expression: Improve the performance of str_to_date

[JaySon-Huang]

What problem does this PR solve?

Issue Number: related to #24928

Problem Summary:
While parsing parts for StrToDate, we first match the length of the digit by regex then parsing the number by strconv.ParseUint. The regex matching is costly and somehow useless, we can rewrite parseDigits to reduce the cost of parsing.

What is changed and how it works?

Depending on #25386, should merge that PR first.

Use parseNDigits instead of matching length by oneOrTwoDigitRegex then parsing number by strconv.ParseUint.

Test the performance by go test -run=^$ -bench=BenchmarkStrToDate -count=5 -v

>  benchstat old.txt new.txt
name                                        old time/op  new time/op  delta
StrToDate/strToDate_yyyyMMdd_hhmmss_ffff-8  1.79µs ± 4%  0.85µs ± 3%  -52.55%  (p=0.008 n=5+5)
StrToDate/strToDate_%r_ddMMyyyy-8           1.21µs ±15%  0.47µs ± 1%  -61.23%  (p=0.008 n=5+5)
StrToDate/strToDate_%T_ddMMyyyy-8           1.18µs ± 2%  0.48µs ± 8%  -59.56%  (p=0.008 n=5+5)

Related changes

• PR to update pingcap/docs/pingcap/docs-cn:
• Need to cherry-pick to the release branch

Check List

Tests

• Unit test

Side effects

• N/A

Release note

• Improve the performance of str_to_date

[JaySon-Huang]

/run-check_dev_2

[wshwsh12]

This pr is depending on #25386, so I cancel the review request now. After that pr merged, feel free to re-request review again.

[JaySon-Huang]

@wshwsh12 PTAL if you are free

[wshwsh12]

Seems you have changed the behavior of dayOfYearThreeDigits.
Please add some test case that len(input) < 3, and compare the result with mysql.

[JaySon-Huang]

MySQL declares that "%j" should be "Day of year (001..366)"[1]. But actually, it accepts a number that is up to three digits[2]. (The same with "%d", "Day of the month, numeric (00..31)" but still accept one digit as the day).
TiDB add parsing function dayOfYearThreeDigits for '%j', but actually it doesn't support parsing format with '%j' [3]. So it is not easy to add cases for testing this function by testing StrToDate.
Instead of adding cases for 'dayOfYearThreeDigits', I think I will add some test cases for parseNDigits, later.

• [1] https://dev.mysql.com/doc/refman/8.0/en/date-and-time-functions.html#function_date-format
• [2] Some result for '%j' in MySQL 5.7

  > select date, format, str_to_date(date, format) as str_to_date from a;
  +-----------+--------+----------------------------+
  | date      | format | str_to_date                |
  +-----------+--------+----------------------------+
  | 015 2021  | %j %Y  | 2021-01-15 00:00:00.000000 |
  | 15 2021   | %j %Y  | 2021-01-15 00:00:00.000000 |
  | 9 2021    | %j %Y  | 2021-01-09 00:00:00.000000 |
  | 900 2021  | %j %Y  | 2023-06-19 00:00:00.000000 |
  | 999 2021  | %j %Y  | 2023-09-26 00:00:00.000000 |
  | 1000 2021 | %j %Y  | 2000-04-09 00:00:00.000000 | -- don't know why, need further investigation
  | 9999 2021 | %j %Y  | 2011-09-26 00:00:00.000000 | -- don't know why, need further investigation
  +-----------+--------+----------------------------+
  > select version();
  +-------------------------+
  | version()               |
  +-------------------------+
  | 5.7.33-0ubuntu0.16.04.1 |
  +-------------------------+
  

• [3]
  

  tidb/types/time.go

  Lines 2752 to 2756 in 2093349

  	// Key of the ctx is the format char, such as `%j` `%p` and so on.
  	if yearOfDay, ok := ctx["%j"]; ok {
  	// TODO: Implement the function that converts day of year to yy:mm:dd.
  	_ = yearOfDay
  	}

[JaySon-Huang]

Instead of adding cases for 'dayOfYearThreeDigits', I think I will add some test cases for parseNDigits

Find it hard to add test cases for the private method parseNDigits, add more test cases for StrToDate instead.

[wshwsh12]

LGTM

[ti-chi-bot]

[REVIEW NOTIFICATION]

This pull request has been approved by:

• guo-shaoge
• wshwsh12

To complete the pull request process, please ask the reviewers in the list to review by filling /cc @reviewer in the comment.
After your PR has acquired the required number of LGTMs, you can assign this pull request to the committer in the list by filling /assign @committer in the comment to help you merge this pull request.

The full list of commands accepted by this bot can be found here.

Reviewer can indicate their review by submitting an approval review.
Reviewer can cancel approval by submitting a request changes review.

[JaySon-Huang]

@wshwsh12 Could you find another reviewer to continue to review this PR when they have spare time? It has not been active for several weeks.

[wshwsh12]

/cc @guo-shaoge

[guo-shaoge]

Will review this today.

[JaySon-Huang]

/merge

[ti-chi-bot]

This pull request has been accepted and is ready to merge.

Commit hash: f7f33e8

[ti-chi-bot]

@JaySon-Huang: Your PR was out of date, I have automatically updated it for you.

At the same time I will also trigger all tests for you:

/run-all-tests

If the CI test fails, you just re-trigger the test that failed and the bot will merge the PR for you after the CI passes.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the ti-community-infra/tichi repository.

[JaySon-Huang]

/run-unit-test

[JaySon-Huang]

/run_check_dev_2

[JaySon-Huang]

/run-check_dev_2