foxhunt/scripts/databento_test.rs

#!/usr/bin/env cargo +nightly -Zscript
//! Databento API Test & Balance Check Script
//! 
//! Purpose: Verify API key, check credit balance, and test minimal data download
//! 
//! Usage:
//!   cargo run --bin databento_test
//! 
//! Requirements:
//!   - DATABENTO_API_KEY environment variable set
//!   - Network connectivity to Databento API
//! 
//! Safety:
//!   - NO automatic downloads
//!   - Displays costs BEFORE any data download
//!   - Requires explicit confirmation

use serde::{Deserialize, Serialize};
use std::error::Error;

#[derive(Debug, Serialize, Deserialize)]
struct AccountInfo {
    balance: f64,
    currency: String,
    organization: String,
}

#[derive(Debug, Serialize, Deserialize)]
struct DatasetInfo {
    dataset: String,
    schema: String,
    cost_per_gb: f64,
    available_symbols: Vec<String>,
}

#[tokio::main]
async fn main() -> Result<(), Box<dyn Error>> {
    println!("🔍 Databento API Test & Balance Checker");
    println!("========================================\n");

    // 1. Check for API key
    let api_key = std::env::var("DATABENTO_API_KEY")
        .map_err(|_| "DATABENTO_API_KEY not set in environment")?;
    
    println!("✅ API Key found: {}...{}", 
             &api_key[..10], 
             &api_key[api_key.len()-10..]);

    // 2. Verify API key and get account info
    println!("\n📊 Checking account balance...");
    let client = reqwest::Client::new();
    
    // Note: Databento API endpoint for account info
    // This is a placeholder - actual endpoint needs to be verified from docs
    let account_url = "https://api.databento.com/v0/account";
    
    let response = client
        .get(account_url)
        .header("Authorization", format!("Bearer {}", api_key))
        .send()
        .await;

    match response {
        Ok(resp) => {
            if resp.status().is_success() {
                println!("✅ API key is valid!");
                
                // Try to parse account info
                if let Ok(text) = resp.text().await {
                    println!("\n📋 Account Info:");
                    println!("{}", text);
                }
            } else {
                let status = resp.status();
                let error_text = resp.text().await.unwrap_or_default();
                println!("❌ API key validation failed:");
                println!("   Status: {}", status);
                println!("   Error: {}", error_text);
                return Err(format!("Invalid API key: {}", status).into());
            }
        }
        Err(e) => {
            println!("❌ Network error connecting to Databento API:");
            println!("   {}", e);
            return Err(e.into());
        }
    }

    // 3. Display pricing information
    println!("\n💰 Dataset Pricing Information:");
    println!("================================");
    println!("\nFrom Databento documentation:");
    println!("• $125 FREE credits for historical data");
    println!("• Pricing is per GB consumed");
    println!("• OHLCV bars (cheapest): ~$0.50-$2.00 per GB");
    println!("• Trades: ~$5-$15 per GB");
    println!("• L2 Order Book (MBP): ~$10-$30 per GB");
    println!("• L3 Order Book (MBO): ~$30-$100 per GB");

    // 4. Recommended minimal test download
    println!("\n🎯 Recommended Minimal Test Download:");
    println!("====================================");
    println!("Symbol:     ES.FUT (E-mini S&P 500 futures)");
    println!("Dataset:    GLBX.MDP3 (CME MDP 3.0)");
    println!("Schema:     ohlcv-1m (1-minute bars)");
    println!("Date Range: 1 day (e.g., 2024-01-02)");
    println!("Est. Size:  ~5-20 MB");
    println!("Est. Cost:  ~$0.01-$0.05 (well under $125 limit)");
    
    println!("\nAlternative (even cheaper):");
    println!("Symbol:     SPY (S&P 500 ETF)");
    println!("Dataset:    XNAS.ITCH (Nasdaq)");
    println!("Schema:     ohlcv-1m");
    println!("Date Range: 1 day");
    println!("Est. Cost:  ~$0.005-$0.02");

    // 5. Cost estimation tool
    println!("\n📐 Cost Estimation Formula:");
    println!("===========================");
    println!("Estimated GB = (symbols × days × data_points × bytes_per_point) / 1GB");
    println!("   OHLCV-1m: ~390 bars/day × 32 bytes = ~12 KB per symbol per day");
    println!("   Trades:   ~10,000 trades/day × 24 bytes = ~240 KB per symbol per day");
    println!("   MBP-1:    ~100,000 updates/day × 48 bytes = ~4.8 MB per symbol per day");

    println!("\n⚠️  IMPORTANT: NO DATA DOWNLOADED YET!");
    println!("   This script only checks your account status.");
    println!("   Use the Databento Python/Rust client to actually download data.");
    println!("   Always check the cost estimate BEFORE downloading!");

    println!("\n📚 Next Steps:");
    println!("==============");
    println!("1. Review the pricing information above");
    println!("2. Use Databento's cost estimator: https://databento.com/pricing");
    println!("3. Start with OHLCV-1m data (cheapest option)");
    println!("4. Download 1-2 days for 1-2 symbols first");
    println!("5. Validate data quality before scaling up");
    println!("6. Monitor your credit balance regularly");

    Ok(())
}