Question

I need help sorting out a solution for a pattern generator.

I have an application that uses exchange formats with different patterns
like

00350-ABA-0NZ0:AXYA-11/11/2012 etc.,

that have numeric and alphanumeric data separated by ‘-‘,’.’,”:” and ‘/’. Now what I want to do is convert this to a generic format like
nnnnn-ccc-nccn:cccc-nn/nn/nnnn where n is a digit and c is a character.

Any help/suggestions/ideas . . .
Thanks
CSK.

Answer 1

You can’t do conditional replaces in a single regex. You need to do it in two steps (here’s a Python example):

>>> s = "00350-ABA-0NZ0:AXYA-11/11/2012"
>>> s = re.sub(r"[A-Za-z]", "c", s)
>>> s
'00350-ccc-0cc0:cccc-11/11/2012'
>>> s = re.sub(r"\d", "n", s)
>>> s
'nnnnn-ccc-nccn:cccc-nn/nn/nnnn'

And you need to do it this way around – I just saw your solution in your comment, and if you look at it again, you’ll see it won’t work. Hint: You’ll get 'ccccc-ccc-cccc:cccc-cc/cc/cccc' as a result…

Another solution would be to use a callback function that examines the match and chooses the replacement string accordingly. But that’s not pure regex anymore:

>>> def replace(m):
...     return "n" if m.group(0).isdigit() else "c"
...
>>> s = re.sub(r"[A-Za-z0-9]", replace, s)
>>> s
'nnnnn-ccc-nccn:cccc-nn/nn/nnnn'