view release on metacpan or  search on metacpan

CAB/Chain.pm  view on Meta::CPAN

## -*- Mode: CPerl -*-
##
## File: DTA::CAB::Chain.pm
## Author: Bryan Jurish <moocow@cpan.org>
## Description: generic analyzer API: analyzer "chains" / "cascades" / "pipelines" / ...

package DTA::CAB::Chain;
use DTA::CAB::Analyzer;
use DTA::CAB::Datum ':all';
use Carp;
use strict;

##==============================================================================
## Globals
##==============================================================================

our @ISA = qw(DTA::CAB::Analyzer);

BEGIN {
  *isa = \&UNIVERSAL::isa;
  *can = \&UNIVERSAL::can;
}

##==============================================================================
## Constructors etc.
##==============================================================================

## $obj = CLASS_OR_OBJ->new(%args)
##  + object structure:
##    (
##     ##-- Analyzers
##     chain => [ $a1, $a2, ..., $aN ],        ##-- default analysis chain; see also chain() method (default: empty)
##
##     ##-- verbose trace
##     logTrace => $level,                     ##-- trace sub-analyzer execution (default: 'none')
##    )
sub new {
  my $that = shift;
  my $ach = bless({
		   ##-- user args
		   chain => [],
		   logTrace => 'none',
		   @_
		  }, ref($that)||$that);
  $ach->initialize();
  $ach->{label} = $ach->defaultLabel if (!defined($ach->{label}));
  return $ach;
}

## undef = $ach->initialize();
##  + default implementation does nothing
##  + INHERITED from DTA::CAB::Analyzer

## undef = $ach->dropClosures();
##  + drops '_analyze*' closures
##  + INHERITED from DTA::CAB::Analyzer

## @keys = $anl->typeKeys(\%opts)
##  + returns list of type-wise keys to be expanded for this analyzer by expandTypes()
##  + default just concatenates keys for sub-analyzers
sub typeKeys {
  my $ach = shift;
  return map {ref($_) ? $_->typeKeys(@_) : qw()} @{$ach->chain(@_)}
}

CAB/Chain.pm  view on Meta::CPAN

=cut

##========================================================================
## NAME
=pod

=head1 NAME

DTA::CAB::Chain - serial multi-analyzer pipeline

=cut

##========================================================================
## SYNOPSIS
=pod

=head1 SYNOPSIS

 use DTA::CAB::Chain;
 
 ##========================================================================
 ## Constructors etc.
 
 $obj = CLASS_OR_OBJ->new(%args);
 @keys = $anl->typeKeys(\%opts);
 
 ##========================================================================
 ## Methods: Chain selection
 
 \@analyzers = $ach->chain();
 \@analyzers = $ach->subAnalyzers();
 
 ##========================================================================
 ## Methods: I/O
 
 $bool = $ach->ensureLoaded();
 
 ##========================================================================
 ## Methods: Analysis
 
 $bool = $ach->canAnalyze();
 $bool = $anl->enabled(\%opts);
 undef = $anl->initInfo();
 
 $doc = $ach->analyzeTypes($doc,$types,\%opts);
 $doc = $ach->analyzeSentences($doc,\%opts);
 $doc = $ach->analyzeLocal($doc,\%opts);
 $doc = $ach->analyzeClean($doc,\%opts);
 

=cut

##========================================================================
## DESCRIPTION
=pod

=head1 DESCRIPTION

DTA::CAB::Chain
is an abstract L<DTA::CAB::Analyzer|DTA::CAB::Analyzer> subclass
for implementing serial document processing "pipelines" or "cascades"
in terms of a flat list of L<DTA::CAB::Analyzer|DTA::CAB::Analyzer> objects.

=cut

##----------------------------------------------------------------
## DESCRIPTION: DTA::CAB::Chain: Constructors etc.
=pod

=head2 Constructors etc.

=over 4

=item new

 $obj = CLASS_OR_OBJ->new(%args);

%$obj, %args:

 chain => [ $a1, $a2, ..., $aN ],        ##-- default analysis chain; see also chain() method (default: empty)

=item typeKeys

 @keys = $anl->typeKeys(\%opts);

Returns list of type-wise keys to be expanded for this analyzer by expandTypes()
Default implementation just concatenates typeKeys() for sub-analyzers.

=back

=cut

##----------------------------------------------------------------
## DESCRIPTION: DTA::CAB::Chain: Methods: Chain selection
=pod

=head2 Methods: Chain selection

=over 4

=item chain

 \@analyzers = $ach->chain();
 \@analyzers = $ach->chain(\%opts)

Get selected analyzer chain.
Default method returns all globally enabled analyzers in $anl-E<gt>{chain}.

=item subAnalyzers

 \@analyzers = $ach->subAnalyzers();
 \@analyzers = $ach->subAnalyzers(\%opts)

Returns a list of all sub-analyzers.
Override just calls chain().

=back

=cut

##----------------------------------------------------------------