[#MDEV-12240] LOAD DATA INFILE binary blobs failing for UTF8

LOAD DATA INFILE with a binary blob fails when using UTF8 char sets. The result is additional escaped characters in the binary data when being accessed.

Test case - run against 10.0/1/2 of MariaDB. Injects binary blob that should only be escape processed for binary content and not treated as utf8; however MariaDB seems to be adding an extra escape to the binary contents - MySQL 5.6.35 is fine. The database and client connection are completely set up to use UTF8.

This might be related to ~~MDEV-11079~~ - "Regression: LOAD DATA INFILE lost BLOB support using utf8 load files"

$ echo -n -e '\xe2\x5c\x30\x0a' > input.bin

> select @@version;

+----------------+

| @@version      |

+----------------+

| 10.2.4-MariaDB |

+----------------+

> show variables like '%char%';

+--------------------------+-----------------------------------------------+

| Variable_name            | Value                                         |

+--------------------------+-----------------------------------------------+

| character_set_client     | utf8                                          |

| character_set_connection | utf8                                          |

| character_set_database   | utf8                                          |

| character_set_filesystem | binary                                        |

| character_set_results    | utf8                                          |

| character_set_server     | utf8                                          |

| character_set_system     | utf8                                          |

| character_sets_dir       | C:\Program Files\MariaDB 10.2\share\charsets\ |

+--------------------------+-----------------------------------------------+

8 rows in set (0.00 sec)

> CREATE TABLE test ( binStuff mediumblob );

> load data local infile 'input.bin' replace into table test;

Query OK, 1 row affected (0.00 sec)

Records: 1  Deleted: 0  Skipped: 0  Warnings: 0

> select * into outfile 'output.bin' from test;

### Dump original input

$ hexdump -C input.bin

00000000  e2 5c 30 0a                                       |.\0.|

00000004

### MariadDB has added an extra back-slash (5c) - probably as a result of treating it as UTF8 rather than binary.

$ hexdump -C output.bin

00000000  e2 5c 5c 30 0a                                    |.\\0.|

00000005

[MDEV-12240] LOAD DATA INFILE binary blobs failing for UTF8 Created: 2017-03-13 Updated: 2021-11-03
Status:	Confirmed
Project:	MariaDB Server
Component/s:	Character Sets
Affects Version/s:	10.0, 10.1, 10.0.30, 10.1.21, 10.2.4, 10.2
Fix Version/s:	10.2

[MDEV-12240] LOAD DATA INFILE binary blobs failing for UTF8 Created: 2017-03-13 Updated: 2021-11-03