Greedy matching in prg2lout

[Darren Bane]

Hi there,

I'm trying to add support for RSL (the RAISE Specification Language - 
http://www.iist.unu.edu/raise/) to prg2lout.  This is somewhat different 
from the other programming languages supported because symbol mode 
should be the default, and there are many more mathematical operators.

One of these is the word 'is', which renders as the equivalence symbol 
(i.e. = with an extra line).  I added FixedToken entries for this, and 
it rendered perfectly.  However, now _any_ sequence of the characters 
i-s is replaced with equivalence, even identifiers like is_empty and 
is_full.

I spent a while hunting around the data structures, and tried

*  Using " is " instead of "is" as the start token.  However, this 
failed when I used is at the end of a line.

*  Using a full token definition to set the start strings to { "is ", 
"is\t", "is\n" }.  This was better, but the previous instance where is 
was at the end of the line now was replaced with two equivalence 
symbols, one in the correct place and another at the start of the next 
line.  I'd like to get rid of the latter.

I'd really rather not change the input format, as I think it's common to 
all extant RAISE CASE tools, and I don't want to mandate particular 
whitespace conventions either as this would mean you can't use the 
pretty printer.  What I'm hoping for is something like Perl's greedy 
pattern matching - prg2lout seems to be minimal; I'm making slow headway 
with the actual code of prg2lout, hence my request.

Thanks in advance.
--
Darren Bane