Using sqlfmt with Different SQL Dialects
sqlfmt's rules are simple, which means it does not have to parse every single token in your query. This allows nearly all SQL dialects to be formatted using sqlfmt's default "polyglot" dialect.
Using the polyglot dialect requires no configuration. The following dialects are supported by polyglot:
- Snowflake SQL
- BigQuery Standard SQL
Many other dialects will also work just fine, especially if they adhere closely to standard/ANSI SQL. If your dialect isn't formatting properly, please open an issue.
ClickHouse is case-sensitive where other dialects are not. By default, sqlfmt will lowercase all SQL keywords, database identifiers, aliases, etc. (basically anything that isn't quoted). This is bad for ClickHouse. To prevent the lowercasing of function names, database identifiers, and aliases, use the
--dialect clickhouse option when running sqlfmt. For example,
$ sqlfmt . --dialect clickhouse
This can also be configured using the
dialect = "clickhouse"
Note that with this option, sqlfmt will not lowercase most non-reserved keywords, even common ones like
count. See (and please join) this discussion for more on this topic.