linguistics 1.0.8
Sign up to get free protection for your applications and to get access to all the features.
- data/ChangeLog +640 -0
- data/LICENSE +27 -0
- data/README +166 -0
- data/README.english +245 -0
- data/Rakefile +338 -0
- data/examples/generalize_sentence.rb +46 -0
- data/lib/linguistics.rb +366 -0
- data/lib/linguistics/en.rb +1728 -0
- data/lib/linguistics/en/infinitive.rb +1145 -0
- data/lib/linguistics/en/linkparser.rb +109 -0
- data/lib/linguistics/en/wordnet.rb +257 -0
- data/lib/linguistics/iso639.rb +461 -0
- data/rake/191_compat.rb +26 -0
- data/rake/dependencies.rb +76 -0
- data/rake/helpers.rb +434 -0
- data/rake/hg.rb +261 -0
- data/rake/manual.rb +782 -0
- data/rake/packaging.rb +144 -0
- data/rake/publishing.rb +318 -0
- data/rake/rdoc.rb +30 -0
- data/rake/style.rb +62 -0
- data/rake/svn.rb +668 -0
- data/rake/testing.rb +187 -0
- data/rake/verifytask.rb +64 -0
- data/rake/win32.rb +190 -0
- data/spec/linguistics/en_spec.rb +215 -0
- data/spec/linguistics/iso639_spec.rb +72 -0
- data/spec/linguistics_spec.rb +107 -0
- data/tests/en/infinitive.tests.rb +207 -0
- data/tests/en/inflect.tests.rb +1389 -0
- data/tests/en/lafcadio.tests.rb +77 -0
- data/tests/en/linkparser.tests.rb +42 -0
- data/tests/en/lprintf.tests.rb +77 -0
- data/tests/en/titlecase.tests.rb +73 -0
- data/tests/en/wordnet.tests.rb +95 -0
- metadata +107 -0
|
@@ -0,0 +1,1145 @@
|
|
|
1
|
+
#!/usr/bin/ruby
|
|
2
|
+
#
|
|
3
|
+
# This file contains functions for deriving the infinitive forms of conjugated
|
|
4
|
+
# English words. Requiring this file adds functions and constants to the
|
|
5
|
+
# Linguistics::EN module.
|
|
6
|
+
#
|
|
7
|
+
# == Authors
|
|
8
|
+
#
|
|
9
|
+
# * Michael Granger <ged@FaerieMUD.org>
|
|
10
|
+
#
|
|
11
|
+
# == Acknowledgments
|
|
12
|
+
#
|
|
13
|
+
# This code was ported from the excellent 'Lingua::EN::Infinitive' Perl module
|
|
14
|
+
# by Ron Savage, which is distributed under the following license:
|
|
15
|
+
#
|
|
16
|
+
# Australian copyright (c) 1999-2002 Ron Savage.
|
|
17
|
+
#
|
|
18
|
+
# All Programs of mine are 'OSI Certified Open Source Software';
|
|
19
|
+
# you can redistribute them and/or modify them under the terms of
|
|
20
|
+
# The Artistic License, a copy of which is available at:
|
|
21
|
+
# http://www.opensource.org/licenses/index.html
|
|
22
|
+
#
|
|
23
|
+
#
|
|
24
|
+
# :include: LICENSE
|
|
25
|
+
#
|
|
26
|
+
#--
|
|
27
|
+
#
|
|
28
|
+
# Please see the file LICENSE in the base directory for licensing details.
|
|
29
|
+
#
|
|
30
|
+
module Linguistics::EN
|
|
31
|
+
|
|
32
|
+
# :stopdoc:
|
|
33
|
+
|
|
34
|
+
# Irregular words => infinitive forms
|
|
35
|
+
IrregularInfinitives = {
|
|
36
|
+
'abided' => 'abide',
|
|
37
|
+
'abode' => 'abide',
|
|
38
|
+
'am' => 'be',
|
|
39
|
+
'are' => 'be',
|
|
40
|
+
'arisen' => 'arise',
|
|
41
|
+
'arose' => 'arise',
|
|
42
|
+
'ate' => 'eat',
|
|
43
|
+
'awaked' => 'awake',
|
|
44
|
+
'awoke' => 'awake',
|
|
45
|
+
'bade' => 'bid',
|
|
46
|
+
'beaten' => 'beat',
|
|
47
|
+
'became' => 'become',
|
|
48
|
+
'been' => 'be',
|
|
49
|
+
'befallen' => 'befall',
|
|
50
|
+
'befell' => 'befall',
|
|
51
|
+
'began' => 'begin',
|
|
52
|
+
'begat' => 'beget',
|
|
53
|
+
'begot' => 'beget',
|
|
54
|
+
'begotten' => 'beget',
|
|
55
|
+
'begun' => 'begin',
|
|
56
|
+
'beheld' => 'behold',
|
|
57
|
+
'bent' => 'bend',
|
|
58
|
+
'bereaved' => 'bereave',
|
|
59
|
+
'bereft' => 'bereave',
|
|
60
|
+
'beseeched' => 'beseech',
|
|
61
|
+
'besought' => 'beseech',
|
|
62
|
+
'bespoke' => 'bespeak',
|
|
63
|
+
'bespoken' => 'bespeak',
|
|
64
|
+
'bestrewed' => 'bestrew',
|
|
65
|
+
'bestrewn' => 'bestrew',
|
|
66
|
+
'bestrid' => 'bestride',
|
|
67
|
+
'bestridden' => 'bestride',
|
|
68
|
+
'bestrode' => 'bestride',
|
|
69
|
+
'betaken' => 'betake',
|
|
70
|
+
'bethought' => 'bethink',
|
|
71
|
+
'betook' => 'betake',
|
|
72
|
+
'betted' => 'bet',
|
|
73
|
+
'bidden' => 'bid',
|
|
74
|
+
'bided' => 'bide',
|
|
75
|
+
'bit' => 'bite',
|
|
76
|
+
'bitten' => 'bite',
|
|
77
|
+
'bled' => 'bleed',
|
|
78
|
+
'blended' => 'blend',
|
|
79
|
+
'blent' => 'blend',
|
|
80
|
+
'blessed' => 'bless',
|
|
81
|
+
'blest' => 'bless',
|
|
82
|
+
'blew' => 'blow',
|
|
83
|
+
'blown' => 'blow',
|
|
84
|
+
'bode' => 'bide',
|
|
85
|
+
'bore' => 'bear',
|
|
86
|
+
'born' => 'bear',
|
|
87
|
+
'borne' => 'bear',
|
|
88
|
+
'bought' => 'buy',
|
|
89
|
+
'bound' => 'bind',
|
|
90
|
+
'bred' => 'breed',
|
|
91
|
+
'broadcasted' => 'broadcast',
|
|
92
|
+
'broke' => 'break',
|
|
93
|
+
'broken' => 'break',
|
|
94
|
+
'brought' => 'bring',
|
|
95
|
+
'browbeaten' => 'browbeat',
|
|
96
|
+
'built' => 'build',
|
|
97
|
+
'burned' => 'burn',
|
|
98
|
+
'burnt' => 'burn',
|
|
99
|
+
'came' => 'come',
|
|
100
|
+
'caught' => 'catch',
|
|
101
|
+
'chid' => 'chide',
|
|
102
|
+
'chidden' => 'chide',
|
|
103
|
+
'chided' => 'chide',
|
|
104
|
+
'chose' => 'choose',
|
|
105
|
+
'chosen' => 'choose',
|
|
106
|
+
'clad' => 'clothe',
|
|
107
|
+
'clave' => 'cleave',
|
|
108
|
+
'cleaved' => 'cleave',
|
|
109
|
+
'cleft' => 'cleave',
|
|
110
|
+
'clothed' => 'clothe',
|
|
111
|
+
'clove' => 'cleave',
|
|
112
|
+
'cloven' => 'cleave',
|
|
113
|
+
'clung' => 'cling',
|
|
114
|
+
'costed' => 'cost',
|
|
115
|
+
'could' => 'can',
|
|
116
|
+
'crept' => 'creep',
|
|
117
|
+
'crew' => 'crow',
|
|
118
|
+
'crowed' => 'crow',
|
|
119
|
+
'dealt' => 'deal',
|
|
120
|
+
'did' => 'do',
|
|
121
|
+
'done' => 'do',
|
|
122
|
+
'dove' => 'dive',
|
|
123
|
+
'drank' => 'drink',
|
|
124
|
+
'drawn' => 'draw',
|
|
125
|
+
'dreamed' => 'dream',
|
|
126
|
+
'dreamt' => 'dream',
|
|
127
|
+
'drew' => 'draw',
|
|
128
|
+
'driven' => 'drive',
|
|
129
|
+
'drove' => 'drive',
|
|
130
|
+
'drunk' => 'drink',
|
|
131
|
+
'dug' => 'dig',
|
|
132
|
+
'dwelled' => 'dwell',
|
|
133
|
+
'dwelt' => 'dwell',
|
|
134
|
+
'eaten' => 'eat',
|
|
135
|
+
'fallen' => 'fall',
|
|
136
|
+
'fed' => 'feed',
|
|
137
|
+
'fell' => 'fall',
|
|
138
|
+
'felt' => 'feel',
|
|
139
|
+
'fled' => 'flee',
|
|
140
|
+
'flew' => 'fly',
|
|
141
|
+
'flown' => 'fly',
|
|
142
|
+
'flung' => 'fling',
|
|
143
|
+
'forbad' => 'forbid',
|
|
144
|
+
'forbade' => 'forbid',
|
|
145
|
+
'forbidden' => 'forbid',
|
|
146
|
+
'forbore' => 'forbear',
|
|
147
|
+
'forborne' => 'forbear',
|
|
148
|
+
'fordid' => 'fordo',
|
|
149
|
+
'fordone' => 'fordo',
|
|
150
|
+
'forecasted' => 'forecast',
|
|
151
|
+
'foregone' => 'forego',
|
|
152
|
+
'foreknew' => 'foreknow',
|
|
153
|
+
'foreknown' => 'foreknow',
|
|
154
|
+
'foreran' => 'forerun',
|
|
155
|
+
'foresaw' => 'foresee',
|
|
156
|
+
'foreshowed' => 'foreshow',
|
|
157
|
+
'foreshown' => 'foreshow',
|
|
158
|
+
'foretold' => 'foretell',
|
|
159
|
+
'forewent' => 'forego',
|
|
160
|
+
'forgave' => 'forgive',
|
|
161
|
+
'forgiven' => 'forgive',
|
|
162
|
+
'forgot' => 'forget',
|
|
163
|
+
'forgotten' => 'forget',
|
|
164
|
+
'forsaken' => 'forsake',
|
|
165
|
+
'forseen' => 'foresee',
|
|
166
|
+
'forsook' => 'forsake',
|
|
167
|
+
'forswore' => 'forswear',
|
|
168
|
+
'forsworn' => 'forswear',
|
|
169
|
+
'fought' => 'fight',
|
|
170
|
+
'found' => 'find',
|
|
171
|
+
'froze' => 'freeze',
|
|
172
|
+
'frozen' => 'freeze',
|
|
173
|
+
'gainsaid' => 'gainsay',
|
|
174
|
+
'gave' => 'give',
|
|
175
|
+
'gilded' => 'gild',
|
|
176
|
+
'gilt' => 'gild',
|
|
177
|
+
'girded' => 'gird',
|
|
178
|
+
'girt' => 'gird',
|
|
179
|
+
'given' => 'give',
|
|
180
|
+
'gone' => 'go',
|
|
181
|
+
'got' => 'get',
|
|
182
|
+
'gotten' => 'get',
|
|
183
|
+
'graved' => 'grave',
|
|
184
|
+
'graven' => 'grave',
|
|
185
|
+
'grew' => 'grow',
|
|
186
|
+
'ground' => 'grind',
|
|
187
|
+
'grown' => 'grow',
|
|
188
|
+
'had' => 'have',
|
|
189
|
+
'hamstringed' => 'hamstring',
|
|
190
|
+
'hamstrung' => 'hamstring',
|
|
191
|
+
'hanged' => 'hang',
|
|
192
|
+
'heard' => 'hear',
|
|
193
|
+
'heaved' => 'heave',
|
|
194
|
+
'held' => 'hold',
|
|
195
|
+
'hewed' => 'hew',
|
|
196
|
+
'hewn' => 'hew',
|
|
197
|
+
'hid' => 'hide',
|
|
198
|
+
'hidden' => 'hide',
|
|
199
|
+
'hove' => 'heave',
|
|
200
|
+
'hung' => 'hang',
|
|
201
|
+
'inlaid' => 'inlay',
|
|
202
|
+
'is' => 'be',
|
|
203
|
+
'kept' => 'keep',
|
|
204
|
+
'kneeled' => 'kneel',
|
|
205
|
+
'knelt' => 'kneel',
|
|
206
|
+
'knew' => 'know',
|
|
207
|
+
'knitted' => 'knit',
|
|
208
|
+
'known' => 'know',
|
|
209
|
+
'laded' => 'lade',
|
|
210
|
+
'laden' => 'lade',
|
|
211
|
+
'laid' => 'lay',
|
|
212
|
+
'lain' => 'lie',
|
|
213
|
+
'lay' => 'lie',
|
|
214
|
+
'leaned' => 'lean',
|
|
215
|
+
'leant' => 'lean',
|
|
216
|
+
'leaped' => 'leap',
|
|
217
|
+
'leapt' => 'leap',
|
|
218
|
+
'learned' => 'learn',
|
|
219
|
+
'learnt' => 'learn',
|
|
220
|
+
'led' => 'lead',
|
|
221
|
+
'left' => 'leave',
|
|
222
|
+
'lent' => 'lend',
|
|
223
|
+
'lighted' => 'light',
|
|
224
|
+
'lit' => 'light',
|
|
225
|
+
'lost' => 'lose',
|
|
226
|
+
'made' => 'make',
|
|
227
|
+
'meant' => 'mean',
|
|
228
|
+
'melted' => 'melt',
|
|
229
|
+
'met' => 'meet',
|
|
230
|
+
'might' => 'may',
|
|
231
|
+
'misdealt' => 'misdeal',
|
|
232
|
+
'misgave' => 'misgive',
|
|
233
|
+
'misgiven' => 'misgive',
|
|
234
|
+
'mislaid' => 'mislay',
|
|
235
|
+
'misled' => 'mislead',
|
|
236
|
+
'mistaken' => 'mistake',
|
|
237
|
+
'mistook' => 'mistake',
|
|
238
|
+
'misunderstood' => 'misunderstand',
|
|
239
|
+
'molten' => 'melt',
|
|
240
|
+
'mowed' => 'mow',
|
|
241
|
+
'mown' => 'mow',
|
|
242
|
+
'outate' => 'outeat',
|
|
243
|
+
'outbade' => 'outbid',
|
|
244
|
+
'outbidden' => 'outbid',
|
|
245
|
+
'outbred' => 'outbreed',
|
|
246
|
+
'outdid' => 'outdo',
|
|
247
|
+
'outdone' => 'outdo',
|
|
248
|
+
'outeaten' => 'outeat',
|
|
249
|
+
'outfought' => 'outfight',
|
|
250
|
+
'outgone' => 'outgo',
|
|
251
|
+
'outgrew' => 'outgrow',
|
|
252
|
+
'outgrown' => 'outgrow',
|
|
253
|
+
'outlaid' => 'outlay',
|
|
254
|
+
'outran' => 'outrun',
|
|
255
|
+
'outridden' => 'outride',
|
|
256
|
+
'outrode' => 'outride',
|
|
257
|
+
'outsat' => 'outsit',
|
|
258
|
+
'outshone' => 'outshine',
|
|
259
|
+
'outshot' => 'outshoot',
|
|
260
|
+
'outsold' => 'outsell',
|
|
261
|
+
'outspent' => 'outspend',
|
|
262
|
+
'outthrew' => 'outthrow',
|
|
263
|
+
'outthrown' => 'outthrow',
|
|
264
|
+
'outwent' => 'outgo',
|
|
265
|
+
'outwore' => 'outwear',
|
|
266
|
+
'outworn' => 'outwear',
|
|
267
|
+
'overate' => 'overeat',
|
|
268
|
+
'overbade' => 'overbid',
|
|
269
|
+
'overbidden' => 'overbid',
|
|
270
|
+
'overblew' => 'overblow',
|
|
271
|
+
'overblown' => 'overblow',
|
|
272
|
+
'overbore' => 'overbear',
|
|
273
|
+
'overborn' => 'overbear',
|
|
274
|
+
'overborne' => 'overbear',
|
|
275
|
+
'overbought' => 'overbuy',
|
|
276
|
+
'overbuilt' => 'overbuild',
|
|
277
|
+
'overcame' => 'overcome',
|
|
278
|
+
'overdid' => 'overdo',
|
|
279
|
+
'overdone' => 'overdo',
|
|
280
|
+
'overdrawn' => 'overdraw',
|
|
281
|
+
'overdrew' => 'overdraw',
|
|
282
|
+
'overdriven' => 'overdrive',
|
|
283
|
+
'overdrove' => 'overdrive',
|
|
284
|
+
'overeaten' => 'overeat',
|
|
285
|
+
'overfed' => 'overfeed',
|
|
286
|
+
'overflew' => 'overfly',
|
|
287
|
+
'overflown' => 'overfly',
|
|
288
|
+
'overgrew' => 'overgrow',
|
|
289
|
+
'overgrown' => 'overgrow',
|
|
290
|
+
'overhanged' => 'overhang',
|
|
291
|
+
'overheard' => 'overhear',
|
|
292
|
+
'overhung' => 'overhang',
|
|
293
|
+
'overladed' => 'overlade',
|
|
294
|
+
'overladen' => 'overlade',
|
|
295
|
+
'overlaid' => 'overlay',
|
|
296
|
+
'overlain' => 'overlie',
|
|
297
|
+
'overlay' => 'overlie',
|
|
298
|
+
'overleaped' => 'overleap',
|
|
299
|
+
'overleapt' => 'overleap',
|
|
300
|
+
'overpaid' => 'overpay',
|
|
301
|
+
'overran' => 'overrun',
|
|
302
|
+
'overridden' => 'override',
|
|
303
|
+
'overrode' => 'override',
|
|
304
|
+
'oversaw' => 'oversee',
|
|
305
|
+
'overseen' => 'oversee',
|
|
306
|
+
'oversewed' => 'oversew',
|
|
307
|
+
'oversewn' => 'oversew',
|
|
308
|
+
'overshot' => 'overshoot',
|
|
309
|
+
'overslept' => 'oversleep',
|
|
310
|
+
'overspent' => 'overspend',
|
|
311
|
+
'overtaken' => 'overtake',
|
|
312
|
+
'overthrew' => 'overthrow',
|
|
313
|
+
'overthrown' => 'overthrow',
|
|
314
|
+
'overtook' => 'overtake',
|
|
315
|
+
'overwinded' => 'overwind',
|
|
316
|
+
'overwound' => 'overwind',
|
|
317
|
+
'overwritten' => 'overwrite',
|
|
318
|
+
'overwrote' => 'overwrite',
|
|
319
|
+
'paid' => 'pay',
|
|
320
|
+
'partaken' => 'partake',
|
|
321
|
+
'partook' => 'partake',
|
|
322
|
+
'prechose' => 'prechoose',
|
|
323
|
+
'prechosen' => 'prechoose',
|
|
324
|
+
'proved' => 'prove',
|
|
325
|
+
'proven' => 'prove',
|
|
326
|
+
'quitted' => 'quit',
|
|
327
|
+
'ran' => 'run',
|
|
328
|
+
'rang' => 'ring',
|
|
329
|
+
'reaved' => 'reave',
|
|
330
|
+
'rebuilt' => 'rebuild',
|
|
331
|
+
'reeved' => 'reeve',
|
|
332
|
+
'reft' => 'reave',
|
|
333
|
+
'relaid' => 'relay',
|
|
334
|
+
'rent' => 'rend',
|
|
335
|
+
'repaid' => 'repay',
|
|
336
|
+
'retold' => 'retell',
|
|
337
|
+
'ridded' => 'rid',
|
|
338
|
+
'ridden' => 'ride',
|
|
339
|
+
'risen' => 'rise',
|
|
340
|
+
'rived' => 'rive',
|
|
341
|
+
'riven' => 'rive',
|
|
342
|
+
'rode' => 'ride',
|
|
343
|
+
'rose' => 'rise',
|
|
344
|
+
'rove' => 'reeve',
|
|
345
|
+
'rung' => 'ring',
|
|
346
|
+
'said' => 'say',
|
|
347
|
+
'sang' => 'sing',
|
|
348
|
+
'sank' => 'sink',
|
|
349
|
+
'sat' => 'sit',
|
|
350
|
+
'saw' => 'see',
|
|
351
|
+
'sawed' => 'saw',
|
|
352
|
+
'sawn' => 'saw',
|
|
353
|
+
'seen' => 'see',
|
|
354
|
+
'sent' => 'send',
|
|
355
|
+
'sewed' => 'sew',
|
|
356
|
+
'sewn' => 'sew',
|
|
357
|
+
'shaken' => 'shake',
|
|
358
|
+
'shaved' => 'shave',
|
|
359
|
+
'shaven' => 'shave',
|
|
360
|
+
'sheared' => 'shear',
|
|
361
|
+
'shined' => 'shine',
|
|
362
|
+
'shod' => 'shoe',
|
|
363
|
+
'shoed' => 'shoe',
|
|
364
|
+
'shone' => 'shine',
|
|
365
|
+
'shook' => 'shake',
|
|
366
|
+
'shorn' => 'shear',
|
|
367
|
+
'shot' => 'shoot',
|
|
368
|
+
'showed' => 'show',
|
|
369
|
+
'shown' => 'show',
|
|
370
|
+
'shrank' => 'shrink',
|
|
371
|
+
'shredded' => 'shred',
|
|
372
|
+
'shrived' => 'shrive',
|
|
373
|
+
'shriven' => 'shrive',
|
|
374
|
+
'shrove' => 'shrive',
|
|
375
|
+
'shrunk' => 'shrink',
|
|
376
|
+
'shrunken' => 'shrink',
|
|
377
|
+
'slain' => 'slay',
|
|
378
|
+
'slept' => 'sleep',
|
|
379
|
+
'slew' => 'slay',
|
|
380
|
+
'slid' => 'slide',
|
|
381
|
+
'slidden' => 'slide',
|
|
382
|
+
'slung' => 'sling',
|
|
383
|
+
'slunk' => 'slink',
|
|
384
|
+
'smelled' => 'smell',
|
|
385
|
+
'smelt' => 'smell',
|
|
386
|
+
'smitten' => 'smite',
|
|
387
|
+
'smote' => 'smite',
|
|
388
|
+
'snuck' => 'sneak',
|
|
389
|
+
'sold' => 'sell',
|
|
390
|
+
'sought' => 'seek',
|
|
391
|
+
'sowed' => 'sow',
|
|
392
|
+
'sown' => 'sow',
|
|
393
|
+
'span' => 'spin',
|
|
394
|
+
'spat' => 'spit',
|
|
395
|
+
'sped' => 'speed',
|
|
396
|
+
'speeded' => 'speed',
|
|
397
|
+
'spelled' => 'spell',
|
|
398
|
+
'spelt' => 'spell',
|
|
399
|
+
'spent' => 'spend',
|
|
400
|
+
'spilled' => 'spill',
|
|
401
|
+
'spilt' => 'spill',
|
|
402
|
+
'spoiled' => 'spoil',
|
|
403
|
+
'spoilt' => 'spoil',
|
|
404
|
+
'spoke' => 'speak',
|
|
405
|
+
'spoken' => 'speak',
|
|
406
|
+
'sprang' => 'spring',
|
|
407
|
+
'sprung' => 'spring',
|
|
408
|
+
'spun' => 'spin',
|
|
409
|
+
'stank' => 'stink',
|
|
410
|
+
'staved' => 'stave',
|
|
411
|
+
'stole' => 'steal',
|
|
412
|
+
'stolen' => 'steal',
|
|
413
|
+
'stood' => 'stand',
|
|
414
|
+
'stove' => 'stave',
|
|
415
|
+
'strewed' => 'strew',
|
|
416
|
+
'strewn' => 'strew',
|
|
417
|
+
'stricken' => 'strike',
|
|
418
|
+
'strid' => 'stride',
|
|
419
|
+
'stridden' => 'stride',
|
|
420
|
+
'strived' => 'strive',
|
|
421
|
+
'striven' => 'strive',
|
|
422
|
+
'strode' => 'stride',
|
|
423
|
+
'strove' => 'strive',
|
|
424
|
+
'struck' => 'strike',
|
|
425
|
+
'strung' => 'string',
|
|
426
|
+
'stuck' => 'stick',
|
|
427
|
+
'stung' => 'sting',
|
|
428
|
+
'stunk' => 'stink',
|
|
429
|
+
'sung' => 'sing',
|
|
430
|
+
'sunk' => 'sink',
|
|
431
|
+
'sunken' => 'sink',
|
|
432
|
+
'swam' => 'swim',
|
|
433
|
+
'sweated' => 'sweat',
|
|
434
|
+
'swelled' => 'swell',
|
|
435
|
+
'swept' => 'sweep',
|
|
436
|
+
'swollen' => 'swell',
|
|
437
|
+
'swore' => 'swear',
|
|
438
|
+
'sworn' => 'swear',
|
|
439
|
+
'swum' => 'swim',
|
|
440
|
+
'swung' => 'swing',
|
|
441
|
+
'taken' => 'take',
|
|
442
|
+
'taught' => 'teach',
|
|
443
|
+
'thought' => 'think',
|
|
444
|
+
'threw' => 'throw',
|
|
445
|
+
'thrived' => 'thrive',
|
|
446
|
+
'thriven' => 'thrive',
|
|
447
|
+
'throve' => 'thrive',
|
|
448
|
+
'thrown' => 'throw',
|
|
449
|
+
'told' => 'tell',
|
|
450
|
+
'took' => 'take',
|
|
451
|
+
'tore' => 'tear',
|
|
452
|
+
'torn' => 'tear',
|
|
453
|
+
'trod' => 'tread',
|
|
454
|
+
'trodden' => 'tread',
|
|
455
|
+
'unbent' => 'unbend',
|
|
456
|
+
'unbound' => 'unbind',
|
|
457
|
+
'unbuilt' => 'unbuild',
|
|
458
|
+
'underbought' => 'underbuy',
|
|
459
|
+
'underfed' => 'underfeed',
|
|
460
|
+
'undergone' => 'undergo',
|
|
461
|
+
'underlaid' => 'underlay',
|
|
462
|
+
'underlain' => 'underlie',
|
|
463
|
+
'underlay' => 'underlie',
|
|
464
|
+
'underpaid' => 'underpay',
|
|
465
|
+
'underran' => 'underrun',
|
|
466
|
+
'undershot' => 'undershoot',
|
|
467
|
+
'undersold' => 'undersell',
|
|
468
|
+
'understood' => 'understand',
|
|
469
|
+
'undertaken' => 'undertake',
|
|
470
|
+
'undertook' => 'undertake',
|
|
471
|
+
'underwent' => 'undergo',
|
|
472
|
+
'underwritten' => 'underwrite',
|
|
473
|
+
'underwrote' => 'underwrite',
|
|
474
|
+
'undid' => 'undo',
|
|
475
|
+
'undone' => 'undo',
|
|
476
|
+
'undrawn' => 'undraw',
|
|
477
|
+
'undrew' => 'undraw',
|
|
478
|
+
'unfroze' => 'unfreeze',
|
|
479
|
+
'unfrozen' => 'unfreeze',
|
|
480
|
+
'ungirded' => 'ungird',
|
|
481
|
+
'ungirt' => 'ungird',
|
|
482
|
+
'unhanged' => 'unhang',
|
|
483
|
+
'unhung' => 'unhang',
|
|
484
|
+
'unknitted' => 'unknit',
|
|
485
|
+
'unladed' => 'unlade',
|
|
486
|
+
'unladen' => 'unlade',
|
|
487
|
+
'unlaid' => 'unlay',
|
|
488
|
+
'unlearned' => 'unlearn',
|
|
489
|
+
'unlearnt' => 'unlearn',
|
|
490
|
+
'unmade' => 'unmake',
|
|
491
|
+
'unreeved' => 'unreeve',
|
|
492
|
+
'unrove' => 'unreeve',
|
|
493
|
+
'unsaid' => 'unsay',
|
|
494
|
+
'unslung' => 'unsling',
|
|
495
|
+
'unspoke' => 'unspeak',
|
|
496
|
+
'unspoken' => 'unspeak',
|
|
497
|
+
'unstrung' => 'unstring',
|
|
498
|
+
'unstuck' => 'unstick',
|
|
499
|
+
'unswore' => 'unswear',
|
|
500
|
+
'unsworn' => 'unswear',
|
|
501
|
+
'untaught' => 'unteach',
|
|
502
|
+
'unthought' => 'unthink',
|
|
503
|
+
'untrod' => 'untread',
|
|
504
|
+
'untrodden' => 'untread',
|
|
505
|
+
'unwinded' => 'unwind',
|
|
506
|
+
'unwound' => 'unwind',
|
|
507
|
+
'unwove' => 'unweave',
|
|
508
|
+
'unwoven' => 'unweave',
|
|
509
|
+
'upbuilt' => 'upbuild',
|
|
510
|
+
'upheld' => 'uphold',
|
|
511
|
+
'uprisen' => 'uprise',
|
|
512
|
+
'uprose' => 'uprise',
|
|
513
|
+
'upswept' => 'upsweep',
|
|
514
|
+
'upswung' => 'upswing',
|
|
515
|
+
'waked' => 'wake',
|
|
516
|
+
'was' => 'be',
|
|
517
|
+
'waylaid' => 'waylay',
|
|
518
|
+
'wedded' => 'wed',
|
|
519
|
+
'went' => 'go',
|
|
520
|
+
'wept' => 'weep',
|
|
521
|
+
'were' => 'be',
|
|
522
|
+
'wetted' => 'wet',
|
|
523
|
+
'winded' => 'wind',
|
|
524
|
+
'wist' => 'wit',
|
|
525
|
+
'wot' => 'wit',
|
|
526
|
+
'withdrawn' => 'withdraw',
|
|
527
|
+
'withdrew' => 'withdraw',
|
|
528
|
+
'withheld' => 'withhold',
|
|
529
|
+
'withstood' => 'withstand',
|
|
530
|
+
'woke' => 'wake',
|
|
531
|
+
'woken' => 'wake',
|
|
532
|
+
'won' => 'win',
|
|
533
|
+
'wore' => 'wear',
|
|
534
|
+
'worked' => 'work',
|
|
535
|
+
'worn' => 'wear',
|
|
536
|
+
'wound' => 'wind',
|
|
537
|
+
'wove' => 'weave',
|
|
538
|
+
'woven' => 'weave',
|
|
539
|
+
'written' => 'write',
|
|
540
|
+
'wrote' => 'write',
|
|
541
|
+
'wrought' => 'work',
|
|
542
|
+
'wrung' => 'wring',
|
|
543
|
+
}
|
|
544
|
+
|
|
545
|
+
# Mapping of word suffixes to infinitive rules.
|
|
546
|
+
InfSuffixRules = {
|
|
547
|
+
# '<suffix>' => {
|
|
548
|
+
# :order => <sort order>,
|
|
549
|
+
# :rule => <rule number>,
|
|
550
|
+
|
|
551
|
+
# :word1 == 0 => Use 0, the index of the longest prefix
|
|
552
|
+
# within @{$prefix{$self->{'suffix'} } }, below.
|
|
553
|
+
|
|
554
|
+
# :word1 == 1 => Use 1, the index of the 2nd longest prefix
|
|
555
|
+
# within @{$prefix{$self->{'suffix'} } }, below.
|
|
556
|
+
|
|
557
|
+
# :word1 == -1 => Use the index of the shortest prefix
|
|
558
|
+
# within @{$prefix{$self->{'suffix'} } }, below + a letter.
|
|
559
|
+
|
|
560
|
+
# :word1 == -2 => Use the index of the shortest prefix
|
|
561
|
+
# within @{$prefix{$self->{'suffix'} } }, below + a letter,
|
|
562
|
+
# and use the shortest prefix as well.
|
|
563
|
+
|
|
564
|
+
# :word1 == -3 => Use the index of the shortest prefix
|
|
565
|
+
# within @{$prefix{$self->{'suffix'} } }, below + meter,
|
|
566
|
+
# and use the shortest prefix + metre as well.
|
|
567
|
+
|
|
568
|
+
# :word1 == -4 => Use the original string.
|
|
569
|
+
'hes' => {
|
|
570
|
+
:order => 1011,
|
|
571
|
+
:rule => '1',
|
|
572
|
+
:word1 => 0, # Longest prefix.
|
|
573
|
+
:suffix1 => '',
|
|
574
|
+
:suffix2 => '',
|
|
575
|
+
},
|
|
576
|
+
'ses' => {
|
|
577
|
+
:order => 1021,
|
|
578
|
+
:rule => '2',
|
|
579
|
+
:word1 => 0, # Longest prefix.
|
|
580
|
+
:suffix1 => '',
|
|
581
|
+
:suffix2 => '',
|
|
582
|
+
},
|
|
583
|
+
'xes' => {
|
|
584
|
+
:order => 1031,
|
|
585
|
+
:rule => '3',
|
|
586
|
+
:word1 => 0, # Longest prefix.
|
|
587
|
+
:suffix1 => '',
|
|
588
|
+
:suffix2 => '',
|
|
589
|
+
},
|
|
590
|
+
'zes' => {
|
|
591
|
+
:order => 1041,
|
|
592
|
+
:rule => '4',
|
|
593
|
+
:word1 => 0, # Longest prefix.
|
|
594
|
+
:suffix1 => '',
|
|
595
|
+
:suffix2 => '',
|
|
596
|
+
},
|
|
597
|
+
'iless' => {
|
|
598
|
+
:order => 1051,
|
|
599
|
+
:rule => '43a',
|
|
600
|
+
:word1 => -1, # Shortest prefix.
|
|
601
|
+
:suffix1 => 'y',
|
|
602
|
+
:suffix2 => '',
|
|
603
|
+
},
|
|
604
|
+
'less' => {
|
|
605
|
+
:order => 1052,
|
|
606
|
+
:rule => '43b',
|
|
607
|
+
:word1 => -1, # Shortest prefix.
|
|
608
|
+
:suffix1 => '',
|
|
609
|
+
:suffix2 => '',
|
|
610
|
+
},
|
|
611
|
+
'iness' => {
|
|
612
|
+
:order => 1053,
|
|
613
|
+
:rule => '44a',
|
|
614
|
+
:word1 => -1, # Shortest prefix.
|
|
615
|
+
:suffix1 => 'y',
|
|
616
|
+
:suffix2 => '',
|
|
617
|
+
},
|
|
618
|
+
'ness' => {
|
|
619
|
+
:order => 1054,
|
|
620
|
+
:rule => '44b',
|
|
621
|
+
:word1 => -1, # Shortest prefix.
|
|
622
|
+
:suffix1 => '',
|
|
623
|
+
:suffix2 => '',
|
|
624
|
+
},
|
|
625
|
+
"'s" => {
|
|
626
|
+
:order => 1055,
|
|
627
|
+
:rule => '7',
|
|
628
|
+
:word1 => -1, # Shortest prefix.
|
|
629
|
+
:suffix1 => '',
|
|
630
|
+
:suffix2 => '',
|
|
631
|
+
},
|
|
632
|
+
'ies' => {
|
|
633
|
+
:order => 1056,
|
|
634
|
+
:rule => '13a',
|
|
635
|
+
:word1 => -1, # Shortest prefix.
|
|
636
|
+
:suffix1 => 'y',
|
|
637
|
+
:suffix2 => '',
|
|
638
|
+
},
|
|
639
|
+
'es' => {
|
|
640
|
+
:order => 1057,
|
|
641
|
+
:rule => '13b',
|
|
642
|
+
:word1 => 0, # Longest prefix.
|
|
643
|
+
:suffix1 => '',
|
|
644
|
+
:suffix2 => '',
|
|
645
|
+
},
|
|
646
|
+
'ss' => {
|
|
647
|
+
:order => 1061,
|
|
648
|
+
:rule => '6a',
|
|
649
|
+
:word1 => -4, # Original string.
|
|
650
|
+
:suffix1 => '',
|
|
651
|
+
:suffix2 => '',
|
|
652
|
+
},
|
|
653
|
+
's' => {
|
|
654
|
+
:order => 1062,
|
|
655
|
+
:rule => '6b',
|
|
656
|
+
:word1 => -1, # Shortest prefix.
|
|
657
|
+
:suffix1 => '',
|
|
658
|
+
:suffix2 => '',
|
|
659
|
+
},
|
|
660
|
+
'ater' => {
|
|
661
|
+
:order => 1081,
|
|
662
|
+
:rule => '8',
|
|
663
|
+
:word1 => -4, # Original string.
|
|
664
|
+
:suffix1 => '',
|
|
665
|
+
:suffix2 => '',
|
|
666
|
+
},
|
|
667
|
+
'cter' => {
|
|
668
|
+
:order => 1091,
|
|
669
|
+
:rule => '9',
|
|
670
|
+
:word1 => -4, # Original string.
|
|
671
|
+
:suffix1 => '',
|
|
672
|
+
:suffix2 => '',
|
|
673
|
+
},
|
|
674
|
+
'ier' => {
|
|
675
|
+
:order => 1101,
|
|
676
|
+
:rule => '10',
|
|
677
|
+
:word1 => -1, # Shortest prefix.
|
|
678
|
+
:suffix1 => 'y',
|
|
679
|
+
:suffix2 => '',
|
|
680
|
+
},
|
|
681
|
+
'er' => {
|
|
682
|
+
:order => 1111,
|
|
683
|
+
:rule => '11',
|
|
684
|
+
:word1 => 0, # Longest prefix.
|
|
685
|
+
:suffix1 => '',
|
|
686
|
+
:suffix2 => '',
|
|
687
|
+
},
|
|
688
|
+
'ied' => {
|
|
689
|
+
:order => 1121,
|
|
690
|
+
:rule => '12a',
|
|
691
|
+
:word1 => -1, # Shortest prefix.
|
|
692
|
+
:suffix1 => 'y',
|
|
693
|
+
:suffix2 => '',
|
|
694
|
+
},
|
|
695
|
+
'ed' => {
|
|
696
|
+
:order => 1122,
|
|
697
|
+
:rule => '12b', # There is extra code for 12b below.
|
|
698
|
+
:word1 => 0, # Longest prefix.
|
|
699
|
+
:suffix1 => '',
|
|
700
|
+
:suffix2 => '',
|
|
701
|
+
},
|
|
702
|
+
'iest' => {
|
|
703
|
+
:order => 1141,
|
|
704
|
+
:rule => '14a',
|
|
705
|
+
:word1 => -1, # Shortest prefix.
|
|
706
|
+
:suffix1 => 'y',
|
|
707
|
+
:suffix2 => '',
|
|
708
|
+
},
|
|
709
|
+
'est' => {
|
|
710
|
+
:order => 1142,
|
|
711
|
+
:rule => '14b',
|
|
712
|
+
:word1 => -2, # Shortest prefix + a letter, and shortest prefix.
|
|
713
|
+
:suffix1 => 'e',
|
|
714
|
+
:suffix2 => '',
|
|
715
|
+
},
|
|
716
|
+
'blity' => {
|
|
717
|
+
:order => 1143,
|
|
718
|
+
:rule => '21',
|
|
719
|
+
:word1 => -4, # Original string.
|
|
720
|
+
:suffix1 => '',
|
|
721
|
+
:suffix2 => '',
|
|
722
|
+
},
|
|
723
|
+
'bility' => {
|
|
724
|
+
:order => 1144,
|
|
725
|
+
:rule => '22',
|
|
726
|
+
:word1 => -1, # Shortest prefix.
|
|
727
|
+
:suffix1 => 'ble',
|
|
728
|
+
:suffix2 => '',
|
|
729
|
+
},
|
|
730
|
+
'fiable' => {
|
|
731
|
+
:order => 1145,
|
|
732
|
+
:rule => '23',
|
|
733
|
+
:word1 => -1, # Shortest prefix.
|
|
734
|
+
:suffix1 => 'fy',
|
|
735
|
+
:suffix2 => '',
|
|
736
|
+
},
|
|
737
|
+
'logist' => {
|
|
738
|
+
:order => 1146,
|
|
739
|
+
:rule => '24',
|
|
740
|
+
:word1 => -1, # Shortest prefix.
|
|
741
|
+
:suffix1 => 'logy',
|
|
742
|
+
:suffix2 => '',
|
|
743
|
+
},
|
|
744
|
+
'ing' => {
|
|
745
|
+
:order => 1151,
|
|
746
|
+
:rule => '15', # There is extra code for 15 below.
|
|
747
|
+
:word1 => -2, # Shortest prefix + a letter, and shortest prefix.
|
|
748
|
+
:suffix1 => 'e',
|
|
749
|
+
:suffix2 => '',
|
|
750
|
+
},
|
|
751
|
+
'ist' => {
|
|
752
|
+
:order => 1161,
|
|
753
|
+
:rule => '16',
|
|
754
|
+
:word1 => -2, # Shortest prefix + a letter, and shortest prefix.
|
|
755
|
+
:suffix1 => 'e',
|
|
756
|
+
:suffix2 => '',
|
|
757
|
+
},
|
|
758
|
+
'ism' => {
|
|
759
|
+
:order => 1171,
|
|
760
|
+
:rule => '17',
|
|
761
|
+
:word1 => -2, # Shortest prefix + a letter, and shortest prefix.
|
|
762
|
+
:suffix1 => 'e',
|
|
763
|
+
:suffix2 => '',
|
|
764
|
+
},
|
|
765
|
+
'ity' => {
|
|
766
|
+
:order => 1181,
|
|
767
|
+
:rule => '18',
|
|
768
|
+
:word1 => -2, # Shortest prefix + a letter, and shortest prefix.
|
|
769
|
+
:suffix1 => 'e',
|
|
770
|
+
:suffix2 => '',
|
|
771
|
+
},
|
|
772
|
+
'ize' => {
|
|
773
|
+
:order => 1191,
|
|
774
|
+
:rule => '19',
|
|
775
|
+
:word1 => -2, # Shortest prefix + a letter, and shortest prefix.
|
|
776
|
+
:suffix1 => 'e',
|
|
777
|
+
:suffix2 => '',
|
|
778
|
+
},
|
|
779
|
+
'cable' => {
|
|
780
|
+
:order => 1201,
|
|
781
|
+
:rule => '20a',
|
|
782
|
+
:word1 => -4, # Original string.
|
|
783
|
+
:suffix1 => '',
|
|
784
|
+
:suffix2 => '',
|
|
785
|
+
},
|
|
786
|
+
'gable' => {
|
|
787
|
+
:order => 1202,
|
|
788
|
+
:rule => '20b',
|
|
789
|
+
:word1 => -4, # Original string.
|
|
790
|
+
:suffix1 => '',
|
|
791
|
+
:suffix2 => '',
|
|
792
|
+
},
|
|
793
|
+
'able' => {
|
|
794
|
+
:order => 1203,
|
|
795
|
+
:rule => '20c',
|
|
796
|
+
:word1 => -2, # Shortest prefix + a letter, and shortest prefix.
|
|
797
|
+
:suffix1 => 'e',
|
|
798
|
+
:suffix2 => '',
|
|
799
|
+
},
|
|
800
|
+
'graphic' => {
|
|
801
|
+
:order => 1251,
|
|
802
|
+
:rule => '25',
|
|
803
|
+
:word1 => -1, # Shortest prefix.
|
|
804
|
+
:suffix1 => 'graphy',
|
|
805
|
+
:suffix2 => '',
|
|
806
|
+
},
|
|
807
|
+
'istic' => {
|
|
808
|
+
:order => 1261,
|
|
809
|
+
:rule => '26',
|
|
810
|
+
:word1 => -1, # Shortest prefix.
|
|
811
|
+
:suffix1 => 'ist',
|
|
812
|
+
:suffix2 => '',
|
|
813
|
+
},
|
|
814
|
+
'itic' => {
|
|
815
|
+
:order => 1271,
|
|
816
|
+
:rule => '27',
|
|
817
|
+
:word1 => -1, # Shortest prefix.
|
|
818
|
+
:suffix1 => 'ite',
|
|
819
|
+
:suffix2 => '',
|
|
820
|
+
},
|
|
821
|
+
'like' => {
|
|
822
|
+
:order => 1281,
|
|
823
|
+
:rule => '28',
|
|
824
|
+
:word1 => -1, # Shortest prefix.
|
|
825
|
+
:suffix1 => '',
|
|
826
|
+
:suffix2 => '',
|
|
827
|
+
},
|
|
828
|
+
'logic' => {
|
|
829
|
+
:order => 1291,
|
|
830
|
+
:rule => '29',
|
|
831
|
+
:word1 => -1, # Shortest prefix.
|
|
832
|
+
:suffix1 => 'logy',
|
|
833
|
+
:suffix2 => '',
|
|
834
|
+
},
|
|
835
|
+
'ment' => {
|
|
836
|
+
:order => 1301,
|
|
837
|
+
:rule => '30',
|
|
838
|
+
:word1 => -1, # Shortest prefix.
|
|
839
|
+
:suffix1 => '',
|
|
840
|
+
:suffix2 => '',
|
|
841
|
+
},
|
|
842
|
+
'mental' => {
|
|
843
|
+
:order => 1311,
|
|
844
|
+
:rule => '31',
|
|
845
|
+
:word1 => -1, # Shortest prefix.
|
|
846
|
+
:suffix1 => 'ment',
|
|
847
|
+
:suffix2 => '',
|
|
848
|
+
},
|
|
849
|
+
'metry' => {
|
|
850
|
+
:order => 1321,
|
|
851
|
+
:rule => '32',
|
|
852
|
+
:word1 => -3, # Shortest prefix + meter, and shortest perfix + metre.
|
|
853
|
+
:suffix1 => 'meter',
|
|
854
|
+
:suffix2 => 'metre',
|
|
855
|
+
},
|
|
856
|
+
'nce' => {
|
|
857
|
+
:order => 1331,
|
|
858
|
+
:rule => '33',
|
|
859
|
+
:word1 => -1, # Shortest prefix.
|
|
860
|
+
:suffix1 => 'nt',
|
|
861
|
+
:suffix2 => '',
|
|
862
|
+
},
|
|
863
|
+
'ncy' => {
|
|
864
|
+
:order => 1341,
|
|
865
|
+
:rule => '34',
|
|
866
|
+
:word1 => -1, # Shortest prefix.
|
|
867
|
+
:suffix1 => 'nt',
|
|
868
|
+
:suffix2 => '',
|
|
869
|
+
},
|
|
870
|
+
'ship' => {
|
|
871
|
+
:order => 1351,
|
|
872
|
+
:rule => '35',
|
|
873
|
+
:word1 => -1, # Shortest prefix.
|
|
874
|
+
:suffix1 => '',
|
|
875
|
+
:suffix2 => '',
|
|
876
|
+
},
|
|
877
|
+
'ical' => {
|
|
878
|
+
:order => 1361,
|
|
879
|
+
:rule => '36',
|
|
880
|
+
:word1 => -1, # Shortest prefix.
|
|
881
|
+
:suffix1 => 'ic',
|
|
882
|
+
:suffix2 => '',
|
|
883
|
+
},
|
|
884
|
+
'ional' => {
|
|
885
|
+
:order => 1371,
|
|
886
|
+
:rule => '37',
|
|
887
|
+
:word1 => -1, # Shortest prefix.
|
|
888
|
+
:suffix1 => 'ion',
|
|
889
|
+
:suffix2 => '',
|
|
890
|
+
},
|
|
891
|
+
'bly' => {
|
|
892
|
+
:order => 1381,
|
|
893
|
+
:rule => '38',
|
|
894
|
+
:word1 => -1, # Shortest prefix.
|
|
895
|
+
:suffix1 => 'ble',
|
|
896
|
+
:suffix2 => '',
|
|
897
|
+
},
|
|
898
|
+
'ily' => {
|
|
899
|
+
:order => 1391,
|
|
900
|
+
:rule => '39',
|
|
901
|
+
:word1 => -1, # Shortest prefix.
|
|
902
|
+
:suffix1 => 'y',
|
|
903
|
+
:suffix2 => '',
|
|
904
|
+
},
|
|
905
|
+
'ly' => {
|
|
906
|
+
:order => 1401,
|
|
907
|
+
:rule => '40',
|
|
908
|
+
:word1 => -1, # Shortest prefix.
|
|
909
|
+
:suffix1 => '',
|
|
910
|
+
:suffix2 => '',
|
|
911
|
+
},
|
|
912
|
+
'iful' => {
|
|
913
|
+
:order => 1411,
|
|
914
|
+
:rule => '41a',
|
|
915
|
+
:word1 => -1, # Shortest prefix.
|
|
916
|
+
:suffix1 => 'y',
|
|
917
|
+
:suffix2 => '',
|
|
918
|
+
},
|
|
919
|
+
'ful' => {
|
|
920
|
+
:order => 1412,
|
|
921
|
+
:rule => '41b',
|
|
922
|
+
:word1 => -1, # Shortest prefix.
|
|
923
|
+
:suffix1 => '',
|
|
924
|
+
:suffix2 => '',
|
|
925
|
+
},
|
|
926
|
+
'ihood' => {
|
|
927
|
+
:order => 1421,
|
|
928
|
+
:rule => '42a',
|
|
929
|
+
:word1 => -1, # Shortest prefix.
|
|
930
|
+
:suffix1 => 'y',
|
|
931
|
+
:suffix2 => '',
|
|
932
|
+
},
|
|
933
|
+
'hood' => {
|
|
934
|
+
:order => 1422,
|
|
935
|
+
:rule => '42b',
|
|
936
|
+
:word1 => -1, # Shortest prefix.
|
|
937
|
+
:suffix1 => '',
|
|
938
|
+
:suffix2 => '',
|
|
939
|
+
},
|
|
940
|
+
'ification' => {
|
|
941
|
+
:order => 1451,
|
|
942
|
+
:rule => '45',
|
|
943
|
+
:word1 => -1, # Shortest prefix.
|
|
944
|
+
:suffix1 => 'ify',
|
|
945
|
+
:suffix2 => '',
|
|
946
|
+
},
|
|
947
|
+
'ization' => {
|
|
948
|
+
:order => 1461,
|
|
949
|
+
:rule => '46',
|
|
950
|
+
:word1 => -1, # Shortest prefix.
|
|
951
|
+
:suffix1 => 'ize',
|
|
952
|
+
:suffix2 => '',
|
|
953
|
+
},
|
|
954
|
+
'ction' => {
|
|
955
|
+
:order => 1471,
|
|
956
|
+
:rule => '47',
|
|
957
|
+
:word1 => -1, # Shortest prefix.
|
|
958
|
+
:suffix1 => 'ct',
|
|
959
|
+
:suffix2 => '',
|
|
960
|
+
},
|
|
961
|
+
'rtion' => {
|
|
962
|
+
:order => 1481,
|
|
963
|
+
:rule => '48',
|
|
964
|
+
:word1 => -1, # Shortest prefix.
|
|
965
|
+
:suffix1 => 'rt',
|
|
966
|
+
:suffix2 => '',
|
|
967
|
+
},
|
|
968
|
+
'ation' => {
|
|
969
|
+
:order => 1491,
|
|
970
|
+
:rule => '49',
|
|
971
|
+
:word1 => -1, # Shortest prefix.
|
|
972
|
+
:suffix1 => 'ate',
|
|
973
|
+
:suffix2 => '',
|
|
974
|
+
},
|
|
975
|
+
'ator' => {
|
|
976
|
+
:order => 1501,
|
|
977
|
+
:rule => '50',
|
|
978
|
+
:word1 => -1, # Shortest prefix.
|
|
979
|
+
:suffix1 => 'ate',
|
|
980
|
+
:suffix2 => '',
|
|
981
|
+
},
|
|
982
|
+
'ctor' => {
|
|
983
|
+
:order => 1511,
|
|
984
|
+
:rule => '51',
|
|
985
|
+
:word1 => -1, # Shortest prefix.
|
|
986
|
+
:suffix1 => 'ct',
|
|
987
|
+
:suffix2 => '',
|
|
988
|
+
},
|
|
989
|
+
'ive' => {
|
|
990
|
+
:order => 1521,
|
|
991
|
+
:rule => '52',
|
|
992
|
+
:word1 => -1, # Shortest prefix.
|
|
993
|
+
:suffix1 => 'ion',
|
|
994
|
+
:suffix2 => '',
|
|
995
|
+
},
|
|
996
|
+
'onian' => {
|
|
997
|
+
:order => 1530,
|
|
998
|
+
:rule => '54',
|
|
999
|
+
:word1 => -1, # Shortest prefix.
|
|
1000
|
+
:suffix1 => 'on',
|
|
1001
|
+
:suffix2 => '',
|
|
1002
|
+
},
|
|
1003
|
+
'an' => {
|
|
1004
|
+
:order => 1531,
|
|
1005
|
+
:rule => '53',
|
|
1006
|
+
:word1 => -1, # Shortest prefix.
|
|
1007
|
+
:suffix1 => 'a',
|
|
1008
|
+
:suffix2 => '',
|
|
1009
|
+
},
|
|
1010
|
+
}
|
|
1011
|
+
InfSuffixRuleOrder = InfSuffixRules.keys.sort_by {|rule| InfSuffixRules[rule][:order]}
|
|
1012
|
+
|
|
1013
|
+
# :startdoc:
|
|
1014
|
+
|
|
1015
|
+
### The object class for the result returned from calling
|
|
1016
|
+
### Linguistics::EN::infinitive.
|
|
1017
|
+
class Infinitive < String
|
|
1018
|
+
|
|
1019
|
+
### Create and return a new Infinitive object.
|
|
1020
|
+
def initialize( word1, word2, suffix, rule )
|
|
1021
|
+
super( word1 )
|
|
1022
|
+
@word2 = word2
|
|
1023
|
+
@suffix = suffix
|
|
1024
|
+
@rule = rule
|
|
1025
|
+
end
|
|
1026
|
+
|
|
1027
|
+
|
|
1028
|
+
######
|
|
1029
|
+
public
|
|
1030
|
+
######
|
|
1031
|
+
|
|
1032
|
+
# The fallback deconjugated form
|
|
1033
|
+
attr_reader :word2
|
|
1034
|
+
|
|
1035
|
+
# The suffix used to to identify the transform rule
|
|
1036
|
+
attr_reader :suffix
|
|
1037
|
+
|
|
1038
|
+
# The rule used
|
|
1039
|
+
attr_reader :rule
|
|
1040
|
+
end
|
|
1041
|
+
|
|
1042
|
+
|
|
1043
|
+
###############
|
|
1044
|
+
module_function
|
|
1045
|
+
###############
|
|
1046
|
+
|
|
1047
|
+
### Return the infinitive form of the given word
|
|
1048
|
+
def infinitive( word )
|
|
1049
|
+
word = word.to_s
|
|
1050
|
+
word1 = word2 = suffix = rule = newword = ''
|
|
1051
|
+
|
|
1052
|
+
if IrregularInfinitives.key?( word )
|
|
1053
|
+
word1 = IrregularInfinitives[ word ]
|
|
1054
|
+
rule = 'irregular'
|
|
1055
|
+
else
|
|
1056
|
+
# Build up $prefix{$suffix} as an array of prefixes, from longest to shortest.
|
|
1057
|
+
prefix, suffix = nil
|
|
1058
|
+
prefixes = Hash::new {|hsh,key| hsh[key] = []}
|
|
1059
|
+
|
|
1060
|
+
# Build the hash of prefixes for the word
|
|
1061
|
+
1.upto( word.length ) {|i|
|
|
1062
|
+
prefix = word[0, i]
|
|
1063
|
+
suffix = word[i..-1]
|
|
1064
|
+
|
|
1065
|
+
(suffix.length - 1).downto( 0 ) {|j|
|
|
1066
|
+
newword = prefix + suffix[0, j]
|
|
1067
|
+
prefixes[ suffix ].push( newword )
|
|
1068
|
+
}
|
|
1069
|
+
}
|
|
1070
|
+
|
|
1071
|
+
$stderr.puts "prefixes: %p" % prefixes if $DEBUG
|
|
1072
|
+
|
|
1073
|
+
# Now check for rules covering the prefixes for this word, picking
|
|
1074
|
+
# the first one if one was found.
|
|
1075
|
+
if (( suffix = ((InfSuffixRuleOrder & prefixes.keys).first) ))
|
|
1076
|
+
rule = InfSuffixRules[ suffix ][:rule]
|
|
1077
|
+
shortestPrefix = InfSuffixRules[ suffix ][:word1]
|
|
1078
|
+
$stderr.puts "Using rule %p (%p) for suffix %p" %
|
|
1079
|
+
[ rule, shortestPrefix, suffix ] if $DEBUG
|
|
1080
|
+
|
|
1081
|
+
case shortestPrefix
|
|
1082
|
+
when 0
|
|
1083
|
+
word1 = prefixes[ suffix ][ 0 ]
|
|
1084
|
+
word2 = prefixes[ suffix ][ 1 ]
|
|
1085
|
+
$stderr.puts "For sp = 0: word1: %p, word2: %p" %
|
|
1086
|
+
[ word1, word2 ] if $DEBUG
|
|
1087
|
+
|
|
1088
|
+
when -1
|
|
1089
|
+
word1 = prefixes[ suffix ].last +
|
|
1090
|
+
InfSuffixRules[ suffix ][:suffix1]
|
|
1091
|
+
word2 = ''
|
|
1092
|
+
$stderr.puts "For sp = -1: word1: %p, word2: %p" %
|
|
1093
|
+
[ word1, word2 ] if $DEBUG
|
|
1094
|
+
|
|
1095
|
+
when -2
|
|
1096
|
+
word1 = prefixes[ suffix ].last +
|
|
1097
|
+
InfSuffixRules[ suffix ][:suffix1]
|
|
1098
|
+
word2 = prefixes[ suffix ].last
|
|
1099
|
+
$stderr.puts "For sp = -2: word1: %p, word2: %p" %
|
|
1100
|
+
[ word1, word2 ] if $DEBUG
|
|
1101
|
+
|
|
1102
|
+
when -3
|
|
1103
|
+
word1 = prefixes[ suffix ].last +
|
|
1104
|
+
InfSuffixRules[ suffix ][:suffix1]
|
|
1105
|
+
word2 = prefixes[ suffix ].last +
|
|
1106
|
+
InfSuffixRules[ suffix ][:suffix2]
|
|
1107
|
+
$stderr.puts "For sp = -3: word1: %p, word2: %p" %
|
|
1108
|
+
[ word1, word2 ] if $DEBUG
|
|
1109
|
+
|
|
1110
|
+
when -4
|
|
1111
|
+
word1 = word
|
|
1112
|
+
word2 = ''
|
|
1113
|
+
$stderr.puts "For sp = -4: word1: %p, word2: %p" %
|
|
1114
|
+
[ word1, word2 ] if $DEBUG
|
|
1115
|
+
|
|
1116
|
+
else
|
|
1117
|
+
raise IndexError,
|
|
1118
|
+
"Couldn't find rule for shortest prefix %p" %
|
|
1119
|
+
shortestPrefix
|
|
1120
|
+
end
|
|
1121
|
+
|
|
1122
|
+
# Rules 12b and 15: Strip off 'ed' or 'ing'.
|
|
1123
|
+
if rule == '12b' or rule == '15'
|
|
1124
|
+
# Do we have a monosyllable of this form:
|
|
1125
|
+
# o 0+ Consonants
|
|
1126
|
+
# o 1+ Vowel
|
|
1127
|
+
# o 2 Non-wx
|
|
1128
|
+
# Eg: tipped => tipp?
|
|
1129
|
+
# Then return tip and tipp.
|
|
1130
|
+
# Eg: swimming => swimm?
|
|
1131
|
+
# Then return tipswim and swimm.
|
|
1132
|
+
|
|
1133
|
+
if /^([^aeiou]*[aeiou]+)([^wx])\2$/ =~ word2
|
|
1134
|
+
word1 = $1 + $2
|
|
1135
|
+
word2 = $1 + $2 + $2
|
|
1136
|
+
end
|
|
1137
|
+
end
|
|
1138
|
+
end
|
|
1139
|
+
end
|
|
1140
|
+
|
|
1141
|
+
return Infinitive::new( word1, word2, suffix, rule )
|
|
1142
|
+
end
|
|
1143
|
+
|
|
1144
|
+
end # module EN::Linguistics
|
|
1145
|
+
|