1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
declare(strict_types=1); |
4
|
|
|
|
5
|
|
|
namespace PHPHtmlParser\Dom\Node; |
6
|
|
|
|
7
|
|
|
use PHPHtmlParser\Dom\Tag; |
8
|
|
|
use PHPHtmlParser\Exceptions\ChildNotFoundException; |
9
|
|
|
use PHPHtmlParser\Exceptions\UnknownChildTypeException; |
10
|
|
|
|
11
|
|
|
/** |
12
|
|
|
* Class HtmlNode. |
13
|
|
|
* |
14
|
|
|
* @property-read string $outerhtml |
15
|
|
|
* @property-read string $innerhtml |
16
|
|
|
* @property-read string $innerText |
17
|
|
|
* @property-read string $text |
18
|
|
|
* @property-read Tag $tag |
19
|
|
|
* @property-read InnerNode $parent |
20
|
|
|
*/ |
21
|
|
|
class HtmlNode extends InnerNode |
22
|
|
|
{ |
23
|
|
|
/** |
24
|
|
|
* Remembers what the innerHtml was if it was scanned previously. |
25
|
|
|
* |
26
|
|
|
* @var ?string |
27
|
|
|
*/ |
28
|
|
|
protected $innerHtml; |
29
|
|
|
|
30
|
|
|
/** |
31
|
|
|
* Remembers what the outerHtml was if it was scanned previously. |
32
|
|
|
* |
33
|
|
|
* @var ?string |
34
|
|
|
*/ |
35
|
|
|
protected $outerHtml; |
36
|
|
|
|
37
|
|
|
/** |
38
|
|
|
* Remembers what the innerText was if it was scanned previously. |
39
|
|
|
* |
40
|
|
|
* @var ?string |
41
|
|
|
*/ |
42
|
|
|
protected $innerText; |
43
|
|
|
|
44
|
|
|
/** |
45
|
|
|
* Remembers what the text was if it was scanned previously. |
46
|
|
|
* |
47
|
|
|
* @var ?string |
48
|
|
|
*/ |
49
|
|
|
protected $text; |
50
|
|
|
|
51
|
|
|
/** |
52
|
|
|
* Remembers what the text was when we looked into all our |
53
|
|
|
* children nodes. |
54
|
|
|
* |
55
|
|
|
* @var ?string |
56
|
|
|
*/ |
57
|
|
|
protected $textWithChildren; |
58
|
|
|
|
59
|
|
|
/** |
60
|
|
|
* Sets up the tag of this node. |
61
|
|
|
* |
62
|
|
|
* @param string|Tag $tag |
63
|
|
|
*/ |
64
|
426 |
|
public function __construct($tag) |
65
|
|
|
{ |
66
|
426 |
|
if (!$tag instanceof Tag) { |
67
|
342 |
|
$tag = new Tag($tag); |
68
|
|
|
} |
69
|
426 |
|
$this->tag = $tag; |
|
|
|
|
70
|
426 |
|
parent::__construct(); |
71
|
426 |
|
} |
72
|
|
|
|
73
|
|
|
/** |
74
|
|
|
* @param bool $htmlSpecialCharsDecode |
75
|
|
|
*/ |
76
|
294 |
|
public function setHtmlSpecialCharsDecode($htmlSpecialCharsDecode = false): void |
77
|
|
|
{ |
78
|
294 |
|
parent::setHtmlSpecialCharsDecode($htmlSpecialCharsDecode); |
79
|
294 |
|
$this->tag->setHtmlSpecialCharsDecode($htmlSpecialCharsDecode); |
80
|
294 |
|
} |
81
|
|
|
|
82
|
|
|
/** |
83
|
|
|
* Gets the inner html of this node. |
84
|
|
|
* |
85
|
|
|
* @throws ChildNotFoundException |
86
|
|
|
* @throws UnknownChildTypeException |
87
|
|
|
*/ |
88
|
174 |
|
public function innerHtml(): string |
89
|
|
|
{ |
90
|
174 |
|
if (!$this->hasChildren()) { |
91
|
|
|
// no children |
92
|
24 |
|
return ''; |
93
|
|
|
} |
94
|
|
|
|
95
|
162 |
|
if ($this->innerHtml !== null) { |
96
|
|
|
// we already know the result. |
97
|
3 |
|
return $this->innerHtml; |
98
|
|
|
} |
99
|
|
|
|
100
|
162 |
|
$child = $this->firstChild(); |
101
|
162 |
|
$string = ''; |
102
|
|
|
|
103
|
|
|
// continue to loop until we are out of children |
104
|
162 |
|
while ($child !== null) { |
105
|
162 |
|
if ($child instanceof TextNode) { |
106
|
150 |
|
$string .= $child->text(); |
107
|
141 |
|
} elseif ($child instanceof HtmlNode) { |
108
|
141 |
|
$string .= $child->outerHtml(); |
109
|
|
|
} else { |
110
|
3 |
|
throw new UnknownChildTypeException('Unknown child type "' . \get_class($child) . '" found in node'); |
111
|
|
|
} |
112
|
|
|
|
113
|
|
|
try { |
114
|
162 |
|
$child = $this->nextChild($child->id()); |
115
|
162 |
|
} catch (ChildNotFoundException $e) { |
116
|
|
|
// no more children |
117
|
162 |
|
unset($e); |
118
|
162 |
|
$child = null; |
119
|
|
|
} |
120
|
|
|
} |
121
|
|
|
|
122
|
|
|
// remember the results |
123
|
162 |
|
$this->innerHtml = $string; |
124
|
|
|
|
125
|
162 |
|
return $string; |
126
|
|
|
} |
127
|
|
|
|
128
|
|
|
/** |
129
|
|
|
* Gets the inner text of this node. |
130
|
|
|
* |
131
|
|
|
* @throws ChildNotFoundException |
132
|
|
|
* @throws UnknownChildTypeException |
133
|
|
|
*/ |
134
|
6 |
|
public function innerText(): string |
135
|
|
|
{ |
136
|
6 |
|
if (\is_null($this->innerText)) { |
|
|
|
|
137
|
6 |
|
$this->innerText = \strip_tags($this->innerHtml()); |
138
|
|
|
} |
139
|
|
|
|
140
|
6 |
|
return $this->innerText; |
141
|
|
|
} |
142
|
|
|
|
143
|
|
|
/** |
144
|
|
|
* Gets the html of this node, including it's own |
145
|
|
|
* tag. |
146
|
|
|
* |
147
|
|
|
* @throws ChildNotFoundException |
148
|
|
|
* @throws UnknownChildTypeException |
149
|
|
|
*/ |
150
|
168 |
|
public function outerHtml(): string |
151
|
|
|
{ |
152
|
|
|
// special handling for root |
153
|
168 |
|
if ($this->tag->name() == 'root') { |
154
|
15 |
|
return $this->innerHtml(); |
155
|
|
|
} |
156
|
|
|
|
157
|
168 |
|
if ($this->outerHtml !== null) { |
158
|
|
|
// we already know the results. |
159
|
6 |
|
return $this->outerHtml; |
160
|
|
|
} |
161
|
|
|
|
162
|
168 |
|
$return = $this->tag->makeOpeningTag(); |
163
|
168 |
|
if ($this->tag->isSelfClosing()) { |
164
|
|
|
// ignore any children... there should not be any though |
165
|
93 |
|
return $return; |
166
|
|
|
} |
167
|
|
|
|
168
|
|
|
// get the inner html |
169
|
150 |
|
$return .= $this->innerHtml(); |
170
|
|
|
|
171
|
|
|
// add closing tag |
172
|
150 |
|
$return .= $this->tag->makeClosingTag(); |
173
|
|
|
|
174
|
|
|
// remember the results |
175
|
150 |
|
$this->outerHtml = $return; |
176
|
|
|
|
177
|
150 |
|
return $return; |
178
|
|
|
} |
179
|
|
|
|
180
|
|
|
/** |
181
|
|
|
* Gets the text of this node (if there is any text). Or get all the text |
182
|
|
|
* in this node, including children. |
183
|
|
|
*/ |
184
|
48 |
|
public function text(bool $lookInChildren = false): string |
185
|
|
|
{ |
186
|
48 |
|
if ($lookInChildren) { |
187
|
6 |
|
if ($this->textWithChildren !== null) { |
188
|
|
|
// we already know the results. |
189
|
6 |
|
return $this->textWithChildren; |
190
|
|
|
} |
191
|
45 |
|
} elseif ($this->text !== null) { |
192
|
|
|
// we already know the results. |
193
|
3 |
|
return $this->text; |
194
|
|
|
} |
195
|
|
|
|
196
|
|
|
// find out if this node has any text children |
197
|
48 |
|
$text = ''; |
198
|
48 |
|
foreach ($this->children as $child) { |
199
|
|
|
/** @var AbstractNode $node */ |
200
|
45 |
|
$node = $child['node']; |
201
|
45 |
|
if ($node instanceof TextNode) { |
202
|
45 |
|
$text .= $child['node']->text; |
203
|
|
|
} elseif ( |
204
|
6 |
|
$lookInChildren && |
205
|
6 |
|
$node instanceof HtmlNode |
206
|
|
|
) { |
207
|
6 |
|
$text .= $node->text($lookInChildren); |
208
|
|
|
} |
209
|
|
|
} |
210
|
|
|
|
211
|
|
|
// remember our result |
212
|
48 |
|
if ($lookInChildren) { |
213
|
6 |
|
$this->textWithChildren = $text; |
214
|
|
|
} else { |
215
|
45 |
|
$this->text = $text; |
|
|
|
|
216
|
|
|
} |
217
|
|
|
|
218
|
48 |
|
return $text; |
219
|
|
|
} |
220
|
|
|
|
221
|
|
|
/** |
222
|
|
|
* Call this when something in the node tree has changed. Like a child has been added |
223
|
|
|
* or a parent has been changed. |
224
|
|
|
*/ |
225
|
408 |
|
protected function clear(): void |
226
|
|
|
{ |
227
|
408 |
|
$this->innerHtml = null; |
228
|
408 |
|
$this->outerHtml = null; |
229
|
408 |
|
$this->text = null; |
|
|
|
|
230
|
408 |
|
$this->textWithChildren = null; |
231
|
|
|
|
232
|
408 |
|
if ($this->parent !== null) { |
233
|
318 |
|
$this->parent->clear(); |
234
|
|
|
} |
235
|
408 |
|
} |
236
|
|
|
|
237
|
|
|
/** |
238
|
|
|
* Returns all children of this html node. |
239
|
|
|
*/ |
240
|
6 |
|
protected function getIteratorArray(): array |
241
|
|
|
{ |
242
|
6 |
|
return $this->getChildren(); |
243
|
|
|
} |
244
|
|
|
} |
245
|
|
|
|