Bug #25865525: BACKPORT 25147988 FIX TO 5.7

Description:
LOAD DATA fails to accept the multibyte character which follows escape mark,
but INSERT can.

Fix:
Change READ_INFO::read_field  to ignore the escape mark if it is followed by
mulitibyte character.

Author: Kailasnath Nagarkar

mysql-test/r/loaddata.result

@@ -597,3 +597,24 @@ LOAD DATA INFILE 'test.dat' INTO TABLE v(@a, @d) SET a= @a, d= @d;
 ERROR HY000: Column 'd' is not updatable
 DROP VIEW v;
 DROP TABLE t;
+#
+# Bug #25147988: LOAD DATA INFILE FAILS WITH AN ESCAPE CHARACTER
+#                FOLLOWED BY A MULTI-BYTE ONE
+#
+CREATE TABLE t1(a VARCHAR(20)) CHARSET utf8mb4;
+LOAD DATA INFILE '../../std_data/loaddata_utf8.dat' INTO TABLE t1 CHARACTER SET utf8mb4;
+SELECT HEX(a) FROM t1;
+HEX(a)
+E4B880E4BA8CE4B889
+E59B9BE4BA94E585AD
+E4B883E585ABE4B99D
+E4B880E4BA8CE4B889
+E59B9BE4BA94E585AD
+E4B883E585ABE4B99D0A
+DROP TABLE t1;
+CREATE TABLE t1(a VARCHAR(20)) CHARSET gb18030;
+LOAD DATA INFILE '../../std_data/loaddata7.dat' INTO TABLE t1 CHARACTER SET gb18030;
+SELECT HEX(a) FROM t1;
+HEX(a)
+815C825C
+DROP TABLE t1;

mysql-test/std_data/loaddata7.dat

@@ -0,0 +1 @@
+\�\\�\

mysql-test/std_data/loaddata_utf8.dat

@@ -1,3 +1,6 @@
 一二三
 四五六
 七八九
+\一二三
+四\五六
+七八九\

mysql-test/suite/rpl/r/rpl_loaddata_charset.result

@@ -57,13 +57,19 @@ hex(cl)
 E4B880E4BA8CE4B889
 E59B9BE4BA94E585AD
 E4B883E585ABE4B99D
+E4B880E4BA8CE4B889
+E59B9BE4BA94E585AD
+E4B883E585ABE4B99D0A
 ----------content on slave----------
 USE mysqltest;
 SELECT hex(cl) FROM t;
 hex(cl)
 E4B880E4BA8CE4B889
 E59B9BE4BA94E585AD
 E4B883E585ABE4B99D
+E4B880E4BA8CE4B889
+E59B9BE4BA94E585AD
+E4B883E585ABE4B99D0A
 DROP DATABASE mysqltest;
 DROP DATABASE IF EXISTS mysqltest;
 CREATE DATABASE mysqltest CHARSET UTF8;
@@ -77,12 +83,18 @@ hex(cl)
 E4B880E4BA8CE4B889
 E59B9BE4BA94E585AD
 E4B883E585ABE4B99D
+E4B880E4BA8CE4B889
+E59B9BE4BA94E585AD
+E4B883E585ABE4B99D0A
 ----------content on slave----------
 USE mysqltest;
 SELECT hex(cl) FROM t;
 hex(cl)
 E4B880E4BA8CE4B889
 E59B9BE4BA94E585AD
 E4B883E585ABE4B99D
+E4B880E4BA8CE4B889
+E59B9BE4BA94E585AD
+E4B883E585ABE4B99D0A
 DROP DATABASE mysqltest;
 include/rpl_end.inc

mysql-test/t/loaddata.test

@@ -709,3 +709,18 @@ LOAD DATA INFILE 'test.dat' INTO TABLE v(@a, @d) SET a= @a, d= @d;
 
 DROP VIEW v;
 DROP TABLE t;
+
+--echo #
+--echo # Bug #25147988: LOAD DATA INFILE FAILS WITH AN ESCAPE CHARACTER
+--echo #                FOLLOWED BY A MULTI-BYTE ONE
+--echo #
+# Test escape mark followed by multibyte character
+CREATE TABLE t1(a VARCHAR(20)) CHARSET utf8mb4;
+LOAD DATA INFILE '../../std_data/loaddata_utf8.dat' INTO TABLE t1 CHARACTER SET utf8mb4;
+SELECT HEX(a) FROM t1;
+DROP TABLE t1;
+# Test multibyte character whose second byte is 0x5C
+CREATE TABLE t1(a VARCHAR(20)) CHARSET gb18030;
+LOAD DATA INFILE '../../std_data/loaddata7.dat' INTO TABLE t1 CHARACTER SET gb18030;
+SELECT HEX(a) FROM t1;
+DROP TABLE t1;

sql/sql_load.cc

@@ -1644,6 +1644,7 @@ int READ_INFO::read_field()
 
   for (;;)
   {
+    bool escaped_mb= false;
     while ( to < end_of_buff)
     {
       chr = GET;
@@ -1665,7 +1666,23 @@ int READ_INFO::read_field()
          */
         if (escape_char != enclosed_char || chr == escape_char)
         {
-          *to++ = (uchar) unescape((char) chr);
+          uint ml;
+          GET_MBCHARLEN(read_charset, chr, ml);
+          /*
+            For escaped multibyte character, push back the first byte,
+            and will handle it below.
+            Because multibyte character's second byte is possible to be
+            0x5C, per Query_result_export::send_data, both head byte and
+            tail byte are escaped for such characters. So mark it if the
+            head byte is escaped and will handle it below.
+          */
+          if (ml == 1)
+            *to++= (uchar) unescape((char) chr);
+          else
+          {
+            escaped_mb= true;
+            PUSH(chr);
+          }
           continue;
         }
         PUSH(chr);
@@ -1757,8 +1774,16 @@ int READ_INFO::read_field()
             to-= i;
             goto found_eof;
           }
+          else if (chr == escape_char && escaped_mb)
+          {
+            // Unescape the second byte if it is escaped.
+            chr= GET;
+            chr= (uchar) unescape((char) chr);
+          }
           *to++ = chr;
         }
+        if (escaped_mb)
+          escaped_mb= false;
         if (my_ismbchar(read_charset,
                         (const char *)p,
                         (const char *)to))