@stll/anonymize-wasm 1.0.8 → 1.1.0

package/dist/address-boundaries.mjs

@@ -0,0 +1,85 @@
+//#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",
+		"zapsán",
+		"zapsaná",
+		"zapsané",
+		"zapsaný",
+		"zastoupená",
+		"zastoupené",
+		"zastoupený"
+	],
+	de: [
+		"amtsgericht",
+		"eingetragen",
+		"geschäftsführer",
+		"handelsregister",
+		"vertreten"
+	],
+	en: [
+		"attention",
+		"attn",
+		"c/o",
+		"care of",
+		"contact person",
+		"incorporated",
+		"registered",
+		"represented"
+	],
+	fr: [
+		"immatriculée",
+		"inscrite",
+		"représenté",
+		"représentée"
+	],
+	hu: [
+		"bejegyzett",
+		"cégjegyzékszám",
+		"képviseli"
+	],
+	it: [
+		"iscritta",
+		"rappresentata",
+		"rappresentato",
+		"registro imprese"
+	],
+	pl: [
+		"reprezentowana",
+		"reprezentowany",
+		"wpisana",
+		"wpisane",
+		"wpisano",
+		"wpisany",
+		"zarejestrowana"
+	],
+	ro: [
+		"înregistrată",
+		"reprezentat",
+		"reprezentată"
+	],
+	sk: [
+		"konajúci",
+		"zapísaná",
+		"zapísaný",
+		"zastúpená",
+		"zastúpený"
+	],
+	universal: [
+		"e-mail",
+		"email",
+		"fax",
+		"http",
+		"https",
+		"tel",
+		"www"
+	]
+};
+//#endregion
+export { address_boundaries_default as default };
+
+//# sourceMappingURL=address-boundaries.mjs.map

package/dist/address-boundaries.mjs.map

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

package/dist/address-prepositions.mjs

@@ -0,0 +1,182 @@
+//#region src/data/address-prepositions.json
+var _comment = "Prepositions used in street/address names by language. Used by backward scan to continue through lowercase words, and by temporal guard to skip date-like patterns.";
+var address = {
+	"cs": [
+		"nad",
+		"pod",
+		"u",
+		"na",
+		"v",
+		"ve",
+		"ke",
+		"za",
+		"při",
+		"do",
+		"od",
+		"mezi"
+	],
+	"sk": [
+		"nad",
+		"pod",
+		"pri",
+		"na",
+		"vo",
+		"ke",
+		"za",
+		"do",
+		"od",
+		"medzi"
+	],
+	"de": [
+		"am",
+		"an",
+		"im",
+		"bei",
+		"zum",
+		"zur",
+		"auf",
+		"in",
+		"vor"
+	],
+	"en": [
+		"upon",
+		"on",
+		"in",
+		"by",
+		"at",
+		"near"
+	],
+	"fr": [
+		"sur",
+		"sous",
+		"de",
+		"du",
+		"des",
+		"aux",
+		"en"
+	],
+	"pl": [
+		"nad",
+		"pod",
+		"przy",
+		"na",
+		"za",
+		"do",
+		"od",
+		"między"
+	],
+	"it": [
+		"di",
+		"del",
+		"della",
+		"sul",
+		"al"
+	],
+	"es": [
+		"de",
+		"del",
+		"en",
+		"sobre"
+	],
+	"hu": [
+		"melletti",
+		"alatti",
+		"feletti"
+	],
+	"ro": [
+		"de",
+		"din",
+		"pe",
+		"la"
+	],
+	"sv": [
+		"vid",
+		"på",
+		"i"
+	]
+};
+var temporal = {
+	"_comment": "Prepositions that indicate temporal context (dates), not addresses. Used as negative guard.",
+	"cs": [
+		"od",
+		"do",
+		"ke",
+		"na",
+		"dne",
+		"ze",
+		"ve",
+		"za",
+		"při",
+		"mezi"
+	],
+	"sk": [
+		"od",
+		"do",
+		"dňa",
+		"na",
+		"vo",
+		"za"
+	],
+	"de": [
+		"ab",
+		"bis",
+		"am",
+		"vom",
+		"seit"
+	],
+	"en": [
+		"from",
+		"to",
+		"on",
+		"at",
+		"since",
+		"until",
+		"by"
+	],
+	"fr": [
+		"du",
+		"au",
+		"dès",
+		"depuis",
+		"le"
+	],
+	"pl": [
+		"od",
+		"do",
+		"dnia",
+		"na",
+		"za"
+	],
+	"it": [
+		"dal",
+		"al",
+		"fino",
+		"da"
+	],
+	"es": [
+		"del",
+		"al",
+		"desde",
+		"hasta"
+	],
+	"hu": ["tól", "ig"],
+	"ro": [
+		"de",
+		"din",
+		"până"
+	],
+	"sv": [
+		"från",
+		"till",
+		"den"
+	]
+};
+var address_prepositions_default = {
+	_comment,
+	address,
+	temporal
+};
+//#endregion
+export { _comment, address, address_prepositions_default as default, temporal };
+
+//# sourceMappingURL=address-prepositions.mjs.map

package/dist/address-prepositions.mjs.map

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

package/dist/address-street-types.mjs

@@ -0,0 +1,162 @@
+//#region src/data/address-street-types.json
+var _comment = "Street type words by language. Used by address seed detector to identify street names via Aho-Corasick.";
+var cs = [
+	"alej",
+	"bulvár",
+	"nábř.",
+	"nábřeží",
+	"nám.",
+	"náměstí",
+	"sady",
+	"tř.",
+	"třída",
+	"ul.",
+	"ulice"
+];
+var de = [
+	"Allee",
+	"Gasse",
+	"Platz",
+	"Ring",
+	"Straße",
+	"Strasse",
+	"Weg"
+];
+var en = [
+	"Ave.",
+	"Avenue",
+	"Blvd.",
+	"Boulevard",
+	"Circle",
+	"Court",
+	"Crescent",
+	"Drive",
+	"Highway",
+	"Lane",
+	"Parkway",
+	"Place",
+	"Rd.",
+	"Road",
+	"Square",
+	"St.",
+	"Street",
+	"Terrace",
+	"Way"
+];
+var es = [
+	"avenida",
+	"bulevar",
+	"calle",
+	"camino",
+	"carretera",
+	"glorieta",
+	"paseo",
+	"plaza",
+	"ronda",
+	"travesía"
+];
+var fr = [
+	"allée",
+	"avenue",
+	"boulevard",
+	"chemin",
+	"impasse",
+	"place",
+	"quai",
+	"rue"
+];
+var hu = [
+	"fasor",
+	"körút",
+	"köz",
+	"krt.",
+	"rakpart",
+	"sétány",
+	"sor",
+	"tér",
+	"u.",
+	"út",
+	"utca"
+];
+var it = [
+	"corso",
+	"largo",
+	"lungarno",
+	"lungotevere",
+	"piazza",
+	"piazzale",
+	"via",
+	"viale",
+	"vicolo"
+];
+var nl = [
+	"gracht",
+	"kade",
+	"laan",
+	"plein",
+	"straat",
+	"weg"
+];
+var pl = [
+	"al.",
+	"aleja",
+	"bulwar",
+	"pl.",
+	"plac",
+	"rondo",
+	"skwer",
+	"ul.",
+	"ulica"
+];
+var pt = [
+	"avenida",
+	"largo",
+	"praça",
+	"rua",
+	"travessa"
+];
+var ro = [
+	"aleea",
+	"bd.",
+	"bulevardul",
+	"calea",
+	"fundătura",
+	"intrare",
+	"piața",
+	"șoseaua",
+	"splaiul",
+	"str.",
+	"strada"
+];
+var sk = [
+	"námest.",
+	"námestie",
+	"ulica"
+];
+var sv = [
+	"allé",
+	"backe",
+	"gata",
+	"gränd",
+	"torg"
+];
+var address_street_types_default = {
+	_comment,
+	cs,
+	de,
+	en,
+	es,
+	fr,
+	hu,
+	it,
+	nl,
+	pl,
+	pt,
+	ro,
+	sk,
+	sv
+};
+//#endregion
+export { _comment, cs, de, address_street_types_default as default, en, es, fr, hu, it, nl, pl, pt, ro, sk, sv };
+
+//# sourceMappingURL=address-street-types.mjs.map

package/dist/address-street-types.mjs.map

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

package/dist/allow-list.mjs

@@ -0,0 +1,183 @@
+//#region src/data/allow-list.json
+var allow_list_default = {
+	_comment: "Words that should never be flagged as PII. Includes common English/Czech words that are also surnames in the name corpus.",
+	words: [
+		"absence",
+		"acceptance",
+		"account",
+		"action",
+		"address",
+		"agreement",
+		"amu",
+		"annual",
+		"applicable",
+		"balance",
+		"base",
+		"benefit",
+		"bic",
+		"board",
+		"bonus",
+		"business",
+		"capital",
+		"case",
+		"cause",
+		"change",
+		"chief",
+		"cisco",
+		"claim",
+		"close",
+		"code",
+		"comfort",
+		"common",
+		"company",
+		"condition",
+		"confidential",
+		"consent",
+		"counsel",
+		"credit",
+		"čr",
+		"čssr",
+		"current",
+		"damages",
+		"day",
+		"default",
+		"dell",
+		"demand",
+		"deposit",
+		"determination",
+		"disability",
+		"discharge",
+		"dph",
+		"effective",
+		"equity",
+		"eu",
+		"event",
+		"evidence",
+		"exchange",
+		"executive",
+		"exercise",
+		"expense",
+		"fair",
+		"federal",
+		"final",
+		"flood",
+		"force",
+		"full",
+		"fund",
+		"gdpr",
+		"general",
+		"good",
+		"grace",
+		"grant",
+		"gross",
+		"holding",
+		"house",
+		"income",
+		"initial",
+		"insurance",
+		"interest",
+		"law",
+		"ldn",
+		"lease",
+		"letter",
+		"lien",
+		"loan",
+		"loss",
+		"major",
+		"mark",
+		"material",
+		"matters",
+		"merit",
+		"mfčr",
+		"minor",
+		"mobile",
+		"moral",
+		"mpsv",
+		"net",
+		"no",
+		"not",
+		"note",
+		"notice",
+		"officer",
+		"option",
+		"order",
+		"otter",
+		"pacht",
+		"page",
+		"parent",
+		"party",
+		"patent",
+		"payment",
+		"pension",
+		"period",
+		"plan",
+		"pledge",
+		"power",
+		"practice",
+		"president",
+		"prime",
+		"principal",
+		"property",
+		"purchase",
+		"rate",
+		"republic",
+		"follows",
+		"lhůta",
+		"read",
+		"reason",
+		"record",
+		"release",
+		"rent",
+		"report",
+		"reserve",
+		"rest",
+		"review",
+		"right",
+		"risk",
+		"rule",
+		"salary",
+		"sale",
+		"seal",
+		"secretary",
+		"section",
+		"seek",
+		"senior",
+		"separation",
+		"server",
+		"service",
+		"severance",
+		"share",
+		"short",
+		"single",
+		"sleva",
+		"smlouva",
+		"sole",
+		"special",
+		"spory",
+		"sr",
+		"standard",
+		"stock",
+		"storage",
+		"subject",
+		"superior",
+		"supplement",
+		"surrender",
+		"tax",
+		"title",
+		"škola",
+		"tato",
+		"term",
+		"to",
+		"total",
+		"transfer",
+		"trust",
+		"value",
+		"vice",
+		"warrant",
+		"will"
+	]
+};
+//#endregion
+export { allow_list_default as default };
+
+//# sourceMappingURL=allow-list.mjs.map

package/dist/allow-list.mjs.map

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

package/dist/amount-words.mjs

@@ -0,0 +1,42 @@
+//#region src/data/amount-words.json
+var amount_words_default = {
+	_comment: "Patterns for written-out monetary amounts in parentheses, by language.",
+	patterns: [
+		{
+			"lang": "cs",
+			"keywords": ["slovy", "slovně"]
+		},
+		{
+			"lang": "sk",
+			"keywords": ["slovom", "slovami"]
+		},
+		{
+			"lang": "de",
+			"keywords": ["in Worten", "in Buchstaben"]
+		},
+		{
+			"lang": "en",
+			"keywords": ["in words", "in letters"]
+		},
+		{
+			"lang": "fr",
+			"keywords": ["en lettres", "en toutes lettres"]
+		},
+		{
+			"lang": "pl",
+			"keywords": ["słownie"]
+		},
+		{
+			"lang": "it",
+			"keywords": ["in lettere"]
+		},
+		{
+			"lang": "es",
+			"keywords": ["en letras"]
+		}
+	]
+};
+//#endregion
+export { amount_words_default as default };
+
+//# sourceMappingURL=amount-words.mjs.map

package/dist/amount-words.mjs.map

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

package/dist/coreference.cs.mjs

@@ -0,0 +1,14 @@
+//#region src/data/coreference.cs.json
+var coreference_cs_default = [{
+	"label": "(dale jen \"X\") or (dale jen 'X') or (dale jen \"X\")",
+	"pattern": "\\(dále\\s+jen\\s+[„\"'‚]([^\"'\"\"]+)[\"\"']\\)",
+	"flags": "gi"
+}, {
+	"label": "dale jen \"X\" without parens",
+	"pattern": "dále\\s+jen\\s+[„\"'‚]([^\"'\"\"]+)[\"\"']",
+	"flags": "gi"
+}];
+//#endregion
+export { coreference_cs_default as default };
+
+//# sourceMappingURL=coreference.cs.mjs.map

package/dist/coreference.cs.mjs.map

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

package/dist/coreference.de.mjs

@@ -0,0 +1,14 @@
+//#region src/data/coreference.de.json
+var coreference_de_default = [{
+	"label": "(nachfolgend \"X\") or (im Folgenden \"X\")",
+	"pattern": "\\((?:nachfolgend|im\\s+Folgenden)\\s+[„\"'‚]([^\"'\"\"]+)[\"\"']\\)",
+	"flags": "gi"
+}, {
+	"label": "nachfolgend \"X\" without parens",
+	"pattern": "(?:nachfolgend|im\\s+Folgenden)\\s+[„\"'‚]([^\"'\"\"]+)[\"\"']",
+	"flags": "gi"
+}];
+//#endregion
+export { coreference_de_default as default };
+
+//# sourceMappingURL=coreference.de.mjs.map

package/dist/coreference.de.mjs.map

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