SQL Formatter

Recognises common clause keywords — SELECT, FROM, WHERE, JOIN, GROUP BY, ORDER BY, and their kin — uppercases them, and breaks the query onto one line per clause with comma-separated lists indented underneath.

A reflow, not a validator

This is a tokenizer, not a parser for any specific SQL dialect. It does not check that the query is syntactically correct for any engine. Anything whose content matters — string literals, "double-quoted", T-SQL [bracketed] and MySQL backtick identifiers, PostgreSQL $$…$$ bodies, and comments — is copied through byte-for-byte, never re-spaced or re-cased.

Comments stay where they can't swallow code

A -- line comment runs to the end of its line, so the formatter always ends the output line after one. Pulling the next clause up onto that line would comment it out and silently change what the query does.

A comma inside a function call, like COUNT(a, b), stays on the same line as the call — only top-level commas (a SELECT column list, an INSERT value list) break onto new lines, tracked by parenthesis depth.