'use strict'; function importPunycode() { try { return process.binding('icu'); } catch (e) { return require('punycode'); } } const { toASCII } = importPunycode(); const internalUrl = require('internal/url'); const encodeAuth = internalUrl.encodeAuth; exports.parse = urlParse; exports.resolve = urlResolve; exports.resolveObject = urlResolveObject; exports.format = urlFormat; exports.URL = internalUrl.URL; exports.originFor = internalUrl.originFor; exports.domainToASCII = internalUrl.domainToASCII; exports.domainToUnicode = internalUrl.domainToUnicode; 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; this.href = 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. const protocolPattern = /^([a-z0-9.+-]+:)/i; const portPattern = /:[0-9]*$/; const hostPattern = /^\/\/[^@/]+@[^@/]+/; // Special case for a simple path URL const simplePathPattern = /^(\/\/?(?!\/)[^?\s]*)(\?[^\s]*)?$/; const hostnameMaxLen = 255; // protocols that can allow "unsafe" and "unwise" chars. const unsafeProtocol = { 'javascript': true, 'javascript:': true }; // protocols that never have a hostname. const hostlessProtocol = { 'javascript': true, 'javascript:': true }; // protocols that always contain a // bit. const slashedProtocol = { 'http': true, 'http:': true, 'https': true, 'https:': true, 'ftp': true, 'ftp:': true, 'gopher': true, 'gopher:': true, 'file': true, 'file:': true }; const querystring = require('querystring'); // This constructor is used to store parsed query string values. Instantiating // this is faster than explicitly calling `Object.create(null)` to get a // "clean" empty object (tested with v8 v4.9). function ParsedQueryString() {} ParsedQueryString.prototype = Object.create(null); function urlParse(url, parseQueryString, slashesDenoteHost) { if (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); } // Copy chrome, IE, opera backslash-handling behavior. // Back slashes before the query string get converted to forward slashes // See: https://code.google.com/p/chromium/issues/detail?id=25916 var hasHash = false; var start = -1; var end = -1; var rest = ''; var lastPos = 0; var i = 0; for (var inWs = false, split = false; i < url.length; ++i) { const code = url.charCodeAt(i); // Find first and last non-whitespace characters for trimming const isWs = code === 32/* */ || code === 9/*\t*/ || code === 13/*\r*/ || code === 10/*\n*/ || code === 12/*\f*/ || code === 160/*\u00A0*/ || code === 65279/*\uFEFF*/; if (start === -1) { if (isWs) continue; lastPos = start = i; } else { if (inWs) { if (!isWs) { end = -1; inWs = false; } } else if (isWs) { end = i; inWs = true; } } // Only convert backslashes while we haven't seen a split character if (!split) { switch (code) { case 35: // '#' hasHash = true; // Fall through case 63: // '?' split = true; break; case 92: // '\\' if (i - lastPos > 0) rest += url.slice(lastPos, i); rest += '/'; lastPos = i + 1; break; } } else if (!hasHash && code === 35/*#*/) { hasHash = true; } } // Check if string was non-empty (including strings with only whitespace) if (start !== -1) { if (lastPos === start) { // We didn't convert any backslashes if (end === -1) { if (start === 0) rest = url; else rest = url.slice(start); } else { rest = url.slice(start, end); } } else if (end === -1 && lastPos < url.length) { // We converted some backslashes and have only part of the entire string rest += url.slice(lastPos); } else if (end !== -1 && lastPos < end) { // We converted some backslashes and have only part of the entire string rest += url.slice(lastPos, end); } } if (!slashesDenoteHost && !hasHash) { // Try fast path regexp const simplePath = simplePathPattern.exec(rest); if (simplePath) { this.path = rest; this.href = rest; this.pathname = simplePath[1]; if (simplePath[2]) { this.search = simplePath[2]; if (parseQueryString) { this.query = querystring.parse(this.search.slice(1)); } else { this.query = this.search.slice(1); } } else if (parseQueryString) { this.search = ''; this.query = new ParsedQueryString(); } return this; } } var proto = protocolPattern.exec(rest); if (proto) { proto = proto[0]; var lowerProto = proto.toLowerCase(); this.protocol = lowerProto; rest = rest.slice(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 || hostPattern.test(rest)) { var slashes = rest.charCodeAt(0) === 47/*/*/ && rest.charCodeAt(1) === 47/*/*/; if (slashes && !(proto && hostlessProtocol[proto])) { rest = rest.slice(2); this.slashes = true; } } if (!hostlessProtocol[proto] && (slashes || (proto && !slashedProtocol[proto]))) { // there's a hostname. // the first instance of /, ?, ;, or # ends the host. // // If there is an @ in the hostname, then non-host chars *are* allowed // to the left of the last @ sign, unless some host-ending character // comes *before* the @-sign. // URLs are obnoxious. // // ex: // http://a@b@c/ => user:a@b host:c // http://a@b?@c => user:a host:b path:/?@c // v0.12 TODO(isaacs): This is not quite how Chrome does things. // Review our test case against browsers more comprehensively. var hostEnd = -1; var atSign = -1; var nonHost = -1; for (i = 0; i < rest.length; ++i) { switch (rest.charCodeAt(i)) { case 9: // '\t' case 10: // '\n' case 13: // '\r' case 32: // ' ' case 34: // '"' case 37: // '%' case 39: // '\'' case 59: // ';' case 60: // '<' case 62: // '>' case 92: // '\\' case 94: // '^' case 96: // '`' case 123: // '{' case 124: // '|' case 125: // '}' // Characters that are never ever allowed in a hostname from RFC 2396 if (nonHost === -1) nonHost = i; break; case 35: // '#' case 47: // '/' case 63: // '?' // Find the first instance of any host-ending characters if (nonHost === -1) nonHost = i; hostEnd = i; break; case 64: // '@' // At this point, either we have an explicit point where the // auth portion cannot go past, or the last @ char is the decider. atSign = i; nonHost = -1; break; } if (hostEnd !== -1) break; } start = 0; if (atSign !== -1) { this.auth = decodeURIComponent(rest.slice(0, atSign)); start = atSign + 1; } if (nonHost === -1) { this.host = rest.slice(start); rest = ''; } else { this.host = rest.slice(start, nonHost); rest = rest.slice(nonHost); } // pull out port. this.parseHost(); // we've indicated that there is a hostname, // so even if it's empty, it has to be present. if (typeof this.hostname !== 'string') this.hostname = ''; var hostname = this.hostname; // if hostname begins with [ and ends with ] // assume that it's an IPv6 address. var ipv6Hostname = hostname.charCodeAt(0) === 91/*[*/ && hostname.charCodeAt(hostname.length - 1) === 93/*]*/; // validate a little. if (!ipv6Hostname) { const result = validateHostname(this, rest, hostname); if (result !== undefined) rest = result; } if (this.hostname.length > hostnameMaxLen) { this.hostname = ''; } else { // hostnames are always lower case. this.hostname = this.hostname.toLowerCase(); } if (!ipv6Hostname) { // IDNA Support: Returns a punycoded representation of "domain". // It only converts parts of the domain name that // have non-ASCII characters, i.e. it doesn't matter if // you call it with a domain that already is ASCII-only. this.hostname = toASCII(this.hostname); } var p = this.port ? ':' + this.port : ''; var h = this.hostname || ''; this.host = h + p; // strip [ and ] from the hostname // the host field still retains them, though if (ipv6Hostname) { this.hostname = this.hostname.slice(1, -1); 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. const result = autoEscapeStr(rest); if (result !== undefined) rest = result; } var questionIdx = -1; var hashIdx = -1; for (i = 0; i < rest.length; ++i) { const code = rest.charCodeAt(i); if (code === 35/*#*/) { this.hash = rest.slice(i); hashIdx = i; break; } else if (code === 63/*?*/ && questionIdx === -1) { questionIdx = i; } } if (questionIdx !== -1) { if (hashIdx === -1) { this.search = rest.slice(questionIdx); this.query = rest.slice(questionIdx + 1); } else { this.search = rest.slice(questionIdx, hashIdx); this.query = rest.slice(questionIdx + 1, hashIdx); } if (parseQueryString) { this.query = querystring.parse(this.query); } } else if (parseQueryString) { // no query string, but parseQueryString still requested this.search = ''; this.query = new ParsedQueryString(); } var firstIdx = (questionIdx !== -1 && (hashIdx === -1 || questionIdx < hashIdx) ? questionIdx : hashIdx); if (firstIdx === -1) { if (rest.length > 0) this.pathname = rest; } else if (firstIdx > 0) { this.pathname = rest.slice(0, firstIdx); } if (slashedProtocol[lowerProto] && this.hostname && !this.pathname) { this.pathname = '/'; } // to support http.request if (this.pathname || this.search) { const p = this.pathname || ''; const s = this.search || ''; this.path = p + s; } // finally, reconstruct the href based on what has been validated. this.href = this.format(); return this; }; function validateHostname(self, rest, hostname) { for (var i = 0; i < hostname.length; ++i) { const code = hostname.charCodeAt(i); const isValid = (code >= 97/*a*/ && code <= 122/*z*/) || code === 46/*.*/ || (code >= 65/*A*/ && code <= 90/*Z*/) || (code >= 48/*0*/ && code <= 57/*9*/) || code === 45/*-*/ || code === 43/*+*/ || code === 95/*_*/ || code > 127; // Invalid host character if (!isValid) { self.hostname = hostname.slice(0, i); return '/' + hostname.slice(i) + rest; } } } // Automatically escape all delimiters and unwise characters from RFC 2396. // Also escape single quotes in case of an XSS attack. // Return undefined if the string doesn't need escaping, // otherwise return the escaped string. function autoEscapeStr(rest) { var escaped = ''; var lastEscapedPos = 0; for (var i = 0; i < rest.length; ++i) { // Manual switching is faster than using a Map/Object. // `escaped` contains substring up to the last escaped cahracter. switch (rest.charCodeAt(i)) { case 9: // '\t' // Concat if there are ordinary characters in the middle. if (i > lastEscapedPos) escaped += rest.slice(lastEscapedPos, i); escaped += '%09'; lastEscapedPos = i + 1; break; case 10: // '\n' if (i > lastEscapedPos) escaped += rest.slice(lastEscapedPos, i); escaped += '%0A'; lastEscapedPos = i + 1; break; case 13: // '\r' if (i > lastEscapedPos) escaped += rest.slice(lastEscapedPos, i); escaped += '%0D'; lastEscapedPos = i + 1; break; case 32: // ' ' if (i > lastEscapedPos) escaped += rest.slice(lastEscapedPos, i); escaped += '%20'; lastEscapedPos = i + 1; break; case 34: // '"' if (i > lastEscapedPos) escaped += rest.slice(lastEscapedPos, i); escaped += '%22'; lastEscapedPos = i + 1; break; case 39: // '\'' if (i > lastEscapedPos) escaped += rest.slice(lastEscapedPos, i); escaped += '%27'; lastEscapedPos = i + 1; break; case 60: // '<' if (i > lastEscapedPos) escaped += rest.slice(lastEscapedPos, i); escaped += '%3C'; lastEscapedPos = i + 1; break; case 62: // '>' if (i > lastEscapedPos) escaped += rest.slice(lastEscapedPos, i); escaped += '%3E'; lastEscapedPos = i + 1; break; case 92: // '\\' if (i > lastEscapedPos) escaped += rest.slice(lastEscapedPos, i); escaped += '%5C'; lastEscapedPos = i + 1; break; case 94: // '^' if (i > lastEscapedPos) escaped += rest.slice(lastEscapedPos, i); escaped += '%5E'; lastEscapedPos = i + 1; break; case 96: // '`' if (i > lastEscapedPos) escaped += rest.slice(lastEscapedPos, i); escaped += '%60'; lastEscapedPos = i + 1; break; case 123: // '{' if (i > lastEscapedPos) escaped += rest.slice(lastEscapedPos, i); escaped += '%7B'; lastEscapedPos = i + 1; break; case 124: // '|' if (i > lastEscapedPos) escaped += rest.slice(lastEscapedPos, i); escaped += '%7C'; lastEscapedPos = i + 1; break; case 125: // '}' if (i > lastEscapedPos) escaped += rest.slice(lastEscapedPos, i); escaped += '%7D'; lastEscapedPos = i + 1; break; } } if (lastEscapedPos === 0) // Nothing has been escaped. return; // There are ordinary characters at the end. if (lastEscapedPos < rest.length) return escaped + rest.slice(lastEscapedPos); else // The last character is escaped. return escaped; } // 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); else if (typeof obj !== 'object' || obj === null) throw new TypeError('Parameter "urlObj" must be an object, not ' + obj === null ? 'null' : typeof obj); else if (!(obj instanceof Url)) return Url.prototype.format.call(obj); return obj.format(); } Url.prototype.format = function() { var auth = this.auth || ''; if (auth) { auth = encodeAuth(auth); auth += '@'; } var protocol = this.protocol || ''; var pathname = this.pathname || ''; var hash = this.hash || ''; var host = ''; var 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 !== null && typeof this.query === 'object') query = querystring.stringify(this.query); var search = this.search || (query && ('?' + query)) || ''; if (protocol && protocol.charCodeAt(protocol.length - 1) !== 58/*:*/) protocol += ':'; var newPathname = ''; var lastPos = 0; for (var i = 0; i < pathname.length; ++i) { switch (pathname.charCodeAt(i)) { case 35: // '#' if (i - lastPos > 0) newPathname += pathname.slice(lastPos, i); newPathname += '%23'; lastPos = i + 1; break; case 63: // '?' if (i - lastPos > 0) newPathname += pathname.slice(lastPos, i); newPathname += '%3F'; lastPos = i + 1; break; } } if (lastPos > 0) { if (lastPos !== pathname.length) pathname = newPathname + pathname.slice(lastPos); else pathname = newPathname; } // only the slashedProtocols get the //. Not mailto:, xmpp:, etc. // unless they had them to begin with. if (this.slashes || slashedProtocol[protocol]) { if (this.slashes || host) { if (pathname && pathname.charCodeAt(0) !== 47/*/*/) pathname = '/' + pathname; host = '//' + host; } else if (protocol.length >= 4 && protocol.charCodeAt(0) === 102/*f*/ && protocol.charCodeAt(1) === 105/*i*/ && protocol.charCodeAt(2) === 108/*l*/ && protocol.charCodeAt(3) === 101/*e*/) { host = '//'; } } search = search.replace(/#/g, '%23'); if (hash && hash.charCodeAt(0) !== 35/*#*/) hash = '#' + hash; if (search && search.charCodeAt(0) !== 63/*?*/) search = '?' + search; 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(); var tkeys = Object.keys(this); for (var tk = 0; tk < tkeys.length; tk++) { var tkey = tkeys[tk]; result[tkey] = this[tkey]; } // 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 var rkeys = Object.keys(relative); for (var rk = 0; rk < rkeys.length; rk++) { var rkey = rkeys[rk]; if (rkey !== 'protocol') result[rkey] = relative[rkey]; } //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]) { var keys = Object.keys(relative); for (var v = 0; v < keys.length; v++) { var k = keys[v]; result[k] = relative[k]; } result.href = result.format(); return result; } result.protocol = relative.protocol; if (!relative.host && !/^file:?$/.test(relative.protocol) && !hostlessProtocol[relative.protocol]) { const 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) === '/'); var isRelAbs = ( relative.host || relative.pathname && relative.pathname.charAt(0) === '/' ); var mustEndAbs = (isRelAbs || isSourceAbs || (result.host && relative.pathname)); var removeAllDots = mustEndAbs; var srcPath = result.pathname && result.pathname.split('/') || []; var relPath = relative.pathname && relative.pathname.split('/') || []; var 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; result.auth = 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. if (relative.host || relative.host === '') { if (result.host !== relative.host) result.auth = null; result.host = relative.host; result.port = relative.port; } if (relative.hostname || relative.hostname === '') { if (result.hostname !== relative.hostname) result.auth = null; result.hostname = relative.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(); //occasionally the auth can get stuck only in host //this especially happens in cases like //url.resolveObject('mailto:local1@domain1', 'local2@domain2') const 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 || srcPath.length > 1) && (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 - 1; i >= 0; i--) { last = srcPath[i]; if (last === '.') { spliceOne(srcPath, i); } else if (last === '..') { spliceOne(srcPath, i); up++; } else if (up) { spliceOne(srcPath, i); 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() : ''; //occasionally the auth can get stuck only in host //this especially happens in cases like //url.resolveObject('mailto:local1@domain1', 'local2@domain2') const 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.slice(1); } host = host.slice(0, host.length - port.length); } if (host) this.hostname = host; }; // About 1.5x faster than the two-arg version of Array#splice(). function spliceOne(list, index) { for (var i = index, k = i + 1, n = list.length; k < n; i += 1, k += 1) list[i] = list[k]; list.pop(); }