使用正则表达式进行部分匹配

Question

I'm building a system that recognizes pattern-based words such as license plates and fiscal codes. My input strings are some times not perfect since deriving from an OCR system, so I'm looking for a solution that "relaxes" my regular expression.

Eg of regex :

^(([a-zA-Z]{2}\d{3}[a-zA-Z]{2})|(([a-zA-Z]{2}|roma)(\d{5}|\d{6})))$

If I have a (italian) license plate AB123CD and I try to match my regex it works but if my input is slightly corruputed eg ABi23CD (the OCR system reads the '1' as an 'i') the regex doens't obviously match.

Is there any way to allow some errors in regex matching? In the example, the second string would match the regex substituting the i with a number so it match but allowing one error. Thanks!

Answer 1

OCR errors are hard to predict, but if you analyze the output you have, you may find regularities, like i recognized instead of 1 , etc.

In that case, you can use character classes: \\d matches a digit, [i\\d] matches i or a digit (but in your case, i is a digit, too, you can replace i with 1 later).

So, your pattern will look like

^([a-zA-Z]{2}[\di]{3}[a-zA-Z]{2}|([a-zA-Z]{2}|roma)[\di]{5,6})$

Note (\\d{5}|\\d{6}) can be shortened as \\d{5,6} .

You can add more characters to the [\\di] once you find them.