| blob | abfb27904f6a1e947ef9c3acf85a51eb8dc10c5e |
1 /**
2 * Library for simple URI parsing and manipulation.
3 *
4 * Intended to be minimal, but featureful; do not expect full RFC 3986 compliance. The use cases we
5 * have in mind are constructing 'next page' or 'previous page' URLs, detecting whether we need to
6 * use cross-domain proxies for an API, constructing simple URL-based API calls, etc. Parsing here
7 * is regex-based, so may not work on all URIs, but is good enough for most.
8 *
9 * You can modify the properties directly, then use the #toString method to extract the full URI
10 * string again. Example:
11 *
12 * var uri = new mw.Uri( 'http://example.com/mysite/mypage.php?quux=2' );
13 *
14 * if ( uri.host == 'example.com' ) {
15 * uri.host = 'foo.example.com';
16 * uri.extend( { bar: 1 } );
17 *
18 * $( 'a#id1' ).attr( 'href', uri );
19 * // anchor with id 'id1' now links to http://foo.example.com/mysite/mypage.php?bar=1&quux=2
20 *
21 * $( 'a#id2' ).attr( 'href', uri.clone().extend( { bar: 3, pif: 'paf' } ) );
22 * // anchor with id 'id2' now links to http://foo.example.com/mysite/mypage.php?bar=3&quux=2&pif=paf
23 * }
24 *
25 * Given a URI like
26 * `http://usr:pwd@www.example.com:81/dir/dir.2/index.htm?q1=0&&test1&test2=&test3=value+%28escaped%29&r=1&r=2#top`
27 * the returned object will have the following properties:
28 *
29 * protocol 'http'
30 * user 'usr'
31 * password 'pwd'
32 * host 'www.example.com'
33 * port '81'
34 * path '/dir/dir.2/index.htm'
35 * query {
36 * q1: '0',
37 * test1: null,
38 * test2: '',
39 * test3: 'value (escaped)'
40 * r: ['1', '2']
41 * }
42 * fragment 'top'
43 *
44 * (N.b., 'password' is technically not allowed for HTTP URIs, but it is possible with other kinds
45 * of URIs.)
46 *
47 * Parsing based on parseUri 1.2.2 (c) Steven Levithan <http://stevenlevithan.com>, MIT License.
48 * <http://stevenlevithan.com/demo/parseuri/js/>
49 *
50 * @class mw.Uri
51 */
53 ( function ( mw, $ ) {
54 /**
55 * Function that's useful when constructing the URI string -- we frequently encounter the pattern
56 * of having to add something to the URI as we go, but only if it's present, and to include a
57 * character before or after if so.
58 *
59 * @private
60 * @static
61 * @param {string|undefined} pre To prepend
62 * @param {string} val To include
63 * @param {string} post To append
64 * @param {boolean} raw If true, val will not be encoded
65 * @return {string} Result
66 */
67 function cat( pre, val, post, raw ) {
68 if ( val === undefined || val === null || val === '' ) {
69 return '';
70 }
71 return pre + ( raw ? val : mw.Uri.encode( val ) ) + post;
72 }
74 /**
75 * Regular expressions to parse many common URIs.
76 *
77 * @private
78 * @static
79 * @property {Object} parser
80 */
81 var parser = {
82 strict: /^(?:([^:\/?#]+):)?(?:\/\/(?:(?:([^:@\/?#]*)(?::([^:@\/?#]*))?)?@)?([^:\/?#]*)(?::(\d*))?)?((?:[^?#\/]*\/)*[^?#]*)(?:\?([^#]*))?(?:#(.*))?/,
83 loose: /^(?:(?![^:@]+:[^:@\/]*@)([^:\/?#.]+):)?(?:\/\/)?(?:(?:([^:@\/?#]*)(?::([^:@\/?#]*))?)?@)?([^:\/?#]*)(?::(\d*))?((?:\/(?:[^?#](?![^?#\/]*\.[^?#\/.]+(?:[?#]|$)))*\/?)?[^?#\/]*)(?:\?([^#]*))?(?:#(.*))?/
84 },
86 /**
87 * The order here matches the order of captured matches in the `parser` property regexes.
88 *
89 * @private
90 * @static
91 * @property {Array} properties
92 */
93 properties = [
94 'protocol',
95 'user',
96 'password',
97 'host',
98 'port',
99 'path',
100 'query',
101 'fragment'
102 ];
104 /**
105 * @property {string} protocol For example `http` (always present)
106 */
107 /**
108 * @property {string|undefined} user For example `usr`
109 */
110 /**
111 * @property {string|undefined} password For example `pwd`
112 */
113 /**
114 * @property {string} host For example `www.example.com` (always present)
115 */
116 /**
117 * @property {string|undefined} port For example `81`
118 */
119 /**
120 * @property {string} path For example `/dir/dir.2/index.htm` (always present)
121 */
122 /**
123 * @property {Object} query For example `{ a: '0', b: '', c: 'value' }` (always present)
124 */
125 /**
126 * @property {string|undefined} fragment For example `top`
127 */
129 /**
130 * A factory method to create a Uri class with a default location to resolve relative URLs
131 * against (including protocol-relative URLs).
132 *
133 * @method
134 * @param {string|Function} documentLocation A full url, or function returning one.
135 * If passed a function, the return value may change over time and this will be honoured. (T74334)
136 * @member mw
137 */
138 mw.UriRelative = function ( documentLocation ) {
139 var getDefaultUri = ( function () {
140 // Cache
141 var href, uri;
143 return function () {
144 var hrefCur = typeof documentLocation === 'string' ? documentLocation : documentLocation();
145 if ( href === hrefCur ) {
146 return uri;
147 }
148 href = hrefCur;
149 uri = new Uri( href );
150 return uri;
151 };
152 }() );
154 /**
155 * @class mw.Uri
156 * @constructor
157 *
158 * Construct a new URI object. Throws error if arguments are illegal/impossible, or
159 * otherwise don't parse.
160 *
161 * @param {Object|string} [uri] URI string, or an Object with appropriate properties (especially
162 * another URI object to clone). Object must have non-blank `protocol`, `host`, and `path`
163 * properties. If omitted (or set to `undefined`, `null` or empty string), then an object
164 * will be created for the default `uri` of this constructor (`location.href` for mw.Uri,
165 * other values for other instances -- see mw.UriRelative for details).
166 * @param {Object|boolean} [options] Object with options, or (backwards compatibility) a boolean
167 * for strictMode
168 * @param {boolean} [options.strictMode=false] Trigger strict mode parsing of the url.
169 * @param {boolean} [options.overrideKeys=false] Whether to let duplicate query parameters
170 * override each other (`true`) or automagically convert them to an array (`false`).
171 */
172 function Uri( uri, options ) {
173 var prop,
174 defaultUri = getDefaultUri();
176 options = typeof options === 'object' ? options : { strictMode: !!options };
177 options = $.extend( {
178 strictMode: false,
179 overrideKeys: false
180 }, options );
182 if ( uri !== undefined && uri !== null && uri !== '' ) {
183 if ( typeof uri === 'string' ) {
184 this.parse( uri, options );
185 } else if ( typeof uri === 'object' ) {
186 // Copy data over from existing URI object
187 for ( prop in uri ) {
188 // Only copy direct properties, not inherited ones
189 if ( uri.hasOwnProperty( prop ) ) {
190 // Deep copy object properties
191 if ( $.isArray( uri[prop] ) || $.isPlainObject( uri[prop] ) ) {
192 this[prop] = $.extend( true, {}, uri[prop] );
193 } else {
194 this[prop] = uri[prop];
195 }
196 }
197 }
198 if ( !this.query ) {
199 this.query = {};
200 }
201 }
202 } else {
203 // If we didn't get a URI in the constructor, use the default one.
204 return defaultUri.clone();
205 }
207 // protocol-relative URLs
208 if ( !this.protocol ) {
209 this.protocol = defaultUri.protocol;
210 }
211 // No host given:
212 if ( !this.host ) {
213 this.host = defaultUri.host;
214 // port ?
215 if ( !this.port ) {
216 this.port = defaultUri.port;
217 }
218 }
219 if ( this.path && this.path.charAt( 0 ) !== '/' ) {
220 // A real relative URL, relative to defaultUri.path. We can't really handle that since we cannot
221 // figure out whether the last path component of defaultUri.path is a directory or a file.
222 throw new Error( 'Bad constructor arguments' );
223 }
224 if ( !( this.protocol && this.host && this.path ) ) {
225 throw new Error( 'Bad constructor arguments' );
226 }
227 }
229 /**
230 * Encode a value for inclusion in a url.
231 *
232 * Standard encodeURIComponent, with extra stuff to make all browsers work similarly and more
233 * compliant with RFC 3986. Similar to rawurlencode from PHP and our JS library
234 * mw.util.rawurlencode, except this also replaces spaces with `+`.
235 *
236 * @static
237 * @param {string} s String to encode
238 * @return {string} Encoded string for URI
239 */
240 Uri.encode = function ( s ) {
241 return encodeURIComponent( s )
242 .replace( /!/g, '%21' ).replace( /'/g, '%27' ).replace( /\(/g, '%28' )
243 .replace( /\)/g, '%29' ).replace( /\*/g, '%2A' )
244 .replace( /%20/g, '+' );
245 };
247 /**
248 * Decode a url encoded value.
249 *
250 * Reversed #encode. Standard decodeURIComponent, with addition of replacing
251 * `+` with a space.
252 *
253 * @static
254 * @param {string} s String to decode
255 * @return {string} Decoded string
256 */
257 Uri.decode = function ( s ) {
258 return decodeURIComponent( s.replace( /\+/g, '%20' ) );
259 };
261 Uri.prototype = {
263 /**
264 * Parse a string and set our properties accordingly.
265 *
266 * @private
267 * @param {string} str URI, see constructor.
268 * @param {Object} options See constructor.
269 */
270 parse: function ( str, options ) {
271 var q, matches,
272 uri = this;
274 // Apply parser regex and set all properties based on the result
275 matches = parser[ options.strictMode ? 'strict' : 'loose' ].exec( str );
276 $.each( properties, function ( i, property ) {
277 uri[ property ] = matches[ i + 1 ];
278 } );
280 // uri.query starts out as the query string; we will parse it into key-val pairs then make
281 // that object the "query" property.
282 // we overwrite query in uri way to make cloning easier, it can use the same list of properties.
283 q = {};
284 // using replace to iterate over a string
285 if ( uri.query ) {
286 uri.query.replace( /(?:^|&)([^&=]*)(?:(=)([^&]*))?/g, function ( $0, $1, $2, $3 ) {
287 var k, v;
288 if ( $1 ) {
289 k = Uri.decode( $1 );
290 v = ( $2 === '' || $2 === undefined ) ? null : Uri.decode( $3 );
292 // If overrideKeys, always (re)set top level value.
293 // If not overrideKeys but this key wasn't set before, then we set it as well.
294 if ( options.overrideKeys || q[ k ] === undefined ) {
295 q[ k ] = v;
297 // Use arrays if overrideKeys is false and key was already seen before
298 } else {
299 // Once before, still a string, turn into an array
300 if ( typeof q[ k ] === 'string' ) {
301 q[ k ] = [ q[ k ] ];
302 }
303 // Add to the array
304 if ( $.isArray( q[ k ] ) ) {
305 q[ k ].push( v );
306 }
307 }
308 }
309 } );
310 }
311 uri.query = q;
312 },
314 /**
315 * Get user and password section of a URI.
316 *
317 * @return {string}
318 */
319 getUserInfo: function () {
320 return cat( '', this.user, cat( ':', this.password, '' ) );
321 },
323 /**
324 * Get host and port section of a URI.
325 *
326 * @return {string}
327 */
328 getHostPort: function () {
329 return this.host + cat( ':', this.port, '' );
330 },
332 /**
333 * Get the userInfo, host and port section of the URI.
334 *
335 * In most real-world URLs this is simply the hostname, but the definition of 'authority' section is more general.
336 *
337 * @return {string}
338 */
339 getAuthority: function () {
340 return cat( '', this.getUserInfo(), '@' ) + this.getHostPort();
341 },
343 /**
344 * Get the query arguments of the URL, encoded into a string.
345 *
346 * Does not preserve the original order of arguments passed in the URI. Does handle escaping.
347 *
348 * @return {string}
349 */
350 getQueryString: function () {
351 var args = [];
352 $.each( this.query, function ( key, val ) {
353 var k = Uri.encode( key ),
354 vals = $.isArray( val ) ? val : [ val ];
355 $.each( vals, function ( i, v ) {
356 if ( v === null ) {
357 args.push( k );
358 } else if ( k === 'title' ) {
359 args.push( k + '=' + mw.util.wikiUrlencode( v ) );
360 } else {
361 args.push( k + '=' + Uri.encode( v ) );
362 }
363 } );
364 } );
365 return args.join( '&' );
366 },
368 /**
369 * Get everything after the authority section of the URI.
370 *
371 * @return {string}
372 */
373 getRelativePath: function () {
374 return this.path + cat( '?', this.getQueryString(), '', true ) + cat( '#', this.fragment, '' );
375 },
377 /**
378 * Get the entire URI string.
379 *
380 * May not be precisely the same as input due to order of query arguments.
381 *
382 * @return {string} The URI string
383 */
384 toString: function () {
385 return this.protocol + '://' + this.getAuthority() + this.getRelativePath();
386 },
388 /**
389 * Clone this URI
390 *
391 * @return {Object} New URI object with same properties
392 */
393 clone: function () {
394 return new Uri( this );
395 },
397 /**
398 * Extend the query section of the URI with new parameters.
399 *
400 * @param {Object} parameters Query parameters to add to ours (or to override ours with) as an
401 * object
402 * @return {Object} This URI object
403 */
404 extend: function ( parameters ) {
405 $.extend( this.query, parameters );
406 return this;
407 }
408 };
410 return Uri;
411 };
413 // Default to the current browsing location (for relative URLs).
414 mw.Uri = mw.UriRelative( function () {
415 return location.href;
416 } );
418 }( mediaWiki, jQuery ) );