2015-06-24 17:26:24 +02:00
|
|
|
var http = require('socks5-http-client');
|
2015-06-06 20:31:35 +02:00
|
|
|
var cheerio = require('cheerio');
|
|
|
|
var iconv = require('iconv-lite');
|
2015-06-14 23:32:57 +02:00
|
|
|
var config = require('./configuration');
|
2015-06-24 17:07:52 +02:00
|
|
|
var url = require('url');
|
2015-08-15 21:48:38 +02:00
|
|
|
var database = require('./')();
|
2015-06-06 20:31:35 +02:00
|
|
|
|
|
|
|
var scheduletypes = [
|
|
|
|
'Klasrooster',
|
|
|
|
'Docentrooster',
|
|
|
|
'Leerlingrooster',
|
|
|
|
'Lokaalrooster'
|
|
|
|
];
|
2015-08-15 21:07:22 +02:00
|
|
|
var schoolID;
|
2015-06-06 20:31:35 +02:00
|
|
|
|
|
|
|
//Function for getting pages with http requests.
|
2015-06-13 14:38:55 +02:00
|
|
|
function get() {
|
|
|
|
database.collection('index').drop();
|
2015-06-07 22:26:24 +02:00
|
|
|
|
2015-06-06 20:31:35 +02:00
|
|
|
for (scheduletype of scheduletypes) {
|
2015-06-13 14:38:55 +02:00
|
|
|
|
2015-06-06 20:31:35 +02:00
|
|
|
(function (scheduletype) {
|
2015-06-07 22:30:48 +02:00
|
|
|
|
2015-08-15 21:07:22 +02:00
|
|
|
var options = url.parse('http://roosters5.gepro-osi.nl/roosters/rooster.php?school=' + schoolID + '&type=' + scheduletype);
|
|
|
|
options.socksPort = config().torPort;
|
|
|
|
options.socksHost = config().torHost;
|
2015-06-06 20:31:35 +02:00
|
|
|
|
2015-06-24 17:07:52 +02:00
|
|
|
http.get(options, function (res) {
|
2015-06-13 14:38:55 +02:00
|
|
|
|
|
|
|
var _download = {};
|
|
|
|
_download.type = scheduletype;
|
2015-06-06 20:31:35 +02:00
|
|
|
|
|
|
|
res.on('data', function (data) {
|
2015-06-13 14:38:55 +02:00
|
|
|
_download.data += data;
|
2015-06-06 20:31:35 +02:00
|
|
|
});
|
|
|
|
|
|
|
|
res.on('end', function () {
|
2015-06-13 14:38:55 +02:00
|
|
|
rip(_download);
|
2015-06-06 20:31:35 +02:00
|
|
|
});
|
2015-06-13 14:38:55 +02:00
|
|
|
|
2015-06-06 20:31:35 +02:00
|
|
|
});
|
|
|
|
})(scheduletype);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
//Function for extracting the lists with useful information from the crawled pages.
|
|
|
|
//(e.g Student names/ids, Teacher codes, Chamber numbers)
|
|
|
|
function extract(page) {
|
|
|
|
var array = cheerio('select', page).text().split('\n');
|
|
|
|
return array.splice(1, array.length - 2);
|
|
|
|
}
|
|
|
|
|
2015-06-13 14:38:55 +02:00
|
|
|
//Function for ripping all of the information
|
|
|
|
function rip(data) {
|
|
|
|
var list = extract(data.data);
|
|
|
|
var collection = database.collection('index');
|
|
|
|
|
|
|
|
if (data.type == 'Leerlingrooster') {
|
|
|
|
|
|
|
|
for(studentcategory of list) {
|
|
|
|
|
|
|
|
(function (studentcategory) {
|
2015-08-15 21:07:22 +02:00
|
|
|
var options = url.parse('http://roosters5.gepro-osi.nl/roosters/rooster.php?school=' + schoolID + '&type=' + data.type + '&afdeling=' + studentcategory);
|
|
|
|
options.socksPort = config().torPort;
|
|
|
|
options.socksHost = config().torHost;
|
2015-06-13 14:38:55 +02:00
|
|
|
|
2015-06-24 17:07:52 +02:00
|
|
|
http.get(options, function (res) {
|
2015-06-13 14:38:55 +02:00
|
|
|
var _download = '';
|
|
|
|
|
|
|
|
res.on('data', function (data) {
|
|
|
|
_download += iconv.decode(data, 'binary');
|
|
|
|
});
|
|
|
|
|
|
|
|
res.on('end', function () {
|
2015-08-15 21:07:22 +02:00
|
|
|
var listOfStudents = cheerio('select', _download).children();
|
2015-06-13 14:38:55 +02:00
|
|
|
|
2015-08-15 21:07:22 +02:00
|
|
|
for (student in listOfStudents) {
|
2015-06-13 14:38:55 +02:00
|
|
|
|
|
|
|
if (!isNaN(student)) {
|
2015-08-15 21:07:22 +02:00
|
|
|
var name = cheerio(listOfStudents[student]).text().split(' - ')[1];
|
|
|
|
var group = cheerio(listOfStudents[student]).text().split(' - ')[0];
|
|
|
|
var id = cheerio(listOfStudents[student]).val();
|
2015-06-13 15:29:42 +02:00
|
|
|
|
2015-08-15 21:07:22 +02:00
|
|
|
var databaseEntry = {
|
2015-06-13 14:38:55 +02:00
|
|
|
'id' : id,
|
2015-07-22 11:35:21 +02:00
|
|
|
'group' : group,
|
2015-06-13 14:38:55 +02:00
|
|
|
'username' : id + name.split(' ')[0].toLowerCase(),
|
2015-06-13 15:29:42 +02:00
|
|
|
'name' : name,
|
2015-06-13 14:38:55 +02:00
|
|
|
'first_name' : name.split(' ')[0],
|
|
|
|
'last_name' : name.split(' ').splice(1).join(' '),
|
|
|
|
'studentcategory' : studentcategory,
|
2015-06-13 15:29:42 +02:00
|
|
|
'type' : data.type.replace(/rooster/g, '').toLowerCase()
|
2015-06-13 14:38:55 +02:00
|
|
|
}
|
|
|
|
|
2015-08-15 21:07:22 +02:00
|
|
|
collection.insert(databaseEntry, showOutput);
|
2015-06-13 14:38:55 +02:00
|
|
|
|
2015-08-15 21:07:22 +02:00
|
|
|
if (studentcategory == list[list.length - 1] && student == listOfStudents.length - 1) {
|
2015-06-24 16:42:09 +02:00
|
|
|
setTimeout(function () {
|
|
|
|
database.close();
|
2015-08-15 21:07:22 +02:00
|
|
|
}, config().spiderTimeout);
|
2015-06-13 14:38:55 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
}
|
|
|
|
}
|
|
|
|
});
|
|
|
|
});
|
|
|
|
})(studentcategory);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
for (entry of list) {
|
2015-08-15 21:07:22 +02:00
|
|
|
var databaseEntry = {
|
2015-06-13 14:38:55 +02:00
|
|
|
'name' : entry,
|
2015-06-13 15:29:42 +02:00
|
|
|
'type' : data.type.replace(/rooster/g, '').toLowerCase()
|
2015-06-13 14:38:55 +02:00
|
|
|
}
|
|
|
|
|
2015-08-15 21:07:22 +02:00
|
|
|
collection.insert(databaseEntry, showOutput);
|
2015-06-13 14:38:55 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2015-06-06 20:31:35 +02:00
|
|
|
//Function being called to access functionality from this module.
|
2015-06-14 23:32:57 +02:00
|
|
|
function crawl() {
|
2015-08-15 21:07:22 +02:00
|
|
|
schoolID = config().schoolID;
|
2015-08-15 21:48:38 +02:00
|
|
|
get();
|
2015-06-07 22:26:24 +02:00
|
|
|
}
|
|
|
|
|
2015-06-13 14:38:55 +02:00
|
|
|
|
2015-06-07 22:30:48 +02:00
|
|
|
//Redundant function for draining native-mongodb-driver output
|
2015-08-15 21:07:22 +02:00
|
|
|
function showOutput(error, message) {
|
2015-06-08 19:58:56 +02:00
|
|
|
if (process.argv[3] == '-v') {
|
|
|
|
// if (error) process.stdout.write(error.toString());
|
|
|
|
if (message != null) console.log(message);
|
2015-06-07 22:26:24 +02:00
|
|
|
}
|
2015-06-06 20:31:35 +02:00
|
|
|
}
|
|
|
|
|
2015-06-07 22:30:48 +02:00
|
|
|
module.exports = {
|
|
|
|
'crawl' : crawl
|
|
|
|
}
|
2015-06-08 19:58:56 +02:00
|
|
|
|
2015-08-15 21:48:38 +02:00
|
|
|
if (process.argv[2] == 'test' || process.argv[2] == 'rip') {
|
2015-06-08 19:58:56 +02:00
|
|
|
module.exports.crawl(934);
|
|
|
|
}
|