view release on metacpan or  search on metacpan

lib/Book/Collate/Report.pm  view on Meta::CPAN

=head2 new

=cut

sub new {
 my ($class, %data) = @_;
  my $self    = {
    _data             => undef,
    _string           => $data{string},
    _words            => [],
    _fry_used         => undef,
    _custom_words     => $data{custom_words},
    _weak_words       => $data{weak_words},
  };
  bless $self, $class;

  
  (my $data_set   = $self->{_string}) =~ s/[;:!'"?.,]/ /g;
  $data_set       =~ s/ (d|ll|m|re|s|t|ve) / /g; 
  $data_set       =~ s/\s*(.*)\s*/$1/;
  $self->{_data}  = $data_set;
  $self->{_words} = [ split(/\s+/, $self->{_data}) ];
  $self->{_word_list}  = $self->word_list($self->{_words});
  $self->{_fry_used} = $self->_generate_fry_stats();
  $self->{_weak_used} = $self->_generate_weak_used();
  
  return $self; 
}

=head2 avg_sentence_length

Returns the average sentence length

=cut

sub avg_sentence_length {
  my $self = shift;
  return $self->word_count / $self->sentence_count ;
}


=head2 avg_word_length

Returns the average word length

=cut

sub avg_word_length {
  my $self      = shift;
  my $character_count = 0;
  foreach my $word ( $self->words() ){
    $character_count += length($word);
  }
  return $character_count / $self->word_count ;
}


=head2 _generate_fry_stats

Gives a percentage of Fry list words used against the total unique words used.

=cut

sub _generate_fry_stats {
  my $self = shift;
  my %word_list = $self->word_list($self->{_words});
  my %custom_words; 
  if ( defined( $self->{_custom_words} ) ){
    %custom_words = %{$self->{_custom_words} };
  }
  my %fry_words = %Book::Collate::Words::fry;
  my %used_words;
  my %missed;
  my %fry_used = ( 
    fry     => 0,
    custom  => 0,
    miss    => 0,
  );  
  foreach my $word ( keys %word_list ){
    $word = Book::Collate::Utils::scrub_word($word);
    $used_words{$word} = 1;
  }   

  foreach my $word ( keys %used_words ){
    if ( defined($fry_words{$word}) ){
      $fry_used{fry}++;
    } elsif ( defined($custom_words{$word}) ){
      $fry_used{custom}++;
    } else {
      $fry_used{miss}++;
      $missed{$word} = 1;
    }   
  }
  return %fry_used;
}

=head2 _generate_weak_used

Returns a hash of weak words used, with the word as key and the count as value.

=cut

sub _generate_weak_used {
  my $self = shift;
  my %weak_used;
  my %weak_words;
  if ( defined( $self->{_weak_words} ) ){
    %weak_words = %{$self->{_weak_words} };
  } 
  foreach my $word ( $self->words() ){
    $word = lc($word);    
    if ( exists( $weak_words{$word} ) ) {
      $weak_used{$word} += 1;
    }
  }
  return \%weak_used;
}

=head2 grade_level