Repository navigation
regexp_match does not support Utf8View #13357
Description
Describe the bug
regexp match's signature:
// TODO: Native support Utf8View for regexp_match.
TypeSignature::Exact(vec![Utf8, Utf8]),
TypeSignature::Exact(vec![LargeUtf8, LargeUtf8]),
TypeSignature::Exact(vec![Utf8, Utf8, Utf8]),
TypeSignature::Exact(vec![LargeUtf8, LargeUtf8, LargeUtf8]),
This udf should be updated to include support for Utf8View. The type signature might want to be updated to use the new TypeSignature::String (see #13301)
To Reproduce
Add the following to the string view tests in regexp.slt
SELECT regexp_match(str, 'a.*', 'i') from t;
This will require #13348 to be in place to test the binary express equivalent
Expected behavior
Test passes, explain of plan verifies that no type coercion occurred to force Utf8View to Utf8
Additional context
No response
Activity
Omega359 commentedon Nov 11, 2024
A good part of the reason why this hasn't been done in DF is that the corresponding functions in arrow-rs do not support StringView - https://github.com/apache/arrow-rs/blob/28878d375c5bb1c2f5adb46b287e16cea768b6e7/arrow-string/src/regexp.rs#L400
The regexp_is_match* functions were updated as part of apache/arrow-rs#6370 / apache/arrow-rs#6376 by @lamb (ticket and review) and @tlm365 (code update) - we will need to file a similar ticket for regexp_match as well
Omega359 commentedon Nov 11, 2024
This ticket is a duplicate of #11911
Omega359 commentedon Nov 11, 2024
Closing as a duplicate