@stll/anonymize-wasm 2.0.0-alpha.1 → 2.0.1

package/package.json

@@ -1,7 +1,18 @@
 {
   "name": "@stll/anonymize-wasm",
-  "version": "2.0.0-alpha.1",
-  "description": "PII detection and anonymization for browsers via WebAssembly. Same API as @stll/anonymize.",
+  "version": "2.0.1",
+  "description": "PII detection and anonymization for browsers via WebAssembly. Same native SDK API as @stll/anonymize.",
+  "keywords": [
+    "anonymization",
+    "pii",
+    "redaction",
+    "ner",
+    "pseudonymization",
+    "gdpr",
+    "privacy",
+    "wasm",
+    "browser"
+  ],
   "type": "module",
   "module": "dist/wasm.mjs",
   "types": "dist/wasm.d.mts",
@@ -30,20 +41,15 @@
     "type": "git",
     "url": "git+https://github.com/stella/anonymize.git"
   },
+  "homepage": "https://github.com/stella/anonymize",
   "license": "MIT",
   "dependencies": {
-    "@huggingface/tokenizers": "^0.1.3",
-    "@stll/stdnum": "^2.1.1",
-    "@stll/text-search-wasm": "^1.0.7"
+    "@napi-rs/wasm-runtime": "^1.1.6"
   },
   "peerDependencies": {
-    "@stll/anonymize-data": "^0.0.6",
     "vite": ">=5 <10"
   },
   "peerDependenciesMeta": {
-    "@stll/anonymize-data": {
-      "optional": true
-    },
     "vite": {
       "optional": true
     }

package/dist/address-boundaries.mjs

@@ -1,197 +0,0 @@
-//#region src/data/address-boundaries.json
-var address_boundaries_default = {
-	_comment: "Boundary words that signal the end of an address span. Multi-language. Used by address seed detector to stop expansion.",
-	cs: [
-		"jednající",
-		"jednajícím",
-		"jejímž jménem",
-		"kontaktní osoba",
-		"pokud",
-		"zapsán",
-		"zapsaná",
-		"zapsané",
-		"zapsaný",
-		"zastoupená",
-		"zastoupené",
-		"zastoupený"
-	],
-	de: [
-		"amtsgericht",
-		"eingetragen",
-		"geschäftsführer",
-		"handelsregister",
-		"vertreten"
-	],
-	en: [
-		"attention",
-		"attn",
-		"c/o",
-		"care of",
-		"concerning",
-		"contact person",
-		"for any",
-		"incorporated",
-		"postmarked",
-		"registered",
-		"represented",
-		"shall govern",
-		"shall be governed",
-		"to be enforced",
-		"with a copy",
-		"with the intention",
-		"without reference",
-		"without regard"
-	],
-	es: [
-		"con CIF",
-		"con C.I.F.",
-		"con NIF",
-		"con N.I.F.",
-		"con DNI",
-		"con D.N.I.",
-		"con NIE",
-		"con N.I.E.",
-		"inscrita",
-		"inscrito",
-		"registro mercantil",
-		"registro",
-		"representada por",
-		"representado por",
-		"actuando en nombre",
-		"número de identificación fiscal",
-		"código postal"
-	],
-	fr: [
-		"agissant",
-		"ayant pour",
-		"ayant son",
-		"ci-après",
-		"ci-dessous",
-		"ci-dessus",
-		"désigné",
-		"désignée",
-		"dénommé",
-		"dénommée",
-		"immatriculé",
-		"immatriculée",
-		"inscrit",
-		"inscrite",
-		"RCS",
-		"représenté",
-		"représentée",
-		"SIREN",
-		"SIRET",
-		"TVA"
-	],
-	hu: [
-		"bejegyzett",
-		"cégjegyzékszám",
-		"képviseli"
-	],
-	it: [
-		"C.F.",
-		"P. IVA",
-		"P.IVA",
-		"codice fiscale",
-		"iscritta",
-		"iscritto",
-		"partita IVA",
-		"punti di contatto",
-		"rappresentata",
-		"rappresentato",
-		"registro imprese",
-		"responsabile del procedimento"
-	],
-	pl: [
-		"kapitał zakładowy",
-		"kapitale zakładowym",
-		"KRS",
-		"NIP",
-		"REGON",
-		"PESEL",
-		"reprezentowana",
-		"reprezentowany",
-		"reprezentowaną",
-		"reprezentowanego",
-		"reprezentowanym",
-		"reprezentowanej",
-		"wpisana",
-		"wpisane",
-		"wpisano",
-		"wpisany",
-		"wpisaną",
-		"wpisanego",
-		"wpisanej",
-		"zarejestrowana",
-		"zarejestrowany",
-		"zarejestrowaną",
-		"zarejestrowanego",
-		"zwana dalej",
-		"zwany dalej",
-		"zwaną dalej",
-		"zwanego dalej"
-	],
-	"pt-br": [
-		"CNPJ",
-		"CPF",
-		"RG",
-		"OAB",
-		"inscrita",
-		"inscrito",
-		"Inscrição Estadual",
-		"Inscrição Municipal",
-		"neste ato representada",
-		"neste ato representado",
-		"doravante denominada",
-		"doravante denominado",
-		"registrada",
-		"registrado",
-		"portador",
-		"portadora"
-	],
-	ro: [
-		"înregistrată",
-		"reprezentat",
-		"reprezentată"
-	],
-	sk: [
-		"IČO",
-		"DIČ",
-		"IČ DPH",
-		"bankové spojenie",
-		"číslo účtu",
-		"konajúci",
-		"konajúca",
-		"kontaktná osoba",
-		"registrovaný",
-		"registrovaná",
-		"splnomocnený zástupca",
-		"splnomocnená zástupkyňa",
-		"štatutárny orgán",
-		"štatutárny zástupca",
-		"v zastúpení",
-		"zapísaná",
-		"zapísaný",
-		"zapísané",
-		"zaregistrovaná",
-		"zaregistrovaný",
-		"zastúpená",
-		"zastúpený",
-		"zastúpené"
-	],
-	universal: [
-		"e-mail",
-		"email",
-		"fax",
-		"http",
-		"https",
-		"phone",
-		"tel",
-		"telephone",
-		"www"
-	]
-};
-//#endregion
-export { address_boundaries_default as default };
-
-//# sourceMappingURL=address-boundaries.mjs.map

package/dist/address-boundaries.mjs.map

@@ -1 +0,0 @@
-{"version":3,"file":"address-boundaries.mjs","names":[],"sources":["../../src/data/address-boundaries.json"],"sourcesContent":[""],"mappings":""}

package/dist/address-jurisdiction-prefixes.mjs

@@ -1,16 +0,0 @@
-//#region src/data/address-jurisdiction-prefixes.json
-var _comment = "Address-like jurisdiction prefixes that are valid location/address spans without digits or street-type words. Lowercased and organized per language.";
-var en = [
-	"commonwealth of",
-	"district of",
-	"state of",
-	"territory of"
-];
-var address_jurisdiction_prefixes_default = {
-	_comment,
-	en
-};
-//#endregion
-export { _comment, address_jurisdiction_prefixes_default as default, en };
-
-//# sourceMappingURL=address-jurisdiction-prefixes.mjs.map

package/dist/address-jurisdiction-prefixes.mjs.map

@@ -1 +0,0 @@
-{"version":3,"file":"address-jurisdiction-prefixes.mjs","names":[],"sources":["../../src/data/address-jurisdiction-prefixes.json"],"sourcesContent":[""],"mappings":""}

package/dist/address-stop-keywords.mjs

@@ -1,148 +0,0 @@
-//#region src/data/address-stop-keywords.json
-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.";
-var cs = [
-	"číslo účtu",
-	"registrační",
-	"zastoupen",
-	"bankovní",
-	"datová",
-	"vložka",
-	"sp.zn.",
-	"jednatel",
-	"ředitel",
-	"oddíl",
-	"telefon",
-	"č.ú.",
-	"dič",
-	"ičo",
-	"ič"
-];
-var de = [
-	"bank",
-	"bic",
-	"iban",
-	"steuer-id",
-	"steueridentifikationsnummer",
-	"steuernummer",
-	"ust-idnr",
-	"ust-idnr."
-];
-var en = [
-	"e-mail",
-	"email",
-	"tel",
-	"swift",
-	"iban",
-	"bic"
-];
-var pl = [
-	"nip",
-	"krs",
-	"regon",
-	"pesel",
-	"numer konta",
-	"konto bankowe",
-	"rachunek bankowy",
-	"tel.",
-	"telefon",
-	"e-mail",
-	"email"
-];
-var fr = [
-	"e-mail",
-	"email",
-	"courriel",
-	"téléphone",
-	"tél.",
-	"tél",
-	"portable",
-	"fax",
-	"siren",
-	"siret",
-	"tva",
-	"rcs",
-	"naf",
-	"ape",
-	"iban",
-	"bic",
-	"swift",
-	"code postal",
-	"cp",
-	"nir"
-];
-var sk = [
-	"číslo účtu",
-	"bankové spojenie",
-	"ičo",
-	"ič dph",
-	"ičdph",
-	"dič",
-	"konateľ",
-	"konateľka",
-	"konateľom",
-	"konateľkou",
-	"riaditeľ",
-	"riaditeľka",
-	"vložka",
-	"oddiel",
-	"sp.zn.",
-	"sp. zn.",
-	"spisovej značky",
-	"č.ú.",
-	"č. účtu"
-];
-var pt_br = [
-	"bic",
-	"cnpj",
-	"cnpj/mf",
-	"conta",
-	"cpf",
-	"cpf/mf",
-	"doravante",
-	"e-mail",
-	"email",
-	"fax",
-	"iban",
-	"inscrição estadual",
-	"inscrição municipal",
-	"neste ato",
-	"nire",
-	"oab",
-	"portador",
-	"portadora",
-	"representada",
-	"representado",
-	"rg",
-	"swift",
-	"tel",
-	"telefone"
-];
-var es = [];
-var it = [
-	"c.f.",
-	"c.f. e p.iva",
-	"codice fiscale",
-	"p. iva",
-	"p.iva",
-	"partita iva",
-	"pec",
-	"registro imprese",
-	"rappresentata",
-	"rappresentato"
-];
-var address_stop_keywords_default = {
-	_comment,
-	cs,
-	de,
-	en,
-	pl,
-	fr,
-	sk,
-	"pt-br": pt_br,
-	es,
-	it
-};
-//#endregion
-export { _comment, cs, de, address_stop_keywords_default as default, en, es, fr, it, pl, pt_br as "pt-br", sk };
-
-//# sourceMappingURL=address-stop-keywords.mjs.map

package/dist/address-stop-keywords.mjs.map

@@ -1 +0,0 @@
-{"version":3,"file":"address-stop-keywords.mjs","names":[],"sources":["../../src/data/address-stop-keywords.json"],"sourcesContent":[""],"mappings":""}

package/dist/address-stopwords.mjs

@@ -1,84 +0,0 @@
-//#region src/data/address-stopwords.json
-var address_stopwords_default = {
-	_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.",
-	_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.",
-	words: [
-		"agenda",
-		"agent",
-		"amount",
-		"basic",
-		"brent",
-		"central",
-		"centro",
-		"client",
-		"code",
-		"company",
-		"consent",
-		"contract",
-		"customer",
-		"default",
-		"delivery",
-		"event",
-		"exhibit",
-		"general",
-		"gordon",
-		"interest",
-		"license",
-		"memorandum",
-		"notice",
-		"owner",
-		"party",
-		"payment",
-		"post",
-		"price",
-		"principal",
-		"rate",
-		"rent",
-		"registro",
-		"schedule",
-		"se",
-		"section",
-		"server",
-		"tenant",
-		"term",
-		"termination",
-		"time",
-		"union",
-		"urbana",
-		"ventosa",
-		"vendor",
-		"waiver",
-		"amministrazione",
-		"appaltante",
-		"appalto",
-		"articolo",
-		"cliente",
-		"concedente",
-		"concessionario",
-		"conduttore",
-		"denominazione",
-		"esecuzione",
-		"fornitore",
-		"indirizzo",
-		"informazioni",
-		"lavoratore",
-		"lavoro",
-		"locatore",
-		"luogo",
-		"magazzino",
-		"mandante",
-		"mandatario",
-		"oggetto",
-		"piano",
-		"procedimento",
-		"procedura",
-		"responsabile",
-		"sede",
-		"sezione",
-		"venditore"
-	]
-};
-//#endregion
-export { address_stopwords_default as default };
-
-//# sourceMappingURL=address-stopwords.mjs.map

package/dist/address-stopwords.mjs.map

@@ -1 +0,0 @@
-{"version":3,"file":"address-stopwords.mjs","names":[],"sources":["../../src/data/address-stopwords.json"],"sourcesContent":[""],"mappings":""}

package/dist/address-unit-abbreviations.mjs

@@ -1,15 +0,0 @@
-//#region src/data/address-unit-abbreviations.json
-var address_unit_abbreviations_default = {
-	_comment: "Dotted address unit abbreviations that should not terminate address seed expansion. Organised per language because abbreviations are locale-specific.",
-	en: [
-		"apt.",
-		"bldg.",
-		"fl.",
-		"ste.",
-		"unit."
-	]
-};
-//#endregion
-export { address_unit_abbreviations_default as default };
-
-//# sourceMappingURL=address-unit-abbreviations.mjs.map

package/dist/address-unit-abbreviations.mjs.map

@@ -1 +0,0 @@
-{"version":3,"file":"address-unit-abbreviations.mjs","names":[],"sources":["../../src/data/address-unit-abbreviations.json"],"sourcesContent":[""],"mappings":""}