#!/usr/bin/perl
# Reads in grades on exam problems and spits out stats formatted as an
# html table.  Each line consists of the grades for a single student (tab
# separated).  The first line consists of the headers; the second line is
# the maximum score for each column.  Use this by copying & pasting the
# gradebook entries from Excel or OpenOffice into a textfile (it should
# automatically tab-delimit them) and then pointing the script at the file.
# This assumes you are storing each student as a row and each grade item
# as a column.


use lib "/home/math/gbutler/lib/perl5";
use Statistics::Descriptive;

$indent = 4 x " ";

$headerLine = <>;
chomp($headerLine);
@headerList = split(/\t/,$headerLine);

$maxLine = <>;
chomp($maxLine);
@maxList = split(/\t/,$maxLine);

@stud = ();
while(<>)  {
    chomp;
    @line = split(/\t/);
    push(@stud,[@line]);
}


@stat = ();
@notAttempted = ();
for($i = 0; $i <= $#headerList; $i++)  { #assumes all lines have the same numer of fields, expect runtime errors otherwise
    @gradeCol = (); #list of all student's grades for a single gradebook entry
    $numNotSubmitted[$i] = 0;
    for($j = 0; $j <= $#stud; $j++)  {
        if(defined($stud[$j][$i]) && $stud[$j][$i] ne "")  {
            push(@gradeCol,$stud[$j][$i]);
        }
        else  {
            $numNotSubmitted[$i]++;
        }
    }
    push(@stat,Statistics::Descriptive::Full->new());
    $stat[$#stat]->add_data(@gradeCol);
}

print "<table border=1>\n";
print $indent . "<tr><th>&nbsp;</th><th>Points Possible</th><th>A+ (>100)</th><th>A (90-100)</th><th>A- (87-89)</th><th>B+ (83-86)</th><th>B (80-82)</th><th>B- (77-79)</th><th>C+ (73-76)</th><th>C (70-72)</th><th>C- (67-69)</th><th>D+ (63-66)</th><th>D (60-62)</th><th>D- (57-59)</th><th>F (0-56)</th><th>Not Submitted</th><th>Mean</th><th>Standard Deviation</th><th>Max</th><th>Q3</th><th>Median</th><th>Q1</th><th>Min</th><th></tr>\n";

for($i = 0; $i <= $#stat; $i++)  {
    $stat[$i]->sort_data();
    @gradeCol = $stat[$i]->get_data();

    $indf  =  0;
    $inddm = -1;
    $indd  = -1;
    $inddp = -1;
    $indcm = -1;
    $indc  = -1;
    $indcp = -1;
    $indbm = -1;
    $indb  = -1;
    $indbp = -1;
    $indam = -1;
    $inda  = -1;
    $indap = -1;

    for($j = 0; $j <= $#gradeCol; $j++)  {
        if($inddm == -1 && $gradeCol[$j] >= 0.57*$maxList[$i])  { $inddm = $j; }
        if($indd  == -1 && $gradeCol[$j] >= 0.60*$maxList[$i])  { $indd  = $j; }
        if($inddp == -1 && $gradeCol[$j] >= 0.63*$maxList[$i])  { $inddp = $j; }
        if($indcm == -1 && $gradeCol[$j] >= 0.67*$maxList[$i])  { $indcm = $j; }
        if($indc  == -1 && $gradeCol[$j] >= 0.70*$maxList[$i])  { $indc  = $j; }
        if($indcp == -1 && $gradeCol[$j] >= 0.73*$maxList[$i])  { $indcp = $j; }
        if($indbm == -1 && $gradeCol[$j] >= 0.77*$maxList[$i])  { $indbm = $j; }
        if($indb  == -1 && $gradeCol[$j] >= 0.80*$maxList[$i])  { $indb  = $j; }
        if($indbp == -1 && $gradeCol[$j] >= 0.83*$maxList[$i])  { $indbp = $j; }
        if($indam == -1 && $gradeCol[$j] >= 0.87*$maxList[$i])  { $indam = $j; }
        if($inda  == -1 && $gradeCol[$j] >= 0.90*$maxList[$i])  { $inda  = $j; }
        if($indap == -1 && $gradeCol[$j] >  $maxList[$i])       { $indap = $j; }
    }

    $numf  = ($indf  >= 0) ? (($inddm >= 0) ? ($inddm - $indf ) : (1 + $#gradeCol - $indf )) : 0;
    $numdm = ($inddm >= 0) ? (($indd  >= 0) ? ($indd  - $inddm) : (1 + $#gradeCol - $inddm)) : 0;
    $numd  = ($indd  >= 0) ? (($inddp >= 0) ? ($inddp - $indd ) : (1 + $#gradeCol - $indd )) : 0;
    $numdp = ($inddp >= 0) ? (($indcm >= 0) ? ($indcm - $inddp) : (1 + $#gradeCol - $inddp)) : 0;
    $numcm = ($indcm >= 0) ? (($indc  >= 0) ? ($indc  - $indcm) : (1 + $#gradeCol - $indcm)) : 0;
    $numc  = ($indc  >= 0) ? (($indcp >= 0) ? ($indcp - $indc ) : (1 + $#gradeCol - $indc )) : 0;
    $numcp = ($indcp >= 0) ? (($indbm >= 0) ? ($indbm - $indcp) : (1 + $#gradeCol - $indcp)) : 0;
    $numbm = ($indbm >= 0) ? (($indb  >= 0) ? ($indb  - $indbm) : (1 + $#gradeCol - $indbm)) : 0;
    $numb  = ($indb  >= 0) ? (($indbp >= 0) ? ($indbp - $indb ) : (1 + $#gradeCol - $indb )) : 0;
    $numbp = ($indbp >= 0) ? (($indam >= 0) ? ($indam - $indbp) : (1 + $#gradeCol - $indbp)) : 0;
    $numam = ($indam >= 0) ? (($inda  >= 0) ? ($inda  - $indam) : (1 + $#gradeCol - $indam)) : 0;
    $numa  = ($inda  >= 0) ? (($indap >= 0) ? ($indap - $inda ) : (1 + $#gradeCol - $inda )) : 0;
    $numap = ($indap >= 0) ? (1 + $#gradeCol - $indap) : 0;

    ($p75,$p75i)=$stat[$i]->percentile(75);
    ($p25,$p25i)=$stat[$i]->percentile(25);
    $p25 = $stat[$i]->median()  if($p25 == "");
    $p75 = $stat[$i]->median()  if($p75 == "");

    print $indent . "<tr><th>" . $headerList[$i] . "</th><td>" . join("</td><td>",($maxList[$i],$numap,$numa,$numam,$numbp,$numb,$numbm,$numcp,$numc,$numcm,$numdp,$numd,$numdm,$numf,$numNotSubmitted[$i],sprintf("%.3lf",$stat[$i]->mean()),sprintf("%.3lf",$stat[$i]->standard_deviation()),$stat[$i]->max(),$p75,$stat[$i]->median(),$p25,$stat[$i]->min())) . "</td></tr>\n";
}
print "</table>\n";