Regular expression as dictionary key?

[Daniel Yoo]

Luke <LLoeffler at home.com> wrote:
: Let's say you have a dict of words like
: words = {"dogs":1,"cats":2}

: Let's say I'm given a singular form of a word and want to match it with 
: plurals still, but words["dog"] will obviously raise a KeyError.

: Without doing a linear substring search on every element in words.key(), 
: is there a way to take advantage of the dict's binary tree layout 
: properties (e.g. speed) like:

: words["dog*"]   # where dog* is a Regex

: Nonexistant syntax, I know, but you get the idea...  I smell a PEP

Hmmm... Interesting!  Let's see:


###
import re

from UserDict import UserDict
from types import StringType

class RegexDict(UserDict):
    """A dictionary that supports regular expressions to get at
    a key.  Response to a post on c.l.py."""
    def __getitem__(self, key):
        if self.data.has_key(key): return self.data[key]
        regex = key
        if isinstance(key, StringType):
            regex = re.compile(key)
        for k in self.data.keys():
            if regex.search(k):
                return self.data[k]
        raise KeyError, key
###


Let's see if this works:

###
>>> words = RegexDict({'dogs' : 1, "cats" : 2})
>>> words['cat']                    ## Normal lookups work
2
>>> words['dog*']                   ## And so do regular expressions!
1
>>> words['dog*r']                  ## But we can still get KeyErrors.
Traceback (most recent call last):
  File "<stdin>", line 1, in ?
  File "/tmp/python-1176RQL", line 17, in __getitem__
KeyError: dog*r
###


Hope this helps!