From: Charles Hixson <charleshixsn@...>
Date: 2004-09-20T06:13:35+09:00
Subject: Re: Method improvement request .--

David A. Black wrote:

>Hi --
>
>On Sun, 19 Sep 2004, Charles Hixson wrote:
>
>  
>
>>That's a very interesting rewrite of the parse1 match patterns.
>>It's a bit more complicated than that, e.g., at the match boundary 
>>apostrophe's aren't a part of the middle, but in the middle (of the 
>>middle) they are.  Think about 'don't'. 
>>    
>>
>Think about "'tis" :-)
>  
>
OUCH!  An excellent point.  I don't know how to handle that except by 
another special case pre-processor.  Sigh!  I was *SO* hoping to 
minimize special cases.  (Of course, I knew that would be impossible, 
but still...)

>>P.S.:  Does scan keep recycling ...it with (an elaboration of)
>>chunks   =   line.scan(/^((\W*)(\w*)(/W*)s+/)$).flatten
>>    
>>
>scan keeps recycling its pattern, but note that ^ and $ will be
>included each time.  Note also that ^ and $ apply to lines, not the
>entire string.  Thus:
>
>  $ ruby -e 'p "a\nb\ncde\n".scan(/./)'
>  ["a", "b", "c", "d", "e"]
>  $ ruby -e 'p "a\nb\ncde\n".scan(/^.$/)'
>  ["a", "b"]
>
>You can use \A and \Z to indicate start and end of string (or \z to
>match end-except-for-possible-newline).  
>David
>
Mmmph... This may take a bit of experimentation to get right.  OTOH, \z 
could save a chomp invocation on each line.  (I do think that handling a 
line at a time is the appropriate choice.  The other reasonable choice 
would be to accumulate text until I encounter a blank line, and then 
parsing it all in one go.  But that's a bit more sensitive to the 
formatting of the text, and could result in excessively large buffers.   
(You can certainly tell I'm no expert at regexps.)