from dataclasses import dataclass, field
from typing import List, Optional


@dataclass
class DataConfig:
    """Configuration for data loading and basic filtering."""
    csv_path: str
    min_date: str = "2017-06-01"
    max_date: Optional[str] = None
    drop_volume_zero: bool = True


@dataclass
class FeatureConfig:
    """Configuration for feature engineering."""
    ohlcv_cols: List[str] = field(default_factory=lambda: ["Open", "High", "Low", "Close", "Volume"])
    lags: int = 3
    window_sizes: List[int] = field(default_factory=lambda: [5, 15, 30])


@dataclass
class PreprocessConfig:
    """Configuration for preprocessing and NaN handling."""
    impute_nans: bool = True


@dataclass
class PruningConfig:
    """Configuration for feature pruning and CV."""
    do_walk_forward_cv: bool = True
    n_splits: int = 5
    auto_prune: bool = True
    top_k: int = 150
    known_low_features: List[str] = field(
        default_factory=lambda: [
            "supertrend_12_3.0",
            "supertrend_10_1.0",
            "supertrend_11_2.0",
            "supertrend_trend_12_3.0",
            "supertrend_trend_10_1.0",
            "supertrend_trend_11_2.0",
            "hour",
        ]
    )


@dataclass
class OutputConfig:
    """Configuration for outputs and artifacts."""
    charts_dir: str = "charts"
    results_csv: str = "../data/cumulative_feature_results.csv"
    model_output_path: str = "../data/xgboost_model_all_features.json"


@dataclass
class RunConfig:
    """Top-level configuration grouping for a pipeline run."""
    data: DataConfig
    features: FeatureConfig = field(default_factory=FeatureConfig)
    preprocess: PreprocessConfig = field(default_factory=PreprocessConfig)
    pruning: PruningConfig = field(default_factory=PruningConfig)
    output: OutputConfig = field(default_factory=OutputConfig)