Commit a616c3df authored by unknown's avatar unknown
Browse files

utf8 contraction fix. 

Polish collation fix.
Slovak collation fix.
Testing all language orders.
parent 10ce930c

mysql-test/r/ctype_uca.result

0 → 100644
+1657 −0

File added.

Preview size limit exceeded, changes collapsed.

mysql-test/t/ctype_uca.test

0 → 100644
+181 −0
Original line number Diff line number Diff line 
-- source include/have_ucs2.inc



--disable_warnings

DROP TABLE IF EXISTS t1;

--enable_warnings



#

# Test Unicode collations.

#



set names utf8; 

create table t1 (c1 char(10) character set utf8 collate utf8_bin);



#

# Basic Latin

#

insert into t1 values ('A'),('a');

insert into t1 values ('B'),('b');

insert into t1 values ('C'),('c');

insert into t1 values ('D'),('d');

insert into t1 values ('E'),('e');

insert into t1 values ('F'),('f');

insert into t1 values ('G'),('g');

insert into t1 values ('H'),('h');

insert into t1 values ('I'),('i');

insert into t1 values ('J'),('j');

insert into t1 values ('K'),('k');

insert into t1 values ('L'),('l');

insert into t1 values ('M'),('m');

insert into t1 values ('N'),('n');

insert into t1 values ('O'),('o');

insert into t1 values ('P'),('p');

insert into t1 values ('Q'),('q');

insert into t1 values ('R'),('r');

insert into t1 values ('S'),('s');

insert into t1 values ('T'),('t');

insert into t1 values ('U'),('u');

insert into t1 values ('V'),('v');

insert into t1 values ('W'),('w');

insert into t1 values ('X'),('x');

insert into t1 values ('Y'),('y');

insert into t1 values ('Z'),('z');



#

# Latin1 suppliment

#

insert into t1 values (_ucs2 0x00e0),(_ucs2 0x00c0);

insert into t1 values (_ucs2 0x00e1),(_ucs2 0x00c1);

insert into t1 values (_ucs2 0x00e2),(_ucs2 0x00c2);

insert into t1 values (_ucs2 0x00e3),(_ucs2 0x00c3);

insert into t1 values (_ucs2 0x00e4),(_ucs2 0x00c4);

insert into t1 values (_ucs2 0x00e5),(_ucs2 0x00c5);

insert into t1 values (_ucs2 0x00e6),(_ucs2 0x00c6);

insert into t1 values (_ucs2 0x00e7),(_ucs2 0x00c7);

insert into t1 values (_ucs2 0x00e8),(_ucs2 0x00c8);

insert into t1 values (_ucs2 0x00e9),(_ucs2 0x00c9);

insert into t1 values (_ucs2 0x00ea),(_ucs2 0x00ca);

insert into t1 values (_ucs2 0x00eb),(_ucs2 0x00cb);

insert into t1 values (_ucs2 0x00ec),(_ucs2 0x00cc);

insert into t1 values (_ucs2 0x00ed),(_ucs2 0x00cd);

insert into t1 values (_ucs2 0x00ee),(_ucs2 0x00ce);

insert into t1 values (_ucs2 0x00ef),(_ucs2 0x00cf);



insert into t1 values (_ucs2 0x00f0),(_ucs2 0x00d0);

insert into t1 values (_ucs2 0x00f1),(_ucs2 0x00d1);

insert into t1 values (_ucs2 0x00f2),(_ucs2 0x00d2);

insert into t1 values (_ucs2 0x00f3),(_ucs2 0x00d3);

insert into t1 values (_ucs2 0x00f4),(_ucs2 0x00d4);

insert into t1 values (_ucs2 0x00f5),(_ucs2 0x00d5);

insert into t1 values (_ucs2 0x00f6),(_ucs2 0x00d6);

insert into t1 values (_ucs2 0x00f7),(_ucs2 0x00d7);

insert into t1 values (_ucs2 0x00f8),(_ucs2 0x00d8);

insert into t1 values (_ucs2 0x00f9),(_ucs2 0x00d9);

insert into t1 values (_ucs2 0x00fa),(_ucs2 0x00da);

insert into t1 values (_ucs2 0x00fb),(_ucs2 0x00db);

insert into t1 values (_ucs2 0x00fc),(_ucs2 0x00dc);

insert into t1 values (_ucs2 0x00fd),(_ucs2 0x00dd);

insert into t1 values (_ucs2 0x00fe),(_ucs2 0x00de);

insert into t1 values (_ucs2 0x00ff),(_ucs2 0x00df);



#

# Latin extended-A, 0100-017F

#

insert into t1 values (_ucs2 0x0100),(_ucs2 0x0101),(_ucs2 0x0102),(_ucs2 0x0103);

insert into t1 values (_ucs2 0x0104),(_ucs2 0x0105),(_ucs2 0x0106),(_ucs2 0x0107);

insert into t1 values (_ucs2 0x0108),(_ucs2 0x0109),(_ucs2 0x010a),(_ucs2 0x010b);

insert into t1 values (_ucs2 0x010c),(_ucs2 0x010d),(_ucs2 0x010e),(_ucs2 0x010f);

insert into t1 values (_ucs2 0x0110),(_ucs2 0x0111),(_ucs2 0x0112),(_ucs2 0x0113);

insert into t1 values (_ucs2 0x0114),(_ucs2 0x0115),(_ucs2 0x0116),(_ucs2 0x0117);

insert into t1 values (_ucs2 0x0118),(_ucs2 0x0119),(_ucs2 0x011a),(_ucs2 0x011b);

insert into t1 values (_ucs2 0x011c),(_ucs2 0x011d),(_ucs2 0x011e),(_ucs2 0x011f);

insert into t1 values (_ucs2 0x0120),(_ucs2 0x0121),(_ucs2 0x0122),(_ucs2 0x0123);

insert into t1 values (_ucs2 0x0124),(_ucs2 0x0125),(_ucs2 0x0126),(_ucs2 0x0127);

insert into t1 values (_ucs2 0x0128),(_ucs2 0x0129),(_ucs2 0x012a),(_ucs2 0x012b);

insert into t1 values (_ucs2 0x012c),(_ucs2 0x012d),(_ucs2 0x012e),(_ucs2 0x012f);

insert into t1 values (_ucs2 0x0130),(_ucs2 0x0131),(_ucs2 0x0132),(_ucs2 0x0133);

insert into t1 values (_ucs2 0x0134),(_ucs2 0x0135),(_ucs2 0x0136),(_ucs2 0x0137);

insert into t1 values (_ucs2 0x0138),(_ucs2 0x0139),(_ucs2 0x013a),(_ucs2 0x013b);

insert into t1 values (_ucs2 0x013c),(_ucs2 0x013d),(_ucs2 0x013e),(_ucs2 0x013f);

insert into t1 values (_ucs2 0x0140),(_ucs2 0x0141),(_ucs2 0x0142),(_ucs2 0x0143);

insert into t1 values (_ucs2 0x0144),(_ucs2 0x0145),(_ucs2 0x0146),(_ucs2 0x0147);

insert into t1 values (_ucs2 0x0148),(_ucs2 0x0149),(_ucs2 0x014a),(_ucs2 0x014b);

insert into t1 values (_ucs2 0x014c),(_ucs2 0x014d),(_ucs2 0x014e),(_ucs2 0x014f);

insert into t1 values (_ucs2 0x0150),(_ucs2 0x0151),(_ucs2 0x0152),(_ucs2 0x0153);

insert into t1 values (_ucs2 0x0154),(_ucs2 0x0155),(_ucs2 0x0156),(_ucs2 0x0157);

insert into t1 values (_ucs2 0x0158),(_ucs2 0x0159),(_ucs2 0x015a),(_ucs2 0x015b);

insert into t1 values (_ucs2 0x015c),(_ucs2 0x015d),(_ucs2 0x015e),(_ucs2 0x015f);

insert into t1 values (_ucs2 0x0160),(_ucs2 0x0161),(_ucs2 0x0162),(_ucs2 0x0163);

insert into t1 values (_ucs2 0x0164),(_ucs2 0x0165),(_ucs2 0x0166),(_ucs2 0x0167);

insert into t1 values (_ucs2 0x0168),(_ucs2 0x0169),(_ucs2 0x016a),(_ucs2 0x016b);

insert into t1 values (_ucs2 0x016c),(_ucs2 0x016d),(_ucs2 0x016e),(_ucs2 0x016f);

insert into t1 values (_ucs2 0x0170),(_ucs2 0x0171),(_ucs2 0x0172),(_ucs2 0x0173);

insert into t1 values (_ucs2 0x0174),(_ucs2 0x0175),(_ucs2 0x0176),(_ucs2 0x0177);

insert into t1 values (_ucs2 0x0178),(_ucs2 0x0179),(_ucs2 0x017a),(_ucs2 0x017b);

insert into t1 values (_ucs2 0x017c),(_ucs2 0x017d),(_ucs2 0x017e),(_ucs2 0x017f);



#

# Latin extended-B, 0180-024F

#

insert into t1 values (_ucs2 0x0180),(_ucs2 0x0181),(_ucs2 0x0182),(_ucs2 0x0183);

insert into t1 values (_ucs2 0x0184),(_ucs2 0x0185),(_ucs2 0x0186),(_ucs2 0x0187);

insert into t1 values (_ucs2 0x0188),(_ucs2 0x0189),(_ucs2 0x018a),(_ucs2 0x018b);

insert into t1 values (_ucs2 0x018c),(_ucs2 0x018d),(_ucs2 0x018e),(_ucs2 0x018f);

insert into t1 values (_ucs2 0x0190),(_ucs2 0x0191),(_ucs2 0x0192),(_ucs2 0x0193);

insert into t1 values (_ucs2 0x0194),(_ucs2 0x0195),(_ucs2 0x0196),(_ucs2 0x0197);

insert into t1 values (_ucs2 0x0198),(_ucs2 0x0199),(_ucs2 0x019a),(_ucs2 0x019b);

insert into t1 values (_ucs2 0x019c),(_ucs2 0x019d),(_ucs2 0x019e),(_ucs2 0x019f);

insert into t1 values (_ucs2 0x01a0),(_ucs2 0x01a1),(_ucs2 0x01a2),(_ucs2 0x01a3);

insert into t1 values (_ucs2 0x01a4),(_ucs2 0x01a5),(_ucs2 0x01a6),(_ucs2 0x01a7);

insert into t1 values (_ucs2 0x01a8),(_ucs2 0x01a9),(_ucs2 0x01aa),(_ucs2 0x01ab);

insert into t1 values (_ucs2 0x01ac),(_ucs2 0x01ad),(_ucs2 0x01ae),(_ucs2 0x01af);

insert into t1 values (_ucs2 0x01b0),(_ucs2 0x01b1),(_ucs2 0x01b2),(_ucs2 0x01b3);

insert into t1 values (_ucs2 0x01b4),(_ucs2 0x01b5),(_ucs2 0x01b6),(_ucs2 0x01b7);

insert into t1 values (_ucs2 0x01b8),(_ucs2 0x01b9),(_ucs2 0x01ba),(_ucs2 0x01bb);

insert into t1 values (_ucs2 0x01bc),(_ucs2 0x01bd),(_ucs2 0x01be),(_ucs2 0x01bf);

insert into t1 values (_ucs2 0x01c0),(_ucs2 0x01c1),(_ucs2 0x01c2),(_ucs2 0x01c3);

insert into t1 values (_ucs2 0x01c4),(_ucs2 0x01c5),(_ucs2 0x01c6),(_ucs2 0x01c7);

insert into t1 values (_ucs2 0x01c8),(_ucs2 0x01c9),(_ucs2 0x01ca),(_ucs2 0x01cb);

insert into t1 values (_ucs2 0x01cc),(_ucs2 0x01cd),(_ucs2 0x01ce),(_ucs2 0x01cf);

insert into t1 values (_ucs2 0x01d0),(_ucs2 0x01d1),(_ucs2 0x01d2),(_ucs2 0x01d3);

insert into t1 values (_ucs2 0x01d4),(_ucs2 0x01d5),(_ucs2 0x01d6),(_ucs2 0x01d7);

insert into t1 values (_ucs2 0x01d8),(_ucs2 0x01d9),(_ucs2 0x01da),(_ucs2 0x01db);

insert into t1 values (_ucs2 0x01dc),(_ucs2 0x01dd),(_ucs2 0x01de),(_ucs2 0x01df);

insert into t1 values (_ucs2 0x01e0),(_ucs2 0x01e1),(_ucs2 0x01e2),(_ucs2 0x01e3);

insert into t1 values (_ucs2 0x01e4),(_ucs2 0x01e5),(_ucs2 0x01e6),(_ucs2 0x01e7);

insert into t1 values (_ucs2 0x01e8),(_ucs2 0x01e9),(_ucs2 0x01ea),(_ucs2 0x01eb);

insert into t1 values (_ucs2 0x01ec),(_ucs2 0x01ed),(_ucs2 0x01ee),(_ucs2 0x01ef);

insert into t1 values (_ucs2 0x01f0),(_ucs2 0x01f1),(_ucs2 0x01f2),(_ucs2 0x01f3);

insert into t1 values (_ucs2 0x01f4),(_ucs2 0x01f5),(_ucs2 0x01f6),(_ucs2 0x01f7);

insert into t1 values (_ucs2 0x01f8),(_ucs2 0x01f9),(_ucs2 0x01fa),(_ucs2 0x01fb);

insert into t1 values (_ucs2 0x01fc),(_ucs2 0x01fd),(_ucs2 0x01fe),(_ucs2 0x01ff);





insert into t1 values ('AA'),('Aa'),('aa'),('aA');

insert into t1 values ('CH'),('Ch'),('ch'),('cH');

insert into t1 values ('DZ'),('Dz'),('dz'),('dZ');

insert into t1 values ('IJ'),('Ij'),('ij'),('iJ');

insert into t1 values ('LJ'),('Lj'),('lj'),('lJ');

insert into t1 values ('LL'),('Ll'),('ll'),('lL');

insert into t1 values ('NJ'),('Nj'),('nj'),('nJ');

insert into t1 values ('OE'),('Oe'),('oe'),('oE');

insert into t1 values ('SS'),('Ss'),('ss'),('sS');

insert into t1 values ('RR'),('Rr'),('rr'),('rR');



select group_concat(c1 order by c1) from t1 group by c1 collate utf8_unicode_ci;

select group_concat(c1 order by c1) from t1 group by c1 collate utf8_icelandic_ci;

select group_concat(c1 order by c1) from t1 group by c1 collate utf8_latvian_ci;

select group_concat(c1 order by c1) from t1 group by c1 collate utf8_romanian_ci;

select group_concat(c1 order by c1) from t1 group by c1 collate utf8_slovenian_ci;

select group_concat(c1 order by c1) from t1 group by c1 collate utf8_polish_ci;

select group_concat(c1 order by c1) from t1 group by c1 collate utf8_estonian_ci;

select group_concat(c1 order by c1) from t1 group by c1 collate utf8_spanish_ci;

select group_concat(c1 order by c1) from t1 group by c1 collate utf8_swedish_ci;

select group_concat(c1 order by c1) from t1 group by c1 collate utf8_turkish_ci;



select group_concat(c1 order by c1) from t1 group by c1 collate utf8_czech_ci;

select group_concat(c1 order by c1) from t1 group by c1 collate utf8_danish_ci;

select group_concat(c1 order by c1) from t1 group by c1 collate utf8_lithuanian_ci;

--select group_concat(c1 order by c1) from t1 group by c1 collate utf8_slovak_ci;

select group_concat(c1 order by c1) from t1 group by c1 collate utf8_spanish2_ci;

strings/ctype-uca.c

+7 −6
Original line number Diff line number Diff line
@@ -6586,7 +6586,7 @@ static const char polish[]= 
    "& N < \\u0144 <<< \\u0143 "

    "& O < \\u00F3 <<< \\u00D3 "

    "& S < \\u015B <<< \\u015A "

    "& Z < \\u017A <<< \\u017B ";

    "& Z < \\u017A <<< \\u0179 < \\u017C <<< \\u017B";



static const char estonian[]=

    "& S < \\u0161 <<< \\u0160 "
@@ -6889,9 +6889,9 @@ static int my_uca_scanner_next_any(my_uca_scanner *scanner) 
                                            scanner->sbeg, 

                                            scanner->send)) >=0) &&

           (!(page1= (wc >> 8))) &&

           ((code1= wc & 0xFF) > 0x40) &&

           ((code1= (wc & 0xFF)) > 0x40) &&

           (code1 < 0x80) && 

           (cweight= scanner->contractions[(scanner->code-0x40)*0x40+scanner->sbeg[1]-0x40]))

           (cweight= scanner->contractions[(scanner->code-0x40)*0x40 + code1-0x40]))

      {

        scanner->implicit[0]= 0;

        scanner->wbeg= scanner->implicit;
@@ -7137,6 +7137,7 @@ static int my_strnxfrm_uca(CHARSET_INFO *cs, 
    dst[1]= s_res & 0xFF;

    dst+= 2;

  }

  for ( ; dst < de; *dst++='\0');

  return dst - dst_orig;

}
@@ -7919,7 +7920,7 @@ CHARSET_INFO my_charset_ucs2_general_uca= 
    "ucs2",		/* cs name    */

    "ucs2_unicode_ci",	/* name         */

    "",			/* comment      */

    NULL,		/* tailoring    */

    "",			/* tailoring    */

    NULL,		/* ctype        */

    NULL,		/* to_lower     */

    NULL,		/* to_upper     */
@@ -8271,7 +8272,7 @@ CHARSET_INFO my_charset_ucs2_slovak_uca_ci= 
    "ucs2",		/* cs name    */

    "ucs2_slovak_ci",	/* name         */

    "",			/* comment      */

    lithuanian,		/* tailoring    */

    slovak,		/* tailoring    */

    NULL,		/* ctype        */

    NULL,		/* to_lower     */

    NULL,		/* to_upper     */
@@ -8723,7 +8724,7 @@ CHARSET_INFO my_charset_utf8_slovak_uca_ci= 
    "utf8",		/* cs name    */

    "utf8_slovak_ci",	/* name         */

    "",			/* comment      */

    lithuanian,		/* tailoring    */

    slovak,		/* tailoring    */

    ctype_utf8,		/* ctype        */

    NULL,		/* to_lower     */

    NULL,		/* to_upper     */