summary refs log tree commit diff
diff options
context:
space:
mode:
authorRemita Amine <remitamine@gmail.com>2021-01-04 01:04:20 +0100
committerRemita Amine <remitamine@gmail.com>2021-01-04 01:14:26 +0100
commit8487e8b98afd1b469c2b9d29ee53bd173ff9a7e0 (patch)
tree8a8f8e59986f472f25c8e5a1dad4fa54b3d4f8c9
parent9c484c0019ba16cd4b4d686930bfa038f1bf9320 (diff)
downloadyoutube-dl-8487e8b98afd1b469c2b9d29ee53bd173ff9a7e0.tar.gz
youtube-dl-8487e8b98afd1b469c2b9d29ee53bd173ff9a7e0.tar.xz
youtube-dl-8487e8b98afd1b469c2b9d29ee53bd173ff9a7e0.zip
[googlepodcasts] Add new extractor
-rw-r--r--youtube_dl/extractor/extractors.py4
-rw-r--r--youtube_dl/extractor/googlepodcasts.py88
2 files changed, 92 insertions, 0 deletions
diff --git a/youtube_dl/extractor/extractors.py b/youtube_dl/extractor/extractors.py
index 82d440a6d..3668197b1 100644
--- a/youtube_dl/extractor/extractors.py
+++ b/youtube_dl/extractor/extractors.py
@@ -423,6 +423,10 @@ from .godtube import GodTubeIE
 from .golem import GolemIE
 from .googledrive import GoogleDriveIE
 from .googleplus import GooglePlusIE
+from .googlepodcasts import (
+    GooglePodcastsIE,
+    GooglePodcastsFeedIE,
+)
 from .googlesearch import GoogleSearchIE
 from .goshgay import GoshgayIE
 from .gputechconf import GPUTechConfIE
diff --git a/youtube_dl/extractor/googlepodcasts.py b/youtube_dl/extractor/googlepodcasts.py
new file mode 100644
index 000000000..31ad79907
--- /dev/null
+++ b/youtube_dl/extractor/googlepodcasts.py
@@ -0,0 +1,88 @@
+# coding: utf-8
+from __future__ import unicode_literals
+
+import json
+import re
+
+from .common import InfoExtractor
+from ..utils import (
+    clean_podcast_url,
+    int_or_none,
+    try_get,
+    urlencode_postdata,
+)
+
+
+class GooglePodcastsBaseIE(InfoExtractor):
+    _VALID_URL_BASE = r'https?://podcasts\.google\.com/feed/'
+
+    def _batch_execute(self, func_id, video_id, params):
+        return json.loads(self._download_json(
+            'https://podcasts.google.com/_/PodcastsUi/data/batchexecute',
+            video_id, data=urlencode_postdata({
+                'f.req': json.dumps([[[func_id, json.dumps(params), None, '1']]]),
+            }), transform_source=lambda x: self._search_regex(r'(?s)(\[.+\])', x, 'data'))[0][2])
+
+    def _extract_episode(self, episode):
+        return {
+            'id': episode[4][3],
+            'title': episode[8],
+            'url': clean_podcast_url(episode[13]),
+            'thumbnail': episode[2],
+            'description': episode[9],
+            'creator': try_get(episode, lambda x: x[14]),
+            'timestamp': int_or_none(episode[11]),
+            'duration': int_or_none(episode[12]),
+            'series': episode[1],
+        }
+
+
+class GooglePodcastsIE(GooglePodcastsBaseIE):
+    IE_NAME = 'google:podcasts'
+    _VALID_URL = GooglePodcastsBaseIE._VALID_URL_BASE + r'(?P<feed_url>[^/]+)/episode/(?P<id>[^/?&#]+)'
+    _TEST = {
+        'url': 'https://podcasts.google.com/feed/aHR0cHM6Ly9mZWVkcy5ucHIub3JnLzM0NDA5ODUzOS9wb2RjYXN0LnhtbA/episode/MzBlNWRlN2UtOWE4Yy00ODcwLTk2M2MtM2JlMmUyNmViOTRh',
+        'md5': 'fa56b2ee8bd0703e27e42d4b104c4766',
+        'info_dict': {
+            'id': '30e5de7e-9a8c-4870-963c-3be2e26eb94a',
+            'ext': 'mp3',
+            'title': 'WWDTM New Year 2021',
+            'description': 'We say goodbye to 2020 with Christine Baranksi, Doug Jones, Jonna Mendez, and Kellee Edwards.',
+            'upload_date': '20210102',
+            'timestamp': 1609606800,
+            'duration': 2901,
+            'series': "Wait Wait... Don't Tell Me!",
+        }
+    }
+
+    def _real_extract(self, url):
+        b64_feed_url, b64_guid = re.match(self._VALID_URL, url).groups()
+        episode = self._batch_execute(
+            'oNjqVe', b64_guid, [b64_feed_url, b64_guid])[1]
+        return self._extract_episode(episode)
+
+
+class GooglePodcastsFeedIE(GooglePodcastsBaseIE):
+    IE_NAME = 'google:podcasts:feed'
+    _VALID_URL = GooglePodcastsBaseIE._VALID_URL_BASE + r'(?P<id>[^/?&#]+)/?(?:[?#&]|$)'
+    _TEST = {
+        'url': 'https://podcasts.google.com/feed/aHR0cHM6Ly9mZWVkcy5ucHIub3JnLzM0NDA5ODUzOS9wb2RjYXN0LnhtbA',
+        'info_dict': {
+            'title': "Wait Wait... Don't Tell Me!",
+            'description': "NPR's weekly current events quiz. Have a laugh and test your news knowledge while figuring out what's real and what we've made up.",
+        },
+        'playlist_mincount': 20,
+    }
+
+    def _real_extract(self, url):
+        b64_feed_url = self._match_id(url)
+        data = self._batch_execute('ncqJEe', b64_feed_url, [b64_feed_url])
+
+        entries = []
+        for episode in (try_get(data, lambda x: x[1][0]) or []):
+            entries.append(self._extract_episode(episode))
+
+        feed = try_get(data, lambda x: x[3]) or []
+        return self.playlist_result(
+            entries, playlist_title=try_get(feed, lambda x: x[0]),
+            playlist_description=try_get(feed, lambda x: x[2]))