expression: fix builtin 'CharLength' for binary string input

[spongedu]

What problem does this PR solve?

Currently in TiDB, builtin CharLength's behavior is not consistent with MySQL when deal with binary strings.

In MySQL:

mysql> select char_length(binary("数 据 库"));
+------------------------------------+
| char_length(binary("数 据 库"))    |
+------------------------------------+
|                                 11 |
+------------------------------------+
1 row in set (0.00 sec)

While in TiDB:

tidb> select char_length(binary("数 据 库"));
+------------------------------------+
| char_length(binary("数 据 库"))    |
+------------------------------------+
|                                  5 |
+------------------------------------+
1 row in set (0.00 sec)

What is changed and how it works?

take special care for binary strings when create builtinSig

Check List

Tests

• Unit test
• Integration test

Code changes

Side effects

Related changes

[shenli]

LGTM

[spongedu]

/run-all-tests

[coocood]

Just len(val) will do.

[spongedu]

ok

[shenli]

/run-all-tests

[coocood]

@spongedu
In this way, If we support more charset in the future, we may need to define a signature for each charset.

Have you considered adding a field to charLengthSig that indicates if it is binary?

[spongedu]

At first sight I think we need only distinguish binary strings and non-binary strings. I'm not sure about this for now. I'll look into it and make some tests today or so.

[winkyao]

LGTM

[zz-jason]

LGTM

[zz-jason]

/run-all-tests