Details
Description
mysql> SET NAMES utf8;
Query OK, 0 rows affected (0.00 sec)
mysql> SELECT hex('😁'), char_length('😁'), octet_length('😁');
+-------------+---------------------+----------------------+
| hex('😁') | char_length('😁') | octet_length('😁') |
+-------------+---------------------+----------------------+
| F09F9881 | 4 | 4 |
+-------------+---------------------+----------------------+
1 row in set (0.00 sec)
Notice, I use "SET NAMES utf8" (which is a 3-byte character set
and supports only BMP characters), but then input a 4-byte character.
The result of CHAR_LENGTH() is wrong.
0xF09F9881 is a wrong byte sequence of utf8 (it's correct for utf8mb4 only)
The expected result would be:
- either return error for the entire query
- or replace the character to '?' and thus make CHAR_LENGTH() return 1.
Gliffy Diagrams
Attachments
Issue Links
- is blocked by
-
MDEV-6643 Improve performance of string processing in the parser
-
- In Review
-
Activity
- All
- Comments
- Work Log
- History
- Activity
- Transitions
LEFT also returns a wrong result:
MariaDB [test]> SELECT hex(left('😁',2));
+---------------------+
| hex(left('😁',2)) |
+---------------------+
| F09F |
+---------------------+
1 row in set (0.00 sec)
RIGHT returns a wrong result:
MariaDB [test]> SELECT hex(right('😁',2));
+----------------------+
| hex(right('😁',2)) |
+----------------------+
| 9881 |
+----------------------+
SUBSTRING returns a wrong result:
MariaDB [test]> SELECT hex(substring('😁',2,1));
+----------------------------+
| hex(substring('😁',2,1)) |
+----------------------------+
| 9F |
+----------------------------+
In this example, the returned string is also bad formed:
MariaDB [test]> SELECT '11😁2😁22222'; +------------------+ | 11😁2😁22222 | +------------------+ | 11😁2😁22222 | +------------------+
It should probably replace unknown bytes to question marks.
Repeatable on mysql-5.6.17