1 package Text::Tradition::Parser::CTE;
6 use Encode qw/ decode /;
7 use Text::Tradition::Error;
8 use Text::Tradition::Parser::Util qw/ collate_variants /;
10 use XML::LibXML::XPathContext;
15 Text::Tradition::Parser::CTE
19 Parser module for Text::Tradition, given a TEI file exported from
20 Classical Text Editor.
26 my @apparatus = read( $xml_file );
28 Takes a Tradition object and a TEI file exported from Classical Text
29 Editor using double-endpoint-attachment critical apparatus encoding;
30 initializes the Tradition from the file.
34 my %sigil_for; # Save the XML IDs for witnesses.
35 my %apps; # Save the apparatus XML for a given ID.
36 my %has_ac; # Keep track of witnesses that have corrections.
39 my( $tradition, $opts ) = @_;
40 my $c = $tradition->collation; # Some shorthand
43 $opts->{interpret_transposition} = 1;
45 # First, parse the XML.
46 my( $tei, $xpc ) = _remove_formatting( $opts );
47 return unless $tei; # we have already warned.
49 # CTE uses a DTD rather than any xmlns-based parsing. Thus we
50 # need no namespace handling.
51 # Get the witnesses and create the witness objects.
52 foreach my $wit_el ( $xpc->findnodes( '//sourceDesc/listWit/witness' ) ) {
53 # The witness xml:id is used internally, and is *not* the sigil name.
54 my $id= $wit_el->getAttribute( 'xml:id' );
55 # If the witness element has an abbr element, that is the sigil. Otherwise
56 # the whole thing is the sigil.
57 my $sig = $xpc->findvalue( 'abbr', $wit_el );
58 my $identifier = 'CTE witness';
60 # The sigil is what is in the <abbr/> tag; the identifier is anything
62 $identifier = _tidy_identifier(
63 $xpc->findvalue( 'child::text()', $wit_el ) );
65 my @sig_parts = $xpc->findnodes( 'descendant::text()', $wit_el );
66 $sig = _stringify_sigil( @sig_parts );
68 say STDERR "Adding witness $sig ($identifier)";
69 $tradition->add_witness( sigil => $sig, identifier => $identifier,
70 sourcetype => 'collation' );
71 $sigil_for{'#'.$id} = $sig; # Make life easy by keying on the ID ref syntax
74 # Now go through the text and find the base tokens, apparatus tags, and
75 # anchors. Make a giant array of all of these things in sequence.
76 # TODO consider combining this with creation of graph below
78 foreach my $pg_el ( $xpc->findnodes( '/TEI/text/body/p' ) ) {
79 foreach my $xn ( $pg_el->childNodes ) {
80 push( @base_text, _get_base( $xn ) );
83 # We now have to work through this array applying the alternate
84 # apparatus readings to the base text. Essentially we will put
85 # everything on the graph, from which we will delete the apps and
86 # anchors when we are done.
88 # First, put the base tokens, apps, and anchors in the graph. Save the
89 # app siglorum separately as it has to be processed in order.
94 foreach my $item ( @base_text ) {
96 if( $item->{'type'} eq 'token' ) {
97 $r = $c->add_reading( { id => 'n'.$counter++,
98 text => $item->{'content'} } );
99 } elsif ( $item->{'type'} eq 'anchor' ) {
100 $r = $c->add_reading( { id => '__ANCHOR_' . $item->{'content'} . '__',
102 } elsif ( $item->{'type'} eq 'app' ) {
103 my $tag = '__APP_' . $counter++ . '__';
104 $r = $c->add_reading( { id => $tag, is_ph => 1 } );
105 my $app = $item->{'content'};
107 # Apparatus should be differentiable by type attribute; apparently
108 # it is not. Peek at the content to categorize it.
109 # Apparatus criticus is type a1; app siglorum is type a2
110 my @sigtags = $xpc->findnodes( 'descendant::*[name(witStart) or name(witEnd)]', $app );
112 push( @app_sig, $tag );
114 push( @app_crit, $tag );
117 $c->add_path( $last, $r, $c->baselabel );
120 $c->add_path( $last, $c->end, $c->baselabel );
122 # Now we can parse the apparatus entries, and add the variant readings
124 foreach my $app_id ( @app_crit ) {
125 _add_readings( $c, $app_id, $opts );
127 _add_lacunae( $c, @app_sig );
129 # Finally, add explicit witness paths, remove the base paths, and remove
130 # the app/anchor tags.
132 _expand_all_paths( $c );
133 } catch( Text::Tradition::Error $e ) {
134 throw( $e->message );
139 # Save the text for each witness so that we can ensure consistency
141 unless( $opts->{'nocalc'} ) {
143 $tradition->collation->text_from_paths();
144 $tradition->collation->calculate_ranks();
145 $tradition->collation->flatten_ranks();
146 } catch( Text::Tradition::Error $e ) {
147 throw( $e->message );
154 sub _stringify_sigil {
156 my @parts = grep { /\w/ } map { $_->data } @nodes;
157 my $whole = join( '', @parts );
162 sub _tidy_identifier {
168 # Get rid of all the formatting elements that get in the way of tokenization.
169 sub _remove_formatting {
172 # First, parse the original XML
173 my $parser = XML::LibXML->new();
175 if( exists $opts->{'string'} ) {
176 $doc = $parser->parse_string( $opts->{'string'} );
177 } elsif ( exists $opts->{'file'} ) {
178 $doc = $parser->parse_file( $opts->{'file'} );
179 } elsif ( exists $opts->{'xmlobj'} ) {
180 $doc = $opts->{'xmlobj'};
182 warn "Could not find string or file option to parse";
186 # Second, remove the formatting
187 my $xpc = XML::LibXML::XPathContext->new( $doc->documentElement );
188 my @useless = $xpc->findnodes( '//hi' );
189 foreach my $n ( @useless ) {
190 my $parent = $n->parentNode();
191 my @children = $n->childNodes();
192 my $first = shift @children;
194 $parent->replaceChild( $first, $n );
195 foreach my $c ( @children ) {
196 $parent->insertAfter( $c, $first );
200 $parent->removeChild( $n );
204 # Third, write out and reparse to merge the text nodes.
205 my $enc = $doc->encoding || 'UTF-8';
206 my $result = decode( $enc, $doc->toString() );
207 my $tei = $parser->parse_string( $result )->documentElement;
208 unless( $tei->nodeName =~ /^tei(corpus)?$/i ) {
209 throw( "Parsed document has non-TEI root element " . $tei->nodeName );
211 $xpc = XML::LibXML::XPathContext->new( $tei );
212 return( $tei, $xpc );
215 ## Helper function to help us navigate through nested XML, picking out
216 ## the words, the apparatus, and the anchors.
221 if( $xn->nodeType == XML_TEXT_NODE ) {
222 # Base text, just split the words on whitespace and add them
226 my @tokens = split( /\s+/, $str );
227 push( @readings, map { { type => 'token', content => $_ } } @tokens );
228 } elsif( $xn->nodeName eq 'app' ) {
229 # Apparatus, just save the entire XML node.
230 push( @readings, { type => 'app', content => $xn } );
231 } elsif( $xn->nodeName eq 'anchor' ) {
232 # Anchor to mark the end of some apparatus; save its ID.
233 if( $xn->hasAttribute('xml:id') ) {
234 push( @readings, { type => 'anchor',
235 content => $xn->getAttribute( 'xml:id' ) } );
236 } # if the anchor has no XML ID, it is not relevant to us.
237 } elsif( $xn->nodeName !~ /^(note|seg|milestone|emph)$/ ) { # Any tag we don't know to disregard
238 say STDERR "Unrecognized tag " . $xn->nodeName;
244 my( $list, @tokens ) = @_;
245 if( @$list && $list->[-1]->{'content'} =~ /\#JOIN\#$/ ) {
246 # The list evidently ended mid-word; join the next token onto it.
247 my $t = shift @tokens;
248 if( ref $t && $t->{'type'} eq 'token' ) {
250 $t = $t->{'content'};
252 # An app or anchor intervened; end the word.
253 unshift( @tokens, $t );
256 $list->[-1]->{'content'} =~ s/\#JOIN\#$/$t/;
258 foreach my $t ( @tokens ) {
259 unless( ref( $t ) ) {
260 $t = { 'type' => 'token', 'content' => $t };
267 my( $c, $app_id, $opts ) = @_;
268 my $xn = $apps{$app_id};
269 my $anchor = _anchor_name( $xn->getAttribute( 'to' ) );
271 # Get the lemma, which is all the readings between app and anchor,
272 # excluding other apps or anchors.
273 my @lemma = _return_lemma( $c, $app_id, $anchor );
275 # For each reading, send its text to 'interpret' along with the lemma,
276 # and then save the list of witnesses that these tokens belong to.
277 my %wit_rdgs; # Maps from witnesses to the variant text
280 $tag =~ s/^\__APP_(.*)\__$/$1/;
282 foreach my $rdg ( $xn->getChildrenByTagName( 'rdg' ) ) {
283 # Get the relevant witnesses.
284 my @witlist = map { $sigil_for{$_} }
285 split( /\s+/, $rdg->getAttribute( 'wit' ) );
287 # Does the reading have an ID? If so it probably has a witDetail
288 # attached, and we need to read it. If an A.C. or P.C. reading is
289 # declared explicity, this is where it will be dealt with.
290 if( $rdg->hasAttribute( 'xml:id' ) ) {
291 my $rid = $rdg->getAttribute( 'xml:id' );
292 my $xpc = XML::LibXML::XPathContext->new( $xn );
293 my @details = $xpc->findnodes( './witDetail[@target="'.$rid.'"]' );
294 foreach my $d ( @details ) {
295 @witlist = _parse_wit_detail( $d, @witlist );
299 # Now we have our list of relevant witnesses for the reading, annotated
300 # with AC or PC if applicable. Interpret the reading in light of the
301 # lemma, anything we already have for the witness, etc.
302 # If an A.C. or P.C. reading is implied rather than explicitly noted,
303 # this is where it will be dealt with.
305 foreach my $wit ( @witlist ) {
306 # The lemma for this witness is either the actual lemma, or the
307 # reading that we have already determined.
309 if( $wit =~ /^(.*)_pc$/ ) {
313 ## TODO think through ac/pc interaction from these specs
314 my $wit_lemma = $wit_rdgs{$wit} || \@lemma;
316 ( $wit, @rdg_nodes )= _read_reading( $rdg, $wit_lemma, $wit,
317 $tag, $ctr, $anchor, $opts );
318 $wit_rdgs{$wit} = \@rdg_nodes;
319 # If the PC flag is set, there is a corresponding AC that
320 # follows the lemma and has to be explicitly declared.
321 if( $flag->{'PC'} ) {
322 $wit_rdgs{$wit.'_ac'} = $wit_lemma;
327 # Now collate the variant readings, since it is not done for us.
328 collate_variants( $c, \@lemma, values %wit_rdgs );
330 # Now add the witness paths for each reading.
331 foreach my $wit_id ( keys %wit_rdgs ) {
332 my $witstr = _get_sigil( $wit_id, $c->ac_label );
333 my $rdg_list = $wit_rdgs{$wit_id};
334 _add_wit_path( $c, $rdg_list, $app_id, $anchor, $witstr );
341 return sprintf( "__ANCHOR_%s__", $xmlid );
345 my( $c, $app, $anchor ) = @_;
346 my @nodes = grep { $_->id !~ /^__A(PP|NCHOR)/ }
347 $c->reading_sequence( $c->reading( $app ), $c->reading( $anchor ),
352 sub _parse_wit_detail {
355 map { $wits{$_} = $_ } @_;
356 my @changewits = map { $sigil_for{$_} }
357 split( /\s+/, $detail->getAttribute( 'wit' ) );
358 my $content = $detail->textContent;
359 if( $content =~ /^a\.?\s*c(orr)?\.$/ ) {
360 # Replace the key in the $readings hash
361 map { $wits{$_} = $_.'_ac' } @changewits;
362 } elsif( $content =~ /^p\.?\s*c(orr)?\.$/ || $content =~ /^s\.?\s*l\.$/
363 || $content =~ /^in marg\.?$/ ) {
364 # If no key for the wit a.c. exists, add one pointing to the lemma
365 map { $wits{$_} = $_.'_pc' } @changewits;
366 } else { #...not sure what it is?
367 say STDERR "WARNING: Unrecognized sigil annotation $content";
369 my @newwits = sort values %wits;
374 my( $rdg, $lemma, $witness, $tag, $ctr, $anchor, $opts ) = @_;
376 # Get the text of the lemma.
377 my $lemma_str = join( ' ', map { $_->text } grep { !$_->is_ph } @$lemma );
380 foreach ( $rdg->childNodes ) {
381 push( @text, _get_base( $_ ) );
383 my( $interpreted, $flag ) = ( '', undef );
385 ( $interpreted, $flag ) = interpret(
386 join( ' ', map { $_->{'content'} } @text ), $lemma_str, $anchor, $opts );
388 if( ( $interpreted eq $lemma_str || $interpreted eq '__LEMMA__' )
390 # The reading is the lemma. Pass it back.
391 return( $wit, @$lemma );
395 if( $interpreted eq '#LACUNA#' ) {
396 push( @rdg_nodes, $c->add_reading( { id => 'r'.$tag.".".$ctr++,
397 is_lacuna => 1 } ) );
398 } elsif( $flag->{'TR'} ) {
399 # Our reading is transposed to after the given string. Look
400 # down the collation base text and try to find it.
401 # The @rdg_nodes should remain blank here, so that the correct
402 # omission goes into the graph.
404 foreach my $w ( split( /\s+/, $interpreted ) ) {
405 my $r = $c->add_reading( { id => 'r'.$tag.".".$ctr++,
407 push( @transp_nodes, $r );
409 if( $anchor && @lemma ) {
410 my $success = _attach_transposition( $c, $lemma, $anchor,
411 \@transp_nodes, $witlist, $flag->{'TR'} );
413 # If we didn't manage to insert the displaced reading,
414 # then restore it here rather than silently deleting it.
415 push( @rdg_nodes, @transp_nodes );
419 foreach my $w ( split( /\s+/, $interpreted ) ) {
420 if( $w eq '__LEMMA__' ) {
421 push( @rdg_nodes, @lemma );
423 my $r = $c->add_reading( { id => 'r'.$tag.".".$ctr++,
425 push( @rdg_nodes, $r );
430 # Note if the interpretation said that we're dealing with a correction.
431 if( $flag->{'AC'} ) {
433 } elsif( $flag->{'PC'} ) {
436 return( $wit, @rdg_nodes );
439 # Make a best-effort attempt to attach a transposition farther down the line.
440 # $lemmaseq contains the Reading objects of the lemma
441 # $anchor contains the point at which we should start scanning for a match
442 # $rdgseq contains the Reading objects of the transposed reading
443 # (should be identical to the lemma)
444 # $witlist contains the list of applicable witnesses
445 # $reftxt contains the text to match, after which the $rdgseq should go.
446 sub _attach_transposition {
447 my( $c, $lemmaseq, $anchor, $rdgseq, $witlist, $reftxt ) = @_;
448 my @refwords = split( /\s+/, $reftxt );
449 my $checked = $c->reading( $anchor );
452 while( $checked ne $c->end && !$found ) {
453 my $next = $c->next_reading( $checked, $c->baselabel );
454 if( $next->text eq $refwords[0] ) {
455 # See if the entire sequence of words matches.
457 foreach my $w ( 1..$#refwords ) {
458 $found = $c->next_reading( $next, $c->baselabel );
459 unless( $found->text eq $refwords[$w] ) {
468 # The $found variable should now contain the reading after which we
469 # should stick the transposition.
470 my $fnext = $c->next_reading( $found, $c->baselabel );
471 my $aclabel = $c->ac_label;
472 foreach my $wit_id ( @$witlist ) {
473 my $witstr = _get_sigil( $wit_id, $aclabel );
474 _add_wit_path( $c, $rdgseq, $found->id, $fnext->id, $witstr );
476 # ...and add the transposition relationship between lemma and rdgseq.
477 if( @$lemmaseq == @$rdgseq ) {
478 foreach my $i ( 0..$#{$lemmaseq} ) {
479 $c->add_relationship( $lemmaseq->[$i], $rdgseq->[$i],
480 { type => 'transposition', annotation => 'Detected by CTE' } );
484 throw( "Lemma at $found and transposed sequence different lengths?!" );
487 say STDERR "WARNING: Unable to find $reftxt in base text for transposition";
492 =head2 interpret( $reading, $lemma )
494 Given a string in $reading and a corresponding lemma in $lemma, interpret what
495 the actual reading should be. Used to deal with apparatus-ese shorthands for
496 marking transpositions, prefixed or suffixed words, and the like.
501 # A utility function to change apparatus-ese into a full variant.
502 my( $reading, $lemma, $anchor, $opts ) = @_;
503 return $reading if $reading eq $lemma;
504 my $oldreading = $reading;
505 # $lemma =~ s/\s+[[:punct:]]+$//;
506 my $flag = {}; # To pass back extra info about the interpretation
507 my @words = split( /\s+/, $lemma );
508 # Discard any 'sic' notation - that rather goes without saying.
509 $reading =~ s/([[:punct:]]+)?sic([[:punct:]]+)?//g;
511 # Look to see if there is an implied add or praem masked by the XML.
512 # If so, undo it for purposes of reading identity.
513 $reading =~ s/^$lemma\b/__LEMMA__/;
514 $reading =~ s/\b$lemma$/__LEMMA__/;
516 # Now look for common jargon.
517 if( $reading =~ /^(.*) praem.$/ || $reading =~ /^praem\. (.*)$/ ) {
518 $reading = "$1 __LEMMA__";
519 } elsif( $reading =~ /^(.*) add.$/ || $reading =~ /^add\. (.*)$/ ) {
520 $reading = "__LEMMA__ $1";
521 } elsif( $reading =~ /locus [uv]acuus/
522 || $reading eq 'def.'
523 || $reading eq 'illeg.'
524 || $reading eq 'desunt'
526 $reading = '#LACUNA#';
527 } elsif( $reading eq 'om.' ) {
529 } elsif( $reading =~ /^in[uv]\.$/
530 || $reading =~ /^tr(ans(p)?)?\.$/ ) {
531 # Hope it is two words.
532 say STDERR "WARNING: want to invert a lemma that is not two words"
533 unless scalar( @words ) == 2;
534 $reading = join( ' ', reverse( @words ) );
535 } elsif( $reading =~ /^iter(\.|at)$/ ) {
537 $reading = "__LEMMA__ $lemma";
538 } elsif( $reading =~ /^(.*?)\s*\(?in marg\.\)?$/ ) {
541 # The given text is a correction.
544 # The lemma itself was the correction; the witness carried
545 # no reading pre-correction.
548 } elsif( $reading =~ /^(.*) \.\.\. (.*)$/ ) {
549 # The first and last N words captured should replace the first and
550 # last N words of the lemma.
551 my @begin = split( /\s+/, $1 );
552 my @end = split( /\s+/, $2 );
553 if( scalar( @begin ) + scalar ( @end ) > scalar( @words ) ) {
554 # Something is wrong and we can't do the splice.
555 throw( "$lemma is too short to accommodate $oldreading" );
557 splice( @words, 0, scalar @begin, @begin );
558 splice( @words, -(scalar @end), scalar @end, @end );
559 $reading = join( ' ', @words );
561 } elsif( $opts->{interpret_transposition} &&
562 ( $reading =~ /^post\s*(?<lem>.*?)\s+tr(ans(p)?)?\.$/ ||
563 $reading =~ /^tr(ans(p)?)?\. post\s*(?<lem>.*)$/) ) {
564 # Try to deal with transposed readings
566 say STDERR "Will attempt transposition: $reading at $anchor";
567 # Copy the lemma into the reading string for insertion later
570 $flag->{'TR'} = $+{lem};
572 return( $reading, $flag );
576 my( $c, @app_id ) = @_;
577 # Go through the apparatus entries in order, noting where to start and stop our
581 foreach my $tag ( @app_id ) {
582 my $app = $apps{$tag};
583 # Find the anchor, if any. This marks the point where the text starts
585 my $anchor = $app->getAttribute( 'to' );
589 $aname = _anchor_name( $anchor );
592 foreach my $rdg ( $app->getChildrenByTagName( 'rdg' ) ) {
593 my @witlist = map { _get_sigil( $_, $c->ac_label ) }
594 split( /\s+/, $rdg->getAttribute( 'wit' ) );
595 my @start = $rdg->getChildrenByTagName( 'witStart' );
596 my @end = $rdg->getChildrenByTagName( 'witEnd' );
597 if( @start && @end ) {
598 throw( "App sig entry at $anchor has both witStart and witEnd!" );
600 if( @start && $anchor &&
601 $c->prior_reading( $aname, $c->baselabel ) ne $c->start ) {
602 # We are picking back up after a hiatus. Find the last end and
603 # add a lacuna link between there and here.
604 foreach my $wit ( @witlist ) {
605 my $stoppoint = delete $lacunose{$wit};
606 my $stopname = $stoppoint ? _anchor_name( $stoppoint ) : $c->start->id;
607 say STDERR "Adding lacuna for $wit between $stopname and $anchor";
608 my $lacuna = $c->add_reading( { id => "as_$anchor.".$ctr++,
610 _add_wit_path( $c, [ $lacuna ], $stopname, $aname, $wit );
612 } elsif( @end && $anchor &&
613 $c->next_reading( $aname, $c->baselabel ) ne $c->end ) {
614 # We are stopping. If we've already stopped for the given witness,
615 # flag an error; otherwise record the stopping point.
616 foreach my $wit ( @witlist ) {
617 if( $lacunose{$wit} ) {
618 throw( "Trying to end $wit at $anchor when already ended at "
621 $lacunose{$wit} = $anchor;
627 # For whatever remains in the %lacunose hash, add a lacuna between that spot and
628 # $c->end for each of the witnesses.
629 foreach my $wit ( keys %lacunose ) {
630 next unless $lacunose{$wit};
631 my $aname = _anchor_name( $lacunose{$wit} );
632 say STDERR "Adding lacuna for $wit from $aname to end";
633 my $lacuna = $c->add_reading( { id => 'as_'.$lacunose{$wit}.'.'.$ctr++,
635 _add_wit_path( $c, [ $lacuna ], $aname, $c->end, $wit );
640 my( $xml_id, $layerlabel ) = @_;
641 if( $xml_id =~ /^(.*)_ac$/ ) {
643 return $sigil_for{$real_id} . $layerlabel;
645 return $sigil_for{$xml_id};
649 sub _expand_all_paths {
652 # Walk the collation and fish out the paths for each witness
653 foreach my $wit ( $c->tradition->witnesses ) {
654 my $sig = $wit->sigil;
655 my @path = grep { !$_->is_ph }
656 $c->reading_sequence( $c->start, $c->end, $sig );
657 $wit->path( \@path );
658 if( $has_ac{$sig} ) {
659 my @ac_path = grep { !$_->is_ph }
660 $c->reading_sequence( $c->start, $c->end, $sig.$c->ac_label );
661 $wit->uncorrected_path( \@ac_path );
666 foreach my $anchor ( grep { $_->is_ph } $c->readings ) {
667 $c->del_reading( $anchor );
669 # Delete the base edges
670 map { $c->del_path( $_, $c->baselabel ) } $c->paths;
672 # Make the path edges
673 $c->make_witness_paths();
675 # Now remove any orphan nodes, and warn that we are doing so.
677 while( $c->sequence->predecessorless_vertices > 1 ) {
678 foreach my $v ( $c->sequence->predecessorless_vertices ) {
679 my $r = $c->reading( $v );
680 next if $r->is_start;
682 $tag =~ s/^r(\d+)\.\d+/$1/;
683 say STDERR "Deleting orphan reading $r / " . $r->text;
684 push( @{$suspect_apps{$tag}}, $r->id ) if $tag =~ /^\d+$/;
685 $c->del_reading( $r );
688 if( $c->sequence->successorless_vertices > 1 ) {
689 my @bad = grep { $_ ne $c->end->id } $c->sequence->successorless_vertices;
692 next unless $tag =~ /^r/;
693 $tag =~ s/^r(\d+)\.\d+/$1/;
694 push( @{$suspect_apps{$tag}}, $_ );
696 _dump_suspects( %suspect_apps );
697 throw( "Remaining hanging readings: @bad" );
699 _dump_suspects( %suspect_apps ) if keys %suspect_apps;
703 my( $c, $rdg, $app, $anchor, $wit ) = @_;
705 push( @nodes, $c->reading( $anchor ) );
707 my $cur = $c->reading( $app );
708 foreach my $n ( @nodes ) {
709 $c->add_path( $cur, $n, $wit );
716 say STDERR "Suspect apparatus entries:";
717 foreach my $suspect ( sort { $a <=> $b } keys %list ) {
718 my @badrdgs = @{$list{$suspect}};
719 say STDERR print_apparatus( $suspect );
720 say STDERR "\t(Linked to readings @badrdgs)";
724 sub print_apparatus {
726 my $tag = '__APP_' . $appid . '__';
727 my $app = $apps{$tag};
729 # Interpret the XML - get the lemma and readings and print them out.
730 my $xpc = XML::LibXML::XPathContext->new( $app );
731 my $anchor = $app->getAttribute('to');
733 # We have a lemma, so we construct it.
735 $appstring .= "(Anchor $anchor) ";
738 last if $curr->nodeType eq XML_ELEMENT_NODE
739 && $curr->hasAttribute( 'xml:id' )
740 && $curr->getAttribute( 'xml:id' ) eq $anchor;
741 $appstring .= $curr->data if $curr->nodeType eq XML_TEXT_NODE;
742 $curr = $curr->nextSibling;
747 foreach my $rdg_el ( $xpc->findnodes( 'child::rdg' ) ) {
751 foreach my $child_el ( $rdg_el->childNodes ) {
752 if( $child_el->nodeType eq XML_TEXT_NODE ) {
753 $rdgtext .= $child_el->data;
754 } elsif( $child_el->nodeName =~ /^wit(Start|End)$/ ) {
755 my $startend = lc( $1 );
756 } elsif( $child_el->nodeName eq 'witDetail' ) {
757 foreach my $wit ( map { _get_sigil( $_ ) }
758 split( /\s+/, $child_el->getAttribute('wit') ) ) {
759 $detail{$wit} = $child_el->textContent;
765 foreach my $witrep ( map { _get_sigil( $_ ) }
766 split( /\s+/, $rdg_el->getAttribute('wit') ) ) {
767 if( exists $detail{$witrep} ) {
768 $witrep .= '(' . $detail{$witrep} . ')'
770 if( $startend eq 'start' ) {
771 $witrep = '*' . $witrep;
772 } elsif( $startend eq 'end' ) {
775 push( @witlist, $witrep );
777 $rdgtext .= " @witlist";
778 push( @readings, $rdgtext );
780 $appstring .= join( ' ', @readings );
785 Text::Tradition::Error->throw(
786 'ident' => 'Parser::CTE error',
793 This package is free software and is provided "as is" without express
794 or implied warranty. You can redistribute it and/or modify it under
795 the same terms as Perl itself.
799 Tara L Andrews, aurum@cpan.org