GH-49186: [R] Support dplyr::filter_out() in Arrow dplyr backend

[larry77]

Rationale for this change

New function in dplyr not yet implemented in Arrow

What changes are included in this PR?

This PR adds support for dplyr::filter_out() in the Arrow R dplyr backend.

The implementation reuses the existing filter() machinery and extends
set_filters() with an exclude flag. When exclude = TRUE, the predicate
is transformed to match dplyr semantics (drop rows where predicate is TRUE,
keep rows where predicate is FALSE or NA).

Multiple filter_out() predicates are combined before exclusion so that
filter_out(a, b) matches dplyr semantics (i.e. drop rows where a & b is TRUE).

This works for arrow_table(), RecordBatchReader, and open_dataset(), and
preserves lazy evaluation for larger-than-memory datasets.

Tests are added to verify basic behavior, NA handling, and multiple predicates.

Note: local test run hits a with_language() locale issue ('.cache' not found),
which appears environment-specific and unrelated to these changes.

Are these changes tested?

Yes

Are there any user-facing changes?

Just the new function

• GitHub Issue: [R] Support dplyr::filter_out() in Arrow dplyr backend #49257
• GitHub Issue: [R]: add support to the new dplyr verb filter_out #49186

[github-actions]

Thanks for opening a pull request!

If this is not a minor PR. Could you open an issue for this pull request on GitHub? https://github.com/apache/arrow/issues/new/choose

Opening GitHub issues ahead of time contributes to the Openness of the Apache Arrow project.

Then could you also rename the pull request title in the following format?

GH-${GITHUB_ISSUE_ID}: [${COMPONENT}] ${SUMMARY}

or

MINOR: [${COMPONENT}] ${SUMMARY}

See also:

• Other pull requests
• Contribution Guidelines - Contributing Overview

[github-actions]

⚠️ GitHub issue #49257 has been automatically assigned in GitHub to PR creator.

[thisisnic]

Thanks for making this PR @larry77! This is looking good, a few notes from me:

• Would you mind adding filter_out to supported_dplyr_methods in arrow/r/R/arrow-package.R?
• Tests look good; I went to see if the core dplyr ones did much different but I think those cases you included cover everything.
• The code in the body of filter_out() appears to be copied from filter(); could we instead extract out that code into its own function and then use it in both?

[github-actions]

⚠️ GitHub issue #49186 has been automatically assigned in GitHub to PR creator.

[larry77]

Thanks for making this PR @larry77! This is looking good, a few notes from me:

* Would you mind adding `filter_out` to `supported_dplyr_methods` in `arrow/r/R/arrow-package.R`?

* Tests look good; I went to see if the core dplyr ones did much different but I think those cases you included cover everything.

* The code in the body of `filter_out()` appears to be copied from `filter()`; could we instead extract out that code into its own function and then use it in both?

Thanks for the review!

– Added filter_out to supported_dplyr_methods in r/R/arrow-package.R.
– Refactored filter() and filter_out() to share a common implementation (no duplicated body).

Local tests pass aside from the existing locale-related .cache flake in the “missing column” test, which seems environment-specific.