<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<html>
<head>
<title>LEXER Package</title>
</head>
<body>
<p><em>Text copied from
<a href="http://www.geocities.com/mparker762/clawk">http://www.geocities.com/mparker762/clawk</a></em></p>
<h2>LEXER package</h2>
<p>The LEXER package implements a lexical-analyzer-generator
called <em>DEFLEXER</em>, which is built on top of both
REGEX and CLAWK. Many of the optimizations in the recent
rewrite of the regex engine went into optimizing the sorts
of patterns generated by DEFLEX.</p>
<p>The default lexer doesn't implement full greediness. If you have a rule for ints followed by a rule for floats, the int rule will match on the part before the decimal before the float rule gets a change to look at it. You can fix this by specifying <code>:flex-compatible</code> as the first rule. This gives all patterns a chance to examine the text and takes the one that matches the longest string (first pattern wins in case of a tie). The down side of this option is that it slows down the analyser. If you can solve the issue by reordering your rules that's the way to do it.</p>
<p>I'm currently writing an AWK->CLAWK translator using this
as the lexer, and it's working fine. As far as I can tell,
the DEFLEXER-generated lexing functions should be fast
enough for production use.</p>
<p>Currently, the LEX/FLEX/BISON feature of switching
productions on and off using state variables is not
supported, but it's a pretty simple feature to add. If
you're using LEXER and discover you need this feature,
let me know.</p>
<p>It also doesn't yet support prefix and postfix context
patterns. This isn't quite so trivial to add, but it's
planned for a future release of regex, so LEXER will be
getting it someday.</p>
<p>Anyway, Here's a simple <em>DEFLEXER</em> example:</p>
<code><pre>
(deflexer test-lexer
("[0-9]+([.][0-9]+([Ee][0-9]+)?)"
(return (values 'flt (num %0))))
("[0-9]+"
(return (values 'int (int %0))))
("[:alpha:][:alnum:]*"
(return (values 'name %0)))
("[:space:]+") )
> (setq *lex* (test-lexer "1.0 12 fred 10.23e45"))
<closure>
> (funcall *lex*)
FLT
1.0
> (funcall *lex*)
INT
12
> (funcall *lex*)
NAME
"fred"
> (funcall *lex*)
FLT
1.0229999999999997E46
> (funcall *lex*)
NIL
NIL
</pre></code>
<p>
You can also write this lexer using the <code>:flex-compatible</code> option, in which case you can write the <em>int</em> and <em>flt</em> rules in any order.
</p>
<code><pre>
(deflexer test-lexer
:flex-compatible
("[0-9]+"
(return (values 'int (int %0))))
("[0-9]+([.][0-9]+([Ee][0-9]+)?)"
(return (values 'flt (num %0))))
("[:space:]+")
)
</pre></code>
<hr>
<address><a href="mailto:mparker762@hotmail.com">Michael Parker</a></address>
</body>
</html>
