Here's a perl version of the lisp evaluator from Paul Graham's
The Roots of Lisp.
Update: Added another test case, fixed a bug in the code, discovered a strange way to cause a segfault (uncomment out the "die" statement below).
#!/usr/bin/perl -w
#
# Evaluate lisp expressions
#
# From Paul Graham's _The Roots of Lisp_
# http://paulgraham.com/rootsoflisp.html
#
# Pretty close to a direct transliteration,
# not simplified to make comparison easier.
use re 'eval'; #use recursive regex for parsing
use strict;
# From the paper, a lisp expressions is either an atom, which is
# a sequence of letters, or a list of zero or more expressions,
# separated by whitespace and enclosed by parentheses.
my $atom = qr{[a-zA-Z]+};
my $exp; $exp = qr{\(\)|$atom|\((??{$exp})(?:\s+(??{$exp}))*\)}s;
# Something to test the evaluator with
my @tests = (
"(quote A)","(quote (D E F))","(atom (quote B))","(atom (quote (c b a)
+))",
"(eq (quote a) (quote a))","(eq (quote b) (quote a))","(eq (quote ())
+(quote ()))",
"(car (quote (a b c)))","(cdr (quote (a b c)))",
"(cons (quote a) (quote (b c (d))))",
"(cond ((atom (quote (a b))) (quote TRUE))
((atom (quote a)) (quote NOT)))",
"((lambda (x) (cons x (quote (a b)))) (quote z))",
# 3+2?
"((label add (lambda (x y)
(cond ((eq (car x) (quote ())) y)
((quote t) (add (cdr x) (cons (quote s) y))))))
(quote (s s s ())) (quote (s s ())))",
# Fibonacci
"((label fib (lambda (n)
(cond ((eq (car n) (quote ())) (quote (s ())))
((eq (car (cdr n)) (quote ())) (quote (s ())))
((quote t) ((label add (lambda (x y)
(cond ((eq (car x) (quote ())) y)
((quote t) (add (cdr x) (cons (quote s)
+y))))))
(fib (cdr n)) (fib (cdr (cdr n))))))))
(quote (s s s s s ())))");
for (@tests)
{
print "$_ => ";
print ev($_,())."\n";
}
sub ev
{
local $_ = shift; my %a = @_;
if(atom($_) eq "t")
{
return $a{$_};
}
elsif(/^\(($exp)(.*)\)$/s and atom($1) eq "t")
{
if(/^\(quote\s+($exp)\)$/s)
{
return $1;
}
elsif(/^\(atom\s+($exp)\)$/s)
{
return atom(ev($1,%a));
}
elsif(/^\(eq\s+($exp)\s*($exp)\)$/s)
{
my $a = ev($1,%a);
my $c = ev($2,%a);
return "t" if ($a eq $c and atom($a));
return "()";
}
elsif(/^\(car\s+($exp)\)$/s)
{
my $x = ev($1, %a);
$x =~ /^\(($exp).*\)$/s;
return $1;
}
elsif(/^\(cdr\s+($exp)\)$/s)
{
my $x = ev($1, %a);
$x =~ /^\($exp\s+(.*)\)$/s;
return "($1)";
}
elsif(/^\(cons\s+($exp)\s+($exp)\)$/s)
{
my $h = ev($1, %a);
my $t = ev($2, %a);
$t=~/^\((.*)\)$/;
return "($h $1)";
}
elsif(/^\(cond\s+(.*)\)$/s)
{
return evcon($1, %a);
}
else
{
/^\(($exp)\s+(.*)\)$/s;
return ev("($a{$1} $2)",%a);
}
}
elsif(/^\(\(label\s+($atom)\s*\((lambda\s+$exp\s*$exp)\)\s*\)\s*(.
+*)\)$/s)
{
$a{$1}="($2)";
return ev("($1 $3)",%a);
}
elsif(/^\(\(lambda\s+($exp)\s+($exp)\)\s*(.*)\)$/s)
{
my ($params, $f, $vals) = ($1,$2,$3);
$vals =~ s/($exp)\s*/ev($1,%a)." "/ges;
$vals =~ s/\s$//s; #strip off trailing space
$vals = "($vals)";
my %b = evlis($params,$vals,%a);
return ev($f, %b);
}
else
{
# Uncommenting this line causes a segfault on perl5.8.4
# even if nothing ever gets here. Strange...
#die "Syntax Error: $_\n";
}
}
sub atom
{
my $x = shift;
return "t" if $x=~/^$atom$/;
return "t" if $x=~/^()$/;
return "()";
}
sub evlis
{
my ($p, $v, %a) = @_;
if($p=~/^\(($exp)\s*(.*)\)$/s)
{
my ($h, $t) = ($1,$2);
if($v=~/^\(($exp)\s*(.*)\)$/s)
{
$a{$h} = $1;
return evlis("($t)","($2)",%a);
}
}
return %a;
}
sub evcon
{
my $c = shift; my %a = @_;
my ($h, $t) = $c=~/^($exp)\s*(.*)$/s;
my ($p, $q) = $h=~/^\(($exp)\s+(.*)\)$/s;
if(ev($p,%a) eq "t")
{
return ev($q,%a);
}
else
{
return evcon($t,%a);
}
}
-
Are you posting in the right place? Check out Where do I post X? to know for sure.
-
Posts may use any of the Perl Monks Approved HTML tags. Currently these include the following:
<code> <a> <b> <big>
<blockquote> <br /> <dd>
<dl> <dt> <em> <font>
<h1> <h2> <h3> <h4>
<h5> <h6> <hr /> <i>
<li> <nbsp> <ol> <p>
<small> <strike> <strong>
<sub> <sup> <table>
<td> <th> <tr> <tt>
<u> <ul>
-
Snippets of code should be wrapped in
<code> tags not
<pre> tags. In fact, <pre>
tags should generally be avoided. If they must
be used, extreme care should be
taken to ensure that their contents do not
have long lines (<70 chars), in order to prevent
horizontal scrolling (and possible janitor
intervention).
-
Want more info? How to link
or How to display code and escape characters
are good places to start.