requireLiteRtLmAudioEncoder function
Throws EmbedderLoadException unless the .litertlm bundle at
modelPath has an audio encoder, read from its container header (D23:
ModelInfo claims audio on the 270m and the 440m). A header this reader
cannot parse is a load failure, never "no audio".
LiteRtEmbeddingBackend.describe() calls it for declared audio, and the
pass's load() again as a backstop. In describe() is where it matters:
core calls describe() before it closes the embedder it holds, so a refusal
there leaves that embedder working, while one from load() comes after
the close.
Implementation
void requireLiteRtLmAudioEncoder(String modelPath) {
final file = _fileName(modelPath);
Never cannotRead(Object e, String reason) =>
throw liteRtLmUnreadableContainer(file, reason, e);
final List<String> types;
try {
types = readLitertlmModelTypes(modelPath);
} on FileSystemException catch (e) {
cannotRead(e, e.osError?.message ?? e.message);
} on FormatException catch (e) {
cannotRead(e, e.message);
}
if (!litertlmHasAudioEncoder(types)) throw liteRtLmNoAudioEncoder(file);
}