From 04e4bfa0ebb32b3f86d1128c242754f7929bdf1f Mon Sep 17 00:00:00 2001
From: lost <lost@negativezero.link>
Date: Mon, 8 Nov 2021 00:00:00 +0000
Subject: [PATCH] Add Castbox support.

---
 scripts/determine_site.py     |  2 ++
 scripts/extractors/castbox.py | 28 ++++++++++++++++++++++++++++
 tests/castbox.txt             |  7 +++++++
 3 files changed, 37 insertions(+)
 create mode 100644 scripts/extractors/castbox.py
 create mode 100644 tests/castbox.txt

diff --git a/scripts/determine_site.py b/scripts/determine_site.py
index 97eed1e..c415fd5 100644
--- a/scripts/determine_site.py
+++ b/scripts/determine_site.py
@@ -3,6 +3,8 @@
 supported_sites = {
 	"apple_podcasts" : [ "podcasts.apple.com" ],
 
+	"castbox" : ["castbox.fm"],
+
 	"castro_fm" : ["castro.fm"],
 
 	"chirbit" : [
diff --git a/scripts/extractors/castbox.py b/scripts/extractors/castbox.py
new file mode 100644
index 0000000..7700f82
--- /dev/null
+++ b/scripts/extractors/castbox.py
@@ -0,0 +1,28 @@
+#!/usr/bin/python3
+
+from utils import *
+from download_page import download
+from urllib.parse import unquote
+
+# portable code to get filename
+import os
+platform = os.path.basename(__file__)
+if platform.endswith(".py"):
+	platform = platform[:(-3)]
+
+def extract_from_page (page, verbosity):
+	# it's like %22%2C%22rss_url%22%3A%22https%3A%2F%2Ffeeds.npr.org%2F510310%2Fpodcast.xml%22%2C%22
+	# and we have to fix the codes
+	feed = search (page, '%22%2C%22rss_url%22%3A%22', '%22%2C%22')
+	if not feed is None:
+		return unquote(feed)
+
+def extract (url, page=None, network=False, verbosity=3, args={}):
+	# cannot get feed from URL alone
+	if not network:
+		return None
+
+	page = download (platform, url, args, verbosity)
+	feed = extract_from_page (page, verbosity)
+	if not feed is None:
+		return feed
diff --git a/tests/castbox.txt b/tests/castbox.txt
new file mode 100644
index 0000000..0aed193
--- /dev/null
+++ b/tests/castbox.txt
@@ -0,0 +1,7 @@
+# channel pages
+https://castbox.fm/channel/id2698771 https://feeds.npr.org/510310/podcast.xml
+https://castbox.fm/channel/id1238764 https://rss.art19.com/generation-why-podcast
+
+# episode pages
+https://castbox.fm/episode/Congress-Passes-Biden's-Trillion-Dollar-Transit%2C-Broadband%2C-And-Power-Bill-id2698771-id440269324 https://feeds.npr.org/510310/podcast.xml
+https://castbox.fm/episode/Introducing%3A-Operator-id1238764-id440095234 https://rss.art19.com/generation-why-podcast