package PubMed::Dain;

use strict;
use warnings;

use Carp;
use Data::Dumper;
use List::Util qw(shuffle);
use File::Basename;
use File::Compare;
use File::Copy;
use File::Slurp;
use File::Temp;

use PubMed::Common;
use PubMed::Paths;
use PubMed::Index;

my $dain_dir=$PubMed::Paths::dain_dir;
## my $dich_dir=$PubMed::Paths::dich_dir;
my $cut=$PubMed::Paths::cut;
my $test=&PubMed::Common::isatty();

## just a wrapper for update_line
sub update_file {
  my $file=shift;
  if(not -f $file) {
    if($test) {
      print "I don't have the file $file\n";
    }
    return;
  }
  open(F,"< $file");
  my $tmp = File::Temp->new(UNLINK => 0);
  my $tmp_filename=$tmp->filename();
  my $line;
  my $count_changes=0;
  while($line=<F>) {
    chomp $line;
    my $check=&update_line($line,$file);
    ## no check is a success
    if(not $check) {
      print $tmp "$line\n";
      next;
    }
    if($check ne $line) {
      if($test) {
        print "I replace '$line' by '$check'\n";
      }
      $count_changes++;
    }
    print $tmp "$check\n";
  }
  if($test) {
    print "I counted $count_changes changes.\n";
  }
  if(not $count_changes) {
    if($test) {
      print "The file $file is not changed.\n";
    }
    unlink $tmp_filename;
    return;
  }
  if($test) {
    print "I copy $tmp_filename to $file.\n";
  }
  $tmp->close();
  copy($tmp_filename,$file);
  unlink($tmp_filename);
}

sub update_line {
  my $line=shift;
  ## just for reporting
  my $file=shift;
  if($line=~m|^(\d+)$|) {
    my $pmid=$1;
    ## fapi case, obsolete
    #my $purd=&PubMed::Api::get_purd_by_pmid($pmid);
    #if($purd) {
    #  ## return '' means line is ok
    #  return '';
    #}
    my $loc=&PubMed::Index::check_purd_by_pmid($pmid);
    if(not $loc) {
      confess "I have an unknown pmid in the file $file.";
    }
    if(not $loc=~m|^\d+ (\d+ \d+ \d{2}n\d{4}\s*)+$|) {
      confess "I found a bad loc '$loc'";
    }
    return $loc;
  }
  $line=~m|^(\d+) | // confess "I don't like line '$line'";
  my $pmid=$1;
  my $latest_line=&PubMed::Index::check_purd_by_pmid($pmid);
  my $loc=&PubMed::Index::get_latest_loc($latest_line);
  return $loc;
}

## ### dich files contain an index what pmid is in what dain file
## #sub refresh_dich_files {
## #  unlink glob("$dich_dir/*");
## #  foreach my $dain_fufi (glob("$dain_dir/*")) {
## #    my $date=basename($dain_fufi);
## #    ## deletes some member of $date_infi
## #    #&deal_with_date($date);
## #    my $s="$cut -d ' ' -f 1 $dain_fufi";
## #    foreach my $pmid (`$s`) {
## #      chomp $pmid;
## #      my $pemi=&PubMed::Common::pmid_to_pemi($pmid); 
## #      my $dich_file="$dich_dir/$pemi";
## #      open(F,">> $dich_file");
## #      print F "$pmid $date\n";
## #      close(F);
## #    }  
## #  }
## #}

## #sub check_dich {
## #  unlink glob("$dich_dir/????-??-??");
## #  foreach my $dich_fufi (glob("$dich_dir/*")) {
## #    my $dich_bana=basename($dich_fufi);
## #    if(not $dich_bana=~m|^\d+$|) {
## #      next;
## #    }
## #    &check_dich_file($dich_fufi);
## #  }
## #}
## #
## #sub check_dich_file {
## #  my $dich_file=shift;
## #  my $seen={};
## #  open(F,"< $dich_file");
## #  my $line;
## #  while($line=<F>) {
## #    $line=~m|^(\d+) (\d{4}-\d{2}-\d{2})| or confess "This is a bad line '$line'";
## #    my $pmid=$1;
## #    my $date=$2;
## #    my $old_date=$seen->{$pmid} // '';
## #    if($seen->{$pmid}) {
## #      print "pmid $pmid is duplicated in $old_date and $date\n";
## #      open(O,">> $dich_dir/$date");
## #      print O "$pmid\n";
## #      close O;
## #    }
## #    $seen->{$pmid}=$date;
## #  }
## #}
## #
## #sub correct_dain_from_dich {
## #  foreach my $dich_fufi (glob("$dich_dir/*")) {
## #    my $dich_bana=basename($dich_fufi);
## #    if(not $dich_bana=~m|^\d{4}-\d{2}-\d{2}$|) {
## #      next;
## #    }
## #    &delete_from_dain($dich_fufi);
## #  }
## #}
## #
## #sub delete_from_dain {
## #  my $dich_file=shift;
## #  open(F,"< $dich_file"); 
## #  my $pmid;
## #  my $to_delete={};
## #  my $date=basename($dich_file);
## #  while($pmid=<F>) {
## #    chomp $pmid;
## #    $to_delete->{$pmid}=1;
## #  }
## #  close F;
## #  my $tmp = File::Temp->new(UNLINK => 0);
## #  my $tmp_filename=$tmp->filename();
## #  my $dain_file="$dain_dir/$date";
## #  if(not -f $dain_file) {
## #    confess "I don't see the file '$dain_file'";
## #  }
## #  open(F,"< $dain_file");
## #  my $dain;
## #  while($dain=<F>) {
## #    chomp $dain;
## #    $dain=~m|(\d+)| or confess "I have a bad line.";
## #    my $pmid=$1;
## #    if(defined($to_delete->{$pmid})) {
## #      if($test) {
## #        #print "I remove $pmid from $dain_file\n";
## #      }
## #      next;
## #    }
## #    print $tmp "$dain\n";
## #  }
## #  $tmp->close();
## #  copy($tmp_filename,$dain_file);
## #  unlink $tmp_filename;
## #}


1;