->getPage($current->getPage() - 1); |
|
132
|
|
|
$page->addWord($current->setParent($page)); |
|
133
|
|
|
} |
|
134
|
|
|
|
|
135
|
|
|
return $document; |
|
136
|
|
|
} |
|
137
|
|
|
|
|
138
|
|
|
/** |
|
139
|
|
|
* Process the headers. |
|
140
|
|
|
* |
|
141
|
|
|
* @param string $rawData The raw data. |
|
142
|
|
|
* @return string[] Returns the headers. |
|
143
|
|
|
*/ |
|
144
|
|
|
protected static function processHeaders($rawData) { |
|
145
|
|
|
|
|
146
|
|
|
$data = explode(DerDeserializer::DER_DELIMITER, $rawData); |
|
147
|
|
|
if (6 === count($data)) { |
|
148
|
|
|
return [$rawData]; |
|
149
|
|
|
} |
|
150
|
|
|
|
|
151
|
|
|
$rows = []; |
|
152
|
|
|
$buffer = []; |
|
153
|
|
|
|
|
154
|
|
|
foreach ($data as $current) { |
|
155
|
|
|
|
|
156
|
|
|
$buffer[] = $current; |
|
157
|
|
|
if (6 === count($buffer)) { |
|
158
|
|
|
|
|
159
|
|
|
$rows[] = implode(DerDeserializer::DER_DELIMITER, $buffer); |
|
160
|
|
|
$buffer = []; |
|
161
|
|
|
} |
|
162
|
|
|
} |
|
163
|
|
|
|
|
164
|
|
|
return $rows; |
|
165
|
|
|
} |
|
166
|
|
|
} |
webeweb /
core-library
