generateAll method

@override Future<void> generateAll(Iterable<GenerativeSource> sources, { void onProgress(GenerativeProgress progress)?, })

Produces every source in sources (cache hit or generate) so that mediaFor, audioFor, and metaFor can answer synchronously afterwards.

Runs before the first frame and is idempotent: producing the same source (by GenerativeSource.cacheKey) twice does the work once. Reports coarse progress through onProgress. Throws a FluvieGenerativeException when a source cannot be produced (provider error, offline cache miss, or budget).

Implementation

@override
Future<void> generateAll(
  Iterable<GenerativeSource> sources, {
  void Function(GenerativeProgress progress)? onProgress,
}) async {
  final list = sources.toList();
  var generated = 0;
  for (var i = 0; i < list.length; i++) {
    final source = list[i];
    if (_meta.containsKey(source)) continue;
    final key = source.cacheKey;
    if (_cache.has(source.providerId, key)) {
      _loadFromCache(source, key);
      _report(onProgress, i, list.length, source, 'cached');
      continue;
    }
    if (_config.offline) {
      throw FluvieGenerativeException(
        'Offline: no cached asset for $source. Pre-warm the cache (run once '
        'online and commit .fluvie/generative) or unset FLUVIE_GENERATIVE_OFFLINE.',
      );
    }
    final budget = _config.maxGenerations;
    if (budget != null && generated >= budget) {
      throw FluvieGenerativeException(
        'Generation budget ($budget) exceeded while resolving $source.',
      );
    }
    _report(onProgress, i, list.length, source, 'generating');
    await _storeProduced(source, key, await _runGenerate(source));
    generated++;
    _report(onProgress, i, list.length, source, 'done');
  }
}