#!@WHICHPERL@
=head1 NAME

momo_webservice - Run MOMO program

=head1 SYNOPSIS

momo_webservice [options] [-psm <psm file> 
                            [ -filter_field <encoded field name> 
                              -filter_type lt|lteq|eq|gteq|gt 
                              -filter_thresh <threshold> ]? ]+

  PSM arguments:
    -psm <psm file>                       Specify a PSM file for processing
    -filter_field <encoded field name>    Specify a field of the previously
                                           specified PSM file to filter on.
    -filter_type lt|lteq|eq|gteq|gt       Specify the filtering type for the
                                           previously specified PSM file as: 
                                            lt   - less than,
                                            le   - less than or equal,
                                            eq   - equal,
                                            ge   - greater than or equal, or
                                            gt   - greater than.
    -filter_thresh <threshold>            Specify the number used to filter the
                                           previously specified PSM file.

  Flank Sequence arguments:
    -flank <FASTA file>                   Specify a protein FASTA file
                                           containing flanking sequence;
                                           any paths begining with db/ are
                                           pre-existing databases.
    -purge <score>                        Run purge on an uploaded flanking
                                           sequence with the given threshold.
  
  Options: 
    -occurs <num>     Set the minimum number of occurences.
    -width <width>     Set the width of the motif.
    -single_per_mass  A single motif will be generated for each distinct
                       modification mass.
    -help             Show this brief help message.

=cut

use strict;
use warnings;
# load standard perl libraries
use Getopt::Long qw(:config permute auto_help);
use Pod::Usage;
# load custom perl libraries
use lib qw(@PERLLIBDIR@);
use StatusPage qw(arg_checks arg_end parse_uploaded opt_db_or_uploaded opt_choice opt_encoded opt_integer opt_number);

# constants
my $bin_dir = '@BINDIR@';
my $fasta_db_dir = '@MEMEDB@/fasta_databases';
# required parameters
my @psms = (); # list of PSM objects with optional filters
my ($filter_field, $filter_type, $filter_thresh);
# Flank arguments
my ($flank_file, $flank_purge);
# option defaults
my $eliminate_width = 0;
my $width = 7;
my $occurs;
my $fg_filetype;
my $psm_column_name;
my $single_per_mass = 0; # FALSE
my ($count_threshold, $score_threshold);
my $algorithm;
#status page
my $status = new StatusPage('MOMO', \@ARGV);
$status->add_message('Parsing arguments');
# parse options
my @arg_errors = ();
my $opts_ok = do {
  local $SIG{__WARN__} = sub { my ($msg) = @_; chomp($msg); push(@arg_errors, $msg); };
  GetOptions(
  '<>' => arg_checks(arg_end()),
  'psm=s' => sub {
    my ($opt, $value) = @_; my $file;
    parse_uploaded($opt, $value, \$file);
    push(@psms, {file => $file});
  },
#  'filter_field=s' => sub {
#    my ($opt, $value) = @_;
#    die("Option -$opt must only appear after a -psm option.") unless (@psms);
#    die("Option -$opt must only be listed once per -psm option.") if (defined($psms[-1]->{field}));
#    parse_encoded($opt, $value, \$psms[-1]->{field});
#  },
#  'filter_type=s' => sub {
#    my ($opt, $value) = @_;
#    die("Option -$opt must only appear after a -psm option.") unless (@psms);
#    die("Option -$opt must only be listed once per -psm option.") if (defined($psms[-1]->{type}));
#    parse_choice($opt, $value, \$psms[-1]->{type}, 'lt', 'le', 'eq', 'ge', 'gt');
#  },
#  'filter_thresh=f' => sub {
#    my ($opt, $value) = @_;
#    die("Option -$opt must only appear after a -psm option.") unless (@psms);
#    die("Option -$opt must only be listed once per -psm option.") if (defined($psms[-1]->{thresh}));
#    $psms[-1]->{thresh} = $value;
#  },
  'filter_field=s' => opt_encoded(\$filter_field),
  'filter_type=s' => opt_choice(\$filter_type, 'lt', 'le', 'eq', 'ge', 'gt'),
  'filter_thresh=f' => opt_number(\$filter_thresh, '>', 0),
  'fg_filetype=s' => opt_choice(\$fg_filetype, 'psm', 'prealigned', 'fasta'),
  'psm_column_name=s' => opt_encoded(\$psm_column_name),
  'flank=s' => opt_db_or_uploaded(\$flank_file, $fasta_db_dir, 'db'),
  'occurs=i' => opt_integer(\$occurs, 1),
  'eliminate_width=i' => opt_integer(\$eliminate_width, 0, 50),
  'width=i' => opt_integer(\$width, 1, 51),
  'count_threshold=i' => opt_integer(\$count_threshold, 1, 50),
  'score_threshold=f' => opt_number(\$score_threshold, '<', 1),
  'score_threshold=f' => opt_number(\$score_threshold, '>', 0),
  'algorithm=s' => opt_choice(\$algorithm, 'alg_simp', 'alg_mtfx', 'alg_modl'),
  'single_per_mass' => \$single_per_mass
  );
};
# add additional error messages for missing sequences and motifs
push(@arg_errors, "No PSMs provided.") unless @psms;
for (my $i = 0; $i < scalar(@psms); $i++) {
  my $psm = $psms[$i];
#  if (defined($psm->{field}) || defined($psm->{type}) || defined($psm->{thresh})) {
#    unless (defined($psm->{field}) && defined($psm->{type}) && defined($psm->{thresh})) {
#      push(@arg_errors, "Missing filter fields for \"" . $psm->{file} . "\".");
#    }
#  }
}
# display the error messages both to the status page and stdout
foreach my $arg_error (@arg_errors) {
  print STDERR $arg_error, "\n";
  $status->add_message($arg_error);
}
$opts_ok = 0 if (scalar(@arg_errors) > 0);
# setup status page

$status->add_file('html', 'momo.html', 'MOMO HTML output');
$status->add_file('txt', 'momo.txt', 'MOMO TXT output');

for (my $i = 0; $i < scalar(@psms); $i++) {
  $status->add_file('psm'.$i, $psms[$i]->{file}, 'Peptide-spectrum matches file ' . ($i + 1));
}
$status->add_file('flank', $flank_file, 'Match flanks');
$status->add_message($opts_ok ? 'Arguments ok' : "Error parsing arguments");
$status->update($opts_ok ? 'Starting' : 'Error');
# exit if there was an error reading the arguments
unless ($opts_ok) {
  $status->write_log();
  pod2usage(2);
}

#TODO put call to actual MOMO program here

my $alg_string;
$alg_string = "simple" if ($algorithm eq 'alg_simp');
$alg_string = "motifx" if ($algorithm eq 'alg_mtfx');
$alg_string = "modl" if ($algorithm eq 'alg_modl');

my @momo_args = ($alg_string, '-oc', '.', '--verbosity', 1);
push(@momo_args, '--protein-database', $flank_file) if (defined($flank_file));
push(@momo_args, '--min-occurrences', $occurs) if (defined($occurs));
push(@momo_args, '--width', $width);
if (defined($filter_field) && defined($filter_type) && defined($filter_thresh)) {
  my $filter_string = "$filter_field,$filter_type,$filter_thresh";
  push(@momo_args, '--filter', $filter_string) 
}
push(@momo_args, '--count-threshold', $count_threshold) if (defined($count_threshold));
push(@momo_args, '--score-threshold', $score_threshold) if (defined($score_threshold));
push(@momo_args, '--fg-filetype', $fg_filetype) if (defined($fg_filetype));
push(@momo_args, '--sequence-column', $psm_column_name) if (defined($psm_column_name));
push(@momo_args, '--eliminate-repeats', $eliminate_width);
push(@momo_args, '--single-motif-per-mass', 'T') if ($single_per_mass != 0);
for (my $i = 0; $i < scalar(@psms); $i++) {
  push(@momo_args, ($psms[$i])->{file});
}
$status->run(PROG => 'momo', BIN => $bin_dir, ARGS => \@momo_args);

# done
$status->add_message("Done");
$status->update();
$status->write_log();
1;