Add EGUSD board meeting scraper

Refs #1

index.js

@@ -1,6 +1,7 @@
 const express = require("express")
 const ics = require("ics")
 
+const scrapeEGUSD = require("./scrapers/egusd-board.js")
 const scrapeSacBoardOfSupervisors = require("./scrapers/sac-board-of-supervisors.js")
 const scrapeSacCityCouncil = require("./scrapers/sac-city-council.js")
 const scrapeSCUSD = require("./scrapers/scusd-board.js")
@@ -12,27 +13,38 @@ app.get("/", (req, res) => {
   res.send("OK")
 })
 
-app.get("/calendar.ics", (req, res) => {
+app.get("/calendar.:format", (req, res) => {
+  const { format } = req.params
   Promise.all([
+    scrapeSacBoardOfSupervisors(),
     scrapeSacCityCouncil(),
     scrapeSCUSD(),
-    scrapeSacBoardOfSupervisors(),
+    scrapeEGUSD(),
   ]).then((data) => {
-    const [cityCouncilMeetings, scusdMeetings, supervisorMeetings] = data
+    const [supervisorMeetings, cityCouncilMeetings, scusdMeetings, egusdMeetings] = data
+    // const [egusdMeetings] = data
     const meetings = [].concat(
+      supervisorMeetings,
       cityCouncilMeetings,
       scusdMeetings,
-      supervisorMeetings
+      egusdMeetings,
     )
     const { error, value } = ics.createEvents(meetings)
 
     if (error) {
       res.json({ error })
       return
     }
-    res.json(meetings)
-    return
-    res.send(value)
+
+    if (format === 'json') {
+      return res.json(meetings)
+    }
+
+    if (format === 'ics') {
+      return res.send(value)
+    }
+
+    return res.status(404)
   })
 })
 

scrapers/egusd-board.js

@@ -0,0 +1,50 @@
+const cheerio = require("cheerio")
+const { timeParse } = require("d3-time-format")
+const fetch = require("isomorphic-fetch")
+
+const url = 'http://www.egusd.net/about/leadership/board-of-education/board-meeting-schedule/'
+const parseDate = timeParse('%B %e, %Y')
+
+async function scrapeEGUSDBoard() {
+    const req = await fetch(url)
+    const text = await req.text()
+    const $ = cheerio.load(text)
+    const calendarIcon = $('.wp-svg-calendar.calendar')
+    const h3 = calendarIcon.closest('h3')
+    const ul = h3.next('ul')
+    const lis = ul.find('li')
+    const meetings = []
+
+    lis.each(function(i, li) {
+        const text = $(li).text()
+        const [date, ...match] = text.match(/^(\w+\s\d+,\s\d{4})/)
+        const isRescheduled = text.includes('Rescheduled')
+        const isWorkshop = text.includes('Board Workshop')
+
+        if (isRescheduled) return
+
+        const parsedDate = parseDate(date)
+        const title = isWorkshop ? 'EGUSD Board Workshop' : 'EGUSD Board Meeting'
+        const start = [
+            parsedDate.getFullYear(),
+            parsedDate.getMonth() + 1,
+            parsedDate.getDate(),
+            isWorkshop ? 8 : 18,
+            isWorkshop ? 30 : 0,
+        ]
+        const duration = {
+            minutes: isWorkshop ? 390 : 240
+        }
+
+        meetings.push({
+            title,
+            description: '',
+            start,
+            duration,
+        })
+    })
+
+    return meetings
+}
+
+module.exports = scrapeEGUSDBoard

scrapers/scusd-board.js

@@ -40,9 +40,8 @@ async function scrapeSCUSDBoard() {
       parsedDate.getMinutes(),
     ]
 
-    // debugger
     data.push({
-      title,
+      title: 'SCUSD Board Meeting',
       start,
       description: "",
       duration: { minutes: 90 },