|
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\Helper; |
|
13
|
|
|
|
|
14
|
|
|
use WBW\Library\Core\Geometry\Point; |
|
15
|
|
|
use WBW\Library\Core\ThirdParty\OcrLad\Model\Document; |
|
16
|
|
|
use WBW\Library\Core\ThirdParty\OcrLad\Model\Word; |
|
17
|
|
|
|
|
18
|
|
|
/** |
|
19
|
|
|
* Document helper. |
|
20
|
|
|
* |
|
21
|
|
|
* @author webeweb <https://github.com/webeweb> |
|
22
|
|
|
* @package WBW\Library\Core\ThirdParty\OcrLad\Helper |
|
23
|
|
|
*/ |
|
24
|
|
|
class DocumentHelper { |
|
25
|
|
|
|
|
26
|
|
|
/** |
|
27
|
|
|
* Document. |
|
28
|
|
|
* |
|
29
|
|
|
* @var Document |
|
30
|
|
|
*/ |
|
31
|
|
|
private $document; |
|
32
|
|
|
|
|
33
|
|
|
/** |
|
34
|
|
|
* Constructor. |
|
35
|
|
|
* |
|
36
|
|
|
* @param Document $document The document. |
|
37
|
|
|
*/ |
|
38
|
|
|
public function __construct(Document $document) { |
|
39
|
|
|
$this->setDocument($document); |
|
40
|
|
|
} |
|
41
|
|
|
|
|
42
|
|
|
/** |
|
43
|
|
|
* Find a word. |
|
44
|
|
|
* |
|
45
|
|
|
* @param Point $position The position. |
|
46
|
|
|
* @param int|null $page The page. |
|
47
|
|
|
* @return Word Returns the word in case of success, null otherwise. |
|
48
|
|
|
*/ |
|
49
|
|
|
public function findWordAt(Point $position, int $page = null): ?Word { |
|
50
|
|
|
|
|
51
|
|
|
foreach ($this->getWords($page) as $current) { |
|
52
|
|
|
|
|
53
|
|
|
$dX = $position->getXInt() === $current->getPoint1()->getXInt(); |
|
54
|
|
|
$dY = $position->getYInt() === $current->getPoint1()->getYInt(); |
|
55
|
|
|
|
|
56
|
|
|
if (true === $dX && true === $dY) { |
|
57
|
|
|
return $current; |
|
58
|
|
|
} |
|
59
|
|
|
} |
|
60
|
|
|
|
|
61
|
|
|
return null; |
|
62
|
|
|
} |
|
63
|
|
|
|
|
64
|
|
|
/** |
|
65
|
|
|
* Find the words. |
|
66
|
|
|
* |
|
67
|
|
|
* @param Point $start The start position. |
|
68
|
|
|
* @param Point $end The end position. |
|
69
|
|
|
* @param int|null $page The page. |
|
70
|
|
|
* @return Word[] Returns the words. |
|
71
|
|
|
*/ |
|
72
|
|
|
public function findWordsIn(Point $start, Point $end, int $page = null): array { |
|
73
|
|
|
|
|
74
|
|
|
$matches = []; |
|
75
|
|
|
|
|
76
|
|
|
foreach ($this->getWords($page) as $current) { |
|
77
|
|
|
|
|
78
|
|
|
$dX1 = $start->getXInt() <= $current->getPoint1()->getXInt(); |
|
79
|
|
|
$dY1 = $start->getYInt() <= $current->getPoint1()->getYInt(); |
|
80
|
|
|
|
|
81
|
|
|
$dX2 = $current->getPoint2()->getXInt() <= $end->getXInt(); |
|
82
|
|
|
$dY2 = $current->getPoint2()->getYInt() <= $end->getYInt(); |
|
83
|
|
|
|
|
84
|
|
|
if (true === $dX1 && true === $dY1 && true === $dX2 && true === $dY2) { |
|
85
|
|
|
$matches[] = $current; |
|
86
|
|
|
} |
|
87
|
|
|
} |
|
88
|
|
|
|
|
89
|
|
|
return $matches; |
|
90
|
|
|
} |
|
91
|
|
|
|
|
92
|
|
|
/** |
|
93
|
|
|
* Find the words. |
|
94
|
|
|
* |
|
95
|
|
|
* @param Point $start The start position. |
|
96
|
|
|
* @param int|null $page The page. |
|
97
|
|
|
* @return Word[] Returns the words. |
|
98
|
|
|
*/ |
|
99
|
|
|
public function findWordsStartAt(Point $start, int $page = null): array { |
|
100
|
|
|
|
|
101
|
|
|
$matches = []; |
|
102
|
|
|
|
|
103
|
|
|
foreach ($this->getWords($page) as $current) { |
|
104
|
|
|
|
|
105
|
|
|
$dX = $start->getXInt() <= $current->getPoint1()->getXInt(); |
|
106
|
|
|
$dY = $start->getYInt() === $current->getPoint1()->getYInt(); |
|
107
|
|
|
|
|
108
|
|
|
if (true === $dX && true === $dY) { |
|
109
|
|
|
$matches[] = $current; |
|
110
|
|
|
} |
|
111
|
|
|
} |
|
112
|
|
|
|
|
113
|
|
|
return $matches; |
|
114
|
|
|
} |
|
115
|
|
|
|
|
116
|
|
|
/** |
|
117
|
|
|
* Get the document. |
|
118
|
|
|
* |
|
119
|
|
|
* @return Document Returns the document. |
|
120
|
|
|
*/ |
|
121
|
|
|
public function getDocument(): Document { |
|
122
|
|
|
return $this->document; |
|
123
|
|
|
} |
|
124
|
|
|
|
|
125
|
|
|
/** |
|
126
|
|
|
* Get the words. |
|
127
|
|
|
* |
|
128
|
|
|
* @param int|null $p The page. |
|
129
|
|
|
* @return Word[] Returns the words. |
|
130
|
|
|
*/ |
|
131
|
|
|
protected function getWords(?int $p): array { |
|
132
|
|
|
|
|
133
|
|
|
$page = $this->getDocument()->getPage($p); |
|
134
|
|
|
if (null !== $page) { |
|
135
|
|
|
return $page->getWords(); |
|
136
|
|
|
} |
|
137
|
|
|
|
|
138
|
|
|
return $this->getDocument()->getWords(); |
|
139
|
|
|
} |
|
140
|
|
|
|
|
141
|
|
|
/** |
|
142
|
|
|
* Set the document. |
|
143
|
|
|
* |
|
144
|
|
|
* @param Document $document The document. |
|
145
|
|
|
* @return DocumentHelper Returns this document helper. |
|
146
|
|
|
*/ |
|
147
|
|
|
protected function setDocument(Document $document): DocumentHelper { |
|
148
|
|
|
$this->document = $document; |
|
149
|
|
|
return $this; |
|
150
|
|
|
} |
|
151
|
|
|
} |
webeweb /
core-library
