@stll/anonymize-wasm 1.3.0 → 1.4.1

package/dist/address-street-types.mjs

@@ -1,4 +1,23 @@
+import { t as __exportAll } from "./rolldown-runtime.mjs";
 //#region src/data/address-street-types.json
+var address_street_types_exports = /* @__PURE__ */ __exportAll({
+	_comment: () => _comment,
+	cs: () => cs,
+	de: () => de,
+	default: () => address_street_types_default,
+	en: () => en,
+	es: () => es,
+	fr: () => fr,
+	hu: () => hu,
+	it: () => it,
+	nl: () => nl,
+	pl: () => pl,
+	pt: () => pt,
+	"pt-br": () => pt_br,
+	ro: () => ro,
+	sk: () => sk,
+	sv: () => sv
+});
 var _comment = "Street type words by language. Used by address seed detector to identify street names via Aho-Corasick.";
 var cs = [
 	"alej",
@@ -220,6 +239,6 @@ var address_street_types_default = {
 	sv
 };
 //#endregion
-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 };
+export { address_street_types_exports as n, address_street_types_default as t };
 
 //# sourceMappingURL=address-street-types.mjs.map

package/dist/allow-list.mjs

@@ -17,6 +17,7 @@ var allow_list_default = {
 		"bic",
 		"board",
 		"bonus",
+		"branch",
 		"business",
 		"capital",
 		"case",
@@ -59,6 +60,8 @@ var allow_list_default = {
 		"expense",
 		"fair",
 		"federal",
+		"fee",
+		"fees",
 		"final",
 		"flood",
 		"force",
@@ -77,9 +80,11 @@ var allow_list_default = {
 		"insurance",
 		"interest",
 		"law",
+		"laws",
 		"ldn",
 		"lease",
 		"letter",
+		"letters",
 		"lien",
 		"loan",
 		"loss",
@@ -87,6 +92,8 @@ var allow_list_default = {
 		"mark",
 		"material",
 		"matters",
+		"measure",
+		"measures",
 		"merit",
 		"mfčr",
 		"minor",
@@ -99,6 +106,7 @@ var allow_list_default = {
 		"note",
 		"notice",
 		"officer",
+		"oil",
 		"option",
 		"order",
 		"otter",
@@ -134,6 +142,7 @@ var allow_list_default = {
 		"review",
 		"right",
 		"risk",
+		"road",
 		"rule",
 		"salary",
 		"sale",
@@ -147,6 +156,7 @@ var allow_list_default = {
 		"service",
 		"severance",
 		"share",
+		"shares",
 		"short",
 		"single",
 		"sleva",
@@ -173,6 +183,8 @@ var allow_list_default = {
 		"trust",
 		"value",
 		"vice",
+		"vote",
+		"votes",
 		"warrant",
 		"will"
 	]

package/dist/amount-words.mjs

@@ -1,42 +1,478 @@
+import { t as __exportAll } from "./rolldown-runtime.mjs";
 //#region src/data/amount-words.json
+var amount_words_exports = /* @__PURE__ */ __exportAll({
+	_comment: () => _comment,
+	default: () => amount_words_default,
+	magnitudeSuffixes: () => magnitudeSuffixes,
+	patterns: () => patterns,
+	percentages: () => percentages,
+	shareQuantityTerms: () => shareQuantityTerms
+});
+var _comment = "Patterns for written-out monetary amounts, magnitude suffixes, share-quantity guards, and numeric percentage phrases, by language.";
+var 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"]
+	}
+];
+var percentages = [{
+	"lang": "en",
+	"keywords": ["percent"],
+	"standalone": ["zero", "one hundred"],
+	"ones": [
+		"one",
+		"two",
+		"three",
+		"four",
+		"five",
+		"six",
+		"seven",
+		"eight",
+		"nine"
+	],
+	"teens": [
+		"ten",
+		"eleven",
+		"twelve",
+		"thirteen",
+		"fourteen",
+		"fifteen",
+		"sixteen",
+		"seventeen",
+		"eighteen",
+		"nineteen"
+	],
+	"tens": [
+		"twenty",
+		"thirty",
+		"forty",
+		"fifty",
+		"sixty",
+		"seventy",
+		"eighty",
+		"ninety"
+	],
+	"allowSpaceCompoundSeparator": true
+}];
+var magnitudeSuffixes = [
+	{
+		"lang": "cs",
+		"words": [
+			"sto",
+			"sta",
+			"stě",
+			"sty",
+			"stovka",
+			"stovky",
+			"stovek",
+			"tisíc",
+			"tisíce",
+			"tisíců",
+			"milion",
+			"miliony",
+			"milionů",
+			"miliarda",
+			"miliardy",
+			"miliard",
+			"bilion",
+			"biliony",
+			"bilionů"
+		],
+		"abbreviationsCaseInsensitive": ["mld"],
+		"abbreviationsCaseSensitive": ["K", "M"]
+	},
+	{
+		"lang": "de",
+		"words": [
+			"hundert",
+			"hunderte",
+			"hunderten",
+			"tausend",
+			"million",
+			"millionen",
+			"milliarde",
+			"milliarden",
+			"billion",
+			"billionen"
+		],
+		"abbreviationsCaseInsensitive": ["mio", "mrd"],
+		"abbreviationsCaseSensitive": ["K", "M"]
+	},
+	{
+		"lang": "en",
+		"words": [
+			"hundred",
+			"hundreds",
+			"thousand",
+			"thousands",
+			"million",
+			"millions",
+			"billion",
+			"billions",
+			"trillion",
+			"trillions",
+			"quadrillion",
+			"quadrillions"
+		],
+		"abbreviationsCaseInsensitive": [
+			"bn",
+			"bln",
+			"tn",
+			"trn"
+		],
+		"abbreviationsCaseSensitive": ["K", "M"]
+	},
+	{
+		"lang": "es",
+		"words": [
+			"cien",
+			"ciento",
+			"cientos",
+			"millón",
+			"millones",
+			"billón",
+			"billones"
+		],
+		"abbreviationsCaseInsensitive": ["mln"],
+		"abbreviationsCaseSensitive": ["K", "M"]
+	},
+	{
+		"lang": "fr",
+		"words": [
+			"cent",
+			"cents",
+			"centaine",
+			"centaines",
+			"mille",
+			"million",
+			"millions",
+			"milliard",
+			"milliards",
+			"billion",
+			"billions"
+		],
+		"abbreviationsCaseInsensitive": [
+			"mio",
+			"md",
+			"mrd"
+		],
+		"abbreviationsCaseSensitive": ["K", "M"]
+	},
+	{
+		"lang": "it",
+		"words": [
+			"cento",
+			"centinaio",
+			"centinaia",
+			"mille",
+			"milione",
+			"milioni",
+			"miliardo",
+			"miliardi",
+			"bilione",
+			"bilioni"
+		],
+		"abbreviationsCaseInsensitive": ["mln", "mld"],
+		"abbreviationsCaseSensitive": ["K", "M"]
+	},
+	{
+		"lang": "pl",
+		"words": [
+			"sto",
+			"sta",
+			"stów",
+			"stoma",
+			"setki",
+			"setek",
+			"tysiąc",
+			"tysiące",
+			"tysięcy",
+			"milion",
+			"miliony",
+			"milionów",
+			"miliard",
+			"miliardy",
+			"miliardów",
+			"bilion",
+			"biliony",
+			"bilionów"
+		],
+		"abbreviationsCaseInsensitive": ["mln", "mld"],
+		"abbreviationsCaseSensitive": ["K", "M"]
+	},
+	{
+		"lang": "pt-br",
+		"words": [
+			"cem",
+			"cento",
+			"centena",
+			"centenas",
+			"milhão",
+			"milhões",
+			"bilhão",
+			"bilhões",
+			"trilhão",
+			"trilhões"
+		],
+		"abbreviationsCaseInsensitive": [],
+		"abbreviationsCaseSensitive": ["K", "M"]
+	},
+	{
+		"lang": "sk",
+		"words": [
+			"sto",
+			"sta",
+			"stoviek",
+			"stovka",
+			"stovky",
+			"tisíc",
+			"tisíce",
+			"tisícov",
+			"milión",
+			"milióny",
+			"miliónov",
+			"milion",
+			"miliony",
+			"milionov",
+			"miliarda",
+			"miliardy",
+			"miliárd",
+			"bilión",
+			"bilióny",
+			"biliónov"
+		],
+		"abbreviationsCaseInsensitive": ["mld"],
+		"abbreviationsCaseSensitive": ["K", "M"]
+	}
+];
+var shareQuantityTerms = [
+	{
+		"lang": "cs",
+		"modifiers": [
+			"kmenové",
+			"obyčejné",
+			"prioritní",
+			"registrované",
+			"vlastní",
+			"hlasovací",
+			"nehlasovací",
+			"omezené",
+			"třída",
+			"série"
+		],
+		"nouns": [
+			"akcie",
+			"cenné papíry",
+			"podíly",
+			"jednotky"
+		]
+	},
+	{
+		"lang": "de",
+		"modifiers": [
+			"stamm",
+			"vorzugs",
+			"namens",
+			"eigene",
+			"stimmberechtigte",
+			"stimmrechtslose",
+			"beschränkte",
+			"klasse",
+			"serie"
+		],
+		"nouns": [
+			"aktien",
+			"wertpapiere",
+			"anteile",
+			"einheiten"
+		]
+	},
+	{
+		"lang": "en",
+		"modifiers": [
+			"ordinary",
+			"common",
+			"preferred",
+			"registered",
+			"treasury",
+			"voting",
+			"non-voting",
+			"restricted",
+			"class",
+			"series"
+		],
+		"nouns": [
+			"share",
+			"shares",
+			"stock",
+			"stocks",
+			"security",
+			"securities",
+			"unit",
+			"units"
+		]
+	},
+	{
+		"lang": "es",
+		"modifiers": [
+			"ordinarias",
+			"comunes",
+			"preferentes",
+			"nominativas",
+			"propias",
+			"con derecho a voto",
+			"sin derecho a voto",
+			"restringidas",
+			"clase",
+			"serie"
+		],
+		"nouns": [
+			"acciones",
+			"valores",
+			"participaciones",
+			"unidades"
+		]
+	},
+	{
+		"lang": "fr",
+		"modifiers": [
+			"ordinaires",
+			"communes",
+			"privilégiées",
+			"nominatives",
+			"propres",
+			"votantes",
+			"sans droit de vote",
+			"restreintes",
+			"classe",
+			"série"
+		],
+		"nouns": [
+			"actions",
+			"titres",
+			"unités"
+		]
+	},
+	{
+		"lang": "it",
+		"modifiers": [
+			"ordinarie",
+			"comuni",
+			"privilegiate",
+			"nominative",
+			"proprie",
+			"votanti",
+			"senza diritto di voto",
+			"vincolate",
+			"classe",
+			"serie"
+		],
+		"nouns": [
+			"azioni",
+			"titoli",
+			"quote",
+			"unità"
+		]
+	},
+	{
+		"lang": "pl",
+		"modifiers": [
+			"zwykłe",
+			"imienne",
+			"uprzywilejowane",
+			"własne",
+			"głosujące",
+			"nieme",
+			"ograniczone",
+			"klasa",
+			"seria"
+		],
+		"nouns": [
+			"akcje",
+			"papiery wartościowe",
+			"udziały",
+			"jednostki"
+		]
+	},
+	{
+		"lang": "pt-br",
+		"modifiers": [
+			"ordinárias",
+			"comuns",
+			"preferenciais",
+			"nominativas",
+			"próprias",
+			"votantes",
+			"sem direito a voto",
+			"restritas",
+			"classe",
+			"série"
+		],
+		"nouns": [
+			"ações",
+			"valores mobiliários",
+			"participações",
+			"unidades"
+		]
+	},
+	{
+		"lang": "sk",
+		"modifiers": [
+			"kmeňové",
+			"obyčajné",
+			"prioritné",
+			"registrované",
+			"vlastné",
+			"hlasovacie",
+			"nehlasovacie",
+			"obmedzené",
+			"trieda",
+			"séria"
+		],
+		"nouns": [
+			"akcie",
+			"cenné papiere",
+			"podiely",
+			"jednotky"
+		]
+	}
+];
 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"]
-		}
-	]
+	_comment,
+	patterns,
+	percentages,
+	magnitudeSuffixes,
+	shareQuantityTerms
 };
 //#endregion
-export { amount_words_default as default };
+export { amount_words_exports as n, amount_words_default as t };
 
 //# sourceMappingURL=amount-words.mjs.map

package/dist/constants.d.mts

@@ -0,0 +1,2 @@
+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";
+export { DEFAULT_ENTITY_LABELS, DETECTION_SOURCES, DETECTOR_PRIORITY, DetectionSource, OPERATOR_TYPES, OperatorType };

package/dist/constants.mjs

@@ -0,0 +1,85 @@
+//#region src/constants.ts
+/**
+* Runtime-free constants for the anonymization pipeline.
+*
+* This module is the SSR-safe / browser-safe entrypoint:
+* importing it must not pull in `@stll/text-search`,
+* `@stll/anonymize-wasm`, or any other runtime-bearing
+* module. Consumers that only need the static label list,
+* detection-source identifiers, or operator names import
+* from `@stll/anonymize/constants` (or
+* `@stll/anonymize-wasm/constants`) without paying the
+* wasm / regex-set startup cost.
+*
+* `types.ts` re-exports these for back-compat, so existing
+* `import { DEFAULT_ENTITY_LABELS } from "@stll/anonymize"`
+* call sites keep working.
+*/
+/**
+* Source of a detected entity span.
+* Ordered by detection layer in the pipeline.
+*/
+const DETECTION_SOURCES = {
+	TRIGGER: "trigger",
+	REGEX: "regex",
+	DENY_LIST: "deny-list",
+	LEGAL_FORM: "legal-form",
+	GAZETTEER: "gazetteer",
+	NER: "ner",
+	COREFERENCE: "coreference"
+};
+/**
+* Priority levels for detection sources.
+* Higher = more structurally reliable. Used during
+* overlap resolution so deterministic detectors beat
+* probabilistic ones regardless of raw score.
+*/
+const DETECTOR_PRIORITY = {
+	[DETECTION_SOURCES.GAZETTEER]: 5,
+	[DETECTION_SOURCES.TRIGGER]: 4,
+	[DETECTION_SOURCES.LEGAL_FORM]: 3,
+	[DETECTION_SOURCES.REGEX]: 3,
+	[DETECTION_SOURCES.DENY_LIST]: 2,
+	[DETECTION_SOURCES.COREFERENCE]: 2,
+	[DETECTION_SOURCES.NER]: 1
+};
+/**
+* Anonymization operator types. Each operator defines
+* how a confirmed entity is replaced in the output.
+*/
+const OPERATOR_TYPES = ["replace", "redact"];
+/**
+* Canonical entity labels used across the pipeline.
+* NER models may use different native labels; the bench
+* NER wrapper maps model output to these canonical names.
+*
+* These labels are ephemeral: entities are regenerated on
+* every pipeline run and never persisted to the database.
+* Renaming a label here requires no migration.
+*/
+const DEFAULT_ENTITY_LABELS = [
+	"person",
+	"organization",
+	"phone number",
+	"address",
+	"email address",
+	"date",
+	"date of birth",
+	"bank account number",
+	"iban",
+	"tax identification number",
+	"identity card number",
+	"birth number",
+	"national identification number",
+	"social security number",
+	"registration number",
+	"credit card number",
+	"passport number",
+	"monetary amount",
+	"land parcel",
+	"misc"
+];
+//#endregion
+export { DEFAULT_ENTITY_LABELS, DETECTION_SOURCES, DETECTOR_PRIORITY, OPERATOR_TYPES };
+
+//# sourceMappingURL=constants.mjs.map

package/dist/constants.mjs.map

@@ -0,0 +1 @@
+{"version":3,"file":"constants.mjs","names":[],"sources":["../../src/constants.ts"],"sourcesContent":["/**\n * Runtime-free constants for the anonymization pipeline.\n *\n * This module is the SSR-safe / browser-safe entrypoint:\n * importing it must not pull in `@stll/text-search`,\n * `@stll/anonymize-wasm`, or any other runtime-bearing\n * module. Consumers that only need the static label list,\n * detection-source identifiers, or operator names import\n * from `@stll/anonymize/constants` (or\n * `@stll/anonymize-wasm/constants`) without paying the\n * wasm / regex-set startup cost.\n *\n * `types.ts` re-exports these for back-compat, so existing\n * `import { DEFAULT_ENTITY_LABELS } from \"@stll/anonymize\"`\n * call sites keep working.\n */\n\n/**\n * Source of a detected entity span.\n * Ordered by detection layer in the pipeline.\n */\nexport const DETECTION_SOURCES = {\n  TRIGGER: \"trigger\",\n  REGEX: \"regex\",\n  DENY_LIST: \"deny-list\",\n  LEGAL_FORM: \"legal-form\",\n  GAZETTEER: \"gazetteer\",\n  NER: \"ner\",\n  COREFERENCE: \"coreference\",\n} as const;\n\nexport type DetectionSource =\n  (typeof DETECTION_SOURCES)[keyof typeof DETECTION_SOURCES];\n\n/**\n * Priority levels for detection sources.\n * Higher = more structurally reliable. Used during\n * overlap resolution so deterministic detectors beat\n * probabilistic ones regardless of raw score.\n */\nexport const DETECTOR_PRIORITY = {\n  [DETECTION_SOURCES.GAZETTEER]: 5,\n  [DETECTION_SOURCES.TRIGGER]: 4,\n  [DETECTION_SOURCES.LEGAL_FORM]: 3,\n  [DETECTION_SOURCES.REGEX]: 3,\n  [DETECTION_SOURCES.DENY_LIST]: 2,\n  [DETECTION_SOURCES.COREFERENCE]: 2,\n  [DETECTION_SOURCES.NER]: 1,\n} as const satisfies Record<DetectionSource, number>;\n\n/**\n * Anonymization operator types. Each operator defines\n * how a confirmed entity is replaced in the output.\n */\nexport const OPERATOR_TYPES = [\"replace\", \"redact\"] as const;\n\nexport type OperatorType = (typeof OPERATOR_TYPES)[number];\n\n/**\n * Canonical entity labels used across the pipeline.\n * NER models may use different native labels; the bench\n * NER wrapper maps model output to these canonical names.\n *\n * These labels are ephemeral: entities are regenerated on\n * every pipeline run and never persisted to the database.\n * Renaming a label here requires no migration.\n */\nexport const DEFAULT_ENTITY_LABELS = [\n  \"person\",\n  \"organization\",\n  \"phone number\",\n  \"address\",\n  \"email address\",\n  \"date\",\n  \"date of birth\",\n  \"bank account number\",\n  \"iban\",\n  \"tax identification number\",\n  \"identity card number\",\n  \"birth number\",\n  \"national identification number\",\n  \"social security number\",\n  \"registration number\",\n  \"credit card number\",\n  \"passport number\",\n  \"monetary amount\",\n  \"land parcel\",\n  \"misc\",\n] as const;\n"],"mappings":";;;;;;;;;;;;;;;;;;;;;AAqBA,MAAa,oBAAoB;CAC/B,SAAS;CACT,OAAO;CACP,WAAW;CACX,YAAY;CACZ,WAAW;CACX,KAAK;CACL,aAAa;CACd;;;;;;;AAWD,MAAa,oBAAoB;EAC9B,kBAAkB,YAAY;EAC9B,kBAAkB,UAAU;EAC5B,kBAAkB,aAAa;EAC/B,kBAAkB,QAAQ;EAC1B,kBAAkB,YAAY;EAC9B,kBAAkB,cAAc;EAChC,kBAAkB,MAAM;CAC1B;;;;;AAMD,MAAa,iBAAiB,CAAC,WAAW,SAAS;;;;;;;;;;AAanD,MAAa,wBAAwB;CACnC;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACD"}