package Ernad::Presort::Predi;

use strict;
use warnings;

use Carp qw(cluck longmess shortmess croak confess);
use Data::Dumper;
use File::Basename;
use File::Slurper;
use URI::Escape;
use Krichel::File;

use base ('Ernad::Presorter');

use Ernad::Presort::Class;
use Ernad::Presort::Ranfi;

binmode(STDOUT,':utf8');

## returns the repcode
sub setup {
  my $p=shift;
  #my $a=Ernad::Presort::Class->new({'e'=>$p->{'e'}});
  #my $p->{'dir'}->{'class'}=$
  my $learn_dir=$p->{'e'}->{'dir'}->{'learn'};
  ## set_dir can't be used here, as we use the report
  my $repcode=$p->get_repcode();
  $p->{'dir'}->{'class'}=$learn_dir.'/class/'.$repcode;
  my $m=Ernad::Presort::Model->new({'e'=>$p->{'e'}});
  $p->{'m'}=$m;
  $p->{'ranid'}=$p->{'m'}->last_ranid();
  $p->{'file'}->{'model'}=$p->{'m'}->most_recent();
}

## callable externally
sub type {
  my $p=shift;
  ## restrict
  my $given_papid=shift // '';
  if(not $p->{'ranid'}) {
    $p->setup();
  }
  if(not $p->{'ranid'}) {
    confess "I don't see the ranid.";
  }
  my $ranid=$p->{'ranid'};
  ## should be defined externally
  my $issuedate=$main::issuedate // confess "I don't have an issuedate.";
  my $test_file=$p->{'dir'}->{'class'}.'/'.$issuedate.'_'.$p->{'ranid'}.".test";
  if(not -f $test_file) {
    confess "I don't see your test file $test_file.";
  }
  my $r=Ernad::Presort::Ranfi->new({'e'=>$p->{'e'}});
  my $firan=$p->invert_ranfi($r->load($ranid));
  open(T,"< $test_file") or confess "I can't open your test file $test_file.";
  my $vemli;
  while($vemli=<T>) {
    $vemli=~m|0\s+([^#]+)\s+# (\S+)$| or confess "I can't parse $vemli.";
    my $data=$1;
    ## handle
    my $papid=uri_unescape($2);
    if($given_papid and $papid ne $given_papid) {
      next;
    }
    my $out='';
    foreach my $fitpart (split(/\s+/,$data)) {
      $fitpart=~m|^(\d+):([.\d]+)$| or confess "I don't get your fitpart $fitpart";
      my $fit=$1 or confess "I need a fit here.";
      my $weight=$2;
      if(not $firan->{$fit}) {
        confess "I have no fit rank for fit $fit";
      }
      $out.=$firan->{$fit}.' ';
    }
    chop $out;
    print "$out\n";
  }
}

## maybe should go to ranfi
sub invert_ranfi {
  my $p=shift;
  my $ranfi=shift;
  my $firan={};
  foreach my $fit (keys %$ranfi) {
    my $rank=$ranfi->{$fit};
    $firan->{$rank}=$fit;
  }
  return $firan;
}

sub run {
  my $p=shift;
  $p->setup;
  my $glob=$p->{'dir'}->{'class'}.'/*'.$p->{'ranid'}.".test";
  foreach my $test_file (glob($glob)) {
    $p->run_file($test_file);
  }
  ## clear the old files in the directory
  $glob=$p->{'dir'}->{'class'}.'/*';
  my $regx='^\d{4}-\d{2}-\d{2}_'.$p->{'ranid'};
  foreach my $file (glob($glob)) {
    my $bana=basename($file);
    if(not $bana=~m|$regx|) {
      $p->echo(__LINE__,"I remove the non-current file $bana.");
      unlink $file;
    }
  }
}

## gather results from teh side file
sub gather {
  my $p=shift;
  $p->setup;
  my $ranid=$p->{'ranid'} // confess "I need a ranid here.";
  my $glob=$p->{'dir'}->{'class'}.'/*';
  $p->{'data'}=undef;
  foreach my $file (glob($glob)) {
    my $bana=basename($file);
    if($ranid and not $bana=~m|^\d{4}-\d{2}-\d{2}_$ranid|) {
      $p->echo(__LINE__,"I remove a non-current ranid file $bana.");
      #unlink $file;
      next;
    }
    if(not $bana=~m|\.side$|) {
      next;
    }
    my $e=$p->{'e'} // $main::e // confess 'Where is my erimp?';
    #my $issuedate=&Ernad::Common::find_issuedate_from_file($file) //
    #  confess "I need an issuedate here.";
    my $issuedate=$e->{'f'}->issuedate($file) // confess "I need an issuedate here.";
    $p->{'data'}->{'files'}->{$issuedate}=$bana;
    my $side_fh=IO::File->new("< $file");
    my $line;
    while($line=<$side_fh>) {
      chomp $line;
      my @pair=split("\t",$line);
      my $value=$pair[0];
      my $handle=uri_unescape($pair[1]);
      $p->{'data'}->{'scores'}->{$issuedate}->{$handle}=$value;
    }
  }
  return $p->{'data'};
}

## code that was in Presort.pm
sub check {
  ##
  #my $class_file=$m->{'class_file'}->{$repcode}->{$issuedate} // '';
  #if(not $class_file or not -f $class_file) {
  #  confess "I can't see class_file $class_file. It looks like the prediction failed.";
  #}
  #$e->echo(__LINE__,"I found the class file $class_file");
  #my $count_lines_in_class_file=`grep -c ^ $class_file`;
  #chomp $count_lines_in_class_file;
  #$e->echo(__LINE__,"issuedate is $issuedate");
  #my $items_classed=$count_lines_in_class_file-1;
  #$e->echo(__LINE__,"class_file $class_file contains $items_classed items");
  #my $total_unsorted_texts=&Ernad::Common::count_texts_in_rif($unsorted_doc);
  #$e->echo(__LINE__,"The \$rerc->{amf_doc} contains $total_unsorted_texts texts");
  #if($items_classed != $total_unsorted_texts) {
  #  confess "$items_classed are classed but there are $total_unsorted_texts unsorted texts";
  #}
}

sub run_file {
  my $p=shift;
  my $test_file=shift;
  if(not -f $test_file) {
    confess "I don't see your file $test_file";
  }
  my $flags=shift // '';
  ## $p->check_test_file($test_file);
  my $model_file=$p->{'file'}->{'model'} or confess "I can't see a model file.";
  my $model_dir=dirname($model_file);
  my $class_file=$test_file;
  $class_file=~s|\.test$|\.class|;
  my $out_file=$class_file;
  $out_file=~s|\.class|.out|;
  my $err_file=$class_file;
  $err_file=~s|\.class|.err|;
  $flags.=" -b 1";
  if(not &Ernad::Common::does_file_need_renewal($class_file,$model_file,$test_file)) {
    $p->echo(__LINE__,"I skip the renewal of $class_file, it needs no renewal.",2);
    return 0;
  }
  else {
    ## request probababilites
    my $s="svm-predict $flags $test_file $model_file $class_file > $out_file 2> $err_file";
    $p->echo(__LINE__,"I run $s");
    system($s);
  }
  if(not -z $err_file) {
    my $error=&File::Slurper::read_text($err_file);
    chomp $error;
    confess "smv_predict gives me an error '$error'";
  }
  ## conduct a check
  my $count_lines_in_test=&Krichel::File::count_lines($test_file);
  my $count_lines_in_class=&Krichel::File::count_lines($class_file);
  if($count_lines_in_test != ($count_lines_in_class-1)) {
    confess "$test_file has $count_lines_in_test but $class_file has $count_lines_in_class";
  }
  ## used for a check in the caller
  #$p->{'class_file'}->{$repcode}->{$issuedate}="$class_file";
  ## this is a debugging tool
  my $scores=$p->write_side_file($class_file,$test_file);
  return $scores;
}

sub write_side_file {
  my $p=shift;
  my $class_file=shift;
  my $test_file=shift;
  my @class_lines=&File::Slurper::read_lines($class_file);
  ## first line is a comment
  my $first_line=shift @class_lines;
  chomp $first_line;
  if($first_line eq 'labels -1 1') {
    map ~s|\S+\s+\S+\s+(\S+)|$1|g, @class_lines;
  }
  elsif($first_line eq 'labels 1 -1') {
    map ~s|\S+\s+(\S+)\s+\S+|$1|g, @class_lines;
  }
  else {
    confess "I found '$first_line' in class_file $class_file.";
  }
  my $side_file=$class_file;
  $side_file=~s|\.class$|\.side|;
  my @test_lines=&File::Slurper::read_lines($test_file);
  #print Dumper @test_lines;
  #print "\n\n";
  map ~s|[^#]+#\s*||g, @test_lines;
  #print Dumper @test_lines;
  my $count=0;
  my $out;
  my $scores;
  foreach my $line (@test_lines) {
    chomp $line;
    my $handle=uri_unescape($line);
    my $score=$class_lines[$count];
    $out.="$score\t$line\n";
    $scores->{$handle}=$score;
    $count++;
  }
  &File::Slurper::write_text($side_file,$out);
  return $scores;
}


1;