nojhan/weboob-devel
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

radiofrance: Support franceinfo archives

Browse source
This commit is contained in:
Laurent Bachelier 2012-10-03 17:39:12 +02:00 • committed by Romain Bignon
commit 9d5f808de9
1 changed files with 17 additions and 6 deletions
Download patch file Download diff file Expand all files Collapse all files

17
modules/radiofrance/browser.py
View file

@ -36,7 +36,7 @@ __all__ = ['RadioFranceBrowser', 'RadioFranceVideo']
class RadioFranceVideo(BaseVideo): class RadioFranceVideo(BaseVideo):
RADIO_DOMAINS = ('franceinter', 'franceculture', 'fipradio', ) RADIO_DOMAINS = ('franceinter', 'franceculture', 'fipradio', 'franceinfo')
def __init__(self, *args, **kwargs): def __init__(self, *args, **kwargs):
BaseVideo.__init__(self, *args, **kwargs) BaseVideo.__init__(self, *args, **kwargs)
@ -65,6 +65,9 @@ class PlayerPage(BasePage):
class ReplayPage(BasePage): class ReplayPage(BasePage):
URL = r'^http://www\.(?P<rdomain>%s)\.fr/(?:emission|diffusion)-.+$' \ URL = r'^http://www\.(?P<rdomain>%s)\.fr/(?:emission|diffusion)-.+$' \
% '|'.join(RadioFranceVideo.RADIO_DOMAINS) % '|'.join(RadioFranceVideo.RADIO_DOMAINS)
# the url does not always end with id-yyy-mm-dd, sometimes no mm or dd
URL2 = r'^http://www\.(?P<rdomain>%s)\.fr/[a-z\-]+/[0-9a-z\-]+/[0-9a-z\-]+-[0-9\-]+' \
% 'franceinfo'
def get_id(self): def get_id(self):
radio_domain = self.groups[0] radio_domain = self.groups[0]
@ -82,6 +85,13 @@ class ReplayPage(BasePage):
if match: if match:
player_id = match.groups()[0] player_id = match.groups()[0]
return (radio_domain, player_id) return (radio_domain, player_id)
# at least for franceinfo
for node in self.parser.select(self.document.getroot(), '#article .emission-player a.play'):
match = re.match(r'^song-(\d+)$', node.attrib.get('rel', ''))
if match:
player_id = match.groups()[0]
return (radio_domain, player_id)
class DataPage(BasePage): class DataPage(BasePage):
@ -123,7 +133,8 @@ class RadioFranceBrowser(BaseBrowser):
PAGES = {r'/playerjs/direct/donneesassociees/html\?guid=$': DataPage, PAGES = {r'/playerjs/direct/donneesassociees/html\?guid=$': DataPage,
r'http://players.tv-radio.com/radiofrance/metadatas/([a-z]+)RSS.html': RssPage, r'http://players.tv-radio.com/radiofrance/metadatas/([a-z]+)RSS.html': RssPage,
PlayerPage.URL: PlayerPage, PlayerPage.URL: PlayerPage,
ReplayPage.URL: ReplayPage} ReplayPage.URL: ReplayPage,
ReplayPage.URL2: ReplayPage}
def id2domain(self, _id): def id2domain(self, _id):
""" """
@ -179,7 +190,7 @@ class RadioFranceBrowser(BaseBrowser):
match = re.match(PlayerPage.URL, url) match = re.match(PlayerPage.URL, url)
if match: if match:
radio_domain, replay_id = match.groups() radio_domain, replay_id = match.groups()
elif re.match(ReplayPage.URL, url): elif re.match(ReplayPage.URL, url) or re.match(ReplayPage.URL2, url):
self.location(url) self.location(url)
assert self.is_on_page(ReplayPage) assert self.is_on_page(ReplayPage)
radio_domain, replay_id = self.page.get_id() radio_domain, replay_id = self.page.get_id()