Hugging Face's logo

Spaces:
HuggingFaceFW-Dev
/
lang-word-tokenizers
Running

App Files Community
guipenedo HF Staff commited on
Commit
6a75090
·
unverified ·
1 Parent(s): d43fb2c

improved viz to include scripts

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. data/Abkhaz-Adyghe.json +53 -27
  2. data/Afro-Asiatic.json +0 -0
  3. data/Algic.json +325 -177
  4. data/Amto-Musan.json +15 -9
  5. data/Andamanese.json +95 -57
  6. data/Arafundi.json +20 -12
  7. data/Arai (Left May).json +37 -21
  8. data/Arauan.json +46 -24
  9. data/Australian.json +0 -0
  10. data/Austro-Asiatic.json +0 -0
  11. data/Austronesian.json +0 -0
  12. data/Aymaran.json +32 -18
  13. data/Barbacoan.json +36 -18
  14. data/Bayono-Awbono.json +15 -9
  15. data/Border.json +99 -57
  16. data/Bororoan.json +22 -12
  17. data/Botocudoan.json +10 -6
  18. data/Caddoan.json +45 -27
  19. data/Cahuapanan.json +17 -9
  20. data/Cariban.json +283 -159
  21. data/Central Solomons.json +25 -15
  22. data/Chapacuran.json +35 -21
  23. data/Chibchan.json +198 -108
  24. data/Chimakuan.json +15 -9
  25. data/Chinookan.json +20 -12
  26. data/Chipaya-Uru.json +17 -9
  27. data/Chocoan.json +63 -33
  28. data/Cholonan.json +15 -9
  29. data/Chon.json +20 -12
  30. data/Chukotko-Kamchatkan.json +54 -30
  31. data/Chumashan.json +45 -27
  32. data/Cochimí-Yuman.json +75 -45
  33. data/Comecrudan.json +30 -18
  34. data/Constructed language.json +12 -6
  35. data/Coosan.json +15 -9
  36. data/Creole.json +0 -0
  37. data/Dravidian.json +0 -0
  38. data/East Bird’s Head-Sentani.json +86 -48
  39. data/East Geelvink Bay.json +72 -42
  40. data/East New Britain.json +52 -30
  41. data/Eastern Trans-Fly.json +29 -15
  42. data/Eskimo-Aleut.json +99 -51
  43. data/Eyak-Athabaskan.json +324 -186
  44. data/Fas.json +15 -9
  45. data/Guajiboan.json +41 -21
  46. data/Guaykuruan.json +48 -24
  47. data/Gum.json +5 -3
  48. data/Haida.json +15 -9
  49. data/Harákmbut.json +17 -9
  50. data/Hmong-Mien.json +261 -153
data/Abkhaz-Adyghe.json CHANGED
@@ -2,75 +2,101 @@
2
  "name": "Abkhaz-Adyghe",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Abkhaz-Abazin",
10
  "iso_1_code": null,
11
  "iso_3_code": null,
12
- "tokenizer": null,
13
- "source": null,
14
  "children": [
15
  {
16
  "name": "Abkhaz",
17
  "iso_1_code": "ab",
18
  "iso_3_code": "abk",
19
- "tokenizer": null,
20
- "source": null,
21
- "children": []
 
 
 
 
22
  },
23
  {
24
  "name": "Abaza",
25
  "iso_1_code": null,
26
  "iso_3_code": "abq",
27
- "tokenizer": null,
28
- "source": null,
29
- "children": []
 
 
 
 
30
  }
31
- ]
 
 
 
32
  },
33
  {
34
  "name": "Circassian",
35
  "iso_1_code": null,
36
  "iso_3_code": null,
37
- "tokenizer": null,
38
- "source": null,
39
  "children": [
40
  {
41
  "name": "Adyghe",
42
  "iso_1_code": null,
43
  "iso_3_code": "ady",
44
- "tokenizer": null,
45
- "source": null,
46
- "children": []
 
 
 
 
47
  },
48
  {
49
  "name": "Kabardian",
50
  "iso_1_code": null,
51
  "iso_3_code": "kbd",
52
- "tokenizer": null,
53
- "source": null,
54
- "children": []
 
 
 
 
55
  }
56
- ]
 
 
 
57
  },
58
  {
59
  "name": "Ubyx",
60
  "iso_1_code": null,
61
  "iso_3_code": null,
62
- "tokenizer": null,
63
- "source": null,
64
  "children": [
65
  {
66
  "name": "Ubykh",
67
  "iso_1_code": null,
68
  "iso_3_code": "uby",
69
- "tokenizer": null,
70
- "source": null,
71
- "children": []
 
 
72
  }
73
- ]
 
 
 
74
  }
75
- ]
 
 
 
76
  }
 
2
  "name": "Abkhaz-Adyghe",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Abkhaz-Abazin",
9
  "iso_1_code": null,
10
  "iso_3_code": null,
11
+ "tokenizers": {},
 
12
  "children": [
13
  {
14
  "name": "Abkhaz",
15
  "iso_1_code": "ab",
16
  "iso_3_code": "abk",
17
+ "tokenizers": {},
18
+ "children": [],
19
+ "node_i": "2",
20
+ "scripts": [
21
+ "Cyrl"
22
+ ],
23
+ "own_tokenizer": false
24
  },
25
  {
26
  "name": "Abaza",
27
  "iso_1_code": null,
28
  "iso_3_code": "abq",
29
+ "tokenizers": {},
30
+ "children": [],
31
+ "node_i": "3",
32
+ "scripts": [
33
+ "Cyrl"
34
+ ],
35
+ "own_tokenizer": false
36
  }
37
+ ],
38
+ "node_i": "1",
39
+ "scripts": [],
40
+ "own_tokenizer": false
41
  },
42
  {
43
  "name": "Circassian",
44
  "iso_1_code": null,
45
  "iso_3_code": null,
46
+ "tokenizers": {},
 
47
  "children": [
48
  {
49
  "name": "Adyghe",
50
  "iso_1_code": null,
51
  "iso_3_code": "ady",
52
+ "tokenizers": {},
53
+ "children": [],
54
+ "node_i": "5",
55
+ "scripts": [
56
+ "Cyrl"
57
+ ],
58
+ "own_tokenizer": false
59
  },
60
  {
61
  "name": "Kabardian",
62
  "iso_1_code": null,
63
  "iso_3_code": "kbd",
64
+ "tokenizers": {},
65
+ "children": [],
66
+ "node_i": "6",
67
+ "scripts": [
68
+ "Cyrl"
69
+ ],
70
+ "own_tokenizer": false
71
  }
72
+ ],
73
+ "node_i": "4",
74
+ "scripts": [],
75
+ "own_tokenizer": false
76
  },
77
  {
78
  "name": "Ubyx",
79
  "iso_1_code": null,
80
  "iso_3_code": null,
81
+ "tokenizers": {},
 
82
  "children": [
83
  {
84
  "name": "Ubykh",
85
  "iso_1_code": null,
86
  "iso_3_code": "uby",
87
+ "tokenizers": {},
88
+ "children": [],
89
+ "node_i": "8",
90
+ "scripts": [],
91
+ "own_tokenizer": false
92
  }
93
+ ],
94
+ "node_i": "7",
95
+ "scripts": [],
96
+ "own_tokenizer": false
97
  }
98
+ ],
99
+ "node_i": "0",
100
+ "scripts": [],
101
+ "own_tokenizer": false
102
  }
data/Afro-Asiatic.json CHANGED
The diff for this file is too large to render. See raw diff
 
data/Algic.json CHANGED
@@ -2,482 +2,630 @@
2
  "name": "Algic",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Algonquian",
10
  "iso_1_code": null,
11
  "iso_3_code": null,
12
- "tokenizer": null,
13
- "source": null,
14
  "children": [
15
  {
16
  "name": "Blackfoot",
17
  "iso_1_code": null,
18
  "iso_3_code": "bla",
19
- "tokenizer": null,
20
- "source": null,
21
- "children": []
 
 
 
 
22
  },
23
  {
24
  "name": "Cheyenne",
25
  "iso_1_code": null,
26
  "iso_3_code": "chy",
27
- "tokenizer": null,
28
- "source": null,
29
- "children": []
 
 
30
  },
31
  {
32
  "name": "Menominee",
33
  "iso_1_code": null,
34
  "iso_3_code": "mez",
35
- "tokenizer": null,
36
- "source": null,
37
- "children": []
 
 
38
  },
39
  {
40
  "name": "Miami",
41
  "iso_1_code": null,
42
  "iso_3_code": "mia",
43
- "tokenizer": null,
44
- "source": null,
45
- "children": []
 
 
46
  },
47
  {
48
  "name": "Nawathinehena",
49
  "iso_1_code": null,
50
  "iso_3_code": "nwa",
51
- "tokenizer": null,
52
- "source": null,
53
- "children": []
 
 
54
  },
55
  {
56
  "name": "Shawnee",
57
  "iso_1_code": null,
58
  "iso_3_code": "sjw",
59
- "tokenizer": null,
60
- "source": null,
61
- "children": []
 
 
62
  },
63
  {
64
  "name": "Arapaho",
65
  "iso_1_code": null,
66
  "iso_3_code": null,
67
- "tokenizer": null,
68
- "source": null,
69
  "children": [
70
  {
71
  "name": "Arapaho",
72
  "iso_1_code": null,
73
  "iso_3_code": "arp",
74
- "tokenizer": null,
75
- "source": null,
76
- "children": []
 
 
 
 
77
  },
78
  {
79
  "name": "Gros Ventre",
80
  "iso_1_code": null,
81
  "iso_3_code": "ats",
82
- "tokenizer": null,
83
- "source": null,
84
- "children": []
 
 
85
  }
86
- ]
 
 
 
87
  },
88
  {
89
  "name": "Cree-Montagnais",
90
  "iso_1_code": null,
91
  "iso_3_code": null,
92
- "tokenizer": null,
93
- "source": null,
94
  "children": [
95
  {
96
  "name": "Atikamekw",
97
  "iso_1_code": null,
98
  "iso_3_code": "atj",
99
- "tokenizer": null,
100
- "source": null,
101
- "children": []
 
 
 
 
102
  },
103
  {
104
  "name": "Cree, Southern East",
105
  "iso_1_code": "cr",
106
  "iso_3_code": "crj",
107
- "tokenizer": null,
108
- "source": "macrolanguage",
109
- "children": []
 
 
 
 
110
  },
111
  {
112
  "name": "Cree, Plains",
113
  "iso_1_code": "cr",
114
  "iso_3_code": "crk",
115
- "tokenizer": null,
116
- "source": "macrolanguage",
117
- "children": []
 
 
 
 
 
118
  },
119
  {
120
  "name": "Cree, Northern East",
121
  "iso_1_code": "cr",
122
  "iso_3_code": "crl",
123
- "tokenizer": null,
124
- "source": "macrolanguage",
125
- "children": []
 
 
 
 
126
  },
127
  {
128
  "name": "Cree, Moose",
129
  "iso_1_code": "cr",
130
  "iso_3_code": "crm",
131
- "tokenizer": null,
132
- "source": "macrolanguage",
133
- "children": []
 
 
 
 
134
  },
135
  {
136
  "name": "Cree, Swampy",
137
  "iso_1_code": "cr",
138
  "iso_3_code": "csw",
139
- "tokenizer": null,
140
- "source": "macrolanguage",
141
- "children": []
 
 
 
 
142
  },
143
  {
144
  "name": "Cree, Woods",
145
  "iso_1_code": "cr",
146
  "iso_3_code": "cwd",
147
- "tokenizer": null,
148
- "source": "macrolanguage",
149
- "children": []
 
 
 
 
150
  },
151
  {
152
  "name": "Innu",
153
  "iso_1_code": null,
154
  "iso_3_code": "moe",
155
- "tokenizer": null,
156
- "source": null,
157
- "children": []
 
 
158
  },
159
  {
160
  "name": "Naskapi",
161
  "iso_1_code": null,
162
  "iso_3_code": "nsk",
163
- "tokenizer": null,
164
- "source": null,
165
- "children": []
 
 
166
  }
167
- ]
 
 
 
168
  },
169
  {
170
  "name": "Eastern Algonquian",
171
  "iso_1_code": null,
172
  "iso_3_code": null,
173
- "tokenizer": null,
174
- "source": null,
175
  "children": [
176
  {
177
  "name": "Carolina Algonquian",
178
  "iso_1_code": null,
179
  "iso_3_code": "crr",
180
- "tokenizer": null,
181
- "source": null,
182
- "children": []
 
 
183
  },
184
  {
185
  "name": "Etchemin",
186
  "iso_1_code": null,
187
  "iso_3_code": "etc",
188
- "tokenizer": null,
189
- "source": null,
190
- "children": []
 
 
191
  },
192
  {
193
  "name": "Mi\u2019kmaq",
194
  "iso_1_code": null,
195
  "iso_3_code": "mic",
196
- "tokenizer": null,
197
- "source": null,
198
- "children": []
 
 
 
 
199
  },
200
  {
201
  "name": "Powhatan",
202
  "iso_1_code": null,
203
  "iso_3_code": "pim",
204
- "tokenizer": null,
205
- "source": null,
206
- "children": []
 
 
207
  },
208
  {
209
  "name": "Malecite-Passamaquoddy",
210
  "iso_1_code": null,
211
  "iso_3_code": "pqm",
212
- "tokenizer": null,
213
- "source": null,
214
- "children": []
 
 
215
  },
216
  {
217
  "name": "Quiripi",
218
  "iso_1_code": null,
219
  "iso_3_code": "qyp",
220
- "tokenizer": null,
221
- "source": null,
222
- "children": []
 
 
223
  },
224
  {
225
  "name": "Wampanoag",
226
  "iso_1_code": null,
227
  "iso_3_code": "wam",
228
- "tokenizer": null,
229
- "source": null,
230
- "children": []
 
 
231
  },
232
  {
233
  "name": "Loup B",
234
  "iso_1_code": null,
235
  "iso_3_code": "xlb",
236
- "tokenizer": null,
237
- "source": null,
238
- "children": []
 
 
239
  },
240
  {
241
  "name": "Loup A",
242
  "iso_1_code": null,
243
  "iso_3_code": "xlo",
244
- "tokenizer": null,
245
- "source": null,
246
- "children": []
 
 
247
  },
248
  {
249
  "name": "Narragansett",
250
  "iso_1_code": null,
251
  "iso_3_code": "xnt",
252
- "tokenizer": null,
253
- "source": null,
254
- "children": []
 
 
255
  },
256
  {
257
  "name": "Mohegan-Pequot",
258
  "iso_1_code": null,
259
  "iso_3_code": "xpq",
260
- "tokenizer": null,
261
- "source": null,
262
- "children": []
 
 
263
  },
264
  {
265
  "name": "Abenaki",
266
  "iso_1_code": null,
267
  "iso_3_code": null,
268
- "tokenizer": null,
269
- "source": null,
270
  "children": [
271
  {
272
  "name": "Abenaki, Eastern",
273
  "iso_1_code": null,
274
  "iso_3_code": "aaq",
275
- "tokenizer": null,
276
- "source": null,
277
- "children": []
 
 
278
  },
279
  {
280
  "name": "Abenaki, Western",
281
  "iso_1_code": null,
282
  "iso_3_code": "abe",
283
- "tokenizer": null,
284
- "source": null,
285
- "children": []
 
 
286
  }
287
- ]
 
 
 
288
  },
289
  {
290
  "name": "Delaware",
291
  "iso_1_code": null,
292
  "iso_3_code": null,
293
- "tokenizer": null,
294
- "source": null,
295
  "children": [
296
  {
297
  "name": "Mahican",
298
  "iso_1_code": null,
299
  "iso_3_code": "mjy",
300
- "tokenizer": null,
301
- "source": null,
302
- "children": []
 
 
303
  },
304
  {
305
  "name": "Munsee",
306
  "iso_1_code": null,
307
  "iso_3_code": "umu",
308
- "tokenizer": null,
309
- "source": null,
310
- "children": []
 
 
311
  },
312
  {
313
  "name": "Unami",
314
  "iso_1_code": null,
315
  "iso_3_code": "unm",
316
- "tokenizer": null,
317
- "source": null,
318
- "children": []
 
 
319
  }
320
- ]
 
 
 
321
  },
322
  {
323
  "name": "Nanticoke-Conoy",
324
  "iso_1_code": null,
325
  "iso_3_code": null,
326
- "tokenizer": null,
327
- "source": null,
328
  "children": [
329
  {
330
  "name": "Nanticoke",
331
  "iso_1_code": null,
332
  "iso_3_code": "nnt",
333
- "tokenizer": null,
334
- "source": null,
335
- "children": []
 
 
336
  },
337
  {
338
  "name": "Piscataway",
339
  "iso_1_code": null,
340
  "iso_3_code": "psy",
341
- "tokenizer": null,
342
- "source": null,
343
- "children": []
 
 
344
  }
345
- ]
 
 
 
346
  }
347
- ]
 
 
 
348
  },
349
  {
350
  "name": "Fox",
351
  "iso_1_code": null,
352
  "iso_3_code": null,
353
- "tokenizer": null,
354
- "source": null,
355
  "children": [
356
  {
357
  "name": "Kickapoo",
358
  "iso_1_code": null,
359
  "iso_3_code": "kic",
360
- "tokenizer": null,
361
- "source": null,
362
- "children": []
 
 
363
  },
364
  {
365
  "name": "Meskwaki",
366
  "iso_1_code": null,
367
  "iso_3_code": "sac",
368
- "tokenizer": null,
369
- "source": null,
370
- "children": []
 
 
371
  }
372
- ]
 
 
 
373
  },
374
  {
375
  "name": "Ojibwa-Potawatomi",
376
  "iso_1_code": null,
377
  "iso_3_code": null,
378
- "tokenizer": null,
379
- "source": null,
380
  "children": [
381
  {
382
  "name": "Algonquin",
383
  "iso_1_code": null,
384
  "iso_3_code": "alq",
385
- "tokenizer": null,
386
- "source": null,
387
- "children": []
 
 
 
 
388
  },
389
  {
390
  "name": "Chippewa",
391
  "iso_1_code": "oj",
392
  "iso_3_code": "ciw",
393
- "tokenizer": null,
394
- "source": "macrolanguage",
395
- "children": []
 
 
396
  },
397
  {
398
  "name": "Ojibwa, Northwestern",
399
  "iso_1_code": "oj",
400
  "iso_3_code": "ojb",
401
- "tokenizer": null,
402
- "source": "macrolanguage",
403
- "children": []
 
 
 
 
 
404
  },
405
  {
406
  "name": "Ojibwa, Central",
407
  "iso_1_code": "oj",
408
  "iso_3_code": "ojc",
409
- "tokenizer": null,
410
- "source": "macrolanguage",
411
- "children": []
 
 
412
  },
413
  {
414
  "name": "Ojibwa, Eastern",
415
  "iso_1_code": "oj",
416
  "iso_3_code": "ojg",
417
- "tokenizer": null,
418
- "source": "macrolanguage",
419
- "children": []
 
 
420
  },
421
  {
422
  "name": "Oji-Cree",
423
  "iso_1_code": "oj",
424
  "iso_3_code": "ojs",
425
- "tokenizer": null,
426
- "source": "macrolanguage",
427
- "children": []
 
 
428
  },
429
  {
430
  "name": "Ojibwa, Western",
431
  "iso_1_code": "oj",
432
  "iso_3_code": "ojw",
433
- "tokenizer": null,
434
- "source": "macrolanguage",
435
- "children": []
 
 
436
  },
437
  {
438
  "name": "Ottawa",
439
  "iso_1_code": "oj",
440
  "iso_3_code": "otw",
441
- "tokenizer": null,
442
- "source": "macrolanguage",
443
- "children": []
 
 
 
 
444
  },
445
  {
446
  "name": "Potawatomi",
447
  "iso_1_code": null,
448
  "iso_3_code": "pot",
449
- "tokenizer": null,
450
- "source": null,
451
- "children": []
 
 
 
 
452
  }
453
- ]
 
 
 
454
  }
455
- ]
 
 
 
456
  },
457
  {
458
  "name": "Ritwan",
459
  "iso_1_code": null,
460
  "iso_3_code": null,
461
- "tokenizer": null,
462
- "source": null,
463
  "children": [
464
  {
465
  "name": "Wiyot",
466
  "iso_1_code": null,
467
  "iso_3_code": "wiy",
468
- "tokenizer": null,
469
- "source": null,
470
- "children": []
 
 
471
  },
472
  {
473
  "name": "Yurok",
474
  "iso_1_code": null,
475
  "iso_3_code": "yur",
476
- "tokenizer": null,
477
- "source": null,
478
- "children": []
 
 
479
  }
480
- ]
 
 
 
481
  }
482
- ]
 
 
 
483
  }
 
2
  "name": "Algic",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Algonquian",
9
  "iso_1_code": null,
10
  "iso_3_code": null,
11
+ "tokenizers": {},
 
12
  "children": [
13
  {
14
  "name": "Blackfoot",
15
  "iso_1_code": null,
16
  "iso_3_code": "bla",
17
+ "tokenizers": {},
18
+ "children": [],
19
+ "node_i": "558",
20
+ "scripts": [
21
+ "Latn"
22
+ ],
23
+ "own_tokenizer": false
24
  },
25
  {
26
  "name": "Cheyenne",
27
  "iso_1_code": null,
28
  "iso_3_code": "chy",
29
+ "tokenizers": {},
30
+ "children": [],
31
+ "node_i": "559",
32
+ "scripts": [],
33
+ "own_tokenizer": false
34
  },
35
  {
36
  "name": "Menominee",
37
  "iso_1_code": null,
38
  "iso_3_code": "mez",
39
+ "tokenizers": {},
40
+ "children": [],
41
+ "node_i": "560",
42
+ "scripts": [],
43
+ "own_tokenizer": false
44
  },
45
  {
46
  "name": "Miami",
47
  "iso_1_code": null,
48
  "iso_3_code": "mia",
49
+ "tokenizers": {},
50
+ "children": [],
51
+ "node_i": "561",
52
+ "scripts": [],
53
+ "own_tokenizer": false
54
  },
55
  {
56
  "name": "Nawathinehena",
57
  "iso_1_code": null,
58
  "iso_3_code": "nwa",
59
+ "tokenizers": {},
60
+ "children": [],
61
+ "node_i": "562",
62
+ "scripts": [],
63
+ "own_tokenizer": false
64
  },
65
  {
66
  "name": "Shawnee",
67
  "iso_1_code": null,
68
  "iso_3_code": "sjw",
69
+ "tokenizers": {},
70
+ "children": [],
71
+ "node_i": "563",
72
+ "scripts": [],
73
+ "own_tokenizer": false
74
  },
75
  {
76
  "name": "Arapaho",
77
  "iso_1_code": null,
78
  "iso_3_code": null,
79
+ "tokenizers": {},
 
80
  "children": [
81
  {
82
  "name": "Arapaho",
83
  "iso_1_code": null,
84
  "iso_3_code": "arp",
85
+ "tokenizers": {},
86
+ "children": [],
87
+ "node_i": "565",
88
+ "scripts": [
89
+ "Latn"
90
+ ],
91
+ "own_tokenizer": false
92
  },
93
  {
94
  "name": "Gros Ventre",
95
  "iso_1_code": null,
96
  "iso_3_code": "ats",
97
+ "tokenizers": {},
98
+ "children": [],
99
+ "node_i": "566",
100
+ "scripts": [],
101
+ "own_tokenizer": false
102
  }
103
+ ],
104
+ "node_i": "564",
105
+ "scripts": [],
106
+ "own_tokenizer": false
107
  },
108
  {
109
  "name": "Cree-Montagnais",
110
  "iso_1_code": null,
111
  "iso_3_code": null,
112
+ "tokenizers": {},
 
113
  "children": [
114
  {
115
  "name": "Atikamekw",
116
  "iso_1_code": null,
117
  "iso_3_code": "atj",
118
+ "tokenizers": {},
119
+ "children": [],
120
+ "node_i": "568",
121
+ "scripts": [
122
+ "Latn"
123
+ ],
124
+ "own_tokenizer": false
125
  },
126
  {
127
  "name": "Cree, Southern East",
128
  "iso_1_code": "cr",
129
  "iso_3_code": "crj",
130
+ "tokenizers": {},
131
+ "children": [],
132
+ "node_i": "569",
133
+ "scripts": [
134
+ "Cans"
135
+ ],
136
+ "own_tokenizer": false
137
  },
138
  {
139
  "name": "Cree, Plains",
140
  "iso_1_code": "cr",
141
  "iso_3_code": "crk",
142
+ "tokenizers": {},
143
+ "children": [],
144
+ "node_i": "570",
145
+ "scripts": [
146
+ "Latn",
147
+ "Cans"
148
+ ],
149
+ "own_tokenizer": false
150
  },
151
  {
152
  "name": "Cree, Northern East",
153
  "iso_1_code": "cr",
154
  "iso_3_code": "crl",
155
+ "tokenizers": {},
156
+ "children": [],
157
+ "node_i": "571",
158
+ "scripts": [
159
+ "Cans"
160
+ ],
161
+ "own_tokenizer": false
162
  },
163
  {
164
  "name": "Cree, Moose",
165
  "iso_1_code": "cr",
166
  "iso_3_code": "crm",
167
+ "tokenizers": {},
168
+ "children": [],
169
+ "node_i": "572",
170
+ "scripts": [
171
+ "Cans"
172
+ ],
173
+ "own_tokenizer": false
174
  },
175
  {
176
  "name": "Cree, Swampy",
177
  "iso_1_code": "cr",
178
  "iso_3_code": "csw",
179
+ "tokenizers": {},
180
+ "children": [],
181
+ "node_i": "573",
182
+ "scripts": [
183
+ "Latn"
184
+ ],
185
+ "own_tokenizer": false
186
  },
187
  {
188
  "name": "Cree, Woods",
189
  "iso_1_code": "cr",
190
  "iso_3_code": "cwd",
191
+ "tokenizers": {},
192
+ "children": [],
193
+ "node_i": "574",
194
+ "scripts": [
195
+ "Cans"
196
+ ],
197
+ "own_tokenizer": false
198
  },
199
  {
200
  "name": "Innu",
201
  "iso_1_code": null,
202
  "iso_3_code": "moe",
203
+ "tokenizers": {},
204
+ "children": [],
205
+ "node_i": "575",
206
+ "scripts": [],
207
+ "own_tokenizer": false
208
  },
209
  {
210
  "name": "Naskapi",
211
  "iso_1_code": null,
212
  "iso_3_code": "nsk",
213
+ "tokenizers": {},
214
+ "children": [],
215
+ "node_i": "576",
216
+ "scripts": [],
217
+ "own_tokenizer": false
218
  }
219
+ ],
220
+ "node_i": "567",
221
+ "scripts": [],
222
+ "own_tokenizer": false
223
  },
224
  {
225
  "name": "Eastern Algonquian",
226
  "iso_1_code": null,
227
  "iso_3_code": null,
228
+ "tokenizers": {},
 
229
  "children": [
230
  {
231
  "name": "Carolina Algonquian",
232
  "iso_1_code": null,
233
  "iso_3_code": "crr",
234
+ "tokenizers": {},
235
+ "children": [],
236
+ "node_i": "578",
237
+ "scripts": [],
238
+ "own_tokenizer": false
239
  },
240
  {
241
  "name": "Etchemin",
242
  "iso_1_code": null,
243
  "iso_3_code": "etc",
244
+ "tokenizers": {},
245
+ "children": [],
246
+ "node_i": "579",
247
+ "scripts": [],
248
+ "own_tokenizer": false
249
  },
250
  {
251
  "name": "Mi\u2019kmaq",
252
  "iso_1_code": null,
253
  "iso_3_code": "mic",
254
+ "tokenizers": {},
255
+ "children": [],
256
+ "node_i": "580",
257
+ "scripts": [
258
+ "Latn"
259
+ ],
260
+ "own_tokenizer": false
261
  },
262
  {
263
  "name": "Powhatan",
264
  "iso_1_code": null,
265
  "iso_3_code": "pim",
266
+ "tokenizers": {},
267
+ "children": [],
268
+ "node_i": "581",
269
+ "scripts": [],
270
+ "own_tokenizer": false
271
  },
272
  {
273
  "name": "Malecite-Passamaquoddy",
274
  "iso_1_code": null,
275
  "iso_3_code": "pqm",
276
+ "tokenizers": {},
277
+ "children": [],
278
+ "node_i": "582",
279
+ "scripts": [],
280
+ "own_tokenizer": false
281
  },
282
  {
283
  "name": "Quiripi",
284
  "iso_1_code": null,
285
  "iso_3_code": "qyp",
286
+ "tokenizers": {},
287
+ "children": [],
288
+ "node_i": "583",
289
+ "scripts": [],
290
+ "own_tokenizer": false
291
  },
292
  {
293
  "name": "Wampanoag",
294
  "iso_1_code": null,
295
  "iso_3_code": "wam",
296
+ "tokenizers": {},
297
+ "children": [],
298
+ "node_i": "584",
299
+ "scripts": [],
300
+ "own_tokenizer": false
301
  },
302
  {
303
  "name": "Loup B",
304
  "iso_1_code": null,
305
  "iso_3_code": "xlb",
306
+ "tokenizers": {},
307
+ "children": [],
308
+ "node_i": "585",
309
+ "scripts": [],
310
+ "own_tokenizer": false
311
  },
312
  {
313
  "name": "Loup A",
314
  "iso_1_code": null,
315
  "iso_3_code": "xlo",
316
+ "tokenizers": {},
317
+ "children": [],
318
+ "node_i": "586",
319
+ "scripts": [],
320
+ "own_tokenizer": false
321
  },
322
  {
323
  "name": "Narragansett",
324
  "iso_1_code": null,
325
  "iso_3_code": "xnt",
326
+ "tokenizers": {},
327
+ "children": [],
328
+ "node_i": "587",
329
+ "scripts": [],
330
+ "own_tokenizer": false
331
  },
332
  {
333
  "name": "Mohegan-Pequot",
334
  "iso_1_code": null,
335
  "iso_3_code": "xpq",
336
+ "tokenizers": {},
337
+ "children": [],
338
+ "node_i": "588",
339
+ "scripts": [],
340
+ "own_tokenizer": false
341
  },
342
  {
343
  "name": "Abenaki",
344
  "iso_1_code": null,
345
  "iso_3_code": null,
346
+ "tokenizers": {},
 
347
  "children": [
348
  {
349
  "name": "Abenaki, Eastern",
350
  "iso_1_code": null,
351
  "iso_3_code": "aaq",
352
+ "tokenizers": {},
353
+ "children": [],
354
+ "node_i": "590",
355
+ "scripts": [],
356
+ "own_tokenizer": false
357
  },
358
  {
359
  "name": "Abenaki, Western",
360
  "iso_1_code": null,
361
  "iso_3_code": "abe",
362
+ "tokenizers": {},
363
+ "children": [],
364
+ "node_i": "591",
365
+ "scripts": [],
366
+ "own_tokenizer": false
367
  }
368
+ ],
369
+ "node_i": "589",
370
+ "scripts": [],
371
+ "own_tokenizer": false
372
  },
373
  {
374
  "name": "Delaware",
375
  "iso_1_code": null,
376
  "iso_3_code": null,
377
+ "tokenizers": {},
 
378
  "children": [
379
  {
380
  "name": "Mahican",
381
  "iso_1_code": null,
382
  "iso_3_code": "mjy",
383
+ "tokenizers": {},
384
+ "children": [],
385
+ "node_i": "593",
386
+ "scripts": [],
387
+ "own_tokenizer": false
388
  },
389
  {
390
  "name": "Munsee",
391
  "iso_1_code": null,
392
  "iso_3_code": "umu",
393
+ "tokenizers": {},
394
+ "children": [],
395
+ "node_i": "594",
396
+ "scripts": [],
397
+ "own_tokenizer": false
398
  },
399
  {
400
  "name": "Unami",
401
  "iso_1_code": null,
402
  "iso_3_code": "unm",
403
+ "tokenizers": {},
404
+ "children": [],
405
+ "node_i": "595",
406
+ "scripts": [],
407
+ "own_tokenizer": false
408
  }
409
+ ],
410
+ "node_i": "592",
411
+ "scripts": [],
412
+ "own_tokenizer": false
413
  },
414
  {
415
  "name": "Nanticoke-Conoy",
416
  "iso_1_code": null,
417
  "iso_3_code": null,
418
+ "tokenizers": {},
 
419
  "children": [
420
  {
421
  "name": "Nanticoke",
422
  "iso_1_code": null,
423
  "iso_3_code": "nnt",
424
+ "tokenizers": {},
425
+ "children": [],
426
+ "node_i": "597",
427
+ "scripts": [],
428
+ "own_tokenizer": false
429
  },
430
  {
431
  "name": "Piscataway",
432
  "iso_1_code": null,
433
  "iso_3_code": "psy",
434
+ "tokenizers": {},
435
+ "children": [],
436
+ "node_i": "598",
437
+ "scripts": [],
438
+ "own_tokenizer": false
439
  }
440
+ ],
441
+ "node_i": "596",
442
+ "scripts": [],
443
+ "own_tokenizer": false
444
  }
445
+ ],
446
+ "node_i": "577",
447
+ "scripts": [],
448
+ "own_tokenizer": false
449
  },
450
  {
451
  "name": "Fox",
452
  "iso_1_code": null,
453
  "iso_3_code": null,
454
+ "tokenizers": {},
 
455
  "children": [
456
  {
457
  "name": "Kickapoo",
458
  "iso_1_code": null,
459
  "iso_3_code": "kic",
460
+ "tokenizers": {},
461
+ "children": [],
462
+ "node_i": "600",
463
+ "scripts": [],
464
+ "own_tokenizer": false
465
  },
466
  {
467
  "name": "Meskwaki",
468
  "iso_1_code": null,
469
  "iso_3_code": "sac",
470
+ "tokenizers": {},
471
+ "children": [],
472
+ "node_i": "601",
473
+ "scripts": [],
474
+ "own_tokenizer": false
475
  }
476
+ ],
477
+ "node_i": "599",
478
+ "scripts": [],
479
+ "own_tokenizer": false
480
  },
481
  {
482
  "name": "Ojibwa-Potawatomi",
483
  "iso_1_code": null,
484
  "iso_3_code": null,
485
+ "tokenizers": {},
 
486
  "children": [
487
  {
488
  "name": "Algonquin",
489
  "iso_1_code": null,
490
  "iso_3_code": "alq",
491
+ "tokenizers": {},
492
+ "children": [],
493
+ "node_i": "603",
494
+ "scripts": [
495
+ "Latn"
496
+ ],
497
+ "own_tokenizer": false
498
  },
499
  {
500
  "name": "Chippewa",
501
  "iso_1_code": "oj",
502
  "iso_3_code": "ciw",
503
+ "tokenizers": {},
504
+ "children": [],
505
+ "node_i": "604",
506
+ "scripts": [],
507
+ "own_tokenizer": false
508
  },
509
  {
510
  "name": "Ojibwa, Northwestern",
511
  "iso_1_code": "oj",
512
  "iso_3_code": "ojb",
513
+ "tokenizers": {},
514
+ "children": [],
515
+ "node_i": "605",
516
+ "scripts": [
517
+ "Latn",
518
+ "Cans"
519
+ ],
520
+ "own_tokenizer": false
521
  },
522
  {
523
  "name": "Ojibwa, Central",
524
  "iso_1_code": "oj",
525
  "iso_3_code": "ojc",
526
+ "tokenizers": {},
527
+ "children": [],
528
+ "node_i": "606",
529
+ "scripts": [],
530
+ "own_tokenizer": false
531
  },
532
  {
533
  "name": "Ojibwa, Eastern",
534
  "iso_1_code": "oj",
535
  "iso_3_code": "ojg",
536
+ "tokenizers": {},
537
+ "children": [],
538
+ "node_i": "607",
539
+ "scripts": [],
540
+ "own_tokenizer": false
541
  },
542
  {
543
  "name": "Oji-Cree",
544
  "iso_1_code": "oj",
545
  "iso_3_code": "ojs",
546
+ "tokenizers": {},
547
+ "children": [],
548
+ "node_i": "608",
549
+ "scripts": [],
550
+ "own_tokenizer": false
551
  },
552
  {
553
  "name": "Ojibwa, Western",
554
  "iso_1_code": "oj",
555
  "iso_3_code": "ojw",
556
+ "tokenizers": {},
557
+ "children": [],
558
+ "node_i": "609",
559
+ "scripts": [],
560
+ "own_tokenizer": false
561
  },
562
  {
563
  "name": "Ottawa",
564
  "iso_1_code": "oj",
565
  "iso_3_code": "otw",
566
+ "tokenizers": {},
567
+ "children": [],
568
+ "node_i": "610",
569
+ "scripts": [
570
+ "Latn"
571
+ ],
572
+ "own_tokenizer": false
573
  },
574
  {
575
  "name": "Potawatomi",
576
  "iso_1_code": null,
577
  "iso_3_code": "pot",
578
+ "tokenizers": {},
579
+ "children": [],
580
+ "node_i": "611",
581
+ "scripts": [
582
+ "Latn"
583
+ ],
584
+ "own_tokenizer": false
585
  }
586
+ ],
587
+ "node_i": "602",
588
+ "scripts": [],
589
+ "own_tokenizer": false
590
  }
591
+ ],
592
+ "node_i": "557",
593
+ "scripts": [],
594
+ "own_tokenizer": false
595
  },
596
  {
597
  "name": "Ritwan",
598
  "iso_1_code": null,
599
  "iso_3_code": null,
600
+ "tokenizers": {},
 
601
  "children": [
602
  {
603
  "name": "Wiyot",
604
  "iso_1_code": null,
605
  "iso_3_code": "wiy",
606
+ "tokenizers": {},
607
+ "children": [],
608
+ "node_i": "613",
609
+ "scripts": [],
610
+ "own_tokenizer": false
611
  },
612
  {
613
  "name": "Yurok",
614
  "iso_1_code": null,
615
  "iso_3_code": "yur",
616
+ "tokenizers": {},
617
+ "children": [],
618
+ "node_i": "614",
619
+ "scripts": [],
620
+ "own_tokenizer": false
621
  }
622
+ ],
623
+ "node_i": "612",
624
+ "scripts": [],
625
+ "own_tokenizer": false
626
  }
627
+ ],
628
+ "node_i": "556",
629
+ "scripts": [],
630
+ "own_tokenizer": false
631
  }
data/Amto-Musan.json CHANGED
@@ -2,24 +2,30 @@
2
  "name": "Amto-Musan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Amto",
10
  "iso_1_code": null,
11
  "iso_3_code": "amt",
12
- "tokenizer": null,
13
- "source": null,
14
- "children": []
 
 
15
  },
16
  {
17
  "name": "Siawi",
18
  "iso_1_code": null,
19
  "iso_3_code": "mmp",
20
- "tokenizer": null,
21
- "source": null,
22
- "children": []
 
 
23
  }
24
- ]
 
 
 
25
  }
 
2
  "name": "Amto-Musan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Amto",
9
  "iso_1_code": null,
10
  "iso_3_code": "amt",
11
+ "tokenizers": {},
12
+ "children": [],
13
+ "node_i": "616",
14
+ "scripts": [],
15
+ "own_tokenizer": false
16
  },
17
  {
18
  "name": "Siawi",
19
  "iso_1_code": null,
20
  "iso_3_code": "mmp",
21
+ "tokenizers": {},
22
+ "children": [],
23
+ "node_i": "617",
24
+ "scripts": [],
25
+ "own_tokenizer": false
26
  }
27
+ ],
28
+ "node_i": "615",
29
+ "scripts": [],
30
+ "own_tokenizer": false
31
  }
data/Andamanese.json CHANGED
@@ -2,156 +2,194 @@
2
  "name": "Andamanese",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Great Andamanese",
10
  "iso_1_code": null,
11
  "iso_3_code": null,
12
- "tokenizer": null,
13
- "source": null,
14
  "children": [
15
  {
16
  "name": "Great Andamanese, Mixed",
17
  "iso_1_code": null,
18
  "iso_3_code": "gac",
19
- "tokenizer": null,
20
- "source": null,
21
- "children": []
 
 
22
  },
23
  {
24
  "name": "Central",
25
  "iso_1_code": null,
26
  "iso_3_code": null,
27
- "tokenizer": null,
28
- "source": null,
29
  "children": [
30
  {
31
  "name": "Aka-Bea",
32
  "iso_1_code": null,
33
  "iso_3_code": "abj",
34
- "tokenizer": null,
35
- "source": null,
36
- "children": []
 
 
37
  },
38
  {
39
  "name": "Akar-Bale",
40
  "iso_1_code": null,
41
  "iso_3_code": "acl",
42
- "tokenizer": null,
43
- "source": null,
44
- "children": []
 
 
45
  },
46
  {
47
  "name": "Aka-Kede",
48
  "iso_1_code": null,
49
  "iso_3_code": "akx",
50
- "tokenizer": null,
51
- "source": null,
52
- "children": []
 
 
53
  },
54
  {
55
  "name": "Aka-Kol",
56
  "iso_1_code": null,
57
  "iso_3_code": "aky",
58
- "tokenizer": null,
59
- "source": null,
60
- "children": []
 
 
61
  },
62
  {
63
  "name": "A-Pucikwar",
64
  "iso_1_code": null,
65
  "iso_3_code": "apq",
66
- "tokenizer": null,
67
- "source": null,
68
- "children": []
 
 
69
  },
70
  {
71
  "name": "Oko-Juwoi",
72
  "iso_1_code": null,
73
  "iso_3_code": "okj",
74
- "tokenizer": null,
75
- "source": null,
76
- "children": []
 
 
77
  }
78
- ]
 
 
 
79
  },
80
  {
81
  "name": "Northern",
82
  "iso_1_code": null,
83
  "iso_3_code": null,
84
- "tokenizer": null,
85
- "source": null,
86
  "children": [
87
  {
88
  "name": "Aka-Cari",
89
  "iso_1_code": null,
90
  "iso_3_code": "aci",
91
- "tokenizer": null,
92
- "source": null,
93
- "children": []
 
 
94
  },
95
  {
96
  "name": "Aka-Kora",
97
  "iso_1_code": null,
98
  "iso_3_code": "ack",
99
- "tokenizer": null,
100
- "source": null,
101
- "children": []
 
 
102
  },
103
  {
104
  "name": "Aka-Jeru",
105
  "iso_1_code": null,
106
  "iso_3_code": "akj",
107
- "tokenizer": null,
108
- "source": null,
109
- "children": []
 
 
110
  },
111
  {
112
  "name": "Aka-Bo",
113
  "iso_1_code": null,
114
  "iso_3_code": "akm",
115
- "tokenizer": null,
116
- "source": null,
117
- "children": []
 
 
118
  }
119
- ]
 
 
 
120
  }
121
- ]
 
 
 
122
  },
123
  {
124
  "name": "South Andamanese",
125
  "iso_1_code": null,
126
  "iso_3_code": null,
127
- "tokenizer": null,
128
- "source": null,
129
  "children": [
130
  {
131
  "name": "Jarawa",
132
  "iso_1_code": null,
133
  "iso_3_code": "anq",
134
- "tokenizer": null,
135
- "source": null,
136
- "children": []
 
 
137
  },
138
  {
139
  "name": "\u00d6\u00f1ge",
140
  "iso_1_code": null,
141
  "iso_3_code": "oon",
142
- "tokenizer": null,
143
- "source": null,
144
- "children": []
 
 
145
  },
146
  {
147
  "name": "Sentinel",
148
  "iso_1_code": null,
149
  "iso_3_code": "std",
150
- "tokenizer": null,
151
- "source": null,
152
- "children": []
 
 
153
  }
154
- ]
 
 
 
155
  }
156
- ]
 
 
 
157
  }
 
2
  "name": "Andamanese",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Great Andamanese",
9
  "iso_1_code": null,
10
  "iso_3_code": null,
11
+ "tokenizers": {},
 
12
  "children": [
13
  {
14
  "name": "Great Andamanese, Mixed",
15
  "iso_1_code": null,
16
  "iso_3_code": "gac",
17
+ "tokenizers": {},
18
+ "children": [],
19
+ "node_i": "620",
20
+ "scripts": [],
21
+ "own_tokenizer": false
22
  },
23
  {
24
  "name": "Central",
25
  "iso_1_code": null,
26
  "iso_3_code": null,
27
+ "tokenizers": {},
 
28
  "children": [
29
  {
30
  "name": "Aka-Bea",
31
  "iso_1_code": null,
32
  "iso_3_code": "abj",
33
+ "tokenizers": {},
34
+ "children": [],
35
+ "node_i": "622",
36
+ "scripts": [],
37
+ "own_tokenizer": false
38
  },
39
  {
40
  "name": "Akar-Bale",
41
  "iso_1_code": null,
42
  "iso_3_code": "acl",
43
+ "tokenizers": {},
44
+ "children": [],
45
+ "node_i": "623",
46
+ "scripts": [],
47
+ "own_tokenizer": false
48
  },
49
  {
50
  "name": "Aka-Kede",
51
  "iso_1_code": null,
52
  "iso_3_code": "akx",
53
+ "tokenizers": {},
54
+ "children": [],
55
+ "node_i": "624",
56
+ "scripts": [],
57
+ "own_tokenizer": false
58
  },
59
  {
60
  "name": "Aka-Kol",
61
  "iso_1_code": null,
62
  "iso_3_code": "aky",
63
+ "tokenizers": {},
64
+ "children": [],
65
+ "node_i": "625",
66
+ "scripts": [],
67
+ "own_tokenizer": false
68
  },
69
  {
70
  "name": "A-Pucikwar",
71
  "iso_1_code": null,
72
  "iso_3_code": "apq",
73
+ "tokenizers": {},
74
+ "children": [],
75
+ "node_i": "626",
76
+ "scripts": [],
77
+ "own_tokenizer": false
78
  },
79
  {
80
  "name": "Oko-Juwoi",
81
  "iso_1_code": null,
82
  "iso_3_code": "okj",
83
+ "tokenizers": {},
84
+ "children": [],
85
+ "node_i": "627",
86
+ "scripts": [],
87
+ "own_tokenizer": false
88
  }
89
+ ],
90
+ "node_i": "621",
91
+ "scripts": [],
92
+ "own_tokenizer": false
93
  },
94
  {
95
  "name": "Northern",
96
  "iso_1_code": null,
97
  "iso_3_code": null,
98
+ "tokenizers": {},
 
99
  "children": [
100
  {
101
  "name": "Aka-Cari",
102
  "iso_1_code": null,
103
  "iso_3_code": "aci",
104
+ "tokenizers": {},
105
+ "children": [],
106
+ "node_i": "629",
107
+ "scripts": [],
108
+ "own_tokenizer": false
109
  },
110
  {
111
  "name": "Aka-Kora",
112
  "iso_1_code": null,
113
  "iso_3_code": "ack",
114
+ "tokenizers": {},
115
+ "children": [],
116
+ "node_i": "630",
117
+ "scripts": [],
118
+ "own_tokenizer": false
119
  },
120
  {
121
  "name": "Aka-Jeru",
122
  "iso_1_code": null,
123
  "iso_3_code": "akj",
124
+ "tokenizers": {},
125
+ "children": [],
126
+ "node_i": "631",
127
+ "scripts": [],
128
+ "own_tokenizer": false
129
  },
130
  {
131
  "name": "Aka-Bo",
132
  "iso_1_code": null,
133
  "iso_3_code": "akm",
134
+ "tokenizers": {},
135
+ "children": [],
136
+ "node_i": "632",
137
+ "scripts": [],
138
+ "own_tokenizer": false
139
  }
140
+ ],
141
+ "node_i": "628",
142
+ "scripts": [],
143
+ "own_tokenizer": false
144
  }
145
+ ],
146
+ "node_i": "619",
147
+ "scripts": [],
148
+ "own_tokenizer": false
149
  },
150
  {
151
  "name": "South Andamanese",
152
  "iso_1_code": null,
153
  "iso_3_code": null,
154
+ "tokenizers": {},
 
155
  "children": [
156
  {
157
  "name": "Jarawa",
158
  "iso_1_code": null,
159
  "iso_3_code": "anq",
160
+ "tokenizers": {},
161
+ "children": [],
162
+ "node_i": "634",
163
+ "scripts": [],
164
+ "own_tokenizer": false
165
  },
166
  {
167
  "name": "\u00d6\u00f1ge",
168
  "iso_1_code": null,
169
  "iso_3_code": "oon",
170
+ "tokenizers": {},
171
+ "children": [],
172
+ "node_i": "635",
173
+ "scripts": [],
174
+ "own_tokenizer": false
175
  },
176
  {
177
  "name": "Sentinel",
178
  "iso_1_code": null,
179
  "iso_3_code": "std",
180
+ "tokenizers": {},
181
+ "children": [],
182
+ "node_i": "636",
183
+ "scripts": [],
184
+ "own_tokenizer": false
185
  }
186
+ ],
187
+ "node_i": "633",
188
+ "scripts": [],
189
+ "own_tokenizer": false
190
  }
191
+ ],
192
+ "node_i": "618",
193
+ "scripts": [],
194
+ "own_tokenizer": false
195
  }
data/Arafundi.json CHANGED
@@ -2,32 +2,40 @@
2
  "name": "Arafundi",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Andai",
10
  "iso_1_code": null,
11
  "iso_3_code": "afd",
12
- "tokenizer": null,
13
- "source": null,
14
- "children": []
 
 
15
  },
16
  {
17
  "name": "Nanubae",
18
  "iso_1_code": null,
19
  "iso_3_code": "afk",
20
- "tokenizer": null,
21
- "source": null,
22
- "children": []
 
 
23
  },
24
  {
25
  "name": "Tapei",
26
  "iso_1_code": null,
27
  "iso_3_code": "afp",
28
- "tokenizer": null,
29
- "source": null,
30
- "children": []
 
 
31
  }
32
- ]
 
 
 
33
  }
 
2
  "name": "Arafundi",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Andai",
9
  "iso_1_code": null,
10
  "iso_3_code": "afd",
11
+ "tokenizers": {},
12
+ "children": [],
13
+ "node_i": "638",
14
+ "scripts": [],
15
+ "own_tokenizer": false
16
  },
17
  {
18
  "name": "Nanubae",
19
  "iso_1_code": null,
20
  "iso_3_code": "afk",
21
+ "tokenizers": {},
22
+ "children": [],
23
+ "node_i": "639",
24
+ "scripts": [],
25
+ "own_tokenizer": false
26
  },
27
  {
28
  "name": "Tapei",
29
  "iso_1_code": null,
30
  "iso_3_code": "afp",
31
+ "tokenizers": {},
32
+ "children": [],
33
+ "node_i": "640",
34
+ "scripts": [],
35
+ "own_tokenizer": false
36
  }
37
+ ],
38
+ "node_i": "637",
39
+ "scripts": [],
40
+ "own_tokenizer": false
41
  }
data/Arai (Left May).json CHANGED
@@ -2,56 +2,72 @@
2
  "name": "Arai (Left May)",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Sawiyanu",
10
  "iso_1_code": null,
11
  "iso_3_code": "amm",
12
- "tokenizer": null,
13
- "source": null,
14
- "children": []
 
 
 
 
15
  },
16
  {
17
  "name": "Bo",
18
  "iso_1_code": null,
19
  "iso_3_code": "bpw",
20
- "tokenizer": null,
21
- "source": null,
22
- "children": []
 
 
23
  },
24
  {
25
  "name": "Yawuno Teneyo",
26
  "iso_1_code": null,
27
  "iso_3_code": "itr",
28
- "tokenizer": null,
29
- "source": null,
30
- "children": []
 
 
31
  },
32
  {
33
  "name": "Nakwi",
34
  "iso_1_code": null,
35
  "iso_3_code": "nax",
36
- "tokenizer": null,
37
- "source": null,
38
- "children": []
 
 
39
  },
40
  {
41
  "name": "Nimo",
42
  "iso_1_code": null,
43
  "iso_3_code": "niw",
44
- "tokenizer": null,
45
- "source": null,
46
- "children": []
 
 
47
  },
48
  {
49
  "name": "Owiniga",
50
  "iso_1_code": null,
51
  "iso_3_code": "owi",
52
- "tokenizer": null,
53
- "source": null,
54
- "children": []
 
 
55
  }
56
- ]
 
 
 
57
  }
 
2
  "name": "Arai (Left May)",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Sawiyanu",
9
  "iso_1_code": null,
10
  "iso_3_code": "amm",
11
+ "tokenizers": {},
12
+ "children": [],
13
+ "node_i": "642",
14
+ "scripts": [
15
+ "Latn"
16
+ ],
17
+ "own_tokenizer": false
18
  },
19
  {
20
  "name": "Bo",
21
  "iso_1_code": null,
22
  "iso_3_code": "bpw",
23
+ "tokenizers": {},
24
+ "children": [],
25
+ "node_i": "643",
26
+ "scripts": [],
27
+ "own_tokenizer": false
28
  },
29
  {
30
  "name": "Yawuno Teneyo",
31
  "iso_1_code": null,
32
  "iso_3_code": "itr",
33
+ "tokenizers": {},
34
+ "children": [],
35
+ "node_i": "644",
36
+ "scripts": [],
37
+ "own_tokenizer": false
38
  },
39
  {
40
  "name": "Nakwi",
41
  "iso_1_code": null,
42
  "iso_3_code": "nax",
43
+ "tokenizers": {},
44
+ "children": [],
45
+ "node_i": "645",
46
+ "scripts": [],
47
+ "own_tokenizer": false
48
  },
49
  {
50
  "name": "Nimo",
51
  "iso_1_code": null,
52
  "iso_3_code": "niw",
53
+ "tokenizers": {},
54
+ "children": [],
55
+ "node_i": "646",
56
+ "scripts": [],
57
+ "own_tokenizer": false
58
  },
59
  {
60
  "name": "Owiniga",
61
  "iso_1_code": null,
62
  "iso_3_code": "owi",
63
+ "tokenizers": {},
64
+ "children": [],
65
+ "node_i": "647",
66
+ "scripts": [],
67
+ "own_tokenizer": false
68
  }
69
+ ],
70
+ "node_i": "641",
71
+ "scripts": [],
72
+ "own_tokenizer": false
73
  }
data/Arauan.json CHANGED
@@ -2,65 +2,87 @@
2
  "name": "Arauan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Aru\u00e1",
10
  "iso_1_code": null,
11
  "iso_3_code": "aru",
12
- "tokenizer": null,
13
- "source": null,
14
- "children": []
 
 
15
  },
16
  {
17
  "name": "Kulina",
18
  "iso_1_code": null,
19
  "iso_3_code": "cul",
20
- "tokenizer": null,
21
- "source": null,
22
- "children": []
 
 
 
 
23
  },
24
  {
25
  "name": "Den\u00ed",
26
  "iso_1_code": null,
27
  "iso_3_code": "dny",
28
- "tokenizer": null,
29
- "source": null,
30
- "children": []
 
 
31
  },
32
  {
33
  "name": "Paumar\u00ed",
34
  "iso_1_code": null,
35
  "iso_3_code": "pad",
36
- "tokenizer": null,
37
- "source": null,
38
- "children": []
 
 
 
 
39
  },
40
  {
41
  "name": "Suruah\u00e1",
42
  "iso_1_code": null,
43
  "iso_3_code": "swx",
44
- "tokenizer": null,
45
- "source": null,
46
- "children": []
 
 
47
  },
48
  {
49
  "name": "Jamamadi",
50
  "iso_1_code": null,
51
  "iso_3_code": null,
52
- "tokenizer": null,
53
- "source": null,
54
  "children": [
55
  {
56
  "name": "Jamamad\u00ed",
57
  "iso_1_code": null,
58
  "iso_3_code": "jaa",
59
- "tokenizer": null,
60
- "source": null,
61
- "children": []
 
 
 
 
62
  }
63
- ]
 
 
 
64
  }
65
- ]
 
 
 
66
  }
 
2
  "name": "Arauan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Aru\u00e1",
9
  "iso_1_code": null,
10
  "iso_3_code": "aru",
11
+ "tokenizers": {},
12
+ "children": [],
13
+ "node_i": "649",
14
+ "scripts": [],
15
+ "own_tokenizer": false
16
  },
17
  {
18
  "name": "Kulina",
19
  "iso_1_code": null,
20
  "iso_3_code": "cul",
21
+ "tokenizers": {},
22
+ "children": [],
23
+ "node_i": "650",
24
+ "scripts": [
25
+ "Latn"
26
+ ],
27
+ "own_tokenizer": false
28
  },
29
  {
30
  "name": "Den\u00ed",
31
  "iso_1_code": null,
32
  "iso_3_code": "dny",
33
+ "tokenizers": {},
34
+ "children": [],
35
+ "node_i": "651",
36
+ "scripts": [],
37
+ "own_tokenizer": false
38
  },
39
  {
40
  "name": "Paumar\u00ed",
41
  "iso_1_code": null,
42
  "iso_3_code": "pad",
43
+ "tokenizers": {},
44
+ "children": [],
45
+ "node_i": "652",
46
+ "scripts": [
47
+ "Latn"
48
+ ],
49
+ "own_tokenizer": false
50
  },
51
  {
52
  "name": "Suruah\u00e1",
53
  "iso_1_code": null,
54
  "iso_3_code": "swx",
55
+ "tokenizers": {},
56
+ "children": [],
57
+ "node_i": "653",
58
+ "scripts": [],
59
+ "own_tokenizer": false
60
  },
61
  {
62
  "name": "Jamamadi",
63
  "iso_1_code": null,
64
  "iso_3_code": null,
65
+ "tokenizers": {},
 
66
  "children": [
67
  {
68
  "name": "Jamamad\u00ed",
69
  "iso_1_code": null,
70
  "iso_3_code": "jaa",
71
+ "tokenizers": {},
72
+ "children": [],
73
+ "node_i": "655",
74
+ "scripts": [
75
+ "Latn"
76
+ ],
77
+ "own_tokenizer": false
78
  }
79
+ ],
80
+ "node_i": "654",
81
+ "scripts": [],
82
+ "own_tokenizer": false
83
  }
84
+ ],
85
+ "node_i": "648",
86
+ "scripts": [],
87
+ "own_tokenizer": false
88
  }
data/Australian.json CHANGED
The diff for this file is too large to render. See raw diff
 
data/Austro-Asiatic.json CHANGED
The diff for this file is too large to render. See raw diff
 
data/Austronesian.json CHANGED
The diff for this file is too large to render. See raw diff
 
data/Aymaran.json CHANGED
@@ -2,50 +2,64 @@
2
  "name": "Aymaran",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Aymara",
10
  "iso_1_code": null,
11
  "iso_3_code": null,
12
- "tokenizer": null,
13
- "source": null,
14
  "children": [
15
  {
16
  "name": "Aymara, Southern",
17
  "iso_1_code": "ay",
18
  "iso_3_code": "ayc",
19
- "tokenizer": null,
20
- "source": "macrolanguage",
21
- "children": []
 
 
22
  },
23
  {
24
  "name": "Aymara, Central",
25
  "iso_1_code": "ay",
26
  "iso_3_code": "ayr",
27
- "tokenizer": null,
28
- "source": "macrolanguage",
29
- "children": []
 
 
 
 
30
  }
31
- ]
 
 
 
32
  },
33
  {
34
  "name": "Tupe",
35
  "iso_1_code": null,
36
  "iso_3_code": null,
37
- "tokenizer": null,
38
- "source": null,
39
  "children": [
40
  {
41
  "name": "Jaqaru",
42
  "iso_1_code": null,
43
  "iso_3_code": "jqr",
44
- "tokenizer": null,
45
- "source": null,
46
- "children": []
 
 
47
  }
48
- ]
 
 
 
49
  }
50
- ]
 
 
 
51
  }
 
2
  "name": "Aymaran",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Aymara",
9
  "iso_1_code": null,
10
  "iso_3_code": null,
11
+ "tokenizers": {},
 
12
  "children": [
13
  {
14
  "name": "Aymara, Southern",
15
  "iso_1_code": "ay",
16
  "iso_3_code": "ayc",
17
+ "tokenizers": {},
18
+ "children": [],
19
+ "node_i": "3257",
20
+ "scripts": [],
21
+ "own_tokenizer": false
22
  },
23
  {
24
  "name": "Aymara, Central",
25
  "iso_1_code": "ay",
26
  "iso_3_code": "ayr",
27
+ "tokenizers": {},
28
+ "children": [],
29
+ "node_i": "3258",
30
+ "scripts": [
31
+ "Latn"
32
+ ],
33
+ "own_tokenizer": false
34
  }
35
+ ],
36
+ "node_i": "3256",
37
+ "scripts": [],
38
+ "own_tokenizer": false
39
  },
40
  {
41
  "name": "Tupe",
42
  "iso_1_code": null,
43
  "iso_3_code": null,
44
+ "tokenizers": {},
 
45
  "children": [
46
  {
47
  "name": "Jaqaru",
48
  "iso_1_code": null,
49
  "iso_3_code": "jqr",
50
+ "tokenizers": {},
51
+ "children": [],
52
+ "node_i": "3260",
53
+ "scripts": [],
54
+ "own_tokenizer": false
55
  }
56
+ ],
57
+ "node_i": "3259",
58
+ "scripts": [],
59
+ "own_tokenizer": false
60
  }
61
+ ],
62
+ "node_i": "3255",
63
+ "scripts": [],
64
+ "own_tokenizer": false
65
  }
data/Barbacoan.json CHANGED
@@ -2,50 +2,68 @@
2
  "name": "Barbacoan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Northern",
10
  "iso_1_code": null,
11
  "iso_3_code": null,
12
- "tokenizer": null,
13
- "source": null,
14
  "children": [
15
  {
16
  "name": "Awa-Cuaiquer",
17
  "iso_1_code": null,
18
  "iso_3_code": "kwi",
19
- "tokenizer": null,
20
- "source": null,
21
- "children": []
 
 
 
 
22
  }
23
- ]
 
 
 
24
  },
25
  {
26
  "name": "Southern",
27
  "iso_1_code": null,
28
  "iso_3_code": null,
29
- "tokenizer": null,
30
- "source": null,
31
  "children": [
32
  {
33
  "name": "Chachi",
34
  "iso_1_code": null,
35
  "iso_3_code": "cbi",
36
- "tokenizer": null,
37
- "source": null,
38
- "children": []
 
 
 
 
39
  },
40
  {
41
  "name": "Tsafiki",
42
  "iso_1_code": null,
43
  "iso_3_code": "cof",
44
- "tokenizer": null,
45
- "source": null,
46
- "children": []
 
 
 
 
47
  }
48
- ]
 
 
 
49
  }
50
- ]
 
 
 
51
  }
 
2
  "name": "Barbacoan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Northern",
9
  "iso_1_code": null,
10
  "iso_3_code": null,
11
+ "tokenizers": {},
 
12
  "children": [
13
  {
14
  "name": "Awa-Cuaiquer",
15
  "iso_1_code": null,
16
  "iso_3_code": "kwi",
17
+ "tokenizers": {},
18
+ "children": [],
19
+ "node_i": "3263",
20
+ "scripts": [
21
+ "Latn"
22
+ ],
23
+ "own_tokenizer": false
24
  }
25
+ ],
26
+ "node_i": "3262",
27
+ "scripts": [],
28
+ "own_tokenizer": false
29
  },
30
  {
31
  "name": "Southern",
32
  "iso_1_code": null,
33
  "iso_3_code": null,
34
+ "tokenizers": {},
 
35
  "children": [
36
  {
37
  "name": "Chachi",
38
  "iso_1_code": null,
39
  "iso_3_code": "cbi",
40
+ "tokenizers": {},
41
+ "children": [],
42
+ "node_i": "3265",
43
+ "scripts": [
44
+ "Latn"
45
+ ],
46
+ "own_tokenizer": false
47
  },
48
  {
49
  "name": "Tsafiki",
50
  "iso_1_code": null,
51
  "iso_3_code": "cof",
52
+ "tokenizers": {},
53
+ "children": [],
54
+ "node_i": "3266",
55
+ "scripts": [
56
+ "Latn"
57
+ ],
58
+ "own_tokenizer": false
59
  }
60
+ ],
61
+ "node_i": "3264",
62
+ "scripts": [],
63
+ "own_tokenizer": false
64
  }
65
+ ],
66
+ "node_i": "3261",
67
+ "scripts": [],
68
+ "own_tokenizer": false
69
  }
data/Bayono-Awbono.json CHANGED
@@ -2,24 +2,30 @@
2
  "name": "Bayono-Awbono",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Awbono",
10
  "iso_1_code": null,
11
  "iso_3_code": "awh",
12
- "tokenizer": null,
13
- "source": null,
14
- "children": []
 
 
15
  },
16
  {
17
  "name": "Bayono",
18
  "iso_1_code": null,
19
  "iso_3_code": "byl",
20
- "tokenizer": null,
21
- "source": null,
22
- "children": []
 
 
23
  }
24
- ]
 
 
 
25
  }
 
2
  "name": "Bayono-Awbono",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Awbono",
9
  "iso_1_code": null,
10
  "iso_3_code": "awh",
11
+ "tokenizers": {},
12
+ "children": [],
13
+ "node_i": "3268",
14
+ "scripts": [],
15
+ "own_tokenizer": false
16
  },
17
  {
18
  "name": "Bayono",
19
  "iso_1_code": null,
20
  "iso_3_code": "byl",
21
+ "tokenizers": {},
22
+ "children": [],
23
+ "node_i": "3269",
24
+ "scripts": [],
25
+ "own_tokenizer": false
26
  }
27
+ ],
28
+ "node_i": "3267",
29
+ "scripts": [],
30
+ "own_tokenizer": false
31
  }
data/Border.json CHANGED
@@ -2,155 +2,197 @@
2
  "name": "Border",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Bewani",
10
  "iso_1_code": null,
11
  "iso_3_code": null,
12
- "tokenizer": null,
13
- "source": null,
14
  "children": [
15
  {
16
  "name": "Ainbai",
17
  "iso_1_code": null,
18
  "iso_3_code": "aic",
19
- "tokenizer": null,
20
- "source": null,
21
- "children": []
 
 
22
  },
23
  {
24
  "name": "Kilmeri",
25
  "iso_1_code": null,
26
  "iso_3_code": "kih",
27
- "tokenizer": null,
28
- "source": null,
29
- "children": []
 
 
30
  },
31
  {
32
  "name": "Ningera",
33
  "iso_1_code": null,
34
  "iso_3_code": "nby",
35
- "tokenizer": null,
36
- "source": null,
37
- "children": []
 
 
38
  },
39
  {
40
  "name": "Pagi",
41
  "iso_1_code": null,
42
  "iso_3_code": "pgi",
43
- "tokenizer": null,
44
- "source": null,
45
- "children": []
 
 
46
  },
47
  {
48
  "name": "Umeda",
49
  "iso_1_code": null,
50
  "iso_3_code": "upi",
51
- "tokenizer": null,
52
- "source": null,
53
- "children": []
 
 
54
  }
55
- ]
 
 
 
56
  },
57
  {
58
  "name": "Taikat",
59
  "iso_1_code": null,
60
  "iso_3_code": null,
61
- "tokenizer": null,
62
- "source": null,
63
  "children": [
64
  {
65
  "name": "Taikat",
66
  "iso_1_code": null,
67
  "iso_3_code": "aos",
68
- "tokenizer": null,
69
- "source": null,
70
- "children": []
 
 
71
  },
72
  {
73
  "name": "Awyi",
74
  "iso_1_code": null,
75
  "iso_3_code": "auw",
76
- "tokenizer": null,
77
- "source": null,
78
- "children": []
 
 
79
  }
80
- ]
 
 
 
81
  },
82
  {
83
  "name": "Waris",
84
  "iso_1_code": null,
85
  "iso_3_code": null,
86
- "tokenizer": null,
87
- "source": null,
88
  "children": [
89
  {
90
  "name": "Amanab",
91
  "iso_1_code": null,
92
  "iso_3_code": "amn",
93
- "tokenizer": null,
94
- "source": null,
95
- "children": []
 
 
 
 
96
  },
97
  {
98
  "name": "Daonda",
99
  "iso_1_code": null,
100
  "iso_3_code": "dnd",
101
- "tokenizer": null,
102
- "source": null,
103
- "children": []
 
 
104
  },
105
  {
106
  "name": "Imonda",
107
  "iso_1_code": null,
108
  "iso_3_code": "imn",
109
- "tokenizer": null,
110
- "source": null,
111
- "children": []
 
 
112
  },
113
  {
114
  "name": "Manem",
115
  "iso_1_code": null,
116
  "iso_3_code": "jet",
117
- "tokenizer": null,
118
- "source": null,
119
- "children": []
 
 
120
  },
121
  {
122
  "name": "Auwe",
123
  "iso_1_code": null,
124
  "iso_3_code": "smf",
125
- "tokenizer": null,
126
- "source": null,
127
- "children": []
 
 
128
  },
129
  {
130
  "name": "Viid",
131
  "iso_1_code": null,
132
  "iso_3_code": "snu",
133
- "tokenizer": null,
134
- "source": null,
135
- "children": []
 
 
136
  },
137
  {
138
  "name": "Sowanda",
139
  "iso_1_code": null,
140
  "iso_3_code": "sow",
141
- "tokenizer": null,
142
- "source": null,
143
- "children": []
 
 
144
  },
145
  {
146
  "name": "Waris",
147
  "iso_1_code": null,
148
  "iso_3_code": "wrs",
149
- "tokenizer": null,
150
- "source": null,
151
- "children": []
 
 
 
 
152
  }
153
- ]
 
 
 
154
  }
155
- ]
 
 
 
156
  }
 
2
  "name": "Border",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Bewani",
9
  "iso_1_code": null,
10
  "iso_3_code": null,
11
+ "tokenizers": {},
 
12
  "children": [
13
  {
14
  "name": "Ainbai",
15
  "iso_1_code": null,
16
  "iso_3_code": "aic",
17
+ "tokenizers": {},
18
+ "children": [],
19
+ "node_i": "3272",
20
+ "scripts": [],
21
+ "own_tokenizer": false
22
  },
23
  {
24
  "name": "Kilmeri",
25
  "iso_1_code": null,
26
  "iso_3_code": "kih",
27
+ "tokenizers": {},
28
+ "children": [],
29
+ "node_i": "3273",
30
+ "scripts": [],
31
+ "own_tokenizer": false
32
  },
33
  {
34
  "name": "Ningera",
35
  "iso_1_code": null,
36
  "iso_3_code": "nby",
37
+ "tokenizers": {},
38
+ "children": [],
39
+ "node_i": "3274",
40
+ "scripts": [],
41
+ "own_tokenizer": false
42
  },
43
  {
44
  "name": "Pagi",
45
  "iso_1_code": null,
46
  "iso_3_code": "pgi",
47
+ "tokenizers": {},
48
+ "children": [],
49
+ "node_i": "3275",
50
+ "scripts": [],
51
+ "own_tokenizer": false
52
  },
53
  {
54
  "name": "Umeda",
55
  "iso_1_code": null,
56
  "iso_3_code": "upi",
57
+ "tokenizers": {},
58
+ "children": [],
59
+ "node_i": "3276",
60
+ "scripts": [],
61
+ "own_tokenizer": false
62
  }
63
+ ],
64
+ "node_i": "3271",
65
+ "scripts": [],
66
+ "own_tokenizer": false
67
  },
68
  {
69
  "name": "Taikat",
70
  "iso_1_code": null,
71
  "iso_3_code": null,
72
+ "tokenizers": {},
 
73
  "children": [
74
  {
75
  "name": "Taikat",
76
  "iso_1_code": null,
77
  "iso_3_code": "aos",
78
+ "tokenizers": {},
79
+ "children": [],
80
+ "node_i": "3278",
81
+ "scripts": [],
82
+ "own_tokenizer": false
83
  },
84
  {
85
  "name": "Awyi",
86
  "iso_1_code": null,
87
  "iso_3_code": "auw",
88
+ "tokenizers": {},
89
+ "children": [],
90
+ "node_i": "3279",
91
+ "scripts": [],
92
+ "own_tokenizer": false
93
  }
94
+ ],
95
+ "node_i": "3277",
96
+ "scripts": [],
97
+ "own_tokenizer": false
98
  },
99
  {
100
  "name": "Waris",
101
  "iso_1_code": null,
102
  "iso_3_code": null,
103
+ "tokenizers": {},
 
104
  "children": [
105
  {
106
  "name": "Amanab",
107
  "iso_1_code": null,
108
  "iso_3_code": "amn",
109
+ "tokenizers": {},
110
+ "children": [],
111
+ "node_i": "3281",
112
+ "scripts": [
113
+ "Latn"
114
+ ],
115
+ "own_tokenizer": false
116
  },
117
  {
118
  "name": "Daonda",
119
  "iso_1_code": null,
120
  "iso_3_code": "dnd",
121
+ "tokenizers": {},
122
+ "children": [],
123
+ "node_i": "3282",
124
+ "scripts": [],
125
+ "own_tokenizer": false
126
  },
127
  {
128
  "name": "Imonda",
129
  "iso_1_code": null,
130
  "iso_3_code": "imn",
131
+ "tokenizers": {},
132
+ "children": [],
133
+ "node_i": "3283",
134
+ "scripts": [],
135
+ "own_tokenizer": false
136
  },
137
  {
138
  "name": "Manem",
139
  "iso_1_code": null,
140
  "iso_3_code": "jet",
141
+ "tokenizers": {},
142
+ "children": [],
143
+ "node_i": "3284",
144
+ "scripts": [],
145
+ "own_tokenizer": false
146
  },
147
  {
148
  "name": "Auwe",
149
  "iso_1_code": null,
150
  "iso_3_code": "smf",
151
+ "tokenizers": {},
152
+ "children": [],
153
+ "node_i": "3285",
154
+ "scripts": [],
155
+ "own_tokenizer": false
156
  },
157
  {
158
  "name": "Viid",
159
  "iso_1_code": null,
160
  "iso_3_code": "snu",
161
+ "tokenizers": {},
162
+ "children": [],
163
+ "node_i": "3286",
164
+ "scripts": [],
165
+ "own_tokenizer": false
166
  },
167
  {
168
  "name": "Sowanda",
169
  "iso_1_code": null,
170
  "iso_3_code": "sow",
171
+ "tokenizers": {},
172
+ "children": [],
173
+ "node_i": "3287",
174
+ "scripts": [],
175
+ "own_tokenizer": false
176
  },
177
  {
178
  "name": "Waris",
179
  "iso_1_code": null,
180
  "iso_3_code": "wrs",
181
+ "tokenizers": {},
182
+ "children": [],
183
+ "node_i": "3288",
184
+ "scripts": [
185
+ "Latn"
186
+ ],
187
+ "own_tokenizer": false
188
  }
189
+ ],
190
+ "node_i": "3280",
191
+ "scripts": [],
192
+ "own_tokenizer": false
193
  }
194
+ ],
195
+ "node_i": "3270",
196
+ "scripts": [],
197
+ "own_tokenizer": false
198
  }
data/Bororoan.json CHANGED
@@ -2,32 +2,42 @@
2
  "name": "Bororoan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Bor\u00f4ro",
10
  "iso_1_code": null,
11
  "iso_3_code": "bor",
12
- "tokenizer": null,
13
- "source": null,
14
- "children": []
 
 
 
 
15
  },
16
  {
17
  "name": "Otuke",
18
  "iso_1_code": null,
19
  "iso_3_code": "otu",
20
- "tokenizer": null,
21
- "source": null,
22
- "children": []
 
 
23
  },
24
  {
25
  "name": "Umot\u00edna",
26
  "iso_1_code": null,
27
  "iso_3_code": "umo",
28
- "tokenizer": null,
29
- "source": null,
30
- "children": []
 
 
31
  }
32
- ]
 
 
 
33
  }
 
2
  "name": "Bororoan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Bor\u00f4ro",
9
  "iso_1_code": null,
10
  "iso_3_code": "bor",
11
+ "tokenizers": {},
12
+ "children": [],
13
+ "node_i": "3290",
14
+ "scripts": [
15
+ "Latn"
16
+ ],
17
+ "own_tokenizer": false
18
  },
19
  {
20
  "name": "Otuke",
21
  "iso_1_code": null,
22
  "iso_3_code": "otu",
23
+ "tokenizers": {},
24
+ "children": [],
25
+ "node_i": "3291",
26
+ "scripts": [],
27
+ "own_tokenizer": false
28
  },
29
  {
30
  "name": "Umot\u00edna",
31
  "iso_1_code": null,
32
  "iso_3_code": "umo",
33
+ "tokenizers": {},
34
+ "children": [],
35
+ "node_i": "3292",
36
+ "scripts": [],
37
+ "own_tokenizer": false
38
  }
39
+ ],
40
+ "node_i": "3289",
41
+ "scripts": [],
42
+ "own_tokenizer": false
43
  }
data/Botocudoan.json CHANGED
@@ -2,16 +2,20 @@
2
  "name": "Botocudoan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Krenak",
10
  "iso_1_code": null,
11
  "iso_3_code": "kqq",
12
- "tokenizer": null,
13
- "source": null,
14
- "children": []
 
 
15
  }
16
- ]
 
 
 
17
  }
 
2
  "name": "Botocudoan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Krenak",
9
  "iso_1_code": null,
10
  "iso_3_code": "kqq",
11
+ "tokenizers": {},
12
+ "children": [],
13
+ "node_i": "3294",
14
+ "scripts": [],
15
+ "own_tokenizer": false
16
  }
17
+ ],
18
+ "node_i": "3293",
19
+ "scripts": [],
20
+ "own_tokenizer": false
21
  }
data/Caddoan.json CHANGED
@@ -2,75 +2,93 @@
2
  "name": "Caddoan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Caddo",
10
  "iso_1_code": null,
11
  "iso_3_code": "cad",
12
- "tokenizer": null,
13
- "source": null,
14
- "children": []
 
 
15
  },
16
  {
17
  "name": "Northern Caddoan",
18
  "iso_1_code": null,
19
  "iso_3_code": null,
20
- "tokenizer": null,
21
- "source": null,
22
  "children": [
23
  {
24
  "name": "Wichita",
25
  "iso_1_code": null,
26
  "iso_3_code": "wic",
27
- "tokenizer": null,
28
- "source": null,
29
- "children": []
 
 
30
  },
31
  {
32
  "name": "Kitsai-Proto-Pawnee",
33
  "iso_1_code": null,
34
  "iso_3_code": null,
35
- "tokenizer": null,
36
- "source": null,
37
  "children": [
38
  {
39
  "name": "Kitsai",
40
  "iso_1_code": null,
41
  "iso_3_code": "kii",
42
- "tokenizer": null,
43
- "source": null,
44
- "children": []
 
 
45
  },
46
  {
47
  "name": "Proto-Pawnee",
48
  "iso_1_code": null,
49
  "iso_3_code": null,
50
- "tokenizer": null,
51
- "source": null,
52
  "children": [
53
  {
54
  "name": "Arikara",
55
  "iso_1_code": null,
56
  "iso_3_code": "ari",
57
- "tokenizer": null,
58
- "source": null,
59
- "children": []
 
 
60
  },
61
  {
62
  "name": "Pawnee",
63
  "iso_1_code": null,
64
  "iso_3_code": "paw",
65
- "tokenizer": null,
66
- "source": null,
67
- "children": []
 
 
68
  }
69
- ]
 
 
 
70
  }
71
- ]
 
 
 
72
  }
73
- ]
 
 
 
74
  }
75
- ]
 
 
 
76
  }
 
2
  "name": "Caddoan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Caddo",
9
  "iso_1_code": null,
10
  "iso_3_code": "cad",
11
+ "tokenizers": {},
12
+ "children": [],
13
+ "node_i": "3296",
14
+ "scripts": [],
15
+ "own_tokenizer": false
16
  },
17
  {
18
  "name": "Northern Caddoan",
19
  "iso_1_code": null,
20
  "iso_3_code": null,
21
+ "tokenizers": {},
 
22
  "children": [
23
  {
24
  "name": "Wichita",
25
  "iso_1_code": null,
26
  "iso_3_code": "wic",
27
+ "tokenizers": {},
28
+ "children": [],
29
+ "node_i": "3298",
30
+ "scripts": [],
31
+ "own_tokenizer": false
32
  },
33
  {
34
  "name": "Kitsai-Proto-Pawnee",
35
  "iso_1_code": null,
36
  "iso_3_code": null,
37
+ "tokenizers": {},
 
38
  "children": [
39
  {
40
  "name": "Kitsai",
41
  "iso_1_code": null,
42
  "iso_3_code": "kii",
43
+ "tokenizers": {},
44
+ "children": [],
45
+ "node_i": "3300",
46
+ "scripts": [],
47
+ "own_tokenizer": false
48
  },
49
  {
50
  "name": "Proto-Pawnee",
51
  "iso_1_code": null,
52
  "iso_3_code": null,
53
+ "tokenizers": {},
 
54
  "children": [
55
  {
56
  "name": "Arikara",
57
  "iso_1_code": null,
58
  "iso_3_code": "ari",
59
+ "tokenizers": {},
60
+ "children": [],
61
+ "node_i": "3302",
62
+ "scripts": [],
63
+ "own_tokenizer": false
64
  },
65
  {
66
  "name": "Pawnee",
67
  "iso_1_code": null,
68
  "iso_3_code": "paw",
69
+ "tokenizers": {},
70
+ "children": [],
71
+ "node_i": "3303",
72
+ "scripts": [],
73
+ "own_tokenizer": false
74
  }
75
+ ],
76
+ "node_i": "3301",
77
+ "scripts": [],
78
+ "own_tokenizer": false
79
  }
80
+ ],
81
+ "node_i": "3299",
82
+ "scripts": [],
83
+ "own_tokenizer": false
84
  }
85
+ ],
86
+ "node_i": "3297",
87
+ "scripts": [],
88
+ "own_tokenizer": false
89
  }
90
+ ],
91
+ "node_i": "3295",
92
+ "scripts": [],
93
+ "own_tokenizer": false
94
  }
data/Cahuapanan.json CHANGED
@@ -2,24 +2,32 @@
2
  "name": "Cahuapanan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Shawi",
10
  "iso_1_code": null,
11
  "iso_3_code": "cbt",
12
- "tokenizer": null,
13
- "source": null,
14
- "children": []
 
 
 
 
15
  },
16
  {
17
  "name": "Jebero",
18
  "iso_1_code": null,
19
  "iso_3_code": "jeb",
20
- "tokenizer": null,
21
- "source": null,
22
- "children": []
 
 
23
  }
24
- ]
 
 
 
25
  }
 
2
  "name": "Cahuapanan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Shawi",
9
  "iso_1_code": null,
10
  "iso_3_code": "cbt",
11
+ "tokenizers": {},
12
+ "children": [],
13
+ "node_i": "3305",
14
+ "scripts": [
15
+ "Latn"
16
+ ],
17
+ "own_tokenizer": false
18
  },
19
  {
20
  "name": "Jebero",
21
  "iso_1_code": null,
22
  "iso_3_code": "jeb",
23
+ "tokenizers": {},
24
+ "children": [],
25
+ "node_i": "3306",
26
+ "scripts": [],
27
+ "own_tokenizer": false
28
  }
29
+ ],
30
+ "node_i": "3304",
31
+ "scripts": [],
32
+ "own_tokenizer": false
33
  }
data/Cariban.json CHANGED
@@ -2,445 +2,569 @@
2
  "name": "Cariban",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Carib",
10
  "iso_1_code": null,
11
  "iso_3_code": "car",
12
- "tokenizer": null,
13
- "source": null,
14
- "children": []
 
 
 
 
15
  },
16
  {
17
  "name": "Central",
18
  "iso_1_code": null,
19
  "iso_3_code": null,
20
- "tokenizer": null,
21
- "source": null,
22
  "children": [
23
  {
24
  "name": "Apala\u00ed",
25
  "iso_1_code": null,
26
  "iso_3_code": "apy",
27
- "tokenizer": null,
28
- "source": null,
29
- "children": []
 
 
 
 
30
  },
31
  {
32
  "name": "Cuman\u00e1",
33
  "iso_1_code": null,
34
  "iso_3_code": null,
35
- "tokenizer": null,
36
- "source": null,
37
  "children": [
38
  {
39
  "name": "Chaima",
40
  "iso_1_code": null,
41
  "iso_3_code": "ciy",
42
- "tokenizer": null,
43
- "source": null,
44
- "children": []
 
 
45
  },
46
  {
47
  "name": "Cumanagoto",
48
  "iso_1_code": null,
49
  "iso_3_code": "cuo",
50
- "tokenizer": null,
51
- "source": null,
52
- "children": []
 
 
53
  }
54
- ]
 
 
 
55
  },
56
  {
57
  "name": "Makiritare",
58
  "iso_1_code": null,
59
  "iso_3_code": null,
60
- "tokenizer": null,
61
- "source": null,
62
  "children": [
63
  {
64
  "name": "Maquiritari",
65
  "iso_1_code": null,
66
  "iso_3_code": "mch",
67
- "tokenizer": null,
68
- "source": null,
69
- "children": []
 
 
70
  }
71
- ]
 
 
 
72
  },
73
  {
74
  "name": "Mapoyo-Yavarana",
75
  "iso_1_code": null,
76
  "iso_3_code": null,
77
- "tokenizer": null,
78
- "source": null,
79
  "children": [
80
  {
81
  "name": "Mapoyo",
82
  "iso_1_code": null,
83
  "iso_3_code": "mcg",
84
- "tokenizer": null,
85
- "source": null,
86
- "children": []
 
 
87
  },
88
  {
89
  "name": "P\u00e9mono",
90
  "iso_1_code": null,
91
  "iso_3_code": "pev",
92
- "tokenizer": null,
93
- "source": null,
94
- "children": []
 
 
95
  },
96
  {
97
  "name": "Tamanaku",
98
  "iso_1_code": null,
99
  "iso_3_code": "tmz",
100
- "tokenizer": null,
101
- "source": null,
102
- "children": []
 
 
103
  },
104
  {
105
  "name": "Yabarana",
106
  "iso_1_code": null,
107
  "iso_3_code": "yar",
108
- "tokenizer": null,
109
- "source": null,
110
- "children": []
 
 
111
  }
112
- ]
 
 
 
113
  },
114
  {
115
  "name": "Wayana",
116
  "iso_1_code": null,
117
  "iso_3_code": null,
118
- "tokenizer": null,
119
- "source": null,
120
  "children": [
121
  {
122
  "name": "Kaxui\u00e2na",
123
  "iso_1_code": null,
124
  "iso_3_code": "kbb",
125
- "tokenizer": null,
126
- "source": null,
127
- "children": []
 
 
128
  },
129
  {
130
  "name": "Wayana",
131
  "iso_1_code": null,
132
  "iso_3_code": "way",
133
- "tokenizer": null,
134
- "source": null,
135
- "children": []
 
 
 
 
136
  }
137
- ]
 
 
 
138
  }
139
- ]
 
 
 
140
  },
141
  {
142
  "name": "Kashuyana",
143
  "iso_1_code": null,
144
  "iso_3_code": null,
145
- "tokenizer": null,
146
- "source": null,
147
  "children": [
148
  {
149
  "name": "Sikiana",
150
  "iso_1_code": null,
151
  "iso_3_code": "sik",
152
- "tokenizer": null,
153
- "source": null,
154
- "children": []
 
 
155
  }
156
- ]
 
 
 
157
  },
158
  {
159
  "name": "North Amazonian",
160
  "iso_1_code": null,
161
  "iso_3_code": null,
162
- "tokenizer": null,
163
- "source": null,
164
  "children": [
165
  {
166
  "name": "Pem\u00f3n",
167
  "iso_1_code": null,
168
  "iso_3_code": null,
169
- "tokenizer": null,
170
- "source": null,
171
  "children": [
172
  {
173
  "name": "Pem\u00f3n proper",
174
  "iso_1_code": null,
175
  "iso_3_code": null,
176
- "tokenizer": null,
177
- "source": null,
178
  "children": [
179
  {
180
  "name": "Pemon",
181
  "iso_1_code": null,
182
  "iso_3_code": "aoc",
183
- "tokenizer": null,
184
- "source": null,
185
- "children": []
 
 
186
  },
187
  {
188
  "name": "Macushi",
189
  "iso_1_code": null,
190
  "iso_3_code": "mbc",
191
- "tokenizer": null,
192
- "source": null,
193
- "children": []
 
 
 
 
194
  },
195
  {
196
  "name": "Kapong",
197
  "iso_1_code": null,
198
  "iso_3_code": null,
199
- "tokenizer": null,
200
- "source": null,
201
  "children": [
202
  {
203
  "name": "Akawaio",
204
  "iso_1_code": null,
205
  "iso_3_code": "ake",
206
- "tokenizer": null,
207
- "source": null,
208
- "children": []
 
 
 
 
209
  },
210
  {
211
  "name": "Patamona",
212
  "iso_1_code": null,
213
  "iso_3_code": "pbc",
214
- "tokenizer": null,
215
- "source": null,
216
- "children": []
 
 
 
 
217
  }
218
- ]
 
 
 
219
  }
220
- ]
 
 
 
221
  }
222
- ]
 
 
 
223
  },
224
  {
225
  "name": "Yawaper\u00ed",
226
  "iso_1_code": null,
227
  "iso_3_code": null,
228
- "tokenizer": null,
229
- "source": null,
230
  "children": [
231
  {
232
  "name": "Waimiri-Atroar\u00ed",
233
  "iso_1_code": null,
234
  "iso_3_code": "atr",
235
- "tokenizer": null,
236
- "source": null,
237
- "children": []
 
 
238
  }
239
- ]
 
 
 
240
  }
241
- ]
 
 
 
242
  },
243
  {
244
  "name": "South Amazonian",
245
  "iso_1_code": null,
246
  "iso_3_code": null,
247
- "tokenizer": null,
248
- "source": null,
249
  "children": [
250
  {
251
  "name": "E\u2019\u00f1apa Woromaipu",
252
  "iso_1_code": null,
253
  "iso_3_code": "pbh",
254
- "tokenizer": null,
255
- "source": null,
256
- "children": []
 
 
257
  },
258
  {
259
  "name": "Arara",
260
  "iso_1_code": null,
261
  "iso_3_code": null,
262
- "tokenizer": null,
263
- "source": null,
264
  "children": [
265
  {
266
  "name": "Arara, Par\u00e1",
267
  "iso_1_code": null,
268
  "iso_3_code": "aap",
269
- "tokenizer": null,
270
- "source": null,
271
- "children": []
 
 
272
  },
273
  {
274
  "name": "Ikpeng",
275
  "iso_1_code": null,
276
  "iso_3_code": "txi",
277
- "tokenizer": null,
278
- "source": null,
279
- "children": []
 
 
280
  }
281
- ]
 
 
 
282
  },
283
  {
284
  "name": "Bakair\u00ed",
285
  "iso_1_code": null,
286
  "iso_3_code": null,
287
- "tokenizer": null,
288
- "source": null,
289
  "children": [
290
  {
291
  "name": "Bakair\u00ed",
292
  "iso_1_code": null,
293
  "iso_3_code": "bkq",
294
- "tokenizer": null,
295
- "source": null,
296
- "children": []
 
 
 
 
297
  },
298
  {
299
  "name": "Amonap",
300
  "iso_1_code": null,
301
  "iso_3_code": null,
302
- "tokenizer": null,
303
- "source": null,
304
  "children": [
305
  {
306
  "name": "Kuik\u00faro-Kalap\u00e1lo",
307
  "iso_1_code": null,
308
  "iso_3_code": "kui",
309
- "tokenizer": null,
310
- "source": null,
311
- "children": []
 
 
312
  },
313
  {
314
  "name": "Matipuhy",
315
  "iso_1_code": null,
316
  "iso_3_code": "mzo",
317
- "tokenizer": null,
318
- "source": null,
319
- "children": []
 
 
320
  }
321
- ]
 
 
 
322
  }
323
- ]
 
 
 
324
  }
325
- ]
 
 
 
326
  },
327
  {
328
  "name": "Tiriy\u00f3",
329
  "iso_1_code": null,
330
  "iso_3_code": null,
331
- "tokenizer": null,
332
- "source": null,
333
  "children": [
334
  {
335
  "name": "Salum\u00e1",
336
  "iso_1_code": null,
337
  "iso_3_code": "slj",
338
- "tokenizer": null,
339
- "source": null,
340
- "children": []
 
 
341
  },
342
  {
343
  "name": "Karihona",
344
  "iso_1_code": null,
345
  "iso_3_code": null,
346
- "tokenizer": null,
347
- "source": null,
348
  "children": [
349
  {
350
  "name": "Carijona",
351
  "iso_1_code": null,
352
  "iso_3_code": "cbd",
353
- "tokenizer": null,
354
- "source": null,
355
- "children": []
 
 
356
  }
357
- ]
 
 
 
358
  },
359
  {
360
  "name": "Tiriy\u00f3",
361
  "iso_1_code": null,
362
  "iso_3_code": null,
363
- "tokenizer": null,
364
- "source": null,
365
  "children": [
366
  {
367
  "name": "Akurio",
368
  "iso_1_code": null,
369
  "iso_3_code": "ako",
370
- "tokenizer": null,
371
- "source": null,
372
- "children": []
 
 
373
  },
374
  {
375
  "name": "Tri\u00f3",
376
  "iso_1_code": null,
377
  "iso_3_code": "tri",
378
- "tokenizer": null,
379
- "source": null,
380
- "children": []
 
 
381
  }
382
- ]
 
 
 
383
  }
384
- ]
 
 
 
385
  },
386
  {
387
  "name": "Waiwai",
388
  "iso_1_code": null,
389
  "iso_3_code": null,
390
- "tokenizer": null,
391
- "source": null,
392
  "children": [
393
  {
394
  "name": "Hixkary\u00e1na",
395
  "iso_1_code": null,
396
  "iso_3_code": "hix",
397
- "tokenizer": null,
398
- "source": null,
399
- "children": []
 
 
 
 
400
  },
401
  {
402
  "name": "Waiwai",
403
  "iso_1_code": null,
404
  "iso_3_code": "waw",
405
- "tokenizer": null,
406
- "source": null,
407
- "children": []
 
 
408
  }
409
- ]
 
 
 
410
  },
411
  {
412
  "name": "Yukpa",
413
  "iso_1_code": null,
414
  "iso_3_code": null,
415
- "tokenizer": null,
416
- "source": null,
417
  "children": [
418
  {
419
  "name": "Yucpa-Yapreria",
420
  "iso_1_code": null,
421
  "iso_3_code": null,
422
- "tokenizer": null,
423
- "source": null,
424
  "children": [
425
  {
426
  "name": "Japreria",
427
  "iso_1_code": null,
428
  "iso_3_code": "jru",
429
- "tokenizer": null,
430
- "source": null,
431
- "children": []
 
 
432
  },
433
  {
434
  "name": "Yukpa",
435
  "iso_1_code": null,
436
  "iso_3_code": "yup",
437
- "tokenizer": null,
438
- "source": null,
439
- "children": []
 
 
 
 
440
  }
441
- ]
 
 
 
442
  }
443
- ]
 
 
 
444
  }
445
- ]
 
 
 
446
  }
 
2
  "name": "Cariban",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Carib",
9
  "iso_1_code": null,
10
  "iso_3_code": "car",
11
+ "tokenizers": {},
12
+ "children": [],
13
+ "node_i": "3308",
14
+ "scripts": [
15
+ "Latn"
16
+ ],
17
+ "own_tokenizer": false
18
  },
19
  {
20
  "name": "Central",
21
  "iso_1_code": null,
22
  "iso_3_code": null,
23
+ "tokenizers": {},
 
24
  "children": [
25
  {
26
  "name": "Apala\u00ed",
27
  "iso_1_code": null,
28
  "iso_3_code": "apy",
29
+ "tokenizers": {},
30
+ "children": [],
31
+ "node_i": "3310",
32
+ "scripts": [
33
+ "Latn"
34
+ ],
35
+ "own_tokenizer": false
36
  },
37
  {
38
  "name": "Cuman\u00e1",
39
  "iso_1_code": null,
40
  "iso_3_code": null,
41
+ "tokenizers": {},
 
42
  "children": [
43
  {
44
  "name": "Chaima",
45
  "iso_1_code": null,
46
  "iso_3_code": "ciy",
47
+ "tokenizers": {},
48
+ "children": [],
49
+ "node_i": "3312",
50
+ "scripts": [],
51
+ "own_tokenizer": false
52
  },
53
  {
54
  "name": "Cumanagoto",
55
  "iso_1_code": null,
56
  "iso_3_code": "cuo",
57
+ "tokenizers": {},
58
+ "children": [],
59
+ "node_i": "3313",
60
+ "scripts": [],
61
+ "own_tokenizer": false
62
  }
63
+ ],
64
+ "node_i": "3311",
65
+ "scripts": [],
66
+ "own_tokenizer": false
67
  },
68
  {
69
  "name": "Makiritare",
70
  "iso_1_code": null,
71
  "iso_3_code": null,
72
+ "tokenizers": {},
 
73
  "children": [
74
  {
75
  "name": "Maquiritari",
76
  "iso_1_code": null,
77
  "iso_3_code": "mch",
78
+ "tokenizers": {},
79
+ "children": [],
80
+ "node_i": "3315",
81
+ "scripts": [],
82
+ "own_tokenizer": false
83
  }
84
+ ],
85
+ "node_i": "3314",
86
+ "scripts": [],
87
+ "own_tokenizer": false
88
  },
89
  {
90
  "name": "Mapoyo-Yavarana",
91
  "iso_1_code": null,
92
  "iso_3_code": null,
93
+ "tokenizers": {},
 
94
  "children": [
95
  {
96
  "name": "Mapoyo",
97
  "iso_1_code": null,
98
  "iso_3_code": "mcg",
99
+ "tokenizers": {},
100
+ "children": [],
101
+ "node_i": "3317",
102
+ "scripts": [],
103
+ "own_tokenizer": false
104
  },
105
  {
106
  "name": "P\u00e9mono",
107
  "iso_1_code": null,
108
  "iso_3_code": "pev",
109
+ "tokenizers": {},
110
+ "children": [],
111
+ "node_i": "3318",
112
+ "scripts": [],
113
+ "own_tokenizer": false
114
  },
115
  {
116
  "name": "Tamanaku",
117
  "iso_1_code": null,
118
  "iso_3_code": "tmz",
119
+ "tokenizers": {},
120
+ "children": [],
121
+ "node_i": "3319",
122
+ "scripts": [],
123
+ "own_tokenizer": false
124
  },
125
  {
126
  "name": "Yabarana",
127
  "iso_1_code": null,
128
  "iso_3_code": "yar",
129
+ "tokenizers": {},
130
+ "children": [],
131
+ "node_i": "3320",
132
+ "scripts": [],
133
+ "own_tokenizer": false
134
  }
135
+ ],
136
+ "node_i": "3316",
137
+ "scripts": [],
138
+ "own_tokenizer": false
139
  },
140
  {
141
  "name": "Wayana",
142
  "iso_1_code": null,
143
  "iso_3_code": null,
144
+ "tokenizers": {},
 
145
  "children": [
146
  {
147
  "name": "Kaxui\u00e2na",
148
  "iso_1_code": null,
149
  "iso_3_code": "kbb",
150
+ "tokenizers": {},
151
+ "children": [],
152
+ "node_i": "3322",
153
+ "scripts": [],
154
+ "own_tokenizer": false
155
  },
156
  {
157
  "name": "Wayana",
158
  "iso_1_code": null,
159
  "iso_3_code": "way",
160
+ "tokenizers": {},
161
+ "children": [],
162
+ "node_i": "3323",
163
+ "scripts": [
164
+ "Latn"
165
+ ],
166
+ "own_tokenizer": false
167
  }
168
+ ],
169
+ "node_i": "3321",
170
+ "scripts": [],
171
+ "own_tokenizer": false
172
  }
173
+ ],
174
+ "node_i": "3309",
175
+ "scripts": [],
176
+ "own_tokenizer": false
177
  },
178
  {
179
  "name": "Kashuyana",
180
  "iso_1_code": null,
181
  "iso_3_code": null,
182
+ "tokenizers": {},
 
183
  "children": [
184
  {
185
  "name": "Sikiana",
186
  "iso_1_code": null,
187
  "iso_3_code": "sik",
188
+ "tokenizers": {},
189
+ "children": [],
190
+ "node_i": "3325",
191
+ "scripts": [],
192
+ "own_tokenizer": false
193
  }
194
+ ],
195
+ "node_i": "3324",
196
+ "scripts": [],
197
+ "own_tokenizer": false
198
  },
199
  {
200
  "name": "North Amazonian",
201
  "iso_1_code": null,
202
  "iso_3_code": null,
203
+ "tokenizers": {},
 
204
  "children": [
205
  {
206
  "name": "Pem\u00f3n",
207
  "iso_1_code": null,
208
  "iso_3_code": null,
209
+ "tokenizers": {},
 
210
  "children": [
211
  {
212
  "name": "Pem\u00f3n proper",
213
  "iso_1_code": null,
214
  "iso_3_code": null,
215
+ "tokenizers": {},
 
216
  "children": [
217
  {
218
  "name": "Pemon",
219
  "iso_1_code": null,
220
  "iso_3_code": "aoc",
221
+ "tokenizers": {},
222
+ "children": [],
223
+ "node_i": "3329",
224
+ "scripts": [],
225
+ "own_tokenizer": false
226
  },
227
  {
228
  "name": "Macushi",
229
  "iso_1_code": null,
230
  "iso_3_code": "mbc",
231
+ "tokenizers": {},
232
+ "children": [],
233
+ "node_i": "3330",
234
+ "scripts": [
235
+ "Latn"
236
+ ],
237
+ "own_tokenizer": false
238
  },
239
  {
240
  "name": "Kapong",
241
  "iso_1_code": null,
242
  "iso_3_code": null,
243
+ "tokenizers": {},
 
244
  "children": [
245
  {
246
  "name": "Akawaio",
247
  "iso_1_code": null,
248
  "iso_3_code": "ake",
249
+ "tokenizers": {},
250
+ "children": [],
251
+ "node_i": "3332",
252
+ "scripts": [
253
+ "Latn"
254
+ ],
255
+ "own_tokenizer": false
256
  },
257
  {
258
  "name": "Patamona",
259
  "iso_1_code": null,
260
  "iso_3_code": "pbc",
261
+ "tokenizers": {},
262
+ "children": [],
263
+ "node_i": "3333",
264
+ "scripts": [
265
+ "Latn"
266
+ ],
267
+ "own_tokenizer": false
268
  }
269
+ ],
270
+ "node_i": "3331",
271
+ "scripts": [],
272
+ "own_tokenizer": false
273
  }
274
+ ],
275
+ "node_i": "3328",
276
+ "scripts": [],
277
+ "own_tokenizer": false
278
  }
279
+ ],
280
+ "node_i": "3327",
281
+ "scripts": [],
282
+ "own_tokenizer": false
283
  },
284
  {
285
  "name": "Yawaper\u00ed",
286
  "iso_1_code": null,
287
  "iso_3_code": null,
288
+ "tokenizers": {},
 
289
  "children": [
290
  {
291
  "name": "Waimiri-Atroar\u00ed",
292
  "iso_1_code": null,
293
  "iso_3_code": "atr",
294
+ "tokenizers": {},
295
+ "children": [],
296
+ "node_i": "3335",
297
+ "scripts": [],
298
+ "own_tokenizer": false
299
  }
300
+ ],
301
+ "node_i": "3334",
302
+ "scripts": [],
303
+ "own_tokenizer": false
304
  }
305
+ ],
306
+ "node_i": "3326",
307
+ "scripts": [],
308
+ "own_tokenizer": false
309
  },
310
  {
311
  "name": "South Amazonian",
312
  "iso_1_code": null,
313
  "iso_3_code": null,
314
+ "tokenizers": {},
 
315
  "children": [
316
  {
317
  "name": "E\u2019\u00f1apa Woromaipu",
318
  "iso_1_code": null,
319
  "iso_3_code": "pbh",
320
+ "tokenizers": {},
321
+ "children": [],
322
+ "node_i": "3337",
323
+ "scripts": [],
324
+ "own_tokenizer": false
325
  },
326
  {
327
  "name": "Arara",
328
  "iso_1_code": null,
329
  "iso_3_code": null,
330
+ "tokenizers": {},
 
331
  "children": [
332
  {
333
  "name": "Arara, Par\u00e1",
334
  "iso_1_code": null,
335
  "iso_3_code": "aap",
336
+ "tokenizers": {},
337
+ "children": [],
338
+ "node_i": "3339",
339
+ "scripts": [],
340
+ "own_tokenizer": false
341
  },
342
  {
343
  "name": "Ikpeng",
344
  "iso_1_code": null,
345
  "iso_3_code": "txi",
346
+ "tokenizers": {},
347
+ "children": [],
348
+ "node_i": "3340",
349
+ "scripts": [],
350
+ "own_tokenizer": false
351
  }
352
+ ],
353
+ "node_i": "3338",
354
+ "scripts": [],
355
+ "own_tokenizer": false
356
  },
357
  {
358
  "name": "Bakair\u00ed",
359
  "iso_1_code": null,
360
  "iso_3_code": null,
361
+ "tokenizers": {},
 
362
  "children": [
363
  {
364
  "name": "Bakair\u00ed",
365
  "iso_1_code": null,
366
  "iso_3_code": "bkq",
367
+ "tokenizers": {},
368
+ "children": [],
369
+ "node_i": "3342",
370
+ "scripts": [
371
+ "Latn"
372
+ ],
373
+ "own_tokenizer": false
374
  },
375
  {
376
  "name": "Amonap",
377
  "iso_1_code": null,
378
  "iso_3_code": null,
379
+ "tokenizers": {},
 
380
  "children": [
381
  {
382
  "name": "Kuik\u00faro-Kalap\u00e1lo",
383
  "iso_1_code": null,
384
  "iso_3_code": "kui",
385
+ "tokenizers": {},
386
+ "children": [],
387
+ "node_i": "3344",
388
+ "scripts": [],
389
+ "own_tokenizer": false
390
  },
391
  {
392
  "name": "Matipuhy",
393
  "iso_1_code": null,
394
  "iso_3_code": "mzo",
395
+ "tokenizers": {},
396
+ "children": [],
397
+ "node_i": "3345",
398
+ "scripts": [],
399
+ "own_tokenizer": false
400
  }
401
+ ],
402
+ "node_i": "3343",
403
+ "scripts": [],
404
+ "own_tokenizer": false
405
  }
406
+ ],
407
+ "node_i": "3341",
408
+ "scripts": [],
409
+ "own_tokenizer": false
410
  }
411
+ ],
412
+ "node_i": "3336",
413
+ "scripts": [],
414
+ "own_tokenizer": false
415
  },
416
  {
417
  "name": "Tiriy\u00f3",
418
  "iso_1_code": null,
419
  "iso_3_code": null,
420
+ "tokenizers": {},
 
421
  "children": [
422
  {
423
  "name": "Salum\u00e1",
424
  "iso_1_code": null,
425
  "iso_3_code": "slj",
426
+ "tokenizers": {},
427
+ "children": [],
428
+ "node_i": "3347",
429
+ "scripts": [],
430
+ "own_tokenizer": false
431
  },
432
  {
433
  "name": "Karihona",
434
  "iso_1_code": null,
435
  "iso_3_code": null,
436
+ "tokenizers": {},
 
437
  "children": [
438
  {
439
  "name": "Carijona",
440
  "iso_1_code": null,
441
  "iso_3_code": "cbd",
442
+ "tokenizers": {},
443
+ "children": [],
444
+ "node_i": "3349",
445
+ "scripts": [],
446
+ "own_tokenizer": false
447
  }
448
+ ],
449
+ "node_i": "3348",
450
+ "scripts": [],
451
+ "own_tokenizer": false
452
  },
453
  {
454
  "name": "Tiriy\u00f3",
455
  "iso_1_code": null,
456
  "iso_3_code": null,
457
+ "tokenizers": {},
 
458
  "children": [
459
  {
460
  "name": "Akurio",
461
  "iso_1_code": null,
462
  "iso_3_code": "ako",
463
+ "tokenizers": {},
464
+ "children": [],
465
+ "node_i": "3351",
466
+ "scripts": [],
467
+ "own_tokenizer": false
468
  },
469
  {
470
  "name": "Tri\u00f3",
471
  "iso_1_code": null,
472
  "iso_3_code": "tri",
473
+ "tokenizers": {},
474
+ "children": [],
475
+ "node_i": "3352",
476
+ "scripts": [],
477
+ "own_tokenizer": false
478
  }
479
+ ],
480
+ "node_i": "3350",
481
+ "scripts": [],
482
+ "own_tokenizer": false
483
  }
484
+ ],
485
+ "node_i": "3346",
486
+ "scripts": [],
487
+ "own_tokenizer": false
488
  },
489
  {
490
  "name": "Waiwai",
491
  "iso_1_code": null,
492
  "iso_3_code": null,
493
+ "tokenizers": {},
 
494
  "children": [
495
  {
496
  "name": "Hixkary\u00e1na",
497
  "iso_1_code": null,
498
  "iso_3_code": "hix",
499
+ "tokenizers": {},
500
+ "children": [],
501
+ "node_i": "3354",
502
+ "scripts": [
503
+ "Latn"
504
+ ],
505
+ "own_tokenizer": false
506
  },
507
  {
508
  "name": "Waiwai",
509
  "iso_1_code": null,
510
  "iso_3_code": "waw",
511
+ "tokenizers": {},
512
+ "children": [],
513
+ "node_i": "3355",
514
+ "scripts": [],
515
+ "own_tokenizer": false
516
  }
517
+ ],
518
+ "node_i": "3353",
519
+ "scripts": [],
520
+ "own_tokenizer": false
521
  },
522
  {
523
  "name": "Yukpa",
524
  "iso_1_code": null,
525
  "iso_3_code": null,
526
+ "tokenizers": {},
 
527
  "children": [
528
  {
529
  "name": "Yucpa-Yapreria",
530
  "iso_1_code": null,
531
  "iso_3_code": null,
532
+ "tokenizers": {},
 
533
  "children": [
534
  {
535
  "name": "Japreria",
536
  "iso_1_code": null,
537
  "iso_3_code": "jru",
538
+ "tokenizers": {},
539
+ "children": [],
540
+ "node_i": "3358",
541
+ "scripts": [],
542
+ "own_tokenizer": false
543
  },
544
  {
545
  "name": "Yukpa",
546
  "iso_1_code": null,
547
  "iso_3_code": "yup",
548
+ "tokenizers": {},
549
+ "children": [],
550
+ "node_i": "3359",
551
+ "scripts": [
552
+ "Latn"
553
+ ],
554
+ "own_tokenizer": false
555
  }
556
+ ],
557
+ "node_i": "3357",
558
+ "scripts": [],
559
+ "own_tokenizer": false
560
  }
561
+ ],
562
+ "node_i": "3356",
563
+ "scripts": [],
564
+ "own_tokenizer": false
565
  }
566
+ ],
567
+ "node_i": "3307",
568
+ "scripts": [],
569
+ "own_tokenizer": false
570
  }
data/Central Solomons.json CHANGED
@@ -2,40 +2,50 @@
2
  "name": "Central Solomons",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Bilua",
10
  "iso_1_code": null,
11
  "iso_3_code": "blb",
12
- "tokenizer": null,
13
- "source": null,
14
- "children": []
 
 
15
  },
16
  {
17
  "name": "Lavukaleve",
18
  "iso_1_code": null,
19
  "iso_3_code": "lvk",
20
- "tokenizer": null,
21
- "source": null,
22
- "children": []
 
 
23
  },
24
  {
25
  "name": "Savosavo",
26
  "iso_1_code": null,
27
  "iso_3_code": "svs",
28
- "tokenizer": null,
29
- "source": null,
30
- "children": []
 
 
31
  },
32
  {
33
  "name": "Touo",
34
  "iso_1_code": null,
35
  "iso_3_code": "tqu",
36
- "tokenizer": null,
37
- "source": null,
38
- "children": []
 
 
39
  }
40
- ]
 
 
 
41
  }
 
2
  "name": "Central Solomons",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Bilua",
9
  "iso_1_code": null,
10
  "iso_3_code": "blb",
11
+ "tokenizers": {},
12
+ "children": [],
13
+ "node_i": "3361",
14
+ "scripts": [],
15
+ "own_tokenizer": false
16
  },
17
  {
18
  "name": "Lavukaleve",
19
  "iso_1_code": null,
20
  "iso_3_code": "lvk",
21
+ "tokenizers": {},
22
+ "children": [],
23
+ "node_i": "3362",
24
+ "scripts": [],
25
+ "own_tokenizer": false
26
  },
27
  {
28
  "name": "Savosavo",
29
  "iso_1_code": null,
30
  "iso_3_code": "svs",
31
+ "tokenizers": {},
32
+ "children": [],
33
+ "node_i": "3363",
34
+ "scripts": [],
35
+ "own_tokenizer": false
36
  },
37
  {
38
  "name": "Touo",
39
  "iso_1_code": null,
40
  "iso_3_code": "tqu",
41
+ "tokenizers": {},
42
+ "children": [],
43
+ "node_i": "3364",
44
+ "scripts": [],
45
+ "own_tokenizer": false
46
  }
47
+ ],
48
+ "node_i": "3360",
49
+ "scripts": [],
50
+ "own_tokenizer": false
51
  }
data/Chapacuran.json CHANGED
@@ -2,58 +2,72 @@
2
  "name": "Chapacuran",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Itene",
10
  "iso_1_code": null,
11
  "iso_3_code": null,
12
- "tokenizer": null,
13
- "source": null,
14
  "children": [
15
  {
16
  "name": "Itene",
17
  "iso_1_code": null,
18
  "iso_3_code": "ite",
19
- "tokenizer": null,
20
- "source": null,
21
- "children": []
 
 
22
  },
23
  {
24
  "name": "Tor\u00e1",
25
  "iso_1_code": null,
26
  "iso_3_code": "trz",
27
- "tokenizer": null,
28
- "source": null,
29
- "children": []
 
 
30
  }
31
- ]
 
 
 
32
  },
33
  {
34
  "name": "Wari",
35
  "iso_1_code": null,
36
  "iso_3_code": null,
37
- "tokenizer": null,
38
- "source": null,
39
  "children": [
40
  {
41
  "name": "Oro Win",
42
  "iso_1_code": null,
43
  "iso_3_code": "orw",
44
- "tokenizer": null,
45
- "source": null,
46
- "children": []
 
 
47
  },
48
  {
49
  "name": "Paka\u00e1snovos",
50
  "iso_1_code": null,
51
  "iso_3_code": "pav",
52
- "tokenizer": null,
53
- "source": null,
54
- "children": []
 
 
55
  }
56
- ]
 
 
 
57
  }
58
- ]
 
 
 
59
  }
 
2
  "name": "Chapacuran",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Itene",
9
  "iso_1_code": null,
10
  "iso_3_code": null,
11
+ "tokenizers": {},
 
12
  "children": [
13
  {
14
  "name": "Itene",
15
  "iso_1_code": null,
16
  "iso_3_code": "ite",
17
+ "tokenizers": {},
18
+ "children": [],
19
+ "node_i": "3367",
20
+ "scripts": [],
21
+ "own_tokenizer": false
22
  },
23
  {
24
  "name": "Tor\u00e1",
25
  "iso_1_code": null,
26
  "iso_3_code": "trz",
27
+ "tokenizers": {},
28
+ "children": [],
29
+ "node_i": "3368",
30
+ "scripts": [],
31
+ "own_tokenizer": false
32
  }
33
+ ],
34
+ "node_i": "3366",
35
+ "scripts": [],
36
+ "own_tokenizer": false
37
  },
38
  {
39
  "name": "Wari",
40
  "iso_1_code": null,
41
  "iso_3_code": null,
42
+ "tokenizers": {},
 
43
  "children": [
44
  {
45
  "name": "Oro Win",
46
  "iso_1_code": null,
47
  "iso_3_code": "orw",
48
+ "tokenizers": {},
49
+ "children": [],
50
+ "node_i": "3370",
51
+ "scripts": [],
52
+ "own_tokenizer": false
53
  },
54
  {
55
  "name": "Paka\u00e1snovos",
56
  "iso_1_code": null,
57
  "iso_3_code": "pav",
58
+ "tokenizers": {},
59
+ "children": [],
60
+ "node_i": "3371",
61
+ "scripts": [],
62
+ "own_tokenizer": false
63
  }
64
+ ],
65
+ "node_i": "3369",
66
+ "scripts": [],
67
+ "own_tokenizer": false
68
  }
69
+ ],
70
+ "node_i": "3365",
71
+ "scripts": [],
72
+ "own_tokenizer": false
73
  }
data/Chibchan.json CHANGED
@@ -2,302 +2,392 @@
2
  "name": "Chibchan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Chibchan A",
10
  "iso_1_code": null,
11
  "iso_3_code": null,
12
- "tokenizer": null,
13
- "source": null,
14
  "children": [
15
  {
16
  "name": "Boruca",
17
  "iso_1_code": null,
18
  "iso_3_code": "brn",
19
- "tokenizer": null,
20
- "source": null,
21
- "children": []
 
 
22
  },
23
  {
24
  "name": "Teribe",
25
  "iso_1_code": null,
26
  "iso_3_code": "tfr",
27
- "tokenizer": null,
28
- "source": null,
29
- "children": []
 
 
 
 
30
  },
31
  {
32
  "name": "Guaymi\u00edc",
33
  "iso_1_code": null,
34
  "iso_3_code": null,
35
- "tokenizer": null,
36
- "source": null,
37
  "children": [
38
  {
39
  "name": "Ng\u00e4bere",
40
  "iso_1_code": null,
41
  "iso_3_code": "gym",
42
- "tokenizer": null,
43
- "source": null,
44
- "children": []
 
 
 
 
45
  },
46
  {
47
  "name": "Buglere",
48
  "iso_1_code": null,
49
  "iso_3_code": "sab",
50
- "tokenizer": null,
51
- "source": null,
52
- "children": []
 
 
 
 
53
  }
54
- ]
 
 
 
55
  },
56
  {
57
  "name": "Viceitic",
58
  "iso_1_code": null,
59
  "iso_3_code": null,
60
- "tokenizer": null,
61
- "source": null,
62
  "children": [
63
  {
64
  "name": "Bribri",
65
  "iso_1_code": null,
66
  "iso_3_code": "bzd",
67
- "tokenizer": null,
68
- "source": null,
69
- "children": []
 
 
 
 
70
  },
71
  {
72
  "name": "Cab\u00e9car",
73
  "iso_1_code": null,
74
  "iso_3_code": "cjp",
75
- "tokenizer": null,
76
- "source": null,
77
- "children": []
 
 
 
 
78
  }
79
- ]
 
 
 
80
  }
81
- ]
 
 
 
82
  },
83
  {
84
  "name": "Chibchan B",
85
  "iso_1_code": null,
86
  "iso_3_code": null,
87
- "tokenizer": null,
88
- "source": null,
89
  "children": [
90
  {
91
  "name": "Pech",
92
  "iso_1_code": null,
93
  "iso_3_code": "pay",
94
- "tokenizer": null,
95
- "source": null,
96
- "children": []
 
 
97
  },
98
  {
99
  "name": "Eastern Chibchan",
100
  "iso_1_code": null,
101
  "iso_3_code": null,
102
- "tokenizer": null,
103
- "source": null,
104
  "children": [
105
  {
106
  "name": "Colombian",
107
  "iso_1_code": null,
108
  "iso_3_code": null,
109
- "tokenizer": null,
110
- "source": null,
111
  "children": [
112
  {
113
  "name": "Northern Colombian",
114
  "iso_1_code": null,
115
  "iso_3_code": null,
116
- "tokenizer": null,
117
- "source": null,
118
  "children": [
119
  {
120
  "name": "Chimila",
121
  "iso_1_code": null,
122
  "iso_3_code": "cbg",
123
- "tokenizer": null,
124
- "source": null,
125
- "children": []
 
 
126
  },
127
  {
128
  "name": "Arhuacan",
129
  "iso_1_code": null,
130
  "iso_3_code": null,
131
- "tokenizer": null,
132
- "source": null,
133
  "children": [
134
  {
135
  "name": "Kogi",
136
  "iso_1_code": null,
137
  "iso_3_code": "kog",
138
- "tokenizer": null,
139
- "source": null,
140
- "children": []
 
 
 
 
141
  },
142
  {
143
  "name": "Southern and Eastern Arhuacan",
144
  "iso_1_code": null,
145
  "iso_3_code": null,
146
- "tokenizer": null,
147
- "source": null,
148
  "children": [
149
  {
150
  "name": "Arhuaco",
151
  "iso_1_code": null,
152
  "iso_3_code": "arh",
153
- "tokenizer": null,
154
- "source": null,
155
- "children": []
 
 
156
  },
157
  {
158
  "name": "Guamaca-Atanque",
159
  "iso_1_code": null,
160
  "iso_3_code": null,
161
- "tokenizer": null,
162
- "source": null,
163
  "children": [
164
  {
165
  "name": "Sanka",
166
  "iso_1_code": null,
167
  "iso_3_code": "mbp",
168
- "tokenizer": null,
169
- "source": null,
170
- "children": []
 
 
171
  }
172
- ]
 
 
 
173
  }
174
- ]
 
 
 
175
  }
176
- ]
 
 
 
177
  }
178
- ]
 
 
 
179
  },
180
  {
181
  "name": "Southern Colombian",
182
  "iso_1_code": null,
183
  "iso_3_code": null,
184
- "tokenizer": null,
185
- "source": null,
186
  "children": [
187
  {
188
  "name": "Bar\u00ed",
189
  "iso_1_code": null,
190
  "iso_3_code": "mot",
191
- "tokenizer": null,
192
- "source": null,
193
- "children": []
 
 
194
  },
195
  {
196
  "name": "Cundicocuyese",
197
  "iso_1_code": null,
198
  "iso_3_code": null,
199
- "tokenizer": null,
200
- "source": null,
201
  "children": [
202
  {
203
  "name": "Chibcha",
204
  "iso_1_code": null,
205
  "iso_3_code": "chb",
206
- "tokenizer": null,
207
- "source": null,
208
- "children": []
 
 
209
  },
210
  {
211
  "name": "Tunebo, Barro Negro",
212
  "iso_1_code": null,
213
  "iso_3_code": "tbn",
214
- "tokenizer": null,
215
- "source": null,
216
- "children": []
 
 
217
  },
218
  {
219
  "name": "Tunebo, Western",
220
  "iso_1_code": null,
221
  "iso_3_code": "tnb",
222
- "tokenizer": null,
223
- "source": null,
224
- "children": []
 
 
225
  },
226
  {
227
  "name": "Tunebo, Angosturas",
228
  "iso_1_code": null,
229
  "iso_3_code": "tnd",
230
- "tokenizer": null,
231
- "source": null,
232
- "children": []
 
 
233
  },
234
  {
235
  "name": "Tunebo, Central",
236
  "iso_1_code": null,
237
  "iso_3_code": "tuf",
238
- "tokenizer": null,
239
- "source": null,
240
- "children": []
 
 
 
 
241
  }
242
- ]
 
 
 
243
  }
244
- ]
 
 
 
245
  }
246
- ]
 
 
 
247
  },
248
  {
249
  "name": "Cuna",
250
  "iso_1_code": null,
251
  "iso_3_code": null,
252
- "tokenizer": null,
253
- "source": null,
254
  "children": [
255
  {
256
  "name": "Kuna, San Blas",
257
  "iso_1_code": null,
258
  "iso_3_code": "cuk",
259
- "tokenizer": null,
260
- "source": null,
261
- "children": []
 
 
 
 
262
  },
263
  {
264
  "name": "Kuna, Border",
265
  "iso_1_code": null,
266
  "iso_3_code": "kvn",
267
- "tokenizer": null,
268
- "source": null,
269
- "children": []
 
 
 
 
270
  }
271
- ]
 
 
 
272
  }
273
- ]
 
 
 
274
  },
275
  {
276
  "name": "Votic",
277
  "iso_1_code": null,
278
  "iso_3_code": null,
279
- "tokenizer": null,
280
- "source": null,
281
  "children": [
282
  {
283
  "name": "Mal\u00e9ku Ja\u00edka",
284
  "iso_1_code": null,
285
  "iso_3_code": "gut",
286
- "tokenizer": null,
287
- "source": null,
288
- "children": []
 
 
289
  },
290
  {
291
  "name": "Rama",
292
  "iso_1_code": null,
293
  "iso_3_code": "rma",
294
- "tokenizer": null,
295
- "source": null,
296
- "children": []
 
 
297
  }
298
- ]
 
 
 
299
  }
300
- ]
 
 
 
301
  }
302
- ]
 
 
 
303
  }
 
2
  "name": "Chibchan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Chibchan A",
9
  "iso_1_code": null,
10
  "iso_3_code": null,
11
+ "tokenizers": {},
 
12
  "children": [
13
  {
14
  "name": "Boruca",
15
  "iso_1_code": null,
16
  "iso_3_code": "brn",
17
+ "tokenizers": {},
18
+ "children": [],
19
+ "node_i": "3374",
20
+ "scripts": [],
21
+ "own_tokenizer": false
22
  },
23
  {
24
  "name": "Teribe",
25
  "iso_1_code": null,
26
  "iso_3_code": "tfr",
27
+ "tokenizers": {},
28
+ "children": [],
29
+ "node_i": "3375",
30
+ "scripts": [
31
+ "Latn"
32
+ ],
33
+ "own_tokenizer": false
34
  },
35
  {
36
  "name": "Guaymi\u00edc",
37
  "iso_1_code": null,
38
  "iso_3_code": null,
39
+ "tokenizers": {},
 
40
  "children": [
41
  {
42
  "name": "Ng\u00e4bere",
43
  "iso_1_code": null,
44
  "iso_3_code": "gym",
45
+ "tokenizers": {},
46
+ "children": [],
47
+ "node_i": "3377",
48
+ "scripts": [
49
+ "Latn"
50
+ ],
51
+ "own_tokenizer": false
52
  },
53
  {
54
  "name": "Buglere",
55
  "iso_1_code": null,
56
  "iso_3_code": "sab",
57
+ "tokenizers": {},
58
+ "children": [],
59
+ "node_i": "3378",
60
+ "scripts": [
61
+ "Latn"
62
+ ],
63
+ "own_tokenizer": false
64
  }
65
+ ],
66
+ "node_i": "3376",
67
+ "scripts": [],
68
+ "own_tokenizer": false
69
  },
70
  {
71
  "name": "Viceitic",
72
  "iso_1_code": null,
73
  "iso_3_code": null,
74
+ "tokenizers": {},
 
75
  "children": [
76
  {
77
  "name": "Bribri",
78
  "iso_1_code": null,
79
  "iso_3_code": "bzd",
80
+ "tokenizers": {},
81
+ "children": [],
82
+ "node_i": "3380",
83
+ "scripts": [
84
+ "Latn"
85
+ ],
86
+ "own_tokenizer": false
87
  },
88
  {
89
  "name": "Cab\u00e9car",
90
  "iso_1_code": null,
91
  "iso_3_code": "cjp",
92
+ "tokenizers": {},
93
+ "children": [],
94
+ "node_i": "3381",
95
+ "scripts": [
96
+ "Latn"
97
+ ],
98
+ "own_tokenizer": false
99
  }
100
+ ],
101
+ "node_i": "3379",
102
+ "scripts": [],
103
+ "own_tokenizer": false
104
  }
105
+ ],
106
+ "node_i": "3373",
107
+ "scripts": [],
108
+ "own_tokenizer": false
109
  },
110
  {
111
  "name": "Chibchan B",
112
  "iso_1_code": null,
113
  "iso_3_code": null,
114
+ "tokenizers": {},
 
115
  "children": [
116
  {
117
  "name": "Pech",
118
  "iso_1_code": null,
119
  "iso_3_code": "pay",
120
+ "tokenizers": {},
121
+ "children": [],
122
+ "node_i": "3383",
123
+ "scripts": [],
124
+ "own_tokenizer": false
125
  },
126
  {
127
  "name": "Eastern Chibchan",
128
  "iso_1_code": null,
129
  "iso_3_code": null,
130
+ "tokenizers": {},
 
131
  "children": [
132
  {
133
  "name": "Colombian",
134
  "iso_1_code": null,
135
  "iso_3_code": null,
136
+ "tokenizers": {},
 
137
  "children": [
138
  {
139
  "name": "Northern Colombian",
140
  "iso_1_code": null,
141
  "iso_3_code": null,
142
+ "tokenizers": {},
 
143
  "children": [
144
  {
145
  "name": "Chimila",
146
  "iso_1_code": null,
147
  "iso_3_code": "cbg",
148
+ "tokenizers": {},
149
+ "children": [],
150
+ "node_i": "3387",
151
+ "scripts": [],
152
+ "own_tokenizer": false
153
  },
154
  {
155
  "name": "Arhuacan",
156
  "iso_1_code": null,
157
  "iso_3_code": null,
158
+ "tokenizers": {},
 
159
  "children": [
160
  {
161
  "name": "Kogi",
162
  "iso_1_code": null,
163
  "iso_3_code": "kog",
164
+ "tokenizers": {},
165
+ "children": [],
166
+ "node_i": "3389",
167
+ "scripts": [
168
+ "Latn"
169
+ ],
170
+ "own_tokenizer": false
171
  },
172
  {
173
  "name": "Southern and Eastern Arhuacan",
174
  "iso_1_code": null,
175
  "iso_3_code": null,
176
+ "tokenizers": {},
 
177
  "children": [
178
  {
179
  "name": "Arhuaco",
180
  "iso_1_code": null,
181
  "iso_3_code": "arh",
182
+ "tokenizers": {},
183
+ "children": [],
184
+ "node_i": "3391",
185
+ "scripts": [],
186
+ "own_tokenizer": false
187
  },
188
  {
189
  "name": "Guamaca-Atanque",
190
  "iso_1_code": null,
191
  "iso_3_code": null,
192
+ "tokenizers": {},
 
193
  "children": [
194
  {
195
  "name": "Sanka",
196
  "iso_1_code": null,
197
  "iso_3_code": "mbp",
198
+ "tokenizers": {},
199
+ "children": [],
200
+ "node_i": "3393",
201
+ "scripts": [],
202
+ "own_tokenizer": false
203
  }
204
+ ],
205
+ "node_i": "3392",
206
+ "scripts": [],
207
+ "own_tokenizer": false
208
  }
209
+ ],
210
+ "node_i": "3390",
211
+ "scripts": [],
212
+ "own_tokenizer": false
213
  }
214
+ ],
215
+ "node_i": "3388",
216
+ "scripts": [],
217
+ "own_tokenizer": false
218
  }
219
+ ],
220
+ "node_i": "3386",
221
+ "scripts": [],
222
+ "own_tokenizer": false
223
  },
224
  {
225
  "name": "Southern Colombian",
226
  "iso_1_code": null,
227
  "iso_3_code": null,
228
+ "tokenizers": {},
 
229
  "children": [
230
  {
231
  "name": "Bar\u00ed",
232
  "iso_1_code": null,
233
  "iso_3_code": "mot",
234
+ "tokenizers": {},
235
+ "children": [],
236
+ "node_i": "3395",
237
+ "scripts": [],
238
+ "own_tokenizer": false
239
  },
240
  {
241
  "name": "Cundicocuyese",
242
  "iso_1_code": null,
243
  "iso_3_code": null,
244
+ "tokenizers": {},
 
245
  "children": [
246
  {
247
  "name": "Chibcha",
248
  "iso_1_code": null,
249
  "iso_3_code": "chb",
250
+ "tokenizers": {},
251
+ "children": [],
252
+ "node_i": "3397",
253
+ "scripts": [],
254
+ "own_tokenizer": false
255
  },
256
  {
257
  "name": "Tunebo, Barro Negro",
258
  "iso_1_code": null,
259
  "iso_3_code": "tbn",
260
+ "tokenizers": {},
261
+ "children": [],
262
+ "node_i": "3398",
263
+ "scripts": [],
264
+ "own_tokenizer": false
265
  },
266
  {
267
  "name": "Tunebo, Western",
268
  "iso_1_code": null,
269
  "iso_3_code": "tnb",
270
+ "tokenizers": {},
271
+ "children": [],
272
+ "node_i": "3399",
273
+ "scripts": [],
274
+ "own_tokenizer": false
275
  },
276
  {
277
  "name": "Tunebo, Angosturas",
278
  "iso_1_code": null,
279
  "iso_3_code": "tnd",
280
+ "tokenizers": {},
281
+ "children": [],
282
+ "node_i": "3400",
283
+ "scripts": [],
284
+ "own_tokenizer": false
285
  },
286
  {
287
  "name": "Tunebo, Central",
288
  "iso_1_code": null,
289
  "iso_3_code": "tuf",
290
+ "tokenizers": {},
291
+ "children": [],
292
+ "node_i": "3401",
293
+ "scripts": [
294
+ "Latn"
295
+ ],
296
+ "own_tokenizer": false
297
  }
298
+ ],
299
+ "node_i": "3396",
300
+ "scripts": [],
301
+ "own_tokenizer": false
302
  }
303
+ ],
304
+ "node_i": "3394",
305
+ "scripts": [],
306
+ "own_tokenizer": false
307
  }
308
+ ],
309
+ "node_i": "3385",
310
+ "scripts": [],
311
+ "own_tokenizer": false
312
  },
313
  {
314
  "name": "Cuna",
315
  "iso_1_code": null,
316
  "iso_3_code": null,
317
+ "tokenizers": {},
 
318
  "children": [
319
  {
320
  "name": "Kuna, San Blas",
321
  "iso_1_code": null,
322
  "iso_3_code": "cuk",
323
+ "tokenizers": {},
324
+ "children": [],
325
+ "node_i": "3403",
326
+ "scripts": [
327
+ "Latn"
328
+ ],
329
+ "own_tokenizer": false
330
  },
331
  {
332
  "name": "Kuna, Border",
333
  "iso_1_code": null,
334
  "iso_3_code": "kvn",
335
+ "tokenizers": {},
336
+ "children": [],
337
+ "node_i": "3404",
338
+ "scripts": [
339
+ "Latn"
340
+ ],
341
+ "own_tokenizer": false
342
  }
343
+ ],
344
+ "node_i": "3402",
345
+ "scripts": [],
346
+ "own_tokenizer": false
347
  }
348
+ ],
349
+ "node_i": "3384",
350
+ "scripts": [],
351
+ "own_tokenizer": false
352
  },
353
  {
354
  "name": "Votic",
355
  "iso_1_code": null,
356
  "iso_3_code": null,
357
+ "tokenizers": {},
 
358
  "children": [
359
  {
360
  "name": "Mal\u00e9ku Ja\u00edka",
361
  "iso_1_code": null,
362
  "iso_3_code": "gut",
363
+ "tokenizers": {},
364
+ "children": [],
365
+ "node_i": "3406",
366
+ "scripts": [],
367
+ "own_tokenizer": false
368
  },
369
  {
370
  "name": "Rama",
371
  "iso_1_code": null,
372
  "iso_3_code": "rma",
373
+ "tokenizers": {},
374
+ "children": [],
375
+ "node_i": "3407",
376
+ "scripts": [],
377
+ "own_tokenizer": false
378
  }
379
+ ],
380
+ "node_i": "3405",
381
+ "scripts": [],
382
+ "own_tokenizer": false
383
  }
384
+ ],
385
+ "node_i": "3382",
386
+ "scripts": [],
387
+ "own_tokenizer": false
388
  }
389
+ ],
390
+ "node_i": "3372",
391
+ "scripts": [],
392
+ "own_tokenizer": false
393
  }
data/Chimakuan.json CHANGED
@@ -2,24 +2,30 @@
2
  "name": "Chimakuan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Quileute",
10
  "iso_1_code": null,
11
  "iso_3_code": "qui",
12
- "tokenizer": null,
13
- "source": null,
14
- "children": []
 
 
15
  },
16
  {
17
  "name": "Chemakum",
18
  "iso_1_code": null,
19
  "iso_3_code": "xch",
20
- "tokenizer": null,
21
- "source": null,
22
- "children": []
 
 
23
  }
24
- ]
 
 
 
25
  }
 
2
  "name": "Chimakuan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Quileute",
9
  "iso_1_code": null,
10
  "iso_3_code": "qui",
11
+ "tokenizers": {},
12
+ "children": [],
13
+ "node_i": "3409",
14
+ "scripts": [],
15
+ "own_tokenizer": false
16
  },
17
  {
18
  "name": "Chemakum",
19
  "iso_1_code": null,
20
  "iso_3_code": "xch",
21
+ "tokenizers": {},
22
+ "children": [],
23
+ "node_i": "3410",
24
+ "scripts": [],
25
+ "own_tokenizer": false
26
  }
27
+ ],
28
+ "node_i": "3408",
29
+ "scripts": [],
30
+ "own_tokenizer": false
31
  }
data/Chinookan.json CHANGED
@@ -2,33 +2,41 @@
2
  "name": "Chinookan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Chinook",
10
  "iso_1_code": null,
11
  "iso_3_code": "chh",
12
- "tokenizer": null,
13
- "source": null,
14
- "children": []
 
 
15
  },
16
  {
17
  "name": "Upper Chinookan",
18
  "iso_1_code": null,
19
  "iso_3_code": null,
20
- "tokenizer": null,
21
- "source": null,
22
  "children": [
23
  {
24
  "name": "Wasco-Wishram",
25
  "iso_1_code": null,
26
  "iso_3_code": "wac",
27
- "tokenizer": null,
28
- "source": null,
29
- "children": []
 
 
30
  }
31
- ]
 
 
 
32
  }
33
- ]
 
 
 
34
  }
 
2
  "name": "Chinookan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Chinook",
9
  "iso_1_code": null,
10
  "iso_3_code": "chh",
11
+ "tokenizers": {},
12
+ "children": [],
13
+ "node_i": "3412",
14
+ "scripts": [],
15
+ "own_tokenizer": false
16
  },
17
  {
18
  "name": "Upper Chinookan",
19
  "iso_1_code": null,
20
  "iso_3_code": null,
21
+ "tokenizers": {},
 
22
  "children": [
23
  {
24
  "name": "Wasco-Wishram",
25
  "iso_1_code": null,
26
  "iso_3_code": "wac",
27
+ "tokenizers": {},
28
+ "children": [],
29
+ "node_i": "3414",
30
+ "scripts": [],
31
+ "own_tokenizer": false
32
  }
33
+ ],
34
+ "node_i": "3413",
35
+ "scripts": [],
36
+ "own_tokenizer": false
37
  }
38
+ ],
39
+ "node_i": "3411",
40
+ "scripts": [],
41
+ "own_tokenizer": false
42
  }
data/Chipaya-Uru.json CHANGED
@@ -2,24 +2,32 @@
2
  "name": "Chipaya-Uru",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Chipaya",
10
  "iso_1_code": null,
11
  "iso_3_code": "cap",
12
- "tokenizer": null,
13
- "source": null,
14
- "children": []
 
 
 
 
15
  },
16
  {
17
  "name": "Uru",
18
  "iso_1_code": null,
19
  "iso_3_code": "ure",
20
- "tokenizer": null,
21
- "source": null,
22
- "children": []
 
 
23
  }
24
- ]
 
 
 
25
  }
 
2
  "name": "Chipaya-Uru",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Chipaya",
9
  "iso_1_code": null,
10
  "iso_3_code": "cap",
11
+ "tokenizers": {},
12
+ "children": [],
13
+ "node_i": "3416",
14
+ "scripts": [
15
+ "Latn"
16
+ ],
17
+ "own_tokenizer": false
18
  },
19
  {
20
  "name": "Uru",
21
  "iso_1_code": null,
22
  "iso_3_code": "ure",
23
+ "tokenizers": {},
24
+ "children": [],
25
+ "node_i": "3417",
26
+ "scripts": [],
27
+ "own_tokenizer": false
28
  }
29
+ ],
30
+ "node_i": "3415",
31
+ "scripts": [],
32
+ "own_tokenizer": false
33
  }
data/Chocoan.json CHANGED
@@ -2,91 +2,121 @@
2
  "name": "Chocoan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Woun Meu",
10
  "iso_1_code": null,
11
  "iso_3_code": "noa",
12
- "tokenizer": null,
13
- "source": null,
14
- "children": []
 
 
 
 
15
  },
16
  {
17
  "name": "Ember\u00e1",
18
  "iso_1_code": null,
19
  "iso_3_code": null,
20
- "tokenizer": null,
21
- "source": null,
22
  "children": [
23
  {
24
  "name": "Northern Ember\u00e1",
25
  "iso_1_code": null,
26
  "iso_3_code": null,
27
- "tokenizer": null,
28
- "source": null,
29
  "children": [
30
  {
31
  "name": "Embera Cat\u00edo",
32
  "iso_1_code": null,
33
  "iso_3_code": "cto",
34
- "tokenizer": null,
35
- "source": null,
36
- "children": []
 
 
 
 
37
  },
38
  {
39
  "name": "Ember\u00e1, Northern",
40
  "iso_1_code": null,
41
  "iso_3_code": "emp",
42
- "tokenizer": null,
43
- "source": null,
44
- "children": []
 
 
 
 
45
  }
46
- ]
 
 
 
47
  },
48
  {
49
  "name": "Southern Ember\u00e1",
50
  "iso_1_code": null,
51
  "iso_3_code": null,
52
- "tokenizer": null,
53
- "source": null,
54
  "children": [
55
  {
56
  "name": "Embera Baud\u00f3",
57
  "iso_1_code": null,
58
  "iso_3_code": "bdc",
59
- "tokenizer": null,
60
- "source": null,
61
- "children": []
 
 
62
  },
63
  {
64
  "name": "Embera Cham\u00ed",
65
  "iso_1_code": null,
66
  "iso_3_code": "cmi",
67
- "tokenizer": null,
68
- "source": null,
69
- "children": []
 
 
70
  },
71
  {
72
  "name": "Epena",
73
  "iso_1_code": null,
74
  "iso_3_code": "sja",
75
- "tokenizer": null,
76
- "source": null,
77
- "children": []
 
 
 
 
78
  },
79
  {
80
  "name": "Embera Tad\u00f3",
81
  "iso_1_code": null,
82
  "iso_3_code": "tdc",
83
- "tokenizer": null,
84
- "source": null,
85
- "children": []
 
 
86
  }
87
- ]
 
 
 
88
  }
89
- ]
 
 
 
90
  }
91
- ]
 
 
 
92
  }
 
2
  "name": "Chocoan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Woun Meu",
9
  "iso_1_code": null,
10
  "iso_3_code": "noa",
11
+ "tokenizers": {},
12
+ "children": [],
13
+ "node_i": "3419",
14
+ "scripts": [
15
+ "Latn"
16
+ ],
17
+ "own_tokenizer": false
18
  },
19
  {
20
  "name": "Ember\u00e1",
21
  "iso_1_code": null,
22
  "iso_3_code": null,
23
+ "tokenizers": {},
 
24
  "children": [
25
  {
26
  "name": "Northern Ember\u00e1",
27
  "iso_1_code": null,
28
  "iso_3_code": null,
29
+ "tokenizers": {},
 
30
  "children": [
31
  {
32
  "name": "Embera Cat\u00edo",
33
  "iso_1_code": null,
34
  "iso_3_code": "cto",
35
+ "tokenizers": {},
36
+ "children": [],
37
+ "node_i": "3422",
38
+ "scripts": [
39
+ "Latn"
40
+ ],
41
+ "own_tokenizer": false
42
  },
43
  {
44
  "name": "Ember\u00e1, Northern",
45
  "iso_1_code": null,
46
  "iso_3_code": "emp",
47
+ "tokenizers": {},
48
+ "children": [],
49
+ "node_i": "3423",
50
+ "scripts": [
51
+ "Latn"
52
+ ],
53
+ "own_tokenizer": false
54
  }
55
+ ],
56
+ "node_i": "3421",
57
+ "scripts": [],
58
+ "own_tokenizer": false
59
  },
60
  {
61
  "name": "Southern Ember\u00e1",
62
  "iso_1_code": null,
63
  "iso_3_code": null,
64
+ "tokenizers": {},
 
65
  "children": [
66
  {
67
  "name": "Embera Baud\u00f3",
68
  "iso_1_code": null,
69
  "iso_3_code": "bdc",
70
+ "tokenizers": {},
71
+ "children": [],
72
+ "node_i": "3425",
73
+ "scripts": [],
74
+ "own_tokenizer": false
75
  },
76
  {
77
  "name": "Embera Cham\u00ed",
78
  "iso_1_code": null,
79
  "iso_3_code": "cmi",
80
+ "tokenizers": {},
81
+ "children": [],
82
+ "node_i": "3426",
83
+ "scripts": [],
84
+ "own_tokenizer": false
85
  },
86
  {
87
  "name": "Epena",
88
  "iso_1_code": null,
89
  "iso_3_code": "sja",
90
+ "tokenizers": {},
91
+ "children": [],
92
+ "node_i": "3427",
93
+ "scripts": [
94
+ "Latn"
95
+ ],
96
+ "own_tokenizer": false
97
  },
98
  {
99
  "name": "Embera Tad\u00f3",
100
  "iso_1_code": null,
101
  "iso_3_code": "tdc",
102
+ "tokenizers": {},
103
+ "children": [],
104
+ "node_i": "3428",
105
+ "scripts": [],
106
+ "own_tokenizer": false
107
  }
108
+ ],
109
+ "node_i": "3424",
110
+ "scripts": [],
111
+ "own_tokenizer": false
112
  }
113
+ ],
114
+ "node_i": "3420",
115
+ "scripts": [],
116
+ "own_tokenizer": false
117
  }
118
+ ],
119
+ "node_i": "3418",
120
+ "scripts": [],
121
+ "own_tokenizer": false
122
  }
data/Cholonan.json CHANGED
@@ -2,24 +2,30 @@
2
  "name": "Cholonan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Chol\u00f3n",
10
  "iso_1_code": null,
11
  "iso_3_code": "cht",
12
- "tokenizer": null,
13
- "source": null,
14
- "children": []
 
 
15
  },
16
  {
17
  "name": "Hibito",
18
  "iso_1_code": null,
19
  "iso_3_code": "hib",
20
- "tokenizer": null,
21
- "source": null,
22
- "children": []
 
 
23
  }
24
- ]
 
 
 
25
  }
 
2
  "name": "Cholonan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Chol\u00f3n",
9
  "iso_1_code": null,
10
  "iso_3_code": "cht",
11
+ "tokenizers": {},
12
+ "children": [],
13
+ "node_i": "3430",
14
+ "scripts": [],
15
+ "own_tokenizer": false
16
  },
17
  {
18
  "name": "Hibito",
19
  "iso_1_code": null,
20
  "iso_3_code": "hib",
21
+ "tokenizers": {},
22
+ "children": [],
23
+ "node_i": "3431",
24
+ "scripts": [],
25
+ "own_tokenizer": false
26
  }
27
+ ],
28
+ "node_i": "3429",
29
+ "scripts": [],
30
+ "own_tokenizer": false
31
  }
data/Chon.json CHANGED
@@ -2,33 +2,41 @@
2
  "name": "Chon",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Tehuelche",
10
  "iso_1_code": null,
11
  "iso_3_code": "teh",
12
- "tokenizer": null,
13
- "source": null,
14
- "children": []
 
 
15
  },
16
  {
17
  "name": "Island Chon",
18
  "iso_1_code": null,
19
  "iso_3_code": null,
20
- "tokenizer": null,
21
- "source": null,
22
  "children": [
23
  {
24
  "name": "Ona",
25
  "iso_1_code": null,
26
  "iso_3_code": "ona",
27
- "tokenizer": null,
28
- "source": null,
29
- "children": []
 
 
30
  }
31
- ]
 
 
 
32
  }
33
- ]
 
 
 
34
  }
 
2
  "name": "Chon",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Tehuelche",
9
  "iso_1_code": null,
10
  "iso_3_code": "teh",
11
+ "tokenizers": {},
12
+ "children": [],
13
+ "node_i": "3433",
14
+ "scripts": [],
15
+ "own_tokenizer": false
16
  },
17
  {
18
  "name": "Island Chon",
19
  "iso_1_code": null,
20
  "iso_3_code": null,
21
+ "tokenizers": {},
 
22
  "children": [
23
  {
24
  "name": "Ona",
25
  "iso_1_code": null,
26
  "iso_3_code": "ona",
27
+ "tokenizers": {},
28
+ "children": [],
29
+ "node_i": "3435",
30
+ "scripts": [],
31
+ "own_tokenizer": false
32
  }
33
+ ],
34
+ "node_i": "3434",
35
+ "scripts": [],
36
+ "own_tokenizer": false
37
  }
38
+ ],
39
+ "node_i": "3432",
40
+ "scripts": [],
41
+ "own_tokenizer": false
42
  }
data/Chukotko-Kamchatkan.json CHANGED
@@ -2,84 +2,108 @@
2
  "name": "Chukotko-Kamchatkan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Northern",
10
  "iso_1_code": null,
11
  "iso_3_code": null,
12
- "tokenizer": null,
13
- "source": null,
14
  "children": [
15
  {
16
  "name": "Chukot",
17
  "iso_1_code": null,
18
  "iso_3_code": null,
19
- "tokenizer": null,
20
- "source": null,
21
  "children": [
22
  {
23
  "name": "Chukchi",
24
  "iso_1_code": null,
25
  "iso_3_code": "ckt",
26
- "tokenizer": null,
27
- "source": null,
28
- "children": []
 
 
 
 
29
  }
30
- ]
 
 
 
31
  },
32
  {
33
  "name": "Koryak-Alyutor",
34
  "iso_1_code": null,
35
  "iso_3_code": null,
36
- "tokenizer": null,
37
- "source": null,
38
  "children": [
39
  {
40
  "name": "Alutor",
41
  "iso_1_code": null,
42
  "iso_3_code": "alr",
43
- "tokenizer": null,
44
- "source": null,
45
- "children": []
 
 
46
  },
47
  {
48
  "name": "Koryak",
49
  "iso_1_code": null,
50
  "iso_3_code": "kpy",
51
- "tokenizer": null,
52
- "source": null,
53
- "children": []
 
 
54
  },
55
  {
56
  "name": "Kerek",
57
  "iso_1_code": null,
58
  "iso_3_code": "krk",
59
- "tokenizer": null,
60
- "source": null,
61
- "children": []
 
 
62
  }
63
- ]
 
 
 
64
  }
65
- ]
 
 
 
66
  },
67
  {
68
  "name": "Southern",
69
  "iso_1_code": null,
70
  "iso_3_code": null,
71
- "tokenizer": null,
72
- "source": null,
73
  "children": [
74
  {
75
  "name": "Itelmen",
76
  "iso_1_code": null,
77
  "iso_3_code": "itl",
78
- "tokenizer": null,
79
- "source": null,
80
- "children": []
 
 
 
 
81
  }
82
- ]
 
 
 
83
  }
84
- ]
 
 
 
85
  }
 
2
  "name": "Chukotko-Kamchatkan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Northern",
9
  "iso_1_code": null,
10
  "iso_3_code": null,
11
+ "tokenizers": {},
 
12
  "children": [
13
  {
14
  "name": "Chukot",
15
  "iso_1_code": null,
16
  "iso_3_code": null,
17
+ "tokenizers": {},
 
18
  "children": [
19
  {
20
  "name": "Chukchi",
21
  "iso_1_code": null,
22
  "iso_3_code": "ckt",
23
+ "tokenizers": {},
24
+ "children": [],
25
+ "node_i": "3439",
26
+ "scripts": [
27
+ "Cyrl"
28
+ ],
29
+ "own_tokenizer": false
30
  }
31
+ ],
32
+ "node_i": "3438",
33
+ "scripts": [],
34
+ "own_tokenizer": false
35
  },
36
  {
37
  "name": "Koryak-Alyutor",
38
  "iso_1_code": null,
39
  "iso_3_code": null,
40
+ "tokenizers": {},
 
41
  "children": [
42
  {
43
  "name": "Alutor",
44
  "iso_1_code": null,
45
  "iso_3_code": "alr",
46
+ "tokenizers": {},
47
+ "children": [],
48
+ "node_i": "3441",
49
+ "scripts": [],
50
+ "own_tokenizer": false
51
  },
52
  {
53
  "name": "Koryak",
54
  "iso_1_code": null,
55
  "iso_3_code": "kpy",
56
+ "tokenizers": {},
57
+ "children": [],
58
+ "node_i": "3442",
59
+ "scripts": [],
60
+ "own_tokenizer": false
61
  },
62
  {
63
  "name": "Kerek",
64
  "iso_1_code": null,
65
  "iso_3_code": "krk",
66
+ "tokenizers": {},
67
+ "children": [],
68
+ "node_i": "3443",
69
+ "scripts": [],
70
+ "own_tokenizer": false
71
  }
72
+ ],
73
+ "node_i": "3440",
74
+ "scripts": [],
75
+ "own_tokenizer": false
76
  }
77
+ ],
78
+ "node_i": "3437",
79
+ "scripts": [],
80
+ "own_tokenizer": false
81
  },
82
  {
83
  "name": "Southern",
84
  "iso_1_code": null,
85
  "iso_3_code": null,
86
+ "tokenizers": {},
 
87
  "children": [
88
  {
89
  "name": "Itelmen",
90
  "iso_1_code": null,
91
  "iso_3_code": "itl",
92
+ "tokenizers": {},
93
+ "children": [],
94
+ "node_i": "3445",
95
+ "scripts": [
96
+ "Cyrl"
97
+ ],
98
+ "own_tokenizer": false
99
  }
100
+ ],
101
+ "node_i": "3444",
102
+ "scripts": [],
103
+ "own_tokenizer": false
104
  }
105
+ ],
106
+ "node_i": "3436",
107
+ "scripts": [],
108
+ "own_tokenizer": false
109
  }
data/Chumashan.json CHANGED
@@ -2,74 +2,92 @@
2
  "name": "Chumashan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Obispe\u00f1o",
10
  "iso_1_code": null,
11
  "iso_3_code": "obi",
12
- "tokenizer": null,
13
- "source": null,
14
- "children": []
 
 
15
  },
16
  {
17
  "name": "Central Chumash",
18
  "iso_1_code": null,
19
  "iso_3_code": null,
20
- "tokenizer": null,
21
- "source": null,
22
  "children": [
23
  {
24
  "name": "Barbare\u00f1o",
25
  "iso_1_code": null,
26
  "iso_3_code": "boi",
27
- "tokenizer": null,
28
- "source": null,
29
- "children": []
 
 
30
  },
31
  {
32
  "name": "Inese\u00f1o",
33
  "iso_1_code": null,
34
  "iso_3_code": "inz",
35
- "tokenizer": null,
36
- "source": null,
37
- "children": []
 
 
38
  },
39
  {
40
  "name": "Purisime\u00f1o",
41
  "iso_1_code": null,
42
  "iso_3_code": "puy",
43
- "tokenizer": null,
44
- "source": null,
45
- "children": []
 
 
46
  },
47
  {
48
  "name": "Venture\u00f1o",
49
  "iso_1_code": null,
50
  "iso_3_code": "veo",
51
- "tokenizer": null,
52
- "source": null,
53
- "children": []
 
 
54
  }
55
- ]
 
 
 
56
  },
57
  {
58
  "name": "Island Chumash",
59
  "iso_1_code": null,
60
  "iso_3_code": null,
61
- "tokenizer": null,
62
- "source": null,
63
  "children": [
64
  {
65
  "name": "Cruze\u00f1o",
66
  "iso_1_code": null,
67
  "iso_3_code": "crz",
68
- "tokenizer": null,
69
- "source": null,
70
- "children": []
 
 
71
  }
72
- ]
 
 
 
73
  }
74
- ]
 
 
 
75
  }
 
2
  "name": "Chumashan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Obispe\u00f1o",
9
  "iso_1_code": null,
10
  "iso_3_code": "obi",
11
+ "tokenizers": {},
12
+ "children": [],
13
+ "node_i": "3447",
14
+ "scripts": [],
15
+ "own_tokenizer": false
16
  },
17
  {
18
  "name": "Central Chumash",
19
  "iso_1_code": null,
20
  "iso_3_code": null,
21
+ "tokenizers": {},
 
22
  "children": [
23
  {
24
  "name": "Barbare\u00f1o",
25
  "iso_1_code": null,
26
  "iso_3_code": "boi",
27
+ "tokenizers": {},
28
+ "children": [],
29
+ "node_i": "3449",
30
+ "scripts": [],
31
+ "own_tokenizer": false
32
  },
33
  {
34
  "name": "Inese\u00f1o",
35
  "iso_1_code": null,
36
  "iso_3_code": "inz",
37
+ "tokenizers": {},
38
+ "children": [],
39
+ "node_i": "3450",
40
+ "scripts": [],
41
+ "own_tokenizer": false
42
  },
43
  {
44
  "name": "Purisime\u00f1o",
45
  "iso_1_code": null,
46
  "iso_3_code": "puy",
47
+ "tokenizers": {},
48
+ "children": [],
49
+ "node_i": "3451",
50
+ "scripts": [],
51
+ "own_tokenizer": false
52
  },
53
  {
54
  "name": "Venture\u00f1o",
55
  "iso_1_code": null,
56
  "iso_3_code": "veo",
57
+ "tokenizers": {},
58
+ "children": [],
59
+ "node_i": "3452",
60
+ "scripts": [],
61
+ "own_tokenizer": false
62
  }
63
+ ],
64
+ "node_i": "3448",
65
+ "scripts": [],
66
+ "own_tokenizer": false
67
  },
68
  {
69
  "name": "Island Chumash",
70
  "iso_1_code": null,
71
  "iso_3_code": null,
72
+ "tokenizers": {},
 
73
  "children": [
74
  {
75
  "name": "Cruze\u00f1o",
76
  "iso_1_code": null,
77
  "iso_3_code": "crz",
78
+ "tokenizers": {},
79
+ "children": [],
80
+ "node_i": "3454",
81
+ "scripts": [],
82
+ "own_tokenizer": false
83
  }
84
+ ],
85
+ "node_i": "3453",
86
+ "scripts": [],
87
+ "own_tokenizer": false
88
  }
89
+ ],
90
+ "node_i": "3446",
91
+ "scripts": [],
92
+ "own_tokenizer": false
93
  }
data/Cochimí-Yuman.json CHANGED
@@ -2,125 +2,155 @@
2
  "name": "Cochim\u00ed-Yuman",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Yuman",
10
  "iso_1_code": null,
11
  "iso_3_code": null,
12
- "tokenizer": null,
13
- "source": null,
14
  "children": [
15
  {
16
  "name": "Cochimi",
17
  "iso_1_code": null,
18
  "iso_3_code": "coj",
19
- "tokenizer": null,
20
- "source": null,
21
- "children": []
 
 
22
  },
23
  {
24
  "name": "Kiliwa",
25
  "iso_1_code": null,
26
  "iso_3_code": "klb",
27
- "tokenizer": null,
28
- "source": null,
29
- "children": []
 
 
30
  },
31
  {
32
  "name": "Delta-California",
33
  "iso_1_code": null,
34
  "iso_3_code": null,
35
- "tokenizer": null,
36
- "source": null,
37
  "children": [
38
  {
39
  "name": "Cocopa",
40
  "iso_1_code": null,
41
  "iso_3_code": "coc",
42
- "tokenizer": null,
43
- "source": null,
44
- "children": []
 
 
45
  },
46
  {
47
  "name": "Kumiai",
48
  "iso_1_code": null,
49
  "iso_3_code": "dih",
50
- "tokenizer": null,
51
- "source": null,
52
- "children": []
 
 
53
  }
54
- ]
 
 
 
55
  },
56
  {
57
  "name": "Pai",
58
  "iso_1_code": null,
59
  "iso_3_code": null,
60
- "tokenizer": null,
61
- "source": null,
62
  "children": [
63
  {
64
  "name": "Paipai",
65
  "iso_1_code": null,
66
  "iso_3_code": "ppi",
67
- "tokenizer": null,
68
- "source": null,
69
- "children": []
 
 
70
  },
71
  {
72
  "name": "Havasupai-Walapai-Yavapai",
73
  "iso_1_code": null,
74
  "iso_3_code": "yuf",
75
- "tokenizer": null,
76
- "source": null,
77
- "children": []
 
 
78
  }
79
- ]
 
 
 
80
  },
81
  {
82
  "name": "River",
83
  "iso_1_code": null,
84
  "iso_3_code": null,
85
- "tokenizer": null,
86
- "source": null,
87
  "children": [
88
  {
89
  "name": "Mojave",
90
  "iso_1_code": null,
91
  "iso_3_code": null,
92
- "tokenizer": null,
93
- "source": null,
94
  "children": [
95
  {
96
  "name": "Mohave",
97
  "iso_1_code": null,
98
  "iso_3_code": "mov",
99
- "tokenizer": null,
100
- "source": null,
101
- "children": []
 
 
102
  },
103
  {
104
  "name": "Maricopa",
105
  "iso_1_code": null,
106
  "iso_3_code": "mrc",
107
- "tokenizer": null,
108
- "source": null,
109
- "children": []
 
 
110
  },
111
  {
112
  "name": "Quechan",
113
  "iso_1_code": null,
114
  "iso_3_code": "yum",
115
- "tokenizer": null,
116
- "source": null,
117
- "children": []
 
 
118
  }
119
- ]
 
 
 
120
  }
121
- ]
 
 
 
122
  }
123
- ]
 
 
 
124
  }
125
- ]
 
 
 
126
  }
 
2
  "name": "Cochim\u00ed-Yuman",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Yuman",
9
  "iso_1_code": null,
10
  "iso_3_code": null,
11
+ "tokenizers": {},
 
12
  "children": [
13
  {
14
  "name": "Cochimi",
15
  "iso_1_code": null,
16
  "iso_3_code": "coj",
17
+ "tokenizers": {},
18
+ "children": [],
19
+ "node_i": "3457",
20
+ "scripts": [],
21
+ "own_tokenizer": false
22
  },
23
  {
24
  "name": "Kiliwa",
25
  "iso_1_code": null,
26
  "iso_3_code": "klb",
27
+ "tokenizers": {},
28
+ "children": [],
29
+ "node_i": "3458",
30
+ "scripts": [],
31
+ "own_tokenizer": false
32
  },
33
  {
34
  "name": "Delta-California",
35
  "iso_1_code": null,
36
  "iso_3_code": null,
37
+ "tokenizers": {},
 
38
  "children": [
39
  {
40
  "name": "Cocopa",
41
  "iso_1_code": null,
42
  "iso_3_code": "coc",
43
+ "tokenizers": {},
44
+ "children": [],
45
+ "node_i": "3460",
46
+ "scripts": [],
47
+ "own_tokenizer": false
48
  },
49
  {
50
  "name": "Kumiai",
51
  "iso_1_code": null,
52
  "iso_3_code": "dih",
53
+ "tokenizers": {},
54
+ "children": [],
55
+ "node_i": "3461",
56
+ "scripts": [],
57
+ "own_tokenizer": false
58
  }
59
+ ],
60
+ "node_i": "3459",
61
+ "scripts": [],
62
+ "own_tokenizer": false
63
  },
64
  {
65
  "name": "Pai",
66
  "iso_1_code": null,
67
  "iso_3_code": null,
68
+ "tokenizers": {},
 
69
  "children": [
70
  {
71
  "name": "Paipai",
72
  "iso_1_code": null,
73
  "iso_3_code": "ppi",
74
+ "tokenizers": {},
75
+ "children": [],
76
+ "node_i": "3463",
77
+ "scripts": [],
78
+ "own_tokenizer": false
79
  },
80
  {
81
  "name": "Havasupai-Walapai-Yavapai",
82
  "iso_1_code": null,
83
  "iso_3_code": "yuf",
84
+ "tokenizers": {},
85
+ "children": [],
86
+ "node_i": "3464",
87
+ "scripts": [],
88
+ "own_tokenizer": false
89
  }
90
+ ],
91
+ "node_i": "3462",
92
+ "scripts": [],
93
+ "own_tokenizer": false
94
  },
95
  {
96
  "name": "River",
97
  "iso_1_code": null,
98
  "iso_3_code": null,
99
+ "tokenizers": {},
 
100
  "children": [
101
  {
102
  "name": "Mojave",
103
  "iso_1_code": null,
104
  "iso_3_code": null,
105
+ "tokenizers": {},
 
106
  "children": [
107
  {
108
  "name": "Mohave",
109
  "iso_1_code": null,
110
  "iso_3_code": "mov",
111
+ "tokenizers": {},
112
+ "children": [],
113
+ "node_i": "3467",
114
+ "scripts": [],
115
+ "own_tokenizer": false
116
  },
117
  {
118
  "name": "Maricopa",
119
  "iso_1_code": null,
120
  "iso_3_code": "mrc",
121
+ "tokenizers": {},
122
+ "children": [],
123
+ "node_i": "3468",
124
+ "scripts": [],
125
+ "own_tokenizer": false
126
  },
127
  {
128
  "name": "Quechan",
129
  "iso_1_code": null,
130
  "iso_3_code": "yum",
131
+ "tokenizers": {},
132
+ "children": [],
133
+ "node_i": "3469",
134
+ "scripts": [],
135
+ "own_tokenizer": false
136
  }
137
+ ],
138
+ "node_i": "3466",
139
+ "scripts": [],
140
+ "own_tokenizer": false
141
  }
142
+ ],
143
+ "node_i": "3465",
144
+ "scripts": [],
145
+ "own_tokenizer": false
146
  }
147
+ ],
148
+ "node_i": "3456",
149
+ "scripts": [],
150
+ "own_tokenizer": false
151
  }
152
+ ],
153
+ "node_i": "3455",
154
+ "scripts": [],
155
+ "own_tokenizer": false
156
  }
data/Comecrudan.json CHANGED
@@ -2,48 +2,60 @@
2
  "name": "Comecrudan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Mamulique",
10
  "iso_1_code": null,
11
  "iso_3_code": "emm",
12
- "tokenizer": null,
13
- "source": null,
14
- "children": []
 
 
15
  },
16
  {
17
  "name": "Comecrudo",
18
  "iso_1_code": null,
19
  "iso_3_code": "xcm",
20
- "tokenizer": null,
21
- "source": null,
22
- "children": []
 
 
23
  },
24
  {
25
  "name": "Cotoname",
26
  "iso_1_code": null,
27
  "iso_3_code": "xcn",
28
- "tokenizer": null,
29
- "source": null,
30
- "children": []
 
 
31
  },
32
  {
33
  "name": "Coahuilteco",
34
  "iso_1_code": null,
35
  "iso_3_code": "xcw",
36
- "tokenizer": null,
37
- "source": null,
38
- "children": []
 
 
39
  },
40
  {
41
  "name": "Garza",
42
  "iso_1_code": null,
43
  "iso_3_code": "xgr",
44
- "tokenizer": null,
45
- "source": null,
46
- "children": []
 
 
47
  }
48
- ]
 
 
 
49
  }
 
2
  "name": "Comecrudan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Mamulique",
9
  "iso_1_code": null,
10
  "iso_3_code": "emm",
11
+ "tokenizers": {},
12
+ "children": [],
13
+ "node_i": "3471",
14
+ "scripts": [],
15
+ "own_tokenizer": false
16
  },
17
  {
18
  "name": "Comecrudo",
19
  "iso_1_code": null,
20
  "iso_3_code": "xcm",
21
+ "tokenizers": {},
22
+ "children": [],
23
+ "node_i": "3472",
24
+ "scripts": [],
25
+ "own_tokenizer": false
26
  },
27
  {
28
  "name": "Cotoname",
29
  "iso_1_code": null,
30
  "iso_3_code": "xcn",
31
+ "tokenizers": {},
32
+ "children": [],
33
+ "node_i": "3473",
34
+ "scripts": [],
35
+ "own_tokenizer": false
36
  },
37
  {
38
  "name": "Coahuilteco",
39
  "iso_1_code": null,
40
  "iso_3_code": "xcw",
41
+ "tokenizers": {},
42
+ "children": [],
43
+ "node_i": "3474",
44
+ "scripts": [],
45
+ "own_tokenizer": false
46
  },
47
  {
48
  "name": "Garza",
49
  "iso_1_code": null,
50
  "iso_3_code": "xgr",
51
+ "tokenizers": {},
52
+ "children": [],
53
+ "node_i": "3475",
54
+ "scripts": [],
55
+ "own_tokenizer": false
56
  }
57
+ ],
58
+ "node_i": "3470",
59
+ "scripts": [],
60
+ "own_tokenizer": false
61
  }
data/Constructed language.json CHANGED
@@ -2,16 +2,22 @@
2
  "name": "Constructed language",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Esperanto",
10
  "iso_1_code": "eo",
11
  "iso_3_code": "epo",
12
- "tokenizer": null,
13
- "source": null,
14
- "children": []
 
 
 
 
15
  }
16
- ]
 
 
 
17
  }
 
2
  "name": "Constructed language",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Esperanto",
9
  "iso_1_code": "eo",
10
  "iso_3_code": "epo",
11
+ "tokenizers": {},
12
+ "children": [],
13
+ "node_i": "3477",
14
+ "scripts": [
15
+ "Latn"
16
+ ],
17
+ "own_tokenizer": false
18
  }
19
+ ],
20
+ "node_i": "3476",
21
+ "scripts": [],
22
+ "own_tokenizer": false
23
  }
data/Coosan.json CHANGED
@@ -2,24 +2,30 @@
2
  "name": "Coosan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Coos",
10
  "iso_1_code": null,
11
  "iso_3_code": "csz",
12
- "tokenizer": null,
13
- "source": null,
14
- "children": []
 
 
15
  },
16
  {
17
  "name": "Miluk",
18
  "iso_1_code": null,
19
  "iso_3_code": "iml",
20
- "tokenizer": null,
21
- "source": null,
22
- "children": []
 
 
23
  }
24
- ]
 
 
 
25
  }
 
2
  "name": "Coosan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Coos",
9
  "iso_1_code": null,
10
  "iso_3_code": "csz",
11
+ "tokenizers": {},
12
+ "children": [],
13
+ "node_i": "3479",
14
+ "scripts": [],
15
+ "own_tokenizer": false
16
  },
17
  {
18
  "name": "Miluk",
19
  "iso_1_code": null,
20
  "iso_3_code": "iml",
21
+ "tokenizers": {},
22
+ "children": [],
23
+ "node_i": "3480",
24
+ "scripts": [],
25
+ "own_tokenizer": false
26
  }
27
+ ],
28
+ "node_i": "3478",
29
+ "scripts": [],
30
+ "own_tokenizer": false
31
  }
data/Creole.json CHANGED
The diff for this file is too large to render. See raw diff
 
data/Dravidian.json CHANGED
The diff for this file is too large to render. See raw diff
 
data/East Bird’s Head-Sentani.json CHANGED
@@ -2,135 +2,173 @@
2
  "name": "East Bird\u2019s Head-Sentani",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Burmeso",
10
  "iso_1_code": null,
11
  "iso_3_code": null,
12
- "tokenizer": null,
13
- "source": null,
14
  "children": [
15
  {
16
  "name": "Burmeso",
17
  "iso_1_code": null,
18
  "iso_3_code": "bzu",
19
- "tokenizer": null,
20
- "source": null,
21
- "children": []
 
 
22
  }
23
- ]
 
 
 
24
  },
25
  {
26
  "name": "East Bird\u2019s Head",
27
  "iso_1_code": null,
28
  "iso_3_code": null,
29
- "tokenizer": null,
30
- "source": null,
31
  "children": [
32
  {
33
  "name": "Mantion",
34
  "iso_1_code": null,
35
  "iso_3_code": null,
36
- "tokenizer": null,
37
- "source": null,
38
  "children": [
39
  {
40
  "name": "Sougb",
41
  "iso_1_code": null,
42
  "iso_3_code": "mnx",
43
- "tokenizer": null,
44
- "source": null,
45
- "children": []
 
 
 
 
46
  }
47
- ]
 
 
 
48
  },
49
  {
50
  "name": "Meax",
51
  "iso_1_code": null,
52
  "iso_3_code": null,
53
- "tokenizer": null,
54
- "source": null,
55
  "children": [
56
  {
57
  "name": "Meyah",
58
  "iso_1_code": null,
59
  "iso_3_code": "mej",
60
- "tokenizer": null,
61
- "source": null,
62
- "children": []
 
 
 
 
63
  },
64
  {
65
  "name": "Moskona",
66
  "iso_1_code": null,
67
  "iso_3_code": "mtj",
68
- "tokenizer": null,
69
- "source": null,
70
- "children": []
 
 
 
 
71
  }
72
- ]
 
 
 
73
  }
74
- ]
 
 
 
75
  },
76
  {
77
  "name": "Sentani",
78
  "iso_1_code": null,
79
  "iso_3_code": null,
80
- "tokenizer": null,
81
- "source": null,
82
  "children": [
83
  {
84
  "name": "Demta",
85
  "iso_1_code": null,
86
  "iso_3_code": null,
87
- "tokenizer": null,
88
- "source": null,
89
  "children": [
90
  {
91
  "name": "Sowari",
92
  "iso_1_code": null,
93
  "iso_3_code": "dmy",
94
- "tokenizer": null,
95
- "source": null,
96
- "children": []
 
 
97
  }
98
- ]
 
 
 
99
  },
100
  {
101
  "name": "Sentani Proper",
102
  "iso_1_code": null,
103
  "iso_3_code": null,
104
- "tokenizer": null,
105
- "source": null,
106
  "children": [
107
  {
108
  "name": "Nafri",
109
  "iso_1_code": null,
110
  "iso_3_code": "nxx",
111
- "tokenizer": null,
112
- "source": null,
113
- "children": []
 
 
114
  },
115
  {
116
  "name": "Sentani",
117
  "iso_1_code": null,
118
  "iso_3_code": "set",
119
- "tokenizer": null,
120
- "source": null,
121
- "children": []
 
 
122
  },
123
  {
124
  "name": "Tabla",
125
  "iso_1_code": null,
126
  "iso_3_code": "tnm",
127
- "tokenizer": null,
128
- "source": null,
129
- "children": []
 
 
130
  }
131
- ]
 
 
 
132
  }
133
- ]
 
 
 
134
  }
135
- ]
 
 
 
136
  }
 
2
  "name": "East Bird\u2019s Head-Sentani",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Burmeso",
9
  "iso_1_code": null,
10
  "iso_3_code": null,
11
+ "tokenizers": {},
 
12
  "children": [
13
  {
14
  "name": "Burmeso",
15
  "iso_1_code": null,
16
  "iso_3_code": "bzu",
17
+ "tokenizers": {},
18
+ "children": [],
19
+ "node_i": "3716",
20
+ "scripts": [],
21
+ "own_tokenizer": false
22
  }
23
+ ],
24
+ "node_i": "3715",
25
+ "scripts": [],
26
+ "own_tokenizer": false
27
  },
28
  {
29
  "name": "East Bird\u2019s Head",
30
  "iso_1_code": null,
31
  "iso_3_code": null,
32
+ "tokenizers": {},
 
33
  "children": [
34
  {
35
  "name": "Mantion",
36
  "iso_1_code": null,
37
  "iso_3_code": null,
38
+ "tokenizers": {},
 
39
  "children": [
40
  {
41
  "name": "Sougb",
42
  "iso_1_code": null,
43
  "iso_3_code": "mnx",
44
+ "tokenizers": {},
45
+ "children": [],
46
+ "node_i": "3719",
47
+ "scripts": [
48
+ "Latn"
49
+ ],
50
+ "own_tokenizer": false
51
  }
52
+ ],
53
+ "node_i": "3718",
54
+ "scripts": [],
55
+ "own_tokenizer": false
56
  },
57
  {
58
  "name": "Meax",
59
  "iso_1_code": null,
60
  "iso_3_code": null,
61
+ "tokenizers": {},
 
62
  "children": [
63
  {
64
  "name": "Meyah",
65
  "iso_1_code": null,
66
  "iso_3_code": "mej",
67
+ "tokenizers": {},
68
+ "children": [],
69
+ "node_i": "3721",
70
+ "scripts": [
71
+ "Latn"
72
+ ],
73
+ "own_tokenizer": false
74
  },
75
  {
76
  "name": "Moskona",
77
  "iso_1_code": null,
78
  "iso_3_code": "mtj",
79
+ "tokenizers": {},
80
+ "children": [],
81
+ "node_i": "3722",
82
+ "scripts": [
83
+ "Latn"
84
+ ],
85
+ "own_tokenizer": false
86
  }
87
+ ],
88
+ "node_i": "3720",
89
+ "scripts": [],
90
+ "own_tokenizer": false
91
  }
92
+ ],
93
+ "node_i": "3717",
94
+ "scripts": [],
95
+ "own_tokenizer": false
96
  },
97
  {
98
  "name": "Sentani",
99
  "iso_1_code": null,
100
  "iso_3_code": null,
101
+ "tokenizers": {},
 
102
  "children": [
103
  {
104
  "name": "Demta",
105
  "iso_1_code": null,
106
  "iso_3_code": null,
107
+ "tokenizers": {},
 
108
  "children": [
109
  {
110
  "name": "Sowari",
111
  "iso_1_code": null,
112
  "iso_3_code": "dmy",
113
+ "tokenizers": {},
114
+ "children": [],
115
+ "node_i": "3725",
116
+ "scripts": [],
117
+ "own_tokenizer": false
118
  }
119
+ ],
120
+ "node_i": "3724",
121
+ "scripts": [],
122
+ "own_tokenizer": false
123
  },
124
  {
125
  "name": "Sentani Proper",
126
  "iso_1_code": null,
127
  "iso_3_code": null,
128
+ "tokenizers": {},
 
129
  "children": [
130
  {
131
  "name": "Nafri",
132
  "iso_1_code": null,
133
  "iso_3_code": "nxx",
134
+ "tokenizers": {},
135
+ "children": [],
136
+ "node_i": "3727",
137
+ "scripts": [],
138
+ "own_tokenizer": false
139
  },
140
  {
141
  "name": "Sentani",
142
  "iso_1_code": null,
143
  "iso_3_code": "set",
144
+ "tokenizers": {},
145
+ "children": [],
146
+ "node_i": "3728",
147
+ "scripts": [],
148
+ "own_tokenizer": false
149
  },
150
  {
151
  "name": "Tabla",
152
  "iso_1_code": null,
153
  "iso_3_code": "tnm",
154
+ "tokenizers": {},
155
+ "children": [],
156
+ "node_i": "3729",
157
+ "scripts": [],
158
+ "own_tokenizer": false
159
  }
160
+ ],
161
+ "node_i": "3726",
162
+ "scripts": [],
163
+ "own_tokenizer": false
164
  }
165
+ ],
166
+ "node_i": "3723",
167
+ "scripts": [],
168
+ "own_tokenizer": false
169
  }
170
+ ],
171
+ "node_i": "3714",
172
+ "scripts": [],
173
+ "own_tokenizer": false
174
  }
data/East Geelvink Bay.json CHANGED
@@ -2,113 +2,143 @@
2
  "name": "East Geelvink Bay",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Anasi",
10
  "iso_1_code": null,
11
  "iso_3_code": "bpo",
12
- "tokenizer": null,
13
- "source": null,
14
- "children": []
 
 
15
  },
16
  {
17
  "name": "Barapasi",
18
  "iso_1_code": null,
19
  "iso_3_code": "brp",
20
- "tokenizer": null,
21
- "source": null,
22
- "children": []
 
 
23
  },
24
  {
25
  "name": "Burate",
26
  "iso_1_code": null,
27
  "iso_3_code": "bti",
28
- "tokenizer": null,
29
- "source": null,
30
- "children": []
 
 
31
  },
32
  {
33
  "name": "Kehu",
34
  "iso_1_code": null,
35
  "iso_3_code": "khh",
36
- "tokenizer": null,
37
- "source": null,
38
- "children": []
 
 
39
  },
40
  {
41
  "name": "Kofei",
42
  "iso_1_code": null,
43
  "iso_3_code": "kpi",
44
- "tokenizer": null,
45
- "source": null,
46
- "children": []
 
 
47
  },
48
  {
49
  "name": "Nisa",
50
  "iso_1_code": null,
51
  "iso_3_code": "njs",
52
- "tokenizer": null,
53
- "source": null,
54
- "children": []
 
 
55
  },
56
  {
57
  "name": "Sauri",
58
  "iso_1_code": null,
59
  "iso_3_code": "srt",
60
- "tokenizer": null,
61
- "source": null,
62
- "children": []
 
 
63
  },
64
  {
65
  "name": "Tefaro",
66
  "iso_1_code": null,
67
  "iso_3_code": "tfo",
68
- "tokenizer": null,
69
- "source": null,
70
- "children": []
 
 
71
  },
72
  {
73
  "name": "Tunggare",
74
  "iso_1_code": null,
75
  "iso_3_code": "trt",
76
- "tokenizer": null,
77
- "source": null,
78
- "children": []
 
 
79
  },
80
  {
81
  "name": "Woria",
82
  "iso_1_code": null,
83
  "iso_3_code": "wor",
84
- "tokenizer": null,
85
- "source": null,
86
- "children": []
 
 
87
  },
88
  {
89
  "name": "Bauzi",
90
  "iso_1_code": null,
91
  "iso_3_code": null,
92
- "tokenizer": null,
93
- "source": null,
94
  "children": [
95
  {
96
  "name": "Bauzi",
97
  "iso_1_code": null,
98
  "iso_3_code": "bvz",
99
- "tokenizer": null,
100
- "source": null,
101
- "children": []
 
 
 
 
102
  },
103
  {
104
  "name": "Demisa",
105
  "iso_1_code": null,
106
  "iso_3_code": "dei",
107
- "tokenizer": null,
108
- "source": null,
109
- "children": []
 
 
110
  }
111
- ]
 
 
 
112
  }
113
- ]
 
 
 
114
  }
 
2
  "name": "East Geelvink Bay",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Anasi",
9
  "iso_1_code": null,
10
  "iso_3_code": "bpo",
11
+ "tokenizers": {},
12
+ "children": [],
13
+ "node_i": "3731",
14
+ "scripts": [],
15
+ "own_tokenizer": false
16
  },
17
  {
18
  "name": "Barapasi",
19
  "iso_1_code": null,
20
  "iso_3_code": "brp",
21
+ "tokenizers": {},
22
+ "children": [],
23
+ "node_i": "3732",
24
+ "scripts": [],
25
+ "own_tokenizer": false
26
  },
27
  {
28
  "name": "Burate",
29
  "iso_1_code": null,
30
  "iso_3_code": "bti",
31
+ "tokenizers": {},
32
+ "children": [],
33
+ "node_i": "3733",
34
+ "scripts": [],
35
+ "own_tokenizer": false
36
  },
37
  {
38
  "name": "Kehu",
39
  "iso_1_code": null,
40
  "iso_3_code": "khh",
41
+ "tokenizers": {},
42
+ "children": [],
43
+ "node_i": "3734",
44
+ "scripts": [],
45
+ "own_tokenizer": false
46
  },
47
  {
48
  "name": "Kofei",
49
  "iso_1_code": null,
50
  "iso_3_code": "kpi",
51
+ "tokenizers": {},
52
+ "children": [],
53
+ "node_i": "3735",
54
+ "scripts": [],
55
+ "own_tokenizer": false
56
  },
57
  {
58
  "name": "Nisa",
59
  "iso_1_code": null,
60
  "iso_3_code": "njs",
61
+ "tokenizers": {},
62
+ "children": [],
63
+ "node_i": "3736",
64
+ "scripts": [],
65
+ "own_tokenizer": false
66
  },
67
  {
68
  "name": "Sauri",
69
  "iso_1_code": null,
70
  "iso_3_code": "srt",
71
+ "tokenizers": {},
72
+ "children": [],
73
+ "node_i": "3737",
74
+ "scripts": [],
75
+ "own_tokenizer": false
76
  },
77
  {
78
  "name": "Tefaro",
79
  "iso_1_code": null,
80
  "iso_3_code": "tfo",
81
+ "tokenizers": {},
82
+ "children": [],
83
+ "node_i": "3738",
84
+ "scripts": [],
85
+ "own_tokenizer": false
86
  },
87
  {
88
  "name": "Tunggare",
89
  "iso_1_code": null,
90
  "iso_3_code": "trt",
91
+ "tokenizers": {},
92
+ "children": [],
93
+ "node_i": "3739",
94
+ "scripts": [],
95
+ "own_tokenizer": false
96
  },
97
  {
98
  "name": "Woria",
99
  "iso_1_code": null,
100
  "iso_3_code": "wor",
101
+ "tokenizers": {},
102
+ "children": [],
103
+ "node_i": "3740",
104
+ "scripts": [],
105
+ "own_tokenizer": false
106
  },
107
  {
108
  "name": "Bauzi",
109
  "iso_1_code": null,
110
  "iso_3_code": null,
111
+ "tokenizers": {},
 
112
  "children": [
113
  {
114
  "name": "Bauzi",
115
  "iso_1_code": null,
116
  "iso_3_code": "bvz",
117
+ "tokenizers": {},
118
+ "children": [],
119
+ "node_i": "3742",
120
+ "scripts": [
121
+ "Latn"
122
+ ],
123
+ "own_tokenizer": false
124
  },
125
  {
126
  "name": "Demisa",
127
  "iso_1_code": null,
128
  "iso_3_code": "dei",
129
+ "tokenizers": {},
130
+ "children": [],
131
+ "node_i": "3743",
132
+ "scripts": [],
133
+ "own_tokenizer": false
134
  }
135
+ ],
136
+ "node_i": "3741",
137
+ "scripts": [],
138
+ "own_tokenizer": false
139
  }
140
+ ],
141
+ "node_i": "3730",
142
+ "scripts": [],
143
+ "own_tokenizer": false
144
  }
data/East New Britain.json CHANGED
@@ -2,82 +2,104 @@
2
  "name": "East New Britain",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Baining",
10
  "iso_1_code": null,
11
  "iso_3_code": null,
12
- "tokenizer": null,
13
- "source": null,
14
  "children": [
15
  {
16
  "name": "Qaqet",
17
  "iso_1_code": null,
18
  "iso_3_code": "byx",
19
- "tokenizer": null,
20
- "source": null,
21
- "children": []
 
 
 
 
22
  },
23
  {
24
  "name": "Kairak",
25
  "iso_1_code": null,
26
  "iso_3_code": "ckr",
27
- "tokenizer": null,
28
- "source": null,
29
- "children": []
 
 
30
  },
31
  {
32
  "name": "Mali",
33
  "iso_1_code": null,
34
  "iso_3_code": "gcc",
35
- "tokenizer": null,
36
- "source": null,
37
- "children": []
 
 
38
  },
39
  {
40
  "name": "Simbali",
41
  "iso_1_code": null,
42
  "iso_3_code": "smg",
43
- "tokenizer": null,
44
- "source": null,
45
- "children": []
 
 
46
  },
47
  {
48
  "name": "Ura",
49
  "iso_1_code": null,
50
  "iso_3_code": "uro",
51
- "tokenizer": null,
52
- "source": null,
53
- "children": []
 
 
54
  },
55
  {
56
  "name": "Makolkol",
57
  "iso_1_code": null,
58
  "iso_3_code": "zmh",
59
- "tokenizer": null,
60
- "source": null,
61
- "children": []
 
 
62
  }
63
- ]
 
 
 
64
  },
65
  {
66
  "name": "Taulil",
67
  "iso_1_code": null,
68
  "iso_3_code": null,
69
- "tokenizer": null,
70
- "source": null,
71
  "children": [
72
  {
73
  "name": "Tulil",
74
  "iso_1_code": null,
75
  "iso_3_code": "tuh",
76
- "tokenizer": null,
77
- "source": null,
78
- "children": []
 
 
79
  }
80
- ]
 
 
 
81
  }
82
- ]
 
 
 
83
  }
 
2
  "name": "East New Britain",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Baining",
9
  "iso_1_code": null,
10
  "iso_3_code": null,
11
+ "tokenizers": {},
 
12
  "children": [
13
  {
14
  "name": "Qaqet",
15
  "iso_1_code": null,
16
  "iso_3_code": "byx",
17
+ "tokenizers": {},
18
+ "children": [],
19
+ "node_i": "3746",
20
+ "scripts": [
21
+ "Latn"
22
+ ],
23
+ "own_tokenizer": false
24
  },
25
  {
26
  "name": "Kairak",
27
  "iso_1_code": null,
28
  "iso_3_code": "ckr",
29
+ "tokenizers": {},
30
+ "children": [],
31
+ "node_i": "3747",
32
+ "scripts": [],
33
+ "own_tokenizer": false
34
  },
35
  {
36
  "name": "Mali",
37
  "iso_1_code": null,
38
  "iso_3_code": "gcc",
39
+ "tokenizers": {},
40
+ "children": [],
41
+ "node_i": "3748",
42
+ "scripts": [],
43
+ "own_tokenizer": false
44
  },
45
  {
46
  "name": "Simbali",
47
  "iso_1_code": null,
48
  "iso_3_code": "smg",
49
+ "tokenizers": {},
50
+ "children": [],
51
+ "node_i": "3749",
52
+ "scripts": [],
53
+ "own_tokenizer": false
54
  },
55
  {
56
  "name": "Ura",
57
  "iso_1_code": null,
58
  "iso_3_code": "uro",
59
+ "tokenizers": {},
60
+ "children": [],
61
+ "node_i": "3750",
62
+ "scripts": [],
63
+ "own_tokenizer": false
64
  },
65
  {
66
  "name": "Makolkol",
67
  "iso_1_code": null,
68
  "iso_3_code": "zmh",
69
+ "tokenizers": {},
70
+ "children": [],
71
+ "node_i": "3751",
72
+ "scripts": [],
73
+ "own_tokenizer": false
74
  }
75
+ ],
76
+ "node_i": "3745",
77
+ "scripts": [],
78
+ "own_tokenizer": false
79
  },
80
  {
81
  "name": "Taulil",
82
  "iso_1_code": null,
83
  "iso_3_code": null,
84
+ "tokenizers": {},
 
85
  "children": [
86
  {
87
  "name": "Tulil",
88
  "iso_1_code": null,
89
  "iso_3_code": "tuh",
90
+ "tokenizers": {},
91
+ "children": [],
92
+ "node_i": "3753",
93
+ "scripts": [],
94
+ "own_tokenizer": false
95
  }
96
+ ],
97
+ "node_i": "3752",
98
+ "scripts": [],
99
+ "own_tokenizer": false
100
  }
101
+ ],
102
+ "node_i": "3744",
103
+ "scripts": [],
104
+ "own_tokenizer": false
105
  }
data/Eastern Trans-Fly.json CHANGED
@@ -2,40 +2,54 @@
2
  "name": "Eastern Trans-Fly",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Bine",
10
  "iso_1_code": null,
11
  "iso_3_code": "bon",
12
- "tokenizer": null,
13
- "source": null,
14
- "children": []
 
 
 
 
15
  },
16
  {
17
  "name": "Wipi",
18
  "iso_1_code": null,
19
  "iso_3_code": "gdr",
20
- "tokenizer": null,
21
- "source": null,
22
- "children": []
 
 
 
 
23
  },
24
  {
25
  "name": "Gizrra",
26
  "iso_1_code": null,
27
  "iso_3_code": "tof",
28
- "tokenizer": null,
29
- "source": null,
30
- "children": []
 
 
31
  },
32
  {
33
  "name": "Meriam Mir",
34
  "iso_1_code": null,
35
  "iso_3_code": "ulk",
36
- "tokenizer": null,
37
- "source": null,
38
- "children": []
 
 
39
  }
40
- ]
 
 
 
41
  }
 
2
  "name": "Eastern Trans-Fly",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Bine",
9
  "iso_1_code": null,
10
  "iso_3_code": "bon",
11
+ "tokenizers": {},
12
+ "children": [],
13
+ "node_i": "3755",
14
+ "scripts": [
15
+ "Latn"
16
+ ],
17
+ "own_tokenizer": false
18
  },
19
  {
20
  "name": "Wipi",
21
  "iso_1_code": null,
22
  "iso_3_code": "gdr",
23
+ "tokenizers": {},
24
+ "children": [],
25
+ "node_i": "3756",
26
+ "scripts": [
27
+ "Latn"
28
+ ],
29
+ "own_tokenizer": false
30
  },
31
  {
32
  "name": "Gizrra",
33
  "iso_1_code": null,
34
  "iso_3_code": "tof",
35
+ "tokenizers": {},
36
+ "children": [],
37
+ "node_i": "3757",
38
+ "scripts": [],
39
+ "own_tokenizer": false
40
  },
41
  {
42
  "name": "Meriam Mir",
43
  "iso_1_code": null,
44
  "iso_3_code": "ulk",
45
+ "tokenizers": {},
46
+ "children": [],
47
+ "node_i": "3758",
48
+ "scripts": [],
49
+ "own_tokenizer": false
50
  }
51
+ ],
52
+ "node_i": "3754",
53
+ "scripts": [],
54
+ "own_tokenizer": false
55
  }
data/Eskimo-Aleut.json CHANGED
@@ -2,141 +2,189 @@
2
  "name": "Eskimo-Aleut",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Aleut",
10
  "iso_1_code": null,
11
  "iso_3_code": null,
12
- "tokenizer": null,
13
- "source": null,
14
  "children": [
15
  {
16
  "name": "Aleut",
17
  "iso_1_code": null,
18
  "iso_3_code": "ale",
19
- "tokenizer": null,
20
- "source": null,
21
- "children": []
 
 
22
  }
23
- ]
 
 
 
24
  },
25
  {
26
  "name": "Eskimo",
27
  "iso_1_code": null,
28
  "iso_3_code": null,
29
- "tokenizer": null,
30
- "source": null,
31
  "children": [
32
  {
33
  "name": "Inuit-Inupiaq",
34
  "iso_1_code": null,
35
  "iso_3_code": null,
36
- "tokenizer": null,
37
- "source": null,
38
  "children": [
39
  {
40
  "name": "Inupiatun, North Alaskan",
41
  "iso_1_code": "ik",
42
  "iso_3_code": "esi",
43
- "tokenizer": null,
44
- "source": "macrolanguage",
45
- "children": []
 
 
 
 
46
  },
47
  {
48
  "name": "Inupiatun, Northwest Alaska",
49
  "iso_1_code": "ik",
50
  "iso_3_code": "esk",
51
- "tokenizer": null,
52
- "source": "macrolanguage",
53
- "children": []
 
 
 
 
54
  },
55
  {
56
  "name": "Inuktitut, Eastern Canadian",
57
  "iso_1_code": "iu",
58
  "iso_3_code": "ike",
59
- "tokenizer": null,
60
- "source": "macrolanguage",
61
- "children": []
 
 
 
 
62
  },
63
  {
64
  "name": "Inuinnaqtun",
65
  "iso_1_code": "iu",
66
  "iso_3_code": "ikt",
67
- "tokenizer": null,
68
- "source": "macrolanguage",
69
- "children": []
 
 
 
 
70
  },
71
  {
72
  "name": "Greenlandic",
73
  "iso_1_code": "kl",
74
  "iso_3_code": "kal",
75
- "tokenizer": null,
76
- "source": null,
77
- "children": []
 
 
 
 
78
  }
79
- ]
 
 
 
80
  },
81
  {
82
  "name": "Yupik",
83
  "iso_1_code": null,
84
  "iso_3_code": null,
85
- "tokenizer": null,
86
- "source": null,
87
  "children": [
88
  {
89
  "name": "Yupik, Saint Lawrence Island",
90
  "iso_1_code": null,
91
  "iso_3_code": "ess",
92
- "tokenizer": null,
93
- "source": null,
94
- "children": []
 
 
 
 
95
  },
96
  {
97
  "name": "Yupik, Naukan",
98
  "iso_1_code": null,
99
  "iso_3_code": "ynk",
100
- "tokenizer": null,
101
- "source": null,
102
- "children": []
 
 
103
  },
104
  {
105
  "name": "Yupik, Sirenik",
106
  "iso_1_code": null,
107
  "iso_3_code": "ysr",
108
- "tokenizer": null,
109
- "source": null,
110
- "children": []
 
 
111
  },
112
  {
113
  "name": "Alaskan Yupik",
114
  "iso_1_code": null,
115
  "iso_3_code": null,
116
- "tokenizer": null,
117
- "source": null,
118
  "children": [
119
  {
120
  "name": "Yupik, Pacific Gulf",
121
  "iso_1_code": null,
122
  "iso_3_code": "ems",
123
- "tokenizer": null,
124
- "source": null,
125
- "children": []
 
 
126
  },
127
  {
128
  "name": "Yupik, Central",
129
  "iso_1_code": null,
130
  "iso_3_code": "esu",
131
- "tokenizer": null,
132
- "source": null,
133
- "children": []
 
 
 
 
134
  }
135
- ]
 
 
 
136
  }
137
- ]
 
 
 
138
  }
139
- ]
 
 
 
140
  }
141
- ]
 
 
 
142
  }
 
2
  "name": "Eskimo-Aleut",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Aleut",
9
  "iso_1_code": null,
10
  "iso_3_code": null,
11
+ "tokenizers": {},
 
12
  "children": [
13
  {
14
  "name": "Aleut",
15
  "iso_1_code": null,
16
  "iso_3_code": "ale",
17
+ "tokenizers": {},
18
+ "children": [],
19
+ "node_i": "3761",
20
+ "scripts": [],
21
+ "own_tokenizer": false
22
  }
23
+ ],
24
+ "node_i": "3760",
25
+ "scripts": [],
26
+ "own_tokenizer": false
27
  },
28
  {
29
  "name": "Eskimo",
30
  "iso_1_code": null,
31
  "iso_3_code": null,
32
+ "tokenizers": {},
 
33
  "children": [
34
  {
35
  "name": "Inuit-Inupiaq",
36
  "iso_1_code": null,
37
  "iso_3_code": null,
38
+ "tokenizers": {},
 
39
  "children": [
40
  {
41
  "name": "Inupiatun, North Alaskan",
42
  "iso_1_code": "ik",
43
  "iso_3_code": "esi",
44
+ "tokenizers": {},
45
+ "children": [],
46
+ "node_i": "3764",
47
+ "scripts": [
48
+ "Latn"
49
+ ],
50
+ "own_tokenizer": false
51
  },
52
  {
53
  "name": "Inupiatun, Northwest Alaska",
54
  "iso_1_code": "ik",
55
  "iso_3_code": "esk",
56
+ "tokenizers": {},
57
+ "children": [],
58
+ "node_i": "3765",
59
+ "scripts": [
60
+ "Latn"
61
+ ],
62
+ "own_tokenizer": false
63
  },
64
  {
65
  "name": "Inuktitut, Eastern Canadian",
66
  "iso_1_code": "iu",
67
  "iso_3_code": "ike",
68
+ "tokenizers": {},
69
+ "children": [],
70
+ "node_i": "3766",
71
+ "scripts": [
72
+ "Cans"
73
+ ],
74
+ "own_tokenizer": false
75
  },
76
  {
77
  "name": "Inuinnaqtun",
78
  "iso_1_code": "iu",
79
  "iso_3_code": "ikt",
80
+ "tokenizers": {},
81
+ "children": [],
82
+ "node_i": "3767",
83
+ "scripts": [
84
+ "Latn"
85
+ ],
86
+ "own_tokenizer": false
87
  },
88
  {
89
  "name": "Greenlandic",
90
  "iso_1_code": "kl",
91
  "iso_3_code": "kal",
92
+ "tokenizers": {},
93
+ "children": [],
94
+ "node_i": "3768",
95
+ "scripts": [
96
+ "Latn"
97
+ ],
98
+ "own_tokenizer": false
99
  }
100
+ ],
101
+ "node_i": "3763",
102
+ "scripts": [],
103
+ "own_tokenizer": false
104
  },
105
  {
106
  "name": "Yupik",
107
  "iso_1_code": null,
108
  "iso_3_code": null,
109
+ "tokenizers": {},
 
110
  "children": [
111
  {
112
  "name": "Yupik, Saint Lawrence Island",
113
  "iso_1_code": null,
114
  "iso_3_code": "ess",
115
+ "tokenizers": {},
116
+ "children": [],
117
+ "node_i": "3770",
118
+ "scripts": [
119
+ "Latn"
120
+ ],
121
+ "own_tokenizer": false
122
  },
123
  {
124
  "name": "Yupik, Naukan",
125
  "iso_1_code": null,
126
  "iso_3_code": "ynk",
127
+ "tokenizers": {},
128
+ "children": [],
129
+ "node_i": "3771",
130
+ "scripts": [],
131
+ "own_tokenizer": false
132
  },
133
  {
134
  "name": "Yupik, Sirenik",
135
  "iso_1_code": null,
136
  "iso_3_code": "ysr",
137
+ "tokenizers": {},
138
+ "children": [],
139
+ "node_i": "3772",
140
+ "scripts": [],
141
+ "own_tokenizer": false
142
  },
143
  {
144
  "name": "Alaskan Yupik",
145
  "iso_1_code": null,
146
  "iso_3_code": null,
147
+ "tokenizers": {},
 
148
  "children": [
149
  {
150
  "name": "Yupik, Pacific Gulf",
151
  "iso_1_code": null,
152
  "iso_3_code": "ems",
153
+ "tokenizers": {},
154
+ "children": [],
155
+ "node_i": "3774",
156
+ "scripts": [],
157
+ "own_tokenizer": false
158
  },
159
  {
160
  "name": "Yupik, Central",
161
  "iso_1_code": null,
162
  "iso_3_code": "esu",
163
+ "tokenizers": {},
164
+ "children": [],
165
+ "node_i": "3775",
166
+ "scripts": [
167
+ "Latn"
168
+ ],
169
+ "own_tokenizer": false
170
  }
171
+ ],
172
+ "node_i": "3773",
173
+ "scripts": [],
174
+ "own_tokenizer": false
175
  }
176
+ ],
177
+ "node_i": "3769",
178
+ "scripts": [],
179
+ "own_tokenizer": false
180
  }
181
+ ],
182
+ "node_i": "3762",
183
+ "scripts": [],
184
+ "own_tokenizer": false
185
  }
186
+ ],
187
+ "node_i": "3759",
188
+ "scripts": [],
189
+ "own_tokenizer": false
190
  }
data/Eyak-Athabaskan.json CHANGED
@@ -2,510 +2,648 @@
2
  "name": "Eyak-Athabaskan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Eyak",
10
  "iso_1_code": null,
11
  "iso_3_code": "eya",
12
- "tokenizer": null,
13
- "source": null,
14
- "children": []
 
 
15
  },
16
  {
17
  "name": "Athabaskan",
18
  "iso_1_code": null,
19
  "iso_3_code": null,
20
- "tokenizer": null,
21
- "source": null,
22
  "children": [
23
  {
24
  "name": "Apachean",
25
  "iso_1_code": null,
26
  "iso_3_code": null,
27
- "tokenizer": null,
28
- "source": null,
29
  "children": [
30
  {
31
  "name": "Navajo",
32
  "iso_1_code": "nv",
33
  "iso_3_code": "nav",
34
- "tokenizer": null,
35
- "source": null,
36
- "children": []
 
 
 
 
37
  },
38
  {
39
  "name": "Apache",
40
  "iso_1_code": null,
41
  "iso_3_code": null,
42
- "tokenizer": null,
43
- "source": null,
44
  "children": [
45
  {
46
  "name": "Apache, Jicarilla",
47
  "iso_1_code": null,
48
  "iso_3_code": "apj",
49
- "tokenizer": null,
50
- "source": null,
51
- "children": []
 
 
52
  },
53
  {
54
  "name": "Apache, Kiowa",
55
  "iso_1_code": null,
56
  "iso_3_code": "apk",
57
- "tokenizer": null,
58
- "source": null,
59
- "children": []
 
 
60
  },
61
  {
62
  "name": "Apache, Lipan",
63
  "iso_1_code": null,
64
  "iso_3_code": "apl",
65
- "tokenizer": null,
66
- "source": null,
67
- "children": []
 
 
68
  },
69
  {
70
  "name": "Apache, Mescalero-Chiricahua",
71
  "iso_1_code": null,
72
  "iso_3_code": "apm",
73
- "tokenizer": null,
74
- "source": null,
75
- "children": []
 
 
76
  },
77
  {
78
  "name": "Apache, Western",
79
  "iso_1_code": null,
80
  "iso_3_code": "apw",
81
- "tokenizer": null,
82
- "source": null,
83
- "children": []
 
 
 
 
84
  }
85
- ]
 
 
 
86
  }
87
- ]
 
 
 
88
  },
89
  {
90
  "name": "Northern Athabaskan",
91
  "iso_1_code": null,
92
  "iso_3_code": null,
93
- "tokenizer": null,
94
- "source": null,
95
  "children": [
96
  {
97
  "name": "Ahtena",
98
  "iso_1_code": null,
99
  "iso_3_code": "aht",
100
- "tokenizer": null,
101
- "source": null,
102
- "children": []
 
 
103
  },
104
  {
105
  "name": "Babine",
106
  "iso_1_code": null,
107
  "iso_3_code": "bcr",
108
- "tokenizer": null,
109
- "source": null,
110
- "children": []
 
 
111
  },
112
  {
113
  "name": "Beaver",
114
  "iso_1_code": null,
115
  "iso_3_code": "bea",
116
- "tokenizer": null,
117
- "source": null,
118
- "children": []
 
 
 
 
119
  },
120
  {
121
  "name": "Dene",
122
  "iso_1_code": null,
123
  "iso_3_code": "chp",
124
- "tokenizer": null,
125
- "source": null,
126
- "children": []
 
 
127
  },
128
  {
129
  "name": "Chilcotin",
130
  "iso_1_code": null,
131
  "iso_3_code": "clc",
132
- "tokenizer": null,
133
- "source": null,
134
- "children": []
 
 
135
  },
136
  {
137
  "name": "Tlicho",
138
  "iso_1_code": null,
139
  "iso_3_code": "dgr",
140
- "tokenizer": null,
141
- "source": null,
142
- "children": []
 
 
 
 
143
  },
144
  {
145
  "name": "Gwich\u2019in",
146
  "iso_1_code": null,
147
  "iso_3_code": "gwi",
148
- "tokenizer": null,
149
- "source": null,
150
- "children": []
 
 
 
 
151
  },
152
  {
153
  "name": "Han",
154
  "iso_1_code": null,
155
  "iso_3_code": "haa",
156
- "tokenizer": null,
157
- "source": null,
158
- "children": []
 
 
159
  },
160
  {
161
  "name": "Holikachuk",
162
  "iso_1_code": null,
163
  "iso_3_code": "hoi",
164
- "tokenizer": null,
165
- "source": null,
166
- "children": []
 
 
167
  },
168
  {
169
  "name": "Deg Xinag",
170
  "iso_1_code": null,
171
  "iso_3_code": "ing",
172
- "tokenizer": null,
173
- "source": null,
174
- "children": []
 
 
175
  },
176
  {
177
  "name": "Koyukon",
178
  "iso_1_code": null,
179
  "iso_3_code": "koy",
180
- "tokenizer": null,
181
- "source": null,
182
- "children": []
 
 
183
  },
184
  {
185
  "name": "Kuskokwim, Upper",
186
  "iso_1_code": null,
187
  "iso_3_code": "kuu",
188
- "tokenizer": null,
189
- "source": null,
190
- "children": []
 
 
191
  },
192
  {
193
  "name": "Sekani",
194
  "iso_1_code": null,
195
  "iso_3_code": "sek",
196
- "tokenizer": null,
197
- "source": null,
198
- "children": []
 
 
199
  },
200
  {
201
  "name": "Sarsi",
202
  "iso_1_code": null,
203
  "iso_3_code": "srs",
204
- "tokenizer": null,
205
- "source": null,
206
- "children": []
 
 
207
  },
208
  {
209
  "name": "Tanana, Lower",
210
  "iso_1_code": null,
211
  "iso_3_code": "taa",
212
- "tokenizer": null,
213
- "source": null,
214
- "children": []
 
 
215
  },
216
  {
217
  "name": "Tanana, Upper",
218
  "iso_1_code": null,
219
  "iso_3_code": "tau",
220
- "tokenizer": null,
221
- "source": null,
222
- "children": []
 
 
223
  },
224
  {
225
  "name": "Tanacross",
226
  "iso_1_code": null,
227
  "iso_3_code": "tcb",
228
- "tokenizer": null,
229
- "source": null,
230
- "children": []
 
 
231
  },
232
  {
233
  "name": "Tanaina",
234
  "iso_1_code": null,
235
  "iso_3_code": "tfn",
236
- "tokenizer": null,
237
- "source": null,
238
- "children": []
 
 
239
  },
240
  {
241
  "name": "Tsetsaut",
242
  "iso_1_code": null,
243
  "iso_3_code": "txc",
244
- "tokenizer": null,
245
- "source": null,
246
- "children": []
 
 
247
  },
248
  {
249
  "name": "Carrier",
250
  "iso_1_code": null,
251
  "iso_3_code": null,
252
- "tokenizer": null,
253
- "source": null,
254
  "children": [
255
  {
256
  "name": "Carrier, Southern",
257
  "iso_1_code": null,
258
  "iso_3_code": "caf",
259
- "tokenizer": null,
260
- "source": null,
261
- "children": []
 
 
 
 
262
  },
263
  {
264
  "name": "Carrier",
265
  "iso_1_code": null,
266
  "iso_3_code": "crx",
267
- "tokenizer": null,
268
- "source": null,
269
- "children": []
 
 
 
 
270
  }
271
- ]
 
 
 
272
  },
273
  {
274
  "name": "Slavey-Hare",
275
  "iso_1_code": null,
276
  "iso_3_code": null,
277
- "tokenizer": null,
278
- "source": null,
279
  "children": [
280
  {
281
  "name": "Slavey, North",
282
  "iso_1_code": null,
283
  "iso_3_code": "scs",
284
- "tokenizer": null,
285
- "source": null,
286
- "children": []
 
 
287
  },
288
  {
289
  "name": "Slavey, South",
290
  "iso_1_code": null,
291
  "iso_3_code": "xsl",
292
- "tokenizer": null,
293
- "source": null,
294
- "children": []
 
 
295
  }
296
- ]
 
 
 
297
  },
298
  {
299
  "name": "Tahltan",
300
  "iso_1_code": null,
301
  "iso_3_code": null,
302
- "tokenizer": null,
303
- "source": null,
304
  "children": [
305
  {
306
  "name": "Kaska",
307
  "iso_1_code": null,
308
  "iso_3_code": "kkz",
309
- "tokenizer": null,
310
- "source": null,
311
- "children": []
 
 
312
  },
313
  {
314
  "name": "Tagish",
315
  "iso_1_code": null,
316
  "iso_3_code": "tgx",
317
- "tokenizer": null,
318
- "source": null,
319
- "children": []
 
 
320
  },
321
  {
322
  "name": "Tahltan",
323
  "iso_1_code": null,
324
  "iso_3_code": "tht",
325
- "tokenizer": null,
326
- "source": null,
327
- "children": []
 
 
328
  }
329
- ]
 
 
 
330
  },
331
  {
332
  "name": "Tuchone",
333
  "iso_1_code": null,
334
  "iso_3_code": null,
335
- "tokenizer": null,
336
- "source": null,
337
  "children": [
338
  {
339
  "name": "Tutchone, Southern",
340
  "iso_1_code": null,
341
  "iso_3_code": "tce",
342
- "tokenizer": null,
343
- "source": null,
344
- "children": []
 
 
345
  },
346
  {
347
  "name": "Tutchone, Northern",
348
  "iso_1_code": null,
349
  "iso_3_code": "ttm",
350
- "tokenizer": null,
351
- "source": null,
352
- "children": []
 
 
353
  }
354
- ]
 
 
 
355
  }
356
- ]
 
 
 
357
  },
358
  {
359
  "name": "Pacific Coast Athabaskan",
360
  "iso_1_code": null,
361
  "iso_3_code": null,
362
- "tokenizer": null,
363
- "source": null,
364
  "children": [
365
  {
366
  "name": "Kwalhioqua-Tlatskanai",
367
  "iso_1_code": null,
368
  "iso_3_code": "qwt",
369
- "tokenizer": null,
370
- "source": null,
371
- "children": []
 
 
372
  },
373
  {
374
  "name": "California Athabaskan",
375
  "iso_1_code": null,
376
  "iso_3_code": null,
377
- "tokenizer": null,
378
- "source": null,
379
  "children": [
380
  {
381
  "name": "Hupa",
382
  "iso_1_code": null,
383
  "iso_3_code": "hup",
384
- "tokenizer": null,
385
- "source": null,
386
- "children": []
 
 
387
  },
388
  {
389
  "name": "Kato",
390
  "iso_1_code": null,
391
  "iso_3_code": "ktw",
392
- "tokenizer": null,
393
- "source": null,
394
- "children": []
 
 
395
  },
396
  {
397
  "name": "Mattole",
398
  "iso_1_code": null,
399
  "iso_3_code": "mvb",
400
- "tokenizer": null,
401
- "source": null,
402
- "children": []
 
 
403
  },
404
  {
405
  "name": "Wailaki",
406
  "iso_1_code": null,
407
  "iso_3_code": "wlk",
408
- "tokenizer": null,
409
- "source": null,
410
- "children": []
 
 
411
  }
412
- ]
 
 
 
413
  },
414
  {
415
  "name": "Oregon Athabaskan",
416
  "iso_1_code": null,
417
  "iso_3_code": null,
418
- "tokenizer": null,
419
- "source": null,
420
  "children": [
421
  {
422
  "name": "Galice",
423
  "iso_1_code": null,
424
  "iso_3_code": "gce",
425
- "tokenizer": null,
426
- "source": null,
427
- "children": []
 
 
428
  },
429
  {
430
  "name": "Upper Umpqua",
431
  "iso_1_code": null,
432
  "iso_3_code": "xup",
433
- "tokenizer": null,
434
- "source": null,
435
- "children": []
 
 
436
  },
437
  {
438
  "name": "Tolowa-Chetco",
439
  "iso_1_code": null,
440
  "iso_3_code": null,
441
- "tokenizer": null,
442
- "source": null,
443
  "children": [
444
  {
445
  "name": "Chetco",
446
  "iso_1_code": null,
447
  "iso_3_code": "ctc",
448
- "tokenizer": null,
449
- "source": null,
450
- "children": []
 
 
451
  },
452
  {
453
  "name": "Tolowa",
454
  "iso_1_code": null,
455
  "iso_3_code": "tol",
456
- "tokenizer": null,
457
- "source": null,
458
- "children": []
 
 
459
  }
460
- ]
 
 
 
461
  },
462
  {
463
  "name": "Tututni-Chasta Costa-Coquille",
464
  "iso_1_code": null,
465
  "iso_3_code": null,
466
- "tokenizer": null,
467
- "source": null,
468
  "children": [
469
  {
470
  "name": "Coquille",
471
  "iso_1_code": null,
472
  "iso_3_code": "coq",
473
- "tokenizer": null,
474
- "source": null,
475
- "children": []
 
 
476
  },
477
  {
478
  "name": "Tututni",
479
  "iso_1_code": null,
480
  "iso_3_code": "tuu",
481
- "tokenizer": null,
482
- "source": null,
483
- "children": []
 
 
484
  }
485
- ]
 
 
 
486
  }
487
- ]
 
 
 
488
  }
489
- ]
 
 
 
490
  }
491
- ]
 
 
 
492
  },
493
  {
494
  "name": "Tlingit",
495
  "iso_1_code": null,
496
  "iso_3_code": null,
497
- "tokenizer": null,
498
- "source": null,
499
  "children": [
500
  {
501
  "name": "Tlingit",
502
  "iso_1_code": null,
503
  "iso_3_code": "tli",
504
- "tokenizer": null,
505
- "source": null,
506
- "children": []
 
 
507
  }
508
- ]
 
 
 
509
  }
510
- ]
 
 
 
511
  }
 
2
  "name": "Eyak-Athabaskan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Eyak",
9
  "iso_1_code": null,
10
  "iso_3_code": "eya",
11
+ "tokenizers": {},
12
+ "children": [],
13
+ "node_i": "3777",
14
+ "scripts": [],
15
+ "own_tokenizer": false
16
  },
17
  {
18
  "name": "Athabaskan",
19
  "iso_1_code": null,
20
  "iso_3_code": null,
21
+ "tokenizers": {},
 
22
  "children": [
23
  {
24
  "name": "Apachean",
25
  "iso_1_code": null,
26
  "iso_3_code": null,
27
+ "tokenizers": {},
 
28
  "children": [
29
  {
30
  "name": "Navajo",
31
  "iso_1_code": "nv",
32
  "iso_3_code": "nav",
33
+ "tokenizers": {},
34
+ "children": [],
35
+ "node_i": "3780",
36
+ "scripts": [
37
+ "Latn"
38
+ ],
39
+ "own_tokenizer": false
40
  },
41
  {
42
  "name": "Apache",
43
  "iso_1_code": null,
44
  "iso_3_code": null,
45
+ "tokenizers": {},
 
46
  "children": [
47
  {
48
  "name": "Apache, Jicarilla",
49
  "iso_1_code": null,
50
  "iso_3_code": "apj",
51
+ "tokenizers": {},
52
+ "children": [],
53
+ "node_i": "3782",
54
+ "scripts": [],
55
+ "own_tokenizer": false
56
  },
57
  {
58
  "name": "Apache, Kiowa",
59
  "iso_1_code": null,
60
  "iso_3_code": "apk",
61
+ "tokenizers": {},
62
+ "children": [],
63
+ "node_i": "3783",
64
+ "scripts": [],
65
+ "own_tokenizer": false
66
  },
67
  {
68
  "name": "Apache, Lipan",
69
  "iso_1_code": null,
70
  "iso_3_code": "apl",
71
+ "tokenizers": {},
72
+ "children": [],
73
+ "node_i": "3784",
74
+ "scripts": [],
75
+ "own_tokenizer": false
76
  },
77
  {
78
  "name": "Apache, Mescalero-Chiricahua",
79
  "iso_1_code": null,
80
  "iso_3_code": "apm",
81
+ "tokenizers": {},
82
+ "children": [],
83
+ "node_i": "3785",
84
+ "scripts": [],
85
+ "own_tokenizer": false
86
  },
87
  {
88
  "name": "Apache, Western",
89
  "iso_1_code": null,
90
  "iso_3_code": "apw",
91
+ "tokenizers": {},
92
+ "children": [],
93
+ "node_i": "3786",
94
+ "scripts": [
95
+ "Latn"
96
+ ],
97
+ "own_tokenizer": false
98
  }
99
+ ],
100
+ "node_i": "3781",
101
+ "scripts": [],
102
+ "own_tokenizer": false
103
  }
104
+ ],
105
+ "node_i": "3779",
106
+ "scripts": [],
107
+ "own_tokenizer": false
108
  },
109
  {
110
  "name": "Northern Athabaskan",
111
  "iso_1_code": null,
112
  "iso_3_code": null,
113
+ "tokenizers": {},
 
114
  "children": [
115
  {
116
  "name": "Ahtena",
117
  "iso_1_code": null,
118
  "iso_3_code": "aht",
119
+ "tokenizers": {},
120
+ "children": [],
121
+ "node_i": "3788",
122
+ "scripts": [],
123
+ "own_tokenizer": false
124
  },
125
  {
126
  "name": "Babine",
127
  "iso_1_code": null,
128
  "iso_3_code": "bcr",
129
+ "tokenizers": {},
130
+ "children": [],
131
+ "node_i": "3789",
132
+ "scripts": [],
133
+ "own_tokenizer": false
134
  },
135
  {
136
  "name": "Beaver",
137
  "iso_1_code": null,
138
  "iso_3_code": "bea",
139
+ "tokenizers": {},
140
+ "children": [],
141
+ "node_i": "3790",
142
+ "scripts": [
143
+ "Latn"
144
+ ],
145
+ "own_tokenizer": false
146
  },
147
  {
148
  "name": "Dene",
149
  "iso_1_code": null,
150
  "iso_3_code": "chp",
151
+ "tokenizers": {},
152
+ "children": [],
153
+ "node_i": "3791",
154
+ "scripts": [],
155
+ "own_tokenizer": false
156
  },
157
  {
158
  "name": "Chilcotin",
159
  "iso_1_code": null,
160
  "iso_3_code": "clc",
161
+ "tokenizers": {},
162
+ "children": [],
163
+ "node_i": "3792",
164
+ "scripts": [],
165
+ "own_tokenizer": false
166
  },
167
  {
168
  "name": "Tlicho",
169
  "iso_1_code": null,
170
  "iso_3_code": "dgr",
171
+ "tokenizers": {},
172
+ "children": [],
173
+ "node_i": "3793",
174
+ "scripts": [
175
+ "Latn"
176
+ ],
177
+ "own_tokenizer": false
178
  },
179
  {
180
  "name": "Gwich\u2019in",
181
  "iso_1_code": null,
182
  "iso_3_code": "gwi",
183
+ "tokenizers": {},
184
+ "children": [],
185
+ "node_i": "3794",
186
+ "scripts": [
187
+ "Latn"
188
+ ],
189
+ "own_tokenizer": false
190
  },
191
  {
192
  "name": "Han",
193
  "iso_1_code": null,
194
  "iso_3_code": "haa",
195
+ "tokenizers": {},
196
+ "children": [],
197
+ "node_i": "3795",
198
+ "scripts": [],
199
+ "own_tokenizer": false
200
  },
201
  {
202
  "name": "Holikachuk",
203
  "iso_1_code": null,
204
  "iso_3_code": "hoi",
205
+ "tokenizers": {},
206
+ "children": [],
207
+ "node_i": "3796",
208
+ "scripts": [],
209
+ "own_tokenizer": false
210
  },
211
  {
212
  "name": "Deg Xinag",
213
  "iso_1_code": null,
214
  "iso_3_code": "ing",
215
+ "tokenizers": {},
216
+ "children": [],
217
+ "node_i": "3797",
218
+ "scripts": [],
219
+ "own_tokenizer": false
220
  },
221
  {
222
  "name": "Koyukon",
223
  "iso_1_code": null,
224
  "iso_3_code": "koy",
225
+ "tokenizers": {},
226
+ "children": [],
227
+ "node_i": "3798",
228
+ "scripts": [],
229
+ "own_tokenizer": false
230
  },
231
  {
232
  "name": "Kuskokwim, Upper",
233
  "iso_1_code": null,
234
  "iso_3_code": "kuu",
235
+ "tokenizers": {},
236
+ "children": [],
237
+ "node_i": "3799",
238
+ "scripts": [],
239
+ "own_tokenizer": false
240
  },
241
  {
242
  "name": "Sekani",
243
  "iso_1_code": null,
244
  "iso_3_code": "sek",
245
+ "tokenizers": {},
246
+ "children": [],
247
+ "node_i": "3800",
248
+ "scripts": [],
249
+ "own_tokenizer": false
250
  },
251
  {
252
  "name": "Sarsi",
253
  "iso_1_code": null,
254
  "iso_3_code": "srs",
255
+ "tokenizers": {},
256
+ "children": [],
257
+ "node_i": "3801",
258
+ "scripts": [],
259
+ "own_tokenizer": false
260
  },
261
  {
262
  "name": "Tanana, Lower",
263
  "iso_1_code": null,
264
  "iso_3_code": "taa",
265
+ "tokenizers": {},
266
+ "children": [],
267
+ "node_i": "3802",
268
+ "scripts": [],
269
+ "own_tokenizer": false
270
  },
271
  {
272
  "name": "Tanana, Upper",
273
  "iso_1_code": null,
274
  "iso_3_code": "tau",
275
+ "tokenizers": {},
276
+ "children": [],
277
+ "node_i": "3803",
278
+ "scripts": [],
279
+ "own_tokenizer": false
280
  },
281
  {
282
  "name": "Tanacross",
283
  "iso_1_code": null,
284
  "iso_3_code": "tcb",
285
+ "tokenizers": {},
286
+ "children": [],
287
+ "node_i": "3804",
288
+ "scripts": [],
289
+ "own_tokenizer": false
290
  },
291
  {
292
  "name": "Tanaina",
293
  "iso_1_code": null,
294
  "iso_3_code": "tfn",
295
+ "tokenizers": {},
296
+ "children": [],
297
+ "node_i": "3805",
298
+ "scripts": [],
299
+ "own_tokenizer": false
300
  },
301
  {
302
  "name": "Tsetsaut",
303
  "iso_1_code": null,
304
  "iso_3_code": "txc",
305
+ "tokenizers": {},
306
+ "children": [],
307
+ "node_i": "3806",
308
+ "scripts": [],
309
+ "own_tokenizer": false
310
  },
311
  {
312
  "name": "Carrier",
313
  "iso_1_code": null,
314
  "iso_3_code": null,
315
+ "tokenizers": {},
 
316
  "children": [
317
  {
318
  "name": "Carrier, Southern",
319
  "iso_1_code": null,
320
  "iso_3_code": "caf",
321
+ "tokenizers": {},
322
+ "children": [],
323
+ "node_i": "3808",
324
+ "scripts": [
325
+ "Latn"
326
+ ],
327
+ "own_tokenizer": false
328
  },
329
  {
330
  "name": "Carrier",
331
  "iso_1_code": null,
332
  "iso_3_code": "crx",
333
+ "tokenizers": {},
334
+ "children": [],
335
+ "node_i": "3809",
336
+ "scripts": [
337
+ "Latn"
338
+ ],
339
+ "own_tokenizer": false
340
  }
341
+ ],
342
+ "node_i": "3807",
343
+ "scripts": [],
344
+ "own_tokenizer": false
345
  },
346
  {
347
  "name": "Slavey-Hare",
348
  "iso_1_code": null,
349
  "iso_3_code": null,
350
+ "tokenizers": {},
 
351
  "children": [
352
  {
353
  "name": "Slavey, North",
354
  "iso_1_code": null,
355
  "iso_3_code": "scs",
356
+ "tokenizers": {},
357
+ "children": [],
358
+ "node_i": "3811",
359
+ "scripts": [],
360
+ "own_tokenizer": false
361
  },
362
  {
363
  "name": "Slavey, South",
364
  "iso_1_code": null,
365
  "iso_3_code": "xsl",
366
+ "tokenizers": {},
367
+ "children": [],
368
+ "node_i": "3812",
369
+ "scripts": [],
370
+ "own_tokenizer": false
371
  }
372
+ ],
373
+ "node_i": "3810",
374
+ "scripts": [],
375
+ "own_tokenizer": false
376
  },
377
  {
378
  "name": "Tahltan",
379
  "iso_1_code": null,
380
  "iso_3_code": null,
381
+ "tokenizers": {},
 
382
  "children": [
383
  {
384
  "name": "Kaska",
385
  "iso_1_code": null,
386
  "iso_3_code": "kkz",
387
+ "tokenizers": {},
388
+ "children": [],
389
+ "node_i": "3814",
390
+ "scripts": [],
391
+ "own_tokenizer": false
392
  },
393
  {
394
  "name": "Tagish",
395
  "iso_1_code": null,
396
  "iso_3_code": "tgx",
397
+ "tokenizers": {},
398
+ "children": [],
399
+ "node_i": "3815",
400
+ "scripts": [],
401
+ "own_tokenizer": false
402
  },
403
  {
404
  "name": "Tahltan",
405
  "iso_1_code": null,
406
  "iso_3_code": "tht",
407
+ "tokenizers": {},
408
+ "children": [],
409
+ "node_i": "3816",
410
+ "scripts": [],
411
+ "own_tokenizer": false
412
  }
413
+ ],
414
+ "node_i": "3813",
415
+ "scripts": [],
416
+ "own_tokenizer": false
417
  },
418
  {
419
  "name": "Tuchone",
420
  "iso_1_code": null,
421
  "iso_3_code": null,
422
+ "tokenizers": {},
 
423
  "children": [
424
  {
425
  "name": "Tutchone, Southern",
426
  "iso_1_code": null,
427
  "iso_3_code": "tce",
428
+ "tokenizers": {},
429
+ "children": [],
430
+ "node_i": "3818",
431
+ "scripts": [],
432
+ "own_tokenizer": false
433
  },
434
  {
435
  "name": "Tutchone, Northern",
436
  "iso_1_code": null,
437
  "iso_3_code": "ttm",
438
+ "tokenizers": {},
439
+ "children": [],
440
+ "node_i": "3819",
441
+ "scripts": [],
442
+ "own_tokenizer": false
443
  }
444
+ ],
445
+ "node_i": "3817",
446
+ "scripts": [],
447
+ "own_tokenizer": false
448
  }
449
+ ],
450
+ "node_i": "3787",
451
+ "scripts": [],
452
+ "own_tokenizer": false
453
  },
454
  {
455
  "name": "Pacific Coast Athabaskan",
456
  "iso_1_code": null,
457
  "iso_3_code": null,
458
+ "tokenizers": {},
 
459
  "children": [
460
  {
461
  "name": "Kwalhioqua-Tlatskanai",
462
  "iso_1_code": null,
463
  "iso_3_code": "qwt",
464
+ "tokenizers": {},
465
+ "children": [],
466
+ "node_i": "3821",
467
+ "scripts": [],
468
+ "own_tokenizer": false
469
  },
470
  {
471
  "name": "California Athabaskan",
472
  "iso_1_code": null,
473
  "iso_3_code": null,
474
+ "tokenizers": {},
 
475
  "children": [
476
  {
477
  "name": "Hupa",
478
  "iso_1_code": null,
479
  "iso_3_code": "hup",
480
+ "tokenizers": {},
481
+ "children": [],
482
+ "node_i": "3823",
483
+ "scripts": [],
484
+ "own_tokenizer": false
485
  },
486
  {
487
  "name": "Kato",
488
  "iso_1_code": null,
489
  "iso_3_code": "ktw",
490
+ "tokenizers": {},
491
+ "children": [],
492
+ "node_i": "3824",
493
+ "scripts": [],
494
+ "own_tokenizer": false
495
  },
496
  {
497
  "name": "Mattole",
498
  "iso_1_code": null,
499
  "iso_3_code": "mvb",
500
+ "tokenizers": {},
501
+ "children": [],
502
+ "node_i": "3825",
503
+ "scripts": [],
504
+ "own_tokenizer": false
505
  },
506
  {
507
  "name": "Wailaki",
508
  "iso_1_code": null,
509
  "iso_3_code": "wlk",
510
+ "tokenizers": {},
511
+ "children": [],
512
+ "node_i": "3826",
513
+ "scripts": [],
514
+ "own_tokenizer": false
515
  }
516
+ ],
517
+ "node_i": "3822",
518
+ "scripts": [],
519
+ "own_tokenizer": false
520
  },
521
  {
522
  "name": "Oregon Athabaskan",
523
  "iso_1_code": null,
524
  "iso_3_code": null,
525
+ "tokenizers": {},
 
526
  "children": [
527
  {
528
  "name": "Galice",
529
  "iso_1_code": null,
530
  "iso_3_code": "gce",
531
+ "tokenizers": {},
532
+ "children": [],
533
+ "node_i": "3828",
534
+ "scripts": [],
535
+ "own_tokenizer": false
536
  },
537
  {
538
  "name": "Upper Umpqua",
539
  "iso_1_code": null,
540
  "iso_3_code": "xup",
541
+ "tokenizers": {},
542
+ "children": [],
543
+ "node_i": "3829",
544
+ "scripts": [],
545
+ "own_tokenizer": false
546
  },
547
  {
548
  "name": "Tolowa-Chetco",
549
  "iso_1_code": null,
550
  "iso_3_code": null,
551
+ "tokenizers": {},
 
552
  "children": [
553
  {
554
  "name": "Chetco",
555
  "iso_1_code": null,
556
  "iso_3_code": "ctc",
557
+ "tokenizers": {},
558
+ "children": [],
559
+ "node_i": "3831",
560
+ "scripts": [],
561
+ "own_tokenizer": false
562
  },
563
  {
564
  "name": "Tolowa",
565
  "iso_1_code": null,
566
  "iso_3_code": "tol",
567
+ "tokenizers": {},
568
+ "children": [],
569
+ "node_i": "3832",
570
+ "scripts": [],
571
+ "own_tokenizer": false
572
  }
573
+ ],
574
+ "node_i": "3830",
575
+ "scripts": [],
576
+ "own_tokenizer": false
577
  },
578
  {
579
  "name": "Tututni-Chasta Costa-Coquille",
580
  "iso_1_code": null,
581
  "iso_3_code": null,
582
+ "tokenizers": {},
 
583
  "children": [
584
  {
585
  "name": "Coquille",
586
  "iso_1_code": null,
587
  "iso_3_code": "coq",
588
+ "tokenizers": {},
589
+ "children": [],
590
+ "node_i": "3834",
591
+ "scripts": [],
592
+ "own_tokenizer": false
593
  },
594
  {
595
  "name": "Tututni",
596
  "iso_1_code": null,
597
  "iso_3_code": "tuu",
598
+ "tokenizers": {},
599
+ "children": [],
600
+ "node_i": "3835",
601
+ "scripts": [],
602
+ "own_tokenizer": false
603
  }
604
+ ],
605
+ "node_i": "3833",
606
+ "scripts": [],
607
+ "own_tokenizer": false
608
  }
609
+ ],
610
+ "node_i": "3827",
611
+ "scripts": [],
612
+ "own_tokenizer": false
613
  }
614
+ ],
615
+ "node_i": "3820",
616
+ "scripts": [],
617
+ "own_tokenizer": false
618
  }
619
+ ],
620
+ "node_i": "3778",
621
+ "scripts": [],
622
+ "own_tokenizer": false
623
  },
624
  {
625
  "name": "Tlingit",
626
  "iso_1_code": null,
627
  "iso_3_code": null,
628
+ "tokenizers": {},
 
629
  "children": [
630
  {
631
  "name": "Tlingit",
632
  "iso_1_code": null,
633
  "iso_3_code": "tli",
634
+ "tokenizers": {},
635
+ "children": [],
636
+ "node_i": "3837",
637
+ "scripts": [],
638
+ "own_tokenizer": false
639
  }
640
+ ],
641
+ "node_i": "3836",
642
+ "scripts": [],
643
+ "own_tokenizer": false
644
  }
645
+ ],
646
+ "node_i": "3776",
647
+ "scripts": [],
648
+ "own_tokenizer": false
649
  }
data/Fas.json CHANGED
@@ -2,24 +2,30 @@
2
  "name": "Fas",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Baibai",
10
  "iso_1_code": null,
11
  "iso_3_code": "bbf",
12
- "tokenizer": null,
13
- "source": null,
14
- "children": []
 
 
15
  },
16
  {
17
  "name": "Momu",
18
  "iso_1_code": null,
19
  "iso_3_code": "fqs",
20
- "tokenizer": null,
21
- "source": null,
22
- "children": []
 
 
23
  }
24
- ]
 
 
 
25
  }
 
2
  "name": "Fas",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Baibai",
9
  "iso_1_code": null,
10
  "iso_3_code": "bbf",
11
+ "tokenizers": {},
12
+ "children": [],
13
+ "node_i": "3839",
14
+ "scripts": [],
15
+ "own_tokenizer": false
16
  },
17
  {
18
  "name": "Momu",
19
  "iso_1_code": null,
20
  "iso_3_code": "fqs",
21
+ "tokenizers": {},
22
+ "children": [],
23
+ "node_i": "3840",
24
+ "scripts": [],
25
+ "own_tokenizer": false
26
  }
27
+ ],
28
+ "node_i": "3838",
29
+ "scripts": [],
30
+ "own_tokenizer": false
31
  }
data/Guajiboan.json CHANGED
@@ -2,57 +2,77 @@
2
  "name": "Guajiboan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Cuiba",
10
  "iso_1_code": null,
11
  "iso_3_code": "cui",
12
- "tokenizer": null,
13
- "source": null,
14
- "children": []
 
 
 
 
15
  },
16
  {
17
  "name": "Guayabero",
18
  "iso_1_code": null,
19
  "iso_3_code": "guo",
20
- "tokenizer": null,
21
- "source": null,
22
- "children": []
 
 
 
 
23
  },
24
  {
25
  "name": "Guajibo",
26
  "iso_1_code": null,
27
  "iso_3_code": null,
28
- "tokenizer": null,
29
- "source": null,
30
  "children": [
31
  {
32
  "name": "Playero",
33
  "iso_1_code": null,
34
  "iso_3_code": "gob",
35
- "tokenizer": null,
36
- "source": null,
37
- "children": []
 
 
38
  },
39
  {
40
  "name": "Guahibo",
41
  "iso_1_code": null,
42
  "iso_3_code": "guh",
43
- "tokenizer": null,
44
- "source": null,
45
- "children": []
 
 
 
 
46
  },
47
  {
48
  "name": "Macagu\u00e1n",
49
  "iso_1_code": null,
50
  "iso_3_code": "mbn",
51
- "tokenizer": null,
52
- "source": null,
53
- "children": []
 
 
54
  }
55
- ]
 
 
 
56
  }
57
- ]
 
 
 
58
  }
 
2
  "name": "Guajiboan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Cuiba",
9
  "iso_1_code": null,
10
  "iso_3_code": "cui",
11
+ "tokenizers": {},
12
+ "children": [],
13
+ "node_i": "3842",
14
+ "scripts": [
15
+ "Latn"
16
+ ],
17
+ "own_tokenizer": false
18
  },
19
  {
20
  "name": "Guayabero",
21
  "iso_1_code": null,
22
  "iso_3_code": "guo",
23
+ "tokenizers": {},
24
+ "children": [],
25
+ "node_i": "3843",
26
+ "scripts": [
27
+ "Latn"
28
+ ],
29
+ "own_tokenizer": false
30
  },
31
  {
32
  "name": "Guajibo",
33
  "iso_1_code": null,
34
  "iso_3_code": null,
35
+ "tokenizers": {},
 
36
  "children": [
37
  {
38
  "name": "Playero",
39
  "iso_1_code": null,
40
  "iso_3_code": "gob",
41
+ "tokenizers": {},
42
+ "children": [],
43
+ "node_i": "3845",
44
+ "scripts": [],
45
+ "own_tokenizer": false
46
  },
47
  {
48
  "name": "Guahibo",
49
  "iso_1_code": null,
50
  "iso_3_code": "guh",
51
+ "tokenizers": {},
52
+ "children": [],
53
+ "node_i": "3846",
54
+ "scripts": [
55
+ "Latn"
56
+ ],
57
+ "own_tokenizer": false
58
  },
59
  {
60
  "name": "Macagu\u00e1n",
61
  "iso_1_code": null,
62
  "iso_3_code": "mbn",
63
+ "tokenizers": {},
64
+ "children": [],
65
+ "node_i": "3847",
66
+ "scripts": [],
67
+ "own_tokenizer": false
68
  }
69
+ ],
70
+ "node_i": "3844",
71
+ "scripts": [],
72
+ "own_tokenizer": false
73
  }
74
+ ],
75
+ "node_i": "3841",
76
+ "scripts": [],
77
+ "own_tokenizer": false
78
  }
data/Guaykuruan.json CHANGED
@@ -2,66 +2,90 @@
2
  "name": "Guaykuruan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Guaykur\u00fa",
10
  "iso_1_code": null,
11
  "iso_3_code": null,
12
- "tokenizer": null,
13
- "source": null,
14
  "children": [
15
  {
16
  "name": "Abipon",
17
  "iso_1_code": null,
18
  "iso_3_code": "axb",
19
- "tokenizer": null,
20
- "source": null,
21
- "children": []
 
 
22
  },
23
  {
24
  "name": "Kadiw\u00e9u",
25
  "iso_1_code": null,
26
  "iso_3_code": "kbc",
27
- "tokenizer": null,
28
- "source": null,
29
- "children": []
 
 
 
 
30
  }
31
- ]
 
 
 
32
  },
33
  {
34
  "name": "Southern",
35
  "iso_1_code": null,
36
  "iso_3_code": null,
37
- "tokenizer": null,
38
- "source": null,
39
  "children": [
40
  {
41
  "name": "Mocov\u00ed",
42
  "iso_1_code": null,
43
  "iso_3_code": "moc",
44
- "tokenizer": null,
45
- "source": null,
46
- "children": []
 
 
 
 
47
  },
48
  {
49
  "name": "Pilag\u00e1",
50
  "iso_1_code": null,
51
  "iso_3_code": "plg",
52
- "tokenizer": null,
53
- "source": null,
54
- "children": []
 
 
 
 
55
  },
56
  {
57
  "name": "Toba",
58
  "iso_1_code": null,
59
  "iso_3_code": "tob",
60
- "tokenizer": null,
61
- "source": null,
62
- "children": []
 
 
 
 
63
  }
64
- ]
 
 
 
65
  }
66
- ]
 
 
 
67
  }
 
2
  "name": "Guaykuruan",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Guaykur\u00fa",
9
  "iso_1_code": null,
10
  "iso_3_code": null,
11
+ "tokenizers": {},
 
12
  "children": [
13
  {
14
  "name": "Abipon",
15
  "iso_1_code": null,
16
  "iso_3_code": "axb",
17
+ "tokenizers": {},
18
+ "children": [],
19
+ "node_i": "3850",
20
+ "scripts": [],
21
+ "own_tokenizer": false
22
  },
23
  {
24
  "name": "Kadiw\u00e9u",
25
  "iso_1_code": null,
26
  "iso_3_code": "kbc",
27
+ "tokenizers": {},
28
+ "children": [],
29
+ "node_i": "3851",
30
+ "scripts": [
31
+ "Latn"
32
+ ],
33
+ "own_tokenizer": false
34
  }
35
+ ],
36
+ "node_i": "3849",
37
+ "scripts": [],
38
+ "own_tokenizer": false
39
  },
40
  {
41
  "name": "Southern",
42
  "iso_1_code": null,
43
  "iso_3_code": null,
44
+ "tokenizers": {},
 
45
  "children": [
46
  {
47
  "name": "Mocov\u00ed",
48
  "iso_1_code": null,
49
  "iso_3_code": "moc",
50
+ "tokenizers": {},
51
+ "children": [],
52
+ "node_i": "3853",
53
+ "scripts": [
54
+ "Latn"
55
+ ],
56
+ "own_tokenizer": false
57
  },
58
  {
59
  "name": "Pilag\u00e1",
60
  "iso_1_code": null,
61
  "iso_3_code": "plg",
62
+ "tokenizers": {},
63
+ "children": [],
64
+ "node_i": "3854",
65
+ "scripts": [
66
+ "Latn"
67
+ ],
68
+ "own_tokenizer": false
69
  },
70
  {
71
  "name": "Toba",
72
  "iso_1_code": null,
73
  "iso_3_code": "tob",
74
+ "tokenizers": {},
75
+ "children": [],
76
+ "node_i": "3855",
77
+ "scripts": [
78
+ "Latn"
79
+ ],
80
+ "own_tokenizer": false
81
  }
82
+ ],
83
+ "node_i": "3852",
84
+ "scripts": [],
85
+ "own_tokenizer": false
86
  }
87
+ ],
88
+ "node_i": "3848",
89
+ "scripts": [],
90
+ "own_tokenizer": false
91
  }
data/Gum.json CHANGED
@@ -2,7 +2,9 @@
2
  "name": "Gum",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
- "children": []
 
 
8
  }
 
2
  "name": "Gum",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
6
+ "children": [],
7
+ "node_i": "3856",
8
+ "scripts": [],
9
+ "own_tokenizer": false
10
  }
data/Haida.json CHANGED
@@ -2,24 +2,30 @@
2
  "name": "Haida",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Haida, Southern",
10
  "iso_1_code": null,
11
  "iso_3_code": "hax",
12
- "tokenizer": null,
13
- "source": null,
14
- "children": []
 
 
15
  },
16
  {
17
  "name": "Haida, Northern",
18
  "iso_1_code": null,
19
  "iso_3_code": "hdn",
20
- "tokenizer": null,
21
- "source": null,
22
- "children": []
 
 
23
  }
24
- ]
 
 
 
25
  }
 
2
  "name": "Haida",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Haida, Southern",
9
  "iso_1_code": null,
10
  "iso_3_code": "hax",
11
+ "tokenizers": {},
12
+ "children": [],
13
+ "node_i": "3858",
14
+ "scripts": [],
15
+ "own_tokenizer": false
16
  },
17
  {
18
  "name": "Haida, Northern",
19
  "iso_1_code": null,
20
  "iso_3_code": "hdn",
21
+ "tokenizers": {},
22
+ "children": [],
23
+ "node_i": "3859",
24
+ "scripts": [],
25
+ "own_tokenizer": false
26
  }
27
+ ],
28
+ "node_i": "3857",
29
+ "scripts": [],
30
+ "own_tokenizer": false
31
  }
data/Harákmbut.json CHANGED
@@ -2,24 +2,32 @@
2
  "name": "Har\u00e1kmbut",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Amarakaeri",
10
  "iso_1_code": null,
11
  "iso_3_code": "amr",
12
- "tokenizer": null,
13
- "source": null,
14
- "children": []
 
 
 
 
15
  },
16
  {
17
  "name": "Huachipaeri",
18
  "iso_1_code": null,
19
  "iso_3_code": "hug",
20
- "tokenizer": null,
21
- "source": null,
22
- "children": []
 
 
23
  }
24
- ]
 
 
 
25
  }
 
2
  "name": "Har\u00e1kmbut",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Amarakaeri",
9
  "iso_1_code": null,
10
  "iso_3_code": "amr",
11
+ "tokenizers": {},
12
+ "children": [],
13
+ "node_i": "3861",
14
+ "scripts": [
15
+ "Latn"
16
+ ],
17
+ "own_tokenizer": false
18
  },
19
  {
20
  "name": "Huachipaeri",
21
  "iso_1_code": null,
22
  "iso_3_code": "hug",
23
+ "tokenizers": {},
24
+ "children": [],
25
+ "node_i": "3862",
26
+ "scripts": [],
27
+ "own_tokenizer": false
28
  }
29
+ ],
30
+ "node_i": "3860",
31
+ "scripts": [],
32
+ "own_tokenizer": false
33
  }
data/Hmong-Mien.json CHANGED
@@ -2,419 +2,527 @@
2
  "name": "Hmong-Mien",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
- "tokenizer": null,
6
- "source": null,
7
  "children": [
8
  {
9
  "name": "Hmongic",
10
  "iso_1_code": null,
11
  "iso_3_code": null,
12
- "tokenizer": null,
13
- "source": null,
14
  "children": [
15
  {
16
  "name": "Bunu",
17
  "iso_1_code": null,
18
  "iso_3_code": null,
19
- "tokenizer": null,
20
- "source": null,
21
  "children": [
22
  {
23
  "name": "Bunu, Younuo",
24
  "iso_1_code": null,
25
  "iso_3_code": "buh",
26
- "tokenizer": null,
27
- "source": null,
28
- "children": []
 
 
29
  },
30
  {
31
  "name": "Bunu, Wunai",
32
  "iso_1_code": null,
33
  "iso_3_code": "bwn",
34
- "tokenizer": null,
35
- "source": null,
36
- "children": []
 
 
37
  },
38
  {
39
  "name": "Bunu, Bu-Nao",
40
  "iso_1_code": null,
41
  "iso_3_code": "bwx",
42
- "tokenizer": null,
43
- "source": null,
44
- "children": []
 
 
45
  },
46
  {
47
  "name": "Bunu, Jiongnai",
48
  "iso_1_code": null,
49
  "iso_3_code": "pnu",
50
- "tokenizer": null,
51
- "source": null,
52
- "children": []
 
 
53
  }
54
- ]
 
 
 
55
  },
56
  {
57
  "name": "Chuanqiandian",
58
  "iso_1_code": null,
59
  "iso_3_code": null,
60
- "tokenizer": null,
61
- "source": null,
62
  "children": [
63
  {
64
  "name": "Miao, Chuanqiandian Cluster",
65
  "iso_1_code": null,
66
  "iso_3_code": "cqd",
67
- "tokenizer": null,
68
- "source": null,
69
- "children": []
 
 
70
  },
71
  {
72
  "name": "Miao, Southern Mashan",
73
  "iso_1_code": null,
74
  "iso_3_code": "hma",
75
- "tokenizer": null,
76
- "source": null,
77
- "children": []
 
 
78
  },
79
  {
80
  "name": "Miao, Central Huishui",
81
  "iso_1_code": null,
82
  "iso_3_code": "hmc",
83
- "tokenizer": null,
84
- "source": null,
85
- "children": []
 
 
86
  },
87
  {
88
  "name": "Miao, Large Flowery",
89
  "iso_1_code": null,
90
  "iso_3_code": "hmd",
91
- "tokenizer": null,
92
- "source": null,
93
- "children": []
 
 
94
  },
95
  {
96
  "name": "Miao, Eastern Huishui",
97
  "iso_1_code": null,
98
  "iso_3_code": "hme",
99
- "tokenizer": null,
100
- "source": null,
101
- "children": []
 
 
102
  },
103
  {
104
  "name": "Hmong Don",
105
  "iso_1_code": null,
106
  "iso_3_code": "hmf",
107
- "tokenizer": null,
108
- "source": null,
109
- "children": []
 
 
110
  },
111
  {
112
  "name": "Miao, Southwestern Guiyang",
113
  "iso_1_code": null,
114
  "iso_3_code": "hmg",
115
- "tokenizer": null,
116
- "source": null,
117
- "children": []
 
 
118
  },
119
  {
120
  "name": "Miao, Southwestern Huishui",
121
  "iso_1_code": null,
122
  "iso_3_code": "hmh",
123
- "tokenizer": null,
124
- "source": null,
125
- "children": []
 
 
126
  },
127
  {
128
  "name": "Miao, Northern Huishui",
129
  "iso_1_code": null,
130
  "iso_3_code": "hmi",
131
- "tokenizer": null,
132
- "source": null,
133
- "children": []
 
 
134
  },
135
  {
136
  "name": "Ge",
137
  "iso_1_code": null,
138
  "iso_3_code": "hmj",
139
- "tokenizer": null,
140
- "source": null,
141
- "children": []
 
 
142
  },
143
  {
144
  "name": "Miao, Luopohe",
145
  "iso_1_code": null,
146
  "iso_3_code": "hml",
147
- "tokenizer": null,
148
- "source": null,
149
- "children": []
 
 
150
  },
151
  {
152
  "name": "Miao, Central Mashan",
153
  "iso_1_code": null,
154
  "iso_3_code": "hmm",
155
- "tokenizer": null,
156
- "source": null,
157
- "children": []
 
 
158
  },
159
  {
160
  "name": "Miao, Northern Mashan",
161
  "iso_1_code": null,
162
  "iso_3_code": "hmp",
163
- "tokenizer": null,
164
- "source": null,
165
- "children": []
 
 
166
  },
167
  {
168
  "name": "Hmong D\u00f4",
169
  "iso_1_code": null,
170
  "iso_3_code": "hmv",
171
- "tokenizer": null,
172
- "source": null,
173
- "children": []
 
 
174
  },
175
  {
176
  "name": "Miao, Western Mashan",
177
  "iso_1_code": null,
178
  "iso_3_code": "hmw",
179
- "tokenizer": null,
180
- "source": null,
181
- "children": []
 
 
182
  },
183
  {
184
  "name": "Miao, Southern Guiyang",
185
  "iso_1_code": null,
186
  "iso_3_code": "hmy",
187
- "tokenizer": null,
188
- "source": null,
189
- "children": []
 
 
190
  },
191
  {
192
  "name": "Sinicized Miao",
193
  "iso_1_code": null,
194
  "iso_3_code": "hmz",
195
- "tokenizer": null,
196
- "source": null,
197
- "children": []
 
 
198
  },
199
  {
200
  "name": "Hmong Njua",
201
  "iso_1_code": null,
202
  "iso_3_code": "hnj",
203
- "tokenizer": null,
204
- "source": null,
205
- "children": []
 
 
 
 
206
  },
207
  {
208
  "name": "Miao, Horned",
209
  "iso_1_code": null,
210
  "iso_3_code": "hrm",
211
- "tokenizer": null,
212
- "source": null,
213
- "children": []
 
 
214
  },
215
  {
216
  "name": "Miao, Northern Guiyang",
217
  "iso_1_code": null,
218
  "iso_3_code": "huj",
219
- "tokenizer": null,
220
- "source": null,
221
- "children": []
 
 
222
  },
223
  {
224
  "name": "Hmong Daw",
225
  "iso_1_code": null,
226
  "iso_3_code": "mww",
227
- "tokenizer": null,
228
- "source": null,
229
- "children": []
 
 
 
 
230
  },
231
  {
232
  "name": "Miao, Small Flowery",
233
  "iso_1_code": null,
234
  "iso_3_code": "sfm",
235
- "tokenizer": null,
236
- "source": null,
237
- "children": []
 
 
238
  }
239
- ]
 
 
 
240
  },
241
  {
242
  "name": "Pa-hng",
243
  "iso_1_code": null,
244
  "iso_3_code": null,
245
- "tokenizer": null,
246
- "source": null,
247
  "children": [
248
  {
249
  "name": "Pa-Hng",
250
  "iso_1_code": null,
251
  "iso_3_code": "pha",
252
- "tokenizer": null,
253
- "source": null,
254
- "children": []
 
 
255
  }
256
- ]
 
 
 
257
  },
258
  {
259
  "name": "Qiandong",
260
  "iso_1_code": null,
261
  "iso_3_code": null,
262
- "tokenizer": null,
263
- "source": null,
264
  "children": [
265
  {
266
  "name": "Miao, Northern Qiandong",
267
  "iso_1_code": null,
268
  "iso_3_code": "hea",
269
- "tokenizer": null,
270
- "source": null,
271
- "children": []
 
 
272
  },
273
  {
274
  "name": "Miao, Eastern Qiandong",
275
  "iso_1_code": null,
276
  "iso_3_code": "hmq",
277
- "tokenizer": null,
278
- "source": null,
279
- "children": []
 
 
280
  },
281
  {
282
  "name": "Miao, Southern Qiandong",
283
  "iso_1_code": null,
284
  "iso_3_code": "hms",
285
- "tokenizer": null,
286
- "source": null,
287
- "children": []
 
 
288
  },
289
  {
290
  "name": "N\u00e1-Meo",
291
  "iso_1_code": null,
292
  "iso_3_code": "neo",
293
- "tokenizer": null,
294
- "source": null,
295
- "children": []
 
 
296
  }
297
- ]
 
 
 
298
  },
299
  {
300
  "name": "Xiangxi",
301
  "iso_1_code": null,
302
  "iso_3_code": null,
303
- "tokenizer": null,
304
- "source": null,
305
  "children": [
306
  {
307
  "name": "Miao, Western Xiangxi",
308
  "iso_1_code": null,
309
  "iso_3_code": "mmr",
310
- "tokenizer": null,
311
- "source": null,
312
- "children": []
 
 
313
  },
314
  {
315
  "name": "Miao, Eastern Xiangxi",
316
  "iso_1_code": null,
317
  "iso_3_code": "muq",
318
- "tokenizer": null,
319
- "source": null,
320
- "children": []
 
 
321
  }
322
- ]
 
 
 
323
  }
324
- ]
 
 
 
325
  },
326
  {
327
  "name": "Ho Nte",
328
  "iso_1_code": null,
329
  "iso_3_code": null,
330
- "tokenizer": null,
331
- "source": null,
332
  "children": [
333
  {
334
  "name": "She",
335
  "iso_1_code": null,
336
  "iso_3_code": "shx",
337
- "tokenizer": null,
338
- "source": null,
339
- "children": []
 
 
340
  }
341
- ]
 
 
 
342
  },
343
  {
344
  "name": "Mienic",
345
  "iso_1_code": null,
346
  "iso_3_code": null,
347
- "tokenizer": null,
348
- "source": null,
349
  "children": [
350
  {
351
  "name": "Biao-Jiao",
352
  "iso_1_code": null,
353
  "iso_3_code": null,
354
- "tokenizer": null,
355
- "source": null,
356
  "children": [
357
  {
358
  "name": "Biao-Jiao Mien",
359
  "iso_1_code": null,
360
  "iso_3_code": "bje",
361
- "tokenizer": null,
362
- "source": null,
363
- "children": []
 
 
364
  }
365
- ]
 
 
 
366
  },
367
  {
368
  "name": "Mian-Jin",
369
  "iso_1_code": null,
370
  "iso_3_code": null,
371
- "tokenizer": null,
372
- "source": null,
373
  "children": [
374
  {
375
  "name": "Biao Mon",
376
  "iso_1_code": null,
377
  "iso_3_code": "bmt",
378
- "tokenizer": null,
379
- "source": null,
380
- "children": []
 
 
381
  },
382
  {
383
  "name": "Iu Mien",
384
  "iso_1_code": null,
385
  "iso_3_code": "ium",
386
- "tokenizer": null,
387
- "source": null,
388
- "children": []
 
 
 
 
389
  },
390
  {
391
  "name": "Kim Mun",
392
  "iso_1_code": null,
393
  "iso_3_code": "mji",
394
- "tokenizer": null,
395
- "source": null,
396
- "children": []
 
 
397
  }
398
- ]
 
 
 
399
  },
400
  {
401
  "name": "Zaomin",
402
  "iso_1_code": null,
403
  "iso_3_code": null,
404
- "tokenizer": null,
405
- "source": null,
406
  "children": [
407
  {
408
  "name": "Dzao Min",
409
  "iso_1_code": null,
410
  "iso_3_code": "bpn",
411
- "tokenizer": null,
412
- "source": null,
413
- "children": []
 
 
414
  }
415
- ]
 
 
 
416
  }
417
- ]
 
 
 
418
  }
419
- ]
 
 
 
420
  }
 
2
  "name": "Hmong-Mien",
3
  "iso_1_code": null,
4
  "iso_3_code": null,
5
+ "tokenizers": {},
 
6
  "children": [
7
  {
8
  "name": "Hmongic",
9
  "iso_1_code": null,
10
  "iso_3_code": null,
11
+ "tokenizers": {},
 
12
  "children": [
13
  {
14
  "name": "Bunu",
15
  "iso_1_code": null,
16
  "iso_3_code": null,
17
+ "tokenizers": {},
 
18
  "children": [
19
  {
20
  "name": "Bunu, Younuo",
21
  "iso_1_code": null,
22
  "iso_3_code": "buh",
23
+ "tokenizers": {},
24
+ "children": [],
25
+ "node_i": "3866",
26
+ "scripts": [],
27
+ "own_tokenizer": false
28
  },
29
  {
30
  "name": "Bunu, Wunai",
31
  "iso_1_code": null,
32
  "iso_3_code": "bwn",
33
+ "tokenizers": {},
34
+ "children": [],
35
+ "node_i": "3867",
36
+ "scripts": [],
37
+ "own_tokenizer": false
38
  },
39
  {
40
  "name": "Bunu, Bu-Nao",
41
  "iso_1_code": null,
42
  "iso_3_code": "bwx",
43
+ "tokenizers": {},
44
+ "children": [],
45
+ "node_i": "3868",
46
+ "scripts": [],
47
+ "own_tokenizer": false
48
  },
49
  {
50
  "name": "Bunu, Jiongnai",
51
  "iso_1_code": null,
52
  "iso_3_code": "pnu",
53
+ "tokenizers": {},
54
+ "children": [],
55
+ "node_i": "3869",
56
+ "scripts": [],
57
+ "own_tokenizer": false
58
  }
59
+ ],
60
+ "node_i": "3865",
61
+ "scripts": [],
62
+ "own_tokenizer": false
63
  },
64
  {
65
  "name": "Chuanqiandian",
66
  "iso_1_code": null,
67
  "iso_3_code": null,
68
+ "tokenizers": {},
 
69
  "children": [
70
  {
71
  "name": "Miao, Chuanqiandian Cluster",
72
  "iso_1_code": null,
73
  "iso_3_code": "cqd",
74
+ "tokenizers": {},
75
+ "children": [],
76
+ "node_i": "3871",
77
+ "scripts": [],
78
+ "own_tokenizer": false
79
  },
80
  {
81
  "name": "Miao, Southern Mashan",
82
  "iso_1_code": null,
83
  "iso_3_code": "hma",
84
+ "tokenizers": {},
85
+ "children": [],
86
+ "node_i": "3872",
87
+ "scripts": [],
88
+ "own_tokenizer": false
89
  },
90
  {
91
  "name": "Miao, Central Huishui",
92
  "iso_1_code": null,
93
  "iso_3_code": "hmc",
94
+ "tokenizers": {},
95
+ "children": [],
96
+ "node_i": "3873",
97
+ "scripts": [],
98
+ "own_tokenizer": false
99
  },
100
  {
101
  "name": "Miao, Large Flowery",
102
  "iso_1_code": null,
103
  "iso_3_code": "hmd",
104
+ "tokenizers": {},
105
+ "children": [],
106
+ "node_i": "3874",
107
+ "scripts": [],
108
+ "own_tokenizer": false
109
  },
110
  {
111
  "name": "Miao, Eastern Huishui",
112
  "iso_1_code": null,
113
  "iso_3_code": "hme",
114
+ "tokenizers": {},
115
+ "children": [],
116
+ "node_i": "3875",
117
+ "scripts": [],
118
+ "own_tokenizer": false
119
  },
120
  {
121
  "name": "Hmong Don",
122
  "iso_1_code": null,
123
  "iso_3_code": "hmf",
124
+ "tokenizers": {},
125
+ "children": [],
126
+ "node_i": "3876",
127
+ "scripts": [],
128
+ "own_tokenizer": false
129
  },
130
  {
131
  "name": "Miao, Southwestern Guiyang",
132
  "iso_1_code": null,
133
  "iso_3_code": "hmg",
134
+ "tokenizers": {},
135
+ "children": [],
136
+ "node_i": "3877",
137
+ "scripts": [],
138
+ "own_tokenizer": false
139
  },
140
  {
141
  "name": "Miao, Southwestern Huishui",
142
  "iso_1_code": null,
143
  "iso_3_code": "hmh",
144
+ "tokenizers": {},
145
+ "children": [],
146
+ "node_i": "3878",
147
+ "scripts": [],
148
+ "own_tokenizer": false
149
  },
150
  {
151
  "name": "Miao, Northern Huishui",
152
  "iso_1_code": null,
153
  "iso_3_code": "hmi",
154
+ "tokenizers": {},
155
+ "children": [],
156
+ "node_i": "3879",
157
+ "scripts": [],
158
+ "own_tokenizer": false
159
  },
160
  {
161
  "name": "Ge",
162
  "iso_1_code": null,
163
  "iso_3_code": "hmj",
164
+ "tokenizers": {},
165
+ "children": [],
166
+ "node_i": "3880",
167
+ "scripts": [],
168
+ "own_tokenizer": false
169
  },
170
  {
171
  "name": "Miao, Luopohe",
172
  "iso_1_code": null,
173
  "iso_3_code": "hml",
174
+ "tokenizers": {},
175
+ "children": [],
176
+ "node_i": "3881",
177
+ "scripts": [],
178
+ "own_tokenizer": false
179
  },
180
  {
181
  "name": "Miao, Central Mashan",
182
  "iso_1_code": null,
183
  "iso_3_code": "hmm",
184
+ "tokenizers": {},
185
+ "children": [],
186
+ "node_i": "3882",
187
+ "scripts": [],
188
+ "own_tokenizer": false
189
  },
190
  {
191
  "name": "Miao, Northern Mashan",
192
  "iso_1_code": null,
193
  "iso_3_code": "hmp",
194
+ "tokenizers": {},
195
+ "children": [],
196
+ "node_i": "3883",
197
+ "scripts": [],
198
+ "own_tokenizer": false
199
  },
200
  {
201
  "name": "Hmong D\u00f4",
202
  "iso_1_code": null,
203
  "iso_3_code": "hmv",
204
+ "tokenizers": {},
205
+ "children": [],
206
+ "node_i": "3884",
207
+ "scripts": [],
208
+ "own_tokenizer": false
209
  },
210
  {
211
  "name": "Miao, Western Mashan",
212
  "iso_1_code": null,
213
  "iso_3_code": "hmw",
214
+ "tokenizers": {},
215
+ "children": [],
216
+ "node_i": "3885",
217
+ "scripts": [],
218
+ "own_tokenizer": false
219
  },
220
  {
221
  "name": "Miao, Southern Guiyang",
222
  "iso_1_code": null,
223
  "iso_3_code": "hmy",
224
+ "tokenizers": {},
225
+ "children": [],
226
+ "node_i": "3886",
227
+ "scripts": [],
228
+ "own_tokenizer": false
229
  },
230
  {
231
  "name": "Sinicized Miao",
232
  "iso_1_code": null,
233
  "iso_3_code": "hmz",
234
+ "tokenizers": {},
235
+ "children": [],
236
+ "node_i": "3887",
237
+ "scripts": [],
238
+ "own_tokenizer": false
239
  },
240
  {
241
  "name": "Hmong Njua",
242
  "iso_1_code": null,
243
  "iso_3_code": "hnj",
244
+ "tokenizers": {},
245
+ "children": [],
246
+ "node_i": "3888",
247
+ "scripts": [
248
+ "Latn"
249
+ ],
250
+ "own_tokenizer": false
251
  },
252
  {
253
  "name": "Miao, Horned",
254
  "iso_1_code": null,
255
  "iso_3_code": "hrm",
256
+ "tokenizers": {},
257
+ "children": [],
258
+ "node_i": "3889",
259
+ "scripts": [],
260
+ "own_tokenizer": false
261
  },
262
  {
263
  "name": "Miao, Northern Guiyang",
264
  "iso_1_code": null,
265
  "iso_3_code": "huj",
266
+ "tokenizers": {},
267
+ "children": [],
268
+ "node_i": "3890",
269
+ "scripts": [],
270
+ "own_tokenizer": false
271
  },
272
  {
273
  "name": "Hmong Daw",
274
  "iso_1_code": null,
275
  "iso_3_code": "mww",
276
+ "tokenizers": {},
277
+ "children": [],
278
+ "node_i": "3891",
279
+ "scripts": [
280
+ "Latn"
281
+ ],
282
+ "own_tokenizer": false
283
  },
284
  {
285
  "name": "Miao, Small Flowery",
286
  "iso_1_code": null,
287
  "iso_3_code": "sfm",
288
+ "tokenizers": {},
289
+ "children": [],
290
+ "node_i": "3892",
291
+ "scripts": [],
292
+ "own_tokenizer": false
293
  }
294
+ ],
295
+ "node_i": "3870",
296
+ "scripts": [],
297
+ "own_tokenizer": false
298
  },
299
  {
300
  "name": "Pa-hng",
301
  "iso_1_code": null,
302
  "iso_3_code": null,
303
+ "tokenizers": {},
 
304
  "children": [
305
  {
306
  "name": "Pa-Hng",
307
  "iso_1_code": null,
308
  "iso_3_code": "pha",
309
+ "tokenizers": {},
310
+ "children": [],
311
+ "node_i": "3894",
312
+ "scripts": [],
313
+ "own_tokenizer": false
314
  }
315
+ ],
316
+ "node_i": "3893",
317
+ "scripts": [],
318
+ "own_tokenizer": false
319
  },
320
  {
321
  "name": "Qiandong",
322
  "iso_1_code": null,
323
  "iso_3_code": null,
324
+ "tokenizers": {},
 
325
  "children": [
326
  {
327
  "name": "Miao, Northern Qiandong",
328
  "iso_1_code": null,
329
  "iso_3_code": "hea",
330
+ "tokenizers": {},
331
+ "children": [],
332
+ "node_i": "3896",
333
+ "scripts": [],
334
+ "own_tokenizer": false
335
  },
336
  {
337
  "name": "Miao, Eastern Qiandong",
338
  "iso_1_code": null,
339
  "iso_3_code": "hmq",
340
+ "tokenizers": {},
341
+ "children": [],
342
+ "node_i": "3897",
343
+ "scripts": [],
344
+ "own_tokenizer": false
345
  },
346
  {
347
  "name": "Miao, Southern Qiandong",
348
  "iso_1_code": null,
349
  "iso_3_code": "hms",
350
+ "tokenizers": {},
351
+ "children": [],
352
+ "node_i": "3898",
353
+ "scripts": [],
354
+ "own_tokenizer": false
355
  },
356
  {
357
  "name": "N\u00e1-Meo",
358
  "iso_1_code": null,
359
  "iso_3_code": "neo",
360
+ "tokenizers": {},
361
+ "children": [],
362
+ "node_i": "3899",
363
+ "scripts": [],
364
+ "own_tokenizer": false
365
  }
366
+ ],
367
+ "node_i": "3895",
368
+ "scripts": [],
369
+ "own_tokenizer": false
370
  },
371
  {
372
  "name": "Xiangxi",
373
  "iso_1_code": null,
374
  "iso_3_code": null,
375
+ "tokenizers": {},
 
376
  "children": [
377
  {
378
  "name": "Miao, Western Xiangxi",
379
  "iso_1_code": null,
380
  "iso_3_code": "mmr",
381
+ "tokenizers": {},
382
+ "children": [],
383
+ "node_i": "3901",
384
+ "scripts": [],
385
+ "own_tokenizer": false
386
  },
387
  {
388
  "name": "Miao, Eastern Xiangxi",
389
  "iso_1_code": null,
390
  "iso_3_code": "muq",
391
+ "tokenizers": {},
392
+ "children": [],
393
+ "node_i": "3902",
394
+ "scripts": [],
395
+ "own_tokenizer": false
396
  }
397
+ ],
398
+ "node_i": "3900",
399
+ "scripts": [],
400
+ "own_tokenizer": false
401
  }
402
+ ],
403
+ "node_i": "3864",
404
+ "scripts": [],
405
+ "own_tokenizer": false
406
  },
407
  {
408
  "name": "Ho Nte",
409
  "iso_1_code": null,
410
  "iso_3_code": null,
411
+ "tokenizers": {},
 
412
  "children": [
413
  {
414
  "name": "She",
415
  "iso_1_code": null,
416
  "iso_3_code": "shx",
417
+ "tokenizers": {},
418
+ "children": [],
419
+ "node_i": "3904",
420
+ "scripts": [],
421
+ "own_tokenizer": false
422
  }
423
+ ],
424
+ "node_i": "3903",
425
+ "scripts": [],
426
+ "own_tokenizer": false
427
  },
428
  {
429
  "name": "Mienic",
430
  "iso_1_code": null,
431
  "iso_3_code": null,
432
+ "tokenizers": {},
 
433
  "children": [
434
  {
435
  "name": "Biao-Jiao",
436
  "iso_1_code": null,
437
  "iso_3_code": null,
438
+ "tokenizers": {},
 
439
  "children": [
440
  {
441
  "name": "Biao-Jiao Mien",
442
  "iso_1_code": null,
443
  "iso_3_code": "bje",
444
+ "tokenizers": {},
445
+ "children": [],
446
+ "node_i": "3907",
447
+ "scripts": [],
448
+ "own_tokenizer": false
449
  }
450
+ ],
451
+ "node_i": "3906",
452
+ "scripts": [],
453
+ "own_tokenizer": false
454
  },
455
  {
456
  "name": "Mian-Jin",
457
  "iso_1_code": null,
458
  "iso_3_code": null,
459
+ "tokenizers": {},
 
460
  "children": [
461
  {
462
  "name": "Biao Mon",
463
  "iso_1_code": null,
464
  "iso_3_code": "bmt",
465
+ "tokenizers": {},
466
+ "children": [],
467
+ "node_i": "3909",
468
+ "scripts": [],
469
+ "own_tokenizer": false
470
  },
471
  {
472
  "name": "Iu Mien",
473
  "iso_1_code": null,
474
  "iso_3_code": "ium",
475
+ "tokenizers": {},
476
+ "children": [],
477
+ "node_i": "3910",
478
+ "scripts": [
479
+ "Latn"
480
+ ],
481
+ "own_tokenizer": false
482
  },
483
  {
484
  "name": "Kim Mun",
485
  "iso_1_code": null,
486
  "iso_3_code": "mji",
487
+ "tokenizers": {},
488
+ "children": [],
489
+ "node_i": "3911",
490
+ "scripts": [],
491
+ "own_tokenizer": false
492
  }
493
+ ],
494
+ "node_i": "3908",
495
+ "scripts": [],
496
+ "own_tokenizer": false
497
  },
498
  {
499
  "name": "Zaomin",
500
  "iso_1_code": null,
501
  "iso_3_code": null,
502
+ "tokenizers": {},
 
503
  "children": [
504
  {
505
  "name": "Dzao Min",
506
  "iso_1_code": null,
507
  "iso_3_code": "bpn",
508
+ "tokenizers": {},
509
+ "children": [],
510
+ "node_i": "3913",
511
+ "scripts": [],
512
+ "own_tokenizer": false
513
  }
514
+ ],
515
+ "node_i": "3912",
516
+ "scripts": [],
517
+ "own_tokenizer": false
518
  }
519
+ ],
520
+ "node_i": "3905",
521
+ "scripts": [],
522
+ "own_tokenizer": false
523
  }
524
+ ],
525
+ "node_i": "3863",
526
+ "scripts": [],
527
+ "own_tokenizer": false
528
  }