#!/usr/local/bin/perl

###############################################################################
# Program     : GetAtlasVariants
# Author      : David Campbell  
# $Id: $
#
# Description : Allows user to query SNPs in various Atlas builds.
#
# SBEAMS is Copyright (C) 2000-2015 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.
#
###############################################################################


###############################################################################
# Set up all needed modules and objects
###############################################################################
use strict;
use Getopt::Long;
use FindBin;
use Data::Dumper;

use lib "$FindBin::Bin/../../lib/perl";
use vars qw ($sbeams $sbeamsMOD $q $current_contact_id $current_username
             $PROG_NAME $USAGE %OPTIONS $QUIET $VERBOSE $DEBUG $DATABASE
             $TABLE_NAME $PROGRAM_FILE_NAME $CATEGORY $DB_TABLE_NAME
             @MENU_OPTIONS);

use SBEAMS::Connection qw($q $log);
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;
$sbeams->setSBEAMS_SUBDIR($SBEAMS_SUBDIR);

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


###############################################################################
# Set program name and usage banner for command like use
###############################################################################
$PROG_NAME = $FindBin::Script;
$USAGE = <<EOU;
Usage: $PROG_NAME [OPTIONS] key=value key=value ...
Options:
  --verbose n         Set verbosity level.  default is 0
  --quiet             Set flag to print nothing at all except errors
  --debug n           Set debug flag

 e.g.:  $PROG_NAME [OPTIONS] [keyword=value],...

EOU

#### Process options
unless (GetOptions(\%OPTIONS,"verbose:s","quiet","debug:s")) {
  print "$USAGE";
  exit;
}

$VERBOSE = $OPTIONS{"verbose"} || 0;
$QUIET = $OPTIONS{"quiet"} || 0;
$DEBUG = $OPTIONS{"debug"} || 0;
if ($DEBUG) {
  print "Options settings:\n";
  print "  VERBOSE = $VERBOSE\n";
  print "  QUIET = $QUIET\n";
  print "  DEBUG = $DEBUG\n";
}


###############################################################################
# Set Global Variables and execute main()
###############################################################################
main();
exit(0);


###############################################################################
# Main Program:
#
# Call $sbeams->Authenticate() and exit if it fails or continue if it works.
###############################################################################
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'],
    # connect_read_only=>1,
    allow_anonymous_access=>1,
  ));


  #### Read in the default input parameters
  my %parameters;

  $parameters{uploaded_file_not_saved} = 1;

  my $n_params_found = $sbeams->parse_input_parameters(
    q=>$q,parameters_ref=>\%parameters);
     #$sbeams->printDebuggingInfo($q);

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

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

    # Some action

  } else {

    $sbeamsMOD->display_page_header(
      use_tabbed_panes=> '1',
    );

    handle_request(ref_parameters=>\%parameters);

    $sbeamsMOD->display_page_footer(
       use_tabbed_panes=> '1',
     );

  }


} # end main


###############################################################################
# Handle Request
###############################################################################
sub handle_request {
  my %args = @_;

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


  #### Show current user context information
  print "<BR>\n" if ($sbeams->output_mode() eq 'html');
  #$sbeams->printUserContext();

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

  #my  @accessibleBuilds = $sbeamsMOD->getAccessibleBuilds();

  ### Define some generic variables
  my ($i,$element,$key,$value,$line,$result,$sql);


  #### Define some variables for a query and resultset
  my %resultset = ();
  my $resultset_ref = \%resultset;
  my (%url_cols,%hidden_cols,%max_widths,$show_sql);

  #### 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 $CATEGORY="Get Atlas Variants";
  $TABLE_NAME="AT_GetAtlasVariants" unless ($TABLE_NAME);
  ($PROGRAM_FILE_NAME) =
    $sbeamsMOD->returnTableInfo($TABLE_NAME,"PROGRAM_FILE_NAME");
  my $base_url = "$CGI_BASE_DIR/$SBEAMS_SUBDIR/$PROGRAM_FILE_NAME";

  #### Get the columns and input types for this table/query
  my @columns = $sbeamsMOD->returnTableInfo($TABLE_NAME,"ordered_columns");

  my %input_types = 
    $sbeamsMOD->returnTableInfo($TABLE_NAME,"input_types");

  #### Read the input parameters for each column
  my $n_params_found = $sbeams->parse_input_parameters(
    q=>$q,parameters_ref=>\%parameters,
    columns_ref=>\@columns,input_types_ref=>\%input_types);
   #$sbeams->printDebuggingInfo($q);


  #### Set some reasonable defaults if no parameters supplied
  unless ($n_params_found) {
    $parameters{input_form_format} = "minimum_detail";
  }
  if( not defined $parameters{organism_id}){
    $parameters{organism_id} = 2;
  }
  #### If the apply action was to recall a previous resultset, do it
  my %rs_params = $sbeams->parseResultSetParams(q=>$q);
  if ($apply_action eq "VIEWRESULTSET"  || $apply_action eq "VIEWPLOT") {
    $sbeams->readResultSet(
        resultset_file=>$rs_params{set_name},
        resultset_ref=>$resultset_ref,
        query_parameters_ref=>\%parameters
    );
    $n_params_found = 99;
  }



  my %sacrificial_params = %parameters;
  #### Display the user-interaction input form
  $sbeams->display_input_form(
    TABLE_NAME=>$TABLE_NAME,
    CATEGORY=>$CATEGORY,
    apply_action=>$apply_action,
    PROGRAM_FILE_NAME=>$PROG_NAME,
    parameters_ref=>\%sacrificial_params,
    input_types_ref=>\%input_types,
    mask_user_context=> '1',
    use_tabbed_panes=> '1',
  );

  #### Display the form action buttons
  $sbeams->display_form_buttons(
     TABLE_NAME=>$TABLE_NAME,
     use_tabbed_panes=> '1'
     );

  ### validate form input 
  if ($sbeams->output_mode() eq 'html'){
		print qq~
			<SCRIPT LANGUAGE="JavaScript" TYPE="text/javascript">
			 var newDiv = null;
			 validate_value = function (){
				 if(this.value > 50 || this.value < 6){
					 newDiv = document.createElement ("B"); 
					 newDiv.innerHTML = '<font color="red">number must be >=6 and <=50 </font>';
					 this.parentNode.appendChild(newDiv);
				 }
			 }
			document.getElementsByName("peptide_length_constraint")[0].onblur = validate_value;

		</SCRIPT>
  ~;
  }
  #### Finish the upper part of the page and go begin the full-width
  #### data portion of the page
  $sbeams->display_page_footer(
    close_tables=>'NO',
    separator_bar=>'NO',
    display_footer=>'NO',
    use_tabbed_panes=> '1',
  );

  #########################################################################
  #### Process all the constraints
  ### check required field -- need to remove the empty selection from the optionlist.
  my ($organism_id) = $parameters{organism_id};
  

  my $protein_name_clause = $sbeams->parseConstraint2SQL(
			constraint_column=>"ASNP.protein",
			constraint_type=>"plain_text",
			constraint_name=>"protein_name",
			constraint_value=>$parameters{protein_name} );
		return if ($protein_name_clause eq '-1');

  my $source_clause = '';
  if ( $parameters{snp_source} eq 'NP' ) {
    $source_clause = "AND snp_source = 'NP'"
  } elsif  ( $parameters{snp_source} eq 'SP' ) {
    $source_clause = "AND snp_source = 'UP'"
  } elsif ( $parameters{snp_source} =~ /NP../ ) {
    $source_clause = "AND snp_source = 'NP' AND annotation NOT LIKE '%Cosmic%'"
  }

  #ad and clause for sql for $parameters{protein_file} - supercedes protein_name
  my %protein_hash;
  if ( $parameters{protein_file} ) {
     ## upload the file to a file handler
     my $fh = $q->upload('protein_file');

     if (!$fh && $q->cgi_error) {
       print $q->header(-status=>$q->cgi_error) if ($sbeams->output_mode() eq 'html');
     }
     # size constraint of 1 MB, restrict $count < 30000
     if ( (-T $fh) && (-s $fh < 1000000)) {
       my $count = 0;
       my $read_file=0;
       my $prt;
       ## protein list
       while ($prt=<$fh>) {
         chomp($prt);
         $prt =~ s/\s+$//;
         if ($prt) {
           $protein_hash{$prt} = $prt;
           $count = $count + 1;
         }
         last if ($count > 30000);
      }
    }
    ## join with a commas:
    
    my $protein_list = '';
    $protein_list = "'" . join( "','", keys( %protein_hash ) ) . "'" if scalar( keys( %protein_hash ) );
    my $protein_names_clause = ( $protein_list ) ? " AND ASNP.protein IN ( $protein_list )" : '';
  } # if upload file

  #### Build ROWCOUNT constraint
  $parameters{row_limit} = 50000 unless ($parameters{row_limit} > 0 && $parameters{row_limit}<=1000000);
  my $limit_clause = "TOP $parameters{row_limit}";

  #### Define some variables needed to build the query
  my @column_array;
    # Set up base column defs
    @column_array = (
    ["atlas_build_id", "ASNP.atlas_build_id", "Atlas Build"],
    ["snp_source","CASE WHEN ASNP.snp_source = 'UP' THEN 'Swiss Prot' ELSE 'NeXtProt' END", "SNP Source"],
    ["protein","ASNP.protein", "Protein"],
    ["site","ASNP.site", "SNP site"],
    ["pe","ASNP.pe", "Protein existence level"],
    ["pre_aa","ASNP.pre_aa", "Original AA"],
    ["post_aa","ASNP.post_aa", "Variant AA"],
    ["total_nobs","ASNP.total_nobs", "Total N observations (site)"],
    ["post_nobs","ASNP.post_nobs", "N observations (SNP)"],
    ["post_uniq_nobs","ASNP.post_uniq_nobs", "N unique observations (SNP)"],
    ["category","ASNP.category", "Category"],
    ["annotation","ASNP.annotation", "Annotation"],

    );


  #### Set flag to display SQL statement if user selected
  if ( $parameters{display_options} =~ /ShowSQL/ ) {
    $show_sql = 1;
  }

  
  #### Build the columns part of the SQL statement
  my %colnameidx = ();
  my @column_titles = ();


  ## Sends @column_array_ref to build_SQL_columns_list, which
  ## (1) appends the 2nd element in array to $columns_clause
  ## (2) fills %colnameidx_ref as a hash with key = 1st element
  ## and value = 3rd element, and (3) fills @column_titles_ref
  ## array with the 3rd element
  my $columns_clause = $sbeams->build_SQL_columns_list(
    column_array_ref=>\@column_array,
    colnameidx_ref=>\%colnameidx,
    column_titles_ref=>\@column_titles
  );

  #### Define the SQL statement
  my $sql = qq~
     SELECT $limit_clause
     $columns_clause
     FROM $TBAT_ATLAS_SNPS ASNP
    WHERE 1 = 1
    $protein_name_clause
    $source_clause
    ORDER BY protein ASC, site ASC
  ~;
 
  #$sbeams->display_sql(sql=>$sql);

  #### Certain types of actions should be passed to links
  my $pass_action = "QUERY";
  $pass_action = $apply_action if ($apply_action =~ /QUERY/i); 

  #### Pass nearly all of the constraints down to a child query
  my @parameters_to_pass;
  my $parameters_list = '';
  while ( ($key,$value) = each %input_types ) {
    if ($key ne 'sort_order' && $key ne 'display_options') {
      if ($parameters{$key}) {
        push(@parameters_to_pass,"$key=$parameters{$key}");
      }
    }
  }
  if (@parameters_to_pass) {
    $parameters_list = join('&',@parameters_to_pass);
  }



  #### Define columns that should be hidden in the output table

  #########################################################################
  #### If QUERY or VIEWRESULTSET was selected, display the data
  if ($apply_action =~ /(QUERY|VIEWRESULTSET|VIEWPLOT)/i ){

    #### If the action contained QUERY, then fetch the results from
    #### the database
    if ($apply_action =~ /QUERY/i) {
      #### Show the SQL that will be or was executed
      $sbeams->display_sql(sql=>$sql,use_tabbed_panes=> '1') if ($show_sql);

      #### Fetch the results from the database server
      $sbeams->fetchResultSet(
        sql_query=>$sql,
        resultset_ref=>$resultset_ref,
      );

      if($parameters{n_peptide_per_protein_constraint}){
        postProcessResultset(
          rs_params_ref=>\%rs_params,
          resultset_ref=>$resultset_ref,
          query_parameters_ref=>\%parameters,
          column_titles_ref=>\@column_titles,
          sql => $sql
        );
      }
      #### Store the resultset and parameters to disk resultset cache
      $rs_params{set_name} = "SETME";
      $sbeams->writeResultSet(
        resultset_file_ref=>\$rs_params{set_name},
        resultset_ref=>$resultset_ref,
        query_parameters_ref=>\%parameters,
        resultset_params_ref=>\%rs_params,
        query_name=>"$SBEAMS_SUBDIR/$PROGRAM_FILE_NAME",
      );
    }


#		my @headings = ('Biosequence Name',
#		);

#		my $col_info = $sbeamsMOD->get_column_defs( labels => \@headings );
#		my $help_text = $sbeamsMOD->make_table_help( entries => $col_info,);
#    $help_text =~ s/<span.*<\/span>//g;
#    $help_text =~ s/<BR>//g;

    my $use_tabbed_panes = 1;
#		if ( $help_text ) {
#       my $prefix = $sbeams->addTabbedPane(label => "Result");
#       print "$prefix$help_text\n" if ($sbeams->output_mode() eq 'html');
#       $use_tabbed_panes = 0;
#    }
#    die Dumper( $help_text );

    #### Display the resultset
    $sbeams->displayResultSet(
      resultset_ref=>$resultset_ref,
      query_parameters_ref=>\%parameters,
      rs_params_ref=>\%rs_params,
      url_cols_ref=>\%url_cols,
      hidden_cols_ref=>\%hidden_cols,
      max_widths=>\%max_widths,
      column_titles_ref=>\@column_titles,
      base_url=>$base_url,
      use_tabbed_panes=>$use_tabbed_panes,
    );
      

    #### Display the resultset controls
    $sbeams->displayResultSetControls(
      resultset_ref=>$resultset_ref,
      query_parameters_ref=>\%parameters,
      rs_params_ref=>\%rs_params,
      base_url=>$base_url,
      use_tabbed_panes=>$use_tabbed_panes,
    );
    if(! $use_tabbed_panes ){
      print $sbeams->closeTabbedPane() , "\n" if ($sbeams->output_mode() eq 'html');
    }


    #print "$parameters{organism_id}<BR>";
    #### Display a plot of data from the resultset
#    $sbeams->displayResultSetPlot(
#      rs_params_ref=>\%rs_params,
#      resultset_ref=>$resultset_ref,
#      query_parameters_ref=>\%parameters,
#      column_titles_ref=>\@column_titles,
#      base_url=>$base_url,
#      use_tabbed_panes=>'1',
#      apply_action=>$apply_action,
#    );

  #### If QUERY was not selected, then tell the user to enter some parameters
  } else {
    if ($sbeams->invocation_mode() eq 'http') {
      print "<H4>Select parameters above and press QUERY</H4>\n";
    } else {
      print "You need to supply some parameters to contrain the query\n";
    }
  }


} # end handle_request

###############################################################################
# postProcessResultset
#
# Perform some additional processing on the resultset that would otherwise
# be very awkward to do in SQL.
###############################################################################
sub postProcessResultset {
  my %args = @_;
  #### Process the arguments list
  my $resultset_ref = $args{'resultset_ref'};
  my $rs_params_ref = $args{'rs_params_ref'};
  my $query_parameters_ref = $args{'query_parameters_ref'};

  my %rs_params = %{$rs_params_ref};
  my %parameters = %{$query_parameters_ref};

  my $n_peptide_per_protein = $parameters{n_peptide_per_protein_constraint};
  $n_peptide_per_protein =~ s/\s+//g;
  if($n_peptide_per_protein =~ /(\D)/){
     print "<H4>only interger number allowed in Number of Peptides Per Protein Constraint.</H4>";
     return -1;
  }
  
  my $n_rows = scalar(@{$resultset_ref->{data_ref}}); 
  my $cols = $resultset_ref->{column_hash_ref};

  my @processed_result =();
  my %counts = ();
  for (my $i=0; $i<$n_rows; $i++) {
    my $protein_accession = $resultset_ref->{data_ref}->[$i]->[$cols->{biosequence_name}];
    if ($counts{$protein_accession} < $n_peptide_per_protein){
      push @processed_result , $resultset_ref->{data_ref}->[$i] ;
    }
    $counts{$protein_accession}++;
  }
  $resultset_ref->{data_ref} = \@processed_result;

}
###############################################################################
# evalSQL
#
# Callback for translating Perl variables into their values,
# especially the global table variables to table names
###############################################################################
sub evalSQL {
  my $sql = shift;

  return eval "\"$sql\"";

} # end evalSQL

#######################################################################
sub getOrganismName 
{
    my %args = @_;
    
    my $organism_id = $args{organism_id};

    my $sql = qq~
        SELECT organism_name
        FROM $TB_ORGANISM
        WHERE organism_id = '$organism_id'
        AND record_status != 'D'
    ~;

    my ($name) = $sbeams->selectOneColumn($sql);

    ## replace spaces with _
    $name =~ s/ /\_/g;

    return $name;
}