Merge 6f19bb1e0c into 47ea6b5092

server/utils/parsers/parseDate.js

@@ -0,0 +1,54 @@
+/**
+ * Parse a date string with multiple fallback formats
+ *
+ * @example
+ * parse('2024-01-15') => Date object for Jan 15, 2024
+ * parse('20240325')   => Date object for Mar 25, 2024
+ * parse('250312')     => Date object for Mar 12, 2025 (year <= 50 as 20xx)
+ * parse('750312')     => Date object for Mar 12, 1975 (year > 50 as 19xx)
+ *
+ * @param {string} dateString The date string to parse
+ * @returns {Date|null} Date object or null if unparseable
+ */
+function parseDate(dateString) {
+  if (!dateString || typeof dateString !== 'string') return null
+
+  const date = new Date(dateString)
+  if (!isNaN(date.getTime())) {
+    const year = date.getFullYear()
+    if (year >= 0 && year <= 9999) {
+      return date
+    }
+  }
+
+  const yyyymmddMatch = dateString.match(/^(\d{4})(\d{2})(\d{2})$/)
+  if (yyyymmddMatch) {
+    const [, year, month, day] = yyyymmddMatch
+    const monthNum = parseInt(month)
+    const dayNum = parseInt(day)
+    if (monthNum >= 1 && monthNum <= 12 && dayNum >= 1 && dayNum <= 31) {
+      const parsedDate = new Date(parseInt(year), monthNum - 1, dayNum)
+      if (!isNaN(parsedDate.getTime())) {
+        return parsedDate
+      }
+    }
+  }
+
+  const yymmddMatch = dateString.match(/^(\d{2})(\d{2})(\d{2})$/)
+  if (yymmddMatch) {
+    const [, year, month, day] = yymmddMatch
+    const monthNum = parseInt(month)
+    const dayNum = parseInt(day)
+    if (monthNum >= 1 && monthNum <= 12 && dayNum >= 1 && dayNum <= 31) {
+      const fullYear = parseInt(year) > 50 ? 1900 + parseInt(year) : 2000 + parseInt(year)
+      const parsedDate = new Date(fullYear, monthNum - 1, dayNum)
+      if (!isNaN(parsedDate.getTime())) {
+        return parsedDate
+      }
+    }
+  }
+
+  return null
+}
+
+module.exports.parse = parseDate

server/utils/podcastUtils.js

@@ -4,6 +4,7 @@ const Logger = require('../Logger')
 const { xmlToJSON, timestampToSeconds } = require('./index')
 const htmlSanitizer = require('../utils/htmlSanitizer')
 const Fuse = require('../libs/fusejs')
+const parseDate = require('../utils/parsers/parseDate')
 
 /**
  * @typedef RssPodcastChapter
@@ -147,6 +148,12 @@ function extractPodcastMetadata(channel) {
     if (value?.['_']) value = value['_']
     metadata[cleanKey] = value
   })
+
+  if (!metadata['pubDate']) {
+    const dateVal = extractFirstArrayItem(channel, 'date')
+    if (dateVal) metadata['pubDate'] = dateVal
+  }
+
   return metadata
 }
 
@@ -198,6 +205,15 @@ function extractEpisodeData(item) {
     } else {
       Logger.error(`[podcastUtils] Invalid pubDate ${item['pubDate']} for ${episode.enclosure.url}`)
     }
+  } else if (item['date']) {
+    const date = extractFirstArrayItem(item, 'date')
+    if (typeof date === 'string') {
+      episode.pubDate = date
+    } else if (typeof date?._ === 'string') {
+      episode.pubDate = date._
+    } else {
+      Logger.error(`[podcastUtils] Invalid date ${item['date']} for ${episode.enclosure.url}`)
+    }
   }
 
   if (item['guid']) {
@@ -265,7 +281,7 @@ function extractEpisodeData(item) {
 }
 
 function cleanEpisodeData(data) {
-  const pubJsDate = data.pubDate ? new Date(data.pubDate) : null
+  const pubJsDate = data.pubDate ? parseDate.parse(data.pubDate) : null
   const publishedAt = pubJsDate && !isNaN(pubJsDate) ? pubJsDate.valueOf() : null
 
   return {