|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace PHPHtmlParser\Dom\Node; |
|
6
|
|
|
|
|
7
|
|
|
use PHPHtmlParser\Dom\Tag; |
|
8
|
|
|
use PHPHtmlParser\Exceptions\LogicalException; |
|
9
|
|
|
|
|
10
|
|
|
/** |
|
11
|
|
|
* Class TextNode. |
|
12
|
|
|
*/ |
|
13
|
|
|
class TextNode extends LeafNode |
|
14
|
|
|
{ |
|
15
|
|
|
/** |
|
16
|
|
|
* This is a text node. |
|
17
|
|
|
* |
|
18
|
|
|
* @var Tag |
|
19
|
|
|
*/ |
|
20
|
|
|
protected $tag; |
|
21
|
|
|
|
|
22
|
|
|
/** |
|
23
|
|
|
* This is the text in this node. |
|
24
|
|
|
* |
|
25
|
|
|
* @var string |
|
26
|
|
|
*/ |
|
27
|
|
|
protected $text; |
|
28
|
|
|
|
|
29
|
|
|
/** |
|
30
|
|
|
* This is the converted version of the text. |
|
31
|
|
|
* |
|
32
|
|
|
* @var ?string |
|
33
|
|
|
*/ |
|
34
|
|
|
protected $convertedText; |
|
35
|
|
|
|
|
36
|
|
|
/** |
|
37
|
|
|
* Sets the text for this node. |
|
38
|
|
|
* |
|
39
|
|
|
* @param bool $removeDoubleSpace |
|
40
|
|
|
*/ |
|
41
|
321 |
|
public function __construct(string $text, $removeDoubleSpace = true) |
|
42
|
|
|
{ |
|
43
|
321 |
|
if ($removeDoubleSpace) { |
|
44
|
|
|
// remove double spaces |
|
45
|
318 |
|
$replacedText = \mb_ereg_replace('\s+', ' ', $text); |
|
46
|
318 |
|
if ($replacedText === false) { |
|
47
|
|
|
throw new LogicalException('mb_ereg_replace returns false when attempting to clean white space from "' . $text . '".'); |
|
48
|
|
|
} |
|
49
|
318 |
|
$text = $replacedText; |
|
50
|
|
|
} |
|
51
|
|
|
|
|
52
|
|
|
// restore line breaks |
|
53
|
321 |
|
$text = \str_replace(' ', "\n", $text); |
|
54
|
|
|
|
|
55
|
321 |
|
$this->text = $text; |
|
56
|
321 |
|
$this->tag = new Tag('text'); |
|
57
|
321 |
|
parent::__construct(); |
|
58
|
321 |
|
} |
|
59
|
|
|
|
|
60
|
|
|
/** |
|
61
|
|
|
* @param bool $htmlSpecialCharsDecode |
|
62
|
|
|
*/ |
|
63
|
246 |
|
public function setHtmlSpecialCharsDecode($htmlSpecialCharsDecode = false): void |
|
64
|
|
|
{ |
|
65
|
246 |
|
parent::setHtmlSpecialCharsDecode($htmlSpecialCharsDecode); |
|
66
|
246 |
|
$this->tag->setHtmlSpecialCharsDecode($htmlSpecialCharsDecode); |
|
67
|
246 |
|
} |
|
68
|
|
|
|
|
69
|
|
|
/** |
|
70
|
|
|
* Returns the text of this node. |
|
71
|
|
|
*/ |
|
72
|
213 |
|
public function text(): string |
|
73
|
|
|
{ |
|
74
|
213 |
|
if ($this->htmlSpecialCharsDecode) { |
|
75
|
3 |
|
$text = \htmlspecialchars_decode($this->text); |
|
76
|
|
|
} else { |
|
77
|
210 |
|
$text = $this->text; |
|
78
|
|
|
} |
|
79
|
|
|
// convert charset |
|
80
|
213 |
|
if (!\is_null($this->encode)) { |
|
81
|
156 |
|
if (!\is_null($this->convertedText)) { |
|
82
|
|
|
// we already know the converted value |
|
83
|
6 |
|
return $this->convertedText; |
|
84
|
|
|
} |
|
85
|
153 |
|
$text = $this->encode->convert($text); |
|
86
|
|
|
|
|
87
|
|
|
// remember the conversion |
|
88
|
153 |
|
$this->convertedText = $text; |
|
89
|
|
|
|
|
90
|
153 |
|
return $text; |
|
91
|
|
|
} |
|
92
|
|
|
|
|
93
|
57 |
|
return $text; |
|
94
|
|
|
} |
|
95
|
|
|
|
|
96
|
|
|
/** |
|
97
|
|
|
* Sets the text for this node. |
|
98
|
|
|
* |
|
99
|
|
|
* @var string |
|
100
|
|
|
*/ |
|
101
|
9 |
|
public function setText(string $text): void |
|
102
|
|
|
{ |
|
103
|
9 |
|
$this->text = $text; |
|
104
|
9 |
|
if (!\is_null($this->encode)) { |
|
105
|
6 |
|
$text = $this->encode->convert($text); |
|
106
|
|
|
|
|
107
|
|
|
// remember the conversion |
|
108
|
6 |
|
$this->convertedText = $text; |
|
109
|
|
|
} |
|
110
|
9 |
|
} |
|
111
|
|
|
|
|
112
|
|
|
/** |
|
113
|
|
|
* This node has no html, just return the text. |
|
114
|
|
|
* |
|
115
|
|
|
* @uses $this->text() |
|
116
|
|
|
*/ |
|
117
|
3 |
|
public function innerHtml(): string |
|
118
|
|
|
{ |
|
119
|
3 |
|
return $this->text(); |
|
120
|
|
|
} |
|
121
|
|
|
|
|
122
|
|
|
/** |
|
123
|
|
|
* This node has no html, just return the text. |
|
124
|
|
|
* |
|
125
|
|
|
* @uses $this->text() |
|
126
|
|
|
*/ |
|
127
|
6 |
|
public function outerHtml(): string |
|
128
|
|
|
{ |
|
129
|
6 |
|
return $this->text(); |
|
130
|
|
|
} |
|
131
|
|
|
|
|
132
|
|
|
/** |
|
133
|
|
|
* Checks if the current node is a text node. |
|
134
|
|
|
*/ |
|
135
|
3 |
|
public function isTextNode(): bool |
|
136
|
|
|
{ |
|
137
|
3 |
|
return true; |
|
138
|
|
|
} |
|
139
|
|
|
|
|
140
|
|
|
/** |
|
141
|
|
|
* Call this when something in the node tree has changed. Like a child has been added |
|
142
|
|
|
* or a parent has been changed. |
|
143
|
|
|
*/ |
|
144
|
|
|
protected function clear(): void |
|
145
|
|
|
{ |
|
146
|
|
|
$this->convertedText = null; |
|
147
|
|
|
} |
|
148
|
|
|
} |
|
149
|
|
|
|
paquettg /
php-html-parser
