Automate sqllogictest for String, LargeString and StringView behavior

[goldmedal]

Which issue does this PR close?

Part of #12415

Rationale for this change

I am implementing a testing framework for string types to ensure consistent behavior across different string types. I have extracted test cases from datafusion/sqllogictest/test_files/string_view.slt, which contains many good tests for string types.

Currently, I haven’t moved all the tests from string_view.slt because I don’t want to make this PR too large (I want to confirm if I’m on the right way 🤔). For now, I’ve only moved the following case:

• column comparison as filters
• column comparison
• column to StringView scalar comparison
• column to String scalar
• column to LargeString scalar
• substr function
• distinct aggregate
• STARTS_WITH function
• TRANSLATE function
• REGEXP_REPLACE function
• Initcap function

Some TODO items should be finished in the follow-up PR

• The remaining tests in the string_view.slt
  • LIKE/ILIKE
  • SUBSTR planing
  • ASCII
  • BTRIM
  • LTRIM
  • RTRIM
  • CHARACTER_LENGTH
  • CONCAT
  • CONTAINS
  • ENDS_WITH
  • LEVENSHTEIN
  • LOWER
  • UPPER
  • LPAD
  • OCTET_LENGTH
  • OVERLAY
  • REGEXP_LIKE
  • REGEXP_MATCH
  • REPEAT
  • REPLACE
  • REVERSE
  • RIGHT
  • LEFT
  • RPAD
  • SPLIT_PART
  • STRPOS
  • SUBSTRINDEX
  • FIND_IN_SET
  • || operator
  • ~ operator (regex match)
  • ~* operator (regex match case-insensitive)
  • !~~ operator (not like match)
  • !~~* operator (not like match case-insensitive)
  • Move StringView specific tests to string/string_view.slt
• Move string literal tests from function.slt to string/string_literal.rs

What changes are included in this PR?

Refer to #12433, I made some changes. I regroup the test case by the data type. The file structure is

string/
    - init_data.slt.part        // generate the testing data
    - string_query.slt.part     // the sharing tests for all string type
    - string.slt                // the entrypoint for string type
    - large_string.slt          // the entrypoint for large_string type
    - string_view.slt           // the entrypoint for string_view type and the string_view specific tests
    - string_literal.slt        // the tests for string literal

Any entry point file should include init_data.slt.part and string_query.slt.part.

For planning tests for StringView (e.g. EXPLAIN ...), I moved to the string/string_view.slt because it's only used to assert the planning behavior of StringView.

Are these changes tested?

Refactor and enhance the original tests.

Are there any user-facing changes?

no

[goldmedal]

Because we started to implement some ASCII-faster paths, I think we should always consider testing both ASCII-only and Unicode cases.

[alamb]

Thank you so much @goldmedal

I really like this structured, and I think the tests are super well commented and easy to follow. Really nice ❤️ 🏆

@2010YOUY01 do you perhaps have time to take a look at this PR as well?

cc @tshauck

[alamb]

❤️

[2010YOUY01]

This is well organized 👍🏼

I have one minor suggestion: we should add a separate README under test_files/string/ to explain the structure

[goldmedal]

This is well organized 👍🏼

I have one minor suggestion: we should add a separate README under test_files/string/ to explain the structure

Nice idea! I have added the README for them. Thanks for the suggestion 🙇

[alamb]

THANK YOU SO MUCH @goldmedal and @2010YOUY01 -- this looks really great

[goldmedal]

Thanks @alamb @2010YOUY01 🙇‍♂️