Core Logic.py

import pandas as pd
import logging

def load_and_validate_data(filepath: str) -> pd.DataFrame:
    """Loads CSV and validates the required structure."""
    try:
        # Load the dataset
        df = pd.read_csv(filepath)
    except FileNotFoundError:
        logging.error(f"Input file missing: {filepath}")
        raise
    except pd.errors.EmptyDataError:
        logging.error("Input CSV is completely empty or invalid.")
        raise
        
    # Validate that the 'close' column exists [cite: 27, 38]
    if 'close' not in df.columns:
        logging.error("Required column 'close' is missing.")
        raise ValueError("Missing 'close' column.")
        
    # Validate the dataframe actually has rows [cite: 27]
    if df.empty:
        logging.error("Dataset contains no data rows.")
        raise ValueError("Empty dataset.")
        
    return df

def compute_signals(df: pd.DataFrame, window: int) -> pd.DataFrame:
    """Computes the rolling mean and generates the binary signal."""
    # Compute rolling mean on 'close' [cite: 41]
    df['rolling_mean'] = df['close'].rolling(window=window).mean()
    
    # Generate signal: 1 if close > rolling_mean, else 0 [cite: 45]
    # IMPORTANT: The assignment asks how we handle the first window-1 rows.
    # Pandas naturally puts NaNs in the rolling_mean for the first (window-1) rows. 
    # In Python, (Any Number > NaN) evaluates to False. 
    # Therefore, those initial rows will safely and consistently evaluate to a signal of 0.
    df['signal'] = (df['close'] > df['rolling_mean']).astype(int)
    
    return df