@stll/anonymize-wasm 1.1.0 → 1.2.0

package/dist/wasm.d.mts

@@ -35,6 +35,7 @@ type Entity = {
   text: string;
   score: number;
   source: DetectionSource;
+  sourceDetail?: "custom-deny-list" | "custom-regex";
 };
 /**
  * Entity after human review. Extends the base Entity
@@ -177,25 +178,45 @@ type DenyListCategory = "Names" | "Places" | "Addresses" | "Courts" | "Financial
 /**
  * Metadata for a single dictionary entry in the
  * deny-list system. Mirrors the shape from
- * `@stll/anonymize-data` so consumers can pass
- * pre-loaded data without the runtime dependency.
+ * the anonymize-data package so consumers can pass
+ * pre-loaded data without a runtime dependency.
  */
 type DictionaryMeta = {
   label: string;
   category: DenyListCategory;
   country: string | null;
 };
+/**
+ * Caller-supplied exact terms for deny-list matching.
+ * These entries are merged with the published deny-list
+ * dictionaries when `enableDenyList` is enabled.
+ */
+type CustomDenyListEntry = {
+  value: string;
+  label: string;
+  variants?: readonly string[];
+};
+/**
+ * Caller-supplied regex detector. The pattern is passed
+ * to the underlying text-search regex engine, so use its
+ * supported regex syntax. Inline flags such as `(?i)` are
+ * accepted when supported by that engine.
+ */
+type CustomRegexPattern = {
+  pattern: string;
+  label: string;
+  score?: number;
+};
 /**
  * Pre-loaded dictionary data for dependency injection.
  * Consumers that want name/city/deny-list detection
- * import from `@stll/anonymize-data` themselves and
- * pass the data here; the anonymize package itself
- * has zero cross-package imports.
+ * load dictionaries themselves (e.g. from the
+ * anonymize-data package) and pass them here; the
+ * anonymize package has zero cross-package imports.
  *
  * All fields are optional. When a field is absent,
  * the corresponding detection path is skipped (same
- * behavior as when `@stll/anonymize-data` was not
- * installed).
+ * behavior as when no dictionaries are available).
  */
 type Dictionaries = {
   /**
@@ -249,6 +270,16 @@ type PipelineConfig = {
   denyListCountries?: string[];
   denyListRegions?: string[];
   denyListExcludeCategories?: string[];
+  /**
+   * Caller-owned exact terms to match through the
+   * deny-list layer. Requires `enableDenyList: true`.
+   */
+  customDenyList?: readonly CustomDenyListEntry[];
+  /**
+   * Caller-owned regex detectors. Requires
+   * `enableRegex: true`.
+   */
+  customRegexes?: readonly CustomRegexPattern[];
   enableGazetteer: boolean;
   enableNer: boolean;
   enableConfidenceBoost: boolean;
@@ -266,7 +297,7 @@ type PipelineConfig = {
    * Pre-loaded dictionary data for name, deny-list,
    * and city detection. When omitted, dictionary-based
    * detection paths are skipped. Consumers load from
-   * `@stll/anonymize-data` and pass the data here.
+   * the anonymize-data package and pass the data here.
    */
   dictionaries?: Dictionaries;
 };
@@ -284,7 +315,8 @@ declare const DEFAULT_ENTITY_LABELS: readonly ["person", "organization", "phone
 //#region src/detectors/regex.d.ts
 type RegexMeta = {
   label: string;
-  score: number; /** Post-match stdnum validator for confirmation. */
+  score: number;
+  sourceDetail?: Entity["sourceDetail"]; /** Post-match stdnum validator for confirmation. */
   validator?: Validator;
 };
 /** Flat pattern array for text-search. */
@@ -321,15 +353,16 @@ declare const CURRENCY_PATTERN_META: Readonly<RegexMeta>;
 declare const processRegexMatches: (allMatches: Match[], sliceStart: number, sliceEnd: number, meta_: readonly RegexMeta[]) => Entity[];
 //#endregion
 //#region src/detectors/deny-list.d.ts
-type DenyListConfig = Pick<PipelineConfig, "enableDenyList" | "enableNameCorpus" | "denyListCountries" | "denyListRegions" | "denyListExcludeCategories" | "dictionaries">;
+type DenyListConfig = Pick<PipelineConfig, "enableDenyList" | "enableNameCorpus" | "denyListCountries" | "denyListRegions" | "denyListExcludeCategories" | "customDenyList" | "dictionaries">;
 /**
  * Source tag for each pattern in the automaton.
  * "deny-list" = standard deny list entry
+ * "custom-deny-list" = caller-owned exact term
  * "first-name" = name corpus first name
  * "surname" = name corpus surname
  * "title" = academic/professional title
  */
-type PatternSource = "deny-list" | "first-name" | "surname" | "title";
+type PatternSource = "deny-list" | "custom-deny-list" | "first-name" | "surname" | "title";
 /**
  * Pre-built deny list data. Constructed once by
  * `buildDenyList`, reused across `processDenyListMatches`
@@ -343,7 +376,8 @@ type DenyListData = {
    * appears in multiple dictionaries (e.g., "Denver"
    * is both a person name and a city name).
    */
-  labels: string[][]; /** Maps pattern index → original pattern text. */
+  labels: string[][]; /** Maps pattern index → labels contributed by custom entries. */
+  customLabels: string[][]; /** Maps pattern index → original pattern text. */
   originals: string[]; /** Maps pattern index → source types (plural). */
   sources: PatternSource[][];
 };
@@ -397,10 +431,12 @@ type GazetteerData = {
   isFuzzy: boolean[];
 };
 type UnifiedSearchInstance = {
-  /** Regex + triggers + legal-forms. */tsRegex: TextSearch; /** Deny-list + street-types + gazetteer. */
+  /** Regex + triggers + legal-forms. */tsRegex: TextSearch; /** Caller-owned custom regexes, isolated for overlap preservation. */
+  tsCustomRegex: TextSearch; /** Deny-list + street-types + gazetteer. */
   tsLiterals: TextSearch;
   slices: {
     regex: PatternSlice;
+    customRegex: PatternSlice;
     legalForms: PatternSlice;
     triggers: PatternSlice;
     denyList: PatternSlice;
@@ -408,6 +444,7 @@ type UnifiedSearchInstance = {
     gazetteer: PatternSlice;
   };
   regexMeta: readonly RegexMeta[];
+  customRegexMeta: readonly RegexMeta[];
   triggerRules: readonly TriggerRule[];
   denyListData: DenyListData | null;
   gazetteerData: GazetteerData | null;
@@ -740,6 +777,7 @@ declare const detectNameCorpus: (fullText: string, ctx?: PipelineContext) => Ent
 //#region src/unified-search.d.ts
 type UnifiedResult = {
   /** All matches from both instances combined. */regexMatches: Match[];
+  customRegexMatches: Match[];
   literalMatches: Match[];
 };
 declare const runUnifiedSearch: (instance: UnifiedSearchInstance, fullText: string) => UnifiedResult;
@@ -997,5 +1035,5 @@ declare const levenshtein: (rawA: string, rawB: string) => number;
  */
 declare const normalizeForSearch: (text: string) => string;
 //#endregion
-export { type AnonymisationOperator, CURRENCY_PATTERN_META, type CountryCode, 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, initHotwordRules, initNameCorpus, initZoneClassifier, levenshtein, mergeAndDedup, mergeChunkEntities, normalizeForSearch, prepareBatch, processAddressSeeds, processDenyListMatches, processGazetteerMatches, processLegalFormMatches, processRegexMatches, processTriggerMatches, propagateOrgNames, redactText, resolveCountries, resolveOperator, runPipeline, runUnifiedSearch, sanitizeEntities, tokenizeText };
 //# sourceMappingURL=wasm.d.mts.map