[utils,franceinter] Add french months' names and fix extraction

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.
This commit is contained in:
renalid 2016-09-02 18:31:52 +02:00 committed by Sergey M․
parent 961516bfd1
commit a942d6cb48
No known key found for this signature in database
GPG Key ID: 2C393E0F18A9236D
2 changed files with 32 additions and 19 deletions

View File

@ -2,20 +2,24 @@
from __future__ import unicode_literals from __future__ import unicode_literals
from .common import InfoExtractor from .common import InfoExtractor
from ..utils import int_or_none from ..utils import (
unified_timestamp,
month_by_name,
)
class FranceInterIE(InfoExtractor): class FranceInterIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?franceinter\.fr/player/reecouter\?play=(?P<id>[0-9]+)' _VALID_URL = r'https?://(?:www\.)?franceinter\.fr/emissions/(?P<id>[^?#]+)'
_TEST = { _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', 'md5': '4764932e466e6f6c79c317d2e74f6884',
'info_dict': { 'info_dict': {
'id': '793962', 'id': 'la-marche-de-l-histoire/la-marche-de-l-histoire-18-decembre-2013',
'ext': 'mp3', 'ext': 'mp3',
'title': 'LHistoire dans les jeux vidéo', 'title': 'LHistoire dans les jeux vidéo du 18 décembre 2013 - France Inter',
'description': 'md5:7e93ddb4451e7530022792240a3049c7', 'description': 'LHistoire 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': 1387369800, 'timestamp': 1387324800,
'upload_date': '20131218', 'upload_date': '20131218',
}, },
} }
@ -25,17 +29,17 @@ def _real_extract(self, url):
webpage = self._download_webpage(url, video_id) webpage = self._download_webpage(url, video_id)
path = self._search_regex( video_url = self._search_regex(
r'<a id="player".+?href="([^"]+)"', webpage, 'video url') r'<button class="replay-button playable" data-is-aod="1" data-url="([^"]+)"', webpage, 'video url')
video_url = 'http://www.franceinter.fr/' + path
title = self._html_search_regex( title = self._og_search_title(webpage)
r'<span class="title-diffusion">(.+?)</span>', webpage, 'title') description = self._og_search_description(webpage)
description = self._html_search_regex(
r'<span class="description">(.*?)</span>', extractdate = self._search_regex('(\d{2}-([a-zA-Z\s]+)-\d{4}$)', url, 'extractdate', fatal=False)
webpage, 'description', fatal=False) extractdate = extractdate.split('-')
timestamp = int_or_none(self._search_regex( extractdate = extractdate[2] + "," + str(month_by_name(extractdate[1], 'fr')) + "," + extractdate[0]
r'data-date="(\d+)"', webpage, 'upload date', fatal=False))
timestamp = unified_timestamp(extractdate)
return { return {
'id': video_id, 'id': video_id,

View File

@ -91,6 +91,10 @@ def register_socks_protocols():
'January', 'February', 'March', 'April', 'May', 'June', 'January', 'February', 'March', 'April', 'May', 'June',
'July', 'August', 'September', 'October', 'November', 'December'] 'July', 'August', 'September', 'October', 'November', 'December']
FRENCH_MONTH_NAMES = [
'janvier', 'fevrier', 'mars', 'avril', 'mai', 'juin',
'juillet', 'aout', 'septembre', 'octobre', 'novembre', 'decembre']
KNOWN_EXTENSIONS = ( KNOWN_EXTENSIONS = (
'mp4', 'm4a', 'm4p', 'm4b', 'm4r', 'm4v', 'aac', 'mp4', 'm4a', 'm4p', 'm4b', 'm4r', 'm4v', 'aac',
'flv', 'f4v', 'f4a', 'f4b', 'flv', 'f4v', 'f4a', 'f4b',
@ -1587,11 +1591,16 @@ def parse_count(s):
return lookup_unit_table(_UNIT_TABLE, 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 """ """ Return the number of a month by (locale-independently) English name """
name_list = ENGLISH_MONTH_NAMES
if lang == 'fr':
name_list = FRENCH_MONTH_NAMES
try: try:
return ENGLISH_MONTH_NAMES.index(name) + 1 return name_list.index(name) + 1
except ValueError: except ValueError:
return None return None