debug_rag.py

import os
from langchain_chroma import Chroma
from get_embedding_function import get_embedding_function

# Direct paths without config
CHROMA_PATH = "chroma1"
DATA_PATH = "data"
EMBEDDING_MODEL = "all-mpnet-base-v2"

def debug_rag_database():
    """Debug what's actually in your RAG database"""
    
    print("=" * 80)
    print("RAG DATABASE DEBUGGING")
    print("=" * 80)
    
    # Check if database exists
    if not os.path.exists(CHROMA_PATH):
        print(f"❌ Database not found at: {CHROMA_PATH}")
        return
    
    print(f"✅ Database found at: {CHROMA_PATH}")
    
    # Check database contents
    try:
        embedding_function = get_embedding_function(EMBEDDING_MODEL)
        db = Chroma(persist_directory=CHROMA_PATH, embedding_function=embedding_function)
        
        # Get all documents
        all_docs = db.get()
        print(f"📊 Total documents in database: {len(all_docs['ids'])}")
        
        if len(all_docs['ids']) == 0:
            print("❌ Database is empty!")
            return
        
        # Check content
        print("\n📄 SAMPLE DOCUMENTS:")
        for i in range(min(5, len(all_docs['documents']))):
            doc_content = all_docs['documents'][i]
            metadata = all_docs['metadatas'][i] if all_docs['metadatas'] else {}
            
            print(f"\nDocument {i+1}:")
            print(f"  Source: {metadata.get('source', 'Unknown')}")
            print(f"  Content (first 200 chars): {doc_content[:600]}...")
            
            # Check for our target content
            content_lower = doc_content.lower()
            has_transformer = 'transformer' in content_lower
            has_gpt = 'gpt' in content_lower or 'gpt-3' in content_lower
            has_vit = 'vision transformer' in content_lower or 'vit' in content_lower
            
            print(f"  Contains 'transformer': {has_transformer}")
            print(f"  Contains 'gpt': {has_gpt}")
            print(f"  Contains 'vision transformer': {has_vit}")
        
        # Test retrieval
        print("\n🔍 TESTING RETRIEVAL:")
        test_queries = [
            "transformer",
            "GPT-3",
            "attention mechanism", 
            "BLEU score"
        ]
        
        for query in test_queries:
            results = db.similarity_search(query, k=3)
            print(f"\nQuery: '{query}' → {len(results)} results")
            if results:
                print(f"  Best match: {results[0].page_content[:100]}...")
            else:
                print("  No results found")
                
    except Exception as e:
        print(f"❌ Error accessing database: {e}")

def check_pdf_files():
    """Check if PDF files exist and are readable"""
    
    print("\n" + "=" * 80)
    print("PDF FILES CHECK")
    print("=" * 80)
    
    expected_files = [
        "attention_is_all_you_need.pdf",
        "gpt3_paper.pdf", 
        "vision_transformer.pdf"
    ]
    
    if not os.path.exists(DATA_PATH):
        print(f"❌ Data directory not found: {DATA_PATH}")
        return
    
    for filename in expected_files:
        filepath = os.path.join(DATA_PATH, filename)
        if os.path.exists(filepath):
            size = os.path.getsize(filepath)
            print(f"✅ {filename}: {size:,} bytes")
        else:
            print(f"❌ {filename}: NOT FOUND")
    
    # List all files in data directory
    print(f"\nAll files in {DATA_PATH}:")
    try:
        for file in os.listdir(DATA_PATH):
            filepath = os.path.join(DATA_PATH, file)
            if os.path.isfile(filepath):
                size = os.path.getsize(filepath)
                print(f"  {file}: {size:,} bytes")
    except Exception as e:
        print(f"Error listing files: {e}")

if __name__ == "__main__":
    check_pdf_files()
    debug_rag_database()