#!/usr/local/bin/perl -w
use strict;

use Bio::Index::Fasta;
use Bio::SeqIO;
my $out = new Bio::SeqIO(-format => 'fasta');

my $indexfile = 'ecoli.idx';
my $dbfile    = '/usr/local/bioperl_course/db/ecoli.aa';

my $dbh = new Bio::Index::Fasta(-filename => $indexfile,
	                        -write_flag=>1);
$dbh->id_parser(\&parse_id);
$dbh->make_index($dbfile);

my $acc1 = 'gi|1786201|gb|AAC73130.1|';
my $acc2 = 'AAC73130.1';
foreach my $acc ($acc1,$acc2) {
 my $seq = $dbh->get_Seq_by_acc($acc);
 if( $seq ) { 
  $seq->display_id($acc);
  $out->write_seq($seq);
 } else { 
  print STDERR "cannot find $acc\n";
 }
}

sub parse_id {
  my $header = shift;
  if( $header =~ /^>\s*(\S+)/ ) {
    my @fields = split(/\|/,$1);
    my ($noacc) = ($fields[3] =~ /(\S+)\./);
    return ($fields[1],$fields[3],$noacc);
  }
}