nameparser 1.2.0__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -0,0 +1,635 @@
1
+ FIRST_NAME_TITLES = set([
2
+ 'aunt',
3
+ 'auntie',
4
+ 'brother',
5
+ 'dame',
6
+ 'father',
7
+ 'king',
8
+ 'maid',
9
+ 'master',
10
+ 'mother',
11
+ 'pope',
12
+ 'queen',
13
+ 'sir',
14
+ 'sister',
15
+ 'uncle',
16
+ 'sheikh',
17
+ 'sheik',
18
+ 'shaik',
19
+ 'shayk',
20
+ 'shaykh',
21
+ 'shaikh',
22
+ 'cheikh',
23
+ 'shekh',
24
+ ])
25
+ """
26
+ When these titles appear with a single other name, that name is a first name, e.g.
27
+ "Sir John", "Sister Mary", "Queen Elizabeth".
28
+ """
29
+
30
+ #: **Cannot include things that could also be first names**, e.g. "dean".
31
+ #: Many of these from wikipedia: https://en.wikipedia.org/wiki/Title.
32
+ #: The parser recognizes chains of these including conjunctions allowing
33
+ #: recognition titles like "Deputy Secretary of State".
34
+ TITLES = FIRST_NAME_TITLES | set([
35
+ "attaché",
36
+ "chargé d'affaires",
37
+ "king's",
38
+ "marchioness",
39
+ "marquess",
40
+ "marquis",
41
+ "marquise",
42
+ "queen's",
43
+ '10th',
44
+ '1lt',
45
+ '1sgt',
46
+ '1st',
47
+ '1stlt',
48
+ '1stsgt',
49
+ '2lt',
50
+ '2nd',
51
+ '2ndlt',
52
+ '3rd',
53
+ '4th',
54
+ '5th',
55
+ '6th',
56
+ '7th',
57
+ '8th',
58
+ '9th',
59
+ 'a1c',
60
+ 'ab',
61
+ 'abbess',
62
+ 'abbot',
63
+ 'abolitionist',
64
+ 'academic',
65
+ 'acolyte',
66
+ 'activist',
67
+ 'actor ',
68
+ 'actress',
69
+ 'adept',
70
+ 'adjutant',
71
+ 'adm',
72
+ 'admiral',
73
+ 'advertising',
74
+ 'adviser',
75
+ 'advocate',
76
+ 'air',
77
+ 'akhoond',
78
+ 'alderman',
79
+ 'almoner',
80
+ 'ambassador',
81
+ 'amn',
82
+ 'analytics',
83
+ 'anarchist',
84
+ 'animator',
85
+ 'anthropologist',
86
+ 'appellate',
87
+ 'apprentice',
88
+ 'arbitrator',
89
+ 'archbishop',
90
+ 'archdeacon',
91
+ 'archdruid',
92
+ 'archduchess',
93
+ 'archduke',
94
+ 'archeologist',
95
+ 'architect',
96
+ 'arhat',
97
+ 'army',
98
+ 'arranger',
99
+ 'assistant',
100
+ 'assoc',
101
+ 'associate',
102
+ 'asst',
103
+ 'astronomer',
104
+ 'attache',
105
+ 'attorney',
106
+ 'author',
107
+ 'award-winning',
108
+ 'ayatollah',
109
+ 'baba',
110
+ 'bailiff',
111
+ 'ballet',
112
+ 'bandleader',
113
+ 'banker',
114
+ 'banner',
115
+ 'bard',
116
+ 'baron',
117
+ 'baroness',
118
+ 'barrister',
119
+ 'baseball',
120
+ 'bearer',
121
+ 'behavioral',
122
+ 'bench',
123
+ 'bg',
124
+ 'bgen',
125
+ 'biblical',
126
+ 'bibliographer',
127
+ 'biochemist',
128
+ 'biographer',
129
+ 'biologist',
130
+ 'bishop',
131
+ 'blessed',
132
+ 'blogger',
133
+ 'blues',
134
+ 'bodhisattva',
135
+ 'bookseller',
136
+ 'botanist',
137
+ 'bp',
138
+ 'brigadier',
139
+ 'briggen',
140
+ 'british',
141
+ 'broadcaster',
142
+ 'buddha',
143
+ 'burgess',
144
+ 'burlesque',
145
+ 'business',
146
+ 'businessman',
147
+ 'businesswoman',
148
+ 'bwana',
149
+ 'canon',
150
+ 'capt',
151
+ 'captain',
152
+ 'cardinal',
153
+ 'cartographer',
154
+ 'cartoonist',
155
+ 'catholicos',
156
+ 'ccmsgt',
157
+ 'cdr',
158
+ 'celebrity',
159
+ 'ceo',
160
+ 'cfo',
161
+ 'chair',
162
+ 'chairs',
163
+ 'chancellor',
164
+ 'chaplain',
165
+ 'chef',
166
+ 'chemist',
167
+ 'chief',
168
+ 'chieftain',
169
+ 'choreographer',
170
+ 'civil',
171
+ 'classical',
172
+ 'clergyman',
173
+ 'clerk',
174
+ 'cmsaf',
175
+ 'cmsgt',
176
+ 'co-chair',
177
+ 'co-chairs',
178
+ 'co-founder',
179
+ 'coach',
180
+ 'col',
181
+ 'collector',
182
+ 'colonel',
183
+ 'comedian',
184
+ 'comedienne',
185
+ 'comic',
186
+ 'commander',
187
+ 'commander-in-chief',
188
+ 'commodore',
189
+ 'composer',
190
+ 'compositeur',
191
+ 'comptroller',
192
+ 'computer',
193
+ 'comtesse',
194
+ 'conductor',
195
+ 'consultant',
196
+ 'controller',
197
+ 'corporal',
198
+ 'corporate',
199
+ 'correspondent',
200
+ 'councillor',
201
+ 'counselor',
202
+ 'count',
203
+ 'countess',
204
+ 'courtier',
205
+ 'cpl',
206
+ 'cpo',
207
+ 'cpt',
208
+ 'credit',
209
+ 'criminal',
210
+ 'criminologist',
211
+ 'critic',
212
+ 'csm',
213
+ 'curator',
214
+ 'customs',
215
+ 'cwo-2',
216
+ 'cwo-3',
217
+ 'cwo-4',
218
+ 'cwo-5',
219
+ 'cwo2',
220
+ 'cwo3',
221
+ 'cwo4',
222
+ 'cwo5',
223
+ 'cyclist',
224
+ 'dancer',
225
+ 'dcn',
226
+ 'deacon',
227
+ 'delegate',
228
+ 'deputy',
229
+ 'designated',
230
+ 'designer',
231
+ 'detective',
232
+ 'developer',
233
+ 'diplomat',
234
+ 'dir',
235
+ 'director',
236
+ 'discovery',
237
+ 'dissident',
238
+ 'district',
239
+ 'division',
240
+ 'do',
241
+ 'docent',
242
+ 'docket',
243
+ 'doctor',
244
+ 'doyen',
245
+ 'dpty',
246
+ 'dr',
247
+ 'dra',
248
+ 'dramatist',
249
+ 'druid',
250
+ 'drummer',
251
+ 'duchesse',
252
+ # 'duke', # a common first name
253
+ 'dutchess',
254
+ 'ecologist',
255
+ 'economist',
256
+ 'editor',
257
+ 'edmi',
258
+ 'edohen',
259
+ 'educator',
260
+ 'effendi',
261
+ 'ekegbian',
262
+ 'elerunwon',
263
+ 'eminence',
264
+ 'emperor',
265
+ 'empress',
266
+ 'engineer',
267
+ 'english',
268
+ 'ens',
269
+ 'entertainer',
270
+ 'entrepreneur',
271
+ 'envoy',
272
+ 'essayist',
273
+ 'evangelist',
274
+ 'excellency',
275
+ 'excellent',
276
+ 'exec',
277
+ 'executive',
278
+ 'expert',
279
+ 'fadm',
280
+ 'family',
281
+ 'federal',
282
+ 'field',
283
+ 'film',
284
+ 'financial',
285
+ 'first',
286
+ 'flag',
287
+ 'flying',
288
+ 'foreign',
289
+ 'forester',
290
+ 'founder',
291
+ 'fr',
292
+ 'friar',
293
+ 'gaf',
294
+ 'gen',
295
+ 'general',
296
+ 'generalissimo',
297
+ 'gentiluomo',
298
+ 'giani',
299
+ 'goodman',
300
+ 'goodwife',
301
+ 'governor',
302
+ 'graf',
303
+ 'grand',
304
+ 'group',
305
+ 'guitarist',
306
+ 'guru',
307
+ 'gyani',
308
+ 'gysgt',
309
+ 'hajji',
310
+ 'headman',
311
+ 'heir',
312
+ 'heiress',
313
+ 'her',
314
+ 'hereditary',
315
+ 'high',
316
+ 'highness',
317
+ 'his',
318
+ 'historian',
319
+ 'historicus',
320
+ 'historien',
321
+ 'holiness',
322
+ 'hon', # sorry Hon Solo, but judges seem more common.
323
+ 'honorable',
324
+ 'honourable',
325
+ 'host',
326
+ 'illustrator',
327
+ 'imam',
328
+ 'industrialist',
329
+ 'information',
330
+ 'instructor',
331
+ 'intelligence',
332
+ 'intendant',
333
+ 'inventor',
334
+ 'investigator',
335
+ 'investor',
336
+ 'journalist',
337
+ 'journeyman',
338
+ 'jr',
339
+ 'judge',
340
+ 'judicial',
341
+ 'junior',
342
+ 'jurist',
343
+ 'keyboardist',
344
+ 'kingdom',
345
+ 'knowledge',
346
+ 'lady',
347
+ 'lama',
348
+ 'lamido',
349
+ 'law',
350
+ 'lawyer',
351
+ 'lcdr',
352
+ 'lcpl',
353
+ 'leader',
354
+ 'lecturer',
355
+ 'legal',
356
+ 'librarian',
357
+ 'lieutenant',
358
+ 'linguist',
359
+ 'literary',
360
+ 'lord',
361
+ 'lt',
362
+ 'ltc',
363
+ 'ltcol',
364
+ 'ltg',
365
+ 'ltgen',
366
+ 'ltjg',
367
+ 'lyricist',
368
+ 'madam',
369
+ 'madame',
370
+ 'mademoiselle',
371
+ 'mag',
372
+ 'mag-judge',
373
+ 'mag/judge',
374
+ 'magistrate',
375
+ 'magistrate-judge',
376
+ 'magnate',
377
+ 'maharajah',
378
+ 'maharani',
379
+ 'mahdi',
380
+ 'maj',
381
+ 'majesty',
382
+ 'majgen',
383
+ 'manager',
384
+ 'marcher',
385
+ 'marchess',
386
+ 'marketing',
387
+ 'marquis',
388
+ 'mathematician',
389
+ 'mathematics',
390
+ 'matriarch',
391
+ 'mayor',
392
+ 'mcpo',
393
+ 'mcpoc',
394
+ 'mcpon',
395
+ 'md',
396
+ 'member',
397
+ 'memoirist',
398
+ 'merchant',
399
+ 'met',
400
+ 'metropolitan',
401
+ 'mg',
402
+ 'mgr',
403
+ 'mgysgt',
404
+ 'military',
405
+ 'minister',
406
+ 'miss',
407
+ 'misses',
408
+ 'missionary',
409
+ 'mister',
410
+ 'mlle',
411
+ 'mme',
412
+ 'mobster',
413
+ 'model',
414
+ 'monk',
415
+ 'monsignor',
416
+ 'most',
417
+ 'mountaineer',
418
+ 'mpco-cg',
419
+ 'mr',
420
+ 'mrs',
421
+ 'ms',
422
+ 'msg',
423
+ 'msgt',
424
+ 'mufti',
425
+ 'mullah',
426
+ 'municipal',
427
+ 'murshid',
428
+ 'musician',
429
+ 'musicologist',
430
+ 'mx',
431
+ 'mystery',
432
+ 'nanny',
433
+ 'narrator',
434
+ 'national',
435
+ 'naturalist',
436
+ 'navy',
437
+ 'neuroscientist',
438
+ 'novelist',
439
+ 'nurse',
440
+ 'obstetritian',
441
+ 'officer',
442
+ 'opera',
443
+ 'operating',
444
+ 'ornithologist',
445
+ 'painter',
446
+ 'paleontologist',
447
+ 'pastor',
448
+ 'patriarch',
449
+ 'pediatrician',
450
+ 'personality',
451
+ 'petty',
452
+ 'pfc',
453
+ 'pharaoh',
454
+ 'phd',
455
+ 'philantropist',
456
+ 'philosopher',
457
+ 'photographer',
458
+ 'physician',
459
+ 'physicist',
460
+ 'pianist',
461
+ 'pilot',
462
+ 'pioneer',
463
+ 'pir',
464
+ 'player',
465
+ 'playwright',
466
+ 'po1',
467
+ 'po2',
468
+ 'po3',
469
+ 'poet',
470
+ 'police',
471
+ 'political',
472
+ 'politician',
473
+ 'prefect',
474
+ 'prelate',
475
+ 'premier',
476
+ 'pres',
477
+ 'presbyter',
478
+ 'president',
479
+ 'presiding',
480
+ 'priest',
481
+ 'priestess',
482
+ 'primate',
483
+ 'prime',
484
+ 'prin',
485
+ 'prince',
486
+ 'princess',
487
+ 'principal',
488
+ 'printer',
489
+ 'printmaker',
490
+ 'prior',
491
+ 'private',
492
+ 'pro',
493
+ 'producer',
494
+ 'prof',
495
+ 'professor',
496
+ 'provost',
497
+ 'pslc',
498
+ 'psychiatrist',
499
+ 'psychologist',
500
+ 'publisher',
501
+ 'pursuivant',
502
+ 'pv2',
503
+ 'pvt',
504
+ 'rabbi',
505
+ 'radio',
506
+ 'radm',
507
+ 'rangatira',
508
+ 'ranger',
509
+ 'rdml',
510
+ 'rear',
511
+ 'rebbe',
512
+ 'registrar',
513
+ 'rep',
514
+ 'representative',
515
+ 'researcher',
516
+ 'resident',
517
+ 'rev',
518
+ 'revenue',
519
+ 'reverend',
520
+ 'right',
521
+ 'risk',
522
+ 'rock',
523
+ 'royal',
524
+ 'rt',
525
+ 'sa',
526
+ 'sailor',
527
+ 'saint',
528
+ 'sainte',
529
+ 'saoshyant',
530
+ 'satirist',
531
+ 'scholar',
532
+ 'schoolmaster',
533
+ 'scientist',
534
+ 'scpo',
535
+ 'screenwriter',
536
+ 'se',
537
+ 'secretary',
538
+ 'security',
539
+ 'seigneur',
540
+ 'senator',
541
+ 'senior',
542
+ 'senior-judge',
543
+ 'sergeant',
544
+ 'servant',
545
+ 'sfc',
546
+ 'sgm',
547
+ 'sgt',
548
+ 'sgtmaj',
549
+ 'sgtmajmc',
550
+ 'shehu',
551
+ 'sheikh',
552
+ 'sheriff',
553
+ 'siddha',
554
+ 'singer',
555
+ 'singer-songwriter',
556
+ 'sma',
557
+ 'smsgt',
558
+ 'sn',
559
+ 'soccer',
560
+ 'social',
561
+ 'sociologist',
562
+ 'software',
563
+ 'soldier',
564
+ 'solicitor',
565
+ 'soprano',
566
+ 'spc',
567
+ 'speaker',
568
+ 'special',
569
+ 'sr',
570
+ 'sra',
571
+ 'srta',
572
+ 'ssg',
573
+ 'ssgt',
574
+ 'st',
575
+ 'staff',
576
+ 'state',
577
+ 'states',
578
+ 'strategy',
579
+ 'subaltern',
580
+ 'subedar',
581
+ 'suffragist',
582
+ 'sultan',
583
+ 'sultana',
584
+ 'superior',
585
+ 'supreme',
586
+ 'surgeon',
587
+ 'swami',
588
+ 'swordbearer',
589
+ 'sysselmann',
590
+ 'tax',
591
+ 'teacher',
592
+ 'technical',
593
+ 'technologist',
594
+ 'television ',
595
+ 'tenor',
596
+ 'theater',
597
+ 'theatre',
598
+ 'theologian',
599
+ 'theorist',
600
+ 'timi',
601
+ 'tirthankar',
602
+ 'translator',
603
+ 'travel',
604
+ 'treasurer',
605
+ 'tsar',
606
+ 'tsarina',
607
+ 'tsgt',
608
+ 'uk',
609
+ 'united',
610
+ 'us',
611
+ 'vadm',
612
+ 'vardapet',
613
+ 'vc',
614
+ 'venerable',
615
+ 'verderer',
616
+ 'vicar',
617
+ 'vice',
618
+ 'viscount',
619
+ 'vizier',
620
+ 'vocalist',
621
+ 'voice',
622
+ 'warden',
623
+ 'warrant',
624
+ 'wing',
625
+ 'wm',
626
+ 'wo-1',
627
+ 'wo1',
628
+ 'wo2',
629
+ 'wo3',
630
+ 'wo4',
631
+ 'wo5',
632
+ 'woodman',
633
+ 'writer',
634
+ 'zoologist',
635
+ ])