@stll/anonymize-wasm 2.1.0 → 2.3.0

package/dist/wasm.mjs

@@ -1,5 +1,5 @@
 import { DEFAULT_ENTITY_LABELS, DETECTION_SOURCES, DETECTOR_PRIORITY, ENTITY_CAPABILITIES, ENTITY_LABELS, ENTITY_SELECTIONS, OPERATOR_TYPES } from "./constants.mjs";
-import { CAPABILITY_MANIFEST, CAPABILITY_MANIFEST_SCHEMA_VERSION, CAPABILITY_RUNTIMES } from "./capabilities.mjs";
+import { CAPABILITY_MANIFEST, CAPABILITY_MANIFEST_SCHEMA_VERSION, CAPABILITY_PARITY_PROFILES, CAPABILITY_RUNTIMES, CAPABILITY_SURFACES } from "./capabilities.mjs";
 //#region src/native.ts
 const CALLER_DETECTION_CONTRACT_VERSION = 2;
 const callerDetectionRequestJson = (detections) => JSON.stringify({
@@ -706,12 +706,31 @@ const pipelineConfigKey = (config, gazetteerEntries) => {
 	const gazFingerprint = config.enableGazetteer && gazetteerEntries.length > 0 ? gazetteerEntries.map((entry) => `${entry.id}:${entry.canonical}:${entry.label}:${[...entry.variants].sort().join(",")}`).toSorted().join(";") : "";
 	return `${config.enableDenyList}:${config.enableTriggerPhrases}:${legalFormsEnabled}:${config.enableNameCorpus}:${contentLanguageFingerprint(config)}:${config.nameCorpusLanguages?.toSorted().join(",") ?? ""}:${config.enableRegex}:${config.threshold}:${config.enableConfidenceBoost}:${config.enableHotwordRules === true}:${config.enableCoreference === true}:${config.enableZoneClassification === true}:${config.labels.toSorted().join(",")}:${config.denyListCountries?.toSorted().join(",") ?? ""}:${config.denyListRegions?.toSorted().join(",") ?? ""}:${config.denyListExcludeCategories?.toSorted().join(",") ?? ""}:${customDenyFingerprint}:${customRegexFingerprint}:${config.enableGazetteer}:${gazFingerprint}:${config.enableCountries !== false}`;
 };
-//#endregion
-//#region src/native-pipeline.ts
 const sharedPackageByDictionaries = /* @__PURE__ */ new WeakMap();
 const sharedPackageWithoutDictionaries = /* @__PURE__ */ new Map();
 const dictionaryCacheIds = /* @__PURE__ */ new WeakMap();
 let nextDictionaryCacheId = 0;
+/** Record `key` as most-recently-used in `cache`, evicting the
+* least-recently-used entry first once the cache is at capacity. A `Map`'s
+* insertion order doubles as recency order here: touching an existing key
+* deletes then re-sets it to move it to the end, and eviction drops the
+* first (oldest) key.
+*
+* Evicting a still-in-flight build only drops the cache's reference to its
+* promise; the caller that started the build (and any concurrent caller that
+* already read the promise before eviction) still resolves it correctly via
+* the guarded `sharedCache.get(key) === promise` checks in
+* `getCachedNativePipelinePackage`. A later caller for the same key just
+* misses the dedupe and starts a fresh build — bounded memory takes priority
+* over perfect dedupe under cache pressure. */
+const touchSharedPackageCacheEntry = (cache, key, value) => {
+	cache.delete(key);
+	if (cache.size >= 32) {
+		const oldestKey = cache.keys().next().value;
+		if (oldestKey !== void 0) cache.delete(oldestKey);
+	}
+	cache.set(key, value);
+};
 const dictionaryCacheKey = (dictionaries) => {
 	if (dictionaries === void 0) return "none";
 	const existing = dictionaryCacheIds.get(dictionaries);
@@ -730,7 +749,7 @@ const sharedPackageCacheFor = (dictionaries) => {
 };
 const getNativePipelineCompatibility = (config) => {
 	const unsupportedFeatures = [];
-	if (config.enableNer) unsupportedFeatures.push("enableNer");
+	if ("enableNer" in config && Boolean(config.enableNer)) unsupportedFeatures.push("enableNer");
 	if (unsupportedFeatures.length === 0) return { status: "supported" };
 	return {
 		status: "unsupported",
@@ -749,11 +768,8 @@ const encoder = new TextEncoder();
 * the separate bundle preferentially, and keeping the (large) dictionaries out
 * of the config JSON avoids serializing them twice.
 */
-const toAssembleInputs = ({ dictionaries, enableNer = false, ...config }, gazetteerEntries) => ({
-	pipelineConfigJson: encoder.encode(JSON.stringify({
-		...config,
-		enableNer
-	})),
+const toAssembleInputs = ({ dictionaries, ...config }, gazetteerEntries) => ({
+	pipelineConfigJson: encoder.encode(JSON.stringify(config)),
 	dictionariesJson: dictionaries === void 0 ? void 0 : encoder.encode(JSON.stringify(dictionaries)),
 	gazetteerJson: gazetteerEntries.length === 0 ? void 0 : encoder.encode(JSON.stringify(gazetteerEntries))
 });
@@ -807,6 +823,7 @@ const getCachedNativePipelinePackage = async ({ binding, config, gazetteerEntrie
 	const sharedCache = sharedPackageCacheFor(scopedConfig.dictionaries);
 	const shared = sharedCache.get(key);
 	if (shared !== void 0) {
+		touchSharedPackageCacheEntry(sharedCache, key, shared);
 		const packageBytes = await shared;
 		ctx.nativePipelinePackage = packageBytes;
 		ctx.nativePipelinePackageKey = key;
@@ -822,7 +839,7 @@ const getCachedNativePipelinePackage = async ({ binding, config, gazetteerEntrie
 		compressed
 	});
 	ctx.nativePipelinePackagePromise = promise;
-	sharedCache.set(key, promise);
+	touchSharedPackageCacheEntry(sharedCache, key, promise);
 	let packageBytes;
 	try {
 		packageBytes = await promise;
@@ -856,6 +873,8 @@ const NODE_FS_MODULE = "node:fs/promises";
 const NATIVE_ASSET_DIR = "native";
 const DEFAULT_PACKAGE_FILE = "native-pipeline.stlanonpkg";
 const LANGUAGE_PATTERN = /^[a-z0-9]+(?:-[a-z0-9]+)*$/u;
+const DEFAULT_PIPELINE_CACHE_KEY = "<default>";
+const DEFAULT_PIPELINE_CACHE_MAX_ENTRIES = 32;
 let bindingPromise;
 const defaultPipelineCache = /* @__PURE__ */ new Map();
 /** Instantiate (once) and return the wasm binding. Safe to call repeatedly:
@@ -933,6 +952,26 @@ const loadDefaultPipeline = async (language, options) => {
 		return loadPipeline(defaultPackageUrl(baseLanguage), options);
 	}
 };
+/** Normalized cache key for {@link defaultPipelineCache}: `undefined` maps to
+* the bundled-default sentinel, everything else goes through the same
+* {@link normalizeLanguage} helper `loadDefaultPipeline`/`defaultPackageUrl`
+* already validate against, so aliases that differ only by case or whitespace
+* (e.g. `"EN"` vs `"en"`) share one cached pipeline instead of each minting
+* their own. */
+const defaultPipelineCacheKey = (language) => language === void 0 ? DEFAULT_PIPELINE_CACHE_KEY : normalizeLanguage(language);
+/** Record `key` as most-recently-used in {@link defaultPipelineCache},
+* evicting the least-recently-used entry first once the cache is at
+* capacity. A `Map`'s insertion order doubles as recency order here: touching
+* an existing key deletes then re-sets it to move it to the end, and
+* eviction drops the first (oldest) key. */
+const touchDefaultPipelineCacheEntry = (key, pipeline) => {
+	defaultPipelineCache.delete(key);
+	if (defaultPipelineCache.size >= DEFAULT_PIPELINE_CACHE_MAX_ENTRIES) {
+		const oldestKey = defaultPipelineCache.keys().next().value;
+		if (oldestKey !== void 0) defaultPipelineCache.delete(oldestKey);
+	}
+	defaultPipelineCache.set(key, pipeline);
+};
 /** Cached variant of {@link loadDefaultPipeline}: the default pipeline for a
 * given language is fetched and prepared once, then reused.
 *
@@ -943,14 +982,22 @@ const loadDefaultPipeline = async (language, options) => {
 * alive. Injected-binding callers get a fresh pipeline each call. */
 const getDefaultPipeline = (language, options) => {
 	if (options?.binding) return loadDefaultPipeline(language, options);
-	const key = language ?? "<default>";
+	let key;
+	try {
+		key = defaultPipelineCacheKey(language);
+	} catch (error) {
+		return Promise.reject(error);
+	}
 	const cached = defaultPipelineCache.get(key);
-	if (cached !== void 0) return cached;
+	if (cached !== void 0) {
+		touchDefaultPipelineCacheEntry(key, cached);
+		return cached;
+	}
 	const pipeline = loadDefaultPipeline(language).catch((error) => {
 		defaultPipelineCache.delete(key);
 		throw error;
 	});
-	defaultPipelineCache.set(key, pipeline);
+	touchDefaultPipelineCacheEntry(key, pipeline);
 	return pipeline;
 };
 const redactDefaultText = async (fullText, operators, language) => (await getDefaultPipeline(language)).redactText(fullText, operators);
@@ -1030,6 +1077,6 @@ const normalizeLanguage = (language) => {
 	return normalized;
 };
 //#endregion
-export { CALLER_DETECTION_CONTRACT_VERSION, CAPABILITY_MANIFEST, CAPABILITY_MANIFEST_SCHEMA_VERSION, CAPABILITY_RUNTIMES, DEFAULT_ENTITY_LABELS, DETECTION_SOURCES, DETECTOR_PRIORITY, ENTITY_CAPABILITIES, ENTITY_LABELS, ENTITY_SELECTIONS, OPERATOR_TYPES, PreparedAnonymizer, PreparedNativeAnonymizer, PreparedNativePipeline, PreparedNativeRedactionSession, PreparedNativeSessionRedactionPlan, PreparedSearch, assertNativeBindingVersion, assertNativePipelineSupported, createNativeAnonymizerFromConfig, createNativeAnonymizerFromPackage, createNativePipelineFromConfig, createNativePipelineFromPackage, createPipelineContext, deanonymise, defaultPackageUrl, diagnostics_json, diagnostics_stream_json, encodeNativeSearchConfig, encodeNativeSearchConfigInput, exportRedactionKey, getBinding, getDefaultPipeline, getNativeBindingVersion, getNativePipelineCompatibility, loadDefaultPipeline, loadPipeline, load_prepared_package, native_package_version, normalize_for_search, prepareNativePipelineConfig, prepareNativePipelinePackage, prepareNativeSearchPackage, prepare_search_package, redactDefaultText, redactDefaultTextJson, redact_text, redact_text_json, redact_text_stream_json, summary_diagnostics_json };
+export { CALLER_DETECTION_CONTRACT_VERSION, CAPABILITY_MANIFEST, CAPABILITY_MANIFEST_SCHEMA_VERSION, CAPABILITY_PARITY_PROFILES, CAPABILITY_RUNTIMES, CAPABILITY_SURFACES, DEFAULT_ENTITY_LABELS, DETECTION_SOURCES, DETECTOR_PRIORITY, ENTITY_CAPABILITIES, ENTITY_LABELS, ENTITY_SELECTIONS, OPERATOR_TYPES, PreparedAnonymizer, PreparedNativeAnonymizer, PreparedNativePipeline, PreparedNativeRedactionSession, PreparedNativeSessionRedactionPlan, PreparedSearch, assertNativeBindingVersion, assertNativePipelineSupported, createNativeAnonymizerFromConfig, createNativeAnonymizerFromPackage, createNativePipelineFromConfig, createNativePipelineFromPackage, createPipelineContext, deanonymise, defaultPackageUrl, diagnostics_json, diagnostics_stream_json, encodeNativeSearchConfig, encodeNativeSearchConfigInput, exportRedactionKey, getBinding, getDefaultPipeline, getNativeBindingVersion, getNativePipelineCompatibility, loadDefaultPipeline, loadPipeline, load_prepared_package, native_package_version, normalize_for_search, prepareNativePipelineConfig, prepareNativePipelinePackage, prepareNativeSearchPackage, prepare_search_package, redactDefaultText, redactDefaultTextJson, redact_text, redact_text_json, redact_text_stream_json, summary_diagnostics_json };
 
 //# sourceMappingURL=wasm.mjs.map