@stll/anonymize-wasm 1.2.0 → 1.4.0
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.
- package/dist/address-boundaries.mjs +113 -5
- package/dist/address-stop-keywords.mjs +137 -0
- package/dist/address-stop-keywords.mjs.map +1 -0
- package/dist/address-stopwords.mjs +81 -0
- package/dist/address-stopwords.mjs.map +1 -0
- package/dist/address-street-types.mjs +67 -4
- package/dist/clause-noun-heads.mjs +75 -0
- package/dist/clause-noun-heads.mjs.map +1 -0
- package/dist/constants.d.mts +2 -0
- package/dist/constants.mjs +85 -0
- package/dist/constants.mjs.map +1 -0
- package/dist/constants2.d.mts +65 -0
- package/dist/coreference.es.mjs +22 -0
- package/dist/coreference.es.mjs.map +1 -0
- package/dist/coreference.fr.mjs +32 -0
- package/dist/coreference.fr.mjs.map +1 -0
- package/dist/coreference.it.mjs +27 -0
- package/dist/coreference.it.mjs.map +1 -0
- package/dist/coreference.pl.mjs +27 -0
- package/dist/coreference.pl.mjs.map +1 -0
- package/dist/coreference.pt-br.mjs +14 -0
- package/dist/coreference.pt-br.mjs.map +1 -0
- package/dist/coreference.sk.mjs +22 -5
- package/dist/generic-roles.mjs +179 -1
- package/dist/hotword-rules.mjs +7 -3
- package/dist/legal-forms.mjs +3 -0
- package/dist/legal-role-heads.cs.mjs +42 -0
- package/dist/legal-role-heads.cs.mjs.map +1 -0
- package/dist/legal-role-heads.de.mjs +33 -0
- package/dist/legal-role-heads.de.mjs.map +1 -0
- package/dist/legal-role-heads.en.mjs +37 -0
- package/dist/legal-role-heads.en.mjs.map +1 -0
- package/dist/legal-role-heads.es.mjs +54 -0
- package/dist/legal-role-heads.es.mjs.map +1 -0
- package/dist/legal-role-heads.fr.mjs +72 -0
- package/dist/legal-role-heads.fr.mjs.map +1 -0
- package/dist/legal-role-heads.it.mjs +68 -0
- package/dist/legal-role-heads.it.mjs.map +1 -0
- package/dist/legal-role-heads.pl.mjs +84 -0
- package/dist/legal-role-heads.pl.mjs.map +1 -0
- package/dist/legal-role-heads.pt-br.mjs +63 -0
- package/dist/legal-role-heads.pt-br.mjs.map +1 -0
- package/dist/legal-role-heads.sk.mjs +80 -0
- package/dist/legal-role-heads.sk.mjs.map +1 -0
- package/dist/manifest.mjs +21 -8
- package/dist/names-exclusions.mjs +21 -1
- package/dist/person-stopwords.mjs +121 -1
- package/dist/section-headings.mjs +15 -3
- package/dist/sentence-verb-indicators.mjs +223 -0
- package/dist/sentence-verb-indicators.mjs.map +1 -0
- package/dist/triggers.cs.mjs +107 -3
- package/dist/triggers.en.mjs +32 -0
- package/dist/triggers.es.mjs +43 -4
- package/dist/triggers.fr.mjs +206 -7
- package/dist/triggers.it.mjs +26 -4
- package/dist/triggers.pl.mjs +191 -4
- package/dist/triggers.pt-br.mjs +193 -0
- package/dist/triggers.pt-br.mjs.map +1 -0
- package/dist/triggers.sk.mjs +493 -0
- package/dist/vite.mjs.map +1 -1
- package/dist/wasm.d.mts +56 -39
- package/dist/wasm.mjs +2382 -1408
- package/dist/wasm.mjs.map +1 -1
- package/package.json +7 -2
|
@@ -26,16 +26,60 @@ var address_boundaries_default = {
|
|
|
26
26
|
"attn",
|
|
27
27
|
"c/o",
|
|
28
28
|
"care of",
|
|
29
|
+
"concerning",
|
|
29
30
|
"contact person",
|
|
31
|
+
"for any",
|
|
30
32
|
"incorporated",
|
|
33
|
+
"postmarked",
|
|
31
34
|
"registered",
|
|
32
|
-
"represented"
|
|
35
|
+
"represented",
|
|
36
|
+
"shall govern",
|
|
37
|
+
"shall be governed",
|
|
38
|
+
"to be enforced",
|
|
39
|
+
"with the intention",
|
|
40
|
+
"without reference",
|
|
41
|
+
"without regard"
|
|
42
|
+
],
|
|
43
|
+
es: [
|
|
44
|
+
"con CIF",
|
|
45
|
+
"con C.I.F.",
|
|
46
|
+
"con NIF",
|
|
47
|
+
"con N.I.F.",
|
|
48
|
+
"con DNI",
|
|
49
|
+
"con D.N.I.",
|
|
50
|
+
"con NIE",
|
|
51
|
+
"con N.I.E.",
|
|
52
|
+
"inscrita",
|
|
53
|
+
"inscrito",
|
|
54
|
+
"registro mercantil",
|
|
55
|
+
"registro",
|
|
56
|
+
"representada por",
|
|
57
|
+
"representado por",
|
|
58
|
+
"actuando en nombre",
|
|
59
|
+
"número de identificación fiscal",
|
|
60
|
+
"código postal"
|
|
33
61
|
],
|
|
34
62
|
fr: [
|
|
63
|
+
"agissant",
|
|
64
|
+
"ayant pour",
|
|
65
|
+
"ayant son",
|
|
66
|
+
"ci-après",
|
|
67
|
+
"ci-dessous",
|
|
68
|
+
"ci-dessus",
|
|
69
|
+
"désigné",
|
|
70
|
+
"désignée",
|
|
71
|
+
"dénommé",
|
|
72
|
+
"dénommée",
|
|
73
|
+
"immatriculé",
|
|
35
74
|
"immatriculée",
|
|
75
|
+
"inscrit",
|
|
36
76
|
"inscrite",
|
|
77
|
+
"RCS",
|
|
37
78
|
"représenté",
|
|
38
|
-
"représentée"
|
|
79
|
+
"représentée",
|
|
80
|
+
"SIREN",
|
|
81
|
+
"SIRET",
|
|
82
|
+
"TVA"
|
|
39
83
|
],
|
|
40
84
|
hu: [
|
|
41
85
|
"bejegyzett",
|
|
@@ -43,19 +87,65 @@ var address_boundaries_default = {
|
|
|
43
87
|
"képviseli"
|
|
44
88
|
],
|
|
45
89
|
it: [
|
|
90
|
+
"C.F.",
|
|
91
|
+
"P. IVA",
|
|
92
|
+
"P.IVA",
|
|
93
|
+
"codice fiscale",
|
|
46
94
|
"iscritta",
|
|
95
|
+
"iscritto",
|
|
96
|
+
"partita IVA",
|
|
97
|
+
"punti di contatto",
|
|
47
98
|
"rappresentata",
|
|
48
99
|
"rappresentato",
|
|
49
|
-
"registro imprese"
|
|
100
|
+
"registro imprese",
|
|
101
|
+
"responsabile del procedimento"
|
|
50
102
|
],
|
|
51
103
|
pl: [
|
|
104
|
+
"kapitał zakładowy",
|
|
105
|
+
"kapitale zakładowym",
|
|
106
|
+
"KRS",
|
|
107
|
+
"NIP",
|
|
108
|
+
"REGON",
|
|
109
|
+
"PESEL",
|
|
52
110
|
"reprezentowana",
|
|
53
111
|
"reprezentowany",
|
|
112
|
+
"reprezentowaną",
|
|
113
|
+
"reprezentowanego",
|
|
114
|
+
"reprezentowanym",
|
|
115
|
+
"reprezentowanej",
|
|
54
116
|
"wpisana",
|
|
55
117
|
"wpisane",
|
|
56
118
|
"wpisano",
|
|
57
119
|
"wpisany",
|
|
58
|
-
"
|
|
120
|
+
"wpisaną",
|
|
121
|
+
"wpisanego",
|
|
122
|
+
"wpisanej",
|
|
123
|
+
"zarejestrowana",
|
|
124
|
+
"zarejestrowany",
|
|
125
|
+
"zarejestrowaną",
|
|
126
|
+
"zarejestrowanego",
|
|
127
|
+
"zwana dalej",
|
|
128
|
+
"zwany dalej",
|
|
129
|
+
"zwaną dalej",
|
|
130
|
+
"zwanego dalej"
|
|
131
|
+
],
|
|
132
|
+
"pt-br": [
|
|
133
|
+
"CNPJ",
|
|
134
|
+
"CPF",
|
|
135
|
+
"RG",
|
|
136
|
+
"OAB",
|
|
137
|
+
"inscrita",
|
|
138
|
+
"inscrito",
|
|
139
|
+
"Inscrição Estadual",
|
|
140
|
+
"Inscrição Municipal",
|
|
141
|
+
"neste ato representada",
|
|
142
|
+
"neste ato representado",
|
|
143
|
+
"doravante denominada",
|
|
144
|
+
"doravante denominado",
|
|
145
|
+
"registrada",
|
|
146
|
+
"registrado",
|
|
147
|
+
"portador",
|
|
148
|
+
"portadora"
|
|
59
149
|
],
|
|
60
150
|
ro: [
|
|
61
151
|
"înregistrată",
|
|
@@ -63,11 +153,29 @@ var address_boundaries_default = {
|
|
|
63
153
|
"reprezentată"
|
|
64
154
|
],
|
|
65
155
|
sk: [
|
|
156
|
+
"IČO",
|
|
157
|
+
"DIČ",
|
|
158
|
+
"IČ DPH",
|
|
159
|
+
"bankové spojenie",
|
|
160
|
+
"číslo účtu",
|
|
66
161
|
"konajúci",
|
|
162
|
+
"konajúca",
|
|
163
|
+
"kontaktná osoba",
|
|
164
|
+
"registrovaný",
|
|
165
|
+
"registrovaná",
|
|
166
|
+
"splnomocnený zástupca",
|
|
167
|
+
"splnomocnená zástupkyňa",
|
|
168
|
+
"štatutárny orgán",
|
|
169
|
+
"štatutárny zástupca",
|
|
170
|
+
"v zastúpení",
|
|
67
171
|
"zapísaná",
|
|
68
172
|
"zapísaný",
|
|
173
|
+
"zapísané",
|
|
174
|
+
"zaregistrovaná",
|
|
175
|
+
"zaregistrovaný",
|
|
69
176
|
"zastúpená",
|
|
70
|
-
"zastúpený"
|
|
177
|
+
"zastúpený",
|
|
178
|
+
"zastúpené"
|
|
71
179
|
],
|
|
72
180
|
universal: [
|
|
73
181
|
"e-mail",
|
|
@@ -0,0 +1,137 @@
|
|
|
1
|
+
//#region src/data/address-stop-keywords.json
|
|
2
|
+
var _comment = "Field-label keywords that terminate address-strategy scanning. When the address strategy hits a comma followed by one of these labels, the address span stops before the keyword so 'IČO:', 'Bankové spojenie:', 'IBAN:' don't get absorbed into the address. Lowercased; matched case-insensitively. Organised per-language because the labels are language-specific; at runtime the trigger detector unions every language's list into a single stop set because address spans don't carry a language tag.";
|
|
3
|
+
var cs = [
|
|
4
|
+
"číslo účtu",
|
|
5
|
+
"registrační",
|
|
6
|
+
"zastoupen",
|
|
7
|
+
"bankovní",
|
|
8
|
+
"datová",
|
|
9
|
+
"vložka",
|
|
10
|
+
"sp.zn.",
|
|
11
|
+
"jednatel",
|
|
12
|
+
"ředitel",
|
|
13
|
+
"oddíl",
|
|
14
|
+
"telefon",
|
|
15
|
+
"č.ú.",
|
|
16
|
+
"dič",
|
|
17
|
+
"ičo",
|
|
18
|
+
"ič"
|
|
19
|
+
];
|
|
20
|
+
var en = [
|
|
21
|
+
"e-mail",
|
|
22
|
+
"email",
|
|
23
|
+
"tel",
|
|
24
|
+
"swift",
|
|
25
|
+
"iban",
|
|
26
|
+
"bic"
|
|
27
|
+
];
|
|
28
|
+
var pl = [
|
|
29
|
+
"nip",
|
|
30
|
+
"krs",
|
|
31
|
+
"regon",
|
|
32
|
+
"pesel",
|
|
33
|
+
"numer konta",
|
|
34
|
+
"konto bankowe",
|
|
35
|
+
"rachunek bankowy",
|
|
36
|
+
"tel.",
|
|
37
|
+
"telefon",
|
|
38
|
+
"e-mail",
|
|
39
|
+
"email"
|
|
40
|
+
];
|
|
41
|
+
var fr = [
|
|
42
|
+
"e-mail",
|
|
43
|
+
"email",
|
|
44
|
+
"courriel",
|
|
45
|
+
"téléphone",
|
|
46
|
+
"tél.",
|
|
47
|
+
"tél",
|
|
48
|
+
"portable",
|
|
49
|
+
"fax",
|
|
50
|
+
"siren",
|
|
51
|
+
"siret",
|
|
52
|
+
"tva",
|
|
53
|
+
"rcs",
|
|
54
|
+
"naf",
|
|
55
|
+
"ape",
|
|
56
|
+
"iban",
|
|
57
|
+
"bic",
|
|
58
|
+
"swift",
|
|
59
|
+
"code postal",
|
|
60
|
+
"cp",
|
|
61
|
+
"nir"
|
|
62
|
+
];
|
|
63
|
+
var sk = [
|
|
64
|
+
"číslo účtu",
|
|
65
|
+
"bankové spojenie",
|
|
66
|
+
"ičo",
|
|
67
|
+
"ič dph",
|
|
68
|
+
"ičdph",
|
|
69
|
+
"dič",
|
|
70
|
+
"konateľ",
|
|
71
|
+
"konateľka",
|
|
72
|
+
"konateľom",
|
|
73
|
+
"konateľkou",
|
|
74
|
+
"riaditeľ",
|
|
75
|
+
"riaditeľka",
|
|
76
|
+
"vložka",
|
|
77
|
+
"oddiel",
|
|
78
|
+
"sp.zn.",
|
|
79
|
+
"sp. zn.",
|
|
80
|
+
"spisovej značky",
|
|
81
|
+
"č.ú.",
|
|
82
|
+
"č. účtu"
|
|
83
|
+
];
|
|
84
|
+
var pt_br = [
|
|
85
|
+
"bic",
|
|
86
|
+
"cnpj",
|
|
87
|
+
"cnpj/mf",
|
|
88
|
+
"conta",
|
|
89
|
+
"cpf",
|
|
90
|
+
"cpf/mf",
|
|
91
|
+
"doravante",
|
|
92
|
+
"e-mail",
|
|
93
|
+
"email",
|
|
94
|
+
"fax",
|
|
95
|
+
"iban",
|
|
96
|
+
"inscrição estadual",
|
|
97
|
+
"inscrição municipal",
|
|
98
|
+
"neste ato",
|
|
99
|
+
"nire",
|
|
100
|
+
"oab",
|
|
101
|
+
"portador",
|
|
102
|
+
"portadora",
|
|
103
|
+
"representada",
|
|
104
|
+
"representado",
|
|
105
|
+
"rg",
|
|
106
|
+
"swift",
|
|
107
|
+
"tel",
|
|
108
|
+
"telefone"
|
|
109
|
+
];
|
|
110
|
+
var es = [];
|
|
111
|
+
var it = [
|
|
112
|
+
"c.f.",
|
|
113
|
+
"c.f. e p.iva",
|
|
114
|
+
"codice fiscale",
|
|
115
|
+
"p. iva",
|
|
116
|
+
"p.iva",
|
|
117
|
+
"partita iva",
|
|
118
|
+
"pec",
|
|
119
|
+
"registro imprese",
|
|
120
|
+
"rappresentata",
|
|
121
|
+
"rappresentato"
|
|
122
|
+
];
|
|
123
|
+
var address_stop_keywords_default = {
|
|
124
|
+
_comment,
|
|
125
|
+
cs,
|
|
126
|
+
en,
|
|
127
|
+
pl,
|
|
128
|
+
fr,
|
|
129
|
+
sk,
|
|
130
|
+
"pt-br": pt_br,
|
|
131
|
+
es,
|
|
132
|
+
it
|
|
133
|
+
};
|
|
134
|
+
//#endregion
|
|
135
|
+
export { _comment, cs, address_stop_keywords_default as default, en, es, fr, it, pl, pt_br as "pt-br", sk };
|
|
136
|
+
|
|
137
|
+
//# sourceMappingURL=address-stop-keywords.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"address-stop-keywords.mjs","names":[],"sources":["../../src/data/address-stop-keywords.json"],"sourcesContent":[""],"mappings":""}
|
|
@@ -0,0 +1,81 @@
|
|
|
1
|
+
//#region src/data/address-stopwords.json
|
|
2
|
+
var address_stopwords_default = {
|
|
3
|
+
_comment: "Single-token English/legal words that collide with tiny entries in city dictionaries (GeoNames-based). When the deny-list pass would emit one of these as a standalone `address` hit, it is dropped — almost certainly a Purchase Price / European Union / Effective Time false positive, not a real place. Multi-token matches (e.g. 'Lake Union, WA') are unaffected because the filter only fires on single-token spans.",
|
|
4
|
+
_it_comment: "Italian additions are conservative: only common-noun Italian words that collide with city dictionary entries. Real Italian cities are NOT listed; multi-token spans (`Via Trieste`, `41012 Modena`) are unaffected.",
|
|
5
|
+
words: [
|
|
6
|
+
"agenda",
|
|
7
|
+
"agent",
|
|
8
|
+
"amount",
|
|
9
|
+
"brent",
|
|
10
|
+
"central",
|
|
11
|
+
"centro",
|
|
12
|
+
"client",
|
|
13
|
+
"code",
|
|
14
|
+
"company",
|
|
15
|
+
"consent",
|
|
16
|
+
"contract",
|
|
17
|
+
"customer",
|
|
18
|
+
"default",
|
|
19
|
+
"delivery",
|
|
20
|
+
"event",
|
|
21
|
+
"exhibit",
|
|
22
|
+
"general",
|
|
23
|
+
"gordon",
|
|
24
|
+
"interest",
|
|
25
|
+
"license",
|
|
26
|
+
"memorandum",
|
|
27
|
+
"notice",
|
|
28
|
+
"owner",
|
|
29
|
+
"party",
|
|
30
|
+
"payment",
|
|
31
|
+
"price",
|
|
32
|
+
"principal",
|
|
33
|
+
"rate",
|
|
34
|
+
"registro",
|
|
35
|
+
"schedule",
|
|
36
|
+
"se",
|
|
37
|
+
"section",
|
|
38
|
+
"server",
|
|
39
|
+
"tenant",
|
|
40
|
+
"term",
|
|
41
|
+
"termination",
|
|
42
|
+
"time",
|
|
43
|
+
"union",
|
|
44
|
+
"urbana",
|
|
45
|
+
"ventosa",
|
|
46
|
+
"vendor",
|
|
47
|
+
"waiver",
|
|
48
|
+
"amministrazione",
|
|
49
|
+
"appaltante",
|
|
50
|
+
"appalto",
|
|
51
|
+
"articolo",
|
|
52
|
+
"cliente",
|
|
53
|
+
"concedente",
|
|
54
|
+
"concessionario",
|
|
55
|
+
"conduttore",
|
|
56
|
+
"denominazione",
|
|
57
|
+
"esecuzione",
|
|
58
|
+
"fornitore",
|
|
59
|
+
"indirizzo",
|
|
60
|
+
"informazioni",
|
|
61
|
+
"lavoratore",
|
|
62
|
+
"lavoro",
|
|
63
|
+
"locatore",
|
|
64
|
+
"luogo",
|
|
65
|
+
"magazzino",
|
|
66
|
+
"mandante",
|
|
67
|
+
"mandatario",
|
|
68
|
+
"oggetto",
|
|
69
|
+
"piano",
|
|
70
|
+
"procedimento",
|
|
71
|
+
"procedura",
|
|
72
|
+
"responsabile",
|
|
73
|
+
"sede",
|
|
74
|
+
"sezione",
|
|
75
|
+
"venditore"
|
|
76
|
+
]
|
|
77
|
+
};
|
|
78
|
+
//#endregion
|
|
79
|
+
export { address_stopwords_default as default };
|
|
80
|
+
|
|
81
|
+
//# sourceMappingURL=address-stopwords.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"address-stopwords.mjs","names":[],"sources":["../../src/data/address-stopwords.json"],"sourcesContent":[""],"mappings":""}
|
|
@@ -44,26 +44,48 @@ var en = [
|
|
|
44
44
|
"Way"
|
|
45
45
|
];
|
|
46
46
|
var es = [
|
|
47
|
+
"Av.",
|
|
48
|
+
"Avda.",
|
|
47
49
|
"avenida",
|
|
48
50
|
"bulevar",
|
|
51
|
+
"C/",
|
|
49
52
|
"calle",
|
|
50
53
|
"camino",
|
|
54
|
+
"carrera",
|
|
51
55
|
"carretera",
|
|
56
|
+
"Cra.",
|
|
57
|
+
"Ctra.",
|
|
52
58
|
"glorieta",
|
|
53
59
|
"paseo",
|
|
54
60
|
"plaza",
|
|
61
|
+
"Pza.",
|
|
55
62
|
"ronda",
|
|
56
63
|
"travesía"
|
|
57
64
|
];
|
|
58
65
|
var fr = [
|
|
59
66
|
"allée",
|
|
67
|
+
"av",
|
|
68
|
+
"av.",
|
|
60
69
|
"avenue",
|
|
70
|
+
"bd",
|
|
71
|
+
"bd.",
|
|
61
72
|
"boulevard",
|
|
62
73
|
"chemin",
|
|
74
|
+
"cours",
|
|
63
75
|
"impasse",
|
|
76
|
+
"passage",
|
|
77
|
+
"pl",
|
|
78
|
+
"pl.",
|
|
64
79
|
"place",
|
|
65
80
|
"quai",
|
|
66
|
-
"
|
|
81
|
+
"résidence",
|
|
82
|
+
"route",
|
|
83
|
+
"rte",
|
|
84
|
+
"rte.",
|
|
85
|
+
"rue",
|
|
86
|
+
"ruelle",
|
|
87
|
+
"square",
|
|
88
|
+
"voie"
|
|
67
89
|
];
|
|
68
90
|
var hu = [
|
|
69
91
|
"fasor",
|
|
@@ -79,15 +101,27 @@ var hu = [
|
|
|
79
101
|
"utca"
|
|
80
102
|
];
|
|
81
103
|
var it = [
|
|
104
|
+
"borgo",
|
|
105
|
+
"calle",
|
|
106
|
+
"contrada",
|
|
82
107
|
"corso",
|
|
83
108
|
"largo",
|
|
84
109
|
"lungarno",
|
|
110
|
+
"lungomare",
|
|
85
111
|
"lungotevere",
|
|
112
|
+
"passeggiata",
|
|
86
113
|
"piazza",
|
|
87
114
|
"piazzale",
|
|
115
|
+
"salita",
|
|
116
|
+
"strada",
|
|
117
|
+
"stradone",
|
|
118
|
+
"traversa",
|
|
88
119
|
"via",
|
|
89
120
|
"viale",
|
|
90
|
-
"
|
|
121
|
+
"vico",
|
|
122
|
+
"vicolo",
|
|
123
|
+
"v.le",
|
|
124
|
+
"v.lo"
|
|
91
125
|
];
|
|
92
126
|
var nl = [
|
|
93
127
|
"gracht",
|
|
@@ -100,13 +134,21 @@ var nl = [
|
|
|
100
134
|
var pl = [
|
|
101
135
|
"al.",
|
|
102
136
|
"aleja",
|
|
137
|
+
"aleje",
|
|
138
|
+
"aleją",
|
|
139
|
+
"alei",
|
|
103
140
|
"bulwar",
|
|
141
|
+
"os.",
|
|
142
|
+
"osiedle",
|
|
104
143
|
"pl.",
|
|
105
144
|
"plac",
|
|
145
|
+
"placu",
|
|
106
146
|
"rondo",
|
|
147
|
+
"rynek",
|
|
107
148
|
"skwer",
|
|
108
149
|
"ul.",
|
|
109
|
-
"ulica"
|
|
150
|
+
"ulica",
|
|
151
|
+
"ulicy"
|
|
110
152
|
];
|
|
111
153
|
var pt = [
|
|
112
154
|
"avenida",
|
|
@@ -115,6 +157,18 @@ var pt = [
|
|
|
115
157
|
"rua",
|
|
116
158
|
"travessa"
|
|
117
159
|
];
|
|
160
|
+
var pt_br = [
|
|
161
|
+
"alameda",
|
|
162
|
+
"av.",
|
|
163
|
+
"avenida",
|
|
164
|
+
"estrada",
|
|
165
|
+
"largo",
|
|
166
|
+
"praça",
|
|
167
|
+
"praca",
|
|
168
|
+
"rodovia",
|
|
169
|
+
"rua",
|
|
170
|
+
"travessa"
|
|
171
|
+
];
|
|
118
172
|
var ro = [
|
|
119
173
|
"aleea",
|
|
120
174
|
"bd.",
|
|
@@ -129,8 +183,16 @@ var ro = [
|
|
|
129
183
|
"strada"
|
|
130
184
|
];
|
|
131
185
|
var sk = [
|
|
186
|
+
"alej",
|
|
187
|
+
"nábr.",
|
|
188
|
+
"nábrežie",
|
|
189
|
+
"nám.",
|
|
132
190
|
"námest.",
|
|
133
191
|
"námestie",
|
|
192
|
+
"sady",
|
|
193
|
+
"tr.",
|
|
194
|
+
"trieda",
|
|
195
|
+
"ul.",
|
|
134
196
|
"ulica"
|
|
135
197
|
];
|
|
136
198
|
var sv = [
|
|
@@ -152,11 +214,12 @@ var address_street_types_default = {
|
|
|
152
214
|
nl,
|
|
153
215
|
pl,
|
|
154
216
|
pt,
|
|
217
|
+
"pt-br": pt_br,
|
|
155
218
|
ro,
|
|
156
219
|
sk,
|
|
157
220
|
sv
|
|
158
221
|
};
|
|
159
222
|
//#endregion
|
|
160
|
-
export { _comment, cs, de, address_street_types_default as default, en, es, fr, hu, it, nl, pl, pt, ro, sk, sv };
|
|
223
|
+
export { _comment, cs, de, address_street_types_default as default, en, es, fr, hu, it, nl, pl, pt, pt_br as "pt-br", ro, sk, sv };
|
|
161
224
|
|
|
162
225
|
//# sourceMappingURL=address-street-types.mjs.map
|
|
@@ -0,0 +1,75 @@
|
|
|
1
|
+
//#region src/data/clause-noun-heads.json
|
|
2
|
+
var _comment = "Common contract/clause nouns that appear in legal prose between a sentence-verb and the company name (e.g. 'Vendor signed Agreement with Acme Inc.'). The legal-form trim skips these when scanning forward for the org's first Cap word, so the anchor lands on the real company name rather than the contract noun. Lowercased; matched case-insensitively. Organised per-language; at runtime the legal-form detector unions every language's list into a single set because match spans don't carry a language tag.";
|
|
3
|
+
var en = [
|
|
4
|
+
"agreement",
|
|
5
|
+
"agreements",
|
|
6
|
+
"contract",
|
|
7
|
+
"contracts",
|
|
8
|
+
"license",
|
|
9
|
+
"licence",
|
|
10
|
+
"lease",
|
|
11
|
+
"memorandum",
|
|
12
|
+
"notice",
|
|
13
|
+
"exhibit",
|
|
14
|
+
"schedule",
|
|
15
|
+
"addendum",
|
|
16
|
+
"amendment",
|
|
17
|
+
"appendix",
|
|
18
|
+
"attachment"
|
|
19
|
+
];
|
|
20
|
+
var cs = [
|
|
21
|
+
"smlouva",
|
|
22
|
+
"smlouvy",
|
|
23
|
+
"smlouvu",
|
|
24
|
+
"smlouvou",
|
|
25
|
+
"dohoda",
|
|
26
|
+
"dohody",
|
|
27
|
+
"dohodu",
|
|
28
|
+
"dohodou",
|
|
29
|
+
"licence",
|
|
30
|
+
"licenci",
|
|
31
|
+
"příloha",
|
|
32
|
+
"přílohy",
|
|
33
|
+
"přílohu",
|
|
34
|
+
"dodatek",
|
|
35
|
+
"dodatku",
|
|
36
|
+
"oznámení"
|
|
37
|
+
];
|
|
38
|
+
var de = [
|
|
39
|
+
"vertrag",
|
|
40
|
+
"vertrages",
|
|
41
|
+
"vereinbarung",
|
|
42
|
+
"vereinbarungen",
|
|
43
|
+
"lizenz",
|
|
44
|
+
"anlage",
|
|
45
|
+
"anhang"
|
|
46
|
+
];
|
|
47
|
+
var es = [
|
|
48
|
+
"acuerdo",
|
|
49
|
+
"acuerdos",
|
|
50
|
+
"contrato",
|
|
51
|
+
"contratos",
|
|
52
|
+
"convenio",
|
|
53
|
+
"convenios",
|
|
54
|
+
"licencia",
|
|
55
|
+
"licencias",
|
|
56
|
+
"memorándum",
|
|
57
|
+
"notificación",
|
|
58
|
+
"anexo",
|
|
59
|
+
"anexos",
|
|
60
|
+
"adenda",
|
|
61
|
+
"apéndice",
|
|
62
|
+
"cláusula",
|
|
63
|
+
"cláusulas"
|
|
64
|
+
];
|
|
65
|
+
var clause_noun_heads_default = {
|
|
66
|
+
_comment,
|
|
67
|
+
en,
|
|
68
|
+
cs,
|
|
69
|
+
de,
|
|
70
|
+
es
|
|
71
|
+
};
|
|
72
|
+
//#endregion
|
|
73
|
+
export { _comment, cs, de, clause_noun_heads_default as default, en, es };
|
|
74
|
+
|
|
75
|
+
//# sourceMappingURL=clause-noun-heads.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"clause-noun-heads.mjs","names":[],"sources":["../../src/data/clause-noun-heads.json"],"sourcesContent":[""],"mappings":""}
|
|
@@ -0,0 +1,2 @@
|
|
|
1
|
+
import { a as OPERATOR_TYPES, i as DetectionSource, n as DETECTION_SOURCES, o as OperatorType, r as DETECTOR_PRIORITY, t as DEFAULT_ENTITY_LABELS } from "./constants2.mjs";
|
|
2
|
+
export { DEFAULT_ENTITY_LABELS, DETECTION_SOURCES, DETECTOR_PRIORITY, DetectionSource, OPERATOR_TYPES, OperatorType };
|
|
@@ -0,0 +1,85 @@
|
|
|
1
|
+
//#region src/constants.ts
|
|
2
|
+
/**
|
|
3
|
+
* Runtime-free constants for the anonymization pipeline.
|
|
4
|
+
*
|
|
5
|
+
* This module is the SSR-safe / browser-safe entrypoint:
|
|
6
|
+
* importing it must not pull in `@stll/text-search`,
|
|
7
|
+
* `@stll/anonymize-wasm`, or any other runtime-bearing
|
|
8
|
+
* module. Consumers that only need the static label list,
|
|
9
|
+
* detection-source identifiers, or operator names import
|
|
10
|
+
* from `@stll/anonymize/constants` (or
|
|
11
|
+
* `@stll/anonymize-wasm/constants`) without paying the
|
|
12
|
+
* wasm / regex-set startup cost.
|
|
13
|
+
*
|
|
14
|
+
* `types.ts` re-exports these for back-compat, so existing
|
|
15
|
+
* `import { DEFAULT_ENTITY_LABELS } from "@stll/anonymize"`
|
|
16
|
+
* call sites keep working.
|
|
17
|
+
*/
|
|
18
|
+
/**
|
|
19
|
+
* Source of a detected entity span.
|
|
20
|
+
* Ordered by detection layer in the pipeline.
|
|
21
|
+
*/
|
|
22
|
+
const DETECTION_SOURCES = {
|
|
23
|
+
TRIGGER: "trigger",
|
|
24
|
+
REGEX: "regex",
|
|
25
|
+
DENY_LIST: "deny-list",
|
|
26
|
+
LEGAL_FORM: "legal-form",
|
|
27
|
+
GAZETTEER: "gazetteer",
|
|
28
|
+
NER: "ner",
|
|
29
|
+
COREFERENCE: "coreference"
|
|
30
|
+
};
|
|
31
|
+
/**
|
|
32
|
+
* Priority levels for detection sources.
|
|
33
|
+
* Higher = more structurally reliable. Used during
|
|
34
|
+
* overlap resolution so deterministic detectors beat
|
|
35
|
+
* probabilistic ones regardless of raw score.
|
|
36
|
+
*/
|
|
37
|
+
const DETECTOR_PRIORITY = {
|
|
38
|
+
[DETECTION_SOURCES.GAZETTEER]: 5,
|
|
39
|
+
[DETECTION_SOURCES.TRIGGER]: 4,
|
|
40
|
+
[DETECTION_SOURCES.LEGAL_FORM]: 3,
|
|
41
|
+
[DETECTION_SOURCES.REGEX]: 3,
|
|
42
|
+
[DETECTION_SOURCES.DENY_LIST]: 2,
|
|
43
|
+
[DETECTION_SOURCES.COREFERENCE]: 2,
|
|
44
|
+
[DETECTION_SOURCES.NER]: 1
|
|
45
|
+
};
|
|
46
|
+
/**
|
|
47
|
+
* Anonymization operator types. Each operator defines
|
|
48
|
+
* how a confirmed entity is replaced in the output.
|
|
49
|
+
*/
|
|
50
|
+
const OPERATOR_TYPES = ["replace", "redact"];
|
|
51
|
+
/**
|
|
52
|
+
* Canonical entity labels used across the pipeline.
|
|
53
|
+
* NER models may use different native labels; the bench
|
|
54
|
+
* NER wrapper maps model output to these canonical names.
|
|
55
|
+
*
|
|
56
|
+
* These labels are ephemeral: entities are regenerated on
|
|
57
|
+
* every pipeline run and never persisted to the database.
|
|
58
|
+
* Renaming a label here requires no migration.
|
|
59
|
+
*/
|
|
60
|
+
const DEFAULT_ENTITY_LABELS = [
|
|
61
|
+
"person",
|
|
62
|
+
"organization",
|
|
63
|
+
"phone number",
|
|
64
|
+
"address",
|
|
65
|
+
"email address",
|
|
66
|
+
"date",
|
|
67
|
+
"date of birth",
|
|
68
|
+
"bank account number",
|
|
69
|
+
"iban",
|
|
70
|
+
"tax identification number",
|
|
71
|
+
"identity card number",
|
|
72
|
+
"birth number",
|
|
73
|
+
"national identification number",
|
|
74
|
+
"social security number",
|
|
75
|
+
"registration number",
|
|
76
|
+
"credit card number",
|
|
77
|
+
"passport number",
|
|
78
|
+
"monetary amount",
|
|
79
|
+
"land parcel",
|
|
80
|
+
"misc"
|
|
81
|
+
];
|
|
82
|
+
//#endregion
|
|
83
|
+
export { DEFAULT_ENTITY_LABELS, DETECTION_SOURCES, DETECTOR_PRIORITY, OPERATOR_TYPES };
|
|
84
|
+
|
|
85
|
+
//# sourceMappingURL=constants.mjs.map
|