tree 3eff9cfdca8a8ccbd269247c0e9288d34c5e8a0c
parent 1757ba2dede35bf9e2dd09e7527e593307a70d4f
author Florian Müller <florian@tomueller.de> 1617278336 -0400
committer Andrew Lamb <andrew@nerdnetworks.org> 1617278336 -0400

ARROW-10354: [Rust][DataFusion] regexp_extract function to select regex groups from strings

Adds a regexp_extract compute kernel to select a substring based on a regular expression.

Some things I did that I may be doing wrong:

* I exposed `GenericStringBuilder`
* I build the resulting Array using a builder - this looks quite different from e.g. the substring kernel. Should I change it accordingly, e.g. because of performance considerations?
* In order to apply the new function in datafusion, I did not see a better solution than to handle the pattern string as `StringArray` and take the first record to compile the regex pattern from it and apply it to all values. Is there a way to define that an argument has to be a literal/scalar and cannot be filled by e.g. another column? I consider my current implementation quite error prone and would like to make this a bit more robust.

Closes #9428 from sweb/ARROW-10354/regexp_extract

Authored-by: Florian Müller <florian@tomueller.de>
Signed-off-by: Andrew Lamb <andrew@nerdnetworks.org>