@stll/anonymize-wasm 1.4.0 → 1.4.3

package/dist/common-words-en.mjs.map

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

package/dist/generic-roles.mjs

@@ -27,6 +27,7 @@ var roles = [
 	"compradora",
 	"committee",
 	"company",
+	"companies",
 	"contractor",
 	"contratante",
 	"contratista",

package/dist/legal-form-leading-clauses.mjs

@@ -0,0 +1,18 @@
+//#region src/data/legal-form-leading-clauses.json
+var _comment = "Per-language clause text that may precede an organization name before a legal-form suffix. `phrases` are trimmed unconditionally when found inside an over-captured legal-form span. `directPrefixes` are trimmed only when the text before the prefix already looks like prose, so company names such as `Stand By Me LLC` are preserved.";
+var en = {
+	"phrases": [
+		"by and among",
+		"by and between",
+		"is between"
+	],
+	"directPrefixes": ["by"]
+};
+var legal_form_leading_clauses_default = {
+	_comment,
+	en
+};
+//#endregion
+export { _comment, legal_form_leading_clauses_default as default, en };
+
+//# sourceMappingURL=legal-form-leading-clauses.mjs.map

package/dist/legal-form-leading-clauses.mjs.map

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

package/dist/rolldown-runtime.mjs

@@ -0,0 +1,13 @@
+//#region \0rolldown/runtime.js
+var __defProp = Object.defineProperty;
+var __exportAll = (all, no_symbols) => {
+	let target = {};
+	for (var name in all) __defProp(target, name, {
+		get: all[name],
+		enumerable: true
+	});
+	if (!no_symbols) __defProp(target, Symbol.toStringTag, { value: "Module" });
+	return target;
+};
+//#endregion
+export { __exportAll as t };

package/dist/structural-single-cap-prefixes.mjs

@@ -0,0 +1,99 @@
+//#region src/data/structural-single-cap-prefixes.json
+var _comment = "Structural document-unit words that can precede a single-letter or letter-number marker before a legal form word, e.g. 'Schedule A LLC Members' or 'Exhibit A-1 LLC Agreement'. Lowercased; matched case-insensitively. Organised per-language; runtime unions all languages because match spans do not carry a language tag.";
+var en = [
+	"annex",
+	"appendix",
+	"article",
+	"attachment",
+	"clause",
+	"exhibit",
+	"paragraph",
+	"part",
+	"schedule",
+	"section"
+];
+var cs = [
+	"bod",
+	"paragraf",
+	"podbod",
+	"ustanovení",
+	"článek"
+];
+var sk = [
+	"bod",
+	"hlava",
+	"oddiel",
+	"odsek",
+	"pododdiel",
+	"sekcia",
+	"časť",
+	"článok"
+];
+var de = [
+	"absatz",
+	"abschnitt",
+	"abteilung",
+	"anlage",
+	"anhang",
+	"artikel",
+	"kapitel",
+	"klausel",
+	"punkt",
+	"sektion",
+	"unterabschnitt",
+	"unterpunkt"
+];
+var es = [
+	"anexo",
+	"apartado",
+	"artículo",
+	"cláusula",
+	"sección"
+];
+var fr = [
+	"annexe",
+	"article",
+	"clause",
+	"paragraphe",
+	"partie",
+	"section"
+];
+var it = [
+	"allegato",
+	"appendice",
+	"articolo",
+	"clausola",
+	"paragrafo",
+	"sezione"
+];
+var pl = [
+	"aneksem",
+	"artykuł",
+	"paragraf",
+	"punkt",
+	"sekcja",
+	"załącznik"
+];
+var pt_br = [
+	"anexo",
+	"artigo",
+	"cláusula",
+	"parágrafo",
+	"seção"
+];
+var structural_single_cap_prefixes_default = {
+	_comment,
+	en,
+	cs,
+	sk,
+	de,
+	es,
+	fr,
+	it,
+	pl,
+	"pt-br": pt_br
+};
+//#endregion
+export { _comment, cs, de, structural_single_cap_prefixes_default as default, en, es, fr, it, pl, pt_br as "pt-br", sk };
+
+//# sourceMappingURL=structural-single-cap-prefixes.mjs.map

package/dist/structural-single-cap-prefixes.mjs.map

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

package/dist/triggers.en.mjs

@@ -33,7 +33,22 @@ var triggers_en_default = [
 	{
 		"id": "en-jurisdiction",
 		"label": "address",
-		"strategy": { "type": "to-next-comma" },
+		"strategy": {
+			"type": "to-next-comma",
+			"maxLength": 30,
+			"stopWords": [
+				"or",
+				"and",
+				"in",
+				"sitting",
+				"shall",
+				"that",
+				"which",
+				"without",
+				"for",
+				"to"
+			]
+		},
 		"triggers": [
 			"State of",
 			"Commonwealth of",

package/dist/wasm.d.mts

@@ -15,7 +15,7 @@ type Entity = {
   text: string;
   score: number;
   source: DetectionSource;
-  sourceDetail?: "custom-deny-list" | "custom-regex";
+  sourceDetail?: "custom-deny-list" | "custom-regex" | "gazetteer-extension";
 };
 /**
  * Entity after human review. Extends the base Entity
@@ -52,6 +52,16 @@ type TriggerStrategy = {
    * word-boundary, case-insensitive.
    */
   stopWords?: string[];
+  /**
+   * Hard cap on the captured span length, in characters,
+   * regardless of where the next comma / stop char sits.
+   * Use for triggers that label short formulaic phrases
+   * ("State of Delaware") and must not absorb the rest
+   * of a long forum-selection clause when the comma is
+   * sentences away. Falls back to the default 100-char
+   * fallback when omitted.
+   */
+  maxLength?: number;
 } | {
   type: "to-end-of-line";
 } | {
@@ -259,8 +269,19 @@ type Dictionaries = {
   /**
    * Pre-loaded city names, already merged across
    * all desired countries.
+   *
+   * Prefer `citiesByCountry` when callers also pass
+   * `denyListCountries` / `denyListRegions`; merged
+   * city arrays cannot be scoped after injection.
    */
   cities?: readonly string[];
+  /**
+   * Pre-loaded city names keyed by ISO 3166-1 alpha-2
+   * country code. When provided, the deny-list builder
+   * applies `denyListCountries` / `denyListRegions`
+   * before adding city patterns to the search automaton.
+   */
+  citiesByCountry?: Readonly<Record<string, readonly string[]>>;
 };
 type PipelineConfig = {
   threshold: number;
@@ -280,6 +301,15 @@ type PipelineConfig = {
    * deny-list search automaton.
    */
   enableNameCorpus: boolean;
+  /**
+   * Optional language scope for first-name/surname
+   * dictionaries, using the keys present in
+   * `dictionaries.firstNames` / `dictionaries.surnames`
+   * (for example `["en", "de"]`). When omitted, all
+   * injected name languages are used for backward
+   * compatibility.
+   */
+  nameCorpusLanguages?: string[];
   enableDenyList: boolean;
   denyListCountries?: string[];
   denyListRegions?: string[];
@@ -357,16 +387,17 @@ 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" | "customDenyList" | "dictionaries">;
+type DenyListConfig = Pick<PipelineConfig, "enableDenyList" | "enableNameCorpus" | "nameCorpusLanguages" | "denyListCountries" | "denyListRegions" | "denyListExcludeCategories" | "customDenyList" | "dictionaries">;
 /**
  * Source tag for each pattern in the automaton.
  * "deny-list" = standard deny list entry
+ * "city" = city dictionary 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" | "custom-deny-list" | "first-name" | "surname" | "title";
+type PatternSource = "deny-list" | "city" | "custom-deny-list" | "first-name" | "surname" | "title";
 /**
  * Pre-built deny list data. Constructed once by
  * `buildDenyList`, reused across `processDenyListMatches`
@@ -404,7 +435,7 @@ declare const buildDenyList: (config: DenyListConfig, ctx?: PipelineContext) =>
  * the search instance was built on a different context
  * (e.g. cachedSearch).
  */
-declare const ensureDenyListData: (ctx?: PipelineContext, dictionaries?: Dictionaries) => Promise<void>;
+declare const ensureDenyListData: (ctx?: PipelineContext, dictionaries?: Dictionaries, nameCorpusLanguages?: readonly string[]) => Promise<void>;
 /**
  * Process deny list matches from the unified search.
  * Receives all matches; filters to the deny list slice
@@ -500,6 +531,7 @@ type PipelineContext = {
   searchKey: string;
   searchPromise: Promise<UnifiedSearchInstance> | null;
   nameCorpus: NameCorpusData | null;
+  nameCorpusKey: string;
   nameCorpusPromise: Promise<void> | null;
   stopwords: ReadonlySet<string> | null;
   stopwordsPromise: Promise<ReadonlySet<string>> | null;
@@ -541,6 +573,22 @@ declare const createPipelineContext: () => PipelineContext;
 declare const sanitizeEntities: (entities: Entity[]) => Entity[];
 declare const mergeAndDedup: (...layers: Entity[][]) => Entity[];
 type NerInferenceFn = (fullText: string, labels: string[], threshold: number, signal?: AbortSignal) => Promise<Entity[]>;
+type PipelineSearchOptions = {
+  config: PipelineConfig;
+  gazetteerEntries?: GazetteerEntry[];
+  context?: PipelineContext;
+};
+/**
+ * Pre-build and cache the unified search instance for a
+ * pipeline configuration. Use the same context in
+ * `runPipeline` to reuse the prepared automata without
+ * passing `cachedSearch` around manually.
+ */
+declare const preparePipelineSearch: ({
+  config,
+  gazetteerEntries,
+  context
+}: PipelineSearchOptions) => Promise<UnifiedSearchInstance>;
 /**
  * Options for {@link runPipeline}.
  *
@@ -770,7 +818,7 @@ declare const propagateOrgNames: (entities: Entity[], fullText: string) => Entit
  *   with per-language first names and surnames. When
  *   omitted, only legacy config files are used.
  */
-declare const initNameCorpus: (ctx?: PipelineContext, dictionaries?: Dictionaries) => Promise<void>;
+declare const initNameCorpus: (ctx?: PipelineContext, dictionaries?: Dictionaries, languages?: readonly string[]) => Promise<void>;
 /**
  * Detect person names by looking up tokens against the
  * name corpus, then chaining adjacent name-like tokens.
@@ -1052,5 +1100,5 @@ 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, initAddressComponents, initHotwordRules, initNameCorpus, initZoneClassifier, levenshtein, mergeAndDedup, mergeChunkEntities, normalizeForSearch, prepareBatch, processAddressSeeds, processDenyListMatches, processGazetteerMatches, processLegalFormMatches, processRegexMatches, processTriggerMatches, propagateOrgNames, redactText, resolveCountries, resolveOperator, runPipeline, runUnifiedSearch, sanitizeEntities, tokenizeText, warmLegalRoleHeads };
+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, 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 };
 //# sourceMappingURL=wasm.d.mts.map