162 lines
3.8 KiB
JavaScript
162 lines
3.8 KiB
JavaScript
'use strict';
|
|
var url = require('url');
|
|
var punycode = require('punycode');
|
|
var queryString = require('query-string');
|
|
var prependHttp = require('prepend-http');
|
|
var sortKeys = require('sort-keys');
|
|
var objectAssign = require('object-assign');
|
|
|
|
var DEFAULT_PORTS = {
|
|
'http:': 80,
|
|
'https:': 443,
|
|
'ftp:': 21
|
|
};
|
|
|
|
// protocols that always contain a `//`` bit
|
|
var slashedProtocol = {
|
|
'http': true,
|
|
'https': true,
|
|
'ftp': true,
|
|
'gopher': true,
|
|
'file': true,
|
|
'http:': true,
|
|
'https:': true,
|
|
'ftp:': true,
|
|
'gopher:': true,
|
|
'file:': true
|
|
};
|
|
|
|
function testParameter(name, filters) {
|
|
return filters.some(function (filter) {
|
|
return filter instanceof RegExp ? filter.test(name) : filter === name;
|
|
});
|
|
}
|
|
|
|
module.exports = function (str, opts) {
|
|
opts = objectAssign({
|
|
normalizeProtocol: true,
|
|
normalizeHttps: false,
|
|
stripFragment: true,
|
|
stripWWW: true,
|
|
removeQueryParameters: [/^utm_\w+/i],
|
|
removeTrailingSlash: true,
|
|
removeDirectoryIndex: false
|
|
}, opts);
|
|
|
|
if (typeof str !== 'string') {
|
|
throw new TypeError('Expected a string');
|
|
}
|
|
|
|
var hasRelativeProtocol = str.indexOf('//') === 0;
|
|
|
|
// prepend protocol
|
|
str = prependHttp(str.trim()).replace(/^\/\//, 'http://');
|
|
|
|
var urlObj = url.parse(str);
|
|
|
|
if (opts.normalizeHttps && urlObj.protocol === 'https:') {
|
|
urlObj.protocol = 'http:';
|
|
}
|
|
|
|
if (!urlObj.hostname && !urlObj.pathname) {
|
|
throw new Error('Invalid URL');
|
|
}
|
|
|
|
// prevent these from being used by `url.format`
|
|
delete urlObj.host;
|
|
delete urlObj.query;
|
|
|
|
// remove fragment
|
|
if (opts.stripFragment) {
|
|
delete urlObj.hash;
|
|
}
|
|
|
|
// remove default port
|
|
var port = DEFAULT_PORTS[urlObj.protocol];
|
|
if (Number(urlObj.port) === port) {
|
|
delete urlObj.port;
|
|
}
|
|
|
|
// remove duplicate slashes
|
|
if (urlObj.pathname) {
|
|
urlObj.pathname = urlObj.pathname.replace(/\/{2,}/g, '/');
|
|
}
|
|
|
|
// decode URI octets
|
|
if (urlObj.pathname) {
|
|
urlObj.pathname = decodeURI(urlObj.pathname);
|
|
}
|
|
|
|
// remove directory index
|
|
if (opts.removeDirectoryIndex === true) {
|
|
opts.removeDirectoryIndex = [/^index\.[a-z]+$/];
|
|
}
|
|
|
|
if (Array.isArray(opts.removeDirectoryIndex) && opts.removeDirectoryIndex.length) {
|
|
var pathComponents = urlObj.pathname.split('/');
|
|
var lastComponent = pathComponents[pathComponents.length - 1];
|
|
|
|
if (testParameter(lastComponent, opts.removeDirectoryIndex)) {
|
|
pathComponents = pathComponents.slice(0, pathComponents.length - 1);
|
|
urlObj.pathname = pathComponents.slice(1).join('/') + '/';
|
|
}
|
|
}
|
|
|
|
// resolve relative paths, but only for slashed protocols
|
|
if (slashedProtocol[urlObj.protocol]) {
|
|
var domain = urlObj.protocol + '//' + urlObj.hostname;
|
|
var relative = url.resolve(domain, urlObj.pathname);
|
|
urlObj.pathname = relative.replace(domain, '');
|
|
}
|
|
|
|
if (urlObj.hostname) {
|
|
// IDN to Unicode
|
|
urlObj.hostname = punycode.toUnicode(urlObj.hostname).toLowerCase();
|
|
|
|
// remove trailing dot
|
|
urlObj.hostname = urlObj.hostname.replace(/\.$/, '');
|
|
|
|
// remove `www.`
|
|
if (opts.stripWWW) {
|
|
urlObj.hostname = urlObj.hostname.replace(/^www\./, '');
|
|
}
|
|
}
|
|
|
|
// remove URL with empty query string
|
|
if (urlObj.search === '?') {
|
|
delete urlObj.search;
|
|
}
|
|
|
|
var queryParameters = queryString.parse(urlObj.search);
|
|
|
|
// remove query unwanted parameters
|
|
if (Array.isArray(opts.removeQueryParameters)) {
|
|
for (var key in queryParameters) {
|
|
if (testParameter(key, opts.removeQueryParameters)) {
|
|
delete queryParameters[key];
|
|
}
|
|
}
|
|
}
|
|
|
|
// sort query parameters
|
|
urlObj.search = queryString.stringify(sortKeys(queryParameters));
|
|
|
|
// decode query parameters
|
|
urlObj.search = decodeURIComponent(urlObj.search);
|
|
|
|
// take advantage of many of the Node `url` normalizations
|
|
str = url.format(urlObj);
|
|
|
|
// remove ending `/`
|
|
if (opts.removeTrailingSlash || urlObj.pathname === '/') {
|
|
str = str.replace(/\/$/, '');
|
|
}
|
|
|
|
// restore relative protocol, if applicable
|
|
if (hasRelativeProtocol && !opts.normalizeProtocol) {
|
|
str = str.replace(/^http:\/\//, '//');
|
|
}
|
|
|
|
return str;
|
|
};
|