Commit | Line | Data |
6f4946fb |
1 | package Text::Tradition::Parser::CTE; |
2 | |
3 | use strict; |
4 | use warnings; |
4d85a60e |
5 | use Text::Tradition::Parser::Util qw/ collate_variants /; |
6f4946fb |
6 | use XML::LibXML; |
7 | use XML::LibXML::XPathContext; |
8 | |
9 | =head1 NAME |
10 | |
11 | Text::Tradition::Parser::CTE |
12 | |
13 | =head1 DESCRIPTION |
14 | |
15 | Parser module for Text::Tradition, given a TEI file exported from |
16 | Classical Text Editor. |
17 | |
18 | =head1 METHODS |
19 | |
20 | =over |
21 | |
22 | =item B<parse> |
23 | |
24 | my @apparatus = read( $xml_file ); |
25 | |
26 | Takes a Tradition object and a TEI file exported from Classical Text |
4d85a60e |
27 | Editor using double-endpoint-attachment critical apparatus encoding; |
28 | initializes the Tradition from the file. |
6f4946fb |
29 | |
30 | =cut |
31 | |
4d85a60e |
32 | my %sigil_for; # Save the XML IDs for witnesses. |
33 | my %apps; # Save the apparatus XML for a given ID. |
f6e19c7c |
34 | my %has_ac; # Keep track of witnesses that have corrections. |
6f4946fb |
35 | |
36 | sub parse { |
dfc37e38 |
37 | my( $tradition, $opts ) = @_; |
4d85a60e |
38 | my $c = $tradition->collation; # Some shorthand |
39 | |
40 | # First, parse the XML. |
41 | my $parser = XML::LibXML->new(); |
dfc37e38 |
42 | my $doc; |
43 | if( exists $opts->{'string'} ) { |
44 | $doc = $parser->parse_string( $opts->{'string'} ); |
45 | } elsif ( exists $opts->{'file'} ) { |
46 | $doc = $parser->parse_file( $opts->{'file'} ); |
47 | } else { |
48 | warn "Could not find string or file option to parse"; |
49 | return; |
50 | } |
4d85a60e |
51 | my $tei = $doc->documentElement(); |
52 | my $xpc = XML::LibXML::XPathContext->new( $tei ); |
53 | |
54 | # CTE uses a DTD rather than any xmlns-based parsing. Thus we |
55 | # need no namespace handling. |
56 | |
57 | # Get the witnesses and create the witness objects. |
58 | foreach my $wit_el ( $xpc->findnodes( '//sourceDesc/listWit/witness' ) ) { |
59 | # The witness xml:id is used internally, and is *not* the sigil name. |
60 | my $id= $wit_el->getAttribute( 'xml:id' ); |
61 | my @sig_parts = $xpc->findnodes( './abbr/descendant::text()', $wit_el ); |
62 | my $sig = _stringify_sigil( @sig_parts ); |
b8f262e8 |
63 | $tradition->add_witness( sigil => $sig, source => $wit_el->toString() ); |
64 | $sigil_for{'#'.$id} = $sig; # Make life easy by keying on the ID ref syntax |
4d85a60e |
65 | } |
6f4946fb |
66 | |
4d85a60e |
67 | # Now go through the text and find the base tokens, apparatus tags, and |
68 | # anchors. Make a giant array of all of these things in sequence. |
69 | # TODO consider combining this with creation of graph below |
70 | my @base_text; |
71 | foreach my $pg_el ( $xpc->findnodes( '/TEI/text/body/p' ) ) { |
72 | foreach my $xn ( $pg_el->childNodes ) { |
73 | push( @base_text, _get_base( $xn ) ); |
74 | } |
6f4946fb |
75 | } |
4d85a60e |
76 | |
77 | # We now have to work through this array applying the alternate |
78 | # apparatus readings to the base text. Essentially we will put |
79 | # everything on the graph, from which we will delete the apps and |
80 | # anchors when we are done. |
f6e19c7c |
81 | |
82 | # First, put the base tokens, apps, and anchors in the graph. |
4d85a60e |
83 | my $counter = 0; |
84 | my $last = $c->start; |
85 | foreach my $item ( @base_text ) { |
86 | my $r; |
87 | if( $item->{'type'} eq 'token' ) { |
12720144 |
88 | $r = $c->add_reading( { id => 'n'.$counter++, |
89 | text => $item->{'content'} } ); |
4d85a60e |
90 | } elsif ( $item->{'type'} eq 'anchor' ) { |
12720144 |
91 | $r = $c->add_reading( { id => '#ANCHOR_' . $item->{'content'} . '#', |
92 | is_ph => 1 } ); |
4d85a60e |
93 | } elsif ( $item->{'type'} eq 'app' ) { |
94 | my $tag = '#APP_' . $counter++ . '#'; |
12720144 |
95 | $r = $c->add_reading( { id => $tag, is_ph => 1 } ); |
4d85a60e |
96 | $apps{$tag} = $item->{'content'}; |
97 | } |
f6e19c7c |
98 | $c->add_path( $last, $r, $c->baselabel ); |
4d85a60e |
99 | $last = $r; |
6f4946fb |
100 | } |
f6e19c7c |
101 | $c->add_path( $last, $c->end, $c->baselabel ); |
4d85a60e |
102 | |
103 | # Now we can parse the apparatus entries, and add the variant readings |
104 | # to the graph. |
105 | |
106 | foreach my $app_id ( keys %apps ) { |
107 | _add_readings( $c, $app_id ); |
6f4946fb |
108 | } |
4d85a60e |
109 | |
f6e19c7c |
110 | # Finally, add explicit witness paths, remove the base paths, and remove |
111 | # the app/anchor tags. |
112 | expand_all_paths( $c ); |
6f4946fb |
113 | } |
114 | |
4d85a60e |
115 | sub _stringify_sigil { |
116 | my( @nodes ) = @_; |
117 | my @parts = grep { /\w/ } map { $_->data } @nodes; |
118 | return join( '', @parts ); |
119 | } |
6f4946fb |
120 | |
121 | ## Recursive little helper function to help us navigate through nested |
4d85a60e |
122 | ## XML, picking out the words, the apparatus, and the anchors. |
123 | |
124 | sub _get_base { |
125 | my( $xn ) = @_; |
126 | my @readings; |
127 | if( $xn->nodeType == XML_TEXT_NODE ) { |
128 | # Base text, just split the words on whitespace and add them |
129 | # to our sequence. |
f6e19c7c |
130 | # TODO consider that XML markup might appear mid-token. |
4d85a60e |
131 | my $str = $xn->data; |
132 | $str =~ s/^\s+//; |
133 | foreach my $w ( split( /\s+/, $str ) ) { |
134 | push( @readings, { 'type' => 'token', 'content' => $w } ); |
135 | } |
136 | } elsif( $xn->nodeName eq 'hi' ) { |
137 | # Recurse as if the hi weren't there. |
138 | foreach( $xn->childNodes ) { |
139 | push( @readings, _get_base( $_ ) ); |
140 | } |
141 | } elsif( $xn->nodeName eq 'app' ) { |
142 | # Apparatus, just save the entire XML node. |
143 | push( @readings, { 'type' => 'app', 'content' => $xn } ); |
144 | } elsif( $xn->nodeName eq 'anchor' ) { |
145 | # Anchor to mark the end of some apparatus; save its ID. |
146 | push( @readings, { 'type' => 'anchor', |
147 | 'content' => $xn->getAttribute( 'xml:id' ) } ); |
148 | } elsif ( $xn->nodeName ne 'note' ) { # Any tag we don't know to disregard |
149 | print STDERR "Unrecognized tag " . $xn->nodeName . "\n"; |
6f4946fb |
150 | } |
4d85a60e |
151 | return @readings; |
6f4946fb |
152 | } |
153 | |
4d85a60e |
154 | sub _add_readings { |
155 | my( $c, $app_id ) = @_; |
156 | my $xn = $apps{$app_id}; |
157 | my $anchor = _anchor_name( $xn->getAttribute( 'to' ) ); |
158 | # Get the lemma, which is all the readings between app and anchor, |
159 | # excluding other apps or anchors. |
160 | my @lemma = _return_lemma( $c, $app_id, $anchor ); |
161 | my $lemma_str = join( ' ', grep { $_ !~ /^\#/ } map { $_->text } @lemma ); |
162 | |
163 | # For each reading, send its text to 'interpret' along with the lemma, |
164 | # and then save the list of witnesses that these tokens belong to. |
3a5d151b |
165 | my %wit_rdgs; # Maps from witnesses to the variant text |
4d85a60e |
166 | my $ctr = 0; |
167 | my $tag = $app_id; |
168 | $tag =~ s/^\#APP_(.*)\#$/$1/; |
169 | foreach my $rdg ( $xn->getChildrenByTagName( 'rdg' ) ) { |
170 | my @text; |
4d85a60e |
171 | foreach ( $rdg->childNodes ) { |
172 | push( @text, _get_base( $_ ) ); |
173 | } |
12720144 |
174 | my( $interpreted, $flag ) = ( '', undef ); |
175 | if( @text ) { |
176 | ( $interpreted, $flag ) = interpret( |
177 | join( ' ', map { $_->{'content'} } @text ), $lemma_str ); |
178 | } |
b8f262e8 |
179 | next if( $interpreted eq $lemma_str ) && !$flag; # Reading is lemma. |
180 | |
4d85a60e |
181 | my @rdg_nodes; |
b8f262e8 |
182 | if( $interpreted eq '#LACUNA#' ) { |
183 | push( @rdg_nodes, $c->add_reading( { id => $tag . "/" . $ctr++, |
184 | is_lacuna => 1 } ) ); |
185 | } else { |
186 | foreach my $w ( split( /\s+/, $interpreted ) ) { |
187 | my $r = $c->add_reading( { id => $tag . "/" . $ctr++, |
188 | text => $w } ); |
189 | push( @rdg_nodes, $r ); |
190 | } |
4d85a60e |
191 | } |
f6e19c7c |
192 | # For each listed wit, save the reading. |
193 | foreach my $wit ( split( /\s+/, $rdg->getAttribute( 'wit' ) ) ) { |
12720144 |
194 | $wit .= $flag if $flag; |
f6e19c7c |
195 | $wit_rdgs{$wit} = \@rdg_nodes; |
196 | } |
12720144 |
197 | |
3a5d151b |
198 | # Does the reading have an ID? If so it probably has a witDetail |
f6e19c7c |
199 | # attached, and we need to read it. |
3a5d151b |
200 | if( $rdg->hasAttribute( 'xml:id' ) ) { |
12720144 |
201 | warn "Witdetail on meta reading" if $flag; # this could get complicated. |
f6e19c7c |
202 | my $rid = $rdg->getAttribute( 'xml:id' ); |
203 | my $xpc = XML::LibXML::XPathContext->new( $xn ); |
204 | my @details = $xpc->findnodes( './witDetail[@target="'.$rid.'"]' ); |
205 | foreach my $d ( @details ) { |
206 | _parse_wit_detail( $d, \%wit_rdgs, \@lemma ); |
207 | } |
3a5d151b |
208 | } |
f6e19c7c |
209 | } |
210 | |
4d85a60e |
211 | # Now collate the variant readings, since it is not done for us. |
12720144 |
212 | collate_variants( $c, \@lemma, values %wit_rdgs ); |
b8f262e8 |
213 | |
4d85a60e |
214 | # Now add the witness paths for each reading. |
f6e19c7c |
215 | foreach my $wit_id ( keys %wit_rdgs ) { |
216 | my $witstr = get_sigil( $wit_id, $c ); |
217 | my $rdg_list = $wit_rdgs{$wit_id}; |
218 | _add_wit_path( $c, $rdg_list, $app_id, $anchor, $witstr ); |
6f4946fb |
219 | } |
4d85a60e |
220 | } |
6f4946fb |
221 | |
4d85a60e |
222 | sub _anchor_name { |
223 | my $xmlid = shift; |
224 | $xmlid =~ s/^\#//; |
225 | return sprintf( "#ANCHOR_%s#", $xmlid ); |
6f4946fb |
226 | } |
227 | |
4d85a60e |
228 | sub _return_lemma { |
229 | my( $c, $app, $anchor ) = @_; |
12720144 |
230 | my @nodes = grep { $_->id !~ /^\#A(PP|NCHOR)/ } |
231 | $c->reading_sequence( $c->reading( $app ), $c->reading( $anchor ), |
232 | $c->baselabel ); |
4d85a60e |
233 | return @nodes; |
234 | } |
6f4946fb |
235 | |
236 | sub interpret { |
4d85a60e |
237 | # A utility function to change apparatus-ese into a full variant. |
238 | my( $reading, $lemma ) = @_; |
239 | return $reading if $reading eq $lemma; |
240 | my $oldreading = $reading; |
241 | # $lemma =~ s/\s+[[:punct:]]+$//; |
242 | # $reading =~ s/\s*\(?sic([\s\w.]+)?\)?$//; |
12720144 |
243 | my $flag; # In case of p.c. indications |
4d85a60e |
244 | my @words = split( /\s+/, $lemma ); |
245 | if( $reading =~ /^(.*) praem.$/ ) { |
246 | $reading = "$1 $lemma"; |
247 | } elsif( $reading =~ /^(.*) add.$/ ) { |
248 | $reading = "$lemma $1"; |
b8f262e8 |
249 | } elsif( $reading =~ /add. alia manu/ |
250 | || $reading =~ /inscriptionem compegi e/ # TODO huh? |
251 | || $reading eq 'inc.' # TODO huh? |
252 | ) { |
12720144 |
253 | # Ignore it. |
254 | $reading = $lemma; |
b8f262e8 |
255 | } elsif( $reading =~ /locus [uv]acuus/ |
256 | || $reading eq 'def.' |
f6e19c7c |
257 | ) { |
b8f262e8 |
258 | $reading = '#LACUNA#'; |
259 | } elsif( $reading eq 'om.' ) { |
4d85a60e |
260 | $reading = ''; |
12720144 |
261 | } elsif( $reading =~ /^in[uv]\.$/ ) { |
4d85a60e |
262 | # Hope it is two words. |
263 | print STDERR "WARNING: want to invert a lemma that is not two words\n" |
264 | unless scalar( @words ) == 2; |
265 | $reading = join( ' ', reverse( @words ) ); |
12720144 |
266 | } elsif( $reading =~ /^iter(\.|at)$/ ) { |
4d85a60e |
267 | # Repeat the lemma |
268 | $reading = "$lemma $lemma"; |
12720144 |
269 | } elsif( $reading eq 'in marg.' ) { |
270 | # There was nothing before a correction. |
271 | $reading = ''; |
272 | $flag = '_ac'; |
273 | } elsif( $reading =~ /^(.*) \.\.\. (.*)$/ ) { |
4d85a60e |
274 | # The first and last N words captured should replace the first and |
275 | # last N words of the lemma. |
276 | my @begin = split( /\s+/, $1 ); |
277 | my @end = split( /\s+/, $2 ); |
278 | if( scalar( @begin ) + scalar ( @end ) > scalar( @words ) ) { |
279 | # Something is wrong and we can't do the splice. |
280 | print STDERR "ERROR: $lemma is too short to accommodate $oldreading\n"; |
281 | } else { |
282 | splice( @words, 0, scalar @begin, @begin ); |
283 | splice( @words, -(scalar @end), scalar @end, @end ); |
284 | $reading = join( ' ', @words ); |
285 | } |
6f4946fb |
286 | } |
12720144 |
287 | if( $oldreading ne $reading || $flag || $oldreading =~ /\./ ) { |
288 | my $int = $reading; |
289 | $int .= " ($flag)" if $flag; |
290 | print STDERR "Interpreted $oldreading as $int given $lemma\n"; |
291 | } |
292 | return( $reading, $flag ); |
4d85a60e |
293 | } |
294 | |
f6e19c7c |
295 | sub _parse_wit_detail { |
296 | my( $detail, $readings, $lemma ) = @_; |
297 | my $wit = $detail->getAttribute( 'wit' ); |
298 | my $content = $detail->textContent; |
299 | if( $content =~ /a\.\s*c\./ ) { |
300 | # Replace the key in the $readings hash |
301 | my $rdg = delete $readings->{$wit}; |
302 | $readings->{$wit.'_ac'} = $rdg; |
303 | $has_ac{$sigil_for{$wit}} = 1; |
304 | } elsif( $content =~ /p\.\s*c\./ ) { |
305 | # If no key for the wit a.c. exists, add one pointing to the lemma |
306 | unless( exists $readings->{$wit.'_ac'} ) { |
307 | $readings->{$wit.'_ac'} = $lemma; |
308 | } |
309 | $has_ac{$sigil_for{$wit}} = 1; |
310 | } # else don't bother just yet |
311 | } |
312 | |
313 | sub get_sigil { |
314 | my( $xml_id, $c ) = @_; |
315 | if( $xml_id =~ /^(.*)_ac$/ ) { |
316 | my $real_id = $1; |
317 | return $sigil_for{$real_id} . $c->ac_label; |
318 | } else { |
319 | return $sigil_for{$xml_id}; |
320 | } |
321 | } |
322 | |
323 | sub expand_all_paths { |
324 | my( $c ) = @_; |
325 | |
326 | # Walk the collation and fish out the paths for each witness |
327 | foreach my $wit ( $c->tradition->witnesses ) { |
328 | my $sig = $wit->sigil; |
12720144 |
329 | my @path = grep { !$_->is_ph } |
f6e19c7c |
330 | $c->reading_sequence( $c->start, $c->end, $sig ); |
331 | $wit->path( \@path ); |
332 | if( $has_ac{$sig} ) { |
12720144 |
333 | my @ac_path = grep { !$_->is_ph } |
f6e19c7c |
334 | $c->reading_sequence( $c->start, $c->end, $sig.$c->ac_label, $sig ); |
335 | $wit->uncorrected_path( \@ac_path ); |
336 | } |
337 | } |
338 | |
339 | # Delete the anchors |
12720144 |
340 | foreach my $anchor ( grep { $_->is_ph } $c->readings ) { |
f6e19c7c |
341 | $c->del_reading( $anchor ); |
342 | } |
12720144 |
343 | # Delete the base edges |
344 | map { $c->del_path( $_, $c->baselabel ) } $c->paths; |
f6e19c7c |
345 | |
346 | # Make the path edges |
347 | $c->make_witness_paths(); |
4d85a60e |
348 | } |
349 | |
350 | sub _add_wit_path { |
f6e19c7c |
351 | my( $c, $rdg, $app, $anchor, $wit ) = @_; |
4d85a60e |
352 | my @nodes = @$rdg; |
f6e19c7c |
353 | push( @nodes, $c->reading( $anchor ) ); |
4d85a60e |
354 | |
f6e19c7c |
355 | my $cur = $c->reading( $app ); |
4d85a60e |
356 | foreach my $n ( @nodes ) { |
f6e19c7c |
357 | $c->add_path( $cur, $n, $wit ); |
4d85a60e |
358 | $cur = $n; |
6f4946fb |
359 | } |
6f4946fb |
360 | } |
361 | |
362 | =back |
363 | |
364 | =head1 LICENSE |
365 | |
366 | This package is free software and is provided "as is" without express |
367 | or implied warranty. You can redistribute it and/or modify it under |
368 | the same terms as Perl itself. |
369 | |
370 | =head1 AUTHOR |
371 | |
372 | Tara L Andrews, aurum@cpan.org |
373 | |
374 | =cut |
375 | |
376 | 1; |
377 | |