MDEV-8844 Unreadable control characters printed as is in warnings

include/m_ctype.h

@@ -45,6 +45,19 @@ extern "C" {
 
 #define MY_CS_REPLACEMENT_CHARACTER 0xFFFD
 
+/**
+  Maximum character length of a string produced by wc_to_printable().
+  Note, wc_to_printable() is currently limited to BMP.
+  One non-printable or non-convertable character can produce a string
+  with at most 5 characters: \hhhh.
+  If we ever modify wc_to_printable() to support supplementary characters,
+  e.g. \+hhhhhh, this constant should be changed to 8.
+  Note, maximum octet length of a wc_to_printable() result can be calculated
+  as: (MY_CS_PRINTABLE_CHAR_LENGTH*cs->mbminlen).
+*/
+#define MY_CS_PRINTABLE_CHAR_LENGTH  5
+
+
 /*
   On i386 we store Unicode->CS conversion tables for
   some character sets using Big-endian order,
@@ -740,6 +753,9 @@ int my_wc_mb_bin(CHARSET_INFO *cs,my_wc_t wc, uchar *s, uchar *e);
 int my_mb_ctype_8bit(CHARSET_INFO *,int *, const uchar *,const uchar *);
 int my_mb_ctype_mb(CHARSET_INFO *,int *, const uchar *,const uchar *);
 
+int my_wc_to_printable_generic(CHARSET_INFO *cs, my_wc_t wc,
+                               uchar *s, uchar *e);
+
 size_t my_scan_8bit(CHARSET_INFO *cs, const char *b, const char *e, int sq);
 
 size_t my_snprintf_8bit(CHARSET_INFO *, char *to, size_t n,

mysql-test/main/ctype_binary.result

@@ -3174,3 +3174,29 @@ DROP TABLE t1;
 #
 # End of 10.1 tests
 #
+#
+# Start of 10.5 tests
+#
+#
+# MDEV-8844 Unreadable control characters printed as is in warnings
+#
+SET NAMES binary;
+CREATE TABLE t1 (a VARCHAR(20) CHARACTER SET latin1, UNIQUE(a));
+INSERT INTO t1 VALUES (0x61000162FF);
+INSERT INTO t1 VALUES (0x61000162FF);
+ERROR 23000: Duplicate entry 'a\0000\0001bÿ' for key 'a'
+INSERT IGNORE INTO t1 VALUES (0x61000162FF);
+Warnings:
+Warning	1062	Duplicate entry 'a\0000\0001bÿ' for key 'a'
+DROP TABLE t1;
+CREATE TABLE t1 (a VARCHAR(20) CHARACTER SET utf8, UNIQUE(a));
+INSERT INTO t1 VALUES (_latin1 0x61000162FF);
+INSERT INTO t1 VALUES (_latin1 0x61000162FF);
+ERROR 23000: Duplicate entry 'a\0000\0001bÿ' for key 'a'
+INSERT IGNORE INTO t1 VALUES (_latin1 0x61000162FF);
+Warnings:
+Warning	1062	Duplicate entry 'a\0000\0001bÿ' for key 'a'
+DROP TABLE t1;
+#
+# End of 10.5 tests
+#

mysql-test/main/ctype_binary.test

@@ -77,3 +77,31 @@ DROP TABLE t1;
 --echo #
 --echo # End of 10.1 tests
 --echo #
+
+--echo #
+--echo # Start of 10.5 tests
+--echo #
+
+--echo #
+--echo # MDEV-8844 Unreadable control characters printed as is in warnings
+--echo #
+
+SET NAMES binary;
+CREATE TABLE t1 (a VARCHAR(20) CHARACTER SET latin1, UNIQUE(a));
+INSERT INTO t1 VALUES (0x61000162FF);
+--error ER_DUP_ENTRY
+INSERT INTO t1 VALUES (0x61000162FF);
+INSERT IGNORE INTO t1 VALUES (0x61000162FF);
+DROP TABLE t1;
+
+CREATE TABLE t1 (a VARCHAR(20) CHARACTER SET utf8, UNIQUE(a));
+INSERT INTO t1 VALUES (_latin1 0x61000162FF);
+--error ER_DUP_ENTRY
+INSERT INTO t1 VALUES (_latin1 0x61000162FF);
+INSERT IGNORE INTO t1 VALUES (_latin1 0x61000162FF);
+DROP TABLE t1;
+
+--echo #
+--echo # End of 10.5 tests
+--echo #
+

mysql-test/main/ctype_latin1.result

@@ -8907,5 +8907,44 @@ t1	CREATE TABLE `t1` (
 ) ENGINE=MyISAM DEFAULT CHARSET=latin1
 DROP TABLE t1;
 #
+# MDEV-8844 Unreadable control characters printed as is in warnings
+#
+SET NAMES latin1;
+SELECT CAST(_latin1 0x610062 AS INT);
+CAST(_latin1 0x610062 AS INT)
+0
+Warnings:
+Warning	1292	Truncated incorrect INTEGER value: 'a\0000b'
+SELECT CAST(_latin1 0x610162 AS INT);
+CAST(_latin1 0x610162 AS INT)
+0
+Warnings:
+Warning	1292	Truncated incorrect INTEGER value: 'a\0001b'
+SELECT CAST(_latin1 0x611F62 AS INT);
+CAST(_latin1 0x611F62 AS INT)
+0
+Warnings:
+Warning	1292	Truncated incorrect INTEGER value: 'a\001Fb'
+SELECT CAST(_latin1 0x617F62 AS INT);
+CAST(_latin1 0x617F62 AS INT)
+0
+Warnings:
+Warning	1292	Truncated incorrect INTEGER value: 'a\007Fb'
+SELECT CAST(_latin1 0x612062 AS INT);
+CAST(_latin1 0x612062 AS INT)
+0
+Warnings:
+Warning	1292	Truncated incorrect INTEGER value: 'a b'
+SELECT CAST(_latin1 0x617E62 AS INT);
+CAST(_latin1 0x617E62 AS INT)
+0
+Warnings:
+Warning	1292	Truncated incorrect INTEGER value: 'a~b'
+SELECT CAST(_latin1 0x61FF62 AS INT);
+CAST(_latin1 0x61FF62 AS INT)
+0
+Warnings:
+Warning	1292	Truncated incorrect INTEGER value: 'ab'
+#
 # End of 10.5 tests
 #

mysql-test/main/ctype_latin1.test

@@ -456,6 +456,22 @@ CREATE OR REPLACE TABLE t1 AS SELECT CAST(1 AS BINARY), CAST(@a AS BINARY), CAST
 SHOW CREATE TABLE t1;
 DROP TABLE t1;
 
+
+--echo #
+--echo # MDEV-8844 Unreadable control characters printed as is in warnings
+--echo #
+SET NAMES latin1;
+# control
+SELECT CAST(_latin1 0x610062 AS INT);
+SELECT CAST(_latin1 0x610162 AS INT);
+SELECT CAST(_latin1 0x611F62 AS INT);
+SELECT CAST(_latin1 0x617F62 AS INT);
+# normal characters
+SELECT CAST(_latin1 0x612062 AS INT);
+SELECT CAST(_latin1 0x617E62 AS INT);
+SELECT CAST(_latin1 0x61FF62 AS INT);
+
+
 --echo #
 --echo # End of 10.5 tests
 --echo #

mysql-test/main/ctype_ucs.result

@@ -6431,3 +6431,45 @@ SET NAMES utf8;
 #
 # End of 10.4 tests
 #
+#
+# Start of 10.5 tests
+#
+#
+# MDEV-8844 Unreadable control characters printed as is in warnings
+#
+# control
+SELECT CAST(_ucs2 0x006100000062 AS INT);
+CAST(_ucs2 0x006100000062 AS INT)
+0
+Warnings:
+Warning	1292	Truncated incorrect INTEGER value: 'a\0000b'
+SELECT CAST(_ucs2 0x006100010062 AS INT);
+CAST(_ucs2 0x006100010062 AS INT)
+0
+Warnings:
+Warning	1292	Truncated incorrect INTEGER value: 'a\0001b'
+# surrogate halfs
+SELECT CAST(_ucs2 0x0061D8000062 AS INT);
+CAST(_ucs2 0x0061D8000062 AS INT)
+0
+Warnings:
+Warning	1292	Truncated incorrect INTEGER value: 'a\D800b'
+SELECT CAST(_ucs2 0x0061DFFF0062 AS INT);
+CAST(_ucs2 0x0061DFFF0062 AS INT)
+0
+Warnings:
+Warning	1292	Truncated incorrect INTEGER value: 'a\DFFFb'
+# normal characters
+SELECT CAST(_ucs2 0x0061D7000062 AS INT);
+CAST(_ucs2 0x0061D7000062 AS INT)
+0
+Warnings:
+Warning	1292	Truncated incorrect INTEGER value: 'a휀b'
+SELECT CAST(_ucs2 0x0061E0030062 AS INT);
+CAST(_ucs2 0x0061E0030062 AS INT)
+0
+Warnings:
+Warning	1292	Truncated incorrect INTEGER value: 'ab'
+#
+# End of 10.5 tests
+#

mysql-test/main/ctype_ucs.test

@@ -1124,3 +1124,28 @@ SET NAMES utf8;
 --echo #
 --echo # End of 10.4 tests
 --echo #
+
+
+--echo #
+--echo # Start of 10.5 tests
+--echo #
+
+--echo #
+--echo # MDEV-8844 Unreadable control characters printed as is in warnings
+--echo #
+
+--echo # control
+SELECT CAST(_ucs2 0x006100000062 AS INT);
+SELECT CAST(_ucs2 0x006100010062 AS INT);
+
+--echo # surrogate halfs
+SELECT CAST(_ucs2 0x0061D8000062 AS INT);
+SELECT CAST(_ucs2 0x0061DFFF0062 AS INT);
+
+--echo # normal characters
+SELECT CAST(_ucs2 0x0061D7000062 AS INT);
+SELECT CAST(_ucs2 0x0061E0030062 AS INT);
+
+--echo #
+--echo # End of 10.5 tests
+--echo #

mysql-test/main/ctype_utf16.result

@@ -2814,3 +2814,18 @@ SET STORAGE_ENGINE=Default;
 #
 # End of 10.2 tests
 #
+#
+# Start of 10.5 tests
+#
+#
+# MDEV-8844 Unreadable control characters printed as is in warnings
+#
+SET NAMES utf8;
+SELECT CAST(_utf16 0x0061D83DDE0E0062 AS INT);
+CAST(_utf16 0x0061D83DDE0E0062 AS INT)
+0
+Warnings:
+Warning	1292	Truncated incorrect INTEGER value: 'a?b'
+#
+# End of 10.5 tests
+#

mysql-test/main/ctype_utf16.test

@@ -934,3 +934,23 @@ let $coll_pad='utf16_bin';
 --echo #
 --echo # End of 10.2 tests
 --echo #
+
+
+--echo #
+--echo # Start of 10.5 tests
+--echo #
+
+--echo #
+--echo # MDEV-8844 Unreadable control characters printed as is in warnings
+--echo #
+
+SET NAMES utf8;
+# Make sure surrogate halfs (when a part of a full utf16 character)
+# are not escaped and the entire utf16 character consisting of two
+# surrogate pairs is replaced to a single question mark.
+SELECT CAST(_utf16 0x0061D83DDE0E0062 AS INT);
+
+
+--echo #
+--echo # End of 10.5 tests
+--echo #

mysql-test/main/ctype_utf8.result

@@ -5407,18 +5407,21 @@ DROP TABLE t1;
 #
 # Bug#11764503 (Bug#57341) Query in EXPLAIN EXTENDED shows wrong characters
 #
+# Emulate utf8 client erroneously started with --default-character-set=latin1,
+# # as in the bug report. EXPLAIN output should still be pretty readable
 SET NAMES latin1;
 EXPLAIN EXTENDED SELECT 'abcdÁÂÃÄÅ', _latin1'abcdÁÂÃÄÅ', _utf8'abcdÁÂÃÄÅ' AS u;
 id	select_type	table	type	possible_keys	key	key_len	ref	rows	filtered	Extra
 1	SIMPLE	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	No tables used
 Warnings:
-Note	1003	select 'abcdÁÂÃÄÅ' AS `abcdÁÂÃÄÅ`,_latin1'abcd\xC3\x81\xC3\x82\xC3\x83\xC3\x84\xC3\x85' AS `abcdÁÂÃÄÅ`,_utf8'abcd\xC3\x81\xC3\x82\xC3\x83\xC3\x84\xC3\x85' AS `u`
+Note	1003	select 'abcd\0081ÂÃÄÅ' AS `abcd\0081ÂÃÄÅ`,_latin1'abcd\xC3\x81\xC3\x82\xC3\x83\xC3\x84\xC3\x85' AS `abcd\0081ÂÃÄÅ`,_utf8'abcd\xC3\x81\xC3\x82\xC3\x83\xC3\x84\xC3\x85' AS `u`
+# Test normal utf8
 SET NAMES utf8;
 EXPLAIN EXTENDED SELECT 'abcdÁÂÃÄÅ', _latin1'abcdÁÂÃÄÅ', _utf8'abcdÁÂÃÄÅ';
 id	select_type	table	type	possible_keys	key	key_len	ref	rows	filtered	Extra
 1	SIMPLE	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	No tables used
 Warnings:
-Note	1003	select 'abcdÁÂÃÄÅ' AS `abcdÁÂÃÄÅ`,_latin1'abcd\xC3\x81\xC3\x82\xC3\x83\xC3\x84\xC3\x85' AS `abcdÁÂÃÄÅ`,_utf8'abcd\xC3\x81\xC3\x82\xC3\x83\xC3\x84\xC3\x85' AS `abcdÁÂÃÄÅ`
+Note	1003	select 'abcdÁÂÃÄÅ' AS `abcdÁÂÃÄÅ`,_latin1'abcd\xC3\x81\xC3\x82\xC3\x83\xC3\x84\xC3\x85' AS `abcdÃ\0081ÂÃÄÅ`,_utf8'abcd\xC3\x81\xC3\x82\xC3\x83\xC3\x84\xC3\x85' AS `abcdÁÂÃÄÅ`
 #
 # Bug#11750518 41090: ORDER BY TRUNCATES GROUP_CONCAT RESULT
 #
@@ -11348,5 +11351,72 @@ SELECT uuid()>'';
 uuid()>''
 1
 #
+# MDEV-8844 Unreadable control characters printed as is in warnings
+#
+SET NAMES utf8;
+# control, part1
+SELECT CAST(_utf8 0x610062 AS INT);
+CAST(_utf8 0x610062 AS INT)
+0
+Warnings:
+Warning	1292	Truncated incorrect INTEGER value: 'a\0000b'
+SELECT CAST(_utf8 0x610162 AS INT);
+CAST(_utf8 0x610162 AS INT)
+0
+Warnings:
+Warning	1292	Truncated incorrect INTEGER value: 'a\0001b'
+SELECT CAST(_utf8 0x611F62 AS INT);
+CAST(_utf8 0x611F62 AS INT)
+0
+Warnings:
+Warning	1292	Truncated incorrect INTEGER value: 'a\001Fb'
+# control, part2: U+0080..U+009F
+SELECT CAST(_utf8 0x617F62 AS INT);
+CAST(_utf8 0x617F62 AS INT)
+0
+Warnings:
+Warning	1292	Truncated incorrect INTEGER value: 'a\007Fb'
+SELECT CAST(_utf8 0x61C28062 AS INT);
+CAST(_utf8 0x61C28062 AS INT)
+0
+Warnings:
+Warning	1292	Truncated incorrect INTEGER value: 'a\0080b'
+SELECT CAST(_utf8 0x61C29F62 AS INT);
+CAST(_utf8 0x61C29F62 AS INT)
+0
+Warnings:
+Warning	1292	Truncated incorrect INTEGER value: 'a\009Fb'
+# normal characters
+SELECT CAST(_utf8 0x612062 AS INT);
+CAST(_utf8 0x612062 AS INT)
+0
+Warnings:
+Warning	1292	Truncated incorrect INTEGER value: 'a b'
+SELECT CAST(_utf8 0x617E62 AS INT);
+CAST(_utf8 0x617E62 AS INT)
+0
+Warnings:
+Warning	1292	Truncated incorrect INTEGER value: 'a~b'
+SELECT CAST(_utf8 0x61C2BF62 AS INT);
+CAST(_utf8 0x61C2BF62 AS INT)
+0
+Warnings:
+Warning	1292	Truncated incorrect INTEGER value: 'a¿b'
+SELECT CAST(_utf8 'ëëë' AS INT);
+CAST(_utf8 'ëëë' AS INT)
+0
+Warnings:
+Warning	1292	Truncated incorrect INTEGER value: 'ëëë'
+SELECT CAST(_utf8 'œœœ' AS INT);
+CAST(_utf8 'œœœ' AS INT)
+0
+Warnings:
+Warning	1292	Truncated incorrect INTEGER value: 'œœœ'
+SELECT CAST(_utf8 'яяя' AS INT);
+CAST(_utf8 'яяя' AS INT)
+0
+Warnings:
+Warning	1292	Truncated incorrect INTEGER value: 'яяя'
+#
 # End of 10.5 tests
 #

mysql-test/main/ctype_utf8.test

@@ -1598,11 +1598,12 @@ DROP TABLE t1;
 --echo #
 --echo # Bug#11764503 (Bug#57341) Query in EXPLAIN EXTENDED shows wrong characters
 --echo #
-# Emulate utf8 client erroneously started with --default-character-set=latin1,
-# as in the bug report. EXPLAIN output should still be pretty readable
+
+--echo # Emulate utf8 client erroneously started with --default-character-set=latin1,
+--echo # # as in the bug report. EXPLAIN output should still be pretty readable
 SET NAMES latin1;
 EXPLAIN EXTENDED SELECT 'abcdÁÂÃÄÅ', _latin1'abcdÁÂÃÄÅ', _utf8'abcdÁÂÃÄÅ' AS u;
-# Test normal utf8
+--echo # Test normal utf8
 SET NAMES utf8;
 EXPLAIN EXTENDED SELECT 'abcdÁÂÃÄÅ', _latin1'abcdÁÂÃÄÅ', _utf8'abcdÁÂÃÄÅ';
 
@@ -2283,6 +2284,30 @@ DROP TABLE t1;
 SET NAMES utf8 COLLATE utf8_unicode_ci;
 SELECT uuid()>'';
 
+
+--echo #
+--echo # MDEV-8844 Unreadable control characters printed as is in warnings
+--echo #
+SET NAMES utf8;
+--echo # control, part1
+SELECT CAST(_utf8 0x610062 AS INT);
+SELECT CAST(_utf8 0x610162 AS INT);
+SELECT CAST(_utf8 0x611F62 AS INT);
+
+--echo # control, part2: U+0080..U+009F
+SELECT CAST(_utf8 0x617F62 AS INT);
+SELECT CAST(_utf8 0x61C28062 AS INT);
+SELECT CAST(_utf8 0x61C29F62 AS INT);
+
+--echo # normal characters
+SELECT CAST(_utf8 0x612062 AS INT);
+SELECT CAST(_utf8 0x617E62 AS INT);
+SELECT CAST(_utf8 0x61C2BF62 AS INT);
+SELECT CAST(_utf8 'ëëë' AS INT);
+SELECT CAST(_utf8 'œœœ' AS INT);
+SELECT CAST(_utf8 'яяя' AS INT);
+
+
 --echo #
 --echo # End of 10.5 tests
 --echo #

mysql-test/main/func_str.result

@@ -967,17 +967,17 @@ explain extended select length('\n\t\r\b\0\_\%\\');
 id	select_type	table	type	possible_keys	key	key_len	ref	rows	filtered	Extra
 1	SIMPLE	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	No tables used
 Warnings:
-Note	1003	select octet_length('\n	\r\0\\_\\%\\') AS `length('\n\t\r\b\0\_\%\\')`
+Note	1003	select octet_length('\n	\r\0008\0\\_\\%\\') AS `length('\n\t\r\b\0\_\%\\')`
 explain extended select bit_length('\n\t\r\b\0\_\%\\');
 id	select_type	table	type	possible_keys	key	key_len	ref	rows	filtered	Extra
 1	SIMPLE	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	No tables used
 Warnings:
-Note	1003	select bit_length('\n	\r\0\\_\\%\\') AS `bit_length('\n\t\r\b\0\_\%\\')`
+Note	1003	select bit_length('\n	\r\0008\0\\_\\%\\') AS `bit_length('\n\t\r\b\0\_\%\\')`
 explain extended select bit_length('\n\t\r\b\0\_\%\\');
 id	select_type	table	type	possible_keys	key	key_len	ref	rows	filtered	Extra
 1	SIMPLE	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	No tables used
 Warnings:
-Note	1003	select bit_length('\n	\r\0\\_\\%\\') AS `bit_length('\n\t\r\b\0\_\%\\')`
+Note	1003	select bit_length('\n	\r\0008\0\\_\\%\\') AS `bit_length('\n\t\r\b\0\_\%\\')`
 explain extended select concat('monty',' was here ','again');
 id	select_type	table	type	possible_keys	key	key_len	ref	rows	filtered	Extra
 1	SIMPLE	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	No tables used

mysql-test/main/get_diagnostics.result

+SET NAMES utf8;
 #
 # WL#2111: GET DIAGNOSTICS tests
 #

mysql-test/main/get_diagnostics.test

+SET NAMES utf8;
+
 --echo #
 --echo # WL#2111: GET DIAGNOSTICS tests
 --echo #

mysql-test/main/mix2_myisam.result

@@ -1552,7 +1552,7 @@ alter table t1 add unique(v);
 ERROR 23000: Duplicate entry '{ ' for key 'v_2'
 show warnings;
 Level	Code	Message
-Error	1062	Duplicate entry 'a
' for key 'v_2'
+Error	1062	Duplicate entry 'a\0001' for key 'v_2'
 alter table t1 add key(v);
 Warnings:
 Note	1831	Duplicate index `v_2`. This is deprecated and will be disallowed in a future release

mysql-test/main/mrr_icp_extra.result

@@ -353,7 +353,7 @@ alter table t1 add unique(v);
 ERROR 23000: Duplicate entry '{ ' for key 'v_2'
 show warnings;
 Level	Code	Message
-Error	1062	Duplicate entry 'a
' for key 'v_2'
+Error	1062	Duplicate entry 'a\0001' for key 'v_2'
 alter table t1 add key(v);
 Warnings:
 Note	1831	Duplicate index `v_2`. This is deprecated and will be disallowed in a future release

mysql-test/main/myisam.result

@@ -1268,7 +1268,7 @@ alter table t1 add unique(v);
 ERROR 23000: Duplicate entry '{ ' for key 'v_2'
 show warnings;
 Level	Code	Message
-Error	1062	Duplicate entry 'a
' for key 'v_2'
+Error	1062	Duplicate entry 'a\0001' for key 'v_2'
 alter table t1 add key(v);
 Warnings:
 Note	1831	Duplicate index `v_2`. This is deprecated and will be disallowed in a future release

mysql-test/main/not_embedded_server.result

@@ -68,7 +68,7 @@ ERROR HY000: Operation CREATE USER failed for 'user\"s_12601974'@'localhost'
 DROP USER 'user\"s_12601974'@'localhost';
 CREATE USER 'user\bs_12601974'@'localhost';
 CREATE USER 'user\bs_12601974'@'localhost';
-ERROR HY000: Operation CREATE USER failed for 'users_12601974'@'localhost'
+ERROR HY000: Operation CREATE USER failed for 'user\0008s_12601974'@'localhost'
 DROP USER 'user\bs_12601974'@'localhost';
 CREATE USER 'user\ns_12601974'@'localhost';
 CREATE USER 'user\ns_12601974'@'localhost';

mysql-test/suite/maria/maria.result

@@ -1156,7 +1156,7 @@ alter table t1 add unique(v);
 ERROR 23000: Duplicate entry '{ ' for key 'v_2'
 show warnings;
 Level	Code	Message
-Error	1062	Duplicate entry 'a
' for key 'v_2'
+Error	1062	Duplicate entry 'a\0001' for key 'v_2'
 alter table t1 add key(v);
 Warnings:
 Note	1831	Duplicate index `v_2`. This is deprecated and will be disallowed in a future release

mysql-test/suite/sys_vars/r/general_log_file_basic.result

@@ -19,7 +19,7 @@ ERROR 42000: Variable 'general_log_file' can't be set to the value of '/tmp/my.c
 SET @@global.general_log_file = '.my.cnf';
 ERROR 42000: Variable 'general_log_file' can't be set to the value of '.my.cnf'
 SET @@global.general_log_file = 'my.cnf\0foo';
-ERROR 42000: Variable 'general_log_file' can't be set to the value of 'my.cnf'
+ERROR 42000: Variable 'general_log_file' can't be set to the value of 'my.cnf\0000foo'
 SET @@global.general_log_file = 'my.ini';
 ERROR 42000: Variable 'general_log_file' can't be set to the value of 'my.ini'
 '#----------------------FN_DYNVARS_004_03------------------------#'

mysql-test/suite/sys_vars/r/slow_query_log_file_basic.result

@@ -16,7 +16,7 @@ ERROR 42000: Variable 'slow_query_log_file' can't be set to the value of '/tmp/m
 SET @@global.general_log_file = '.my.cnf';
 ERROR 42000: Variable 'general_log_file' can't be set to the value of '.my.cnf'
 SET @@global.general_log_file = 'my.cnf\0foo';
-ERROR 42000: Variable 'general_log_file' can't be set to the value of 'my.cnf'
+ERROR 42000: Variable 'general_log_file' can't be set to the value of 'my.cnf\0000foo'
 SET @@global.general_log_file = 'my.ini';
 ERROR 42000: Variable 'general_log_file' can't be set to the value of 'my.ini'
 '#----------------------FN_DYNVARS_004_03------------------------#'

sql/protocol.cc

@@ -1143,6 +1143,18 @@ bool Protocol::store_string_aux(const char *from, size_t length,
 }
 
 
+bool Protocol::store_warning(const char *from, size_t length)
+{
+  BinaryStringBuffer<MYSQL_ERRMSG_SIZE> tmp;
+  CHARSET_INFO *cs= thd->variables.character_set_results;
+  if (cs == &my_charset_bin)
+    cs= system_charset_info;
+  if (tmp.copy_printable_hhhh(cs, system_charset_info, from, length))
+    return net_store_data((const uchar*)"", 0);
+  return net_store_data((const uchar *) tmp.ptr(), tmp.length());
+}
+
+
 bool Protocol_text::store(const char *from, size_t length,
                           CHARSET_INFO *fromcs, CHARSET_INFO *tocs)
 {

sql/protocol.h

@@ -90,6 +90,7 @@ class Protocol
 
   bool store(I_List<i_string> *str_list);
   bool store(const char *from, CHARSET_INFO *cs);
+  bool store_warning(const char *from, size_t length);
   String *storage_packet() { return packet; }
   inline void free() { packet->free(); }
   virtual bool write();

sql/sql_error.cc

@@ -822,9 +822,8 @@ bool mysqld_show_warnings(THD *thd, ulong levels_to_show)
 		    warning_level_names[err->get_level()].length,
                     system_charset_info);
     protocol->store((uint32) err->get_sql_errno());
-    protocol->store(err->get_message_text(),
-                    err->get_message_octet_length(),
-                    system_charset_info);
+    protocol->store_warning(err->get_message_text(),
+                            err->get_message_octet_length());
     if (protocol->write())
       DBUG_RETURN(TRUE);
   }
@@ -836,6 +835,26 @@ bool mysqld_show_warnings(THD *thd, ulong levels_to_show)
 }
 
 
+/**
+  This replaces U+0000 to '\0000', so the result error message string:
+  - is a good null-terminated string
+  - presents the entire data
+  For example:
+    SELECT CAST(_latin1 0x610062 AS SIGNED);
+  returns a warning:
+    Truncated incorrect INTEGER value: 'a\0000b'
+  Notice, the 0x00 byte is replaced to a 5-byte long string '\0000',
+  while 'a' and 'b' are printed as is.
+*/
+extern "C" int my_wc_mb_utf8_null_terminated(CHARSET_INFO *cs,
+                                             my_wc_t wc, uchar *r, uchar *e)
+{
+  return wc == '\0' ?
+         my_wc_to_printable_generic(cs, wc, r, e) :
+         my_charset_utf8mb3_handler.wc_mb(cs, wc, r, e);
+}
+
+
 /**
    Convert value for dispatch to error message(see WL#751).
 
@@ -894,8 +913,11 @@ char *err_conv(char *buff, uint to_length, const char *from,
   else
   {
     uint errors;
-    res= copy_and_convert(to, to_length, system_charset_info,
-                          from, from_length, from_cs, &errors);
+    res= my_convert_using_func(to, to_length, system_charset_info,
+                               my_wc_mb_utf8_null_terminated,
+                               from, from_length, from_cs,
+                               from_cs->cset->mb_wc,
+                               &errors);
     to[res]= 0;
   }
   return buff;
@@ -921,64 +943,21 @@ size_t convert_error_message(char *to, size_t to_length, CHARSET_INFO *to_cs,
                              const char *from, size_t from_length,
                              CHARSET_INFO *from_cs, uint *errors)
 {
-  int  cnvres;
-  my_wc_t     wc;
-  const uchar *from_end= (const uchar*) from+from_length;
-  char *to_start= to;
-  uchar *to_end;
-  my_charset_conv_mb_wc mb_wc= from_cs->cset->mb_wc;
-  my_charset_conv_wc_mb wc_mb;
-  uint error_count= 0;
-  size_t length;
-
   DBUG_ASSERT(to_length > 0);
   /* Make room for the null terminator. */
   to_length--;
-  to_end= (uchar*) (to + to_length);
-
-  if (!to_cs || from_cs == to_cs || to_cs == &my_charset_bin)
-  {
-    length= MY_MIN(to_length, from_length);
-    memmove(to, from, length);
-    to[length]= 0;
-    return length;
-  }
-
-  wc_mb= to_cs->cset->wc_mb;
-  while (1)
-  {
-    if ((cnvres= (*mb_wc)(from_cs, &wc, (uchar*) from, from_end)) > 0)
-    {
-      if (!wc)
-        break;
-      from+= cnvres;
-    }
-    else if (cnvres == MY_CS_ILSEQ)
-    {
-      wc= (ulong) (uchar) *from;
-      from+=1;
-    }
-    else
-      break;
-
-    if ((cnvres= (*wc_mb)(to_cs, wc, (uchar*) to, to_end)) > 0)
-      to+= cnvres;
-    else if (cnvres == MY_CS_ILUNI)
-    {
-      length= (wc <= 0xFFFF) ? 6/* '\1234' format*/ : 9 /* '\+123456' format*/;
-      if ((uchar*)(to + length) >= to_end)
-        break;
-      cnvres= (int)my_snprintf(to, 9,
-                          (wc <= 0xFFFF) ? "\\%04X" : "\\+%06X", (uint) wc);
-      to+= cnvres;
-    }
-    else
-      break;
-  }
 
-  *to= 0;
-  *errors= error_count;
-  return (size_t) (to - to_start);
+  if (to_cs == &my_charset_bin)
+    to_cs= system_charset_info;
+  uint32 cnv_length= my_convert_using_func(to, to_length,
+                                           to_cs,
+                                           my_wc_to_printable_generic,
+                                           from, from_length,
+                                           from_cs, from_cs->cset->mb_wc,
+                                           errors);
+  DBUG_ASSERT(to_length >= cnv_length);
+  to[cnv_length]= '\0';
+  return cnv_length;
 }
 
 

sql/sql_string.cc

@@ -777,6 +777,28 @@ void Static_binary_string::qs_append(ulonglong i)
   str_length+= (int) (end-buff);
 }
 
+
+bool Binary_string::copy_printable_hhhh(CHARSET_INFO *to_cs,
+                                        CHARSET_INFO *from_cs,
+                                        const char *from,
+                                        uint32 from_length)
+{
+  uint errors;
+  uint one_escaped_char_length= MY_CS_PRINTABLE_CHAR_LENGTH * to_cs->mbminlen;
+  uint one_char_length= MY_MAX(one_escaped_char_length, to_cs->mbmaxlen);
+  uint32 bytes_needed= (uint32) from_length * one_char_length;
+  if (alloc(bytes_needed))
+    return true;
+  str_length= my_convert_using_func(Ptr, Alloced_length, to_cs,
+                                    my_wc_to_printable_generic,
+                                    from, from_length,
+                                    from_cs,
+                                    from_cs->cset->mb_wc,
+                                    &errors);
+  return false;
+}
+
+
 /*
   Compare strings according to collation, without end space.
 

sql/sql_string.h

@@ -524,6 +524,15 @@ class Binary_string: public Static_binary_string
   bool copy(const char *s, size_t arg_length);	// Allocate new string
   bool copy_or_move(const char *s,size_t arg_length);
 
+  /**
+    Convert a string to a printable format.
+    All non-convertable and control characters are replaced to 5-character
+    sequences '\hhhh'.
+  */
+  bool copy_printable_hhhh(CHARSET_INFO *to_cs,
+                           CHARSET_INFO *from_cs,
+                           const char *from, uint32 from_length);
+
   bool append_ulonglong(ulonglong val);
   bool append_longlong(longlong val);
 

storage/mroonga/mysql-test/mroonga/wrapper/r/transaction_rollback_delete_delete.result

@@ -28,7 +28,7 @@ MATCH(body) AGAINST("groonga");
 id	title	body
 DELETE FROM diaries WHERE id = 1;
 Warnings:
-Warning	1026	failed to get record ID for deleting from groonga: key=<
>
+Warning	1026	failed to get record ID for deleting from groonga: key=<\0001>
 SELECT * FROM diaries;
 id	title	body
 2	groonga (1)	starting groonga...

strings/ctype.c

@@ -985,6 +985,100 @@ my_charset_is_ascii_based(CHARSET_INFO *cs)
 }
 
 
+/**
+  Detect if a Unicode code point is printable.
+*/
+static inline my_bool
+my_is_printable(my_wc_t wc)
+{
+  /*
+    Blocks:
+      U+0000 .. U+001F     control
+      U+0020 .. U+007E     printable
+      U+007F .. U+009F     control
+      U+00A0 .. U+00FF     printable
+      U+0100 .. U+10FFFF   As of Unicode-6.1.0, this range does not have any
+                           characters of the "Cc" (Other, control) category.
+                           Should be mostly safe to print.
+                           Except for the surrogate halfs,
+                           which are encoding components, not real characters.
+  */
+  if (wc >= 0x20 && wc <= 0x7E) /* Quickly detect ASCII printable */
+    return TRUE;
+  if (wc <= 0x9F)    /* The rest of U+0000..U+009F are control characters */
+  {
+    /* NL, CR, TAB are Ok */
+    return (wc == '\r' || wc == '\n' || wc == '\t');
+  }
+  /*
+    Surrogate halfs (when alone) print badly in terminals:
+      SELECT _ucs2 0xD800;
+    Let's escape them as well.
+  */
+  if (wc >= 0xD800 && wc <= 0xDFFF)
+    return FALSE;
+  return TRUE;
+}
+
+
+static uint to_printable_8bit(uchar *dst, my_wc_t wc)
+{
+  /*
+    This function is used only in context of error messages for now.
+    All non-BMP characters are currently replaced to question marks
+    when a message is put into diagnostics area.
+  */
+  DBUG_ASSERT(wc < 0x10000);
+  *dst++= '\\';
+  *dst++= _dig_vec_upper[(wc >> 12) & 0x0F];
+  *dst++= _dig_vec_upper[(wc >> 8) & 0x0F];
+  *dst++= _dig_vec_upper[(wc >> 4) & 0x0F];
+  *dst++= _dig_vec_upper[wc & 0x0F];
+  return MY_CS_PRINTABLE_CHAR_LENGTH;
+}
+
+
+/**
+  Encode an Unicode character "wc" into a printable string.
+  This function is suitable for any character set, including
+  ASCII-incompatible multi-byte character sets, e.g. ucs2, utf16, utf32.
+*/
+int
+my_wc_to_printable_generic(CHARSET_INFO *cs, my_wc_t wc,
+                           uchar *str, uchar *end)
+{
+  uchar *str0;
+  uint i, length;
+  uchar tmp[MY_CS_PRINTABLE_CHAR_LENGTH];
+
+  if (my_is_printable(wc))
+  {
+    int mblen= cs->cset->wc_mb(cs, wc, str, end);
+    if (mblen > 0)
+      return mblen;
+  }
+
+  if (str + MY_CS_PRINTABLE_CHAR_LENGTH * cs->mbminlen > end)
+    return MY_CS_TOOSMALLN(MY_CS_PRINTABLE_CHAR_LENGTH * cs->mbminlen);
+
+  if ((cs->state & MY_CS_NONASCII) == 0)
+    return to_printable_8bit(str, wc);
+
+  length= to_printable_8bit(tmp, wc);
+  str0= str;
+  for (i= 0; i < length; i++)
+  {
+    if (cs->cset->wc_mb(cs, tmp[i], str, end) != (int) cs->mbminlen)
+    {
+      DBUG_ASSERT(0);
+      return MY_CS_ILSEQ;
+    }
+    str+= cs->mbminlen;
+  }
+  return str - str0;
+}
+
+
 /*
   Convert a string between two character sets.
   'to' must be large enough to store (form_length * to_cs->mbmaxlen) bytes.