| blob | 95263ec4517c7c4d8b33c43cc69fa03f559ef6af |
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 /* eslint-disable no-use-before-define */
55 ( function ( mw, $ ) {
56 var parser, properties;
58 /**
59 * Function that's useful when constructing the URI string -- we frequently encounter the pattern
60 * of having to add something to the URI as we go, but only if it's present, and to include a
61 * character before or after if so.
62 *
63 * @private
64 * @static
65 * @param {string|undefined} pre To prepend
66 * @param {string} val To include
67 * @param {string} post To append
68 * @param {boolean} raw If true, val will not be encoded
69 * @return {string} Result
70 */
71 function cat( pre, val, post, raw ) {
72 if ( val === undefined || val === null || val === '' ) {
73 return '';
74 }
76 return pre + ( raw ? val : mw.Uri.encode( val ) ) + post;
77 }
79 /**
80 * Regular expressions to parse many common URIs.
81 *
82 * As they are gnarly, they have been moved to separate files to allow us to format them in the
83 * 'extended' regular expression format (which JavaScript normally doesn't support). The subset of
84 * features handled is minimal, but just the free whitespace gives us a lot.
85 *
86 * @private
87 * @static
88 * @property {Object} parser
89 */
90 parser = {
91 strict: mw.template.get( 'mediawiki.Uri', 'strict.regexp' ).render(),
92 loose: mw.template.get( 'mediawiki.Uri', 'loose.regexp' ).render()
93 };
95 /**
96 * The order here matches the order of captured matches in the `parser` property regexes.
97 *
98 * @private
99 * @static
100 * @property {Array} properties
101 */
102 properties = [
103 'protocol',
104 'user',
105 'password',
106 'host',
107 'port',
108 'path',
109 'query',
110 'fragment'
111 ];
113 /**
114 * @property {string} protocol For example `http` (always present)
115 */
116 /**
117 * @property {string|undefined} user For example `usr`
118 */
119 /**
120 * @property {string|undefined} password For example `pwd`
121 */
122 /**
123 * @property {string} host For example `www.example.com` (always present)
124 */
125 /**
126 * @property {string|undefined} port For example `81`
127 */
128 /**
129 * @property {string} path For example `/dir/dir.2/index.htm` (always present)
130 */
131 /**
132 * @property {Object} query For example `{ a: '0', b: '', c: 'value' }` (always present)
133 */
134 /**
135 * @property {string|undefined} fragment For example `top`
136 */
138 /**
139 * A factory method to create a Uri class with a default location to resolve relative URLs
140 * against (including protocol-relative URLs).
141 *
142 * @method
143 * @param {string|Function} documentLocation A full url, or function returning one.
144 * If passed a function, the return value may change over time and this will be honoured. (T74334)
145 * @member mw
146 * @return {Function} Uri class
147 */
148 mw.UriRelative = function ( documentLocation ) {
149 var getDefaultUri = ( function () {
150 // Cache
151 var href, uri;
153 return function () {
154 var hrefCur = typeof documentLocation === 'string' ? documentLocation : documentLocation();
155 if ( href === hrefCur ) {
156 return uri;
157 }
158 href = hrefCur;
159 uri = new Uri( href );
160 return uri;
161 };
162 }() );
164 /**
165 * Construct a new URI object. Throws error if arguments are illegal/impossible, or
166 * otherwise don't parse.
167 *
168 * @class mw.Uri
169 * @constructor
170 * @param {Object|string} [uri] URI string, or an Object with appropriate properties (especially
171 * another URI object to clone). Object must have non-blank `protocol`, `host`, and `path`
172 * properties. If omitted (or set to `undefined`, `null` or empty string), then an object
173 * will be created for the default `uri` of this constructor (`location.href` for mw.Uri,
174 * other values for other instances -- see mw.UriRelative for details).
175 * @param {Object|boolean} [options] Object with options, or (backwards compatibility) a boolean
176 * for strictMode
177 * @param {boolean} [options.strictMode=false] Trigger strict mode parsing of the url.
178 * @param {boolean} [options.overrideKeys=false] Whether to let duplicate query parameters
179 * override each other (`true`) or automagically convert them to an array (`false`).
180 */
181 function Uri( uri, options ) {
182 var prop, hrefCur,
183 hasOptions = ( options !== undefined ),
184 defaultUri = getDefaultUri();
186 options = typeof options === 'object' ? options : { strictMode: !!options };
187 options = $.extend( {
188 strictMode: false,
189 overrideKeys: false
190 }, options );
192 if ( uri !== undefined && uri !== null && uri !== '' ) {
193 if ( typeof uri === 'string' ) {
194 this.parse( uri, options );
195 } else if ( typeof uri === 'object' ) {
196 // Copy data over from existing URI object
197 for ( prop in uri ) {
198 // Only copy direct properties, not inherited ones
199 if ( uri.hasOwnProperty( prop ) ) {
200 // Deep copy object properties
201 if ( $.isArray( uri[ prop ] ) || $.isPlainObject( uri[ prop ] ) ) {
202 this[ prop ] = $.extend( true, {}, uri[ prop ] );
203 } else {
204 this[ prop ] = uri[ prop ];
205 }
206 }
207 }
208 if ( !this.query ) {
209 this.query = {};
210 }
211 }
212 } else if ( hasOptions ) {
213 // We didn't get a URI in the constructor, but we got options.
214 hrefCur = typeof documentLocation === 'string' ? documentLocation : documentLocation();
215 this.parse( hrefCur, options );
216 } else {
217 // We didn't get a URI or options in the constructor, use the default instance.
218 return defaultUri.clone();
219 }
221 // protocol-relative URLs
222 if ( !this.protocol ) {
223 this.protocol = defaultUri.protocol;
224 }
225 // No host given:
226 if ( !this.host ) {
227 this.host = defaultUri.host;
228 // port ?
229 if ( !this.port ) {
230 this.port = defaultUri.port;
231 }
232 }
233 if ( this.path && this.path[ 0 ] !== '/' ) {
234 // A real relative URL, relative to defaultUri.path. We can't really handle that since we cannot
235 // figure out whether the last path component of defaultUri.path is a directory or a file.
236 throw new Error( 'Bad constructor arguments' );
237 }
238 if ( !( this.protocol && this.host && this.path ) ) {
239 throw new Error( 'Bad constructor arguments' );
240 }
241 }
243 /**
244 * Encode a value for inclusion in a url.
245 *
246 * Standard encodeURIComponent, with extra stuff to make all browsers work similarly and more
247 * compliant with RFC 3986. Similar to rawurlencode from PHP and our JS library
248 * mw.util.rawurlencode, except this also replaces spaces with `+`.
249 *
250 * @static
251 * @param {string} s String to encode
252 * @return {string} Encoded string for URI
253 */
254 Uri.encode = function ( s ) {
255 return encodeURIComponent( s )
256 .replace( /!/g, '%21' ).replace( /'/g, '%27' ).replace( /\(/g, '%28' )
257 .replace( /\)/g, '%29' ).replace( /\*/g, '%2A' )
258 .replace( /%20/g, '+' );
259 };
261 /**
262 * Decode a url encoded value.
263 *
264 * Reversed #encode. Standard decodeURIComponent, with addition of replacing
265 * `+` with a space.
266 *
267 * @static
268 * @param {string} s String to decode
269 * @return {string} Decoded string
270 */
271 Uri.decode = function ( s ) {
272 return decodeURIComponent( s.replace( /\+/g, '%20' ) );
273 };
275 Uri.prototype = {
277 /**
278 * Parse a string and set our properties accordingly.
279 *
280 * @private
281 * @param {string} str URI, see constructor.
282 * @param {Object} options See constructor.
283 */
284 parse: function ( str, options ) {
285 var q, matches,
286 uri = this,
287 hasOwn = Object.prototype.hasOwnProperty;
289 // Apply parser regex and set all properties based on the result
290 matches = parser[ options.strictMode ? 'strict' : 'loose' ].exec( str );
291 $.each( properties, function ( i, property ) {
292 uri[ property ] = matches[ i + 1 ];
293 } );
295 // uri.query starts out as the query string; we will parse it into key-val pairs then make
296 // that object the "query" property.
297 // we overwrite query in uri way to make cloning easier, it can use the same list of properties.
298 q = {};
299 // using replace to iterate over a string
300 if ( uri.query ) {
301 uri.query.replace( /(?:^|&)([^&=]*)(?:(=)([^&]*))?/g, function ( $0, $1, $2, $3 ) {
302 var k, v;
303 if ( $1 ) {
304 k = Uri.decode( $1 );
305 v = ( $2 === '' || $2 === undefined ) ? null : Uri.decode( $3 );
307 // If overrideKeys, always (re)set top level value.
308 // If not overrideKeys but this key wasn't set before, then we set it as well.
309 if ( options.overrideKeys || !hasOwn.call( q, k ) ) {
310 q[ k ] = v;
312 // Use arrays if overrideKeys is false and key was already seen before
313 } else {
314 // Once before, still a string, turn into an array
315 if ( typeof q[ k ] === 'string' ) {
316 q[ k ] = [ q[ k ] ];
317 }
318 // Add to the array
319 if ( $.isArray( q[ k ] ) ) {
320 q[ k ].push( v );
321 }
322 }
323 }
324 } );
325 }
326 uri.query = q;
327 },
329 /**
330 * Get user and password section of a URI.
331 *
332 * @return {string}
333 */
334 getUserInfo: function () {
335 return cat( '', this.user, cat( ':', this.password, '' ) );
336 },
338 /**
339 * Get host and port section of a URI.
340 *
341 * @return {string}
342 */
343 getHostPort: function () {
344 return this.host + cat( ':', this.port, '' );
345 },
347 /**
348 * Get the userInfo, host and port section of the URI.
349 *
350 * In most real-world URLs this is simply the hostname, but the definition of 'authority' section is more general.
351 *
352 * @return {string}
353 */
354 getAuthority: function () {
355 return cat( '', this.getUserInfo(), '@' ) + this.getHostPort();
356 },
358 /**
359 * Get the query arguments of the URL, encoded into a string.
360 *
361 * Does not preserve the original order of arguments passed in the URI. Does handle escaping.
362 *
363 * @return {string}
364 */
365 getQueryString: function () {
366 var args = [];
367 $.each( this.query, function ( key, val ) {
368 var k = Uri.encode( key ),
369 vals = $.isArray( val ) ? val : [ val ];
370 $.each( vals, function ( i, v ) {
371 if ( v === null ) {
372 args.push( k );
373 } else if ( k === 'title' ) {
374 args.push( k + '=' + mw.util.wikiUrlencode( v ) );
375 } else {
376 args.push( k + '=' + Uri.encode( v ) );
377 }
378 } );
379 } );
380 return args.join( '&' );
381 },
383 /**
384 * Get everything after the authority section of the URI.
385 *
386 * @return {string}
387 */
388 getRelativePath: function () {
389 return this.path + cat( '?', this.getQueryString(), '', true ) + cat( '#', this.fragment, '' );
390 },
392 /**
393 * Get the entire URI string.
394 *
395 * May not be precisely the same as input due to order of query arguments.
396 *
397 * @return {string} The URI string
398 */
399 toString: function () {
400 return this.protocol + '://' + this.getAuthority() + this.getRelativePath();
401 },
403 /**
404 * Clone this URI
405 *
406 * @return {Object} New URI object with same properties
407 */
408 clone: function () {
409 return new Uri( this );
410 },
412 /**
413 * Extend the query section of the URI with new parameters.
414 *
415 * @param {Object} parameters Query parameters to add to ours (or to override ours with) as an
416 * object
417 * @return {Object} This URI object
418 */
419 extend: function ( parameters ) {
420 $.extend( this.query, parameters );
421 return this;
422 }
423 };
425 return Uri;
426 };
428 // Default to the current browsing location (for relative URLs).
429 mw.Uri = mw.UriRelative( function () {
430 return location.href;
431 } );
433 }( mediaWiki, jQuery ) );