perl regex for a string

Question

I have a string from where I need to extract street , city , state , zip .

The string may look like

    a)$str1  ="2500 South 3850 West Suite A Salt Lake City, UT 84120-7225";
    b)$str2 ="19701 DaVinci  Lake Forest, CA 92610";
    c)$str3="abc ,def ,ca 1234"; -->(note there are two commas in this one)

I am currently using split function to get an array and taking those values to do my work but I want a single regex to do the task for me.I would appreciate any help. Thanks

Please show the code you are using now. That would help a lot. The problem as you have defined it looks ambiguous. For example, how do you know the city name is "Salt Lake City" rather than "West Suite A Salt Lake City", "Lake City", or "City"? — dan1111, Nov 13 '12 at 12:08
The format is not fixed, regex is not likely to be the solution. — neevek, Nov 13 '12 at 12:08
Hm. Address and name parsing is an extremely complex field in the sense that the approach is basically brute force. There are companies that make a fair living doing this stuff. — Leonardo Herrera, Nov 13 '12 at 12:54

score 2 · Answer 1 · answered Nov 13 '12 at 12:08

Locations in general are far from simple and I am very confident that, even when you find one that works for your test cases, you will most likely stumple upon a case where your regex doesn't match correctly.

Addresses are extremely difficult to parse, because you can never be completely sure, what string will match to what part.

There are cities with numbers. Streets with special characters. The only thing close to regular is the zipcode, you could extract that fairly safely, but only if it is always at the end of your string.

Other than that, you will have to match the string against some kind of address database to know wether the city/street you extracted even exists.

Yes Florian thanks for the insight. I think that is more appropriate. — jnanchak, Nov 14 '12 at 16:25

score 2 · Answer 2 · answered Nov 13 '12 at 15:40

I agree w/Florian: as soon as you get a regex to work, you'll introduce a new string that it won't work on, and your prog will bork.

Having said that, here is a regex that will work on your example:

#!/usr/bin/perl
use strict;
use warnings;

my $addr1 = "2500 South 3850 West Suite A Salt Lake City, UT 84120-7225";
my $addr2 = "19701 DaVinci  Lake Forest, CA 92610";
my $addr3 = "abc ,def ,ca 1234";

for my $addr($addr1,$addr2,$addr3){
  $addr =~ m/^(.*),[ \t]*([a-zA-Z]{2})[ \t]([0-9\-]*)$/;
  print "\naddr : '$1'\n";
  print "state: '$2'\n";
  print "zip  : '$3'\n";
}

It cannot parse the City, though, as the commas in your addresses are inconsistent.

`[ \t]*` can be replaced with simply `\s*`. – dan1111 Nov 13 '12 at 15:45 — dan1111, Nov 13 '12 at 15:45

perl regex for a string

2 Answers2