hf_sync.py

"""Auto-sync evaluations CSV to HF Space git (self or external)."""
import os
from pathlib import Path
from huggingface_hub import CommitScheduler, HfApi
from datetime import datetime

# Config
CSV_FILE = "evaluations.csv"
SPACE_ID = os.getenv("SPACE_ID")  # Auto-set by HF Spaces
HF_TOKEN = os.getenv("HF_TOKEN")  # Optional for external sync

# Global state
_scheduler = None
_sync_mode = None


def init_hf_sync():
    """Initialize HF sync. Auto-detects if running in Space.
    
    Modes:
    - 'self': Running in HF Space, commits to same Space
    - 'external': Local run, commits to external Space (needs HF_TOKEN + HF_SPACE_REPO)
    - None: No sync
    
    Returns:
        bool: True if sync enabled
    """
    global _scheduler, _sync_mode
    
    # Mode 1: Running inside HF Space - commit to self
    if SPACE_ID:
        try:
            from huggingface_hub import CommitScheduler
            _scheduler = CommitScheduler(
                repo_id=SPACE_ID,
                repo_type="space",
                folder_path=".",
                path_in_repo="data",
                every=1,  # Commit every second if changes exist
            )
            _sync_mode = 'self'
            print(f"✅ HF Space auto-commit enabled: {SPACE_ID}")
            print(f"📁 CSV saves to Space git automatically")
            return True
        except Exception as e:
            print(f"⚠️  Space auto-commit failed: {e}")
            return False
    
    # Mode 2: Running locally - sync to external Space
    external_repo = os.getenv("HF_SPACE_REPO")
    if external_repo and HF_TOKEN:
        try:
            _scheduler = CommitScheduler(
                repo_id=external_repo,
                repo_type="space",
                folder_path=".",
                path_in_repo="data",
                every=1,
                token=HF_TOKEN
            )
            _sync_mode = 'external'
            print(f"✅ External sync enabled: {external_repo}")
            print(f"📁 CSV syncs to external Space")
            return True
        except Exception as e:
            print(f"⚠️  External sync failed: {e}")
            return False
    
    # Mode 3: No sync
    print("📁 Local-only mode (no HF sync)")
    return False


def sync_to_hf():
    """Trigger immediate sync. CommitScheduler handles actual upload.
    
    With CommitScheduler, just saving the file triggers auto-commit.
    This function is a no-op but kept for API compatibility.
    """
    # CommitScheduler watches file changes automatically
    # No manual action needed
    return _scheduler is not None


if __name__ == "__main__":
    # Test sync
    if init_hf_sync():
        print("\n🧪 Testing event-based sync...")
        success = sync_to_hf()
        if success:
            print(f"✅ Sync successful: https://huggingface.co/spaces/{HF_SPACE_REPO}/blob/main/data/{CSV_FILE}")
        else:
            print("❌ Sync failed")
    else:
        print("❌ Sync not configured")