@stll/anonymize-wasm 1.0.8 → 1.1.0

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 (78) hide show
  1. package/dist/address-boundaries.mjs +85 -0
  2. package/dist/address-boundaries.mjs.map +1 -0
  3. package/dist/address-prepositions.mjs +182 -0
  4. package/dist/address-prepositions.mjs.map +1 -0
  5. package/dist/address-street-types.mjs +162 -0
  6. package/dist/address-street-types.mjs.map +1 -0
  7. package/dist/allow-list.mjs +183 -0
  8. package/dist/allow-list.mjs.map +1 -0
  9. package/dist/amount-words.mjs +42 -0
  10. package/dist/amount-words.mjs.map +1 -0
  11. package/dist/coreference.cs.mjs +14 -0
  12. package/dist/coreference.cs.mjs.map +1 -0
  13. package/dist/coreference.de.mjs +14 -0
  14. package/dist/coreference.de.mjs.map +1 -0
  15. package/dist/coreference.en.mjs +14 -0
  16. package/dist/coreference.en.mjs.map +1 -0
  17. package/dist/coreference.sk.mjs +10 -0
  18. package/dist/coreference.sk.mjs.map +1 -0
  19. package/dist/currencies.mjs +231 -0
  20. package/dist/currencies.mjs.map +1 -0
  21. package/dist/date-months.mjs +618 -0
  22. package/dist/date-months.mjs.map +1 -0
  23. package/dist/generic-roles.mjs +65 -0
  24. package/dist/generic-roles.mjs.map +1 -0
  25. package/dist/hotword-rules.mjs +145 -0
  26. package/dist/hotword-rules.mjs.map +1 -0
  27. package/dist/legal-forms.mjs +2112 -0
  28. package/dist/legal-forms.mjs.map +1 -0
  29. package/dist/manifest.mjs +56 -0
  30. package/dist/manifest.mjs.map +1 -0
  31. package/dist/names-exclusions.mjs +191 -0
  32. package/dist/names-exclusions.mjs.map +1 -0
  33. package/dist/names-first.mjs +418 -0
  34. package/dist/names-first.mjs.map +1 -0
  35. package/dist/names-surnames.mjs +113 -0
  36. package/dist/names-surnames.mjs.map +1 -0
  37. package/dist/names-title-tokens.mjs +40 -0
  38. package/dist/names-title-tokens.mjs.map +1 -0
  39. package/dist/person-stopwords.mjs +69 -0
  40. package/dist/person-stopwords.mjs.map +1 -0
  41. package/dist/section-headings.mjs +52 -0
  42. package/dist/section-headings.mjs.map +1 -0
  43. package/dist/signing-clauses.mjs +78 -0
  44. package/dist/signing-clauses.mjs.map +1 -0
  45. package/dist/stopwords.mjs +9915 -0
  46. package/dist/stopwords.mjs.map +1 -0
  47. package/dist/triggers.cs.mjs +465 -0
  48. package/dist/triggers.cs.mjs.map +1 -0
  49. package/dist/triggers.de.mjs +139 -0
  50. package/dist/triggers.de.mjs.map +1 -0
  51. package/dist/triggers.en.mjs +72 -0
  52. package/dist/triggers.en.mjs.map +1 -0
  53. package/dist/triggers.es.mjs +57 -0
  54. package/dist/triggers.es.mjs.map +1 -0
  55. package/dist/triggers.fr.mjs +76 -0
  56. package/dist/triggers.fr.mjs.map +1 -0
  57. package/dist/triggers.global.mjs +34 -0
  58. package/dist/triggers.global.mjs.map +1 -0
  59. package/dist/triggers.hu.mjs +41 -0
  60. package/dist/triggers.hu.mjs.map +1 -0
  61. package/dist/triggers.it.mjs +52 -0
  62. package/dist/triggers.it.mjs.map +1 -0
  63. package/dist/triggers.pl.mjs +84 -0
  64. package/dist/triggers.pl.mjs.map +1 -0
  65. package/dist/triggers.ro.mjs +59 -0
  66. package/dist/triggers.ro.mjs.map +1 -0
  67. package/dist/triggers.sk.mjs +62 -0
  68. package/dist/triggers.sk.mjs.map +1 -0
  69. package/dist/triggers.sv.mjs +58 -0
  70. package/dist/triggers.sv.mjs.map +1 -0
  71. package/dist/vite.mjs +4 -2
  72. package/dist/vite.mjs.map +1 -1
  73. package/dist/wasm.d.mts +75 -11
  74. package/dist/wasm.mjs +635 -180
  75. package/dist/wasm.mjs.map +1 -1
  76. package/dist/year-words.mjs +62 -0
  77. package/dist/year-words.mjs.map +1 -0
  78. package/package.json +1 -5
@@ -0,0 +1 @@
1
+ {"version":3,"file":"legal-forms.mjs","names":[],"sources":["../../src/data/legal-forms.json"],"sourcesContent":[""],"mappings":""}
@@ -0,0 +1,56 @@
1
+ //#region src/data/manifest.json
2
+ var _comment = "Language manifest. Drives auto-discovery of config files. Adding a new language requires: (1) add JSON config files, (2) update this manifest, (3) add a static import entry in lang-loader.ts.";
3
+ var languages = {
4
+ "cs": {
5
+ "triggers": true,
6
+ "coreference": true
7
+ },
8
+ "de": {
9
+ "triggers": true,
10
+ "coreference": true
11
+ },
12
+ "en": {
13
+ "triggers": true,
14
+ "coreference": true
15
+ },
16
+ "es": {
17
+ "triggers": true,
18
+ "coreference": false
19
+ },
20
+ "fr": {
21
+ "triggers": true,
22
+ "coreference": false
23
+ },
24
+ "hu": {
25
+ "triggers": true,
26
+ "coreference": false
27
+ },
28
+ "it": {
29
+ "triggers": true,
30
+ "coreference": false
31
+ },
32
+ "pl": {
33
+ "triggers": true,
34
+ "coreference": false
35
+ },
36
+ "ro": {
37
+ "triggers": true,
38
+ "coreference": false
39
+ },
40
+ "sk": {
41
+ "triggers": true,
42
+ "coreference": true
43
+ },
44
+ "sv": {
45
+ "triggers": true,
46
+ "coreference": false
47
+ }
48
+ };
49
+ var manifest_default = {
50
+ _comment,
51
+ languages
52
+ };
53
+ //#endregion
54
+ export { _comment, manifest_default as default, languages };
55
+
56
+ //# sourceMappingURL=manifest.mjs.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"manifest.mjs","names":[],"sources":["../../src/data/manifest.json"],"sourcesContent":[""],"mappings":""}
@@ -0,0 +1,191 @@
1
+ //#region src/data/names-exclusions.json
2
+ var _comment = "Words that look like names but are common legal/business terms. Used to suppress false positives in name detection.";
3
+ var words = [
4
+ "above",
5
+ "advocate",
6
+ "after",
7
+ "agentura",
8
+ "agreement",
9
+ "akce",
10
+ "annex",
11
+ "appendix",
12
+ "applicant",
13
+ "arbeitgeber",
14
+ "arbeitnehmer",
15
+ "article",
16
+ "assembly",
17
+ "auftraggeber",
18
+ "auftragnehmer",
19
+ "balíček",
20
+ "before",
21
+ "below",
22
+ "between",
23
+ "bonus",
24
+ "borrower",
25
+ "buyer",
26
+ "cena",
27
+ "centrum",
28
+ "chamber",
29
+ "claimant",
30
+ "client",
31
+ "commission",
32
+ "company",
33
+ "contract",
34
+ "contractor",
35
+ "council",
36
+ "court",
37
+ "customer",
38
+ "darlehensgeber",
39
+ "darlehensnehmer",
40
+ "defendant",
41
+ "division",
42
+ "divize",
43
+ "dlužník",
44
+ "dodávka",
45
+ "dohoda",
46
+ "during",
47
+ "email",
48
+ "employee",
49
+ "employer",
50
+ "exhibit",
51
+ "faktura",
52
+ "fax",
53
+ "firma",
54
+ "freedom",
55
+ "from",
56
+ "gesellschaft",
57
+ "government",
58
+ "guarantor",
59
+ "hereby",
60
+ "herein",
61
+ "http",
62
+ "https",
63
+ "including",
64
+ "institut",
65
+ "kancelář",
66
+ "komise",
67
+ "komisi",
68
+ "komisí",
69
+ "kraj",
70
+ "kraje",
71
+ "krajem",
72
+ "kredit",
73
+ "kupující",
74
+ "käufer",
75
+ "landlord",
76
+ "lender",
77
+ "lhůta",
78
+ "licensee",
79
+ "licensor",
80
+ "mieter",
81
+ "ministerstva",
82
+ "ministerstvo",
83
+ "ministry",
84
+ "mobil",
85
+ "nabídka",
86
+ "nájemce",
87
+ "objednatel",
88
+ "objednávka",
89
+ "odbor",
90
+ "oddělení",
91
+ "over",
92
+ "parlament",
93
+ "parlamentu",
94
+ "parliament",
95
+ "parties",
96
+ "party",
97
+ "plaintiff",
98
+ "platba",
99
+ "pobočka",
100
+ "podnik",
101
+ "police",
102
+ "prodávající",
103
+ "program",
104
+ "pronajímatel",
105
+ "prosecutor",
106
+ "provided",
107
+ "provozovna",
108
+ "pursuant",
109
+ "předmět",
110
+ "rada",
111
+ "radou",
112
+ "rady",
113
+ "radě",
114
+ "rector",
115
+ "registrar",
116
+ "republic",
117
+ "republika",
118
+ "republiky",
119
+ "respondent",
120
+ "schedule",
121
+ "section",
122
+ "sekce",
123
+ "seller",
124
+ "senát",
125
+ "senátu",
126
+ "since",
127
+ "sleva",
128
+ "služba",
129
+ "smlouva",
130
+ "smluvní",
131
+ "sněmovna",
132
+ "sněmovny",
133
+ "soud",
134
+ "soudem",
135
+ "soudu",
136
+ "společnost",
137
+ "správa",
138
+ "správou",
139
+ "správy",
140
+ "state",
141
+ "strana",
142
+ "strany",
143
+ "středisko",
144
+ "subject",
145
+ "supplier",
146
+ "tarif",
147
+ "tel",
148
+ "telefon",
149
+ "tenant",
150
+ "termín",
151
+ "that",
152
+ "the",
153
+ "therefore",
154
+ "thereof",
155
+ "this",
156
+ "tribunal",
157
+ "under",
158
+ "until",
159
+ "upon",
160
+ "url",
161
+ "vendor",
162
+ "verkäufer",
163
+ "vermieter",
164
+ "vláda",
165
+ "vládou",
166
+ "vlády",
167
+ "věřitel",
168
+ "web",
169
+ "whereas",
170
+ "with",
171
+ "www",
172
+ "zaměstnanec",
173
+ "zaměstnavatel",
174
+ "zhotovitel",
175
+ "záloha",
176
+ "závod",
177
+ "útvar",
178
+ "úřad",
179
+ "úřadem",
180
+ "úřadu",
181
+ "článek",
182
+ "částka"
183
+ ];
184
+ var names_exclusions_default = {
185
+ _comment,
186
+ words
187
+ };
188
+ //#endregion
189
+ export { _comment, names_exclusions_default as default, words };
190
+
191
+ //# sourceMappingURL=names-exclusions.mjs.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"names-exclusions.mjs","names":[],"sources":["../../src/data/names-exclusions.json"],"sourcesContent":[""],"mappings":""}
@@ -0,0 +1,418 @@
1
+ //#region src/data/names-first.json
2
+ var _comment = "First names for person detection. Merged from Czech/Slovak and international sources. Sorted alphabetically, deduplicated.";
3
+ var names = [
4
+ "Adam",
5
+ "Adrián",
6
+ "Adéla",
7
+ "Agnieszka",
8
+ "Ahmed",
9
+ "Ahmet",
10
+ "Alain",
11
+ "Alena",
12
+ "Alexander",
13
+ "Aleš",
14
+ "Ali",
15
+ "Alice",
16
+ "Alžběta",
17
+ "Amanda",
18
+ "Andrea",
19
+ "Andreas",
20
+ "Andrew",
21
+ "Andrzej",
22
+ "André",
23
+ "Anežka",
24
+ "Angelika",
25
+ "Anna",
26
+ "Anne",
27
+ "Anthony",
28
+ "Antonio",
29
+ "Antonín",
30
+ "Arnošt",
31
+ "Ayşe",
32
+ "Barbara",
33
+ "Barbora",
34
+ "Beata",
35
+ "Bedřich",
36
+ "Bedřiška",
37
+ "Benjamin",
38
+ "Bernard",
39
+ "Bernd",
40
+ "Bernhard",
41
+ "Birgit",
42
+ "Blanka",
43
+ "Blažej",
44
+ "Blažena",
45
+ "Bohdan",
46
+ "Bohumil",
47
+ "Bohumila",
48
+ "Bořivoj",
49
+ "Božena",
50
+ "Branislav",
51
+ "Brigitte",
52
+ "Bronislav",
53
+ "Břetislav",
54
+ "Carlos",
55
+ "Carmen",
56
+ "Caroline",
57
+ "Catherine",
58
+ "Cecílie",
59
+ "Charles",
60
+ "Charlotte",
61
+ "Christa",
62
+ "Christina",
63
+ "Christine",
64
+ "Christophe",
65
+ "Christopher",
66
+ "Claude",
67
+ "Claudia",
68
+ "Ctibor",
69
+ "Dagmar",
70
+ "Dalibor",
71
+ "Dana",
72
+ "Daniel",
73
+ "Daniela",
74
+ "Darina",
75
+ "David",
76
+ "Denisa",
77
+ "Detlef",
78
+ "Dieter",
79
+ "Dobromil",
80
+ "Dobromila",
81
+ "Dolores",
82
+ "Dominik",
83
+ "Dominique",
84
+ "Dorota",
85
+ "Drahomíra",
86
+ "Dušan",
87
+ "Dáša",
88
+ "Edita",
89
+ "Eduard",
90
+ "Edward",
91
+ "Elfriede",
92
+ "Elif",
93
+ "Elizabeth",
94
+ "Eliška",
95
+ "Emil",
96
+ "Emilie",
97
+ "Emily",
98
+ "Emine",
99
+ "Emma",
100
+ "Erika",
101
+ "Ervín",
102
+ "Eva",
103
+ "Evžen",
104
+ "Ewa",
105
+ "Fatima",
106
+ "Fatma",
107
+ "Fernando",
108
+ "Filip",
109
+ "Florián",
110
+ "František",
111
+ "Františka",
112
+ "François",
113
+ "Françoise",
114
+ "Friedrich",
115
+ "Gabriela",
116
+ "Gabriele",
117
+ "George",
118
+ "Gerhard",
119
+ "Gertruda",
120
+ "Giovanni",
121
+ "Gisela",
122
+ "Giuseppe",
123
+ "Grace",
124
+ "Grażyna",
125
+ "Grzegorz",
126
+ "Gustav",
127
+ "Gérard",
128
+ "Günter",
129
+ "Hacer",
130
+ "Hana",
131
+ "Hannah",
132
+ "Hans",
133
+ "Hanuš",
134
+ "Hartmut",
135
+ "Hasan",
136
+ "Hassan",
137
+ "Hatice",
138
+ "Hedvika",
139
+ "Heike",
140
+ "Heinrich",
141
+ "Helen",
142
+ "Helena",
143
+ "Helga",
144
+ "Helmut",
145
+ "Henry",
146
+ "Herbert",
147
+ "Hermína",
148
+ "Hildegard",
149
+ "Horst",
150
+ "Hubert",
151
+ "Hynek",
152
+ "Hüseyin",
153
+ "Ibrahim",
154
+ "Igor",
155
+ "Ilja",
156
+ "Ingrid",
157
+ "Irena",
158
+ "Isabelle",
159
+ "Ivan",
160
+ "Ivana",
161
+ "Iveta",
162
+ "Ivo",
163
+ "Izabela",
164
+ "Jacques",
165
+ "Jakub",
166
+ "James",
167
+ "Jan",
168
+ "Jana",
169
+ "Jane",
170
+ "Janusz",
171
+ "Jarmila",
172
+ "Jaroslav",
173
+ "Jaroslava",
174
+ "Jean",
175
+ "Jennifer",
176
+ "Jeroným",
177
+ "Jessica",
178
+ "Jindřich",
179
+ "Jitka",
180
+ "Jiřina",
181
+ "Jiří",
182
+ "Joanna",
183
+ "John",
184
+ "Josef",
185
+ "Josefa",
186
+ "Joseph",
187
+ "Jozef",
188
+ "João",
189
+ "Julia",
190
+ "Jáchym",
191
+ "Ján",
192
+ "Jürgen",
193
+ "Kamil",
194
+ "Karel",
195
+ "Karin",
196
+ "Karol",
197
+ "Karolína",
198
+ "Katarzyna",
199
+ "Katarína",
200
+ "Kateřina",
201
+ "Khalid",
202
+ "Klaus",
203
+ "Klára",
204
+ "Kornel",
205
+ "Kornélie",
206
+ "Kristýna",
207
+ "Kryštof",
208
+ "Krzysztof",
209
+ "Květa",
210
+ "Květoslav",
211
+ "Květoslava",
212
+ "Ladislav",
213
+ "Laura",
214
+ "Laurent",
215
+ "Lenka",
216
+ "Leontýna",
217
+ "Leopold",
218
+ "Leoš",
219
+ "Libor",
220
+ "Libuše",
221
+ "Louise",
222
+ "Lubomír",
223
+ "Lucia",
224
+ "Lucie",
225
+ "Ludmila",
226
+ "Luigi",
227
+ "Lukáš",
228
+ "Magdalena",
229
+ "Manfred",
230
+ "Marcin",
231
+ "Marco",
232
+ "Marek",
233
+ "Margaret",
234
+ "Marie",
235
+ "Marián",
236
+ "Mark",
237
+ "Markéta",
238
+ "Maroš",
239
+ "Marta",
240
+ "Martin",
241
+ "Martina",
242
+ "Mary",
243
+ "María",
244
+ "Matouš",
245
+ "Matthew",
246
+ "Matthias",
247
+ "Matúš",
248
+ "Matěj",
249
+ "Małgorzata",
250
+ "Mehmet",
251
+ "Melanie",
252
+ "Meryem",
253
+ "Michael",
254
+ "Michaela",
255
+ "Michal",
256
+ "Michel",
257
+ "Miguel",
258
+ "Milada",
259
+ "Milan",
260
+ "Milena",
261
+ "Miloslav",
262
+ "Miroslav",
263
+ "Miroslava",
264
+ "Mohammad",
265
+ "Mojmír",
266
+ "Monika",
267
+ "Monique",
268
+ "Mustafa",
269
+ "Mária",
270
+ "Naděžda",
271
+ "Natalie",
272
+ "Nathalie",
273
+ "Natálie",
274
+ "Nela",
275
+ "Nicolas",
276
+ "Nicole",
277
+ "Nikola",
278
+ "Norbert",
279
+ "Nuriye",
280
+ "Oldřich",
281
+ "Olga",
282
+ "Oliver",
283
+ "Olivia",
284
+ "Omar",
285
+ "Ondřej",
286
+ "Osman",
287
+ "Otakar",
288
+ "Otýlie",
289
+ "Pablo",
290
+ "Patrick",
291
+ "Patrik",
292
+ "Paul",
293
+ "Pavel",
294
+ "Pavla",
295
+ "Paweł",
296
+ "Pedro",
297
+ "Peter",
298
+ "Petr",
299
+ "Petra",
300
+ "Philip",
301
+ "Philippe",
302
+ "Pierre",
303
+ "Pilar",
304
+ "Piotr",
305
+ "Přemysl",
306
+ "Rachel",
307
+ "Radek",
308
+ "Radim",
309
+ "Radka",
310
+ "Radomír",
311
+ "Radomíra",
312
+ "Radovan",
313
+ "Rainer",
314
+ "Rebecca",
315
+ "Renata",
316
+ "Renate",
317
+ "Renáta",
318
+ "René",
319
+ "Ricardo",
320
+ "Richard",
321
+ "Robert",
322
+ "Roberto",
323
+ "Roger",
324
+ "Roman",
325
+ "Rosa",
326
+ "Rostislav",
327
+ "Rudolf",
328
+ "Róbert",
329
+ "Růžena",
330
+ "Sabine",
331
+ "Samantha",
332
+ "Sandra",
333
+ "Sarah",
334
+ "Sergio",
335
+ "Siegfried",
336
+ "Silvia",
337
+ "Simon",
338
+ "Simona",
339
+ "Sophie",
340
+ "Soňa",
341
+ "Stanislav",
342
+ "Stanislava",
343
+ "Stefan",
344
+ "Stefanie",
345
+ "Stephanie",
346
+ "Stephen",
347
+ "Stéphane",
348
+ "Susanne",
349
+ "Svatava",
350
+ "Svatopluk",
351
+ "Světlana",
352
+ "Sylvie",
353
+ "Sébastien",
354
+ "Tatiana",
355
+ "Teodor",
356
+ "Teresa",
357
+ "Tereza",
358
+ "Thierry",
359
+ "Thomas",
360
+ "Tibor",
361
+ "Timothy",
362
+ "Tomasz",
363
+ "Tomáš",
364
+ "Ulrich",
365
+ "Ursula",
366
+ "Uwe",
367
+ "Vendula",
368
+ "Veronika",
369
+ "Victoria",
370
+ "Viera",
371
+ "Viktor",
372
+ "Vilém",
373
+ "Vladimír",
374
+ "Vlasta",
375
+ "Vlastimil",
376
+ "Vlastimila",
377
+ "Vojtěch",
378
+ "Volker",
379
+ "Vratislav",
380
+ "Václav",
381
+ "Véronique",
382
+ "Vít",
383
+ "Vítězslav",
384
+ "Věra",
385
+ "Věroslava",
386
+ "Waltraud",
387
+ "Werner",
388
+ "William",
389
+ "Wojciech",
390
+ "Wolfgang",
391
+ "Yusuf",
392
+ "Yves",
393
+ "Zbigniew",
394
+ "Zbyněk",
395
+ "Zdeněk",
396
+ "Zdeňka",
397
+ "Zdislava",
398
+ "Zehra",
399
+ "Zeynep",
400
+ "Zikmund",
401
+ "Zuzana",
402
+ "Čeněk",
403
+ "Čestmír",
404
+ "İbrahim",
405
+ "İsmail",
406
+ "Ľudmila",
407
+ "Šimon",
408
+ "Štefan",
409
+ "Šárka"
410
+ ];
411
+ var names_first_default = {
412
+ _comment,
413
+ names
414
+ };
415
+ //#endregion
416
+ export { _comment, names_first_default as default, names };
417
+
418
+ //# sourceMappingURL=names-first.mjs.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"names-first.mjs","names":[],"sources":["../../src/data/names-first.json"],"sourcesContent":[""],"mappings":""}