[fix] yahoo news date parsing

This commit is contained in:
Adam Tauber 2017-04-08 19:42:50 +02:00
parent 12d91c1d67
commit 46a2c63f8e
1 changed files with 11 additions and 8 deletions

View File

@ -80,16 +80,19 @@ def response(resp):
# still useful ? # still useful ?
if re.match("^[0-9]+ minute(s|) ago$", publishedDate): if re.match("^[0-9]+ minute(s|) ago$", publishedDate):
publishedDate = datetime.now() - timedelta(minutes=int(re.match(r'\d+', publishedDate).group())) # noqa publishedDate = datetime.now() - timedelta(minutes=int(re.match(r'\d+', publishedDate).group()))
elif re.match("^[0-9]+ days? ago$", publishedDate):
publishedDate = datetime.now() - timedelta(days=int(re.match(r'\d+', publishedDate).group()))
elif re.match("^[0-9]+ hour(s|), [0-9]+ minute(s|) ago$", publishedDate):
timeNumbers = re.findall(r'\d+', publishedDate)
publishedDate = datetime.now()\
- timedelta(hours=int(timeNumbers[0]))\
- timedelta(minutes=int(timeNumbers[1]))
else: else:
if re.match("^[0-9]+ hour(s|), [0-9]+ minute(s|) ago$", try:
publishedDate):
timeNumbers = re.findall(r'\d+', publishedDate)
publishedDate = datetime.now()\
- timedelta(hours=int(timeNumbers[0]))\
- timedelta(minutes=int(timeNumbers[1]))
else:
publishedDate = parser.parse(publishedDate) publishedDate = parser.parse(publishedDate)
except:
publishedDate = datetime.now()
if publishedDate.year == 1900: if publishedDate.year == 1900:
publishedDate = publishedDate.replace(year=datetime.now().year) publishedDate = publishedDate.replace(year=datetime.now().year)