--- loncom/interface/statistics/lonproblemstatistics.pm	2004/04/01 21:35:52	1.81
+++ loncom/interface/statistics/lonproblemstatistics.pm	2005/01/11 19:45:18	1.96
@@ -1,6 +1,6 @@
 # The LearningOnline Network with CAPA
 #
-# $Id: lonproblemstatistics.pm,v 1.81 2004/04/01 21:35:52 matthew Exp $
+# $Id: lonproblemstatistics.pm,v 1.96 2005/01/11 19:45:18 matthew Exp $
 #
 # Copyright Michigan State University Board of Trustees
 #
@@ -55,6 +55,7 @@ use Apache::loncommon();
 use Apache::lonhtmlcommon;
 use Apache::loncoursedata;
 use Apache::lonstatistics;
+use LONCAPA::lonmetadata();
 use Apache::lonlocal;
 use Spreadsheet::WriteExcel;
 use Apache::lonstathelpers();
@@ -95,6 +96,10 @@ my %SeqStat;    # keys are symbs, values
 ##                                      statistics display?
 ## selected        yes     (yes|no)     Is the column selected by default?
 ##
+## format          no      sprintf format string
+##
+## excel_format    no      excel format type 
+##                               (see &Apache::loncommon::define_excel_formats
 my @Fields = (
            { name => 'problem_num',
              title => 'P#',
@@ -226,6 +231,17 @@ my @Fields = (
              selectable => 'yes',
              defaultselected => 'yes',
            },
+           { name   => 'tries_per_correct',
+             title  => 'tries/correct',
+             align  => 'right',
+             color  => '#FFDDDD',
+             format => '%4.1f',
+             sortable  => 'yes',
+             graphable => 'yes',
+             long_title => 'Tries per Correct Answer',
+             selectable => 'yes',
+             defaultselected => 'yes',
+           },
            { name   => 'num_wrong',
              title  => '#Wrng',
              align  => 'right',
@@ -233,6 +249,17 @@ my @Fields = (
              format => '%4.1f',
              sortable  => 'yes',
              graphable => 'yes',
+             long_title => 'Number of students whose final answer is wrong',
+             selectable => 'yes',
+             defaultselected => 'yes',
+           },
+           { name   => 'per_wrong',
+             title  => '%Wrng',
+             align  => 'right',
+             color  => '#FFDDDD',
+             format => '%4.1f',
+             sortable  => 'yes',
+             graphable => 'yes',
              long_title => 'Percent of students whose final answer is wrong',
              selectable => 'yes',
              defaultselected => 'yes',
@@ -258,8 +285,85 @@ my @Fields = (
              graphable => 'yes',
              long_title => 'Degree of Discrimination',
              selectable => 'yes',
-             defaultselected => 'no',
+             defaultselected => 'yes',
            },
+##   duedate included for research purposes.  Commented out most of the time.
+#           { name => 'duedate',
+#             title => 'Due Date',
+#             align => 'left',
+#             color => '#FFFFFF',
+#             sortable => 'yes',
+#             graphable => 'no',
+#             long_title => 'Due date of resource for instructor',
+#             selectable => 'no',
+#             defaultselected => 'yes',
+#            },
+##   opendate included for research purposes.  Commented out most of the time.
+#           { name => 'opendate',
+#             title => 'Open Date',
+#             align => 'left',
+#             color => '#FFFFFF',
+#             sortable => 'yes',
+#             graphable => 'no',
+#             long_title => 'date resource became answerable',
+#             selectable => 'no',
+#             defaultselected => 'yes',
+#            },
+##   symb included for research purposes.  Commented out most of the time.
+#           { name => 'symb',
+#             title => 'Symb',
+#             align => 'left',
+#             color => '#FFFFFF',
+#             sortable => 'yes',
+#             graphable => 'no',
+#             long_title => 'Unique LON-CAPA identifier for problem',
+#             selectable => 'no',
+#             defaultselected => 'yes',
+#            },
+##   resptypes included for research purposes.  Commented out most of the time.
+#           { name => 'resptypes',
+#             title => 'Response Types',
+#             align => 'left',
+#             color => '#FFFFFF',
+#             sortable => 'no',
+#             graphable => 'no',
+#             long_title => 'Response Types used in this problem',
+#             selectable => 'no',
+#             defaultselected => 'yes',
+#            },
+##   maxtries included for research purposes.  Commented out most of the time.
+#           { name => 'maxtries',
+#             title => 'Maxtries',
+#             align => 'left',
+#             color => '#FFFFFF',
+#             sortable => 'no',
+#             graphable => 'no',
+#             long_title => 'Maximum number of tries',
+#             selectable => 'no',
+#             defaultselected => 'yes',
+#            },
+##   hinttries included for research purposes.  Commented out most of the time.
+#           { name => 'hinttries',
+#             title => 'hinttries',
+#             align => 'left',
+#             color => '#FFFFFF',
+#             sortable => 'no',
+#             graphable => 'no',
+#             long_title => 'Number of tries before a hint appears',
+#             selectable => 'no',
+#             defaultselected => 'yes',
+#            },
+##   weight included for research purposes.  Commented out most of the time.
+#           { name => 'weight',
+#             title => 'weight',
+#             align => 'right',
+#             color => '#FFFFFF',
+#             sortable => 'no',
+#             graphable => 'no',
+#             long_title => 'Problem weight (for instructor)',
+#             selectable => 'no',
+#             defaultselected => 'yes',
+#            },
 );
 
 my @SeqFields = (
@@ -461,6 +565,11 @@ sub parse_field_selection {
             }
         }
     }
+    #
+    # Always show all the sequence statistics (for now)
+    foreach my $field (@SeqFields) {
+        $field->{'selected'} = 'yes';
+    }
     return;
 }
 
@@ -493,6 +602,7 @@ select sections, maps, and output.
 ###############################################
 ###############################################
 sub CreateInterface {
+    my ($r) = @_;
     #
     &parse_field_selection();
     #
@@ -528,19 +638,18 @@ sub CreateInterface {
     $Str .= '</td><td>'.&field_selection_input();
     $Str .= '</td></tr>'."\n";
     $Str .= '</table>'."\n";
+    #
+    $Str .= '<p>'.&mt('Status: [_1]',
+                         '<input type="text" '.
+                         'name="stats_status" size="60" value="" />'
+                         ).
+                         '</nobr></p>';
+    #
     $Str .= '<input type="submit" name="GenerateStatistics" value="'.
         &mt('Generate Statistics').'" />';
     $Str .= '&nbsp;'x5;
     $Str .= 'Plot '.&plot_dropdown().('&nbsp;'x10);
-    $Str .= '<input type="submit" name="ClearCache" value="'.
-        &mt('Clear Caches').'" />';
-    $Str .= '&nbsp;'x5;
-    $Str .= '<input type="submit" name="UpdateCache" value="'.
-        &mt('Update Student Data').'" />';
-    $Str .= '&nbsp;'x5;
-    $Str .= '<input type="submit" name="Excel" value="'.
-        &mt('Produce Excel Output').'" />';
-    $Str .= '&nbsp;'x5;
+    #
     return $Str;
 }
 
@@ -578,13 +687,20 @@ sub BuildProblemStatisticsPage {
     undef(%SeqStat);
     #
     # Finally let the user know we are here
-    my $interface = &CreateInterface();
+    my $interface = &CreateInterface($r);
     $r->print($interface);
     $r->print('<input type="hidden" name="sortby" value="'.$ENV{'form.sortby'}.
               '" />');
     #
-    if (! exists($ENV{'form.statsfirstcall'})) {
-        $r->print('<input type="hidden" name="statsfirstcall" value="yes" />');
+    my @CacheButtonHTML = 
+        &Apache::lonstathelpers::manage_caches($r,'Statistics','stats_status');
+    my $Str;
+    foreach my $html (@CacheButtonHTML) {
+        $Str.=$html.('&nbsp;'x5);
+    }
+    #
+    $r->print($Str);
+    if (! exists($ENV{'form.firstrun'})) {
         $r->print('<h3>'.
                   &mt('Press "Generate Statistics" when you are ready.').
                   '</h3><p>'.
@@ -593,13 +709,6 @@ sub BuildProblemStatisticsPage {
                       ' will not have this delay.').
                   '</p>');
         return;
-    } elsif ($ENV{'form.statsfirstcall'} eq 'yes' || 
-             exists($ENV{'form.UpdateCache'}) ||
-             exists($ENV{'form.ClearCache'}) ) {
-        $r->print('<input type="hidden" name="statsfirstcall" value="no" />');
-        &Apache::lonstatistics::Gather_Student_Data($r);
-    } else {
-        $r->print('<input type="hidden" name="statsfirstcall" value="no" />');
     }
     $r->rflush();
     #
@@ -609,7 +718,10 @@ sub BuildProblemStatisticsPage {
     #
     if (exists($ENV{'form.Excel'})) {
         &Excel_output($r);
-    } else {
+    } else { 
+        $r->print('<input type="submit" name="Excel" value="'.
+                  &mt('Produce Excel Output').'" />'.'&nbsp;'x5);
+        $r->rflush();
         my $count = 0;
         foreach my $seq (&Apache::lonstatistics::Sequences_with_Assess()) {
             $count += $seq->{'num_assess_parts'};
@@ -647,7 +759,9 @@ sub BuildProblemStatisticsPage {
 sub output_sequence_statistics {
     my ($r) = @_;
     my $c=$r->connection();
-    $r->print('<h2>'.&mt('Sequence Statistics').'</h2>');
+    $r->print('<h2>'.&mt('Sequence Statistics').
+	      &Apache::loncommon::help_open_topic('Statistics_Sequence').
+	      '</h2>');
     $r->print('<table border="0"><tr><td bgcolor="#777777">'."\n".
               '<table border="0" cellpadding="3">'."\n".
               '<tr bgcolor="#FFFFE6">');
@@ -760,7 +874,7 @@ sub statistics_html_table_data {
         if (exists($field->{'special'}) && $field->{'special'} eq 'link') {
             $row .= '<a href="'.$data->{$field->{'name'}.'.link'}.'">';
         }
-        if (exists($field->{'format'})) {
+        if (exists($field->{'format'}) && $data->{$field->{'name'}} !~ /[A-Z]/i) {
             $row .= sprintf($field->{'format'},$data->{$field->{'name'}});
         } else {
             $row .= $data->{$field->{'name'}};
@@ -818,15 +932,9 @@ sub sequence_html_header {
 sub sequence_html_output {
     my ($seq) = @_;
     my $data = $SeqStat{$seq->{'symb'}};
-#    $SeqStat{$symb}->{'max'}
-#    $SeqStat{$symb}->{'min'}
-#    $SeqStat{$symb}->{'mean'}
-#    $SeqStat{$symb}->{'std'}
-#    $SeqStat{$symb}->{'count'}
-#    $SeqStat{$symb}->{'max_possible'}
     my $row = '<tr>';
     foreach my $field (@SeqFields) {
-#        next if ($field->{'selected'} ne 'yes');
+        next if ($field->{'selected'} ne 'yes');
         $row .= '<td bgcolor="'.$field->{'color'}.'"';
         if (exists($field->{'align'})) {
             $row .= ' align="'.$field->{'align'}.'"';
@@ -1184,6 +1292,8 @@ sub Excel_output {
     }
     my $excel_sheet = $excel_workbook->addworksheet(
         &Apache::loncommon::clean_excel_name($sheetname));
+    #
+    my $format = &Apache::loncommon::define_excel_formats($excel_workbook);
     ##
     ## Begin creating excel sheet
     ##
@@ -1191,28 +1301,17 @@ sub Excel_output {
     #
     # Put the course description in the header
     $excel_sheet->write($rows_output,$cols_output++,
-                   $ENV{'course.'.$ENV{'request.course.id'}.'.description'});
+                   $ENV{'course.'.$ENV{'request.course.id'}.'.description'},
+                        $format->{'h1'});
     $cols_output += 3;
     #
     # Put a description of the sections listed
     my $sectionstring = '';
-    my @Sections = @Apache::lonstatistics::SelectedSections;
-    if (scalar(@Sections) > 1) {
-        if (scalar(@Sections) > 2) {
-            my $last = pop(@Sections);
-            $sectionstring = "Sections ".join(', ',@Sections).', and '.$last;
-        } else {
-            $sectionstring = "Sections ".join(' and ',@Sections);
-        }
-    } else {
-        if ($Sections[0] eq 'all') {
-            $sectionstring = "All sections";
-        } else {
-            $sectionstring = "Section ".$Sections[0];
-        }
-    }
-    $excel_sheet->write($rows_output,$cols_output++,$sectionstring);
-    $cols_output += scalar(@Sections);
+    $excel_sheet->write($rows_output,$cols_output++,
+                        &Apache::lonstathelpers::sections_description
+                            (@Apache::lonstatistics::SelectedSections),
+                        $format->{'h3'});
+    $cols_output += scalar(@Apache::lonstatistics::SelectedSections);
     #
     # Time restrictions
     my $time_string;
@@ -1228,6 +1327,10 @@ sub Excel_output {
         # See note above about lonlocal:locallocaltime
         $time_string .= 'Data collected before '.localtime($endtime).'.';
     }
+    if (defined($time_string)) {
+        $excel_sheet->write($rows_output,$cols_output++,$time_string);
+        $cols_output+= 5;
+    }
     #
     # Put the date in there too
     $excel_sheet->write($rows_output,$cols_output++,
@@ -1235,38 +1338,49 @@ sub Excel_output {
     #
     $rows_output++; 
     $cols_output=0;
-    #
-    # Long Headers
-    foreach my $field (@Fields) {
-        next if ($field->{'name'} eq 'problem_num');
-        next if ($field->{'selected'} ne 'yes');
-        if (exists($field->{'long_title'})) {
+    ##
+    ## Sequence Statistics
+    ## 
+    &write_headers($excel_sheet,$format,\$rows_output,\$cols_output,
+                   \@SeqFields);
+    foreach my $seq (&Apache::lonstatistics::Sequences_with_Assess()) {
+        next if ($seq->{'num_assess'} < 1);
+        my $data = $SeqStat{$seq->{'symb'}};
+        $cols_output=0;
+        foreach my $field (@SeqFields) {
+            next if ($field->{'selected'} ne 'yes');
+            my $fieldformat = undef;
+            if (exists($field->{'excel_format'})) {
+                $fieldformat = $format->{$field->{'excel_format'}};
+            }
             $excel_sheet->write($rows_output,$cols_output++,
-                                $field->{'long_title'});
-        } else {
-            $excel_sheet->write($rows_output,$cols_output++,'');
+                                $data->{$field->{'name'}},$fieldformat);
         }
+        $rows_output++;
+        $cols_output=0;
     }
+    ##
+    ## Resource Statistics
+    ##
     $rows_output++;
     $cols_output=0;
-    # Brief headers
-    foreach my $field (@Fields) {
-        next if ($field->{'selected'} ne 'yes');
-        next if ($field->{'name'} eq 'problem_num');
-        # Use english for excel as I am not sure how well excel handles 
-        # other character sets....
-        $excel_sheet->write($rows_output,$cols_output++,$field->{'title'});
-    }
-    $rows_output++;
+    &write_headers($excel_sheet,$format,\$rows_output,\$cols_output,
+                   \@Fields);
+    #
     foreach my $data (@StatsArray) {
         $cols_output=0;
         foreach my $field (@Fields) {
             next if ($field->{'selected'} ne 'yes');
             next if ($field->{'name'} eq 'problem_num');
+            my $fieldformat = undef;
+            if (exists($field->{'excel_format'})) {
+                $fieldformat = $format->{$field->{'excel_format'}};
+            }
             $excel_sheet->write($rows_output,$cols_output++,
-                                $data->{$field->{'name'}});
+                                $data->{$field->{'name'}},$fieldformat);
         }
         $rows_output++;
+        $cols_output=0;
     }
     #
     $excel_workbook->close();
@@ -1279,6 +1393,44 @@ sub Excel_output {
     return;
 }
 
+##
+## &write_headers
+##
+sub write_headers {
+    my ($excel_sheet,$format,$rows_output,$cols_output,$Fields) = @_;
+    ##
+    ## First the long titles
+    foreach my $field (@{$Fields}) {
+        next if ($field->{'name'} eq 'problem_num');
+        next if ($field->{'selected'} ne 'yes');
+        if (exists($field->{'long_title'})) {
+            $excel_sheet->write($$rows_output,${$cols_output},
+                                $field->{'long_title'},
+                                $format->{'bold'});
+        } else {
+            $excel_sheet->write($$rows_output,${$cols_output},'');
+        }
+        ${$cols_output}+= 1;
+    }
+    ${$cols_output} =0;
+    ${$rows_output}+=1;
+    ##
+    ## Then the short titles
+    foreach my $field (@{$Fields}) {
+        next if ($field->{'selected'} ne 'yes');
+        next if ($field->{'name'} eq 'problem_num');
+        # Use english for excel as I am not sure how well excel handles 
+        # other character sets....
+        $excel_sheet->write($$rows_output,$$cols_output,
+                            $field->{'title'},
+                            $format->{'bold'});
+        $$cols_output+=1;
+    }
+    ${$cols_output} =0;
+    ${$rows_output}+=1;
+    return;
+}
+
 ##################################################
 ##################################################
 ##
@@ -1292,6 +1444,7 @@ sub compute_statistics_on_sequence {
     foreach my $res (@{$seq->{'contents'}}) {
         next if ($res->{'type'} ne 'assessment');
         foreach my $part (@{$res->{'parts'}}) {
+            next if ($res->{'partdata'}->{$part}->{'Survey'});
             #
             # This is where all the work happens
             my $data = &get_statistics($seq,$res,$part,scalar(@StatsArray)+1);
@@ -1312,6 +1465,7 @@ sub compute_all_statistics {
     foreach my $seq (&Apache::lonstatistics::Sequences_with_Assess()) {
         last if ($c->aborted);
         next if ($seq->{'num_assess'} < 1);
+        &compute_sequence_statistics($seq);
         &compute_statistics_on_sequence($seq);
     }
 }
@@ -1404,6 +1558,7 @@ sub get_statistics {
                         (\@Apache::lonstatistics::SelectedSections,
                          $Apache::lonstatistics::enrollment_status,
                          $symb,$part,$courseid,$starttime,$endtime);
+    $data->{'symb'}        = $symb;
     $data->{'part'}        = $part;
     $data->{'problem_num'} = $problem_num;
     $data->{'container'}   = $sequence->{'title'};
@@ -1415,6 +1570,37 @@ sub get_statistics {
         $data->{'deg_of_disc'} = 
             &compute_discrimination_factor($resource,$part,$sequence);
     }
+    #
+    # Store in metadata if computations were done for all students
+    if ($data->{'num_students'} > 1) {
+        my @Sections = @Apache::lonstatistics::SelectedSections;
+        my $sections = '"'.join(' ',@Sections).'"';
+        $sections =~ s/&+/_/g;  # Ensure no special characters
+        $data->{'sections'}=$sections;
+        $data->{'course'} = $ENV{'request.course.id'};
+        my $urlres=(&Apache::lonnet::decode_symb($resource->{'symb'}))[2];
+        $data->{'urlres'}=$urlres;
+        my %storestats = 
+            &LONCAPA::lonmetadata::dynamic_metadata_storage($data);
+        my ($dom,$user) = $urlres=~/^(\w+)\/(\w+)/; 
+        &Apache::lonnet::put('nohist_resevaldata',\%storestats,$dom,$user);
+    }
+    #
+    $data->{'tries_per_correct'} = $data->{'tries'} / 
+        ($data->{'num_solved'}+0.1);
+    #
+    # Get the due date for research purposes (commented out most of the time)
+#    $data->{'duedate'} = 
+#        &Apache::lonnet::EXT('resource.'.$part.'.duedate',$symb);
+#    $data->{'opendate'} = 
+#        &Apache::lonnet::EXT('resource.'.$part.'.opendate',$symb);
+#    $data->{'maxtries'} = 
+#        &Apache::lonnet::EXT('resource.'.$part.'.maxtries',$symb);
+#    $data->{'hinttries'} =
+#        &Apache::lonnet::EXT('resource.'.$part.'.hinttries',$symb);
+#    $data->{'weight'} =
+#        &Apache::lonnet::EXT('resource.'.$part.'.weight',$symb);
+#    $data->{'resptypes'} = join(',',@{$resource->{'partdata'}->{$part}->{'ResponseTypes'}});
     return $data;
 }
 
@@ -1442,11 +1628,13 @@ sub compute_discrimination_factor {
     }
     #
     # rank
+    my ($starttime,$endtime) = &Apache::lonstathelpers::get_time_limits();
     my $ranking = 
         &Apache::loncoursedata::rank_students_by_scores_on_resources
         (\@Resources,
          \@Apache::lonstatistics::SelectedSections,
-         $Apache::lonstatistics::enrollment_status,undef);
+         $Apache::lonstatistics::enrollment_status,undef,
+         $starttime,$endtime);
     #
     # compute their percent scores on the problems in the sequence,
     my $number_to_grab = int(scalar(@{$ranking})/4);
@@ -1457,10 +1645,16 @@ sub compute_discrimination_factor {
         map { 
             $_->[&Apache::loncoursedata::RNK_student()]; 
           } @{$ranking}[($num_students-$number_to_grab)..($num_students-1)];
+    if (! @BottomSet || (@BottomSet == 1 && $BottomSet[0] eq '') ||
+        ! @TopSet    || (@TopSet    == 1 && $TopSet[0]    eq '')) {
+        return 'nan';
+    }
     my ($bottom_sum,$bottom_max) = 
-        &Apache::loncoursedata::get_sum_of_scores($resource,$part,\@BottomSet);
+        &Apache::loncoursedata::get_sum_of_scores($resource,$part,\@BottomSet,
+                                                  undef,$starttime,$endtime);
     my ($top_sum,$top_max) = 
-        &Apache::loncoursedata::get_sum_of_scores($resource,$part,\@TopSet);
+        &Apache::loncoursedata::get_sum_of_scores($resource,$part,\@TopSet,
+                                                  undef,$starttime,$endtime);
     my $deg_of_disc;
     if ($top_max == 0 || $bottom_max==0) {
         $deg_of_disc = 'nan';
@@ -1530,10 +1724,10 @@ sub compute_sequence_statistics {
     $SeqStat{$symb}->{'countmax'} = $cmax;
     $SeqStat{$symb}->{'countmin'} = $cmin;
     $SeqStat{$symb}->{'countstd'} = $cSTD;
+    $SeqStat{$symb}->{'countmean'} = $cMean;
     $SeqStat{$symb}->{'count'} = $ccount;
     $SeqStat{$symb}->{'items'} = $K;
     $SeqStat{$symb}->{'KR-21'}=$kr_21;
-
     return;
 }