@stll/anonymize-wasm 1.4.7 → 1.4.9

package/dist/triggers.global.mjs

@@ -1,33 +1,78 @@
 //#region src/data/triggers.global.json
-var triggers_global_default = [{
-	"id": "swift-bic",
-	"triggers": ["SWIFT", "BIC"],
-	"label": "bank account number",
-	"strategy": {
-		"type": "n-words",
-		"count": 1
+var triggers_global_default = [
+	{
+		"id": "swift-bic",
+		"triggers": ["SWIFT", "BIC"],
+		"label": "bank account number",
+		"strategy": {
+			"type": "n-words",
+			"count": 1
+		},
+		"extensions": ["add-colon"],
+		"validations": [{
+			"type": "matches-pattern",
+			"pattern": "^[A-Za-z]{4}[A-Za-z]{2}[A-Za-z0-9]{2}(?:[A-Za-z0-9]{3})?$"
+		}]
 	},
-	"extensions": ["add-colon"],
-	"validations": [{
-		"type": "matches-pattern",
-		"pattern": "^[A-Za-z]{4}[A-Za-z]{2}[A-Za-z0-9]{2}(?:[A-Za-z0-9]{3})?$"
-	}]
-}, {
-	"id": "amount-prefix",
-	"triggers": [
-		"ve výši",
-		"in the amount of",
-		"in Höhe von",
-		"d'un montant de",
-		"w wysokości",
-		"nell'importo di",
-		"por un importe de"
-	],
-	"label": "monetary amount",
-	"strategy": { "type": "to-next-comma" },
-	"extensions": ["add-colon"],
-	"validations": [{ "type": "has-digits" }]
-}];
+	{
+		"id": "us-aba-routing",
+		"triggers": [
+			"ABA",
+			"ABA Number",
+			"RTN",
+			"Routing Number",
+			"Routing No.",
+			"Routing Transit Number"
+		],
+		"label": "bank account number",
+		"strategy": {
+			"type": "n-words",
+			"count": 1
+		},
+		"extensions": ["add-colon"],
+		"validations": [{ "type": "has-digits" }, {
+			"type": "valid-id",
+			"validator": "us.rtn"
+		}]
+	},
+	{
+		"id": "intl-labeled-phone",
+		"triggers": [
+			"fax",
+			"telecopy",
+			"telecopier",
+			"facsimile",
+			"telephone",
+			"phone number",
+			"contact phone number"
+		],
+		"label": "phone number",
+		"strategy": {
+			"type": "match-pattern",
+			"pattern": "(?:\\+\\d{1,3}[\\s.\\-]?)?\\(?\\d{1,4}\\)?(?:[\\s.\\-]\\d{1,4}){2,4}"
+		},
+		"validations": [{
+			"type": "matches-pattern",
+			"pattern": "(?:\\d[^\\d]*){6,}"
+		}]
+	},
+	{
+		"id": "amount-prefix",
+		"triggers": [
+			"ve výši",
+			"in the amount of",
+			"in Höhe von",
+			"d'un montant de",
+			"w wysokości",
+			"nell'importo di",
+			"por un importe de"
+		],
+		"label": "monetary amount",
+		"strategy": { "type": "to-next-comma" },
+		"extensions": ["add-colon"],
+		"validations": [{ "type": "has-digits" }]
+	}
+];
 //#endregion
 export { triggers_global_default as default };
 

package/dist/wasm.d.mts

@@ -5,18 +5,45 @@ import { Tokenizer } from "@huggingface/tokenizers";
 
 //#region src/types.d.ts
 /**
- * A detected PII entity span in the source text.
- * Every detection layer produces these.
+ * Fields shared by every entity span in the source text.
  */
-type Entity = {
+type EntityBase = {
   start: number;
   end: number;
   label: string;
   text: string;
   score: number;
-  source: DetectionSource;
   sourceDetail?: "custom-deny-list" | "custom-regex" | "gazetteer-extension";
 };
+/**
+ * A PII entity span found by a primary detection layer
+ * (regex, NER, legal forms, deny list, ...).
+ */
+type DetectedEntity = EntityBase & {
+  source: Exclude<DetectionSource, typeof DETECTION_SOURCES.COREFERENCE>;
+};
+/**
+ * An alias mention of a previously detected entity: a
+ * defined term ("the Seller") or a propagated bare
+ * mention ("Acme" after "Acme Corp.").
+ *
+ * `corefSourceText` is required by construction, so an
+ * alias cannot exist without the link back to its source
+ * entity. Placeholder numbering reads it to give the
+ * alias the same placeholder as the source. The link
+ * travels with the entity instead of living in a
+ * side-channel map that a producer could forget to
+ * write — or that a later pass could clear.
+ */
+type CorefAliasEntity = EntityBase & {
+  source: typeof DETECTION_SOURCES.COREFERENCE; /** Full text of the source entity this alias refers to. */
+  corefSourceText: string;
+};
+/**
+ * A detected PII entity span in the source text.
+ * Every detection layer produces these.
+ */
+type Entity = DetectedEntity | CorefAliasEntity;
 /**
  * Entity after human review. Extends the base Entity
  * with a review decision.
@@ -117,7 +144,7 @@ type TriggerValidation = {
 };
 /** Built-in stdnum validators that can be referenced
  *  by `valid-id` validations. */
-type ValidIdValidator = "br.cpf" | "br.cnpj";
+type ValidIdValidator = "br.cpf" | "br.cnpj" | "us.rtn";
 /** Auto-generated trigger variants — closed set. */
 type TriggerExtension = "add-colon" | "add-trailing-space" | "add-colon-space" | "normalize-spaces";
 /** V2 trigger config entry (JSON shape). */
@@ -521,6 +548,10 @@ declare const buildUnifiedSearch: (config: PipelineConfig, gazetteerEntries?: Ga
  * shallow copies (spread). Uses position + label so the
  * key is identical for the original object and any
  * `{ ...entity }` copy produced by mergeAndDedup.
+ *
+ * @deprecated No longer used internally: coref alias
+ *   links travel on the entities themselves
+ *   (`corefSourceText`). Kept for API compatibility.
  */
 declare const corefKey: (e: Entity) => string;
 /**
@@ -582,17 +613,6 @@ type PipelineContext = {
   zoneHeadingPatterns: RegExp[] | null;
   zoneSigningPatterns: RegExp[] | null;
   zoneInitPromise: Promise<void> | null;
-  /**
-   * Maps coreference entities to their source entity
-   * text. Populated by findCoreferenceSpans, consumed
-   * by buildPlaceholderMap for consistent placeholder
-   * numbering across aliases and source entities.
-   *
-   * Keyed by `start:end:label` composite string so
-   * lookups survive shallow copies (e.g. from
-   * mergeAndDedup's spread operator).
-   */
-  corefSourceMap: Map<string, string>;
 };
 /** Create a fresh, empty pipeline context. */
 declare const createPipelineContext: () => PipelineContext;
@@ -663,12 +683,11 @@ declare const runPipeline: (options: PipelineOptions) => Promise<Entity[]>;
  * Placeholder format: [LABEL_N] where LABEL is uppercase
  * and N is a 1-based counter per label.
  *
- * @param ctx Pipeline context. Must be the same instance
- *   passed to `runPipeline` (or `findCoreferenceSpans`)
- *   so coreference placeholder links are preserved.
- *   Defaults to `defaultContext` for single-tenant usage.
+ * @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>;
+declare const buildPlaceholderMap: (entities: Entity[], _ctx?: PipelineContext) => Map<string, string>;
 /**
  * Apply redactions to the source text, replacing each
  * confirmed entity span using the configured operator.
@@ -811,11 +830,14 @@ declare const extractDefinedTerms: (fullText: string, entities: Entity[], ctx?:
  * character before the start and after the end are NOT
  * word characters (letter/digit).
  *
- * Populates `ctx.corefSourceMap` with entries linking
- * each coref entity to its source entity text, for
- * consistent placeholder numbering.
+ * Each returned alias carries `corefSourceText` linking
+ * it to its source entity text, for consistent
+ * placeholder numbering.
+ *
+ * @param _ctx Unused. Kept for signature compatibility;
+ *   alias links now travel on the entities themselves.
  */
-declare const findCoreferenceSpans: (fullText: string, terms: DefinedTerm[], ctx?: PipelineContext) => Entity[];
+declare const findCoreferenceSpans: (fullText: string, terms: DefinedTerm[], _ctx?: PipelineContext) => Entity[];
 //#endregion
 //#region src/detectors/org-propagation.d.ts
 /**
@@ -824,6 +846,12 @@ declare const findCoreferenceSpans: (fullText: string, terms: DefinedTerm[], ctx
  * to get the base name, and re-scan the full text for
  * bare mentions of that base name. Returns new entities
  * for occurrences not already covered.
+ *
+ * Propagated mentions are coref aliases: each carries
+ * `corefSourceText` linking it to the full seed entity
+ * text, so placeholder numbering assigns the bare
+ * mention the same placeholder as its source ("Acme"
+ * and "Acme Corp." both become [ORGANIZATION_1]).
  */
 declare const propagateOrgNames: (entities: Entity[], fullText: string) => Entity[];
 //#endregion