BeckOK/crawler.js

70 lines
1.7 KiB
JavaScript

var request = require('request');
var cheerio = require('cheerio');
var URL = require('url-parse');
var plsStop = 0;
var pagesVisited = {};
// set a default starting page (which is the title of Grundgesetz for arbitrary reasons) but take input from CLI
var argv = process.argv.slice(2);
var START_URL = "https://beck-online.beck.de/?vpath=bibdata/komm/BeckOKGG_44/cont/BECKOKGG.glUD1.htm";
if (argv[0]){
START_URL = argv[0];
}
console.log(START_URL);
var url = new URL(START_URL);
var baseUrl = url.protocol + "//" + url.hostname;
pageToVisit = START_URL;
crawl();
// function declarations:
function crawl() {
console.log(plsStop);
if (plsStop === 1) {
crawl();
}
if (pageToVisit in pagesVisited) {
// We've already visited this page, so repeat the crawl
crawl();
} else {
// New page we haven't visited
visitPage(pageToVisit, crawl);
}
}
function visitPage(url, callback){
console.log("Visiting page " + url);
pagesVisited[url] = true;
request(url, function(error, response, body) {
if(error) {
console.log("Error: " + error);
}
// Check status code (200 is HTTP OK)
console.log("Status code: " + response.statusCode);
if(response.statusCode === 200) {
// Parse the document body
var $ = cheerio.load(body);
console.log("Page title: " + $('title').text());
// savePage($,crawl);
// prepare next page
var nextPage = $('#next').attr('href');
console.log("Next up: " + nextPage);
if (nextPage){
pageToVisit = baseUrl + nextPage;
callback();
} else {
plsStop = 1;
}
}
});
}
function savePage($,callback){
console.log($);
}