package Ernad::Presort::Train;

use strict;
use warnings;

use Carp qw(cluck longmess shortmess croak confess);
use Data::Dumper;
use File::Basename;
use File::Path;

use base ('Ernad::Presorter');

use Ernad::Presort::Chopa;
use Ernad::Presort::Ranfi;
use Ernad::Presort::Vemli;

binmode(STDOUT,':utf8');

## returns the repcode
sub setup {
  my $a=shift;
  my $h=Ernad::Presort::Chopa->new({'e'=>$a->{'e'}});
  $a->{'h'}=$h;
  my $r=Ernad::Presort::Ranfi->new({'e'=>$a->{'e'}});
  $a->{'r'}=$r;
  my $v=Ernad::Presort::Vemli->new({'e'=>$a->{'e'}});
  $a->{'v'}=$v;
  $a->{'v'}->motto('train');
}

## callable external
sub build_file {
  my $a=shift;
  ## check for setup
  if(not $a->{'h'}) {
    $a->setup();
  }
  $a->setup_for_report();
  my $ranid=$a->{'ranid'};
  my $shoti=$a->{'shoti'};
  my $train_file=$a->{'dir'}->{'model'}."/$shoti"."_$ranid.train";
  $a->{'file'}->{'train'}=$train_file;
  ## FixMe: this needs more careful checking if the data is up-to-date
  if(-f $train_file and not -z $train_file) {
    $a->echo(__LINE__,"I use the existing $train_file");
    return $train_file;
  }
  $a->{'fh'}->{'train'}=IO::File->new("> $train_file");
  my $count_papers=0;
  ## find the files
  my $files=$a->{'v'}->find($ranid);
  my $l=$a->{'v'}->next_paper();
  while(defined($l)) {
    ## end of a file reached
    if(not ref $l) {
      $l=$a->{'v'}->next_paper();
      next;
    }
    if(not defined($l)) {
      last;
    }
    if(not $l) {
      next;
    }
    ## code to deal with paper
    $a->deal_with_paper($l);
    ##
    $l=$a->{'v'}->next_paper();
  }
  $a->echo(__LINE__,"I wrote $train_file");
  $a->{'fh'}->{'train'}->close();
  return $train_file;
}

sub deal_with_paper {
  my $a=shift;
  my $l=shift // confess "I need a paper here.";
  if(not $l->{'i'}) {
    confess "My papers needs a date.";
  }
  ## this function will take care of error handling and return '' on error
  my $status=$a->{'h'}->status_by_handle_and_issuedate($l->{'i'},$l->{'h'}) or return;
  my $fh=$a->{'fh'}->{'train'};
  print $fh "$status ". $l->{'d'}.' # '.$l->{'u'}."\n";
}

sub setup_for_report {
  my $a=shift;
  my $repcode=$a->get_repcode() // confess "I can't get the repcode.";
  ## list the vemli files.
  my $e=$a->{'e'} // confess "I need an erimp here.";
  my $learn_dir=$e->{'dir'}->{'learn'};
  ## set_dir can't be used here, as we use the report
  my $model_dir=$learn_dir.'/model/'.$repcode;
  if(not -d $model_dir) {
    $a->echo(__LINE__,"I make $model_dir.");
    mkpath($model_dir);
  }
  $a->{'dir'}->{'model'}=$model_dir;
  $a->{'ranid'}=$a->{'r'}->find_last_ranid() // confess "I need a ranid here.";
  $a->{'shoti'}=$a->{'r'}->find_last_shoti() // confess "I need a shoti here.";
}

sub train_flags {
  my $a=shift;
  my $train_flags=$a->{'e'}->{'conf'}->{'svm_train_flags'} // '';
  if($train_flags) {
    $a->{'train_flags'}=$train_flags;
    $a->echo(__LINE__,"train_flags set to $train_flags",4);
  }
  else {
    confess "I need svm_train_flags in the ernad configuration.";
  }
}

1;