From: Adam Chamberlain <awchamberlain@...>
Date: 2008-02-09T04:12:07+09:00
Subject: Parsing a spice netlist, need to be faster

Hey guys!  I am writing some code that has gotten far too slow.  I am
reading files that have the following types of statements:

This is all test written in a file
+bob = 2
+roger = 3
+barn = 'bob'
+frank = '3 + roger'
+nancy = 'roger*6 '  gary =4 jason= 'bob+roger'

It cannot be assumed that bob will be declared before barn, though I
would like to know if it isn't in certain modes.  I want to be able to
read all these parameters in (there will be roughly 20,000 of them) and
be able to dump them into an object.  That way I can do something like
a["nancy"] and get 18.  The code is currently too slow.  I need to read
parse strings really fast and be able to evaluate these statements.  For
example, if I read in '3 + roger' and do an eval(a), there will be a
nomethod error because roger is not defined.  I cannot have roger as a
variable because I can't keep it in scope.  Instead I have to pick the
line apart and get 'roger' out of it, then I search through the hash
table to find where roger is.  This implementation works but is
incredibly slow.  I need someone to point me to some good practices in
Ruby for parsing text and trying to evaluate what's found.

I also recognize the danger of the eval command.  I have taken steps to
limit what can and can't reach the eval statement.  For example, it is
impossible for backticks to reach the eval command, removing one method
of making a system call within the evla statement.

Adam
-- 
Posted via http://www.ruby-forum.com/.