Issue #6882: REGEXP_EXTRACT Capture Groups #6918
Conversation
Add a new variant that takes a list of field names and produces a STRUCT of all the capture groups.
Fix release unused variable.
There was a problem hiding this comment.
Looks good, just had a comment about an optimization we could do
(could be considered not an optimization as well, in the case where you have a really big string and the groups are only a small portion of the string, then keeping this string alive needlessly might be wasteful)
Reuse the input string memory instead of allocating new buffers for each group column.
A good point, but I have to say that having worked on a text shaping engine, you don't usually end up with a lot of filler data, so I think reuse is the way to go. |
Switch to CreateStringPiece to make sure strings get created correctly.
Fix formatting from merge.
Use the new safe casting infrastructure.
CR feedback: Case-insensitivity, validity masks, more tests.
More validity masks.
Fix result output for vsize=2
|
Thanks! LGTM |
Add a new variant that takes a list of field names and produces a STRUCT of all the capture groups.