@stll/anonymize 1.2.0 → 1.3.0

package/dist/coreference.it.mjs

@@ -0,0 +1,27 @@
+//#region src/data/coreference.it.json
+var coreference_it_default = [
+	{
+		"label": "(in seguito \"X\") or (di seguito \"X\")",
+		"pattern": "\\((?:in\\s+seguito|di\\s+seguito|qui\\s+di\\s+seguito|nel\\s+seguito)\\s+(?:denominata|denominato|detta|detto|indicata|indicato|anche|la|il)?\\s*[\"'“”«»‘’‚]([^\"'“”«»‘’‚]+)[\"'“”«»‘’‚]\\)",
+		"flags": "gi"
+	},
+	{
+		"label": "(denominato/denominata \"X\")",
+		"pattern": "\\((?:denominata|denominato|qui\\s+di\\s+seguito\\s+denominata|qui\\s+di\\s+seguito\\s+denominato)\\s+[\"'“”«»‘’‚]([^\"'“”«»‘’‚]+)[\"'“”«»‘’‚]\\)",
+		"flags": "gi"
+	},
+	{
+		"label": "(la/il \"X\")",
+		"pattern": "\\((?:la|il|lo|le|gli|l')\\s*[\"'“”«»‘’‚]([^\"'“”«»‘’‚]+)[\"'“”«»‘’‚]\\)",
+		"flags": "gi"
+	},
+	{
+		"label": "denominata/denominato \"X\" without parens",
+		"pattern": "(?:in\\s+seguito|di\\s+seguito)\\s+(?:denominata|denominato|detta|detto|indicata|indicato)\\s+[\"'“”«»‘’‚]([^\"'“”«»‘’‚]+)[\"'“”«»‘’‚]",
+		"flags": "gi"
+	}
+];
+//#endregion
+export { coreference_it_default as default };
+
+//# sourceMappingURL=coreference.it.mjs.map

package/dist/coreference.it.mjs.map

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

package/dist/coreference.pl.mjs

@@ -0,0 +1,27 @@
+//#region src/data/coreference.pl.json
+var coreference_pl_default = [
+	{
+		"label": "(zwana/y dalej \"X\") parenthetical define-term",
+		"pattern": "zwan(?:ego|emu|ymi|ym|ej|[ayąei])?\\s+dalej\\s+[\"„‚']([^\"”’']+)[\"”’']",
+		"flags": "gi"
+	},
+	{
+		"label": "dalej zwana/y \"X\"",
+		"pattern": "dalej\\s+zwan(?:ego|emu|ymi|ym|ej|[ayąei])?\\s+[\"„‚']([^\"”’']+)[\"”’']",
+		"flags": "gi"
+	},
+	{
+		"label": "(\"X\") parenthetical define-term",
+		"pattern": "\\(\\s*[\"„‚']([^\"”’']+)[\"”’']\\s*\\)",
+		"flags": "gi"
+	},
+	{
+		"label": "zwanymi dalej łącznie \"X\"",
+		"pattern": "zwanymi\\s+dalej\\s+(?:łącznie\\s+)?[\"„‚']([^\"”’']+)[\"”’']",
+		"flags": "gi"
+	}
+];
+//#endregion
+export { coreference_pl_default as default };
+
+//# sourceMappingURL=coreference.pl.mjs.map

package/dist/coreference.pl.mjs.map

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

package/dist/coreference.pt-br.mjs

@@ -0,0 +1,14 @@
+//#region src/data/coreference.pt-br.json
+var coreference_pt_br_default = [{
+	"label": "(doravante denominada \"X\") / (doravante denominado \"X\")",
+	"pattern": "doravante\\s+denominad[ao]s?(?:\\s+simplesmente)?\\s+[\"'“”„‟‚‘’]?([^\"'“”„‟‚‘’,;.\\n)]+?)[\"'“”„‟‚‘’,;.\\n)]",
+	"flags": "gi"
+}, {
+	"label": "doravante referida como \"X\"",
+	"pattern": "doravante\\s+(?:referid[ao]|design[ao]d[ao])s?(?:\\s+como)?\\s+[\"'“”„‟‚‘’]?([^\"'“”„‟‚‘’,;.\\n)]+?)[\"'“”„‟‚‘’,;.\\n)]",
+	"flags": "gi"
+}];
+//#endregion
+export { coreference_pt_br_default as default };
+
+//# sourceMappingURL=coreference.pt-br.mjs.map

package/dist/coreference.pt-br.mjs.map

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

package/dist/coreference.sk.mjs

@@ -1,9 +1,26 @@
 //#region src/data/coreference.sk.json
-var coreference_sk_default = [{
-	"label": "(dalej len \"X\")",
-	"pattern": "\\(ďalej\\s+len\\s+[„\"'‚]([^\"'\"\"]+)[\"\"']\\)",
-	"flags": "gi"
-}];
+var coreference_sk_default = [
+	{
+		"label": "(ďalej len \"X\") in parentheses",
+		"pattern": "\\(ďalej\\s+len\\s+[„\"'‚,]*([^\"'„“,;.\\n)]+?)[\"'„“,;.]*\\s*\\)",
+		"flags": "gi"
+	},
+	{
+		"label": "ďalej len \"X\" without parens",
+		"pattern": "ďalej\\s+len\\s+[„\"'‚,]([^\"'„“,;.\\n]+?)[\"'„“]",
+		"flags": "gi"
+	},
+	{
+		"label": "(ďalej spolu ako \"X\")",
+		"pattern": "\\(ďalej\\s+spolu\\s+(?:len|ako)\\s+[„\"'‚,]*([^\"'„“,;.\\n)]+?)[\"'„“,;.]*\\s*\\)",
+		"flags": "gi"
+	},
+	{
+		"label": "(ďalej tiež \"X\")",
+		"pattern": "\\(ďalej\\s+tiež\\s+[„\"'‚,]*([^\"'„“,;.\\n)]+?)[\"'„“,;.]*\\s*\\)",
+		"flags": "gi"
+	}
+];
 //#endregion
 export { coreference_sk_default as default };
 

package/dist/generic-roles.mjs

@@ -1,59 +1,237 @@
 //#region src/data/generic-roles.json
 var _comment = "Generic legal/business role terms that are not PII. Used to filter false positives from name detection.";
 var roles = [
+	"acreedor",
+	"acreedora",
+	"adjudicatario",
+	"adjudicataria",
 	"arbeitgeber",
 	"arbeitnehmer",
+	"arrendador",
+	"arrendadora",
+	"arrendatario",
+	"arrendataria",
 	"auftraggeber",
 	"auftragnehmer",
 	"bonus",
 	"borrower",
 	"buyer",
+	"capitán",
+	"cedente",
 	"cena",
+	"cesionario",
+	"cesionaria",
 	"client",
+	"cliente",
+	"comprador",
+	"compradora",
 	"committee",
 	"company",
 	"contractor",
+	"contratante",
+	"contratista",
+	"coronel",
 	"customer",
 	"darlehensgeber",
 	"darlehensnehmer",
+	"deudor",
+	"deudora",
 	"dlužník",
+	"dlžník",
 	"dodavatel",
+	"dodávateľ",
+	"dostawca",
+	"dostawcą",
+	"dostawcy",
+	"empleado",
+	"empleador",
+	"empleadora",
+	"empleada",
+	"empresa",
+	"empresario",
+	"empresaria",
 	"employee",
 	"employer",
 	"fax",
+	"fiador",
+	"fiadora",
 	"gesellschaft",
 	"guarantor",
 	"kupující",
+	"kupujúci",
 	"käufer",
 	"landlord",
+	"licitador",
+	"licitadora",
 	"lender",
 	"licensee",
 	"licensor",
 	"mieter",
 	"nájemce",
+	"nájomca",
 	"objednatel",
+	"objednávateľ",
 	"odběratel",
+	"odberateľ",
+	"otorgante",
+	"otorgado",
+	"otorgada",
+	"parte",
+	"partes",
 	"parties",
 	"party",
 	"plátce",
+	"poskytovateľ",
+	"prestador",
+	"prestadora",
+	"prestatario",
+	"prestataria",
+	"prijímateľ",
 	"prodávající",
+	"predávajúci",
 	"pronajímatel",
+	"prenajímateľ",
+	"proveedor",
+	"proveedora",
 	"příjemce",
 	"seller",
+	"sociedad",
+	"spoločnosť",
 	"společnost",
 	"strana",
 	"strany",
 	"supplier",
 	"tenant",
+	"vendedor",
+	"vendedora",
 	"vendor",
 	"vermieter",
 	"verkäufer",
 	"věřitel",
+	"veriteľ",
 	"uchazeč",
+	"ubytovateľ",
+	"usporiadateľ",
 	"zadavatel",
+	"zamawiający",
+	"zamawiającego",
+	"zamawiającym",
 	"zaměstnanec",
+	"zamestnanec",
 	"zaměstnavatel",
-	"zhotovitel"
+	"zamestnávateľ",
+	"zhotovitel",
+	"zhotoviteľ",
+	"acquirente",
+	"amministrazione",
+	"appaltatore",
+	"appaltatrice",
+	"cedente",
+	"cessionario",
+	"cliente",
+	"committente",
+	"compratore",
+	"compratrice",
+	"concedente",
+	"concessionario",
+	"conduttore",
+	"conduttrice",
+	"creditore",
+	"datore",
+	"debitore",
+	"denominazione",
+	"dipendente",
+	"fideiussore",
+	"fornitore",
+	"fornitrice",
+	"garante",
+	"lavoratore",
+	"lavoratrice",
+	"licenziante",
+	"licenziatario",
+	"locatario",
+	"locataria",
+	"locatore",
+	"locatrice",
+	"mandante",
+	"mandatario",
+	"mandataria",
+	"mutuante",
+	"mutuatario",
+	"parte",
+	"parti",
+	"stazione",
+	"subappaltatore",
+	"venditore",
+	"venditrice",
+	"zleceniodawca",
+	"zleceniodawcą",
+	"zleceniodawcy",
+	"zleceniobiorca",
+	"zleceniobiorcą",
+	"zleceniobiorcy",
+	"wykonawca",
+	"wykonawcą",
+	"wykonawcy",
+	"wynajmujący",
+	"wynajmująca",
+	"wynajmującym",
+	"wynajmującej",
+	"najemca",
+	"najemcy",
+	"sprzedawca",
+	"sprzedawcy",
+	"sprzedawcą",
+	"sprzedający",
+	"kupujący",
+	"kupującego",
+	"pracodawca",
+	"pracodawcą",
+	"pracownik",
+	"pracownikiem",
+	"pożyczkodawca",
+	"pożyczkodawcą",
+	"pożyczkobiorca",
+	"pożyczkobiorcą",
+	"odbiorca",
+	"odbiorcą",
+	"acheteur",
+	"acheteuse",
+	"acquéreur",
+	"acquéreuse",
+	"bailleresse",
+	"bailleur",
+	"cédant",
+	"cédante",
+	"cessionnaire",
+	"concédant",
+	"concédante",
+	"concessionnaire",
+	"donneur d'ordre",
+	"emprunteur",
+	"emprunteuse",
+	"employé",
+	"employée",
+	"employeur",
+	"employeuse",
+	"fournisseur",
+	"fournisseuse",
+	"garant",
+	"garante",
+	"locataire",
+	"mandant",
+	"mandante",
+	"mandataire",
+	"preneur",
+	"preneuse",
+	"prestataire",
+	"prêteur",
+	"prêteuse",
+	"salarié",
+	"salariée",
+	"société",
+	"vendeur",
+	"vendeuse"
 ];
 var generic_roles_default = {
 	_comment,

package/dist/hotword-rules.mjs

@@ -7,7 +7,7 @@ var rules = [
 			"birth number",
 			"Geburtsnummer"
 		],
-		"targetLabels": ["czech birth number"],
+		"targetLabels": ["birth number"],
 		"scoreAdjustment": .25,
 		"proximityBefore": 60,
 		"proximityAfter": 60
@@ -87,7 +87,8 @@ var rules = [
 		"hotwords": [
 			"e-mail",
 			"email",
-			"E-Mail"
+			"E-Mail",
+			"pec"
 		],
 		"targetLabels": ["email address"],
 		"scoreAdjustment": .1,
@@ -101,7 +102,10 @@ var rules = [
 			"sídlem",
 			"bytem",
 			"Adresse",
-			"address"
+			"address",
+			"indirizzo",
+			"sede legale",
+			"sede"
 		],
 		"targetLabels": ["address"],
 		"scoreAdjustment": .15,

package/dist/index.d.mts

@@ -62,6 +62,16 @@ type GazetteerEntry = {
 /** Extraction strategy — closed discriminated union. */
 type TriggerStrategy = {
   type: "to-next-comma";
+  /**
+   * Optional list of lowercase keywords that terminate
+   * the value scan, in addition to commas/newlines. Useful
+   * for triggers like court names that may continue past
+   * a missing comma into adjacent clause text ("Městským
+   * soudem v Praze dne 1. 1. 2020"); listing `"dne"` here
+   * stops the scan at the date boundary. Matched on a
+   * word-boundary, case-insensitive.
+   */
+  stopWords?: string[];
 } | {
   type: "to-end-of-line";
 } | {
@@ -72,6 +82,19 @@ type TriggerStrategy = {
 } | {
   type: "address";
   maxChars?: number;
+} | {
+  /**
+   * Extract the first regex match in the value text.
+   * Useful for shape-bounded values that follow a
+   * label on the same line as other fields, where
+   * `to-end-of-line` would over-capture. The pattern
+   * is anchored to the start of the (already
+   * leading-whitespace-stripped) value, so use
+   * `(?:.*?)` prefix only when intentional.
+   */
+  type: "match-pattern";
+  pattern: string;
+  flags?: string;
 };
 /** Validation rules — closed discriminated union. */
 type TriggerValidation = {
@@ -90,7 +113,21 @@ type TriggerValidation = {
   type: "matches-pattern";
   pattern: string;
   flags?: string;
+}
+/**
+ * Run a named stdnum validator (checksum + length)
+ * against the captured value. Keeps the trigger
+ * path symmetrical with the formatted-regex
+ * detectors so e.g. `CPF nº 00000000000` does not
+ * survive as a tax-ID entity.
+ */
+| {
+  type: "valid-id";
+  validator: ValidIdValidator;
 };
+/** Built-in stdnum validators that can be referenced
+ *  by `valid-id` validations. */
+type ValidIdValidator = "br.cpf" | "br.cnpj";
 /** Auto-generated trigger variants — closed set. */
 type TriggerExtension = "add-colon" | "add-trailing-space" | "add-colon-space" | "normalize-spaces";
 /** V2 trigger config entry (JSON shape). */
@@ -124,6 +161,9 @@ type CompiledValidation = {
 } | {
   type: "matches-pattern";
   re: RegExp;
+} | {
+  type: "valid-id";
+  check: (value: string) => boolean;
 };
 /**
  * Runtime rule — one per trigger string after
@@ -310,7 +350,7 @@ type PipelineConfig = {
  * every pipeline run and never persisted to the database.
  * Renaming a label here requires no migration.
  */
-declare const DEFAULT_ENTITY_LABELS: readonly ["person", "organization", "phone number", "address", "email address", "date", "date of birth", "bank account number", "iban", "tax identification number", "identity card number", "registration number", "credit card number", "passport number", "monetary amount", "land parcel"];
+declare const DEFAULT_ENTITY_LABELS: readonly ["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
 //#region src/detectors/regex.d.ts
 type RegexMeta = {
@@ -502,7 +542,9 @@ type PipelineContext = {
   allowList: ReadonlySet<string> | null;
   allowListPromise: Promise<ReadonlySet<string>> | null;
   personStopwords: ReadonlySet<string> | null;
-  personStopwordsPromise: Promise<ReadonlySet<string>> | null; /** First-name exclusions for stopword filtering. */
+  personStopwordsPromise: Promise<ReadonlySet<string>> | null;
+  addressStopwords: ReadonlySet<string> | null;
+  addressStopwordsPromise: Promise<ReadonlySet<string>> | null; /** First-name exclusions for stopword filtering. */
   firstNameExclusions: ReadonlySet<string> | null;
   firstNameExclusionCorpusLen: number;
   genericRoles: ReadonlySet<string> | null;
@@ -627,6 +669,7 @@ declare const DEFAULT_OPERATOR_CONFIG: OperatorConfig;
 declare const resolveOperator: (config: OperatorConfig, label: string) => OperatorType;
 //#endregion
 //#region src/detectors/legal-forms.d.ts
+declare const warmLegalRoleHeads: () => Promise<void>;
 /**
  * Build legal form regex pattern strings.
  * Returns an array of regex strings for the unified
@@ -638,6 +681,14 @@ declare const buildLegalFormPatterns: () => Promise<string[]>;
  * Process legal form matches from the unified search.
  * Receives all matches; filters to the legal forms
  * slice via sliceStart/sliceEnd.
+ *
+ * The role-head trimming step reads per-language data from
+ * a cache that `runPipeline` warms via `warmLegalRoleHeads()`
+ * before calling this. Callers that invoke
+ * `processLegalFormMatches` directly (without going through
+ * `runPipeline`) must `await warmLegalRoleHeads()` first;
+ * 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[];
 //#endregion
@@ -819,6 +870,8 @@ type CountryCode = NonNullRegion[keyof NonNullRegion][number];
 declare const resolveCountries: (regions?: string[], countries?: string[]) => Set<string> | null;
 //#endregion
 //#region src/filters/false-positives.d.ts
+/** Ensure street-type vocabulary is loaded. */
+declare const initAddressComponents: () => Promise<void>;
 /**
  * Filter out entities that are likely false positives:
  * template placeholders, clause/section numbers,
@@ -1035,5 +1088,5 @@ declare const levenshtein: (rawA: string, rawB: string) => number;
  */
 declare const normalizeForSearch: (text: string) => string;
 //#endregion
-export { type AnonymisationOperator, CURRENCY_PATTERN_META, type CountryCode, type CustomDenyListEntry, type CustomRegexPattern, DATE_PATTERN_META, DEFAULT_ENTITY_LABELS, DEFAULT_OPERATOR_CONFIG, DETECTION_SOURCES, DETECTOR_PRIORITY, type DefinitionPattern, type DenyListCategory, type DenyListData, type DetectionSource, type Dictionaries, type DictionaryMeta, type DocumentZone, type Entity, type EntityResult, type GazetteerData, type GazetteerEntry, type HotwordRule, type NameCorpusData, type NerInferenceFn, OPERATOR_REGISTRY, OPERATOR_TYPES, type OperatorConfig, type OperatorType, type PipelineConfig, type PipelineContext, type PipelineOptions, REGEX_META, REGEX_PATTERNS, REGIONS, type RawInferenceResult, type RedactionResult, type RegexMeta, type RegionId, type ReviewDecision, type ReviewedEntity, type TriggerExtension, type TriggerGroupConfig, type TriggerRule, type TriggerStrategy, type TriggerValidation, type UnifiedResult, type UnifiedSearchInstance, ZONE_SCORE_ADJUSTMENTS, type ZoneSpan, applyHotwordRules, applyZoneAdjustments, boostNearMissEntities, buildDenyList, buildGazetteerPatterns, buildLegalFormPatterns, buildPlaceholderMap, buildStreetTypePatterns, buildTriggerPatterns, buildUnifiedSearch, chunkText, classifyZones, computeChunkOffsets, corefKey, createPipelineContext, deanonymise, decodeSpans, decodeTokenSpans, detectNameCorpus, ensureDenyListData, exportRedactionKey, extractDefinedTerms, filterFalsePositives, findCoreferenceSpans, getCurrencyPatterns, getDatePatterns, initHotwordRules, initNameCorpus, initZoneClassifier, levenshtein, mergeAndDedup, mergeChunkEntities, normalizeForSearch, prepareBatch, processAddressSeeds, processDenyListMatches, processGazetteerMatches, processLegalFormMatches, processRegexMatches, processTriggerMatches, propagateOrgNames, redactText, resolveCountries, resolveOperator, runPipeline, runUnifiedSearch, sanitizeEntities, tokenizeText };
+export { type AnonymisationOperator, CURRENCY_PATTERN_META, type CountryCode, type CustomDenyListEntry, type CustomRegexPattern, DATE_PATTERN_META, DEFAULT_ENTITY_LABELS, DEFAULT_OPERATOR_CONFIG, DETECTION_SOURCES, DETECTOR_PRIORITY, type DefinitionPattern, type DenyListCategory, type DenyListData, type DetectionSource, type Dictionaries, type DictionaryMeta, type DocumentZone, type Entity, type EntityResult, type GazetteerData, type GazetteerEntry, type HotwordRule, type NameCorpusData, type NerInferenceFn, OPERATOR_REGISTRY, OPERATOR_TYPES, type OperatorConfig, type OperatorType, type PipelineConfig, type PipelineContext, type PipelineOptions, REGEX_META, REGEX_PATTERNS, REGIONS, type RawInferenceResult, type RedactionResult, type RegexMeta, type RegionId, type ReviewDecision, type ReviewedEntity, type TriggerExtension, type TriggerGroupConfig, type TriggerRule, type TriggerStrategy, type TriggerValidation, type UnifiedResult, type UnifiedSearchInstance, ZONE_SCORE_ADJUSTMENTS, type ZoneSpan, applyHotwordRules, applyZoneAdjustments, boostNearMissEntities, buildDenyList, buildGazetteerPatterns, buildLegalFormPatterns, buildPlaceholderMap, buildStreetTypePatterns, buildTriggerPatterns, buildUnifiedSearch, chunkText, classifyZones, computeChunkOffsets, corefKey, createPipelineContext, deanonymise, decodeSpans, decodeTokenSpans, detectNameCorpus, ensureDenyListData, exportRedactionKey, extractDefinedTerms, filterFalsePositives, findCoreferenceSpans, getCurrencyPatterns, getDatePatterns, initAddressComponents, initHotwordRules, initNameCorpus, initZoneClassifier, levenshtein, mergeAndDedup, mergeChunkEntities, normalizeForSearch, prepareBatch, processAddressSeeds, processDenyListMatches, processGazetteerMatches, processLegalFormMatches, processRegexMatches, processTriggerMatches, propagateOrgNames, redactText, resolveCountries, resolveOperator, runPipeline, runUnifiedSearch, sanitizeEntities, tokenizeText, warmLegalRoleHeads };
 //# sourceMappingURL=index.d.mts.map