embeddings:
  - name: local_embeddings
    from: huggingface:huggingface.co/sentence-transformers/all-MiniLM-L6-v2

  - name: openai_embeddings
    from: openai:text-embedding-3-small
    params:
      openai_api_key: ${ secrets:OPENAI_API_KEY }

datasets:
  - from: postgres:documents
    name: docs
    acceleration:
      enabled: true
    columns:
      - name: content
        embeddings:
          - from: local_embeddings
            row_id: id
            chunking:
              enabled: true
              target_chunk_size: 512
              overlap_size: 64

curl -X POST http://localhost:8090/v1/search \
  -H 'Content-Type: application/json' \
  -d '{
    "datasets": ["docs"],
    "text": "cutting edge AI",
    "where": "author=\"jeadie\"",
    "additional_columns": ["title", "state"],
    "limit": 5
  }'

SELECT id, title, score
FROM vector_search(docs, 'cutting edge AI')
WHERE state = 'Open'
ORDER BY score DESC
LIMIT 5;

vector_search(
  table STRING,          -- Dataset name (required)
  query STRING,          -- Search text (required)
  col STRING,            -- Column (optional if single embedding column)
  limit INTEGER,         -- Max results (default: 1000)
  include_score BOOLEAN  -- Include score column (default: TRUE)
) RETURNS TABLE

datasets:
  - from: postgres:articles
    name: articles
    acceleration:
      enabled: true
    columns:
      - name: title
        full_text_search:
          enabled: true
          row_id:
            - id
      - name: body
        full_text_search:
          enabled: true

SELECT id, title, score
FROM text_search(articles, 'search keywords', body)
ORDER BY score DESC
LIMIT 5;

text_search(
  table STRING,          -- Dataset name (required)
  query STRING,          -- Keywords/phrase (required)
  col STRING,            -- Column (required if multiple indexed columns)
  limit INTEGER,         -- Max results (default: 1000)
  include_score BOOLEAN  -- Include score column (default: TRUE)
) RETURNS TABLE

RRF Score = Σ(rank_weight / (k + rank))

SELECT id, title, content, fused_score
FROM rrf(
    vector_search(documents, 'machine learning algorithms'),
    text_search(documents, 'neural networks deep learning', content),
    join_key => 'id'
)
ORDER BY fused_score DESC
LIMIT 5;

SELECT fused_score, title, content
FROM rrf(
    text_search(posts, 'artificial intelligence', rank_weight => 50.0),
    vector_search(posts, 'AI machine learning', rank_weight => 200.0)
)
ORDER BY fused_score DESC
LIMIT 10;

-- Exponential decay (1-hour scale)
SELECT fused_score, title, created_at
FROM rrf(
    text_search(news, 'breaking news'),
    vector_search(news, 'latest updates'),
    time_column => 'created_at',
    recency_decay => 'exponential',
    decay_constant => 0.05,
    decay_scale_secs => 3600
)
ORDER BY fused_score DESC
LIMIT 10;

-- Linear decay (24-hour window)
SELECT fused_score, content
FROM rrf(
    text_search(posts, 'trending'),
    vector_search(posts, 'viral popular'),
    time_column => 'created_at',
    recency_decay => 'linear',
    decay_window_secs => 86400
)
ORDER BY fused_score DESC;

SELECT fused_score, text, langs
FROM rrf(
    vector_search(posts, 'ultimas noticias', rank_weight => 100),
    text_search(posts, 'news'),
    time_column => 'created_at',
    recency_decay => 'exponential',
    decay_constant => 0.05,
    decay_scale_secs => 3600
)
WHERE trim(text) != ''
ORDER BY fused_score DESC LIMIT 15;

datasets:
  - from: postgres:documents
    name: docs
    acceleration:
      enabled: true
    columns:
      - name: content
        embeddings:
          - from: embed_model
            row_id: id
        metadata:
          vectors: non-filterable
      - name: category
        metadata:
          vectors: filterable # enable filtering on this column
    vectors:
      enabled: true
      engine: s3_vectors
      params:
        s3_vectors_bucket: my-bucket
        s3_vectors_region: us-east-1

SELECT * FROM my_table WHERE column LIKE '%substring%';
SELECT * FROM my_table WHERE column = 'exact value';
SELECT * FROM my_table WHERE regexp_like(column, '^spice.*ai$');

spice search "cutting edge AI" --dataset docs --limit 5
spice search --cache-control no-cache "search terms"

version: v1
kind: Spicepod
name: search_app

secrets:
  - from: env
    name: env

embeddings:
  - name: embeddings
    from: huggingface:huggingface.co/sentence-transformers/all-MiniLM-L6-v2

datasets:
  - from: file:articles.parquet
    name: articles
    acceleration:
      enabled: true
      engine: duckdb
    columns:
      - name: title
        full_text_search:
          enabled: true
          row_id:
            - id
      - name: content
        embeddings:
          - from: embeddings
        full_text_search:
          enabled: true

SELECT id, title, content, fused_score
FROM rrf(
    vector_search(articles, 'machine learning best practices'),
    text_search(articles, 'neural network training', content),
    join_key => 'id',
    time_column => 'published_at',
    recency_decay => 'exponential',
    decay_constant => 0.01,
    decay_scale_secs => 86400
)
WHERE fused_score > 0.01
ORDER BY fused_score DESC
LIMIT 10;