[crypto] add blake2b and blake2s functions

[jimexist]

Which issue does this PR close?

add blake2b and blake2s functions, as provided in https://docs.rs/blake2/0.9.2/blake2/

must be merged after #1090 is merged.

Closes #1093
Closes #1094

Rationale for this change

blake2s is a better alternative for Md5 and blake2b is more secure, modern, etc.

What changes are included in this PR?

see https://en.wikipedia.org/wiki/BLAKE_(hash_function)

Are there any user-facing changes?

❯ cargo run -p datafusion-cli
    Finished dev [unoptimized + debuginfo] target(s) in 0.18s
     Running `target/debug/datafusion-cli`
DataFusion CLI v5.1.0-SNAPSHOT

> select blake2b(''), blake2s('');
+----------------------------------------------------------------------------------------------------------------------------------+------------------------------------------------------------------+
| blake2b(Utf8(""))                                                                                                                | blake2s(Utf8(""))                                                |
+----------------------------------------------------------------------------------------------------------------------------------+------------------------------------------------------------------+
| 786a02f742015903c6c6fd852552d272912f4740e15847618a86e217f71f5419d25e1031afee585313896444934eb04b903a685b1448b755d56f701afe9be2ce | 69217a3079908094e11121d042354a7c1f55b6482ca1a51e1b250dfd1ed0eef9 |
+----------------------------------------------------------------------------------------------------------------------------------+------------------------------------------------------------------+
1 row in set. Query took 0.007 seconds.
>

[jimexist]

so strange that i cannot reproduce the errors in CI

[xudong963]

so strange that i cannot reproduce the errors in CI

Maybe CI doesn't run with the newest code

[alamb]

💯 for making these functions optional

[alamb]

🤔 there seems to be a build failures but otherwise this looks good to me

https://github.com/apache/arrow-datafusion/pull/1081/checks?check_run_id=3840276390

error[E0004]: non-exhaustive patterns: `&Blake2b` and `&Blake2s` not covered
   --> /home/runner/work/arrow-datafusion/arrow-datafusion/datafusion/src/physical_plan/functions.rs:319:15
    |
174 | / pub enum BuiltinScalarFunction {
175 | |     // math functions
176 | |     /// abs
177 | |     Abs,
...   |
275 | |     Blake2b,
    | |     ------- not covered
276 | |     /// Blake2s
277 | |     Blake2s,
    | |     ------- not covered
...   |
305 | |     RegexpMatch,
306 | | }
    | |_- `BuiltinScalarFunction` defined here
...
319 |           match self {
    |                 ^^^^ patterns `&Blake2b` and `&Blake2s` not covered
    |
    = help: ensure that all possible cases are being handled, possibly by adding wildcards or more match arms
    = note: the matched value is of type `&BuiltinScalarFunction`

FWIW I looked at postgres and what it does is rather than having individual functions for different hashing algorithms, it appears to have a single digest function that takes the requested algorithm as a string parameter

F.26.1.1. Digest()
digest(data text, type text) returns bytea
digest(data bytea, type text) returns bytea

https://www.postgresql.org/docs/current/pgcrypto.html

[houqp]

so strange that i cannot reproduce the errors in CI

@jimexist are you using a rust toolchain with the same version that's being used in our CI?

[jimexist]

🤔 there seems to be a build failures but otherwise this looks good to me

https://github.com/apache/arrow-datafusion/pull/1081/checks?check_run_id=3840276390

error[E0004]: non-exhaustive patterns: `&Blake2b` and `&Blake2s` not covered
   --> /home/runner/work/arrow-datafusion/arrow-datafusion/datafusion/src/physical_plan/functions.rs:319:15
    |
174 | / pub enum BuiltinScalarFunction {
175 | |     // math functions
176 | |     /// abs
177 | |     Abs,
...   |
275 | |     Blake2b,
    | |     ------- not covered
276 | |     /// Blake2s
277 | |     Blake2s,
    | |     ------- not covered
...   |
305 | |     RegexpMatch,
306 | | }
    | |_- `BuiltinScalarFunction` defined here
...
319 |           match self {
    |                 ^^^^ patterns `&Blake2b` and `&Blake2s` not covered
    |
    = help: ensure that all possible cases are being handled, possibly by adding wildcards or more match arms
    = note: the matched value is of type `&BuiltinScalarFunction`

FWIW I looked at postgres and what it does is rather than having individual functions for different hashing algorithms, it appears to have a single digest function that takes the requested algorithm as a string parameter

F.26.1.1. Digest()
digest(data text, type text) returns bytea
digest(data bytea, type text) returns bytea

https://www.postgresql.org/docs/current/pgcrypto.html

let's address this in #1090 so merge that first

[alamb]

looks good to me

[alamb]

Thanks @jimexist