#!/usr/local/bin/perl
#
# checktables.pl - Verify your HTML code so your tables will work and look
# good in Netscape (which does not obey the HTML specification).
#
# Written by Bjorn Stenberg (bjorn.stenberg@sth.frontec.se) 980725
# 

$argc=@ARGV;

boo:
    while ( $x < $argc ) {
        if ( $ARGV[$x] eq "-h" ) {
            &usage;
        }
        elsif ( $ARGV[$x] eq "-v" ) {
            $verbose = 1;
        }
        else {
            $filename = $ARGV[$x];
            last boo;
        }
        $x++;
    }
if ( $filename eq "" ) {
    &usage;
    exit;
}

sub usage {
    print "usage: checktables.pl [-v] [-h] htmlfile\n";
    print "\t-v  produces a \"graph\" of the table structure\n";
    print "\t-h  produces this help text\n";
    exit;
}

sub add {
    $a = "$a  ";
}

sub del {
    $a = substr( $a, 0,length($a)-2);
}

%ok=("table",1,"tr",1,"td",1,"/table",1,"/tr",1,"/td",1);

#
# Allowed sequences of tags (on the same level)
#
#            table, /table, tr, /tr, td, /td
# table        0      1     0   0    1    0
# /table       1      0     0   1    0    0
# tr           1      0     0   1    0    0
# /tr          0      0     1   0    0    1
# td           0      0     1   0    0    1
# /td          0      0     0   1    1    0

%allow=("table","/table,td", "/table", "table,/tr", "tr", "table,/tr",
        "/tr", "tr,/td", "td", "tr,/td", "/td", "/tr,td" );

$help{"table","table"}  ="/table";
$help{"tr","table"}     ="td";
$help{"/tr","table"}    ="/table";
$help{"/td","table"}    ="td";
$help{"/table","/table"}="table";
$help{"tr","/table"}    ="/tr";
$help{"td","/table"}    ="table";
$help{"/td","/table"}   ="/tr";
$help{"/table","tr"}    ="table";
$help{"tr","tr"}        ="/tr";
$help{"td","tr"}="/td";
$help{"/td","tr"}       ="/tr";
$help{"table","/tr"}    = "tr";
$help{"/table","/tr"}   = "table";
$help{"/tr","/tr"}      = "tr";
$help{"td","/tr"}       = "/td";
$help{"table","td"}     = "tr";
$help{"/table","td"}    = "table"; # -> table+td -> table+tr+td
$help{"/tr","td"}       = "tr";
$help{"td","td"}        = "/td";
$help{"table","/td"}    = "td";    # -> table+td -> table+tr+td
$help{"/table","/td"}   = "table"; # -> table+/td -> table+td+/td -> t+tr+td
$help{"tr","/td"}       = "td";
$help{"/td","/td"}      = "td";



open(FILE, "<$filename");

if ($verbose){
    print "line  tag  level\n";
}

loop:
    while(<FILE>) {
        $in = $_;

        $line++;

#    print "\rLine $line\n";

hej:
    # print "In: $in";
        if ( ($pos = index( $in, "<" )) >=0 ) {
            $tag = substr($in,$pos+1);
            if ( "/" eq substr( $tag, 0, 1 ) ) {
                $dir = -1;
            }
            else {
                $dir = 1;
            }
            if( $tag =~ /([^> ]*)/ ) {
                $tag = $1;
#    print "Tag: $tag\n";
            }
        }
        else {
            $type = "";
            next loop;
        }

#    print "$line: In1: $in\n";

    # skip a bit, brother
        $in = substr( $in, $pos+length($tag)+2 );

#    print "$line: In2: $in\n";

    # skip all but desired tags
        if ( $ok{$tag} != 1 ) {
            goto hej;
        }

        $tags++;

#    print "$line: Tag <$tag>, level $lvl\n";

    # is this tag allowed?
        if ( $lvl >0 ) {
            @options=split(",",$allow{$tag});
            for ( @options ) {
#    print "$line: $tag allowed after $_\. lasttag is $lasttag[$lvl]\n";
                if ( ( "" ne $lasttag[$lvl] ) && ( $_ eq $lasttag[$lvl] ) ) {
                    goto ok;
                }
            }
            if ($verbose){
                printf("%3d !! <%s> %d !!\n",$line,$tag,$lvl);
            }
            print "Bad sequence on line $line (level $lvl): A <$tag> after the <$lasttag[$lvl]> on line $lastline[$lvl]\n";
            $l1 = ($lastline[$lvl+1]+1);
            if ( $l1 < $lastline[$lvl] ) { $l1 = $lastline[$lvl] }
            $l2 = ($line-1);
            if ( $l2 < $l1 ) { $l2 = $l1 }

            print "It looks like you've missed a <$help{$lasttag[$lvl],$tag}> ";
            if ( $l1 != $l2 ) {
                printf "somewhere between line %d and %d.\n", $l1, $l2;
            }
            else {
                print "on line $l1.\n";
            }
            close(FILE);
            exit;
        }

ok:
        if ( $dir > 0 ) { 
            $lvl += $dir;
            $lasttag[$lvl] = $tag;
            $lastline[$lvl] = $line;
            if ($verbose){
                printf("%3d $a<%s> %d \n",$line,$tag,$lvl);
            }
            &add;
        }
        else {
            &del;
            if ($verbose){
                printf("%3d $a<%s> %d \n",$line,$tag,$lvl);
            }
            $lasttag[$lvl] = ($tag,$line);
            $lastline[$lvl] = $line;
            $lvl += $dir;
        }

        goto hej;
    }

close(FILE);

if (!$tags) {
    print "There are no tables in this page.\n";
}
else {
    print "All tables appear to be correct.\n";
}