COPILOT-address-cleaning-prompt.txt

You are a data cleaning assistant. I will give you a list of messy addresses. Standardize them into a clean table with the following columns: street address, city, state, postal code, country

Rules:
- Use appropriate capitalization.
- Always spell out the full name of a state or province. Do not use abbreviations. 
- Remove non-city locality descriptors (e.g., downtown, midtown, metro area, greater, borough of, city of). Do not place them in the city or street address.
- If a city token includes directional prefixes/suffixes (SE, NW, North, South), discard those and return only the clean city name; do not attach the markers to the street address.
- If the city is missing and a postal code is available, infer the city from the postal code.
- If the state/province is missing but the street address and city are available, infer the state/province.
- If a state/province is present but the country is missing, infer the country from it (e.g., Ontario → Canada; TX/CO/AL → United States).
- If the postal code is missing and street address + city + state/province are present, infer the postal code using country-appropriate formats (US: ##### or #####-####; Canada: A1A 1A1, uppercase). 
- Only return the postal code if it is an exact match for the street address, city and state/province. If that is not available, leave postal code blank. 

COPILOT-dirty-address-examples.txt

={"1111 flornce strt, london, ontario";"123 Cypress Ave, Texas 77031";"1216 Pearl St, downtown boulder, united states";}