Subliminal update

This commit is contained in:
Ruud committed 2012-04-08 17:35:39 +02:00
1 parent 5321037f08
commit 148e1940ac
85 files changed
+4375 -3420

No files matched your search

+31 -28
View File
@@ -21,6 +21,7 @@
import datetime
import re
def search_year(string):
"""Looks for year patterns, and if found return the year and group span.
Assumes there are sentinels at the beginning and end of the string that
@@ -62,34 +63,35 @@ def search_date(string):
dsep = r'[-/ \.]'
date_rexps = [ # 20010823
r'[^0-9]' +
r'(?P<year>[0-9]{4})' +
r'(?P<month>[0-9]{2})' +
r'(?P<day>[0-9]{2})' +
r'[^0-9]',
date_rexps = [
# 20010823
r'[^0-9]' +
r'(?P<year>[0-9]{4})' +
r'(?P<month>[0-9]{2})' +
r'(?P<day>[0-9]{2})' +
r'[^0-9]',
# 2001-08-23
r'[^0-9]' +
r'(?P<year>[0-9]{4})' + dsep +
r'(?P<month>[0-9]{2})' + dsep +
r'(?P<day>[0-9]{2})' +
r'[^0-9]',
# 2001-08-23
r'[^0-9]' +
r'(?P<year>[0-9]{4})' + dsep +
r'(?P<month>[0-9]{2})' + dsep +
r'(?P<day>[0-9]{2})' +
r'[^0-9]',
# 23-08-2001
r'[^0-9]' +
r'(?P<day>[0-9]{2})' + dsep +
r'(?P<month>[0-9]{2})' + dsep +
r'(?P<year>[0-9]{4})' +
r'[^0-9]',
# 23-08-2001
r'[^0-9]' +
r'(?P<day>[0-9]{2})' + dsep +
r'(?P<month>[0-9]{2})' + dsep +
r'(?P<year>[0-9]{4})' +
r'[^0-9]',
# 23-08-01
r'[^0-9]' +
r'(?P<day>[0-9]{2})' + dsep +
r'(?P<month>[0-9]{2})' + dsep +
r'(?P<year>[0-9]{2})' +
r'[^0-9]',
]
# 23-08-01
r'[^0-9]' +
r'(?P<day>[0-9]{2})' + dsep +
r'(?P<month>[0-9]{2})' + dsep +
r'(?P<year>[0-9]{2})' +
r'[^0-9]',
]
for drexp in date_rexps:
match = re.search(drexp, string)
@@ -98,7 +100,7 @@ def search_date(string):
year, month, day = int(d['year']), int(d['month']), int(d['day'])
# years specified as 2 digits should be adjusted here
if year < 100:
if year > (datetime.date.today().year % 100)+ 5:
if year > (datetime.date.today().year % 100) + 5:
year = 1900 + year
else:
year = 2000 + year
@@ -120,8 +122,9 @@ def search_date(string):
continue
# looks like we have a valid date
# note: span is [+1,-1] because we don't want to include the non-digit char
# note: span is [+1,-1] because we don't want to include the
# non-digit char
start, end = match.span()
return (date, (start+1, end-1))
return (date, (start + 1, end - 1))
return None, None