fromNetwork method

InferenceInstallationBuilder fromNetwork(
  1. String url, {
  2. String? token,
  3. bool? foreground,
  4. String? filename,
  5. int? expectedSize,
})

Set model source from network URL (HTTP/HTTPS)

Parameters:

  • url: The HTTP/HTTPS URL to download from
  • token: Optional authentication token (e.g., HuggingFace token)
  • foreground: Android foreground service mode (shows notification, no timeout)
    • null (default): NO foreground service — this branch configures no notification, and the platform needs one before it starts the service
    • true: always use foreground
    • false: never use foreground
  • filename: the name the file is installed under instead of the URL's file name, with the same extension. isModelInstalled, getModelPath and uninstallModel then take this name.
  • expectedSize: the file's size in bytes once downloaded, checked on every install before the model becomes active, a file installed earlier included. A file of another size is deleted with its record and install throws ModelValidationException; a negative size throws ArgumentError here. With it set, install throws StateError before the download when it would replace a file of this name that an active model of any kind reads (one not installed yet, or installed from another source): uninstall that model first. On the web neither is checked.

Implementation

InferenceInstallationBuilder fromNetwork(
  String url, {
  String? token,
  bool? foreground,
  String? filename,
  int? expectedSize,
}) {
  final expected = FileExpectation.checked(size: expectedSize);
  _clearSource();
  _modelSource = ModelSource.network(
    url,
    authToken: token,
    foreground: foreground,
  );
  _filename = filename;
  _expected = expected;
  return this;
}