|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Felix\Scraper; |
|
4
|
|
|
|
|
5
|
|
|
use URL\Normalizer; |
|
6
|
|
|
|
|
7
|
|
|
class Url |
|
8
|
|
|
{ |
|
9
|
|
|
/** @var array */ |
|
10
|
|
|
private static $parts; |
|
11
|
|
|
|
|
12
|
|
|
/** |
|
13
|
|
|
* Obtener una parte de la URL. |
|
14
|
|
|
* |
|
15
|
|
|
* @param $part string Parte de la URL (Ej: host). |
|
16
|
|
|
* |
|
17
|
|
|
* @return string|null |
|
18
|
|
|
*/ |
|
19
|
1 |
|
public static function part($part) |
|
20
|
|
|
{ |
|
21
|
1 |
|
return array_key_exists($part, self::$parts) ? self::$parts[$part] : null; |
|
22
|
|
|
} |
|
23
|
|
|
|
|
24
|
|
|
/** |
|
25
|
|
|
* Url tiene una parte. |
|
26
|
|
|
* |
|
27
|
|
|
* @return bool true|false |
|
28
|
|
|
*/ |
|
29
|
|
|
public static function has($part) |
|
30
|
|
|
{ |
|
31
|
|
|
return self::part($part) !== null; |
|
32
|
|
|
} |
|
33
|
|
|
|
|
34
|
|
|
/** |
|
35
|
|
|
* Agregar host a url. |
|
36
|
|
|
* |
|
37
|
|
|
* @param $url string Url |
|
38
|
|
|
* @param $host string Dominio |
|
39
|
|
|
* |
|
40
|
|
|
* @return string |
|
41
|
|
|
*/ |
|
42
|
1 |
|
public static function addHost($url, $host) |
|
43
|
|
|
{ |
|
44
|
1 |
|
return rtrim($host, '/') . '/' . $url; |
|
45
|
|
|
} |
|
46
|
|
|
|
|
47
|
|
|
/** |
|
48
|
|
|
* Agregar http a url. |
|
49
|
|
|
* |
|
50
|
|
|
* @param $url string Url |
|
51
|
|
|
* @param $scheme string Esquema |
|
52
|
|
|
* |
|
53
|
|
|
* @return string $url |
|
54
|
|
|
*/ |
|
55
|
|
|
public static function addScheme($url, $scheme = 'http://') |
|
56
|
|
|
{ |
|
57
|
|
|
$url = $scheme . ltrim($url, '/'); |
|
58
|
|
|
|
|
59
|
|
|
return $url; |
|
60
|
|
|
} |
|
61
|
|
|
|
|
62
|
|
|
/** |
|
63
|
|
|
* Decode unreserved characters |
|
64
|
|
|
* |
|
65
|
|
|
* @param $url string Url. |
|
66
|
|
|
* |
|
67
|
|
|
* @return string |
|
68
|
|
|
*/ |
|
69
|
1 |
|
public static function encode($url) |
|
70
|
|
|
{ |
|
71
|
1 |
|
$components = explode("/", $url); |
|
72
|
1 |
|
$end = array_pop($components); |
|
73
|
1 |
|
array_push($components, rawurlencode($end)); |
|
74
|
1 |
|
$url = implode("/", $components); |
|
75
|
|
|
|
|
76
|
1 |
|
return $url; |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
|
|
/** |
|
80
|
|
|
* Normalizar URL. |
|
81
|
|
|
* |
|
82
|
|
|
* @param $url string Url a normalizar. |
|
83
|
|
|
* @param $schemeAndHost string Esquema y dominio base (Ej. http://example.com) |
|
84
|
|
|
* |
|
85
|
|
|
* @return string |
|
86
|
|
|
*/ |
|
87
|
1 |
|
public static function normalize($url, $host = null) |
|
88
|
|
|
{ |
|
89
|
1 |
|
self::$parts = parse_url($url); |
|
90
|
|
|
|
|
91
|
1 |
|
if (self::$parts === false) { |
|
92
|
|
|
throw new \Exception($url.' es una URL mal formada y no se puede procesar'); |
|
93
|
|
|
} |
|
94
|
|
|
|
|
95
|
1 |
|
if (! self::part('host') && $host !== null) { |
|
|
|
|
|
|
96
|
1 |
|
$url = self::addHost($url, $host); |
|
97
|
|
|
} |
|
98
|
|
|
|
|
99
|
1 |
|
if (self::part('host') && self::part('host')) { |
|
|
|
|
|
|
100
|
|
|
$url = self::addScheme($url); |
|
101
|
|
|
} |
|
102
|
|
|
|
|
103
|
1 |
|
$url = self::encode($url); |
|
104
|
|
|
|
|
105
|
1 |
|
return $url; |
|
106
|
|
|
} |
|
107
|
|
|
} |
|
108
|
|
|
|
In PHP, under loose comparison (like
==, or!=, orswitchconditions), values of different types might be equal.For
stringvalues, the empty string''is a special case, in particular the following results might be unexpected: