Same Socrata API, Four Incompatible Schemas: Lessons from Scraping State License Data

Four U.S. states — New York, Connecticut, Colorado, and Texas — expose professional licensing registries through the same Socrata SODA API standard, but each uses entirely different column names, data structures, and filtering logic. Cosmetology licenses are not stored in dedicated datasets; they are buried within large multi-profession tables alongside electricians, dentists, and notaries, requiring precise server-side filtering before pagination. Texas lacks a license status column entirely, meaning any active-only filter applied there silently returns no results rather than raising an error. Colorado stores licensee names in separate first and last name columns, unlike every other supported state, requiring the full name to be assembled at read time. A developer building a unified scraper across all four states must implement per-state configuration objects that map each state's unique column names and filter syntax to a single canonical output format.