pre-process.py

import argparse
import os
import zipfile
import shutil

# -----------------------------
# Argument parsing
# -----------------------------
parser = argparse.ArgumentParser()
parser.add_argument("--data_path", type=str, required=True)
args = parser.parse_args()

DATA_PATH = args.data_path

ZIP_NAME = "sentiment140"
TARGET_NAME = "sentiment_data_file.csv"

print("Data path:", DATA_PATH)

if not os.path.isdir(DATA_PATH):
    raise FileNotFoundError(f"{DATA_PATH} is not a directory")

zip_path = os.path.join(DATA_PATH, ZIP_NAME)

if not os.path.isfile(zip_path):
    raise FileNotFoundError(f"{ZIP_NAME} not found in {DATA_PATH}")

# -----------------------------
# Unzip
# -----------------------------
print("Unzipping:", zip_path)

with zipfile.ZipFile(zip_path, "r") as zip_ref:
    print("Zip contents:")
    zip_ref.printdir()
    zip_ref.extractall(DATA_PATH)

# -----------------------------
# Find extracted CSV
# -----------------------------
extracted_csv = None

for root, _, files in os.walk(DATA_PATH):
    for f in files:
        if f.endswith(".csv") and f != TARGET_NAME:
            extracted_csv = os.path.join(root, f)
            break

if not extracted_csv:
    raise RuntimeError("No CSV file found after unzip")

print("Extracted CSV found:", extracted_csv)

# -----------------------------
# Rename (or copy if rename fails)
# -----------------------------
target_path = os.path.join(DATA_PATH, TARGET_NAME)

try:
    os.rename(extracted_csv, target_path)
    print("Renamed to:", target_path)
except PermissionError:
    print("Rename not allowed, copying instead...")
    shutil.copyfile(extracted_csv, target_path)
    print("Copied to:", target_path)

# -----------------------------
# Final verification
# -----------------------------
print("\nFinal files in dataset directory:")
for f in os.listdir(DATA_PATH):
    print(" -", f)