1version: v1beta1
2kind: Spicepod
3name: react
4
5runtime:
6  dataset_load_parallelism: 1
7
8datasets:
9  - from: github:github.com/reactjs/react.dev/files/main
10    name: react.docs
11    description: React.js documentation and reference, from https://react.dev
12    params:
13      file_format: md
14      github_client_id: ${secrets:GITHUB_CLIENT_ID}
15      github_private_key: ${secrets:GITHUB_PRIVATE_KEY}
16      github_installation_id: ${secrets:GITHUB_INSTALLATION_ID}
17      include: "src/content/learn/**/*.md; src/content/reference/**/*.md"
18    acceleration:
19      enabled: true
20      refresh_check_interval: 4h
21      refresh_jitter_enabled: true
22      refresh_jitter_max: 15m
23    columns:
24      - name: content
25        embeddings:
26          - from: openai_embeddings
27            row_id:
28              - path
29            chunking:
30              enabled: true
31              target_chunk_size: 512
32              overlap_size: 128
33              trim_whitespace: true
34
35  - from: github:github.com/facebook/react/issues
36    name: react.issues
37    description: React.js GitHub issues from github.com/facebook/react
38    time_column: created_at
39    params:
40      file_format: md
41      github_client_id: ${secrets:GITHUB_CLIENT_ID}
42      github_private_key: ${secrets:GITHUB_PRIVATE_KEY}
43      github_installation_id: ${secrets:GITHUB_INSTALLATION_ID}
44    acceleration:
45      enabled: true
46      refresh_sql: "SELECT * FROM react.issues LIMIT 10"
47      refresh_mode: full
48      refresh_check_interval: 4h
49      refresh_jitter_enabled: true
50      refresh_jitter_max: 15m
51    # embeddings:
52    #   - column: title
53    #     column_pk:
54    #       - id
55    #     use: openai_embeddings
56
57embeddings:
58  - from: openai
59    name: openai_embeddings
60    params:
61      openai_api_key: ${ secrets:OPENAI_API_KEY }
62

1version: v1beta1
2kind: Spicepod
3name: react
4
5runtime:
6  dataset_load_parallelism: 1
7
8datasets:
9  - from: github:github.com/reactjs/react.dev/files/main
10    name: react.docs
11    description: React.js documentation and reference, from https://react.dev
12    params:
13      file_format: md
14      github_client_id: ${secrets:GITHUB_CLIENT_ID}
15      github_private_key: ${secrets:GITHUB_PRIVATE_KEY}
16      github_installation_id: ${secrets:GITHUB_INSTALLATION_ID}
17      include: "src/content/learn/**/*.md; src/content/reference/**/*.md"
18    acceleration:
19      enabled: true
20      refresh_check_interval: 4h
21      refresh_jitter_enabled: true
22      refresh_jitter_max: 15m
23    columns:
24      - name: content
25        embeddings:
26          - from: openai_embeddings
27            row_id:
28              - path
29            chunking:
30              enabled: true
31              target_chunk_size: 512
32              overlap_size: 128
33              trim_whitespace: true
34
35  - from: github:github.com/facebook/react/issues
36    name: react.issues
37    description: React.js GitHub issues from github.com/facebook/react
38    time_column: created_at
39    params:
40      file_format: md
41      github_client_id: ${secrets:GITHUB_CLIENT_ID}
42      github_private_key: ${secrets:GITHUB_PRIVATE_KEY}
43      github_installation_id: ${secrets:GITHUB_INSTALLATION_ID}
44    acceleration:
45      enabled: true
46      refresh_sql: "SELECT * FROM react.issues LIMIT 10"
47      refresh_mode: full
48      refresh_check_interval: 4h
49      refresh_jitter_enabled: true
50      refresh_jitter_max: 15m
51    # embeddings:
52    #   - column: title
53    #     column_pk:
54    #       - id
55    #     use: openai_embeddings
56
57embeddings:
58  - from: openai
59    name: openai_embeddings
60    params:
61      openai_api_key: ${ secrets:OPENAI_API_KEY }
62