leave meta readings out of equiv graph; test for warnings on relationshipstore.t

[scpubgit/stemmatology.git] / lib / Text / Tradition / Collation / Reading.pm

diff --git a/lib/Text/Tradition/Collation/Reading.pm b/lib/Text/Tradition/Collation/Reading.pm
index f090b34..108548b 100644 (file)
--- a/lib/Text/Tradition/Collation/Reading.pm
+++ b/lib/Text/Tradition/Collation/Reading.pm
@@ -1,11 +1,21 @@
 package Text::Tradition::Collation::Reading;
 
 use Moose;
+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;
 
+subtype 'ReadingID',
+       as 'Str',
+       where { $_ =~ /\A$xml10_name_rx\z/ },
+       message { 'Reading ID must be a valid XML attribute string' };
+       
+no Moose::Util::TypeConstraints;
+
 =head1 NAME
 
 Text::Tradition::Collation::Reading - represents a reading (usually a word)
@@ -78,7 +88,7 @@ has 'collation' => (
 
 has 'id' => (
        is => 'ro',
-       isa => 'Str',
+       isa => 'ReadingID',
        required => 1,
        );
 
@@ -134,6 +144,18 @@ has 'rank' => (
     
 ## 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',
@@ -145,6 +167,7 @@ has 'reading_lexemes' => (
        traits => ['Array'],
        isa => 'ArrayRef[Text::Tradition::Collation::Reading::Lexeme]',
        handles => {
+               lexeme => 'get',
                lexemes => 'elements',
                has_lexemes => 'count',
                clear_lexemes => 'clear',
@@ -183,16 +206,25 @@ around BUILDARGS => sub {
        if( exists $args->{'is_lacuna'} && !exists $args->{'text'} ) {
                $args->{'text'} = '#LACUNA#';
        } elsif( exists $args->{'is_start'} ) {
-               $args->{'id'} = '#START#';  # Change the ID to ensure we have only one
+               $args->{'id'} = '__START__';  # Change the ID to ensure we have only one
                $args->{'text'} = '#START#';
                $args->{'rank'} = 0;
        } elsif( exists $args->{'is_end'} ) {
-               $args->{'id'} = '#END#';        # Change the ID to ensure we have only one
+               $args->{'id'} = '__END__';      # Change the ID to ensure we have only one
                $args->{'text'} = '#END#';
        } elsif( exists $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 );
 };
 
@@ -217,6 +249,17 @@ sub is_meta {
        return $self->is_start || $self->is_end || $self->is_lacuna || $self->is_ph;    
 }
 
+=head2 is_nonrel
+
+Similar to is_meta, but returns false for the start and end readings.
+
+=cut
+
+sub is_nonrel {
+       my $self = shift;
+       return $self->is_lacuna || $self->is_ph;
+}
+
 =head1 Convenience methods
 
 =head2 related_readings
@@ -328,55 +371,30 @@ sub lemmatize {
 
 }
 
-# For graph serialization. Return a string representation of the associated
+# For graph serialization. Return a JSON representation of the associated
 # reading lexemes.
-# TODO Push this in to the Lexeme package.
 sub _serialize_lexemes {
        my $self = shift;
-       my @lexstrs;
-       foreach my $l ( $self->lexemes ) {
-               my @mf;
-               foreach my $wf ( $l->matching_forms ) {
-                       push( @mf, $wf->to_string );
-               }
-               my $form = $l->form ? $l->form->to_string : '';
-               push( @lexstrs, join( '|L|', $l->language, $l->string, $form, 
-                       join( '|M|', @mf ) ) );
-       }
-       return join( '|R|', @lexstrs );
+       my $json = JSON->new->allow_blessed(1)->convert_blessed(1);
+       return $json->encode( [ $self->lexemes ] );
 }
 
+# Given a JSON representation of the lexemes, instantiate them and add
+# them to the reading.
 sub _deserialize_lexemes {
-       my( $self, $data ) = @_;
-       return unless $data;
+       my( $self, $json ) = @_;
+       my $data = from_json( $json );
+       return unless @$data;
        
-       # Need to have the lexeme modules in order to have lexemes.
-       eval { 
-               use Text::Tradition::Collation::Reading::Lexeme; 
-               use Text::Tradition::Collation::Reading::WordForm;
-       };
+       # 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 $lexdata ( split( /\|R\|/, $data ) ) {
-               my( $lang, $lstring, $form, $allforms ) = split( /\|L\|/, $lexdata );
-               my @wfdata;
-               push( @wfdata, $form ) if $form;
-               push( @wfdata, split( /\|M\|/, $allforms ) );
-               my @wforms;
-               foreach my $wd ( @wfdata ) {
-                       my $wf = Text::Tradition::Collation::Reading::WordForm->new(
-                               'serial' => $wd );
-                       push( @wforms, $wf );
-               }
-               my %largs = ( 'language' => $lang, 'string' => $lstring );
-               if( $form ) {
-                       $largs{'form'} = shift @wforms;
-                       $largs{'is_disambiguated'} = 1;
-               }
-               $largs{'wordform_matchlist'} = \@wforms;
-               push( @lexemes, Text::Tradition::Collation::Reading::Lexeme->new( %largs ) );
+       foreach my $lexhash ( @$data ) {
+               push( @lexemes, Text::Tradition::Collation::Reading::Lexeme->new(
+                       'JSON' => $lexhash ) );
        }
        $self->clear_lexemes;
        $self->add_lexeme( @lexemes );