1
|
|
|
<?php declare(strict_types=1); |
2
|
|
|
namespace PHPHtmlParser\Dom; |
3
|
|
|
|
4
|
|
|
use PHPHtmlParser\Exceptions\UnknownChildTypeException; |
5
|
|
|
use PHPHtmlParser\Exceptions\ChildNotFoundException; |
6
|
|
|
|
7
|
|
|
/** |
8
|
|
|
* Class HtmlNode |
9
|
|
|
* |
10
|
|
|
* @package PHPHtmlParser\Dom |
11
|
|
|
*/ |
12
|
|
|
class HtmlNode extends InnerNode |
13
|
|
|
{ |
14
|
|
|
|
15
|
|
|
/** |
16
|
|
|
* Remembers what the innerHtml was if it was scanned previously. |
17
|
|
|
* |
18
|
|
|
* @var string |
19
|
|
|
*/ |
20
|
|
|
protected $innerHtml = null; |
21
|
|
|
|
22
|
|
|
/** |
23
|
|
|
* Remembers what the outerHtml was if it was scanned previously. |
24
|
|
|
* |
25
|
|
|
* @var string |
26
|
|
|
*/ |
27
|
|
|
protected $outerHtml = null; |
28
|
|
|
|
29
|
|
|
/** |
30
|
|
|
* Remembers what the text was if it was scanned previously. |
31
|
|
|
* |
32
|
|
|
* @var string |
33
|
|
|
*/ |
34
|
|
|
protected $text = null; |
35
|
|
|
|
36
|
|
|
/** |
37
|
|
|
* Remembers what the text was when we looked into all our |
38
|
|
|
* children nodes. |
39
|
|
|
* |
40
|
|
|
* @var string |
41
|
|
|
*/ |
42
|
|
|
protected $textWithChildren = null; |
43
|
|
|
|
44
|
|
|
/** |
45
|
|
|
* Sets up the tag of this node. |
46
|
|
|
* |
47
|
|
|
* @param string|Tag $tag |
48
|
|
|
*/ |
49
|
354 |
|
public function __construct($tag) |
50
|
|
|
{ |
51
|
354 |
|
if ( ! $tag instanceof Tag) { |
52
|
270 |
|
$tag = new Tag($tag); |
53
|
|
|
} |
54
|
354 |
|
$this->tag = $tag; |
55
|
354 |
|
parent::__construct(); |
56
|
354 |
|
} |
57
|
|
|
|
58
|
|
|
/** |
59
|
|
|
* @param bool $htmlSpecialCharsDecode |
60
|
|
|
* @return void |
61
|
|
|
*/ |
62
|
228 |
|
public function setHtmlSpecialCharsDecode($htmlSpecialCharsDecode = false): void |
63
|
|
|
{ |
64
|
228 |
|
parent::setHtmlSpecialCharsDecode($htmlSpecialCharsDecode); |
65
|
228 |
|
$this->tag->setHtmlSpecialCharsDecode($htmlSpecialCharsDecode); |
66
|
228 |
|
} |
67
|
|
|
|
68
|
|
|
/** |
69
|
|
|
* Gets the inner html of this node. |
70
|
|
|
* @return string |
71
|
|
|
* @throws ChildNotFoundException |
72
|
|
|
* @throws UnknownChildTypeException |
73
|
|
|
*/ |
74
|
138 |
|
public function innerHtml(): string |
75
|
|
|
{ |
76
|
138 |
|
if ( ! $this->hasChildren()) { |
77
|
|
|
// no children |
78
|
18 |
|
return ''; |
79
|
|
|
} |
80
|
|
|
|
81
|
129 |
|
if ( ! is_null($this->innerHtml)) { |
|
|
|
|
82
|
|
|
// we already know the result. |
83
|
3 |
|
return $this->innerHtml; |
84
|
|
|
} |
85
|
|
|
|
86
|
129 |
|
$child = $this->firstChild(); |
87
|
129 |
|
$string = ''; |
88
|
|
|
|
89
|
|
|
// continue to loop until we are out of children |
90
|
129 |
|
while ( ! is_null($child)) { |
91
|
129 |
|
if ($child instanceof TextNode) { |
92
|
123 |
|
$string .= $child->text(); |
93
|
114 |
|
} elseif ($child instanceof HtmlNode) { |
94
|
114 |
|
$string .= $child->outerHtml(); |
95
|
|
|
} else { |
96
|
3 |
|
throw new UnknownChildTypeException('Unknown child type "'.get_class($child).'" found in node'); |
97
|
|
|
} |
98
|
|
|
|
99
|
|
|
try { |
100
|
129 |
|
$child = $this->nextChild($child->id()); |
101
|
129 |
|
} catch (ChildNotFoundException $e) { |
102
|
|
|
// no more children |
103
|
129 |
|
$child = null; |
104
|
|
|
} |
105
|
|
|
} |
106
|
|
|
|
107
|
|
|
// remember the results |
108
|
129 |
|
$this->innerHtml = $string; |
109
|
|
|
|
110
|
129 |
|
return $string; |
111
|
|
|
} |
112
|
|
|
|
113
|
|
|
/** |
114
|
|
|
* Gets the html of this node, including it's own |
115
|
|
|
* tag. |
116
|
|
|
* @return string |
117
|
|
|
* @throws ChildNotFoundException |
118
|
|
|
* @throws UnknownChildTypeException |
119
|
|
|
*/ |
120
|
138 |
|
public function outerHtml(): string |
121
|
|
|
{ |
122
|
|
|
// special handling for root |
123
|
138 |
|
if ($this->tag->name() == 'root') { |
124
|
3 |
|
return $this->innerHtml(); |
125
|
|
|
} |
126
|
|
|
|
127
|
138 |
|
if ( ! is_null($this->outerHtml)) { |
|
|
|
|
128
|
|
|
// we already know the results. |
129
|
6 |
|
return $this->outerHtml; |
130
|
|
|
} |
131
|
|
|
|
132
|
138 |
|
$return = $this->tag->makeOpeningTag(); |
133
|
138 |
|
if ($this->tag->isSelfClosing()) { |
134
|
|
|
// ignore any children... there should not be any though |
135
|
84 |
|
return $return; |
136
|
|
|
} |
137
|
|
|
|
138
|
|
|
// get the inner html |
139
|
120 |
|
$return .= $this->innerHtml(); |
140
|
|
|
|
141
|
|
|
// add closing tag |
142
|
120 |
|
$return .= $this->tag->makeClosingTag(); |
143
|
|
|
|
144
|
|
|
// remember the results |
145
|
120 |
|
$this->outerHtml = $return; |
146
|
|
|
|
147
|
120 |
|
return $return; |
148
|
|
|
} |
149
|
|
|
|
150
|
|
|
/** |
151
|
|
|
* Gets the text of this node (if there is any text). Or get all the text |
152
|
|
|
* in this node, including children. |
153
|
|
|
* |
154
|
|
|
* @param bool $lookInChildren |
155
|
|
|
* @return string |
156
|
|
|
*/ |
157
|
48 |
|
public function text(bool $lookInChildren = false): string |
158
|
|
|
{ |
159
|
48 |
|
if ($lookInChildren) { |
160
|
6 |
|
if ( ! is_null($this->textWithChildren)) { |
|
|
|
|
161
|
|
|
// we already know the results. |
162
|
6 |
|
return $this->textWithChildren; |
163
|
|
|
} |
164
|
45 |
|
} elseif ( ! is_null($this->text)) { |
|
|
|
|
165
|
|
|
// we already know the results. |
166
|
3 |
|
return $this->text; |
167
|
|
|
} |
168
|
|
|
|
169
|
|
|
// find out if this node has any text children |
170
|
48 |
|
$text = ''; |
171
|
48 |
|
foreach ($this->children as $child) { |
172
|
|
|
/** @var AbstractNode $node */ |
173
|
45 |
|
$node = $child['node']; |
174
|
45 |
|
if ($node instanceof TextNode) { |
175
|
45 |
|
$text .= $child['node']->text; |
176
|
6 |
|
} elseif ($lookInChildren && |
177
|
6 |
|
$node instanceof HtmlNode |
178
|
|
|
) { |
179
|
32 |
|
$text .= $node->text($lookInChildren); |
180
|
|
|
} |
181
|
|
|
} |
182
|
|
|
|
183
|
|
|
// remember our result |
184
|
48 |
|
if ($lookInChildren) { |
185
|
6 |
|
$this->textWithChildren = $text; |
186
|
|
|
} else { |
187
|
45 |
|
$this->text = $text; |
188
|
|
|
} |
189
|
|
|
|
190
|
48 |
|
return $text; |
191
|
|
|
} |
192
|
|
|
|
193
|
|
|
/** |
194
|
|
|
* Call this when something in the node tree has changed. Like a child has been added |
195
|
|
|
* or a parent has been changed. |
196
|
|
|
*/ |
197
|
336 |
|
protected function clear(): void |
198
|
|
|
{ |
199
|
336 |
|
$this->innerHtml = null; |
200
|
336 |
|
$this->outerHtml = null; |
201
|
336 |
|
$this->text = null; |
202
|
336 |
|
$this->textWithChildren = null; |
203
|
|
|
|
204
|
336 |
|
if (is_null($this->parent) === false) { |
205
|
282 |
|
$this->parent->clear(); |
|
|
|
|
206
|
|
|
} |
207
|
336 |
|
} |
208
|
|
|
|
209
|
|
|
/** |
210
|
|
|
* Returns all children of this html node. |
211
|
|
|
* |
212
|
|
|
* @return array |
213
|
|
|
*/ |
214
|
6 |
|
protected function getIteratorArray(): array |
215
|
|
|
{ |
216
|
6 |
|
return $this->getChildren(); |
217
|
|
|
} |
218
|
|
|
} |
219
|
|
|
|