toc_extract

#!/usr/bin/perl
use warnings;
use strict;
use File::Basename;

# Jekyll's Markdown doesn't support [toc]
# It does generate the anchor tags.
# Look thru the generated HTML and produce a TOC.
# It can be inserted manually.
# Set $mode = 0 to skip the first H1; to -1 to include all.

my $mode = -1;

my $path = dirname( $0 ) . "/docs/_site/simdocs/%s.html";
my $n = @ARGV;

while( @ARGV ) {
    my $fn = shift;
    $fn = sprintf( $path, basename( $fn, ( qw(.html .htm) ) ) )
      unless( $fn =~ m,/, );
    my $first = $mode;

    open( my $f, '<', $fn ) or die( "$fn:$!\n" ) ;
    printf STDERR ( "%s\n", $fn ) if( $n > 1 );
    while( <$f> ) {
        next unless( /^\s*<h([1-6])\s+id="([^"]+)".*?>(.*?)<\/h/ );
        my($lvl, $tag, $text) = ( $1, $2, $3 );
        if( $lvl == 1 ) {
            if( $first > 0 ) {
                $first = 0;
                next;
            } else {
                printf( " %s- [%s](#%s)\n", '  ' x ($lvl -1), $text, $tag );
            }
        } else {
            printf( " %s- [%s](#%s)\n", '  ' x ($lvl -2), $text, $tag );
            $first = 0 if( $first > 0 );
        }
    }
}