foxhunt/services/backtesting_service/tests/data_replay.rs

//! Tests for data replay functionality in backtesting service
//!
//! Target Coverage: 50%+ for historical data replay, timestamp handling, and data validation

use anyhow::Result;
use chrono::{Duration, Utc};
use rust_decimal::Decimal;
use std::sync::Arc;

mod mock_repositories;

use backtesting_service::repositories::{MarketDataRepository, NewsRepository};
use mock_repositories::*;

/// Test loading historical market data
#[tokio::test]
async fn test_load_historical_data() -> Result<()> {
    let market_data = generate_sample_market_data("AAPL", 100, 150.0, 0.02);
    let repo = MockMarketDataRepository::with_data(market_data.clone());

    let start_time = market_data.first().unwrap().timestamp.timestamp_nanos_opt().unwrap_or(0);
    let end_time = market_data.last().unwrap().timestamp.timestamp_nanos_opt().unwrap_or(0);

    let loaded = repo
        .load_historical_data(&["AAPL".to_string()], start_time, end_time)
        .await?;

    assert_eq!(loaded.len(), 100, "Should load all 100 data points");
    assert_eq!(loaded[0].symbol, "AAPL");
    assert!(loaded[0].close > Decimal::ZERO);

    Ok(())
}

/// Test data filtering by symbol
#[tokio::test]
async fn test_data_filtering_by_symbol() -> Result<()> {
    let mut all_data = Vec::new();
    all_data.extend(generate_sample_market_data("AAPL", 50, 150.0, 0.02));
    all_data.extend(generate_sample_market_data("MSFT", 50, 200.0, 0.015));
    all_data.extend(generate_sample_market_data("GOOGL", 50, 120.0, 0.025));

    let repo = MockMarketDataRepository::with_data(all_data.clone());

    let start_time = all_data.first().unwrap().timestamp.timestamp_nanos_opt().unwrap_or(0);
    let end_time = all_data.last().unwrap().timestamp.timestamp_nanos_opt().unwrap_or(0);

    // Load only AAPL data
    let aapl_data = repo
        .load_historical_data(&["AAPL".to_string()], start_time, end_time)
        .await?;

    assert_eq!(aapl_data.len(), 50, "Should load only AAPL data");
    assert!(aapl_data.iter().all(|d| d.symbol == "AAPL"));

    // Load multiple symbols
    let multi_data = repo
        .load_historical_data(
            &["AAPL".to_string(), "MSFT".to_string()],
            start_time,
            end_time,
        )
        .await?;

    assert_eq!(multi_data.len(), 100, "Should load AAPL and MSFT data");

    Ok(())
}

/// Test timestamp range filtering
#[tokio::test]
async fn test_timestamp_range_filtering() -> Result<()> {
    let market_data = generate_sample_market_data("AAPL", 100, 150.0, 0.02);
    let repo = MockMarketDataRepository::with_data(market_data.clone());

    // Get middle 50 days
    let start_time = market_data[25].timestamp.timestamp_nanos_opt().unwrap_or(0);
    let end_time = market_data[74].timestamp.timestamp_nanos_opt().unwrap_or(0);

    let filtered = repo
        .load_historical_data(&["AAPL".to_string()], start_time, end_time)
        .await?;

    assert_eq!(filtered.len(), 50, "Should load middle 50 data points");
    assert!(filtered[0].timestamp >= market_data[25].timestamp);
    assert!(filtered.last().unwrap().timestamp <= market_data[74].timestamp);

    Ok(())
}

/// Test data availability check
#[tokio::test]
async fn test_data_availability_check() -> Result<()> {
    let market_data = generate_sample_market_data("AAPL", 50, 150.0, 0.02);
    let repo = MockMarketDataRepository::with_data(market_data.clone());

    let start_time = market_data.first().unwrap().timestamp.timestamp_nanos_opt().unwrap_or(0);
    let end_time = market_data.last().unwrap().timestamp.timestamp_nanos_opt().unwrap_or(0);

    let availability = repo
        .check_data_availability(
            &["AAPL".to_string(), "MSFT".to_string()],
            start_time,
            end_time,
        )
        .await?;

    assert_eq!(availability.len(), 2);
    assert_eq!(availability.get("AAPL"), Some(&true));
    assert_eq!(availability.get("MSFT"), Some(&true));

    Ok(())
}

/// Test empty data range
#[tokio::test]
async fn test_empty_data_range() -> Result<()> {
    let market_data = generate_sample_market_data("AAPL", 50, 150.0, 0.02);
    let repo = MockMarketDataRepository::with_data(market_data.clone());

    // Request data from future (no data available)
    let future_start = Utc::now().timestamp_nanos_opt().unwrap_or(0) + 1_000_000_000_000;
    let future_end = future_start + 1_000_000_000_000;

    let loaded = repo
        .load_historical_data(&["AAPL".to_string()], future_start, future_end)
        .await?;

    assert_eq!(loaded.len(), 0, "Future data should be empty");

    Ok(())
}

/// Test chronological order of replayed data
#[tokio::test]
async fn test_chronological_order() -> Result<()> {
    let market_data = generate_sample_market_data("AAPL", 100, 150.0, 0.02);
    let repo = MockMarketDataRepository::with_data(market_data.clone());

    let start_time = market_data.first().unwrap().timestamp.timestamp_nanos_opt().unwrap_or(0);
    let end_time = market_data.last().unwrap().timestamp.timestamp_nanos_opt().unwrap_or(0);

    let loaded = repo
        .load_historical_data(&["AAPL".to_string()], start_time, end_time)
        .await?;

    // Verify data is in chronological order
    for i in 1..loaded.len() {
        assert!(
            loaded[i].timestamp >= loaded[i - 1].timestamp,
            "Data should be in chronological order"
        );
    }

    Ok(())
}

/// Test news event replay
#[tokio::test]
async fn test_news_event_replay() -> Result<()> {
    let symbols = vec!["AAPL".to_string()];
    let news_events = generate_sample_news_events(&symbols, 50);
    let repo = MockNewsRepository::with_events(news_events.clone());

    let start_time = news_events.first().unwrap().timestamp;
    let end_time = news_events.last().unwrap().timestamp;

    let loaded = repo.load_news_events(&symbols, start_time, end_time).await?;

    assert_eq!(loaded.len(), 50, "Should load all news events");
    assert!(loaded.iter().all(|e| e.symbols.contains(&"AAPL".to_string())));

    Ok(())
}

/// Test news event filtering by time range
#[tokio::test]
async fn test_news_event_time_filtering() -> Result<()> {
    let symbols = vec!["AAPL".to_string()];
    let news_events = generate_sample_news_events(&symbols, 100);
    let repo = MockNewsRepository::with_events(news_events.clone());

    // Get middle portion
    let start_time = news_events[30].timestamp;
    let end_time = news_events[69].timestamp;

    let loaded = repo.load_news_events(&symbols, start_time, end_time).await?;

    assert!(loaded.len() >= 30 && loaded.len() <= 50, "Should load middle portion of events");
    assert!(loaded.iter().all(|e| e.timestamp >= start_time && e.timestamp <= end_time));

    Ok(())
}

/// Test sentiment data aggregation
#[tokio::test]
async fn test_sentiment_data_aggregation() -> Result<()> {
    let symbols = vec!["AAPL".to_string(), "MSFT".to_string()];
    let news_events = generate_sample_news_events(&symbols, 50);
    let repo = MockNewsRepository::with_events(news_events.clone());

    let timestamp = Utc::now();
    let lookback_hours = 24;

    let sentiment = repo
        .get_sentiment_data(&symbols, timestamp, lookback_hours)
        .await?;

    assert!(sentiment.contains_key("AAPL"));
    assert!(sentiment.contains_key("MSFT"));

    // Sentiment should be in valid range
    for (_, value) in &sentiment {
        assert!(
            *value >= -1.0 && *value <= 1.0,
            "Sentiment should be between -1 and 1"
        );
    }

    Ok(())
}

/// Test mixed timeframe data replay
#[tokio::test]
async fn test_mixed_timeframe_data() -> Result<()> {
    use backtesting_service::strategy_engine::{MarketData, TimeFrame};

    let mut market_data = Vec::new();
    let base_time = Utc::now() - Duration::days(100);

    // Create data with different timeframes
    for i in 0..30 {
        market_data.push(MarketData {
            symbol: "AAPL".to_string(),
            timestamp: base_time + Duration::days(i),
            open: Decimal::from(150),
            high: Decimal::from(152),
            low: Decimal::from(148),
            close: Decimal::from(151),
            volume: Decimal::from(1000000),
            timeframe: TimeFrame::Daily,
        });
    }

    for i in 0..24 {
        market_data.push(MarketData {
            symbol: "AAPL".to_string(),
            timestamp: base_time + Duration::hours(i),
            open: Decimal::from(150),
            high: Decimal::from(151),
            low: Decimal::from(149),
            close: Decimal::from(150),
            volume: Decimal::from(100000),
            timeframe: TimeFrame::Hour,
        });
    }

    let repo = MockMarketDataRepository::with_data(market_data.clone());

    let start_time = base_time.timestamp_nanos_opt().unwrap_or(0);
    let end_time = (base_time + Duration::days(50)).timestamp_nanos_opt().unwrap_or(0);

    let loaded = repo
        .load_historical_data(&["AAPL".to_string()], start_time, end_time)
        .await?;

    // Should load all data regardless of timeframe
    assert!(!loaded.is_empty(), "Should load mixed timeframe data");

    Ok(())
}

/// Test data integrity validation
#[tokio::test]
async fn test_data_integrity_validation() -> Result<()> {
    let market_data = generate_sample_market_data("AAPL", 50, 150.0, 0.02);
    let repo = MockMarketDataRepository::with_data(market_data.clone());

    let start_time = market_data.first().unwrap().timestamp.timestamp_nanos_opt().unwrap_or(0);
    let end_time = market_data.last().unwrap().timestamp.timestamp_nanos_opt().unwrap_or(0);

    let loaded = repo
        .load_historical_data(&["AAPL".to_string()], start_time, end_time)
        .await?;

    // Validate data integrity
    for data_point in &loaded {
        // OHLC validation
        assert!(data_point.high >= data_point.open, "High should be >= open");
        assert!(data_point.high >= data_point.close, "High should be >= close");
        assert!(data_point.low <= data_point.open, "Low should be <= open");
        assert!(data_point.low <= data_point.close, "Low should be <= close");
        assert!(data_point.volume >= Decimal::ZERO, "Volume should be non-negative");
    }

    Ok(())
}

/// Test concurrent data loading
#[tokio::test]
async fn test_concurrent_data_loading() -> Result<()> {
    let market_data = generate_sample_market_data("AAPL", 100, 150.0, 0.02);
    let repo = Arc::new(MockMarketDataRepository::with_data(market_data.clone()));

    let start_time = market_data.first().unwrap().timestamp.timestamp_nanos_opt().unwrap_or(0);
    let end_time = market_data.last().unwrap().timestamp.timestamp_nanos_opt().unwrap_or(0);

    // Spawn multiple concurrent load tasks
    let mut handles = Vec::new();
    for _ in 0..10 {
        let repo_clone = repo.clone();
        let handle = tokio::spawn(async move {
            repo_clone
                .load_historical_data(&["AAPL".to_string()], start_time, end_time)
                .await
        });
        handles.push(handle);
    }

    // Wait for all tasks
    for handle in handles {
        let result = handle.await??;
        assert_eq!(result.len(), 100, "Each concurrent load should return all data");
    }

    Ok(())
}