From 5db1a3a73f08a545eaf54fdae81e054b08f60ebd Mon Sep 17 00:00:00 2001 From: renalid Date: Fri, 2 Sep 2016 18:31:52 +0200 Subject: [PATCH 1/3] [Extractor Update] France Inter Update of the "FranceInter" radio extractor : webpages HTML structure had changed, the extractor didn't work. So I updated this extractor to get the mp3 URL and all details. --- youtube_dl/extractor/franceinter.py | 38 ++++++++++++++++------------- youtube_dl/utils.py | 13 ++++++++++ 2 files changed, 34 insertions(+), 17 deletions(-) diff --git a/youtube_dl/extractor/franceinter.py b/youtube_dl/extractor/franceinter.py index 2369f868d..ec57e3e67 100644 --- a/youtube_dl/extractor/franceinter.py +++ b/youtube_dl/extractor/franceinter.py @@ -2,20 +2,24 @@ from __future__ import unicode_literals from .common import InfoExtractor -from ..utils import int_or_none +from ..utils import ( + unified_timestamp, + month_by_french_name, +) class FranceInterIE(InfoExtractor): - _VALID_URL = r'https?://(?:www\.)?franceinter\.fr/player/reecouter\?play=(?P[0-9]+)' + _VALID_URL = r'https?://(?:www\.)?franceinter\.fr/emissions/(?P[^?#]+)' + _TEST = { - 'url': 'http://www.franceinter.fr/player/reecouter?play=793962', + 'url': 'https://www.franceinter.fr/emissions/la-marche-de-l-histoire/la-marche-de-l-histoire-18-decembre-2013', 'md5': '4764932e466e6f6c79c317d2e74f6884', 'info_dict': { - 'id': '793962', + 'id': 'la-marche-de-l-histoire/la-marche-de-l-histoire-18-decembre-2013', 'ext': 'mp3', - 'title': 'L’Histoire dans les jeux vidéo', - 'description': 'md5:7e93ddb4451e7530022792240a3049c7', - 'timestamp': 1387369800, + 'title': 'L’Histoire dans les jeux vidéo du 18 décembre 2013 - France Inter', + 'description': 'L’Histoire dans les jeux vidéo du 18 décembre 2013 par Jean Lebrun en replay sur France Inter. Retrouvez l\'émission en réécoute gratuite et abonnez-vous au podcast !', + 'timestamp': 1387324800, 'upload_date': '20131218', }, } @@ -25,17 +29,17 @@ class FranceInterIE(InfoExtractor): webpage = self._download_webpage(url, video_id) - path = self._search_regex( - r'(.+?)', webpage, 'title') - description = self._html_search_regex( - r'(.*?)', - webpage, 'description', fatal=False) - timestamp = int_or_none(self._search_regex( - r'data-date="(\d+)"', webpage, 'upload date', fatal=False)) + title = self._og_search_title(webpage) + description = self._og_search_description(webpage) + + extractdate = self._search_regex('(\d{2}-([a-zA-Z\s]+)-\d{4}$)', url, 'extractdate', fatal=False) + extractdate = extractdate.split('-') + extractdate = extractdate[2] + "," + str(month_by_french_name(extractdate[1])) + "," + extractdate[0] + + timestamp = unified_timestamp(extractdate) return { 'id': video_id, diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py index 1091f17f3..847c3bc96 100644 --- a/youtube_dl/utils.py +++ b/youtube_dl/utils.py @@ -91,6 +91,10 @@ ENGLISH_MONTH_NAMES = [ 'January', 'February', 'March', 'April', 'May', 'June', 'July', 'August', 'September', 'October', 'November', 'December'] +FRENCH_MONTH_NAMES = [ + 'janvier', 'fevrier', 'mars', 'avril', 'mai', 'juin', + 'juillet', 'août', 'septembre', 'octobre', 'novembre', 'decembre'] + KNOWN_EXTENSIONS = ( 'mp4', 'm4a', 'm4p', 'm4b', 'm4r', 'm4v', 'aac', 'flv', 'f4v', 'f4a', 'f4b', @@ -1596,6 +1600,15 @@ def month_by_name(name): return None +def month_by_french_name(name): + """ Return the number of a month by French name """ + + try: + return FRENCH_MONTH_NAMES.index(name) + 1 + except ValueError: + return None + + def month_by_abbreviation(abbrev): """ Return the number of a month by (locale-independently) English abbreviations """ From d330e57c2c0d8fef2d8ac5257003faf1230366de Mon Sep 17 00:00:00 2001 From: renalid Date: Sun, 4 Sep 2016 00:00:44 +0200 Subject: [PATCH 2/3] FranceInter update : merge to similar functions on utils --- youtube_dl/extractor/franceinter.py | 4 ++-- youtube_dl/utils.py | 16 ++++++---------- 2 files changed, 8 insertions(+), 12 deletions(-) diff --git a/youtube_dl/extractor/franceinter.py b/youtube_dl/extractor/franceinter.py index ec57e3e67..fd2f18e25 100644 --- a/youtube_dl/extractor/franceinter.py +++ b/youtube_dl/extractor/franceinter.py @@ -4,7 +4,7 @@ from __future__ import unicode_literals from .common import InfoExtractor from ..utils import ( unified_timestamp, - month_by_french_name, + month_by_name, ) @@ -37,7 +37,7 @@ class FranceInterIE(InfoExtractor): extractdate = self._search_regex('(\d{2}-([a-zA-Z\s]+)-\d{4}$)', url, 'extractdate', fatal=False) extractdate = extractdate.split('-') - extractdate = extractdate[2] + "," + str(month_by_french_name(extractdate[1])) + "," + extractdate[0] + extractdate = extractdate[2] + "," + str(month_by_name(extractdate[1],'fr')) + "," + extractdate[0] timestamp = unified_timestamp(extractdate) diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py index 847c3bc96..a8f5f1876 100644 --- a/youtube_dl/utils.py +++ b/youtube_dl/utils.py @@ -93,7 +93,7 @@ ENGLISH_MONTH_NAMES = [ FRENCH_MONTH_NAMES = [ 'janvier', 'fevrier', 'mars', 'avril', 'mai', 'juin', - 'juillet', 'août', 'septembre', 'octobre', 'novembre', 'decembre'] + 'juillet', 'aout', 'septembre', 'octobre', 'novembre', 'decembre'] KNOWN_EXTENSIONS = ( 'mp4', 'm4a', 'm4p', 'm4b', 'm4r', 'm4v', 'aac', @@ -1591,20 +1591,16 @@ def parse_count(s): return lookup_unit_table(_UNIT_TABLE, s) -def month_by_name(name): +def month_by_name(name, lang='en'): """ Return the number of a month by (locale-independently) English name """ - try: - return ENGLISH_MONTH_NAMES.index(name) + 1 - except ValueError: - return None + name_list = ENGLISH_MONTH_NAMES - -def month_by_french_name(name): - """ Return the number of a month by French name """ + if lang == 'fr': + name_list = FRENCH_MONTH_NAMES try: - return FRENCH_MONTH_NAMES.index(name) + 1 + return name_list.index(name) + 1 except ValueError: return None From 7d61189a20c255839c4dabcedbc1ce8a5360abd1 Mon Sep 17 00:00:00 2001 From: renalid Date: Sun, 4 Sep 2016 00:05:12 +0200 Subject: [PATCH 3/3] Syntax correction --- youtube_dl/extractor/franceinter.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/youtube_dl/extractor/franceinter.py b/youtube_dl/extractor/franceinter.py index fd2f18e25..6dad8d712 100644 --- a/youtube_dl/extractor/franceinter.py +++ b/youtube_dl/extractor/franceinter.py @@ -37,7 +37,7 @@ class FranceInterIE(InfoExtractor): extractdate = self._search_regex('(\d{2}-([a-zA-Z\s]+)-\d{4}$)', url, 'extractdate', fatal=False) extractdate = extractdate.split('-') - extractdate = extractdate[2] + "," + str(month_by_name(extractdate[1],'fr')) + "," + extractdate[0] + extractdate = extractdate[2] + "," + str(month_by_name(extractdate[1], 'fr')) + "," + extractdate[0] timestamp = unified_timestamp(extractdate)