[FEA] Regex: Provide option for ^ and $ to only match beginning and end of input string

[andygrove]

Is your feature request related to a problem? Please describe.
I would like the ability to use contains_re in a way that is compatible with Java and Python's default handling of multi-line inputs. The default cuDF behavior is that ^ will match at the beginning of the input and also match after each newline in the input. The default Python and Java/Spark behavior is to only match at the start of the input.

Python default behavior:

>>> print(re.compile('^A').search("A\nB"))
<re.Match object; span=(0, 1), match='A'>
>>> print(re.compile('^B').search("A\nB"))
None

cuDF default behavior:

>>> print(cudf.Series(['A\nB']).str.contains('^A'))
0    True
dtype: bool
>>> print(cudf.Series(['A\nB']).str.contains('^B'))
0    True
dtype: bool

Describe the solution you'd like
I would like to be able to specify how ^ and $ behave with multi-line inputs.

Describe alternatives you've considered
In the RAPIDS Accelerator for Apache Spark, we could potentially parse the regex and translate ^ and $ to \A and \Z.

Additional context
This requirement is being driven by NVIDIA/spark-rapids#3797

[davidwendt]

Python regex library includes a MULTILINE flag that can be passed on various regex functions (e.g. regex.match) to specify how the ^ and $ instructions are interpretted.

This feature looks like it may be possible to add into libcudf by adding a similar flags parameter to the cudf::strings APIs (and Python cudf strings APIs) that accept regex patterns.