| """Upload RAG artifacts (FAISS index, metadata DB) to Modal Volume. |
| |
| Carica i file dell'indice RAG sul Volume 'boccaccio-rag-data' usato da |
| modal_app.py. Di default carica solo l'indice FAISS; con --metadata carica |
| anche metadata.db. |
| |
| Uso: |
| python scripts/upload_index_modal.py |
| python scripts/upload_index_modal.py --input rag/index/faiss/index.faiss |
| python scripts/upload_index_modal.py --metadata |
| """ |
| from __future__ import annotations |
|
|
| import argparse |
| from pathlib import Path |
|
|
| import modal |
|
|
|
|
| VOLUME_NAME = "boccaccio-rag-data" |
|
|
|
|
| def parse_args() -> argparse.Namespace: |
| p = argparse.ArgumentParser(description="Upload RAG index to Modal Volume.") |
| p.add_argument("--input", default="rag/index/faiss/index.faiss", |
| help="Path locale dell'indice FAISS da caricare.") |
| p.add_argument("--remote-name", default="index.faiss", |
| help="Nome del file sul Volume.") |
| p.add_argument("--metadata", action="store_true", |
| help="Carica anche metadata.db.") |
| p.add_argument("--metadata-input", default="rag/index/faiss/metadata.db", |
| help="Path locale di metadata.db.") |
| return p.parse_args() |
|
|
|
|
| def main() -> None: |
| args = parse_args() |
|
|
| local = Path(args.input) |
| if not local.exists(): |
| raise FileNotFoundError(f"File not found: {local}") |
| size_gb = local.stat().st_size / 1024**3 |
| print(f"Uploading {local} ({size_gb:.2f} GB) -> Volume '{VOLUME_NAME}'/{args.remote_name}") |
|
|
| vol = modal.Volume.from_name(VOLUME_NAME, create_if_missing=True) |
| with vol.batch_upload(force=True) as batch: |
| batch.put_file(str(local), args.remote_name) |
| print(f" index uploaded ({size_gb:.2f} GB).") |
|
|
| if args.metadata: |
| meta = Path(args.metadata_input) |
| if not meta.exists(): |
| raise FileNotFoundError(f"File not found: {meta}") |
| m_gb = meta.stat().st_size / 1024**3 |
| print(f"Uploading {meta} ({m_gb:.2f} GB) -> Volume '{VOLUME_NAME}'/metadata.db") |
| with vol.batch_upload(force=True) as batch: |
| batch.put_file(str(meta), "metadata.db") |
| print(f" metadata uploaded ({m_gb:.2f} GB).") |
|
|
| print("Done.") |
|
|
|
|
| if __name__ == "__main__": |
| main() |
|
|
