Cell magics starting with single '%' are not ignored

[danielsparing]

For better or worse, some systems use single '%' headers for cell magics, notably Databricks, e.g. %sql and %r.

I successfully used nbQA with %sql cells for a long time (they magically just get ignored), however, with %r cells, nbqa-ruff-check and nbqa-ruff-format has a false positive at the below example (in R, <- is the assignment operator):

nbqa-ruff-check..........................................................Failed
- hook id: nbqa-ruff-check
- exit code: 1

r\install sf.ipynb:cell_3:2:1: B015 Pointless comparison. Did you mean to assign a value? Otherwise, prepend `assert` or remove it.
   |
 9 | # %%NBQA-CELL-SEPe5ddb4
10 | hash(0x96DAFA9D)
11 | var <- 1 + 1
   | ^^^^^^^^^^^^ B015
   |

Found 1 error.

nbqa-ruff-format.........................................................Failed
- hook id: nbqa-ruff-format
- files were modified by this hook

1 file reformatted

and the reformat is:

-    "var <- 1 + 1"
+    "var < -1 + 1"

(if I use %%r, they are ignored -- but %r is the right cell magic in this env.)

I wonder if there's a way to work around this?

[danielsparing]

I guess what happens here is SQL code is skipped for invalid syntax, while R is just too close to Python that it gets worked on.

[MarcoGorelli]

thanks for the report, fixes welcome