test_processing.py

import os
import sys
from ingest import load_and_chunk
from langchain_core.documents import Document

def test_ingest():
    print("--- Testing Document Processing ---")
    
    file_path = "test_doc_a.pdf"
    if not os.path.exists(file_path):
        print(f"❌ Test file not found: {file_path}")
        # Create a dummy PDF if needed or just fail
        return False
        
    try:
        print(f"Processing {file_path}...")
        chunks = load_and_chunk(file_path, "doc_a")
        
        if not chunks:
            print("❌ No chunks generated")
            return False
            
        print(f"✅ Generated {len(chunks)} chunks")
        print(f"   First chunk content preview: {chunks[0].page_content[:100]}...")
        
        # Verify metadata
        if chunks[0].metadata.get("doc_source") == "doc_a":
             print("✅ Metadata verified")
        else:
             print("❌ Metadata incorrect")
             return False
             
        return True
        
    except Exception as e:
        print(f"❌ Processing failed: {e}")
        return False

if __name__ == "__main__":
    if test_ingest():
        print("\n✅ Document Processing Verification Successful")
        sys.exit(0)
    else:
        print("\n❌ Document Processing Verification Failed")
        sys.exit(1)