#!/usr/bin/perl

use strict 'vars';
use LWP::UserAgent;
use URI::Escape;
use Getopt::Long;
use strict;
use constant GENBANK_URL=>"http://www.ncbi.nlm.nih.gov/htbin-post/Entrez/query?form=4&html=no";
use constant CHUNKSIZE=>10000;
my ($TERM,$AGE,$MAX,$DB,$VERBOSE,$BUFFER,$COUNT,$GOT,$GOT_TOTAL);

GetOptions('max=i'  => \$MAX,
	   'db=s'   =>\$DB,
	   'age=i'  => \$AGE,
	   'count'  => \$COUNT,
	   'verbose!'=>\$VERBOSE) || die <<USAGE;
Usage: $0 [options] <query string>
Query GenBank for a list of accession numbers.  The query
string should be in the form recognized by  NCBI\'s term parser. 
See http://www.ncbi.nlm.nih.gov/Entrez/linking.html
for examples.

Options:
       -db  <database> Database to search (n)
       -max <max>      Max entries to return (100)
       -age <days>     Only fetch accessions entered <days> ago
       -verbose        Show brief description line
       -count          Just retrieve the count that would be retrieved

Database specifiers:
  m  MEDLINE
  p  Protein
  n  Nucleotide
  t  3-D structure
  c  Genome
Example search:
  gb_search -verbose -db n 'Oryza sativa[Organism] AND EST[Keyword]'
Some common field modifiers:
  [All Fields]
  [Accession]
  [Author Name]
  [Feature Key]
  [Gene Name]
  [Keyword]
  [Organism]
USAGE
;

$MAX ||= 100;
$DB  ||= 'n';
$DB = lc(substr($DB,0,1));  # first character, whatever it is
$BUFFER = '';
my $max = $MAX > CHUNKSIZE ? CHUNKSIZE : $MAX;

$TERM = GENBANK_URL . "&db=$DB&dispmax=$max&term=" . uri_escape("@ARGV");
$TERM .= "&relpubdate=$AGE" if $AGE > 0;
$TERM .= "&Dopt=" . ($COUNT ? 'q' : 'd');
$GOT_TOTAL=0;
my $agent = LWP::UserAgent->new();

while ($GOT_TOTAL < $MAX) {
  $GOT = 0;
  my $req   = HTTP::Request->new('GET' => "$TERM&dispstart=$GOT_TOTAL");
  my $response = $agent->request($req,\&process_text);
  die "Request failure: ",$response->status_line 
    unless $response->is_success;
  last if $GOT < CHUNKSIZE;
  $GOT_TOTAL += $GOT;
}
exit 0;

sub process_text ($$$) {
  my ($chunk,$response,$protocol) = @_;
  my ($position,$remainder);
  $BUFFER .= $chunk;

  if ($COUNT) {
    print $1,"\n" if $BUFFER=~/^(\d+)\n/;
    $GOT = $MAX;
    return;
  }
  elsif ($DB =~ /^[mt]$/) {
    while ($BUFFER=~/^\s+(\d+) (.+)$/mg) {
      print $VERBOSE ? "$1\t$2" : $1,"\n";
      $position = pos($BUFFER);
      $GOT++;
    }
  }
  
  else {
    while ($BUFFER=~/^\s+\d+ (\w+)\n\n(.*)\ngi/mg) {
      print $VERBOSE ? "$1\t$2" : $1,"\n";
      $position = pos($BUFFER);
      $GOT++;
    }
  }
  substr($BUFFER,0,$position) = '';
  pos($BUFFER)=0;
}