Function refreshIdentifierEmbeddings

src/tools/semantic-identifiers.ts:431–462

(options: { rootDir: string; relativePaths: string[] })

}

export async function refreshIdentifierEmbeddings(options: { rootDir: string; relativePaths: string[] }): Promise<number> {
  const uniquePaths = Array.from(new Set(options.relativePaths.map(normalizeRelativePath).filter(Boolean)));
  if (uniquePaths.length === 0) return 0;

  const cache = await loadEmbeddingCache(options.rootDir, IDENTIFIER_CACHE_FILE);
  const pending: { key: string; hash: string; text: string }[] = [];

  for (const relativePath of uniquePaths) {
    removeFileScopedCacheEntries(cache, relativePath);
    const docs = await buildIdentifierDocsForFile(options.rootDir, relativePath);
    for (const doc of docs) {
      const key = `id:${doc.id}`;
      const hash = hashContent(doc.text);
      pending.push({ key, hash, text: doc.text });
    }
  }

  if (pending.length > 0) {
    const batchSize = getEmbeddingBatchSize();
    for (let i = 0; i < pending.length; i += batchSize) {
      const batch = pending.slice(i, i + batchSize);
      const vectors = await fetchEmbedding(batch.map((entry) => entry.text));
      for (let j = 0; j < batch.length; j++) {
        cache[batch[j].key] = { hash: batch[j].hash, vector: vectors[j] };
      }
    }
  }

  await saveEmbeddingCache(options.rootDir, cache, IDENTIFIER_CACHE_FILE);
  invalidateIdentifierSearchCache();
  return pending.length;
}