fix forumgroningen (website changed from NL to EN)

ckuijjer · Mar 22, 2024 · 4ca404d · 4ca404d
1 parent ba2e7a1
commit 4ca404d
Show file tree

Hide file tree

Showing 2 changed files with 44 additions and 54 deletions.
diff --git a/cloud/scrapers/forumgroningen.ts b/cloud/scrapers/forumgroningen.ts
@@ -4,7 +4,7 @@ import { DateTime } from 'luxon'
 import guessYear from './utils/guessYear'
 import { Screening } from '../types'
 import { logger as parentLogger } from '../powertools'
-import { fullMonthToNumberDutch } from './utils/monthToNumber'
+import { fullMonthToNumberEnglish } from './utils/monthToNumber'
 import splitTime from './utils/splitTime'
 
 const logger = parentLogger.createChild({
@@ -67,9 +67,9 @@ const extractFromMoviePage = async ({
         .map(({ time }) => {
           const [dayOfWeek, dayString, monthString] = date.split(/\s+/)
           const day = Number(dayString)
-          const month = fullMonthToNumberDutch(monthString)
+          const month = fullMonthToNumberEnglish(monthString)
           const [startTime, endTime] = time.split(/ tot | till /)
-          const [hour, minute] = splitTime(startTime)
+          const { hour, minute } = DateTime.fromFormat(startTime, 'h:mm a')
 
           const year = guessYear({
             day,
@@ -104,10 +104,10 @@ const extractFromMainPage = async (): Promise<Screening[]> => {
   const url = 'https://forum.nl/en/whats-on/international-movie-night'
 
   const movies = (
-    await xray(url, '.content-row-medium.text-and-image', [
+    await xray(url, '.calendar-list .ticket-row', [
       {
-        title: 'h2 | cleanTitle | trim',
-        url: 'a@href',
+        title: '.content .main | cleanTitle | trim',
+        url: '@href',
       },
     ])
   ).filter(({ url }) => url !== undefined) // remove movies without url (e.g. in the past)
@@ -125,10 +125,7 @@ const extractFromMainPage = async (): Promise<Screening[]> => {
 
 if (require.main === module) {
   // extractFromMoviePage({
-  //   // url: 'https://forum.nl/nl/agenda/gaia',
-  //   // url: 'https://forum.nl/nl/agenda/ringu',
-  //   // url: 'https://forum.nl/nl/agenda/passages',p
-  //   url: 'https://forum.nl/nl/agenda/pans-labyrinth',
+  //   url: 'https://forum.nl/en/whats-on/the-zone-of-interest',
   // })
 
   extractFromMainPage()

diff --git a/cloud/scrapers/utils/monthToNumber.ts b/cloud/scrapers/utils/monthToNumber.ts
@@ -1,5 +1,19 @@
-export const fullMonthToNumberDutch = (month: string) => {
-  const monthNumber =
+const monthToNumber = (name: string, listOfMonths: string[], month: string) => {
+  const lowerCaseMonth = month.toLowerCase()
+
+  const monthNumber = listOfMonths.indexOf(lowerCaseMonth) + 1
+  if (monthNumber === 0) {
+    throw new Error(
+      `${name}: invalid month ${lowerCaseMonth}, might you need another function?`,
+    )
+  }
+
+  return monthNumber
+}
+
+export function fullMonthToNumberDutch(month: string) {
+  return monthToNumber(
+    'fullMonthToNumberDutch',
     [
       'januari',
       'februari',
@@ -13,19 +27,14 @@ export const fullMonthToNumberDutch = (month: string) => {
       'oktober',
       'november',
       'december',
-    ].indexOf(month.toLowerCase()) + 1
-
-  if (monthNumber === 0) {
-    throw new Error(
-      `invalid month ${month}, might you need the English version of this function?`,
-    )
-  }
-
-  return monthNumber
+    ],
+    month,
+  )
 }
 
-export const shortMonthToNumberDutch = (month: string) => {
-  const monthNumber =
+export function shortMonthToNumberDutch(month: string) {
+  return monthToNumber(
+    'shortMonthToNumberDutch',
     [
       'jan',
       'feb',
@@ -39,19 +48,14 @@ export const shortMonthToNumberDutch = (month: string) => {
       'okt',
       'nov',
       'dec',
-    ].indexOf(month.toLowerCase()) + 1
-
-  if (monthNumber === 0) {
-    throw new Error(
-      `invalid month ${month}, might you need the English version of this function?`,
-    )
-  }
-
-  return monthNumber
+    ],
+    month,
+  )
 }
 
-export const fullMonthToNumberEnglish = (month: string) => {
-  const monthNumber =
+export function fullMonthToNumberEnglish(month: string) {
+  return monthToNumber(
+    'fullMonthToNumberEnglish',
     [
       'january',
       'february',
@@ -65,19 +69,14 @@ export const fullMonthToNumberEnglish = (month: string) => {
       'october',
       'november',
       'december',
-    ].indexOf(month.toLowerCase()) + 1
-
-  if (monthNumber === 0) {
-    throw new Error(
-      `invalid month ${month}, might you need the Dutch version of this function?`,
-    )
-  }
-
-  return monthNumber
+    ],
+    month,
+  )
 }
 
-export const shortMonthToNumberEnglish = (month: string) => {
-  const monthNumber =
+export function shortMonthToNumberEnglish(month: string) {
+  return monthToNumber(
+    'shortMonthToNumberEnglish',
     [
       'jan',
       'feb',
@@ -91,13 +90,7 @@ export const shortMonthToNumberEnglish = (month: string) => {
       'oct',
       'nov',
       'dec',
-    ].indexOf(month.toLowerCase()) + 1
-
-  if (monthNumber === 0) {
-    throw new Error(
-      `invalid month ${month}, might you need the Dutch version of this function?`,
-    )
-  }
-
-  return monthNumber
+    ],
+    month,
+  )
 }