feat(query): optimize stringsearch like

[TCeason]

I hereby agree to the terms of the CLA available at: https://databend.rs/dev/policies/cla/

Summary

optimize like parttern

0.15s -> 0.10s

-- before optimize:
mysql> select count() from orders where o_clerk like '%lerk#00000177%';
+---------+
| count() |
+---------+
|   15086 |
+---------+
1 row in set (0.15 sec)
Read 15000000 rows, 329.02 MiB in 0.145 sec., 103.29 million rows/sec., 2.21 GiB/sec.

--after optimize:

mysql> select count() from orders where o_clerk like '%lerk#00000177%';
+---------+
| count() |
+---------+
|   15086 |
+---------+
1 row in set (0.10 sec)
Read 15000000 rows, 329.02 MiB in 0.101 sec., 147.92 million rows/sec., 3.17 GiB/sec.

Closes #8458

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

1 Ignored Deployment

Name	Status	Preview	Updated
databend	⬜️ Ignored (Inspect)		Nov 9, 2022 at 11:41AM (UTC)

[TCeason]

In ClickHouse != string cost 0.055 sec, in databend != string cost 0.250 sec, And I retest the int comparison and modify the result in there.

!= string and int comparison ck 4x faster than databend.

And in this pr, the ck like function is also 4x faster than databend.

databend-arch :) select version();

SELECT version()

Query id: 9e68b4d6-c26c-40b5-9aee-646da7b573b0

┌─version()─┐
│ 22.7.1.1  │
└───────────┘

1 row in set. Elapsed: 0.004 sec. 

databend-arch :) select count() from orders where o_comment != '%s%p%';

SELECT count()
FROM orders
WHERE o_comment != '%s%p%'

Query id: 41b001f2-bf71-4a46-9815-bb3f73460d99

┌──count()─┐
│ 15000000 │
└──────────┘

1 row in set. Elapsed: 0.055 sec. Processed 15.00 million rows, 858.40 MB (274.07 million rows/s., 15.68 GB/s.)

mysql> select count() from orders where o_comment != '%s%p%';
+----------+
| count()  |
+----------+
| 15000000 |
+----------+
1 row in set (0.25 sec)
Read 15000000 rows, 808.13 MiB in 0.250 sec., 60 million rows/sec., 3.16 GiB/sec.