diff options
author | Noah Loomans <noahloomans@gmail.com> | 2018-02-17 18:11:46 +0100 |
---|---|---|
committer | Noah Loomans <noahloomans@gmail.com> | 2018-02-17 18:11:46 +0100 |
commit | 2e4b9dc1cc48e5d35243bcef32b8acd29171bf23 (patch) | |
tree | 619cb2bfd1f41430d7b834a54738a8ca512fa36a /src/server/lib/schools/hetmml | |
parent | 4da5def0e14dbbbbdd7396921799b4753f073c26 (diff) |
Move school spesific files to school spesifc folders
Diffstat (limited to 'src/server/lib/schools/hetmml')
-rw-r--r-- | src/server/lib/schools/hetmml/getScheduleData.js | 127 | ||||
-rw-r--r-- | src/server/lib/schools/hetmml/getURLOfUser.js | 8 |
2 files changed, 135 insertions, 0 deletions
diff --git a/src/server/lib/schools/hetmml/getScheduleData.js b/src/server/lib/schools/hetmml/getScheduleData.js new file mode 100644 index 0000000..303af6d --- /dev/null +++ b/src/server/lib/schools/hetmml/getScheduleData.js @@ -0,0 +1,127 @@ +const Promise = require('bluebird'); +const cheerio = require('cheerio'); +const iconv = require('iconv-lite'); +const debounce = require('promise-debounce'); +const _ = require('lodash'); +const request = Promise.promisify(require('request')); + +const getUrlOfUser = require('./getURLOfUser'); + +let meetingpointData; +let lastUpdate; + +function scrapeUsers(html) { + const page = cheerio.load(html); + const script = page('script').eq(1).text(); + + const regexs = [/var classes = \[(.+)\];/, /var teachers = \[(.+)\];/, /var rooms = \[(.+)\];/, /var students = \[(.+)\];/]; + const items = regexs.map(regex => script.match(regex)[1].split(',').map(item => item.replace(/"/g, ''))); + + const classes = items[0].map((item, index) => ({ + type: 'c', + value: item, + index, + })); + + const teachers = items[1].map((item, index) => ({ + type: 't', + value: item, + index, + })); + + const rooms = items[2].map((item, index) => ({ + type: 'r', + value: item, + index, + })); + + const students = items[3].map((item, index) => ({ + type: 's', + value: item, + index, + })); + + return _.flatten([classes, teachers, rooms, students]); +} + +function scrapeWeeks(html) { + const page = cheerio.load(html); + const weekSelector = page('select[name="week"]'); + const weeks = _.map(weekSelector.children(), option => ({ + id: cheerio(option).attr('value'), + text: cheerio(option).text(), + })); + + return weeks; +} + +function scrapeAltText(html) { + const page = cheerio.load(html); + return page('center > font').eq(2).text().trim(); +} + +function combineUsers(usersArrays) { + return _.uniqBy(_.flatten(usersArrays), user => `${user.type}/${user.value}`); +} + +function getAlts(users) { + const requests = users.map(user => + request(getUrlOfUser('dag', user.type, user.index, 7), { timeout: 8000, encoding: null })); + + return Promise.all(requests).then(teacherResponses => + teacherResponses.map((teacherResponse, index) => { + const teacherResponseBody = iconv.decode(teacherResponse.body, 'iso-8859-1'); + + const teacherName = scrapeAltText(teacherResponseBody); + + return { + ...users[index], + alt: teacherName, + }; + })); +} + +function getScheduleData() { + const navbarRequests = [ + request('http://www.meetingpointmco.nl/Roosters-AL/doc/dagroosters/frames/navbar.htm', { timeout: 5000 }), + request('http://www.meetingpointmco.nl/Roosters-AL/doc/basisroosters/frames/navbar.htm', { timeout: 5000 }), + ]; + + return Promise.all(navbarRequests) + .then(([dailyScheduleResponse, basisScheduleResponse]) => { + const users = scrapeUsers(dailyScheduleResponse.body); + const dailyScheduleWeeks = scrapeWeeks(dailyScheduleResponse.body); + const basisScheduleWeeks = scrapeWeeks(basisScheduleResponse.body); + + const teachers = users.filter(user => user.type === 't'); + + return getAlts(teachers) + .then(teachersWithAlts => ({ + users: combineUsers([teachersWithAlts, users]), + dailyScheduleWeeks, + basisScheduleWeeks, + })) + .catch(() => ({ + // Just return the user data without the alts if getAlts fails, since + // the alts are non-essential. + users, + dailyScheduleWeeks, + basisScheduleWeeks, + })); + }); +} + +function getScheduleDataCacheWrapper() { + if (meetingpointData == null || new Date() - lastUpdate > 30 * 60 * 1000) { // 30 minutes + return getScheduleData().then((meetingpointData_) => { + lastUpdate = new Date(); + meetingpointData = meetingpointData_; + + return meetingpointData; + }); + } + + return Promise.resolve(meetingpointData); +} + +module.exports = debounce(getScheduleDataCacheWrapper); diff --git a/src/server/lib/schools/hetmml/getURLOfUser.js b/src/server/lib/schools/hetmml/getURLOfUser.js new file mode 100644 index 0000000..f44d1a3 --- /dev/null +++ b/src/server/lib/schools/hetmml/getURLOfUser.js @@ -0,0 +1,8 @@ +const leftPad = require('left-pad'); // I imported this just to piss you off ;) + +function getURLOfUser(scheduleType, type, index, week) { + return `http://www.meetingpointmco.nl/Roosters-AL/doc/${scheduleType}roosters/` + + `${leftPad(week, 2, '0')}/${type}/${type}${leftPad(index + 1, 5, '0')}.htm`; +} + +module.exports = getURLOfUser; |