[FEA] Add support for SHA256 and SHA512 to cudf::hash

[MikeChenfu]

Is your feature request related to a problem? Please describe.
Hello, currently I am thinking if the cudf supports more hash functions like hashlib.sha256, hashlib.sha512. Thanks for the consideration.

Describe the solution you'd like
df['sha256'] = df['h'].hash_func(method='sha256')

Describe alternatives you've considered

import pandas as pd
df['sha256'] = df['h'].apply(lambda x: (hashlib.sha256(str(x).encode()).hexdigest().upper()))

Additional context
Add any other context, code examples, or references to existing implementations about the feature request here.

[beckernick]

Hi @MikeChenfu , thanks for the request. Does the existing Series.hash_values (murmur3) in Python work for your use case?

[MikeChenfu]

Hi @beckernick, thanks for the reply. Currently, murmur3 is not the choice for our use case. It would be great if we have multiple choices for the hash function.

[shwina]

Likely will need support for this at the C++ (libcudf) layer. cc: @jrhemstad @harrism

[jrhemstad]

We do have some other hash functions other than murmur3 that aren't exposed to Python:

cudf/cpp/include/cudf/types.hpp

Lines 333 to 338 in 3ee264c

	enum class hash_id {
	HASH_IDENTITY = 0, ///< Identity hash function that simply returns the key to be hashed
	HASH_MURMUR3, ///< Murmur3 hash function
	HASH_MD5, ///< MD5 hash function
	HASH_SERIAL_MURMUR3, ///< Serial Murmur3 hash function
	HASH_SPARK_MURMUR3 ///< Spark Murmur3 hash function

[MikeChenfu]

@shwina @jrhemstad Thanks for the information. Can we have the sha256 into the cudf lib?

[davidwendt]

Reference #6020

[github-actions]

This issue has been labeled inactive-90d due to no recent activity in the past 90 days. Please close this issue if no further response or action is needed. Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed.

[bdice]

Status update on this issue: I have been working on several refactors, fixes, and performance enhancements for libcudf's hashing functionality. I expect this feature to land in the 22.06 release with PR #9215.