RunAnywhereStorage class

Static helpers for storage + low-level download + model registration.

Mirrors Swift RunAnywhere+Storage.swift — every public method in the Swift extension has a corresponding entry point here.

Properties

hashCode → int

The hash code for this object.

no setter, inherited

runtimeType → Type

A representation of the runtime type of the object.

no setter, inherited

Methods

noSuchMethod(Invocation invocation) → dynamic

Invoked when a nonexistent method or property is accessed.

inherited

toString() → String

A string representation of this object.

inherited

Operators

operator ==(Object other) → bool

The equality operator.

inherited

Static Methods

cleanTempFiles() → Future<void>

Clear the SDK's Temp directory. Mirrors Swift's cleanTempFiles().

deleteModel(String modelId) → Future<StorageDeleteResult>

Delete one downloaded model end-to-end: unload it if loaded, remove its files through the platform adapter, and clear its registry path so the entry returns to registered-not-downloaded (re-downloadable). Convenience over deleteStorage with the canonical flag set — mirrors Swift RunAnywhere.deleteModel(_:).

deleteStorage(StorageDeleteRequest request) → Future<StorageDeleteResult>

Execute a generated-proto storage delete request.

importModel(ModelImportRequest request) → Future<ModelImportResult>

Import a stable, platform-normalized local model path into the generated registry. Also the public local-import entry point for file-picker / bookmark flows after the platform has handled sandbox access.

registerArchiveModel({required String archiveUrl, required ArchiveStructure structure, String? id, required String name, required InferenceFramework framework, ModelCategory modality = ModelCategory.MODEL_CATEGORY_LANGUAGE, ArchiveType? archiveType, int? memoryRequirement, bool supportsThinking = false, bool supportsLora = false}) → Future<ModelInfo>

Register an archive-packaged model (tar.gz / tar.bz2 / tar.xz / zip) where the caller needs to specify the on-disk layout (ArchiveStructure.ARCHIVE_STRUCTURE_DIRECTORY_BASED, ArchiveStructure.ARCHIVE_STRUCTURE_NESTED_DIRECTORY, etc.) the URL-form registerModel cannot infer.

registerModel({String? id, required String name, required String url, required InferenceFramework framework, ModelCategory modality = ModelCategory.MODEL_CATEGORY_LANGUAGE, ModelArtifactType? artifactType, int? memoryRequirement, bool supportsThinking = false, bool supportsLora = false}) → Future<ModelInfo>

Register a remote model with the in-memory model registry from a download URL. Delegates the full build-and-save flow to the commons single-call factory rac_register_model_from_url_proto, which derives the canonical id from the URL (rac_model_generate_id), defaults format/framework/category/context-length, infers the artifact type from the URL extension (archive vs single-file), overlays the caller-supplied capability fields, and persists through the registry save path.

registerMultiFileModel({required List<ModelFileDescriptor> files, required String id, required String name, required InferenceFramework framework, ModelCategory modality = ModelCategory.MODEL_CATEGORY_LANGUAGE, int? memoryRequirement, int? contextLength, bool supportsThinking = false, ModelSource source = ModelSource.MODEL_SOURCE_REMOTE}) → Future<ModelInfo>

Register a multi-file model (e.g., VLMs with a separate mmproj, MiniLM embedding with vocab.txt) through the canonical commons factory (rac_register_multi_file_model_proto) — no URL is involved at the model level because each ModelFileDescriptor carries its own URL.