]>
Commit | Line | Data |
---|---|---|
1 | <?php | |
2 | /** | |
3 | * Converts an array-represented URL to a string | |
4 | * | |
5 | * Source: http://php.net/manual/en/function.parse-url.php#106731 | |
6 | * | |
7 | * @see http://php.net/manual/en/function.parse-url.php | |
8 | * | |
9 | * @param array $parsedUrl an array-represented URL | |
10 | * | |
11 | * @return string the string representation of the URL | |
12 | */ | |
13 | function unparse_url($parsedUrl) | |
14 | { | |
15 | $scheme = isset($parsedUrl['scheme']) ? $parsedUrl['scheme'].'://' : ''; | |
16 | $host = isset($parsedUrl['host']) ? $parsedUrl['host'] : ''; | |
17 | $port = isset($parsedUrl['port']) ? ':'.$parsedUrl['port'] : ''; | |
18 | $user = isset($parsedUrl['user']) ? $parsedUrl['user'] : ''; | |
19 | $pass = isset($parsedUrl['pass']) ? ':'.$parsedUrl['pass'] : ''; | |
20 | $pass = ($user || $pass) ? "$pass@" : ''; | |
21 | $path = isset($parsedUrl['path']) ? $parsedUrl['path'] : ''; | |
22 | $query = isset($parsedUrl['query']) ? '?'.$parsedUrl['query'] : ''; | |
23 | $fragment = isset($parsedUrl['fragment']) ? '#'.$parsedUrl['fragment'] : ''; | |
24 | ||
25 | return "$scheme$user$pass$host$port$path$query$fragment"; | |
26 | } | |
27 | ||
28 | /** | |
29 | * URL representation and cleanup utilities | |
30 | * | |
31 | * Form | |
32 | * scheme://[username:password@]host[:port][/path][?query][#fragment] | |
33 | * | |
34 | * Examples | |
35 | * http://username:password@hostname:9090/path?arg1=value1&arg2=value2#anchor | |
36 | * https://host.name.tld | |
37 | * https://h2.g2/faq/?vendor=hitchhiker&item=guide&dest=galaxy#answer | |
38 | * | |
39 | * @see http://www.faqs.org/rfcs/rfc3986.html | |
40 | */ | |
41 | class Url | |
42 | { | |
43 | private static $annoyingQueryParams = array( | |
44 | ||
45 | 'action_object_map=', | |
46 | 'action_ref_map=', | |
47 | 'action_type_map=', | |
48 | 'fb_', | |
49 | 'fb=', | |
50 | ||
51 | // Scoop.it | |
52 | '__scoop', | |
53 | ||
54 | // Google Analytics & FeedProxy | |
55 | 'utm_', | |
56 | ||
57 | // ATInternet | |
58 | 'xtor=' | |
59 | ); | |
60 | ||
61 | private static $annoyingFragments = array( | |
62 | // ATInternet | |
63 | 'xtor=RSS-', | |
64 | ||
65 | // Misc. | |
66 | 'tk.rss_all' | |
67 | ); | |
68 | ||
69 | /* | |
70 | * URL parts represented as an array | |
71 | * | |
72 | * @see http://php.net/parse_url | |
73 | */ | |
74 | protected $parts; | |
75 | ||
76 | /** | |
77 | * Parses a string containing a URL | |
78 | * | |
79 | * @param string $url a string containing a URL | |
80 | */ | |
81 | public function __construct($url) | |
82 | { | |
83 | $this->parts = parse_url($url); | |
84 | ||
85 | if (!empty($url) && empty($this->parts['scheme'])) { | |
86 | $this->parts['scheme'] = 'http'; | |
87 | } | |
88 | } | |
89 | ||
90 | /** | |
91 | * Returns a string representation of this URL | |
92 | */ | |
93 | public function __toString() | |
94 | { | |
95 | return unparse_url($this->parts); | |
96 | } | |
97 | ||
98 | /** | |
99 | * Removes undesired query parameters | |
100 | */ | |
101 | protected function cleanupQuery() | |
102 | { | |
103 | if (! isset($this->parts['query'])) { | |
104 | return; | |
105 | } | |
106 | ||
107 | $queryParams = explode('&', $this->parts['query']); | |
108 | ||
109 | foreach (self::$annoyingQueryParams as $annoying) { | |
110 | foreach ($queryParams as $param) { | |
111 | if (startsWith($param, $annoying)) { | |
112 | $queryParams = array_diff($queryParams, array($param)); | |
113 | continue; | |
114 | } | |
115 | } | |
116 | } | |
117 | ||
118 | if (count($queryParams) == 0) { | |
119 | unset($this->parts['query']); | |
120 | return; | |
121 | } | |
122 | ||
123 | $this->parts['query'] = implode('&', $queryParams); | |
124 | } | |
125 | ||
126 | /** | |
127 | * Removes undesired fragments | |
128 | */ | |
129 | protected function cleanupFragment() | |
130 | { | |
131 | if (! isset($this->parts['fragment'])) { | |
132 | return; | |
133 | } | |
134 | ||
135 | foreach (self::$annoyingFragments as $annoying) { | |
136 | if (startsWith($this->parts['fragment'], $annoying)) { | |
137 | unset($this->parts['fragment']); | |
138 | break; | |
139 | } | |
140 | } | |
141 | } | |
142 | ||
143 | /** | |
144 | * Removes undesired query parameters and fragments | |
145 | * | |
146 | * @return string the string representation of this URL after cleanup | |
147 | */ | |
148 | public function cleanup() | |
149 | { | |
150 | $this->cleanupQuery(); | |
151 | $this->cleanupFragment(); | |
152 | return $this->__toString(); | |
153 | } | |
154 | ||
155 | /** | |
156 | * Get URL scheme. | |
157 | * | |
158 | * @return string the URL scheme or false if none is provided. | |
159 | */ | |
160 | public function getScheme() { | |
161 | if (!isset($this->parts['scheme'])) { | |
162 | return false; | |
163 | } | |
164 | return $this->parts['scheme']; | |
165 | } | |
166 | } |