package Ernad::Presort::Ranfi;

use strict;
use warnings;

use base ('Ernad::Presorter');

use Carp qw(confess);

use Data::Dumper;
use File::Basename;

use Krichel::Shonu;
use Krichel::Shoti;
use Krichel::File;

## main
sub update_for_report {
  my $r=shift;
  my $e=$r->{'e'} // $main::e // confess "I have no erimp.";
  my $report=shift // confess "I need a report set here.";
  my $k=Ernad::Presort::Fidek->new({'e'=>$r->{'e'}});
  $r->{'k'}=$k;
  $r->set_report($report);
  $k->set_report($report);
  my $i_stored=0;
  if(not $r->does_it_require_renewal()) {
    $e->echo(__LINE__,"The ranfi needs no update.");
    return 0;
  }
  my $can_rank_be_formed=$r->calculate_fitrank();
  if(not $can_rank_be_formed) {
    $e->echo(__LINE__,"The rank can not be found.");
    return 0;
    #next;
  }
  $i_stored++;
  $e->echo(__LINE__,"I store the rank.");
  $r->store();
  ## indicator of success
  return $i_stored;
}

## main
# sub update_for_report_with_fitcla {
#   my $r=shift;
#   my $report=shift // confess "I need a report set here.";
#   my $k=Ernad::Presort::Fidek->new({'e'=>$r->{'e'}});
#   $r->{'k'}=$k;
#   $r->set_report($report);
#   $k->set_report($report);
#   my $fitclas=$r->get_fitclas();
#   my $i_stored_for_a_fitcla=0;
#   foreach my $fitcla (keys %$fitclas) {
#     $k->set_fitcla($fitcla);
#     $r->{'k'}->set_fitcla($fitcla);
#     $r->setup_fidek();
#     if(not $r->does_it_require_renewal()) {
#       next;
#     }
#     my $can_rank_be_formed=$r->calculate_fitrank();
#     if(not $can_rank_be_formed) {
#       next;
#     }
#     $i_stored_for_a_fitcla++;
#     $r->store();
#   }
#   ## indicator of success
#   return $i_stored_for_a_fitcla;
# }

## returns 0 on failure
sub calculate_fitrank {
  my $r=shift;
  ## FixMe: only works for term
  my $fidek_files=$r->{'k'}->list('train');
  my $count_fits=0;
  foreach my $fidek_file (@$fidek_files) {
    my $fidek=$r->load_json_with_gz($fidek_file);
    $r->echo(__LINE__,"I load $fidek_file");
    foreach my $handle (keys %$fidek) {
      foreach my $fit (keys %{$fidek->{$handle}}) {
        ## count features rather than files to make sure
        ## we don't take account of empty fideks
        $count_fits++;
        if(not defined($r->{'f'}->{$fit})) {
          $r->{'f'}->{$fit}=$fidek->{$handle}->{$fit};
        }
        else {
          $r->{'f'}->{$fit}+=$fidek->{$handle}->{$fit};
        }
      }
    }
  }
  ## no features, return 0
  if(not $count_fits) {
    return 0;
  }
  ## trim
  $r->{'total'}=0;
  foreach my $fit (keys %{$r->{'f'}}) {
    if($r->{'f'}->{$fit}==1) {
      delete $r->{'f'}->{$fit};
      next;
    }
    $r->{'total'}+=$r->{'f'}->{$fit};
  }
  my  $sorter = sub {
    if($r->{'f'}->{$a} < $r->{'f'}->{$b}) {
      return 1;
    }
    if($r->{'f'}->{$a} > $r->{'f'}->{$b}) {
      return -1;
    }
    return $a cmp $b;
  };
  @{$r->{'r'}}=sort $sorter keys %{$r->{'f'}};
  delete $r->{'f'};
  my $count=1;
  my $rank={};
  foreach my $fit (@{$r->{'r'}}) {
    $rank->{$fit}=$count++;
  }
  $r->{'rank'}=$rank;
  delete $r->{'r'};
  $r->{'total'}=&Krichel::Shonu::make($r->{'total'});
  $r->{'time'}=&Krichel::Shoti::now();
  return 1;
}

sub store {
  my $r=shift;
  my $e=$r->{'e'} // $main::e // confess "I have no erimp.";
  my $learnport=$r->{'learnport'};
  if(not $learnport) {
    $learnport=$r->get_learnport() // confess "I need this here.";
  }
  my $out_dir=$r->set_dir('ranfi');
  if(not -d $out_dir) {
    mkpath $out_dir;
  }
  my $total=$r->{'total'} // confess "I must have a total here.";
  my $there_already=glob("$out_dir/*$total.json*");
  if($there_already) {
    delete $r->{'rank'};
    $e->echo(__LINE__,"It's already there $there_already.");
    return '';
  }
  my $time=$r->{'time'};
  my $out_file="$out_dir/$time"."_$total.json";
  $e->echo(__LINE__,"I write $out_file");
  $r->save_json_with_gz($r->{'rank'},$out_file);
  delete $r->{'rank'};
  return $out_file;
}

sub setup_fidek {
  my $r=shift;
  if(not $r->{'k'}) {
    $r->{'k'}=Ernad::Presort::Fidek->new({'e'=>$r->{'e'}});
    # $r->{'k'}->set_fitcla($r->{'fitcla'});
    $r->{'k'}->set_report($r->{'report'});
    $r->{'k'}->set_dir('fidek');
  }
  my $fidek_dir=$r->{'k'}->set_dir('fidek');
  my $list=$r->{'k'}->list('train');
  $r->{'fidek_files'}=$list;
  return $list;
}

## checks if there is a recent fidek
## that would require the ranfi to be updated
sub does_it_require_renewal {
  my $r=shift;
  my $e=$r->{'e'} // $main::e // confess "I have no erimp.";
  my $fidek_files=$r->setup_fidek();
  my @fidek_files=@$fidek_files;
  my $last=$r->find_last();
  if(not $last) {
    $e->echo(__LINE__,"There is no last ranfi.");
    return 1;
  }
  if(&Ernad::Common::does_file_need_renewal($last,@fidek_files)) {
    $e->echo(__LINE__,"The last ranfi $last needs renewal.");
    return 1;
  }
  $e->echo(__LINE__,"I don't need to renew the ranfi.");
  return 0;
}

sub find_last {
  my $r=shift;
  my $ranfi_dir=$r->set_dir('ranfi');
  my @ranfi_files=glob("$ranfi_dir/*json*.gz");
  ## since the time comes first in the bana, the pop should give us the last element
  my $last=pop(@ranfi_files);
  ## the array has been shortened.
  foreach my $file (@ranfi_files) {
    my $bana=basename($file);
    $r->echo(__LINE__,"I remove the non-current file $bana.");
    unlink $file;
  }
  $r->{'last'}=$last;
  return $last;
}

sub find_last_by_ranid {
  my $r=shift;
  my $ranid=shift // confess "I need to know what ranid you want.";
  my $ranfi_dir=$r->set_dir('ranfi');
  my @ranfi_files=glob("$ranfi_dir/*$ranid*json*.gz");
  my $last=pop(@ranfi_files) // return '';
  $r->{'last'}=$last;
  return $last;
}

sub find_last_ranid {
  my $r=shift;
  my $fufi;
  if(not $r->{'last'}) {
    $fufi=$r->find_last();
  }
  else {
    $fufi=$r->{'last'};
  }
  if(not $fufi) {
    return '';
  }
  my $bana=basename($fufi);
  if(not $bana=~m|^[a-z0-9]{6}_([a-z0-9]+)\.json\.gz$|) {
    $r->echo(__LINE__,"This $fufi does not look like a ranfi to me.");
  }
  my $ranid=$1;
  return $ranid;
}

sub find_last_shoti {
  my $r=shift;
  my $fufi;
  if(not $r->{'last'}) {
    $fufi=$r->find_last();
  }
  else {
    $fufi=$r->{'last'};
  }
  if(not $fufi) {
    return '';
  }
  my $bana=basename($fufi);
  if(not $bana=~m|^([a-z0-9]{6})_([a-z0-9]+)\.json\.gz$|) {
    confess "This $fufi does not look like a ranfi to me.";
  }
  my $shoti=$1;
  return $shoti;
}

#sub get_by_ranid {
#  my $r=shift;
#  my $ranid=shift // confess "I need to have a ranid here\n";
#  my $ranfi_dir=$r->{'k'}->set_dir('ranfi');
#  my @files=glob("$ranfi_dir/*_$ranid.json*");
#  if(not @files) {
#    confess "I don't see a ranfi of ranid $ranid\n";
#  }
#  ## just in case there are several ... I am being lax here.
#  my $last=pop(@files);
#  return $last;
#}

sub load {
  my $r=shift;
  my $ranid=shift // '';
  my $ranfi_file;
  if($ranid) {
    $ranfi_file=$r->find_last_by_ranid($ranid);
  }
  else {
    $ranfi_file=$r->find_last();
  }
  if(not $ranfi_file) {
    $r->echo(__LINE__,"I have no ranfi file.");
    return 0;
  }
  $r->{'rank'}=&Krichel::File::load($ranfi_file);
  return $r->{'rank'};
}


sub list_by {
  my $r=shift;
  my $what=shift // confess "by what?";
  my $ranfi_dir=$r->set_dir('ranfi');
  my @ranfi_files=glob("$ranfi_dir/*.json.gz");
  my $out;
  foreach my $ranfi_file (@ranfi_files) {
    my $bana=basename($ranfi_file);
    if(not $bana=~m|^([0-9a-z]{6})_([0-9a-z]+)\.json.gz|) {
      next;
    }
    if($what eq 'shoti') {
      $out->{$1}->{'file'}=$ranfi_file;
      $out->{$1}->{'ranid'}=$2;
    }
    elsif($what eq 'ranid') {
      $out->{$2}->{'file'}=$ranfi_file;
      $out->{$2}->{'shoti'}=$1;
    }
    else {
      confess "I don't know by what to list."
    }
  }
  return $out;
}

1;