Regex Refinement

Question

I'm trying to parse addresses out of blocks of text, and have the following expression to do so:

/\d+\s(?:[sewnSEWN]\.?\s)?[\d\w]+\s(?:(?:[\d\w]+\s){0,3})?\w+\.?/

It will currently parse addresses such as:

300 E. Randolph St. Chicago, IL>> Returns 300 E. Randolph St.

5553 Bay Shore Drive>> Returns input

23 Joseph E Lowery Boulevard>> Returns input

513 Martin Luther King Jr Boulevard>> Returns input

This is exactly what I want. I was wondering, as this is the first expression I have ever written, if there was a way to shorten down the expression or refine it a little?

Answer 1

I don't know which implementation you are using, so translate this to relevant language when needed.

\w = [a-zA-Z0-9] so [\d\w] is same as [\w]

Note that (?(?:[\w]+\s){0,3})? is same as (?:[\w]+\s){0,3} because the expression inside is matched zero or more times.

You can also add in the \w+\s at the beginning to the above expression, and make it repeat from 1 to 4.

Here is a matching for your example, Knowing not much about your format, here is what I find odd.

/
 \d+ # 513
 \s
 (?:[sewnSEWN]\.?\s)? #
 (?:\w+\s){1,4} # Martin Luther King Jr 
 \w+ # Boulevard
 \.?
/

• The spaces are restricted to a single space \s is this that strict? Perhaps you want \s+
• If I understand you right, the portion after NSEW. directions are that there has to be atleast two words, and atmost 5 words separated by spaces. is this a correct interpretation?

• ayyp: I have it as single spaces because I assumed that's how standard addresses are broken up. The way I was trying to do it, the street name could contain at least one part (such as Randalph), and at most 4 different parts, and after the street name there would be the descripter, i.e. street, park, drive, etc. But they can be abbreviated, which is why I searched for the period after.
• Rahul Gopinath: @AndrewPeacock makes sense. If you are trying to match generic street names, what about the directions like NW, SE etc?
• ayyp: I hadn't thought about that... Also, when I change (?:(?:[\w]+\s){0,3})? to (?:[\w]+\s){0,3}?, the second doesn't return the same result as the first.
• Rahul Gopinath: @AndrewPeacock remove the ? in your second expression.