allow quick and dirty reading merge by relationship type in tab output
[scpubgit/stemmatology.git] / base / t / text_tradition_collation.t
CommitLineData
0e47f4f6 1#!/usr/bin/perl -w
2
3use strict;
4use Test::More 'no_plan';
5$| = 1;
6
7
8
9# =begin testing
10{
11use Text::Tradition;
58568d5c 12use TryCatch;
0e47f4f6 13
4e483aa5 14my $cxfile = 't/data/Collatex-16.xml';
15my $t = Text::Tradition->new(
16 'name' => 'inline',
17 'input' => 'CollateX',
18 'file' => $cxfile,
19 );
20my $c = $t->collation;
21
22my $rno = scalar $c->readings;
3c234eb6 23# Split n21 ('unto') for testing purposes
4e483aa5 24my $new_r = $c->add_reading( { 'id' => 'n21p0', 'text' => 'un', 'join_next' => 1 } );
25my $old_r = $c->reading( 'n21' );
26$old_r->alter_text( 'to' );
27$c->del_path( 'n20', 'n21', 'A' );
28$c->add_path( 'n20', 'n21p0', 'A' );
29$c->add_path( 'n21p0', 'n21', 'A' );
7a0956c1 30$c->add_relationship( 'n21', 'n22', { type => 'collated', scope => 'local' } );
4e483aa5 31$c->flatten_ranks();
32ok( $c->reading( 'n21p0' ), "New reading exists" );
33is( scalar $c->readings, $rno, "Reading add offset by flatten_ranks" );
34
679f17e1 35# Combine n3 and n4 ( with his )
4e483aa5 36$c->merge_readings( 'n3', 'n4', 1 );
37ok( !$c->reading('n4'), "Reading n4 is gone" );
38is( $c->reading('n3')->text, 'with his', "Reading n3 has both words" );
39
679f17e1 40# Collapse n9 and n10 ( rood / root )
41$c->merge_readings( 'n9', 'n10' );
42ok( !$c->reading('n10'), "Reading n10 is gone" );
43is( $c->reading('n9')->text, 'rood', "Reading n9 has an unchanged word" );
4e483aa5 44
58568d5c 45# Try to combine n21 and n21p0. This should break.
4e483aa5 46my $remaining = $c->reading('n21');
47$remaining ||= $c->reading('n22'); # one of these should still exist
58568d5c 48try {
49 $c->merge_readings( 'n21p0', $remaining, 1 );
50 ok( 0, "Bad reading merge changed the graph" );
51} catch( Text::Tradition::Error $e ) {
52 like( $e->message, qr/neither concatenated nor collated/, "Expected exception from bad concatenation" );
53} catch {
54 ok( 0, "Unexpected error on bad reading merge: $@" );
55}
56
57try {
58 $c->calculate_ranks();
59 ok( 1, "Graph is still evidently whole" );
60} catch( Text::Tradition::Error $e ) {
61 ok( 0, "Caught a rank exception: " . $e->message );
62}
4e483aa5 63}
64
65
66
67# =begin testing
68{
58568d5c 69use Test::Warn;
8d9494a8 70use Text::Tradition;
71use TryCatch;
72
58568d5c 73my $t;
74warnings_exist {
75 $t = Text::Tradition->new( 'input' => 'Self', 'file' => 't/data/legendfrag.xml' );
76} [qr/Cannot set relationship on a meta reading/],
77 "Got expected relationship drop warning on parse";
78
8d9494a8 79my $c = $t->collation;
80
81my %rdg_ids;
82map { $rdg_ids{$_} = 1 } $c->readings;
83$c->merge_related( 'orthographic' );
751ee528 84is( scalar( $c->readings ), keys( %rdg_ids ) - 9,
8d9494a8 85 "Successfully collapsed orthographic variation" );
751ee528 86map { $rdg_ids{$_} = undef } qw/ r13.3 r11.4 r8.5 r8.2 r7.7 r7.5 r7.4 r7.3 r7.1 /;
8d9494a8 87foreach my $rid ( keys %rdg_ids ) {
88 my $exp = $rdg_ids{$rid};
89 is( !$c->reading( $rid ), !$exp, "Reading $rid correctly " .
90 ( $exp ? "retained" : "removed" ) );
91}
92ok( $c->linear, "Graph is still linear" );
93try {
94 $c->calculate_ranks; # This should succeed
95 ok( 1, "Can still calculate ranks on the new graph" );
96} catch {
97 ok( 0, "Rank calculation on merged graph failed: $@" );
98}
99
100# Now add some transpositions
101$c->add_relationship( 'r8.4', 'r10.4', { type => 'transposition' } );
102$c->merge_related( 'transposition' );
751ee528 103is( scalar( $c->readings ), keys( %rdg_ids ) - 10,
8d9494a8 104 "Transposed relationship is merged away" );
105ok( !$c->reading('r8.4'), "Correct transposed reading removed" );
106ok( !$c->linear, "Graph is no longer linear" );
107try {
108 $c->calculate_ranks; # This should fail
109 ok( 0, "Rank calculation happened on nonlinear graph?!" );
110} catch ( Text::Tradition::Error $e ) {
111 is( $e->message, 'Cannot calculate ranks on a non-linear graph',
112 "Rank calculation on merged graph threw an error" );
113}
114}
115
116
117
118# =begin testing
119{
68e48c06 120use Test::More::UTF8;
4e483aa5 121use Text::Tradition;
68e48c06 122use TryCatch;
f97ef19e 123
124my $st = Text::Tradition->new( 'input' => 'Self', 'file' => 't/data/collatecorr.xml' );
125is( ref( $st ), 'Text::Tradition', "Got a tradition from test file" );
126ok( $st->has_witness('Ba96'), "Tradition has the affected witness" );
127
128my $sc = $st->collation;
129my $numr = 17;
130ok( $sc->reading('n131'), "Tradition has the affected reading" );
131is( scalar( $sc->readings ), $numr, "There are $numr readings in the graph" );
132is( $sc->end->rank, 14, "There are fourteen ranks in the graph" );
133
134# Detach the erroneously collated reading
2dcb5d11 135my( $newr, @del_rdgs ) = $sc->duplicate_reading( 'n131', 'Ba96' );
ef73c20a 136ok( $newr, "New reading was created" );
f97ef19e 137ok( $sc->reading('n131_0'), "Detached the bad collation with a new reading" );
138is( scalar( $sc->readings ), $numr + 1, "A reading was added to the graph" );
139is( $sc->end->rank, 10, "There are now only ten ranks in the graph" );
3c234eb6 140my $csucc = $sc->common_successor( 'n131', 'n131_0' );
141is( $csucc->id, 'n136', "Found correct common successor to duped reading" );
f97ef19e 142
143# Check that the bad transposition is gone
2dcb5d11 144is( scalar @del_rdgs, 1, "Deleted reading was returned by API call" );
f97ef19e 145is( $sc->get_relationship( 'n130', 'n135' ), undef, "Bad transposition relationship is gone" );
146
e19635f8 147# The collation should not be fixed
148my @pairs = $sc->identical_readings();
149is( scalar @pairs, 0, "Not re-collated yet" );
f97ef19e 150# Fix the collation
3c234eb6 151ok( $sc->merge_readings( 'n124', 'n131_0' ), "Collated the readings correctly" );
e19635f8 152@pairs = $sc->identical_readings( start => 'n124', end => $csucc->id );
3c234eb6 153is( scalar @pairs, 3, "Found three more identical readings" );
e19635f8 154is( $sc->end->rank, 11, "The ranks shifted appropriately" );
3c234eb6 155$sc->flatten_ranks();
f97ef19e 156is( scalar( $sc->readings ), $numr - 3, "Now we are collated correctly" );
68e48c06 157
158# Check that we can't "duplicate" a reading with no wits or with all wits
159try {
160 my( $badr, @del_rdgs ) = $sc->duplicate_reading( 'n124' );
161 ok( 0, "Reading duplication without witnesses throws an error" );
162} catch( Text::Tradition::Error $e ) {
163 like( $e->message, qr/Must specify one or more witnesses/,
164 "Reading duplication without witnesses throws the expected error" );
165} catch {
166 ok( 0, "Reading duplication without witnesses threw the wrong error" );
167}
168
169try {
170 my( $badr, @del_rdgs ) = $sc->duplicate_reading( 'n124', 'Ba96', 'Mü11475' );
171 ok( 0, "Reading duplication with all witnesses throws an error" );
172} catch( Text::Tradition::Error $e ) {
173 like( $e->message, qr/Cannot join all witnesses/,
174 "Reading duplication with all witnesses throws the expected error" );
175} catch {
176 ok( 0, "Reading duplication with all witnesses threw the wrong error" );
177}
58568d5c 178
179try {
180 $sc->calculate_ranks();
181 ok( 1, "Graph is still evidently whole" );
182} catch( Text::Tradition::Error $e ) {
183 ok( 0, "Caught a rank exception: " . $e->message );
184}
f97ef19e 185}
186
187
188
189# =begin testing
190{
191use Text::Tradition;
951ddfe8 192use TryCatch;
4e483aa5 193
56eefa04 194my $READINGS = 311;
195my $PATHS = 361;
196
197my $datafile = 't/data/florilegium_tei_ps.xml';
198my $tradition = Text::Tradition->new( 'input' => 'TEI',
199 'name' => 'test0',
200 'file' => $datafile,
201 'linear' => 1 );
202
203ok( $tradition, "Got a tradition object" );
204is( scalar $tradition->witnesses, 13, "Found all witnesses" );
205ok( $tradition->collation, "Tradition has a collation" );
206
207my $c = $tradition->collation;
208is( scalar $c->readings, $READINGS, "Collation has all readings" );
209is( scalar $c->paths, $PATHS, "Collation has all paths" );
210is( scalar $c->relationships, 0, "Collation has all relationships" );
211
212# Add a few relationships
213$c->add_relationship( 'w123', 'w125', { 'type' => 'collated' } );
214$c->add_relationship( 'w193', 'w196', { 'type' => 'collated' } );
b71e7ea8 215$c->add_relationship( 'w257', 'w262', { 'type' => 'transposition',
216 'is_significant' => 'yes' } );
56eefa04 217
218# Now write it to GraphML and parse it again.
219
220my $graphml = $c->as_graphml;
221my $st = Text::Tradition->new( 'input' => 'Self', 'string' => $graphml );
222is( scalar $st->collation->readings, $READINGS, "Reparsed collation has all readings" );
223is( scalar $st->collation->paths, $PATHS, "Reparsed collation has all paths" );
224is( scalar $st->collation->relationships, 3, "Reparsed collation has new relationships" );
b71e7ea8 225my $sigrel = $st->collation->get_relationship( 'w257', 'w262' );
226is( $sigrel->is_significant, 'yes', "Ternary attribute value was restored" );
2a812726 227
9fef629b 228# Now add a stemma, write to GraphML, and look at the output.
951ddfe8 229SKIP: {
37bf09f4 230 skip "Analysis module not present", 3 unless $tradition->can( 'add_stemma' );
951ddfe8 231 my $stemma = $tradition->add_stemma( 'dotfile' => 't/data/florilegium.dot' );
232 is( ref( $stemma ), 'Text::Tradition::Stemma', "Parsed dotfile into stemma" );
233 is( $tradition->stemmata, 1, "Tradition now has the stemma" );
234 $graphml = $c->as_graphml;
235 like( $graphml, qr/digraph/, "Digraph declaration exists in GraphML" );
236}
56eefa04 237}
238
239
240
241# =begin testing
242{
16203db5 243use Text::Tradition;
34ca808b 244use Text::CSV;
16203db5 245
246my $READINGS = 311;
247my $PATHS = 361;
34ca808b 248my $WITS = 13;
249my $WITAC = 4;
16203db5 250
251my $datafile = 't/data/florilegium_tei_ps.xml';
252my $tradition = Text::Tradition->new( 'input' => 'TEI',
253 'name' => 'test0',
254 'file' => $datafile,
255 'linear' => 1 );
256
257my $c = $tradition->collation;
258# Export the thing to CSV
259my $csvstr = $c->as_csv();
34ca808b 260# Count the columns
261my $csv = Text::CSV->new({ sep_char => ',', binary => 1 });
262my @lines = split(/\n/, $csvstr );
263ok( $csv->parse( $lines[0] ), "Successfully parsed first line of CSV" );
264is( scalar( $csv->fields ), $WITS + $WITAC, "CSV has correct number of witness columns" );
cbc8e08f 265my @q_ac = grep { $_ eq 'Q'.$c->ac_label } $csv->fields;
266ok( @q_ac, "Found a layered witness" );
267
16203db5 268my $t2 = Text::Tradition->new( input => 'Tabular',
269 name => 'test2',
270 string => $csvstr,
271 sep_char => ',' );
272is( scalar $t2->collation->readings, $READINGS, "Reparsed CSV collation has all readings" );
273is( scalar $t2->collation->paths, $PATHS, "Reparsed CSV collation has all paths" );
274
275# Now do it with TSV
276my $tsvstr = $c->as_tsv();
277my $t3 = Text::Tradition->new( input => 'Tabular',
278 name => 'test3',
279 string => $tsvstr,
280 sep_char => "\t" );
281is( scalar $t3->collation->readings, $READINGS, "Reparsed TSV collation has all readings" );
282is( scalar $t3->collation->paths, $PATHS, "Reparsed TSV collation has all paths" );
34ca808b 283
4e64b669 284my $table = $c->alignment_table;
34ca808b 285my $noaccsv = $c->as_csv({ noac => 1 });
286my @noaclines = split(/\n/, $noaccsv );
287ok( $csv->parse( $noaclines[0] ), "Successfully parsed first line of no-ac CSV" );
288is( scalar( $csv->fields ), $WITS, "CSV has correct number of witness columns" );
4e64b669 289is( $c->alignment_table, $table, "Request for CSV did not alter the alignment table" );
cbc8e08f 290
291my $safecsv = $c->as_csv({ safe_ac => 1});
292my @safelines = split(/\n/, $safecsv );
293ok( $csv->parse( $safelines[0] ), "Successfully parsed first line of safe CSV" );
294is( scalar( $csv->fields ), $WITS + $WITAC, "CSV has correct number of witness columns" );
295@q_ac = grep { $_ eq 'Q__L' } $csv->fields;
296ok( @q_ac, "Found a sanitized layered witness" );
297is( $c->alignment_table, $table, "Request for CSV did not alter the alignment table" );
de20588d 298
299# Test relationship collapse
300$c->add_relationship( $c->readings_at_rank( 37 ), { type => 'spelling' } );
301$c->add_relationship( $c->readings_at_rank( 60 ), { type => 'spelling' } );
302
303my $mergedtsv = $c->as_tsv({mergetypes => [ 'spelling', 'orthographic' ] });
304my $t4 = Text::Tradition->new( input => 'Tabular',
305 name => 'test4',
306 string => $mergedtsv,
307 sep_char => "\t" );
308is( scalar $t4->collation->readings, $READINGS - 2, "Reparsed TSV merge collation has fewer readings" );
309is( scalar $t4->collation->paths, $PATHS - 4, "Reparsed TSV merge collation has fewer paths" );
16203db5 310}
311
312
313
314# =begin testing
315{
56eefa04 316use Text::Tradition;
317
0e47f4f6 318my $cxfile = 't/data/Collatex-16.xml';
319my $t = Text::Tradition->new(
320 'name' => 'inline',
321 'input' => 'CollateX',
322 'file' => $cxfile,
323 );
324my $c = $t->collation;
4633f9e4 325
b365fbae 326# Make an svg
bfcbcecb 327my $table = $c->alignment_table;
328ok( $c->has_cached_table, "Alignment table was cached" );
329is( $c->alignment_table, $table, "Cached table returned upon second call" );
b365fbae 330$c->calculate_ranks;
bfcbcecb 331is( $c->alignment_table, $table, "Cached table retained with no rank change" );
864ee4bf 332$c->add_relationship( 'n13', 'n23', { type => 'repetition' } );
333is( $c->alignment_table, $table, "Alignment table unchanged after non-colo relationship add" );
334$c->add_relationship( 'n24', 'n23', { type => 'spelling' } );
335isnt( $c->alignment_table, $table, "Alignment table changed after colo relationship add" );
b365fbae 336}
337
338
339
340# =begin testing
341{
342use Text::Tradition;
343
344my $cxfile = 't/data/Collatex-16.xml';
345my $t = Text::Tradition->new(
346 'name' => 'inline',
347 'input' => 'CollateX',
348 'file' => $cxfile,
349 );
350my $c = $t->collation;
0e47f4f6 351
d4b75f44 352my @common = $c->calculate_common_readings();
353is( scalar @common, 8, "Found correct number of common readings" );
354my @marked = sort $c->common_readings();
355is( scalar @common, 8, "All common readings got marked as such" );
679f17e1 356my @expected = qw/ n1 n11 n16 n19 n20 n5 n6 n7 /;
d4b75f44 357is_deeply( \@marked, \@expected, "Found correct list of common readings" );
358}
359
360
361
362# =begin testing
363{
364use Text::Tradition;
365
366my $cxfile = 't/data/Collatex-16.xml';
367my $t = Text::Tradition->new(
368 'name' => 'inline',
369 'input' => 'CollateX',
370 'file' => $cxfile,
371 );
372my $c = $t->collation;
373
679f17e1 374is( $c->common_predecessor( 'n24', 'n23' )->id,
0e47f4f6 375 'n20', "Found correct common predecessor" );
679f17e1 376is( $c->common_successor( 'n24', 'n23' )->id,
10e4b1ac 377 '__END__', "Found correct common successor" );
0e47f4f6 378
4e5a7b2c 379is( $c->common_predecessor( 'n19', 'n17' )->id,
0e47f4f6 380 'n16', "Found correct common predecessor for readings on same path" );
679f17e1 381is( $c->common_successor( 'n21', 'n10' )->id,
10e4b1ac 382 '__END__', "Found correct common successor for readings on same path" );
0e47f4f6 383}
384
385
386
387
3881;