From: Robert Klemme <bob.news@...>
Date: 2005-12-15T19:07:39+09:00
Subject: Re: string mangling

Martin Pirker wrote:
> Robert Klemme <bob.news@gmx.net> wrote:
>>> Imagine an input string
>>> aaaaabbccccceeebbbbbbbbbbeaaabaacccceeee...
>>>
>>> I have regexp for the parts a,b,c
>>> and e can be considered as else.
>>>
>>> So how can I efficiently search/step through the string from left to
>>> right, while calling for each section the fitting handler, kind of
>>>
>>> case section
>>>   /aaaa/ ...
>>>
>>>   /bbb/  ...
>>>
>>>   /cccc/ ..
>>>
>>>   else
>>>
>>> end
> [...]
>> Or, if you want to avoid a second match:
>
> of course I want :-)
>
>>>> s.scan /(a+)|(b+)|(c+)/ do |m|
>>>   case ( m.inject(0) {|i,e| break i if e; i + 1} )
>>>>     when 0
>>>>       puts "A"
>>>>     when 1
>>>>       puts "B"
>>>>     when 2
>>>>       puts "C"
>>>>   end
>>>> end
>> A
>> B
>> C
>> B
>> A
>> B
>> A
>> C
>> => "aaaaabbccccceeebbbbbbbbbbeaaabaacccceeee"
>
> ...but this doesn't allow a processing step in the "else" case?

You can have an else clause - but it will never be called.  I guess you
will process entries between matches.  In that case scan won't help - at
least not as used in my example.

A simple option would be to use #split with a group around the whole
regexp and then operate on the array of strings you get.  Whether that's
feasible (volume?) in you case I cannot decide.

s.split(/((?:a+)|(?:b+)|(?:c+))/.each do |m|
 case section
   /aaaa/ ...

   /bbb/  ...

   /cccc/ ..
   else
 end
end


Kind regards

    robert