BigQuery: Option to have L054 ignore GROUP BY/ORDER BY clauses in aggegate functions (e.g. ARRAY_AGG())

Search before asking

• I searched the issues and found no similar issues.

What Happened

In this query, an L054 (Inconsistent column references in ‘GROUP BY/ORDER BY’ clauses.) is detected at the order by within the array_agg array function on line 4. The L054 documentation says “ORDER BY clauses from WINDOW clauses are ignored by this rule”, so it probably doesn’t take this odd order by inside an array function into account.

    select
        user_uuid,
        max(started_at) as updated_at,
        to_json_string(array_agg(product_id order by started_at desc)) as products

    from purchased
    group by 1

Expected Behaviour

L054 should probably ignore order by in array functions.

Observed Behaviour

See “What Happened”.

How to reproduce

stick that query in a file 😃

Dialect

bigquery

Version

1.2.1

Configuration

[sqlfluff]
dialect = bigquery
templater = dbt

# https://docs.sqlfluff.com/en/stable/rules.html
# L016: Line is too long
# L034: Use wildcards then simple targets before calculations and aggregates in select statements
exclude_rules = L016,L034

[sqlfluff:rules:L014]
extended_capitalisation_policy = lower
ignore_words = _TABLE_SUFFIX

[sqlfluff:rules:L064]
preferred_quoted_literal_style = single_quotes

Are you willing to work on and submit a PR to address the issue?

• Yes I am willing to submit a PR!

Code of Conduct

• I agree to follow this project’s Code of Conduct

[tunetheweb]

L054 already skips enforcing this role in windowing functions as noted in the docs: https://docs.sqlfluff.com/en/stable/rules.html#sqlfluff.core.rules.Rule_L054 This is done for precisely the reasons you note.

The issue here is an array function is technically not a windowing function, so that skipping logic needs to be increased to include that.

[remilepriol]

Note that this is not specific to BigQuery. Many engines support array_agg (presto, postgre, snowflake …)