biodiversity 0.5.16 → 0.6.0
Sign up to get free protection for your applications and to get access to all the features.
- data/.gitignore +2 -0
- data/README.rdoc +5 -5
- data/Rakefile +8 -3
- data/VERSION +1 -1
- data/bin/nnparse +7 -3
- data/bin/parserver +1 -0
- data/lib/biodiversity/parser.rb +50 -5
- data/lib/biodiversity/parser/scientific_name_clean.treetop +131 -40
- data/spec/parser/scientific_name_clean.spec.rb +53 -27
- data/spec/parser/test_data.txt +73 -21
- metadata +4 -8
- data/biodiversity.gemspec +0 -88
- data/lib/biodiversity/parser/scientific_name_canonical.rb +0 -481
- data/lib/biodiversity/parser/scientific_name_clean.rb +0 -6118
- data/lib/biodiversity/parser/scientific_name_dirty.rb +0 -1309
@@ -1,1309 +0,0 @@
|
|
1
|
-
# encoding: UTF-8
|
2
|
-
module ScientificNameDirty
|
3
|
-
include Treetop::Runtime
|
4
|
-
|
5
|
-
def root
|
6
|
-
@root || :root
|
7
|
-
end
|
8
|
-
|
9
|
-
include ScientificNameClean
|
10
|
-
|
11
|
-
module Root0
|
12
|
-
def space1
|
13
|
-
elements[0]
|
14
|
-
end
|
15
|
-
|
16
|
-
def a
|
17
|
-
elements[1]
|
18
|
-
end
|
19
|
-
|
20
|
-
def space2
|
21
|
-
elements[2]
|
22
|
-
end
|
23
|
-
end
|
24
|
-
|
25
|
-
module Root1
|
26
|
-
def value
|
27
|
-
a.value.gsub(/\s{2,}/, ' ').strip
|
28
|
-
end
|
29
|
-
|
30
|
-
def canonical
|
31
|
-
a.canonical.gsub(/\s{2,}/, ' ').strip
|
32
|
-
end
|
33
|
-
|
34
|
-
def pos
|
35
|
-
a.pos
|
36
|
-
end
|
37
|
-
|
38
|
-
def hybrid
|
39
|
-
a.hybrid
|
40
|
-
end
|
41
|
-
|
42
|
-
def details
|
43
|
-
a.details.class == Array ? a.details : [a.details]
|
44
|
-
end
|
45
|
-
|
46
|
-
def parser_run
|
47
|
-
2
|
48
|
-
end
|
49
|
-
end
|
50
|
-
|
51
|
-
def _nt_root
|
52
|
-
start_index = index
|
53
|
-
if node_cache[:root].has_key?(index)
|
54
|
-
cached = node_cache[:root][index]
|
55
|
-
if cached
|
56
|
-
cached = SyntaxNode.new(input, index...(index + 1)) if cached == true
|
57
|
-
@index = cached.interval.end
|
58
|
-
end
|
59
|
-
return cached
|
60
|
-
end
|
61
|
-
|
62
|
-
i0, s0 = index, []
|
63
|
-
r1 = _nt_space
|
64
|
-
s0 << r1
|
65
|
-
if r1
|
66
|
-
r2 = _nt_scientific_name_5
|
67
|
-
s0 << r2
|
68
|
-
if r2
|
69
|
-
r3 = _nt_space
|
70
|
-
s0 << r3
|
71
|
-
end
|
72
|
-
end
|
73
|
-
if s0.last
|
74
|
-
r0 = instantiate_node(SyntaxNode,input, i0...index, s0)
|
75
|
-
r0.extend(Root0)
|
76
|
-
r0.extend(Root1)
|
77
|
-
else
|
78
|
-
@index = i0
|
79
|
-
r0 = nil
|
80
|
-
end
|
81
|
-
|
82
|
-
node_cache[:root][start_index] = r0
|
83
|
-
|
84
|
-
r0
|
85
|
-
end
|
86
|
-
|
87
|
-
module ScientificName50
|
88
|
-
def a
|
89
|
-
elements[0]
|
90
|
-
end
|
91
|
-
|
92
|
-
def garbage
|
93
|
-
elements[1]
|
94
|
-
end
|
95
|
-
end
|
96
|
-
|
97
|
-
module ScientificName51
|
98
|
-
def value
|
99
|
-
a.value
|
100
|
-
end
|
101
|
-
|
102
|
-
def canonical
|
103
|
-
a.canonical
|
104
|
-
end
|
105
|
-
|
106
|
-
def pos
|
107
|
-
a.pos
|
108
|
-
end
|
109
|
-
|
110
|
-
def details
|
111
|
-
a.details
|
112
|
-
end
|
113
|
-
end
|
114
|
-
|
115
|
-
def _nt_scientific_name_5
|
116
|
-
start_index = index
|
117
|
-
if node_cache[:scientific_name_5].has_key?(index)
|
118
|
-
cached = node_cache[:scientific_name_5][index]
|
119
|
-
if cached
|
120
|
-
cached = SyntaxNode.new(input, index...(index + 1)) if cached == true
|
121
|
-
@index = cached.interval.end
|
122
|
-
end
|
123
|
-
return cached
|
124
|
-
end
|
125
|
-
|
126
|
-
i0 = index
|
127
|
-
i1, s1 = index, []
|
128
|
-
r2 = _nt_scientific_name_4
|
129
|
-
s1 << r2
|
130
|
-
if r2
|
131
|
-
r3 = _nt_garbage
|
132
|
-
s1 << r3
|
133
|
-
end
|
134
|
-
if s1.last
|
135
|
-
r1 = instantiate_node(SyntaxNode,input, i1...index, s1)
|
136
|
-
r1.extend(ScientificName50)
|
137
|
-
r1.extend(ScientificName51)
|
138
|
-
else
|
139
|
-
@index = i1
|
140
|
-
r1 = nil
|
141
|
-
end
|
142
|
-
if r1
|
143
|
-
r0 = r1
|
144
|
-
else
|
145
|
-
r4 = super
|
146
|
-
if r4
|
147
|
-
r0 = r4
|
148
|
-
else
|
149
|
-
@index = i0
|
150
|
-
r0 = nil
|
151
|
-
end
|
152
|
-
end
|
153
|
-
|
154
|
-
node_cache[:scientific_name_5][start_index] = r0
|
155
|
-
|
156
|
-
r0
|
157
|
-
end
|
158
|
-
|
159
|
-
module Infraspecies0
|
160
|
-
def a
|
161
|
-
elements[0]
|
162
|
-
end
|
163
|
-
|
164
|
-
def space
|
165
|
-
elements[1]
|
166
|
-
end
|
167
|
-
|
168
|
-
def b
|
169
|
-
elements[2]
|
170
|
-
end
|
171
|
-
end
|
172
|
-
|
173
|
-
module Infraspecies1
|
174
|
-
def value
|
175
|
-
a.value + " " + b.value
|
176
|
-
end
|
177
|
-
|
178
|
-
def canonical
|
179
|
-
a.canonical
|
180
|
-
end
|
181
|
-
|
182
|
-
def pos
|
183
|
-
a.pos.merge(b.pos)
|
184
|
-
end
|
185
|
-
|
186
|
-
def details
|
187
|
-
{:infraspecies => a.details[:infraspecies].merge(b.details)}
|
188
|
-
end
|
189
|
-
end
|
190
|
-
|
191
|
-
module Infraspecies2
|
192
|
-
def a
|
193
|
-
elements[0]
|
194
|
-
end
|
195
|
-
|
196
|
-
def space1
|
197
|
-
elements[1]
|
198
|
-
end
|
199
|
-
|
200
|
-
def string_authorship_inconsistencies
|
201
|
-
elements[2]
|
202
|
-
end
|
203
|
-
|
204
|
-
def space2
|
205
|
-
elements[3]
|
206
|
-
end
|
207
|
-
|
208
|
-
def b
|
209
|
-
elements[4]
|
210
|
-
end
|
211
|
-
end
|
212
|
-
|
213
|
-
module Infraspecies3
|
214
|
-
def value
|
215
|
-
a.value + " " + b.value
|
216
|
-
end
|
217
|
-
|
218
|
-
def canonical
|
219
|
-
a.canonical
|
220
|
-
end
|
221
|
-
|
222
|
-
def pos
|
223
|
-
a.pos.merge(b.pos)
|
224
|
-
end
|
225
|
-
|
226
|
-
def details
|
227
|
-
{:infraspecies => a.details[:infraspecies].merge(b.details)}
|
228
|
-
end
|
229
|
-
end
|
230
|
-
|
231
|
-
def _nt_infraspecies
|
232
|
-
start_index = index
|
233
|
-
if node_cache[:infraspecies].has_key?(index)
|
234
|
-
cached = node_cache[:infraspecies][index]
|
235
|
-
if cached
|
236
|
-
cached = SyntaxNode.new(input, index...(index + 1)) if cached == true
|
237
|
-
@index = cached.interval.end
|
238
|
-
end
|
239
|
-
return cached
|
240
|
-
end
|
241
|
-
|
242
|
-
i0 = index
|
243
|
-
i1, s1 = index, []
|
244
|
-
r2 = _nt_infraspecies_string
|
245
|
-
s1 << r2
|
246
|
-
if r2
|
247
|
-
r3 = _nt_space
|
248
|
-
s1 << r3
|
249
|
-
if r3
|
250
|
-
r4 = _nt_year
|
251
|
-
s1 << r4
|
252
|
-
end
|
253
|
-
end
|
254
|
-
if s1.last
|
255
|
-
r1 = instantiate_node(SyntaxNode,input, i1...index, s1)
|
256
|
-
r1.extend(Infraspecies0)
|
257
|
-
r1.extend(Infraspecies1)
|
258
|
-
else
|
259
|
-
@index = i1
|
260
|
-
r1 = nil
|
261
|
-
end
|
262
|
-
if r1
|
263
|
-
r0 = r1
|
264
|
-
else
|
265
|
-
i5, s5 = index, []
|
266
|
-
r6 = _nt_infraspecies_string
|
267
|
-
s5 << r6
|
268
|
-
if r6
|
269
|
-
r7 = _nt_space
|
270
|
-
s5 << r7
|
271
|
-
if r7
|
272
|
-
r8 = _nt_string_authorship_inconsistencies
|
273
|
-
s5 << r8
|
274
|
-
if r8
|
275
|
-
r9 = _nt_space
|
276
|
-
s5 << r9
|
277
|
-
if r9
|
278
|
-
r10 = _nt_authorship
|
279
|
-
s5 << r10
|
280
|
-
end
|
281
|
-
end
|
282
|
-
end
|
283
|
-
end
|
284
|
-
if s5.last
|
285
|
-
r5 = instantiate_node(SyntaxNode,input, i5...index, s5)
|
286
|
-
r5.extend(Infraspecies2)
|
287
|
-
r5.extend(Infraspecies3)
|
288
|
-
else
|
289
|
-
@index = i5
|
290
|
-
r5 = nil
|
291
|
-
end
|
292
|
-
if r5
|
293
|
-
r0 = r5
|
294
|
-
else
|
295
|
-
r11 = super
|
296
|
-
if r11
|
297
|
-
r0 = r11
|
298
|
-
else
|
299
|
-
@index = i0
|
300
|
-
r0 = nil
|
301
|
-
end
|
302
|
-
end
|
303
|
-
end
|
304
|
-
|
305
|
-
node_cache[:infraspecies][start_index] = r0
|
306
|
-
|
307
|
-
r0
|
308
|
-
end
|
309
|
-
|
310
|
-
module Species0
|
311
|
-
def a
|
312
|
-
elements[0]
|
313
|
-
end
|
314
|
-
|
315
|
-
def space
|
316
|
-
elements[1]
|
317
|
-
end
|
318
|
-
|
319
|
-
def b
|
320
|
-
elements[2]
|
321
|
-
end
|
322
|
-
end
|
323
|
-
|
324
|
-
module Species1
|
325
|
-
def value
|
326
|
-
a.value + " " + b.value
|
327
|
-
end
|
328
|
-
|
329
|
-
def canonical
|
330
|
-
a.canonical
|
331
|
-
end
|
332
|
-
|
333
|
-
def pos
|
334
|
-
a.pos.merge(b.pos)
|
335
|
-
end
|
336
|
-
|
337
|
-
def details
|
338
|
-
{:species => a.details[:species].merge(b.details)}
|
339
|
-
end
|
340
|
-
end
|
341
|
-
|
342
|
-
def _nt_species
|
343
|
-
start_index = index
|
344
|
-
if node_cache[:species].has_key?(index)
|
345
|
-
cached = node_cache[:species][index]
|
346
|
-
if cached
|
347
|
-
cached = SyntaxNode.new(input, index...(index + 1)) if cached == true
|
348
|
-
@index = cached.interval.end
|
349
|
-
end
|
350
|
-
return cached
|
351
|
-
end
|
352
|
-
|
353
|
-
i0 = index
|
354
|
-
i1, s1 = index, []
|
355
|
-
r2 = _nt_species_string
|
356
|
-
s1 << r2
|
357
|
-
if r2
|
358
|
-
r3 = _nt_space
|
359
|
-
s1 << r3
|
360
|
-
if r3
|
361
|
-
r4 = _nt_year
|
362
|
-
s1 << r4
|
363
|
-
end
|
364
|
-
end
|
365
|
-
if s1.last
|
366
|
-
r1 = instantiate_node(SyntaxNode,input, i1...index, s1)
|
367
|
-
r1.extend(Species0)
|
368
|
-
r1.extend(Species1)
|
369
|
-
else
|
370
|
-
@index = i1
|
371
|
-
r1 = nil
|
372
|
-
end
|
373
|
-
if r1
|
374
|
-
r0 = r1
|
375
|
-
else
|
376
|
-
r5 = super
|
377
|
-
if r5
|
378
|
-
r0 = r5
|
379
|
-
else
|
380
|
-
@index = i0
|
381
|
-
r0 = nil
|
382
|
-
end
|
383
|
-
end
|
384
|
-
|
385
|
-
node_cache[:species][start_index] = r0
|
386
|
-
|
387
|
-
r0
|
388
|
-
end
|
389
|
-
|
390
|
-
module LatinWord0
|
391
|
-
def a
|
392
|
-
elements[0]
|
393
|
-
end
|
394
|
-
|
395
|
-
def b
|
396
|
-
elements[1]
|
397
|
-
end
|
398
|
-
end
|
399
|
-
|
400
|
-
module LatinWord1
|
401
|
-
def value
|
402
|
-
res = ''
|
403
|
-
text_value.split('').each do |l|
|
404
|
-
l = 'ae' if l == 'æ'
|
405
|
-
l = 'oe' if l == 'œ'
|
406
|
-
res << l
|
407
|
-
end
|
408
|
-
res.tr('àâåãäáçčéèíìïňññóòôøõöúùürŕřŗššşž',
|
409
|
-
'aaaaaacceeiiinnnoooooouuurrrrsssz')
|
410
|
-
end
|
411
|
-
end
|
412
|
-
|
413
|
-
def _nt_latin_word
|
414
|
-
start_index = index
|
415
|
-
if node_cache[:latin_word].has_key?(index)
|
416
|
-
cached = node_cache[:latin_word][index]
|
417
|
-
if cached
|
418
|
-
cached = SyntaxNode.new(input, index...(index + 1)) if cached == true
|
419
|
-
@index = cached.interval.end
|
420
|
-
end
|
421
|
-
return cached
|
422
|
-
end
|
423
|
-
|
424
|
-
i0, s0 = index, []
|
425
|
-
if has_terminal?('\G[a-z\\-ëæœàâåãäáçčéèíìïňññóòôøõöúùürŕřŗššşž]', true, index)
|
426
|
-
r1 = true
|
427
|
-
@index += 1
|
428
|
-
else
|
429
|
-
r1 = nil
|
430
|
-
end
|
431
|
-
s0 << r1
|
432
|
-
if r1
|
433
|
-
r2 = _nt_valid_name_letters
|
434
|
-
s0 << r2
|
435
|
-
end
|
436
|
-
if s0.last
|
437
|
-
r0 = instantiate_node(SyntaxNode,input, i0...index, s0)
|
438
|
-
r0.extend(LatinWord0)
|
439
|
-
r0.extend(LatinWord1)
|
440
|
-
else
|
441
|
-
@index = i0
|
442
|
-
r0 = nil
|
443
|
-
end
|
444
|
-
|
445
|
-
node_cache[:latin_word][start_index] = r0
|
446
|
-
|
447
|
-
r0
|
448
|
-
end
|
449
|
-
|
450
|
-
module ValidNameLetters0
|
451
|
-
def value
|
452
|
-
res = ''
|
453
|
-
text_value.split('').each do |l|
|
454
|
-
l = 'ae' if l == 'æ'
|
455
|
-
l = 'oe' if l == 'œ'
|
456
|
-
res << l
|
457
|
-
end
|
458
|
-
res.tr('àâåãäáçčéèíìïňññóòôøõöúùürŕřŗššşž',
|
459
|
-
'aaaaaacceeiiinnnoooooouuurrrrsssz')
|
460
|
-
end
|
461
|
-
end
|
462
|
-
|
463
|
-
def _nt_valid_name_letters
|
464
|
-
start_index = index
|
465
|
-
if node_cache[:valid_name_letters].has_key?(index)
|
466
|
-
cached = node_cache[:valid_name_letters][index]
|
467
|
-
if cached
|
468
|
-
cached = SyntaxNode.new(input, index...(index + 1)) if cached == true
|
469
|
-
@index = cached.interval.end
|
470
|
-
end
|
471
|
-
return cached
|
472
|
-
end
|
473
|
-
|
474
|
-
s0, i0 = [], index
|
475
|
-
loop do
|
476
|
-
if has_terminal?('\G[a-z\\-ëæœàâåãäáçčéèíìïňññóòôøõöúùürŕřŗššşž]', true, index)
|
477
|
-
r1 = true
|
478
|
-
@index += 1
|
479
|
-
else
|
480
|
-
r1 = nil
|
481
|
-
end
|
482
|
-
if r1
|
483
|
-
s0 << r1
|
484
|
-
else
|
485
|
-
break
|
486
|
-
end
|
487
|
-
end
|
488
|
-
if s0.empty?
|
489
|
-
@index = i0
|
490
|
-
r0 = nil
|
491
|
-
else
|
492
|
-
r0 = instantiate_node(SyntaxNode,input, i0...index, s0)
|
493
|
-
r0.extend(ValidNameLetters0)
|
494
|
-
end
|
495
|
-
|
496
|
-
node_cache[:valid_name_letters][start_index] = r0
|
497
|
-
|
498
|
-
r0
|
499
|
-
end
|
500
|
-
|
501
|
-
module ValidNameLetters0
|
502
|
-
def value
|
503
|
-
text_value
|
504
|
-
end
|
505
|
-
end
|
506
|
-
|
507
|
-
def _nt_valid_name_letters
|
508
|
-
start_index = index
|
509
|
-
if node_cache[:valid_name_letters].has_key?(index)
|
510
|
-
cached = node_cache[:valid_name_letters][index]
|
511
|
-
if cached
|
512
|
-
cached = SyntaxNode.new(input, index...(index + 1)) if cached == true
|
513
|
-
@index = cached.interval.end
|
514
|
-
end
|
515
|
-
return cached
|
516
|
-
end
|
517
|
-
|
518
|
-
s0, i0 = [], index
|
519
|
-
loop do
|
520
|
-
if has_terminal?('\G[a-z\\-ëüäöïéåóç]', true, index)
|
521
|
-
r1 = true
|
522
|
-
@index += 1
|
523
|
-
else
|
524
|
-
r1 = nil
|
525
|
-
end
|
526
|
-
if r1
|
527
|
-
s0 << r1
|
528
|
-
else
|
529
|
-
break
|
530
|
-
end
|
531
|
-
end
|
532
|
-
if s0.empty?
|
533
|
-
@index = i0
|
534
|
-
r0 = nil
|
535
|
-
else
|
536
|
-
r0 = instantiate_node(SyntaxNode,input, i0...index, s0)
|
537
|
-
r0.extend(ValidNameLetters0)
|
538
|
-
end
|
539
|
-
|
540
|
-
node_cache[:valid_name_letters][start_index] = r0
|
541
|
-
|
542
|
-
r0
|
543
|
-
end
|
544
|
-
|
545
|
-
module RightParen0
|
546
|
-
def space
|
547
|
-
elements[1]
|
548
|
-
end
|
549
|
-
|
550
|
-
end
|
551
|
-
|
552
|
-
def _nt_right_paren
|
553
|
-
start_index = index
|
554
|
-
if node_cache[:right_paren].has_key?(index)
|
555
|
-
cached = node_cache[:right_paren][index]
|
556
|
-
if cached
|
557
|
-
cached = SyntaxNode.new(input, index...(index + 1)) if cached == true
|
558
|
-
@index = cached.interval.end
|
559
|
-
end
|
560
|
-
return cached
|
561
|
-
end
|
562
|
-
|
563
|
-
i0 = index
|
564
|
-
i1, s1 = index, []
|
565
|
-
if has_terminal?(")", false, index)
|
566
|
-
r2 = instantiate_node(SyntaxNode,input, index...(index + 1))
|
567
|
-
@index += 1
|
568
|
-
else
|
569
|
-
terminal_parse_failure(")")
|
570
|
-
r2 = nil
|
571
|
-
end
|
572
|
-
s1 << r2
|
573
|
-
if r2
|
574
|
-
r3 = _nt_space
|
575
|
-
s1 << r3
|
576
|
-
if r3
|
577
|
-
if has_terminal?(")", false, index)
|
578
|
-
r4 = instantiate_node(SyntaxNode,input, index...(index + 1))
|
579
|
-
@index += 1
|
580
|
-
else
|
581
|
-
terminal_parse_failure(")")
|
582
|
-
r4 = nil
|
583
|
-
end
|
584
|
-
s1 << r4
|
585
|
-
end
|
586
|
-
end
|
587
|
-
if s1.last
|
588
|
-
r1 = instantiate_node(SyntaxNode,input, i1...index, s1)
|
589
|
-
r1.extend(RightParen0)
|
590
|
-
else
|
591
|
-
@index = i1
|
592
|
-
r1 = nil
|
593
|
-
end
|
594
|
-
if r1
|
595
|
-
r0 = r1
|
596
|
-
else
|
597
|
-
r5 = super
|
598
|
-
if r5
|
599
|
-
r0 = r5
|
600
|
-
else
|
601
|
-
@index = i0
|
602
|
-
r0 = nil
|
603
|
-
end
|
604
|
-
end
|
605
|
-
|
606
|
-
node_cache[:right_paren][start_index] = r0
|
607
|
-
|
608
|
-
r0
|
609
|
-
end
|
610
|
-
|
611
|
-
module LeftParen0
|
612
|
-
def space
|
613
|
-
elements[1]
|
614
|
-
end
|
615
|
-
|
616
|
-
end
|
617
|
-
|
618
|
-
def _nt_left_paren
|
619
|
-
start_index = index
|
620
|
-
if node_cache[:left_paren].has_key?(index)
|
621
|
-
cached = node_cache[:left_paren][index]
|
622
|
-
if cached
|
623
|
-
cached = SyntaxNode.new(input, index...(index + 1)) if cached == true
|
624
|
-
@index = cached.interval.end
|
625
|
-
end
|
626
|
-
return cached
|
627
|
-
end
|
628
|
-
|
629
|
-
i0 = index
|
630
|
-
i1, s1 = index, []
|
631
|
-
if has_terminal?("(", false, index)
|
632
|
-
r2 = instantiate_node(SyntaxNode,input, index...(index + 1))
|
633
|
-
@index += 1
|
634
|
-
else
|
635
|
-
terminal_parse_failure("(")
|
636
|
-
r2 = nil
|
637
|
-
end
|
638
|
-
s1 << r2
|
639
|
-
if r2
|
640
|
-
r3 = _nt_space
|
641
|
-
s1 << r3
|
642
|
-
if r3
|
643
|
-
if has_terminal?("(", false, index)
|
644
|
-
r4 = instantiate_node(SyntaxNode,input, index...(index + 1))
|
645
|
-
@index += 1
|
646
|
-
else
|
647
|
-
terminal_parse_failure("(")
|
648
|
-
r4 = nil
|
649
|
-
end
|
650
|
-
s1 << r4
|
651
|
-
end
|
652
|
-
end
|
653
|
-
if s1.last
|
654
|
-
r1 = instantiate_node(SyntaxNode,input, i1...index, s1)
|
655
|
-
r1.extend(LeftParen0)
|
656
|
-
else
|
657
|
-
@index = i1
|
658
|
-
r1 = nil
|
659
|
-
end
|
660
|
-
if r1
|
661
|
-
r0 = r1
|
662
|
-
else
|
663
|
-
r5 = super
|
664
|
-
if r5
|
665
|
-
r0 = r5
|
666
|
-
else
|
667
|
-
@index = i0
|
668
|
-
r0 = nil
|
669
|
-
end
|
670
|
-
end
|
671
|
-
|
672
|
-
node_cache[:left_paren][start_index] = r0
|
673
|
-
|
674
|
-
r0
|
675
|
-
end
|
676
|
-
|
677
|
-
module Year0
|
678
|
-
def a
|
679
|
-
elements[0]
|
680
|
-
end
|
681
|
-
|
682
|
-
def space
|
683
|
-
elements[1]
|
684
|
-
end
|
685
|
-
|
686
|
-
def b
|
687
|
-
elements[2]
|
688
|
-
end
|
689
|
-
end
|
690
|
-
|
691
|
-
module Year1
|
692
|
-
def value
|
693
|
-
a.value + " " + b.value
|
694
|
-
end
|
695
|
-
|
696
|
-
def pos
|
697
|
-
a.pos.merge(b.pos)
|
698
|
-
end
|
699
|
-
|
700
|
-
def details
|
701
|
-
{:year => a.value, :approximate_year => b.value}
|
702
|
-
end
|
703
|
-
end
|
704
|
-
|
705
|
-
module Year2
|
706
|
-
def a
|
707
|
-
elements[0]
|
708
|
-
end
|
709
|
-
|
710
|
-
def space
|
711
|
-
elements[1]
|
712
|
-
end
|
713
|
-
|
714
|
-
def page_number
|
715
|
-
elements[2]
|
716
|
-
end
|
717
|
-
end
|
718
|
-
|
719
|
-
module Year3
|
720
|
-
def value
|
721
|
-
a.text_value
|
722
|
-
end
|
723
|
-
|
724
|
-
def pos
|
725
|
-
{a.interval.begin => ['year', a.interval.end]}
|
726
|
-
end
|
727
|
-
|
728
|
-
def details
|
729
|
-
{:year => value}
|
730
|
-
end
|
731
|
-
end
|
732
|
-
|
733
|
-
def _nt_year
|
734
|
-
start_index = index
|
735
|
-
if node_cache[:year].has_key?(index)
|
736
|
-
cached = node_cache[:year][index]
|
737
|
-
if cached
|
738
|
-
cached = SyntaxNode.new(input, index...(index + 1)) if cached == true
|
739
|
-
@index = cached.interval.end
|
740
|
-
end
|
741
|
-
return cached
|
742
|
-
end
|
743
|
-
|
744
|
-
i0 = index
|
745
|
-
i1, s1 = index, []
|
746
|
-
r2 = _nt_year_number
|
747
|
-
s1 << r2
|
748
|
-
if r2
|
749
|
-
r3 = _nt_space
|
750
|
-
s1 << r3
|
751
|
-
if r3
|
752
|
-
r4 = _nt_approximate_year
|
753
|
-
s1 << r4
|
754
|
-
end
|
755
|
-
end
|
756
|
-
if s1.last
|
757
|
-
r1 = instantiate_node(SyntaxNode,input, i1...index, s1)
|
758
|
-
r1.extend(Year0)
|
759
|
-
r1.extend(Year1)
|
760
|
-
else
|
761
|
-
@index = i1
|
762
|
-
r1 = nil
|
763
|
-
end
|
764
|
-
if r1
|
765
|
-
r0 = r1
|
766
|
-
else
|
767
|
-
i5, s5 = index, []
|
768
|
-
r6 = _nt_year_number
|
769
|
-
s5 << r6
|
770
|
-
if r6
|
771
|
-
r7 = _nt_space
|
772
|
-
s5 << r7
|
773
|
-
if r7
|
774
|
-
r8 = _nt_page_number
|
775
|
-
s5 << r8
|
776
|
-
end
|
777
|
-
end
|
778
|
-
if s5.last
|
779
|
-
r5 = instantiate_node(SyntaxNode,input, i5...index, s5)
|
780
|
-
r5.extend(Year2)
|
781
|
-
r5.extend(Year3)
|
782
|
-
else
|
783
|
-
@index = i5
|
784
|
-
r5 = nil
|
785
|
-
end
|
786
|
-
if r5
|
787
|
-
r0 = r5
|
788
|
-
else
|
789
|
-
r9 = _nt_year_number_with_punctuation
|
790
|
-
if r9
|
791
|
-
r0 = r9
|
792
|
-
else
|
793
|
-
r10 = _nt_approximate_year
|
794
|
-
if r10
|
795
|
-
r0 = r10
|
796
|
-
else
|
797
|
-
r11 = _nt_double_year
|
798
|
-
if r11
|
799
|
-
r0 = r11
|
800
|
-
else
|
801
|
-
r12 = super
|
802
|
-
if r12
|
803
|
-
r0 = r12
|
804
|
-
else
|
805
|
-
@index = i0
|
806
|
-
r0 = nil
|
807
|
-
end
|
808
|
-
end
|
809
|
-
end
|
810
|
-
end
|
811
|
-
end
|
812
|
-
end
|
813
|
-
|
814
|
-
node_cache[:year][start_index] = r0
|
815
|
-
|
816
|
-
r0
|
817
|
-
end
|
818
|
-
|
819
|
-
module ApproximateYear0
|
820
|
-
def space1
|
821
|
-
elements[1]
|
822
|
-
end
|
823
|
-
|
824
|
-
def a
|
825
|
-
elements[2]
|
826
|
-
end
|
827
|
-
|
828
|
-
def space2
|
829
|
-
elements[3]
|
830
|
-
end
|
831
|
-
|
832
|
-
end
|
833
|
-
|
834
|
-
module ApproximateYear1
|
835
|
-
def value
|
836
|
-
"(" + a.text_value + ")"
|
837
|
-
end
|
838
|
-
|
839
|
-
def pos
|
840
|
-
{a.interval.begin => ['year', a.interval.end]}
|
841
|
-
end
|
842
|
-
|
843
|
-
def details
|
844
|
-
{:approximate_year => value}
|
845
|
-
end
|
846
|
-
end
|
847
|
-
|
848
|
-
def _nt_approximate_year
|
849
|
-
start_index = index
|
850
|
-
if node_cache[:approximate_year].has_key?(index)
|
851
|
-
cached = node_cache[:approximate_year][index]
|
852
|
-
if cached
|
853
|
-
cached = SyntaxNode.new(input, index...(index + 1)) if cached == true
|
854
|
-
@index = cached.interval.end
|
855
|
-
end
|
856
|
-
return cached
|
857
|
-
end
|
858
|
-
|
859
|
-
i0, s0 = index, []
|
860
|
-
if has_terminal?("[", false, index)
|
861
|
-
r1 = instantiate_node(SyntaxNode,input, index...(index + 1))
|
862
|
-
@index += 1
|
863
|
-
else
|
864
|
-
terminal_parse_failure("[")
|
865
|
-
r1 = nil
|
866
|
-
end
|
867
|
-
s0 << r1
|
868
|
-
if r1
|
869
|
-
r2 = _nt_space
|
870
|
-
s0 << r2
|
871
|
-
if r2
|
872
|
-
r3 = _nt_year_number
|
873
|
-
s0 << r3
|
874
|
-
if r3
|
875
|
-
r4 = _nt_space
|
876
|
-
s0 << r4
|
877
|
-
if r4
|
878
|
-
s5, i5 = [], index
|
879
|
-
loop do
|
880
|
-
if has_terminal?("]", false, index)
|
881
|
-
r6 = instantiate_node(SyntaxNode,input, index...(index + 1))
|
882
|
-
@index += 1
|
883
|
-
else
|
884
|
-
terminal_parse_failure("]")
|
885
|
-
r6 = nil
|
886
|
-
end
|
887
|
-
if r6
|
888
|
-
s5 << r6
|
889
|
-
else
|
890
|
-
break
|
891
|
-
end
|
892
|
-
end
|
893
|
-
if s5.empty?
|
894
|
-
@index = i5
|
895
|
-
r5 = nil
|
896
|
-
else
|
897
|
-
r5 = instantiate_node(SyntaxNode,input, i5...index, s5)
|
898
|
-
end
|
899
|
-
s0 << r5
|
900
|
-
end
|
901
|
-
end
|
902
|
-
end
|
903
|
-
end
|
904
|
-
if s0.last
|
905
|
-
r0 = instantiate_node(SyntaxNode,input, i0...index, s0)
|
906
|
-
r0.extend(ApproximateYear0)
|
907
|
-
r0.extend(ApproximateYear1)
|
908
|
-
else
|
909
|
-
@index = i0
|
910
|
-
r0 = nil
|
911
|
-
end
|
912
|
-
|
913
|
-
node_cache[:approximate_year][start_index] = r0
|
914
|
-
|
915
|
-
r0
|
916
|
-
end
|
917
|
-
|
918
|
-
module DoubleYear0
|
919
|
-
def year_number
|
920
|
-
elements[0]
|
921
|
-
end
|
922
|
-
|
923
|
-
end
|
924
|
-
|
925
|
-
module DoubleYear1
|
926
|
-
def value
|
927
|
-
text_value
|
928
|
-
end
|
929
|
-
|
930
|
-
def pos
|
931
|
-
{interval.begin => ['year', interval.end]}
|
932
|
-
end
|
933
|
-
|
934
|
-
def details
|
935
|
-
{:year => value}
|
936
|
-
end
|
937
|
-
end
|
938
|
-
|
939
|
-
def _nt_double_year
|
940
|
-
start_index = index
|
941
|
-
if node_cache[:double_year].has_key?(index)
|
942
|
-
cached = node_cache[:double_year][index]
|
943
|
-
if cached
|
944
|
-
cached = SyntaxNode.new(input, index...(index + 1)) if cached == true
|
945
|
-
@index = cached.interval.end
|
946
|
-
end
|
947
|
-
return cached
|
948
|
-
end
|
949
|
-
|
950
|
-
i0, s0 = index, []
|
951
|
-
r1 = _nt_year_number
|
952
|
-
s0 << r1
|
953
|
-
if r1
|
954
|
-
if has_terminal?("-", false, index)
|
955
|
-
r2 = instantiate_node(SyntaxNode,input, index...(index + 1))
|
956
|
-
@index += 1
|
957
|
-
else
|
958
|
-
terminal_parse_failure("-")
|
959
|
-
r2 = nil
|
960
|
-
end
|
961
|
-
s0 << r2
|
962
|
-
if r2
|
963
|
-
s3, i3 = [], index
|
964
|
-
loop do
|
965
|
-
if has_terminal?('\G[0-9]', true, index)
|
966
|
-
r4 = true
|
967
|
-
@index += 1
|
968
|
-
else
|
969
|
-
r4 = nil
|
970
|
-
end
|
971
|
-
if r4
|
972
|
-
s3 << r4
|
973
|
-
else
|
974
|
-
break
|
975
|
-
end
|
976
|
-
end
|
977
|
-
if s3.empty?
|
978
|
-
@index = i3
|
979
|
-
r3 = nil
|
980
|
-
else
|
981
|
-
r3 = instantiate_node(SyntaxNode,input, i3...index, s3)
|
982
|
-
end
|
983
|
-
s0 << r3
|
984
|
-
if r3
|
985
|
-
if has_terminal?('\G[A-Za-z]', true, index)
|
986
|
-
r6 = true
|
987
|
-
@index += 1
|
988
|
-
else
|
989
|
-
r6 = nil
|
990
|
-
end
|
991
|
-
if r6
|
992
|
-
r5 = r6
|
993
|
-
else
|
994
|
-
r5 = instantiate_node(SyntaxNode,input, index...index)
|
995
|
-
end
|
996
|
-
s0 << r5
|
997
|
-
if r5
|
998
|
-
if has_terminal?('\G[\\?]', true, index)
|
999
|
-
r8 = true
|
1000
|
-
@index += 1
|
1001
|
-
else
|
1002
|
-
r8 = nil
|
1003
|
-
end
|
1004
|
-
if r8
|
1005
|
-
r7 = r8
|
1006
|
-
else
|
1007
|
-
r7 = instantiate_node(SyntaxNode,input, index...index)
|
1008
|
-
end
|
1009
|
-
s0 << r7
|
1010
|
-
end
|
1011
|
-
end
|
1012
|
-
end
|
1013
|
-
end
|
1014
|
-
if s0.last
|
1015
|
-
r0 = instantiate_node(SyntaxNode,input, i0...index, s0)
|
1016
|
-
r0.extend(DoubleYear0)
|
1017
|
-
r0.extend(DoubleYear1)
|
1018
|
-
else
|
1019
|
-
@index = i0
|
1020
|
-
r0 = nil
|
1021
|
-
end
|
1022
|
-
|
1023
|
-
node_cache[:double_year][start_index] = r0
|
1024
|
-
|
1025
|
-
r0
|
1026
|
-
end
|
1027
|
-
|
1028
|
-
module YearNumberWithPunctuation0
|
1029
|
-
def a
|
1030
|
-
elements[0]
|
1031
|
-
end
|
1032
|
-
|
1033
|
-
end
|
1034
|
-
|
1035
|
-
module YearNumberWithPunctuation1
|
1036
|
-
def value
|
1037
|
-
a.text_value
|
1038
|
-
end
|
1039
|
-
|
1040
|
-
def pos
|
1041
|
-
{interval.begin => ['year', interval.end]}
|
1042
|
-
end
|
1043
|
-
|
1044
|
-
def details
|
1045
|
-
{:year => value}
|
1046
|
-
end
|
1047
|
-
end
|
1048
|
-
|
1049
|
-
def _nt_year_number_with_punctuation
|
1050
|
-
start_index = index
|
1051
|
-
if node_cache[:year_number_with_punctuation].has_key?(index)
|
1052
|
-
cached = node_cache[:year_number_with_punctuation][index]
|
1053
|
-
if cached
|
1054
|
-
cached = SyntaxNode.new(input, index...(index + 1)) if cached == true
|
1055
|
-
@index = cached.interval.end
|
1056
|
-
end
|
1057
|
-
return cached
|
1058
|
-
end
|
1059
|
-
|
1060
|
-
i0, s0 = index, []
|
1061
|
-
r1 = _nt_year_number
|
1062
|
-
s0 << r1
|
1063
|
-
if r1
|
1064
|
-
if has_terminal?(".", false, index)
|
1065
|
-
r2 = instantiate_node(SyntaxNode,input, index...(index + 1))
|
1066
|
-
@index += 1
|
1067
|
-
else
|
1068
|
-
terminal_parse_failure(".")
|
1069
|
-
r2 = nil
|
1070
|
-
end
|
1071
|
-
s0 << r2
|
1072
|
-
end
|
1073
|
-
if s0.last
|
1074
|
-
r0 = instantiate_node(SyntaxNode,input, i0...index, s0)
|
1075
|
-
r0.extend(YearNumberWithPunctuation0)
|
1076
|
-
r0.extend(YearNumberWithPunctuation1)
|
1077
|
-
else
|
1078
|
-
@index = i0
|
1079
|
-
r0 = nil
|
1080
|
-
end
|
1081
|
-
|
1082
|
-
node_cache[:year_number_with_punctuation][start_index] = r0
|
1083
|
-
|
1084
|
-
r0
|
1085
|
-
end
|
1086
|
-
|
1087
|
-
module PageNumber0
|
1088
|
-
def space
|
1089
|
-
elements[1]
|
1090
|
-
end
|
1091
|
-
|
1092
|
-
end
|
1093
|
-
|
1094
|
-
module PageNumber1
|
1095
|
-
def value
|
1096
|
-
end
|
1097
|
-
end
|
1098
|
-
|
1099
|
-
def _nt_page_number
|
1100
|
-
start_index = index
|
1101
|
-
if node_cache[:page_number].has_key?(index)
|
1102
|
-
cached = node_cache[:page_number][index]
|
1103
|
-
if cached
|
1104
|
-
cached = SyntaxNode.new(input, index...(index + 1)) if cached == true
|
1105
|
-
@index = cached.interval.end
|
1106
|
-
end
|
1107
|
-
return cached
|
1108
|
-
end
|
1109
|
-
|
1110
|
-
i0, s0 = index, []
|
1111
|
-
if has_terminal?(":", false, index)
|
1112
|
-
r1 = instantiate_node(SyntaxNode,input, index...(index + 1))
|
1113
|
-
@index += 1
|
1114
|
-
else
|
1115
|
-
terminal_parse_failure(":")
|
1116
|
-
r1 = nil
|
1117
|
-
end
|
1118
|
-
s0 << r1
|
1119
|
-
if r1
|
1120
|
-
r2 = _nt_space
|
1121
|
-
s0 << r2
|
1122
|
-
if r2
|
1123
|
-
s3, i3 = [], index
|
1124
|
-
loop do
|
1125
|
-
if has_terminal?('\G[\\d]', true, index)
|
1126
|
-
r4 = true
|
1127
|
-
@index += 1
|
1128
|
-
else
|
1129
|
-
r4 = nil
|
1130
|
-
end
|
1131
|
-
if r4
|
1132
|
-
s3 << r4
|
1133
|
-
else
|
1134
|
-
break
|
1135
|
-
end
|
1136
|
-
end
|
1137
|
-
if s3.empty?
|
1138
|
-
@index = i3
|
1139
|
-
r3 = nil
|
1140
|
-
else
|
1141
|
-
r3 = instantiate_node(SyntaxNode,input, i3...index, s3)
|
1142
|
-
end
|
1143
|
-
s0 << r3
|
1144
|
-
end
|
1145
|
-
end
|
1146
|
-
if s0.last
|
1147
|
-
r0 = instantiate_node(SyntaxNode,input, i0...index, s0)
|
1148
|
-
r0.extend(PageNumber0)
|
1149
|
-
r0.extend(PageNumber1)
|
1150
|
-
else
|
1151
|
-
@index = i0
|
1152
|
-
r0 = nil
|
1153
|
-
end
|
1154
|
-
|
1155
|
-
node_cache[:page_number][start_index] = r0
|
1156
|
-
|
1157
|
-
r0
|
1158
|
-
end
|
1159
|
-
|
1160
|
-
def _nt_string_authorship_inconsistencies
|
1161
|
-
start_index = index
|
1162
|
-
if node_cache[:string_authorship_inconsistencies].has_key?(index)
|
1163
|
-
cached = node_cache[:string_authorship_inconsistencies][index]
|
1164
|
-
if cached
|
1165
|
-
cached = SyntaxNode.new(input, index...(index + 1)) if cached == true
|
1166
|
-
@index = cached.interval.end
|
1167
|
-
end
|
1168
|
-
return cached
|
1169
|
-
end
|
1170
|
-
|
1171
|
-
if has_terminal?("corrig.", false, index)
|
1172
|
-
r0 = instantiate_node(SyntaxNode,input, index...(index + 7))
|
1173
|
-
@index += 7
|
1174
|
-
else
|
1175
|
-
terminal_parse_failure("corrig.")
|
1176
|
-
r0 = nil
|
1177
|
-
end
|
1178
|
-
|
1179
|
-
node_cache[:string_authorship_inconsistencies][start_index] = r0
|
1180
|
-
|
1181
|
-
r0
|
1182
|
-
end
|
1183
|
-
|
1184
|
-
module Garbage0
|
1185
|
-
def space1
|
1186
|
-
elements[0]
|
1187
|
-
end
|
1188
|
-
|
1189
|
-
def space2
|
1190
|
-
elements[2]
|
1191
|
-
end
|
1192
|
-
|
1193
|
-
end
|
1194
|
-
|
1195
|
-
module Garbage1
|
1196
|
-
def space_hard
|
1197
|
-
elements[0]
|
1198
|
-
end
|
1199
|
-
|
1200
|
-
end
|
1201
|
-
|
1202
|
-
def _nt_garbage
|
1203
|
-
start_index = index
|
1204
|
-
if node_cache[:garbage].has_key?(index)
|
1205
|
-
cached = node_cache[:garbage][index]
|
1206
|
-
if cached
|
1207
|
-
cached = SyntaxNode.new(input, index...(index + 1)) if cached == true
|
1208
|
-
@index = cached.interval.end
|
1209
|
-
end
|
1210
|
-
return cached
|
1211
|
-
end
|
1212
|
-
|
1213
|
-
i0 = index
|
1214
|
-
i1, s1 = index, []
|
1215
|
-
r2 = _nt_space
|
1216
|
-
s1 << r2
|
1217
|
-
if r2
|
1218
|
-
if has_terminal?('\G["\',.]', true, index)
|
1219
|
-
r3 = true
|
1220
|
-
@index += 1
|
1221
|
-
else
|
1222
|
-
r3 = nil
|
1223
|
-
end
|
1224
|
-
s1 << r3
|
1225
|
-
if r3
|
1226
|
-
r4 = _nt_space
|
1227
|
-
s1 << r4
|
1228
|
-
if r4
|
1229
|
-
s5, i5 = [], index
|
1230
|
-
loop do
|
1231
|
-
if has_terminal?('\G[^щ]', true, index)
|
1232
|
-
r6 = true
|
1233
|
-
@index += 1
|
1234
|
-
else
|
1235
|
-
r6 = nil
|
1236
|
-
end
|
1237
|
-
if r6
|
1238
|
-
s5 << r6
|
1239
|
-
else
|
1240
|
-
break
|
1241
|
-
end
|
1242
|
-
end
|
1243
|
-
r5 = instantiate_node(SyntaxNode,input, i5...index, s5)
|
1244
|
-
s1 << r5
|
1245
|
-
end
|
1246
|
-
end
|
1247
|
-
end
|
1248
|
-
if s1.last
|
1249
|
-
r1 = instantiate_node(SyntaxNode,input, i1...index, s1)
|
1250
|
-
r1.extend(Garbage0)
|
1251
|
-
else
|
1252
|
-
@index = i1
|
1253
|
-
r1 = nil
|
1254
|
-
end
|
1255
|
-
if r1
|
1256
|
-
r0 = r1
|
1257
|
-
else
|
1258
|
-
i7, s7 = index, []
|
1259
|
-
r8 = _nt_space_hard
|
1260
|
-
s7 << r8
|
1261
|
-
if r8
|
1262
|
-
s9, i9 = [], index
|
1263
|
-
loop do
|
1264
|
-
if has_terminal?('\G[^ш]', true, index)
|
1265
|
-
r10 = true
|
1266
|
-
@index += 1
|
1267
|
-
else
|
1268
|
-
r10 = nil
|
1269
|
-
end
|
1270
|
-
if r10
|
1271
|
-
s9 << r10
|
1272
|
-
else
|
1273
|
-
break
|
1274
|
-
end
|
1275
|
-
end
|
1276
|
-
if s9.empty?
|
1277
|
-
@index = i9
|
1278
|
-
r9 = nil
|
1279
|
-
else
|
1280
|
-
r9 = instantiate_node(SyntaxNode,input, i9...index, s9)
|
1281
|
-
end
|
1282
|
-
s7 << r9
|
1283
|
-
end
|
1284
|
-
if s7.last
|
1285
|
-
r7 = instantiate_node(SyntaxNode,input, i7...index, s7)
|
1286
|
-
r7.extend(Garbage1)
|
1287
|
-
else
|
1288
|
-
@index = i7
|
1289
|
-
r7 = nil
|
1290
|
-
end
|
1291
|
-
if r7
|
1292
|
-
r0 = r7
|
1293
|
-
else
|
1294
|
-
@index = i0
|
1295
|
-
r0 = nil
|
1296
|
-
end
|
1297
|
-
end
|
1298
|
-
|
1299
|
-
node_cache[:garbage][start_index] = r0
|
1300
|
-
|
1301
|
-
r0
|
1302
|
-
end
|
1303
|
-
|
1304
|
-
end
|
1305
|
-
|
1306
|
-
class ScientificNameDirtyParser < Treetop::Runtime::CompiledParser
|
1307
|
-
include ScientificNameDirty
|
1308
|
-
end
|
1309
|
-
|