Merge branch 'master' of github.com:tla/stemmatology

[scpubgit/stemmatology.git] / lib / Text / Tradition / Analysis.pm

diff --git a/lib/Text/Tradition/Analysis.pm b/lib/Text/Tradition/Analysis.pm
index 56b943c..910f994 100644 (file)
--- a/lib/Text/Tradition/Analysis.pm
+++ b/lib/Text/Tradition/Analysis.pm
@@ -8,7 +8,7 @@ use Text::Tradition;
 use Text::Tradition::Stemma;
 
 use vars qw/ @EXPORT_OK /;
-@EXPORT_OK = qw/ run_analysis group_variants wit_stringify /;
+@EXPORT_OK = qw/ run_analysis group_variants analyze_variant_location wit_stringify /;
 
 sub run_analysis {
        my( $tradition ) = @_;
@@ -17,11 +17,11 @@ sub run_analysis {
        my $data = {};
        
        # We need a stemma in order to run this...
-       unless( $tradition->has_stemma ) {
+       unless( $tradition->stemma_count ) {
                warn "Tradition '" . $tradition->name . "' has no stemma to analyze";
                return undef;
        }
-       my $stemma = $tradition->stemma;
+       my $stemma = $tradition->stemma(0); # TODO allow multiple
                
        # We have the collation, so get the alignment table with witnesses in rows.
        # Also return the reading objects in the table, rather than just the words.
@@ -31,9 +31,8 @@ sub run_analysis {
        # groupings of witnesses match our stemma hypothesis. We also need to keep
        # track of the maximum number of variants at any one location.
        my $max_variants = 0;
-       my ( $total, $genealogical, $conflicts ) = ( 0, 0, 0 );
+       my ( $genealogical, $conflicts ) = ( 0, 0, 0 );
        
-    my $t0 = Benchmark->new();
     my $variant_groups = group_variants( $tradition->collation, $wits );
     foreach my $rank ( 0 .. $#{$variant_groups} ) {
         my $groups = $variant_groups->[$rank]->{'groups'};
@@ -72,10 +71,7 @@ sub run_analysis {
                # Record that we used this variant in an analysis
                push( @$variants, $variant_loc );
        }
-    my $t1 = Benchmark->new();
-    print STDERR "Analysis of graph for " . $tradition->name . " took " 
-        . timestr( timediff( $t1, $t0 ) ) . "seconds\n";
-       
+
        # Go through our variant locations, after we have seen all of them once,
        # and add the number of empty columns needed by each.
        foreach my $row ( @$variants ) {
@@ -84,7 +80,7 @@ sub run_analysis {
        }
        
        $data->{'variants'} = $variants;
-       $data->{'variant_count'} = $total;
+       $data->{'variant_count'} = $tradition->collation->end->rank - 1;
        $data->{'conflict_count'} = $conflicts;
        $data->{'genealogical_count'} = $genealogical;
        return $data;
@@ -98,33 +94,33 @@ sub group_variants {
        # Also return the reading objects in the table, rather than just the words.
        my $all_wits_table = $c->make_alignment_table( 'refs', $wits );
        # Strip the list of sigla and save it for correlation to the readings.
-       my $col_wits = shift @$all_wits_table;
+       my @table_wits = map { $_->{'witness'} } @{$all_wits_table->{'alignment'}};
        # Any witness in the stemma that has no row should be noted.
-    foreach ( @$col_wits ) {
+    foreach ( @table_wits ) {
         $wits->{$_}++; # Witnesses present in table and stemma now have value 2.
     }
     my @not_collated = grep { $wits->{$_} == 1 } keys %$wits;
-       foreach my $i ( 0 .. $#$all_wits_table ) {
+       foreach my $i ( 0 .. $all_wits_table->{'length'} - 1 ) {
                # For each column in the table, group the readings by witness.
                my $rdg_wits = {};
-               my $col_rdgs = shift @$all_wits_table;
+               my @col_rdgs = map { $_->{tokens}->[$i] } @{$all_wits_table->{'alignment'}};
                my $lacunose = [ @not_collated ];
-               foreach my $j ( 0 .. $#{$col_rdgs} ) {
-                       my $rdg = $col_rdgs->[$j];
+               foreach my $j ( 0 .. $#col_rdgs ) {
+                       my $rdg = $col_rdgs[$j];
                        my $rdg_text = '(omitted)';  # Initialize in case of empty reading
                        if( $rdg ) {
-                           if( $rdg->is_lacuna ) {
+                           if( $rdg->{'t'}->is_lacuna ) {
                                $rdg_text = undef;   # Don't count lacunae
-                               push( @$lacunose, $col_wits->[$j] );
+                               push( @$lacunose, $table_wits[$j] );
                            } else {
-                               $rdg_text = $rdg->text; 
+                               $rdg_text = $rdg->{'t'}->text; 
                                }
                        }
                        if( defined $rdg_text ) {
                                # Initialize the witness array if we haven't got one yet
                                $rdg_wits->{$rdg_text} = [] unless $rdg_wits->{$rdg_text};
                                # Add the relevant witness, subject to a.c. logic
-                               add_variant_wit( $rdg_wits->{$rdg_text}, $col_wits->[$j],
+                               add_variant_wit( $rdg_wits->{$rdg_text}, $table_wits[$j],
                                        $c->ac_label );
                        }
                }