Coverage for /pythoncovmergedfiles/medio/medio/usr/local/lib/python3.8/site-packages/nameparser/config/titles.py: 100%
3 statements
« prev ^ index » next coverage.py v7.2.7, created at 2023-06-07 06:08 +0000
« prev ^ index » next coverage.py v7.2.7, created at 2023-06-07 06:08 +0000
1# -*- coding: utf-8 -*-
2from __future__ import unicode_literals
4FIRST_NAME_TITLES = set([
5 'aunt',
6 'auntie',
7 'brother',
8 'dame',
9 'father',
10 'king',
11 'maid',
12 'master',
13 'mother',
14 'pope',
15 'queen',
16 'sir',
17 'sister',
18 'uncle',
19 'sheikh',
20 'sheik',
21 'shaik',
22 'shayk',
23 'shaykh',
24 'shaikh',
25 'cheikh',
26 'shekh',
27])
28"""
29When these titles appear with a single other name, that name is a first name, e.g.
30"Sir John", "Sister Mary", "Queen Elizabeth".
31"""
33#: **Cannot include things that could also be first names**, e.g. "dean".
34#: Many of these from wikipedia: https://en.wikipedia.org/wiki/Title.
35#: The parser recognizes chains of these including conjunctions allowing
36#: recognition titles like "Deputy Secretary of State".
37TITLES = FIRST_NAME_TITLES | set([
38 "attaché",
39 "chargé d'affaires",
40 "king's",
41 "marchioness",
42 "marquess",
43 "marquis",
44 "marquise",
45 "queen's",
46 '10th',
47 '1lt',
48 '1sgt',
49 '1st',
50 '1stlt',
51 '1stsgt',
52 '2lt',
53 '2nd',
54 '2ndlt',
55 '3rd',
56 '4th',
57 '5th',
58 '6th',
59 '7th',
60 '8th',
61 '9th',
62 'a1c',
63 'ab',
64 'abbess',
65 'abbot',
66 'abolitionist',
67 'academic',
68 'acolyte',
69 'activist',
70 'actor ',
71 'actress',
72 'adept',
73 'adjutant',
74 'adm',
75 'admiral',
76 'advertising',
77 'adviser',
78 'advocate',
79 'air',
80 'akhoond',
81 'alderman',
82 'almoner',
83 'ambassador',
84 'amn',
85 'analytics',
86 'anarchist',
87 'animator',
88 'anthropologist',
89 'appellate',
90 'apprentice',
91 'arbitrator',
92 'archbishop',
93 'archdeacon',
94 'archdruid',
95 'archduchess',
96 'archduke',
97 'archeologist',
98 'architect',
99 'arhat',
100 'army',
101 'arranger',
102 'assistant',
103 'assoc',
104 'associate',
105 'asst',
106 'astronomer',
107 'attache',
108 'attorney',
109 'author',
110 'award-winning',
111 'ayatollah',
112 'baba',
113 'bailiff',
114 'ballet',
115 'bandleader',
116 'banker',
117 'banner',
118 'bard',
119 'baron',
120 'baroness',
121 'barrister',
122 'baseball',
123 'bearer',
124 'behavioral',
125 'bench',
126 'bg',
127 'bgen',
128 'biblical',
129 'bibliographer',
130 'biochemist',
131 'biographer',
132 'biologist',
133 'bishop',
134 'blessed',
135 'blogger',
136 'blues',
137 'bodhisattva',
138 'bookseller',
139 'botanist',
140 'bp',
141 'brigadier',
142 'briggen',
143 'british',
144 'broadcaster',
145 'buddha',
146 'burgess',
147 'burlesque',
148 'business',
149 'businessman',
150 'businesswoman',
151 'bwana',
152 'canon',
153 'capt',
154 'captain',
155 'cardinal',
156 'cartographer',
157 'cartoonist',
158 'catholicos',
159 'ccmsgt',
160 'cdr',
161 'celebrity',
162 'ceo',
163 'cfo',
164 'chair',
165 'chairs',
166 'chancellor',
167 'chaplain',
168 'chef',
169 'chemist',
170 'chief',
171 'chieftain',
172 'choreographer',
173 'civil',
174 'classical',
175 'clergyman',
176 'clerk',
177 'cmsaf',
178 'cmsgt',
179 'co-chair',
180 'co-chairs',
181 'co-founder',
182 'coach',
183 'col',
184 'collector',
185 'colonel',
186 'comedian',
187 'comedienne',
188 'comic',
189 'commander',
190 'commander-in-chief',
191 'commodore',
192 'composer',
193 'compositeur',
194 'comptroller',
195 'computer',
196 'comtesse',
197 'conductor',
198 'consultant',
199 'controller',
200 'corporal',
201 'corporate',
202 'correspondent',
203 'councillor',
204 'counselor',
205 'count',
206 'countess',
207 'courtier',
208 'cpl',
209 'cpo',
210 'cpt',
211 'credit',
212 'criminal',
213 'criminologist',
214 'critic',
215 'csm',
216 'curator',
217 'customs',
218 'cwo-2',
219 'cwo-3',
220 'cwo-4',
221 'cwo-5',
222 'cwo2',
223 'cwo3',
224 'cwo4',
225 'cwo5',
226 'cyclist',
227 'dancer',
228 'dcn',
229 'deacon',
230 'delegate',
231 'deputy',
232 'designated',
233 'designer',
234 'detective',
235 'developer',
236 'diplomat',
237 'dir',
238 'director',
239 'discovery',
240 'dissident',
241 'district',
242 'division',
243 'do',
244 'docent',
245 'docket',
246 'doctor',
247 'doyen',
248 'dpty',
249 'dr',
250 'dra',
251 'dramatist',
252 'druid',
253 'drummer',
254 'duchesse',
255 # 'duke', # a common first name
256 'dutchess',
257 'ecologist',
258 'economist',
259 'editor',
260 'edmi',
261 'edohen',
262 'educator',
263 'effendi',
264 'ekegbian',
265 'elerunwon',
266 'eminence',
267 'emperor',
268 'empress',
269 'engineer',
270 'english',
271 'ens',
272 'entertainer',
273 'entrepreneur',
274 'envoy',
275 'essayist',
276 'evangelist',
277 'excellency',
278 'excellent',
279 'exec',
280 'executive',
281 'expert',
282 'fadm',
283 'family',
284 'federal',
285 'field',
286 'film',
287 'financial',
288 'first',
289 'flag',
290 'flying',
291 'foreign',
292 'forester',
293 'founder',
294 'fr',
295 'friar',
296 'gaf',
297 'gen',
298 'general',
299 'generalissimo',
300 'gentiluomo',
301 'giani',
302 'goodman',
303 'goodwife',
304 'governor',
305 'graf',
306 'grand',
307 'group',
308 'guitarist',
309 'guru',
310 'gyani',
311 'gysgt',
312 'hajji',
313 'headman',
314 'heir',
315 'heiress',
316 'her',
317 'hereditary',
318 'high',
319 'highness',
320 'his',
321 'historian',
322 'historicus',
323 'historien',
324 'holiness',
325 'hon', # sorry Hon Solo, but judges seem more common.
326 'honorable',
327 'honourable',
328 'host',
329 'illustrator',
330 'imam',
331 'industrialist',
332 'information',
333 'instructor',
334 'intelligence',
335 'intendant',
336 'inventor',
337 'investigator',
338 'investor',
339 'journalist',
340 'journeyman',
341 'jr',
342 'judge',
343 'judicial',
344 'junior',
345 'jurist',
346 'keyboardist',
347 'kingdom',
348 'knowledge',
349 'lady',
350 'lama',
351 'lamido',
352 'law',
353 'lawyer',
354 'lcdr',
355 'lcpl',
356 'leader',
357 'lecturer',
358 'legal',
359 'librarian',
360 'lieutenant',
361 'linguist',
362 'literary',
363 'lord',
364 'lt',
365 'ltc',
366 'ltcol',
367 'ltg',
368 'ltgen',
369 'ltjg',
370 'lyricist',
371 'madam',
372 'madame',
373 'mademoiselle',
374 'mag',
375 'mag-judge',
376 'mag/judge',
377 'magistrate',
378 'magistrate-judge',
379 'magnate',
380 'maharajah',
381 'maharani',
382 'mahdi',
383 'maj',
384 'majesty',
385 'majgen',
386 'manager',
387 'marcher',
388 'marchess',
389 'marketing',
390 'marquis',
391 'mathematician',
392 'mathematics',
393 'matriarch',
394 'mayor',
395 'mcpo',
396 'mcpoc',
397 'mcpon',
398 'md',
399 'member',
400 'memoirist',
401 'merchant',
402 'met',
403 'metropolitan',
404 'mg',
405 'mgr',
406 'mgysgt',
407 'military',
408 'minister',
409 'miss',
410 'misses',
411 'missionary',
412 'mister',
413 'mlle',
414 'mme',
415 'mobster',
416 'model',
417 'monk',
418 'monsignor',
419 'most',
420 'mountaineer',
421 'mpco-cg',
422 'mr',
423 'mrs',
424 'ms',
425 'msg',
426 'msgt',
427 'mufti',
428 'mullah',
429 'municipal',
430 'murshid',
431 'musician',
432 'musicologist',
433 'mx',
434 'mystery',
435 'nanny',
436 'narrator',
437 'national',
438 'naturalist',
439 'navy',
440 'neuroscientist',
441 'novelist',
442 'nurse',
443 'obstetritian',
444 'officer',
445 'opera',
446 'operating',
447 'ornithologist',
448 'painter',
449 'paleontologist',
450 'pastor',
451 'patriarch',
452 'pediatrician',
453 'personality',
454 'petty',
455 'pfc',
456 'pharaoh',
457 'phd',
458 'philantropist',
459 'philosopher',
460 'photographer',
461 'physician',
462 'physicist',
463 'pianist',
464 'pilot',
465 'pioneer',
466 'pir',
467 'player',
468 'playwright',
469 'po1',
470 'po2',
471 'po3',
472 'poet',
473 'police',
474 'political',
475 'politician',
476 'prefect',
477 'prelate',
478 'premier',
479 'pres',
480 'presbyter',
481 'president',
482 'presiding',
483 'priest',
484 'priestess',
485 'primate',
486 'prime',
487 'prin',
488 'prince',
489 'princess',
490 'principal',
491 'printer',
492 'printmaker',
493 'prior',
494 'private',
495 'pro',
496 'producer',
497 'prof',
498 'professor',
499 'provost',
500 'pslc',
501 'psychiatrist',
502 'psychologist',
503 'publisher',
504 'pursuivant',
505 'pv2',
506 'pvt',
507 'rabbi',
508 'radio',
509 'radm',
510 'rangatira',
511 'ranger',
512 'rdml',
513 'rear',
514 'rebbe',
515 'registrar',
516 'rep',
517 'representative',
518 'researcher',
519 'resident',
520 'rev',
521 'revenue',
522 'reverend',
523 'right',
524 'risk',
525 'rock',
526 'royal',
527 'rt',
528 'sa',
529 'sailor',
530 'saint',
531 'sainte',
532 'saoshyant',
533 'satirist',
534 'scholar',
535 'schoolmaster',
536 'scientist',
537 'scpo',
538 'screenwriter',
539 'se',
540 'secretary',
541 'security',
542 'seigneur',
543 'senator',
544 'senior',
545 'senior-judge',
546 'sergeant',
547 'servant',
548 'sfc',
549 'sgm',
550 'sgt',
551 'sgtmaj',
552 'sgtmajmc',
553 'shehu',
554 'sheikh',
555 'sheriff',
556 'siddha',
557 'singer',
558 'singer-songwriter',
559 'sma',
560 'smsgt',
561 'sn',
562 'soccer',
563 'social',
564 'sociologist',
565 'software',
566 'soldier',
567 'solicitor',
568 'soprano',
569 'spc',
570 'speaker',
571 'special',
572 'sr',
573 'sra',
574 'srta',
575 'ssg',
576 'ssgt',
577 'st',
578 'staff',
579 'state',
580 'states',
581 'strategy',
582 'subaltern',
583 'subedar',
584 'suffragist',
585 'sultan',
586 'sultana',
587 'superior',
588 'supreme',
589 'surgeon',
590 'swami',
591 'swordbearer',
592 'sysselmann',
593 'tax',
594 'teacher',
595 'technical',
596 'technologist',
597 'television ',
598 'tenor',
599 'theater',
600 'theatre',
601 'theologian',
602 'theorist',
603 'timi',
604 'tirthankar',
605 'translator',
606 'travel',
607 'treasurer',
608 'tsar',
609 'tsarina',
610 'tsgt',
611 'uk',
612 'united',
613 'us',
614 'vadm',
615 'vardapet',
616 'vc',
617 'venerable',
618 'verderer',
619 'vicar',
620 'vice',
621 'viscount',
622 'vizier',
623 'vocalist',
624 'voice',
625 'warden',
626 'warrant',
627 'wing',
628 'wm',
629 'wo-1',
630 'wo1',
631 'wo2',
632 'wo3',
633 'wo4',
634 'wo5',
635 'woodman',
636 'writer',
637 'zoologist',
638])