X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=lib%2FText%2FTradition%2FCollation.pm;h=8344288f7118986a3f286bf671eef291b46e3804;hb=fae07016dc0e300ca65f3ca03d408c47d80c9217;hp=8f3d6d2cb972d6671831e61f6e07a96f761ddc08;hpb=e247aad18368a4927b54e96a99aec4a442a5196a;p=scpubgit%2Fstemmatology.git diff --git a/lib/Text/Tradition/Collation.pm b/lib/Text/Tradition/Collation.pm index 8f3d6d2..8344288 100644 --- a/lib/Text/Tradition/Collation.pm +++ b/lib/Text/Tradition/Collation.pm @@ -2,6 +2,7 @@ package Text::Tradition::Collation; use Encode qw( decode_utf8 ); use File::Temp; +use File::Which; use Graph; use IPC::Run qw( run binary ); use Text::CSV_XS; @@ -27,6 +28,8 @@ has 'relations' => ( handles => { relationships => 'relationships', related_readings => 'related_readings', + get_relationship => 'get_relationship', + del_relationship => 'del_relationship', }, writer => '_set_relations', ); @@ -74,6 +77,12 @@ has 'ac_label' => ( default => ' (a.c.)', ); +has 'wordsep' => ( + is => 'rw', + isa => 'Str', + default => ' ', + ); + has 'start' => ( is => 'ro', isa => 'Text::Tradition::Collation::Reading', @@ -94,6 +103,19 @@ has 'cached_svg' => ( predicate => 'has_cached_svg', clearer => 'wipe_svg', ); + +has 'cached_table' => ( + is => 'rw', + isa => 'HashRef', + predicate => 'has_cached_table', + clearer => 'wipe_table', + ); + +has '_graphcalc_done' => ( + is => 'rw', + isa => 'Bool', + default => undef, + ); =head1 NAME @@ -147,6 +169,9 @@ representing another layer of path for the given witness - that is, when a text has more than one possible reading due to scribal corrections or the like. Defaults to ' (a.c.)'. +=item * wordsep - The string used to separate words in the original text. +Defaults to ' '. + =back =head1 ACCESSORS @@ -161,6 +186,8 @@ the like. Defaults to ' (a.c.)'. =head2 ac_label +=head2 wordsep + Simple accessors for collation attributes. =head2 start @@ -189,10 +216,14 @@ See L for the available arguments. Removes the given reading from the collation, implicitly removing its paths and relationships. -=head2 merge_readings( $main, $second ) +=head2 merge_readings( $main, $second, $concatenate, $with_str ) -Merges the $second reading into the $main one. -The arguments may be either readings or reading IDs. +Merges the $second reading into the $main one. If $concatenate is true, then +the merged node will carry the text of both readings, concatenated with either +$with_str (if specified) or a sensible default (the empty string if the +appropriate 'join_*' flag is set on either reading, or else $self->wordsep.) + +The first two arguments may be either readings or reading IDs. =head2 has_reading( $id ) @@ -257,6 +288,7 @@ sub add_reading { if( $self->reading( $reading->id ) ) { throw( "Collation already has a reading with id " . $reading->id ); } + $self->_graphcalc_done(0); $self->_add_reading( $reading->id => $reading ); # Once the reading has been added, put it in both graphs. $self->sequence->add_vertex( $reading->id ); @@ -273,6 +305,8 @@ around del_reading => sub { $arg = $arg->id; } # Remove the reading from the graphs. + $self->_graphcalc_done(0); + $self->_clear_cache; # Explicitly clear caches to GC the reading $self->sequence->delete_vertex( $arg ); $self->relations->delete_reading( $arg ); @@ -280,14 +314,58 @@ around del_reading => sub { $self->$orig( $arg ); }; -# merge_readings( $main, $to_be_deleted ); +=begin testing + +use Text::Tradition; + +my $cxfile = 't/data/Collatex-16.xml'; +my $t = Text::Tradition->new( + 'name' => 'inline', + 'input' => 'CollateX', + 'file' => $cxfile, + ); +my $c = $t->collation; + +my $rno = scalar $c->readings; +# Split n21 for testing purposes +my $new_r = $c->add_reading( { 'id' => 'n21p0', 'text' => 'un', 'join_next' => 1 } ); +my $old_r = $c->reading( 'n21' ); +$old_r->alter_text( 'to' ); +$c->del_path( 'n20', 'n21', 'A' ); +$c->add_path( 'n20', 'n21p0', 'A' ); +$c->add_path( 'n21p0', 'n21', 'A' ); +$c->flatten_ranks(); +ok( $c->reading( 'n21p0' ), "New reading exists" ); +is( scalar $c->readings, $rno, "Reading add offset by flatten_ranks" ); + +# Combine n3 and n4 +$c->merge_readings( 'n3', 'n4', 1 ); +ok( !$c->reading('n4'), "Reading n4 is gone" ); +is( $c->reading('n3')->text, 'with his', "Reading n3 has both words" ); + +# Collapse n25 and n26 +$c->merge_readings( 'n25', 'n26' ); +ok( !$c->reading('n26'), "Reading n26 is gone" ); +is( $c->reading('n25')->text, 'rood', "Reading n25 has an unchanged word" ); + +# Combine n21 and n21p0 +my $remaining = $c->reading('n21'); +$remaining ||= $c->reading('n22'); # one of these should still exist +$c->merge_readings( 'n21p0', $remaining, 1 ); +ok( !$c->reading('n21'), "Reading $remaining is gone" ); +is( $c->reading('n21p0')->text, 'unto', "Reading n21p0 merged correctly" ); + +=end testing + +=cut sub merge_readings { my $self = shift; # We only need the IDs for adding paths to the graph, not the reading # objects themselves. - my( $kept, $deleted, $combine_char ) = $self->_stringify_args( @_ ); + my( $kept, $deleted, $combine, $combine_char ) = $self->_stringify_args( @_ ); + $self->_graphcalc_done(0); # The kept reading should inherit the paths and the relationships # of the deleted reading. @@ -305,11 +383,15 @@ sub merge_readings { $self->relations->merge_readings( $kept, $deleted, $combine_char ); # Do the deletion deed. - if( $combine_char ) { + if( $combine ) { my $kept_obj = $self->reading( $kept ); - my $new_text = join( $combine_char, $kept_obj->text, - $self->reading( $deleted )->text ); - $kept_obj->alter_text( $new_text ); + my $del_obj = $self->reading( $deleted ); + my $joinstr = $combine_char; + unless( defined $joinstr ) { + $joinstr = '' if $kept_obj->join_next || $del_obj->join_prior; + $joinstr = $self->wordsep unless defined $joinstr; + } + $kept_obj->alter_text( join( $joinstr, $kept_obj->text, $del_obj->text ) ); } $self->del_reading( $deleted ); } @@ -317,12 +399,12 @@ sub merge_readings { # Helper function for manipulating the graph. sub _stringify_args { - my( $self, $first, $second, $arg ) = @_; + my( $self, $first, $second, @args ) = @_; $first = $first->id if ref( $first ) eq 'Text::Tradition::Collation::Reading'; $second = $second->id if ref( $second ) eq 'Text::Tradition::Collation::Reading'; - return( $first, $second, $arg ); + return( $first, $second, @args ); } # Helper function for manipulating the graph. @@ -343,6 +425,7 @@ sub add_path { # objects themselves. my( $source, $target, $wit ) = $self->_stringify_args( @_ ); + $self->_graphcalc_done(0); # Connect the readings $self->sequence->add_edge( $source, $target ); # Note the witness in question @@ -363,6 +446,7 @@ sub del_path { # objects themselves. my( $source, $target, $wit ) = $self->_stringify_args( @args ); + $self->_graphcalc_done(0); if( $self->sequence->has_edge_attribute( $source, $target, $wit ) ) { $self->sequence->delete_edge_attribute( $source, $target, $wit ); } @@ -391,6 +475,7 @@ be called via $tradition->del_witness. sub clear_witness { my( $self, @sigils ) = @_; + $self->_graphcalc_done(0); # Clear the witness(es) out of the paths foreach my $e ( $self->paths ) { foreach my $sig ( @sigils ) { @@ -411,11 +496,21 @@ sub add_relationship { my( $source, $target, $opts ) = $self->_stringify_args( @_ ); my( @vectors ) = $self->relations->add_relationship( $source, $self->reading( $source ), $target, $self->reading( $target ), $opts ); - # Force a full rank recalculation every time. Yuck. - $self->calculate_ranks() if $self->end->has_rank; + $self->_graphcalc_done(0); return @vectors; } +around qw/ get_relationship del_relationship / => sub { + my $orig = shift; + my $self = shift; + my @args = @_; + if( @args == 1 && ref( $args[0] ) eq 'ARRAY' ) { + @args = @{$_[0]}; + } + my( $source, $target ) = $self->_stringify_args( @args ); + $self->$orig( $source, $target ); +}; + =head2 reading_witnesses( $reading ) Return a list of sigils corresponding to the witnesses in which the reading appears. @@ -434,6 +529,10 @@ sub reading_witnesses { my $wits = $self->sequence->get_edge_attributes( @$e ); @all_witnesses{ keys %$wits } = 1; } + my $acstr = $self->ac_label; + foreach my $acwit ( grep { $_ =~ s/^(.*)\Q$acstr\E$/$1/ } keys %all_witnesses ) { + delete $all_witnesses{$acwit.$acstr} if exists $all_witnesses{$acwit}; + } return keys %all_witnesses; } @@ -442,13 +541,16 @@ sub reading_witnesses { =head2 as_svg( \%options ) Returns an SVG string that represents the graph, via as_dot and graphviz. -See as_dot for a list of options. +See as_dot for a list of options. Must have GraphViz (dot) installed to run. =cut sub as_svg { my( $self, $opts ) = @_; + throw( "Need GraphViz installed to output SVG" ) + unless File::Which::which( 'dot' ); my $want_subgraph = exists $opts->{'from'} || exists $opts->{'to'}; + $self->calculate_ranks() unless $self->_graphcalc_done; if( !$self->has_cached_svg || $opts->{'recalc'} || $want_subgraph ) { my @cmd = qw/dot -Tsvg/; my( $svg, $err ); @@ -554,6 +656,8 @@ sub as_dot { next if $reading->id eq $reading->text; my $rattrs; my $label = $reading->text; + $label .= '-' if $reading->join_next; + $label = "-$label" if $reading->join_prior; $label =~ s/\"/\\\"/g; $rattrs->{'label'} = $label; $rattrs->{'fillcolor'} = '#b3f36d' if $reading->is_common && $color_common; @@ -687,7 +791,16 @@ sub _path_display_label { return join( ', ', @wits ); } } - + +=head2 witnesses_at_rank + +Returns a list of witnesses that are not lacunose, for a given rank. + +=cut + +sub witnesses_at_rank { + my( $self, $rank ) = @_; +} =head2 as_graphml @@ -737,7 +850,8 @@ is( scalar $st->collation->relationships, 3, "Reparsed collation has new relatio sub as_graphml { my( $self ) = @_; - + $self->calculate_ranks unless $self->_graphcalc_done; + # Some namespaces my $graphml_ns = 'http://graphml.graphdrawing.org/xmlns'; my $xsi_ns = 'http://www.w3.org/2001/XMLSchema-instance'; @@ -759,7 +873,7 @@ sub as_graphml { $graph_data_keys{$datum} = 'dg'.$gdi++; my $key = $root->addNewChild( $graphml_ns, 'key' ); $key->setAttribute( 'attr.name', $datum ); - $key->setAttribute( 'attr.type', $key eq 'linear' ? 'boolean' : 'string' ); + $key->setAttribute( 'attr.type', $datum eq 'linear' ? 'boolean' : 'string' ); $key->setAttribute( 'for', 'graph' ); $key->setAttribute( 'id', $graph_data_keys{$datum} ); } @@ -774,6 +888,9 @@ sub as_graphml { is_start => 'boolean', is_end => 'boolean', is_lacuna => 'boolean', + is_common => 'boolean', + join_prior => 'boolean', + join_next => 'boolean', ); foreach my $datum ( keys %node_data ) { $node_data_keys{$datum} = 'dn'.$ndi++; @@ -793,6 +910,7 @@ sub as_graphml { relationship => 'string', # ID/label for a relationship extra => 'boolean', # Path key scope => 'string', # Relationship key + annotation => 'string', # Relationship key non_correctable => 'boolean', # Relationship key non_independent => 'boolean', # Relationship key ); @@ -892,7 +1010,7 @@ row per witness (or witness uncorrected.) sub as_csv { my( $self ) = @_; - my $table = $self->make_alignment_table; + my $table = $self->alignment_table; my $csv = Text::CSV_XS->new( { binary => 1, quote_null => 0 } ); my @result; # Make the header row @@ -901,14 +1019,14 @@ sub as_csv { # Make the rest of the rows foreach my $idx ( 0 .. $table->{'length'} - 1 ) { my @rowobjs = map { $_->{'tokens'}->[$idx] } @{$table->{'alignment'}}; - my @row = map { $_ ? $_->{'t'} : $_ } @rowobjs; + my @row = map { $_ ? $_->{'t'}->text : $_ } @rowobjs; $csv->combine( @row ); push( @result, decode_utf8( $csv->string ) ); } return join( "\n", @result ); } -=head2 make_alignment_table( $use_refs, $include_witnesses ) +=head2 alignment_table( $use_refs, $include_witnesses ) Return a reference to an alignment table, in a slightly enhanced CollateX format which looks like this: @@ -928,8 +1046,11 @@ keys have a true hash value will be included. =cut -sub make_alignment_table { - my( $self, $noderefs, $include ) = @_; +sub alignment_table { + my( $self ) = @_; + $self->calculate_ranks() unless $self->_graphcalc_done; + return $self->cached_table if $self->has_cached_table; + # Make sure we can do this throw( "Need a linear graph in order to make an alignment table" ) unless $self->linear; @@ -938,27 +1059,25 @@ sub make_alignment_table { my $table = { 'alignment' => [], 'length' => $self->end->rank - 1 }; my @all_pos = ( 1 .. $self->end->rank - 1 ); foreach my $wit ( sort { $a->sigil cmp $b->sigil } $self->tradition->witnesses ) { - if( $include ) { - next unless $include->{$wit->sigil}; - } # print STDERR "Making witness row(s) for " . $wit->sigil . "\n"; my @wit_path = $self->reading_sequence( $self->start, $self->end, $wit->sigil ); - my @row = _make_witness_row( \@wit_path, \@all_pos, $noderefs ); + my @row = _make_witness_row( \@wit_path, \@all_pos ); push( @{$table->{'alignment'}}, { 'witness' => $wit->sigil, 'tokens' => \@row } ); if( $wit->is_layered ) { my @wit_ac_path = $self->reading_sequence( $self->start, $self->end, $wit->sigil.$self->ac_label ); - my @ac_row = _make_witness_row( \@wit_ac_path, \@all_pos, $noderefs ); + my @ac_row = _make_witness_row( \@wit_ac_path, \@all_pos ); push( @{$table->{'alignment'}}, { 'witness' => $wit->sigil.$self->ac_label, 'tokens' => \@ac_row } ); } } - return $table; + $self->cached_table( $table ); + return $table; } sub _make_witness_row { - my( $path, $positions, $noderefs ) = @_; + my( $path, $positions ) = @_; my %char_hash; map { $char_hash{$_} = undef } @$positions; my $debug = 0; @@ -967,8 +1086,7 @@ sub _make_witness_row { $rtext = '#LACUNA#' if $rdg->is_lacuna; print STDERR "rank " . $rdg->rank . "\n" if $debug; # print STDERR "No rank for " . $rdg->id . "\n" unless defined $rdg->rank; - $char_hash{$rdg->rank} = $noderefs ? { 't' => $rdg } - : { 't' => $rtext }; + $char_hash{$rdg->rank} = { 't' => $rdg }; } my @row = map { $char_hash{$_} } @$positions; # Fill in lacuna markers for undef spots in the row @@ -977,8 +1095,8 @@ sub _make_witness_row { foreach my $el ( @row ) { # If we are using node reference, make the lacuna node appear many times # in the table. If not, use the lacuna tag. - if( $last_el && _el_is_lacuna( $last_el ) && !defined $el ) { - $el = $noderefs ? $last_el : { 't' => '#LACUNA#' }; + if( $last_el && $last_el->{'t'}->is_lacuna && !defined $el ) { + $el = $last_el; } push( @filled_row, $el ); $last_el = $el; @@ -986,30 +1104,6 @@ sub _make_witness_row { return @filled_row; } -# Tiny utility function to say if a table element is a lacuna -sub _el_is_lacuna { - my $el = shift; - return 1 if $el->{'t'} eq '#LACUNA#'; - return 1 if ref( $el->{'t'} ) eq 'Text::Tradition::Collation::Reading' - && $el->{'t'}->is_lacuna; - return 0; -} - -# Helper to turn the witnesses along columns rather than rows. Assumes -# equal-sized rows. -sub _turn_table { - my( $table ) = @_; - my $result = []; - return $result unless scalar @$table; - my $nrows = scalar @{$table->[0]}; - foreach my $idx ( 0 .. $nrows - 1 ) { - foreach my $wit ( 0 .. $#{$table} ) { - $result->[$idx]->[$wit] = $table->[$wit]->[$idx]; - } - } - return $result; -} - =head1 NAVIGATION METHODS =head2 reading_sequence( $first, $last, $sigil, $backup ) @@ -1178,7 +1272,17 @@ sub path_text { $start = $self->start unless $start; $end = $self->end unless $end; my @path = grep { !$_->is_meta } $self->reading_sequence( $start, $end, $wit ); - return join( ' ', map { $_->text } @path ); + my $pathtext = ''; + my $last; + foreach my $r ( @path ) { + if( $r->join_prior || !$last || $last->join_next ) { + $pathtext .= $r->text; + } else { + $pathtext .= ' ' . $r->text; + } + $last = $r; + } + return $pathtext; } =head1 INITIALIZATION METHODS @@ -1249,14 +1353,13 @@ my $t = Text::Tradition->new( my $c = $t->collation; # Make an svg -my $svg = $c->as_svg; -is( substr( $svg, 0, 5 ), 'has_cached_svg, "SVG was cached" ); -is( $c->as_svg, $svg, "Cached SVG returned upon second call" ); +my $table = $c->alignment_table; +ok( $c->has_cached_table, "Alignment table was cached" ); +is( $c->alignment_table, $table, "Cached table returned upon second call" ); $c->calculate_ranks; -is( $c->as_svg, $svg, "Cached SVG retained with no rank change" ); +is( $c->alignment_table, $table, "Cached table retained with no rank change" ); $c->add_relationship( 'n9', 'n23', { 'type' => 'spelling' } ); -isnt( $c->as_svg, $svg, "SVG changed after relationship add" ); +isnt( $c->alignment_table, $table, "Alignment table changed after relationship add" ); =end testing @@ -1323,14 +1426,20 @@ sub calculate_ranks { throw( "Ranks not calculated after $last - do you have a cycle in the graph?" ); } } - # Do we need to invalidate the cached SVG? - if( $self->has_cached_svg ) { + # Do we need to invalidate the cached data? + if( $self->has_cached_svg || $self->has_cached_table ) { foreach my $r ( $self->readings ) { - next if $existing_ranks{$r} == $r->rank; - $self->wipe_svg; + next if defined( $existing_ranks{$r} ) + && $existing_ranks{$r} == $r->rank; + # Something has changed, so clear the cache + $self->_clear_cache; + # ...and recalculate the common readings. + $self->calculate_common_readings(); last; } } + # The graph calculation information is now up to date. + $self->_graphcalc_done(1); } sub _assign_rank { @@ -1368,6 +1477,13 @@ sub _assign_rank { return @next_nodes; } +sub _clear_cache { + my $self = shift; + $self->wipe_svg if $self->has_cached_svg; + $self->wipe_table if $self->has_cached_table; +} + + =head2 flatten_ranks A convenience method for parsing collation data. Searches the graph for readings @@ -1391,44 +1507,6 @@ sub flatten_ranks { } } } - -=head2 remove_collations - -Another convenience method for parsing. Removes all 'collation' relationships -that were defined in order to get the reading ranks to be correct. - -=begin testing - -use Text::Tradition; - -my $cxfile = 't/data/Collatex-16.xml'; -my $t = Text::Tradition->new( - 'name' => 'inline', - 'input' => 'CollateX', - 'file' => $cxfile, - ); -my $c = $t->collation; - -isnt( $c->reading('n23')->rank, $c->reading('n9')->rank, "Rank skew exists" ); -$c->add_relationship( 'n23', 'n9', { 'type' => 'collated', 'scope' => 'local' } ); -is( scalar $c->relationships, 4, "Found all expected relationships" ); -$c->remove_collations; -is( scalar $c->relationships, 3, "Collated relationships now gone" ); -is( $c->reading('n23')->rank, $c->reading('n9')->rank, "Aligned ranks were preserved" ); - -=end testing - -=cut - -sub remove_collations { - my $self = shift; - foreach my $reledge ( $self->relationships ) { - my $relobj = $self->relations->get_relationship( $reledge ); - if( $relobj && $relobj->type eq 'collated' ) { - $self->relations->delete_relationship( $reledge ); - } - } -} =head2 calculate_common_readings @@ -1463,9 +1541,13 @@ is_deeply( \@marked, \@expected, "Found correct list of common readings" ); sub calculate_common_readings { my $self = shift; my @common; - my $table = $self->make_alignment_table( 1 ); + map { $_->is_common( 0 ) } $self->readings; + # Implicitly calls calculate_ranks + my $table = $self->alignment_table; foreach my $idx ( 0 .. $table->{'length'} - 1 ) { - my @row = map { $_->{'tokens'}->[$idx]->{'t'} } @{$table->{'alignment'}}; + my @row = map { $_->{'tokens'}->[$idx] + ? $_->{'tokens'}->[$idx]->{'t'} : '' } + @{$table->{'alignment'}}; my %hash; foreach my $r ( @row ) { if( $r ) {