@stll/anonymize-wasm 1.4.5 → 1.4.6

package/dist/address-boundaries.mjs

@@ -183,7 +183,9 @@ var address_boundaries_default = {
 		"fax",
 		"http",
 		"https",
+		"phone",
 		"tel",
+		"telephone",
 		"www"
 	]
 };

package/dist/constants.mjs.map

@@ -1 +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  COUNTRY: \"country\",\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.COUNTRY]: 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  \"country\",\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,SAAS;CACT,KAAK;CACL,aAAa;CACd;;;;;;;AAWD,MAAa,oBAAoB;EAC9B,kBAAkB,YAAY;EAC9B,kBAAkB,UAAU;EAC5B,kBAAkB,aAAa;EAC/B,kBAAkB,QAAQ;EAC1B,kBAAkB,UAAU;EAC5B,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;CACA;CACD"}
+{"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  COUNTRY: \"country\",\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.COUNTRY]: 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  \"country\",\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,SAAS;CACT,KAAK;CACL,aAAa;AACf;;;;;;;AAWA,MAAa,oBAAoB;EAC9B,kBAAkB,YAAY;EAC9B,kBAAkB,UAAU;EAC5B,kBAAkB,aAAa;EAC/B,kBAAkB,QAAQ;EAC1B,kBAAkB,UAAU;EAC5B,kBAAkB,YAAY;EAC9B,kBAAkB,cAAc;EAChC,kBAAkB,MAAM;AAC3B;;;;;AAMA,MAAa,iBAAiB,CAAC,WAAW,QAAQ;;;;;;;;;;AAalD,MAAa,wBAAwB;CACnC;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;AACF"}

package/dist/document-structure-headings.mjs

@@ -0,0 +1,90 @@
+//#region src/data/document-structure-headings.json
+var _comment = "Document-structure heading words per language. Captured by the trigger detector as organizations when followed by 'no. N' / 'č. N' / 'Nr. N' / similar, they are scaffolding (appendix titles, schedule headings, exhibit labels), not party names. Each entry lists the heading words; the false-positive filter rejects an organization span whose first word matches one of these followed by a numerical ordinal abbreviation and a digit. Lowercased; matched case-insensitively. Organized per-language because the heading vocabulary is language-specific; at runtime the filter unions every language's list into a single set because match spans don't carry a language tag.";
+var cs = [
+	"příloha",
+	"dodatek",
+	"schéma",
+	"exhibit",
+	"tabulka"
+];
+var sk = [
+	"príloha",
+	"dodatok",
+	"exhibit",
+	"tabuľka"
+];
+var pl = [
+	"załącznik",
+	"aneks",
+	"schemat",
+	"tabela"
+];
+var de = [
+	"anlage",
+	"anhang",
+	"anlage",
+	"exhibit",
+	"tabelle"
+];
+var en = [
+	"appendix",
+	"annex",
+	"exhibit",
+	"schedule",
+	"attachment",
+	"table"
+];
+var fr = [
+	"annexe",
+	"appendice",
+	"tableau",
+	"exhibit"
+];
+var it = [
+	"allegato",
+	"appendice",
+	"tabella",
+	"exhibit"
+];
+var es = [
+	"anexo",
+	"apéndice",
+	"tabla",
+	"exhibit"
+];
+var pt_br = [
+	"anexo",
+	"apêndice",
+	"tabela",
+	"exhibit"
+];
+var ro = [
+	"anexa",
+	"apendice",
+	"tabel",
+	"exhibit"
+];
+var hu = [
+	"melléklet",
+	"függelék",
+	"táblázat",
+	"exhibit"
+];
+var document_structure_headings_default = {
+	_comment,
+	cs,
+	sk,
+	pl,
+	de,
+	en,
+	fr,
+	it,
+	es,
+	"pt-br": pt_br,
+	ro,
+	hu
+};
+//#endregion
+export { _comment, cs, de, document_structure_headings_default as default, en, es, fr, hu, it, pl, pt_br as "pt-br", ro, sk };
+
+//# sourceMappingURL=document-structure-headings.mjs.map

package/dist/document-structure-headings.mjs.map

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

package/dist/sentence-verb-indicators.mjs

@@ -47,7 +47,16 @@ var en = [
 	"signs",
 	"signed",
 	"leases",
-	"rents"
+	"rents",
+	"entered",
+	"made",
+	"executed",
+	"retained",
+	"hired",
+	"engaged",
+	"appointed",
+	"named",
+	"sued"
 ];
 var de = [
 	"ist",

package/dist/triggers.de.mjs

@@ -49,8 +49,8 @@ var triggers_de_default = [
 		"id": "de-iban",
 		"label": "iban",
 		"strategy": {
-			"type": "n-words",
-			"count": 1
+			"type": "match-pattern",
+			"pattern": "\\b[A-Z]{2}\\d{2}(?=(?:\\s?[\\dA-Z]){11,30}\\b)(?:\\s?[\\dA-Z]{1,4}){3,8}\\b"
 		},
 		"triggers": ["IBAN"],
 		"extensions": ["add-colon"]

package/dist/vite.mjs.map

@@ -1 +1 @@
-{"version":3,"file":"vite.mjs","names":[],"sources":["../../src/vite.ts"],"sourcesContent":["/* Vite plugin that wires up @stll/anonymize-wasm so its transitive\n * napi-rs-generated wasm loaders survive Vite's dep pre-bundler. Adds\n * this package plus its @stll/*-wasm dependencies to optimizeDeps.exclude\n * so the loader modules are served with their original paths, keeping\n * `new URL(\"./foo.wasm\", import.meta.url)` correct. */\nimport type { Plugin } from \"vite\";\n\n// Only the napi-rs WASM packages need excluding — they use\n// `new URL(\"./foo.wasm\", import.meta.url)` which Vite's optimizer\n// would rewrite. @stll/anonymize-wasm itself is pure JS and should\n// NOT be excluded.\nconst PACKAGES = [\n  \"@stll/text-search-wasm\",\n  \"@stll/aho-corasick-wasm\",\n  \"@stll/aho-corasick-wasm32-wasi\",\n  \"@stll/fuzzy-search-wasm\",\n  \"@stll/fuzzy-search-wasm32-wasi\",\n  \"@stll/regex-set-wasm\",\n  \"@stll/regex-set-wasm32-wasi\",\n];\n\nexport default function stllAnonymizeWasmVite(): Plugin {\n  return {\n    name: \"stll-anonymize-wasm\",\n    apply: \"serve\",\n    config() {\n      return {\n        optimizeDeps: { exclude: PACKAGES },\n      };\n    },\n  };\n}\n"],"mappings":";AAWA,MAAM,WAAW;CACf;CACA;CACA;CACA;CACA;CACA;CACA;CACD;AAED,SAAwB,wBAAgC;CACtD,OAAO;EACL,MAAM;EACN,OAAO;EACP,SAAS;GACP,OAAO,EACL,cAAc,EAAE,SAAS,UAAU,EACpC;;EAEJ"}
+{"version":3,"file":"vite.mjs","names":[],"sources":["../../src/vite.ts"],"sourcesContent":["/* Vite plugin that wires up @stll/anonymize-wasm so its transitive\n * napi-rs-generated wasm loaders survive Vite's dep pre-bundler. Adds\n * this package plus its @stll/*-wasm dependencies to optimizeDeps.exclude\n * so the loader modules are served with their original paths, keeping\n * `new URL(\"./foo.wasm\", import.meta.url)` correct. */\nimport type { Plugin } from \"vite\";\n\n// Only the napi-rs WASM packages need excluding — they use\n// `new URL(\"./foo.wasm\", import.meta.url)` which Vite's optimizer\n// would rewrite. @stll/anonymize-wasm itself is pure JS and should\n// NOT be excluded.\nconst PACKAGES = [\n  \"@stll/text-search-wasm\",\n  \"@stll/aho-corasick-wasm\",\n  \"@stll/aho-corasick-wasm32-wasi\",\n  \"@stll/fuzzy-search-wasm\",\n  \"@stll/fuzzy-search-wasm32-wasi\",\n  \"@stll/regex-set-wasm\",\n  \"@stll/regex-set-wasm32-wasi\",\n];\n\nexport default function stllAnonymizeWasmVite(): Plugin {\n  return {\n    name: \"stll-anonymize-wasm\",\n    apply: \"serve\",\n    config() {\n      return {\n        optimizeDeps: { exclude: PACKAGES },\n      };\n    },\n  };\n}\n"],"mappings":";AAWA,MAAM,WAAW;CACf;CACA;CACA;CACA;CACA;CACA;CACA;AACF;AAEA,SAAwB,wBAAgC;CACtD,OAAO;EACL,MAAM;EACN,OAAO;EACP,SAAS;GACP,OAAO,EACL,cAAc,EAAE,SAAS,SAAS,EACpC;EACF;CACF;AACF"}

package/dist/wasm.d.mts

@@ -405,6 +405,8 @@ type DenyListConfig = Pick<PipelineConfig, "enableDenyList" | "enableNameCorpus"
  * "title" = academic/professional title
  */
 type PatternSource = "deny-list" | "city" | "custom-deny-list" | "first-name" | "surname" | "title";
+type PatternLabels = string | string[];
+type PatternSources = PatternSource | PatternSource[];
 /**
  * Pre-built deny list data. Constructed once by
  * `buildDenyList`, reused across `processDenyListMatches`
@@ -418,10 +420,10 @@ type DenyListData = {
    * appears in multiple dictionaries (e.g., "Denver"
    * is both a person name and a city name).
    */
-  labels: string[][]; /** Maps pattern index → labels contributed by custom entries. */
-  customLabels: string[][]; /** Maps pattern index → original pattern text. */
+  labels: PatternLabels[]; /** Maps pattern index → labels contributed by custom entries. */
+  customLabels: (PatternLabels | undefined)[]; /** Maps pattern index → original pattern text. */
   originals: string[]; /** Maps pattern index → source types (plural). */
-  sources: PatternSource[][];
+  sources: PatternSources[];
 };
 /**
  * Resolve which dictionaries to load based on country
@@ -729,18 +731,16 @@ declare const buildLegalFormPatterns: () => Promise<string[]>;
  * otherwise the trim falls back to a no-op and sentence-
  * fragment fixes do not apply.
  */
-declare const processLegalFormMatches: (allMatches: Match[], sliceStart: number, sliceEnd: number, fullText?: string) => Entity[];
+declare const processLegalFormMatches: (allMatches: Match[], sliceStart: number, sliceEnd: number, fullText?: string, options?: {
+  suppressExtendBackward?: boolean;
+}) => Entity[];
 //#endregion
 //#region src/detectors/triggers.d.ts
-/**
- * Build trigger patterns and rules from data configs.
- * Returns string[] for the unified TextSearch
- * builder and the parallel rules array.
- */
-declare const buildTriggerPatterns: () => Promise<{
+type TriggerPatterns = {
   patterns: string[];
   rules: TriggerRule[];
-}>;
+};
+declare const buildTriggerPatterns: () => Promise<TriggerPatterns>;
 /**
  * Process trigger matches from the unified search.
  * Receives all matches; filters to the trigger slice
@@ -751,11 +751,6 @@ declare const buildTriggerPatterns: () => Promise<{
 declare const processTriggerMatches: (allMatches: Match[], sliceStart: number, sliceEnd: number, fullText: string, rules: readonly TriggerRule[]) => Entity[];
 //#endregion
 //#region src/detectors/address-seeds.d.ts
-/**
- * Build street type patterns for the unified search.
- * Returns string[] for the unified TextSearch
- * builder. Empty if data package is not installed.
- */
 declare const buildStreetTypePatterns: () => Promise<string[]>;
 /**
  * Process address seeds from the unified search.