|
1
|
|
|
# Copyright 2018-2020 by Christopher C. Little. |
|
2
|
|
|
# This file is part of Abydos. |
|
3
|
|
|
# |
|
4
|
|
|
# Abydos is free software: you can redistribute it and/or modify |
|
5
|
|
|
# it under the terms of the GNU General Public License as published by |
|
6
|
|
|
# the Free Software Foundation, either version 3 of the License, or |
|
7
|
|
|
# (at your option) any later version. |
|
8
|
|
|
# |
|
9
|
|
|
# Abydos is distributed in the hope that it will be useful, |
|
10
|
|
|
# but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
11
|
|
|
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
|
12
|
|
|
# GNU General Public License for more details. |
|
13
|
|
|
# |
|
14
|
|
|
# You should have received a copy of the GNU General Public License |
|
15
|
|
|
# along with Abydos. If not, see <http://www.gnu.org/licenses/>. |
|
16
|
|
|
|
|
17
|
|
|
"""abydos.fingerprint._synoname. |
|
18
|
|
|
|
|
19
|
1 |
|
Synoname toolcode |
|
20
|
|
|
""" |
|
21
|
|
|
|
|
22
|
|
|
from typing import Tuple |
|
23
|
|
|
|
|
24
|
1 |
|
from ._fingerprint import _Fingerprint |
|
25
|
|
|
|
|
26
|
|
|
__all__ = ['SynonameToolcode'] |
|
27
|
|
|
|
|
28
|
|
|
|
|
29
|
|
|
class SynonameToolcode(_Fingerprint): |
|
30
|
|
|
"""Synoname Toolcode. |
|
31
|
1 |
|
|
|
32
|
|
|
Cf. :cite:`Getty:1991,Gross:1991`. |
|
33
|
1 |
|
|
|
34
|
1 |
|
.. versionadded:: 0.3.6 |
|
35
|
|
|
""" |
|
36
|
1 |
|
|
|
37
|
|
|
_synoname_special_table = ( |
|
38
|
|
|
# Roman, match, extra, method |
|
39
|
1 |
|
(False, 'NONE', '', 0), |
|
40
|
|
|
(False, 'aine', '', 3), |
|
41
|
|
|
(False, 'also erroneously', '', 4), |
|
42
|
|
|
(False, 'also identified with the', '', 2), |
|
43
|
|
|
(False, 'also identified with', '', 2), |
|
44
|
|
|
(False, 'archbishop', '', 7), |
|
45
|
|
|
(False, 'atelier', '', 7), |
|
46
|
|
|
(False, 'baron', '', 7), |
|
47
|
1 |
|
(False, 'cadet', '', 3), |
|
48
|
|
|
(False, 'cardinal', '', 7), |
|
49
|
|
|
(False, 'circle of', '', 5), |
|
50
|
|
|
(False, 'circle', '', 5), |
|
51
|
|
|
(False, 'class of', '', 5), |
|
52
|
|
|
(False, 'conde de', '', 7), |
|
53
|
|
|
(False, 'countess', '', 7), |
|
54
|
|
|
(False, 'count', '', 7), |
|
55
|
|
|
(False, "d'", " d'", 15), |
|
56
|
|
|
(False, 'dai', '', 15), |
|
57
|
|
|
(False, "dall'", " dall'", 15), |
|
58
|
|
|
(False, 'dalla', '', 15), |
|
59
|
|
|
(False, 'dalle', '', 15), |
|
60
|
|
|
(False, 'dal', '', 15), |
|
61
|
|
|
(False, 'da', '', 15), |
|
62
|
|
|
(False, 'degli', '', 15), |
|
63
|
|
|
(False, 'della', '', 15), |
|
64
|
|
|
(False, 'del', '', 15), |
|
65
|
|
|
(False, 'den', '', 15), |
|
66
|
|
|
(False, 'der altere', '', 3), |
|
67
|
|
|
(False, 'der jungere', '', 3), |
|
68
|
|
|
(False, 'der', '', 15), |
|
69
|
|
|
(False, 'de la', '', 15), |
|
70
|
|
|
(False, 'des', '', 15), |
|
71
|
|
|
(False, "de'", " de'", 15), |
|
72
|
|
|
(False, 'de', '', 15), |
|
73
|
|
|
(False, 'di ser', '', 7), |
|
74
|
|
|
(False, 'di', '', 15), |
|
75
|
|
|
(False, 'dos', '', 15), |
|
76
|
|
|
(False, 'du', '', 15), |
|
77
|
|
|
(False, 'duke of', '', 7), |
|
78
|
|
|
(False, 'earl of', '', 7), |
|
79
|
|
|
(False, 'el', '', 15), |
|
80
|
|
|
(False, 'fils', '', 3), |
|
81
|
|
|
(False, 'florentine follower of', '', 5), |
|
82
|
|
|
(False, 'follower of', '', 5), |
|
83
|
|
|
(False, 'fra', '', 7), |
|
84
|
|
|
(False, 'freiherr von', '', 7), |
|
85
|
|
|
(False, 'giovane', '', 7), |
|
86
|
|
|
(False, 'group', '', 5), |
|
87
|
|
|
(True, 'iii', '', 3), |
|
88
|
|
|
(True, 'ii', '', 3), |
|
89
|
|
|
(False, 'il giovane', '', 7), |
|
90
|
|
|
(False, 'il vecchio', '', 7), |
|
91
|
|
|
(False, 'il', '', 15), |
|
92
|
|
|
(False, "in't", '', 7), |
|
93
|
|
|
(False, 'in het', '', 7), |
|
94
|
|
|
(True, 'iv', '', 3), |
|
95
|
|
|
(True, 'ix', '', 3), |
|
96
|
|
|
(True, 'i', '', 3), |
|
97
|
|
|
(False, 'jr.', '', 3), |
|
98
|
|
|
(False, 'jr', '', 3), |
|
99
|
|
|
(False, 'juniore', '', 3), |
|
100
|
|
|
(False, 'junior', '', 3), |
|
101
|
|
|
(False, 'king of', '', 7), |
|
102
|
|
|
(False, "l'", " l'", 15), |
|
103
|
|
|
(False, "l'aine", '', 3), |
|
104
|
|
|
(False, 'la', '', 15), |
|
105
|
|
|
(False, 'le jeune', '', 3), |
|
106
|
|
|
(False, 'le', '', 15), |
|
107
|
|
|
(False, 'lo', '', 15), |
|
108
|
|
|
(False, 'maestro', '', 7), |
|
109
|
|
|
(False, 'maitre', '', 7), |
|
110
|
|
|
(False, 'marchioness', '', 7), |
|
111
|
|
|
(False, 'markgrafin von', '', 7), |
|
112
|
|
|
(False, 'marquess', '', 7), |
|
113
|
|
|
(False, 'marquis', '', 7), |
|
114
|
|
|
(False, 'master of the', '', 7), |
|
115
|
|
|
(False, 'master of', '', 7), |
|
116
|
|
|
(False, 'master known as the', '', 7), |
|
117
|
|
|
(False, 'master with the', '', 7), |
|
118
|
|
|
(False, 'master with', '', 7), |
|
119
|
|
|
(False, 'masters', '', 7), |
|
120
|
|
|
(False, 'master', '', 7), |
|
121
|
|
|
(False, 'meister', '', 7), |
|
122
|
|
|
(False, 'met de', '', 7), |
|
123
|
|
|
(False, 'met', '', 7), |
|
124
|
|
|
(False, 'mlle.', '', 7), |
|
125
|
|
|
(False, 'mlle', '', 7), |
|
126
|
|
|
(False, 'monogrammist', '', 7), |
|
127
|
|
|
(False, 'monsu', '', 7), |
|
128
|
|
|
(False, 'nee', '', 2), |
|
129
|
|
|
(False, 'of', '', 3), |
|
130
|
|
|
(False, 'oncle', '', 3), |
|
131
|
|
|
(False, 'op den', '', 15), |
|
132
|
|
|
(False, 'op de', '', 15), |
|
133
|
|
|
(False, 'or', '', 2), |
|
134
|
|
|
(False, 'over den', '', 15), |
|
135
|
|
|
(False, 'over de', '', 15), |
|
136
|
|
|
(False, 'over', '', 7), |
|
137
|
|
|
(False, 'p.re', '', 7), |
|
138
|
|
|
(False, 'p.r.a.', '', 1), |
|
139
|
|
|
(False, 'padre', '', 7), |
|
140
|
|
|
(False, 'painter', '', 7), |
|
141
|
|
|
(False, 'pere', '', 3), |
|
142
|
|
|
(False, 'possibly identified with', '', 6), |
|
143
|
|
|
(False, 'possibly', '', 6), |
|
144
|
|
|
(False, 'pseudo', '', 15), |
|
145
|
|
|
(False, 'r.a.', '', 1), |
|
146
|
|
|
(False, 'reichsgraf von', '', 7), |
|
147
|
|
|
(False, 'ritter von', '', 7), |
|
148
|
|
|
(False, 'sainte-', ' sainte-', 8), |
|
149
|
|
|
(False, 'sainte', '', 7), |
|
150
|
|
|
(False, 'saint-', ' saint-', 8), |
|
151
|
|
|
(False, 'saint', '', 7), |
|
152
|
|
|
(False, 'santa', '', 15), |
|
153
|
|
|
(False, "sant'", " sant'", 15), |
|
154
|
|
|
(False, 'san', '', 15), |
|
155
|
|
|
(False, 'ser', '', 7), |
|
156
|
|
|
(False, 'seniore', '', 3), |
|
157
|
|
|
(False, 'senior', '', 3), |
|
158
|
|
|
(False, 'sir', '', 5), |
|
159
|
|
|
(False, 'sr.', '', 3), |
|
160
|
|
|
(False, 'sr', '', 3), |
|
161
|
|
|
(False, 'ss.', ' ss.', 14), |
|
162
|
|
|
(False, 'ss', '', 6), |
|
163
|
|
|
(False, 'st-', ' st-', 8), |
|
164
|
|
|
(False, 'st.', ' st.', 15), |
|
165
|
|
|
(False, 'ste-', ' ste-', 8), |
|
166
|
|
|
(False, 'ste.', ' ste.', 15), |
|
167
|
|
|
(False, 'studio', '', 7), |
|
168
|
|
|
(False, 'sub-group', '', 5), |
|
169
|
|
|
(False, 'sultan of', '', 7), |
|
170
|
|
|
(False, 'ten', '', 15), |
|
171
|
|
|
(False, 'ter', '', 15), |
|
172
|
|
|
(False, 'the elder', '', 3), |
|
173
|
|
|
(False, 'the younger', '', 3), |
|
174
|
|
|
(False, 'the', '', 7), |
|
175
|
|
|
(False, 'tot', '', 15), |
|
176
|
|
|
(False, 'unidentified', '', 1), |
|
177
|
|
|
(False, 'van den', '', 15), |
|
178
|
|
|
(False, 'van der', '', 15), |
|
179
|
|
|
(False, 'van de', '', 15), |
|
180
|
|
|
(False, 'vanden', '', 15), |
|
181
|
|
|
(False, 'vander', '', 15), |
|
182
|
|
|
(False, 'van', '', 15), |
|
183
|
|
|
(False, 'vecchia', '', 7), |
|
184
|
|
|
(False, 'vecchio', '', 7), |
|
185
|
|
|
(True, 'viii', '', 3), |
|
186
|
|
|
(True, 'vii', '', 3), |
|
187
|
|
|
(True, 'vi', '', 3), |
|
188
|
|
|
(True, 'v', '', 3), |
|
189
|
|
|
(False, 'vom', '', 7), |
|
190
|
|
|
(False, 'von', '', 15), |
|
191
|
|
|
(False, 'workshop', '', 7), |
|
192
|
|
|
(True, 'xiii', '', 3), |
|
193
|
|
|
(True, 'xii', '', 3), |
|
194
|
|
|
(True, 'xiv', '', 3), |
|
195
|
|
|
(True, 'xix', '', 3), |
|
196
|
|
|
(True, 'xi', '', 3), |
|
197
|
|
|
(True, 'xviii', '', 3), |
|
198
|
|
|
(True, 'xvii', '', 3), |
|
199
|
|
|
(True, 'xvi', '', 3), |
|
200
|
|
|
(True, 'xv', '', 3), |
|
201
|
|
|
(True, 'xx', '', 3), |
|
202
|
|
|
(True, 'x', '', 3), |
|
203
|
|
|
(False, 'y', '', 7), |
|
204
|
|
|
) |
|
205
|
|
|
|
|
206
|
|
|
_method_dict = { |
|
207
|
|
|
'end': 1, |
|
208
|
|
|
'middle': 2, |
|
209
|
|
|
'beginning': 4, |
|
210
|
|
|
'beginning_no_space': 8, |
|
211
|
|
|
} |
|
212
|
|
|
|
|
213
|
|
|
# Fill field 0 (qualifier) |
|
214
|
|
|
_qual_3 = { |
|
215
|
|
|
'adaptation after', |
|
216
|
1 |
|
'after', |
|
217
|
|
|
'assistant of', |
|
218
|
|
|
'assistants of', |
|
219
|
|
|
'circle of', |
|
220
|
|
|
'follower of', |
|
221
|
|
|
'imitator of', |
|
222
|
|
|
'in the style of', |
|
223
|
|
|
'manner of', |
|
224
|
1 |
|
'pupil of', |
|
225
|
|
|
'school of', |
|
226
|
|
|
'studio of', |
|
227
|
|
|
'style of', |
|
228
|
|
|
'workshop of', |
|
229
|
|
|
} |
|
230
|
|
|
_qual_2 = {'copy after', 'copy after?', 'copy of'} |
|
231
|
|
|
_qual_1 = { |
|
232
|
|
|
'ascribed to', |
|
233
|
|
|
'attributed to or copy after', |
|
234
|
|
|
'attributed to', |
|
235
|
|
|
'possibly', |
|
236
|
|
|
} |
|
237
|
|
|
|
|
238
|
|
|
# Fill field 2 (generation) |
|
239
|
|
|
_gen_1 = ( |
|
240
|
1 |
|
'the elder', |
|
241
|
1 |
|
' sr.', |
|
242
|
|
|
' sr', |
|
243
|
|
|
'senior', |
|
244
|
|
|
'der altere', |
|
245
|
|
|
'il vecchio', |
|
246
|
|
|
"l'aine", |
|
247
|
|
|
'p.re', |
|
248
|
|
|
'padre', |
|
249
|
1 |
|
'seniore', |
|
250
|
|
|
'vecchia', |
|
251
|
|
|
'vecchio', |
|
252
|
|
|
) |
|
253
|
|
|
_gen_2 = ( |
|
254
|
|
|
' jr.', |
|
255
|
|
|
' jr', |
|
256
|
|
|
'der jungere', |
|
257
|
|
|
'il giovane', |
|
258
|
|
|
'giovane', |
|
259
|
|
|
'juniore', |
|
260
|
|
|
'junior', |
|
261
|
|
|
'le jeune', |
|
262
|
|
|
'the younger', |
|
263
|
1 |
|
) |
|
264
|
|
|
|
|
265
|
|
|
def fingerprint( |
|
266
|
|
|
self, lname: str, fname: str = '', qual: str = '', normalize: int = 0 |
|
267
|
|
|
) -> str: |
|
268
|
|
|
"""Build the Synoname toolcode. |
|
269
|
|
|
|
|
270
|
|
|
Parameters |
|
271
|
|
|
---------- |
|
272
|
|
|
lname : str |
|
273
|
|
|
Last name |
|
274
|
|
|
fname : str |
|
275
|
1 |
|
First name (can be blank) |
|
276
|
|
|
qual : str |
|
277
|
|
|
Qualifier |
|
278
|
|
|
normalize : int |
|
279
|
|
|
Normalization mode (0, 1, or 2) |
|
280
|
|
|
|
|
281
|
|
|
Returns |
|
282
|
|
|
------- |
|
283
|
|
|
str |
|
284
|
|
|
The transformed names and the synoname toolcode, separated by |
|
285
|
|
|
commas |
|
286
|
|
|
|
|
287
|
|
|
Examples |
|
288
|
|
|
-------- |
|
289
|
|
|
>>> st = SynonameToolcode() |
|
290
|
|
|
>>> st.fingerprint('hat') |
|
291
|
|
|
'hat,,0000000003$$h' |
|
292
|
|
|
>>> st.fingerprint('niall') |
|
293
|
|
|
'niall,,0000000005$$n' |
|
294
|
|
|
>>> st.fingerprint('colin') |
|
295
|
|
|
'colin,,0000000005$$c' |
|
296
|
|
|
>>> st.fingerprint('atcg') |
|
297
|
|
|
'atcg,,0000000004$$a' |
|
298
|
|
|
>>> st.fingerprint('entreatment') |
|
299
|
|
|
'entreatment,,0000000011$$e' |
|
300
|
|
|
|
|
301
|
|
|
>>> st.fingerprint('Ste.-Marie', 'Count John II', normalize=2) |
|
302
|
|
|
'ste.-marie ii,count john,0200491310$015b049a127c$smcji' |
|
303
|
|
|
>>> st.fingerprint('Michelangelo IV', '', 'Workshop of') |
|
304
|
|
|
'michelangelo iv,,3000550015$055b$mi' |
|
305
|
|
|
|
|
306
|
|
|
|
|
307
|
|
|
.. versionadded:: 0.3.0 |
|
308
|
|
|
.. versionchanged:: 0.3.6 |
|
309
|
|
|
Encapsulated in class |
|
310
|
|
|
.. versionchanged:: 0.6.0 |
|
311
|
|
|
Changed to return a comma-separated string instead of 3-tuple of |
|
312
|
|
|
strs |
|
313
|
|
|
|
|
314
|
|
|
""" |
|
315
|
|
|
return ','.join(self.fingerprint_tuple(lname, fname, qual, normalize)) |
|
316
|
|
|
|
|
317
|
|
|
def fingerprint_tuple( |
|
318
|
|
|
self, lname: str, fname: str = '', qual: str = '', normalize: int = 0 |
|
319
|
1 |
|
) -> Tuple[str, str, str]: |
|
320
|
1 |
|
"""Build the Synoname toolcode. |
|
321
|
1 |
|
|
|
322
|
|
|
Parameters |
|
323
|
|
|
---------- |
|
324
|
1 |
|
lname : str |
|
325
|
|
|
Last name |
|
326
|
1 |
|
fname : str |
|
327
|
|
|
First name (can be blank) |
|
328
|
1 |
|
qual : str |
|
329
|
1 |
|
Qualifier |
|
330
|
1 |
|
normalize : int |
|
331
|
1 |
|
Normalization mode (0, 1, or 2) |
|
332
|
1 |
|
|
|
333
|
1 |
|
Returns |
|
334
|
|
|
------- |
|
335
|
|
|
tuple |
|
336
|
1 |
|
The transformed names and the synoname toolcode |
|
337
|
1 |
|
|
|
338
|
|
|
Examples |
|
339
|
1 |
|
-------- |
|
340
|
1 |
|
>>> st = SynonameToolcode() |
|
341
|
1 |
|
>>> st.fingerprint_tuple('hat') |
|
342
|
1 |
|
('hat', '', '0000000003$$h') |
|
343
|
|
|
>>> st.fingerprint_tuple('niall') |
|
344
|
1 |
|
('niall', '', '0000000005$$n') |
|
345
|
1 |
|
>>> st.fingerprint_tuple('colin') |
|
346
|
1 |
|
('colin', '', '0000000005$$c') |
|
347
|
1 |
|
>>> st.fingerprint_tuple('atcg') |
|
348
|
1 |
|
('atcg', '', '0000000004$$a') |
|
349
|
1 |
|
>>> st.fingerprint_tuple('entreatment') |
|
350
|
|
|
('entreatment', '', '0000000011$$e') |
|
351
|
1 |
|
|
|
352
|
1 |
|
>>> st.fingerprint_tuple('Ste.-Marie', 'Count John II', normalize=2) |
|
353
|
1 |
|
('ste.-marie ii', 'count john', '0200491310$015b049a127c$smcji') |
|
354
|
1 |
|
>>> st.fingerprint_tuple('Michelangelo IV', '', 'Workshop of') |
|
355
|
1 |
|
('michelangelo iv', '', '3000550015$055b$mi') |
|
356
|
|
|
|
|
357
|
|
|
|
|
358
|
1 |
|
.. versionadded:: 0.6.0 |
|
359
|
1 |
|
|
|
360
|
1 |
|
""" |
|
361
|
1 |
|
lname = lname.lower() |
|
362
|
1 |
|
fname = fname.lower() |
|
363
|
1 |
|
qual = qual.lower() |
|
364
|
1 |
|
|
|
365
|
1 |
|
# Start with the basic code |
|
366
|
|
|
toolcode = ['0', '0', '0', '000', '00', '00', '$', '', '$', ''] |
|
367
|
|
|
|
|
368
|
1 |
|
full_name = ' '.join((lname, fname)) |
|
369
|
1 |
|
|
|
370
|
1 |
|
if qual in self._qual_3: |
|
371
|
1 |
|
toolcode[0] = '3' |
|
372
|
1 |
|
elif qual in self._qual_2: |
|
373
|
|
|
toolcode[0] = '2' |
|
374
|
|
|
elif qual in self._qual_1: |
|
375
|
|
|
toolcode[0] = '1' |
|
376
|
|
|
|
|
377
|
|
|
# Fill field 1 (punctuation) |
|
378
|
|
|
if '.' in full_name: |
|
379
|
1 |
|
toolcode[1] = '2' |
|
380
|
1 |
|
else: |
|
381
|
|
|
for punct in ',-/:;"&\'()!{|}?$%*+<=>[\\]^_`~': |
|
382
|
|
|
if punct in full_name: |
|
383
|
1 |
|
toolcode[1] = '1' |
|
384
|
1 |
|
break |
|
385
|
1 |
|
|
|
386
|
1 |
|
elderyounger = '' # save elder/younger for possible movement later |
|
387
|
1 |
|
for gen in self._gen_1: |
|
388
|
|
|
if gen in full_name: |
|
389
|
|
|
toolcode[2] = '1' |
|
390
|
1 |
|
elderyounger = gen |
|
391
|
1 |
|
break |
|
392
|
1 |
|
else: |
|
393
|
1 |
|
for gen in self._gen_2: |
|
394
|
1 |
|
if gen in full_name: |
|
395
|
|
|
toolcode[2] = '2' |
|
396
|
1 |
|
elderyounger = gen |
|
397
|
|
|
break |
|
398
|
|
|
|
|
399
|
|
|
# do comma flip |
|
400
|
|
|
if normalize: |
|
401
|
|
|
comma = lname.find(',') |
|
402
|
|
|
if comma != -1: |
|
403
|
|
|
lname_end = lname[comma + 1 :] |
|
404
|
|
|
while lname_end[0] in {' ', ','}: |
|
405
|
|
|
lname_end = lname_end[1:] |
|
406
|
|
|
fname = lname_end + ' ' + fname |
|
407
|
|
|
lname = lname[:comma].strip() |
|
408
|
|
|
|
|
409
|
|
|
# do elder/younger move |
|
410
|
|
|
if normalize == 2 and elderyounger: |
|
411
|
|
|
elderyounger_loc = fname.find(elderyounger) |
|
412
|
|
|
if elderyounger_loc != -1: |
|
413
|
|
|
lname = ' '.join((lname, elderyounger.strip())) |
|
414
|
|
|
fname = ' '.join( |
|
415
|
|
|
( |
|
416
|
1 |
|
fname[:elderyounger_loc].strip(), |
|
417
|
1 |
|
fname[elderyounger_loc + len(elderyounger) :], |
|
418
|
|
|
) |
|
419
|
|
|
).strip() |
|
420
|
|
|
|
|
421
|
|
|
toolcode[4] = '{:02d}'.format(len(fname)) |
|
422
|
1 |
|
toolcode[5] = '{:02d}'.format(len(lname)) |
|
423
|
1 |
|
|
|
424
|
|
|
# strip punctuation |
|
425
|
|
|
for char in ',/:;"&()!{|}?$%*+<=>[\\]^_`~': |
|
426
|
|
|
full_name = full_name.replace(char, '') |
|
427
|
|
|
for pos, char in enumerate(full_name): |
|
428
|
|
|
if char == '-' and full_name[pos - 1 : pos + 2] != 'b-g': |
|
429
|
1 |
|
full_name = full_name[:pos] + ' ' + full_name[pos + 1 :] |
|
430
|
|
|
|
|
431
|
|
|
# Fill field 9 (search range) |
|
432
|
1 |
|
for letter in [_[0] for _ in full_name.split()]: |
|
433
|
1 |
|
if letter not in toolcode[9]: |
|
434
|
1 |
|
toolcode[9] += letter |
|
435
|
1 |
|
if len(toolcode[9]) == 15: |
|
436
|
1 |
|
break |
|
437
|
1 |
|
|
|
438
|
|
|
def roman_check( |
|
439
|
|
|
numeral: str, fname: str, lname: str |
|
440
|
1 |
|
) -> Tuple[str, str]: |
|
441
|
1 |
|
"""Move Roman numerals from first name to last. |
|
442
|
|
|
|
|
443
|
|
|
Parameters |
|
444
|
1 |
|
---------- |
|
445
|
1 |
|
numeral : str |
|
446
|
1 |
|
Roman numeral |
|
447
|
1 |
|
fname : str |
|
448
|
1 |
|
First name |
|
449
|
1 |
|
lname : str |
|
450
|
|
|
Last name |
|
451
|
1 |
|
|
|
452
|
1 |
|
Returns |
|
453
|
1 |
|
------- |
|
454
|
1 |
|
tuple |
|
455
|
1 |
|
First and last names with Roman numeral moved |
|
456
|
1 |
|
|
|
457
|
1 |
|
.. versionadded:: 0.3.0 |
|
458
|
1 |
|
|
|
459
|
1 |
|
""" |
|
460
|
1 |
|
loc = fname.find(numeral) |
|
461
|
|
|
if fname and ( |
|
462
|
|
|
loc != -1 |
|
463
|
1 |
|
and (len(fname[loc:]) == len(numeral)) |
|
464
|
|
|
or fname[loc + len(numeral)] in {' ', ','} |
|
465
|
|
|
): |
|
466
|
|
|
lname = ' '.join((lname, numeral)) |
|
467
|
1 |
|
fname = ' '.join( |
|
468
|
1 |
|
( |
|
469
|
1 |
|
fname[:loc].strip(), |
|
470
|
1 |
|
fname[loc + len(numeral) :].lstrip(' ,'), |
|
471
|
1 |
|
) |
|
472
|
|
|
) |
|
473
|
|
|
return fname.strip(), lname.strip() |
|
474
|
|
|
|
|
475
|
1 |
|
# Fill fields 7 (specials) and 3 (roman numerals) |
|
476
|
|
|
for num, special in enumerate(self._synoname_special_table): |
|
477
|
|
|
roman, match, extra, method = special |
|
478
|
|
|
if method & self._method_dict['end']: |
|
479
|
1 |
|
match_context = ' ' + match |
|
480
|
1 |
|
loc = full_name.find(match_context) |
|
481
|
1 |
|
if (len(full_name) > len(match_context)) and ( |
|
482
|
1 |
|
loc == len(full_name) - len(match_context) |
|
483
|
1 |
|
): |
|
484
|
1 |
|
if roman: |
|
485
|
1 |
|
if not any( |
|
486
|
1 |
|
abbr in fname for abbr in ('i.', 'v.', 'x.') |
|
487
|
1 |
|
): |
|
488
|
1 |
|
full_name = full_name[:loc] |
|
489
|
1 |
|
toolcode[7] += '{:03d}'.format(num) + 'a' |
|
490
|
1 |
|
if toolcode[3] == '000': |
|
491
|
1 |
|
toolcode[3] = '{:03d}'.format(num) |
|
492
|
|
|
if normalize == 2: |
|
493
|
1 |
|
fname, lname = roman_check(match, fname, lname) |
|
494
|
1 |
|
else: |
|
495
|
1 |
|
full_name = full_name[:loc] |
|
496
|
1 |
|
toolcode[7] += '{:03d}'.format(num) + 'a' |
|
497
|
|
|
if method & self._method_dict['middle']: |
|
498
|
|
|
match_context = ' ' + match + ' ' |
|
499
|
|
|
loc = 0 |
|
500
|
|
|
while loc != -1: |
|
501
|
|
|
loc = full_name.find(match_context, loc + 1) |
|
502
|
1 |
|
if loc > 0: |
|
503
|
|
|
if roman: |
|
504
|
1 |
|
if not any( |
|
505
|
|
|
abbr in fname for abbr in ('i.', 'v.', 'x.') |
|
506
|
|
|
): |
|
507
|
1 |
|
full_name = ( |
|
508
|
|
|
full_name[:loc] |
|
509
|
|
|
+ full_name[loc + len(match) + 1 :] |
|
510
|
|
|
) |
|
511
|
|
|
toolcode[7] += '{:03d}'.format(num) + 'b' |
|
512
|
|
|
if toolcode[3] == '000': |
|
513
|
1 |
|
toolcode[3] = '{:03d}'.format(num) |
|
514
|
|
|
if normalize == 2: |
|
515
|
|
|
fname, lname = roman_check( |
|
516
|
|
|
match, fname, lname |
|
517
|
|
|
) |
|
518
|
|
|
else: |
|
519
|
|
|
full_name = ( |
|
520
|
|
|
full_name[:loc] |
|
521
|
|
|
+ full_name[loc + len(match) + 1 :] |
|
522
|
|
|
) |
|
523
|
|
|
toolcode[7] += '{:03d}'.format(num) + 'b' |
|
524
|
|
|
if method & self._method_dict['beginning']: |
|
525
|
|
|
match_context = match + ' ' |
|
526
|
|
|
loc = full_name.find(match_context) |
|
527
|
|
|
if loc == 0: |
|
528
|
|
|
full_name = full_name[len(match) + 1 :] |
|
529
|
|
|
toolcode[7] += '{:03d}'.format(num) + 'c' |
|
530
|
|
|
if method & self._method_dict['beginning_no_space']: |
|
531
|
|
|
loc = full_name.find(match) |
|
532
|
|
|
if loc == 0: |
|
533
|
|
|
toolcode[7] += '{:03d}'.format(num) + 'd' |
|
534
|
|
|
if full_name[: len(match)] not in toolcode[9]: |
|
535
|
|
|
toolcode[9] += full_name[: len(match)] |
|
536
|
|
|
|
|
537
|
|
|
if extra: |
|
538
|
|
|
loc = full_name.find(extra) |
|
539
|
|
|
if loc != -1: |
|
540
|
|
|
toolcode[7] += '{:03d}'.format(num) + 'X' |
|
541
|
|
|
# Since extras are unique, we only look for each of them |
|
542
|
|
|
# once, and they include otherwise impossible characters |
|
543
|
|
|
# for this field, it's not possible for the following line |
|
544
|
|
|
# to have ever been false. |
|
545
|
|
|
# if full_name[loc:loc+len(extra)] not in toolcode[9]: |
|
546
|
|
|
toolcode[9] += full_name[loc : loc + len(match)] |
|
547
|
|
|
|
|
548
|
|
|
return lname, fname, ''.join(toolcode) |
|
549
|
|
|
|
|
550
|
|
|
|
|
551
|
|
|
if __name__ == '__main__': |
|
552
|
|
|
import doctest |
|
553
|
|
|
|
|
554
|
|
|
doctest.testmod() |
|
555
|
|
|
|
chrislit /
abydos
