tson-sql 🚀

Advanced TSON to SQL Query Generator & ORM Payload Parser.

Convert TSON (Token-Structured Object Notation) responses from LLMs like Gemini, OpenAI GPT-4o, Claude, and Llama directly into Parameterized SQL Queries (INSERT, UPDATE, WHERE) and ORM payloads (Knex, Prisma, Drizzle).

Cut AI output generation costs & latency by 30% to 50%!

💡 Why tson-sql?

When forcing LLMs to output database payloads or SQL inserts:

• ❌ Expensive Output Tokens: LLM output tokens cost 3x to 4x more than input prompt tokens.
• ❌ Slow Output Speed: Generating full JSON arrays or raw SQL strings with repeating column names doubles generation latency.
• ❌ SQL Injection Risk: Direct raw SQL string output from LLMs can expose your database to injection vulnerabilities.

tson-sql instructs your AI model to reply in compact TSON format, and safely generates parameterized SQL queries ($1, $2, ?) for PostgreSQL, MySQL, SQLite, and MSSQL.

📊 Token Savings & Speed Benchmark

| LLM Output Target | Standard JSON Output | TSON Output | Output Tokens Saved | Speedup | | :--- | :--- | :--- | :--- | :--- | | Extract 10 Rows | ~1,200 tokens | ~610 tokens | 590 tokens | ~2.0x Faster ⚡ | | Extract 50 Rows | ~5,800 tokens | ~2,350 tokens | 3,450 tokens | ~2.2x Faster ⚡ | | SQL Update Payload | ~250 tokens | ~120 tokens | 130 tokens | ~1.9x Faster ⚡ |

📦 Installation

npm install tson-sql
# or
yarn add tson-sql
# or
pnpm add tson-sql

⚡ Quick Start

1. Generate Parameterized PostgreSQL INSERT Query

import { tsonToSqlInsert } from 'tson-sql';
import { Pool } from 'pg';

const pool = new Pool();

// TSON response received from LLM
const llmTsonOutput = `
first_name: Abhi, Sarah
last_name: Asok, Chen
email: [email protected], [email protected]
role: Architect, AI Engineer
`;

// Generate Parameterized Query
const query = tsonToSqlInsert(llmTsonOutput, {
  tableName: 'users',
  dialect: 'postgres',
  returning: true
});

console.log(query.text);
// Output: INSERT INTO "users" ("first_name", "last_name", "email", "role") VALUES ($1, $2, $3, $4), ($5, $6, $7, $8) RETURNING *;

console.log(query.values);
// Output: ['Abhi', 'Asok', '[email protected]', 'Architect', 'Sarah', 'Chen', '[email protected]', 'AI Engineer']

// Execute safely with pg pool!
const result = await pool.query(query.text, query.values);

2. Generate Parameterized MySQL INSERT Query

import { tsonToSqlInsert } from 'tson-sql';

const query = tsonToSqlInsert(llmTsonOutput, {
  tableName: 'users',
  dialect: 'mysql'
});

console.log(query.text);
// Output: INSERT INTO `users` (`first_name`, `last_name`, `email`, `role`) VALUES (?, ?, ?, ?), (?, ?, ?, ?);

3. Generate Parameterized UPDATE Query

import { tsonToSqlUpdate } from 'tson-sql';

const tsonUpdate = `
id: 42
status: completed
paidAt: 2026-08-27T16:50:00.000Z
`;

const updateQuery = tsonToSqlUpdate(tsonUpdate, {
  tableName: 'orders',
  whereColumns: ['id'],
  dialect: 'postgres'
});

console.log(updateQuery.text);
// Output: UPDATE "orders" SET "status" = $1, "paidAt" = $2 WHERE "id" = $3;

4. Generate SQL WHERE Clauses from LLMs

import { tsonToSqlWhere } from 'tson-sql';

const tsonWhere = `
status: active, pending
role: Architect
`;

const where = tsonToSqlWhere(tsonWhere, { dialect: 'postgres' });
console.log(where.text);
// Output: WHERE "status" IN ($1, $2) AND "role" = $3

5. Interactive Terminal Demo

Run the interactive CLI demonstration:

npm run demo

🛠️ API Reference

| Function | Output | Description | | :--- | :--- | :--- | | tsonToSql(tsonStr, options?) | Record<string, any>[] | Parses TSON string into JavaScript SQL row objects. | | tsonToSqlInsert(tsonStr, options?) | { text, values, columns } | Generates parameterized SQL INSERT statements ($1, ?, @p1). | | tsonToSqlUpdate(tsonStr, options?) | { text, values } | Generates parameterized SQL UPDATE statements. | | tsonToSqlWhere(tsonStr, options?) | { text, values } | Generates parameterized SQL WHERE filter clauses. |

👤 Author & Contact

Developed with ❤️ by Abhi Asok.

For business inquiries, collaboration, or support:

• 📧 Email: [email protected]
• 📞 Phone / WhatsApp: +91 9142125724
• 💼 LinkedIn: linkedin.com/in/abhi-asok-09439788
• 🐙 GitHub: github.com/AbhiArvension

📄 License

MIT © Abhi Asok