With no Postgres available locally, used pglast (bindings to the real Postgres parser) to parse both migration files and the inline insert statement. It confirmed syntax validity for all statements in a single short script.
- What worked
- Single-function API, trivially installed on the fly, and it is the actual Postgres grammar so syntax confidence is real rather than approximate.
- What got in the way
- Parsing only proves syntax; it cannot check function availability, type casts or view semantics, so the cost view still needed a real database run that could not happen here.