Awesome Open Source
Awesome Open Source


build Documentation

ROAPI automatically spins up read-only APIs for static datasets without requiring you to write a single line of code. It builds on top of Apache Arrow and Datafusion. The core of its design can be boiled down to the following:

  • Query frontends to translate SQL, GraphQL and REST API queries into Datafusion plans.
  • Datafusion for query plan execution.
  • Data layer to load datasets from a variety of sources and formats with automatic schema inference.
  • Response encoding layer to serialize intermediate Arrow record batch into various formats requested by client.

See below for a high level diagram:



Install pre-built binary

pip install roapi-http

Check out Github release page for pre-built binaries for each platform. Pre-built docker images are also available at

Install from source

cargo install --locked --git --branch main --bin roapi-http


Quick start

Spin up APIs for test_data/uk_cities_with_headers.csv and test_data/spacex-launches.json:

roapi-http \
    --table 'uk_cities:test_data/uk_cities_with_headers.csv' \
    --table 'spacex_launches:test_data/spacex-launches.json'

Or using docker:

docker run -t --rm -p 8080:8080 --addr \
    --table 'uk_cities:test_data/uk_cities_with_headers.csv' \
    --table 'spacex_launches:test_data/spacex-launches.json'

Query tables using SQL, GraphQL or REST:

curl -X POST -d "SELECT city, lat, lng FROM uk_cities LIMIT 2" localhost:8080/api/sql
curl -X POST -d "query { uk_cities(limit: 2) {city, lat, lng} }" localhost:8080/api/graphql
curl "localhost:8080/api/tables/uk_cities?columns=city,lat,lng&limit=2"

Get inferred schema for all tables:

curl 'localhost:8080/api/schema'

Config file

You can also configure multiple table sources using YAML config, which supports more advanced format specific table options:

  - name: "blogs"
    uri: "test_data/blogs.parquet"

  - name: "ubuntu_ami"
    uri: "test_data/ubuntu-ami.json"
      format: "json"
      pointer: "/aaData"
      array_encoded: true
        - name: "zone"
          data_type: "Utf8"
        - name: "name"
          data_type: "Utf8"
        - name: "version"
          data_type: "Utf8"
        - name: "arch"
          data_type: "Utf8"
        - name: "instance_type"
          data_type: "Utf8"
        - name: "release"
          data_type: "Utf8"
        - name: "ami_id"
          data_type: "Utf8"
        - name: "aki_id"
          data_type: "Utf8"

  - name: "spacex_launches"
    uri: ""
      format: "json"

  - name: "github_jobs"
    uri: ""

To run serve tables using config file:

roapi-http -c ./roapi.yml

See config documentation for more options including using Google spreadsheet as a table source.

Response serialization

By default, ROAPI encodes responses in JSON format, but you can request different encodings by specifying the ACCEPT header:

curl -X POST \
    -H 'ACCEPT: application/' \
    -d "SELECT launch_library_id FROM spacex_launches WHERE launch_library_id IS NOT NULL" \

REST API query interface

You can query tables through REST API by sending GET requests to /api/tables/{table_name}. Query operators are specified as query params.

REST query frontend currently supports the following query operators:

  • columns
  • sort
  • limit
  • filter

To sort column col1 in ascending order and col2 in descending order, set query param to: sort=col1,-col2.

To find all rows with col1 equal to string 'foo', set query param to: filter[col1]='foo'. You can also do basic comparisons with filters, for example predicate 0 <= col2 < 5 can be expressed as filter[col2]gte=0&filter[col2]lt=5.

GraphQL query interface

To query tables using GraphQL, send the query through POST request to /api/graphql endpoint.

GraphQL query frontend supports the same set of operators supported by REST query frontend. Here how is you can apply various operators in a query:

        filter: {
            col1: false
            col2: { gteq: 4, lt: 1000 }
        sort: [
            { field: "col2", order: "desc" }
            { field: "col3" }
        limit: 100
    ) {

SQL query interface

To query tables using a subset of standard SQL, send the query through POST request to /api/sql endpoint. This is the only query interface that supports table joins.


Query layer:

  • [x] REST API GET
  • [x] GraphQL
  • [x] SQL
  • [ ] join between tables
  • [ ] support filter on nested struct fields
  • [ ] index
  • protocol
    • [ ] gRPC
    • [ ] MySQL
    • [ ] Postgres

Response serialization:

  • [x] JSON application/json
  • [x] Arrow application/
  • [ ] msgpack

Data layer:

  • [x] filesystem
  • [x] HTTP/HTTPS
  • [x] S3
  • [ ] GCS
  • [x] Google spreadsheet
  • [ ] MySQL
  • [ ] Postgres
  • [ ] Airtable
  • Data format


  • [ ] auto gen OpenAPI doc for rest layer
  • [ ] query input type conversion based on table schema
  • [ ] stream arrow encoding response
  • [ ] authentication layer


The core of ROAPI, including query frontends and data layer, lives in the self-contained columnq crate. It takes queries and outputs Arrow record batches. Data sources will also be loaded and stored in memory as Arrow record batches.

The roapi-http crate wraps columnq with a HTTP based API layer. It serializes Arrow record batches produced by columnq into different formats based on client request.

Building ROAPI with simd optimization requires nightly rust toolchain.

Build Docker image

docker build --rm -t .

You can set RELEASE variable to any git reference to build for a specific version.

Get A Weekly Email With Trending Projects For These Topics
No Spam. Unsubscribe easily at any time.
rust (4,736
graphql (1,076
sql (722
rest-api (559
analytics (326
cloud-native (209
s3 (203
datasets (108
query (99
parquet (39
arrow (27