executor: fix escape for select into outfile

[ichn-hu]

What problem does this PR solve?

Issue Number: close #19517

Problem Summary:

SelectInto executor does not handle escape.

What is changed and how it works?

Proposal: xxx

What's Changed:

Added the handling, and also some more test cases.

How it Works:

Scan the field, and add escape.

Related changes

• Need to cherry-pick to the release 4.0

Check List

Tests

• Unit test

Side effects

• Performance regression
  • Consumes more CPU

Release note

• N/A

[ichn-hu]

/uncc @lzmhhh123

[ichn-hu]

/cc @qw4990 , @SunRunAway

[ichn-hu]

/run-all-tests

[ichn-hu]

/run-all-tests

[ichn-hu]

/run-unit-test

[qw4990]

LGTM

[breezewish]

What about other special characters like \r\n?

[ichn-hu]

What about other special characters like \r\n?

What do you mean? If they are used as line terminator, then it will be appended, if they appeared in a string, they will probably be enclosed (otherwise, we can't do anything).

Tested in MySQL:

mysql> insert into tx values ('\r\n', 123);
mysql> select * from tx into outfile '/tmp/test-r-n.txt' fields enclosed by '' lines terminated by '\r\n';

ichn-arch-pc# cat test-r-n.txt 
a       1
\N      \N
\
        123
ichn-arch-pc# hexdump -c test-r-n.txt 
0000000   a  \t   1  \r  \n   \   N  \t   \   N  \r  \n   \  \r  \n  \t
0000010   1   2   3  \r  \n                                            
0000015
ichn-arch-pc# cat test-r-n-enclosed.txt 
"a"     "1"
\N      \N
"\
"       "123"
ichn-arch-pc# hexdump -c test-r-n-enclosed.txt
0000000   "   a   "  \t   "   1   "  \r  \n   \   N  \t   \   N  \r  \n
0000010   "   \  \r  \n   "  \t   "   1   2   3   "  \r  \n            
000001d

TiDB:

mysql root@localhost:test> create table tx (a varchar(30), b int);
mysql root@localhost:test> insert into tx values ('\r\n', 123);
mysql root@localhost:test> select * from tx into outfile 'tidb-r-n.txt' fields enclosed by '' lines terminated by '\r\n';
mysql root@localhost:test> select * from tx into outfile 'tidb-r-n-enclosed.txt' fields enclosed by '"' lines terminated by '\r\n';


╭─ichn@ichn-arch-pc ~/go/src/github.com/pingcap/tidb ‹fix-select-into*› 
╰─$ cat tidb-r-n.txt

        123
╭─ichn@ichn-arch-pc ~/go/src/github.com/pingcap/tidb ‹fix-select-into*› 
╰─$ hexdump -c tidb-r-n.txt
0000000  \r  \n  \t   1   2   3  \r  \n                                
0000008
╭─ichn@ichn-arch-pc ~/go/src/github.com/pingcap/tidb ‹fix-select-into*› 
╰─$ cat tidb-r-n-enclosed.txt 
"
"       "123"
╭─ichn@ichn-arch-pc ~/go/src/github.com/pingcap/tidb ‹fix-select-into*› 
╰─$ hexdump -c tidb-r-n-enclosed.txt 
0000000   "  \r  \n   "  \t   "   1   2   3   "  \r  \n                
000000c

[ichn-hu]

/cc @breeswish

[breezewish]

@ichn-hu Thank you for pointing out! Could you further verify that the results of the following statements are identical?

CREATE TABLE `tx` (
  `a` varbinary(20) DEFAULT NULL
);

insert into tx values ('d","e",');
insert into tx values (unhex("00"));
insert into tx values ("\r\n\b\Z\t");
insert into tx values (null);

select * from tx into outfile '~/1.csv' FIELDS TERMINATED BY ',' ENCLOSED BY '"' LINES TERMINATED BY '\n';

[ichn-hu]

/run-all-tests

[ichn-hu]

/run-all-tests

[ichn-hu]

/run-sqllogic-test-1

[ichn-hu]

/run-all-tests

[ichn-hu]

/run-unit-test

[ichn-hu]

/run-check-dev

[SunRunAway]

/run-all-tests

[ichn-hu]

/run-all-tests

[ichn-hu]

/run-unit-test

[ichn-hu]

/run-unit-test

[ichn-hu]

/run-check-dev

[ichn-hu]

/run-all-tests

[ichn-hu]

/run-all-tests

[ichn-hu]

/run-all-tests

[ichn-hu]

/run-all-tests

[ti-srebot]

cherry pick to release-4.0 in PR #19905