diagnose_hf.py

"""
Diagnostic script for HuggingFace Spaces deployment
Run this to check what's happening in the Space
"""

import os
import sys
from pathlib import Path

print("=" * 60)
print("🔍 HuggingFace Space Diagnostic Report")
print("=" * 60)

# Check Python version
print(f"\n📌 Python Version: {sys.version}")

# Check current directory
print(f"\n📁 Current Directory: {os.getcwd()}")
print(f"📂 Directory Contents:")
for item in sorted(os.listdir(".")):
    print(f"   - {item}")

# Check if data directory exists
data_path = Path("data")
if data_path.exists():
    print(f"\n✅ Data directory found")
    print(f"📂 Data directory contents:")
    for item in sorted(data_path.rglob("*")):
        if item.is_file():
            print(f"   - {item.relative_to(data_path)} ({item.stat().st_size} bytes)")
else:
    print(f"\n❌ Data directory NOT found!")

# Check if src directory exists
src_path = Path("src")
if src_path.exists():
    print(f"\n✅ Src directory found")
    print(f"📂 Src directory structure:")
    for item in sorted(src_path.rglob("*.py")):
        print(f"   - {item.relative_to(src_path)}")
else:
    print(f"\n❌ Src directory NOT found!")

# Check environment variables
print(f"\n🔑 Environment Variables:")
required_vars = [
    "AZURE_OPENAI_API_KEY",
    "AZURE_OPENAI_ENDPOINT",
    "AZURE_OPENAI_DEPLOYMENT_NAME",
    "AZURE_OPENAI_EMBEDDING_DEPLOYMENT_NAME",
    "AZURE_OPENAI_API_VERSION",
]

for var in required_vars:
    value = os.getenv(var)
    if value:
        # Mask sensitive values
        if "KEY" in var:
            display = f"{value[:10]}...{value[-5:]}" if len(value) > 15 else "***"
        elif "ENDPOINT" in var:
            display = f"{value[:40]}..." if len(value) > 40 else value
        else:
            display = value
        print(f"   ✅ {var}: {display}")
    else:
        print(f"   ❌ {var}: NOT SET")

# Try importing dependencies
print(f"\n📦 Checking Dependencies:")
dependencies = [
    "gradio",
    "langchain",
    "langchain_openai",
    "langchain_community",
    "faiss",
    "rank_bm25",
    "dotenv",
]

for dep in dependencies:
    try:
        if dep == "faiss":
            import faiss
        elif dep == "dotenv":
            import dotenv
        else:
            __import__(dep)
        print(f"   ✅ {dep}")
    except ImportError as e:
        print(f"   ❌ {dep}: {e}")

# Try importing project modules
print(f"\n🔧 Checking Project Modules:")
sys.path.insert(0, str(Path(__file__).parent / "src"))

try:
    from setup_retrieval import setup_retrieval_system
    print(f"   ✅ setup_retrieval")
except Exception as e:
    print(f"   ❌ setup_retrieval: {e}")

try:
    from rag.chain import UPBRAGChain
    print(f"   ✅ UPBRAGChain")
except Exception as e:
    print(f"   ❌ UPBRAGChain: {e}")

try:
    from metadata.metadata_manager import MetadataManager
    print(f"   ✅ MetadataManager")
except Exception as e:
    print(f"   ❌ MetadataManager: {e}")

# Try minimal RAG initialization
print(f"\n🚀 Attempting RAG Initialization:")
try:
    from setup_retrieval import setup_retrieval_system
    print("   - Loading documents...")
    retriever, vectorstore_manager, chunks = setup_retrieval_system()
    print(f"   ✅ Success! Loaded {len(chunks)} chunks")
except Exception as e:
    import traceback
    print(f"   ❌ Failed:")
    print(f"\n{traceback.format_exc()}")

print("\n" + "=" * 60)
print("✅ Diagnostic Complete")
print("=" * 60)