@stll/anonymize-wasm 1.1.1 → 1.3.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (59) hide show
  1. package/dist/address-boundaries.mjs +113 -5
  2. package/dist/address-stop-keywords.mjs +137 -0
  3. package/dist/address-stop-keywords.mjs.map +1 -0
  4. package/dist/address-stopwords.mjs +81 -0
  5. package/dist/address-stopwords.mjs.map +1 -0
  6. package/dist/address-street-types.mjs +67 -4
  7. package/dist/clause-noun-heads.mjs +75 -0
  8. package/dist/clause-noun-heads.mjs.map +1 -0
  9. package/dist/coreference.es.mjs +22 -0
  10. package/dist/coreference.es.mjs.map +1 -0
  11. package/dist/coreference.fr.mjs +32 -0
  12. package/dist/coreference.fr.mjs.map +1 -0
  13. package/dist/coreference.it.mjs +27 -0
  14. package/dist/coreference.it.mjs.map +1 -0
  15. package/dist/coreference.pl.mjs +27 -0
  16. package/dist/coreference.pl.mjs.map +1 -0
  17. package/dist/coreference.pt-br.mjs +14 -0
  18. package/dist/coreference.pt-br.mjs.map +1 -0
  19. package/dist/coreference.sk.mjs +22 -5
  20. package/dist/generic-roles.mjs +179 -1
  21. package/dist/hotword-rules.mjs +7 -3
  22. package/dist/legal-forms.mjs +3 -0
  23. package/dist/legal-role-heads.cs.mjs +42 -0
  24. package/dist/legal-role-heads.cs.mjs.map +1 -0
  25. package/dist/legal-role-heads.de.mjs +33 -0
  26. package/dist/legal-role-heads.de.mjs.map +1 -0
  27. package/dist/legal-role-heads.en.mjs +37 -0
  28. package/dist/legal-role-heads.en.mjs.map +1 -0
  29. package/dist/legal-role-heads.es.mjs +54 -0
  30. package/dist/legal-role-heads.es.mjs.map +1 -0
  31. package/dist/legal-role-heads.fr.mjs +72 -0
  32. package/dist/legal-role-heads.fr.mjs.map +1 -0
  33. package/dist/legal-role-heads.it.mjs +68 -0
  34. package/dist/legal-role-heads.it.mjs.map +1 -0
  35. package/dist/legal-role-heads.pl.mjs +84 -0
  36. package/dist/legal-role-heads.pl.mjs.map +1 -0
  37. package/dist/legal-role-heads.pt-br.mjs +63 -0
  38. package/dist/legal-role-heads.pt-br.mjs.map +1 -0
  39. package/dist/legal-role-heads.sk.mjs +80 -0
  40. package/dist/legal-role-heads.sk.mjs.map +1 -0
  41. package/dist/manifest.mjs +21 -8
  42. package/dist/names-exclusions.mjs +21 -1
  43. package/dist/person-stopwords.mjs +121 -1
  44. package/dist/section-headings.mjs +15 -3
  45. package/dist/sentence-verb-indicators.mjs +223 -0
  46. package/dist/sentence-verb-indicators.mjs.map +1 -0
  47. package/dist/triggers.cs.mjs +107 -3
  48. package/dist/triggers.es.mjs +43 -4
  49. package/dist/triggers.fr.mjs +206 -7
  50. package/dist/triggers.it.mjs +26 -4
  51. package/dist/triggers.pl.mjs +191 -4
  52. package/dist/triggers.pt-br.mjs +193 -0
  53. package/dist/triggers.pt-br.mjs.map +1 -0
  54. package/dist/triggers.sk.mjs +493 -0
  55. package/dist/vite.mjs.map +1 -1
  56. package/dist/wasm.d.mts +100 -8
  57. package/dist/wasm.mjs +2520 -1360
  58. package/dist/wasm.mjs.map +1 -1
  59. package/package.json +2 -2
@@ -1,5 +1,357 @@
1
1
  //#region src/data/triggers.sk.json
2
2
  var triggers_sk_default = [
3
+ {
4
+ "id": "sk-person-to-next-comma",
5
+ "label": "person",
6
+ "strategy": { "type": "to-next-comma" },
7
+ "triggers": [
8
+ "zastúpený",
9
+ "zastúpená",
10
+ "zastúpené",
11
+ "zastúpeným",
12
+ "zastúpenou",
13
+ "zastúpení",
14
+ "konajúci",
15
+ "konajúca",
16
+ "konajúcim",
17
+ "v zastúpení",
18
+ "splnomocnený zástupca",
19
+ "splnomocnená zástupkyňa",
20
+ "štatutárny orgán",
21
+ "štatutárny zástupca",
22
+ "konateľ spoločnosti",
23
+ "konateľka spoločnosti",
24
+ "konateľom",
25
+ "konateľkou",
26
+ "prokurista",
27
+ "prokuristom",
28
+ "predseda predstavenstva",
29
+ "podpredseda predstavenstva",
30
+ "člen predstavenstva",
31
+ "členka predstavenstva",
32
+ "riaditeľ",
33
+ "riaditeľka",
34
+ "riaditeľom",
35
+ "riaditeľkou",
36
+ "kontaktná osoba"
37
+ ],
38
+ "extensions": ["add-colon"],
39
+ "validations": [{ "type": "starts-uppercase" }]
40
+ },
41
+ {
42
+ "id": "sk-person-trailing-space",
43
+ "label": "person",
44
+ "strategy": { "type": "to-next-comma" },
45
+ "triggers": [
46
+ "pán",
47
+ "pani",
48
+ "pána",
49
+ "panu",
50
+ "pánom",
51
+ "paniam"
52
+ ],
53
+ "extensions": ["add-trailing-space"],
54
+ "validations": [{ "type": "starts-uppercase" }]
55
+ },
56
+ {
57
+ "id": "sk-address-full",
58
+ "label": "address",
59
+ "strategy": {
60
+ "type": "address",
61
+ "maxChars": 120
62
+ },
63
+ "triggers": [
64
+ "sídlo",
65
+ "sídlom",
66
+ "so sídlom",
67
+ "miesto podnikania",
68
+ "miestom podnikania",
69
+ "trvalý pobyt",
70
+ "trvalým pobytom",
71
+ "bytom",
72
+ "adresa",
73
+ "adresa trvalého pobytu",
74
+ "doručovacia adresa",
75
+ "korešpondenčná adresa",
76
+ "fakturačná adresa",
77
+ "miesto dodania",
78
+ "miesto plnenia"
79
+ ],
80
+ "extensions": ["add-colon"]
81
+ },
82
+ {
83
+ "id": "sk-address-house-number",
84
+ "label": "address",
85
+ "strategy": {
86
+ "type": "n-words",
87
+ "count": 1
88
+ },
89
+ "triggers": [
90
+ "č.d.",
91
+ "súp. č.",
92
+ "súpisné číslo",
93
+ "or. č.",
94
+ "orientačné číslo"
95
+ ]
96
+ },
97
+ {
98
+ "id": "sk-address-psc",
99
+ "label": "address",
100
+ "strategy": {
101
+ "type": "n-words",
102
+ "count": 2
103
+ },
104
+ "triggers": ["PSČ"],
105
+ "extensions": ["add-colon", "add-trailing-space"]
106
+ },
107
+ {
108
+ "id": "sk-organization-roles",
109
+ "label": "organization",
110
+ "strategy": { "type": "to-next-comma" },
111
+ "triggers": [
112
+ "zhotoviteľ",
113
+ "zhotoviteľom",
114
+ "objednávateľ",
115
+ "objednávateľom",
116
+ "dodávateľ",
117
+ "dodávateľom",
118
+ "odberateľ",
119
+ "odberateľom",
120
+ "poskytovateľ",
121
+ "poskytovateľom",
122
+ "prijímateľ",
123
+ "prijímateľom"
124
+ ],
125
+ "extensions": ["add-colon"],
126
+ "validations": [{ "type": "starts-uppercase" }]
127
+ },
128
+ {
129
+ "id": "sk-organization-municipalities",
130
+ "label": "organization",
131
+ "strategy": { "type": "to-next-comma" },
132
+ "triggers": [
133
+ "mestská časť",
134
+ "mestskej časti",
135
+ "mestská polícia",
136
+ "hlavné mesto",
137
+ "hlavného mesta",
138
+ "obec",
139
+ "obce",
140
+ "obci"
141
+ ],
142
+ "extensions": ["add-trailing-space"],
143
+ "validations": [{ "type": "starts-uppercase" }]
144
+ },
145
+ {
146
+ "id": "sk-organization-trailing-space",
147
+ "label": "organization",
148
+ "strategy": { "type": "to-next-comma" },
149
+ "triggers": [
150
+ "mesto",
151
+ "mesta",
152
+ "mestom",
153
+ "mestu"
154
+ ],
155
+ "extensions": ["add-trailing-space"],
156
+ "validations": [{ "type": "starts-uppercase" }]
157
+ },
158
+ {
159
+ "id": "sk-organization-institutions",
160
+ "label": "organization",
161
+ "strategy": { "type": "to-next-comma" },
162
+ "triggers": [
163
+ "základná škola",
164
+ "základnej školy",
165
+ "stredná škola",
166
+ "strednej školy",
167
+ "materská škola",
168
+ "nemocnica",
169
+ "nemocnice",
170
+ "fakultná nemocnica",
171
+ "univerzita",
172
+ "univerzity",
173
+ "nadácia",
174
+ "príspevková organizácia",
175
+ "rozpočtová organizácia",
176
+ "občianske združenie"
177
+ ],
178
+ "extensions": ["add-trailing-space"],
179
+ "validations": [{ "type": "starts-uppercase" }]
180
+ },
181
+ {
182
+ "id": "sk-organization-fixed-name-institutions",
183
+ "label": "organization",
184
+ "strategy": { "type": "to-next-comma" },
185
+ "triggers": [
186
+ "Najvyšší kontrolný úrad",
187
+ "Slovenský pozemkový fond",
188
+ "Sociálna poisťovňa",
189
+ "Všeobecná zdravotná poisťovňa",
190
+ "Národná banka Slovenska"
191
+ ],
192
+ "includeTrigger": true,
193
+ "validations": [{
194
+ "type": "min-length",
195
+ "min": 3
196
+ }]
197
+ },
198
+ {
199
+ "id": "sk-date-of-birth",
200
+ "label": "date of birth",
201
+ "strategy": { "type": "to-next-comma" },
202
+ "triggers": [
203
+ "nar.",
204
+ "narodený dňa",
205
+ "narodená dňa",
206
+ "narodený/á dňa",
207
+ "dát. nar.",
208
+ "dátum narodenia"
209
+ ],
210
+ "extensions": ["add-colon"],
211
+ "validations": [{
212
+ "type": "max-length",
213
+ "max": 30
214
+ }]
215
+ },
216
+ {
217
+ "id": "sk-birth-number",
218
+ "label": "birth number",
219
+ "strategy": {
220
+ "type": "n-words",
221
+ "count": 1
222
+ },
223
+ "triggers": [
224
+ "r.č.",
225
+ "r. č.",
226
+ "rodné číslo"
227
+ ],
228
+ "extensions": ["add-colon"]
229
+ },
230
+ {
231
+ "id": "sk-bank-account-eol",
232
+ "label": "bank account number",
233
+ "strategy": { "type": "to-end-of-line" },
234
+ "triggers": ["bankové spojenie"],
235
+ "extensions": ["add-colon"],
236
+ "validations": [{ "type": "has-digits" }, {
237
+ "type": "min-length",
238
+ "min": 5
239
+ }]
240
+ },
241
+ {
242
+ "id": "sk-bank-account-nwords",
243
+ "label": "bank account number",
244
+ "strategy": {
245
+ "type": "n-words",
246
+ "count": 1
247
+ },
248
+ "triggers": [
249
+ "č. účtu",
250
+ "č.ú.",
251
+ "číslo účtu"
252
+ ],
253
+ "extensions": ["add-colon"],
254
+ "validations": [{ "type": "has-digits" }, {
255
+ "type": "min-length",
256
+ "min": 5
257
+ }]
258
+ },
259
+ {
260
+ "id": "sk-company-id",
261
+ "label": "registration number",
262
+ "strategy": { "type": "company-id-value" },
263
+ "triggers": [
264
+ "IČO",
265
+ "1ČO",
266
+ "lČO",
267
+ "identifikačné číslo",
268
+ "identifikačné číslo organizácie"
269
+ ]
270
+ },
271
+ {
272
+ "id": "sk-tax-id",
273
+ "label": "tax identification number",
274
+ "strategy": { "type": "company-id-value" },
275
+ "triggers": [
276
+ "DIČ",
277
+ "D1Č",
278
+ "daňové identifikačné číslo"
279
+ ]
280
+ },
281
+ {
282
+ "id": "sk-vat-id",
283
+ "label": "tax identification number",
284
+ "strategy": { "type": "company-id-value" },
285
+ "triggers": ["IČ DPH", "IČDPH"]
286
+ },
287
+ {
288
+ "id": "sk-vehicle-registration",
289
+ "label": "registration number",
290
+ "strategy": {
291
+ "type": "n-words",
292
+ "count": 1
293
+ },
294
+ "triggers": [
295
+ "EČV",
296
+ "ŠPZ",
297
+ "evidenčné číslo vozidla"
298
+ ],
299
+ "extensions": ["add-colon"]
300
+ },
301
+ {
302
+ "id": "sk-oddiel",
303
+ "label": "registration number",
304
+ "strategy": {
305
+ "type": "n-words",
306
+ "count": 1
307
+ },
308
+ "triggers": [
309
+ "oddiel",
310
+ "oddiele",
311
+ "oddielu",
312
+ "oddielom"
313
+ ],
314
+ "extensions": ["add-colon", "add-trailing-space"],
315
+ "validations": [{
316
+ "type": "matches-pattern",
317
+ "pattern": "^(?:Sa|Sro|Sr|Po|Dr|Pšn|Insre|Po-Re)[,.;:]?$"
318
+ }]
319
+ },
320
+ {
321
+ "id": "sk-vlozka",
322
+ "label": "registration number",
323
+ "strategy": {
324
+ "type": "n-words",
325
+ "count": 1
326
+ },
327
+ "triggers": [
328
+ "vložka číslo",
329
+ "vložka č.",
330
+ "vložky číslo",
331
+ "vložky č.",
332
+ "vložka",
333
+ "vložky",
334
+ "vložke",
335
+ "vložkou"
336
+ ],
337
+ "extensions": ["add-colon", "add-trailing-space"]
338
+ },
339
+ {
340
+ "id": "sk-sp-zn",
341
+ "label": "registration number",
342
+ "strategy": {
343
+ "type": "n-words",
344
+ "count": 2
345
+ },
346
+ "triggers": [
347
+ "sp. zn.",
348
+ "sp.zn.",
349
+ "spisovej značky",
350
+ "spisovou značkou",
351
+ "pod sp. zn.",
352
+ "pod sp.zn."
353
+ ]
354
+ },
3
355
  {
4
356
  "id": "sk-parcel-number",
5
357
  "label": "land parcel",
@@ -54,6 +406,147 @@ var triggers_sk_default = [
54
406
  "type": "min-length",
55
407
  "min": 3
56
408
  }]
409
+ },
410
+ {
411
+ "id": "sk-ministry",
412
+ "label": "organization",
413
+ "strategy": { "type": "to-next-comma" },
414
+ "triggers": [
415
+ "ministerstvo",
416
+ "ministerstva",
417
+ "ministerstvu",
418
+ "ministerstvom"
419
+ ],
420
+ "extensions": ["add-trailing-space"],
421
+ "includeTrigger": true,
422
+ "validations": [{
423
+ "type": "min-length",
424
+ "min": 6
425
+ }]
426
+ },
427
+ {
428
+ "id": "sk-courts-nominative",
429
+ "label": "organization",
430
+ "strategy": {
431
+ "type": "to-next-comma",
432
+ "stopWords": [
433
+ "dňa",
434
+ "oddiel",
435
+ "oddiele",
436
+ "vložka",
437
+ "vložky",
438
+ "sp.zn.",
439
+ "sp. zn.",
440
+ "pod"
441
+ ]
442
+ },
443
+ "triggers": [
444
+ "okresný súd",
445
+ "krajský súd",
446
+ "mestský súd",
447
+ "najvyšší súd",
448
+ "ústavný súd",
449
+ "špecializovaný trestný súd"
450
+ ],
451
+ "includeTrigger": true,
452
+ "extensions": [],
453
+ "validations": [{
454
+ "type": "min-length",
455
+ "min": 3
456
+ }]
457
+ },
458
+ {
459
+ "id": "sk-courts-genitive",
460
+ "label": "organization",
461
+ "strategy": {
462
+ "type": "to-next-comma",
463
+ "stopWords": [
464
+ "dňa",
465
+ "oddiel",
466
+ "oddiele",
467
+ "vložka",
468
+ "vložky",
469
+ "sp.zn.",
470
+ "sp. zn.",
471
+ "pod"
472
+ ]
473
+ },
474
+ "triggers": [
475
+ "okresného súdu",
476
+ "krajského súdu",
477
+ "mestského súdu",
478
+ "najvyššieho súdu",
479
+ "ústavného súdu",
480
+ "špecializovaného trestného súdu"
481
+ ],
482
+ "includeTrigger": true,
483
+ "extensions": [],
484
+ "validations": [{
485
+ "type": "min-length",
486
+ "min": 3
487
+ }]
488
+ },
489
+ {
490
+ "id": "sk-courts-instrumental",
491
+ "label": "organization",
492
+ "strategy": {
493
+ "type": "to-next-comma",
494
+ "stopWords": [
495
+ "dňa",
496
+ "oddiel",
497
+ "oddiele",
498
+ "vložka",
499
+ "vložky",
500
+ "sp.zn.",
501
+ "sp. zn.",
502
+ "pod"
503
+ ]
504
+ },
505
+ "triggers": [
506
+ "okresným súdom",
507
+ "krajským súdom",
508
+ "mestským súdom",
509
+ "najvyšším súdom",
510
+ "ústavným súdom",
511
+ "špecializovaným trestným súdom"
512
+ ],
513
+ "includeTrigger": true,
514
+ "extensions": [],
515
+ "validations": [{
516
+ "type": "min-length",
517
+ "min": 3
518
+ }]
519
+ },
520
+ {
521
+ "id": "sk-courts-locative",
522
+ "label": "organization",
523
+ "strategy": {
524
+ "type": "to-next-comma",
525
+ "stopWords": [
526
+ "dňa",
527
+ "oddiel",
528
+ "oddiele",
529
+ "vložka",
530
+ "vložky",
531
+ "sp.zn.",
532
+ "sp. zn.",
533
+ "pod"
534
+ ]
535
+ },
536
+ "triggers": [
537
+ "okresnom súde",
538
+ "krajskom súde",
539
+ "mestskom súde",
540
+ "najvyššom súde",
541
+ "ústavnom súde",
542
+ "špecializovanom trestnom súde"
543
+ ],
544
+ "includeTrigger": true,
545
+ "extensions": [],
546
+ "validations": [{
547
+ "type": "min-length",
548
+ "min": 3
549
+ }]
57
550
  }
58
551
  ];
59
552
  //#endregion
package/dist/vite.mjs.map CHANGED
@@ -1 +1 @@
1
- {"version":3,"file":"vite.mjs","names":[],"sources":["../../src/vite.ts"],"sourcesContent":["/* Vite plugin that wires up @stll/anonymize-wasm so its transitive\n * napi-rs-generated wasm loaders survive Vite's dep pre-bundler. Adds\n * this package plus its @stll/*-wasm dependencies to optimizeDeps.exclude\n * so the loader modules are served with their original paths, keeping\n * `new URL(\"./foo.wasm\", import.meta.url)` correct. */\nimport type { Plugin } from \"vite\";\n\n// Only the napi-rs WASM packages need excluding — they use\n// `new URL(\"./foo.wasm\", import.meta.url)` which Vite's optimizer\n// would rewrite. @stll/anonymize-wasm itself is pure JS and should\n// NOT be excluded.\nconst PACKAGES = [\n \"@stll/text-search-wasm\",\n \"@stll/aho-corasick-wasm\",\n \"@stll/aho-corasick-wasm32-wasi\",\n \"@stll/fuzzy-search-wasm\",\n \"@stll/fuzzy-search-wasm32-wasi\",\n \"@stll/regex-set-wasm\",\n \"@stll/regex-set-wasm32-wasi\",\n];\n\nexport default function stllAnonymizeWasmVite(): Plugin {\n return {\n name: \"stll-anonymize-wasm\",\n apply: \"serve\",\n config() {\n return {\n optimizeDeps: { exclude: PACKAGES },\n };\n },\n };\n}\n"],"mappings":";AAWA,MAAM,WAAW;CACf;CACA;CACA;CACA;CACA;CACA;CACA;CACD;AAED,SAAwB,wBAAgC;AACtD,QAAO;EACL,MAAM;EACN,OAAO;EACP,SAAS;AACP,UAAO,EACL,cAAc,EAAE,SAAS,UAAU,EACpC;;EAEJ"}
1
+ {"version":3,"file":"vite.mjs","names":[],"sources":["../../src/vite.ts"],"sourcesContent":["/* Vite plugin that wires up @stll/anonymize-wasm so its transitive\n * napi-rs-generated wasm loaders survive Vite's dep pre-bundler. Adds\n * this package plus its @stll/*-wasm dependencies to optimizeDeps.exclude\n * so the loader modules are served with their original paths, keeping\n * `new URL(\"./foo.wasm\", import.meta.url)` correct. */\nimport type { Plugin } from \"vite\";\n\n// Only the napi-rs WASM packages need excluding — they use\n// `new URL(\"./foo.wasm\", import.meta.url)` which Vite's optimizer\n// would rewrite. @stll/anonymize-wasm itself is pure JS and should\n// NOT be excluded.\nconst PACKAGES = [\n \"@stll/text-search-wasm\",\n \"@stll/aho-corasick-wasm\",\n \"@stll/aho-corasick-wasm32-wasi\",\n \"@stll/fuzzy-search-wasm\",\n \"@stll/fuzzy-search-wasm32-wasi\",\n \"@stll/regex-set-wasm\",\n \"@stll/regex-set-wasm32-wasi\",\n];\n\nexport default function stllAnonymizeWasmVite(): Plugin {\n return {\n name: \"stll-anonymize-wasm\",\n apply: \"serve\",\n config() {\n return {\n optimizeDeps: { exclude: PACKAGES },\n };\n },\n };\n}\n"],"mappings":";AAWA,MAAM,WAAW;CACf;CACA;CACA;CACA;CACA;CACA;CACA;CACD;AAED,SAAwB,wBAAgC;CACtD,OAAO;EACL,MAAM;EACN,OAAO;EACP,SAAS;GACP,OAAO,EACL,cAAc,EAAE,SAAS,UAAU,EACpC;;EAEJ"}