cvs: php-lang / syntax_spec_scanner.pl

sas		Tue Nov 21 14:47:37 2000 EDT

  Added files:
    /php-lang	syntax_spec_scanner.pl 
  Log:
  A script which scans the syntax as found in c-like-syntax-spec
  and outputs XML.
  
  

Index: php-lang/syntax_spec_scanner.pl
+++ php-lang/syntax_spec_scanner.pl
#! /usr/bin/perl
#
# This script converts the syntax to XML
#
# $Id: syntax_spec_scanner.pl,v 1.1 2000/11/21 22:47:36 sas Exp $

use strict;

my %nts;
my $nt;
my @nt_list;
my $lineno = 0;
my $flags;
my $one_of;
my $match;

while (<>) {
	chop;

	$lineno++;
	
	# Skip empty and comment lines
	next if (/^(\s*|#.*)$/);

	if (/^([-a-z]+):(.*)/) {
		$nt = $1;
		$flags = $2;
			
		$one_of = 0;
		$match = 0;
	
		if ($flags =~ /one of/) {
			$one_of = 1;
		} elsif ($flags =~ /match/) {
			$match = 1;
		}
		
		# We want to keep the order. keys %nts does not
		# maintain the order..
		push @nt_list, $nt;
	} elsif (/^\s{4}(.+)/) {
		if ($one_of) {
			foreach my $e (split(" ", $1)) {
				push @{$nts{$nt}}, $e;
			}
		} elsif ($match) {
			push @{$nts{$nt}}, "MATCH$1";
		} else {
			push @{$nts{$nt}}, $1;
		}
	} else {
		print "Unrecognized line $lineno\n";
	}
}

foreach $nt (@nt_list) {
	print "<defsym>$nt\n";
	
	foreach my $opt (@{$nts{$nt}}) {
		print "   <def>";
		foreach my $t (split(" ", $opt)) {
			if ($t =~ /^'(.*)'$/) {
				print "<literal>$1</literal>";
			} elsif ($t =~ /^\[(.*)\]$/) {
				print "<opt-non-term>$1</opt-non-term>";
			} elsif ($t =~ /^MATCH(.*)$/) {
				print "<match>$1</match>";
			} else {
				print "<non-terminal>$t</non-terminal>";
			}
		}
		print "</def>\n";
	}
	print "</defsym>\n";
}