@stll/anonymize-wasm 2.8.2 → 2.9.0

package/README.md

@@ -18,14 +18,18 @@ and local document-provider tools remain Node-only. The binding is instantiated
 lazily on first use.
 
 ```ts
-import { loadDefaultPipeline } from "@stll/anonymize-wasm";
+import { createPipeline } from "@stll/anonymize-wasm";
 
-// Loads the compressed default package bundled in the tarball.
-const pipeline = await loadDefaultPipeline();
+const pipeline = await createPipeline({ language: "en" });
 const { redaction } = pipeline.redactText("A contract signed by Jan Novak.");
 console.log(redaction.redactedText);
 ```
 
+`createPipeline()` accepts one supported language, an exact non-empty
+combination such as `{ language: ["cs", "en"] }`, or `{ language: "all" }`.
+Language-scoped pipelines are assembled and cached from lazy dictionary chunks;
+`"all"` loads the bundled default prepared package.
+
 Bring your own prepared package (bytes, an `ArrayBuffer`, or a URL to fetch):
 
 ```ts
@@ -66,11 +70,16 @@ stllAnonymizeWasm({ packages: "none" });
 stllAnonymizeWasm({ packages: "all" });
 ```
 
-`"default"` selects the full-dictionary package that `loadDefaultPipeline()`
+`"default"` selects the full-dictionary package that `createPipeline()` with
+`language: "all"` or `loadDefaultPipeline()`
 (no argument) loads; a language code selects the scoped package that
 `loadDefaultPipeline("cs")` loads. Requesting a package that is not bundled
 fails the build.
 
+The semantic factory supports `cs`, `de`, `en`, `es`, `fr`, `hu`, `it`, `lv`,
+`pl`, `pt-br`, `ro`, `sk`, and `sv`. The lower-level artifact loader has
+bundled scoped packages for `cs`, `de`, and `en` only.
+
 Interplay with the runtime loaders: the plugin only controls which assets ship,
 not which the code asks for. Calling `loadDefaultPipeline(language)` for a
 package you did not emit resolves to a missing asset URL and rejects with a

package/dist/wasm.d.mts

@@ -813,6 +813,11 @@ type NativeOperatorConfig = {
   redactString?: string;
 };
 declare const CALLER_DETECTION_CONTRACT_VERSION = 2;
+declare const CALLER_DETECTION_MAX_COUNT = 1000000;
+declare const CALLER_DETECTION_TEXT_MAX_BYTES: number;
+declare const CALLER_DETECTION_REQUEST_JSON_MAX_BYTES: number;
+declare const SESSION_CALLER_MAX_INPUTS = 100000;
+declare const SESSION_CALLER_INPUTS_JSON_MAX_BYTES: number;
 declare const EXTERNAL_DETECTION_BATCH_VERSION: 1;
 declare const EXTERNAL_DETECTION_BATCH_MAX_BYTES: number;
 declare const EXTERNAL_DETECTION_DOCUMENT_MAX_BYTES: number;
@@ -1071,6 +1076,11 @@ type PreparedSearch = PreparedNativeAnonymizer;
 declare const PreparedAnonymizer: typeof PreparedNativeAnonymizer;
 type PreparedAnonymizer = PreparedNativeAnonymizer;
 //#endregion
+//#region src/pipeline-language.d.ts
+type SupportedLanguage = "cs" | "de" | "en" | "es" | "fr" | "hu" | "it" | "lv" | "pl" | "pt-br" | "ro" | "sk" | "sv";
+declare const SUPPORTED_LANGUAGES: readonly SupportedLanguage[];
+type PipelineLanguageSelection = SupportedLanguage | readonly [SupportedLanguage, ...SupportedLanguage[]] | "all";
+//#endregion
 //#region src/context.d.ts
 /**
  * Cached state for a single pipeline run (or a sequence of runs sharing the
@@ -1137,6 +1147,9 @@ type WasmBindingOptions = {
  * the underlying wasm module is instantiated a single time and cached. */
 declare const getBinding: () => Promise<NativeAnonymizeBinding>;
 type LoadPreparedPackageOptions = WasmBindingOptions;
+type CreatePipelineOptions = WasmBindingOptions & {
+  language?: PipelineLanguageSelection;
+};
 /** Load a prepared package and return a pipeline ready to redact text. */
 declare const loadPipeline: (source: PreparedPackageSource, options?: LoadPreparedPackageOptions) => Promise<PreparedNativePipeline>;
 /** Load a prepared package and return the lower-level anonymizer. */
@@ -1160,6 +1173,8 @@ declare const loadDefaultPipeline: (language?: string, options?: LoadPreparedPac
  * folding the binding into the key would keep unbounded per-binding entries
  * alive. Injected-binding callers get a fresh pipeline each call. */
 declare const getDefaultPipeline: (language?: string, options?: LoadPreparedPackageOptions) => Promise<PreparedNativePipeline>;
+declare const createPipeline: ({ language, ...bindingOptions }?: CreatePipelineOptions) => Promise<PreparedNativePipeline>;
+declare const create_pipeline: typeof createPipeline;
 declare const redactDefaultText: (fullText: string, operators?: NativeOperatorConfig, language?: string) => Promise<NativeStaticRedactionResult>;
 declare const redactDefaultTextJson: (fullText: string, operators?: NativeOperatorConfig, language?: string) => Promise<string>;
 declare const native_package_version: (options?: WasmBindingOptions) => Promise<string>;
@@ -1179,5 +1194,5 @@ declare const summary_diagnostics_json: (config: NativeSearchPackageInput, fullT
  * fail-closed core used by Node and Python. This does not redact the PDF. */
 declare const inspect_pdf_json: (document: Uint8Array, observationsJson?: string, options?: WasmBindingOptions) => Promise<string>;
 //#endregion
-export { type AnonymisationOperator, CALLER_DETECTION_CONTRACT_VERSION, CAPABILITY_MANIFEST, CAPABILITY_MANIFEST_SCHEMA_VERSION, CAPABILITY_PARITY_PROFILES, CAPABILITY_RUNTIMES, CAPABILITY_SURFACES, type CapabilityManifest, type CapabilityParityProfile, type CapabilityRuntime, type CapabilitySurface, type CapabilitySurfaceId, ConvertExternalDetectionBatchOptions, DEFAULT_ENTITY_LABELS, DETECTION_SOURCES, DETECTOR_PRIORITY, type DefaultEntityLabel, type DetectionSource, type Dictionaries, ENTITY_CAPABILITIES, ENTITY_LABELS, ENTITY_SELECTIONS, EXTERNAL_DETECTION_BATCH_MAX_BYTES, EXTERNAL_DETECTION_BATCH_VERSION, EXTERNAL_DETECTION_DOCUMENT_MAX_BYTES, EXTERNAL_DETECTION_MAX_DETECTIONS, EXTERNAL_DETECTION_MAX_LABEL_MAPPINGS, EXTERNAL_DETECTION_MAX_METADATA_BYTES, EXTERNAL_DETECTION_OFFSET_UNITS, EXTERNAL_DETECTION_PROVIDER_ID_MAX_BYTES, type Entity, type EntityCapability, type EntityLabel, type EntitySelection, ExternalDetectionBatch, ExternalDetectionOffsetUnit, type GazetteerEntry, LoadPreparedPackageOptions, NATIVE_BINDING_PARITY_MEMBERS, NativeAnonymizeBinding, NativeAnonymizerFromConfigOptions, NativeAnonymizerFromPackageOptions, NativeBindingVersionOptions, NativeCallerDetection, NativeCallerRedactionOptions, NativeCreateSessionWithLifecycleOptions, NativeDiagnosticsBatchCallback, NativeNormalizeOptions, NativeOpenSessionArchiveOptions, NativeOperatorConfig, type NativePipelineBuildOptions, type NativePipelineCompatibility, NativePipelineEntity, NativePipelineFromPackageOptions, type NativePipelinePackageOptions, type NativePipelineUnsupportedFeature, NativePreparedRedactionSessionBinding, NativePreparedSearchBinding, type NativePreparedSearchConfig, NativePreparedSessionRedactionPlanBinding, NativeRedactionResult, NativeResultEventCallback, NativeSearchPackageInput, NativeSearchPackageOptions, NativeSessionBlockRedactionPlan, NativeSessionCallerRedactionInput, NativeSessionCallerRedactionPlanOptions, NativeSessionDeletionSummary, NativeSessionLifecycle, NativeSessionMetadata, NativeSessionRedactionAtOptions, NativeSessionStatus, NativeStaticRedactionResult, NativeTextReplacement, OPERATOR_TYPES, type OperatorConfig, type OperatorType, type PipelineConfig, type PipelineContext, PrepareSearchPackageOptions, PreparedAnonymizer, PreparedNativeAnonymizer, PreparedNativePipeline, PreparedNativeRedactionSession, PreparedNativeSessionRedactionPlan, PreparedPackageSource, PreparedSearch, type RedactionResult, type ReviewDecision, type ReviewedEntity, SharedNativeDiagnosticsJsonOptions, SharedNativeDiagnosticsStreamJsonOptions, SharedNativePreparedPackageOptions, SharedNativeRedactTextJsonOptions, SharedNativeRedactTextOptions, SharedNativeRedactTextStreamJsonOptions, SharedNativeSearchPackageOptions, WasmBindingOptions, assertNativeBindingVersion, assertNativePipelineSupported, convert_external_detection_batch, createNativeAnonymizerFromConfig, createNativeAnonymizerFromPackage, createNativePipelineFromConfig, createNativePipelineFromPackage, createPipelineContext, deanonymise, defaultPackageUrl, diagnostics_json, diagnostics_stream_json, encodeNativeSearchConfig, encodeNativeSearchConfigInput, exportRedactionKey, getBinding, getDefaultPipeline, getNativeBindingVersion, getNativePipelineCompatibility, inspect_pdf_json, isNativeAnonymizeBinding, 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 { type AnonymisationOperator, CALLER_DETECTION_CONTRACT_VERSION, CALLER_DETECTION_MAX_COUNT, CALLER_DETECTION_REQUEST_JSON_MAX_BYTES, CALLER_DETECTION_TEXT_MAX_BYTES, CAPABILITY_MANIFEST, CAPABILITY_MANIFEST_SCHEMA_VERSION, CAPABILITY_PARITY_PROFILES, CAPABILITY_RUNTIMES, CAPABILITY_SURFACES, type CapabilityManifest, type CapabilityParityProfile, type CapabilityRuntime, type CapabilitySurface, type CapabilitySurfaceId, ConvertExternalDetectionBatchOptions, CreatePipelineOptions, DEFAULT_ENTITY_LABELS, DETECTION_SOURCES, DETECTOR_PRIORITY, type DefaultEntityLabel, type DetectionSource, type Dictionaries, ENTITY_CAPABILITIES, ENTITY_LABELS, ENTITY_SELECTIONS, EXTERNAL_DETECTION_BATCH_MAX_BYTES, EXTERNAL_DETECTION_BATCH_VERSION, EXTERNAL_DETECTION_DOCUMENT_MAX_BYTES, EXTERNAL_DETECTION_MAX_DETECTIONS, EXTERNAL_DETECTION_MAX_LABEL_MAPPINGS, EXTERNAL_DETECTION_MAX_METADATA_BYTES, EXTERNAL_DETECTION_OFFSET_UNITS, EXTERNAL_DETECTION_PROVIDER_ID_MAX_BYTES, type Entity, type EntityCapability, type EntityLabel, type EntitySelection, ExternalDetectionBatch, ExternalDetectionOffsetUnit, type GazetteerEntry, LoadPreparedPackageOptions, NATIVE_BINDING_PARITY_MEMBERS, NativeAnonymizeBinding, NativeAnonymizerFromConfigOptions, NativeAnonymizerFromPackageOptions, NativeBindingVersionOptions, NativeCallerDetection, NativeCallerRedactionOptions, NativeCreateSessionWithLifecycleOptions, NativeDiagnosticsBatchCallback, NativeNormalizeOptions, NativeOpenSessionArchiveOptions, NativeOperatorConfig, type NativePipelineBuildOptions, type NativePipelineCompatibility, NativePipelineEntity, NativePipelineFromPackageOptions, type NativePipelinePackageOptions, type NativePipelineUnsupportedFeature, NativePreparedRedactionSessionBinding, NativePreparedSearchBinding, type NativePreparedSearchConfig, NativePreparedSessionRedactionPlanBinding, NativeRedactionResult, NativeResultEventCallback, NativeSearchPackageInput, NativeSearchPackageOptions, NativeSessionBlockRedactionPlan, NativeSessionCallerRedactionInput, NativeSessionCallerRedactionPlanOptions, NativeSessionDeletionSummary, NativeSessionLifecycle, NativeSessionMetadata, NativeSessionRedactionAtOptions, NativeSessionStatus, NativeStaticRedactionResult, NativeTextReplacement, OPERATOR_TYPES, type OperatorConfig, type OperatorType, type PipelineConfig, type PipelineContext, type PipelineLanguageSelection, PrepareSearchPackageOptions, PreparedAnonymizer, PreparedNativeAnonymizer, PreparedNativePipeline, PreparedNativeRedactionSession, PreparedNativeSessionRedactionPlan, PreparedPackageSource, PreparedSearch, type RedactionResult, type ReviewDecision, type ReviewedEntity, SESSION_CALLER_INPUTS_JSON_MAX_BYTES, SESSION_CALLER_MAX_INPUTS, SUPPORTED_LANGUAGES, SharedNativeDiagnosticsJsonOptions, SharedNativeDiagnosticsStreamJsonOptions, SharedNativePreparedPackageOptions, SharedNativeRedactTextJsonOptions, SharedNativeRedactTextOptions, SharedNativeRedactTextStreamJsonOptions, SharedNativeSearchPackageOptions, type SupportedLanguage, WasmBindingOptions, assertNativeBindingVersion, assertNativePipelineSupported, convert_external_detection_batch, createNativeAnonymizerFromConfig, createNativeAnonymizerFromPackage, createNativePipelineFromConfig, createNativePipelineFromPackage, createPipeline, createPipelineContext, create_pipeline, deanonymise, defaultPackageUrl, diagnostics_json, diagnostics_stream_json, encodeNativeSearchConfig, encodeNativeSearchConfigInput, exportRedactionKey, getBinding, getDefaultPipeline, getNativeBindingVersion, getNativePipelineCompatibility, inspect_pdf_json, isNativeAnonymizeBinding, 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.d.mts.map