utils/runtime_fetcher.py

#!/usr/bin/env python3
"""
Runtime artifact fetcher for Dressify.
Downloads pre-processed artifacts from Hugging Face Hub to avoid reprocessing.
"""

import os
import json
import shutil
import tarfile
import zipfile
from pathlib import Path
from typing import Dict, List, Any, Optional
from huggingface_hub import hf_hub_download, snapshot_download

class RuntimeArtifactFetcher:
    """Fetches artifacts from HF Hub at runtime to avoid reprocessing."""
    
    def __init__(self, base_dir: str = "/home/user/app"):
        self.base_dir = base_dir
        self.data_dir = os.path.join(base_dir, "data/Polyvore")
        self.splits_dir = os.path.join(self.data_dir, "splits")
        self.export_dir = os.getenv("EXPORT_DIR", "models/exports")
        
        # Default HF repositories - updated to use your specific repos
        self.default_repos = {
            "splits": "Stylique/Dressify-Helper",
            "models": "Stylique/dressify-models",
            "metadata": "Stylique/Dressify-Helper"
        }
    
    def check_artifacts_needed(self) -> Dict[str, Any]:
        """Check what artifacts need to be fetched."""
        needs = {
            "splits": False,
            "models": False,
            "metadata": False,
            "total_size_mb": 0
        }
        
        # Check splits
        if not os.path.exists(self.splits_dir) or not self._has_complete_splits():
            needs["splits"] = True
            needs["total_size_mb"] += 50  # Estimate splits size
        
        # Check models
        if not os.path.exists(self.export_dir) or not self._has_trained_models():
            needs["models"] = True
            needs["total_size_mb"] += 200  # Estimate models size
        
        # Check metadata
        if not self._has_complete_metadata():
            needs["metadata"] = True
            needs["total_size_mb"] += 100  # Estimate metadata size
        
        return needs
    
    def _has_complete_splits(self) -> bool:
        """Check if complete splits are available."""
        required_files = [
            "train.json", "valid.json", "test.json",
            "outfit_triplets_train.json", "outfit_triplets_valid.json", "outfit_triplets_test.json"
        ]
        
        for file in required_files:
            if not os.path.exists(os.path.join(self.splits_dir, file)):
                return False
        return True
    
    def _has_trained_models(self) -> bool:
        """Check if trained models are available."""
        required_files = [
            "resnet_item_embedder_best.pth",
            "vit_outfit_model_best.pth"
        ]
        
        for file in required_files:
            if not os.path.exists(os.path.join(self.export_dir, file)):
                return False
        return True
    
    def _has_complete_metadata(self) -> bool:
        """Check if complete metadata is available."""
        required_files = [
            "polyvore_item_metadata.json",
            "polyvore_outfit_titles.json",
            "categories.csv"
        ]
        
        for file in required_files:
            if not os.path.exists(os.path.join(self.data_dir, file)):
                return False
        return True
    
    def fetch_splits_from_hf(self, repo: str = None, token: str = None) -> bool:
        """Fetch dataset splits from HF Hub."""
        if repo is None:
            repo = self.default_repos["splits"]
        
        try:
            print(f"🔄 Fetching splits from {repo}...")
            
            # Create splits directory
            os.makedirs(self.splits_dir, exist_ok=True)
            
            # Download splits files
            split_files = [
                "train.json", "valid.json", "test.json",
                "outfits_train.json", "outfits_valid.json", "outfits_test.json",
                "outfit_triplets_train.json", "outfit_triplets_valid.json", "outfit_triplets_test.json"
            ]
            
            for file in split_files:
                try:
                    local_path = hf_hub_download(
                        repo_id=repo,
                        filename=f"splits/{file}",
                        local_dir=self.splits_dir,
                        token=token
                    )
                    print(f"✅ Downloaded: {file}")
                except Exception as e:
                    print(f"⚠️  Failed to download {file}: {e}")
            
            print(f"✅ Splits fetched successfully to {self.splits_dir}")
            return True
            
        except Exception as e:
            print(f"❌ Failed to fetch splits: {e}")
            return False
    
    def fetch_models_from_hf(self, repo: str = None, token: str = None) -> bool:
        """Fetch trained models from HF Hub."""
        if repo is None:
            repo = self.default_repos["models"]
        
        try:
            print(f"🔄 Fetching models from {repo}...")
            
            # Create export directory
            os.makedirs(self.export_dir, exist_ok=True)
            
            # Download model files
            model_files = [
                "resnet_item_embedder_best.pth",
                "vit_outfit_model_best.pth",
                "resnet_metrics.json",
                "vit_metrics.json"
            ]
            
            for file in model_files:
                try:
                    local_path = hf_hub_download(
                        repo_id=repo,
                        filename=file,
                        local_dir=self.export_dir,
                        token=token
                    )
                    print(f"✅ Downloaded: {file}")
                except Exception as e:
                    print(f"⚠️  Failed to download {file}: {e}")
            
            print(f"✅ Models fetched successfully to {self.export_dir}")
            return True
            
        except Exception as e:
            print(f"❌ Failed to fetch models: {e}")
            return False
    
    def fetch_metadata_from_hf(self, repo: str = None, token: str = None) -> bool:
        """Fetch metadata from HF Hub."""
        if repo is None:
            repo = self.default_repos["metadata"]
        
        try:
            print(f"🔄 Fetching metadata from {repo}...")
            
            # Create data directory
            os.makedirs(self.data_dir, exist_ok=True)
            
            # Download metadata files
            metadata_files = [
                "polyvore_item_metadata.json",
                "polyvore_outfit_titles.json",
                "categories.csv"
            ]
            
            for file in metadata_files:
                try:
                    local_path = hf_hub_download(
                        repo_id=repo,
                        filename=f"metadata/{file}",
                        local_dir=self.data_dir,
                        token=token
                    )
                    print(f"✅ Downloaded: {file}")
                except Exception as e:
                    print(f"⚠️  Failed to download {file}: {e}")
            
            print(f"✅ Metadata fetched successfully to {self.data_dir}")
            return True
            
        except Exception as e:
            print(f"❌ Failed to fetch metadata: {e}")
            return False
    
    def fetch_everything_from_hf(self, splits_repo: str = None, models_repo: str = None, 
                                metadata_repo: str = None, token: str = None) -> Dict[str, bool]:
        """Fetch all artifacts from HF Hub."""
        results = {}
        
        print("🚀 Starting comprehensive artifact fetch from HF Hub...")
        
        # Fetch splits
        results["splits"] = self.fetch_splits_from_hf(splits_repo, token)
        
        # Fetch models
        results["models"] = self.fetch_models_from_hf(models_repo, token)
        
        # Fetch metadata
        results["metadata"] = self.fetch_metadata_from_hf(metadata_repo, token)
        
        # Summary
        success_count = sum(results.values())
        total_count = len(results)
        
        print(f"\n📊 Fetch Summary: {success_count}/{total_count} successful")
        for artifact, success in results.items():
            status = "✅" if success else "❌"
            print(f"  {status} {artifact}")
        
        return results
    
    def download_and_extract_package(self, package_path: str, extract_to: str = None) -> bool:
        """Download and extract a package from HF Hub."""
        try:
            if extract_to is None:
                extract_to = self.base_dir
            
            print(f"🔄 Downloading and extracting package: {package_path}")
            
            # Download the package
            local_path = hf_hub_download(
                repo_id="Stylique/Dressify-Helper",
                filename=f"packages/{os.path.basename(package_path)}",
                local_dir=extract_to,
                token=None
            )
            
            # Extract based on file type
            if package_path.endswith(".tar.gz"):
                with tarfile.open(local_path, 'r:gz') as tar:
                    tar.extractall(extract_to)
            elif package_path.endswith(".zip"):
                with zipfile.ZipFile(local_path, 'r') as zipf:
                    zipf.extractall(extract_to)
            
            print(f"✅ Package extracted to {extract_to}")
            return True
            
        except Exception as e:
            print(f"❌ Failed to download/extract package: {e}")
            return False
    
    def get_fetch_status(self) -> Dict[str, Any]:
        """Get current fetch status."""
        return {
            "splits_available": self._has_complete_splits(),
            "models_available": self._has_trained_models(),
            "metadata_available": self._has_complete_metadata(),
            "artifacts_needed": self.check_artifacts_needed(),
            "base_dir": self.base_dir,
            "splits_dir": self.splits_dir,
            "export_dir": self.export_dir,
            "hf_repos": self.default_repos
        }

def create_runtime_fetcher() -> RuntimeArtifactFetcher:
    """Create a runtime fetcher instance."""
    return RuntimeArtifactFetcher()

def auto_fetch_if_needed(token: str = None) -> Dict[str, bool]:
    """Automatically fetch artifacts if they're needed."""
    fetcher = create_runtime_fetcher()
    
    # Check what's needed
    needs = fetcher.check_artifacts_needed()
    
    if not any([needs["splits"], needs["models"], needs["metadata"]]):
        print("✅ All artifacts are already available - no fetching needed")
        return {"splits": True, "models": True, "metadata": True}
    
    print(f"🔄 Auto-fetching needed artifacts (estimated size: {needs['total_size_mb']} MB)")
    
    # Fetch what's needed
    results = {}
    if needs["splits"]:
        results["splits"] = fetcher.fetch_splits_from_hf(token=token)
    
    if needs["models"]:
        results["models"] = fetcher.fetch_models_from_hf(token=token)
    
    if needs["metadata"]:
        results["metadata"] = fetcher.fetch_metadata_from_hf(token=token)
    
    return results

if __name__ == "__main__":
    # Test the fetcher
    fetcher = create_runtime_fetcher()
    status = fetcher.get_fetch_status()
    print("Current fetch status:", json.dumps(status, indent=2))