Fix wrong output when using pattern expression as the filter in MATCH

[Aiee]

What type of PR is this?

• bug
• feature
• enhancement

What problem(s) does this PR solve?

Issue(s) number:

Close #4760

Description:

The roolUp executor produces wrong results.

After fix:

(root@nebula) [nba]> match p = (v:player{name:"Tim Duncan"})-[e]->(t) where not (v)-[]->(t:player) return t
+--------------------------------+
| t                              |
+--------------------------------+
| ("Spurs" :team{name: "Spurs"}) |
+--------------------------------+
Got 1 rows (time spent 22.629ms/23.301331ms)

How do you solve it?

Special notes for your reviewer, ex. impact of this fix, design document, etc:

Checklist:

Tests:

• Unit test(positive and negative cases)
• Function test
• Performance test
• N/A

Affects:

• Documentation affected (Please add the label if documentation needs to be modified.)
• Incompatibility (If it breaks the compatibility, please describe it and add the label.）
• If it's needed to cherry-pick (If cherry-pick to some branches is required, please label the destination version(s).)
• Performance impacted: Consumes more CPU/Memory

Release notes:

Please confirm whether to be reflected in release notes and how to describe:

ex. Fixed the bug .....

[Shylock-Hg]

Good job.

[Shylock-Hg]

The header of input table is fixed, why propIndex_ will be incorrect?

[Aiee]

For example, the column t is the 3rd column in the left iterator and is the 2nd column in the right iterator.

I20221026 16:00:58.585834 2228773 RollUpApplyExecutor.cpp:39] left iter size: 7
I20221026 16:00:58.585844 2228773 RollUpApplyExecutor.cpp:40] left iter value: v|e|t|p|

I20221026 16:00:58.586164 2228773 RollUpApplyExecutor.cpp:42] right iter size: 6
I20221026 16:00:58.586175 2228773 RollUpApplyExecutor.cpp:43] right iter value: v|t|(v)-->(t:player)|

[Shylock-Hg]

build and probe will eval the shared expression object on different input tables.

[Aiee]

build and probe will eval the shared expression object on different input tables.

This is correct, but the propIndex_ is cached in the expression, and the same column name may have a different column index in different input tables.

[Shylock-Hg]

build and probe will eval the shared expression object on different input tables.

This is correct, but the propIndex_ is cached in the expression, and the same column name may have a different column index in different input tables.

Yes, that's the point of this bug, so don't need to clone expr in each row.

[Shylock-Hg]

I think you better clone in rollUpApply function.

[Aiee]

I think you better clone in rollUpApply function.

Moved the copy into rollUpApply()

[Shylock-Hg]

LGTM, but this copy is useless.

[Aiee]

I'll remove this in another bug fix PR.

[xtcyclist]

Looks good.

[foesa-yang]

vesoft-inc/nebula-docs-cn#2519