Strange behaviour while replacing string in python

Question

I have a date given in format "October 28, 2010" (or simillar). Now I want to change the full name of month to shorter version (in this case Oct). For this I prepared a dictionary:

_mapping = {'January': 'Jan', 
            'February': 'Feb', 
            'March': 'Mar', 
            'April': 'Apr', 
            'May': 'May', 
            'June': 'Jun', 
            'July': 'Jul', 
            'August': 'Aug', 
            'September': 'Sep', 
            'October': 'Oct', 
            'November': 'Nov', 
            'December': 'Dec'}

and in the method where the substition goes I wrote following:

def fetch(self, pno):
    ...
    date = #get data (working fine)
    for l, s in self._mapping.iteritems():
        pubdate = date.replace(l, s)
        print l + " -> " + pubdate #this is only for debug
    (pubd_month, self.pubd_day, self.pubd_year) = pubdate.split(' ')
    print pubd_month, self.pubd_day, self.pubd_year
    print pubdate

The result of execution is:

February -> October 28, 2008
October -> Oct 28, 2008
January -> October 28, 2008
April -> October 28, 2008
November -> October 28, 2008
March -> October 28, 2008
August -> October 28, 2008
May -> October 28, 2008
December -> October 28, 2008
June -> October 28, 2008
September -> October 28, 2008
July -> October 28, 2008
October
October 28, 2008

As you can see it seems that the replacement goes ok when it finds October but outside the loop I get again full month name. What do I do wrong?

Another question: is there shorter way to do this?

score 3 · Answer 1 · answered Sep 21 '13 at 13:06

Because you are always replacing pubdate with a replacement on date, you overwrite any successful matching.

Either test first if the month is in the string before replacing, or replace pubdate with pubdate.replace() results:

pubdate = #get data (working fine)
for l, s in self._mapping.iteritems():
    pubdate = pubdate.replace(l, s)

or, using testing:

date = #get data (working fine)
for l, s in self._mapping.iteritems():
    if l in date:
        pubdate = date.replace(l, s)

The latter still will result in pubdate being overwritten if more than one month name is found. Alternatively, if there is only ever going to be one month match, use break to end the loop early:

date = #get data (working fine)
for l, s in self._mapping.iteritems():
    if l in date:
        pubdate = date.replace(l, s)
        break

Caveat: if no match is ever found, pubdate will never have been assigned to and you'll get a UnboundLocal exception.

Huh, silly mistake of mine. Thanks :) However though in this specific situation ubutbu's solution is cleaner so I'll mark his as accepted. — Tomasz Kapłoński, Sep 21 '13 at 13:16

score 3 · Accepted Answer · answered Sep 21 '13 at 13:10

3

There is no need for _mapping. If you use datetime.datetime objects, the strftime method can return the abbreviated month for you:

import datetime as DT

def short_date(date):
    date = DT.datetime.strptime(date, '%B %d, %Y')
    return date.strftime('%b %d, %Y')

print(short_date('October 28, 2010'))

prints

Oct 28, 2010

If your date strings come in a variety of formats then instead of parsing them with strptime(date, '%B %d, %Y') you could use dateutil's parser.parse method. For greatest flexibility, you'll probably be best served parsing the date strings into datetime.datetime objects as early as possible, and using strftime to format them as late as possible.

answered Sep 21 '13 at 13:10

unutbu

842,883
184
1,785
1,677

Note that the `%b` and `%B` month names are locale dependent. I've found that if your code has to run on different environments where you do not control the locale, you have to do manual month replacements *anyway*. – Martijn Pieters Sep 21 '13 at 13:17
Is it possible to force using english locale then? – Tomasz Kapłoński Sep 21 '13 at 13:20
Not really; it is far easier to just replace the strings, to be honest. See [Locale date formatting in Python](http://stackoverflow.com/q/985505) for a solution that *may* work on POSIX systems, provided the locale has been generated. – Martijn Pieters Sep 21 '13 at 13:21

Strange behaviour while replacing string in python

2 Answers2