#!/usr/local/bin/perl

###############################################################################
# Program     : buildInfo
# Author      : Terry Farrah and Zhi Sun (tfarrah@systemsbiology.org)
# $Id: main.cgi 5994 2009-03-06 08:01:38Z dcampbel $
#
# Description : This script displays details about all default Atlas builds
#
# SBEAMS is Copyright (C) 2000-2021 Institute for Systems Biology
# This program is governed by the terms of the GNU General Public License (GPL)
# version 2 as published by the Free Software Foundation.  It is provided
# WITHOUT ANY WARRANTY.  See the full description of GPL terms in the
# LICENSE file distributed with this software.
#
###############################################################################


###############################################################################
# Get the script set up with everything it will need
###############################################################################
use strict;
use vars qw ($q $sbeams $sbeamsMOD $PROG_NAME
             $current_contact_id $current_username);
use lib qw (../../lib/perl);
use CGI::Carp qw(fatalsToBrowser croak);
use Data::Dumper;

use SBEAMS::Connection qw($q $log);
use SBEAMS::Connection::DataTable;
use SBEAMS::Connection::Settings;
use SBEAMS::Connection::Tables;
use SBEAMS::Connection::TabMenu;

use SBEAMS::PeptideAtlas;
use SBEAMS::PeptideAtlas::Settings;
use SBEAMS::PeptideAtlas::Tables;

$sbeams = new SBEAMS::Connection;
$sbeamsMOD = new SBEAMS::PeptideAtlas;
$sbeamsMOD->setSBEAMS($sbeams);


###############################################################################
# Global Variables
###############################################################################
$PROG_NAME = 'buildInfo';
main();


###############################################################################
# Main Program:
#
# Call $sbeams->Authentication and stop immediately if authentication
# fails else continue.
###############################################################################
sub main
{
    #### Do the SBEAMS authentication and exit if a username is not returned
    exit unless ($current_username = $sbeams->Authenticate(
        permitted_work_groups_ref=>['PeptideAtlas_user','PeptideAtlas_admin',
        'PeptideAtlas_readonly', 'PeptideAtlas_exec'],
        #connect_read_only=>1,
        allow_anonymous_access=>1,
    ));


    #### Read in the default input parameters
    my %parameters;
    my $n_params_found = $sbeams->parse_input_parameters(
        q=>$q,
        parameters_ref=>\%parameters
        );


    ## get project_id to send to HTMLPrinter display
    my $project_id = $sbeamsMOD->getProjectID(
        atlas_build_name => $parameters{atlas_build_name},
        atlas_build_id => $parameters{atlas_build_id}
        );


    #### Process generic "state" parameters before we start
    $sbeams->processStandardParameters(parameters_ref=>\%parameters);
#    $sbeams->printCGIParams($q);


    #### Decide what action to take based on information so far
    if ($parameters{action} eq "???") {

        # Some action
 
    } else {

      # zhi says maybe make a new version of this subroutine to put in new style sheet
      if ($sbeams->output_mode() =~ /tsv/i ) {
        my $header = $sbeams->get_http_header( mode => 'tsv', filename => 'build_information_table.tsv'  );
        print $header;
        handle_request(ref_parameters=>\%parameters);
      } else {
        $sbeamsMOD->display_page_header(project_id => $project_id);
        handle_request(ref_parameters=>\%parameters);
        $sbeamsMOD->display_page_footer();
      }

    }




} # end main


###############################################################################
# Show the page
###############################################################################
sub handle_request {
  my %args = @_;

    #### Process the arguments list
  my $ref_parameters = $args{'ref_parameters'}
        || die "ref_parameters not passed";

  my %parameters = %{$ref_parameters};


  #### Get the current atlas_build_id based on parameters or session
  my $atlas_build_id = $sbeamsMOD->getCurrentAtlasBuildID(
    parameters_ref => \%parameters,
  );
  if (defined($atlas_build_id) && $atlas_build_id < 0) {
    #### Don't return. Let the user pick from a valid one.
    #return;
  }


  #### Get the HTML to display the tabs
  my $tabMenu = $sbeamsMOD->getTabMenu(
    parameters_ref => \%parameters,
    program_name => $PROG_NAME,
  );
  if ($sbeams->output_mode() eq 'html') {
    print $tabMenu->asHTML() if ($sbeams->output_mode() eq 'html');
    print "<BR>\n";
  }


    #### Read in the standard form values
    my $apply_action  = $parameters{'action'} || $parameters{'apply_action'};
    my $TABLE_NAME = $parameters{'QUERY_NAME'};


    #### Set some specific settings for this program
    my $PROGRAM_FILE_NAME = $PROG_NAME;
    my $base_url = "$CGI_BASE_DIR/$SBEAMS_SUBDIR/$PROGRAM_FILE_NAME";
    my $help_url = "$CGI_BASE_DIR/help_popup.cgi";
    my $ppt = $parameters{'ppt'} || 0;
    my $table_width = $parameters{'width'} || 800;


    #### Get a list of accessible project_ids
    my @accessible_project_ids = $sbeams->getAccessibleProjects();
    my $accessible_project_ids = join( ",", @accessible_project_ids ) || '0';

    #### Get a list of accessible atlas builds
    my $sql = qq~
    SELECT AB.atlas_build_id, atlas_build_name, default_atlas_build_id
    FROM $TBAT_ATLAS_BUILD AB
    LEFT JOIN $TBAT_DEFAULT_ATLAS_BUILD DAB 
    ON DAB.atlas_build_id = AB.atlas_build_id
    WHERE AB.project_id IN ( $accessible_project_ids )
	  --AND AB.atlas_build_id IN ( SELECT DISTINCT atlas_build_id FROM $TBAT_PEPTIDE_INSTANCE )
	  AND ( DAB.record_status IS NULL OR DAB.record_status != 'D' )
	  AND AB.record_status != 'D'
	  AND NOT ( DAB.organism_id IS NULL AND default_atlas_build_id IS NOT NULL ) -- keep global default from showing up 2x
    ~;
    my @atlas_builds = $sbeams->selectSeveralColumns($sql);
    my $atlas_atlas_build_id_output_idx = 0;
    my $atlas_build_name_idx = 1;
    my $def_atlas_build_id_output_idx = 2;

    # Get name of current (default) atlas build
    my $default_build_name = '';
    foreach my $atlas_build ( @atlas_builds ) {
      if ($atlas_build->[$atlas_atlas_build_id_output_idx] == $atlas_build_id) {
	$default_build_name = $atlas_build->[$atlas_build_name_idx];
      }
    }

  # Make hash out of accessible atlas builds
  my %atlas_build_hash;
  foreach my $atlas_build ( @atlas_builds ) {
    my $id = $atlas_build->[$atlas_atlas_build_id_output_idx];
    $atlas_build_hash{$id} = $atlas_build;
  }

  my %accessible_default_builds;
  my $adb_sql = qq~
  SELECT AB.atlas_build_id, atlas_build_name 
  FROM $TBAT_DEFAULT_ATLAS_BUILD DAB 
  JOIN $TBAT_ATLAS_BUILD AB ON AB.atlas_build_id = DAB.atlas_build_id
  WHERE organism_specialized_build IS NULL
  AND DAB.organism_id IS NOT NULL
  ~;
  my $sth = $sbeams->get_statement_handle( $adb_sql );
  while ( my @row = $sth->fetchrow_array() ) {
    $accessible_default_builds{$row[0]} = $row[1];
  }

  my $table = SBEAMS::Connection::DataTable->new();
#  print $table->asTSV();

  my $is_html = ( $sbeams->output_mode =~ /html/i ) ? 1 : 0;

  #### If the output_mode is HTML, then display the form
  if ( 1 ) {

    print qq~
      <script LANGUAGE="Javascript">
        function switchAtlasBuild() {
          document.AtlasBuildList.apply_action.value = "GO";
          document.AtlasBuildList.submit();
        }
      document.title = 'PeptideAtlas: Build Stats & Lists';
      </script>
    ~ if $is_html;

    print $q->start_form(-method=>"POST",
                         -action=>"$base_url",
			     -name=>"AtlasBuildList",
                            ) if $is_html;

    unless ($default_build_name) {
      $default_build_name = qq~<FONT COLOR="red"> - NONE - </FONT>~;
    }
    my ( $tr, $link ) = $sbeams->make_table_toggle( name    => 'atlas_build_select',
                                                  visible => 0,
                                                  tooltip => 'Show/Hide Section',
                                                  imglink => 1,
                                                  textlink => 1,
                                                  tr_asref => 1,
                                                  hidetext => 'View default builds',
                                                  showtext => 'View all builds',
                                                  sticky  => 1 );
    my $earl = $q->self_url();
    $earl .= '?output_mode=tsv';
    my $download_link = "Download table as a <a href='$earl'> TSV file </a>\n";
    if ( !$ppt ) {
      print qq~
    <div style='padding-left:25px;margin-top:10px;margin-bottom:10px;'>
    Number of proteins returned on click may differ from tallies shown here:
    <li>Protein tallies exclude decoy, unmapped, and some contaminant proteins.</li>
    <li>Swiss-Prot click returns only 6-char core identifiers, not splice variants.</li>
    See <i>Protein ID terms</i> in sidebar menu for nomenclature.  $download_link
    </div>
    ~ if $is_html;
    } else {
      print "<p>To hide any row, click its Build Name. To reset, reload page. </p>\n" if $is_html;
    }


  ### Construct headers for info table

  my $this_row = $table->getRowNum();

  ### IF YOU CHANGE THE INDICES BELOW, CHANGE IN EXACTLY THE SAME WAY
  ### IN BuildInfo.pm
  my $build_name_idx = 0;
  my $org_idx = 1;
  my $peptide_inclusion_idx = 2;
  my $smpl_count_idx = 3;
  my $spectra_searched_idx = 4;
  my $psm_count_idx = 5;
  my $distinct_peps_idx = 6;
  my $n_canonicals_idx = 7;
  my $n_canon_dist_idx = 8;
  my $n_disting_prots_idx = 9;
  my $n_seq_unique_prots_idx = 10;
  my $n_swiss_idx = 11;
  my $n_covering_idx = 12;
  my $descr_idx = 13;
  my $atlas_build_id_output_idx = 14;
  my $ncols = 15;

  my @ppt_cols;
  my $build_name_ppt_idx;
  my $peptide_inclusion_ppt_idx;
  my $smpl_count_ppt_idx;
  my $spectra_searched_ppt_idx;
  my $psm_count_ppt_idx;
  my $distinct_peps_ppt_idx;
  my $n_canonicals_ppt_idx;
  my $n_swiss_ppt_idx;
  my $descr_ppt_idx;
  my $ppt_ncols;
  if ($ppt) {
    @ppt_cols = (0,2,3,4,5,6,7,11);
    $build_name_ppt_idx = 0;
    $peptide_inclusion_ppt_idx = 1;
    $smpl_count_ppt_idx = 2;
    $spectra_searched_ppt_idx = 3;
    $psm_count_ppt_idx = 4;
    $distinct_peps_ppt_idx = 5;
    $n_canonicals_ppt_idx = 6;
    $n_swiss_ppt_idx = 7;
    $ppt_ncols = 8;
#--------------------------------------------------
#   # for testing
#     @ppt_cols = (0,2,3,6,7,11);
#     $build_name_ppt_idx = 0;
#     $peptide_inclusion_ppt_idx = 1;
#     $smpl_count_ppt_idx = 2;
#     $distinct_peps_ppt_idx = 3;
#     $n_canonicals_ppt_idx = 4;
#     $n_swiss_ppt_idx = 5;
#     $ppt_ncols = 6;
#-------------------------------------------------- 
  }


  #Open and read the cached .tsv file.
  my $tsv_file = "$PHYSICAL_BASE_DIR/tmp/buildInfo.tsv";

  open (TSV, $tsv_file) || die "$PROG_NAME: cannot open $tsv_file for reading.";
  my $header = <TSV>;
  chomp $header;
  my @headers = split("\t", $header);
  # Remove unwanted headers for ppt
  if ($ppt) {
    my @ppt_headers=();
    for my $col (@ppt_cols) {
      my $header = $headers[$col];
      if (($header =~ /spectra searched/i) ||
          ($header =~ /observations/i) ||
          ($header =~ /psms/i) ||
          ($header =~ /peptide spectrum matches/i)) {
	$header .= " (1000's)";
      }
      push (@ppt_headers, $header);
    }
    @headers = @ppt_headers;
  } else {
    if ( $parameters{protease_info} ) {
      push @headers, 'Build Enzyme Usage';
    }
  }

  # Define header row
  $table ->addRow( \@headers );
  # setRowAttr does not take a COLS arg.
  $table->setRowAttr( ROWS => [1], BGCOLOR => '#002664', ALIGN=>'CENTER', NOWRAP => 0, HEIGHT => 35 );
  $table->setHeaderAttr(WHITE_TEXT=>1,BOLD=>1);   # this actually has effect via sub formatHeader

  # Read in the rows
  my $selected = '';
  my @rows;
  my @build_ids;
  my $col_n = 0;
  while (my $line = <TSV>) {
    chomp $line;
    my @row = split("\t", $line);
    my $build_name = $row[$build_name_idx];
    $build_name =~ s/.*">//;
    $build_name =~ s/<.*//;
    push @row, $build_name;
    push (@rows, \@row);
    $col_n = scalar @row;
    push @build_ids, $row[$#row]; # Assumes build id is last column
  
  }

#  my %accessible_default_builds;
#  my %atlas_build_hash;
  my $accessible_string = join( ',', keys( %atlas_build_hash ) );
  my $enzyme_info = get_enzyme_info( $accessible_string );

  # Sort the rows based on organism.
  my %organisms = (
    'Human'=>0,
    'Chimpanzee'=>1,
    'Mouse'=>2,
    'Cow'=>5,
    'Pig'=>6,
    'Horse'=>7,
    'Canine'=>8,
    'Dog'=>8,
    'Drosophila'=>10,
    'Honeybee'=>11,
    'C Elegans'=>20,
    'Yeast'=>25,
    'Fission Yeast'=>26,
    'Leptospira interrogans'=>31,
    'Candida albicans'=>33,
    'Ecoli'=>39,
    'Streptococcus'=>40,
    'Halobacterium'=>41,
    'Mtuberculosis'=>42,
    'Dengue virus'=>50,
  );

  # Put any additional organisms at the end
  for my $row (@rows) {
    if (! defined $organisms{$row->[$org_idx]} ) {
      $organisms{$row->[$org_idx]} = 100;
    }
  }

  @rows = sort { $a->[$col_n -1] cmp  $b->[$col_n -1]} @rows;
    # This is supposed to do an alphabetical sub-sort, but it's not working
    #|| $a->[$build_name_idx] cmp  $b->[$build_name_idx]
	      

  # Process and output the rows
  for my $row_aref (@rows) {
    my @row = @{$row_aref};
    pop @row;
    my @trinfo;
    my $selected = '';
    my $bgcolor = '#f3f1e4';

    # Check atlas build ID
    my $id = $row[$atlas_build_id_output_idx];
    my $atlas_build = $atlas_build_hash{$id};
    next if (! defined $atlas_build);

    $row[$spectra_searched_idx]   = $sbeams->commifyNumber($row[$spectra_searched_idx]);
    $row[$psm_count_idx]          = $sbeams->commifyNumber($row[$psm_count_idx]);
    $row[$distinct_peps_idx]      = $sbeams->commifyNumber($row[$distinct_peps_idx]);
    $row[$n_canonicals_idx]       = $sbeams->commifyNumber($row[$n_canonicals_idx]);
    $row[$n_canon_dist_idx]       = $sbeams->commifyNumber($row[$n_canon_dist_idx]);
    $row[$n_disting_prots_idx]    = $sbeams->commifyNumber($row[$n_disting_prots_idx]);
    $row[$n_seq_unique_prots_idx] = $sbeams->commifyNumber($row[$n_seq_unique_prots_idx]);
    $row[$n_swiss_idx]            = $sbeams->commifyNumber($row[$n_swiss_idx]);
    $row[$n_covering_idx]         = $sbeams->commifyNumber($row[$n_covering_idx]);

    if ( $parameters{organism_default_builds} ) {
      next if !$accessible_default_builds{$id};
    }
    if ( $parameters{protease_info} ) {
      push @row, $enzyme_info->{$id}->{enz_info};
    }

    # Post-processing for ppt format
    my $row_id;
    if ($ppt) {
      # Remove unwanted columns for ppt
      my @ppt_row=();
      for my $col (@ppt_cols) {
	my $content = $row[$col];
	# Remove any hyperlink
	if ($content =~ /<a href=.*>(.*)<\/a>/i ) {
	  $content = $1;
	}
	# Tidy build names
	if ($col == $build_name_idx) {
	  $content =~ s/ PeptideAtlas//;   # remove PeptideAtlas
	  $content =~ s/ Ens..//;          # remove Ensembl version number
	  $content =~ s/ Milk//;           # "Cow Milk" atlas has much more than milk
	  $content =~ s/Prob 0.99 //;      # prob is shown in its own column
	  $content =~ s/Prob 0.9 //;
	  # tidy species names
	  $content =~ s/HoneyBee/Honeybee/;
	  $content =~ s/Pyogenes/pyogenes/;
	  $content =~ s/Tuberculosis/tuberculosis/;
	  $content =~ s/CelegansPhospho/C\. elegans Phospho/;
	  $content =~ s/Celegans/C\. elegans/;
	  $content =~ s/Leptospira/L\./;
	  # general tidy
	  $content =~ s/Yeast_MRM_Atlas_2008-03_P0.9/Yeast SRMAtlas 2008-03/;
	  # Add padding whitespace
	  $content .= "&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;";
	  # Add an element for hiding the row.
	  $row_id = "row_$id";
	  $content = qq~<span onClick="document.getElementById('$row_id').style.display='none'">$content</span>~;
	}
	# Show spectra, obs in 1000's
	if (($col == $spectra_searched_idx ) || ($col == $psm_count_idx)) {
	  $content = sprintf("%0.0f", $content/1000);
	}
	push (@ppt_row, $content);
      }
      @row = @ppt_row;
    }
    # Add the row to the table.
    $table->addRow( \@row );
    $this_row = $table->getRowNum();
    # This manages which builds are shown for "View default builds"
    if ($id == $atlas_build_id) {
      $selected = 'CHECKED ';
    }   
    if ( !$atlas_build->[$def_atlas_build_id_output_idx] ) {
      if ( $selected ne 'CHECKED ' ) { # We will show the current build regardless
	$log->debug( "checking is $atlas_build->[$atlas_atlas_build_id_output_idx]" );
	@trinfo = ( $tr =~ /(NAME)=('[^']+')\s+(ID)=('[^']+')\s+(CLASS)=('[^']+')/ );
      }
      $bgcolor = '#d3d1c4' unless ($ppt);
    }
    if (! $ppt) {
      $table->setRowAttr( ROWS => [$this_row], BGCOLOR => $bgcolor, @trinfo );
    } else {
      # overwrites special ID for current atlas build, but should be OK.
      $table->setRowAttr( ROWS => [$this_row], BGCOLOR => $bgcolor, @trinfo , NOWRAP => 0, HEIGHT => 25, ID=>qq~"$row_id"~,);
    }
  }
  close TSV;

  # Set display attributes for the various columns.
  # 05/02/12: value below is always zero!
  my $nrows = $this_row;

  if (!$ppt) {
    $table->setColAttr(  COLS => [$build_name_idx+1], ROWS => [2..$nrows], NOWRAP => 1 );
    $table->setColAttr(  COLS => [$descr_idx+1], ROWS => [2..$nrows], NOWRAP => 1 );
    $table->setColAttr(  COLS => [$build_name_idx+1], ROWS => [1..$nrows], ALIGN => 'LEFT' );
    $table->setColAttr(  COLS => [$build_name_idx+2..$descr_idx], ROWS => [1..$nrows], ALIGN => 'CENTER' );
    $table->setColAttr(  COLS => [$descr_idx+1], ROWS => [1..$nrows], ALIGN => 'LEFT' );
    $table->setColAttr(  COLS => [$atlas_build_id_output_idx+1], ROWS => [1..$nrows], HIDDEN => 1 );

    # ppt
  } else {

    $table->setColAttr(COLS=>[$build_name_ppt_idx+1],ROWS=>[2..$nrows],NOWRAP=>1);
    $table->setColAttr(COLS=>[$peptide_inclusion_ppt_idx+1],ROWS=>[2..$nrows],NOWRAP=>1);
    $table->setColAttr(COLS=>[$build_name_ppt_idx+1..$peptide_inclusion_ppt_idx+1],ROWS=>[1..$nrows],ALIGN=>'LEFT');
    $table->setColAttr(  COLS => [$peptide_inclusion_ppt_idx+2..$n_swiss_ppt_idx+1], ROWS => [1..$nrows], ALIGN => 'RIGHT' );
    $table->setColAttr(  COLS => [$atlas_build_id_output_idx+1], ROWS => [1..$nrows], HIDDEN => 1 );
  }
  if ( $is_html ) {
    print $link unless $ppt;  #table toggle
  }

  }

  # Display the table!
  if ( $is_html ) {

    my $style = qq~<style type="text/css">
    table.buildtable { font-size:20px; }
    </style>
    ~;

    # This stringify is implicitly calling the asHTML method...
    $table = $style . "$table";
    $table =~ s/<table/<table width=${table_width}/i if $ppt;
    $table =~ s/<table/<table class="buildtable"/i if $ppt;

    print "$table";
    print $q->hidden( "apply_action", '');
    print $q->end_form;

  } elsif ( $sbeams->output_mode =~ /tsv/i ) {
    print $table->asTSV( strip_markup => 1 );
  }
} # end showMainPage



sub get_enzyme_info {
  my $build_str = shift || return {};
  my $enz_sql = qq~
  SELECT atlas_build_id, name, COUNT(*) 
  FROM $TBAT_ATLAS_BUILD_SEARCH_BATCH ABSB 
  JOIN $TBAT_ATLAS_SEARCH_BATCH ASB ON ASB.atlas_search_batch_id = ABSB.atlas_search_batch_id
  JOIN $TBAT_SAMPLE S on S.sample_id = ASB.sample_id
  JOIN $TBAT_PROTEASES P on P.id = S.protease_id
  WHERE atlas_build_id IN ( $build_str )
  GROUP BY P.name, atlas_build_id
  ORDER BY atlas_build_id, name
  ~;
  my %builds;
  my $sth = $sbeams->get_statement_handle( $enz_sql );
  my %enz;
  while ( my @row = $sth->fetchrow_array() ) {
    $builds{$row[0]} ||= {};
    $builds{$row[0]}->{$row[1]} = $row[2];
    $builds{$row[0]}->{total} += $row[2];
    $enz{$row[1]}++;
  }
  for my $id ( keys( %builds ) ) {
    for my $enz ( sort( keys( %{$builds{$id}} ) ) ) {
      next if $enz eq 'total';
      $builds{$id}->{$enz} = sprintf( "%0.1f", 100*($builds{$id}->{$enz}/$builds{$id}->{total}) );
      my $sep = ( $builds{$id}->{enz_info} ) ? ', ' : '';
      $builds{$id}->{enz_info} .= $sep . "$enz: $builds{$id}->{$enz}" . '%';
    }
  }
  return \%builds;
}