From 0c01a551739975d3d1ba57190b2cd137dc0760f0 Mon Sep 17 00:00:00 2001 From: Nino Burini Date: Tue, 4 Feb 2025 11:38:09 +0100 Subject: [PATCH 1/6] Update _extractors.py Add entries for IlPost --- yt_dlp/extractor/_extractors.py | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/yt_dlp/extractor/_extractors.py b/yt_dlp/extractor/_extractors.py index c331bab78..f23ce76fe 100644 --- a/yt_dlp/extractor/_extractors.py +++ b/yt_dlp/extractor/_extractors.py @@ -851,7 +851,10 @@ IHeartRadioIE, IHeartRadioPodcastIE, ) -from .ilpost import IlPostIE +from .ilpost import ( + IlPostIE, + IlPostPodcastIE, +) from .iltalehti import IltalehtiIE from .imdb import ( ImdbIE, From 50ddaf1b85fee7561d29b68d1be2e7e59caf22e1 Mon Sep 17 00:00:00 2001 From: Nino Burini Date: Tue, 4 Feb 2025 11:38:59 +0100 Subject: [PATCH 2/6] Update ilpost.py Fix IlPost and add playlist support --- yt_dlp/extractor/ilpost.py | 85 ++++++++++++++++++++++++++++---------- 1 file changed, 63 insertions(+), 22 deletions(-) diff --git a/yt_dlp/extractor/ilpost.py b/yt_dlp/extractor/ilpost.py index da203cf5f..d758ab41c 100644 --- a/yt_dlp/extractor/ilpost.py +++ b/yt_dlp/extractor/ilpost.py @@ -9,11 +9,12 @@ ) from ..utils.traversal import traverse_obj +PODCAST_API = 'https://api-prod.ilpost.it/podcast/v1/podcast/%s?hits=20' class IlPostIE(InfoExtractor): - _VALID_URL = r'https?://(?:www\.)?ilpost\.it/episodes/(?P[^/?#]+)' + _VALID_URL = r'https?://(?:www\.)?ilpost\.it/podcasts/(?:.*?)/(?P[^/?#]+)' _TESTS = [{ - 'url': 'https://www.ilpost.it/episodes/1-avis-akvasas-ka/', + 'url': 'https://www.ilpost.it/podcasts/l-invasione/1-avis-akvasas-ka/', 'md5': '43649f002d85e1c2f319bb478d479c40', 'info_dict': { 'id': '2972047', @@ -26,7 +27,7 @@ class IlPostIE(InfoExtractor): 'duration': 2495.0, 'availability': 'public', 'series_id': '235598', - 'description': '', + 'thumbnail': 'https://www.ilpost.it/wp-content/uploads/2023/12/22/1703238848-copertina500x500.jpg', }, }] @@ -35,34 +36,74 @@ def _real_extract(self, url): webpage = self._download_webpage(url, display_id) endpoint_metadata = self._search_json( - r'var\s+ilpostpodcast\s*=', webpage, 'metadata', display_id) - episode_id = endpoint_metadata['post_id'] - podcast_id = endpoint_metadata['podcast_id'] - podcast_metadata = self._download_json( - endpoint_metadata['ajax_url'], display_id, data=urlencode_postdata({ - 'action': 'checkpodcast', - 'cookie': endpoint_metadata['cookie'], - 'post_id': episode_id, - 'podcast_id': podcast_id, - })) + r'{"props":{"pageProps":', webpage, 'metadata', display_id) + episode_id = endpoint_metadata['data']['data']['episode']['data'][0]['id'] + podcast_id = traverse_obj(endpoint_metadata, ('data', 'data', 'episode', 'data', 0, 'parent', 'id')) + podcast_metadata = traverse_obj(endpoint_metadata, ('data', 'data', 'episode', 'data', 0)) - episode = traverse_obj(podcast_metadata, ( - 'data', 'postcastList', lambda _, v: str(v['id']) == episode_id, {dict}), get_all=False) + episode = podcast_metadata if not episode: raise ExtractorError('Episode could not be extracted') return { - 'id': episode_id, - 'display_id': display_id, - 'series_id': podcast_id, - 'vcodec': 'none', + 'id': str(episode_id), + 'display_id': str(display_id), + 'series_id': str(podcast_id), **traverse_obj(episode, { 'title': ('title', {str}), - 'description': ('description', {str}), - 'url': ('podcast_raw_url', {url_or_none}), + 'url': ('episode_raw_url', {url_or_none}), 'thumbnail': ('image', {url_or_none}), + 'description': ('content_html', {str}), 'timestamp': ('timestamp', {int_or_none}), 'duration': ('milliseconds', {float_or_none(scale=1000)}), - 'availability': ('free', {lambda v: 'public' if v else 'subscriber_only'}), + 'availability': ('access_level', {lambda v: 'public' if v else 'subscriber_only'}), }), } + +class IlPostPodcastIE(InfoExtractor): + _VALID_URL = r'https?://(?:www\.)?ilpost\.it/podcasts/(?P[^/?#]+)' + _TESTS = [{ + 'url': 'https://www.ilpost.it/podcasts/morning/', + 'info_dict': { + 'id': 'morning', + 'display_id': 'morning', + 'title': 'Morning', + 'series': 'Morning', + 'season_number': 1, + }, + 'playlist_mincount': 20, + }, { + 'url': 'https://www.ilpost.it/podcasts/basaglia-e-i-suoi/', + 'info_dict': { + 'id': 'basaglia-e-i-suoi', + 'display_id': 'basaglia-e-i-suoi', + 'title': 'Basaglia e i suoi', + 'series': 'Basaglia e i suoi', + 'season_number': 1, + }, + 'playlist_mincount': 5, + }] + + def _real_extract(self, url): + id = self._match_valid_url(url).group('id') + data = self._download_json(PODCAST_API % id, id) + + entries = [{ + '_type': 'url', + 'url': episode['url'], + 'title': episode.get('title'), + 'description': episode.get('content_html'), + 'series': traverse_obj(data, ('data', 0, 'parent', 'title')), + 'season_number': 1, + 'episode_number': episode['id'], + } for episode in traverse_obj(data, ('data'))] + + return { + '_type': 'playlist', + 'id': id, + 'display_id': id, + 'title': traverse_obj(data, ('data', 0, 'parent', 'title')), + 'series': traverse_obj(data, ('data', 0, 'parent', 'title')), + 'entries': entries, + 'season_number': 1, + } From a487aa9379672cf53291fa62437d3a9b53973030 Mon Sep 17 00:00:00 2001 From: Nino Burini Date: Tue, 4 Feb 2025 11:47:41 +0100 Subject: [PATCH 3/6] Update ilpost.py Fix format --- yt_dlp/extractor/ilpost.py | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) diff --git a/yt_dlp/extractor/ilpost.py b/yt_dlp/extractor/ilpost.py index d758ab41c..f7f0cb920 100644 --- a/yt_dlp/extractor/ilpost.py +++ b/yt_dlp/extractor/ilpost.py @@ -5,7 +5,6 @@ float_or_none, int_or_none, url_or_none, - urlencode_postdata, ) from ..utils.traversal import traverse_obj @@ -24,6 +23,7 @@ class IlPostIE(InfoExtractor): 'url': 'https://www.ilpost.it/wp-content/uploads/2023/12/28/1703781217-l-invasione-pt1-v6.mp3', 'timestamp': 1703835014, 'upload_date': '20231229', + 'description': '

Circa tre miliardi di persone, oggi, parlano lingue che hanno un’unica antenata: dall’italiano all’inglese, passando per il farsi e l’islandese, queste lingue discendono tutte da una lingua arrivata in Europa circa cinquemila anni fa, insieme a un gruppo di persone ben preciso.
\nCon la loro lingua queste persone si portarono dietro anche alcuni oggetti, miti e leggende, e una certa visione della società, lasciando tracce indelebili ancora oggi.

\n

Per approfondire gli argomenti trattati nel podcast abbiamo raccolto in questa pagina le cose da leggere e da guardare dopo aver ascoltato le puntate.

\n', 'duration': 2495.0, 'availability': 'public', 'series_id': '235598', @@ -85,8 +85,8 @@ class IlPostPodcastIE(InfoExtractor): }] def _real_extract(self, url): - id = self._match_valid_url(url).group('id') - data = self._download_json(PODCAST_API % id, id) + display_id = self._match_valid_url(url).group('id') + data = self._download_json(PODCAST_API % display_id, display_id) entries = [{ '_type': 'url', @@ -100,8 +100,8 @@ def _real_extract(self, url): return { '_type': 'playlist', - 'id': id, - 'display_id': id, + 'id': display_id, + 'display_id': display_id, 'title': traverse_obj(data, ('data', 0, 'parent', 'title')), 'series': traverse_obj(data, ('data', 0, 'parent', 'title')), 'entries': entries, From 54aec0ce21ee74afd7feac06cf2d1a09eac83de1 Mon Sep 17 00:00:00 2001 From: Nino Burini Date: Tue, 4 Feb 2025 11:48:49 +0100 Subject: [PATCH 4/6] Update ilpost.py Add missing newlines --- yt_dlp/extractor/ilpost.py | 2 ++ 1 file changed, 2 insertions(+) diff --git a/yt_dlp/extractor/ilpost.py b/yt_dlp/extractor/ilpost.py index f7f0cb920..9206e511b 100644 --- a/yt_dlp/extractor/ilpost.py +++ b/yt_dlp/extractor/ilpost.py @@ -10,6 +10,7 @@ PODCAST_API = 'https://api-prod.ilpost.it/podcast/v1/podcast/%s?hits=20' + class IlPostIE(InfoExtractor): _VALID_URL = r'https?://(?:www\.)?ilpost\.it/podcasts/(?:.*?)/(?P[^/?#]+)' _TESTS = [{ @@ -60,6 +61,7 @@ def _real_extract(self, url): }), } + class IlPostPodcastIE(InfoExtractor): _VALID_URL = r'https?://(?:www\.)?ilpost\.it/podcasts/(?P[^/?#]+)' _TESTS = [{ From 3c0f2a64e666d345fb13309a7d6773bb16e1adaf Mon Sep 17 00:00:00 2001 From: Nino Burini Date: Tue, 4 Feb 2025 13:33:05 +0100 Subject: [PATCH 5/6] Update ilpost.py Correct _VALID_URL regex for IlPostPodcast --- yt_dlp/extractor/ilpost.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/yt_dlp/extractor/ilpost.py b/yt_dlp/extractor/ilpost.py index 9206e511b..dba66c3c8 100644 --- a/yt_dlp/extractor/ilpost.py +++ b/yt_dlp/extractor/ilpost.py @@ -63,7 +63,7 @@ def _real_extract(self, url): class IlPostPodcastIE(InfoExtractor): - _VALID_URL = r'https?://(?:www\.)?ilpost\.it/podcasts/(?P[^/?#]+)' + _VALID_URL = r'https?://(?:www\.)?ilpost\.it/podcasts/(?P[\-a-zA-Z0-9]+)' _TESTS = [{ 'url': 'https://www.ilpost.it/podcasts/morning/', 'info_dict': { From 933ff47afaee38a1fcf8b405eab161a7c84f17be Mon Sep 17 00:00:00 2001 From: Nino Burini Date: Tue, 4 Feb 2025 23:07:46 +0100 Subject: [PATCH 6/6] Update ilpost.py Update regex for IlPostPodcast --- yt_dlp/extractor/ilpost.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/yt_dlp/extractor/ilpost.py b/yt_dlp/extractor/ilpost.py index dba66c3c8..cee60d4ad 100644 --- a/yt_dlp/extractor/ilpost.py +++ b/yt_dlp/extractor/ilpost.py @@ -63,7 +63,7 @@ def _real_extract(self, url): class IlPostPodcastIE(InfoExtractor): - _VALID_URL = r'https?://(?:www\.)?ilpost\.it/podcasts/(?P[\-a-zA-Z0-9]+)' + _VALID_URL = r'https?://(?:www\.)?ilpost\.it/podcasts/(?P[a-zA-Z0-9\-]+)/?$' _TESTS = [{ 'url': 'https://www.ilpost.it/podcasts/morning/', 'info_dict': {