euredact 0.3.9 → 0.5.0

package/dist/esm/types.d.ts

@@ -1,5 +1,20 @@
 export declare enum EntityType {
-    NAME = "NAME",
+    /**
+     * Canonical name for a person's name. `NAME` is kept below as a legacy alias
+     * with the same value, exactly as `IBAN` aliases `BANK_ACCOUNT`, so
+     * `EntityType.NAME === EntityType.PERSON_NAME` and existing code keeps
+     * working — but the emitted value, and therefore the `[PERSON_NAME]`
+     * placeholder written into redacted text, is canonical.
+     *
+     * Renamed rather than translated at the cloud boundary: two names for one
+     * type is how a whole category goes missing when someone filters on the
+     * spelling they happened to know. Nothing could have depended on the old
+     * value — the type is cloud-only and the cloud tier was stubbed, so it was
+     * never emitted.
+     */
+    PERSON_NAME = "PERSON_NAME",
+    /** @deprecated legacy alias of {@link EntityType.PERSON_NAME} */
+    NAME = "PERSON_NAME",
     ADDRESS = "ADDRESS",
     /**
      * Canonical name for a bank account identifier. `IBAN` is kept below as a
@@ -47,6 +62,15 @@ export declare enum EntityType {
     UUID = "UUID",
     SOCIAL_HANDLE = "SOCIAL_HANDLE",
     SECRET = "SECRET",
+    ORGANISATION_NAME = "ORGANISATION_NAME",
+    JOB_TITLE = "JOB_TITLE",
+    MEDICAL_CONDITION = "MEDICAL_CONDITION",
+    SENSITIVE_ATTRIBUTE = "SENSITIVE_ATTRIBUTE",
+    BIOMETRIC_REF = "BIOMETRIC_REF",
+    FINANCIAL_AMOUNT = "FINANCIAL_AMOUNT",
+    QUASI_IDENTIFIER = "QUASI_IDENTIFIER",
+    CREDENTIAL = "CREDENTIAL",
+    URL = "URL",
     OTHER = "OTHER"
 }
 /**
@@ -55,6 +79,14 @@ export declare enum EntityType {
  * legacy_aliases). The engine never *emits* these names.
  */
 export declare const LEGACY_TYPE_ALIASES: Record<string, string>;
+/**
+ * Resolve a possibly-legacy type name to its canonical form.
+ *
+ * Mirrors `EntityType._missing_` in the Python SDK. An unrecognised name is
+ * returned unchanged rather than dropped: a client one release behind the
+ * service must not silently lose a whole category of PII.
+ */
+export declare function canonicalType(name: string): string;
 export declare enum DetectionSource {
     RULES = "rules",
     CLOUD = "cloud"
@@ -101,6 +133,24 @@ export interface Detection {
      */
     outOfScope?: boolean;
 }
+/**
+ * A detection the allowlist kept out of the redacted output.
+ *
+ * An exemption is a deliberate decision to leave a direct identifier in a
+ * document that otherwise claims to be redacted, so it is reported rather than
+ * silently dropped: without this, "never detected" and "detected and kept"
+ * look identical on the result.
+ */
+export interface Exemption {
+    entityType: EntityType | string;
+    start: number;
+    end: number;
+    text: string;
+    /** The allowlist entry that matched, as the caller wrote it. */
+    rule: string;
+    /** `"value"` for the literal allowlist, `"domain"` for allowlistDomains. */
+    ruleKind: "value" | "domain";
+}
 export interface RedactResult {
     redactedText: string;
     detections: Detection[];
@@ -126,6 +176,18 @@ export interface RedactResult {
      * already means the tier selector, which {@link RedactResult.source} reports.
      */
     detectionMode: string;
+    /**
+     * Token -> original text, populated by `redact(text, { tokenize: true })`.
+     *
+     * Hand it to `restore()` with the text that came back from wherever the
+     * redacted text went. Every token is unique to this call: the same value in
+     * another document gets a different token. Treat the mapping as read-only —
+     * a cached result shares it with every caller.
+     */
+    tokens: Record<string, string>;
+    /** Detections the allowlist kept out of the output, with the rule that
+     *  matched. Empty unless an allowlist was in force. */
+    exempted: Exemption[];
 }
 export interface PatternDef {
     entityType: EntityType | string;

package/dist/esm/types.js

@@ -1,6 +1,21 @@
 export var EntityType;
 (function (EntityType) {
-    EntityType["NAME"] = "NAME";
+    /**
+     * Canonical name for a person's name. `NAME` is kept below as a legacy alias
+     * with the same value, exactly as `IBAN` aliases `BANK_ACCOUNT`, so
+     * `EntityType.NAME === EntityType.PERSON_NAME` and existing code keeps
+     * working — but the emitted value, and therefore the `[PERSON_NAME]`
+     * placeholder written into redacted text, is canonical.
+     *
+     * Renamed rather than translated at the cloud boundary: two names for one
+     * type is how a whole category goes missing when someone filters on the
+     * spelling they happened to know. Nothing could have depended on the old
+     * value — the type is cloud-only and the cloud tier was stubbed, so it was
+     * never emitted.
+     */
+    EntityType["PERSON_NAME"] = "PERSON_NAME";
+    /** @deprecated legacy alias of {@link EntityType.PERSON_NAME} */
+    EntityType["NAME"] = "PERSON_NAME";
     EntityType["ADDRESS"] = "ADDRESS";
     /**
      * Canonical name for a bank account identifier. `IBAN` is kept below as a
@@ -48,6 +63,20 @@ export var EntityType;
     EntityType["UUID"] = "UUID";
     EntityType["SOCIAL_HANDLE"] = "SOCIAL_HANDLE";
     EntityType["SECRET"] = "SECRET";
+    // ── [CLOUD EXTENSION] ────────────────────────────────────────────────────
+    // Types only the cloud tier can detect. The rule engine never emits these —
+    // there is no shape to match on — which is exactly why they exist: the
+    // model's job is what the rules cannot catch. Named to match the detection
+    // canon the service is trained and evaluated against.
+    EntityType["ORGANISATION_NAME"] = "ORGANISATION_NAME";
+    EntityType["JOB_TITLE"] = "JOB_TITLE";
+    EntityType["MEDICAL_CONDITION"] = "MEDICAL_CONDITION";
+    EntityType["SENSITIVE_ATTRIBUTE"] = "SENSITIVE_ATTRIBUTE";
+    EntityType["BIOMETRIC_REF"] = "BIOMETRIC_REF";
+    EntityType["FINANCIAL_AMOUNT"] = "FINANCIAL_AMOUNT";
+    EntityType["QUASI_IDENTIFIER"] = "QUASI_IDENTIFIER";
+    EntityType["CREDENTIAL"] = "CREDENTIAL";
+    EntityType["URL"] = "URL";
     EntityType["OTHER"] = "OTHER";
 })(EntityType || (EntityType = {}));
 /**
@@ -57,7 +86,21 @@ export var EntityType;
  */
 export const LEGACY_TYPE_ALIASES = {
     IBAN: "BANK_ACCOUNT",
+    NAME: "PERSON_NAME",
+    STREET_ADDRESS: "ADDRESS",
+    NATIONALITY_ETHNICITY: "SENSITIVE_ATTRIBUTE",
 };
+/**
+ * Resolve a possibly-legacy type name to its canonical form.
+ *
+ * Mirrors `EntityType._missing_` in the Python SDK. An unrecognised name is
+ * returned unchanged rather than dropped: a client one release behind the
+ * service must not silently lose a whole category of PII.
+ */
+export function canonicalType(name) {
+    const upper = String(name ?? "").trim().toUpperCase();
+    return LEGACY_TYPE_ALIASES[upper] ?? upper;
+}
 export var DetectionSource;
 (function (DetectionSource) {
     DetectionSource["RULES"] = "rules";

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "euredact",
-  "version": "0.3.9",
+  "version": "0.5.0",
   "description": "European PII redaction SDK -- fast, zero-dependency, rule-based detection and redaction of personal data across 31 European countries",
   "homepage": "https://euredact.dev",
   "license": "Apache-2.0",
@@ -54,7 +54,7 @@
     "build:cjs": "tsc -p tsconfig.cjs.json",
     "postbuild:cjs": "echo '{\"type\":\"commonjs\"}' > dist/cjs/package.json",
     "build": "rm -rf dist && npm run build:esm && npm run build:cjs",
-    "test": "npm run test:spec && npm run test:inference && npm run test:conformance",
+    "test": "npm run test:spec && npm run test:inference && npm run test:cloud && npm run test:tokenize && npm run test:allowlist && npm run test:eval-scoring && npm run test:conformance",
     "benchmark": "node --import tsx src/__tests__/benchmark.ts",
     "prepublishOnly": "npm run build",
     "test:spec": "node --import tsx src/__tests__/spec.ts",
@@ -62,7 +62,11 @@
     "test:eval:full": "node --max-old-space-size=8192 --import tsx src/__tests__/evalFull.ts",
     "test:conformance": "node --import tsx src/__tests__/conformance.ts",
     "test:inference": "node --import tsx src/__tests__/inference.ts",
-    "dump": "node --import tsx src/__tests__/dump.ts"
+    "dump": "node --import tsx src/__tests__/dump.ts",
+    "test:cloud": "node --import tsx src/__tests__/cloud.ts",
+    "test:tokenize": "node --import tsx src/__tests__/tokenize.ts",
+    "test:allowlist": "node --import tsx src/__tests__/allowlist.ts",
+    "test:eval-scoring": "node --import tsx src/__tests__/evalScoring.ts"
   },
   "devDependencies": {
     "@types/node": "22.10.2",