import argparse

import sys

import pandas as pd

from core.features import FeatureEngineer

from utilities.settings import logger

def main():

parser = argparse.ArgumentParser(description="Generate technical features for crypto 1h data.")

# Arguments

parser.add_argument("-i", "--input_file", type=str, required=True, help="Path to raw CSV file")

parser.add_argument("-o", "--output_file", type=str, required=True, help="Path to save feature CSV")

args = parser.parse_args()

# 1. Init Classes

engineer = FeatureEngineer()

print(f"--- Loading: {args.input_file} ---")

try:

df = pd.read_csv(args.input_file)

except FileNotFoundError:

sys.exit(f"Error: File {args.input_file} not found.")

# Standardize Timestamp

# Assuming 'timestamp' exists as per prompt. If 'timestamp_open' is the main one, adjust here.

if 'timestamp' in df.columns:

df['ts'] = pd.to_datetime(df['timestamp'])

elif 'timestamp_open' in df.columns:

df['ts'] = pd.to_datetime(df['timestamp_open'])

else:

sys.exit("Error: Could not find 'timestamp' or 'timestamp_open' column.")

# Sort

df_raw = df.sort_values('ts').reset_index(drop=True)

df_raw.sort_index(inplace=True)

if df_raw.empty:

sys.exit("Error: No data received from API.")

logger.info(f"loaded {len(df_raw)} rows.")

# 3. Apply Features

logger.info("--- Engineering Features ---")

df_features = engineer.gen_features(df_raw)

# 4. Clean (Drop NaNs from lookback periods)

initial_len = len(df_features)

df_clean = engineer.clean_data(df_features)

dropped = initial_len - len(df_clean)

logger.info(f"Dropped {dropped} rows due to rolling window warmup.")

# 5. Save

df_clean.to_csv(args.output_file, index=False)

# Summary

logger.info(f"--- Success ---")

logger.info(f"Saved to: {args.output_file}")

logger.info(f"Final Shape: {df_clean.shape}")

logger.info("\nSample Data:")

cols = ['uid', 'close', 'log_ret', 'accel', 'rsi14']

print(df_clean[cols].tail())

if __name__ == "__main__":

main()