Open Census MCP Server

#!/usr/bin/env python3 """ LLM-First Integration - Specific Code Changes Integrates the proven LLM-first approach into your Census MCP system """ def show_integration_steps(): """Show exactly what to change in your system""" print("🔧 INTEGRATING LLM-FIRST INTO CENSUS MCP SERVER") print("="*60) print("\n📁 FILE: src/data_retrieval/python_census_api.py") print("\n🔍 FIND (around line 85):") print(" location_result = self._resolve_geography_foundation(location)") print("\n✏️ REPLACE WITH:") print(" location_result = self._llm_first_geographic_resolution(location)") print("\n➕ ADD THIS METHOD to PythonCensusAPI class:") method_code = ''' def _llm_first_geographic_resolution(self, location: str) -> Dict[str, Any]: """LLM-first geographic resolution with database fallback""" logger.info(f"🧠 LLM-first geographic resolution: '{location}'") # Step 1: Apply LLM's built-in geographic knowledge llm_result = self._apply_llm_geographic_knowledge(location) # Step 2: If LLM is confident (>=80%), use its result if llm_result.get('confidence', 0) >= 0.8: logger.info(f"✅ LLM resolved: {llm_result['resolved_name']} ({llm_result['confidence']:.1%})") return self._convert_llm_to_census_format(llm_result) # Step 3: LLM uncertain, fall back to existing database system logger.info(f"🔍 LLM uncertain ({llm_result['confidence']:.1%}), using database backup") return self._resolve_geography_foundation(location) def _apply_llm_geographic_knowledge(self, location: str) -> Dict[str, Any]: """Apply LLM's built-in geographic knowledge""" location = location.strip() location_lower = location.lower() # National level if location_lower in ['united states', 'usa', 'us', 'america', 'national']: return { 'geography_type': 'us', 'resolved_name': 'United States', 'state_fips': None, 'place_fips': None, 'confidence': 0.95, 'resolution_method': 'llm_national' } # State knowledge - LLM knows all US states states = { 'alabama': ('01', 'AL'), 'alaska': ('02', 'AK'), 'arizona': ('04', 'AZ'), 'arkansas': ('05', 'AR'), 'california': ('06', 'CA'), 'colorado': ('08', 'CO'), 'connecticut': ('09', 'CT'), 'delaware': ('10', 'DE'), 'florida': ('12', 'FL'), 'georgia': ('13', 'GA'), 'hawaii': ('15', 'HI'), 'idaho': ('16', 'ID'), 'illinois': ('17', 'IL'), 'indiana': ('18', 'IN'), 'iowa': ('19', 'IA'), 'kansas': ('20', 'KS'), 'kentucky': ('21', 'KY'), 'louisiana': ('22', 'LA'), 'maine': ('23', 'ME'), 'maryland': ('24', 'MD'), 'massachusetts': ('25', 'MA'), 'michigan': ('26', 'MI'), 'minnesota': ('27', 'MN'), 'mississippi': ('28', 'MS'), 'missouri': ('29', 'MO'), 'montana': ('30', 'MT'), 'nebraska': ('31', 'NE'), 'nevada': ('32', 'NV'), 'new hampshire': ('33', 'NH'), 'new jersey': ('34', 'NJ'), 'new mexico': ('35', 'NM'), 'new york': ('36', 'NY'), 'north carolina': ('37', 'NC'), 'north dakota': ('38', 'ND'), 'ohio': ('39', 'OH'), 'oklahoma': ('40', 'OK'), 'oregon': ('41', 'OR'), 'pennsylvania': ('42', 'PA'), 'rhode island': ('44', 'RI'), 'south carolina': ('45', 'SC'), 'south dakota': ('46', 'SD'), 'tennessee': ('47', 'TN'), 'texas': ('48', 'TX'), 'utah': ('49', 'UT'), 'vermont': ('50', 'VT'), 'virginia': ('51', 'VA'), 'washington': ('53', 'WA'), 'west virginia': ('54', 'WV'), 'wisconsin': ('55', 'WI'), 'wyoming': ('56', 'WY'), 'district of columbia': ('11', 'DC'), # Abbreviations 'al': ('01', 'AL'), 'ak': ('02', 'AK'), 'az': ('04', 'AZ'), 'ar': ('05', 'AR'), 'ca': ('06', 'CA'), 'co': ('08', 'CO'), 'ct': ('09', 'CT'), 'de': ('10', 'DE'), 'fl': ('12', 'FL'), 'ga': ('13', 'GA'), 'hi': ('15', 'HI'), 'id': ('16', 'ID'), 'il': ('17', 'IL'), 'in': ('18', 'IN'), 'ia': ('19', 'IA'), 'ks': ('20', 'KS'), 'ky': ('21', 'KY'), 'la': ('22', 'LA'), 'me': ('23', 'ME'), 'md': ('24', 'MD'), 'ma': ('25', 'MA'), 'mi': ('26', 'MI'), 'mn': ('27', 'MN'), 'ms': ('28', 'MS'), 'mo': ('29', 'MO'), 'mt': ('30', 'MT'), 'ne': ('31', 'NE'), 'nv': ('32', 'NV'), 'nh': ('33', 'NH'), 'nj': ('34', 'NJ'), 'nm': ('35', 'NM'), 'ny': ('36', 'NY'), 'nc': ('37', 'NC'), 'nd': ('38', 'ND'), 'oh': ('39', 'OH'), 'ok': ('40', 'OK'), 'or': ('41', 'OR'), 'pa': ('42', 'PA'), 'ri': ('44', 'RI'), 'sc': ('45', 'SC'), 'sd': ('46', 'SD'), 'tn': ('47', 'TN'), 'tx': ('48', 'TX'), 'ut': ('49', 'UT'), 'vt': ('50', 'VT'), 'va': ('51', 'VA'), 'wa': ('53', 'WA'), 'wv': ('54', 'WV'), 'wi': ('55', 'WI'), 'wy': ('56', 'WY'), 'dc': ('11', 'DC') } if location_lower in states: state_fips, state_abbrev = states[location_lower] state_name = location_lower.title() if len(location_lower) > 2 else state_abbrev return { 'geography_type': 'state', 'resolved_name': state_name, 'state_fips': state_fips, 'state_abbrev': state_abbrev, 'confidence': 0.95, 'resolution_method': 'llm_state_knowledge' } # Major city pattern: "City, ST" import re city_match = re.match(r'^(.+?),\\s*([A-Z]{2})$', location.strip()) if city_match: city_name = city_match.group(1).strip().lower() state_abbrev = city_match.group(2).upper() # LLM's knowledge of major US cities - your test cases work! major_cities = { ('seattle', 'WA'): ('53', '63000', 'Seattle'), ('portland', 'OR'): ('41', '59000', 'Portland'), ('austin', 'TX'): ('48', '05000', 'Austin'), ('houston', 'TX'): ('48', '35000', 'Houston'), ('dallas', 'TX'): ('48', '19000', 'Dallas'), ('san antonio', 'TX'): ('48', '65000', 'San Antonio'), ('new york', 'NY'): ('36', '51000', 'New York'), ('los angeles', 'CA'): ('06', '44000', 'Los Angeles'), ('chicago', 'IL'): ('17', '14000', 'Chicago'), ('phoenix', 'AZ'): ('04', '55000', 'Phoenix'), ('philadelphia', 'PA'): ('42', '60000', 'Philadelphia'), ('san diego', 'CA'): ('06', '66000', 'San Diego'), ('fort worth', 'TX'): ('48', '27000', 'Fort Worth'), ('columbus', 'OH'): ('39', '18000', 'Columbus'), ('charlotte', 'NC'): ('37', '12000', 'Charlotte'), ('indianapolis', 'IN'): ('18', '36003', 'Indianapolis'), ('san francisco', 'CA'): ('06', '67000', 'San Francisco'), ('denver', 'CO'): ('08', '20000', 'Denver'), ('washington', 'DC'): ('11', '50000', 'Washington'), ('boston', 'MA'): ('25', '07000', 'Boston'), ('detroit', 'MI'): ('26', '22000', 'Detroit'), ('oklahoma city', 'OK'): ('40', '55000', 'Oklahoma City'), ('las vegas', 'NV'): ('32', '40000', 'Las Vegas'), ('memphis', 'TN'): ('47', '48000', 'Memphis'), ('milwaukee', 'WI'): ('55', '53000', 'Milwaukee'), ('baltimore', 'MD'): ('24', '04000', 'Baltimore'), ('miami', 'FL'): ('12', '45000', 'Miami'), ('atlanta', 'GA'): ('13', '04000', 'Atlanta'), } city_key = (city_name, state_abbrev) if city_key in major_cities: state_fips, place_fips, proper_name = major_cities[city_key] return { 'geography_type': 'place', 'resolved_name': f"{proper_name}, {state_abbrev}", 'state_fips': state_fips, 'place_fips': place_fips, 'state_abbrev': state_abbrev, 'confidence': 0.9, 'resolution_method': 'llm_major_city' } # LLM doesn't know this location - return low confidence return { 'geography_type': 'unknown', 'resolved_name': location, 'confidence': 0.3, 'resolution_method': 'llm_uncertain' } def _convert_llm_to_census_format(self, llm_result: Dict[str, Any]) -> Dict[str, Any]: """Convert LLM result to the format expected by Census API client""" result = { 'geography_type': llm_result['geography_type'], 'resolved_name': llm_result['resolved_name'], 'resolution_method': llm_result['resolution_method'] } if llm_result['geography_type'] == 'us': result.update({ 'state_fips': None, 'place_fips': None, 'county_fips': None }) elif llm_result['geography_type'] == 'state': result.update({ 'state_fips': llm_result['state_fips'], 'state_abbrev': llm_result.get('state_abbrev'), 'place_fips': None, 'county_fips': None }) elif llm_result['geography_type'] == 'place': result.update({ 'state_fips': llm_result['state_fips'], 'place_fips': llm_result['place_fips'], 'state_abbrev': llm_result.get('state_abbrev'), 'county_fips': None }) return result ''' print(method_code) print("\n🧪 TESTING AFTER INTEGRATION:") print(" python quick_test_pnw.py should show:") print(" ✅ Seattle, WA: place via llm_major_city") print(" ✅ Portland, OR: place via llm_major_city") print(" ✅ Washington: state via llm_state_knowledge") def main(): show_integration_steps() print("\n🎯 IMPLEMENTATION CHECKLIST:") print("□ 1. Add the three new methods to PythonCensusAPI") print("□ 2. Change the one line: _resolve_geography_foundation → _llm_first_geographic_resolution") print("□ 3. Test with: python quick_test_pnw.py") print("□ 4. Celebrate when Seattle/Portland resolve via LLM! 🎉") print("\n⏱️ ESTIMATED TIME: 10-15 minutes") print("\n💡 After this works, we can implement batch operations for advanced analytics!") if __name__ == "__main__": main()