| """Sinhronizē Maris atmiņas datus uz lokālo disku.""" |
|
|
| from __future__ import annotations |
|
|
| import argparse |
| import logging |
| import os |
|
|
| logger = logging.getLogger(__name__) |
|
|
|
|
| def sync_datasets( |
| repo_id: str | None = None, |
| local_dir: str = "./data/hf_cache", |
| ) -> None: |
| """Lejupielādē Maris AI atmiņas datus no origin repozitorija.""" |
| repo_id = ( |
| repo_id |
| or os.getenv("MARIS_MEMORY_REPO") |
| or os.getenv("HF_DATASET_REPO", "MarisUK/maris-ai-memory") |
| ) |
| token = os.getenv("MARIS_REPO_TOKEN") or os.getenv("MARIS_TOKEN") or os.getenv("HF_TOKEN") |
|
|
| try: |
| from huggingface_hub import snapshot_download |
|
|
| logger.info("Lejupielādē dataset: %s -> %s", repo_id, local_dir) |
| snapshot_download( |
| repo_id=repo_id, |
| repo_type="dataset", |
| local_dir=local_dir, |
| token=token, |
| ) |
| logger.info("Sinhronizācija pabeigta: %s", local_dir) |
| except Exception as exc: |
| logger.error("Sinhronizācijas kļūda: %s", exc) |
| raise |
|
|
|
|
| if __name__ == "__main__": |
| parser = argparse.ArgumentParser(description="Sinhronizē Maris AI atmiņas repozitoriju") |
| parser.add_argument("--repo-id", help="Atmiņas repo ID") |
| parser.add_argument("--local-dir", default="./data/hf_cache", help="Lokālā direktorija") |
| args = parser.parse_args() |
|
|
| logging.basicConfig(level=logging.INFO) |
| sync_datasets(args.repo_id, args.local_dir) |
|
|
