#!/usr/bin/perl -w

# ---------------------------------------------------------------
# Perl code to convert email delimted data to Timesearcher format   
# ---------------------------------------------------------------   
# To execute the script, type                                       
# perl p_codeTSfmt.pl <inputFileName> <outputFileName>                  
# 
# Input file data format supported: time-katy_iv.txt
# (Available at: http://iv.slis.indiana.edu/lm/lm-time-series.html)
#
# Output file generated can be directly used for visualization in 
# time-searcher application
#
# ---------------------------------------------------------------

# Read the input file
# -------------------
open(IN, $ARGV[0]);
@data = <IN>;
close(IN);

# remove the header line
# ----------------------
shift(@data);

# Split the record based on the delimiter and generate arrays of 
# names, date and combination of both
# ---------------------------------------------------------------
@new_data = ();
foreach(@data) {
    my($time_log, $name, $sub) = split("\\|", $_);
    
    @tmp_time = ();
    @tmp_time = split("\\s", $time_log);
    
    $date = "";
    $date = "$tmp_time[2]"."$tmp_time[1]";
    push(@date, $date);

    $name =~s/(.*)?\<.*/$1/;
    $name =~s/^\s+//;
    $name =~s/\s$//;
    push(@name, $name);

    $new_data = "";
    $new_data = "$date#$name";
    push(@new_data, $new_data);   
}

# Find unique names
# --------------------
$hash_name = ();
foreach(@name) {
    $hash_name{$_}++;
}

@uniq_names = ();
foreach(keys %hash_name){
    push(@uniq_names, $_);
}
$len_uniq_names = @uniq_names;

# Find unique dates
# -----------------
@uniq_dates = ();
$tmp_date = "";
push(@uniq_dates, $tmp_date);
foreach(@date){
    if ($_ ne $tmp_date) {
	$tmp_date = $_;
	push(@uniq_dates, $tmp_date);
    }
}
shift(@uniq_dates);
$len_uniq_dates = @uniq_dates;

# Generate the output file in .tqd format
# --------------------------------------- 
open(OUT, ">$ARGV[1]");
print OUT"# title\n";
print OUT"Enter time searcher title here ...\n";

print OUT"# static attributes\nSender,String\n";
print OUT"# Dynamic atts\nemailNumber,Int\n";

print OUT"# of time points=n\n";
print OUT"$len_uniq_dates\n";

print OUT"# of records k\n";
print OUT"$len_uniq_names\n";

print OUT"# time point labels\n";
$date_string = join(",", @uniq_dates);
print OUT"$date_string\n";

print OUT"#stat1, dynamic \@ t1, dynamic \@t2, ....,dynamic \@tn\n";
print OUT"# each dynamic are the v1, v2, ..vk \@ time tj\.\n";

foreach $name(@uniq_names){
    $name_freq_str = "$name,";

    foreach $date(@uniq_dates){
	$date_name_str = "$date#$name";
	$ctr = 0;
	foreach(@new_data) {
	    if ($_ eq $date_name_str) {
		$ctr++;
	    }
	}
	$name_freq_str .= "$ctr,"

    }
    $name_freq_str =~s/,$//;
    print OUT"$name_freq_str\n";
} 

close(OUT);