#!/usr/bin/env perl

use strict;
use FindBin;
use Bio::SeqIO;
use Bio::Seq;
use Path::Tiny;
use File::Basename;
use File::Spec;
use File::Path qw(make_path remove_tree);
use List::Util qw(first);
use Cwd        qw(abs_path);
use Data::Dumper;
use LWP::Simple;
use JSON;

#..............................................................................
# Globals

my $EXE     = basename($0);
my $ABX_SEP = ';';

my %DATABASE = (
    'resfinder'     => \&get_resfinder,
    'plasmidfinder' => \&get_plasmidfinder,
    'megares'       => \&get_megares,
    'argannot'      => \&get_argannot,
    'card'          => \&get_card,

    #  'ncbibetalactamase' => \&get_ncbibetalactamase,
    'ncbi'     => \&get_ncbi,
    'vfdb'     => \&get_vfdb,
    'ecoli_vf' => \&get_ecoli_vf,    # https://github.com/phac-nml/ecoli_vf
    'ecoh'     => \&get_ecoh,
    'bacmet2'  => \&get_bacmet2,
    'victors'  => \&get_victors,

    #  'serotypefinder'    => \&get_serotypefinder,
);
my $DATABASES = join( ' ', sort keys %DATABASE );

#..............................................................................
# Command line options

my ( @Options, $debug, $outdir, $db, $force );
setOptions();

$db                   or err("Please choose a --db from: $DATABASES");
exists $DATABASE{$db} or err("Unknown --db '$db', choose from: $DATABASES ");
-d $outdir            or err("--outdir '$outdir' does not exist");

my $dir = abs_path( File::Spec->catdir( $outdir, $db ) );
make_path($dir);
msg("Setting up '$db' in '$dir'");

#my $tmpdir = tempdir("$db-XXXXXXXX", DIR=>$dir, CLEANUP=>0);
#my $tmpdir = "/home/tseemann/git/abricate/db/resfinder/resfinder-6Kuphtvv";
my $tmpdir = "$dir/src";
make_path($tmpdir);

# run the specific function from --db
chdir $tmpdir;
my $seq = $DATABASE{$db}->();
map { is_full_gene($_) } @$seq;    # doesn't do anything?
$seq = dedupe_seq($seq);

#print Dumper($seq);
msg("Sorting sequences by ID");
$seq = [ sort { $a->{ID} cmp $b->{ID} } @$seq ];
save_fasta( "$dir/sequences", $seq );

msg("Formatting BLASTN database: $dir/sequences");
my $logfile    = "$tmpdir/makeblastdb.log";
my $ncbi_title = $db;
if ( "$db" eq "ncbi" ) {
    $ncbi_title = "ncbiamrplus";
}
my $ec = system(
"makeblastdb -in '$dir/sequences' -title '$ncbi_title' -dbtype nucl -hash_index -logfile $logfile"
);
if ( $ec != 0 ) {
    system("tail '$logfile'");
    err("Error with makign BLAST database. See $logfile");
}

#msg("Run 'abricate --setupdb' to format the database");

msg("Done.");

#..............................................................................

sub download {
    my ( $url, $dest ) = @_;
    if ( -r $dest and not $force ) {
        msg("Won't re-download existing $dest (use --force)");

        #exit(1);
    }
    else {
        msg("Downloading: $url");
        my $ec = mirror( $url, $dest );
        msg("HTTP Result: $ec");
        ( $ec == 200 or $ec = 304 )
          or err("HTTP $ec | failed to download $url");    # is HTTP OK ?
    }
    msg("Destination: $dest");
    msg( "Filesize:", ( -s $dest ), "bytes" );
}

#..............................................................................
sub trim_spaces {
    my ($s) = @_;
    $s =~ s/^\s+//;
    $s =~ s/\s+$//;
    return $s;
}

#..............................................................................
sub get_resfinder {
    my $name = "resfinder_db";

    # FIXME - can we just get HEAD.zip like in plasmidfinder?
    my $url = "https://bitbucket.org/genomicepidemiology/$name.git";

    #   if (-r $name and not $force) {
    #     msg("Won't overwrite existing $name (use --force)");
    # #    exit(1);
    #   }
    #   else {
    #     msg("Nuking existing folder: $name");
    #     remove_tree("./$name");
    #     msg("Cloning $url to $name");
    #     system("git clone --quiet $url $name");
    #   }

    #<*.fsa>
    #>aac(6')-Ib_2_M23634
    #>blaNDM-19_1_MF370080
    #>mcr-1.1_1_KP347127
    #>fosB1_1_CP001903
    #>fusB_1_AY373761
    #>VanHAX_1_FJ866609
    #>ere(A)_6_DQ157752
    #>nimA_1_X71444
    #>cfr_1_AM408573
    #>catB3_2_U13880
    #>qnrA1_1_AY070235
    #>ARR-2_1_HQ141279
    #>sul1_2_U12338
    #>tet_1_M74049
    #>dfrA19_1_EU855687

    #<notes.txt>
    #aac(6')-Iv:Aminoglycoside resistance:
    #aac(6')-Iw:Aminoglycoside resistance:Alternate name; aac(6')-Ix
    #sul3:Sulphonamide resistance:
    ##Tetracycline:
    #ort(B):Tetracycline resistance:
    #blaCMY-59:Beta-lactam resistance:

#<phenotypes.txt>
#Gene_accession no.      Class   Phenotype       PMID    Mechanism of resistance Notes Required_gene
#ant(2'')-Ia_1_X04555    Aminoglycoside  Gentamicin, Tobramycin  3024112 Enzymatic modification  Alternative name aadB
#ant(2'')-Ia_2_JF826500  Aminoglycoside  Gentamicin, Tobramycin  22271862        Enzymat

    $name = "~/apps/bettercallsal/assets/abricate_dbs/$name";
    my $metafn = "$name/phenotypes.txt";
    my @meta   = path($metafn)->lines( { chomp => 1 } );
    my %anno;
    foreach (@meta) {
        next if m/^#/;
        my @x = split m/\t/;

        #msg("$metafn: @x");
        my ($gene) = ( $x[0] =~ m/^(.*?)_\w+$/ );
        $anno{$gene}{ABX} = [
            map    { trim_spaces($_) }
              grep { !m/(unknown|notes|^none)/i }
              split m/,\s*/,
            $x[2]
        ];

        #msg("$metafn: $gene |", $anno{$gene}{ABX}->@*);
    }
    msg( "get_resfinder: $metafn", scalar( keys %anno ), "genes" );

    #print Dumper(\%anno);

    my @seq;
    for my $fasta (<$name/*.fsa>) {

        # Issue #62 - repair broken fasta files like this:
        # GCTTTAAATTGGAAAAAAGATAGTCAAACTCTTTAA>cmr_1_U43535
        # inline replacement
        system( 'sed', '-i.bak', 's/\([A-Z]\)>/\1\n>/gi', $fasta );
        my $args = load_fasta($fasta);

        # use name of fasta file as antibiotic name
        #my $abx = basename($fasta, '.fsa');
        #msg("$fasta: Assigning '$abx' to all genes");
        #push @{$_->{ABX}}, $abx for (@$args);
        push @seq, @$args;
    }

    # https://github.com/tseemann/abricate/issues/92
    # mcr-9_1_NZ_NAAN01000063.1
    #>mcr-9_1_NZ_NAAN01000063.1
    # mcr-9.1:Colistin resistance:

    for my $seq (@seq) {
        my ( $id, $copy, $acc ) = $seq->{ID} =~ m/^(.*?)_(\d+)_(\S+)$/;

        #msg("resfinder: $1 $2 $3", $anno{$1});
        $seq->{ID}   = "${id}_${copy}";
        $seq->{ACC}  = $acc;
        $seq->{DESC} = $anno{$id}{DESC} || $id;
        push @{ $seq->{ABX} }, @{ $anno{$id}{ABX} } if $anno{$id}{ABX};
    }

    return \@seq;
}

#..............................................................................
sub get_serotypefinder {
    my $name = "serotypefinder_db";
    my $url  = "https://bitbucket.org/genomicepidemiology/$name.git";

    if ( -r $name and not $force ) {
        msg("Won't overwrite existing $name (use --force)");

        #    exit(1);
    }
    else {
        msg("Nuking existing folder: $name");
        remove_tree("./$name");
        msg("Cloning $url to $name");
        system("git clone --quiet $url $name");
    }

    my @seq;
    for my $fasta (<$name/*.fsa>) {
        push @seq, @{ load_fasta($fasta) };
    }

    # >fliC_44444_AY250028_H52
    # FIXME - this is already in EcOH database!
    for my $seq (@seq) {
        my ( $id, $copy, $acc ) = $seq->{ID} =~ m/^(.*)_(\d+)_(\w+)$/;

        #msg("serotypefinder: $1 $2 $3", $anno{$1});
        $seq->{ID}  = "${id}_${copy}";
        $seq->{ACC} = $acc;

        #$seq->{DESC} = $anno{$id} || '';
    }

    return \@seq;
}

#..............................................................................
sub get_tag {
    my ( $f, $tag ) = @_;
    if ( $f->has_tag($tag) ) {
        my ($val) = $f->get_tag_values($tag);
        return $val;
    }
    return '';
}

#..............................................................................
sub get_ncbi {
    my $AFP =
"https://ftp.ncbi.nlm.nih.gov/pathogen/Antimicrobial_resistance/AMRFinderPlus/database/latest";

#my $src = "https://ftp.ncbi.nlm.nih.gov/pathogen/Antimicrobial_resistance/AMRFinderPlus/data/latest/AMR_CDS";
    my $src  = "$AFP/AMR_CDS";
    my $name = "amr_cds.ffn";

#my $src2 = "https://ftp.ncbi.nlm.nih.gov/pathogen/Antimicrobial_resistance/AMRFinderPlus/data/latest/ReferenceGeneCatalog.txt";
    my $src2  = "$AFP/ReferenceGeneCatalog.txt";
    my $name2 = "amr_cds.tsv";

    if ( -r $name and -r $name2 and not $force ) {
        msg("Won't overwrite existing $name/$name2 (use --force)");

        #    exit(1);
    }
    else {
        download( $src,  $name );
        download( $src2, $name2 );
    }

    #1   allele
    #2   gene_family                   ble
    #3   whitelisted_taxa
    #4   product_name                  BLMA family bleomycin binding protein
    #5   scope                         core
    #6   type                          AMR
    #7   subtype                       AMR
    #8   class                         BLEOMYCIN
    #9   subclass                      BLEOMYCIN
    #10  refseq_protein_accession      WP_063842967.1
    #11  refseq_nucleotide_accession   NG_047554.1
    #12  curated_refseq_start          No
    #13  genbank_protein_accession     CAA02068.1
    #14  genbank_nucleotide_accession  A31900.1
    #15  genbank_strand_orientation    +
    #16  genbank_cds_start             6
    #17  genbank_cds_stop              374
    #18  pubmed_reference
    #19  blacklisted_taxa
    #20  db_version                    2019-08-27.1

    my $tsv = load_tabular( $name2, 10 );    # refseq_nucleotide_accession
    msg( "[$name2] Loaded", scalar keys %$tsv, "records" );

# https://github.com/ncbi/amr/wiki/AMRFinderPlus-database#amrprot
#  0          1              2         3 4 5          6      7
# >1000909371|WP_061158039.1|NG_050200|1|1|blaTEM-156|blaTEM|class_A_beta-lactamase_TEM-156 NG_050200:101-961
    my @seq;
    my $in = Bio::SeqIO->new( -file => $name, -format => "fasta" );
    while ( my $rec = $in->next_seq ) {

        # parse ID
        my ( $gi, $pi, $acc, $fp, $fn, $gene, $fam, $prod ) = split m/\|/,
          $rec->id;

        # skip fusion genes
        next unless $fp == 1 and $fn == 1;

        # only keep true ARGs
        $acc .= ".1" unless $acc =~ m/\.\d+$/;
        my $t = $tsv->{$acc} or next;
        next unless ( $t->{scope} eq 'plus' || $t->{scope} eq 'core' );

        # next unless $t->{type} eq 'VIRULENCE';
        # next unless $t->{subtype} eq 'VIRULENCE';
        # construct sequence record
        $prod =~ s/_/ /g;
        err("$pi: gene is empty")    unless $gene;
        err("$pi: product is empty") unless $prod;
        my $s = {
            ID   => $gene,
            ACC  => $acc,
            DESC => $prod,
            SEQ  => $rec->seq,
            ABX  => [ split m'/', $t->{subclass} ]
        };
        push @seq, $s;
        msg( "[$name]", 0 + @seq, "|", $s->{ID}, "|", $s->{ACC}, "|",
            $s->{DESC} );

        #msg(Dumper($s));
        msg( $s->{ID}, " is fusion $fp/$fn" ) if "$fp$fn" ne '11';
    }
    return \@seq;
}

#..............................................................................
sub get_plasmidfinder {
    my $name = "plasmidfinder";
    my $zip  = "$name.zip";

#  download("https://cge.cbs.dtu.dk/cge/download_data.php?folder=$name&filename=$zip&submit=$zip", $zip);
    download(
"https://bitbucket.org/genomicepidemiology/plasmidfinder_db/get/HEAD.zip",
        $zip
    );
    system("unzip -j -u $zip");

    my @seq;
    for my $fasta (<*.fsa>) {
        push @seq, @{ load_fasta($fasta) };
    }

    for my $seq (@seq) {
        $seq->{DESC} = $seq->{ID};    # no desc, so use ORIGINAL ID as desc
        my ( $id, $acc ) =
          ( $seq->{ID} =~ m/^(.*)_(([A-Z]+|NC_)\d+(\.\d+)?)$/ );
        $id =~ s/_+$//g;
        $seq->{ID}  = $id  || $seq->{ID};
        $seq->{ACC} = $acc || '';
        wrn( "Parsed empty ID:",
            $seq->{DESC},
            "=> id='$id' acc='$acc' seq=" . substr( $seq->{SEQ}, 0, 10 ) )
          if not $id;
    }

    return \@seq;
}

#..............................................................................
sub get_megares {
    my $zip = "megares.zip";
    download( 'https://www.meglab.org/downloads/megares_v3.00.zip', $zip );
    system("unzip -j -u $zip");
    my $seqs = load_fasta( glob("megares_drugs_*.fasta") );
    my @okseq;

# >MEG_372|Multi-compound|Drug_and_biocide_resistance|Drug_and_biocide_MATE_efflux_pumps|ABEM
# >MEG_411|Multi-compound|Drug_and_biocide_resistance|Drug_and_biocide_RND_efflux_regulator|ACRR|RequiresSNPConfirmation
# >MEG_7860|Drugs|betalactams|Class_B_betalactamases|ZOG
# >MEG_7439|Drugs|Glycopeptides|VanI-type_resistance_protein|VANI
# >MEG_7245|Drugs|Tetracyclines|Tetracycline_resistance_MFS_efflux_pumps|TETY
# >MEG_9|Drugs|Aminoglycosides|Aminoglycoside-resistant_16S_ribosomal_subunit_protein|A16S|RequiresSNPConfirmation

    for my $s (@$seqs) {
        my ( $id, $type, $class, $mech, $group, $note ) = split m/\|/, $s->{ID};
        if ($note) {

            # "RequiresSNPConfirmation" is the common one; we can't do that
            msg("Skipping $id due to: $note");
            next;
        }
        $s->{ID}   = $group;
        $s->{ACC}  = $id;
        $s->{DESC} = join( ':', $type, $class, $mech, $group );
        push @okseq, $s;
    }
    return [@okseq];

    #return $seqs;
}

#..............................................................................
sub get_argannot {
    my $fasta = 'arg-annot.fa';
    download(
#    'http://www.mediterranee-infection.com/arkotheque/client/ihumed/_depot_arko/articles/1425/argannot-aa-v3-march2017_doc.fasta',
#    'http://www.mediterranee-infection.com/arkotheque/client/ihumed/_depot_arko/articles/691/argannot-nt_doc.fasta',
'https://www.mediterranee-infection.com/wp-content/uploads/2019/06/ARG_ANNOT_V5_Nt_JUNE2019.txt',
        $fasta
    );

    # fix syntax errors in the FASTA file...
    path($fasta)->edit( sub { s/\\//g; $_ } );

    my $seqs = load_fasta($fasta);

    #  0             1         2               3
    # >(AGly)Aac2-Ie:NC_011896:3039059-3039607:549
    for my $s (@$seqs) {
        my @x = split m/:/, $s->{ID};
        $s->{ID}   = $x[0];
        $s->{ACC}  = $x[1] . ':' . $x[2];
        $s->{DESC} = '';
    }

    return $seqs;
}

#..............................................................................
sub get_bacmet2 {
    my $fasta = 'bacmet2.fa';
    download(
        'http://bacmet.biomedicine.gu.se/download/BacMet2_EXP_database.fasta',
        $fasta );

    # This is a PROTEIN file
    my $seqs = load_fasta($fasta);

    #  0       1    2  3      4         ^
    # >BAC0098|ctpC|sp|P0A502|CTPC_MYCTU Probable manganese/zinc-exporting
    for my $s (@$seqs) {
        my @x = split m/\|/, $s->{ID};
        $s->{ID}  = $x[1] . '-' . $x[0];
        $s->{ACC} = $x[2] . ':' . $x[3];
    }

    return $seqs;
}

#..............................................................................
sub get_card {

    # https://github.com/tseemann/abricate/issues/25
    my $tarball = 'card.tar.bz2';
    download(
        #'https://card.mcmaster.ca/download/0/broadstreet-v2.0.2.tar.gz',
        'https://card.mcmaster.ca/latest/data',

        #'https://card.mcmaster.ca/latest/data/card-data.tar.bz2',
        $tarball    # yes, it's really BZ2 not GZ ...
    );

    #  my $fasta = "./nucleotide_fasta_protein_homolog_model.fasta";
    my $jsonfile = "./card.json";
    system( "tar", "xf", $tarball, $jsonfile ) == 0
      or err("Problem with tar xf $tarball $jsonfile");
    -r $jsonfile or err("Could not extract $jsonfile from $tarball");

    # JSON
    my $json = path($jsonfile)->slurp_utf8;
    my $card = from_json( $json, { latin1 => 1 } );
    my @seq;
    for my $g ( values %$card ) {
        next unless ref($g) eq 'HASH';

        #    msg(Dumper($g));
        next
          unless $g->{model_type} eq
          "protein homolog model";    # only 'acquired' genes
        my $id = $g->{model_name};
        err("$id has {model_param}{snp}") if exists $g->{model_param}{snp};

        #    msg("CARD: $id");
        #    print STDERR Dumper($g);
        my $dna = $g->{model_sequences}{sequence}
          or err("$id: no {model_sequences}{sequence} found");
        my ($key) = sort keys %$dna;    # first key
        $dna = $dna->{$key}         or err("$id: invalid key '$key'");
        $dna = $dna->{dna_sequence} or err("$id: no dna_sequence");

        #    msg(Dumper($dna)) if $id eq 'OXA-25';

        # ARO_category => {
        #	'category_aro_name' => 'cephalosporin',
        #	'category_aro_class_name' => 'Drug Class',
        my $is_amr_gene = 0;
        my @abx;
        for my $key ( keys $g->{ARO_category}->%* ) {
            my $c = $g->{ARO_category}{$key};
            if ( $c->{category_aro_class_name} eq 'Drug Class' ) {
                my $abx = $c->{category_aro_name};
                $abx =~ s/ antibiotic//;
                $abx =~ s/\s/_/g;
                push @abx, $abx;
            }
            if ( $c->{category_aro_class_name} eq 'AMR Gene Family' ) {
                $is_amr_gene++;
            }
        }

        #err("CARD | $id | ", Dumper($g->{ARO_category}) ) unless $is_amr_gene;
        #msg("ABX=$_") for @abx;

        # put coordinates into normal form
        my ( $start, $stop ) =
          $dna->{strand} eq '-'
          ? ( $dna->{fmax}, $dna->{fmin} )
          : ( $dna->{fmin}, $dna->{fmax} );

        $id =~ s/\s+/_/g;
        push @seq,
          {
            ID   => $id,
            ACC  => $dna->{accession} . ":$start-$stop",
            DESC => ( $g->{ARO_description} || $g->{ARO_accession} ),
            SEQ  => $dna->{sequence},
            ABX  => [@abx],
          };

        #    msg(Dumper($seq[-1]));
    }

    return \@seq;
}

#..............................................................................
sub get_victors {

    # the CDS data is in .ffn and has source GI and coords
    # the PROT data is in .faa and has the protein ref and /product

  #>gi|115534241:2616-3152 Campylobacter jejuni plasmid pCJ01, complete sequence
    download( 'http://www.phidias.us/victors/downloads/gen_downloads.php',
        'victors.ffn' );

#>gi|115534244|ref|YP_783826.1| hypothetical protein pCJ01p4 [Campylobacter jejuni]
    download(
        'http://www.phidias.us/victors/downloads/gen_downloads_protein.php',
        'victors.faa' );

    my %gi;
    open my $FAA, '<', 'victors.faa';
    while (<$FAA>) {

#>gi|115534244|ref|YP_783826.1| hypothetical protein pCJ01p4 [Campylobacter jejuni]
        next unless m"^>gi.(\d+).ref.([^|]+). ([^[]+)";
        $gi{$1}{ACC}  = $2;
        $gi{$1}{DESC} = $3;
    }

    my $seqs = load_fasta("victors.ffn");

    for my $s (@$seqs) {

  #>gi|115534241:2616-3152 Campylobacter jejuni plasmid pCJ01, complete sequence
        $s->{ID} =~ m/gi.(\d+):(\d+)-(\d+)/;
        $s->{ACC} = $gi{$1}{ACC}    || "gi|$1:$2-$3";
        $s->{DESC} =~ $gi{$1}{DESC} || 'hypothetical protein';
    }

    #  print Dumper($seqs); exit;

    return $seqs;
}

#..............................................................................
sub get_vfdb {
    download( 'http://www.mgc.ac.cn/VFs/Down/VFDB_setA_nt.fas.gz',
        'vfdb.fa.gz' );
    system("gzip -f -d -c vfdb.fa.gz > vfdb.fa");
    my $seqs = load_fasta("vfdb.fa");

# >VFG000676(gb|AAD32411) (lef) anthrax toxin lethal factor precursor [Anthrax toxin (VF0142)] [Bacillus anthracis str. Sterne]
    for my $s (@$seqs) {

# https://github.com/tseemann/abricate/issues/64#issuecomment-421895159 by @VGalata
        $s->{ID} =~ m/^(\w+)\(\w+\|(\w+)(\.\d+)?\)$/;    #
            #$s->{ID} =~ m/^(\w+)\(\w+\|(\w+)\)$/;
        $s->{ACC} = $2 if $2;
        $s->{DESC} =~ m/^\((.*?)\)/;
        $s->{ID} = $1 if $1;

        #    print STDERR Dumper($s); exit;
    }

    return $seqs;
}

#..............................................................................
sub get_ncbibetalactamase {
    my $fasta = "ncbi.fa";
    download(
        'ftp://ftp.ncbi.nlm.nih.gov/pathogen/betalactamases/Allele-dna.fa',
        $fasta );
    my $tab = "ncbi.tab";
    download( 'ftp://ftp.ncbi.nlm.nih.gov/pathogen/betalactamases/Allele.tab',
        $tab );

    # >ACD12694.1 EU650653.1:1-1173
    my $seqs = load_fasta($fasta);

# ACC-1 ACD12694.1 EU650653.1 blaACC-1 1 1173 + cephalosporin-hydrolyzing class C beta-lactamase ACC-1
    my %anno;
    my @anno = grep { !m/^#/ } path($tab)->lines( { chomp => 1 } );
    msg( "Read", 0 + @anno, "annotations" );
    foreach (@anno) {
        my ( $name, $id, $acc, $gene, $begin, $end, undef, $product ) =
          split m/\t/;
        $anno{$id} = {
            ID   => $gene,
            DESC => $product,
            ACC  => "$acc:$begin-$end",
        };
    }

    #  print Dumper(\%anno);

    for my $s (@$seqs) {
        my $id = $s->{ID};
        next unless exists $anno{$id};
        $s->{ID}   = $anno{$id}{ID};
        $s->{ACC}  = $anno{$id}{ACC};
        $s->{DESC} = $anno{$id}{DESC};
    }

    #  print Dumper($seqs);

    return $seqs;
}

#..............................................................................
sub get_ecoh {
    my $fasta = "EcOH.fa";
    download(
'https://raw.githubusercontent.com/katholt/srst2/master/data/EcOH.fasta',
        $fasta
    );

# https://github.com/katholt/srst2#generating-srst2-compatible-clustered-database-from-raw-sequences
# [clusterID]__[gene]__[allele]__[seqID] [other stuff]
# >1__fliC__fliC-H1__1 AB028471.1;flagellin;H1
# >8__wzx__wzx-O41__246 AB811617.1;O antigen flippase;O41
# >9__wzy__wzy-OgN31__597 LC125932.1;O antigen polyermase;OgN31
    my $seqs = load_fasta($fasta);

    for my $s (@$seqs) {
        my @id   = split m/__/, $s->{ID};
        my @desc = split m';',  $s->{DESC};
        $s->{ID}   = $id[2];
        $s->{ACC}  = shift(@desc);
        $s->{DESC} = join( ' ', @desc );
    }

    #  print Dumper($seqs);
    return $seqs;
}

#..............................................................................
sub get_ecoli_vf {
    my $fasta = "ecoli_vf.ffn";
    download(
'https://github.com/phac-nml/ecoli_vf/raw/master/data/repaired_ecoli_vfs_shortnames.ffn',
        $fasta
    );
    my $seqs = load_fasta($fasta);

# >VFG000748(gi:2865308) (espF) EspF [EspF (VF0182)] [Escherichia coli O127:H6 str. E2348/69]
# >VFG000749(gi:6009379) (bfpA) Bundlin [BFP (VF0174)] [Escherichia coli B171]
# >SPG000142 (cvac) Escherichia coli cvi cvaC operon. [X57525 434-745]
# >SPG000143 (iss2) Escherichia coli Iss (iss) gene, complete cds. [AF042279 292-600]

    for my $s (@$seqs) {

        #print STDERR Dumper("IN", $s);
        $s->{ID} =~ m/ ^ (\w+) (?: \( (.*?) \) )? $ /x
          or die "Can't parse $fasta at " . Dumper($s);
        $s->{ID}  = $1 if $1;
        $s->{ACC} = $2 || $1;
        $s->{DESC} =~ s/\s\[.*?\]$//g;               # remove strain name
        $s->{DESC} =~ m/^(?:\((.*?)\)\s+)?(.*?)$/;
        $s->{ID}   = $1 if $1;
        $s->{DESC} = $2;

        #print STDERR Dumper("OUT", $s);
        #print STDERR "="x60, "\n";
    }

    #  print Dumper($seqs);
    return $seqs;
}

#..............................................................................
sub is_full_gene {
    my ($s) = @_;
    my $has_ambig = 0;

    my $id = $s->{ID};
    my $L  = length( $s->{SEQ} );
    if ( $L % 3 != 0 ) {
        wrn("$id - length $L bp is not multiple of 3");
        return;
    }
    if ( $s->{SEQ} !~ m/^[AGCT]+$/ ) {
        wrn("$id - has non-AGTC bases");
        return;
    }

    my $seq = Bio::Seq->new( -id => $s->{ID}, -seq => $s->{SEQ} );

    my $aa = $seq->translate->seq;

    if ( $aa =~ m/\*./ ) {
        wrn("$id - has internal stop codons, trying revcom");
        $aa = $seq->revcom->translate->seq;
        if ( $aa =~ m/\*./ ) {
            wrn("$id - revcom has internal stop codons too");
            return;
        }
        else {
            msg("$id - revcom resolves problem, hooray!");
            $s->{SEQ} = $seq->revcom->seq;
        }
    }

    return $L;
}

#..............................................................................
sub dedupe_seq {
    my ($seq) = @_;
    my %seen;
    my $good = [];
    for my $s (@$seq) {
        if ( $seen{ $s->{SEQ} } ) {
            wrn( "duplicate", length( $s->{SEQ} ),
                "bp sequence:", $s->{ID}, '~', $seen{ $s->{SEQ} } );
        }
        else {
            push @$good, $s;
        }
        $seen{ $s->{SEQ} } .= ' ' . $s->{ID};
    }
    msg( "dedupe_seq: read", scalar(@$seq), "/ kept", scalar(@$good) );
    return $good;
}

#..............................................................................
sub load_tabular {
    my ( $fname, $keycol, $sep ) = @_;
    $keycol //= 0;
    $sep    //= "\t";
    my $hash = {};
    my @hdr;
    my $row = 0;
    open my $TSV, '<', $fname or err("Can't read TSV file: $fname");
    while (<$TSV>) {
        chomp;
        my @col = split m/$sep/;
        $row++;
        if (@hdr) {
            @hdr == @col or err("Header and row $row differ in column count");

       #my $key = $col[$keycol] or wrn("Empty key column $keycol: $_");
       #exists{$hash->{$col[$key}} and wrn("WARNING: dupe key $key at row: $_");
            my $key = $col[$keycol];
            $hash->{$key} ||=
              { map { ( $hdr[$_] => $col[$_] ) } ( 0 .. $#hdr ) };
        }
        else {
            @hdr = @col;
        }
    }
    close $TSV;
    return $hash;
}

#..............................................................................
sub load_fasta {
    my ($fasta) = @_;
    my %seen;
    my $list;
    my $dbtype = 'unknown';
    msg("load_fasta: $fasta");
    my $in = Bio::SeqIO->new( -file => $fasta, -format => 'fasta' );
    while ( my $seq = $in->next_seq ) {
        my $id = $seq->id or err("Empty ID in $fasta");
        if ( $seen{$id} ) {
            wrn("Duplicate ID '$id' in $fasta");
            $id = $id . '_dupe';
        }
        $seen{$id}++;
        my $s = uc( $seq->seq );
        $dbtype = $seq->alphabet eq 'dna' ? 'nucl' : 'prot';
        $dbtype eq 'nucl' ? $s =~ s/[^AGTC]/N/g : $s =~ s/[^A-Z]/X/g;
        push @$list,
          {
            ID   => $id,
            ACC  => '',
            DESC => $seq->desc,
            SEQ  => $s,
            TYPE => $dbtype,
          };
    }
    msg( "load_fasta: read", scalar(@$list), "$dbtype sequences" );
    return $list;
}

#..............................................................................
sub save_fasta {
    my ( $fasta, $seq ) = @_;
    msg("save_fasta: $fasta");
    my %seen;
    my $out = Bio::SeqIO->new( -file => ">$fasta", -format => 'fasta' );
    for my $s (@$seq) {
        $seen{ $s->{ID} }++;
        my $freq = $seen{ $s->{ID} };

        #wrn("seen $s->{ID} now $freq times") if $freq > 1;
        #    print Dumper($s);
        my $ABX =
          defined( $s->{ABX} ) ? join( $ABX_SEP, sort @{ $s->{ABX} } ) : '';
        $ABX =~ s/\s+/_/g;    # remove spaces!
        my $obj = Bio::Seq->new(
            -id   => join( '~~~', $db, $s->{ID}, $s->{ACC}, $ABX ),
            -desc => ( $s->{DESC} || $s->{ID} ),
            -seq  => $s->{SEQ},
        );

        #    $obj->desc( hash_encode($s) );
        $out->write_seq($obj);

        #    $seen{ $s->{ID} }++;
    }
    msg( "save_fasta: wrote", scalar(@$seq), "sequences" );
}

#----------------------------------------------------------------------
sub msg {
    print STDERR "@_\n";
}

#----------------------------------------------------------------------
sub wrn {
    msg( "WARNING:", @_ ) if $debug;
}

#----------------------------------------------------------------------
sub err {
    msg( "ERROR:", @_ );
    exit(1);
}

#----------------------------------------------------------------------
# Option setting routines

sub setOptions {
    use Getopt::Long;

    @Options = (
        { OPT => "help", VAR => \&usage, DESC => "This help" },
        {
            OPT     => "debug!",
            VAR     => \$debug,
            DEFAULT => 0,
            DESC    => "Verbose debug output"
        },
        {
            OPT     => "dbdir=s",
            VAR     => \$outdir,
            DEFAULT => abs_path("$FindBin::RealBin/../db"),
            DESC    => "Parent folder"
        },
        {
            OPT     => "db=s",
            VAR     => \$db,
            DEFAULT => "",
            DESC    => "Choices: $DATABASES"
        },
        {
            OPT     => "force!",
            VAR     => \$force,
            DEFAULT => 0,
            DESC    => "Force download even if exists"
        },
    );

    &GetOptions( map { $_->{OPT}, $_->{VAR} } @Options ) || usage(1);

    # Now setup default values.
    foreach (@Options) {
        if ( defined( $_->{DEFAULT} ) && !defined( ${ $_->{VAR} } ) ) {
            ${ $_->{VAR} } = $_->{DEFAULT};
        }
    }
}

sub usage {
    my ($exitcode) = @_;
    $exitcode = 0 if $exitcode eq 'help';  # what gets passed by getopt func ref
    $exitcode ||= 0;
    select STDERR if $exitcode;    # write to STDERR if exitcode is error

    print "SYNOPIS\n  Download databases for abricate to use\n";
    print "USAGE\n  $EXE [options] --db DATABASE\n";
    print "OPTIONS\n";
    foreach (@Options) {
        printf "  --%-13s %s%s.\n", $_->{OPT}, $_->{DESC},
          defined( $_->{DEFAULT} ) ? " (default '$_->{DEFAULT}')" : "";
    }
    exit($exitcode);
}

#----------------------------------------------------------------------