Open Census MCP Server

#!/usr/bin/env python3 """Debug OpenAI embeddings - check what's actually being used""" import os from pathlib import Path def check_openai_setup(): """Check OpenAI API configuration""" print("🔍 OpenAI Configuration Check") print("=" * 40) # Check API key api_key = os.getenv('OPENAI_API_KEY') if api_key: print(f"✅ OPENAI_API_KEY found: {api_key[:10]}...{api_key[-4:]}") else: print("❌ OPENAI_API_KEY not found") return False # Test OpenAI import try: from openai import OpenAI print("✅ OpenAI library imports successfully") client = OpenAI(api_key=api_key) print("✅ OpenAI client created") # Test embedding call print("\n🧪 Testing embedding generation...") response = client.embeddings.create( input=["test population query"], model="text-embedding-3-large" ) embedding = response.data[0].embedding print(f"✅ OpenAI embedding created: {len(embedding)} dimensions") if len(embedding) == 3072: print("✅ Correct embedding dimension (3072)") return True else: print(f"❌ Wrong embedding dimension: {len(embedding)} (expected 3072)") return False except Exception as e: print(f"❌ OpenAI test failed: {e}") return False def check_database_dimensions(): """Check what dimensions the databases expect""" print("\n🗄️ Database Dimension Check") print("=" * 40) # Check variables FAISS variables_dir = Path("knowledge-base/variables-faiss") build_info_file = variables_dir / "build_info.json" if build_info_file.exists(): import json with open(build_info_file) as f: build_info = json.load(f) dim = build_info.get('embedding_dimension', 'unknown') model = build_info.get('embedding_model', 'unknown') print(f"Variables DB: {dim} dimensions, model: {model}") if dim == 3072: print("✅ Variables built with OpenAI embeddings") else: print(f"❌ Variables built with wrong dimensions: {dim}") else: print("❌ Variables build_info.json not found") # Check table catalog catalog_dir = Path("knowledge-base/table-catalog") mapping_file = catalog_dir / "table_mapping_enhanced.json" if mapping_file.exists(): import json with open(mapping_file) as f: mapping_data = json.load(f) dim = mapping_data.get('embedding_dimension', 'unknown') print(f"Tables DB: {dim} dimensions") if dim == 3072: print("✅ Tables built with OpenAI embeddings") else: print(f"❌ Tables built with wrong dimensions: {dim}") else: print("❌ Tables mapping file not found") if __name__ == "__main__": openai_ok = check_openai_setup() check_database_dimensions() if openai_ok: print("\n🎯 Action Plan:") print("1. Force all search engines to use OpenAI embeddings") print("2. Remove all SentenceTransformers fallbacks") print("3. Ensure dimension consistency (3072 everywhere)") else: print("\n❌ Fix OpenAI setup first!")