calculate common readings when we parse
[scpubgit/stemmatology.git] / t / graph.t
CommitLineData
b49c4318 1#!/usr/bin/perl
2
3use strict; use warnings;
b22576c6 4use Test::More;
b49c4318 5use lib 'lib';
8e1394aa 6use Text::Tradition;
b49c4318 7use XML::LibXML;
8use XML::LibXML::XPathContext;
9
10my $datafile = 't/data/Collatex-16.xml';
11
7035e3a6 12my $tradition = Text::Tradition->new(
13 'name' => 'inline',
14 'input' => 'CollateX',
15 'file' => $datafile,
16 );
8e1394aa 17my $collation = $tradition->collation;
b49c4318 18
19# Test the svg creation
20my $parser = XML::LibXML->new();
21$parser->load_ext_dtd( 0 );
8e1394aa 22my $svg = $parser->parse_string( $collation->as_svg() );
b49c4318 23is( $svg->documentElement->nodeName(), 'svg', 'Got an svg document' );
24
25# Test for the correct number of nodes in the SVG
26my $svg_xpc = XML::LibXML::XPathContext->new( $svg->documentElement() );
27$svg_xpc->registerNs( 'svg', 'http://www.w3.org/2000/svg' );
28my @svg_nodes = $svg_xpc->findnodes( '//svg:g[@class="node"]' );
3265b0ce 29is( scalar @svg_nodes, 26, "Correct number of nodes in the graph" );
b49c4318 30
31# Test for the correct number of edges
32my @svg_edges = $svg_xpc->findnodes( '//svg:g[@class="edge"]' );
3265b0ce 33is( scalar @svg_edges, 32, "Correct number of edges in the graph" );
b49c4318 34
b22576c6 35# Test svg creation for a subgraph
36my $part_svg = $parser->parse_string( $collation->svg_subgraph( 15 ) ); # start, no end
37is( $part_svg->documentElement->nodeName(), 'svg', "Got an svg subgraph to end" );
38my $part_xpc = XML::LibXML::XPathContext->new( $part_svg->documentElement() );
39$part_xpc->registerNs( 'svg', 'http://www.w3.org/2000/svg' );
40@svg_nodes = $part_xpc->findnodes( '//svg:g[@class="node"]' );
41is( scalar( @svg_nodes ), 9,
42 "Correct number of nodes in the subgraph" );
43@svg_edges = $part_xpc->findnodes( '//svg:g[@class="edge"]' );
44is( scalar( @svg_edges ), 10,
45 "Correct number of edges in the subgraph" );
46open( OUT, ">test.svg" );
47print OUT $part_svg->toString();
48close OUT;
49
50$part_svg = $parser->parse_string( $collation->svg_subgraph( 10, 13 ) ); # start, no end
51is( $part_svg->documentElement->nodeName(), 'svg', "Got an svg subgraph in the middle" );
52$part_xpc = XML::LibXML::XPathContext->new( $part_svg->documentElement() );
53$part_xpc->registerNs( 'svg', 'http://www.w3.org/2000/svg' );
54@svg_nodes = $part_xpc->findnodes( '//svg:g[@class="node"]' );
55is( scalar( @svg_nodes ), 9,
56 "Correct number of nodes in the subgraph" );
57@svg_edges = $part_xpc->findnodes( '//svg:g[@class="edge"]' );
58is( scalar( @svg_edges ), 11,
59 "Correct number of edges in the subgraph" );
60
61
62$part_svg = $parser->parse_string( $collation->svg_subgraph( 0, 5 ) ); # start, no end
63is( $part_svg->documentElement->nodeName(), 'svg', "Got an svg subgraph from start" );
64$part_xpc = XML::LibXML::XPathContext->new( $part_svg->documentElement() );
65$part_xpc->registerNs( 'svg', 'http://www.w3.org/2000/svg' );
66@svg_nodes = $part_xpc->findnodes( '//svg:g[@class="node"]' );
67is( scalar( @svg_nodes ), 7,
68 "Correct number of nodes in the subgraph" );
69@svg_edges = $part_xpc->findnodes( '//svg:g[@class="edge"]' );
70is( scalar( @svg_edges ), 7,
71 "Correct number of edges in the subgraph" );
72
73SKIP: {
74 skip "lemmatization disabled for now", 1;
75 # Test for the correct common nodes
76 my @common_nodes = ( '#START#' );
77 push( @common_nodes, qw/ n1 n5 n6 n7 n12 n16 n19 n20 n27 / );
78 my @expected_nodes = map { [ $_, 1 ] } @common_nodes;
79 foreach my $idx ( qw/2 3 4 8 10 11 13 16 17 18/ ) {
80 splice( @expected_nodes, $idx, 0, [ "node_null", undef ] );
b49c4318 81 }
b22576c6 82 my @active_nodes = $collation->lemma_readings();
83 subtest 'Initial common points' => \&compare_active;
84 my $string = '# when ... ... ... showers sweet with ... fruit ... ... of ... has pierced ... ... ... #';
85 is( make_text( @active_nodes ), $string, "Got the right starting text" );
86
87 sub compare_active {
88 is( scalar( @active_nodes ), scalar ( @expected_nodes ),
89 "Arrays are same length" );
90
91 foreach ( 0 .. scalar(@active_nodes)-1 ) {
92 is( $active_nodes[$_]->[1], $expected_nodes[$_]->[1],
93 "Element has same toggle value" );
94 if( defined $active_nodes[$_]->[1] ) {
95 is( $active_nodes[$_]->[0], $expected_nodes[$_]->[0],
96 "Active or toggled element has same node name "
97 . $active_nodes[$_]->[0] );
98 }
99 }
b49c4318 100 }
b22576c6 101
102 sub make_text {
103 my @words;
104 foreach my $n ( @_ ) {
105 if( $n->[1] ) {
106 push( @words, $collation->reading( $n->[0] )->text );
107 } elsif ( !defined $n->[1] ) {
108 push( @words, '...' );
109 }
110 }
111 return join( ' ', @words );
112 }
113
114 # Test that the common nodes are marked common
115 foreach my $cn ( @common_nodes ) {
116 ok( $collation->reading( $cn )->is_common, "Node $cn is marked common" );
117 }
118
119 # Test the manuscript paths
120 my $wit_a = '# when april with his showers sweet with fruit the drought of march has pierced unto the root #';
121 my $wit_b = '# when showers sweet with april fruit the march of drought has pierced to the root #';
122 my $wit_c = '# when showers sweet with april fruit teh drought of march has pierced teh rood #';
123 is( join( ' ', @{$tradition->witness( "A" )->text} ), $wit_a, "Correct path for witness A" );
124 is( join( ' ', @{$tradition->witness( "B" )->text} ), $wit_b, "Correct path for witness B" );
125 is( join( ' ', @{$tradition->witness( "C" )->text} ), $wit_c, "Correct path for witness C" );
126
127 # Test the transposition identifiers
128 my $transposition_pools = [ [ 'n2', 'n11' ], [ 'n14', 'n18' ],
129 [ 'n17', 'n15' ] ];
130 my $transposed_nodes = { 'n2' => $transposition_pools->[0],
131 'n11' => $transposition_pools->[0],
132 'n14' => $transposition_pools->[1],
133 'n15' => $transposition_pools->[2],
134 'n17' => $transposition_pools->[2],
135 'n18' => $transposition_pools->[1],
136 };
137
138 my $real_transposed_nodes = {};
139 foreach my $r ( $collation->readings ) {
140 my @same = map { $_->name } @{$r->same_as};
141 $real_transposed_nodes->{ $r->name } = \@same if @same > 1;
142 }
143
144 is_deeply( $real_transposed_nodes, $transposed_nodes, "Found the right transpositions" );
145
146 # Test turning on a node
147 my @off = $collation->toggle_reading( 'n21' );
148 $expected_nodes[ 16 ] = [ "n21", 1 ];
149 @active_nodes = $collation->lemma_readings( @off );
150 subtest 'Turned on node for new location' => \&compare_active;
151 $string = '# when ... ... ... showers sweet with ... fruit ... ... of ... has pierced unto ... ... #';
152 is( make_text( @active_nodes ), $string, "Got the right text" );
153
154 # Test the toggling effects of same-column
155 @off = $collation->toggle_reading( 'n22' );
156 splice( @expected_nodes, 16, 1, ( [ "n21", 0 ], [ "n22", 1 ] ) );
157 @active_nodes = $collation->lemma_readings( @off );
158 subtest 'Turned on other node in that location' => \&compare_active;
159 $string = '# when ... ... ... showers sweet with ... fruit ... ... of ... has pierced to ... ... #';
160 is( make_text( @active_nodes ), $string, "Got the right text" );
161
162 # Test the toggling effects of transposition
163 @off = $collation->toggle_reading( 'n14' );
164 # Add the turned on node
165 $expected_nodes[ 11 ] = [ "n14", 1 ];
166 # Remove the 'off' for the previous node
167 splice( @expected_nodes, 16, 1 );
168 @active_nodes = $collation->lemma_readings( @off );
169 subtest 'Turned on transposition node' => \&compare_active;
170 $string = '# when ... ... ... showers sweet with ... fruit ... drought of ... has pierced to ... ... #';
171 is( make_text( @active_nodes ), $string, "Got the right text" );
172
173 @off = $collation->toggle_reading( 'n18' );
174 # Toggle on the new node
175 $expected_nodes[ 13 ] = [ "n18", 1 ];
176 # Toggle off the transposed node
177 $expected_nodes[ 11 ] = [ "n14", undef ];
178 @active_nodes = $collation->lemma_readings( @off );
179 subtest 'Turned on that node\'s partner' => \&compare_active;
180 $string = '# when ... ... ... showers sweet with ... fruit ... ... of drought has pierced to ... ... #';
181 is( make_text( @active_nodes ), $string, "Got the right text" );
182
183 @off = $collation->toggle_reading( 'n14' );
184 # Toggle on the new node
185 $expected_nodes[ 11 ] = [ "n14", 1 ];
186 # Toggle off the transposed node
187 $expected_nodes[ 13 ] = [ "n18", undef ];
188 @active_nodes = $collation->lemma_readings( @off );
189 subtest 'Turned on the original node' => \&compare_active;
190 $string = '# when ... ... ... showers sweet with ... fruit ... drought of ... has pierced to ... ... #';
191 is( make_text( @active_nodes ), $string, "Got the right text" );
192
193 @off = $collation->toggle_reading( 'n15' );
194 # Toggle on the new node, and off with the old
195 splice( @expected_nodes, 11, 1, [ "n14", 0 ], [ "n15", 1 ] );
196 @active_nodes = $collation->lemma_readings( @off );
197 subtest 'Turned on the colocated node' => \&compare_active;
198 $string = '# when ... ... ... showers sweet with ... fruit ... march of ... has pierced to ... ... #';
199 is( make_text( @active_nodes ), $string, "Got the right text" );
200
201 @off = $collation->toggle_reading( 'n3' );
202 # Toggle on the new node
203 splice( @expected_nodes, 3, 1, [ "n3", 1 ] );
204 # Remove the old toggle-off
205 splice( @expected_nodes, 11, 1 );
206 @active_nodes = $collation->lemma_readings( @off );
207 subtest 'Turned on a singleton node' => \&compare_active;
208 $string = '# when ... with ... showers sweet with ... fruit ... march of ... has pierced to ... ... #';
209 is( make_text( @active_nodes ), $string, "Got the right text" );
210
211 @off = $collation->toggle_reading( 'n3' );
212 # Toggle off this node
213 splice( @expected_nodes, 3, 1, [ "n3", 0 ] );
214 @active_nodes = $collation->lemma_readings( @off );
215 subtest 'Turned off a singleton node' => \&compare_active;
216 $string = '# when ... ... showers sweet with ... fruit ... march of ... has pierced to ... ... #';
217 is( make_text( @active_nodes ), $string, "Got the right text" );
218
219 @off = $collation->toggle_reading( 'n21' );
220 splice( @expected_nodes, 16, 1, ["n22", 0 ], [ "n21", 1 ] );
221 @active_nodes = $collation->lemma_readings( @off );
222 subtest 'Turned on another node after singleton switchoff' => \&compare_active;
223 $string = '# when ... ... showers sweet with ... fruit ... march of ... has pierced unto ... ... #';
224 is( make_text( @active_nodes ), $string, "Got the right text" );
225
226 # Now start testing some position identifiers
227 # 2. 'april with his' have no colocated
228 # 3. 'april' 2 has no colocated
229 # 4. 'teh' and 'the'
230 # 5. 'drought' & 'march'
231 # 6. 'march' & 'drought'
232 # 7. 'unto' 'the' 'root'...
233 # 'unto can match 'to' or 'teh'
234 # 'the' can match 'teh' or 'rood'
235 # 'root' can mach 'rood'
236
237 foreach my $cn ( @common_nodes ) {
238 my $cnr = $collation->reading( $cn );
239 is( scalar( $collation->same_position_as( $cnr ) ), 0, "Node $cn has no colocations" );
240 }
241
242 my %expected_colocations = (
243 'n2' => [], # april
244 'n3' => [], # with
245 'n4' => [], # his
246 'n11' => [], # april
247 'n8' => [ 'n13' ], # teh -> the
248 'n13' => [ 'n8' ], # the -> teh
249 'n14' => [ 'n15' ], # drought -> march
250 'n18' => [ 'n17' ], # drought -> march
251 'n17' => [ 'n18' ], # march -> drought
252 'n15' => [ 'n14' ], # march -> drought
253 'n21' => [ 'n22', 'n9' ], # unto -> to, teh
254 'n22' => [ 'n21', 'n9' ], # to -> unto, teh
255 'n9' => [ 'n21', 'n22', 'n23' ], # teh -> unto, to, the
256 'n23' => [ 'n25', 'n9' ], # the -> teh, rood
257 'n25' => [ 'n23', 'n26' ], # rood -> the, root
258 'n26' => [ 'n25' ], # root -> rood
259 );
260
261 foreach my $n ( keys %expected_colocations ) {
262 my $nr = $collation->reading( $n );
263 my @colocated = sort( map { $_->name } $collation->same_position_as( $nr ) );
264 is_deeply( \@colocated, $expected_colocations{$n}, "Colocated nodes for $n correct" );
265 }
266
267 # Test strict colocations
268 $expected_colocations{'n9'} = [];
269 $expected_colocations{'n21'} = ['n22'];
270 $expected_colocations{'n22'} = ['n21'];
271 $expected_colocations{'n23'} = [];
272 $expected_colocations{'n25'} = [];
273 $expected_colocations{'n26'} = [];
274
275 foreach my $n ( keys %expected_colocations ) {
276 my $nr = $collation->reading( $n );
277 my @colocated = sort( map { $_->name } $collation->same_position_as( $nr, 1 ) );
278 is_deeply( \@colocated, $expected_colocations{$n}, "Strictly colocated nodes for $n correct" );
279 }
280
281 # Test turning on, then off, an annoyingly overlapping node
282
283 @off = $collation->toggle_reading( 'n9' );
284 # Remove the old toggle-off
285 splice( @expected_nodes, 16, 1 );
286 splice( @expected_nodes, 17, 0, [ "n9", 1 ] );
287 @active_nodes = $collation->lemma_readings( @off );
288 subtest 'Turned on a node without fixed position' => \&compare_active;
289 $string = '# when ... ... showers sweet with ... fruit ... march of ... has pierced unto teh ... ... #';
290 is( make_text( @active_nodes ), $string, "Got the right text" );
291
292 @off = $collation->toggle_reading( 'n23' );
293 splice( @expected_nodes, 18, 1, [ "n23", 1 ] );
294 @active_nodes = $collation->lemma_readings( @off );
295 subtest 'Turned on a node colocated to one without fixed position' => \&compare_active;
296 $string = '# when ... ... showers sweet with ... fruit ... march of ... has pierced unto teh the ... #';
297 is( make_text( @active_nodes ), $string, "Got the right text" );
298
299 @off = $collation->toggle_reading( 'n9' );
300 splice( @expected_nodes, 17, 1, [ "n9", 0 ] );
301 @active_nodes = $collation->lemma_readings( @off );
302 subtest 'Turned on a node colocated to one without fixed position' => \&compare_active;
303 $string = '# when ... ... showers sweet with ... fruit ... march of ... has pierced unto the ... #';
304 is( make_text( @active_nodes ), $string, "Got the right text" );
305
306 ### Now test relationship madness.
307
308 my( $result, @relations ) = $collation->add_relationship( 'n25', 'n23', {'type' => 'lexical'} ); # rood -> the
309 ok( $result, "Added relationship between nodes" );
310 is( scalar @relations, 1, "Returned only the one collapse" );
311 is_deeply( $relations[0], [ 'n25', 'n23' ], "Returned the correct collapse" );
312 is( $collation->reading( 'n25' )->position->reference, '9,3', "Harmonized position for n25 correct" );
313 is( $collation->reading( 'n23' )->position->reference, '9,3', "Harmonized position for n23 correct" );
314 is( $collation->reading( 'n9' )->position->reference, '9,2', "Adjusted position for n9 correct" );
315
316 # Do some yucky hardcoded cleanup to undo this relationship.
317 $collation->reading('n25')->position->max( 4 );
318 $collation->reading('n9')->position->max( 3 );
319 $collation->graph->del_edge( $collation->reading('n25')->edges_to( $collation->reading('n23')) );
320
321 ( $result, @relations ) = $collation->add_relationship( 'n26', 'n25', {'type' => 'spelling'} ); # root -> rood
322 ok( $result, "Added relationship between nodes" );
323 is( scalar @relations, 1, "Returned only the one collapse" );
324 is_deeply( $relations[0], [ 'n26', 'n25' ], "Returned the correct collapse" );
325 is( $collation->reading( 'n26' )->position->reference, '9,4', "Harmonized position for n26 correct" );
326 is( $collation->reading( 'n25' )->position->reference, '9,4', "Harmonized position for n25 correct" );
327 is( $collation->reading( 'n9' )->position->reference, '9,2-3', "Adjusted position for n9 correct" );
328
329 ( $result, @relations ) = $collation->add_relationship( 'n15', 'n9', {'type' => 'lexical'} ); # bogus march -> teh
330 ok( !$result, "Refused to add skewed relationship: " . $relations[0] );
331
332 ( $result, @relations ) = $collation->add_relationship( 'n25', 'n26', {'type' => 'spelling'} ); # root -> rood
333 ok( !$result, "Refused to add dupe relationship: " . $relations[0] );
334
335 ( $result, @relations ) = $collation->add_relationship( 'n8', 'n13', {'type' => 'spelling', 'global' => 1 } ); # teh -> the
336 ok( $result, "Added global relationship between nodes" );
337 is( scalar @relations, 2, "Returned two relationship creations" );
338 is_deeply( $relations[0], [ 'n8', 'n13' ], "Returned the original collapse" );
339 is_deeply( $relations[1], [ 'n9', 'n23' ], "Returned the other collapse" );
340 is( $collation->reading( 'n8' )->position->reference, '6,2', "Harmonized position for n8 correct" );
341 is( $collation->reading( 'n9' )->position->reference, '9,3', "Harmonized position for n9 correct" );
4cdd82f1 342}
343
b49c4318 344done_testing();