fix(crawler/fl): adapt list format introduced on 21/01/29

packages/crawler/lib/fl.js

@@ -5,13 +5,26 @@ const JSDOM = require('jsdom').JSDOM
 const fetchOptions = require('./fetch-options.json')
 
 const rDate = /^([0-9]+)\.([0-9]+)\.([0-9]+)$/
-const rDeaths = /([0-9]+)[\D]+Verst(?:or|ro)?ben/i
-const rInfected = /([0-9]+)[\D]+Infizierte/
-const rQuarantined = /([0-9]+)[\D]+ (?<!Lehrer )in Qua?rantäne/
-const rRecovered = /([0-9]+)[\D]+gen?en?sen/
+const rDeaths = [/verstorben: ([0-9]+)/i, /([0-9]+)[\D]+Verst(?:or|ro)?ben/i]
+const rInfected = [/Positive gesamt: ([0-9]+)/i, /([0-9]+)[\D]+Infizierte/]
+const rQuarantined = [
+  /Quarantänefälle: ([0-9]+)/i,
+  /([0-9]+)[\D]+ (?<!Lehrer )in Qua?rantäne/
+]
+const rRecovered = [/genesen: ([0-9]+)/i, /([0-9]+)[\D]+gen?en?sen/]
+
+const matcher = (str, r) => {
+  if (Array.isArray(r)) {
+    let match
+    r.every(re => !(match = str.match(re)))
+    return match
+  }
+
+  return str.match(r)
+}
 
 const getMatch = (el, regex, isOptional) => {
-  let match = el.textContent.match(regex)
+  let match = matcher(el.textContent, regex)
 
   if (match) {
     return match
@@ -20,7 +33,7 @@ const getMatch = (el, regex, isOptional) => {
   let nextEl = el.previousElementSibling || el.parentElement
 
   while (nextEl && !match) {
-    match = nextEl.textContent.match(regex)
+    match = matcher(nextEl.textContent, regex)
     nextEl = nextEl.previousElementSibling || nextEl.parentElement
   }
 
@@ -62,17 +75,17 @@ const getRecord = el => {
 }
 
 const reducer = (acc, el) => {
-  if (el.childElementCount > 0) {
+  if (el.nodeName !== 'P' && el.childElementCount > 0) {
     return [...el.childNodes].reduce(reducer, acc)
   }
 
   const str = el.textContent
 
   if (
-    rInfected.test(str) &&
-    rRecovered.test(str) &&
-    rQuarantined.test(str) &&
-    rDeaths.test(str)
+    matcher(str, rInfected) &&
+    matcher(str, rRecovered) &&
+    matcher(str, rQuarantined) &&
+    matcher(str, rDeaths)
   ) {
     return [...acc, getRecord(el)]
   }