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
|
@@ -1,481 +0,0 @@
|
|
|
1
|
-
# encoding: UTF-8
|
|
2
|
-
module ScientificNameCanonical
|
|
3
|
-
include Treetop::Runtime
|
|
4
|
-
|
|
5
|
-
def root
|
|
6
|
-
@root || :root
|
|
7
|
-
end
|
|
8
|
-
|
|
9
|
-
include ScientificNameDirty
|
|
10
|
-
|
|
11
|
-
module Root0
|
|
12
|
-
def hybrid
|
|
13
|
-
false
|
|
14
|
-
end
|
|
15
|
-
|
|
16
|
-
def details
|
|
17
|
-
[super]
|
|
18
|
-
end
|
|
19
|
-
|
|
20
|
-
def parser_run
|
|
21
|
-
3
|
|
22
|
-
end
|
|
23
|
-
end
|
|
24
|
-
|
|
25
|
-
module Root1
|
|
26
|
-
def hybrid
|
|
27
|
-
false
|
|
28
|
-
end
|
|
29
|
-
|
|
30
|
-
def details
|
|
31
|
-
[super]
|
|
32
|
-
end
|
|
33
|
-
|
|
34
|
-
def parser_run
|
|
35
|
-
3
|
|
36
|
-
end
|
|
37
|
-
end
|
|
38
|
-
|
|
39
|
-
def _nt_root
|
|
40
|
-
start_index = index
|
|
41
|
-
if node_cache[:root].has_key?(index)
|
|
42
|
-
cached = node_cache[:root][index]
|
|
43
|
-
if cached
|
|
44
|
-
cached = SyntaxNode.new(input, index...(index + 1)) if cached == true
|
|
45
|
-
@index = cached.interval.end
|
|
46
|
-
end
|
|
47
|
-
return cached
|
|
48
|
-
end
|
|
49
|
-
|
|
50
|
-
i0 = index
|
|
51
|
-
r1 = _nt_multinomial_with_garbage
|
|
52
|
-
r1.extend(Root0)
|
|
53
|
-
if r1
|
|
54
|
-
r0 = r1
|
|
55
|
-
else
|
|
56
|
-
r2 = _nt_uninomial_with_garbage
|
|
57
|
-
r2.extend(Root1)
|
|
58
|
-
if r2
|
|
59
|
-
r0 = r2
|
|
60
|
-
else
|
|
61
|
-
@index = i0
|
|
62
|
-
r0 = nil
|
|
63
|
-
end
|
|
64
|
-
end
|
|
65
|
-
|
|
66
|
-
node_cache[:root][start_index] = r0
|
|
67
|
-
|
|
68
|
-
r0
|
|
69
|
-
end
|
|
70
|
-
|
|
71
|
-
module MultinomialWithGarbage0
|
|
72
|
-
def a
|
|
73
|
-
elements[0]
|
|
74
|
-
end
|
|
75
|
-
|
|
76
|
-
def space1
|
|
77
|
-
elements[1]
|
|
78
|
-
end
|
|
79
|
-
|
|
80
|
-
def b
|
|
81
|
-
elements[2]
|
|
82
|
-
end
|
|
83
|
-
|
|
84
|
-
def space2
|
|
85
|
-
elements[3]
|
|
86
|
-
end
|
|
87
|
-
|
|
88
|
-
def c
|
|
89
|
-
elements[4]
|
|
90
|
-
end
|
|
91
|
-
|
|
92
|
-
def garbage
|
|
93
|
-
elements[5]
|
|
94
|
-
end
|
|
95
|
-
end
|
|
96
|
-
|
|
97
|
-
module MultinomialWithGarbage1
|
|
98
|
-
def value
|
|
99
|
-
a.value + " " + b.value + " " + c.value
|
|
100
|
-
end
|
|
101
|
-
|
|
102
|
-
def canonical
|
|
103
|
-
a.canonical + " " + b.canonical + " " + c.canonical
|
|
104
|
-
end
|
|
105
|
-
|
|
106
|
-
def pos
|
|
107
|
-
a.pos.merge(b.pos).merge(c.pos)
|
|
108
|
-
end
|
|
109
|
-
|
|
110
|
-
def details
|
|
111
|
-
a.details.merge(b.details).merge(c.details)
|
|
112
|
-
end
|
|
113
|
-
end
|
|
114
|
-
|
|
115
|
-
module MultinomialWithGarbage2
|
|
116
|
-
def a
|
|
117
|
-
elements[0]
|
|
118
|
-
end
|
|
119
|
-
|
|
120
|
-
def space
|
|
121
|
-
elements[1]
|
|
122
|
-
end
|
|
123
|
-
|
|
124
|
-
def b
|
|
125
|
-
elements[2]
|
|
126
|
-
end
|
|
127
|
-
|
|
128
|
-
def garbage
|
|
129
|
-
elements[3]
|
|
130
|
-
end
|
|
131
|
-
end
|
|
132
|
-
|
|
133
|
-
module MultinomialWithGarbage3
|
|
134
|
-
def value
|
|
135
|
-
a.value + " " + b.value
|
|
136
|
-
end
|
|
137
|
-
|
|
138
|
-
def canonical
|
|
139
|
-
a.canonical + " " + b.canonical
|
|
140
|
-
end
|
|
141
|
-
|
|
142
|
-
def pos
|
|
143
|
-
a.pos.merge(b.pos)
|
|
144
|
-
end
|
|
145
|
-
|
|
146
|
-
def details
|
|
147
|
-
a.details.merge(b.details)
|
|
148
|
-
end
|
|
149
|
-
end
|
|
150
|
-
|
|
151
|
-
module MultinomialWithGarbage4
|
|
152
|
-
def a
|
|
153
|
-
elements[0]
|
|
154
|
-
end
|
|
155
|
-
|
|
156
|
-
def space
|
|
157
|
-
elements[1]
|
|
158
|
-
end
|
|
159
|
-
|
|
160
|
-
def b
|
|
161
|
-
elements[2]
|
|
162
|
-
end
|
|
163
|
-
|
|
164
|
-
def garbage
|
|
165
|
-
elements[3]
|
|
166
|
-
end
|
|
167
|
-
end
|
|
168
|
-
|
|
169
|
-
module MultinomialWithGarbage5
|
|
170
|
-
def value
|
|
171
|
-
a.value + " " + b.value
|
|
172
|
-
end
|
|
173
|
-
|
|
174
|
-
def canonical
|
|
175
|
-
a.canonical + " " + b.canonical
|
|
176
|
-
end
|
|
177
|
-
|
|
178
|
-
def pos
|
|
179
|
-
a.pos.merge(b.pos)
|
|
180
|
-
end
|
|
181
|
-
|
|
182
|
-
def details
|
|
183
|
-
a.details.merge(b.details)
|
|
184
|
-
end
|
|
185
|
-
end
|
|
186
|
-
|
|
187
|
-
def _nt_multinomial_with_garbage
|
|
188
|
-
start_index = index
|
|
189
|
-
if node_cache[:multinomial_with_garbage].has_key?(index)
|
|
190
|
-
cached = node_cache[:multinomial_with_garbage][index]
|
|
191
|
-
if cached
|
|
192
|
-
cached = SyntaxNode.new(input, index...(index + 1)) if cached == true
|
|
193
|
-
@index = cached.interval.end
|
|
194
|
-
end
|
|
195
|
-
return cached
|
|
196
|
-
end
|
|
197
|
-
|
|
198
|
-
i0 = index
|
|
199
|
-
i1, s1 = index, []
|
|
200
|
-
r2 = _nt_genus
|
|
201
|
-
s1 << r2
|
|
202
|
-
if r2
|
|
203
|
-
r3 = _nt_space
|
|
204
|
-
s1 << r3
|
|
205
|
-
if r3
|
|
206
|
-
r4 = _nt_infragenus
|
|
207
|
-
s1 << r4
|
|
208
|
-
if r4
|
|
209
|
-
r5 = _nt_space
|
|
210
|
-
s1 << r5
|
|
211
|
-
if r5
|
|
212
|
-
r6 = _nt_species
|
|
213
|
-
s1 << r6
|
|
214
|
-
if r6
|
|
215
|
-
r7 = _nt_garbage
|
|
216
|
-
s1 << r7
|
|
217
|
-
end
|
|
218
|
-
end
|
|
219
|
-
end
|
|
220
|
-
end
|
|
221
|
-
end
|
|
222
|
-
if s1.last
|
|
223
|
-
r1 = instantiate_node(SyntaxNode,input, i1...index, s1)
|
|
224
|
-
r1.extend(MultinomialWithGarbage0)
|
|
225
|
-
r1.extend(MultinomialWithGarbage1)
|
|
226
|
-
else
|
|
227
|
-
@index = i1
|
|
228
|
-
r1 = nil
|
|
229
|
-
end
|
|
230
|
-
if r1
|
|
231
|
-
r0 = r1
|
|
232
|
-
else
|
|
233
|
-
i8, s8 = index, []
|
|
234
|
-
r9 = _nt_genus
|
|
235
|
-
s8 << r9
|
|
236
|
-
if r9
|
|
237
|
-
r10 = _nt_space
|
|
238
|
-
s8 << r10
|
|
239
|
-
if r10
|
|
240
|
-
r11 = _nt_infragenus
|
|
241
|
-
s8 << r11
|
|
242
|
-
if r11
|
|
243
|
-
r12 = _nt_garbage
|
|
244
|
-
s8 << r12
|
|
245
|
-
end
|
|
246
|
-
end
|
|
247
|
-
end
|
|
248
|
-
if s8.last
|
|
249
|
-
r8 = instantiate_node(SyntaxNode,input, i8...index, s8)
|
|
250
|
-
r8.extend(MultinomialWithGarbage2)
|
|
251
|
-
r8.extend(MultinomialWithGarbage3)
|
|
252
|
-
else
|
|
253
|
-
@index = i8
|
|
254
|
-
r8 = nil
|
|
255
|
-
end
|
|
256
|
-
if r8
|
|
257
|
-
r0 = r8
|
|
258
|
-
else
|
|
259
|
-
i13, s13 = index, []
|
|
260
|
-
r14 = _nt_genus
|
|
261
|
-
s13 << r14
|
|
262
|
-
if r14
|
|
263
|
-
r15 = _nt_space
|
|
264
|
-
s13 << r15
|
|
265
|
-
if r15
|
|
266
|
-
r16 = _nt_species
|
|
267
|
-
s13 << r16
|
|
268
|
-
if r16
|
|
269
|
-
r17 = _nt_garbage
|
|
270
|
-
s13 << r17
|
|
271
|
-
end
|
|
272
|
-
end
|
|
273
|
-
end
|
|
274
|
-
if s13.last
|
|
275
|
-
r13 = instantiate_node(SyntaxNode,input, i13...index, s13)
|
|
276
|
-
r13.extend(MultinomialWithGarbage4)
|
|
277
|
-
r13.extend(MultinomialWithGarbage5)
|
|
278
|
-
else
|
|
279
|
-
@index = i13
|
|
280
|
-
r13 = nil
|
|
281
|
-
end
|
|
282
|
-
if r13
|
|
283
|
-
r0 = r13
|
|
284
|
-
else
|
|
285
|
-
@index = i0
|
|
286
|
-
r0 = nil
|
|
287
|
-
end
|
|
288
|
-
end
|
|
289
|
-
end
|
|
290
|
-
|
|
291
|
-
node_cache[:multinomial_with_garbage][start_index] = r0
|
|
292
|
-
|
|
293
|
-
r0
|
|
294
|
-
end
|
|
295
|
-
|
|
296
|
-
module UninomialWithGarbage0
|
|
297
|
-
def a
|
|
298
|
-
elements[0]
|
|
299
|
-
end
|
|
300
|
-
|
|
301
|
-
def b
|
|
302
|
-
elements[1]
|
|
303
|
-
end
|
|
304
|
-
end
|
|
305
|
-
|
|
306
|
-
module UninomialWithGarbage1
|
|
307
|
-
def value
|
|
308
|
-
a.value
|
|
309
|
-
end
|
|
310
|
-
|
|
311
|
-
def canonical
|
|
312
|
-
a.canonical
|
|
313
|
-
end
|
|
314
|
-
|
|
315
|
-
def pos
|
|
316
|
-
a.pos
|
|
317
|
-
end
|
|
318
|
-
|
|
319
|
-
def details
|
|
320
|
-
{:uninomial => a.details[:uninomial]}
|
|
321
|
-
end
|
|
322
|
-
end
|
|
323
|
-
|
|
324
|
-
def _nt_uninomial_with_garbage
|
|
325
|
-
start_index = index
|
|
326
|
-
if node_cache[:uninomial_with_garbage].has_key?(index)
|
|
327
|
-
cached = node_cache[:uninomial_with_garbage][index]
|
|
328
|
-
if cached
|
|
329
|
-
cached = SyntaxNode.new(input, index...(index + 1)) if cached == true
|
|
330
|
-
@index = cached.interval.end
|
|
331
|
-
end
|
|
332
|
-
return cached
|
|
333
|
-
end
|
|
334
|
-
|
|
335
|
-
i0, s0 = index, []
|
|
336
|
-
r1 = _nt_uninomial_string
|
|
337
|
-
s0 << r1
|
|
338
|
-
if r1
|
|
339
|
-
r2 = _nt_garbage
|
|
340
|
-
s0 << r2
|
|
341
|
-
end
|
|
342
|
-
if s0.last
|
|
343
|
-
r0 = instantiate_node(SyntaxNode,input, i0...index, s0)
|
|
344
|
-
r0.extend(UninomialWithGarbage0)
|
|
345
|
-
r0.extend(UninomialWithGarbage1)
|
|
346
|
-
else
|
|
347
|
-
@index = i0
|
|
348
|
-
r0 = nil
|
|
349
|
-
end
|
|
350
|
-
|
|
351
|
-
node_cache[:uninomial_with_garbage][start_index] = r0
|
|
352
|
-
|
|
353
|
-
r0
|
|
354
|
-
end
|
|
355
|
-
|
|
356
|
-
module Garbage0
|
|
357
|
-
def space1
|
|
358
|
-
elements[0]
|
|
359
|
-
end
|
|
360
|
-
|
|
361
|
-
def space2
|
|
362
|
-
elements[2]
|
|
363
|
-
end
|
|
364
|
-
|
|
365
|
-
end
|
|
366
|
-
|
|
367
|
-
module Garbage1
|
|
368
|
-
def space_hard
|
|
369
|
-
elements[0]
|
|
370
|
-
end
|
|
371
|
-
|
|
372
|
-
end
|
|
373
|
-
|
|
374
|
-
def _nt_garbage
|
|
375
|
-
start_index = index
|
|
376
|
-
if node_cache[:garbage].has_key?(index)
|
|
377
|
-
cached = node_cache[:garbage][index]
|
|
378
|
-
if cached
|
|
379
|
-
cached = SyntaxNode.new(input, index...(index + 1)) if cached == true
|
|
380
|
-
@index = cached.interval.end
|
|
381
|
-
end
|
|
382
|
-
return cached
|
|
383
|
-
end
|
|
384
|
-
|
|
385
|
-
i0 = index
|
|
386
|
-
i1, s1 = index, []
|
|
387
|
-
r2 = _nt_space
|
|
388
|
-
s1 << r2
|
|
389
|
-
if r2
|
|
390
|
-
if has_terminal?('\G["\',.]', true, index)
|
|
391
|
-
r3 = true
|
|
392
|
-
@index += 1
|
|
393
|
-
else
|
|
394
|
-
r3 = nil
|
|
395
|
-
end
|
|
396
|
-
s1 << r3
|
|
397
|
-
if r3
|
|
398
|
-
r4 = _nt_space
|
|
399
|
-
s1 << r4
|
|
400
|
-
if r4
|
|
401
|
-
s5, i5 = [], index
|
|
402
|
-
loop do
|
|
403
|
-
if has_terminal?('\G[^щ]', true, index)
|
|
404
|
-
r6 = true
|
|
405
|
-
@index += 1
|
|
406
|
-
else
|
|
407
|
-
r6 = nil
|
|
408
|
-
end
|
|
409
|
-
if r6
|
|
410
|
-
s5 << r6
|
|
411
|
-
else
|
|
412
|
-
break
|
|
413
|
-
end
|
|
414
|
-
end
|
|
415
|
-
r5 = instantiate_node(SyntaxNode,input, i5...index, s5)
|
|
416
|
-
s1 << r5
|
|
417
|
-
end
|
|
418
|
-
end
|
|
419
|
-
end
|
|
420
|
-
if s1.last
|
|
421
|
-
r1 = instantiate_node(SyntaxNode,input, i1...index, s1)
|
|
422
|
-
r1.extend(Garbage0)
|
|
423
|
-
else
|
|
424
|
-
@index = i1
|
|
425
|
-
r1 = nil
|
|
426
|
-
end
|
|
427
|
-
if r1
|
|
428
|
-
r0 = r1
|
|
429
|
-
else
|
|
430
|
-
i7, s7 = index, []
|
|
431
|
-
r8 = _nt_space_hard
|
|
432
|
-
s7 << r8
|
|
433
|
-
if r8
|
|
434
|
-
s9, i9 = [], index
|
|
435
|
-
loop do
|
|
436
|
-
if has_terminal?('\G[^ш]', true, index)
|
|
437
|
-
r10 = true
|
|
438
|
-
@index += 1
|
|
439
|
-
else
|
|
440
|
-
r10 = nil
|
|
441
|
-
end
|
|
442
|
-
if r10
|
|
443
|
-
s9 << r10
|
|
444
|
-
else
|
|
445
|
-
break
|
|
446
|
-
end
|
|
447
|
-
end
|
|
448
|
-
if s9.empty?
|
|
449
|
-
@index = i9
|
|
450
|
-
r9 = nil
|
|
451
|
-
else
|
|
452
|
-
r9 = instantiate_node(SyntaxNode,input, i9...index, s9)
|
|
453
|
-
end
|
|
454
|
-
s7 << r9
|
|
455
|
-
end
|
|
456
|
-
if s7.last
|
|
457
|
-
r7 = instantiate_node(SyntaxNode,input, i7...index, s7)
|
|
458
|
-
r7.extend(Garbage1)
|
|
459
|
-
else
|
|
460
|
-
@index = i7
|
|
461
|
-
r7 = nil
|
|
462
|
-
end
|
|
463
|
-
if r7
|
|
464
|
-
r0 = r7
|
|
465
|
-
else
|
|
466
|
-
@index = i0
|
|
467
|
-
r0 = nil
|
|
468
|
-
end
|
|
469
|
-
end
|
|
470
|
-
|
|
471
|
-
node_cache[:garbage][start_index] = r0
|
|
472
|
-
|
|
473
|
-
r0
|
|
474
|
-
end
|
|
475
|
-
|
|
476
|
-
end
|
|
477
|
-
|
|
478
|
-
class ScientificNameCanonicalParser < Treetop::Runtime::CompiledParser
|
|
479
|
-
include ScientificNameCanonical
|
|
480
|
-
end
|
|
481
|
-
|