chore: add SQLFluff to pre-commit hooks

[IvanPiatnishin]

This adds SQLFluff with the PostgreSQL dialect to enforce SQL formatting rules via pre-commit.

[CommanderStorm]

the added CI-step is failing though. This needs to be fixed for mergability ^^
https://results.pre-commit.ci/run/github/105363726/1744040351.aYINARI8SCavxUA6wdemiQ

[sharkAndshark]

@IvanPiatnishin You may want to run commands to fix the formatting errors.

sqlfluff fix --dialect postgres demo/db/initdb/01.sql
sqlfluff fix --dialect postgres demo/db/initdb/03.sql
sqlfluff fix --dialect postgres martin/src/pg/scripts/query_available_function.sql

[CommanderStorm]

There are a few things where you will need to look into the formatting. CI is still failing.
https://results.pre-commit.ci/run/github/105363726/1744269374.byjF0W4QT7WxnLg-J1GjkQ and https://github.com/maplibre/martin/actions/runs/14374564930/job/40304086214?pr=1774

[sharkAndshark]

Or just run this to fix all .sql files

sqlfluff fix --dialect postgres

[nyurik]

Note that ideally this should not just verify the style, but just like cargo fmt it should push changes to the pr

[nyurik]

Note that this PR will not auto-run workflows because @IvanPiatnishin is a new github contributor. A workaround for this may be to create a PR in their own fork -- https://github.com/IvanPiatnishin/martin/compare/IvanPiatnishin:main...IvanPiatnishin:setup-sqlfluff-formatting?expand=1

[nyurik]

On a more substantive notes - sqlfluff currently catches these two lints:

== [tests/fixtures/functions/function_special_characters.sql] FAIL
L:   1 | P:  32 | RF05 | Do not use special characters in identifiers.
                       | [references.special_chars]
== [martin/src/pg/scripts/query_available_tables.sql] FAIL
L:  11 | P:  39 | RF04 | Keywords should not be used as identifiers.

Both are actually OK for us because that's the whole idea of those tests - to see if our code will work correctly when the user does something weird. So we should disable these specific lints in the config somewhere

[CommanderStorm]

I changed a few things about the setup in be89425

• enabled fixing
• re-added an removed comment
• changed which rules we ignore (see below for rationale)

I think RF02 (Unqualified reference), RF04 (Keywords should not be used as identifiers) and RF05 (special characters in identifiers) they are interesting to keep as indicating possible correctness issues.
Not strong indicators, but a smell.

I would like them to be ignored instead:
https://docs.sqlfluff.com/en/3.2.3/configuration/ignoring_configuration.html

LT05 (line lenght) is mostly triggering on comments. => lets ignore this one

[nyurik]

looks good! It seems we might need to remove PG 11 support as it is no longer installing? (it should probably be a separate PR to keep things separate)

[nyurik]

Is it possible to ignore certain rules just for a single file, or maybe even for a specific line in a file? We could use something like that for keywords and special character tests.

[sharkAndshark]

Seems sqlfluff support In-File Configuration Directives, we can do that

to ignore certain rules just for a single file, or maybe even for a specific line in a file

And certain line could be configured by this

-- Ignore all errors
SeLeCt  1 from tBl ;    -- noqa

-- Ignore rule CP02 & rule CP03
SeLeCt  1 from tBl ;    -- noqa: CP02,CP03

-- Ignore all parsing errors
SeLeCt from tBl ;       -- noqa: PRS

[CommanderStorm]

Is it possible to ignore certain rules just for a single file

Yes, see for example this file. We already do so:
https://github.com/maplibre/martin/blob/be894254fb564a4b6c3fa5e13ad3c5900588db2f/tests/fixtures/functions/function_special_characters.sql

[CommanderStorm]

Forgot to ACK this PR