|
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\Model; |
|
13
|
|
|
|
|
14
|
|
|
use WBW\Library\Core\Model\Attribute\StringFilenameTrait; |
|
15
|
|
|
use WBW\Library\Core\ThirdParty\OcrLad\Model\Attribute\ArrayWordsTrait; |
|
16
|
|
|
|
|
17
|
|
|
/** |
|
18
|
|
|
* Document. |
|
19
|
|
|
* |
|
20
|
|
|
* @author webeweb <https://github.com/webeweb> |
|
21
|
|
|
* @package WBW\Library\Core\ThirdParty\OcrLad\Model |
|
22
|
|
|
*/ |
|
23
|
|
|
class Document { |
|
24
|
|
|
|
|
25
|
|
|
use ArrayWordsTrait; |
|
26
|
|
|
use StringFilenameTrait; |
|
27
|
|
|
|
|
28
|
|
|
/** |
|
29
|
|
|
* Index. |
|
30
|
|
|
* |
|
31
|
|
|
* @var Word[] |
|
32
|
|
|
*/ |
|
33
|
|
|
private $index; |
|
34
|
|
|
|
|
35
|
|
|
/** |
|
36
|
|
|
* Pages. |
|
37
|
|
|
* |
|
38
|
|
|
* @var Page[] |
|
39
|
|
|
*/ |
|
40
|
|
|
private $pages; |
|
41
|
|
|
|
|
42
|
|
|
/** |
|
43
|
|
|
* Constructor. |
|
44
|
|
|
*/ |
|
45
|
|
|
public function __construct() { |
|
46
|
|
|
$this->setIndex([]); |
|
47
|
|
|
$this->setPages([]); |
|
48
|
|
|
$this->setWords([]); |
|
49
|
|
|
} |
|
50
|
|
|
|
|
51
|
|
|
/** |
|
52
|
|
|
* Add a page. |
|
53
|
|
|
* |
|
54
|
|
|
* @param Page $page The page. |
|
55
|
|
|
* @return Document Returns this document. |
|
56
|
|
|
*/ |
|
57
|
|
|
public function addPage(Page $page) { |
|
58
|
|
|
$this->pages[] = $page; |
|
59
|
|
|
return $this; |
|
60
|
|
|
} |
|
61
|
|
|
|
|
62
|
|
|
/** |
|
63
|
|
|
* Get the index. |
|
64
|
|
|
* |
|
65
|
|
|
* @return Word[] Returns the index. |
|
66
|
|
|
*/ |
|
67
|
|
|
public function getIndex() { |
|
68
|
|
|
return $this->index; |
|
69
|
|
|
} |
|
70
|
|
|
|
|
71
|
|
|
/** |
|
72
|
|
|
* Get the number of pages. |
|
73
|
|
|
* |
|
74
|
|
|
* @return int Returns the number of pages. |
|
75
|
|
|
*/ |
|
76
|
|
|
public function getNumberPages() { |
|
77
|
|
|
return count($this->getPages()); |
|
78
|
|
|
} |
|
79
|
|
|
|
|
80
|
|
|
/** |
|
81
|
|
|
* Get a page. |
|
82
|
|
|
* |
|
83
|
|
|
* @param int $p The page. |
|
84
|
|
|
* @return Page|null Returns the page in case of success, null otherwise. |
|
85
|
|
|
*/ |
|
86
|
|
|
public function getPage($p) { |
|
87
|
|
|
if (false === is_int($p) || $p < 0 || $this->getNumberPages() <= $p) { |
|
88
|
|
|
return null; |
|
89
|
|
|
} |
|
90
|
|
|
return $this->getPages()[$p]; |
|
91
|
|
|
} |
|
92
|
|
|
|
|
93
|
|
|
/** |
|
94
|
|
|
* Get the pages. |
|
95
|
|
|
* |
|
96
|
|
|
* @return Page[] Returns the pages. |
|
97
|
|
|
*/ |
|
98
|
|
|
public function getPages() { |
|
99
|
|
|
return $this->pages; |
|
100
|
|
|
} |
|
101
|
|
|
|
|
102
|
|
|
/** |
|
103
|
|
|
* Determines if this instance has pages. |
|
104
|
|
|
* |
|
105
|
|
|
* @return bool Returns true in case of success, false otherwise. |
|
106
|
|
|
*/ |
|
107
|
|
|
public function hasPages() { |
|
108
|
|
|
return 1 <= $this->getNumberPages(); |
|
109
|
|
|
} |
|
110
|
|
|
|
|
111
|
|
|
/** |
|
112
|
|
|
* Index a word. |
|
113
|
|
|
* |
|
114
|
|
|
* @param Word $word The word. |
|
115
|
|
|
* @return Document Returns this document. |
|
116
|
|
|
*/ |
|
117
|
|
|
public function index(Word $word) { |
|
118
|
|
|
|
|
119
|
|
|
if (false === array_key_exists($word->getContent(), $this->index)) { |
|
120
|
|
|
$this->index[$word->getContent()] = []; |
|
121
|
|
|
} |
|
122
|
|
|
|
|
123
|
|
|
$this->index[$word->getContent()][] = $word; |
|
124
|
|
|
|
|
125
|
|
|
return $this; |
|
126
|
|
|
} |
|
127
|
|
|
|
|
128
|
|
|
/** |
|
129
|
|
|
* Set the index. |
|
130
|
|
|
* |
|
131
|
|
|
* @param Word[] $index The index. |
|
132
|
|
|
* @return Document Returns this document. |
|
133
|
|
|
*/ |
|
134
|
|
|
protected function setIndex($index) { |
|
135
|
|
|
$this->index = $index; |
|
136
|
|
|
return $this; |
|
137
|
|
|
} |
|
138
|
|
|
|
|
139
|
|
|
/** |
|
140
|
|
|
* Set the pages. |
|
141
|
|
|
* |
|
142
|
|
|
* @param Page[] $pages The pages. |
|
143
|
|
|
* @return Document Returns this document. |
|
144
|
|
|
*/ |
|
145
|
|
|
protected function setPages(array $pages) { |
|
146
|
|
|
$this->pages = $pages; |
|
147
|
|
|
return $this; |
|
148
|
|
|
} |
|
149
|
|
|
} |
webeweb /
core-library
