@stll/anonymize-wasm 1.4.11 → 2.0.0-alpha.1

package/dist/wasm.d.mts

@@ -180,6 +180,7 @@ type CompiledValidation = {
   re: RegExp;
 } | {
   type: "valid-id";
+  validator: ValidIdValidator;
   check: (value: string) => boolean;
 };
 /**
@@ -257,6 +258,7 @@ type CustomRegexPattern = {
   pattern: string;
   label: string;
   score?: number;
+  preparedArtifactPolicy?: "include" | "omit";
 };
 /**
  * Pre-loaded dictionary data for dependency injection.
@@ -321,6 +323,18 @@ type PipelineConfig = {
   threshold: number;
   enableTriggerPhrases: boolean;
   enableRegex: boolean;
+  /**
+   * Expected content language codes. When present, these
+   * derive default dictionary scopes for name corpus and
+   * deny-list matching unless the lower-level scope fields
+   * below are set explicitly.
+   */
+  languages?: string[];
+  /**
+   * Convenience form for single-language documents. Ignored
+   * when `languages` is also provided.
+   */
+  language?: string;
   /**
    * Enables legal-form organization detection.
    * Required for typed callers; legacy untyped
@@ -391,21 +405,48 @@ type PipelineConfig = {
 type RegexMeta = {
   label: string;
   score: number;
-  sourceDetail?: Entity["sourceDetail"]; /** Post-match stdnum validator for confirmation. */
+  sourceDetail?: Entity["sourceDetail"];
+  minByteLength?: number; /** Post-match stdnum validator for confirmation. */
   validator?: Validator;
+  validatorId?: string; /** Extract the identifier portion when context is part of the regex span. */
+  validatorInput?: (text: string) => string;
+  validatorInputKind?: "digits-only" | "crypto-wallet-candidate";
 };
 /** Flat pattern array for text-search. */
 declare const REGEX_PATTERNS: readonly string[];
 /** Parallel metadata. Index = pattern index. */
 declare const REGEX_META: readonly RegexMeta[];
+type DateMonthData = Record<string, string[]>;
+type YearWordData = Record<string, string[]>;
 /**
  * Get dynamically built date patterns from
  * date-months.json. Returns a cached promise; the JSON
  * is loaded only once.
  */
-declare const getDatePatterns: () => Promise<string[]>;
+declare const getDatePatterns: (languages?: readonly string[]) => Promise<string[]>;
 /** Date pattern metadata (all are score 1 dates). */
 declare const DATE_PATTERN_META: Readonly<RegexMeta>;
+type MonetaryData = {
+  currencies: {
+    codes: string[];
+    symbols: string[];
+    local_names: string[];
+  };
+  amount_words: {
+    written_amount_patterns: Array<{
+      keywords: string[];
+    }>;
+    magnitude_suffixes: Array<{
+      words: string[];
+      abbreviations_case_insensitive: string[];
+      abbreviations_case_sensitive: string[];
+    }>;
+    share_quantity_terms: Array<{
+      modifiers: string[];
+      nouns: string[];
+    }>;
+  };
+};
 /**
  * Get dynamically built monetary amount patterns from
  * currencies.json. Returns a cached promise; the JSON
@@ -429,6 +470,30 @@ declare const processRegexMatches: (allMatches: Match[], sliceStart: number, sli
 //#endregion
 //#region src/detectors/deny-list.d.ts
 type DenyListConfig = Pick<PipelineConfig, "enableDenyList" | "enableNameCorpus" | "nameCorpusLanguages" | "denyListCountries" | "denyListRegions" | "denyListExcludeCategories" | "customDenyList" | "dictionaries" | "enableCountries">;
+type DenyListFilterData = {
+  stopwords: string[];
+  allowList: string[];
+  personStopwords: string[];
+  personTrailingNouns: string[];
+  addressStopwords: string[];
+  addressJurisdictionPrefixes: string[];
+  streetTypes: string[];
+  addressComponentTerms: string[];
+  ambiguousStreetTypeTerms: string[];
+  firstNames: string[];
+  genericRoles: string[];
+  numberAbbrevPrefixes: string[];
+  sentenceStarters: string[];
+  trailingAddressWordExclusions: string[];
+  documentHeadingWords: string[];
+  documentHeadingOrdinalMarkers: string[];
+  definedTermCues: string[];
+  signingPlaceGuards: DenyListSigningPlaceGuardData[];
+};
+type DenyListSigningPlaceGuardData = {
+  prefixPhrases: string[];
+  suffixPhrases: string[];
+};
 /**
  * Source tag for each pattern in the automaton.
  * "deny-list" = standard deny list entry
@@ -458,6 +523,7 @@ type DenyListData = {
   customLabels: (PatternLabels | undefined)[]; /** Maps pattern index → original pattern text. */
   originals: string[]; /** Maps pattern index → source types (plural). */
   sources: PatternSources[];
+  filters: DenyListFilterData;
 };
 /**
  * Resolve which dictionaries to load based on country
@@ -495,6 +561,25 @@ declare const ensureDenyListData: (ctx?: PipelineContext, dictionaries?: Diction
  */
 declare const processDenyListMatches: (allMatches: Match[], sliceStart: number, sliceEnd: number, fullText: string, data: DenyListData, ctx?: PipelineContext) => Entity[];
 //#endregion
+//#region src/detectors/address-seeds.d.ts
+type AddressSeedData = {
+  boundary_words: string[];
+  br_cep_cue_words: string[];
+  unit_abbreviations: string[];
+};
+declare const buildStreetTypePatterns: () => Promise<string[]>;
+/**
+ * Process address seeds from the unified search.
+ * Receives all matches; filters to the street types
+ * slice via sliceStart/sliceEnd. Uses fullText and
+ * existingEntities for seed collection, clustering,
+ * expansion, and scoring.
+ *
+ * Runs as a post-processor after all other detectors,
+ * using their output as seed sources.
+ */
+declare const processAddressSeeds: (allMatches: Match[], sliceStart: number, sliceEnd: number, fullText: string, existingEntities: Entity[]) => Promise<Entity[]>;
+//#endregion
 //#region src/detectors/countries.d.ts
 /**
  * Pre-built country patterns + parallel label/source
@@ -513,11 +598,289 @@ type CountryData = {
 };
 type CountryVariant = "name" | "alias" | "alpha3" | "alpha2";
 //#endregion
+//#region src/filters/confidence-boost.d.ts
+/**
+ * Boost confidence of near-miss NER entities that appear
+ * near high-confidence detections (regex, trigger phrase).
+ *
+ * If an NER entity scored between (threshold - 0.15) and
+ * threshold, count how many confirmed entities exist within
+ * a 150-char window. Add +0.05 per co-located entity.
+ * If the boosted score crosses the threshold, include it.
+ *
+ * Only mutates score on near-miss entities; high-confidence
+ * entities pass through unchanged.
+ */
+declare const boostNearMissEntities: (entities: Entity[], threshold: number) => Entity[];
+type AddressContextData = {
+  address_prepositions: string[];
+  temporal_prepositions: string[];
+  street_abbreviations: string[];
+  bare_house_stopwords: string[];
+};
+//#endregion
 //#region src/build-unified-search.d.ts
 type PatternSlice = {
   start: number;
   end: number;
 };
+type NativeSearchPatternKind = "literal" | "literal-with-options" | "regex" | "fuzzy";
+type NativeSearchPattern = {
+  kind: NativeSearchPatternKind;
+  pattern: string;
+  distance?: number;
+  case_insensitive?: boolean;
+  whole_words?: boolean;
+  lazy?: boolean;
+  prefilter_any?: string[];
+  prefilter_case_insensitive?: boolean;
+  prefilter_regex?: string;
+  prefilter_window_bytes?: number;
+  prepared_artifact_policy?: "include" | "omit";
+};
+type NativeSearchOptions = {
+  literal_case_insensitive?: boolean;
+  literal_whole_words?: boolean;
+  regex_whole_words?: boolean;
+  regex_overlap_all?: boolean;
+  regex_artifact_policy?: "include" | "omit";
+  fuzzy_case_insensitive?: boolean;
+  fuzzy_whole_words?: boolean;
+  fuzzy_normalize_diacritics?: boolean;
+};
+type NativeRegexMatchMeta = {
+  label: string;
+  score: number;
+  source_detail?: string;
+  requires_validation?: boolean;
+  validator_id?: string;
+  validator_input?: string;
+  min_byte_length?: number;
+};
+type NativeDenyListFilterData = {
+  stopwords: string[];
+  allow_list: string[];
+  person_stopwords: string[];
+  person_trailing_nouns: string[];
+  address_stopwords: string[];
+  address_jurisdiction_prefixes: string[];
+  street_types: string[];
+  address_component_terms: string[];
+  ambiguous_street_type_terms: string[];
+  first_names: string[];
+  generic_roles: string[];
+  number_abbrev_prefixes: string[];
+  sentence_starters: string[];
+  trailing_address_word_exclusions: string[];
+  document_heading_words: string[];
+  document_heading_ordinal_markers: string[];
+  defined_term_cues: string[];
+  signing_place_guards: NativeSigningPlaceGuardData[];
+};
+type NativeSigningPlaceGuardData = {
+  prefix_phrases: string[];
+  suffix_phrases: string[];
+};
+type NativeDenyListMatchData = {
+  labels?: string[][];
+  label_table?: string[];
+  label_indices?: number[][];
+  custom_labels?: string[][];
+  custom_label_indices?: number[][];
+  originals: string[];
+  sources?: string[][];
+  source_table?: string[];
+  source_indices?: number[][];
+  filters?: NativeDenyListFilterData;
+};
+type NativeTriggerStrategy = {
+  type: "to-next-comma";
+  stop_words?: string[];
+  max_length?: number;
+} | {
+  type: "to-end-of-line";
+} | {
+  type: "n-words";
+  count: number;
+} | {
+  type: "company-id-value";
+} | {
+  type: "address";
+  max_chars?: number;
+} | {
+  type: "match-pattern";
+  pattern: string;
+  flags?: string;
+};
+type NativeTriggerValidation = {
+  type: "starts-uppercase";
+} | {
+  type: "min-length";
+  min: number;
+} | {
+  type: "max-length";
+  max: number;
+} | {
+  type: "no-digits";
+} | {
+  type: "has-digits";
+} | {
+  type: "matches-pattern";
+  pattern: string;
+  flags?: string;
+} | {
+  type: "valid-id";
+  validator: string;
+};
+type NativeTriggerRule = {
+  trigger: string;
+  label: string;
+  strategy: NativeTriggerStrategy;
+  validations: NativeTriggerValidation[];
+  include_trigger: boolean;
+};
+type NativeTriggerData = {
+  rules: NativeTriggerRule[];
+  address_stop_keywords: string[];
+  party_position_terms: string[];
+  post_nominals: string[];
+  sentence_terminal_currency_terms: string[];
+  phone_extension_labels: string[];
+  number_markers: string[];
+  number_labels: string[];
+};
+type NativeLegalFormData = {
+  suffixes: string[];
+  normalized_boundary_suffixes: string[];
+  normalized_in_name_words: string[];
+  normalized_suffix_words: string[];
+  role_heads: string[];
+  sentence_verb_indicators: string[];
+  clause_noun_heads: string[];
+  connector_prose_heads: string[];
+  structural_single_cap_prefixes: string[];
+  leading_clause_phrases: string[];
+  leading_clause_direct_prefixes: string[];
+  connector_words: string[];
+  and_connector_words: string[];
+  in_name_prepositions: string[];
+  company_suffix_words: string[];
+  comma_gated_direct_prefixes: string[];
+};
+type NativeDateData = {
+  month_names_by_language: DateMonthData;
+  year_words_by_language: YearWordData;
+};
+type NativeMonetaryData = MonetaryData;
+type NativeAddressSeedData = AddressSeedData;
+type NativeAddressContextData = AddressContextData;
+type NativeCoreferencePatternData = {
+  pattern: string;
+  flags: string;
+};
+type NativeCoreferenceData = {
+  definition_patterns: NativeCoreferencePatternData[];
+  role_stop_terms: string[];
+  legal_form_aliases: string[];
+  organization_suffixes: string[];
+  organization_determiners: string[];
+};
+type NativeNameCorpusData = {
+  first_names: string[];
+  surnames: string[];
+  title_tokens: string[];
+  title_abbreviations: string[];
+  excluded_words: string[];
+  common_words: string[];
+  non_western_names: string[];
+  excluded_all_caps: string[];
+  ja_suffixes: string[];
+  arabic_connectors: string[];
+  relation_connectors: string[];
+  hyphenated_prefixes: string[];
+  cjk_non_person_terms: string[];
+  cjk_surname_starters: string[];
+  organization_terms: string[];
+};
+type NativeNameCorpusMode = "full" | "supplemental";
+type NativeZonePatternData = {
+  pattern: string;
+  flags: string;
+};
+type NativeZoneSigningClauseData = {
+  prefix: string;
+  suffix: string;
+  prepositions: string[];
+};
+type NativeZoneData = {
+  section_heading_patterns: NativeZonePatternData[];
+  signing_clauses: NativeZoneSigningClauseData[];
+};
+type NativeGazetteerData = {
+  labels: string[];
+  is_fuzzy: boolean[];
+};
+type NativeHotwordRule = {
+  hotwords: string[];
+  target_labels: string[];
+  score_adjustment: number;
+  reclassify_to?: string;
+  proximity_before: number;
+  proximity_after: number;
+};
+type NativeHotwordRuleData = {
+  rules: NativeHotwordRule[];
+  pattern_rule_indices: number[];
+};
+type NativeSignatureData = {
+  labels: string[];
+  witness_phrases: string[];
+  name_particles: string[];
+  post_nominal_suffixes: string[];
+  organization_suffixes: string[];
+  image_stub_prefixes: string[];
+};
+type NativePreparedSearchConfig = {
+  regex_patterns: NativeSearchPattern[];
+  custom_regex_patterns: NativeSearchPattern[];
+  literal_patterns: NativeSearchPattern[];
+  regex_options: NativeSearchOptions;
+  custom_regex_options: NativeSearchOptions;
+  literal_options: NativeSearchOptions;
+  literal_patterns_from_deny_list_data?: boolean;
+  allowed_labels: string[];
+  threshold: number;
+  confidence_boost: boolean;
+  slices: {
+    regex: PatternSlice;
+    custom_regex: PatternSlice;
+    legal_forms?: PatternSlice;
+    triggers?: PatternSlice;
+    deny_list: PatternSlice;
+    street_types?: PatternSlice;
+    gazetteer: PatternSlice;
+    countries: PatternSlice;
+    hotwords?: PatternSlice;
+  };
+  regex_meta: NativeRegexMatchMeta[];
+  custom_regex_meta: NativeRegexMatchMeta[];
+  deny_list_data?: NativeDenyListMatchData;
+  false_positive_filters?: NativeDenyListFilterData;
+  gazetteer_data?: NativeGazetteerData;
+  country_data?: CountryData;
+  hotword_data?: NativeHotwordRuleData;
+  trigger_data?: NativeTriggerData;
+  legal_form_data?: NativeLegalFormData;
+  address_seed_data?: NativeAddressSeedData;
+  zone_data?: NativeZoneData;
+  address_context_data?: NativeAddressContextData;
+  coreference_data?: NativeCoreferenceData;
+  name_corpus_data?: NativeNameCorpusData;
+  signature_data?: NativeSignatureData;
+  name_corpus_mode?: NativeNameCorpusMode;
+  date_data?: NativeDateData;
+  monetary_data?: NativeMonetaryData;
+};
 type GazetteerData = {
   /** Maps local pattern index to entry label. */labels: string[];
   /**
@@ -546,6 +909,7 @@ type UnifiedSearchInstance = {
   denyListData: DenyListData | null;
   gazetteerData: GazetteerData | null;
   countryData: CountryData | null;
+  nativeStaticConfig: NativePreparedSearchConfig;
 };
 declare const buildUnifiedSearch: (config: PipelineConfig, gazetteerEntries?: GazetteerEntry[], ctx?: PipelineContext) => Promise<UnifiedSearchInstance>;
 //#endregion
@@ -582,7 +946,12 @@ type NameCorpusData = {
    *  Contains the lowercase, dot-stripped form
    *  (e.g., "dr", "smt", "atty"). */
   titleAbbreviations: ReadonlySet<string>;
-  excludedWords: ReadonlySet<string>; /** Non-Western name tokens merged across all locales. */
+  excludedWords: ReadonlySet<string>;
+  /** Lowercased common English words. A name chain whose
+   *  every token is a common word (e.g. "Loan Documents",
+   *  where "Loan" coincides with a Vietnamese given name)
+   *  is treated as a common-word phrase, not a person. */
+  commonWords: ReadonlySet<string>; /** Non-Western name tokens merged across all locales. */
   nonWesternNames: ReadonlySet<string>; /** All-caps acronyms excluded from name detection. */
   excludedAllCaps: ReadonlySet<string>; /** Raw arrays exposed for deny-list AC integration. */
   firstNamesList: readonly string[];
@@ -606,6 +975,9 @@ type PipelineContext = {
   search: UnifiedSearchInstance | null;
   searchKey: string;
   searchPromise: Promise<UnifiedSearchInstance> | null;
+  nativePipelinePackage: Uint8Array | null;
+  nativePipelinePackageKey: string;
+  nativePipelinePackagePromise: Promise<Uint8Array> | null;
   nameCorpus: NameCorpusData | null;
   nameCorpusKey: string;
   nameCorpusPromise: Promise<void> | null;
@@ -615,6 +987,8 @@ type PipelineContext = {
   allowListPromise: Promise<ReadonlySet<string>> | null;
   personStopwords: ReadonlySet<string> | null;
   personStopwordsPromise: Promise<ReadonlySet<string>> | null;
+  definedTermHeads: ReadonlySet<string> | null;
+  definedTermHeadsPromise: Promise<ReadonlySet<string>> | null;
   addressStopwords: ReadonlySet<string> | null;
   addressStopwordsPromise: Promise<ReadonlySet<string>> | null; /** First-name exclusions for stopword filtering. */
   firstNameExclusions: ReadonlySet<string> | null;
@@ -622,6 +996,7 @@ type PipelineContext = {
   genericRoles: ReadonlySet<string> | null;
   genericRolesPromise: Promise<ReadonlySet<string>> | null;
   corefPatterns: DefinitionPattern[] | null;
+  corefPatternsKey: string;
   corefPatternsPromise: Promise<DefinitionPattern[]> | null;
   corefLoadAttempted: boolean;
   roleStopSet: ReadonlySet<string> | null;
@@ -689,21 +1064,304 @@ type PipelineOptions = {
  */
 declare const runPipeline: (options: PipelineOptions) => Promise<Entity[]>;
 //#endregion
+//#region src/native.d.ts
+type NativeBindingOperatorConfig = {
+  operators?: Record<string, OperatorType>;
+  redactString?: string;
+};
+type NativeDiagnosticsBatchCallback = (diagnosticsJson: string) => void;
+type NativeResultEventCallback = (eventJson: string) => void;
+type NativeBindingRedactionEntry = {
+  placeholder: string;
+  original: string;
+};
+type NativeBindingOperatorEntry = {
+  placeholder: string;
+  operator: OperatorType;
+};
+type NativeBindingPipelineEntity = {
+  start: number;
+  end: number;
+  label: string;
+  text: string;
+  score: number;
+  source: string;
+  sourceDetail?: string | null;
+};
+type NativeBindingRedactionResult = {
+  redactedText: string;
+  redactionMap: NativeBindingRedactionEntry[];
+  operatorMap: NativeBindingOperatorEntry[];
+  entityCount: number;
+};
+type NativeBindingStaticRedactionResult = {
+  resolvedEntities: NativeBindingPipelineEntity[];
+  redaction: NativeBindingRedactionResult;
+};
+type NativePreparedSearchBinding = {
+  prepareDiagnosticsJson?: () => string;
+  warmLazyRegex?: () => void;
+  warm_lazy_regex?: () => void;
+  warmLazyRegexDiagnosticsJson?: () => string;
+  warm_lazy_regex_diagnostics_json?: () => string;
+  redactStaticEntities: (fullText: string, operators?: NativeBindingOperatorConfig) => NativeBindingStaticRedactionResult;
+  redactStaticEntitiesJson?: (fullText: string, operators?: NativeBindingOperatorConfig) => string;
+  redactStaticEntitiesResultStreamJson?: (fullText: string, operators: NativeBindingOperatorConfig | undefined, onEvent: NativeResultEventCallback) => string;
+  redactStaticEntitiesDiagnosticsJson?: (fullText: string, operators?: NativeBindingOperatorConfig) => string;
+  redactStaticEntitiesDiagnosticsStreamJson?: (fullText: string, operators: NativeBindingOperatorConfig | undefined, onBatch: NativeDiagnosticsBatchCallback) => string;
+  redactStaticEntitiesSummaryDiagnosticsJson?: (fullText: string, operators?: NativeBindingOperatorConfig) => string;
+};
+type NativeAnonymizeBinding = {
+  normalizeForSearch: (text: string) => string;
+  nativePackageVersion: () => string;
+  NativePreparedSearch: {
+    fromConfigJsonBytes: (configJson: Uint8Array) => NativePreparedSearchBinding;
+    fromPreparedPackageBytes: (packageBytes: Uint8Array) => NativePreparedSearchBinding;
+    fromPreparedPackageBytesWithoutCache?: (packageBytes: Uint8Array) => NativePreparedSearchBinding;
+    fromTrustedPreparedPackageBytes?: (packageBytes: Uint8Array) => NativePreparedSearchBinding;
+    fromTrustedPreparedPackageBytesWithoutCache?: (packageBytes: Uint8Array) => NativePreparedSearchBinding;
+  };
+  prepareStaticSearchPackageBytes: (configJson: Uint8Array) => Uint8Array;
+  prepareStaticSearchCompressedPackageBytes: (configJson: Uint8Array) => Uint8Array;
+};
+type NativeOperatorConfig = {
+  operators?: Record<string, OperatorType>;
+  redactString?: string;
+};
+type NativePipelineEntity = {
+  start: number;
+  end: number;
+  label: string;
+  text: string;
+  score: number;
+  source: string;
+  sourceDetail?: string;
+};
+type NativeRedactionResult = {
+  redactedText: string;
+  redactionMap: Map<string, string>;
+  operatorMap: Map<string, OperatorType>;
+  entityCount: number;
+};
+type NativeStaticRedactionResult = {
+  resolvedEntities: NativePipelineEntity[];
+  redaction: NativeRedactionResult;
+};
+type NativeSearchPackageOptions = {
+  binding: NativeAnonymizeBinding;
+  config: NativePreparedSearchConfig;
+  compressed?: boolean;
+};
+type NativeSearchPackageInput = NativePreparedSearchConfig | string | Uint8Array;
+type SharedNativeSearchPackageOptions = {
+  binding: NativeAnonymizeBinding;
+  config: NativeSearchPackageInput;
+  compressed?: boolean;
+};
+type SharedNativePreparedPackageOptions = {
+  binding: NativeAnonymizeBinding;
+  packageBytes: Uint8Array;
+};
+type SharedNativeRedactTextJsonOptions = {
+  binding: NativeAnonymizeBinding;
+  config: NativeSearchPackageInput;
+  fullText: string;
+  operators?: NativeOperatorConfig;
+};
+type SharedNativeRedactTextOptions = SharedNativeRedactTextJsonOptions;
+type SharedNativeDiagnosticsJsonOptions = SharedNativeRedactTextJsonOptions;
+type SharedNativeRedactTextStreamJsonOptions = SharedNativeRedactTextJsonOptions & {
+  onEvent: NativeResultEventCallback;
+};
+type NativeNormalizeOptions = {
+  binding: NativeAnonymizeBinding;
+  text: string;
+};
+type NativeAnonymizerFromConfigOptions = {
+  binding: NativeAnonymizeBinding;
+  config: NativePreparedSearchConfig;
+};
+type NativeAnonymizerFromPackageOptions = {
+  binding: NativeAnonymizeBinding;
+  packageBytes: Uint8Array;
+};
+type NativePipelineFromPackageOptions = NativeAnonymizerFromPackageOptions;
+type NativeBindingVersionOptions = {
+  binding: NativeAnonymizeBinding;
+  expectedVersion: string;
+};
+declare class PreparedNativeAnonymizer {
+  #private;
+  constructor(prepared: NativePreparedSearchBinding);
+  prepareDiagnosticsJson(): string | null;
+  prepare_diagnostics_json(): string | null;
+  warmLazyRegex(): void;
+  warm_lazy_regex(): void;
+  warmLazyRegexDiagnosticsJson(): string | null;
+  warm_lazy_regex_diagnostics_json(): string | null;
+  redactStaticEntities(fullText: string, operators?: NativeOperatorConfig): NativeStaticRedactionResult;
+  redact_text(fullText: string, operators?: NativeOperatorConfig): NativeStaticRedactionResult;
+  redact_text_json(fullText: string, operators?: NativeOperatorConfig): string;
+  redactTextJson(fullText: string, operators?: NativeOperatorConfig): string;
+  redactTextStreamJson(fullText: string, onEvent: NativeResultEventCallback, operators?: NativeOperatorConfig): string | null;
+  redact_text_stream_json(fullText: string, onEvent: NativeResultEventCallback, operators?: NativeOperatorConfig): string | null;
+  redactStaticEntitiesDiagnosticsJson(fullText: string, operators?: NativeOperatorConfig): string | null;
+  diagnostics_json(fullText: string, operators?: NativeOperatorConfig): string | null;
+  diagnosticsStreamJson(fullText: string, onBatch: NativeDiagnosticsBatchCallback, operators?: NativeOperatorConfig): string | null;
+  diagnostics_stream_json(fullText: string, onBatch: NativeDiagnosticsBatchCallback, operators?: NativeOperatorConfig): string | null;
+  redactStaticEntitiesSummaryDiagnosticsJson(fullText: string, operators?: NativeOperatorConfig): string | null;
+  summary_diagnostics_json(fullText: string, operators?: NativeOperatorConfig): string | null;
+}
+declare class PreparedNativePipeline {
+  #private;
+  constructor(anonymizer: PreparedNativeAnonymizer);
+  prepareDiagnosticsJson(): string | null;
+  prepare_diagnostics_json(): string | null;
+  warmLazyRegex(): void;
+  warm_lazy_regex(): void;
+  warmLazyRegexDiagnosticsJson(): string | null;
+  warm_lazy_regex_diagnostics_json(): string | null;
+  redactText(fullText: string, operators?: NativeOperatorConfig): NativeStaticRedactionResult;
+  redact_text(fullText: string, operators?: NativeOperatorConfig): NativeStaticRedactionResult;
+  redact_text_json(fullText: string, operators?: NativeOperatorConfig): string;
+  redactTextJson(fullText: string, operators?: NativeOperatorConfig): string;
+  redactTextStreamJson(fullText: string, onEvent: NativeResultEventCallback, operators?: NativeOperatorConfig): string | null;
+  redact_text_stream_json(fullText: string, onEvent: NativeResultEventCallback, operators?: NativeOperatorConfig): string | null;
+  redactTextDiagnosticsJson(fullText: string, operators?: NativeOperatorConfig): string | null;
+  diagnostics_json(fullText: string, operators?: NativeOperatorConfig): string | null;
+  diagnosticsStreamJson(fullText: string, onBatch: NativeDiagnosticsBatchCallback, operators?: NativeOperatorConfig): string | null;
+  diagnostics_stream_json(fullText: string, onBatch: NativeDiagnosticsBatchCallback, operators?: NativeOperatorConfig): string | null;
+  redactTextSummaryDiagnosticsJson(fullText: string, operators?: NativeOperatorConfig): string | null;
+  summary_diagnostics_json(fullText: string, operators?: NativeOperatorConfig): string | null;
+}
+declare const encodeNativeSearchConfig: (config: NativePreparedSearchConfig) => Uint8Array;
+declare const encodeNativeSearchConfigInput: (config: NativeSearchPackageInput) => Uint8Array;
+declare const getNativeBindingVersion: (binding: NativeAnonymizeBinding) => string;
+declare const native_package_version: (binding: NativeAnonymizeBinding) => string;
+declare const normalize_for_search: ({
+  binding,
+  text
+}: NativeNormalizeOptions) => string;
+declare const assertNativeBindingVersion: ({
+  binding,
+  expectedVersion
+}: NativeBindingVersionOptions) => void;
+declare const prepareNativeSearchPackage: ({
+  binding,
+  config,
+  compressed
+}: NativeSearchPackageOptions) => Uint8Array;
+declare const prepare_search_package: ({
+  binding,
+  config,
+  compressed
+}: SharedNativeSearchPackageOptions) => Uint8Array;
+declare const createNativeAnonymizerFromConfig: ({
+  binding,
+  config
+}: NativeAnonymizerFromConfigOptions) => PreparedNativeAnonymizer;
+declare const createNativeAnonymizerFromPackage: ({
+  binding,
+  packageBytes
+}: NativeAnonymizerFromPackageOptions) => PreparedNativeAnonymizer;
+declare const load_prepared_package: ({
+  binding,
+  packageBytes
+}: SharedNativePreparedPackageOptions) => PreparedNativeAnonymizer;
+declare const redact_text_json: ({
+  binding,
+  config,
+  fullText,
+  operators
+}: SharedNativeRedactTextJsonOptions) => string;
+declare const redact_text: ({
+  binding,
+  config,
+  fullText,
+  operators
+}: SharedNativeRedactTextOptions) => NativeStaticRedactionResult;
+declare const redact_text_stream_json: ({
+  binding,
+  config,
+  fullText,
+  operators,
+  onEvent
+}: SharedNativeRedactTextStreamJsonOptions) => string | null;
+declare const diagnostics_json: ({
+  binding,
+  config,
+  fullText,
+  operators
+}: SharedNativeDiagnosticsJsonOptions) => string | null;
+declare const createNativePipelineFromPackage: ({
+  binding,
+  packageBytes
+}: NativePipelineFromPackageOptions) => PreparedNativePipeline;
+declare const PreparedSearch: typeof PreparedNativeAnonymizer;
+type PreparedSearch = PreparedNativeAnonymizer;
+//#endregion
+//#region src/native-default-config.d.ts
+declare const DEFAULT_NATIVE_PIPELINE_CONFIG: PipelineConfig;
+//#endregion
+//#region src/native-pipeline.d.ts
+type NativePipelineUnsupportedFeature = "enableNer";
+type NativePipelineCompatibility = {
+  status: "supported";
+} | {
+  status: "unsupported";
+  unsupportedFeatures: NativePipelineUnsupportedFeature[];
+};
+type NativePipelineBuildOptions = {
+  binding: NativeAnonymizeBinding;
+  config: PipelineConfig;
+  gazetteerEntries?: GazetteerEntry[];
+  context?: PipelineContext;
+};
+type NativePipelinePackageOptions = NativePipelineBuildOptions & {
+  compressed?: boolean;
+};
+declare const getNativePipelineCompatibility: (config: PipelineConfig) => NativePipelineCompatibility;
+declare const assertNativePipelineSupported: (config: PipelineConfig) => void;
+declare const prepareNativePipelineConfig: ({
+  config,
+  gazetteerEntries,
+  context
+}: Omit<NativePipelineBuildOptions, "binding">) => Promise<NativePreparedSearchConfig>;
+declare const prepareNativePipelinePackage: ({
+  binding,
+  config,
+  gazetteerEntries,
+  context,
+  compressed
+}: NativePipelinePackageOptions) => Promise<Uint8Array>;
+declare const createNativePipelineFromConfig: ({
+  binding,
+  config,
+  gazetteerEntries,
+  context
+}: NativePipelineBuildOptions) => Promise<PreparedNativePipeline>;
+//#endregion
 //#region src/redact.d.ts
 /**
  * Build a stable mapping from entity text to numbered
  * placeholders. Same real-world value always maps to the
  * same placeholder (e.g., "Dr. Muller" and "Dr.  Muller"
- * both become [PERSON_1]).
+ * share one person placeholder).
  *
- * Placeholder format: [LABEL_N] where LABEL is uppercase
- * and N is a 1-based counter per label.
+ * Placeholder format: [LABEL_N] where LABEL is uppercase.
+ * N is allocated per label and skips tokens already present
+ * in reserved text.
  *
  * @param _ctx Unused. Kept for signature compatibility;
  *   coref alias links now travel on the entities
  *   themselves (`corefSourceText`).
  */
-declare const buildPlaceholderMap: (entities: Entity[], _ctx?: PipelineContext) => Map<string, string>;
+type PlaceholderMapOptions = {
+  reservedText?: string;
+};
+declare const buildPlaceholderMap: (entities: Entity[], _ctx?: PipelineContext, {
+  reservedText
+}?: PlaceholderMapOptions) => Map<string, string>;
 /**
  * Apply redactions to the source text, replacing each
  * confirmed entity span using the configured operator.
@@ -775,7 +1433,7 @@ type TriggerPatterns = {
   patterns: string[];
   rules: TriggerRule[];
 };
-declare const buildTriggerPatterns: () => Promise<TriggerPatterns>;
+declare const buildTriggerPatterns: (languages?: readonly string[]) => Promise<TriggerPatterns>;
 /**
  * Process trigger matches from the unified search.
  * Receives all matches; filters to the trigger slice
@@ -785,20 +1443,6 @@ declare const buildTriggerPatterns: () => Promise<TriggerPatterns>;
  */
 declare const processTriggerMatches: (allMatches: Match[], sliceStart: number, sliceEnd: number, fullText: string, rules: readonly TriggerRule[]) => Entity[];
 //#endregion
-//#region src/detectors/address-seeds.d.ts
-declare const buildStreetTypePatterns: () => Promise<string[]>;
-/**
- * Process address seeds from the unified search.
- * Receives all matches; filters to the street types
- * slice via sliceStart/sliceEnd. Uses fullText and
- * existingEntities for seed collection, clustering,
- * expansion, and scoring.
- *
- * Runs as a post-processor after all other detectors,
- * using their output as seed sources.
- */
-declare const processAddressSeeds: (allMatches: Match[], sliceStart: number, sliceEnd: number, fullText: string, existingEntities: Entity[]) => Promise<Entity[]>;
-//#endregion
 //#region src/detectors/gazetteer.d.ts
 /**
  * Build TextSearch-compatible patterns from gazetteer
@@ -830,13 +1474,16 @@ declare const buildGazetteerPatterns: (entries: GazetteerEntry[]) => {
 declare const processGazetteerMatches: (allMatches: Match[], sliceStart: number, sliceEnd: number, fullText: string, data: GazetteerData) => Entity[];
 //#endregion
 //#region src/detectors/coreference.d.ts
+type ExtractDefinedTermsOptions = {
+  languages?: readonly string[];
+};
 type DefinedTerm = {
   alias: string;
   label: string; /** Position of the definition in the source text */
   definitionStart: number; /** Original entity text the alias refers to */
   sourceText: string;
 };
-declare const extractDefinedTerms: (fullText: string, entities: Entity[], ctx?: PipelineContext) => Promise<DefinedTerm[]>;
+declare const extractDefinedTerms: (fullText: string, entities: Entity[], ctx?: PipelineContext, options?: ExtractDefinedTermsOptions) => Promise<DefinedTerm[]>;
 /**
  * Find all occurrences of defined-term aliases in the
  * full text. Returns Entity spans for each match.
@@ -973,21 +1620,6 @@ declare const initAddressComponents: () => Promise<void>;
  */
 declare const filterFalsePositives: (entities: Entity[], ctx?: PipelineContext, fullText?: string) => Entity[];
 //#endregion
-//#region src/filters/confidence-boost.d.ts
-/**
- * Boost confidence of near-miss NER entities that appear
- * near high-confidence detections (regex, trigger phrase).
- *
- * If an NER entity scored between (threshold - 0.15) and
- * threshold, count how many confirmed entities exist within
- * a 150-char window. Add +0.05 per co-located entity.
- * If the boosted score crosses the threshold, include it.
- *
- * Only mutates score on near-miss entities; high-confidence
- * entities pass through unchanged.
- */
-declare const boostNearMissEntities: (entities: Entity[], threshold: number) => Entity[];
-//#endregion
 //#region src/filters/hotword-rules.d.ts
 type HotwordRule = {
   hotwords: string[];
@@ -1110,17 +1742,45 @@ declare const prepareBatch: (tokenizer: Tokenizer, texts: string[], entities: st
 };
 //#endregion
 //#region src/util/chunker.d.ts
+/** A chunk paired with its start offset in the source text. */
+type ChunkSpan = {
+  text: string;
+  offset: number;
+};
+/**
+ * Split text into overlapping chunks, each paired with its
+ * exact start offset in the source text.
+ *
+ * Carrying the offset out of the splitter is the robust way to
+ * map chunk-local entity offsets back to document offsets:
+ * downstream code never has to re-locate a chunk by content
+ * search (which mis-locates when boilerplate repeats; see
+ * computeChunkOffsets).
+ *
+ * Character-based splitting (rough token approximation for
+ * GLiNER's ~512 token window); breaks at sentence boundaries
+ * when possible.
+ */
+declare const chunkTextWithOffsets: (text: string) => ChunkSpan[];
 /**
- * Split text into overlapping chunks for GLiNER's
- * ~512 token context window. Character-based splitting
- * (rough approximation of token limits).
+ * Split text into overlapping chunks for GLiNER's ~512 token
+ * context window. Character-based splitting (rough token
+ * approximation); breaks at sentence boundaries when possible.
  *
- * Tries to break at sentence boundaries when possible.
+ * Prefer chunkTextWithOffsets when you also need each chunk's
+ * document offset.
  */
 declare const chunkText: (text: string) => string[];
 /**
- * Compute the byte offset of each chunk within the
- * original document text.
+ * Compute the start offset of each chunk within the original
+ * document text by content search.
+ *
+ * @deprecated Re-locates each chunk with `indexOf`, which can
+ *   match the wrong position when identical content repeats in
+ *   the document (common in boilerplate-heavy legal text) and
+ *   then desyncs every subsequent offset. Use
+ *   `chunkTextWithOffsets`, which carries exact offsets out of
+ *   the splitter.
  */
 declare const computeChunkOffsets: (fullText: string, chunks: string[]) => number[];
 /**
@@ -1179,5 +1839,5 @@ 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, type PipelineSearchOptions, 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, getNameCorpusNonWesternNames, initAddressComponents, initHotwordRules, initNameCorpus, initZoneClassifier, levenshtein, mergeAndDedup, mergeChunkEntities, normalizeForSearch, prepareBatch, preparePipelineSearch, processAddressSeeds, processDenyListMatches, processGazetteerMatches, processLegalFormMatches, processRegexMatches, processTriggerMatches, propagateOrgNames, redactText, resolveCountries, resolveOperator, runPipeline, runUnifiedSearch, sanitizeEntities, tokenizeText, warmLegalRoleHeads };
+export { type AnonymisationOperator, CURRENCY_PATTERN_META, type ChunkSpan, type CountryCode, type CustomDenyListEntry, type CustomRegexPattern, DATE_PATTERN_META, DEFAULT_ENTITY_LABELS, DEFAULT_NATIVE_PIPELINE_CONFIG, 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 NativeAnonymizeBinding, type NativeAnonymizerFromConfigOptions, type NativeAnonymizerFromPackageOptions, type NativeBindingVersionOptions, type NativeNormalizeOptions, type NativeOperatorConfig, type NativePipelineBuildOptions, type NativePipelineCompatibility, type NativePipelineEntity, type NativePipelineFromPackageOptions, type NativePipelinePackageOptions, type NativePipelineUnsupportedFeature, type NativePreparedSearchBinding, type NativeRedactionResult, type NativeResultEventCallback, type NativeSearchPackageInput, type NativeSearchPackageOptions, type NativeStaticRedactionResult, type NerInferenceFn, OPERATOR_REGISTRY, OPERATOR_TYPES, type OperatorConfig, type OperatorType, type PipelineConfig, type PipelineContext, type PipelineOptions, type PipelineSearchOptions, PreparedNativeAnonymizer, PreparedNativePipeline, PreparedSearch, type PreparedSearch as PreparedSearchInstance, REGEX_META, REGEX_PATTERNS, REGIONS, type RawInferenceResult, type RedactionResult, type RegexMeta, type RegionId, type ReviewDecision, type ReviewedEntity, type SharedNativeDiagnosticsJsonOptions, type SharedNativePreparedPackageOptions, type SharedNativeRedactTextJsonOptions, type SharedNativeRedactTextOptions, type SharedNativeRedactTextStreamJsonOptions, type SharedNativeSearchPackageOptions, type TriggerExtension, type TriggerGroupConfig, type TriggerRule, type TriggerStrategy, type TriggerValidation, type UnifiedResult, type UnifiedSearchInstance, ZONE_SCORE_ADJUSTMENTS, type ZoneSpan, applyHotwordRules, applyZoneAdjustments, assertNativeBindingVersion, assertNativePipelineSupported, boostNearMissEntities, buildDenyList, buildGazetteerPatterns, buildLegalFormPatterns, buildPlaceholderMap, buildStreetTypePatterns, buildTriggerPatterns, buildUnifiedSearch, chunkText, chunkTextWithOffsets, classifyZones, computeChunkOffsets, corefKey, createNativeAnonymizerFromConfig, createNativeAnonymizerFromPackage, createNativePipelineFromConfig, createNativePipelineFromPackage, createPipelineContext, deanonymise, decodeSpans, decodeTokenSpans, detectNameCorpus, diagnostics_json, encodeNativeSearchConfig, encodeNativeSearchConfigInput, ensureDenyListData, exportRedactionKey, extractDefinedTerms, filterFalsePositives, findCoreferenceSpans, getCurrencyPatterns, getDatePatterns, getNameCorpusNonWesternNames, getNativeBindingVersion, getNativePipelineCompatibility, initAddressComponents, initHotwordRules, initNameCorpus, initZoneClassifier, levenshtein, load_prepared_package, mergeAndDedup, mergeChunkEntities, native_package_version, normalizeForSearch, normalize_for_search, prepareBatch, prepareNativePipelineConfig, prepareNativePipelinePackage, prepareNativeSearchPackage, preparePipelineSearch, prepare_search_package, processAddressSeeds, processDenyListMatches, processGazetteerMatches, processLegalFormMatches, processRegexMatches, processTriggerMatches, propagateOrgNames, redactText, redact_text, redact_text_json, redact_text_stream_json, resolveCountries, resolveOperator, runPipeline, runUnifiedSearch, sanitizeEntities, tokenizeText, warmLegalRoleHeads };
 //# sourceMappingURL=wasm.d.mts.map