hackerxian

// This example shows how to render pages that perform AJAX calls
// upon page load.
//
// Instead of waiting a fixed amount of time before doing the render,
// we are keeping track of every resource that is loaded.
//
// Once all resources are loaded, we wait a small amount of time
// (resourceWait) in case these resources load other resources.
//
// The page is rendered after a maximum amount of time (maxRenderTime)

hackerxian

//PhantomJS http://phantomjs.org/ based web crawler Anton Ivanov [email protected] 2012
//UPDATE: This gist has been made into a Node.js module and now can be installed with "npm install js-crawler"
//the Node.js version does not use Phantom.JS, but the API available to the client is similar to the present gist

(function(host) {

    function Crawler() {
        this.visitedURLs = {};
    };
    

hackerxian

/*
 * Copyright (c) 2010 Apple Inc. All rights reserved.
 */
function characterNeedsScoreMultiplier(e) {
    if (!e || e.length === 0)
        return !1;
    var t = e.charCodeAt(0);
    return t > 11904 && t < 12031?!0 : t > 12352 && t < 12543?!0 : t > 12736 && t < 19903?!0 : t > 19968 && t < 40959?!0 : t > 44032 && t < 55215?!0 : t > 63744 && t < 64255?!0 : t > 65072 && t < 65103?!0 : t > 131072 && t < 173791?!0 : t > 194560 && t < 195103?!0 : !1
}
function domDistance(e, t, n) {

hackerxian

//http://www.cnblogs.com/gaizai/p/4233780.html
//https://www.v2ex.com/t/388152?p=2
(function () {
  'use strict';

  function string10to62(number) {
    var chars = '0123456789abcdefghigklmnopqrstuvwxyzABCDEFGHIGKLMNOPQRSTUVWXYZ'.split('');
    var radix = chars.length;
    var qutient = +number;
    var arr = [];