doc/ja/prehtml

#!/usr/bin/perl -w
#
# The FreeBSD Japanese Documentation Project
#
# This is a preprocessor for HTML docs.
#
# usage: prehtml [-revcheck] <localtop> <relative pathname from localtop> <SGML filename>
#   (ex. % prehtml -revcheck ../.. news/1996 index.sgml)
#
# $FreeBSD: www/ja/prehtml,v 1.3 2000/10/31 10:14:38 kuriyama Exp $

my $revcheck;
my $topdir;
my $reldir;

my %file;
my %rev;

### parse options ....................................................
###
sub sOPT {1};
sub sARG {2};

my $opt_state = sOPT;
my $argv;

die "$0: too few arguments.\n" if (scalar @ARGV < 2);

while(defined($_ = $ARGV[0])) {

    if ($opt_state eq sOPT) {
        # option expected
        if(/^-(.+)/) { 
            shift @ARGV; # discard option itself
            local $_ = $1;
            /revcheck/ and do { $revcheck = 1; next; };

            die qq/$0: invalid option "-$1"\n/;
        } else {
            # this is not a option but an argument
            $opt_state = sARG;
            next;
        }
    } elsif ($opt_state eq sARG) {
        die "$0: too few arguments.\n" if (scalar @ARGV < 2);

        $topdir = shift @ARGV;
        $reldir = shift @ARGV;

        ### normalize $reldir into the form "foo/bar/"
        if($reldir ne '' and $reldir !~ /\/$/) {
            $reldir .= "/";
        }

        if(@ARGV) {
            $file{target} = shift @ARGV;
            if($revcheck) {
                local $_ = $file{target};
                s/\.sgml$//;
                $file{cvsweb}  = "www/en/${reldir}$_.sgml";
                $file{orgbase} = "${topdir}/../en/${reldir}$_";
                $rev{org} = get_rev_org($file{orgbase}.".sgml");
            }
        } else {
            ### If not specified SGML filename, then use stdin
            ### (but revcheck facility is disabled).
            ### This is mainly for debugging purpose.
            $file{target} = "-";
            undef $revcheck;
        }
        last;
    }
    die "$0: internal error: option parsing abnormally terminated.\n";
}

### add and replace entities .........................................
###
open TARGET,"<$file{target}" or die "$0: cannot open a target file: $!\n";
my $pos_date;
### first, get date string and rev_target
while(defined($_ = <TARGET>)) {
    if(/<!ENTITY\s+date/) {
        $pos_date = tell(TARGET);
        if(m/\"\x24FreeBSD: [^\s]+ [.0-9]+ ([\/0-9]+)[^\x24]*\x24\"/) {
            $date = $1;
        } else {
            $date = "UNKNOWN";
        }
    }
    if(/<!ENTITY\s+title/) {
        $pos_title = tell(TARGET);
    }

    if($revcheck) {
        if(/[Oo]riginal [Rr]evision:[ \t]*([0-9.]+)/) {
            $rev{target} = $1;
        }
        last if($revcheck and $rev{target} and $date);
    } else {
        last if($date);
    }
}

$rev{target} ||= "TARGET revision not found";
$rev{org}    ||= "ORG revision not found";

### if offset of "date" string is not found,
### use "title"'s instead.
if(not defined $pos_date) {
    if(not defined $pos_title) {
        die qq/$0: element "date" or "title" is not defined.\n/;
    }
    $pos_date = $pos_title;
}

### next, put lines and replace the line with
### $date + entity definitions for revcheck 
seek TARGET,0,0;
while(defined($_ = <TARGET>)) {
    if(tell(TARGET) == $pos_date) {
        print qq|<!ENTITY date CDATA 'Last modified: ${date}'>\n|;
        print qq|<!ENTITY base CDATA '${topdir}'>\n|;
        if($revcheck) {
            #print STDERR "$rev{org} -> $rev{target}\n";
            print qq|<!ENTITY enbase CDATA '${topdir}/../en'>\n|;
            print qq|<!ENTITY file.cvsweb  CDATA '$file{cvsweb}'>\n|;
            print qq|<!ENTITY file.orgbase CDATA '$file{orgbase}'>\n|;
            print qq|<!ENTITY rev.latest CDATA '$rev{org}'>\n|;
            print qq|<!ENTITY rev.target CDATA '$rev{target}'>\n|;
            printf "<!ENTITY %% rev.diff '%s'>\n", ($rev{org} eq $rev{target}) ? "IGNORE" : "INCLUDE";
        }
    } else {
        # for backward compatibility
        s/<!ENTITY \% rev.incl SYSTEM "[-a-zA-Z0-9]+.revinc"> \%rev.incl;//;
        print;
    }
}
close TARGET;

exit 0;

sub get_rev_org 
{
    my $infile = shift @_;
    my $rev_org;

    open ORG,"<$infile" or return undef;
    while(defined($_ = <ORG>)) {
        if(/\x24Free[B]SD: [^\s]+ ([.0-9]+) [\/0-9]+[^\x24]*\x24/) {
            $rev_org = $1;
            last;
        }
    }
    close ORG;
    return $rev_org;
}

__END__