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

ina: Try to handle more videos

Browse source 
Also, the URLs are now closer to what a real browser loads. It still
does not work though.
This commit is contained in:
Laurent Bachelier 2013-01-09 18:18:36 +01:00 • committed by Romain Bignon
commit abe3f3c4a0
1 changed files with 9 additions and 6 deletions
Download patch file Download diff file Expand all files Collapse all files

15
modules/ina/pages/video.py
View file

@ -58,15 +58,16 @@ class BaseVideoPage(BasePage):
def get_url(self): def get_url(self):
qs = parse_qs(self.document.getroot().cssselect('param[name="flashvars"]')[0].attrib['value']) qs = parse_qs(self.document.getroot().cssselect('param[name="flashvars"]')[0].attrib['value'])
s = self.browser.readurl('http://boutique.ina.fr/player/infovideo/id_notice/%s' % qs['id_notice'][0]) s = self.browser.readurl('http://www.ina.fr/player/infovideo/id_notice/%s/module_request/%s' % (qs['id_notice'][0], qs['module'][0]))
s = s[s.find('<Media>')+7:s.find('</Media>')] s = s[s.find('<Media>')+7:s.find('</Media>')]
return u'%s/pkey/%s' % (s, qs['pkey'][0]) return u'%s/id_chaine/%s/module_request/%s/pkey/%s' % \
(s, qs['id_chaine'][0], qs['module'][0], qs['pkey'][0])
def parse_date_and_duration(self, text): def parse_date_and_duration(self, text):
duration_regexp = re.compile('(.* - )?(.+) - ((.+)h)?((.+)min)?(.+)s') duration_regexp = re.compile('(.* - )?(.+) - ((.+)h)?((.+)min)?(.+)s')
m = duration_regexp.match(text) m = duration_regexp.match(text)
if m: if m:
day, month, year = [int(s) for s in m.group(2).split('/')] day, month, year = [abs(int(s)) for s in m.group(2).split('/')]
date = datetime.datetime(year, month, day) date = datetime.datetime(year, month, day)
duration = datetime.timedelta(hours=int(m.group(4) if m.group(4) is not None else 0), duration = datetime.timedelta(hours=int(m.group(4) if m.group(4) is not None else 0),
minutes=int(m.group(6) if m.group(6) is not None else 0), minutes=int(m.group(6) if m.group(6) is not None else 0),
@ -99,11 +100,13 @@ class VideoPage(BaseVideoPage):
return self.parse_date_and_duration(qr.find('h2').tail.strip()) return self.parse_date_and_duration(qr.find('h2').tail.strip())
def get_title(self): def get_title(self):
qr = self.parser.select(self.document.getroot(), 'div.container-global-qr')[0].find('div').findall('div')[1] qr = self.parser.select(self.document.getroot(), 'div.container-global-qr')[0]
return unicode(qr.find('h2').text.strip()) return unicode(qr.cssselect('h2.titre-propre')[0].text.strip())
def get_description(self): def get_description(self):
return unicode(self.parser.select(self.document.getroot(), 'div.container-global-qr')[1].find('div').find('p').text.strip()) desc = self.parser.select(self.document.getroot(), 'div.container-global-qr')[1].find('div').find('p')
if desc:
return unicode(desc.text.strip())
class BoutiqueVideoPage(BaseVideoPage): class BoutiqueVideoPage(BaseVideoPage):