Commit | Line | Data |
d71100ed |
1 | package Text::Tradition::Analysis; |
2 | |
3 | use strict; |
4 | use warnings; |
e4386ba9 |
5 | use Benchmark; |
88a6bac5 |
6 | use Encode qw/ encode_utf8 /; |
d1348d38 |
7 | use Exporter 'import'; |
b4cb2d60 |
8 | use Graph; |
88a6bac5 |
9 | use JSON qw/ encode_json decode_json /; |
10 | use LWP::UserAgent; |
d71100ed |
11 | use Text::Tradition; |
12 | use Text::Tradition::Stemma; |
13 | |
d1348d38 |
14 | use vars qw/ @EXPORT_OK /; |
a2cf85dd |
15 | @EXPORT_OK = qw/ run_analysis group_variants analyze_variant_location wit_stringify /; |
d1348d38 |
16 | |
7f52eac8 |
17 | =head1 NAME |
18 | |
19 | Text::Tradition::Analysis - functions for stemma analysis of a tradition |
20 | |
21 | =head1 SYNOPSIS |
22 | |
23 | use Text::Tradition; |
24 | use Text::Tradition::Analysis qw/ run_analysis analyze_variant_location /; |
25 | my $t = Text::Tradition->new( |
26 | 'name' => 'this is a text', |
27 | 'input' => 'TEI', |
28 | 'file' => '/path/to/tei_parallel_seg_file.xml' ); |
29 | $t->add_stemma( 'dotfile' => $stemmafile ); |
30 | |
31 | my $variant_data = run_analysis( $tradition ); |
32 | # Recalculate rank $n treating all orthographic variants as equivalent |
33 | my $reanalyze = analyze_variant_location( $tradition, $n, 0, 'orthographic' ); |
34 | |
35 | =head1 DESCRIPTION |
36 | |
37 | Text::Tradition is a library for representation and analysis of collated |
38 | texts, particularly medieval ones. The Collation is the central feature of |
39 | a Tradition, where the text, its sequence of readings, and its relationships |
40 | between readings are actually kept. |
41 | |
42 | =head1 SUBROUTINES |
43 | |
88a6bac5 |
44 | =head2 run_analysis( $tradition, %opts ) |
7f52eac8 |
45 | |
88a6bac5 |
46 | Runs the analysis described in analyze_variant_location on every location in the |
47 | collation of the given tradition, with the given options. These include: |
7f52eac8 |
48 | |
88a6bac5 |
49 | =over 4 |
50 | |
51 | =item * stemma_id - Specify which of the tradition's stemmata to use. Default |
52 | is 0 (i.e. the first). |
53 | |
54 | =item * ranks - Specify a list of location ranks to analyze; exclude the rest. |
55 | |
56 | =item * merge_types - Specify a list of relationship types, where related readings |
57 | should be treated as identical for the purposes of analysis. |
58 | |
f2fec47d |
59 | =item * exclude_type1 - Exclude those ranks whose groupings have only type-1 variants. |
60 | |
88a6bac5 |
61 | =back |
7f52eac8 |
62 | |
63 | =begin testing |
64 | |
65 | use Text::Tradition; |
66 | use Text::Tradition::Analysis qw/ run_analysis analyze_variant_location /; |
67 | |
68 | my $datafile = 't/data/florilegium_tei_ps.xml'; |
69 | my $tradition = Text::Tradition->new( 'input' => 'TEI', |
70 | 'name' => 'test0', |
71 | 'file' => $datafile ); |
72 | my $s = $tradition->add_stemma( 'dotfile' => 't/data/florilegium.dot' ); |
73 | is( ref( $s ), 'Text::Tradition::Stemma', "Added stemma to tradition" ); |
74 | |
f00cefe8 |
75 | my %expected_genealogical = ( |
a44aaf2a |
76 | 1 => 0, |
f00cefe8 |
77 | 2 => 1, |
a44aaf2a |
78 | 3 => 0, |
79 | 5 => 0, |
80 | 7 => 0, |
81 | 8 => 0, |
82 | 10 => 0, |
f00cefe8 |
83 | 13 => 1, |
a44aaf2a |
84 | 33 => 0, |
85 | 34 => 0, |
86 | 37 => 0, |
87 | 60 => 0, |
f00cefe8 |
88 | 81 => 1, |
a44aaf2a |
89 | 84 => 0, |
90 | 87 => 0, |
91 | 101 => 0, |
92 | 102 => 0, |
f00cefe8 |
93 | 122 => 1, |
a44aaf2a |
94 | 157 => 0, |
f00cefe8 |
95 | 166 => 1, |
96 | 169 => 1, |
a44aaf2a |
97 | 200 => 0, |
f00cefe8 |
98 | 216 => 1, |
99 | 217 => 1, |
100 | 219 => 1, |
101 | 241 => 1, |
102 | 242 => 1, |
103 | 243 => 1, |
104 | ); |
105 | |
7f52eac8 |
106 | my $data = run_analysis( $tradition ); |
f00cefe8 |
107 | foreach my $row ( @{$data->{'variants'}} ) { |
a44aaf2a |
108 | # Account for rows that used to be "not useful" |
109 | unless( exists $expected_genealogical{$row->{'id'}} ) { |
110 | $expected_genealogical{$row->{'id'}} = 1; |
111 | } |
6d25a3a0 |
112 | is( $row->{'genealogical'}, $expected_genealogical{$row->{'id'}}, |
f00cefe8 |
113 | "Got correct genealogical flag for row " . $row->{'id'} ); |
114 | } |
a44aaf2a |
115 | is( $data->{'variant_count'}, 58, "Got right total variant number" ); |
b4cb2d60 |
116 | # TODO Make something meaningful of conflict count, maybe test other bits |
7f52eac8 |
117 | |
118 | =end testing |
119 | |
120 | =cut |
121 | |
d71100ed |
122 | sub run_analysis { |
88a6bac5 |
123 | my( $tradition, %opts ) = @_; |
f00cefe8 |
124 | my $c = $tradition->collation; |
88a6bac5 |
125 | |
126 | my $stemma_id = $opts{'stemma_id'} || 0; |
1d73ecad |
127 | my @ranks = ref( $opts{'ranks'} ) eq 'ARRAY' ? @{$opts{'ranks'}} : (); |
128 | my @collapse = ref( $opts{'merge_types'} ) eq 'ARRAY' ? @{$opts{'merge_types'}} : (); |
88a6bac5 |
129 | |
130 | # Get the stemma |
131 | my $stemma = $tradition->stemma( $stemma_id ); |
b4cb2d60 |
132 | |
88a6bac5 |
133 | # Figure out which witnesses we are working with |
134 | my @lacunose = $stemma->hypotheticals; |
fae07016 |
135 | my @tradition_wits = map { $_->sigil } $tradition->witnesses; |
b4cb2d60 |
136 | map { push( @tradition_wits, $_->sigil.$c->ac_label ) if $_->is_layered } |
fae07016 |
137 | $tradition->witnesses; |
138 | push( @lacunose, _symmdiff( [ $stemma->witnesses ], \@tradition_wits ) ); |
88a6bac5 |
139 | |
140 | # Find and mark 'common' ranks for exclusion, unless they were |
141 | # explicitly specified. |
142 | unless( @ranks ) { |
143 | my %common_rank; |
a44aaf2a |
144 | foreach my $rdg ( $c->common_readings ) { |
88a6bac5 |
145 | $common_rank{$rdg->rank} = 1; |
146 | } |
147 | @ranks = grep { !$common_rank{$_} } ( 1 .. $c->end->rank-1 ); |
d71100ed |
148 | } |
7f52eac8 |
149 | |
88a6bac5 |
150 | # Group the variants to send to the solver |
151 | my @groups; |
a44aaf2a |
152 | my %lacunae; |
88a6bac5 |
153 | foreach my $rank ( @ranks ) { |
a44aaf2a |
154 | my $missing = [ @lacunose ]; |
f2fec47d |
155 | my $rankgroup = group_variants( $tradition, $rank, $missing, \@collapse ); |
156 | if( $opts{'exclude_type1'} ) { |
157 | # Check to see whether this is a "useful" group. |
158 | my( $rdgs, $grps ) = _useful_variant( $rankgroup ); |
159 | next unless @$rdgs; |
160 | } |
161 | push( @groups, $rankgroup ); |
a44aaf2a |
162 | $lacunae{$rank} = $missing; |
d71100ed |
163 | } |
b4cb2d60 |
164 | $DB::single = 1; |
88a6bac5 |
165 | # Parse the answer |
e59b8faa |
166 | my $answer = solve_variants( $stemma, @groups ); |
fae07016 |
167 | |
88a6bac5 |
168 | # Do further analysis on the answer |
a44aaf2a |
169 | my $conflict_count = 0; |
88a6bac5 |
170 | foreach my $idx ( 0 .. $#ranks ) { |
171 | my $location = $answer->{'variants'}->[$idx]; |
172 | # Add the rank back in |
173 | $location->{'id'} = $ranks[$idx]; |
a44aaf2a |
174 | # Add the lacunae back in |
175 | $location->{'missing'} = $lacunae{$ranks[$idx]}; |
88a6bac5 |
176 | # Run the extra analysis we need. |
88a6bac5 |
177 | analyze_location( $tradition, $stemma->graph, $location ); |
a44aaf2a |
178 | # Add the reading text back in |
179 | foreach my $rdghash ( @{$location->{'readings'}} ) { |
180 | $conflict_count++ |
181 | if exists $rdghash->{'conflict'} && $rdghash->{'conflict'}; |
182 | my $rdg = $c->reading( $rdghash->{'readingid'} ); |
183 | $rdghash->{'text'} = $rdg ? $rdg->text : $rdghash->{'readingid'}; |
184 | } |
88a6bac5 |
185 | } |
a44aaf2a |
186 | $answer->{'conflict_count'} = $conflict_count; |
f00cefe8 |
187 | |
88a6bac5 |
188 | return $answer; |
d71100ed |
189 | } |
190 | |
7f52eac8 |
191 | =head2 group_variants( $tradition, $rank, $lacunose, @merge_relationship_types ) |
192 | |
193 | Groups the variants at the given $rank of the collation, treating any |
194 | relationships in @merge_relationship_types as equivalent. $lacunose should |
195 | be a reference to an array, to which the sigla of lacunose witnesses at this |
196 | rank will be appended. |
197 | |
f2fec47d |
198 | Returns a hash $group_readings where $rdg is attested by the witnesses listed |
199 | in $group_readings->{$rdg}. |
7f52eac8 |
200 | |
201 | =cut |
202 | |
203 | # Return group_readings, groups, lacunose |
d1348d38 |
204 | sub group_variants { |
7f52eac8 |
205 | my( $tradition, $rank, $lacunose, $collapse ) = @_; |
206 | my $c = $tradition->collation; |
b4cb2d60 |
207 | my $aclabel = $c->ac_label; |
7f52eac8 |
208 | # Get the alignment table readings |
209 | my %readings_at_rank; |
210 | my @gap_wits; |
1d73ecad |
211 | foreach my $tablewit ( @{$c->alignment_table->{'alignment'}} ) { |
7f52eac8 |
212 | my $rdg = $tablewit->{'tokens'}->[$rank-1]; |
fae07016 |
213 | my $wit = $tablewit->{'witness'}; |
7f52eac8 |
214 | if( $rdg && $rdg->{'t'}->is_lacuna ) { |
b4cb2d60 |
215 | _add_to_witlist( $wit, $lacunose, $aclabel ); |
7f52eac8 |
216 | } elsif( $rdg ) { |
217 | $readings_at_rank{$rdg->{'t'}->text} = $rdg->{'t'}; |
218 | } else { |
b4cb2d60 |
219 | _add_to_witlist( $wit, \@gap_wits, $aclabel ); |
7f52eac8 |
220 | } |
221 | } |
d1348d38 |
222 | |
7f52eac8 |
223 | # Group the readings, collapsing groups by relationship if needed |
224 | my %grouped_readings; |
b4cb2d60 |
225 | foreach my $rdg ( sort { $b->witnesses <=> $a->witnesses } |
226 | values %readings_at_rank ) { |
7f52eac8 |
227 | # Skip readings that have been collapsed into others. |
f00cefe8 |
228 | next if exists $grouped_readings{$rdg->id} && !$grouped_readings{$rdg->id}; |
7f52eac8 |
229 | my @wits = $rdg->witnesses; |
230 | if( $collapse ) { |
231 | my $filter = sub { my $r = $_[0]; grep { $_ eq $r->type } @$collapse; }; |
232 | foreach my $other ( $rdg->related_readings( $filter ) ) { |
fae07016 |
233 | my @otherwits = $other->witnesses; |
fae07016 |
234 | push( @wits, @otherwits ); |
f00cefe8 |
235 | $grouped_readings{$other->id} = 0; |
d1348d38 |
236 | } |
237 | } |
f00cefe8 |
238 | $grouped_readings{$rdg->id} = \@wits; |
7f52eac8 |
239 | } |
240 | $grouped_readings{'(omitted)'} = \@gap_wits if @gap_wits; |
241 | # Get rid of our collapsed readings |
242 | map { delete $grouped_readings{$_} unless $grouped_readings{$_} } |
243 | keys %grouped_readings |
244 | if $collapse; |
245 | |
5be0cdeb |
246 | return \%grouped_readings; |
d1348d38 |
247 | } |
248 | |
88a6bac5 |
249 | =head2 solve_variants( $graph, @groups ) |
250 | |
251 | Sends the set of groups to the external graph solver service and returns |
252 | a cleaned-up answer, adding the rank IDs back where they belong. |
253 | |
254 | The JSON has the form |
255 | { "graph": [ stemmagraph DOT string without newlines ], |
256 | "groupings": [ array of arrays of groups, one per rank ] } |
257 | |
258 | The answer has the form |
259 | { "variants" => [ array of variant location structures ], |
260 | "variant_count" => total, |
261 | "conflict_count" => number of conflicts detected, |
262 | "genealogical_count" => number of solutions found } |
263 | |
264 | =cut |
265 | |
266 | sub solve_variants { |
e59b8faa |
267 | my( $stemma, @groups ) = @_; |
88a6bac5 |
268 | |
269 | # Make the json with stemma + groups |
b4cb2d60 |
270 | my $groupings = []; |
88a6bac5 |
271 | foreach my $ghash ( @groups ) { |
272 | my @grouping; |
6d25a3a0 |
273 | foreach my $k ( keys %$ghash ) { |
88a6bac5 |
274 | push( @grouping, $ghash->{$k} ); |
275 | } |
b4cb2d60 |
276 | push( @$groupings, \@grouping ); |
88a6bac5 |
277 | } |
b4cb2d60 |
278 | ## Witness map is a HACK to get around limitations in node names from IDP |
279 | my $witness_map = {}; |
280 | my $json = encode_json( _safe_wit_strings( $stemma, $groupings, $witness_map ) ); |
88a6bac5 |
281 | |
282 | # Send it off and get the result |
283 | my $solver_url = 'http://byzantini.st/cgi-bin/graphcalc.cgi'; |
284 | my $ua = LWP::UserAgent->new(); |
285 | my $resp = $ua->post( $solver_url, 'Content-Type' => 'application/json', |
286 | 'Content' => $json ); |
287 | |
288 | my $answer; |
6d25a3a0 |
289 | my $used_idp; |
88a6bac5 |
290 | if( $resp->is_success ) { |
b4cb2d60 |
291 | $answer = _desanitize_names( decode_json( $resp->content ), $witness_map ); |
6d25a3a0 |
292 | $used_idp = 1; |
88a6bac5 |
293 | } else { |
fae07016 |
294 | # Fall back to the old method. |
295 | warn "IDP solver returned " . $resp->status_line . " / " . $resp->content |
296 | . "; falling back to perl method"; |
b4cb2d60 |
297 | $answer = perl_solver( $stemma, @$groupings ); |
88a6bac5 |
298 | } |
299 | |
300 | # Fold the result back into what we know about the groups. |
301 | my $variants = []; |
302 | my $genealogical = 0; |
303 | foreach my $idx ( 0 .. $#groups ) { |
304 | my( $calc_groups, $result ) = @{$answer->[$idx]}; |
6d25a3a0 |
305 | if( $result ) { |
306 | $genealogical++; |
307 | # Prune the calculated groups, in case the IDP solver failed to. |
308 | if( $used_idp ) { |
309 | my @pruned_groups; |
310 | foreach my $cg ( @$calc_groups ) { |
311 | my @pg = _prune_group( $cg, $stemma ); |
312 | push( @pruned_groups, \@pg ); |
313 | } |
314 | $calc_groups = \@pruned_groups; |
315 | } |
316 | } |
88a6bac5 |
317 | my $input_group = $groups[$idx]; |
318 | foreach my $k ( sort keys %$input_group ) { |
319 | my $cg = shift @$calc_groups; |
320 | $input_group->{$k} = $cg; |
321 | } |
322 | my $vstruct = { |
323 | 'genealogical' => $result, |
324 | 'readings' => [], |
fae07016 |
325 | }; |
88a6bac5 |
326 | foreach my $k ( keys %$input_group ) { |
327 | push( @{$vstruct->{'readings'}}, |
fae07016 |
328 | { 'readingid' => $k, 'group' => $input_group->{$k}} ); |
88a6bac5 |
329 | } |
330 | push( @$variants, $vstruct ); |
331 | } |
332 | |
333 | return { 'variants' => $variants, |
334 | 'variant_count' => scalar @$variants, |
335 | 'genealogical_count' => $genealogical }; |
336 | } |
337 | |
b4cb2d60 |
338 | #### HACKERY to cope with IDP's limited idea of what a node name looks like ### |
339 | |
340 | sub _safe_wit_strings { |
341 | my( $stemma, $groupings, $witness_map ) = @_; |
342 | my $safegraph = Graph->new(); |
343 | # Convert the graph to a safe representation and store the conversion. |
344 | foreach my $n ( $stemma->graph->vertices ) { |
345 | my $sn = _safe_witstr( $n ); |
346 | warn "Ambiguous stringification $sn for $n and " . $witness_map->{$sn} |
347 | if exists $witness_map->{$sn}; |
348 | $witness_map->{$sn} = $n; |
349 | $safegraph->add_vertex( $sn ); |
350 | $safegraph->set_vertex_attributes( $sn, |
351 | $stemma->graph->get_vertex_attributes( $n ) ); |
352 | } |
353 | foreach my $e ( $stemma->graph->edges ) { |
354 | my @safe_e = ( _safe_witstr( $e->[0] ), _safe_witstr( $e->[1] ) ); |
355 | $safegraph->add_edge( @safe_e ); |
356 | } |
357 | my $safe_stemma = Text::Tradition::Stemma->new( |
358 | 'collation' => $stemma->collation, 'graph' => $safegraph ); |
359 | |
360 | # Now convert the witness groupings to a safe representation. |
361 | my $safe_groupings = []; |
362 | foreach my $grouping ( @$groupings ) { |
363 | my $safe_grouping = []; |
364 | foreach my $group ( @$grouping ) { |
365 | my $safe_group = []; |
366 | foreach my $n ( @$group ) { |
367 | my $sn = _safe_witstr( $n ); |
368 | warn "Ambiguous stringification $sn for $n and " . $witness_map->{$sn} |
369 | if exists $witness_map->{$sn} && $witness_map->{$sn} ne $n; |
370 | $witness_map->{$sn} = $n; |
371 | push( @$safe_group, $sn ); |
372 | } |
373 | push( @$safe_grouping, $safe_group ); |
374 | } |
375 | push( @$safe_groupings, $safe_grouping ); |
376 | } |
377 | |
378 | # Return it all in the struct we expect. We have stored the reductions |
379 | # in the $witness_map that we were passed. |
380 | return { 'graph' => $safe_stemma->editable( ' ' ), 'groupings' => $safe_groupings }; |
381 | } |
382 | |
383 | sub _safe_witstr { |
384 | my $witstr = shift; |
385 | $witstr =~ s/\s+/_/g; |
386 | $witstr =~ s/[^\w\d-]//g; |
387 | return $witstr; |
388 | } |
389 | |
390 | sub _desanitize_names { |
391 | my( $jsonstruct, $witness_map ) = @_; |
392 | my $result = []; |
393 | foreach my $grouping ( @$jsonstruct ) { |
394 | my $real_grouping = []; |
395 | foreach my $element ( @$grouping ) { |
396 | if( ref( $element ) eq 'ARRAY' ) { |
397 | # it's the groupset. |
398 | my $real_groupset = []; |
399 | foreach my $group ( @$element ) { |
400 | my $real_group = []; |
401 | foreach my $n ( @$group ) { |
402 | my $rn = $witness_map->{$n}; |
403 | push( @$real_group, $rn ); |
404 | } |
405 | push( @$real_groupset, $real_group ); |
406 | } |
407 | push( @$real_grouping, $real_groupset ); |
408 | } else { |
409 | # It is the boolean, not actually a group. |
410 | push( @$real_grouping, $element ); |
411 | } |
412 | } |
413 | push( @$result, $real_grouping ); |
414 | } |
415 | return $result; |
416 | } |
417 | |
418 | ### END HACKERY ### |
419 | |
fae07016 |
420 | =head2 analyze_location ( $tradition, $graph, $location_hash ) |
7f52eac8 |
421 | |
fae07016 |
422 | Given the tradition, its stemma graph, and the solution from the graph solver, |
423 | work out the rest of the information we want. For each reading we need missing, |
424 | conflict, reading_parents, independent_occurrence, followed, not_followed, and follow_unknown. Alters the location_hash in place. |
7f52eac8 |
425 | |
426 | =cut |
732152b1 |
427 | |
fae07016 |
428 | sub analyze_location { |
429 | my ( $tradition, $graph, $variant_row ) = @_; |
430 | |
431 | # Make a hash of all known node memberships, and make the subgraphs. |
432 | my $contig = {}; |
433 | my $reading_roots = {}; |
434 | my $subgraph = {}; |
435 | foreach my $rdghash ( @{$variant_row->{'readings'}} ) { |
436 | my $rid = $rdghash->{'readingid'}; |
437 | map { $contig->{$_} = $rid } @{$rdghash->{'group'}}; |
c4a4fb1b |
438 | |
fae07016 |
439 | # Make the subgraph. |
440 | my $part = $graph->copy; |
441 | my %these_vertices; |
442 | map { $these_vertices{$_} = 1 } @{$rdghash->{'group'}}; |
443 | $part->delete_vertices( grep { !$these_vertices{$_} } $part->vertices ); |
444 | $subgraph->{$rid} = $part; |
445 | # Get the reading roots. |
446 | map { $reading_roots->{$_} = $rid } $part->predecessorless_vertices; |
bebec0e9 |
447 | } |
448 | |
fae07016 |
449 | # Now that we have all the node group memberships, calculate followed/ |
bebec0e9 |
450 | # non-followed/unknown values for each reading. Also figure out the |
451 | # reading's evident parent(s). |
452 | foreach my $rdghash ( @{$variant_row->{'readings'}} ) { |
fae07016 |
453 | # Group string key - TODO do we need this? |
454 | my $gst = wit_stringify( $rdghash->{'group'} ); |
455 | my $rid = $rdghash->{'readingid'}; |
456 | # Get the subgraph |
457 | my $part = $subgraph->{$rid}; |
458 | |
459 | # Start figuring things out. |
bebec0e9 |
460 | my @roots = $part->predecessorless_vertices; |
461 | $rdghash->{'independent_occurrence'} = scalar @roots; |
462 | $rdghash->{'followed'} = scalar( $part->vertices ) - scalar( @roots ); |
463 | # Find the parent readings, if any, of this reading. |
f00cefe8 |
464 | my %rdgparents; |
bebec0e9 |
465 | foreach my $wit ( @roots ) { |
f00cefe8 |
466 | # Look in the main stemma to find this witness's extant or known-reading |
467 | # immediate ancestor(s), and look up the reading that each ancestor olds. |
468 | my @check = $graph->predecessors( $wit ); |
469 | while( @check ) { |
470 | my @next; |
471 | foreach my $wparent( @check ) { |
fae07016 |
472 | my $preading = $contig->{$wparent}; |
473 | if( $preading ) { |
474 | $rdgparents{$preading} = 1; |
f00cefe8 |
475 | } else { |
476 | push( @next, $graph->predecessors( $wparent ) ); |
477 | } |
478 | } |
479 | @check = @next; |
480 | } |
bebec0e9 |
481 | } |
f00cefe8 |
482 | $rdghash->{'reading_parents'} = [ keys %rdgparents ]; |
bebec0e9 |
483 | |
484 | # Find the number of times this reading was altered, and the number of |
485 | # times we're not sure. |
486 | my( %nofollow, %unknownfollow ); |
487 | foreach my $wit ( $part->vertices ) { |
488 | foreach my $wchild ( $graph->successors( $wit ) ) { |
489 | next if $part->has_vertex( $wchild ); |
fae07016 |
490 | if( $reading_roots->{$wchild} && $contig->{$wchild} ) { |
bebec0e9 |
491 | # It definitely changed here. |
492 | $nofollow{$wchild} = 1; |
493 | } elsif( !($contig->{$wchild}) ) { |
494 | # The child is a hypothetical node not definitely in |
495 | # any group. Answer is unknown. |
496 | $unknownfollow{$wchild} = 1; |
497 | } # else it's a non-root node in a known group, and therefore |
498 | # is presumed to have its reading from its group, not this link. |
499 | } |
500 | } |
501 | $rdghash->{'not_followed'} = keys %nofollow; |
502 | $rdghash->{'follow_unknown'} = keys %unknownfollow; |
fae07016 |
503 | |
504 | # Now say whether this reading represents a conflict. |
505 | unless( $variant_row->{'genealogical'} ) { |
506 | $rdghash->{'conflict'} = @roots != 1; |
507 | } |
c4a4fb1b |
508 | } |
d71100ed |
509 | } |
510 | |
fae07016 |
511 | |
512 | =head2 perl_solver( $tradition, $rank, $stemma_id, @merge_relationship_types ) |
513 | |
514 | ** NOTE ** This method should hopefully not be called - it is not guaranteed |
515 | to be correct. Serves as a backup for the real solver. |
516 | |
517 | Runs an analysis of the given tradition, at the location given in $rank, |
518 | against the graph of the stemma specified in $stemma_id. The argument |
519 | @merge_relationship_types is an optional list of relationship types for |
520 | which readings so related should be treated as equivalent. |
521 | |
522 | Returns a nested array data structure as follows: |
523 | |
524 | [ [ group_list, is_genealogical ], [ group_list, is_genealogical ] ... ] |
525 | |
526 | where the group list is the array of arrays passed in for each element of @groups, |
527 | possibly with the addition of hypothetical readings. |
528 | |
529 | |
530 | =cut |
531 | |
532 | sub perl_solver { |
e59b8faa |
533 | my( $stemma, @groups ) = @_; |
534 | my $graph = $stemma->graph; |
535 | my @answer; |
536 | foreach my $g ( @groups ) { |
537 | push( @answer, _solve_variant_location( $graph, $g ) ); |
538 | } |
539 | return \@answer; |
fae07016 |
540 | } |
541 | |
e59b8faa |
542 | sub _solve_variant_location { |
543 | my( $graph, $groups ) = @_; |
fae07016 |
544 | # Now do the work. |
e59b8faa |
545 | my $contig = {}; |
546 | my $subgraph = {}; |
547 | my $is_conflicted; |
548 | my $conflict = {}; |
549 | |
550 | # Mark each ms as in its own group, first. |
551 | foreach my $g ( @$groups ) { |
552 | my $gst = wit_stringify( $g ); |
553 | map { $contig->{$_} = $gst } @$g; |
554 | } |
555 | |
556 | # Now for each unmarked node in the graph, initialize an array |
557 | # for possible group memberships. We will use this later to |
558 | # resolve potential conflicts. |
559 | map { $contig->{$_} = [] unless $contig->{$_} } $graph->vertices; |
560 | foreach my $g ( sort { scalar @$b <=> scalar @$a } @$groups ) { |
561 | my $gst = wit_stringify( $g ); # This is the group name |
562 | # Copy the graph, and delete all non-members from the new graph. |
563 | my $part = $graph->copy; |
564 | my @group_roots; |
565 | $part->delete_vertices( |
566 | grep { !ref( $contig->{$_} ) && $contig->{$_} ne $gst } $graph->vertices ); |
567 | |
568 | # Now look to see if our group is connected. |
569 | if( @$g > 1 ) { |
570 | # We have to take directionality into account. |
571 | # How many root nodes do we have? |
572 | my @roots = grep { ref( $contig->{$_} ) || $contig->{$_} eq $gst } |
573 | $part->predecessorless_vertices; |
574 | # Assuming that @$g > 1, find the first root node that has at |
575 | # least one successor belonging to our group. If this reading |
576 | # is genealogical, there should be only one, but we will check |
577 | # that implicitly later. |
578 | foreach my $root ( @roots ) { |
579 | # Prune the tree to get rid of extraneous hypotheticals. |
580 | $root = _prune_subtree( $part, $root, $contig ); |
581 | next unless $root; |
582 | # Save this root for our group. |
583 | push( @group_roots, $root ); |
584 | # Get all the successor nodes of our root. |
585 | } |
586 | } else { |
587 | # Dispense with the trivial case of one reading. |
588 | my $wit = $g->[0]; |
589 | @group_roots = ( $wit ); |
590 | foreach my $v ( $part->vertices ) { |
591 | $part->delete_vertex( $v ) unless $v eq $wit; |
592 | } |
593 | } |
594 | |
595 | if( @group_roots > 1 ) { |
596 | $conflict->{$gst} = 1; |
597 | $is_conflicted = 1; |
598 | } |
599 | # Paint the 'hypotheticals' with our group. |
600 | foreach my $wit ( $part->vertices ) { |
601 | if( ref( $contig->{$wit} ) ) { |
602 | push( @{$contig->{$wit}}, $gst ); |
603 | } elsif( $contig->{$wit} ne $gst ) { |
604 | warn "How did we get here?"; |
605 | } |
606 | } |
607 | |
608 | |
609 | # Save the relevant subgraph. |
610 | $subgraph->{$gst} = $part; |
611 | } |
612 | |
613 | # For each of our hypothetical readings, flatten its 'contig' array if |
614 | # the array contains zero or one group. If we have any unflattened arrays, |
615 | # we may need to run the resolution process. If the reading is already known |
616 | # to have a conflict, flatten the 'contig' array to nothing; we won't resolve |
617 | # it. |
618 | my @resolve; |
619 | foreach my $wit ( keys %$contig ) { |
620 | next unless ref( $contig->{$wit} ); |
621 | if( @{$contig->{$wit}} > 1 ) { |
622 | if( $is_conflicted ) { |
623 | $contig->{$wit} = ''; # We aren't going to decide. |
624 | } else { |
625 | push( @resolve, $wit ); |
626 | } |
627 | } else { |
628 | my $gst = pop @{$contig->{$wit}}; |
629 | $contig->{$wit} = $gst || ''; |
630 | } |
631 | } |
632 | |
633 | if( @resolve ) { |
634 | my $still_contig = {}; |
635 | foreach my $h ( @resolve ) { |
636 | # For each of the hypothetical readings with more than one possibility, |
637 | # try deleting it from each of its member subgraphs in turn, and see |
638 | # if that breaks the contiguous grouping. |
639 | # TODO This can still break in a corner case where group A can use |
640 | # either vertex 1 or 2, and group B can use either vertex 2 or 1. |
641 | # Revisit this if necessary; it could get brute-force nasty. |
642 | foreach my $gst ( @{$contig->{$h}} ) { |
643 | my $gpart = $subgraph->{$gst}->copy(); |
644 | # If we have come this far, there is only one root and everything |
645 | # is reachable from it. |
646 | my( $root ) = $gpart->predecessorless_vertices; |
647 | my $reachable = {}; |
648 | map { $reachable->{$_} = 1 } $gpart->vertices; |
649 | |
650 | # Try deleting the hypothetical node. |
651 | $gpart->delete_vertex( $h ); |
652 | if( $h eq $root ) { |
653 | # See if we still have a single root. |
654 | my @roots = $gpart->predecessorless_vertices; |
655 | warn "This shouldn't have happened" unless @roots; |
656 | if( @roots > 1 ) { |
657 | # $h is needed by this group. |
658 | if( exists( $still_contig->{$h} ) ) { |
659 | # Conflict! |
660 | $conflict->{$gst} = 1; |
661 | $still_contig->{$h} = ''; |
662 | } else { |
663 | $still_contig->{$h} = $gst; |
664 | } |
665 | } |
666 | } else { |
667 | # $h is somewhere in the middle. See if everything |
668 | # else can still be reached from the root. |
669 | my %still_reachable = ( $root => 1 ); |
670 | map { $still_reachable{$_} = 1 } |
671 | $gpart->all_successors( $root ); |
672 | foreach my $v ( keys %$reachable ) { |
673 | next if $v eq $h; |
674 | if( !$still_reachable{$v} |
675 | && ( $contig->{$v} eq $gst |
676 | || ( exists $still_contig->{$v} |
677 | && $still_contig->{$v} eq $gst ) ) ) { |
678 | # We need $h. |
679 | if( exists $still_contig->{$h} ) { |
680 | # Conflict! |
681 | $conflict->{$gst} = 1; |
682 | $still_contig->{$h} = ''; |
683 | } else { |
684 | $still_contig->{$h} = $gst; |
685 | } |
686 | last; |
687 | } # else we don't need $h in this group. |
688 | } # end foreach $v |
689 | } # endif $h eq $root |
690 | } # end foreach $gst |
691 | } # end foreach $h |
692 | |
693 | # Now we have some hypothetical vertices in $still_contig that are the |
694 | # "real" group memberships. Replace these in $contig. |
695 | foreach my $v ( keys %$contig ) { |
696 | next unless ref $contig->{$v}; |
697 | $contig->{$v} = $still_contig->{$v}; |
698 | } |
699 | } # end if @resolve |
700 | |
701 | my $is_genealogical = keys %$conflict ? JSON::false : JSON::true; |
702 | my $variant_row = [ [], $is_genealogical ]; |
703 | # Fill in the groupings from $contig. |
704 | foreach my $g ( @$groups ) { |
705 | my $gst = wit_stringify( $g ); |
706 | my @realgroup = grep { $contig->{$_} eq $gst } keys %$contig; |
707 | push( @{$variant_row->[0]}, \@realgroup ); |
708 | } |
709 | return $variant_row; |
710 | } |
fae07016 |
711 | |
6d25a3a0 |
712 | sub _prune_group { |
713 | my( $group, $stemma ) = @_; |
714 | # Get these into a form prune_subtree will recognize. Make a "contighash" |
715 | my $hypohash = {}; |
716 | map { $hypohash->{$_} = 1 } @$group; |
717 | # ...with reference values for hypotheticals. |
718 | map { $hypohash->{$_} = [] } $stemma->hypotheticals; |
719 | # Make our subgraph |
720 | my $subgraph = $stemma->graph->copy; |
721 | map { $subgraph->delete_vertex( $_ ) unless exists $hypohash->{$_} } |
722 | $subgraph->vertices; |
723 | # ...and find the root. |
724 | my( $root ) = $subgraph->predecessorless_vertices; |
725 | # Now prune and return the remaining vertices. |
726 | _prune_subtree( $subgraph, $root, $hypohash ); |
727 | return $subgraph->vertices; |
728 | } |
729 | |
7f52eac8 |
730 | sub _prune_subtree { |
231d71fc |
731 | my( $tree, $root, $contighash ) = @_; |
732 | # First, delete hypothetical leaves / orphans until there are none left. |
733 | my @orphan_hypotheticals = grep { ref( $contighash->{$_} ) } |
734 | $tree->successorless_vertices; |
735 | while( @orphan_hypotheticals ) { |
736 | $tree->delete_vertices( @orphan_hypotheticals ); |
737 | @orphan_hypotheticals = grep { ref( $contighash->{$_} ) } |
738 | $tree->successorless_vertices; |
739 | } |
740 | # Then delete a hypothetical root with only one successor, moving the |
bebec0e9 |
741 | # root to the first child that has no other predecessors. |
231d71fc |
742 | while( $tree->successors( $root ) == 1 && ref $contighash->{$root} ) { |
743 | my @nextroot = $tree->successors( $root ); |
744 | $tree->delete_vertex( $root ); |
bebec0e9 |
745 | ( $root ) = grep { $tree->is_predecessorless_vertex( $_ ) } @nextroot; |
231d71fc |
746 | } |
747 | # The tree has been modified in place, but we need to know the new root. |
bebec0e9 |
748 | $root = undef unless $root && $tree->has_vertex( $root ); |
231d71fc |
749 | return $root; |
750 | } |
d71100ed |
751 | # Add the variant, subject to a.c. representation logic. |
752 | # This assumes that we will see the 'main' version before the a.c. version. |
753 | sub add_variant_wit { |
754 | my( $arr, $wit, $acstr ) = @_; |
755 | my $skip; |
756 | if( $wit =~ /^(.*)\Q$acstr\E$/ ) { |
757 | my $real = $1; |
758 | $skip = grep { $_ =~ /^\Q$real\E$/ } @$arr; |
759 | } |
760 | push( @$arr, $wit ) unless $skip; |
761 | } |
762 | |
5be0cdeb |
763 | sub _useful_variant { |
764 | my( $group_readings, $graph, $acstr ) = @_; |
765 | |
766 | # TODO Decide what to do with AC witnesses |
767 | |
768 | # Sort by group size and return |
769 | my $is_useful = 0; |
770 | my( @readings, @groups ); # The sorted groups for our answer. |
771 | foreach my $rdg ( sort { @{$group_readings->{$b}} <=> @{$group_readings->{$a}} } |
772 | keys %$group_readings ) { |
773 | push( @readings, $rdg ); |
774 | push( @groups, $group_readings->{$rdg} ); |
775 | if( @{$group_readings->{$rdg}} > 1 ) { |
776 | $is_useful++; |
777 | } else { |
778 | my( $wit ) = @{$group_readings->{$rdg}}; |
779 | $wit =~ s/^(.*)\Q$acstr\E$/$1/; |
780 | $is_useful++ unless( $graph->is_sink_vertex( $wit ) ); |
781 | } |
782 | } |
783 | if( $is_useful > 1 ) { |
784 | return( \@readings, \@groups ); |
785 | } else { |
786 | return( [], [] ); |
787 | } |
788 | } |
789 | |
7f52eac8 |
790 | =head2 wit_stringify( $groups ) |
791 | |
792 | Takes an array of witness groupings and produces a string like |
793 | ['A','B'] / ['C','D','E'] / ['F'] |
d71100ed |
794 | |
7f52eac8 |
795 | =cut |
d71100ed |
796 | |
797 | sub wit_stringify { |
798 | my $groups = shift; |
799 | my @gst; |
800 | # If we were passed an array of witnesses instead of an array of |
801 | # groupings, then "group" the witnesses first. |
802 | unless( ref( $groups->[0] ) ) { |
803 | my $mkgrp = [ $groups ]; |
804 | $groups = $mkgrp; |
805 | } |
806 | foreach my $g ( @$groups ) { |
807 | push( @gst, '[' . join( ',', map { "'$_'" } @$g ) . ']' ); |
808 | } |
809 | return join( ' / ', @gst ); |
810 | } |
7f52eac8 |
811 | |
5be0cdeb |
812 | # Helper function to ensure that X and X a.c. never appear in the same list. |
813 | sub _add_to_witlist { |
814 | my( $wit, $list, $acstr ) = @_; |
815 | my %inlist; |
816 | my $idx = 0; |
817 | map { $inlist{$_} = $idx++ } @$list; |
818 | if( $wit =~ /^(.*)\Q$acstr\E$/ ) { |
819 | my $acwit = $1; |
820 | unless( exists $inlist{$acwit} ) { |
821 | push( @$list, $acwit.$acstr ); |
822 | } |
823 | } else { |
824 | if( exists( $inlist{$wit.$acstr} ) ) { |
825 | # Replace the a.c. version with the main witness |
826 | my $i = $inlist{$wit.$acstr}; |
827 | $list->[$i] = $wit; |
828 | } else { |
829 | push( @$list, $wit ); |
830 | } |
831 | } |
832 | } |
833 | |
bebec0e9 |
834 | sub _symmdiff { |
835 | my( $lista, $listb ) = @_; |
7f52eac8 |
836 | my %union; |
837 | my %scalars; |
838 | map { $union{$_} = 1; $scalars{$_} = $_ } @$lista; |
839 | map { $union{$_} += 1; $scalars{$_} = $_ } @$listb; |
bebec0e9 |
840 | my @set = grep { $union{$_} == 1 } keys %union; |
7f52eac8 |
841 | return map { $scalars{$_} } @set; |
842 | } |
843 | |
844 | 1; |
845 | |
846 | =head1 LICENSE |
847 | |
848 | This package is free software and is provided "as is" without express |
849 | or implied warranty. You can redistribute it and/or modify it under |
850 | the same terms as Perl itself. |
851 | |
852 | =head1 AUTHOR |
853 | |
854 | Tara L Andrews E<lt>aurum@cpan.orgE<gt> |