version: v2
kind: Spicepod
name: nextjs

datasets:
  - from: github:github.com/vercel/next.js/files/canary
    name: nextjs.docs
    description: Next.js documentation from https://nextjs.org/docs
    params:
      file_format: md
      github_client_id: ${secrets:GITHUB_CLIENT_ID}
      github_private_key: ${secrets:GITHUB_PRIVATE_KEY}
      github_installation_id: ${secrets:GITHUB_INSTALLATION_ID}
      include: "docs/**/*.mdx"
    acceleration:
      enabled: true
      refresh_check_interval: 4h
      refresh_jitter_enabled: true
      refresh_jitter_max: 15m
    columns:
      - name: content
        embeddings:
          - from: openai_embeddings
            row_id:
              - path
            chunking:
              enabled: true
              target_chunk_size: 512
              overlap_size: 128
              trim_whitespace: true

  - from: github:github.com/vercel/next.js/issues
    name: nextjs.issues
    description: Next.js GitHub issues from github.com/vercel/next.js
    time_column: created_at
    params:
      file_format: md
      github_client_id: ${secrets:GITHUB_CLIENT_ID}
      github_private_key: ${secrets:GITHUB_PRIVATE_KEY}
      github_installation_id: ${secrets:GITHUB_INSTALLATION_ID}
    acceleration:
      enabled: true
      refresh_check_interval: 4h
      refresh_mode: full
      refresh_jitter_enabled: true
      refresh_jitter_max: 15m
      refresh_sql: "SELECT * from nextjs.issues LIMIT 10"
    # embeddings:
    #   - column: title
    #     column_pk:
    #       - id
    #     use: openai_embeddings

embeddings:
  - from: openai
    name: openai_embeddings
    params:
      openai_api_key: ${ secrets:OPENAI_API_KEY }