Feature/scrape

index.js

@@ -8,39 +8,19 @@ const cache = require('memory-cache')
 const microCors = require('micro-cors')
 const { send } = require('micro')
 
-const getEvents = require('./lib/get-events')
-const getMeetups = require('./lib/get-meetups')
+const scrapeEvents = require('./lib/scrape-events')
 
-const blackList = process.env.BLACK_LIST ? process.env.BLACK_LIST.split(',') : []
 const cacheExpiration = parseInt(process.env.CACHE_EXPIRATION)
 const cors = microCors({
     allowMethods: ['GET']
 })
-const whiteList = process.env.WHITE_LIST ? process.env.WHITE_LIST.split(',') : []
 
 async function handler(req, res) {
     try {
         // si el resultado del API no fue previamente cacheado
         if (!cache.get('data')) {
             // obtenemos un array de meetups que corresponden al rango es búsqueda
-            const data = await getMeetups()
-                // filtramos los eventos que no queremos que aparezcan
-                .then(eventsList =>
-                    eventsList.filter(event => !blackList.includes(event.id.toString()))
-                )
-                // agregamos los meetups que queremos que aparezcan que no entran en el rango de búsqueda
-                .then(meetups => meetups.concat(whiteList.map(m => ({ urlname: m }))))
-                // buscamos los eventos de esos meetups
-                .then(meetups => meetups.map(m => getEvents(m.urlname)))
-                // cuando obtengamos toda la información, vamos a tener un array de arrays,
-                // donde cada uno es una lista de eventos
-                .then(eventsProms => Promise.all(eventsProms))
-                // generamos un array de 1 solo nivel por medio de un reduce
-                // que solo concatena todos los eventos de los diferentes meetups
-                .then(eventsLists =>
-                    eventsLists.reduce((output, events) => output.concat(events), [])
-                )
-
+            const data = await scrapeEvents()
             // guardamos los datos en cache por el tiempo indicado por configuración
             cache.put('data', data, cacheExpiration)
         }

lib/scrape-events.js

@@ -0,0 +1,46 @@
+const cheerio = require('cheerio')
+const got = require('got')
+const moment = require('moment')
+
+module.exports = async function scrapeEvents() {
+    const { RADIUS } = process.env
+    const { body } = await got(
+        `https://www.meetup.com/es-ES/find/events/tech/?allMeetups=false&radius=${RADIUS}&userFreeform=buenos+ai&mcId=c1000296&mcName=Buenos+Aires%2C+AR`
+    )
+    const $ = cheerio.load(body)
+    const days = $('li.event-listing-container-li')
+    const dayInfo = $('li.date-indicator')
+    const eventsArray = []
+    days.each((dayIndex, li) => {
+        const events = $(li).find('ul.event-listing-container > li')
+        events.each((_, li) => {
+            $(li).each((_, e) => {
+                const eventDateElement = dayInfo.get(dayIndex)
+                let eventDate
+                if (eventDateElement) {
+                    const {
+                        'data-year': year,
+                        'data-month': month,
+                        'data-day': day
+                    } = dayInfo.get(dayIndex).attribs
+                    eventDate = `${year}-${month}-${day}`
+                } else {
+                    eventDate = moment().format('YYYY-MM-DD')
+                }                
+                eventsArray.push({
+                    date:
+                        moment.utc(
+                            `${eventDate} ${$(e).find('div a').first().text().trim()}`,
+                            'YYYY-MM-DD HH:mm'
+                        ),
+                    eventName:
+                        `${$(e).find('div[itemprop="location"]').text().trim()} - ${$(e).find('a.event span[itemprop="name"]').text().trim()}`,
+                    eventLink: $(e).find('a.event').prop('href'),
+                    place: '',
+                    attendeeCount: $(e).find('div.attendee-count').text().replace(/\n+/g, ' ')        
+                })
+            })
+        })
+    })
+    return eventsArray
+}

package.json

@@ -18,6 +18,7 @@
         "micro-dev": "^3.0.0"
     },
     "dependencies": {
+        "cheerio": "^1.0.0-rc.3",
         "dotenv": "^8.0.0",
         "got": "^9.6.0",
         "memory-cache": "^0.2.0",