]>
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$fragment"; | |
26 | } | |
27 | ||
28 | /** | |
29 | * Removes undesired query parameters and fragments | |
30 | * | |
31 | * @param string url Url to be cleaned | |
32 | * | |
33 | * @return string the string representation of this URL after cleanup | |
34 | */ | |
35 | function cleanup_url($url) | |
36 | { | |
37 | $obj_url = new Url($url); | |
38 | return $obj_url->cleanup(); | |
39 | } | |
40 | ||
41 | /** | |
42 | * Get URL scheme. | |
43 | * | |
44 | * @param string url Url for which the scheme is requested | |
45 | * | |
46 | * @return mixed the URL scheme or false if none is provided. | |
47 | */ | |
48 | function get_url_scheme($url) | |
49 | { | |
50 | $obj_url = new Url($url); | |
51 | return $obj_url->getScheme(); | |
52 | } | |
53 | ||
54 | /** | |
55 | * Adds a trailing slash at the end of URL if necessary. | |
56 | * | |
57 | * @param string $url URL to check/edit. | |
58 | * | |
59 | * @return string $url URL with a end trailing slash. | |
60 | */ | |
61 | function add_trailing_slash($url) | |
62 | { | |
63 | return $url . (!endsWith($url, '/') ? '/' : ''); | |
64 | } | |
65 | ||
66 | /** | |
67 | * URL representation and cleanup utilities | |
68 | * | |
69 | * Form | |
70 | * scheme://[username:password@]host[:port][/path][?query][#fragment] | |
71 | * | |
72 | * Examples | |
73 | * http://username:password@hostname:9090/path?arg1=value1&arg2=value2#anchor | |
74 | * https://host.name.tld | |
75 | * https://h2.g2/faq/?vendor=hitchhiker&item=guide&dest=galaxy#answer | |
76 | * | |
77 | * @see http://www.faqs.org/rfcs/rfc3986.html | |
78 | */ | |
79 | class Url | |
80 | { | |
81 | private static $annoyingQueryParams = array( | |
82 | ||
83 | 'action_object_map=', | |
84 | 'action_ref_map=', | |
85 | 'action_type_map=', | |
86 | 'fb_', | |
87 | 'fb=', | |
88 | 'PHPSESSID=', | |
89 | ||
90 | // Scoop.it | |
91 | '__scoop', | |
92 | ||
93 | // Google Analytics & FeedProxy | |
94 | 'utm_', | |
95 | ||
96 | // ATInternet | |
97 | 'xtor=', | |
98 | ||
99 | // Other | |
100 | 'campaign_' | |
101 | ); | |
102 | ||
103 | private static $annoyingFragments = array( | |
104 | // ATInternet | |
105 | 'xtor=RSS-', | |
106 | ||
107 | // Misc. | |
108 | 'tk.rss_all' | |
109 | ); | |
110 | ||
111 | /* | |
112 | * URL parts represented as an array | |
113 | * | |
114 | * @see http://php.net/parse_url | |
115 | */ | |
116 | protected $parts; | |
117 | ||
118 | /** | |
119 | * Parses a string containing a URL | |
120 | * | |
121 | * @param string $url a string containing a URL | |
122 | */ | |
123 | public function __construct($url) | |
124 | { | |
125 | $url = self::cleanupUnparsedUrl(trim($url)); | |
126 | $this->parts = parse_url($url); | |
127 | ||
128 | if (!empty($url) && empty($this->parts['scheme'])) { | |
129 | $this->parts['scheme'] = 'http'; | |
130 | } | |
131 | } | |
132 | ||
133 | /** | |
134 | * Clean up URL before it's parsed. | |
135 | * ie. handle urlencode, url prefixes, etc. | |
136 | * | |
137 | * @param string $url URL to clean. | |
138 | * | |
139 | * @return string cleaned URL. | |
140 | */ | |
141 | protected static function cleanupUnparsedUrl($url) | |
142 | { | |
143 | return self::removeFirefoxAboutReader($url); | |
144 | } | |
145 | ||
146 | /** | |
147 | * Remove Firefox Reader prefix if it's present. | |
148 | * | |
149 | * @param string $input url | |
150 | * | |
151 | * @return string cleaned url | |
152 | */ | |
153 | protected static function removeFirefoxAboutReader($input) | |
154 | { | |
155 | $firefoxPrefix = 'about://reader?url='; | |
156 | if (startsWith($input, $firefoxPrefix)) { | |
157 | return urldecode(ltrim($input, $firefoxPrefix)); | |
158 | } | |
159 | return $input; | |
160 | } | |
161 | ||
162 | /** | |
163 | * Returns a string representation of this URL | |
164 | */ | |
165 | public function toString() | |
166 | { | |
167 | return unparse_url($this->parts); | |
168 | } | |
169 | ||
170 | /** | |
171 | * Removes undesired query parameters | |
172 | */ | |
173 | protected function cleanupQuery() | |
174 | { | |
175 | if (! isset($this->parts['query'])) { | |
176 | return; | |
177 | } | |
178 | ||
179 | $queryParams = explode('&', $this->parts['query']); | |
180 | ||
181 | foreach (self::$annoyingQueryParams as $annoying) { | |
182 | foreach ($queryParams as $param) { | |
183 | if (startsWith($param, $annoying)) { | |
184 | $queryParams = array_diff($queryParams, array($param)); | |
185 | continue; | |
186 | } | |
187 | } | |
188 | } | |
189 | ||
190 | if (count($queryParams) == 0) { | |
191 | unset($this->parts['query']); | |
192 | return; | |
193 | } | |
194 | ||
195 | $this->parts['query'] = implode('&', $queryParams); | |
196 | } | |
197 | ||
198 | /** | |
199 | * Removes undesired fragments | |
200 | */ | |
201 | protected function cleanupFragment() | |
202 | { | |
203 | if (! isset($this->parts['fragment'])) { | |
204 | return; | |
205 | } | |
206 | ||
207 | foreach (self::$annoyingFragments as $annoying) { | |
208 | if (startsWith($this->parts['fragment'], $annoying)) { | |
209 | unset($this->parts['fragment']); | |
210 | break; | |
211 | } | |
212 | } | |
213 | } | |
214 | ||
215 | /** | |
216 | * Removes undesired query parameters and fragments | |
217 | * | |
218 | * @return string the string representation of this URL after cleanup | |
219 | */ | |
220 | public function cleanup() | |
221 | { | |
222 | $this->cleanupQuery(); | |
223 | $this->cleanupFragment(); | |
224 | return $this->toString(); | |
225 | } | |
226 | ||
227 | /** | |
228 | * Converts an URL with an International Domain Name host to a ASCII one. | |
229 | * This requires PHP-intl. If it's not available, just returns this->cleanup(). | |
230 | * | |
231 | * @return string converted cleaned up URL. | |
232 | */ | |
233 | public function idnToAscii() | |
234 | { | |
235 | $out = $this->cleanup(); | |
236 | if (! function_exists('idn_to_ascii') || ! isset($this->parts['host'])) { | |
237 | return $out; | |
238 | } | |
239 | $asciiHost = idn_to_ascii($this->parts['host']); | |
240 | return str_replace($this->parts['host'], $asciiHost, $out); | |
241 | } | |
242 | ||
243 | /** | |
244 | * Get URL scheme. | |
245 | * | |
246 | * @return string the URL scheme or false if none is provided. | |
247 | */ | |
248 | public function getScheme() { | |
249 | if (!isset($this->parts['scheme'])) { | |
250 | return false; | |
251 | } | |
252 | return $this->parts['scheme']; | |
253 | } | |
254 | ||
255 | /** | |
256 | * Get URL host. | |
257 | * | |
258 | * @return string the URL host or false if none is provided. | |
259 | */ | |
260 | public function getHost() { | |
261 | if (empty($this->parts['host'])) { | |
262 | return false; | |
263 | } | |
264 | return $this->parts['host']; | |
265 | } | |
266 | ||
267 | /** | |
268 | * Test if the Url is an HTTP one. | |
269 | * | |
270 | * @return true is HTTP, false otherwise. | |
271 | */ | |
272 | public function isHttp() { | |
273 | return strpos(strtolower($this->parts['scheme']), 'http') !== false; | |
274 | } | |
275 | } |