UNSW Timetable Converter for 12s2 only

#!/usr/bin/perl

# funswap.pl
# EDIT: modified by Calvin,
# Changes include:
#   -changed $weekStart, starting date of semester (now works for 12s2)
#   -changed $OUTFILE, output file name to timetable12s2.ics
#   -changed $weekBeforeBreak
#
# written by Rowan Katekar (rowan.katekar@gmail.com)
# released under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported License.
# http://creativecommons.org/licenses/by-nc-sa/3.0/

# USAGE SUGGESTION:
# When importing into google calendar, create a special calendar just for this timetable.
# That way, if something goes wrong you can delete the events from the calendar all at once, rather
# than having to delete them one by one.

# uses DateTime module - install from cpan using
# sudo apt-get install libdatetime-perl

# reads in UNSW timetable html file
# outputs .ics file for importing into Outlook, Evolution, Google Calendar, ... etc.
# assumes UNSW timetable html file is in the same folder as the perl script and called "timetable.html"

## Edited by William Pickering (@YodaDaCoda)
## * Changed from cat timetable.html (linux only file handling) to native perl file handling.
##      Solves problems with things being printed to the terminal and generally not working
##      properly on Windows.
## * Changed date logic
##      Every single class was set to be on the next week from the previous class,
##      leading to classes being held years into the future.

# updated by Rowan Katekar again on 11-12-2011
# *  Changed dates to semester 1 2012 dates.
# *  Also changed the lecture name to the code of the subject
# *  Subject title is now part of the description


use DateTime;

# convert time of the format /(\d)?\d:\d\d[AP]M/ to a number from 0..23
sub conv24($);

# convert a weekday abbreviation to a number from 1-5
sub dayToNumber($);

# substitutes event values into iCal string & returns resultant string
sub subInValues($$$$$);

## Gets the minutes of a specified time string.
sub getMins($);

# First day of semester (currently set to first day of semester for 2012 semester 1)
$weekStart = DateTime->new( year=>2012, month=>7, day=>16, time_zone => 'Australia/Sydney' );

# Last week before mid-semester break
$weekBeforeBreak = 7;

# destination of ical file
$OUTFILE = "timetable12s2.ics";

# file to read from
$INFILE = "timetable.html";

# crap to ignore
@ignore =  ("Teaching", "Status", "Enrolled", "Units", "Campus", "Activity", "Section",
            "Weeks", "Location", "Instructor", "bsdsSequence", "outerTable", "Key",
            "printBt", "Print", "onClick", "Site Map", "Site Feedback", "Privacy Statement",
            "Provider", "Copyright");

##open files for reading/writing
open OUTFILE, ">", $OUTFILE or die $!;
open INFILE, "<", $INFILE or die $!;

# read in page
@lines = <INFILE>;
close INFILE;

##convert from array to one long string
$page = join (' ', @lines);

# split into sections based on subject
@sections = split (/([A-Z]{4}[0-9]{4}\s-.*?)</, $page);
shift @sections;

# do for each subject
foreach $section (@sections) {
   @useful = ();
   # print subject title
   if ($section =~ m/[A-Z]{4}[0-9]{4}\s-/) {
      $subject = $section;
   } else {
      # extract everything that's not in html tags
      @info = ($section =~ m/>([\w-:;,\s\(\)\&]*?[A-Z0-9]+[\w-:;,\s\)\&]*?)</g);
      foreach (@info) {
         push @useful, $_ if !(m/(data|row|table|sectionHeading)/ || m/^\s*\d+\s*$/);
      }
      # find times, classes, locations
      foreach (@useful) {
         $classes{$subject} .=  "\n" if  m/^Lecture|Tutorial|Laboratory$/;
         s/^\s*// && s/\s*$//;
         # ignore garbage
         if (!m/^[A-Z0-9]+$/ or m/Lecture|Tutorial|Laboratory|Mon|Tue|Wed|Thu|Fri/) {
            $garbage = 0;
            foreach $word (@ignore) {
               $garbage = 1 if ($_ =~ m/$word/);
            }
            $classes{$subject} .= "$_\n" if not $garbage;
         }
      }
   }
}

# go through classes for each subject
foreach $subject (keys %classes) {
   @lines = split /\n/, $classes{$subject};
   $lesson = 0;
   for $i (0..$#lines) {
      # get day
      if ($lines[$i] =~ m/\b(Mon|Tue|Wed|Thu|Fri)\b/) {
         $classDays{$subject}[$lesson] = $1;
         $lesson++;
      }
      # get weeks
      if ($lines[$i] =~ m/^[\d,-]+$/) {
         while ($lines[$i] =~ m/(\d+-\d+)/) {
            $replacement = "";
            $numbers = $1;
            $numbers =~ m/(\d+)-(\d+)/;
            $from = $1;
            $to = $2;
            $replacement .= "$_," for ($from..$to);
            $replacement =~ s/,$//;
            $lines[$i] =~ s/$numbers/$replacement/;
         }
         $classWeeks{$subject}[$lesson - 1] = $lines[$i]; # hax
         $classLocation{$subject}[$lesson - 1] = $lines[$i + 1]; # get location
      }
      # get time
      $classTimes{$subject}[$lesson - 1] = $1 if ($lines[$i] =~ m/(\d+:\d+[AP]M - \d+:\d+[AP]M)/);
      # get class type (lecture/tute/lab)
      $classType = $1 if ($lines[$i] =~ m/(Lecture|Tutorial|Laboratory)/ or $lines[$i] =~ m/(.*\bof\b.*)/ );
      $classTypes{$subject}[$lesson] = $classType;
   }
   $classTypes{$subject}[$lesson] = ""; # moar hax
}


# ical 'header'
$STARTCALENDAR = <<START;
BEGIN:VCALENDAR
PRODID:-//fun swap//hacked together perl//EN
VERSION:2.0
CALSCALE:GREGORIAN
METHOD:PUBLISH
START

# *START*, *END*, *EVENT* etc. are substituted with actual class details later on
$EVENTSTRING = <<EVENT;
BEGIN:VEVENT
DTSTART:*START*
DTEND:*END*
SUMMARY:*TITLE*
LOCATION:*LOCATION*
DESCRIPTION:*DESCRIPTION*
SEQUENCE:0
END:VEVENT
EVENT

# goes at the end of the produced file
$ENDCALENDAR = "END:VCALENDAR";

print OUTFILE $STARTCALENDAR;

foreach $subject (keys %classes) {
   foreach $lesson (0..length($classTypes{$subject})) {
      if (defined $classDays{$subject}[$lesson]) {
         @weeksClassIsOn = split (/,/,$classWeeks{$subject}[$lesson]);

         # convert start and end times to 24h time
         @times = split (/ - /, $classTimes{$subject}[$lesson]);
         $startClassH = conv24($times[0]);
         $startClassM = getMins($times[0]);
         $endClassH = conv24($times[1]);
         $endClassM = getMins($times[0]);

         # format the title of the event as "[SUBJECT CODE] [Lecture|Tutorial|Laboratory]"
         $subject;
         @subjectCode = split(" - ", $subject);
         $subjectCode = @subjectCode[0];
         @subjectName = split(" - ", $subject);
         $subjectName = @subjectName[1];
         $subjectCode =~ s/^([A-Z]{4}[0-9]{4}).*$/$1/;
         $classType = $classTypes{$subject}[$lesson];
         $title = "$subjectCode $classType";
         $description = "$subjectName";

         # location of event
         $location = $classLocation{$subject}[$lesson];

         # add classes to calendar
         for $week (1..13) {   #for each week in the timetable


            if ($week ~~ @weeksClassIsOn) {   #if the class is on that week

               ##starting date
               $weekDate = $weekStart->clone();
               $weekDate->add( weeks => ($week -1) );
               # skip the week of mid-session break
               $weekDate->add( days => 7) if ($week > $weekBeforeBreak);

               $weekDate->add( days => dayToNumber($classDays{$subject}[$lesson]) - 1 );

               ##set the time the class starts
               $startTime = $weekDate->clone();
               $startTime->set_hour($startClassH);
               $startTime->set_minute($startClassM);

               ##set the time the class ends
               $endTime = $weekDate->clone();
               $endTime->set_hour($endClassH);
               $endTime->set_minute($endClassM);

               ##DateTime hackery pary one...
               ##Time part of the DateTime stamp needs to be formatted in Zulu time for compatibility with google calendar.
               ##If at least the time part is not in Zulu, all events show up as being one hour in duration.
               ##Convert time to UTC...
               $startTime->set_time_zone('UTC');
               $endTime->set_time_zone('UTC');

               # convert DateTime formatting to iCal formatting
               $startTime =~ s/[^0-9A-Z]//g;
               $endTime =~ s/[^0-9A-Z]//g;

               ##DateTime hackery part 2...
               ##Append Z for Zulu timezone to the times
               $startTime = $startTime."Z";
               $endTime = $endTime."Z";

               # substitute values into iCal event string
               $icalEvent = subInValues($title, $location, $startTime, $endTime, $description);

               ## append to file
               print OUTFILE $icalEvent;
            }
         }
      }
   }
}

## append the final line to the file... and we're done
print OUTFILE $ENDCALENDAR;
close OUTFILE;

##return the minutes of a given time
sub getMins($) {
   my $mins = $_[0];
   $mins =~ m/.*?:(.{2}).*?/;
   $mins = $1;
   return $mins;
}

# convert a weekday abbreviation to a number from 1-5
sub dayToNumber($) {
   my $dayName = $_[0];
   my %days = ("Mon" => 1,
            "Tue" => 2,
            "Wed" => 3,
            "Thu" => 4,
            "Fri" => 5);
   return $days{$dayName};
}

# convert time of the format /(\d)?\d:\d\d[AP]M/ to a number from 0..23
sub conv24($) {
   my $time = $_[0];
   $hour = $time;
   $hour =~ s/:.*$//;
   $hour += 12  if ($time =~ m/PM/ && $hour < 12); # convert to 24h time
   return $hour;
}

# substitutes event values into iCal string & returns resultant string
sub subInValues($$$$$) {
   my ($title, $location, $startTime, $endTime, $description) = @_[0..4];

   my $icalEvent = $EVENTSTRING;
   $icalEvent =~ s/\*START\*/$startTime/;
   $icalEvent =~ s/\*END\*/$endTime/;
   $icalEvent =~ s/\*TITLE\*/$title/;
   $icalEvent =~ s/\*LOCATION\*/$location/;
   $icalEvent =~ s/\*DESCRIPTION\*/$description/;

   return $icalEvent;
}