From b6e4f3a8c5e6ca867fa20de514807a1419980014 Mon Sep 17 00:00:00 2001
From: advplyr <advplyr@protonmail.com>
Date: Sat, 5 Mar 2022 18:54:24 -0600
Subject: [PATCH] Add:Podcast RSS feed parser

---
 server/ApiController.js      |  26 +++++++++
 server/utils/podcastUtils.js | 110 +++++++++++++++++++++++++++++++++++
 2 files changed, 136 insertions(+)
 create mode 100644 server/utils/podcastUtils.js

diff --git a/server/ApiController.js b/server/ApiController.js
index 45969c75..b63cc784 100644
--- a/server/ApiController.js
+++ b/server/ApiController.js
@@ -2,9 +2,11 @@ const express = require('express')
 const Path = require('path')
 const fs = require('fs-extra')
 const date = require('date-and-time')
+const axios = require('axios')
 
 const Logger = require('./Logger')
 const { isObject } = require('./utils/index')
+const { parsePodcastRssFeedXml } = require('./utils/podcastUtils')
 
 const BookController = require('./controllers/BookController')
 const LibraryController = require('./controllers/LibraryController')
@@ -179,6 +181,8 @@ class ApiController {
     this.router.post('/syncLocal', this.syncLocal.bind(this))
 
     this.router.post('/streams/:id/close', this.closeStream.bind(this))
+
+    this.router.post('/getPodcastFeed', this.getPodcastFeed.bind(this))
   }
 
   async findBooks(req, res) {
@@ -527,5 +531,27 @@ class ApiController {
     this.streamManager.closeStreamApiRequest(userId, streamId)
     res.sendStatus(200)
   }
+
+  getPodcastFeed(req, res) {
+    var url = req.body.rssFeed
+    if (!url) {
+      return res.status(400).send('Bad request')
+    }
+
+    axios.get(url).then(async (data) => {
+      if (!data || !data.data) {
+        Logger.error('Invalid podcast feed request response')
+        return res.status(500).send('Bad response from feed request')
+      }
+      var podcast = await parsePodcastRssFeedXml(data.data)
+      if (!podcast) {
+        return res.status(500).send('Invalid podcast RSS feed')
+      }
+      res.json(podcast)
+    }).catch((error) => {
+      console.error('Failed', error)
+      res.status(500).send(error)
+    })
+  }
 }
 module.exports = ApiController
\ No newline at end of file
diff --git a/server/utils/podcastUtils.js b/server/utils/podcastUtils.js
new file mode 100644
index 00000000..d0a2e607
--- /dev/null
+++ b/server/utils/podcastUtils.js
@@ -0,0 +1,110 @@
+const Logger = require('../Logger')
+const { xmlToJSON } = require('./index')
+
+function extractFirstArrayItem(json, key) {
+  if (!json[key] || !json[key].length) return null
+  return json[key][0]
+}
+
+function extractImage(channel) {
+  if (!channel.image || !channel.image.url || !channel.image.url.length) {
+    if (!channel['itunes:image'] || !channel['itunes:image'].length || !channel['itunes:image'][0]['$']) {
+      return null
+    }
+    var itunesImage = channel['itunes:image'][0]['$']
+    return itunesImage.href || null
+  }
+  return channel.image.url[0] || null
+}
+
+function extractCategories(channel) {
+  if (!channel['itunes:category'] || !channel['itunes:category'].length) return []
+  var categories = channel['itunes:category']
+  var cleanedCats = []
+  categories.forEach((cat) => {
+    if (!cat['$'] || !cat['$'].text) return
+    var cattext = cat['$'].text
+    if (cat['itunes:category']) {
+      var subcats = extractCategories(cat)
+      if (subcats.length) {
+        cleanedCats = cleanedCats.concat(subcats.map((subcat) => `${cattext}:${subcat}`))
+      } else {
+        cleanedCats.push(cattext)
+      }
+    } else {
+      cleanedCats.push(cattext)
+    }
+  })
+  return cleanedCats
+}
+
+function extractPodcastMetadata(channel) {
+  var arrayFields = ['title', 'language', 'description', 'itunes:explicit', 'itunes:author']
+  var metadata = {
+    image: extractImage(channel),
+    categories: extractCategories(channel)
+  }
+  arrayFields.forEach((key) => {
+    var cleanKey = key.split(':').pop()
+    metadata[cleanKey] = extractFirstArrayItem(channel, key)
+  })
+  return metadata
+}
+
+function extractEpisodeData(item) {
+  // Episode must have url
+  if (!item.enclosure || !item.enclosure.length || !item.enclosure[0]['$'] || !item.enclosure[0]['$'].url) {
+    Logger.error(`[podcastUtils] Invalid podcast episode data`)
+    return null
+  }
+  var arrayFields = ['title', 'pubDate', 'description', 'itunes:episodeType', 'itunes:episode', 'itunes:author', 'itunes:duration', 'itunes:explicit']
+  var episode = {
+    enclosure: {
+      ...item.enclosure[0]['$']
+    }
+  }
+  arrayFields.forEach((key) => {
+    var cleanKey = key.split(':').pop()
+    episode[cleanKey] = extractFirstArrayItem(item, key)
+  })
+  return episode
+}
+
+function extractPodcastEpisodes(items) {
+  var episodes = []
+  items.forEach((item) => {
+    var cleaned = extractEpisodeData(item)
+    if (cleaned) {
+      episodes.push(cleaned)
+    }
+  })
+  return episodes
+}
+
+function cleanPodcastJson(rssJson) {
+  if (!rssJson.channel || !rssJson.channel.length) {
+    Logger.error(`[podcastUtil] Invalid podcast no channel object`)
+    return null
+  }
+  var channel = rssJson.channel[0]
+  if (!channel.item || !channel.item.length) {
+    Logger.error(`[podcastUtil] Invalid podcast no episodes`)
+    return null
+  }
+
+  var podcast = {
+    metadata: extractPodcastMetadata(channel),
+    episodes: extractPodcastEpisodes(channel.item)
+  }
+  return podcast
+}
+
+module.exports.parsePodcastRssFeedXml = async (xml) => {
+  if (!xml) return null
+  var json = await xmlToJSON(xml)
+  if (!json || !json.rss) {
+    Logger.error('[podcastUtils] Invalid XML or RSS feed')
+    return null
+  }
+  return cleanPodcastJson(json.rss)
+}
\ No newline at end of file