@stll/anonymize 2.0.1 → 2.0.2

package/README.md

@@ -6,7 +6,7 @@
 
 Runtime package for multi-layer PII detection and anonymization.
 
-It combines regex detectors, trigger phrases, deny-list matching, coreference handling, and NER into a single pipeline that works in native Node.js and in browser builds through the WASM entrypoint.
+It combines regex detectors, trigger phrases, deny-list matching, and coreference handling in a single deterministic pipeline that works in native Node.js and in browser builds through the WASM entrypoint.
 
 ## Install
 
@@ -68,6 +68,130 @@ const warmDiagnosticsJson = anonymizer.warmLazyRegexDiagnosticsJson();
 const result = anonymizer.redact_text(text, { redactString: "***" });
 ```
 
+For related documents, create an explicit in-memory session from the prepared
+anonymizer. The session reuses placeholders for the same normalized entity while
+keeping its mutable mapping state isolated from other sessions:
+
+```ts
+const session = anonymizer.createRedactionSession("opaque_case_1");
+const first = session.redact_text(firstDocument);
+const second = session.redact_text(secondDocument);
+const restoredText = session.restoreText(first.redaction.redactedText);
+```
+
+`restoreText()` replaces only complete placeholders owned by that active
+session. It performs one non-cascading pass, leaves other session namespaces
+unchanged, and rejects unknown placeholders in its own namespace. Pass the
+caller-supplied observation time as the second argument for lifecycle sessions.
+
+`session.toPlaintextJson()` supports deterministic in-memory transfer between
+runtime instances. Its output contains original personal data in plaintext: do
+not log it or persist it without an application-owned protection layer. Restore
+validated transfer state with `anonymizer.restoreRedactionSession(json)`.
+
+For persistence or transfer, prefer an authenticated encrypted archive. Supply
+a caller-owned 32-byte key; key generation, storage, rotation, and access
+control remain application responsibilities. Restoring also requires the
+expected session ID so an archive cannot be substituted across records:
+
+```ts
+const key = crypto.getRandomValues(new Uint8Array(32));
+const archive = session.toEncryptedArchive(key);
+const restored = anonymizer.restoreEncryptedRedactionSession({
+  archive,
+  key,
+  expectedSessionId: "opaque_case_1",
+});
+```
+
+The archive contains personal data as ciphertext. Do not log the key or derive
+it directly from a password; use a key-management boundary appropriate to the
+deployment.
+
+Sessions can carry explicit lifecycle bounds. The engine never reads the system
+clock; supply the UTC epoch-second observation time for each lifecycle-aware
+operation:
+
+```ts
+const key = crypto.getRandomValues(new Uint8Array(32));
+const session = anonymizer.createRedactionSessionWithLifecycle({
+  sessionId: "opaque_case_2",
+  createdAtEpochSeconds: 1_800_000_000,
+  expiresAtEpochSeconds: 1_800_086_400,
+});
+const result = session.redactTextAt({
+  fullText: document,
+  observedAtEpochSeconds: 1_800_000_100,
+});
+const metadata = session.inspect(1_800_000_100); // contains no entity values
+const archive = session.toEncryptedArchiveAt(key, 1_800_000_100);
+const restored = anonymizer.restoreEncryptedRedactionSession({
+  archive,
+  key,
+  expectedSessionId: "opaque_case_2",
+  observedAtEpochSeconds: 1_800_000_100,
+});
+const deletion = session.delete();
+```
+
+Expiry is fail-closed at its exact boundary. `delete()` performs logical
+deletion: it clears the session mappings and prevents future use, but does not
+revoke earlier exported copies or claim physical erasure of process memory.
+
+Per-label operators support `replace`, `redact`, `keep`, and tagged `mask`
+configuration. `keep` records
+that an entity was processed while leaving its source text unchanged; it
+creates no reversible redaction-key entry:
+
+```ts
+const result = anonymizer.redactText(text, {
+  operators: { organization: "keep" },
+});
+```
+
+`mask` replaces a configured number of visible Unicode grapheme clusters from
+the start or end. A masking character must itself be exactly one grapheme:
+
+```ts
+const result = anonymizer.redactText(text, {
+  operators: {
+    "email address": {
+      type: "mask",
+      maskingCharacter: "*",
+      charactersToMask: 6,
+      direction: "start",
+    },
+  },
+});
+```
+
+Caller-produced spans enter the same resolution and redaction pipeline. Node
+and browser offsets use JavaScript UTF-16 string indexes; matched text is
+derived from the input:
+
+```ts
+const result = anonymizer.redactTextWithCallerDetections("😀Alice signed.", {
+  detections: [
+    {
+      start: 2,
+      end: 7,
+      label: "person",
+      score: 0.95,
+      providerId: "example-ner",
+      detectionId: "person-1",
+    },
+  ],
+});
+```
+
+`providerId` and `detectionId` are required provenance identifiers. They must
+be 1–128 ASCII characters, start with an alphanumeric character, and otherwise
+contain only alphanumerics, `.`, `_`, `:`, or `-`; do not encode personal data
+in them. Retained result entities preserve both IDs. Use
+`redactTextWithCallerDetectionsDiagnosticsJson()` for audit-safe input and
+retained counts. Diagnostic events include provenance, labels, offsets, and
+scores, but never matched text.
+
 The config module may export a `PipelineConfig` directly or `{ config, gazetteerEntries }`. Include `@stll/anonymize-data` dictionaries there if your runtime config uses the deny-list or name-corpus layers; keep the corresponding layers enabled for caller-owned `customDenyList`, `customRegexes`, and gazetteers. Those inputs are part of the prepared package and should be regenerated when they change.
 
 ## Python SDK
@@ -81,10 +205,40 @@ prepared = anonymize.preload_default_native_pipeline(
 )
 result = prepared.redact_text(text, redact_string="***")
 
+session = prepared.create_redaction_session("opaque_case_1")
+redacted = session.redact_text(text)
+restored_text = session.restore_text(redacted.redaction.redacted_text)
+archive = session.to_encrypted_archive(application_owned_32_byte_key)
+restored = prepared.restore_encrypted_redaction_session(
+    archive,
+    application_owned_32_byte_key,
+    "opaque_case_1",
+)
+
 print(result.redaction.redacted_text)
 ```
 
-The Python SDK uses the same Rust core and prepared-package contract as the Node SDK. Prefer `get_default_native_pipeline()`, `preload_default_native_pipeline()`, `load_prepared_package()`, or `load_prepared_package_file()` for repeated calls; top-level `redact_text()` and `redact_text_json()` prepare from config on each call.
+Python caller detections use Python character indexes:
+
+```py
+result = prepared.redact_text_with_caller_detections(
+    "😀Alice signed.",
+    [{"start": 1, "end": 6, "label": "person", "score": 0.95,
+      "provider_id": "example-ner", "detection_id": "person-1"}],
+)
+```
+
+Python preserves `provider_id` and `detection_id` on retained entities. Use
+`redact_text_with_caller_detections_diagnostics_json()` for the same audit-safe
+diagnostics contract.
+
+The Python SDK uses the same Rust core, encrypted session archive format, and
+prepared-package contract as the Node SDK. The application owns archive-key
+generation, storage, rotation, and authorization. Prefer
+`get_default_native_pipeline()`, `preload_default_native_pipeline()`,
+`load_prepared_package()`, or `load_prepared_package_file()` for repeated calls;
+top-level `redact_text()` and `redact_text_json()` prepare from config on each
+call.
 
 ## Caller-Owned Deny Lists and Regexes
 
@@ -143,7 +297,10 @@ export default {
 
 - Native architecture and extension guidance:
   [`ARCHITECTURE.md`](ARCHITECTURE.md).
-- `labels: []` disables deterministic label filtering; when NER is enabled it falls back to the default label set.
+- `labels: []` disables deterministic label filtering.
+- `enableNer` is a compatibility field for the removed TypeScript pipeline. The
+  native pipeline rejects `true`; model-produced spans will use a separate
+  caller-detection API.
 - `enableNameCorpus` also controls whether first names, surnames, and titles are injected into deny-list matching when `enableDenyList` is enabled.
 - The optional `@stll/anonymize-data` package carries the published dictionary and trigger data used when building prepared packages.
 - `customDenyList` and `customRegexes` are part of the prepared package input and should be regenerated when they change.

package/dist/capabilities.d.mts

@@ -0,0 +1,123 @@
+import { o as ENTITY_CAPABILITIES } from "./constants2.mjs";
+//#region src/capabilities.d.ts
+declare const CAPABILITY_MANIFEST_SCHEMA_VERSION: 1;
+declare const CAPABILITY_RUNTIMES: readonly ["node", "python", "wasm"];
+type CapabilityRuntime = (typeof CAPABILITY_RUNTIMES)[number];
+type CapabilityManifest = {
+  schemaVersion: typeof CAPABILITY_MANIFEST_SCHEMA_VERSION;
+  runtimes: readonly CapabilityRuntime[];
+  entities: typeof ENTITY_CAPABILITIES;
+};
+/**
+ * Versioned, runtime-free discovery contract for the deterministic pipeline.
+ * The manifest contains no accuracy claims; it describes available labels,
+ * activation, provenance sources, and runtime parity only.
+ */
+declare const CAPABILITY_MANIFEST: {
+  readonly schemaVersion: 1;
+  readonly runtimes: readonly ["node", "python", "wasm"];
+  readonly entities: readonly [{
+    readonly label: "person";
+    readonly selection: "default";
+    readonly detectionSources: readonly ["trigger", "regex", "deny-list", "coreference"];
+  }, {
+    readonly label: "organization";
+    readonly selection: "default";
+    readonly detectionSources: readonly ["trigger", "deny-list", "legal-form", "gazetteer", "coreference"];
+  }, {
+    readonly label: "phone number";
+    readonly selection: "default";
+    readonly detectionSources: readonly ["regex", "trigger"];
+  }, {
+    readonly label: "address";
+    readonly selection: "default";
+    readonly detectionSources: readonly ["regex", "trigger", "deny-list"];
+  }, {
+    readonly label: "country";
+    readonly selection: "default";
+    readonly detectionSources: readonly ["country"];
+  }, {
+    readonly label: "email address";
+    readonly selection: "default";
+    readonly detectionSources: readonly ["regex"];
+  }, {
+    readonly label: "date";
+    readonly selection: "default";
+    readonly detectionSources: readonly ["regex", "trigger"];
+  }, {
+    readonly label: "date of birth";
+    readonly selection: "default";
+    readonly detectionSources: readonly ["trigger"];
+  }, {
+    readonly label: "bank account number";
+    readonly selection: "default";
+    readonly detectionSources: readonly ["regex", "trigger"];
+  }, {
+    readonly label: "iban";
+    readonly selection: "default";
+    readonly detectionSources: readonly ["regex", "trigger"];
+  }, {
+    readonly label: "tax identification number";
+    readonly selection: "default";
+    readonly detectionSources: readonly ["regex", "trigger"];
+  }, {
+    readonly label: "identity card number";
+    readonly selection: "default";
+    readonly detectionSources: readonly ["regex", "trigger"];
+  }, {
+    readonly label: "birth number";
+    readonly selection: "default";
+    readonly detectionSources: readonly ["regex", "trigger"];
+  }, {
+    readonly label: "national identification number";
+    readonly selection: "default";
+    readonly detectionSources: readonly ["regex", "trigger"];
+  }, {
+    readonly label: "social security number";
+    readonly selection: "default";
+    readonly detectionSources: readonly ["regex", "trigger"];
+  }, {
+    readonly label: "registration number";
+    readonly selection: "default";
+    readonly detectionSources: readonly ["regex", "trigger"];
+  }, {
+    readonly label: "credit card number";
+    readonly selection: "default";
+    readonly detectionSources: readonly ["regex"];
+  }, {
+    readonly label: "passport number";
+    readonly selection: "default";
+    readonly detectionSources: readonly ["regex"];
+  }, {
+    readonly label: "crypto";
+    readonly selection: "default";
+    readonly detectionSources: readonly ["regex"];
+  }, {
+    readonly label: "monetary amount";
+    readonly selection: "default";
+    readonly detectionSources: readonly ["regex", "trigger"];
+  }, {
+    readonly label: "land parcel";
+    readonly selection: "default";
+    readonly detectionSources: readonly ["trigger"];
+  }, {
+    readonly label: "misc";
+    readonly selection: "default";
+    readonly detectionSources: readonly ["regex", "deny-list"];
+  }, {
+    readonly label: "ip address";
+    readonly selection: "opt-in";
+    readonly detectionSources: readonly ["regex"];
+  }, {
+    readonly label: "mac address";
+    readonly selection: "opt-in";
+    readonly detectionSources: readonly ["regex"];
+  }, {
+    readonly label: "url";
+    readonly selection: "opt-in";
+    readonly detectionSources: readonly ["regex"];
+  }];
+};
+//#endregion
+export { CAPABILITY_MANIFEST, CAPABILITY_MANIFEST_SCHEMA_VERSION, CAPABILITY_RUNTIMES, CapabilityManifest, CapabilityRuntime };
+//# sourceMappingURL=capabilities.d.mts.map

package/dist/capabilities.mjs

@@ -0,0 +1,22 @@
+import { ENTITY_CAPABILITIES } from "./constants.mjs";
+//#region src/capabilities.ts
+const CAPABILITY_MANIFEST_SCHEMA_VERSION = 1;
+const CAPABILITY_RUNTIMES = [
+	"node",
+	"python",
+	"wasm"
+];
+/**
+* Versioned, runtime-free discovery contract for the deterministic pipeline.
+* The manifest contains no accuracy claims; it describes available labels,
+* activation, provenance sources, and runtime parity only.
+*/
+const CAPABILITY_MANIFEST = {
+	schemaVersion: 1,
+	runtimes: CAPABILITY_RUNTIMES,
+	entities: ENTITY_CAPABILITIES
+};
+//#endregion
+export { CAPABILITY_MANIFEST, CAPABILITY_MANIFEST_SCHEMA_VERSION, CAPABILITY_RUNTIMES };
+
+//# sourceMappingURL=capabilities.mjs.map

package/dist/capabilities.mjs.map

@@ -0,0 +1 @@
+{"version":3,"file":"capabilities.mjs","names":[],"sources":["../src/capabilities.ts"],"sourcesContent":["import { ENTITY_CAPABILITIES } from \"./constants\";\n\nexport const CAPABILITY_MANIFEST_SCHEMA_VERSION = 1 as const;\n\nexport const CAPABILITY_RUNTIMES = [\"node\", \"python\", \"wasm\"] as const;\n\nexport type CapabilityRuntime = (typeof CAPABILITY_RUNTIMES)[number];\n\nexport type CapabilityManifest = {\n  schemaVersion: typeof CAPABILITY_MANIFEST_SCHEMA_VERSION;\n  runtimes: readonly CapabilityRuntime[];\n  entities: typeof ENTITY_CAPABILITIES;\n};\n\n/**\n * Versioned, runtime-free discovery contract for the deterministic pipeline.\n * The manifest contains no accuracy claims; it describes available labels,\n * activation, provenance sources, and runtime parity only.\n */\nexport const CAPABILITY_MANIFEST = {\n  schemaVersion: CAPABILITY_MANIFEST_SCHEMA_VERSION,\n  runtimes: CAPABILITY_RUNTIMES,\n  entities: ENTITY_CAPABILITIES,\n} as const satisfies CapabilityManifest;\n"],"mappings":";;AAEA,MAAa,qCAAqC;AAElD,MAAa,sBAAsB;CAAC;CAAQ;CAAU;AAAM;;;;;;AAe5D,MAAa,sBAAsB;CACjC,eAAA;CACA,UAAU;CACV,UAAU;AACZ"}

package/dist/constants.d.mts

@@ -1,2 +1,2 @@
-import { a as OPERATOR_TYPES, i as DetectionSource, n as DETECTION_SOURCES, o as OperatorType, r as DETECTOR_PRIORITY, t as DEFAULT_ENTITY_LABELS } from "./constants2.mjs";
-export { DEFAULT_ENTITY_LABELS, DETECTION_SOURCES, DETECTOR_PRIORITY, DetectionSource, OPERATOR_TYPES, OperatorType };
+import { a as DetectionSource, c as ENTITY_SELECTIONS, d as EntitySelection, f as OPERATOR_TYPES, i as DefaultEntityLabel, l as EntityCapability, n as DETECTION_SOURCES, o as ENTITY_CAPABILITIES, p as OperatorType, r as DETECTOR_PRIORITY, s as ENTITY_LABELS, t as DEFAULT_ENTITY_LABELS, u as EntityLabel } from "./constants2.mjs";
+export { DEFAULT_ENTITY_LABELS, DETECTION_SOURCES, DETECTOR_PRIORITY, DefaultEntityLabel, DetectionSource, ENTITY_CAPABILITIES, ENTITY_LABELS, ENTITY_SELECTIONS, EntityCapability, EntityLabel, EntitySelection, OPERATOR_TYPES, OperatorType };

package/dist/constants.mjs

@@ -49,41 +49,171 @@ const DETECTOR_PRIORITY = {
 * Anonymization operator types. Each operator defines
 * how a confirmed entity is replaced in the output.
 */
-const OPERATOR_TYPES = ["replace", "redact"];
+const OPERATOR_TYPES = [
+	"replace",
+	"redact",
+	"keep",
+	"mask"
+];
+const ENTITY_SELECTIONS = {
+	DEFAULT: "default",
+	OPT_IN: "opt-in"
+};
 /**
-* Canonical entity labels used across the pipeline.
-* NER models may use different native labels; the bench
-* NER wrapper maps model output to these canonical names.
+* Canonical entity capabilities exposed by the deterministic native pipeline.
+* `selection` describes whether the default package requests the label; opt-in
+* labels have built-in detection rules but must be requested explicitly.
 *
 * These labels are ephemeral: entities are regenerated on
 * every pipeline run and never persisted to the database.
 * Renaming a label here requires no migration.
 */
-const DEFAULT_ENTITY_LABELS = [
-	"person",
-	"organization",
-	"phone number",
-	"address",
-	"country",
-	"email address",
-	"date",
-	"date of birth",
-	"bank account number",
-	"iban",
-	"tax identification number",
-	"identity card number",
-	"birth number",
-	"national identification number",
-	"social security number",
-	"registration number",
-	"credit card number",
-	"passport number",
-	"crypto",
-	"monetary amount",
-	"land parcel",
-	"misc"
+const ENTITY_CAPABILITIES = [
+	{
+		label: "person",
+		selection: ENTITY_SELECTIONS.DEFAULT,
+		detectionSources: [
+			DETECTION_SOURCES.TRIGGER,
+			DETECTION_SOURCES.REGEX,
+			DETECTION_SOURCES.DENY_LIST,
+			DETECTION_SOURCES.COREFERENCE
+		]
+	},
+	{
+		label: "organization",
+		selection: ENTITY_SELECTIONS.DEFAULT,
+		detectionSources: [
+			DETECTION_SOURCES.TRIGGER,
+			DETECTION_SOURCES.DENY_LIST,
+			DETECTION_SOURCES.LEGAL_FORM,
+			DETECTION_SOURCES.GAZETTEER,
+			DETECTION_SOURCES.COREFERENCE
+		]
+	},
+	{
+		label: "phone number",
+		selection: ENTITY_SELECTIONS.DEFAULT,
+		detectionSources: [DETECTION_SOURCES.REGEX, DETECTION_SOURCES.TRIGGER]
+	},
+	{
+		label: "address",
+		selection: ENTITY_SELECTIONS.DEFAULT,
+		detectionSources: [
+			DETECTION_SOURCES.REGEX,
+			DETECTION_SOURCES.TRIGGER,
+			DETECTION_SOURCES.DENY_LIST
+		]
+	},
+	{
+		label: "country",
+		selection: ENTITY_SELECTIONS.DEFAULT,
+		detectionSources: [DETECTION_SOURCES.COUNTRY]
+	},
+	{
+		label: "email address",
+		selection: ENTITY_SELECTIONS.DEFAULT,
+		detectionSources: [DETECTION_SOURCES.REGEX]
+	},
+	{
+		label: "date",
+		selection: ENTITY_SELECTIONS.DEFAULT,
+		detectionSources: [DETECTION_SOURCES.REGEX, DETECTION_SOURCES.TRIGGER]
+	},
+	{
+		label: "date of birth",
+		selection: ENTITY_SELECTIONS.DEFAULT,
+		detectionSources: [DETECTION_SOURCES.TRIGGER]
+	},
+	{
+		label: "bank account number",
+		selection: ENTITY_SELECTIONS.DEFAULT,
+		detectionSources: [DETECTION_SOURCES.REGEX, DETECTION_SOURCES.TRIGGER]
+	},
+	{
+		label: "iban",
+		selection: ENTITY_SELECTIONS.DEFAULT,
+		detectionSources: [DETECTION_SOURCES.REGEX, DETECTION_SOURCES.TRIGGER]
+	},
+	{
+		label: "tax identification number",
+		selection: ENTITY_SELECTIONS.DEFAULT,
+		detectionSources: [DETECTION_SOURCES.REGEX, DETECTION_SOURCES.TRIGGER]
+	},
+	{
+		label: "identity card number",
+		selection: ENTITY_SELECTIONS.DEFAULT,
+		detectionSources: [DETECTION_SOURCES.REGEX, DETECTION_SOURCES.TRIGGER]
+	},
+	{
+		label: "birth number",
+		selection: ENTITY_SELECTIONS.DEFAULT,
+		detectionSources: [DETECTION_SOURCES.REGEX, DETECTION_SOURCES.TRIGGER]
+	},
+	{
+		label: "national identification number",
+		selection: ENTITY_SELECTIONS.DEFAULT,
+		detectionSources: [DETECTION_SOURCES.REGEX, DETECTION_SOURCES.TRIGGER]
+	},
+	{
+		label: "social security number",
+		selection: ENTITY_SELECTIONS.DEFAULT,
+		detectionSources: [DETECTION_SOURCES.REGEX, DETECTION_SOURCES.TRIGGER]
+	},
+	{
+		label: "registration number",
+		selection: ENTITY_SELECTIONS.DEFAULT,
+		detectionSources: [DETECTION_SOURCES.REGEX, DETECTION_SOURCES.TRIGGER]
+	},
+	{
+		label: "credit card number",
+		selection: ENTITY_SELECTIONS.DEFAULT,
+		detectionSources: [DETECTION_SOURCES.REGEX]
+	},
+	{
+		label: "passport number",
+		selection: ENTITY_SELECTIONS.DEFAULT,
+		detectionSources: [DETECTION_SOURCES.REGEX]
+	},
+	{
+		label: "crypto",
+		selection: ENTITY_SELECTIONS.DEFAULT,
+		detectionSources: [DETECTION_SOURCES.REGEX]
+	},
+	{
+		label: "monetary amount",
+		selection: ENTITY_SELECTIONS.DEFAULT,
+		detectionSources: [DETECTION_SOURCES.REGEX, DETECTION_SOURCES.TRIGGER]
+	},
+	{
+		label: "land parcel",
+		selection: ENTITY_SELECTIONS.DEFAULT,
+		detectionSources: [DETECTION_SOURCES.TRIGGER]
+	},
+	{
+		label: "misc",
+		selection: ENTITY_SELECTIONS.DEFAULT,
+		detectionSources: [DETECTION_SOURCES.REGEX, DETECTION_SOURCES.DENY_LIST]
+	},
+	{
+		label: "ip address",
+		selection: ENTITY_SELECTIONS.OPT_IN,
+		detectionSources: [DETECTION_SOURCES.REGEX]
+	},
+	{
+		label: "mac address",
+		selection: ENTITY_SELECTIONS.OPT_IN,
+		detectionSources: [DETECTION_SOURCES.REGEX]
+	},
+	{
+		label: "url",
+		selection: ENTITY_SELECTIONS.OPT_IN,
+		detectionSources: [DETECTION_SOURCES.REGEX]
+	}
 ];
+const ENTITY_LABELS = ENTITY_CAPABILITIES.map(({ label }) => label);
+const isDefaultEntityCapability = (capability) => capability.selection === ENTITY_SELECTIONS.DEFAULT;
+const DEFAULT_ENTITY_LABELS = ENTITY_CAPABILITIES.filter(isDefaultEntityCapability).map(({ label }) => label);
 //#endregion
-export { DEFAULT_ENTITY_LABELS, DETECTION_SOURCES, DETECTOR_PRIORITY, OPERATOR_TYPES };
+export { DEFAULT_ENTITY_LABELS, DETECTION_SOURCES, DETECTOR_PRIORITY, ENTITY_CAPABILITIES, ENTITY_LABELS, ENTITY_SELECTIONS, OPERATOR_TYPES };
 
 //# sourceMappingURL=constants.mjs.map

package/dist/constants.mjs.map

@@ -1 +1 @@
-{"version":3,"file":"constants.mjs","names":[],"sources":["../src/constants.ts"],"sourcesContent":["/**\n * Runtime-free constants for the anonymization pipeline.\n *\n * This module is the SSR-safe / browser-safe entrypoint:\n * importing it must not pull in `@stll/text-search`,\n * `@stll/anonymize-wasm`, or any other runtime-bearing\n * module. Consumers that only need the static label list,\n * detection-source identifiers, or operator names import\n * from `@stll/anonymize/constants` (or\n * `@stll/anonymize-wasm/constants`) without paying the\n * wasm / regex-set startup cost.\n *\n * `types.ts` re-exports these for back-compat, so existing\n * `import { DEFAULT_ENTITY_LABELS } from \"@stll/anonymize\"`\n * call sites keep working.\n */\n\n/**\n * Source of a detected entity span.\n * Ordered by detection layer in the pipeline.\n */\nexport const DETECTION_SOURCES = {\n  TRIGGER: \"trigger\",\n  REGEX: \"regex\",\n  DENY_LIST: \"deny-list\",\n  LEGAL_FORM: \"legal-form\",\n  GAZETTEER: \"gazetteer\",\n  COUNTRY: \"country\",\n  NER: \"ner\",\n  COREFERENCE: \"coreference\",\n} as const;\n\nexport type DetectionSource =\n  (typeof DETECTION_SOURCES)[keyof typeof DETECTION_SOURCES];\n\n/**\n * Priority levels for detection sources.\n * Higher = more structurally reliable. Used during\n * overlap resolution so deterministic detectors beat\n * probabilistic ones regardless of raw score.\n */\nexport const DETECTOR_PRIORITY = {\n  [DETECTION_SOURCES.GAZETTEER]: 5,\n  [DETECTION_SOURCES.TRIGGER]: 4,\n  [DETECTION_SOURCES.LEGAL_FORM]: 3,\n  [DETECTION_SOURCES.REGEX]: 3,\n  [DETECTION_SOURCES.COUNTRY]: 3,\n  [DETECTION_SOURCES.DENY_LIST]: 2,\n  [DETECTION_SOURCES.COREFERENCE]: 2,\n  [DETECTION_SOURCES.NER]: 1,\n} as const satisfies Record<DetectionSource, number>;\n\n/**\n * Anonymization operator types. Each operator defines\n * how a confirmed entity is replaced in the output.\n */\nexport const OPERATOR_TYPES = [\"replace\", \"redact\"] as const;\n\nexport type OperatorType = (typeof OPERATOR_TYPES)[number];\n\n/**\n * Canonical entity labels used across the pipeline.\n * NER models may use different native labels; the bench\n * NER wrapper maps model output to these canonical names.\n *\n * These labels are ephemeral: entities are regenerated on\n * every pipeline run and never persisted to the database.\n * Renaming a label here requires no migration.\n */\nexport const DEFAULT_ENTITY_LABELS = [\n  \"person\",\n  \"organization\",\n  \"phone number\",\n  \"address\",\n  \"country\",\n  \"email address\",\n  \"date\",\n  \"date of birth\",\n  \"bank account number\",\n  \"iban\",\n  \"tax identification number\",\n  \"identity card number\",\n  \"birth number\",\n  \"national identification number\",\n  \"social security number\",\n  \"registration number\",\n  \"credit card number\",\n  \"passport number\",\n  \"crypto\",\n  \"monetary amount\",\n  \"land parcel\",\n  \"misc\",\n] as const;\n"],"mappings":";;;;;;;;;;;;;;;;;;;;;AAqBA,MAAa,oBAAoB;CAC/B,SAAS;CACT,OAAO;CACP,WAAW;CACX,YAAY;CACZ,WAAW;CACX,SAAS;CACT,KAAK;CACL,aAAa;AACf;;;;;;;AAWA,MAAa,oBAAoB;EAC9B,kBAAkB,YAAY;EAC9B,kBAAkB,UAAU;EAC5B,kBAAkB,aAAa;EAC/B,kBAAkB,QAAQ;EAC1B,kBAAkB,UAAU;EAC5B,kBAAkB,YAAY;EAC9B,kBAAkB,cAAc;EAChC,kBAAkB,MAAM;AAC3B;;;;;AAMA,MAAa,iBAAiB,CAAC,WAAW,QAAQ;;;;;;;;;;AAalD,MAAa,wBAAwB;CACnC;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;AACF"}
+{"version":3,"file":"constants.mjs","names":[],"sources":["../src/constants.ts"],"sourcesContent":["/**\n * Runtime-free constants for the anonymization pipeline.\n *\n * This module is the SSR-safe / browser-safe entrypoint:\n * importing it must not pull in `@stll/text-search`,\n * `@stll/anonymize-wasm`, or any other runtime-bearing\n * module. Consumers that only need the static label list,\n * detection-source identifiers, or operator names import\n * from `@stll/anonymize/constants` (or\n * `@stll/anonymize-wasm/constants`) without paying the\n * wasm / regex-set startup cost.\n *\n * `types.ts` re-exports these for back-compat, so existing\n * `import { DEFAULT_ENTITY_LABELS } from \"@stll/anonymize\"`\n * call sites keep working.\n */\n\n/**\n * Source of a detected entity span.\n * Ordered by detection layer in the pipeline.\n */\nexport const DETECTION_SOURCES = {\n  TRIGGER: \"trigger\",\n  REGEX: \"regex\",\n  DENY_LIST: \"deny-list\",\n  LEGAL_FORM: \"legal-form\",\n  GAZETTEER: \"gazetteer\",\n  COUNTRY: \"country\",\n  NER: \"ner\",\n  COREFERENCE: \"coreference\",\n} as const;\n\nexport type DetectionSource =\n  (typeof DETECTION_SOURCES)[keyof typeof DETECTION_SOURCES];\n\n/**\n * Priority levels for detection sources.\n * Higher = more structurally reliable. Used during\n * overlap resolution so deterministic detectors beat\n * probabilistic ones regardless of raw score.\n */\nexport const DETECTOR_PRIORITY = {\n  [DETECTION_SOURCES.GAZETTEER]: 5,\n  [DETECTION_SOURCES.TRIGGER]: 4,\n  [DETECTION_SOURCES.LEGAL_FORM]: 3,\n  [DETECTION_SOURCES.REGEX]: 3,\n  [DETECTION_SOURCES.COUNTRY]: 3,\n  [DETECTION_SOURCES.DENY_LIST]: 2,\n  [DETECTION_SOURCES.COREFERENCE]: 2,\n  [DETECTION_SOURCES.NER]: 1,\n} as const satisfies Record<DetectionSource, number>;\n\n/**\n * Anonymization operator types. Each operator defines\n * how a confirmed entity is replaced in the output.\n */\nexport const OPERATOR_TYPES = [\"replace\", \"redact\", \"keep\", \"mask\"] as const;\n\nexport type OperatorType = (typeof OPERATOR_TYPES)[number];\n\nexport const ENTITY_SELECTIONS = {\n  DEFAULT: \"default\",\n  OPT_IN: \"opt-in\",\n} as const;\n\nexport type EntitySelection =\n  (typeof ENTITY_SELECTIONS)[keyof typeof ENTITY_SELECTIONS];\n\nexport type EntityCapability = {\n  label: string;\n  selection: EntitySelection;\n  detectionSources: readonly DetectionSource[];\n};\n\n/**\n * Canonical entity capabilities exposed by the deterministic native pipeline.\n * `selection` describes whether the default package requests the label; opt-in\n * labels have built-in detection rules but must be requested explicitly.\n *\n * These labels are ephemeral: entities are regenerated on\n * every pipeline run and never persisted to the database.\n * Renaming a label here requires no migration.\n */\nexport const ENTITY_CAPABILITIES = [\n  {\n    label: \"person\",\n    selection: ENTITY_SELECTIONS.DEFAULT,\n    detectionSources: [\n      DETECTION_SOURCES.TRIGGER,\n      DETECTION_SOURCES.REGEX,\n      DETECTION_SOURCES.DENY_LIST,\n      DETECTION_SOURCES.COREFERENCE,\n    ],\n  },\n  {\n    label: \"organization\",\n    selection: ENTITY_SELECTIONS.DEFAULT,\n    detectionSources: [\n      DETECTION_SOURCES.TRIGGER,\n      DETECTION_SOURCES.DENY_LIST,\n      DETECTION_SOURCES.LEGAL_FORM,\n      DETECTION_SOURCES.GAZETTEER,\n      DETECTION_SOURCES.COREFERENCE,\n    ],\n  },\n  {\n    label: \"phone number\",\n    selection: ENTITY_SELECTIONS.DEFAULT,\n    detectionSources: [DETECTION_SOURCES.REGEX, DETECTION_SOURCES.TRIGGER],\n  },\n  {\n    label: \"address\",\n    selection: ENTITY_SELECTIONS.DEFAULT,\n    detectionSources: [\n      DETECTION_SOURCES.REGEX,\n      DETECTION_SOURCES.TRIGGER,\n      DETECTION_SOURCES.DENY_LIST,\n    ],\n  },\n  {\n    label: \"country\",\n    selection: ENTITY_SELECTIONS.DEFAULT,\n    detectionSources: [DETECTION_SOURCES.COUNTRY],\n  },\n  {\n    label: \"email address\",\n    selection: ENTITY_SELECTIONS.DEFAULT,\n    detectionSources: [DETECTION_SOURCES.REGEX],\n  },\n  {\n    label: \"date\",\n    selection: ENTITY_SELECTIONS.DEFAULT,\n    detectionSources: [DETECTION_SOURCES.REGEX, DETECTION_SOURCES.TRIGGER],\n  },\n  {\n    label: \"date of birth\",\n    selection: ENTITY_SELECTIONS.DEFAULT,\n    detectionSources: [DETECTION_SOURCES.TRIGGER],\n  },\n  {\n    label: \"bank account number\",\n    selection: ENTITY_SELECTIONS.DEFAULT,\n    detectionSources: [DETECTION_SOURCES.REGEX, DETECTION_SOURCES.TRIGGER],\n  },\n  {\n    label: \"iban\",\n    selection: ENTITY_SELECTIONS.DEFAULT,\n    detectionSources: [DETECTION_SOURCES.REGEX, DETECTION_SOURCES.TRIGGER],\n  },\n  {\n    label: \"tax identification number\",\n    selection: ENTITY_SELECTIONS.DEFAULT,\n    detectionSources: [DETECTION_SOURCES.REGEX, DETECTION_SOURCES.TRIGGER],\n  },\n  {\n    label: \"identity card number\",\n    selection: ENTITY_SELECTIONS.DEFAULT,\n    detectionSources: [DETECTION_SOURCES.REGEX, DETECTION_SOURCES.TRIGGER],\n  },\n  {\n    label: \"birth number\",\n    selection: ENTITY_SELECTIONS.DEFAULT,\n    detectionSources: [DETECTION_SOURCES.REGEX, DETECTION_SOURCES.TRIGGER],\n  },\n  {\n    label: \"national identification number\",\n    selection: ENTITY_SELECTIONS.DEFAULT,\n    detectionSources: [DETECTION_SOURCES.REGEX, DETECTION_SOURCES.TRIGGER],\n  },\n  {\n    label: \"social security number\",\n    selection: ENTITY_SELECTIONS.DEFAULT,\n    detectionSources: [DETECTION_SOURCES.REGEX, DETECTION_SOURCES.TRIGGER],\n  },\n  {\n    label: \"registration number\",\n    selection: ENTITY_SELECTIONS.DEFAULT,\n    detectionSources: [DETECTION_SOURCES.REGEX, DETECTION_SOURCES.TRIGGER],\n  },\n  {\n    label: \"credit card number\",\n    selection: ENTITY_SELECTIONS.DEFAULT,\n    detectionSources: [DETECTION_SOURCES.REGEX],\n  },\n  {\n    label: \"passport number\",\n    selection: ENTITY_SELECTIONS.DEFAULT,\n    detectionSources: [DETECTION_SOURCES.REGEX],\n  },\n  {\n    label: \"crypto\",\n    selection: ENTITY_SELECTIONS.DEFAULT,\n    detectionSources: [DETECTION_SOURCES.REGEX],\n  },\n  {\n    label: \"monetary amount\",\n    selection: ENTITY_SELECTIONS.DEFAULT,\n    detectionSources: [DETECTION_SOURCES.REGEX, DETECTION_SOURCES.TRIGGER],\n  },\n  {\n    label: \"land parcel\",\n    selection: ENTITY_SELECTIONS.DEFAULT,\n    detectionSources: [DETECTION_SOURCES.TRIGGER],\n  },\n  {\n    label: \"misc\",\n    selection: ENTITY_SELECTIONS.DEFAULT,\n    detectionSources: [DETECTION_SOURCES.REGEX, DETECTION_SOURCES.DENY_LIST],\n  },\n  {\n    label: \"ip address\",\n    selection: ENTITY_SELECTIONS.OPT_IN,\n    detectionSources: [DETECTION_SOURCES.REGEX],\n  },\n  {\n    label: \"mac address\",\n    selection: ENTITY_SELECTIONS.OPT_IN,\n    detectionSources: [DETECTION_SOURCES.REGEX],\n  },\n  {\n    label: \"url\",\n    selection: ENTITY_SELECTIONS.OPT_IN,\n    detectionSources: [DETECTION_SOURCES.REGEX],\n  },\n] as const satisfies readonly EntityCapability[];\n\ntype KnownEntityCapability = (typeof ENTITY_CAPABILITIES)[number];\n\nexport type EntityLabel = KnownEntityCapability[\"label\"];\n\nexport const ENTITY_LABELS: readonly EntityLabel[] = ENTITY_CAPABILITIES.map(\n  ({ label }) => label,\n);\n\nconst isDefaultEntityCapability = (\n  capability: KnownEntityCapability,\n): capability is Extract<\n  KnownEntityCapability,\n  { selection: typeof ENTITY_SELECTIONS.DEFAULT }\n> => capability.selection === ENTITY_SELECTIONS.DEFAULT;\n\nexport type DefaultEntityLabel = Extract<\n  KnownEntityCapability,\n  { selection: typeof ENTITY_SELECTIONS.DEFAULT }\n>[\"label\"];\n\nexport const DEFAULT_ENTITY_LABELS: readonly DefaultEntityLabel[] =\n  ENTITY_CAPABILITIES.filter(isDefaultEntityCapability).map(\n    ({ label }) => label,\n  );\n"],"mappings":";;;;;;;;;;;;;;;;;;;;;AAqBA,MAAa,oBAAoB;CAC/B,SAAS;CACT,OAAO;CACP,WAAW;CACX,YAAY;CACZ,WAAW;CACX,SAAS;CACT,KAAK;CACL,aAAa;AACf;;;;;;;AAWA,MAAa,oBAAoB;EAC9B,kBAAkB,YAAY;EAC9B,kBAAkB,UAAU;EAC5B,kBAAkB,aAAa;EAC/B,kBAAkB,QAAQ;EAC1B,kBAAkB,UAAU;EAC5B,kBAAkB,YAAY;EAC9B,kBAAkB,cAAc;EAChC,kBAAkB,MAAM;AAC3B;;;;;AAMA,MAAa,iBAAiB;CAAC;CAAW;CAAU;CAAQ;AAAM;AAIlE,MAAa,oBAAoB;CAC/B,SAAS;CACT,QAAQ;AACV;;;;;;;;;;AAoBA,MAAa,sBAAsB;CACjC;EACE,OAAO;EACP,WAAW,kBAAkB;EAC7B,kBAAkB;GAChB,kBAAkB;GAClB,kBAAkB;GAClB,kBAAkB;GAClB,kBAAkB;EACpB;CACF;CACA;EACE,OAAO;EACP,WAAW,kBAAkB;EAC7B,kBAAkB;GAChB,kBAAkB;GAClB,kBAAkB;GAClB,kBAAkB;GAClB,kBAAkB;GAClB,kBAAkB;EACpB;CACF;CACA;EACE,OAAO;EACP,WAAW,kBAAkB;EAC7B,kBAAkB,CAAC,kBAAkB,OAAO,kBAAkB,OAAO;CACvE;CACA;EACE,OAAO;EACP,WAAW,kBAAkB;EAC7B,kBAAkB;GAChB,kBAAkB;GAClB,kBAAkB;GAClB,kBAAkB;EACpB;CACF;CACA;EACE,OAAO;EACP,WAAW,kBAAkB;EAC7B,kBAAkB,CAAC,kBAAkB,OAAO;CAC9C;CACA;EACE,OAAO;EACP,WAAW,kBAAkB;EAC7B,kBAAkB,CAAC,kBAAkB,KAAK;CAC5C;CACA;EACE,OAAO;EACP,WAAW,kBAAkB;EAC7B,kBAAkB,CAAC,kBAAkB,OAAO,kBAAkB,OAAO;CACvE;CACA;EACE,OAAO;EACP,WAAW,kBAAkB;EAC7B,kBAAkB,CAAC,kBAAkB,OAAO;CAC9C;CACA;EACE,OAAO;EACP,WAAW,kBAAkB;EAC7B,kBAAkB,CAAC,kBAAkB,OAAO,kBAAkB,OAAO;CACvE;CACA;EACE,OAAO;EACP,WAAW,kBAAkB;EAC7B,kBAAkB,CAAC,kBAAkB,OAAO,kBAAkB,OAAO;CACvE;CACA;EACE,OAAO;EACP,WAAW,kBAAkB;EAC7B,kBAAkB,CAAC,kBAAkB,OAAO,kBAAkB,OAAO;CACvE;CACA;EACE,OAAO;EACP,WAAW,kBAAkB;EAC7B,kBAAkB,CAAC,kBAAkB,OAAO,kBAAkB,OAAO;CACvE;CACA;EACE,OAAO;EACP,WAAW,kBAAkB;EAC7B,kBAAkB,CAAC,kBAAkB,OAAO,kBAAkB,OAAO;CACvE;CACA;EACE,OAAO;EACP,WAAW,kBAAkB;EAC7B,kBAAkB,CAAC,kBAAkB,OAAO,kBAAkB,OAAO;CACvE;CACA;EACE,OAAO;EACP,WAAW,kBAAkB;EAC7B,kBAAkB,CAAC,kBAAkB,OAAO,kBAAkB,OAAO;CACvE;CACA;EACE,OAAO;EACP,WAAW,kBAAkB;EAC7B,kBAAkB,CAAC,kBAAkB,OAAO,kBAAkB,OAAO;CACvE;CACA;EACE,OAAO;EACP,WAAW,kBAAkB;EAC7B,kBAAkB,CAAC,kBAAkB,KAAK;CAC5C;CACA;EACE,OAAO;EACP,WAAW,kBAAkB;EAC7B,kBAAkB,CAAC,kBAAkB,KAAK;CAC5C;CACA;EACE,OAAO;EACP,WAAW,kBAAkB;EAC7B,kBAAkB,CAAC,kBAAkB,KAAK;CAC5C;CACA;EACE,OAAO;EACP,WAAW,kBAAkB;EAC7B,kBAAkB,CAAC,kBAAkB,OAAO,kBAAkB,OAAO;CACvE;CACA;EACE,OAAO;EACP,WAAW,kBAAkB;EAC7B,kBAAkB,CAAC,kBAAkB,OAAO;CAC9C;CACA;EACE,OAAO;EACP,WAAW,kBAAkB;EAC7B,kBAAkB,CAAC,kBAAkB,OAAO,kBAAkB,SAAS;CACzE;CACA;EACE,OAAO;EACP,WAAW,kBAAkB;EAC7B,kBAAkB,CAAC,kBAAkB,KAAK;CAC5C;CACA;EACE,OAAO;EACP,WAAW,kBAAkB;EAC7B,kBAAkB,CAAC,kBAAkB,KAAK;CAC5C;CACA;EACE,OAAO;EACP,WAAW,kBAAkB;EAC7B,kBAAkB,CAAC,kBAAkB,KAAK;CAC5C;AACF;AAMA,MAAa,gBAAwC,oBAAoB,KACtE,EAAE,YAAY,KACjB;AAEA,MAAM,6BACJ,eAIG,WAAW,cAAc,kBAAkB;AAOhD,MAAa,wBACX,oBAAoB,OAAO,yBAAyB,CAAC,CAAC,KACnD,EAAE,YAAY,KACjB"}