Trino MCP

🚀 Accessing the "bullshit Dataset" with Trino MCP

This project enables users to directly access data from Trino's "bullshit Dataset" via SQL queries through a REST API provided by MCP (Machine Learning Platform). This README will guide you on how to use this interface for data queries and present a simple Python example.

🚀 Quick Start

📦 Installation

To run the Trino MCP service, ensure that Docker is installed and running on your system. Then, you can use the following command to pull the necessary images and start the service:

docker-compose up -d trino-mcp

💻 Usage Examples

Basic Usage

You can execute SQL queries by sending a POST request to the Trino MCP API endpoint. Here is a Python example:

import requests
import json

# Set the API endpoint address (default port 9097 of the Docker container)
api_url = "http://localhost:9097/api/query"

# Define your SQL query
query_data = {
    "query": "SELECT * FROM memory.bullshit.real_bullshit_data LIMIT 5",
    "catalog": "memory",
    "schema": "bullshit"
}

# Send a POST request
response = requests.post(api_url, json=query_data)

# Process the response
if response.status_code == 200:
    results = response.json()
    if results["success"]:
        print(f"Query succeeded, returning {results['results']['row_count']} rows of data")
        for row in results['results']['rows']:
            print(row)
    else:
        print(f"Query failed: {results.get('message', 'Unknown error')}")
else:
    print(f"HTTP request failed, status code: {response.status_code}")

📚 Documentation

Supported API Endpoints

Trino MCP provides the following two main API endpoints:

1. Docker Container API (Port 9097)

import requests
import json

# Set the API endpoint address (default port 9097 of the Docker container)
api_url = "http://localhost:9097/api/query"

# Define your SQL query
query_data = {
    "query": "SELECT * FROM memory.bullshit.real_bullshit_data LIMIT 5",
    "catalog": "memory",
    "schema": "bullshit"
}

# Send a POST request
response = requests.post(api_url, json=query_data)

# Process the response
if response.status_code == 200:
    results = response.json()
    if results["success"]:
        print(f"Query succeeded, returning {results['results']['row_count']} rows of data")
        for row in results['results']['rows']:
            print(row)
    else:
        print(f"Query failed: {results.get('message', 'Unknown error')}")
else:
    print(f"HTTP request failed, status code: {response.status_code}")

2. Standalone Python API (Port 8008)

# The same code, but using a different port
api_url = "http://localhost:8008/query"

These APIs offer the following functions:

• GET /api: Retrieve API documentation and usage examples.
• POST /api/query: Execute SQL queries against Trino.

These APIs eliminate the need for wrapper scripts by directly calling the REST interface, making it easier to integrate into AI systems such as Claude and GPT.

🔧 Troubleshooting

1. API Returns a 503 Service Unavailable Error

If the Docker container API returns a 503 error, follow these steps to troubleshoot:

• Rebuild the container:

  docker-compose stop trino-mcp
  docker-compose rm -f trino-mcp
  docker-compose up -d trino-mcp
  

• Check container logs:

  docker logs trino_mcp_trino-mcp_1
  

• Verify the Trino running status:

  curl -s http://localhost:9095/v1/info | jq
  

2. Port Conflict Issue

If the default port (8008) used by the standalone API is occupied, follow these steps to resolve it:

• Modify the port number in the code: Open the llm_trino_api.py file, find the last line, and change it to the desired port.

  uvicorn.run(app, host="127.0.0.1", port=8009)
  

• Run the modified service:

  python llm_trino_api.py
  

📄 Summary

Through Trino MCP, users can conveniently access the "bullshit Dataset" using SQL queries. This project simplifies the data access process and enhances the efficiency of data integration.