X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=lib%2FText%2FTradition%2FAnalysis.pm;h=90a82d86d163086354b5b1e74557dd326b898439;hb=08e0fb850e4e867ac0be6711ec77b69718cc3958;hp=3b8888b67feddcc59772141a7c644f60c8f46661;hpb=732152b1ae3962d3d27d501e8a687a32fb80e04f;p=scpubgit%2Fstemmatology.git diff --git a/lib/Text/Tradition/Analysis.pm b/lib/Text/Tradition/Analysis.pm index 3b8888b..90a82d8 100644 --- a/lib/Text/Tradition/Analysis.pm +++ b/lib/Text/Tradition/Analysis.pm @@ -9,7 +9,10 @@ sub new { my( $class, $args ) = @_; my $self = {}; bless( $self, $class ); - $self->run_analysis( $args->{'file'}, $args->{'stemmadot'} ); + $self->{'data'} = []; + foreach my $t ( @{$args->{'traditions'}} ) { + $self->run_analysis( $t->{'file'}, $t->{'stemmadot'} ); + } return $self; } @@ -18,6 +21,7 @@ sub run_analysis { # What we will return my $svg; my $variants = []; + my $data = {}; # Read in the file and stemma my $tradition = Text::Tradition->new( @@ -25,22 +29,21 @@ sub run_analysis { 'file' => $file, 'linear' => 1, ); - $self->{'title'} = $tradition->name; + $data->{'title'} = $tradition->name; my $stemma = Text::Tradition::Stemma->new( 'collation' => $tradition->collation, 'dot' => $stemmadot, ); # We will return the stemma picture - $svg = $stemma->as_svg; - ### DIRTY HACK - $svg =~ s/transform=\"scale\(1 1\)/transform=\"scale\(0.7 0.7\)/; - $self->{'svg'} = $svg; + $svg = $stemma->as_svg( { size => "8,7.5" } );; + $data->{'svg'} = $svg; # 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. - - my $all_wits_table = $tradition->collation->make_alignment_table( 'refs' ); + my $wits = {}; + map { $wits->{$_} = 1 } $stemma->witnesses; + my $all_wits_table = $tradition->collation->make_alignment_table( 'refs', $wits ); # For each column in the alignment table, we want to see if the existing # groupings of witnesses match our stemma hypothesis. We also want, at the @@ -100,8 +103,9 @@ sub run_analysis { # For all the groups with more than one member, collect the list of all # contiguous vertices needed to connect them. # TODO: deal with a.c. reading logic + $DB::single = 1 if $rank == 25; my $variant_row = analyze_variant_location( $group_readings, $groups, - $stemma->apsp, $lacunose ); + $stemma->graph, $lacunose ); $variant_row->{'id'} = $rank; $genealogical++ if $variant_row->{'genealogical'}; $conflicts += grep { $_->{'conflict'} } @{$variant_row->{'readings'}}; @@ -110,8 +114,7 @@ sub run_analysis { # my @trees = @{$stemma->distance_trees}; # if( @trees ) { # foreach my $tree ( 0 .. $#trees ) { -# my $dc = analyze_variant_location( $group_readings, $groups, -# $stemma->distance_apsps->[$tree] ); +# my $dc = analyze_variant_location( $group_readings, $groups, $tree ); # foreach my $rdg ( keys %$dc ) { # my $var = $dc->{$rdg}; # # TODO Do something with this @@ -131,38 +134,55 @@ sub run_analysis { } # Populate self with our analysis data. - $self->{'variants'} = $variants; - $self->{'variant_count'} = $total; - $self->{'conflict_count'} = $conflicts; - $self->{'genealogical_count'} = $genealogical; + $data->{'variants'} = $variants; + $data->{'variant_count'} = $total; + $data->{'conflict_count'} = $conflicts; + $data->{'genealogical_count'} = $genealogical; + push( @{$self->{'data'}}, $data ); } # variant_row -> genealogical # -> readings [ { text, group, conflict, missing } ] sub analyze_variant_location { - my( $group_readings, $groups, $apsp, $lacunose ) = @_; + my( $group_readings, $groups, $graph, $lacunose ) = @_; my %contig; my $conflict = {}; my %missing; map { $missing{$_} = 1 } @$lacunose; my $variant_row = { 'readings' => [] }; + # Mark each ms as in its own group, first. + foreach my $g ( @$groups ) { + my $gst = wit_stringify( $g ); + map { $contig{$_} = $gst } @$g; + } foreach my $g ( sort { scalar @$b <=> scalar @$a } @$groups ) { - my @members = @$g; - my $gst = wit_stringify( $g ); # $gst is now the name of this group. - map { $contig{$_} = $gst } @members; # All members are in this group. - while( @members ) { - # Gather the list of vertices that are needed to join all members. - my $curr = pop @members; - foreach my $m ( @members ) { - foreach my $v ( $apsp->path_vertices( $curr, $m ) ) { - $contig{$v} = $gst unless exists $contig{$v}; - next if $contig{$v} eq $gst; - # Record what is conflicting. TODO do we use this? - $conflict->{$group_readings->{$gst}} = $group_readings->{$contig{$v}}; - } + my $gst = wit_stringify( $g ); + # Copy the graph, and delete all non-members from the new graph. + my $part = $graph->undirected_copy; + map { $part->delete_vertex( $_ ) + if $contig{$_} && $contig{$_} ne $gst } $graph->vertices; + # Now all the members of the group should still be reachable + # from the first member. + my %reachable = ( $g->[0] => 1 ); + map { $reachable{$_} = 1 } $part->all_reachable( $g->[0] ); + + # ...and none of these nodes should be marked as being in another + # group. + foreach ( keys %reachable ) { + if( $contig{$_} && $contig{$_} ne $gst ) { + $conflict->{$group_readings->{$gst}} = $group_readings->{$contig{$_}}; + } else { + $contig{$_} = $gst; } } + # None of the unreachable nodes should be in our group either. + foreach ( $part->vertices ) { + next if $reachable{$_}; + $conflict->{$group_readings->{$gst}} = $group_readings->{$gst} + if $contig{$_} && $contig{$_} eq $gst; + } + # Write the reading. my $reading = { 'text' => $group_readings->{$gst}, 'missing' => wit_stringify( $lacunose ), @@ -170,12 +190,12 @@ sub analyze_variant_location { if( $reading->{'conflict'} ) { $reading->{'group'} = $gst; } else { - my @all_vertices = grep { $contig{$_} eq $gst && !$missing{$_} } keys %contig; + my @all_vertices = grep { !$missing{$_} } keys %reachable; $reading->{'group'} = wit_stringify( \@all_vertices ); } push( @{$variant_row->{'readings'}}, $reading ); } - $variant_row->{'genealogical'} = keys %$conflict ? undef : 1; + $variant_row->{'genealogical'} = !( keys %$conflict ); return $variant_row; }