1
|
|
|
<?php |
2
|
|
|
namespace PHPTools\PHPHtmlDom; |
3
|
|
|
|
4
|
|
|
/** |
5
|
|
|
* Esta es la clas eque el usuario debe inicializar para convertir un texto html en un objeto DOM. |
6
|
|
|
*/ |
7
|
|
|
class PHPHtmlDom |
8
|
|
|
{ |
9
|
|
|
/** |
10
|
|
|
* Objeto que permite convertir un selector css en un path. |
11
|
|
|
* @var \Symfony\Component\CssSelector\CssSelector |
12
|
|
|
*/ |
13
|
|
|
private $selector; |
14
|
|
|
|
15
|
|
|
/** |
16
|
|
|
* Objeto que permite escribir logs. |
17
|
|
|
* @var \PHPTools\PHPHtmlDom\Core\PHPHtmlDomLog |
18
|
|
|
*/ |
19
|
|
|
private $logger; |
20
|
|
|
|
21
|
|
|
/** |
22
|
|
|
* Objeto nativo de php que permite crear un documento DOM. |
23
|
|
|
* @var \DOMDocument |
24
|
|
|
*/ |
25
|
|
|
private $dom; |
26
|
|
|
|
27
|
|
|
/** |
28
|
|
|
* Objeto nativo de php que permite buscar un xpath dentro de un documento DOM. |
29
|
|
|
* @var \DOMXPath |
30
|
|
|
*/ |
31
|
|
|
private $xpath; |
32
|
|
|
|
33
|
|
|
/** |
34
|
|
|
* Objeto que permite importar el contennido html. |
35
|
|
|
* @var \PHPTools\PHPHtmlDom\Core\PHPHtmlDomImportHtml |
36
|
|
|
*/ |
37
|
|
|
private $importer; |
38
|
|
|
|
39
|
|
|
/** |
40
|
|
|
* Contenido crudo html. |
41
|
|
|
* @var string |
42
|
|
|
*/ |
43
|
|
|
private $html_content; |
44
|
|
|
|
45
|
|
|
public function __construct() |
46
|
|
|
{ |
47
|
|
|
$this->selector = new \Symfony\Component\CssSelector\CssSelectorConverter; |
|
|
|
|
48
|
|
|
$this->logger = new \PHPTools\PHPHtmlDom\Core\PHPHtmlDomLog; |
49
|
|
|
$this->dom = new \DOMDocument; |
50
|
|
|
$this->importer = new \PHPTools\PHPHtmlDom\Core\PHPHtmlDomImportHtml; |
51
|
|
|
$this->html_content = NULL; |
52
|
|
|
|
53
|
|
|
$this->dom->preserveWhiteSpace = false; |
54
|
|
|
$this->dom->validateOnParse = true; |
55
|
|
|
} |
56
|
|
|
|
57
|
|
|
/** |
58
|
|
|
* Metodo que importa y convierte el contenido html en un objeto DOM. |
59
|
|
|
* @param string $text_html Cadena de texto con la url, path, texto html. |
60
|
|
|
* @return boolean |
61
|
|
|
*/ |
62
|
|
|
final public function importHTML($text_html) |
63
|
|
|
{ |
64
|
|
|
$content = $this->importer->import($text_html); |
65
|
|
|
|
66
|
|
|
if(!!is_null($content)) |
67
|
|
|
{ |
68
|
|
|
$this->logger->logError('E001', array($text_html)); |
69
|
|
|
} |
70
|
|
|
else if(!$content) |
71
|
|
|
{ |
72
|
|
|
$this->logger->logWarn('W001', array($text_html)); |
73
|
|
|
} |
74
|
|
|
else |
75
|
|
|
{ |
76
|
|
|
$this->html_content = $content; |
|
|
|
|
77
|
|
|
} |
78
|
|
|
|
79
|
|
|
return !!$this->domImport(); |
80
|
|
|
} |
81
|
|
|
|
82
|
|
|
/** |
83
|
|
|
* Metodo que permite buscar elementos hijos a partir de un selector css. |
84
|
|
|
* @param string $css_selector Cadena de texto con el selector css. |
85
|
|
|
* @return PHPTools\PHPHtmlDom\Core\PHPHtmlDomList |
86
|
|
|
*/ |
87
|
|
|
final public function e($css_selector) |
88
|
|
|
{ |
89
|
|
|
$xpath = $this->toXPath($css_selector); |
90
|
|
|
|
91
|
|
|
if(!!$xpath) |
|
|
|
|
92
|
|
|
{ |
93
|
|
|
return new \PHPTools\PHPHtmlDom\Core\PHPHtmlDomList($this->xpath->query($xpath)); |
94
|
|
|
} |
95
|
|
|
} |
96
|
|
|
|
97
|
|
|
/** |
98
|
|
|
* Permite importar el contenido dom del texto html. |
99
|
|
|
* @return boolean |
100
|
|
|
*/ |
101
|
|
|
private function domImport() |
102
|
|
|
{ |
103
|
|
|
$dom_import = @$this->dom->loadHTML($this->html_content); |
104
|
|
|
|
105
|
|
|
if(!!$dom_import) |
106
|
|
|
{ |
107
|
|
|
$this->xpath = new \DOMXPath($this->dom); |
108
|
|
|
} |
109
|
|
|
else |
110
|
|
|
{ |
111
|
|
|
$this->logger->logError('E002', array($this->html_content)); |
112
|
|
|
} |
113
|
|
|
|
114
|
|
|
return $dom_import; |
115
|
|
|
} |
116
|
|
|
|
117
|
|
|
/** |
118
|
|
|
* Metodo que convierte un selector css en un xpath |
119
|
|
|
* @param string $css_selector Cadena de texto con el selector css. |
120
|
|
|
* @return string|NULL Devuelve una cadena de texto con formato xpath si la converción e sposible o NULL en caso contrario. |
121
|
|
|
*/ |
122
|
|
|
private function toXPath($css_selector) |
123
|
|
|
{ |
124
|
|
|
$xpath = Null; |
125
|
|
|
|
126
|
|
|
try |
127
|
|
|
{ |
128
|
|
|
$xpath = $this->selector->toXPath($css_selector); |
129
|
|
|
} |
130
|
|
|
catch (\Exception $e) |
131
|
|
|
{ |
132
|
|
|
$this->logger->logError('E003', array($css_selector)); |
133
|
|
|
$this->logger->logError('E000', array($e->getMessage())); |
134
|
|
|
} |
135
|
|
|
|
136
|
|
|
return $xpath; |
137
|
|
|
} |
138
|
|
|
} |
139
|
|
|
|
Our type inference engine has found an assignment to a property that is incompatible with the declared type of that property.
Either this assignment is in error or the assigned type should be added to the documentation/type hint for that property..