Function storeDocumentationVectors

src/api/utils/vectorStore.ts:663–746

(
  owner: string,
  repo: string,
  content: string,
  fileName: string,
  vectorize?: Vectorize,
)

 * @returns Number of vectors stored
 */
export async function storeDocumentationVectors(
  owner: string,
  repo: string,
  content: string,
  fileName: string,
  vectorize?: Vectorize,
): Promise<number> {
  try {
    console.log(`Storing vectors for ${owner}/${repo}`);

    // Check if Vectorize is available
    if (!vectorize) {
      console.warn("Vectorize binding not available. Skipping vector storage.");
      return 0;
    }

    // Generate namespace for this repository
    const namespace = getRepoNamespace(owner, repo);
    console.log(`Using namespace: ${namespace}`);

    // First delete any existing vectors for this repo's namespace
    try {
      // Query existing vectors in this namespace
      const existingVectors = await vectorize.query(
        await getEmbeddings(""), // Empty query will match based on namespace
        {
          namespace: namespace,
          returnValues: false,
          topK: 100, // Respecting Vectorize's limit of 100 max results
        },
      );

      if (existingVectors?.matches?.length > 0) {
        // Extract IDs of vectors to delete
        const idsToDelete = existingVectors.matches.map((match) => match.id);

        // Delete the vectors by IDs
        await vectorize.deleteByIds(idsToDelete);
        console.log(
          `Deleted ${idsToDelete.length} existing vectors for ${namespace}`,
        );
      } else {
        console.log(`No existing vectors found for ${namespace}`);
      }
    } catch (error) {
      console.log(`Error managing existing vectors: ${error}`);
    }

    // Use specialized documentation chunking for better results
    const chunks = chunkDocumentation(content, fileName);
    console.log(`Created ${chunks.length} chunks for ${owner}/${repo}`);

    // Generate embeddings and upsert vectors
    const vectors = [];

    for (let i = 0; i < chunks.length; i++) {
      const chunk = chunks[i];
      const embedding = await getEmbeddings(chunk);