UNPKG

2 kBJavaScriptView Raw
1var assert = require("assert");
2var crawler = require("../index.js");
3var cs = require("../plugins/console-plugin.js");
4var testSite = require("./website-2/start.js").site;
5var proxyLoader = require("simple-proxies/lib/proxyfileloader");
6var log = require("crawler-ninja-logger").Logger;
7
8//var heapdump = require('heapdump');
9
10log.level("debug");
11
12var proxyListTest = null;
13
14describe('Memory leaks', function() {
15
16 before(function(done) {
17 var proxyFile = "./test/proxies-all.txt";
18
19 // Load proxies
20 var config = proxyLoader.config()
21 .setProxyFile(proxyFile)
22 .setCheckProxies(false)
23 .setRemoveInvalidProxies(false);
24
25 proxyLoader.loadProxyFile(config, function(error, proxyList) {
26 if (error) {
27 console.log(error);
28
29 }
30 else {
31 proxyListTest = proxyList;
32 done();
33 }
34
35 });
36 });
37 it.skip('should crawl a big site without memory leaks', function(done) {
38 this.timeout(3000000);
39 /*
40 setInterval(function(){
41 console.log(">>>>> Dump !");
42 heapdump.writeSnapshot('./dump/dump' + Date.now() + '.heapsnapshot');
43 }, 120000);
44 */
45 var options = {
46 skipDuplicates: true,
47 externalDomains: false,
48 scripts : false,
49 links : false,
50 images : false,
51 maxConnections : 10
52 }
53 var consolePlugin = new cs.Plugin();
54 //crawler.init(options, done, proxyListTest);
55 crawler.init(options, done, null);
56 crawler.registerPlugin(consolePlugin);
57
58 //crawler.queue({url : "http://localhost:9991/index.html" });
59 crawler.queue({url : "http://www.rtbf.be"});
60
61 });
62
63});