Implemented UDF Filters

[lamarrr]

Description

Checklist

• [ ] I am familiar with the Contributing Guidelines.
• [ ] New or existing tests cover these changes.
• [ ] The documentation is up to date with these changes.

[copy-pr-bot[bot]]

Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually.

Contributors can view more details about this message here.

[lamarrr]

/ok to test 7f97737

[lamarrr]

I'll follow up with benchmarks once this is merged

[lamarrr]

std::vector<std::unique_ptr<column>> filter(
  std::vector<column_view> const& columns,
  std::string const& predicate_udf,
  bool is_ptx,
  std::optional<void*> user_data             = std::nullopt,
  std::optional<std::vector<bool>> copy_mask = std::nullopt,
  rmm::cuda_stream_view stream               = cudf::get_default_stream(),
  rmm::device_async_resource_ref mr          = cudf::get_current_device_resource_ref());

@davidwendt do you think this should be consistent with other APIs and use a table_view instead?

i.e:

std::unique_ptr<table> filter(
  table_view const& table,
  std::string const& predicate_udf,
  bool is_ptx,
  std::optional<void*> user_data             = std::nullopt,
  std::optional<std::vector<bool>> copy_mask = std::nullopt,
  rmm::cuda_stream_view stream               = cudf::get_default_stream(),
  rmm::device_async_resource_ref mr          = cudf::get_current_device_resource_ref());

[davidwendt]

do you think this should be consistent with other APIs and use a table_view instead?

Yes, a table_view makes sense. Also since all the columns have to be the same size.

[lamarrr]

Yes, a table_view makes sense. Also since all the columns have to be the same size.

Just remembered it won't work since it can have scalars as input as well. I think we can proceed in its current state and make necessary changes later.

[lamarrr]

CI is taking forever

[lamarrr]

/merge

[lamarrr]

/merge