From: MikkelFJ <mikkelfj-anti-spam@...>
Date: 2003-03-12T22:34:21+09:00
Subject: Re: scanning strings


"YANAGAWA Kazuhisa" <kjana@dm4lab.to> wrote in message
news:20030312112248.DC4161EE12@milestones.dm4lab.to...

> > Nope, split will not give me the text where the split takes place
> > and it will also not give me a match object such that I can use
> > subexpressions.
>
> What did you test?
>
>     >ruby -ve 'p "abc def ghi".split(/(\s+)/)'
>     ruby 1.7.2 (2002-05-07) [i386-freebsd]
>     ["abc", " ", "def", " ", "ghi"]

I used 1.7.3, but mostly referred to the information provided by ri.
I assumed that matched string was removed, but this is apparently not always
the case:
From ri String#split

" now's  the time".split(/ /)   #=> ["", "now's", "", "the", "time"]
"1, 2.34,56, 7".split(/,\s*/)   #=> ["1", "2.34", "56", "7"]

My own test in 1.7.3:

"abc def ghi".split(/\s+/) #=> ["abc", "def", "ghi"]


I think my problem is important because it is typical for scanning tags
embedded in text.

On a related note: In the end I solved my problem using a while loop.
But since match do not take an offset as parameter I get to reallocate
strings all the time.

Based on memory, my solution was something like

s = "my string to scan"
RE = "\s+" # actually somewhat more complicated than this
while s.length > 0
  m = RE.match s
  if m
    process_text m.pre_match
    process_space m[0]
    s = m.post_match
 else
   process_text s
   break
 end
end

This is lengthy and not very efficient. The s = m_post_match could result in
a lot of allocation on large strings, and you can't operate directly on a
file.

Mikkel