From: Robert Klemme <shortcutter@...>
Date: 2008-06-18T21:20:38+09:00
Subject: Re: Regular Expressions

2008/6/18 Dave Bass <davebass@musician.org>:
> Robert Klemme wrote:
>> The easiest (but not most efficient approach) would be to create on
>> alternative for each variant you have, like
>>
>> %r{
>>    ^(?:
>>      v(?:\.\d+)+-\d+
>>    | \w+\d+-[\d_]+
>>    | ...
>>    )$
>> }x
>>
>> etc.
>
> The problem with regular expressions is that they can easily get out of
> hand and become incomprehensible, as the above code shows (though
> presumably to RK it's totally transparent).

Actually the RX I presented was not complete and was intended to
convey your point. :-)

> Better to write a number of small regexps, each testing for a specific
> pattern. Then combine the results with a logical OR. This can be done
> using a flag variable, or an if-elsif tree, a case statement, etc.,
> whatever you feel happiest with. This approach will be a lot easier to
> test and debug.

Depends.  If you build the one RX one alternative at a time and test
during each iteration I'd say that works pretty good as well.  And if
the volume of data is hight the performance advantage of a single RX
might pay off.

Kind regards

robert


-- 
use.inject do |as, often| as.you_can - without end