diff options
Diffstat (limited to 'mysql-test/t/ctype_utf32_uca.test')
| -rw-r--r-- | mysql-test/t/ctype_utf32_uca.test | 291 |
1 files changed, 291 insertions, 0 deletions
diff --git a/mysql-test/t/ctype_utf32_uca.test b/mysql-test/t/ctype_utf32_uca.test new file mode 100644 index 00000000000..9386cc9e65e --- /dev/null +++ b/mysql-test/t/ctype_utf32_uca.test @@ -0,0 +1,291 @@ +-- source include/have_ucs2.inc +-- source include/have_utf32.inc + +--disable_warnings +DROP TABLE IF EXISTS t1; +--enable_warnings + +--echo # +--echo # Start of 5.5 tests +--echo # + +set names utf8; +set collation_connection=utf32_unicode_ci; +select hex('a'), hex('a '); +-- source include/endspace.inc + +# +# Bug #6787 LIKE not working properly with _ and utf8 data +# +select 'c' like '\_' as want0; + +# +# Bug #5679 utf8_unicode_ci LIKE--trailing % doesn't equal zero characters +# +CREATE TABLE t ( + c char(20) NOT NULL +) ENGINE=MyISAM DEFAULT CHARACTER SET utf32 COLLATE=utf32_unicode_ci; +INSERT INTO t VALUES ('a'),('ab'),('aba'); +ALTER TABLE t ADD INDEX (c); +SELECT c FROM t WHERE c LIKE 'a%'; +DROP TABLE t; + + +create table t1 (c1 char(10) character set utf32 collate utf32_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 binary c1 separator '') from t1 group by c1 collate utf32_unicode_ci; +select group_concat(c1 order by binary c1 separator '') from t1 group by c1 collate utf32_icelandic_ci; +select group_concat(c1 order by binary c1 separator '') from t1 group by c1 collate utf32_latvian_ci; +select group_concat(c1 order by binary c1 separator '') from t1 group by c1 collate utf32_romanian_ci; +select group_concat(c1 order by binary c1 separator '') from t1 group by c1 collate utf32_slovenian_ci; +select group_concat(c1 order by binary c1 separator '') from t1 group by c1 collate utf32_polish_ci; +select group_concat(c1 order by binary c1 separator '') from t1 group by c1 collate utf32_estonian_ci; +select group_concat(c1 order by binary c1 separator '') from t1 group by c1 collate utf32_spanish_ci; +select group_concat(c1 order by binary c1 separator '') from t1 group by c1 collate utf32_swedish_ci; +select group_concat(c1 order by binary c1 separator '') from t1 group by c1 collate utf32_turkish_ci; +select group_concat(c1 order by binary c1 separator '') from t1 group by c1 collate utf32_czech_ci; +select group_concat(c1 order by binary c1 separator '') from t1 group by c1 collate utf32_danish_ci; +select group_concat(c1 order by binary c1 separator '') from t1 group by c1 collate utf32_lithuanian_ci; +select group_concat(c1 order by binary c1 separator '') from t1 group by c1 collate utf32_slovak_ci; +select group_concat(c1 order by binary c1 separator '') from t1 group by c1 collate utf32_spanish2_ci; +select group_concat(c1 order by binary c1 separator '') from t1 group by c1 collate utf32_roman_ci; +select group_concat(c1 order by binary c1 separator '') from t1 group by c1 collate utf32_esperanto_ci; +select group_concat(c1 order by binary c1 separator '') from t1 group by c1 collate utf32_hungarian_ci; + +drop table t1; + +# +# Bug#5324 +# +SET NAMES utf8; +#test1 +CREATE TABLE t1 (c varchar(200) CHARACTER SET utf32 COLLATE utf32_general_ci NOT NULL, INDEX (c)); +INSERT INTO t1 VALUES (_ucs2 0x039C03C903B403B11F770308); +#Check one row +SELECT * FROM t1 WHERE c LIKE _utf32 0x0000039C00000025 COLLATE utf32_general_ci; +INSERT INTO t1 VALUES (CONVERT(_ucs2 0x039C03C903B4 USING utf8)); +#Check two rows +SELECT * FROM t1 WHERE c LIKE _utf32 0x0000039C00000025 +COLLATE utf32_general_ci ORDER BY c; +DROP TABLE t1; +#test2 +CREATE TABLE t1 (c varchar(200) CHARACTER SET utf32 COLLATE utf32_unicode_ci NOT NULL, INDEX (c)); +INSERT INTO t1 VALUES (_ucs2 0x039C03C903B403B11F770308); +#Check one row +SELECT * FROM t1 WHERE c LIKE _utf32 0x0000039C00000025 COLLATE utf32_unicode_ci; +INSERT INTO t1 VALUES (_ucs2 0x039C03C903B4); +#Check two rows +SELECT * FROM t1 WHERE c LIKE _utf32 0x0000039C00000025 +COLLATE utf32_unicode_ci ORDER BY c; +DROP TABLE t1; +#test 3 +CREATE TABLE t1 (c varchar(200) CHARACTER SET utf32 COLLATE utf32_unicode_ci NOT NULL, INDEX (c)); +INSERT INTO t1 VALUES (_ucs2 0x039C03C903B403B11F770308); +#Check one row row +SELECT * FROM t1 WHERE c LIKE CONVERT(_ucs2 0x039C0025 USING utf32) COLLATE utf32_unicode_ci; +INSERT INTO t1 VALUES (CONVERT(_ucs2 0x039C03C903B4 USING utf8)); +#Check two rows +SELECT * FROM t1 WHERE c LIKE CONVERT(_ucs2 0x039C0025 USING utf32) +COLLATE utf32_unicode_ci ORDER BY c; +DROP TABLE t1; + + +SET NAMES utf8; +SET @test_character_set='utf32'; +SET @test_collation='utf32_swedish_ci'; +-- source include/ctype_common.inc + + +SET collation_connection='utf32_unicode_ci'; +-- source include/ctype_filesort.inc +-- source include/ctype_like_escape.inc + +--echo End of 4.1 tests + +# +# Check UPPER/LOWER changing length +# +# Result shorter than argument +CREATE TABLE t1 (id int, a varchar(30) character set utf32); +INSERT INTO t1 VALUES (1, _ucs2 0x01310069), (2, _ucs2 0x01310131); +INSERT INTO t1 VALUES (3, _ucs2 0x00690069), (4, _ucs2 0x01300049); +INSERT INTO t1 VALUES (5, _ucs2 0x01300130), (6, _ucs2 0x00490049); +SELECT a, length(a) la, @l:=lower(a) l, length(@l) ll, @u:=upper(a) u, length(@u) lu +FROM t1 ORDER BY id; +ALTER TABLE t1 MODIFY a VARCHAR(30) character set utf32 collate utf32_turkish_ci; +SELECT a, length(a) la, @l:=lower(a) l, length(@l) ll, @u:=upper(a) u, length(@u) lu +FROM t1 ORDER BY id; +DROP TABLE t1; + +# +# Bug #27079 Crash while grouping empty ucs2 strings +# +CREATE TABLE t1 ( + c1 text character set utf32 collate utf32_polish_ci NOT NULL +) ENGINE=MyISAM; +insert into t1 values (''),('a'); +SELECT COUNT(*), c1 FROM t1 GROUP BY c1; +DROP TABLE IF EXISTS t1; + + +# +# Test basic regex functionality +# +set collation_connection=utf32_unicode_ci; +--source include/ctype_regex.inc + + +--echo # +--echo # End of 5.5 tests +--echo # |