X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=lib%2FText%2FTradition%2FCollation%2FReading.pm;h=f445420829ba97d6230bc5d2d1f958cc2530ea8a;hb=339786dd0b0c493786af4df1ec3bc7ef2bf497e2;hp=f59841812f172f50b755da7e3d9e310f42c0e241;hpb=eca160573f4f28f53faa7dd97ee5cb6740a4b517;p=scpubgit%2Fstemmatology.git

diff --git a/lib/Text/Tradition/Collation/Reading.pm b/lib/Text/Tradition/Collation/Reading.pm
index f598418..f445420 100644
--- a/lib/Text/Tradition/Collation/Reading.pm
+++ b/lib/Text/Tradition/Collation/Reading.pm
@@ -1,249 +1,433 @@
 package Text::Tradition::Collation::Reading;
 
 use Moose;
-use MooseX::NonMoose;
-use Text::Tradition::Collation::Position;
+use Moose::Util::TypeConstraints;
+use JSON qw/ from_json /;
+use Module::Load;
+use Text::Tradition::Error;
+use XML::Easy::Syntax qw( $xml10_name_rx $xml10_namestartchar_rx );
+use YAML::XS;
+use overload '""' => \&_stringify, 'fallback' => 1;
 
-extends 'Graph::Easy::Node';
+subtype 'ReadingID',
+	as 'Str',
+	where { $_ =~ /\A$xml10_name_rx\z/ },
+	message { 'Reading ID must be a valid XML attribute string' };
+	
+no Moose::Util::TypeConstraints;
 
-has 'position' => (
-    is => 'rw',
-    isa => 'Text::Tradition::Collation::Position',
-    predicate => 'has_position',
+=head1 NAME
+
+Text::Tradition::Collation::Reading - represents a reading (usually a word)
+in a collation.
+
+=head1 DESCRIPTION
+
+Text::Tradition is a library for representation and analysis of collated
+texts, particularly medieval ones.  A 'reading' refers to a unit of text,
+usually a word, that appears in one or more witnesses (manuscripts) of the
+tradition; the text of a given witness is composed of a set of readings in
+a particular sequence
+
+=head1 METHODS
+
+=head2 new
+
+Creates a new reading in the given collation with the given attributes.
+Options include:
+
+=over 4
+
+=item collation - The Text::Tradition::Collation object to which this
+reading belongs.  Required.
+
+=item id - A unique identifier for this reading. Required.
+
+=item text - The word or other text of the reading.
+
+=item is_start - The reading is the starting point for the collation.
+
+=item is_end - The reading is the ending point for the collation.
+
+=item is_lacuna - The 'reading' represents a known gap in the text.
+
+=item is_ph - A temporary placeholder for apparatus parsing purposes.  Do
+not use unless you know what you are doing.
+
+=item rank - The sequence number of the reading. This should probably not
+be set manually.
+
+=back
+
+One of 'text', 'is_start', 'is_end', or 'is_lacuna' is required.
+
+=head2 collation
+
+=head2 id
+
+=head2 text
+
+=head2 is_start
+
+=head2 is_end
+
+=head2 is_lacuna
+
+=head2 rank
+
+Accessor methods for the given attributes.
+
+=cut
+
+has 'collation' => (
+	is => 'ro',
+	isa => 'Text::Tradition::Collation',
+	# required => 1,
+	weak_ref => 1,
+	);
+
+has 'id' => (
+	is => 'ro',
+	isa => 'ReadingID',
+	required => 1,
+	);
+
+has 'text' => (
+	is => 'ro',
+	isa => 'Str',
+	required => 1,
+	writer => 'alter_text',
+	);
+	
+has 'language' => (
+	is => 'ro',
+	isa => 'Str',
+	predicate => 'has_language',
+	);
+	
+has 'is_start' => (
+	is => 'ro',
+	isa => 'Bool',
+	default => undef,
+	);
+
+has 'is_end' => (
+	is => 'ro',
+	isa => 'Bool',
+	default => undef,
+	);
+    
+has 'is_lacuna' => (
+    is => 'ro',
+    isa => 'Bool',
+	default => undef,
     );
     
+has 'is_ph' => (
+	is => 'ro',
+	isa => 'Bool',
+	default => undef,
+	);
+	
+has 'is_common' => (
+	is => 'rw',
+	isa => 'Bool',
+	default => undef,
+	);
+
 has 'rank' => (
     is => 'rw',
     isa => 'Int',
     predicate => 'has_rank',
+    clearer => 'clear_rank',
     );
     
-has 'is_lacuna' => (
-    is => 'rw',
-    isa => 'Bool',
-    );
+## For morphological analysis
+
+has 'grammar_invalid' => (
+	is => 'rw',
+	isa => 'Bool',
+	default => undef,
+	);
+	
+has 'is_nonsense' => (
+	is => 'rw',
+	isa => 'Bool',
+	default => undef,
+	);
+
+has 'normal_form' => (
+	is => 'rw',
+	isa => 'Str',
+	predicate => '_has_normal_form',
+	clearer => '_clear_normal_form',
+	);
+
+# Holds the lexemes for the reading.
+has 'reading_lexemes' => (
+	traits => ['Array'],
+	isa => 'ArrayRef[Text::Tradition::Collation::Reading::Lexeme]',
+	handles => {
+		lexeme => 'get',
+		lexemes => 'elements',
+		has_lexemes => 'count',
+		clear_lexemes => 'clear',
+		add_lexeme => 'push',
+		},
+	default => sub { [] },
+	);
+	
+## For prefix/suffix readings
+
+has 'join_prior' => (
+	is => 'ro',
+	isa => 'Bool',
+	default => undef,
+	writer => '_set_join_prior',
+	);
+	
+has 'join_next' => (
+	is => 'ro',
+	isa => 'Bool',
+	default => undef,
+	writer => '_set_join_next',
+	);
 
-# This contains an array of reading objects; the array is a pool,
-# shared by the reading objects inside the pool.  When a reading is
-# added to the pool, all the same_as attributes should be updated.
-has 'same_as' => (
-    is => 'rw',
-    isa => 'ArrayRef[Text::Tradition::Collation::Reading]',
-    );
 
-# Deal with the non-arg option for Graph::Easy's constructor.
 around BUILDARGS => sub {
-    my $orig = shift;
-    my $class = shift;
-
-    my %args;
-    if( @_ == 1 && ref( $_[0] ) ne 'HASH' ) {
-	return $class->$orig( 'name' => $_[0] );
-    } else {
-	return $class->$orig( @_ );
+	my $orig = shift;
+	my $class = shift;
+	my $args;
+	if( @_ == 1 ) {
+		$args = shift;
+	} else {
+		$args = { @_ };
+	}
+			
+	# If one of our special booleans is set, we change the text and the
+	# ID to match.
+	if( exists $args->{'is_lacuna'} && $args->{'is_lacuna'} && !exists $args->{'text'} ) {
+		$args->{'text'} = '#LACUNA#';
+	} elsif( exists $args->{'is_start'} && $args->{'is_start'} ) {
+		$args->{'id'} = '__START__';  # Change the ID to ensure we have only one
+		$args->{'text'} = '#START#';
+		$args->{'rank'} = 0;
+	} elsif( exists $args->{'is_end'} && $args->{'is_end'} ) {
+		$args->{'id'} = '__END__';	# Change the ID to ensure we have only one
+		$args->{'text'} = '#END#';
+	} elsif( exists $args->{'is_ph'} && $args->{'is_ph'} ) {
+		$args->{'text'} = $args->{'id'};
+	}
+	
+	# Backwards compatibility for non-XMLname IDs
+	my $rid = $args->{'id'};
+	$rid =~ s/\#/__/g;
+	$rid =~ s/[\/,]/./g;
+    if( $rid !~ /^$xml10_namestartchar_rx/ ) {
+    	$rid = 'r'.$rid;
     }
+	$args->{'id'} = $rid;
+	
+	$class->$orig( $args );
 };
 
-# Take constructor args as well as a Position argument.
-around position => sub {
-    my $orig = shift;
-    my $self = shift;
-    return $self->$orig() unless @_;
+# Look for a lexeme-string argument in the build args.
+sub BUILD {
+	my( $self, $args ) = @_;
+	if( exists $args->{'lexemes'} ) {
+		$self->_deserialize_lexemes( $args->{'lexemes'} );
+	}
+}
 
-    my @args = @_;
-    unless( @_ == 1 && ref( $_[0] ) eq 'Text::Tradition::Collation::Position' ) {
-	# We have constructor arguments; pass them to Position.
-	@args = ( Text::Tradition::Collation::Position->new( @_ ) );
-    }
-    $self->$orig( @args );
+# Make normal_form default to text, transparently.
+around 'normal_form' => sub {
+	my $orig = shift;
+	my $self = shift;
+	my( $arg ) = @_;
+	if( $arg && $arg eq $self->text ) {
+		$self->_clear_normal_form;
+		return $arg;
+	} elsif( !$arg && !$self->_has_normal_form ) {
+		return $self->text;
+	} else {
+		$self->$orig( @_ );
+	}
 };
 
-# Initialize the identity pool. 
-sub BUILD {
-    my( $self, $args ) = @_;
-    $self->same_as( [ $self ] );
-}
+=head2 is_meta
 
-sub text {
-    # Wrapper function around 'label' attribute.
-    my $self = shift;
-    if( @_ ) {
-        if( defined $_[0] ) {
-        	$self->set_attribute( 'label', $_[0] );
-        } else {
-            $self->del_attribute( 'label' );
-        }
-    }
-    return $self->label;
+A meta attribute (ha ha), which should be true if any of our 'special'
+booleans are true.  Implies that the reading does not represent a bit 
+of text found in a witness.
+
+=cut
+
+sub is_meta {
+	my $self = shift;
+	return $self->is_start || $self->is_end || $self->is_lacuna || $self->is_ph;	
 }
 
-sub witnessed_by {
-    my( $self, $sigil, $backup ) = @_;
-    my @wits = $self->witnesses;
-    return 1 if grep { $_ eq $sigil } @wits;
-    if( $backup ) {
-        return 1 if grep { $_ eq $backup } @wits;
-    }
-    return 0;
+=head1 Convenience methods
+
+=head2 related_readings
+
+Calls Collation's related_readings with $self as the first argument.
+
+=cut
+
+sub related_readings {
+	my $self = shift;
+	return $self->collation->related_readings( $self, @_ );
 }
-    
+
+=head2 witnesses 
+
+Calls Collation's reading_witnesses with $self as the first argument.
+
+=cut
+
 sub witnesses {
-    my( $self ) = @_;
-    my @paths = grep { $_->get_attribute( 'class' ) eq 'path' } $self->outgoing;
-    push( @paths, grep { $_->get_attribute( 'class' ) eq 'path' } $self->incoming );
-    my %wits;
-    foreach my $p ( @paths ) {
-        if( $p->has_hidden_witnesses ) {
-            foreach ( @{$p->hidden_witnesses} ) {
-                $wits{$_} = 1;
-            }
-        } else {
-            $wits{$p->label} = 1;
-        }
-    }
-    return keys %wits;
+	my $self = shift;
+	return $self->collation->reading_witnesses( $self, @_ );
 }
 
-sub merge_from {
-    my( $self, $merged_node ) = @_;
-    # Adopt the identity pool of the other node.
-    my @now_identical = grep { $_ ne $merged_node } @{$merged_node->same_as};
-    my $new_pool = _merge_array_pool( \@now_identical, $self->same_as )
-	if @now_identical;
+=head2 predecessors
 
-    # TODO Adopt the relationship attributes and segment memberships of the other node.
-}
+Returns a list of Reading objects that immediately precede $self in the collation.
 
-## Dealing with transposed readings.  These methods are only really
-## applicable if we have a linear collation graph.
+=cut
 
-sub set_identical {
-    my( $self, $other_node ) = @_; 
-    my $enlarged_pool = _merge_array_pool( $self->same_as, 
-					   $other_node->same_as );
-
-    # ...and set this node to point to the enlarged pool.
-    $self->same_as( $enlarged_pool );
-}   
-
-sub identical_readings {
-    my $self = shift;
-    my @same = grep { $_ ne $self } @{$self->same_as};
-    return @same;
+sub predecessors {
+	my $self = shift;
+	my @pred = $self->collation->sequence->predecessors( $self->id );
+	return map { $self->collation->reading( $_ ) } @pred;
 }
 
-sub _merge_array_pool {
-    my( $pool, $main_pool ) = @_;
-    my %poolhash;
-    foreach ( @$main_pool ) {
-	# Note which nodes are already in the main pool so that we
-	# don't re-add them.
-	$poolhash{$_->name} = 1;
-    }
+=head2 successors
 
-    foreach( @$pool ) {
-	# Add the remaining nodes to the main pool...
-	push( @$main_pool, $_ ) unless $poolhash{$_->name};
-    }
-    return $main_pool;
-}
+Returns a list of Reading objects that immediately follow $self in the collation.
 
-sub has_primary {
-    my $self = shift;
-    my $pool = $self->same_as;
-    return $pool->[0]->name ne $self->name;
+=cut
+
+sub successors {
+	my $self = shift;
+	my @succ = $self->collation->sequence->successors( $self->id );
+	return map { $self->collation->reading( $_ ) } @succ;
 }
 
-sub primary {
-    my $self = shift;
-    return $self->same_as->[0];
+=head2 set_identical( $other_reading)
+
+Backwards compatibility method, to add a transposition relationship
+between $self and $other_reading.  Don't use this.
+
+=cut
+
+sub set_identical {
+	my( $self, $other ) = @_;
+	return $self->collation->add_relationship( $self, $other, 
+		{ 'type' => 'transposition' } );
 }
 
-sub is_at_position {
-    my $self = shift;
-    return undef unless $self->has_position;
-    return $self->position->is_colocated( @_ );
+sub _stringify {
+	my $self = shift;
+	return $self->id;
 }
 
-# Returns all readings that adjoin this one on any path.
-sub neighbor_readings {
-    my( $self, $direction ) = @_;
-    $direction = 'both' unless $direction;
-    my @paths = grep { $_->isa( 'Text::Tradition::Collation::Path' ) } $self->edges;
-    my %connected;
-    foreach my $p ( @paths ) {
-	if( $p->to eq $self ) {
-	    next if $direction eq 'forward';
-	    $connected{$p->from->name} = $p->from;
-	} else { # $p->from eq $self
-	    next if $direction =~ /^back/;
-	    $connected{$p->to->name} = $p->to;
+=head1 MORPHOLOGY
+
+Methods for the morphological information (if any) attached to readings.
+A reading may be made up of multiple lexemes; the concatenated lexeme
+strings ought to match the reading's normalized form.
+ 
+See L<Text::Tradition::Collation::Reading::Lexeme> for more information
+on Lexeme objects and their attributes.
+
+=head2 has_lexemes
+
+Returns a true value if the reading has any attached lexemes.
+
+=head2 lexemes
+
+Returns the Lexeme objects (if any) attached to the reading.
+
+=head2 clear_lexemes
+
+Wipes any associated Lexeme objects out of the reading.
+
+=head2 add_lexeme( $lexobj )
+
+Adds the Lexeme in $lexobj to the list of lexemes.
+
+=head2 lemmatize
+
+If the language of the reading is set, this method will use the appropriate
+Language model to determine the lexemes that belong to this reading.  See
+L<Text::Tradition::lemmatize> if you wish to lemmatize an entire tradition.
+
+=cut
+
+sub lemmatize {
+	my $self = shift;
+	unless( $self->has_language ) {
+		warn "Please set a language to lemmatize a tradition";
+		return;
 	}
-    }
-    return values( %connected );
-}
+	my $mod = "Text::Tradition::Language::" . $self->language;
+	load( $mod );
+	$mod->can( 'reading_lookup' )->( $self );
 
-# Returns all readings related to the one we've got.
-sub related_readings {
-    my( $self, $colocated ) = @_;
-    my @related;
-    foreach my $e ( $self->edges ) {
-        next unless $e->isa( 'Text::Tradition::Collation::Relationship' );
-        next if $colocated && $e->type eq 'repetition';
-        push( @related, $e->from eq $self ? $e->to : $e->from );
-    }
-    return @related;
 }
 
-## Keep track of which readings are unchanged across witnesses.
-sub is_common {
-    my( $self ) = shift;
-    return $self->get_attribute( 'class' ) eq 'common';
+# For graph serialization. Return a JSON representation of the associated
+# reading lexemes.
+sub _serialize_lexemes {
+	my $self = shift;
+	my $json = JSON->new->allow_blessed(1)->convert_blessed(1);
+	return $json->encode( [ $self->lexemes ] );
 }
 
-sub make_common {
-    my( $self ) = shift;
-    $self->set_attribute( 'class', 'common' );
+# Given a JSON representation of the lexemes, instantiate them and add
+# them to the reading.
+sub _deserialize_lexemes {
+	my( $self, $json ) = @_;
+	my $data = from_json( $json );
+	return unless @$data;
+	
+	# Need to have the lexeme module in order to have lexemes.
+	eval { use Text::Tradition::Collation::Reading::Lexeme; };
+	throw( $@ ) if $@;
+	
+	# Good to go - add the lexemes.
+	my @lexemes;
+	foreach my $lexhash ( @$data ) {
+		push( @lexemes, Text::Tradition::Collation::Reading::Lexeme->new(
+			'JSON' => $lexhash ) );
+	}
+	$self->clear_lexemes;
+	$self->add_lexeme( @lexemes );
 }
 
-sub make_variant {
-    my( $self ) = shift;
-    $self->set_attribute( 'class', 'variant' );
+sub disambiguated {
+	my $self = shift;
+	return 0 unless $self->has_lexemes;
+	return !grep { !$_->is_disambiguated } $self->lexemes;
 }
 
-no Moose;
-__PACKAGE__->meta->make_immutable;
-
-1;
-
-######################################################
-## copied from Graph::Easy::Parser docs
-######################################################
-# when overriding nodes, we also need ::Anon
+## Utility methods
 
-package Text::Tradition::Collation::Reading::Anon;
-use Moose;
-use MooseX::NonMoose;
-extends 'Text::Tradition::Collation::Reading';
-extends 'Graph::Easy::Node::Anon';
-no Moose;
-__PACKAGE__->meta->make_immutable;
-
-1;
-# use base qw/Text::Tradition::Collation::Reading/;
-# use base qw/Graph::Easy::Node::Anon/;
+sub TO_JSON {
+	my $self = shift;
+	return $self->text;
+}
 
-######################################################
-# and :::Empty
+sub throw {
+	Text::Tradition::Error->throw( 
+		'ident' => 'Reading error',
+		'message' => $_[0],
+		);
+}
 
-package Text::Tradition::Collation::Reading::Empty;
-use Moose;
-use MooseX::NonMoose;
-extends 'Graph::Easy::Node::Empty';
 no Moose;
 __PACKAGE__->meta->make_immutable;
 
 1;
-# use base qw/Text::Tradition::Collation::Reading/;
-
-######################################################