[SPARK-41159][SQL] Optimize like any and like all expressions

[wankunde]

What changes were proposed in this pull request?

Optimize like any and like all expressions with startWith, endWith, contains, equalTo methods.

Why are the changes needed?

Now like any and like all expressions will be very slow whether enable or disable LikeSimplification rule.
Refer to org.apache.spark.sql.execution.benchmark.LikeAnyBenchmark

OpenJDK 64-Bit Server VM 1.8.0_352-b08 on Linux 5.15.0-1022-azure
Intel(R) Xeon(R) Platinum 8370C CPU @ 2.80GHz
[info] Multi like query:                         Best Time(ms)   Avg Time(ms)   Stdev(ms)    Rate(M/s)   Per Row(ns)   Relative
[info] ------------------------------------------------------------------------------------------------------------------------
[info] Query with multi like                              1439           1518          78          0.0     1438904.6       1.0X
[info] Query with LikeAny simplification                  1392           1427          30          0.0     1392103.7       1.0X
[info] Query without LikeAny simplification                368            374           5          0.0      368485.2       3.9X

Does this PR introduce any user-facing change?

No

How was this patch tested?

Exists UT

[AmplabJenkins]

Can one of the admins verify this patch?

[wankunde]

Hi, @beliefer @cloud-fan @wangyum Could you help to review this PR? Thanks

[beliefer]

LikeSimplification have the similar optimization. Why need this class ?

[beliefer]

The benchmark cannot prove the performance improvement. Could you test with or without MatchMultiHelper ?

[wankunde]

Before this PR:

[info] Intel(R) Core(TM) i9-9980HK CPU @ 2.40GHz
[info] Multi like query:                         Best Time(ms)   Avg Time(ms)   Stdev(ms)    Rate(M/s)   Per Row(ns)   Relative
[info] ------------------------------------------------------------------------------------------------------------------------
[info] Query with multi like                              1393           1469         119          0.0     1392586.7       1.0X
[info] Query with LikeAny simplification                  1244           1309          97          0.0     1244382.5       1.1X
[info] Query without LikeAny simplification                400            407           8          0.0      399924.3       3.5X

[info] Multi like query:                         Best Time(ms)   Avg Time(ms)   Stdev(ms)    Rate(M/s)   Per Row(ns)   Relative
[info] ------------------------------------------------------------------------------------------------------------------------
[info] Query with multi like                              1476           1576         149          0.0     1475710.1       1.0X
[info] Query with LikeAny simplification                  1387           1429          37          0.0     1386669.1       1.1X
[info] Query without LikeAny simplification                430            470          35          0.0      430435.8       3.4X

After this PR:

[info] Multi like query:                         Best Time(ms)   Avg Time(ms)   Stdev(ms)    Rate(M/s)   Per Row(ns)   Relative
[info] ------------------------------------------------------------------------------------------------------------------------
[info] Query with multi like                              1441           1516          78          0.0     1441335.8       1.0X
[info] Query with LikeAny simplification                  1401           1431          44          0.0     1400743.9       1.0X
[info] Query without LikeAny simplification                357            369          10          0.0      357419.8       4.0X

[info] Multi like query:                         Best Time(ms)   Avg Time(ms)   Stdev(ms)    Rate(M/s)   Per Row(ns)   Relative
[info] ------------------------------------------------------------------------------------------------------------------------
[info] Query with multi like                              1524           1628         117          0.0     1524119.6       1.0X
[info] Query with LikeAny simplification                  1405           1418          18          0.0     1405258.7       1.1X
[info] Query without LikeAny simplification                362            372          12          0.0      361654.4       4.2X

[beliefer]

It means regression ?

[wankunde]

After LikeSimplification, the combination of multiple like expressions with OR can be pushdown to parquet reader, while like any can not.
So close this PR.