fix: lantarenvenster (29 feb next year edge case)

ckuijjer · Dec 6, 2023 · 34dbaef · 34dbaef
1 parent cd7a941
commit 34dbaef
Show file tree

Hide file tree

Showing 14 changed files with 118 additions and 95 deletions.
diff --git a/cloud/scrapers/filmhuislumen.ts b/cloud/scrapers/filmhuislumen.ts
@@ -67,14 +67,12 @@ const extractFromMainPage = async () => {
 
       const [hour, minute] = splitTimeDot(time)
 
-      const year = guessYear(
-        DateTime.fromObject({
-          day,
-          month,
-          hour,
-          minute,
-        }),
-      )
+      const year = guessYear({
+        day,
+        month,
+        hour,
+        minute,
+      })
 
       return {
         title,

diff --git a/cloud/scrapers/forumgroningen.ts b/cloud/scrapers/forumgroningen.ts
@@ -71,14 +71,12 @@ const extractFromMoviePage = async ({
           const [startTime, endTime] = time.split(/ tot /)
           const [hour, minute] = splitTime(startTime)
 
-          const year = guessYear(
-            DateTime.fromObject({
-              day,
-              month,
-              hour,
-              minute,
-            }),
-          )
+          const year = guessYear({
+            day,
+            month,
+            hour,
+            minute,
+          })
 
           return {
             title: scrapeResult.title,

diff --git a/cloud/scrapers/hartlooper.ts b/cloud/scrapers/hartlooper.ts
@@ -97,14 +97,12 @@ const extractFromMoviePage = async ({ url }: { url: string }) => {
         const month = fullMonthToNumberDutch(date.month)
         const [hour, minute] = splitTime(time)
 
-        const year = guessYear(
-          DateTime.fromObject({
-            day,
-            month,
-            hour,
-            minute,
-          }),
-        )
+        const year = guessYear({
+          day,
+          month,
+          hour,
+          minute,
+        })
 
         return {
           title: cleanTitle(movie.title),

diff --git a/cloud/scrapers/lab111.ts b/cloud/scrapers/lab111.ts
@@ -69,14 +69,12 @@ const extractFromMainPage = async () => {
           const day = Number(dayString)
           const month = shortMonthToNumberEnglish(monthString)
           const [hour, minute] = splitTime(time)
-          const year = guessYear(
-            DateTime.fromObject({
-              day,
-              month,
-              hour,
-              minute,
-            }),
-          )
+          const year = guessYear({
+            day,
+            month,
+            hour,
+            minute,
+          })
 
           logger.debug('extracted date', {
             dateString: date,

diff --git a/cloud/scrapers/lantarenvenster.ts b/cloud/scrapers/lantarenvenster.ts
@@ -63,16 +63,16 @@ export const extractFromMoviePage = async (
         .map((time) => {
           const [dayOfWeek, dayString, monthString] = date.split(' ')
           const day = Number(dayString)
+
           const month = shortMonthToNumberDutch(monthString)
           const [hour, minute] = splitTime(time)
-          const year = guessYear(
-            DateTime.fromObject({
-              day,
-              month,
-              hour,
-              minute,
-            }),
-          )
+
+          const year = guessYear({
+            day,
+            month,
+            hour,
+            minute,
+          })
 
           return {
             title: cleanTitle(movie.title),

diff --git a/cloud/scrapers/lumiere.ts b/cloud/scrapers/lumiere.ts
@@ -93,12 +93,10 @@ const extractFromMoviePage = async (url: string): Promise<Screening[]> => {
       const day = Number(dayString)
       const month = shortMonthToNumberDutch(monthString)
 
-      const year = guessYear(
-        DateTime.fromObject({
-          day,
-          month,
-        }),
-      )
+      const year = guessYear({
+        day,
+        month,
+      })
 
       return times.map((time) => {
         const [hour, minute] = splitTime(time)

diff --git a/cloud/scrapers/rialto.ts b/cloud/scrapers/rialto.ts
@@ -5,6 +5,8 @@ import { DateTime } from 'luxon'
 
 import guessYear from './utils/guessYear'
 import { logger as parentLogger } from '../powertools'
+import { fullMonthToNumberEnglish } from './utils/monthToNumber'
+import splitTime from './utils/splitTime'
 
 const logger = parentLogger.createChild({
   persistentLogAttributes: {
@@ -53,13 +55,20 @@ const extractFromMoviePage = async ({
     ([cinema, dates]) => {
       return Object.entries(dates).flatMap(([dateString, times]) => {
         return times.map(({ time, link, text }) => {
-          let date = DateTime.fromFormat(
-            `${dateString} ${time}`,
-            'EEEE d MMMM H:mm',
-          )
-
-          const year = guessYear(date)
-          date = date.set({ year })
+          const [dayOfWeek, dayString, monthString] = dateString.split(/\s+/)
+          const day = Number(dayString)
+          const month = fullMonthToNumberEnglish(monthString)
+          const [hour, minute] = splitTime(time)
+
+          const year = guessYear({ day, month, hour, minute })
+
+          const date = DateTime.fromObject({
+            year,
+            day,
+            month,
+            hour,
+            minute,
+          })
 
           return {
             title,

diff --git a/cloud/scrapers/schuur.ts b/cloud/scrapers/schuur.ts
@@ -62,14 +62,12 @@ const extractFromMainPage = async () => {
         const month = shortMonthToNumberDutch(monthString)
         const [hour, minute] = splitTime(time)
 
-        const year = guessYear(
-          DateTime.fromObject({
-            day,
-            month,
-            hour,
-            minute,
-          }),
-        )
+        const year = guessYear({
+          day,
+          month,
+          hour,
+          minute,
+        })
 
         return {
           title,

diff --git a/cloud/scrapers/slachtstraat.ts b/cloud/scrapers/slachtstraat.ts
@@ -97,14 +97,12 @@ const extractFromMoviePage = async ({ url }: { url: string }) => {
         const month = fullMonthToNumberDutch(date.month)
         const [hour, minute] = splitTime(time)
 
-        const year = guessYear(
-          DateTime.fromObject({
-            day,
-            month,
-            hour,
-            minute,
-          }),
-        )
+        const year = guessYear({
+          day,
+          month,
+          hour,
+          minute,
+        })
 
         return {
           title: cleanTitle(movie.title),

diff --git a/cloud/scrapers/springhaver.ts b/cloud/scrapers/springhaver.ts
@@ -95,14 +95,12 @@ const extractFromMoviePage = async ({ url }: { url: string }) => {
         const month = fullMonthToNumberDutch(date.month)
         const [hour, minute] = splitTime(time)
 
-        const year = guessYear(
-          DateTime.fromObject({
-            day,
-            month,
-            hour,
-            minute,
-          }),
-        )
+        const year = guessYear({
+          day,
+          month,
+          hour,
+          minute,
+        })
 
         return {
           title: cleanTitle(movie.title),

diff --git a/cloud/scrapers/studiok.ts b/cloud/scrapers/studiok.ts
@@ -75,14 +75,12 @@ const extractFromMoviePage = async (url: string) => {
         })
         .map(({ time, title }) => {
           const [hour, minute] = splitTime(time)
-          const year = guessYear(
-            DateTime.fromObject({
-              day,
-              month,
-              hour,
-              minute,
-            }),
-          )
+          const year = guessYear({
+            day,
+            month,
+            hour,
+            minute,
+          })
 
           return {
             title: cleanTitle(title),

diff --git a/cloud/scrapers/themovies.ts b/cloud/scrapers/themovies.ts
@@ -62,14 +62,12 @@ const extractFromMainPage = async () => {
             .split(':')
             .map(Number)
 
-          const year = guessYear(
-            DateTime.fromObject({
-              day,
-              month,
-              hour,
-              minute,
-            }),
-          )
+          const year = guessYear({
+            day,
+            month,
+            hour,
+            minute,
+          })
 
           return {
             title,

diff --git a/cloud/scrapers/utils/guessYear.ts b/cloud/scrapers/utils/guessYear.ts
@@ -1,8 +1,16 @@
-import { DateTime } from 'luxon'
+import { DateTime, DateObjectUnits } from 'luxon'
 
 // if the date is over half a year in the past, it's likely one belonging to next year
-const guessYear = (date: DateTime) => {
+const guessYear = (dateObjects: DateObjectUnits) => {
   const now = DateTime.local()
+
+  let date = DateTime.fromObject(dateObjects)
+
+  // handle edge case where you're looking for 29 feb in a leap year that's next year
+  if (!date.isValid && date.invalidReason === 'unit out of range') {
+    date = DateTime.fromObject({ ...dateObjects, year: now.year + 1 })
+  }
+
   if (date < now.minus({ months: 6 })) {
     return date.year + 1
   } else {

diff --git a/cloud/scrapers/utils/monthToNumber.ts b/cloud/scrapers/utils/monthToNumber.ts
@@ -50,6 +50,32 @@ export const shortMonthToNumberDutch = (month: string) => {
   return monthNumber
 }
 
+export const fullMonthToNumberEnglish = (month: string) => {
+  const monthNumber =
+    [
+      'january',
+      'february',
+      'march',
+      'april',
+      'may',
+      'june',
+      'july',
+      'august',
+      'september',
+      '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
+}
+
 export const shortMonthToNumberEnglish = (month: string) => {
   const monthNumber =
     [