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

fix parsing of duration

Browse source
This commit is contained in:
Romain Bignon 2010-11-17 21:30:02 +01:00
commit 5d923bd935
1 changed files with 8 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

11
weboob/backends/ina/pages/video.py
View file

@ -25,6 +25,7 @@ except ImportError:
from cgi import parse_qs from cgi import parse_qs
from weboob.tools.browser import BasePage from weboob.tools.browser import BasePage
from weboob.tools.parsers.lxmlparser import SelectElementException
from ..video import InaVideo from ..video import InaVideo
@ -60,17 +61,21 @@ class VideoPage(BasePage):
return el.text.strip() return el.text.strip()
def get_date_and_duration(self): def get_date_and_duration(self):
duration_regexp = re.compile('(.+) - (.+)min(.+)s') duration_regexp = re.compile('(.+) - ((.+)h)?((.+)min)?(.+)s')
el = self.document.getroot().cssselect('div.bloc-produit-haut p.date')[0] el = self.document.getroot().cssselect('div.bloc-produit-haut p.date')[0]
if el is not None: if el is not None:
m = duration_regexp.match(el.text.strip()) m = duration_regexp.match(el.text.strip())
if m: if m:
day, month, year = [int(s) for s in m.group(1).split('/')] day, month, year = [int(s) for s in m.group(1).split('/')]
date = datetime.datetime(year, month, day) date = datetime.datetime(year, month, day)
duration = datetime.timedelta(minutes=int(m.group(3)), seconds=int(m.group(2))) duration = datetime.timedelta(hours=int(m.group(3) if m.group(3) is not None else 0),
minutes=int(m.group(5) if m.group(5) is not None else 0),
seconds=int(m.group(6)))
return date, duration return date, duration
else:
raise SelectElementException('Unable to parse date and duration')
else: else:
return None raise SelectElementException('Unable to find date and duration element')
def get_title(self): def get_title(self):
el = self.document.getroot().cssselect('div.bloc-produit-haut h1')[0] el = self.document.getroot().cssselect('div.bloc-produit-haut h1')[0]