#!/usr/bin/perl
#
# Perl script to convert CSV output from glucometer into graphs, using gnuplot.
#
# Author: Timothy Allen <tim@treehouse.org.za>
# License: MIT
#
# TODO Convert to python (see # https://www.physicsforums.com/threads/gnuplot-how-to-find-the-area-under-a-curve-integrate.382070/
# )

use strict;
use warnings;
use Getopt::Long;
use IPC::Open3;
use Time::Piece;
use Data::Dumper;

my $error = "Usage: $0 --input <CSV file> --output <output PDF> [--max <n.n>] [--low <n.n>] [--high <n.n>]\n";
my @lines;
my @sorted_lines;
my @data;
my @avg_data;
my %intervals;
my %seen; 
my $a1c_calc;
my $page_size;
my $gnuplot_data;
my $total_graphs;
my $count_graphs = 0; 
my $page_number  = 0;
my $interval     = 15; # The number of minutes to average points for the area range graph

my $input        = '';
my $output       = '';
# set these values either in mmol/L or mg/dL (don't mix them)
my $max_glucose  = 8;
my $min_glucose  = 4;
my $graph_max    = 21;
my $units        = '';
my $page         = 'a4';
my $days_per_page = 2;
GetOptions ("input=s"    => \$input,          # The name of the CSV file from which to read values
            "output=s"   => \$output,         # The name of the PDF file to output
            "high:f"     => \$max_glucose,    # The high end of your target blood glucose level
            "low:f"      => \$min_glucose,    # The low end of your target blood glucose level
            "max:i"      => \$graph_max,      # The highest displayed glucose level on each graph
            "units:s"    => \$units,          # mmol/L or mg/dL
            "pagesize:s" => \$page,           # size of page to print
            "graphs:i"   => \$days_per_page)  # The number of days printed on each page
    or die $error;

open( my $ifh, '<:encoding(UTF-8)', $input ) 
    or die "Could not open file '$input' $!";
 
while ( my $row = <$ifh> ) {
    chomp( $row );
    push @lines, $row;
}

close( $ifh )
    or warn "close failed: $!";

if ( $page =~ /a4/i ) {
    $page_size = "29.7cm,21.0cm";
} elsif ( $page =~ /letter/i ) {
    $page_size = "11in,8.5in";
} elsif ( $page =~ /\d+(cm|in),\d+/ ) {
    $page_size = $page;
} else { 
    # A4 size default
    $page_size = "29.7cm,21.0cm";
}

# Set up basic gnuplot options for reading the CSV data
push @data, qq(
set terminal pdf size $page_size enhanced font 'Calibri,14' linewidth 1 
#set output '$output'

);

# Get the list of days for which to produce graphs
foreach my $row ( @lines ) {
    if ( $row =~ m#^"(\d{4}-\d{2}-\d{2})#ms ) { 
        my $day = $1; 
        $seen{$day}++; 
    }
}
$total_graphs = scalar keys %seen;

# Read each line into a $Data variable for use by gnuplot
# Then sample into a smoothed plot for each day, and store each smoothed line in a new $SmoothData$date variable
foreach my $d ( sort keys %seen ) { 
    my $label = "$1$2$3" if ( $d =~ m#(\d{4})-(\d{2})-(\d{2})# );
    push @data, qq(
\$Data$label << EOD
"timestamp","blood glucose","meal","method","comment");
    foreach my $row (@lines) {
        if ( $row =~ s#^"($d )#"$1#ms ) {
            push @data, $row;
        }
    }
    push @data, qq(EOD);

    push @data, qq(
set datafile separator ","

set timefmt "%Y-%m-%d %H:%M:%S"
set format x "%s" timedate
set format y "%.2f" numeric 
set samples 10000
set xdata
stats \$Data$label using 2
Mean$label = STATS_mean

set xdata time
set table \$SmoothData$label
#plot \$Data$label using "timestamp":"blood glucose" 
#plot \$Data$label using "timestamp":"blood glucose" smooth frequency
plot \$Data$label using "timestamp":"blood glucose" smooth mcsplines
#plot \$Data$label using "timestamp":"blood glucose" smooth bezier
unset table
undefine \$Data$label
);
}

# Set up output options for gnuplot.
# We don't bother to do this at the start, since the CSV needs a comma separator
# and the new $SmoothData, which contains a table, needs a whitespace separator
push @data, qq(
# change separator from CSV to table
reset
set datafile separator whitespace

set key off
set style data lines
set xdata time
set timefmt "%H:%M:%S"
set format x "%H:%M" timedate
set format y "%.0f" numeric 

set yrange [0:$graph_max]
# If extended to 23:59, the x grid overlaps with the border
set xrange ["00:00":"23:58"]

set style line 100 dt 3 lw 1 lc rgb "#202020"
set style line 101 dt 1 lw 1 lc rgb "#202020"
set linetype 110 lc rgb "red"

set lmargin 12
set rmargin 10
set tmargin 5
set bmargin 5

set multiplot title layout $days_per_page,1 

);

# For each day, generate a graph with some fancy options
foreach my $d ( sort keys %seen ) { 
    my $label = "$1$2$3" if ( $d =~ m#(\d{4})-(\d{2})-(\d{2})# );
    my $time  = Time::Piece->strptime ( $d, "%Y-%m-%d" );
    #my $title = $time->strftime("%a %d %b %Y");
    my $title = $time->strftime("%A, %d %B %Y");

    $count_graphs++;

    push @data, qq(
set title "Daily Glucose Summary for $title" font "Calibri,18"
set xlabel "Time" offset 0,-0.25
set ylabel "Blood glucose" 
set xtics left tc rgb "#000000"
set ytics 2    tc rgb "#000000"
set grid ytics ls 100 front

#set arrow from graph 0,first $min_glucose to graph 1,first $min_glucose ls 6 lw 2 nohead 
#set arrow from graph 0,first $max_glucose to graph 1,first $max_glucose ls 6 lw 2 nohead
set object 1 rect from graph 0, first $min_glucose to graph 1,first $max_glucose fc ls 6 fs solid 0.2 back

AVG = Mean$label
AVG_LABEL = gprintf("Average glucose: %.2f", AVG) 
set object 2 rect at graph 0.9, graph 0.9 fc ls 2 fs transparent solid 0.5 front size char strlen(AVG_LABEL), char 3
set label 2 AVG_LABEL at graph 0.9, graph 0.9 front center

#plot \$SmoothData$label using 1:2:( \$2 > $max_glucose || \$2 < $min_glucose ? 110 : $count_graphs ) with linespoints ls 120 lc variable 
plot \$SmoothData$label using (strftime("%H:%M:%S", \$1)):2:( \$2 > $max_glucose || \$2 < $min_glucose ? 110 : 1 ) with lines lw 3 lc variable 

undefine \$SmoothData$label

# Add an x grid
set multiplot previous
set title "  "
set xlabel " " offset 0,-0.25
set ylabel " " 
set xtics tc rgb "#ffffff00" 
set ytics tc rgb "#ffffff00"
unset grid
unset object 1
set grid xtics ls 101
plot 1/0
);

    if ( $count_graphs % $days_per_page == 0 && $count_graphs < $total_graphs ) {
        push @data, qq(unset multiplot);
        push @data, qq(set multiplot layout $days_per_page,1);
        $page_number++;
    }

}

# Output data averages by hour of the day
@sorted_lines = ();
push @data, qq(
\$DataAvg << EOD
"timestamp","blood glucose","meal","method","comment");
foreach my $row (@lines) {
    if ( $row =~ s#^"\d{4}-\d{2}-\d{2} #"#ms ) {
        push @sorted_lines, $row;
    }
}
push @data, sort @sorted_lines;
push @data, qq(EOD);

# Output min/max for each time interval
foreach my $row ( @sorted_lines ) {
    $row =~ s/"//g;
    my ( $time, $value ) = split /,/, $row;
    my ( $hour, $minute, $second ) = split /:/, $time;
    $time = sprintf( "%02d:%02d:00", $hour, int($minute/$interval)*$interval );
    
    # Override the current minimum values for this interval if it
    # exists; otherwise, set it
        if ( exists ( $intervals{$time}{min} ) ) {
        if ( $intervals{$time}{min} < $value ) {
            $intervals{$time}{min} = $value;
        }
    } else { 
        $intervals{$time}{min} = $value;
    }
    # Override the current maximum values for this interval if it
    # exists; otherwise, set it
        if ( exists ( $intervals{$time}{max} ) ) {
        if ( $intervals{$time}{max} > $value ) {
            $intervals{$time}{max} = $value;
        }
    } else { 
        $intervals{$time}{max} = $value;
    }

}
$Data::Dumper::Sortkeys = 1;
#die Dumper(\%intervals);

push @data, qq(
\$DataMaxMin << EOD
"timestamp","max","min");
foreach my $time  ( sort keys %intervals ) {
    warn $time;
    push @data, qq("$time","$intervals{$time}{max}","$intervals{$time}{min}");
}
push @data, qq(EOD);

# Standardise units for gnuplot's A1C calculations 
if ( $units =~ /mg/i ) {
    $units = 'mg/dL';
} elsif ( $units =~ /mmol/i ) {
    $units = 'mmol/L';
} else {
    $units = '';
}

push @data, qq(
reset
set datafile separator ","

set timefmt "%H:%M:%S"
set format x "%s" timedate
set format y "%.2f" numeric 
set samples 10000
set xdata
stats \$DataAvg using 2
MeanTotal = STATS_mean

set xdata time

set table \$DataAvgTable

#avg(x) = g
#min(x) = x<g
#max(x) = x>g
#f(x) = g
#fit f(x) \$DataAvg using 1:(\$2>MeanTotal?\$2:'') via t, g
#plot f(x) smooth mcsplines

#plot \$DataAvg using 1:(\$2>MeanTotal?\$2:'') every $count_graphs/2 lc 2, \$DataAvg using 1:(\$2<MeanTotal?\$2:'') every $count_graphs/2 lc 1

#samples(x) = \$0 > 4 ? 5 : (\$0+1)
#avg5(x) = (shift5(x), (back1+back2+back3+back4+back5)/samples(\$0))
#shift5(x) = (back5 = back4, back4 = back3, back3 = back2, back2 = back1, back1 = x)
## Initialize a running sum
#init(x) = (back1 = back2 = back3 = back4 = back5 = sum = 0)

#plot sum = init(0), \$DataAvg using 1:2 title 'data' lw 2 lc rgb 'forest-green', '' using 1:(avg5(\$2)) pt 7 ps 0.5 lw 1 lc rgb "blue"

plot \$DataAvg using 1:2 smooth mcsplines

unset table

set table \$SmoothDataAvg
plot \$DataAvg using 1:2 smooth bezier
unset table

undefine \$DataAvg

# Convert DataMaxMin from CSV to table
set table \$DataMaxMinTable
plot \$DataMaxMin using 1:2:3 with table
unset table

reset
set datafile separator whitespace

set key off
set style data lines
set xdata time
set timefmt "%H:%M:%S"
set format x "%H:%M" timedate
set format y "%.0f" numeric 

set yrange [0:$graph_max]
# If extended to 23:59, the x grid overlaps with the border
set xrange ["00:00":"23:58"]

set style line 100 dt 3 lw 1 lc rgb "#202020"
set style line 101 dt 1 lw 1 lc rgb "#202020"
set linetype 110 lc rgb "red"
set linetype 111 lc rgb "#B0B0B0"
set style fill transparent solid 0.5 noborder

set lmargin 12
set rmargin 10
set tmargin 5
set bmargin 5

set multiplot title layout $days_per_page,1 

set title "Average Daily Glucose" font "Calibri,18"
set xlabel "Time" offset 0,-0.25
set ylabel "Blood glucose" 
set xtics left tc rgb "#000000"
set ytics 2    tc rgb "#000000"
set grid ytics ls 100 front

set object 1 rect from graph 0, first $min_glucose to graph 1,first $max_glucose fc ls 6 fs solid 0.05 back

AVG = MeanTotal
AVG_LABEL = gprintf("Average glucose: %.2f", AVG) 
set object 2 rect at graph 0.9, graph 0.9 fc ls 2 fs transparent solid 0.5 front size char strlen(AVG_LABEL), char 3
set label 2 AVG_LABEL at graph 0.9, graph 0.9 front center

A1C = 0
if (A1C == 0 && '$units' eq 'mg/dL') {
    A1C = (MeanTotal + 46.7) / 28.7
}
if (A1C == 0 && '$units' eq 'mmol/L') {
    A1C = (MeanTotal + 2.59) / 1.59
}
# mg/dL numbers tend to be higher than 35
if (A1C == 0 && MeanTotal >= 35) { 
    A1C = (MeanTotal + 46.7) / 28.7
}
# mmol/L numbers tend to be lower than 35
if (A1C == 0 && MeanTotal < 35) { 
    A1C = (MeanTotal + 2.59) / 1.59
}

A1C_LABEL = gprintf("Average A1c: %.1f", A1C) 
set object 3 rect at graph 0.07, graph 0.9 fc ls 4 fs transparent solid 0.5 front size char strlen(A1C_LABEL), char 3
set label 3 A1C_LABEL at graph 0.07, graph 0.9 front center

#plot \$SmoothDataAvg using ( strftime("%H:%M:%S", \$1) ):2:( \$2 > $max_glucose || \$2 < $min_glucose ? 110 : 1 ) with lines lw 3 lc variable 
#plot \$DataAvgTable using (strftime("%H:%M:%S", \$1)):2 with points lc 5 ps 0.5 pt 37, \$SmoothDataAvg using (strftime("%H:%M:%S", \$1)):2:( \$2 > $max_glucose || \$2 < $min_glucose ? 110 : 1 ) with lines lw 3 lc variable 

plot \$DataMaxMinTable using (strftime("%H:%M:%S", \$1)):2:3 with filledcurves lc 111, \$SmoothDataAvg using (strftime("%H:%M:%S", \$1)):2:( \$2 > $max_glucose || \$2 < $min_glucose ? 110 : 1 ) with lines lw 3 lc variable 

undefine \$DataAvg
undefine \$DataMaxMin
undefine \$DataMaxMinTable
undefine \$SmoothDataAvg
undefine \$DataAvgTable

# Add an x grid
set multiplot previous
set title "  "
set xlabel " " offset 0,-0.25
set ylabel " " 
set xtics tc rgb "#ffffff00" 
set ytics tc rgb "#ffffff00"
unset grid
unset object 1
set grid xtics ls 101
plot 1/0

);

push @data, qq(
unset multiplot
test
);

# run the data through gnuplot
$gnuplot_data = join "\n", @data;
print $gnuplot_data;

open( my $ofh, '>', $output ) 
    or die "Could not open file '$output' $!";

my ( $pid, $stdin, $stdout, $stderr );
use Symbol 'gensym'; 
$stderr = gensym;

$pid = open3( $stdin, $stdout, $stderr, 'gnuplot' );

print $stdin $gnuplot_data;
close( $stdin );

while ( <$stdout> ) {
    print $ofh "$_";
}

while ( <$stderr> ) {
    warn $_;
}

close($stdout);
close($stderr);

waitpid( $pid, 0 );
my $child_exit_status = $? >> 8;

close( $ofh )
    or warn "close failed: $!";

#open(GNUPLOT, "|gnuplot");
#print GNUPLOT $gnuplot_data;
#close(GNUPLOT); 

# vim : set expandtab shiftwidth=4 softtabstop=4 tw=1000 :