From aefaba9f947a99e73f908ebc6cb286a93d7884a1 Mon Sep 17 00:00:00 2001 From: Alard Date: Mon, 5 Feb 2024 15:22:10 +0100 Subject: [PATCH 1/6] [extractor/tv5monde] Update for new tvmonde.com website --- yt_dlp/extractor/tv5mondeplus.py | 118 +++++++++++++++---------------- 1 file changed, 59 insertions(+), 59 deletions(-) diff --git a/yt_dlp/extractor/tv5mondeplus.py b/yt_dlp/extractor/tv5mondeplus.py index a445fae85..706097f05 100644 --- a/yt_dlp/extractor/tv5mondeplus.py +++ b/yt_dlp/extractor/tv5mondeplus.py @@ -1,3 +1,4 @@ +import re import urllib.parse from .common import InfoExtractor @@ -14,73 +15,69 @@ from ..utils import ( class TV5MondePlusIE(InfoExtractor): IE_DESC = 'TV5MONDE+' - _VALID_URL = r'https?://(?:www\.)?(?:tv5mondeplus|revoir\.tv5monde)\.com/toutes-les-videos/[^/]+/(?P[^/?#]+)' + _VALID_URL = r'https?://(?:www\.)?tv5monde\.com/tv/video/(?P[^/?#]+)' _TESTS = [{ - # movie - 'url': 'https://revoir.tv5monde.com/toutes-les-videos/cinema/les-novices', - 'md5': 'c86f60bf8b75436455b1b205f9745955', + # documentary + 'url': 'https://www.tv5monde.com/tv/video/65931-baudouin-l-heritage-d-un-roi-baudouin-l-heritage-d-un-roi', + 'md5': 'd2a708902d3df230a357c99701aece05', 'info_dict': { - 'id': 'ZX0ipMyFQq_6D4BA7b', - 'display_id': 'les-novices', + 'id': '3FPa7JMu21_6D4BA7b', + 'display_id': '65931-baudouin-l-heritage-d-un-roi-baudouin-l-heritage-d-un-roi', 'ext': 'mp4', - 'title': 'Les novices', - 'description': 'md5:2e7c33ba3ad48dabfcc2a956b88bde2b', - 'upload_date': '20230821', - 'thumbnail': 'https://revoir.tv5monde.com/uploads/media/video_thumbnail/0738/60/01e952b7ccf36b7c6007ec9131588954ab651de9.jpeg', - 'duration': 5177, - 'episode': 'Les novices', + 'title': "Baudouin, l'héritage d'un roi", + 'thumbnail': 'https://psi.tv5monde.com/upsilon-images/960x540/6f/baudouin-f49c6b0e.jpg', + 'duration': 4842, + 'upload_date': '20240130', + 'episode': "Baudouin, l'héritage d'un roi", + 'description': 'md5:78125c74a5cac06d7743a2d09126edad', }, }, { # series episode - 'url': 'https://revoir.tv5monde.com/toutes-les-videos/series-fictions/opj-les-dents-de-la-terre-2', + 'url': 'https://www.tv5monde.com/tv/video/5639-allo-tribunal-vol-de-projet', + 'md5': '64bbc2fe7e9867d97292eb02b9cd892a', 'info_dict': { - 'id': 'wJ0eeEPozr_6D4BA7b', - 'display_id': 'opj-les-dents-de-la-terre-2', + 'id': 'YAGqEEHqri_6D4BA7b', + 'display_id': '5639-allo-tribunal-vol-de-projet', 'ext': 'mp4', - 'title': "OPJ - Les dents de la Terre (2)", - 'description': 'md5:288f87fd68d993f814e66e60e5302d9d', - 'upload_date': '20230823', - 'series': 'OPJ', - 'episode': 'Les dents de la Terre (2)', - 'duration': 2877, - 'thumbnail': 'https://dl-revoir.tv5monde.com/images/1a/5753448.jpg' + 'title': 'Allô tribunal - Vol de projet', + 'description': "md5:da441d0b328793dfb4059884fbee6af3", + 'thumbnail': 'https://psi.tv5monde.com/media/image/960px/5157125.jpg', + 'duration': 1483, + 'upload_date': '20240205', + 'series': 'Allô tribunal', + 'episode': 'Vol de projet', }, }, { # movie - 'url': 'https://revoir.tv5monde.com/toutes-les-videos/cinema/ceux-qui-travaillent', - 'md5': '32fa0cde16a4480d1251502a66856d5f', + 'url': 'https://www.tv5monde.com/tv/video/47676-l-ennemi', + 'md5': '2013bb2c4c45ec39bdd0ff1df85ee7e7', 'info_dict': { - 'id': 'dc57a011-ec4b-4648-2a9a-4f03f8352ed3', - 'display_id': 'ceux-qui-travaillent', + 'id': 'x3fWbhTPcN_6D4BA7b', + 'display_id': '47676-l-ennemi', 'ext': 'mp4', - 'title': 'Ceux qui travaillent', - 'description': 'md5:570e8bb688036ace873b2d50d24c026d', - 'upload_date': '20210819', + 'title': "L'ennemi", + 'description': 'md5:8474e44ab7f2c7c8984a038b2922c69e', + 'thumbnail': 'https://psi.tv5monde.com/upsilon-images/960x540/08/Ennemi-05b7ee31.jpg', + 'duration': 6005, + 'upload_date': '20240212', + 'episode': "L'ennemi", }, - 'skip': 'no longer available', }, { - # series episode - 'url': 'https://revoir.tv5monde.com/toutes-les-videos/series-fictions/vestiaires-caro-actrice', + # news + 'url': 'https://www.tv5monde.com/tv/video/64797-tv5monde-le-journal-edition-du-05-02-24-11h', + 'md5': 'aacefe2f227e9e369146dbdc1ed048b4', 'info_dict': { - 'id': '9e9d599e-23af-6915-843e-ecbf62e97925', - 'display_id': 'vestiaires-caro-actrice', + 'id': '8HnwWTv6cf_6D4BA7b', + 'display_id': '64797-tv5monde-le-journal-edition-du-05-02-24-11h', 'ext': 'mp4', - 'title': "Vestiaires - Caro actrice", - 'description': 'md5:db15d2e1976641e08377f942778058ea', - 'upload_date': '20210819', - 'series': "Vestiaires", - 'episode': 'Caro actrice', + 'title': 'TV5MONDE, le journal', + 'description': 'md5:777dc209eaa4423b678477c36b0b04a8', + 'thumbnail': 'https://psi.tv5monde.com/media/image/960px/6091980.jpg', + 'duration': 862, + 'upload_date': '20240205', + 'series': 'TV5MONDE, le journal', + 'episode': 'TV5MONDE, le journal', }, - 'params': { - 'skip_download': True, - }, - 'skip': 'no longer available', - }, { - 'url': 'https://revoir.tv5monde.com/toutes-les-videos/series-fictions/neuf-jours-en-hiver-neuf-jours-en-hiver', - 'only_matching': True, - }, { - 'url': 'https://revoir.tv5monde.com/toutes-les-videos/info-societe/le-journal-de-la-rts-edition-du-30-01-20-19h30', - 'only_matching': True, }] _GEO_BYPASS = False @@ -98,7 +95,6 @@ class TV5MondePlusIE(InfoExtractor): if ">Ce programme n'est malheureusement pas disponible pour votre zone géographique.<" in webpage: self.raise_geo_restricted(countries=['FR']) - title = episode = self._html_search_regex(r'

([^<]+)', webpage, 'title') vpl_data = extract_attributes(self._search_regex( r'(<[^>]+class="video_player_loader"[^>]+>)', webpage, 'video player loader')) @@ -151,16 +147,20 @@ class TV5MondePlusIE(InfoExtractor): duration = (int_or_none(try_get(metadata, lambda x: x['content']['duration'])) or parse_duration(self._html_search_meta('duration', webpage))) + title = episode = self._html_search_regex(r'

([^<]+)', webpage, 'title') + series = self._html_search_regex(r'

([^<]+)', webpage, 'title', default=None) + subtitle = self._html_search_regex(r'

([^<]+)', webpage, 'subtitle', default=None) + if series and subtitle: + title = '%s - %s' % (series, subtitle) + episode = subtitle + + ep_summary = self._search_regex( + r']+class="ep-summary"[^>]*>(.+?)', webpage, + 'episode summary', fatal=False, flags=re.DOTALL) + description = self._html_search_regex( - r'(?s)]+class=["\']episode-texte[^>]+>(.+?)', webpage, - 'description', fatal=False) - - series = self._html_search_regex( - r']+class=["\']episode-emission[^>]+>([^<]+)', webpage, - 'series', default=None) - - if series and series != title: - title = '%s - %s' % (series, title) + r'

(.+?)

', ep_summary, + 'description', fatal=False, flags=re.DOTALL) upload_date = self._search_regex( r'(?:date_publication|publish_date)["\']\s*:\s*["\'](\d{4}_\d{2}_\d{2})', From 40a297a6edb71c15645044d237b7eb6666c14aa8 Mon Sep 17 00:00:00 2001 From: Alard Date: Thu, 11 Apr 2024 22:39:20 +0200 Subject: [PATCH 2/6] [ie/TV5MondePlus] Suggested changes. --- yt_dlp/extractor/tv5mondeplus.py | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) diff --git a/yt_dlp/extractor/tv5mondeplus.py b/yt_dlp/extractor/tv5mondeplus.py index 706097f05..87ffc74e2 100644 --- a/yt_dlp/extractor/tv5mondeplus.py +++ b/yt_dlp/extractor/tv5mondeplus.py @@ -14,7 +14,8 @@ from ..utils import ( class TV5MondePlusIE(InfoExtractor): - IE_DESC = 'TV5MONDE+' + IE_NAME = 'TV5MONDE' + IE_DESC = 'TV5MONDE' _VALID_URL = r'https?://(?:www\.)?tv5monde\.com/tv/video/(?P[^/?#]+)' _TESTS = [{ # documentary @@ -40,7 +41,7 @@ class TV5MondePlusIE(InfoExtractor): 'display_id': '5639-allo-tribunal-vol-de-projet', 'ext': 'mp4', 'title': 'Allô tribunal - Vol de projet', - 'description': "md5:da441d0b328793dfb4059884fbee6af3", + 'description': 'md5:da441d0b328793dfb4059884fbee6af3', 'thumbnail': 'https://psi.tv5monde.com/media/image/960px/5157125.jpg', 'duration': 1483, 'upload_date': '20240205', @@ -147,11 +148,10 @@ class TV5MondePlusIE(InfoExtractor): duration = (int_or_none(try_get(metadata, lambda x: x['content']['duration'])) or parse_duration(self._html_search_meta('duration', webpage))) - title = episode = self._html_search_regex(r'

([^<]+)', webpage, 'title') + title = episode = self._html_search_regex(r'

([^<]+)', webpage, 'title', default=None) series = self._html_search_regex(r'

([^<]+)', webpage, 'title', default=None) subtitle = self._html_search_regex(r'

([^<]+)', webpage, 'subtitle', default=None) - if series and subtitle: - title = '%s - %s' % (series, subtitle) + if subtitle: episode = subtitle ep_summary = self._search_regex( From 6b1ceff50b83d918f50eb2dafe5ac2f819c72935 Mon Sep 17 00:00:00 2001 From: Alard Date: Thu, 11 Apr 2024 22:55:01 +0200 Subject: [PATCH 3/6] [ie/TV5MondePlus] Update new tests. --- yt_dlp/extractor/tv5mondeplus.py | 56 ++++++++++++++++---------------- 1 file changed, 28 insertions(+), 28 deletions(-) diff --git a/yt_dlp/extractor/tv5mondeplus.py b/yt_dlp/extractor/tv5mondeplus.py index 87ffc74e2..ea38bf4cf 100644 --- a/yt_dlp/extractor/tv5mondeplus.py +++ b/yt_dlp/extractor/tv5mondeplus.py @@ -34,48 +34,48 @@ class TV5MondePlusIE(InfoExtractor): }, }, { # series episode - 'url': 'https://www.tv5monde.com/tv/video/5639-allo-tribunal-vol-de-projet', - 'md5': '64bbc2fe7e9867d97292eb02b9cd892a', + 'url': 'https://www.tv5monde.com/tv/video/52952-toute-la-vie-mardi-23-mars-2021', + 'md5': 'f5e09637cadd55639c05874e22eb56bf', 'info_dict': { - 'id': 'YAGqEEHqri_6D4BA7b', - 'display_id': '5639-allo-tribunal-vol-de-projet', + 'id': 'obRRZ8m6g9_6D4BA7b', + 'display_id': '52952-toute-la-vie-mardi-23-mars-2021', 'ext': 'mp4', - 'title': 'Allô tribunal - Vol de projet', - 'description': 'md5:da441d0b328793dfb4059884fbee6af3', - 'thumbnail': 'https://psi.tv5monde.com/media/image/960px/5157125.jpg', - 'duration': 1483, - 'upload_date': '20240205', - 'series': 'Allô tribunal', - 'episode': 'Vol de projet', + 'title': 'Toute la vie', + 'description': 'md5:a824a2e1dfd94cf45fa379a1fb43ce65', + 'thumbnail': 'https://psi.tv5monde.com/media/image/960px/5880553.jpg', + 'duration': 2526, + 'upload_date': '20240411', + 'series': 'Toute la vie', + 'episode': 'Mardi 23 mars 2021', }, }, { # movie - 'url': 'https://www.tv5monde.com/tv/video/47676-l-ennemi', - 'md5': '2013bb2c4c45ec39bdd0ff1df85ee7e7', + 'url': 'https://www.tv5monde.com/tv/video/8771-ce-fleuve-qui-nous-charrie-ce-fleuve-qui-nous-charrie-p001-ce-fleuve-qui-nous-charrie', + 'md5': '87cefc34e10a6bf4f7823cccd7b36eb2', 'info_dict': { - 'id': 'x3fWbhTPcN_6D4BA7b', - 'display_id': '47676-l-ennemi', + 'id': 'DOcfvdLKXL_6D4BA7b', + 'display_id': '8771-ce-fleuve-qui-nous-charrie-ce-fleuve-qui-nous-charrie-p001-ce-fleuve-qui-nous-charrie', 'ext': 'mp4', - 'title': "L'ennemi", - 'description': 'md5:8474e44ab7f2c7c8984a038b2922c69e', - 'thumbnail': 'https://psi.tv5monde.com/upsilon-images/960x540/08/Ennemi-05b7ee31.jpg', - 'duration': 6005, - 'upload_date': '20240212', - 'episode': "L'ennemi", + 'title': 'Ce fleuve qui nous charrie', + 'description': 'md5:62ba3f875343c7fc4082bdfbbc1be992', + 'thumbnail': 'https://psi.tv5monde.com/media/image/960px/5476617.jpg', + 'duration': 5300, + 'upload_date': '20240411', + 'episode': 'Ce fleuve qui nous charrie', }, }, { # news - 'url': 'https://www.tv5monde.com/tv/video/64797-tv5monde-le-journal-edition-du-05-02-24-11h', - 'md5': 'aacefe2f227e9e369146dbdc1ed048b4', + 'url': 'https://www.tv5monde.com/tv/video/68764-tv5monde-le-journal-edition-du-11-04-24-11h', + 'md5': 'c62977d6d10754a2ecebba70ad370479', 'info_dict': { - 'id': '8HnwWTv6cf_6D4BA7b', - 'display_id': '64797-tv5monde-le-journal-edition-du-05-02-24-11h', + 'id': '0NU1lYMa0Q_6D4BA7b', + 'display_id': '68764-tv5monde-le-journal-edition-du-11-04-24-11h', 'ext': 'mp4', 'title': 'TV5MONDE, le journal', 'description': 'md5:777dc209eaa4423b678477c36b0b04a8', - 'thumbnail': 'https://psi.tv5monde.com/media/image/960px/6091980.jpg', - 'duration': 862, - 'upload_date': '20240205', + 'thumbnail': 'https://psi.tv5monde.com/media/image/960px/6183997.jpg', + 'duration': 836, + 'upload_date': '20240411', 'series': 'TV5MONDE, le journal', 'episode': 'TV5MONDE, le journal', }, From 1aa522dfc418ef949cc0ee561de809069b72f3ad Mon Sep 17 00:00:00 2001 From: Alard Date: Sat, 13 Apr 2024 22:45:06 +0200 Subject: [PATCH 4/6] [ie/TV5MondePlus] Small fixes. --- yt_dlp/extractor/tv5mondeplus.py | 9 +++------ 1 file changed, 3 insertions(+), 6 deletions(-) diff --git a/yt_dlp/extractor/tv5mondeplus.py b/yt_dlp/extractor/tv5mondeplus.py index ea38bf4cf..18e750328 100644 --- a/yt_dlp/extractor/tv5mondeplus.py +++ b/yt_dlp/extractor/tv5mondeplus.py @@ -5,6 +5,7 @@ from .common import InfoExtractor from ..utils import ( determine_ext, extract_attributes, + get_element_html_by_class, int_or_none, parse_duration, traverse_obj, @@ -15,7 +16,6 @@ from ..utils import ( class TV5MondePlusIE(InfoExtractor): IE_NAME = 'TV5MONDE' - IE_DESC = 'TV5MONDE' _VALID_URL = r'https?://(?:www\.)?tv5monde\.com/tv/video/(?P[^/?#]+)' _TESTS = [{ # documentary @@ -149,14 +149,11 @@ class TV5MondePlusIE(InfoExtractor): or parse_duration(self._html_search_meta('duration', webpage))) title = episode = self._html_search_regex(r'

([^<]+)', webpage, 'title', default=None) - series = self._html_search_regex(r'

([^<]+)', webpage, 'title', default=None) subtitle = self._html_search_regex(r'

([^<]+)', webpage, 'subtitle', default=None) if subtitle: episode = subtitle - ep_summary = self._search_regex( - r']+class="ep-summary"[^>]*>(.+?)', webpage, - 'episode summary', fatal=False, flags=re.DOTALL) + ep_summary = get_element_html_by_class('ep-summary', webpage) description = self._html_search_regex( r'

(.+?)

', ep_summary, @@ -185,6 +182,6 @@ class TV5MondePlusIE(InfoExtractor): 'formats': formats, 'subtitles': self._extract_subtitles(self._parse_json( traverse_obj(vpl_data, ('data-captions', {str}), default='{}'), display_id, fatal=False)), - 'series': series, + 'series': self._html_search_regex(r'

([^<]+)', webpage, 'title', default=None), 'episode': episode, } From 24e9dab5861b49779568916ec854ae393afa5160 Mon Sep 17 00:00:00 2001 From: Alard Date: Sat, 13 Apr 2024 22:56:39 +0200 Subject: [PATCH 5/6] [ie/TV5MondePlus] Use more get_element_by_class. --- yt_dlp/extractor/tv5mondeplus.py | 13 ++++++------- 1 file changed, 6 insertions(+), 7 deletions(-) diff --git a/yt_dlp/extractor/tv5mondeplus.py b/yt_dlp/extractor/tv5mondeplus.py index 18e750328..bab3b0eb4 100644 --- a/yt_dlp/extractor/tv5mondeplus.py +++ b/yt_dlp/extractor/tv5mondeplus.py @@ -3,8 +3,10 @@ import urllib.parse from .common import InfoExtractor from ..utils import ( + clean_html, determine_ext, extract_attributes, + get_element_by_class, get_element_html_by_class, int_or_none, parse_duration, @@ -148,16 +150,13 @@ class TV5MondePlusIE(InfoExtractor): duration = (int_or_none(try_get(metadata, lambda x: x['content']['duration'])) or parse_duration(self._html_search_meta('duration', webpage))) - title = episode = self._html_search_regex(r'

([^<]+)', webpage, 'title', default=None) - subtitle = self._html_search_regex(r'

([^<]+)', webpage, 'subtitle', default=None) + title = episode = clean_html(get_element_by_class('main-title', webpage)) + subtitle = clean_html(get_element_by_class('video-subtitle', webpage)) if subtitle: episode = subtitle ep_summary = get_element_html_by_class('ep-summary', webpage) - - description = self._html_search_regex( - r'

(.+?)

', ep_summary, - 'description', fatal=False, flags=re.DOTALL) + description = clean_html(get_element_by_class('text', ep_summary)) upload_date = self._search_regex( r'(?:date_publication|publish_date)["\']\s*:\s*["\'](\d{4}_\d{2}_\d{2})', @@ -182,6 +181,6 @@ class TV5MondePlusIE(InfoExtractor): 'formats': formats, 'subtitles': self._extract_subtitles(self._parse_json( traverse_obj(vpl_data, ('data-captions', {str}), default='{}'), display_id, fatal=False)), - 'series': self._html_search_regex(r'

([^<]+)', webpage, 'title', default=None), + 'series': clean_html(get_element_by_class('video-title', webpage)), 'episode': episode, } From 8e6ff6f5f0bbbaabc65fd339bde377fa5c7a84d5 Mon Sep 17 00:00:00 2001 From: Alard Date: Sat, 13 Apr 2024 23:07:15 +0200 Subject: [PATCH 6/6] [ie/TV5MondePlus] Remove unused import re. --- yt_dlp/extractor/tv5mondeplus.py | 1 - 1 file changed, 1 deletion(-) diff --git a/yt_dlp/extractor/tv5mondeplus.py b/yt_dlp/extractor/tv5mondeplus.py index bab3b0eb4..99031d750 100644 --- a/yt_dlp/extractor/tv5mondeplus.py +++ b/yt_dlp/extractor/tv5mondeplus.py @@ -1,4 +1,3 @@ -import re import urllib.parse from .common import InfoExtractor