#!/usr/bin/perl

use strict;


topictree_find_unmentioned_pages(web=>$ARGV[0],file_containing_topictree=>$ARGV[1]);
exit(0);

sub topictree_find_unmentioned_pages {
    my %args = @_;

    die "file_containing_topictree does not exist: <<$args{file_containing_topictree}>>" if ! -e $args{file_containing_topictree};
    die "web does not exist: <<$args{web}>>" if ! -e $args{web};

    print "This file automatically generated by script topictree_find_unmentioned_pages.pl\n\n";
    print "Date script last run: " . `date` . "\n\n";
    print "Typically, inspect this, edit it, and copy back to TopicTree\n\n";
    print "\n";

    my $topictree = TopicTree::scan_topictree($args{file_containing_topictree});
    my $topiclist = list_of_topics_in_web($args{web});

    
    #print "topicslist=" . join(' ',@$topiclist);

    my @missing_topic_list;
    foreach my $f ( @$topiclist ) {
	#print "$f\n";
	push(@missing_topic_list, $f) if ! $topictree->topic_is_mentioned($f);
    }

    if( @missing_topic_list ) {
	print "Topics missing from topictree:\n";
	foreach my $m ( @missing_topic_list ) {
	    print "   * [[$m]]\n";
	}
	$topictree->print;
    }
    else {
	print "No files missing from TopicTree\n";
    }


}

sub list_of_topics_in_web {
    my $web = shift;
    die "web does not exist $web" if ! -e $web;
    
    my $ret = [];

    opendir my $webdir,$web;
    while(my $file=readdir($webdir)) {
	next unless ($file =~ /\.txt$/);
	my $topic = $file;
	$topic =~ s/\.txt$//;
        push @$ret, $topic;
    }

    return $ret;
}


{
    package TopicTree;
    
    sub new {
	my $this = {};
	bless $this;
    }

    sub scan_topictree {
	my $file = shift;
	die "scan_topictree does not exist: <<$file>>" if ! -e $file;

        my $topictree = new TopicTree;
        $topictree->{links_mentioned} = {};
	
        my $topictree_linenumber = 0;
        my $in_topictree = 0;

	open my $fh, $file;
        while( my $line = <$fh> ) {
	    #print $line;
            $topictree->{raw_text} .= $line;
	    
            $topictree_linenumber++ if $in_topictree;

	    # Look for %BeginTopicTree% / %EndTopicTree%
	    # TBD: provide %BeginTopicTree{Name}%
	    if( $line =~ /%BeginTopicTree%/ ) { $in_topictree = 1; $topictree_linenumber = 0; }
	    if( $line =~ /%EndTopicTree%/ ) { $in_topictree = 0; }

	    # if between BeginTopicTree and EndTopicTree
            if( $in_topictree && $topictree_linenumber > 0 ) {
		# if looking at a line that looks like a TopicTree line
		# (other lines are just comments, as is rest of a TopicTree line)
		if( $line =~ m/^ *\*/ ) {
		    # delete twiki list syntax.
		    # TBD: remember list depth?
                    my $line_ = $line;
		    $line_ =~ s/^ *\*[ \t]*//;
		    # extract first twikilink on line - that is the link in the topictree.
		    my $link;
		    $link = $1 if $line_ =~ m{^\[\[([^]]+)\]\]};
		    $link = $1 if $line_ =~ m{^\[\[([^]]+)\]\[[^\]]*\]\]};
		    $link = $1 if $line_ =~ m{^([A-Z][a-zA-Z0-9]*)};
		    my $lm = $topictree->{links_mentioned};
		    $$lm{$link}++;

		    #print "DEBUG Raw line = $line\nlink extracted = $link <BR>\n\n"
		    #	if "$link" eq "";
		}
	    }
	}

        return $topictree;
    }

    sub topic_is_mentioned {
	my $this = shift;
	my $topic = shift;
	return $this->{links_mentioned}->{$topic}; 
    }
    
    sub print {
	my $this = shift;
	print $this->{raw_text};
    }

}