Multi-level collation in UCA, Thai sorting with contraction for UTF8.

fb35b9ad · pruet · 9c9747fe · fb35b9ad · fb35b9ad · fb35b9ad
Commit fb35b9ad authored May 26, 2016 by pruet
13 changed files
--- a/include/m_ctype.h
+++ b/include/m_ctype.h
@@ -101,7 +101,7 @@ extern MY_UNICASE_INFO my_unicase_unicode520;
 */
 #define MY_UCA_MAX_WEIGHT_SIZE (8+1)               /* Including 0 terminator */
 #define MY_UCA_CONTRACTION_MAX_WEIGHT_SIZE (2*8+1) /* Including 0 terminator */
-#define MY_UCA_WEIGHT_LEVELS   1
+#define MY_UCA_WEIGHT_LEVELS   2

 typedef struct my_contraction_t
 {

--- a/mysql-test/r/ctype_uca.result
+++ b/mysql-test/r/ctype_uca.result
@@ -2706,6 +2706,261 @@ Z,z,Ź,ź,Ż,ż,Ž,ž
 ǁ
 ǂ
 ǃ
+select group_concat(c1 order by c1) from t1 group by c1 collate utf8_thai_520_w2;
+group_concat(c1 order by c1)
+÷
+×
+A,a
+Á,á
+À,à
+Ă,ă
+Ắ,ắ
+Ằ,ằ
+Ẵ,ẵ
+Ẳ,ẳ
+Â,â
+Ấ,ấ
+Ầ,ầ
+Ẫ,ẫ
+Ẩ,ẩ
+Ǎ,ǎ
+Å,å
+Ǻ,ǻ
+Ä,ä
+Ǟ,ǟ
+Ã,ã
+Ǡ,ǡ
+Ą,ą
+Ā,ā
+Ả,ả
+Ạ,ạ
+Ặ,ặ
+Ậ,ậ
+AA,Aa,aA,aa
+AE,Ae,aE,ae
+Æ,æ
+Ǽ,ǽ
+Ǣ,ǣ
+B,b
+ƀ
+Ɓ
+Ƃ,ƃ
+C,c
+Ć,ć
+Ĉ,ĉ
+Č,č
+Ċ,ċ
+Ç,ç
+CH,Ch,cH,ch
+Ƈ,ƈ
+D,d
+Ď,ď
+Đ,đ
+Ð,ð
+DZ,Dz,dZ,dz,Ǳ,ǲ,ǳ
+DŽ,Dž,dŽ,dž,Ǆ,ǅ,ǆ
+Ɖ
+Ɗ
+Ƌ,ƌ
+E,e
+É,é
+È,è
+Ĕ,ĕ
+Ê,ê
+Ế,ế
+Ề,ề
+Ễ,ễ
+Ể,ể
+Ě,ě
+Ë,ë
+Ẽ,ẽ
+Ė,ė
+Ę,ę
+Ē,ē
+Ẻ,ẻ
+Ẹ,ẹ
+Ệ,ệ
+Ǝ,ǝ
+Ə
+Ɛ
+F,f
+Ƒ,ƒ
+G,g
+Ǵ,ǵ
+Ğ,ğ
+Ĝ,ĝ
+Ǧ,ǧ
+Ġ,ġ
+Ģ,ģ
+Ǥ,ǥ
+Ɠ
+Ɣ
+Ƣ,ƣ
+H,h
+Ĥ,ĥ
+Ħ,ħ
+ƕ,Ƕ
+I,i
+Í,í
+Ì,ì
+Ĭ,ĭ
+Î,î
+Ǐ,ǐ
+Ï,ï
+Ĩ,ĩ
+İ
+Į,į
+Ī,ī
+Ỉ,ỉ
+Ị,ị
+IJ,Ij,iJ,ij,Ĳ,ĳ
+ı
+Ɨ
+Ɩ
+J,j
+Ĵ,ĵ
+ǰ
+K,k
+Ǩ,ǩ
+Ķ,ķ
+Ƙ,ƙ
+L,l
+Ĺ,ĺ
+Ľ,ľ
+Ļ,ļ
+Ł,ł
+Ŀ,ŀ
+LJ,Lj,lJ,lj,Ǉ,ǈ,ǉ
+LL,Ll,lL,ll
+ƚ
+ƛ
+M,m
+N,n
+Ń,ń
+Ǹ,ǹ
+Ň,ň
+Ñ,ñ
+Ņ,ņ
+NJ,Nj,nJ,nj,Ǌ,ǋ,ǌ
+Ɲ
+ƞ
+Ŋ,ŋ
+O,o
+Ó,ó
+Ò,ò
+Ŏ,ŏ
+Ô,ô
+Ố,ố
+Ồ,ồ
+Ỗ,ỗ
+Ổ,ổ
+Ǒ,ǒ
+Ö,ö
+Ő,ő
+Õ,õ
+Ø,ø
+Ǿ,ǿ
+Ǫ,ǫ
+Ǭ,ǭ
+Ō,ō
+Ỏ,ỏ
+Ơ,ơ
+Ớ,ớ
+Ờ,ờ
+Ỡ,ỡ
+Ở,ở
+Ợ,ợ
+Ọ,ọ
+Ộ,ộ
+OE,Oe,oE,oe
+Œ,œ
+Ɔ
+Ɵ
+P,p
+Ƥ,ƥ
+Q,q
+ĸ
+R,r
+Ŕ,ŕ
+Ř,ř
+Ŗ,ŗ
+RR,Rr,rR,rr
+Ʀ
+S,s
+Ś,ś
+Ŝ,ŝ
+Š,š
+Ş,ş
+ſ
+SS,Ss,sS,ss
+ß
+Ʃ
+ƪ
+T,t
+Ť,ť
+Ţ,ţ
+ƾ
+Ŧ,ŧ
+ƫ
+Ƭ,ƭ
+Ʈ
+U,u
+Ú,ú
+Ù,ù
+Ŭ,ŭ
+Û,û
+Ǔ,ǔ
+Ů,ů
+Ü,ü
+Ǘ,ǘ
+Ǜ,ǜ
+Ǚ,ǚ
+Ǖ,ǖ
+Ű,ű
+Ũ,ũ
+Ų,ų
+Ū,ū
+Ủ,ủ
+Ư,ư
+Ứ,ứ
+Ừ,ừ
+Ữ,ữ
+Ử,ử
+Ự,ự
+Ụ,ụ
+Ɯ
+Ʊ
+V,v
+Ʋ
+W,w
+Ŵ,ŵ
+X,x
+Y,y
+Ý,ý
+Ŷ,ŷ
+ÿ,Ÿ
+Ƴ,ƴ
+Z,z
+Ź,ź
+Ž,ž
+Ż,ż
+ƍ
+Ƶ,ƶ
+Ʒ
+Ǯ,ǯ
+Ƹ,ƹ
+ƺ
+Þ,þ
+ƿ,Ƿ
+ƻ
+Ƨ,ƨ
+Ƽ,ƽ
+Ƅ,ƅ
+ŉ
+ǀ
+ǁ
+ǂ
+ǃ
 ALTER TABLE t1 CONVERT TO CHARACTER SET ucs2 COLLATE ucs2_bin;
 SELECT GROUP_CONCAT(c1 ORDER BY c1) FROM t1 GROUP BY c1 COLLATE ucs2_unicode_ci;
 GROUP_CONCAT(c1 ORDER BY c1)

--- a/mysql-test/r/ctype_uca_partitions.result
+++ b/mysql-test/r/ctype_uca_partitions.result
+SET NAMES utf8;
+CREATE TABLE t1 (c1 VARCHAR(10) CHARACTER SET utf8 COLLATE utf8_thai_520_w2)
+PARTITION BY KEY(c1) PARTITIONS 3;
+SHOW CREATE TABLE t1;
+Table	Create Table
+t1	CREATE TABLE `t1` (
+  `c1` varchar(10) CHARACTER SET utf8 COLLATE utf8_thai_520_w2 DEFAULT NULL
+) ENGINE=MyISAM DEFAULT CHARSET=latin1
+/*!50100 PARTITION BY KEY (c1)
+PARTITIONS 3 */
+INSERT INTO t1 VALUES ('A'),('À'),('Á'),('Â'),('Ã'),('Ä'),('Å');
+INSERT INTO t1 VALUES ('B');
+INSERT INTO t1 VALUES ('C');
+INSERT INTO t1 VALUES ('D');
+INSERT INTO t1 VALUES ('E');
+INSERT INTO t1 VALUES ('F');
+INSERT INTO t1 VALUES ('G');
+INSERT INTO t1 VALUES ('H');
+INSERT INTO t1 VALUES ('I');
+INSERT INTO t1 VALUES ('J');
+INSERT INTO t1 VALUES ('K');
+INSERT INTO t1 VALUES ('L'),('Ĺ'),('Ļ'),('Ľ'),('Ŀ'),('Ł');
+INSERT INTO t1 VALUES ('M');
+INSERT INTO t1 VALUES ('N');
+INSERT INTO t1 VALUES ('O'),('Ò'),('Ó'),('Ô'),('Õ'),('Ö'),('Ø');
+INSERT INTO t1 VALUES ('P');
+INSERT INTO t1 VALUES ('Q');
+INSERT INTO t1 VALUES ('R');
+INSERT INTO t1 VALUES ('S');
+INSERT INTO t1 VALUES ('T');
+INSERT INTO t1 VALUES ('U');
+INSERT INTO t1 VALUES ('V');
+INSERT INTO t1 VALUES ('W');
+INSERT INTO t1 VALUES ('X');
+INSERT INTO t1 VALUES ('Y');
+INSERT INTO t1 VALUES ('Z');
+SELECT * FROM t1 PARTITION (p0) ORDER BY c1;
+c1
+J
+K
+L
+Ĺ
+Ľ
+Ļ
+Ł
+Ŀ
+M
+Q
+R
+T
+V
+X
+SELECT * FROM t1 PARTITION (p1) ORDER BY c1;
+c1
+A
+Á
+À
+Â
+Å
+Ä
+Ã
+D
+E
+F
+G
+I
+N
+S
+U
+W
+Z
+SELECT * FROM t1 PARTITION (p2) ORDER BY c1;
+c1
+B
+C
+H
+O
+Ó
+Ò
+Ô
+Ö
+Õ
+Ø
+P
+Y
+DROP TABLE t1;
--- a/mysql-test/r/ctype_uca_th.result
+++ b/mysql-test/r/ctype_uca_th.result
+SET NAMES utf8;
+#
+# Testing strnxfrm
+#
+CREATE TABLE t1 (a VARCHAR(3) CHARACTER SET utf8 COLLATE utf8_thai_520_w2);
+INSERT INTO t1 VALUES (SPACE(0)),(SPACE(1)),(SPACE(2));
+SELECT HEX(a), HEX(WEIGHT_STRING(a AS CHAR(8))) FROM t1;
+HEX(a)	HEX(WEIGHT_STRING(a AS CHAR(8)))
+	020A020A020A020A020A020A020A020A00200020002000200020002000200020
+20	020A020A020A020A020A020A020A020A00200020002000200020002000200020
+2020	020A020A020A020A020A020A020A020A00200020002000200020002000200020
+SELECT HEX(a), HEX(WEIGHT_STRING(a AS CHAR(4))) FROM t1;
+HEX(a)	HEX(WEIGHT_STRING(a AS CHAR(4)))
+	020A020A020A020A0020002000200020
+20	020A020A020A020A0020002000200020
+2020	020A020A020A020A0020002000200020
+SELECT HEX(a), HEX(WEIGHT_STRING(a AS CHAR(4) LEVEL 1)) FROM t1;
+HEX(a)	HEX(WEIGHT_STRING(a AS CHAR(4) LEVEL 1))
+	020A020A020A020A
+20	020A020A020A020A
+2020	020A020A020A020A
+SELECT HEX(a), HEX(WEIGHT_STRING(a AS CHAR(4) LEVEL 2)) FROM t1;
+HEX(a)	HEX(WEIGHT_STRING(a AS CHAR(4) LEVEL 2))
+	0020002000200020
+20	0020002000200020
+2020	0020002000200020
+SELECT HEX(a), HEX(WEIGHT_STRING(a AS CHAR(4) LEVEL 3)) FROM t1;
+HEX(a)	HEX(WEIGHT_STRING(a AS CHAR(4) LEVEL 3))
+	0020002000200020
+20	0020002000200020
+2020	0020002000200020
+DROP TABLE t1;
+CREATE TABLE t1 (a CHAR(10) CHARACTER SET utf8 COLLATE utf8_thai_520_w2);
+INSERT INTO t1 VALUES ('A'),('À'),('Á'),('Â'),('Ã'),('Ä'),('Å');
+SELECT a, HEX(WEIGHT_STRING(a LEVEL 2)) FROM t1 ORDER BY a;
+a	HEX(WEIGHT_STRING(a LEVEL 2))
+A	0020
+Á	00200032
+À	00200035
+Â	0020003C
+Å	00200043
+Ä	00200047
+Ã	0020004E
+SELECT a, HEX(WEIGHT_STRING(a LEVEL 2)) FROM t1 ORDER BY a DESC;
+a	HEX(WEIGHT_STRING(a LEVEL 2))
+Ã	0020004E
+Ä	00200047
+Å	00200043
+Â	0020003C
+À	00200035
+Á	00200032
+A	0020
+ALTER TABLE t1 MODIFY a VARCHAR(10) CHARACTER SET utf8 COLLATE utf8_thai_520_w2;
+SELECT a, HEX(WEIGHT_STRING(a LEVEL 2)) FROM t1 ORDER BY a;
+a	HEX(WEIGHT_STRING(a LEVEL 2))
+A	0020
+Á	00200032
+À	00200035
+Â	0020003C
+Å	00200043
+Ä	00200047
+Ã	0020004E
+SELECT a, HEX(WEIGHT_STRING(a LEVEL 2)) FROM t1 ORDER BY a DESC;
+a	HEX(WEIGHT_STRING(a LEVEL 2))
+Ã	0020004E
+Ä	00200047
+Å	00200043
+Â	0020003C
+À	00200035
+Á	00200032
+A	0020
+DROP TABLE t1;
+CREATE TABLE t1 (a VARCHAR(10) CHARACTER SET utf8 COLLATE utf8_thai_520_w2);
+INSERT INTO t1 VALUES ('cota'),('cote'),('cotz');
+INSERT INTO t1 VALUES ('coté'),('côte'),('côté'),('cotë'),('côtë');
+SELECT * FROM t1 ORDER BY a;
+a
+cota
+cote
+coté
+cotë
+côte
+côté
+côtë
+cotz
+SELECT * FROM t1 ORDER BY a DESC;
+a
+cotz
+côtë
+côté
+côte
+cotë
+coté
+cote
+cota
+DROP TABLE t1;
--- a/mysql-test/r/ctype_utf8_th.result
+++ b/mysql-test/r/ctype_utf8_th.result
--- a/mysql-test/t/ctype_uca.test
+++ b/mysql-test/t/ctype_uca.test
@@ -70,6 +70,7 @@ select group_concat(c1 order by c1) from t1 group by c1 collate utf8_croatian_ci
 select group_concat(c1 order by c1) from t1 group by c1 collate utf8_german2_ci;
 select group_concat(c1 order by c1) from t1 group by c1 collate utf8_unicode_520_ci;
 select group_concat(c1 order by c1) from t1 group by c1 collate utf8_vietnamese_ci;
+select group_concat(c1 order by c1) from t1 group by c1 collate utf8_thai_520_w2;

 ALTER TABLE t1 CONVERT TO CHARACTER SET ucs2 COLLATE ucs2_bin;
 SELECT GROUP_CONCAT(c1 ORDER BY c1) FROM t1 GROUP BY c1 COLLATE ucs2_unicode_ci;

--- a/mysql-test/t/ctype_uca_partitions.test
+++ b/mysql-test/t/ctype_uca_partitions.test
+--source include/have_partition.inc
+
+SET NAMES utf8;
+
+CREATE TABLE t1 (c1 VARCHAR(10) CHARACTER SET utf8 COLLATE utf8_thai_520_w2)
+PARTITION BY KEY(c1) PARTITIONS 3;
+SHOW CREATE TABLE t1;
+INSERT INTO t1 VALUES ('A'),('À'),('Á'),('Â'),('Ã'),('Ä'),('Å');
+INSERT INTO t1 VALUES ('B');
+INSERT INTO t1 VALUES ('C');
+INSERT INTO t1 VALUES ('D');
+INSERT INTO t1 VALUES ('E');
+INSERT INTO t1 VALUES ('F');
+INSERT INTO t1 VALUES ('G');
+INSERT INTO t1 VALUES ('H');
+INSERT INTO t1 VALUES ('I');
+INSERT INTO t1 VALUES ('J');
+INSERT INTO t1 VALUES ('K');
+INSERT INTO t1 VALUES ('L'),('Ĺ'),('Ļ'),('Ľ'),('Ŀ'),('Ł');
+INSERT INTO t1 VALUES ('M');
+INSERT INTO t1 VALUES ('N');
+INSERT INTO t1 VALUES ('O'),('Ò'),('Ó'),('Ô'),('Õ'),('Ö'),('Ø');
+INSERT INTO t1 VALUES ('P');
+INSERT INTO t1 VALUES ('Q');
+INSERT INTO t1 VALUES ('R');
+INSERT INTO t1 VALUES ('S');
+INSERT INTO t1 VALUES ('T');
+INSERT INTO t1 VALUES ('U');
+INSERT INTO t1 VALUES ('V');
+INSERT INTO t1 VALUES ('W');
+INSERT INTO t1 VALUES ('X');
+INSERT INTO t1 VALUES ('Y');
+INSERT INTO t1 VALUES ('Z');
+
+SELECT * FROM t1 PARTITION (p0) ORDER BY c1;
+SELECT * FROM t1 PARTITION (p1) ORDER BY c1;
+SELECT * FROM t1 PARTITION (p2) ORDER BY c1;
+DROP TABLE t1;
--- a/mysql-test/t/ctype_uca_th.test
+++ b/mysql-test/t/ctype_uca_th.test
+
+
+SET NAMES utf8;
+
+
+--echo #
+--echo # Testing strnxfrm
+--echo #
+
+CREATE TABLE t1 (a VARCHAR(3) CHARACTER SET utf8 COLLATE utf8_thai_520_w2);
+INSERT INTO t1 VALUES (SPACE(0)),(SPACE(1)),(SPACE(2));
+SELECT HEX(a), HEX(WEIGHT_STRING(a AS CHAR(8))) FROM t1;
+SELECT HEX(a), HEX(WEIGHT_STRING(a AS CHAR(4))) FROM t1;
+SELECT HEX(a), HEX(WEIGHT_STRING(a AS CHAR(4) LEVEL 1)) FROM t1;
+SELECT HEX(a), HEX(WEIGHT_STRING(a AS CHAR(4) LEVEL 2)) FROM t1;
+SELECT HEX(a), HEX(WEIGHT_STRING(a AS CHAR(4) LEVEL 3)) FROM t1;
+DROP TABLE t1;
+
+CREATE TABLE t1 (a CHAR(10) CHARACTER SET utf8 COLLATE utf8_thai_520_w2);
+INSERT INTO t1 VALUES ('A'),('À'),('Á'),('Â'),('Ã'),('Ä'),('Å');
+SELECT a, HEX(WEIGHT_STRING(a LEVEL 2)) FROM t1 ORDER BY a;
+SELECT a, HEX(WEIGHT_STRING(a LEVEL 2)) FROM t1 ORDER BY a DESC;
+
+ALTER TABLE t1 MODIFY a VARCHAR(10) CHARACTER SET utf8 COLLATE utf8_thai_520_w2;
+SELECT a, HEX(WEIGHT_STRING(a LEVEL 2)) FROM t1 ORDER BY a;
+SELECT a, HEX(WEIGHT_STRING(a LEVEL 2)) FROM t1 ORDER BY a DESC;
+DROP TABLE t1;
+
+CREATE TABLE t1 (a VARCHAR(10) CHARACTER SET utf8 COLLATE utf8_thai_520_w2);
+INSERT INTO t1 VALUES ('cota'),('cote'),('cotz');
+INSERT INTO t1 VALUES ('coté'),('côte'),('côté'),('cotë'),('côtë');
+SELECT * FROM t1 ORDER BY a;
+SELECT * FROM t1 ORDER BY a DESC;
+DROP TABLE t1;
--- a/mysql-test/t/ctype_utf8_th.test
+++ b/mysql-test/t/ctype_utf8_th.test
+--disable_warnings
+DROP TABLE IF EXISTS t1;
+--enable_warnings
+
+#
+# Test Unicode Thai collations based on libthai testcases
+#
+
+SET NAMES utf8;
+
+CREATE TABLE t1 (a VARCHAR(30) CHARACTER SET utf8 COLLATE utf8_bin);
+
+INSERT INTO t1 VALUES ('-กระแย่ง');
+INSERT INTO t1 VALUES ('กก');
+INSERT INTO t1 VALUES ('กราบ');
+INSERT INTO t1 VALUES ('ข่าง');
+INSERT INTO t1 VALUES ('ข้าง');
+INSERT INTO t1 VALUES ('ข้างกระดาน');
+INSERT INTO t1 VALUES ('ข้างขึ้น');
+INSERT INTO t1 VALUES ('ข้างควาย');
+INSERT INTO t1 VALUES ('ข้างเงิน');
+INSERT INTO t1 VALUES ('ข้างแรม');
+INSERT INTO t1 VALUES ('ข้างออก');
+INSERT INTO t1 VALUES ('ข้างๆ');
+INSERT INTO t1 VALUES ('ข้างๆ คูๆ');
+INSERT INTO t1 VALUES ('ขาง');
+INSERT INTO t1 VALUES ('แข็ง');
+INSERT INTO t1 VALUES ('แข่ง');
+INSERT INTO t1 VALUES ('แข่งขัน');
+INSERT INTO t1 VALUES ('แข้ง');
+INSERT INTO t1 VALUES ('แข้งขวา');
+INSERT INTO t1 VALUES ('แข็งขัน');
+INSERT INTO t1 VALUES ('ทูลเกล้า');
+INSERT INTO t1 VALUES ('ทูลเกล้าทูลกระหม่อม');
+INSERT INTO t1 VALUES ('ทูลเกล้าฯ');
+INSERT INTO t1 VALUES ('บุญญา');
+INSERT INTO t1 VALUES ('บุญ-หลง');
+INSERT INTO t1 VALUES ('บุญหลง');
+INSERT INTO t1 VALUES ('ป่า');
+INSERT INTO t1 VALUES ('ป่าน');
+INSERT INTO t1 VALUES ('ป้า');
+INSERT INTO t1 VALUES ('ป้าน');
+INSERT INTO t1 VALUES ('ป๊า');
+INSERT INTO t1 VALUES ('ป๊าน');
+INSERT INTO t1 VALUES ('ป๋า');
+INSERT INTO t1 VALUES ('ป๋าน');
+INSERT INTO t1 VALUES ('ปา');
+INSERT INTO t1 VALUES ('ปาน');
+INSERT INTO t1 VALUES ('แป้ง');
+INSERT INTO t1 VALUES ('พณิชย์');
+INSERT INTO t1 VALUES ('ม้า');
+INSERT INTO t1 VALUES ('ฯพณฯ');
+INSERT INTO t1 VALUES ('A');
+INSERT INTO t1 VALUES ('a');
+INSERT INTO t1 VALUES ('a\'');
+INSERT INTO t1 VALUES ('A-');
+INSERT INTO t1 VALUES ('a-');
+INSERT INTO t1 VALUES ('-a');
+INSERT INTO t1 VALUES ('A.');
+INSERT INTO t1 VALUES ('a.');
+INSERT INTO t1 VALUES ('A-1');
+INSERT INTO t1 VALUES ('aa');
+INSERT INTO t1 VALUES ('AA');
+INSERT INTO t1 VALUES ('A.A.');
+INSERT INTO t1 VALUES ('a.a.');
+INSERT INTO t1 VALUES ('AAA');
+INSERT INTO t1 VALUES ('A.A.A.');
+INSERT INTO t1 VALUES ('AAAA');
+INSERT INTO t1 VALUES ('A.A.A.L.');
+INSERT INTO t1 VALUES ('A.A.A.S.');
+INSERT INTO t1 VALUES ('Aachen');
+INSERT INTO t1 VALUES ('A.A.E.');
+INSERT INTO t1 VALUES ('A.Ae.E.');
+INSERT INTO t1 VALUES ('A.A.E.E.');
+INSERT INTO t1 VALUES ('AAES');
+INSERT INTO t1 VALUES ('AAF');
+INSERT INTO t1 VALUES ('A.Agr');
+INSERT INTO t1 VALUES ('aah');
+INSERT INTO t1 VALUES ('@@@@@');
+INSERT INTO t1 VALUES ('0000');
+INSERT INTO t1 VALUES ('9999');
+INSERT INTO t1 VALUES ('Aalborg');
+INSERT INTO t1 VALUES ('aide');
+INSERT INTO t1 VALUES ('air');
+INSERT INTO t1 VALUES ('@@@air');
+INSERT INTO t1 VALUES ('air@@@');
+INSERT INTO t1 VALUES ('C.A.F');
+INSERT INTO t1 VALUES ('Canon');
+INSERT INTO t1 VALUES ('coop');
+INSERT INTO t1 VALUES ('co-op');
+INSERT INTO t1 VALUES ('COOP');
+INSERT INTO t1 VALUES ('CO-OP');
+INSERT INTO t1 VALUES ('Copenhegen');
+INSERT INTO t1 VALUES ('McArthur');
+INSERT INTO t1 VALUES ('Mc Arthur');
+INSERT INTO t1 VALUES ('Mc Mahon');
+INSERT INTO t1 VALUES ('vice-president');
+INSERT INTO t1 VALUES ('vice versa');
+INSERT INTO t1 VALUES ('vice-versa');
+INSERT INTO t1 VALUES ('10 ลิตร');
+INSERT INTO t1 VALUES ('๑๐ ลิตร');
+INSERT INTO t1 VALUES ('10 litre');
+INSERT INTO t1 VALUES ('10 litre (10 ลิตร)');
+INSERT INTO t1 VALUES ('10 ลิตร (10 litre)');
+INSERT INTO t1 VALUES ('10 litre (๑๐ ลิตร)');
+INSERT INTO t1 VALUES ('๑๐ ลิตร (10 litre)');
+INSERT INTO t1 VALUES ('๑๐ ลิตร [10 litre]');
+INSERT INTO t1 VALUES ('๑๐ ลิตร {10 litre}');
+
+# Sort the table according to binary 'a'
+# To avoid problems with unpredictable order or 'AA' and 'aa'
+ALTER TABLE t1 ORDER BY a;
+
+ALTER TABLE t1
+  MODIFY a VARCHAR(30) CHARACTER SET utf8 COLLATE utf8_thai_520_w2,
+  ADD id INT NOT NULL AUTO_INCREMENT PRIMARY KEY,
+  ADD KEY a_id (a, id);
+
+--echo #
+--echo # Ascending sort, using filesort
+--echo #
+EXPLAIN SELECT a FROM t1 ORDER BY a, BINARY a;
+SELECT a FROM t1 ORDER BY a, BINARY a;
+
+--echo #
+--echo # Descending sort, using filesort
+--echo #
+EXPLAIN SELECT a FROM t1 ORDER BY a DESC, BINARY a DESC;
+SELECT a FROM t1 ORDER BY a DESC, BINARY a DESC;
+
+
+--echo #
+--echo # Ascending sort, using index
+--echo #
+EXPLAIN SELECT a FROM t1 ORDER BY a, id;
+SELECT a FROM t1 ORDER BY a, id;
+
+--echo #
+--echo # Descending sort, using index
+--echo #
+EXPLAIN SELECT a FROM t1 ORDER BY a DESC, id DESC;
+SELECT a FROM t1 ORDER BY a DESC, id DESC;
+
+DROP TABLE t1;
--- a/mysys/charset-def.c
+++ b/mysys/charset-def.c
@@ -133,6 +133,7 @@ extern struct charset_info_st my_charset_utf8_hungarian_uca_ci;
 extern struct charset_info_st my_charset_utf8_croatian_mysql561_uca_ci;
 extern struct charset_info_st my_charset_utf8_sinhala_uca_ci;
 extern struct charset_info_st my_charset_utf8_unicode_520_ci;
+extern struct charset_info_st my_charset_utf8_thai_520_w2;
 extern struct charset_info_st my_charset_utf8_vietnamese_ci;
 extern struct charset_info_st my_charset_utf8_croatian_uca_ci;
 extern struct charset_info_st my_charset_utf8_myanmar_uca_ci;
@@ -300,6 +301,7 @@ my_bool init_compiled_charsets(myf flags __attribute__((unused)))
  add_compiled_collation(&my_charset_utf8_croatian_mysql561_uca_ci);
  add_compiled_collation(&my_charset_utf8_sinhala_uca_ci);
  add_compiled_collation(&my_charset_utf8_unicode_520_ci);
+  add_compiled_collation(&my_charset_utf8_thai_520_w2);
  add_compiled_collation(&my_charset_utf8_vietnamese_ci);
  add_compiled_collation(&my_charset_utf8_croatian_uca_ci);
  add_compiled_collation(&my_charset_utf8_myanmar_uca_ci);

--- a/sql/item_strfunc.cc
+++ b/sql/item_strfunc.cc
@@ -3536,7 +3536,7 @@ void Item_func_weight_string::fix_length_and_dec()
  {
    uint char_length;
    char_length= ((cs->state & MY_CS_STRNXFRM_BAD_NWEIGHTS) || !nweights) ?
-                 args[0]->max_char_length() : nweights;
+                 args[0]->max_char_length() : nweights * cs->levels_for_order;
    max_length= cs->coll->strnxfrmlen(cs, char_length * cs->mbmaxlen);
  }
  maybe_null= 1;

--- a/strings/ctype-tis620.c
+++ b/strings/ctype-tis620.c
@@ -40,7 +40,6 @@

 #ifdef HAVE_CHARSET_tis620

-#define BUFFER_MULTIPLY 4
 #define M  L_MIDDLE
 #define U  L_UPPER
 #define L  L_LOWER

--- a/strings/ctype-uca.c
+++ b/strings/ctype-uca.c