binmode STDOUT, ":utf8";
eval { no warnings; binmode $DB::OUT, ":utf8"; };
-my $informat = 'TEI';
-my $inbase;
-my $linear = 1;
-
# Parse the tradition data
+my $informat = 'Self';
-my $input = $ARGV[0];
-my @lines;
-open( INFILE, "$input" ) or die "Could not read $input";
-binmode INFILE, ':utf8';
-@lines = <INFILE>;
-close INFILE;
-$input = join( '', @lines );
-
-my %args = ( $informat => $input,
- 'linear' => $linear );
-$args{'base'} = $inbase if $inbase;
- my $tradition = Text::Tradition->new( %args );
+my %args = ( 'input' => $informat,
+ 'file' => $ARGV[0] );
+my $tradition = Text::Tradition->new( %args );
# Parse the stemma hypothesis
my $stemma = Text::Tradition::Stemma->new(
);
# 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( 1 );
+my $all_wits_table = $tradition->collation->make_alignment_table( 'refs' );
# For each column in the alignment table, we want to see if the existing
-# groupings of witnesses match our stemma hypothesis. First let's just go
-# through the groupings.
+# groupings of witnesses match our stemma hypothesis. We also want, at the
+# end, to produce an HTML table with all the variants.
+my $html_columns = 0;
+my $html_data = [];
+my $total = 0; # Keep track of the total number of variant locations
# Strip the list of sigla and save it for correlation to the readings.
my $col_wits = shift @$all_wits_table;
-# For each column in the table, group the readings by witness.
-
-my $useful_vars = 0;
foreach my $i ( 0 .. $#$all_wits_table ) {
+ # For each column in the table, group the readings by witness.
my $rdg_wits = {};
my $col_rdgs = shift @$all_wits_table;
+ my $rank;
foreach my $j ( 0 .. $#{$col_rdgs} ) {
my $rdg = $col_rdgs->[$j];
- $rdg = '' unless $rdg; # We care about empty readings
- $rdg = undef if $rdg eq '#LACUNA#'; # ... unless they're lacunas
+ $rank = $rdg->rank if $rdg; # Save the rank for later display
+ my $rdg_text = '(omitted)'; # Initialize in case of empty reading
if( $rdg ) {
- $rdg_wits->{$rdg} = [] unless $rdg_wits->{$rdg};
- add_variant_wit( $rdg_wits->{$rdg}, $col_wits->[$j] );
+ $rdg_text = $rdg->is_lacuna ? undef : $rdg->text; # Don't count lacunae
+ }
+ 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] );
}
}
+ # See if this column has any potentially genealogical variants.
+ # If not, skip to the next.
+ $total++ unless scalar keys %$rdg_wits == 1;
my( $groups, $readings ) = useful_variant( $rdg_wits );
- next unless $groups && $readings;
+ next unless $groups && $readings;
+ $html_columns = scalar @$groups if scalar @$groups > $html_columns;
+
+ # We can already look up witnesses for a reading; we also want to look
+ # up readings for a given witness.
+ my $group_readings = {};
+ foreach my $x ( 0 .. $#$groups ) {
+ $group_readings->{wit_stringify( $groups->[$x] )} = $readings->[$x];
+ }
- # For all the groups with more than one member, make a group that contains
- # all contiguous vertices to connect them.
- # TODO Need to do pairwise comparison of groups - a variant location can
- # have both coincidental and genealogical variants!
+ # 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
+ my $sc = analyze_variant_location( $group_readings, $groups, $stemma->apsp );
+ print wit_stringify( $groups ) . ' - ' . join( " / ", @$readings ) . "\n";
+ foreach my $rdg ( keys %$sc ) {
+ my $var = $sc->{$rdg};
+ print "\tReadings '$rdg' and '$var' are not genealogical\n";
+ }
+
+ # Now run the same analysis given the calculated distance tree(s).
+ foreach my $tree ( 0 .. $#{$stemma->distance_trees} ) {
+ my $dc = analyze_variant_location( $group_readings, $groups,
+ $stemma->distance_apsps->[$tree] );
+ foreach my $rdg ( keys %$dc ) {
+ my $var = $dc->{$rdg};
+ print "\tReadings '$rdg' and '$var' disregarded by parsimony on tree $tree\n";
+ }
+ }
+
+ # Record that we used this variant in an analysis
+ push( @$html_data, [ $rank, $readings, $sc ] );
+}
+
+# Save the stemma picture
+open( STEMMA, ">stemma_graph.svg" ) or die "Could not open stemma graph to write";
+binmode STEMMA, ":utf8";
+print STEMMA $stemma->as_svg;
+close STEMMA;
+
+printf( "Ran analysis on %d / %d variant locations\n", scalar @$html_data, $total );
+
+sub analyze_variant_location {
+ my( $group_readings, $groups, $apsp ) = @_;
my %contig;
- my $conflict;
- foreach my $g ( @$groups ) {
- my @members = split( /,/, $g );
+ my $conflict = {};
+ foreach my $g ( sort { scalar @$b <=> scalar @$a } @$groups ) {
+ my @members = @$g;
+ my $gst = wit_stringify( $g );
+ map { $contig{$_} = $gst } @members; # The witnesses need themselves to be
+ # in their collection.
next unless @members > 1;
- map { $contig{$_} = $g } @members;
my $curr = pop @members;
foreach my $m ( @members ) {
- foreach my $v ( $stemma->apsp->path_vertices( $curr, $m ) ) {
- $contig{$v} = $g unless exists $contig{$v};
- next if $contig{$v} eq $g;
- # print STDERR "Conflict at $v between group $g and group "
+ foreach my $v ( $apsp->path_vertices( $curr, $m ) ) {
+ $contig{$v} = $gst unless exists $contig{$v};
+ next if $contig{$v} eq $gst;
+ # print STDERR "Conflict at $v between group $gst and group "
# . $contig{$v} . "\n";
- $conflict = 1;
+ # Record what is conflicting.
+ $conflict->{$group_readings->{$gst}} = $group_readings->{$contig{$v}};
}
}
}
- print join( " / ", @$groups ) . ' - ' . join( " / ", @$readings ) . ' - ';
- print $conflict ? "coincidental" : "genealogical";
- print "\n";
- $useful_vars++;
-
+ return $conflict;
}
-print "Found $useful_vars useful variants\n";
# Add the variant, subject to a.c. representation logic.
# This assumes that we will see the 'main' version before the a.c. version.
return( undef, undef ) if $total <= 1;
my( $groups, $text );
foreach my $var ( keys %$readings ) {
- push( @$groups, join( ',', @{$readings->{$var}} ) );
+ push( @$groups, $readings->{$var} );
push( @$text, $var );
}
return( $groups, $text );
}
+
+# Take an array of witness groupings and produce a string like
+# A,B / C,D,E / F
+
+sub wit_stringify {
+ my $groups = shift;
+ my @gst;
+ # If we were passed an array of witnesses instead of an array of
+ # groupings, then "group" the witnesses first.
+ unless( ref( $groups->[0] ) ) {
+ my $mkgrp = [ $groups ];
+ $groups = $mkgrp;
+ }
+ foreach my $g ( @$groups ) {
+ push( @gst, join( ',', @$g ) );
+ }
+ return join( ' / ', @gst );
+}
+
\ No newline at end of file
projects / scpubgit/stemmatology.git / blobdiff