Skip to content

sqruff

Default severity: ignore

What it does

Reports what the core lint rules of sqruff find in a query, in the project's dialect. The message starts with the sqruff rule code, such as AM06, and the underline is the part of the query it is about.

[tool.sqlty.rules]
sqruff = "warn"
from sqlty import SQL

# error: [sqruff] "AM06: Inconsistent column references in 'GROUP BY/ORDER BY' clauses."
query = SQL[tuple[str, int]]("""--sql
SELECT name, COUNT(*) AS n
FROM users
GROUP BY 1
ORDER BY name""")
Open in playground

Every query sqlty checks is linted, whatever its string looks like: escape sequences are read, and the underline lands on the source they are written in. Placeholders (%s, %(name)s, ?) are read as names, and a finding on one is dropped. A query of over 4,096 words and symbols is not linted: sqruff takes seconds on one that long.

Why is this bad?

sqruff's core rules catch queries that run but read badly or mean something other than they seem: a DISTINCT that looks like a function, an alias never used, an ORDER BY that mixes positions and names.

How to fix

Follow the rule: the sqruff rules page explains each code.

Configuration

Off by default, as these rules judge style rather than types. Turn it on per project:

[tool.sqlty.rules]
sqruff = "warn"

sqruff's layout rules are left out: sqlty format lays out queries. sqruff reads no configuration file here, so the rules are the same on every machine. # sqlty: ignore[sqruff] on the line a query's string starts silences all its findings.