SQL to CSV

Reads INSERT statements back into CSV: column names from the statement, values un-escaped, and NULL turned back into an empty field. Both one-statement-per-row and batched forms are handled.

This is the reverse of an export, and it comes up more than you would expect: a dump file, a migration someone wrote by hand, a set of statements pasted into a ticket. Getting them into a spreadsheet means parsing the tuples — and the tuples contain commas inside quoted values, which is exactly the case a split on commas gets wrong. What it does not do is parse SQL properly. It looks for `INSERT INTO table (columns) VALUES` and reads the tuples after it, which covers what generators emit and what people write by hand. A statement with an expression in it — `VALUES (1, NOW())` — comes through as the literal text `NOW()`, because evaluating it would need a database. Anything more complicated needs a real parser, and the page says so rather than producing plausible nonsense. The columns come from the first statement. If a dump has statements with different column lists, only the first is used as the header, and that is worth checking before trusting the result.