Files
foxhunt/crates/data/src/lib.rs
jgrusewski cc4e0c5a2d refactor: remove trading_engine dep from ml and risk crates
Re-export HardwareTimestamp through data crate instead of ml/risk
depending directly on trading_engine. Reduces coupling between
the ML pipeline and the trading engine.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-05 23:19:38 +01:00

378 lines
17 KiB
Rust

#![allow(unused_extern_crates)]
#![allow(unused_crate_dependencies)]
#![allow(dead_code)] // Data providers and processors defined but not yet integrated from services
#![allow(unsafe_code)] // Intentional unsafe for high-performance data processing
#![allow(missing_docs)] // Internal implementation details don't require documentation
#![allow(missing_debug_implementations)] // Not all types need Debug
#![allow(clippy::float_arithmetic)]
// Data processing requires float arithmetic
// Performance-critical HFT code - pedantic lints that would require major refactoring
#![allow(clippy::str_to_string)] // High-performance string conversions
#![allow(clippy::as_conversions)] // Low-level data parsing requires type conversions
#![allow(clippy::default_numeric_fallback)] // Protocol parsing with numeric literals
#![allow(clippy::arithmetic_side_effects)] // Market data calculations are performance-critical
#![allow(clippy::indexing_slicing)] // Protocol buffer parsing requires indexing
#![allow(clippy::doc_markdown)] // API names in docs (ICMarkets, InteractiveBrokers, etc.)
#![allow(clippy::module_name_repetitions)] // Broker/provider naming conventions
#![allow(clippy::map_err_ignore)] // Error context not always needed in data pipeline
#![allow(clippy::result_large_err)] // Error types sized for comprehensive error reporting
#![allow(clippy::missing_const_for_fn)] // Runtime dynamic behavior in many functions
#![deny(clippy::unwrap_used, clippy::expect_used)]
#![cfg_attr(test, allow(clippy::unwrap_used, clippy::expect_used))]
#![allow(clippy::clone_on_ref_ptr)] // Arc/Rc clones required for concurrent access
#![allow(clippy::integer_division)] // Price calculations require division
#![allow(clippy::wildcard_enum_match_arm)] // Future-proof protocol extensions
#![allow(clippy::similar_names)] // Domain-specific names (side/size, etc.)
#![allow(clippy::else_if_without_else)] // State machine logic doesn't always need else
#![allow(clippy::single_char_lifetime_names)] // Standard Rust lifetime conventions
#![allow(clippy::useless_conversion)] // Type system conversions for API compatibility
#![allow(clippy::used_underscore_binding)] // Protocol field parsing with underscore prefix
#![allow(clippy::if_then_some_else_none)] // Conditional logic readability
#![allow(clippy::cognitive_complexity)] // Complex protocol state machines
#![allow(clippy::shadow_reuse)] // Variable shadowing in nested scopes
#![allow(clippy::let_underscore_must_use)] // Intentionally discarding results in data pipeline
#![allow(clippy::unnecessary_wraps)] // Consistent Result types across trait implementations
#![allow(clippy::wildcard_imports)] // Internal module imports
#![allow(clippy::shadow_unrelated)] // Shadowing for readability
#![allow(clippy::non_ascii_literal)] // Currency symbols and market data
#![allow(clippy::unwrap_or_default)] // Performance optimizations
#![allow(clippy::unwrap_in_result)] // Verified safe unwraps in error paths
#![allow(clippy::string_slice)] // Protocol parsing requires string slicing
#![allow(clippy::redundant_closure)] // Explicit closures for clarity
#![allow(clippy::new_without_default)] // Builder pattern implementations
#![allow(clippy::upper_case_acronyms)] // API/FIX protocol acronyms
#![allow(clippy::type_complexity)] // Complex generic types in broker traits
#![allow(clippy::same_name_method)] // Trait and inherent method coexistence
#![allow(clippy::string_to_string)] // String conversions in protocol parsing
#![allow(clippy::rc_buffer)] // Rc<Vec> for shared buffer access
#![allow(clippy::infinite_loop)] // Event loop implementations
#![allow(clippy::unnecessary_cast)] // Explicit casts for protocol compatibility
#![allow(clippy::too_many_lines)] // Complex protocol handlers
#![allow(clippy::ptr_arg)] // Vec arguments for builder patterns
#![allow(clippy::needless_range_loop)] // Manual indexing for performance
#![allow(clippy::clone_on_copy)] // Explicit clone calls for clarity
#![allow(clippy::unnecessary_sort_by)] // Custom comparison logic
#![allow(clippy::unnecessary_map_or)] // Explicit mapping for readability
#![allow(clippy::manual_range_contains)] // Explicit range checks
#![allow(clippy::undocumented_unsafe_blocks)] // Unsafe blocks documented inline
#![allow(clippy::unchecked_duration_subtraction)] // Duration arithmetic validated by logic
#![allow(clippy::single_match_else)] // Explicit match for readability
#![allow(clippy::shadow_same)] // Shadowing for progressive refinement
#![allow(clippy::reserve_after_initialization)] // Dynamic capacity management
#![allow(clippy::redundant_clone)] // Clone required for ownership transfer
#![allow(clippy::modulo_arithmetic)] // Modulo for circular buffer indexing
#![allow(clippy::manual_ok_err)] // Explicit Ok/Err construction
#![allow(clippy::manual_clamp)] // Custom clamping logic
#![allow(clippy::len_zero)] // Explicit len() == 0 checks
#![allow(clippy::into_iter_on_ref)] // Explicit iterator creation
#![allow(clippy::impl_trait_in_params)] // Trait object parameters
#![allow(clippy::explicit_counter_loop)] // Manual loop counters for control
#![allow(clippy::doc_lazy_continuation)] // Documentation formatting
#![allow(clippy::await_holding_lock)] // Lock scope carefully managed
#![allow(clippy::assign_op_pattern)] // Explicit assignment patterns
#![allow(clippy::manual_let_else)] // if-let pattern preferred in streaming error paths
//! # Foxhunt Data Module
//!
//! High-performance market data ingestion and broker integration module for HFT systems,
//! including Interactive Brokers, ICMarkets, and other data providers.
//!
//! ## Features
//!
//! - **High-Frequency Data Ingestion**: Sub-millisecond market data processing
//! - **Multiple Data Providers**: Interactive Brokers, ICMarkets
//! - **Real-time WebSocket Streams**: Level 1 and Level 2 market data
//! - **FIX Protocol Support**: Complete FIX 4.4 implementation for broker connectivity
//! - **Order Management**: Order lifecycle management with execution reports
//! - **Resilient Connectivity**: Automatic reconnection with exponential backoff
//! - **Performance Optimized**: Zero-copy parsing and lock-free data structures
//!
//! ## Architecture
//!
//! ```text
//! ┌─────────────────┐ ┌─────────────────┐ ┌─────────────────┐
//! │ Market Data │ │ Broker Conn │ │ Data Store │
//! │ Providers │───▶│ Management │───▶│ & Cache │
//! └─────────────────┘ └─────────────────┘ └─────────────────┘
//! │ │ │
//! ▼ ▼ ▼
//! ┌─────────────────┐ ┌─────────────────┐ ┌─────────────────┐
//! │ WebSocket │ │ FIX Protocol │ │ Event Stream │
//! │ Streams │ │ Engine │ │ Publisher │
//! └─────────────────┘ └─────────────────┘ └─────────────────┘
//! ```
//!
//! ## Core Components
//!
//! ### Market Data Providers
//! - **Databento**: Real-time and historical market data
//! - **ICMarkets**: FIX protocol integration for forex and CFDs
//! - **Interactive Brokers**: TWS API integration for equities and derivatives
//!
//! ### FIX Protocol Engine
//! - Complete FIX 4.4 implementation
//! - Session management with heartbeat
//! - Order lifecycle management
//! - Execution report processing
//! - Message parsing with zero-copy optimization
//!
//! ### Performance Features
//! - Lock-free data structures for high-throughput scenarios
//! - Memory pools for allocation efficiency
//! - SIMD-optimized message parsing
//! - CPU affinity for deterministic latency
//!
//! ## Usage
//!
//! ```rust
//! // NOTE: Broker clients moved to core module in monolithic architecture
//! use core::brokers::{
//! ICMarketsClient, ICMarketsConfig, FixOrder
//! };
//! use core::prelude::{OrderSide, OrderType, ExecutionReport};
//! // REMOVED: Polygon client - replaced with Databento
//! use data::types::{MarketDataEvent, Subscription};
//!
//! #[tokio::main]
//! async fn main() -> anyhow::Result<()> {
//! // Initialize ICMarkets FIX client
//! let config = ICMarketsConfig {
//! host: "fix-demo.icmarkets.com".to_string(),
//! port: 9880,
//! sender_comp_id: "CLIENT".to_string(),
//! target_comp_id: "ICMARKETS".to_string(),
//! username: "your_username".to_string(),
//! password: "your_password".to_string(),
//! heartbeat_interval: 30,
//! ..Default::default()
//! };
//!
//! let mut client = ICMarketsClient::new(config);
//! client.connect().await?;
//!
//! // Submit an order
//! let order = FixOrder {
//! cl_ord_id: "ORDER001".to_string(),
//! symbol: "EURUSD".to_string(),
//! side: OrderSide::Buy,
//! ord_type: OrderType::Market,
//! order_qty: 10000.0,
//! price: None,
//! stop_px: None,
//! time_in_force: 1, // GTC
//! account: None,
//! };
//!
//! let order_id = client.submit_order(order).await?;
//! println!("Order submitted: {}", order_id);
//!
//! Ok(())
//! }
//! ```
//!
//! ## Configuration
//!
//! The module supports environment-based configuration:
//!
//! ```toml
//! [data]
//! // REMOVED: polygon_api_key configuration
//! ib_host = "${IB_TWS_HOST:-127.0.0.1}"
//! ib_port = "${IB_TWS_PORT:-7497}"
//!
//! # ICMarkets configuration
//! icmarkets = {
//! host = "${ICMARKETS_HOST:-fix-demo.icmarkets.com}",
//! port = "${ICMARKETS_PORT:-9880}",
//! username = "${ICMARKETS_USERNAME}",
//! password = "${ICMARKETS_PASSWORD}"
//! }
//! ```
#![warn(rust_2018_idioms, unused_qualifications, clippy::large_enum_variant)]
// Note: cognitive_complexity and type_complexity are allowed at crate-level for HFT protocol code
// dead_code lint intentionally enabled — compiler should flag unused items
// Note: Deprecated fields are kept for backwards compatibility but usage updated
#![allow(unexpected_cfgs)] // Allow unexpected cfg attributes
#![allow(private_bounds)] // Allow private type bounds
#![allow(unreachable_pub)] // Allow unreachable public items
// Note: unwrap/expect/panic lints are handled at crate-level above for HFT performance code
pub mod brokers;
// pub mod config; // Temporarily disabled - complex fixes needed
pub mod dbn_uploader; // DBN file uploader to MinIO
pub mod error;
pub mod features; // Feature engineering for ML models
pub mod parquet_persistence; // Parquet market data persistence for replay
pub mod providers; // Data providers (Databento, Benzinga)
pub mod replay; // Real market data replay infrastructure for tests/benchmarks
pub mod storage;
pub mod training_pipeline; // Training data pipeline for ML models
pub mod types;
pub mod unified_feature_extractor; // Unified feature extraction across systems
pub mod utils;
pub mod validation; // Data validation and quality control
// Temporarily disabled due to compilation errors
// #[cfg(test)]
// mod storage_test;
// #[cfg(test)]
// REMOVED: polygon test module
// Tracing macros
use tracing::{error, info, warn};
// === External Re-exports ===
// Re-export HardwareTimestamp so downstream crates (ml) don't need a direct trading_engine dep
pub use trading_engine::timing::HardwareTimestamp;
// Commonly used external types
use tokio::sync::broadcast;
// Import configuration and event types that are actually used
use config::data_config::DataModuleConfig;
// OrderEvent type is not currently used in data module - removed import
use crate::brokers::{IBConfig, InteractiveBrokersAdapter};
use crate::error::Result;
use ::common::{MarketDataEvent, Subscription};
// Using direct imports from common crate - NO backward compatibility aliases
/// Initialize the data module with configuration
pub async fn initialize(config: DataModuleConfig) -> Result<DataManager> {
DataManager::new(config).await
}
/// Main data manager for coordinating all data providers and brokers
pub struct DataManager {
config: DataModuleConfig,
// REMOVED: polygon_client: Option<crate::polygon::PolygonClient>,
ib_client: Option<InteractiveBrokersAdapter>,
// icmarkets_client moved to core module
market_data_broadcast_tx: broadcast::Sender<MarketDataEvent>,
order_update_broadcast_tx: broadcast::Sender<MarketDataEvent>,
}
impl DataManager {
/// Create a new data manager
pub async fn new(config: DataModuleConfig) -> Result<Self> {
// Initialize broadcast channels with buffer sizes from config
let (market_data_broadcast_tx, _market_data_broadcast_rx) =
broadcast::channel(config.settings.market_data_buffer_size);
let (order_update_broadcast_tx, _order_update_broadcast_rx) =
broadcast::channel::<MarketDataEvent>(config.settings.order_event_buffer_size);
// REMOVED: Polygon client initialization
let ib_client = if let Some(ib_config) = &config.interactive_brokers {
let broker_config = IBConfig {
host: ib_config.host.clone(),
port: ib_config.port,
client_id: ib_config.client_id as i32,
account_id: std::env::var("IB_ACCOUNT_ID")
.unwrap_or_else(|_| "DU123456".to_string()),
connection_timeout: ib_config.timeout_seconds,
heartbeat_interval: 30,
max_reconnect_attempts: 5,
request_timeout: 30,
};
Some(InteractiveBrokersAdapter::new(broker_config))
} else {
None
};
// icmarkets_client initialization moved to core module
Ok(Self {
config,
// REMOVED: polygon_client,
ib_client,
// icmarkets_client field removed
market_data_broadcast_tx,
order_update_broadcast_tx,
})
}
// ICMarkets client methods moved to core module
/// Start all configured data providers and brokers
pub async fn start(&mut self) -> Result<()> {
// ICMarkets connection handling moved to core module
// REMOVED: Polygon client startup code
// Start Interactive Brokers client if configured
if let Some(client) = &mut self.ib_client {
info!("Starting Interactive Brokers connection");
match client.connect().await {
Ok(()) => {
info!("Interactive Brokers connection established successfully");
// Note: IB execution subscription would be implemented here when the
// subscribe_executions method is available in InteractiveBrokersAdapter
info!("Interactive Brokers connection ready - execution subscription not yet implemented");
},
Err(e) => {
error!("Failed to connect to Interactive Brokers: {}", e);
warn!("Continuing startup without Interactive Brokers connection");
},
}
}
Ok(())
}
/// Stop all data providers and brokers
pub async fn stop(&mut self) -> Result<()> {
// ICMarkets disconnect handling moved to core module
// Stop other clients as needed...
Ok(())
}
/// Subscribe to market data events
pub fn subscribe_market_data_events(&self) -> broadcast::Receiver<MarketDataEvent> {
self.market_data_broadcast_tx.subscribe()
}
/// Subscribe to order update events
pub fn subscribe_order_update_events(&self) -> broadcast::Receiver<MarketDataEvent> {
self.order_update_broadcast_tx.subscribe()
}
/// Subscribe to market data for specific symbols and data types
pub async fn subscribe_market_data(&self, subscription: Subscription) -> Result<()> {
// REMOVED: Polygon subscription code
// Subscribe with other clients as needed...
// ICMarkets doesn't typically handle market data subscriptions in the same way
info!(
"Market data subscription request received for: {:?}",
subscription.symbols
);
Ok(())
}
}
// Temporarily disabled due to compilation errors
// #[cfg(test)]
// mod tests {
// use super::*;
//
// #[test]
// fn test_config_default() {
// let config = DataModuleConfig::default();
// assert!(config.interactive_brokers.is_none()); // Default has no IB config
// assert_eq!(config.settings.max_reconnect_attempts, 10);
// }
//
// #[tokio::test]
// async fn test_data_manager_creation() {
// let config = DataModuleConfig::default();
// let data_manager = DataManager::new(config).await;
// assert!(data_manager.is_ok());
// }
// }