0
0
mirror of https://github.com/nodejs/node.git synced 2024-11-22 15:47:56 +01:00
nodejs/lib/url.js
J. Lee Coltrane 54d293da56 url: make url.format escape delimiters in path and query
`url.format` should escape ? and # chars in pathname, and # chars in
search, because they change the semantics of the operation otherwise.
Don't escape % chars, or anything else. (see: #4082)
2012-10-30 09:16:13 -07:00

690 lines
21 KiB
JavaScript

// Copyright Joyent, Inc. and other Node contributors.
//
// Permission is hereby granted, free of charge, to any person obtaining a
// copy of this software and associated documentation files (the
// "Software"), to deal in the Software without restriction, including
// without limitation the rights to use, copy, modify, merge, publish,
// distribute, sublicense, and/or sell copies of the Software, and to permit
// persons to whom the Software is furnished to do so, subject to the
// following conditions:
//
// The above copyright notice and this permission notice shall be included
// in all copies or substantial portions of the Software.
//
// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
// OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
// MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN
// NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM,
// DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
// OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
// USE OR OTHER DEALINGS IN THE SOFTWARE.
var punycode = require('punycode');
exports.parse = urlParse;
exports.resolve = urlResolve;
exports.resolveObject = urlResolveObject;
exports.format = urlFormat;
exports.Url = Url;
function Url() {
this.protocol = null;
this.slashes = null;
this.auth = null;
this.host = null;
this.port = null;
this.hostname = null;
this.hash = null;
this.search = null;
this.query = null;
this.pathname = null;
this.path = null;
}
// Reference: RFC 3986, RFC 1808, RFC 2396
// define these here so at least they only have to be
// compiled once on the first module load.
var protocolPattern = /^([a-z0-9.+-]+:)/i,
portPattern = /:[0-9]*$/,
// RFC 2396: characters reserved for delimiting URLs.
// We actually just auto-escape these.
delims = ['<', '>', '"', '`', ' ', '\r', '\n', '\t'],
// RFC 2396: characters not allowed for various reasons.
unwise = ['{', '}', '|', '\\', '^', '~', '`'].concat(delims),
// Allowed by RFCs, but cause of XSS attacks. Always escape these.
autoEscape = ['\''].concat(delims),
// Characters that are never ever allowed in a hostname.
// Note that any invalid chars are also handled, but these
// are the ones that are *expected* to be seen, so we fast-path
// them.
nonHostChars = ['%', '/', '?', ';', '#']
.concat(unwise).concat(autoEscape),
nonAuthChars = ['/', '@', '?', '#'].concat(delims),
hostnameMaxLen = 255,
hostnamePartPattern = /^[a-z0-9A-Z_-]{0,63}$/,
hostnamePartStart = /^([a-z0-9A-Z_-]{0,63})(.*)$/,
// protocols that can allow "unsafe" and "unwise" chars.
unsafeProtocol = {
'javascript': true,
'javascript:': true
},
// protocols that never have a hostname.
hostlessProtocol = {
'javascript': true,
'javascript:': true
},
// protocols that always have a path component.
pathedProtocol = {
'http': true,
'https': true,
'ftp': true,
'gopher': true,
'file': true,
'http:': true,
'ftp:': true,
'gopher:': true,
'file:': true
},
// protocols that always contain a // bit.
slashedProtocol = {
'http': true,
'https': true,
'ftp': true,
'gopher': true,
'file': true,
'http:': true,
'https:': true,
'ftp:': true,
'gopher:': true,
'file:': true
},
querystring = require('querystring');
function urlParse(url, parseQueryString, slashesDenoteHost) {
if (url && typeof(url) === 'object' && url instanceof Url) return url;
var u = new Url;
u.parse(url, parseQueryString, slashesDenoteHost);
return u;
}
Url.prototype.parse = function(url, parseQueryString, slashesDenoteHost) {
if (typeof url !== 'string') {
throw new TypeError("Parameter 'url' must be a string, not " + typeof url);
}
var rest = url;
// trim before proceeding.
// This is to support parse stuff like " http://foo.com \n"
rest = rest.trim();
var proto = protocolPattern.exec(rest);
if (proto) {
proto = proto[0];
var lowerProto = proto.toLowerCase();
this.protocol = lowerProto;
rest = rest.substr(proto.length);
}
// figure out if it's got a host
// user@server is *always* interpreted as a hostname, and url
// resolution will treat //foo/bar as host=foo,path=bar because that's
// how the browser resolves relative URLs.
if (slashesDenoteHost || proto || rest.match(/^\/\/[^@\/]+@[^@\/]+/)) {
var slashes = rest.substr(0, 2) === '//';
if (slashes && !(proto && hostlessProtocol[proto])) {
rest = rest.substr(2);
this.slashes = true;
}
}
if (!hostlessProtocol[proto] &&
(slashes || (proto && !slashedProtocol[proto]))) {
// there's a hostname.
// the first instance of /, ?, ;, or # ends the host.
// don't enforce full RFC correctness, just be unstupid about it.
// If there is an @ in the hostname, then non-host chars *are* allowed
// to the left of the first @ sign, unless some non-auth character
// comes *before* the @-sign.
// URLs are obnoxious.
var atSign = rest.indexOf('@');
if (atSign !== -1) {
var auth = rest.slice(0, atSign);
// there *may be* an auth
var hasAuth = true;
for (var i = 0, l = nonAuthChars.length; i < l; i++) {
if (auth.indexOf(nonAuthChars[i]) !== -1) {
// not a valid auth. Something like http://foo.com/bar@baz/
hasAuth = false;
break;
}
}
if (hasAuth) {
// pluck off the auth portion.
this.auth = decodeURIComponent(auth);
rest = rest.substr(atSign + 1);
}
}
var firstNonHost = -1;
for (var i = 0, l = nonHostChars.length; i < l; i++) {
var index = rest.indexOf(nonHostChars[i]);
if (index !== -1 &&
(firstNonHost < 0 || index < firstNonHost)) firstNonHost = index;
}
if (firstNonHost !== -1) {
this.host = rest.substr(0, firstNonHost);
rest = rest.substr(firstNonHost);
} else {
this.host = rest;
rest = '';
}
// pull out port.
this.parseHost();
// we've indicated that there is a hostname,
// so even if it's empty, it has to be present.
this.hostname = this.hostname || '';
// if hostname begins with [ and ends with ]
// assume that it's an IPv6 address.
var ipv6Hostname = this.hostname[0] === '[' &&
this.hostname[this.hostname.length - 1] === ']';
// validate a little.
if (!ipv6Hostname) {
var hostparts = this.hostname.split(/\./);
for (var i = 0, l = hostparts.length; i < l; i++) {
var part = hostparts[i];
if (!part) continue;
if (!part.match(hostnamePartPattern)) {
var newpart = '';
for (var j = 0, k = part.length; j < k; j++) {
if (part.charCodeAt(j) > 127) {
// we replace non-ASCII char with a temporary placeholder
// we need this to make sure size of hostname is not
// broken by replacing non-ASCII by nothing
newpart += 'x';
} else {
newpart += part[j];
}
}
// we test again with ASCII char only
if (!newpart.match(hostnamePartPattern)) {
var validParts = hostparts.slice(0, i);
var notHost = hostparts.slice(i + 1);
var bit = part.match(hostnamePartStart);
if (bit) {
validParts.push(bit[1]);
notHost.unshift(bit[2]);
}
if (notHost.length) {
rest = '/' + notHost.join('.') + rest;
}
this.hostname = validParts.join('.');
break;
}
}
}
}
if (this.hostname.length > hostnameMaxLen) {
this.hostname = '';
} else {
// hostnames are always lower case.
this.hostname = this.hostname.toLowerCase();
}
if (!ipv6Hostname) {
// IDNA Support: Returns a puny coded representation of "domain".
// It only converts the part of the domain name that
// has non ASCII characters. I.e. it dosent matter if
// you call it with a domain that already is in ASCII.
var domainArray = this.hostname.split('.');
var newOut = [];
for (var i = 0; i < domainArray.length; ++i) {
var s = domainArray[i];
newOut.push(s.match(/[^A-Za-z0-9_-]/) ?
'xn--' + punycode.encode(s) : s);
}
this.hostname = newOut.join('.');
}
var p = this.port ? ':' + this.port : '';
var h = this.hostname || '';
this.host = h + p;
this.href += this.host;
// strip [ and ] from the hostname
// the host field still retains them, though
if (ipv6Hostname) {
this.hostname = this.hostname.substr(1, this.hostname.length - 2);
if (rest[0] !== '/') {
rest = '/' + rest;
}
}
}
// now rest is set to the post-host stuff.
// chop off any delim chars.
if (!unsafeProtocol[lowerProto]) {
// First, make 100% sure that any "autoEscape" chars get
// escaped, even if encodeURIComponent doesn't think they
// need to be.
for (var i = 0, l = autoEscape.length; i < l; i++) {
var ae = autoEscape[i];
var esc = encodeURIComponent(ae);
if (esc === ae) {
esc = escape(ae);
}
rest = rest.split(ae).join(esc);
}
}
// chop off from the tail first.
var hash = rest.indexOf('#');
if (hash !== -1) {
// got a fragment string.
this.hash = rest.substr(hash);
rest = rest.slice(0, hash);
}
var qm = rest.indexOf('?');
if (qm !== -1) {
this.search = rest.substr(qm);
this.query = rest.substr(qm + 1);
if (parseQueryString) {
this.query = querystring.parse(this.query);
}
rest = rest.slice(0, qm);
} else if (parseQueryString) {
// no query string, but parseQueryString still requested
this.search = '';
this.query = {};
}
if (rest) this.pathname = rest;
if (slashedProtocol[proto] &&
this.hostname && !this.pathname) {
this.pathname = '/';
}
//to support http.request
if (this.pathname || this.search) {
var p = this.pathname || '';
var s = this.search || '';
this.path = p + s;
}
// finally, reconstruct the href based on what has been validated.
this.href = this.format();
return this;
};
// format a parsed object into a url string
function urlFormat(obj) {
// ensure it's an object, and not a string url.
// If it's an obj, this is a no-op.
// this way, you can call url_format() on strings
// to clean up potentially wonky urls.
if (typeof(obj) === 'string') obj = urlParse(obj);
if (!(obj instanceof Url)) return Url.prototype.format.call(obj);
return obj.format();
}
Url.prototype.format = function() {
var auth = this.auth || '';
if (auth) {
auth = encodeURIComponent(auth);
auth = auth.replace(/%3A/i, ':');
auth += '@';
}
var protocol = this.protocol || '',
pathname = this.pathname || '',
hash = this.hash || '',
host = false,
query = '';
if (this.host) {
host = auth + this.host;
} else if (this.hostname) {
host = auth + (this.hostname.indexOf(':') === -1 ?
this.hostname :
'[' + this.hostname + ']');
if (this.port) {
host += ':' + this.port;
}
}
if (this.query && typeof this.query === 'object' &&
Object.keys(this.query).length) {
query = querystring.stringify(this.query);
}
var search = this.search || (query && ('?' + query)) || '';
if (protocol && protocol.substr(-1) !== ':') protocol += ':';
// only the slashedProtocols get the //. Not mailto:, xmpp:, etc.
// unless they had them to begin with.
if (this.slashes ||
(!protocol || slashedProtocol[protocol]) && host !== false) {
host = '//' + (host || '');
if (pathname && pathname.charAt(0) !== '/') pathname = '/' + pathname;
} else if (!host) {
host = '';
}
if (hash && hash.charAt(0) !== '#') hash = '#' + hash;
if (search && search.charAt(0) !== '?') search = '?' + search;
pathname = pathname.replace(/[?#]/g, function(match) {
return encodeURIComponent(match);
});
search = search.replace('#', '%23');
return protocol + host + pathname + search + hash;
};
function urlResolve(source, relative) {
return urlParse(source, false, true).resolve(relative);
}
Url.prototype.resolve = function(relative) {
return this.resolveObject(urlParse(relative, false, true)).format();
};
function urlResolveObject(source, relative) {
if (!source) return relative;
return urlParse(source, false, true).resolveObject(relative);
}
Url.prototype.resolveObject = function(relative) {
if (typeof relative === 'string') {
var rel = new Url();
rel.parse(relative, false, true);
relative = rel;
}
var result = new Url();
Object.keys(this).forEach(function(k) {
result[k] = this[k];
}, this);
// hash is always overridden, no matter what.
// even href="" will remove it.
result.hash = relative.hash;
// if the relative url is empty, then there's nothing left to do here.
if (relative.href === '') {
result.href = result.format();
return result;
}
// hrefs like //foo/bar always cut to the protocol.
if (relative.slashes && !relative.protocol) {
// take everything except the protocol from relative
Object.keys(relative).forEach(function(k) {
if (k !== 'protocol')
result[k] = relative[k];
});
//urlParse appends trailing / to urls like http://www.example.com
if (slashedProtocol[result.protocol] &&
result.hostname && !result.pathname) {
result.path = result.pathname = '/';
}
result.href = result.format();
return result;
}
if (relative.protocol && relative.protocol !== result.protocol) {
// if it's a known url protocol, then changing
// the protocol does weird things
// first, if it's not file:, then we MUST have a host,
// and if there was a path
// to begin with, then we MUST have a path.
// if it is file:, then the host is dropped,
// because that's known to be hostless.
// anything else is assumed to be absolute.
if (!slashedProtocol[relative.protocol]) {
Object.keys(relative).forEach(function(k) {
result[k] = relative[k];
});
result.href = result.format();
return result;
}
result.protocol = relative.protocol;
if (!relative.host && !hostlessProtocol[relative.protocol]) {
var relPath = (relative.pathname || '').split('/');
while (relPath.length && !(relative.host = relPath.shift()));
if (!relative.host) relative.host = '';
if (!relative.hostname) relative.hostname = '';
if (relPath[0] !== '') relPath.unshift('');
if (relPath.length < 2) relPath.unshift('');
result.pathname = relPath.join('/');
} else {
result.pathname = relative.pathname;
}
result.search = relative.search;
result.query = relative.query;
result.host = relative.host || '';
result.auth = relative.auth;
result.hostname = relative.hostname || relative.host;
result.port = relative.port;
// to support http.request
if (result.pathname || result.search) {
var p = result.pathname || '';
var s = result.search || '';
result.path = p + s;
}
result.slashes = result.slashes || relative.slashes;
result.href = result.format();
return result;
}
var isSourceAbs = (result.pathname && result.pathname.charAt(0) === '/'),
isRelAbs = (
relative.host ||
relative.pathname && relative.pathname.charAt(0) === '/'
),
mustEndAbs = (isRelAbs || isSourceAbs ||
(result.host && relative.pathname)),
removeAllDots = mustEndAbs,
srcPath = result.pathname && result.pathname.split('/') || [],
relPath = relative.pathname && relative.pathname.split('/') || [],
psychotic = result.protocol && !slashedProtocol[result.protocol];
// if the url is a non-slashed url, then relative
// links like ../.. should be able
// to crawl up to the hostname, as well. This is strange.
// result.protocol has already been set by now.
// Later on, put the first path part into the host field.
if (psychotic) {
result.hostname = '';
result.port = null;
if (result.host) {
if (srcPath[0] === '') srcPath[0] = result.host;
else srcPath.unshift(result.host);
}
result.host = '';
if (relative.protocol) {
relative.hostname = null;
relative.port = null;
if (relative.host) {
if (relPath[0] === '') relPath[0] = relative.host;
else relPath.unshift(relative.host);
}
relative.host = null;
}
mustEndAbs = mustEndAbs && (relPath[0] === '' || srcPath[0] === '');
}
if (isRelAbs) {
// it's absolute.
result.host = (relative.host || relative.host === '') ?
relative.host : result.host;
result.hostname = (relative.hostname || relative.hostname === '') ?
relative.hostname : result.hostname;
result.search = relative.search;
result.query = relative.query;
srcPath = relPath;
// fall through to the dot-handling below.
} else if (relPath.length) {
// it's relative
// throw away the existing file, and take the new path instead.
if (!srcPath) srcPath = [];
srcPath.pop();
srcPath = srcPath.concat(relPath);
result.search = relative.search;
result.query = relative.query;
} else if (relative.search !== null && relative.search !== undefined) {
// just pull out the search.
// like href='?foo'.
// Put this after the other two cases because it simplifies the booleans
if (psychotic) {
result.hostname = result.host = srcPath.shift();
//occationaly the auth can get stuck only in host
//this especialy happens in cases like
//url.resolveObject('mailto:local1@domain1', 'local2@domain2')
var authInHost = result.host && result.host.indexOf('@') > 0 ?
result.host.split('@') : false;
if (authInHost) {
result.auth = authInHost.shift();
result.host = result.hostname = authInHost.shift();
}
}
result.search = relative.search;
result.query = relative.query;
//to support http.request
if (result.pathname !== null || result.search !== null) {
result.path = (result.pathname ? result.pathname : '') +
(result.search ? result.search : '');
}
result.href = result.format();
return result;
}
if (!srcPath.length) {
// no path at all. easy.
// we've already handled the other stuff above.
result.pathname = null;
//to support http.request
if (result.search) {
result.path = '/' + result.search;
} else {
result.path = null;
}
result.href = result.format();
return result;
}
// if a url ENDs in . or .., then it must get a trailing slash.
// however, if it ends in anything else non-slashy,
// then it must NOT get a trailing slash.
var last = srcPath.slice(-1)[0];
var hasTrailingSlash = (
(result.host || relative.host) && (last === '.' || last === '..') ||
last === '');
// strip single dots, resolve double dots to parent dir
// if the path tries to go above the root, `up` ends up > 0
var up = 0;
for (var i = srcPath.length; i >= 0; i--) {
last = srcPath[i];
if (last == '.') {
srcPath.splice(i, 1);
} else if (last === '..') {
srcPath.splice(i, 1);
up++;
} else if (up) {
srcPath.splice(i, 1);
up--;
}
}
// if the path is allowed to go above the root, restore leading ..s
if (!mustEndAbs && !removeAllDots) {
for (; up--; up) {
srcPath.unshift('..');
}
}
if (mustEndAbs && srcPath[0] !== '' &&
(!srcPath[0] || srcPath[0].charAt(0) !== '/')) {
srcPath.unshift('');
}
if (hasTrailingSlash && (srcPath.join('/').substr(-1) !== '/')) {
srcPath.push('');
}
var isAbsolute = srcPath[0] === '' ||
(srcPath[0] && srcPath[0].charAt(0) === '/');
// put the host back
if (psychotic) {
result.hostname = result.host = isAbsolute ? '' :
srcPath.length ? srcPath.shift() : '';
//occationaly the auth can get stuck only in host
//this especialy happens in cases like
//url.resolveObject('mailto:local1@domain1', 'local2@domain2')
var authInHost = result.host && result.host.indexOf('@') > 0 ?
result.host.split('@') : false;
if (authInHost) {
result.auth = authInHost.shift();
result.host = result.hostname = authInHost.shift();
}
}
mustEndAbs = mustEndAbs || (result.host && srcPath.length);
if (mustEndAbs && !isAbsolute) {
srcPath.unshift('');
}
if (!srcPath.length) {
result.pathname = null;
result.path = null;
} else {
result.pathname = srcPath.join('/');
}
//to support request.http
if (result.pathname !== null || result.search !== null) {
result.path = (result.pathname ? result.pathname : '') +
(result.search ? result.search : '');
}
result.auth = relative.auth || result.auth;
result.slashes = result.slashes || relative.slashes;
result.href = result.format();
return result;
};
Url.prototype.parseHost = function() {
var host = this.host;
var port = portPattern.exec(host);
if (port) {
port = port[0];
if (port !== ':') {
this.port = port.substr(1);
}
host = host.substr(0, host.length - port.length);
}
if (host) this.hostname = host;
};