#!/usr/bin/perl -w
#
# Take one of the raw data files of the form: yyyymmdd  line_of_data
# which may either a plain file or a flat file with multiple tagged entries
# and extract the required data into a simple table (minus the date)
#
# Version 1.0
#
# fc - 31st May 2010

use Date::Calc qw(Delta_Days);

# $date1 = '20090101';
# $date2 = '20090102';
# $diff = compare_yyyymmdd_dates( $date1, $date2 );
# print "$date1 $date2 -> $diff\n";
# exit(1);


$FLAG_DROP_DATE = 1;     # Can ONLY INCLUDE the data when $DOWNSAMP == 1

$FILE_IN      = '';
$FILE_OUT     = '';
$TAG          = '';    # For the flat files
$DATE_ST      = '';
$DATE_END     = '';
$DATE_YEAR    = '';    # These are an alternative to giving full start and end
$DATE_MONTH   = '';    # If given, year is needed; month is optional
$DOWNSAMP     = 1;
$" = "\t";

if( not @ARGV )
{
    print STDERR "\n";
    print STDERR "Usage: table_dat.pl -in in_file -start yyyymmdd -end yyyymmdd [options]\n";
    print STDERR "                    -out out_file      default is to stdout\n";
    print STDERR "                    -tag TAG           flat file tag\n";
    print STDERR "                    -downsamp #        down sample the data; # -> 1\n";
    print STDERR "\n";
    exit( 1 );
}

for( $i=0; $i<@ARGV; $i++ )
{
    if   ( $ARGV[$i] eq '-in'    ) { $FILE_IN  = $ARGV[++$i]; }
    elsif( $ARGV[$i] eq '-out'   ) { $FILE_OUT = $ARGV[++$i]; }
    elsif( $ARGV[$i] eq '-tag'   ) { $TAG      = $ARGV[++$i]; }

    elsif( $ARGV[$i] eq '-start'   ) { $DATE_ST    = $ARGV[++$i]; }
    elsif( $ARGV[$i] eq '-end'     ) { $DATE_END   = $ARGV[++$i]; }
    elsif( $ARGV[$i] eq '-year'    ) { $DATE_YEAR  = $ARGV[++$i]; }
  # elsif( $ARGV[$i] eq '-month'   ) { $DATE_MONTH = $ARGV[++$i]; }     # Not implemented just yet

    elsif( $ARGV[$i] eq '-downsamp') { $DOWNSAMP = $ARGV[++$i]; }
    else  { die "ERROR unknown option: $ARGV[$i]\n"; }
}

-r  $FILE_IN    or die "ERROR: input data is not there or not readable\n";
if( $FILE_OUT ) { open STDOUT, ">$FILE_OUT"  or die "ERROR opening $FILE_OUT for STDOUT: $!\n"; }


# Sort out the requested time slice
if( $DATE_ST )
{
        $DATE_END   or die "ERROR: must provide an end date\n";
    not $DATE_YEAR  or die "ERROR: -start and -year incompatiable\n";
    not $DATE_MONTH or die "ERROR: -start and -month incompatiable\n";
}
else
{
    $DATE_YEAR  or die "ERROR: -start OR -year needed\n";
    
    $DATE_ST  = $DATE_YEAR."0101";
    $DATE_END = $DATE_YEAR."1231";                             # NEED TO implement the month level here
}

$flag_record = 1;

open   IN, $FILE_IN  or die "ERROR trying to open input file: $FILE_IN\n";
while(<IN>)
{
    next if /^%/;
    next if /^\s+/;

    if( /^>(\S+)/ )
    {
        $tag = $1;
        if( $tag eq $TAG ) { $flag_record = 1; }
        else               { $flag_record = 0; }
        
        next;
    }
    
    next unless $flag_record;
    
    
    chomp;
    /^(\S+)\s+(.*)/  or die "ERROR parsing line $. of $FILE_IN\n";
    $ln   = $_;
    $date = $1;
    $data = $2;
    
    if( 0 <= compare_yyyymmdd_dates( $DATE_ST,  $date)  and  0 >= compare_yyyymmdd_dates( $DATE_END, $date) )
    {  
        $ln = $data  if $FLAG_DROP_DATE;
        if( 1 == $DOWNSAMP)
        {
            print "$ln\n";
        }
        else   # Down sample the data
        {
            @nums = split /\s+/, $data;      # ASSUMING there is NO DATE here  -- now enforced by using $data rather than $ln here
            @samp = ();
            
            for( $i=0; $i<@nums; $i+=$DOWNSAMP )
            {
                for( $val=0, $j=0; $j<$DOWNSAMP; $j++ )
                {
                    $val += $nums[$i+$j];
                }
                $val = int( 10 * $val / $DOWNSAMP ) /10;
                push @samp, $val;
            }
            print "@samp\n";
        }
    }
}
close  IN;
close STDOUT;



#=====
#
# return the number of days between the two dates.
#        if date2 is the next day from date1, return 1, etc
#
sub compare_yyyymmdd_dates
{
    my( $date1, $date2 ) = @_;
    my( $y1,$m1,$d1, $y2,$m2,$d2 );
    
    $date1 =~ /^(\d\d\d\d)(\d\d)(\d\d)$/ or die "Error [compare_yyyymmdd_dates] parsing date: $date1\n";
    $y1 = $1;
    $m1 = $2;
    $d1 = $3;

    $date2 =~ /^(\d\d\d\d)(\d\d)(\d\d)$/ or die "Error [compare_yyyymmdd_dates] parsing date: $date2\n";
    $y2 = $1;
    $m2 = $2;
    $d2 = $3;
    
    return( Delta_Days($y1,$m1,$d1, $y2,$m2,$d2) );
}