Thread: Java String Regular Expression Help

Hi All,

I have wriittn below regex for two lines.

String LN1Pattern = "^((?=.{1,35}$)(/([A-Z]{1,5})(|(//[a-zA-Z0-9]+))))$";
System.out.println("/ABC//FGhiJkl012345".matches(LN1Pattern));
 
String LN2Pattern = "^(|((\\s+(//[a-zA-Z0-9]{1,33})){1,2}))$";
System.out.println("".matches(LN2Pattern));

\\s+ is a newline character.

But when I combines both as below, its not giving me expected result.

^(((?=.{1,35}$)(/([A-Z]{1,5})(|(//[a-zA-Z0-9]+))))(|((\\s+(//[a-zA-Z0-9]{1,33})){1,2})))$

For string "/ABC//FGhiJkl012345\n//abCD01EF02" - returns False. Expected is True
I think there is some problem in lookahead placed.

Please help.


Thanks
Raj

Originally Posted by Raj4Smile

I think there is some problem in lookahead placed.

You're right. The problem is in the positive lookahead: (?=.{1,35}$). Quoting from Pattern (Java Platform SE 7 ), "the regular expression . matches any character except a line terminator unless the DOTALL flag is specified," and your test string has exactly that character - \n. To fix this all you need to do is to replace the regex fragment above with

(?=.|\n{1,35}$)

A few comments about your regex:

1. There appears to be an excessive use of capturing groups, i.e., (regex). Are you using them to visually separate parts of your regex? If so...

2. Your regex will be far more readable and maintainable if you break up the regex string and comment each logical part, e.g.,

String LN1Pattern =
    "^(?=.|\n{1,35}$)" +    // Positive lookahead - to match 1-35 characters, including \n, but not other line terminators
    "/[A-Z]{1,5}" +         // Match / and 1-5 characters in the character class 
    "(|//[a-zA-Z0-9]+)$";   // Optionally match // and 1 or more characters in the character class

This can also help spot mistakes by tallying the regex with the associated comment.

3. You appear to use (|regex) to denote optional match. It will be better, imo, to use the optional quantifier (?), e.g., (?:regex)?. In this case I used a non-capturing group, i.e., (?:regex) to convey the message that I'm grouping the regex for the quantifier, and not for capturing for later use.

Thanks Jashburn, but I am still struggling after using (?=.|\n{1,35}$). I am much newer to regex patterns.

I have groups of variable length and total length after combining all groups should not exceed given length say 35.

What I am trying to implement in regex is SWIFT MT103 Field rules. If possible and you get time please have a look at Field 72.

Now I am detailing my requirement here:
Pattern specified for the field value is as below
It can contain either 1 or two lines. Format for two lines are as below

First Line
--------------------
First Line will contain three groups which are optional to each other. Total length of first line should not exceed 35

First group will contain Uppercase alphabate value, length for which may vary from 1 to 5
Second group will contain Uppercase alphanumeric value. Length should not exceed 5.
Third group will contain characters say SWIFT characters.

Second Line
-----------------------
Third group will contain characters say SWIFT characters. Total Length should not exceed 35 characters.



Can you please help me here.



Thanks
Raj

Sure, I can help you with this, but I'll need some help from you too. The requirements you posted are not sufficiently clear, so I'll need a number of test cases - examples of valid field 72 values that exercise the field format per requirement. Also throw in some invalid field 72 values.

The way I develop regex is I always use TDD, where I'd write the unit test cases for a representative set of positive and negative cases. Only then I'd write the regex to pass my test cases.

I did try looking into MT103 field/tag 72, but the info I found isn't good enough. E.g., the info at SWIFTStandards - Category 1 - MT 103 - 22. Field 72: Sender to Receiver Information and SWIFTStandards - Category 1 - MT 103 Examples are not sufficient for me to develop my test cases.

Btw, I found an open source (LGPLv2) library, WIFE - open source library | Free Development software downloads at SourceForge.net, that might be applicable for your purposes. I'm not suggesting that you should use the library (as I like a good regex challenge), but I'd not be fulfilling my duty if I don't mention it...

Hi, I am back from long vacation and will be detailing this requirement shortly.


Thanks
Raj