sync.py

import os
import time
import shutil
from pathlib import Path
from huggingface_hub import HfApi, snapshot_download

# Configuration fetched from Space Secrets
REPO_ID = os.getenv("DATASET_REPO_ID") 
TOKEN = os.getenv("HF_TOKEN")

# Core directories
WORKSPACE_DIR = Path("/home/node/workspace")
OPENCODE_DIR = Path("/home/node/.local/share/opencode") # Default Linux storage
GLOBAL_CONFIG_DIR = Path("/home/node/.config/opencode") # Global AGENTS.md + skills
CACHE_DIR = Path("/home/node/.sync_cache")
INTERVAL = 30 * 60  # 30 minutes in seconds

api = HfApi(token=TOKEN)

def initial_pull():
    if not REPO_ID:
        print("No DATASET_REPO_ID provided. Skipping pull.")
        return
        
    print("Fetching dataset on startup...")
    try:
        # Download the latest snapshot of the private dataset
        local_dir = snapshot_download(repo_id=REPO_ID, repo_type="dataset", token=TOKEN)
        
        # Restore workspace files (includes env.example / env.json.local)
        ds_workspace = Path(local_dir) / "workspace"
        if ds_workspace.exists():
            shutil.copytree(ds_workspace, WORKSPACE_DIR, dirs_exist_ok=True)
            
        # Restore OpenCode sessions and history
        ds_opencode = Path(local_dir) / "opencode"
        if ds_opencode.exists():
            shutil.copytree(ds_opencode, OPENCODE_DIR, dirs_exist_ok=True)
            
        # Restore global config (AGENTS.md and skills)
        ds_global = Path(local_dir) / "global"
        if ds_global.exists():
            shutil.copytree(ds_global, GLOBAL_CONFIG_DIR, dirs_exist_ok=True)
            
        # Create an initial cache to track future deletions
        shutil.copytree(OPENCODE_DIR, CACHE_DIR, dirs_exist_ok=True)
        print("Restoration complete.")
    except Exception as e:
        print(f"Initial pull failed (likely an empty/new dataset): {e}")

def push_to_dataset():
    if not REPO_ID:
        return
        
    print("Initiating 30-minute sync cycle...")
    session_dir = OPENCODE_DIR / "storage" / "session"
    cache_session_dir = CACHE_DIR / "storage" / "session"
    
    # 1. Check for trashed sessions
    if cache_session_dir.exists():
        for cache_file in cache_session_dir.rglob("*.json"):
            relative_path = cache_file.relative_to(cache_session_dir)
            current_file = session_dir / relative_path
            
            if not current_file.exists():
                print(f"Detected deleted session: {relative_path}. Moving to trash...")
                api.upload_file(
                    path_or_fileobj=str(cache_file),
                    path_in_repo=f"trashed/sessions/{relative_path}",
                    repo_id=REPO_ID,
                    repo_type="dataset",
                    token=TOKEN
                )
                
    # 2. Sync workspace and configuration
    # The API automatically ignores the push if files are unchanged/empty.
    # Files like env.json.local will upload securely since the dataset is private.
    api.upload_folder(
        folder_path=str(WORKSPACE_DIR),
        path_in_repo="workspace",
        repo_id=REPO_ID,
        repo_type="dataset",
        token=TOKEN,
        ignore_patterns=["**/__pycache__/*", "**/.git/*"]
    )
    
    api.upload_folder(
        folder_path=str(OPENCODE_DIR),
        path_in_repo="opencode",
        repo_id=REPO_ID,
        repo_type="dataset",
        token=TOKEN,
        ignore_patterns=["**/*.log", "log", "log/**", "**/log/*"]
    )
    
    # 3. Sync global config (AGENTS.md and skills)
    api.upload_folder(
        folder_path=str(GLOBAL_CONFIG_DIR),
        path_in_repo="global",
        repo_id=REPO_ID,
        repo_type="dataset",
        token=TOKEN,
        ignore_patterns=["**/node_modules/**", "node_modules/**", "node_modules", "**/*.log"]
    )
    
    # 4. Rebuild local deletion cache
    if CACHE_DIR.exists():
        shutil.rmtree(CACHE_DIR)
    shutil.copytree(OPENCODE_DIR, CACHE_DIR, dirs_exist_ok=True)
    print("Sync cycle finished successfully.")

if __name__ == "__main__":
    initial_pull()
    while True:
        time.sleep(INTERVAL)
        push_to_dataset()