util: skip always false DNF

[Rustin170506]

What problem does this PR solve?

Issue Number: close #40997

Problem Summary:
For example:

explain
select
  *
from
  t71706696 force index(dt_2)
where
  (
    (
      dt = '20210112'
      and db_id = '62812'
      and tbl_id > '228892694'
    )
    or (
      dt = '20210112'
      and db_id = '62813'
      and tbl_id <= '226785696'
    )
    or (
      dt = '20210112'
      and db_id > '62812'
      and db_id < '62813'
    )
  )

In the where clause we have CNF contains some DNF and the last DNF is always false. But it will cause the optimizer to fall back to a full table scan.

The final plan looks like:

+-------------------------------+----------+-----------+------------------------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| id                            | estRows  | task      | access object                                  | operator info                                                                                                                                                                                |
+-------------------------------+----------+-----------+------------------------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| IndexLookUp_8                 | 0.67     | root      |                                                |                                                                                                                                                                                |
| ├─Selection_7(Build)          | 0.67     | cop[tikv] |                                                | or(and(eq(test.t71706696.dt, "20210112"), and(eq(test.t71706696.db_id, 62812), gt(test.t71706696.tbl_id, 228892694))), or(and(eq(test.t71706696.dt, "20210112"), and(eq(test.t71706696.db_id, 62813), le(test.t71706696.tbl_id, 226785696))), and(eq(test.t71706696.dt, "20210112"), and(gt(test.t71706696.db_id, 62812), lt(test.t71706696.db_id, 62813))))) |
| │ └─IndexFullScan_5           | 10000.00 | cop[tikv] | table:t71706696, index:dt_2(dt, db_id, tbl_id) | keep order:false, stats:pseudo                                                                                                                                                                                |
| └─TableRowIDScan_6(Probe)     | 0.67     | cop[tikv] | table:t71706696                                | keep order:false, stats:pseudo                                                                                                                                                                                |
+-------------------------------+----------+-----------+------------------------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+

Actually, we can ignore this always false filter and do an index scan.

What changed and how does it work?

From the code details, we should ignore the empty range here:

tidb/pkg/util/ranger/detacher.go

Line 739 in 3f24681

ranges := res.Ranges

In the detachCNFCondAndBuildRangeForIndex we will return directly from here if the range is empty:

tidb/pkg/util/ranger/detacher.go

Line 315 in 3f24681

if emptyRange {

But we didn't consider it in the detachDNFCondAndBuildRangeForIndex.

So I skip it here:
https://github.com/hi-rustin/tidb/blob/3f2468115cf4089d1bf47da3da4d06ca2dd31850/pkg/util/ranger/detacher.go#L740

			// If CNF item always false, we can return ignore this CNF item.
			if len(ranges) == 0 {
				continue
			}

I also updated the TestIndexMergeIssue49605 test case. Because in this case we have an empty range: BETWEEN _UTF8MB4'WVz' AND _UTF8MB4'G#駧褉ZC領*lov'. So that it doesn't use index merge anymore.

Check List

Tests

• Unit test
• Integration test
• Manual test (add detailed scripts or steps below)
• No need to test

  • I checked and no code files have been changed.

Side effects

• Performance regression: Consumes more CPU
• Performance regression: Consumes more Memory
• Breaking backward compatibility

Documentation

• Affects user behaviors
• Contains syntax changes
• Contains variable changes
• Contains experimental features
• Changes MySQL compatibility

Release note

Please refer to Release Notes Language Style Guide to write a quality release note.

增强 TiDB 对总是 false 的 DNF 的处理，直接忽略这种 filter
Enhance TiDB's handling of DNFs that are always false by simply ignoring such filters

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 55.2386%. Comparing base (9b6db9d) to head (1a1e545).
Report is 1671 commits behind head on master.

Additional details and impacted files

@@                Coverage Diff                @@
##             master     #51901         +/-   ##
=================================================
- Coverage   70.7435%   55.2386%   -15.5050%     
=================================================
  Files          1477       1592        +115     
  Lines        438550     616970     +178420     
=================================================
+ Hits         310246     340806      +30560     
- Misses       108884     252967     +144083     
- Partials      19420      23197       +3777     

Flag	Coverage Δ
integration	36.9547% <100.0000%> (?)
unit	70.9688% <100.0000%> (+0.4257%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Components	Coverage Δ
dumpling	53.9957% <ø> (ø)
parser	∅ <ø> (∅)
br	50.8987% <ø> (+5.0702%)	⬆️

[Rustin170506]

🔢 Self-check (PR reviewed by myself and ready for feedback.)

[Rustin170506]

-1696020282760139948 > -2619168038882941276. So this is an empty range.

[Rustin170506]

mysql> select 1 where _UTF8MB4'WVz' > _UTF8MB4'G#駧褉ZC領*lov';
+---+
| 1 |
+---+
| 1 |
+---+
1 row in set (0.00 sec)

mysql> select 1 where _UTF8MB4'WVz' < _UTF8MB4'G#駧褉ZC領*lov';
Empty set (0.00 sec)

[Rustin170506]

/unhold

[ti-chi-bot]

@hawkingrei: Your lgtm message is repeated, so it is ignored.

In response to this:

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 kubernetes/test-infra repository.

[ti-chi-bot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: AilinKid, hawkingrei

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

The pull request process is described here

Needs approval from an approver in each of these files:

• OWNERS [AilinKid,hawkingrei]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[ti-chi-bot]

[LGTM Timeline notifier]

Timeline:

• 2024-03-20 03:05:26.020780706 +0000 UTC m=+1432353.043027096: ☑️ agreed by hawkingrei.
• 2024-03-21 09:41:18.392882341 +0000 UTC m=+1542505.415128732: ☑️ agreed by AilinKid.

[ti-chi-bot]

In response to a cherrypick label: new pull request created to branch release-7.5: #53413.

[ti-chi-bot]

In response to a cherrypick label: new pull request created to branch release-7.1: #56993.