Wiki2rem

#!/usr/bin/perl

# wiki2rem 
# Convert the PZI Networked Media course schedule from
# the HTML of Trac's wiki to a remind calendar file

=head1 NAME

wiki2rem - Convert the the PZI Networked Media course schedule from HTML to remind

=head1 SYNOPSIS

wiki2rem [URL]

wiki2rem < [FILE.html]

=head1 DEPENDENCIES

wget

=head1 AUTHOR

FC, 4/2010

=cut

# determine current year
my ($sec,$min,$hour,$mday,$mon,$year,$wday,$yday,$isdst) = localtime time;
$year += 1900;

# fetch web page if URL is given as an argument
if ($ARGV[0]) {
	# sanitize input
	$ARGV[0] =~ s/[;|\s]//g;
	@wiki_lines = split ("\n", `wget -O - $ARGV[0]`);
	}
# or act as a filter on the HTML file supplied via STDIN
else {
	while (<STDIN>) {
		push @wiki_lines, $_;
		}
	}

# transform wiki schedule HTML page into a remind file, print to STDOUT
foreach $_(@wiki_lines) {
	chomp;
	if (/<h3 id/) {
		/.*id=\"([A-Z][a-z]+)([A-Z][a-z]+)([0-9]+)\".*/;
		$day = "$3 $2 $year";
		}
	if (/<li>/) {
		/.*<li>([0-9]+)-([0-9]+) (.*)/;
		$hour = $1.':00';
		$duration = $2-$1.':00';
		$message = $3;
		$message =~ s/<[^>]*>//g; 

		if ($message) {
			print "$day AT $hour DURATION $duration MSG $message\n";
			}
		}
	}

# EOF