bk commit into 4.1 tree (gshchepa:1.2671) BUG#29294

From: <gshchepa(at)mysql.com>
Date: Wed Jun 27 2007 - 14:37:15 EDT

Below is the list of changes that have just been committed into a local 4.1 repository of uchum. When uchum does a push these changes will be propagated to the main repository and, within 24 hours after the push, to the public repository.
For information on how to access the public repository see http://dev.mysql.com/doc/mysql/en/installing-source-tree.html

ChangeSet@1.2671, 2007-06-27 23:37:13+05:00, gshchepa@gleb.loc +5 -0   Fixed bug #29294.
  Actually there are was 2 different bugs:

1. SELECT INTO OUTFILE ... FIELDS ENCLOSED BY <special character> escaped strings with <special character> inside with the value from an ESCAPED BY clause. So, for the FIELDS ENCLOSED BY 'r' the string of value "...r..." was encoded as "...\\r..." (here the value of ESCAPED BY is default '\\' character). On the LOAD DATA query this string was correctly decoded as "...\r...", not as the "...r...".
2. LOAD DATA INFILE was trying to "unescape" strings like "\\r" as the '\r' ignoring the fact that the 'r' character is an escaped value of ENCLOSED BY clause and must be decoded as "r".

  mysql-test/r/loaddata.result@1.18, 2007-06-27 22:51:59+05:00, gshchepa@gleb.loc +121 -1     Update test case for bug #29294.

  mysql-test/t/loaddata.test@1.11, 2007-06-27 22:51:46+05:00, gshchepa@gleb.loc +61 -1     Update test case for bug #29294.

  sql/sql_class.cc@1.215, 2007-06-27 22:51:38+05:00, gshchepa@gleb.loc +5 -1     Fixed bug #29294.
    The select_export::send_data method has been modified to     encode field characters of values equal to the field_sep,     as "nn", "tt", "rr" etc (not as "\\n", "\\t" or "\rr"),     when the vale of field_sep is 'n', 't', 'r' etc.

  sql/sql_class.h@1.294, 2007-06-27 22:51:25+05:00, gshchepa@gleb.loc +16 -0     Fixed bug #29294.
    The select_export::is_ambiguous_field_sep field has been added     to distinguish where it is possible to escape field_sep characters     or not.
    The ESCAPE_CHARS macro constant is defined to enumerate     symbolic names of special characters like '\n', '\t' etc.

  sql/sql_load.cc@1.91, 2007-06-27 22:51:43+05:00, gshchepa@gleb.loc +3 -2     Fixed bug #29294.
    The READ_INFO::read_field method has been modified to not     to apply the unescape function to escaped value of the     escape_char.

diff -Nrup a/mysql-test/r/loaddata.result b/mysql-test/r/loaddata.result

--- a/mysql-test/r/loaddata.result	2005-10-22 07:54:29 +05:00

+++ b/mysql-test/r/loaddata.result	2007-06-27 22:51:59 +05:00

@@ -1,4 +1,4 @@
-drop table if exists t1;
+drop table if exists t1,t2;

 create table t1 (a date, b date, c date not null, d date);  load data infile '../../std_data/loaddata1.dat' into table t1 fields terminated by ',';  Warnings:

@@ -85,3 +85,123 @@ field1	field2
 a"b	cd"ef
 a"b	c"d"e

 drop table t1;
+CREATE TABLE t1 (
+id INT AUTO_INCREMENT PRIMARY KEY,
+c1 VARCHAR(255)
+);
+CREATE TABLE t2 (
+id INT,
+c2 VARCHAR(255)
+);
+INSERT INTO t1 (c1) VALUES
+('r'), ('rr'), ('rrr'), ('rrrr'),
+('.r'), ('.rr'), ('.rrr'), ('.rrrr'),
+('r.'), ('rr.'), ('rrr.'), ('rrrr.'),
+('.r.'), ('.rr.'), ('.rrr.'), ('.rrrr.'),
+('\r'), ('\\rr'), ('\\\rr'), ('\\\\rr');
+SELECT * FROM t1;
+id c1
+1 r
+2 rr
+3 rrr
+4 rrrr
+5 .r
+6 .rr
+7 .rrr
+8 .rrrr
+9 r.
+10 rr.
+11 rrr.
+12 rrrr.
+13 .r.
+14 .rr.
+15 .rrr.
+16 .rrrr.
+17
+18 \rr
+19 \r
+20 \\rr
+SELECT * INTO OUTFILE 'MYSQL_TEST_DIR/var/tmp/t1' FIELDS ENCLOSED BY 'r' FROM t1;
+r1r rrrr
+r2r rrrrrr
+r3r rrrrrrrr
+r4r rrrrrrrrrr
+r5r r.rrr
+r6r r.rrrrr
+r7r r.rrrrrrr
+r8r r.rrrrrrrrr
+r9r rrr.r
+r10r rrrrr.r
+r11r rrrrrrr.r
+r12r rrrrrrrrr.r
+r13r r.rr.r
+r14r r.rrrr.r
+r15r r.rrrrrr.r
+r16r r.rrrrrrrr.r
+r17r rr
+r18r r\\rrrrr
+r19r r\\rrr
+r20r r\\\\rrrrr
+LOAD DATA INFILE 'MYSQL_TEST_DIR/var/tmp/t1' INTO TABLE t2 FIELDS ENCLOSED BY 'r';
+SELECT COUNT(*) FROM t2;
+COUNT(*)
+20
+SELECT t1.id, c1, c2 FROM t1 LEFT JOIN t2 ON t1.id=t2.id WHERE c1 != c2;
+id c1 c2
+TRUNCATE t2;
+SELECT * INTO OUTFILE 'MYSQL_TEST_DIR/var/tmp/t1' FIELDS ESCAPED BY 'r' FROM t1;
+1 rr
+2 rrrr
+3 rrrrrr
+4 rrrrrrrr
+5 .rr
+6 .rrrr
+7 .rrrrrr
+8 .rrrrrrrr
+9 rr.
+10 rrrr.
+11 rrrrrr.
+12 rrrrrrrr.
+13 .rr.
+14 .rrrr.
+15 .rrrrrr.
+16 .rrrrrrrr.
+17
+18 \rrrr
+19 \rr
+20 \\rrrr
+LOAD DATA INFILE 'MYSQL_TEST_DIR/var/tmp/t1' INTO TABLE t2 FIELDS ESCAPED BY 'r';
+SELECT COUNT(*) FROM t2;
+COUNT(*)
+20
+SELECT t1.id, c1, c2 FROM t1 LEFT JOIN t2 ON t1.id=t2.id WHERE c1 != c2;
+id c1 c2
+TRUNCATE t2;
+SELECT * INTO OUTFILE 'MYSQL_TEST_DIR/var/tmp/t1' FIELDS ENCLOSED BY 'r' ESCAPED BY 'r' FROM t1;
+r1r rrrr
+r2r rrrrrr
+r3r rrrrrrrr
+r4r rrrrrrrrrr
+r5r r.rrr
+r6r r.rrrrr
+r7r r.rrrrrrr
+r8r r.rrrrrrrrr
+r9r rrr.r
+r10r rrrrr.r
+r11r rrrrrrr.r
+r12r rrrrrrrrr.r
+r13r r.rr.r
+r14r r.rrrr.r
+r15r r.rrrrrr.r
+r16r r.rrrrrrrr.r
+r17r rr
+r18r r\rrrrr
+r19r r\rrr
+r20r r\\rrrrr
+LOAD DATA INFILE 'MYSQL_TEST_DIR/var/tmp/t1' INTO TABLE t2 FIELDS ENCLOSED BY 'r' ESCAPED BY 'r';
+SELECT COUNT(*) FROM t2;
+COUNT(*)
+20
+SELECT t1.id, c1, c2 FROM t1 LEFT JOIN t2 ON t1.id=t2.id WHERE c1 != c2;
+id c1 c2
+DROP TABLE t1,t2;

diff -Nrup a/mysql-test/t/loaddata.test b/mysql-test/t/loaddata.test

--- a/mysql-test/t/loaddata.test	2005-10-22 07:54:29 +05:00

+++ b/mysql-test/t/loaddata.test	2007-06-27 22:51:46 +05:00

@@ -3,7 +3,7 @@
 #  

 --disable_warnings
-drop table if exists t1;
+drop table if exists t1,t2;

 --enable_warnings  

 create table t1 (a date, b date, c date not null, d date); @@ -66,5 +66,65 @@ create table t1 (a varchar(20), b varcha  load data infile '../../std_data/loaddata5.dat' into table t1 fields terminated by ',' enclosed by '"' escaped by '"' (a,b);  select * from t1;
 drop table t1;
+
+#
+# Bug #29294 SELECT INTO OUTFILE/LOAD DATA INFILE with special
+# characters in the FIELDS ENCLOSED BY clause
+#
+
+CREATE TABLE t1 (
+ id INT AUTO_INCREMENT PRIMARY KEY,
+ c1 VARCHAR(255)
+);
+
+CREATE TABLE t2 (
+ id INT,
+ c2 VARCHAR(255)
+);
+
+INSERT INTO t1 (c1) VALUES
+ ('r'), ('rr'), ('rrr'), ('rrrr'),
+ ('.r'), ('.rr'), ('.rrr'), ('.rrrr'),
+ ('r.'), ('rr.'), ('rrr.'), ('rrrr.'),
+ ('.r.'), ('.rr.'), ('.rrr.'), ('.rrrr.'),
+ ('\r'), ('\\rr'), ('\\\rr'), ('\\\\rr');
+SELECT * FROM t1;
+
+--exec rm -f $MYSQL_TEST_DIR/var/tmp/t1
+--replace_result $MYSQL_TEST_DIR MYSQL_TEST_DIR
+eval SELECT * INTO OUTFILE '$MYSQL_TEST_DIR/var/tmp/t1' FIELDS ENCLOSED BY 'r' FROM t1;
+--exec cat $MYSQL_TEST_DIR/var/tmp/t1
+
+--replace_result $MYSQL_TEST_DIR MYSQL_TEST_DIR
+eval LOAD DATA INFILE '$MYSQL_TEST_DIR/var/tmp/t1' INTO TABLE t2 FIELDS ENCLOSED BY 'r';
+SELECT COUNT(*) FROM t2;
+SELECT t1.id, c1, c2 FROM t1 LEFT JOIN t2 ON t1.id=t2.id WHERE c1 != c2;
+
+TRUNCATE t2;
+
+--exec rm -f $MYSQL_TEST_DIR/var/tmp/t1
+--replace_result $MYSQL_TEST_DIR MYSQL_TEST_DIR
+eval SELECT * INTO OUTFILE '$MYSQL_TEST_DIR/var/tmp/t1' FIELDS ESCAPED BY 'r' FROM t1;
+--exec cat $MYSQL_TEST_DIR/var/tmp/t1
+
+--replace_result $MYSQL_TEST_DIR MYSQL_TEST_DIR
+eval LOAD DATA INFILE '$MYSQL_TEST_DIR/var/tmp/t1' INTO TABLE t2 FIELDS ESCAPED BY 'r';
+SELECT COUNT(*) FROM t2;
+SELECT t1.id, c1, c2 FROM t1 LEFT JOIN t2 ON t1.id=t2.id WHERE c1 != c2;
+
+TRUNCATE t2;
+
+--exec rm -f $MYSQL_TEST_DIR/var/tmp/t1
+--replace_result $MYSQL_TEST_DIR MYSQL_TEST_DIR
+eval SELECT * INTO OUTFILE '$MYSQL_TEST_DIR/var/tmp/t1' FIELDS ENCLOSED BY 'r' ESCAPED BY 'r' FROM t1;
+--exec cat $MYSQL_TEST_DIR/var/tmp/t1
+
+--replace_result $MYSQL_TEST_DIR MYSQL_TEST_DIR
+eval LOAD DATA INFILE '$MYSQL_TEST_DIR/var/tmp/t1' INTO TABLE t2 FIELDS ENCLOSED BY 'r' ESCAPED BY 'r';
+SELECT COUNT(*) FROM t2;
+SELECT t1.id, c1, c2 FROM t1 LEFT JOIN t2 ON t1.id=t2.id WHERE c1 != c2;
+
+--exec rm $MYSQL_TEST_DIR/var/tmp/t1
+DROP TABLE t1,t2;
 

 # End of 4.1 tests
diff -Nrup a/sql/sql_class.cc b/sql/sql_class.cc

--- a/sql/sql_class.cc	2007-03-20 21:59:23 +04:00

+++ b/sql/sql_class.cc	2007-06-27 22:51:38 +05:00

@@ -1020,6 +1020,7 @@ select_export::prepare(List &list,
   field_sep_char= (exchange->enclosed->length() ? (*exchange->enclosed)[0] :
 		   field_term_length ? (*exchange->field_term)[0] : INT_MAX);
   escape_char=	(exchange->escaped->length() ? (*exchange->escaped)[0] : -1);

+ is_ambiguous_field_sep= strchr(ESCAPE_CHARS, field_sep_char);

   line_sep_char= (exchange->line_term->length() ?

                   (*exchange->line_term)[0] : INT_MAX);    if (!field_term_length)
@@ -1113,7 +1114,10 @@ bool select_export::send_data(List<Item>

 	      (int) *pos == line_sep_char || !*pos)
 	  {
 	    char tmp_buff[2];
-	    tmp_buff[0]= escape_char;

+            if ((int) *pos == field_sep_char && is_ambiguous_field_sep)
+              tmp_buff[0]= field_sep_char;
+            else
+              tmp_buff[0]= escape_char;

 	    tmp_buff[1]= *pos ? *pos : '0';
 	    if (my_b_write(&cache,(byte*) start,(uint) (pos-start)) ||
 		my_b_write(&cache,(byte*) tmp_buff,2))
diff -Nrup a/sql/sql_class.h b/sql/sql_class.h
--- a/sql/sql_class.h	2007-01-15 13:58:02 +04:00

+++ b/sql/sql_class.h	2007-06-27 22:51:25 +05:00

@@ -1224,9 +1224,25 @@ public:
 };    

+#define ESCAPE_CHARS "ntrb0ZN" // keep synchronous with READ_INFO::unescaped
+
+

 class select_export :public select_to_file {    uint field_term_length;
   int field_sep_char,escape_char,line_sep_char;
+ /*
+ The is_ambiguous_field_sep field is true if the value of field_sep_char
+ field is one of the 'n', 't', 'r' etc characters
+ (see the READ_INFO::unescape method and the ESCAPE_CHARS constant value).
+
+ For example, if the value of field_sep_char is 'r', it is incorrect to
+ escape an 'r' character in the string "...r..." with the value of
+ escape_char field, because the combination of escape_char and the 'r'
+ character will be decoded as '\r'. Thus, we should encode the "...r..."
+ string as the "...rr...", not as the "...\\r..." where '\' is a value of
+ escape_char.
+ */
+ bool is_ambiguous_field_sep;

   bool fixed_row_size;
 public:
   select_export(sql_exchange *ex) :select_to_file(ex) {} diff -Nrup a/sql/sql_load.cc b/sql/sql_load.cc

--- a/sql/sql_load.cc	2007-02-26 23:33:55 +04:00

+++ b/sql/sql_load.cc	2007-06-27 22:51:43 +05:00

@@ -611,6 +611,7 @@ read_sep_field(THD *thd,COPY_INFO &info,  char
 READ_INFO::unescape(char chr)
 {
+ /* Keep this switch statement synchronous with ESCAPE_CHARS constant */

   switch(chr) {
   case 'n': return '\n';
   case 't': return '\t';
@@ -824,9 +825,9 @@ int READ_INFO::read_field()

           like \n. This allows: LOAD DATA ... ENCLOSED BY '"' ESCAPED BY '"'
           with data like: "fie""ld1", "field2"
          */
-        if (escape_char != enclosed_char || chr == escape_char)

+        if (escape_char != enclosed_char)

         {
-          *to++ = (byte) unescape((char) chr);

+          *to++= chr == escape_char ? (byte) chr : (byte) unescape((char) chr);

           continue;
         }
         PUSH(chr);

-- 
MySQL Code Commits Mailing List
For list archives: 
http://lists.mysql.com/commits
To unsubscribe:    
http://lists.mysql.com/commits?unsub=lists@pantek.com

Received on Wed Jun 27 14:38:05 2007