package PubMed::Search;

use strict;
use warnings;

use Carp qw(confess);
use Data::Dumper;
use Fcntl qw(SEEK_SET O_RDONLY SEEK_CUR);
use IO::File;

## go to the start of the line. I don't seem to
## find a build-in function to do that
sub get_line {
  my $fh=shift;
  my $offset=shift // 0;
  if($offset) {
    $fh->seek($offset,SEEK_SET);
  }
  my $start = $fh->tell();
  my $count=0;
  my $line;
  while($count<$start) {
    if(not $fh->seek(-2,SEEK_CUR)) {      
      $fh->seek(-1,SEEK_CUR);
      my $line=$fh->getline();
      chomp $line;
      return $line;
    }
    my $char=$fh->getc();
    if(not defined($char)) {
      my $file_info=Dumper $fh;            
      confess "I could not get the char from $file_info.";
    }
    my $ord=ord($char);
    if($ord == 10) {
      $line=$fh->getline;
      chomp $line;
      return $line;
    }
  }
  return '';
}

sub get_pmid {
  my $in=shift;
  my $blank_pos=index($in,' ') // confess "I need a blank in '$in'";
  my $pmid=substr($in,0,$blank_pos);
  return $pmid;
}

## This searches for the target pmid, supported to be followed by
## a blank. The file's lines start with the pmids. Pmids are 
## sorted by string order. 
sub search {
  my $file=shift;
  my $target=shift;
  my $fh = IO::File->new($file, O_RDONLY)
    or confess "I could not open the file '$file': $!";
  my @stat=$fh->stat();
  my ($lo, $hi) = (0,$stat[7]);
  my $old_mid=0;
  while(1) {
    my $mid = int(($lo + $hi)/2);
    $old_mid=$mid;
    $fh->seek($mid,SEEK_SET);
    my $line = &get_line($fh);
    my $pmid = &get_pmid($line);
    if($pmid eq $target) {
      return $line;
    }
    my $diff=$pmid cmp $target;
    my $start = $fh->tell();
    my $length = length($line);
    if($diff < 0) {
      $lo=$mid+1;
      my $lo_line=get_line($fh,$lo);
    }
    else {
      $hi=$mid;
      my $hi_line=get_line($fh,$hi);
    }
    if($hi <= $lo) {
      ## failed to find the pmid
      return '';
    }
  }
}

1;