#!/usr/bin/env python3
"""
Test Statistical Advisor Integration - Minimal Test
This is the actual integration test for the statistical advisor.
"""
import sys
import os
import logging
from pathlib import Path
# Setup paths
current_dir = Path(__file__).parent
kb_path = current_dir / "knowledge-base"
src_path = current_dir / "src"
sys.path.insert(0, str(kb_path))
sys.path.insert(0, str(src_path))
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)
def test_integration():
"""Test the full integration like the new MCP server does"""
logger.info("🧠 Testing LLM Statistical Advisor Integration (v3.0)...")
try:
# Step 1: Import and create search engine (like MCP server does)
from kb_search import create_search_engine
knowledge_base_dir = str(kb_path)
gazetteer_path = kb_path / "geo-db" / "geography.db"
logger.info(f"Creating search engine with:")
logger.info(f" Knowledge base: {knowledge_base_dir}")
logger.info(f" Gazetteer: {gazetteer_path} (exists: {gazetteer_path.exists()})")
search_engine = create_search_engine(
knowledge_base_dir=knowledge_base_dir,
gazetteer_db_path=str(gazetteer_path) if gazetteer_path.exists() else None
)
logger.info("✅ Search engine created")
# Step 2: Import and create LLM statistical advisor
from llm_statistical_advisor import LLMStatisticalAdvisor
advisor = LLMStatisticalAdvisor()
logger.info("✅ LLM statistical advisor created")
# Step 3: Wire them together (like MCP server does)
geo_parser = getattr(search_engine, 'geo_parser', None)
variable_search = getattr(search_engine, 'variable_search', None)
methodology_search = getattr(search_engine, 'methodology_search', None)
logger.info(f"Tool extraction:")
logger.info(f" geo_parser: {'✅' if geo_parser else '❌'}")
logger.info(f" variable_search: {'✅' if variable_search else '❌'}")
logger.info(f" methodology_search: {'✅' if methodology_search else '❌'}")
advisor.set_tools(
geo_parser=geo_parser,
variable_search=variable_search,
methodology_search=methodology_search
)
logger.info("✅ Tools wired to LLM advisor")
# Step 4: Test the wiring
has_geo = advisor.geo_parser is not None
has_var = advisor.variable_search is not None
has_method = advisor.methodology_search is not None
logger.info(f"Final wiring check:")
logger.info(f" advisor.geo_parser: {'✅' if has_geo else '❌'}")
logger.info(f" advisor.variable_search: {'✅' if has_var else '❌'}")
logger.info(f" advisor.methodology_search: {'✅' if has_method else '❌'}")
if has_geo and has_var:
logger.info("🎯 INTEGRATION SUCCESS! LLM advisor is wired and ready.")
# Step 5: Quick functional test
logger.info("Testing LLM consultation...")
try:
# Use existing consult method with GeographicContext
from geographic_parsing import GeographicContext
geo_context = GeographicContext(
location_mentioned=True,
location_text="Austin, Texas",
geography_level="place",
confidence=0.9
)
result = advisor.consult("What variables for median income analysis?", geo_context=geo_context)
logger.info(f"Consultation result: {type(result).__name__} with confidence: {result.confidence:.1%}")
logger.info("✅ LLM functional test passed")
except Exception as e:
logger.warning(f"⚠️ LLM functional test failed: {e}")
return True
else:
logger.error("❌ INTEGRATION FAILED: Missing critical tools (geo_parser and variable_search)")
return False
except Exception as e:
logger.error(f"❌ Integration test failed: {e}")
import traceback
traceback.print_exc()
return False
if __name__ == "__main__":
success = test_integration()
if success:
print("\n🎉 READY TO ROCK! The LLM statistical advisor integration is working.")
print("You can now run the new MCP server (v3.0) and use the LLM statistical consultation tool.")
else:
print("\n❌ Integration issues detected. Check the logs above.")
sys.exit(0 if success else 1)
