fix usage of gdown and add --no-upsert flag

pypi_scout/scripts/download_dataset.py

@@ -21,7 +21,7 @@ def download_dataset():
 
     logging.info(f"⬇️ Downloading raw dataset from Google Drive to {target_path}...")
     url = f"https://drive.google.com/uc?id={config.GOOGLE_FILE_ID}"
-    gdown.download(url, target_path, quiet=False)
+    gdown.download(url, str(target_path), quiet=False)
     logging.info("✅ Done!")
 
 

pypi_scout/scripts/setup.py

@@ -1,13 +1,22 @@
+import argparse
 from pypi_scout.scripts.download_dataset import download_dataset
 from pypi_scout.scripts.process_dataset import process_dataset
 from pypi_scout.scripts.setup_pinecone import setup_pinecone
 from pypi_scout.scripts.upsert_data import upsert_data
 from pypi_scout.utils.logging import setup_logging
 
-setup_logging()
-
-if __name__ == "__main__":
+def main(no_upsert):
+    setup_logging()
     setup_pinecone()
     download_dataset()
     process_dataset()
-    upsert_data()
+    if not no_upsert:
+        upsert_data()
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(description="Run the setup script with optional flags.")
+    parser.add_argument('--no-upsert', action='store_true', help='If set, do not upsert data to the Pinecone database.')
+
+    args = parser.parse_args()
+
+    main(no_upsert=args.no_upsert)