139 lines
3.6 KiB
JavaScript
139 lines
3.6 KiB
JavaScript
var http = require('http');
|
|
var cheerio = require('cheerio');
|
|
var iconv = require('iconv-lite');
|
|
var mongodb = require('mongodb').MongoClient;
|
|
|
|
var scheduletypes = [
|
|
'Klasrooster',
|
|
'Docentrooster',
|
|
'Leerlingrooster',
|
|
'Lokaalrooster'
|
|
];
|
|
var school_id;
|
|
var database;
|
|
|
|
//Function for getting pages with http requests.
|
|
function get() {
|
|
database.collection('index').drop();
|
|
|
|
for (scheduletype of scheduletypes) {
|
|
|
|
(function (scheduletype) {
|
|
|
|
var link = 'http://roosters5.gepro-osi.nl/roosters/rooster.php?school=' + school_id + '&type=' + scheduletype;
|
|
|
|
http.get(link, function (res) {
|
|
|
|
var _download = {};
|
|
_download.type = scheduletype;
|
|
|
|
res.on('data', function (data) {
|
|
_download.data += data;
|
|
});
|
|
|
|
res.on('end', function () {
|
|
rip(_download);
|
|
});
|
|
|
|
});
|
|
})(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);
|
|
}
|
|
|
|
//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) {
|
|
|
|
http.get('http://roosters5.gepro-osi.nl/roosters/rooster.php?school=' + school_id + '&type=' + data.type + '&afdeling=' + studentcategory, function (res) {
|
|
var _download = '';
|
|
|
|
res.on('data', function (data) {
|
|
_download += iconv.decode(data, 'binary');
|
|
});
|
|
|
|
res.on('end', function () {
|
|
var list_students = cheerio('select', _download).children();
|
|
|
|
for (student in list_students) {
|
|
|
|
if (!isNaN(student)) {
|
|
var name = cheerio(list_students[student]).text().split(' - ')[1];
|
|
var id = cheerio(list_students[student]).val();
|
|
|
|
var database_entry = {
|
|
'id' : id,
|
|
'username' : id + name.split(' ')[0].toLowerCase(),
|
|
'name' : name,
|
|
'first_name' : name.split(' ')[0],
|
|
'last_name' : name.split(' ').splice(1).join(' '),
|
|
'studentcategory' : studentcategory,
|
|
'type' : data.type.replace(/rooster/g, '').toLowerCase()
|
|
}
|
|
|
|
collection.insert(database_entry, show_output);
|
|
|
|
if (studentcategory == list[list.length - 1] && student == list_students.length - 1) {
|
|
database.close();
|
|
}
|
|
|
|
}
|
|
}
|
|
});
|
|
});
|
|
})(studentcategory);
|
|
}
|
|
}
|
|
else {
|
|
for (entry of list) {
|
|
var database_entry = {
|
|
'name' : entry,
|
|
'type' : data.type.replace(/rooster/g, '').toLowerCase()
|
|
}
|
|
|
|
collection.insert(database_entry, show_output);
|
|
}
|
|
}
|
|
}
|
|
|
|
//Function being called to access functionality from this module.
|
|
function crawl(sid) {
|
|
school_id = sid;
|
|
mongodb.connect('mongodb://wallpiece/roosterio', function (error, db) {
|
|
if (error) console.warn(error);
|
|
database = db;
|
|
|
|
get();
|
|
});
|
|
}
|
|
|
|
|
|
//Redundant function for draining native-mongodb-driver output
|
|
function show_output(error, message) {
|
|
if (process.argv[3] == '-v') {
|
|
// if (error) process.stdout.write(error.toString());
|
|
if (message != null) console.log(message);
|
|
}
|
|
}
|
|
|
|
module.exports = {
|
|
'crawl' : crawl
|
|
}
|
|
|
|
if (process.argv[2] == 'test') {
|
|
module.exports.crawl(934);
|
|
}
|