/natuql

Allows you to use natural language as a query to a relational database.

Primary LanguageGo

natuql enables you to query databases using natural language queries.

This tool is a proof of concept. Do not use in production.

Features

  • query databases using natural language.
  • supported language: English, Japanese, Chinese, French
  • supported database: MySQL only

Install

go install github.com/dotneet/natuql@latest

Usage

Create index and Query

CAUTION: OpenAPI is used to build the index. This is expensive, consuming roughly 2000-4000 tokens per 5 tables.

export OPENAI_API_KEY=YOUR_API_KEY
export DATABASE_CONNECTION="root:root@tcp(127.0.0.1:3306)/yourdb"
natuql index-create
natuql query "count users over 30 years old"

Output:
Token usage: 1103.000000
SQL: SELECT COUNT(*) FROM users INNER JOIN profiles ON users.id = profiles.user_id WHERE profiles.age > 30;
Result:
4

natuql query "2022年の販売件数を取得して。"

Output:
Token usage: 1808.000000
SQL: SELECT COUNT(*) FROM sales WHERE YEAR(sold_at) = 2022;
Result:
2

Rebuilding Index

If the DB schema is changed, the indexes must be rebuilt.

natuql index-remove
natuql index-create

Configuration

Environment Variables

Name Description Default
OPENAI_API_KEY OpenAI API Key empty
DATABASE_CONNECTION Database connection string empty

Toml

A toml file must be stored in `$HOME/.config/natuql/config.toml'.

# Example
apikey=your_secret_key
dbconn=root:root@(tcp:127.0.0.1)/dbname

# language is used for building index file.
# supported language: English, Japanese, Chinese, French
# Default: Auto detection by environment variables.
language=Japanese

# context-tables-count is the number of tables to be used for query context.
# if set larger value, the query will be more accurate, but it will make the query expensive.
# Default: 8
context-tables-count=8