#!/usr/local/bin/perl

################################################################################
# Name:  Bridget Thomson-McInnes
#
# Summary: Takes in a file and parses it at the '.START'. Then uses regular
#          expressions to find dates/times and tags them accordingly.
#
# Usage:   This program takes in a file on the command line. 
#              Ex: ./tagging.pl <filename>
#################################################################################
#variables
$timeTag = " [time]";  
$timeEnd = "[/time] ";
$dateTag = " [date] ";
$dateEnd = "[/date] ";

$week = "(Monday|Tuesday|Wednesday|Thursday|Friday|Saturday|Sunday|Mon|Tue|Wed|Thur|Fri|Sat|Sun).?";
$month = "(January|February|March|April|May|June|July|August|September|October|November|December|Jan|Feb|Mar|Apr|Jun|Jul|Aug|Sept|Sep|Oct|Nov|Dec).?";
$holidays = "(Christmas|Easter|Halloween|Hanauka|Thanksgiving|Valentine Day|President\'s Day|Memorial Day|Labor Day|Father\'s Day|Mother\'s Day|Independence Day|New Years)";
$deicticExp = "(yesterday|tomorrow)";
$writtenNum = "(one|two|three|four|five|six|seven|eight|nine|ten|eleven|twelve)";
$writtencom = "(first|second|third|fourth|fifth|sixth|seventh|eighth|ninth|tenth|eleventh|twelfth)";

#Load file
$inputLine = "";
while (<>) {  $inputLine = $inputLine . $_; }

#Split at the .START expression
$expression = ".START";
@blockArray = split /$expression/, $inputLine;

#for every block find the time/dates and tag them
foreach (@blockArray) {
  #midnight afternoon or morning
  s/(M|m)idnight|(A|a)fternoon|(M|m)orning/$timeTag$&$timeEnd/g;
  #time
  s/1?[0-9]:[0-5][0-9] (am|pm)?/$timeTag$&$timeEnd/g;
  #time o clock
  s/[0-9] o\'clock|1[012] o\'clock|$writtenNum o\'clock /$timeTag$&$timeEnd/g;
  #quarter to or half past an hour
  s/quarter to $writtenNum|half past $writtenNum/$timeTag$&$timeEnd/g;
  #date
  s/([123]?[0-9])? ($week|$month.?|$week $month) ([123]?[0-9],?)? ([123]?[0-9]?[0-9][0-9])?/$dateTag$&$dateEnd/g;
  #years
  s/[123][0-9][0-9][0-9][.,\s]?\s/$dateTag$&$dateEnd/g;
  #year
  s/('[0-9][0-9])[^(yrs|years)]?/$dateTag$&$dateEnd/g;
  #month the number
  s/(T|t)he [123]?[0-9](rd|st|nd|th) of $month/$dateTag$&$dateEnd/g;
  #holidays
  #s/$holdiay/$dateTag$&$dateEnd/g;
  #the day after or before tomorrow or today
  s/((T|t)he day before)? (yesterday|tomorrow)/$dateTag$&$dateEnd/g;
  #number dates
  s/[123]?[0-9][\/][0123]?[0-9][\/][123]?[0-9]?[0-9][0-9]/$dateTag$&$dateEnd/g;
}

print "@blockArray\n";