From b62123e0d517a63fb7a1192093fd3cafcfe9d480 Mon Sep 17 00:00:00 2001
From: Alexander Barkov <bar@mariadb.com>
Date: Thu, 23 Feb 2023 22:19:26 +0400
Subject: MDEV-30716 Wrong casefolding in xxx_unicode_520_ci for U+0700..U+07FF

The array my_unicase_pages_unicode520[7] erroneously mapped to plane06
instead of plane07.
---
 mysql-test/include/ctype_unicode_casefold_bmp.inc  | 23 +++++++++++++++++++
 .../ctype_unicode_casefold_supplementary.inc       | 22 ++++++++++++++++++
 mysql-test/include/ctype_unicode_ws_bmp.inc        | 26 ++++++++++++++++++++++
 .../include/ctype_unicode_ws_supplementary.inc     | 19 ++++++++++++++++
 4 files changed, 90 insertions(+)
 create mode 100644 mysql-test/include/ctype_unicode_casefold_bmp.inc
 create mode 100644 mysql-test/include/ctype_unicode_casefold_supplementary.inc
 create mode 100644 mysql-test/include/ctype_unicode_ws_bmp.inc
 create mode 100644 mysql-test/include/ctype_unicode_ws_supplementary.inc

(limited to 'mysql-test/include')

diff --git a/mysql-test/include/ctype_unicode_casefold_bmp.inc b/mysql-test/include/ctype_unicode_casefold_bmp.inc
new file mode 100644
index 00000000000..e5b16b9334e
--- /dev/null
+++ b/mysql-test/include/ctype_unicode_casefold_bmp.inc
@@ -0,0 +1,23 @@
+--source include/have_utf32.inc
+--source include/have_ucs2.inc
+--source include/have_sequence.inc
+
+EXECUTE IMMEDIATE SFORMAT('
+CREATE VIEW v_bmp AS
+SELECT
+  seq AS codepoint,
+  LPAD(HEX(seq),4,''0'') AS codepoint_hex4,
+  CONVERT(CHAR(seq USING utf32) USING {}) COLLATE {} AS c
+FROM
+  seq_0_to_65535', @@character_set_connection, @@collation_connection);
+
+SELECT COLLATION(c) FROM v_bmp LIMIT 1;
+
+SELECT
+  codepoint_hex4,
+  HEX(CAST(LOWER(c) AS CHAR CHARACTER SET ucs2)),
+  HEX(CAST(UPPER(c) AS CHAR CHARACTER SET ucs2))
+FROM v_bmp
+WHERE BINARY(c)<>BINARY(LOWER(c)) OR BINARY(c)<>BINARY(UPPER(c));
+
+DROP VIEW v_bmp;
diff --git a/mysql-test/include/ctype_unicode_casefold_supplementary.inc b/mysql-test/include/ctype_unicode_casefold_supplementary.inc
new file mode 100644
index 00000000000..bb603118db6
--- /dev/null
+++ b/mysql-test/include/ctype_unicode_casefold_supplementary.inc
@@ -0,0 +1,22 @@
+--source include/have_utf32.inc
+--source include/have_sequence.inc
+
+EXECUTE IMMEDIATE SFORMAT('
+CREATE VIEW v_supplementary AS
+SELECT
+  seq AS codepoint,
+  LPAD(HEX(seq),8,''0'') AS codepoint_hex8,
+  CONVERT(CHAR(seq USING utf32) USING {}) COLLATE {} AS c
+FROM
+  seq_65536_to_1114111', @@character_set_connection, @@collation_connection);
+
+SELECT COLLATION(c) FROM v_supplementary LIMIT 1;
+
+SELECT
+  codepoint_hex8,
+  HEX(CAST(LOWER(c) AS CHAR CHARACTER SET utf32)),
+  HEX(CAST(UPPER(c) AS CHAR CHARACTER SET utf32))
+FROM v_supplementary
+WHERE BINARY(c)<>BINARY(LOWER(c)) OR BINARY(c)<>BINARY(UPPER(c));
+
+DROP VIEW v_supplementary;
diff --git a/mysql-test/include/ctype_unicode_ws_bmp.inc b/mysql-test/include/ctype_unicode_ws_bmp.inc
new file mode 100644
index 00000000000..31166b5d31a
--- /dev/null
+++ b/mysql-test/include/ctype_unicode_ws_bmp.inc
@@ -0,0 +1,26 @@
+--source include/have_utf32.inc
+--source include/have_sequence.inc
+
+EXECUTE IMMEDIATE SFORMAT('
+CREATE VIEW v_bmp AS
+SELECT
+  seq AS codepoint,
+  LPAD(HEX(seq),4,''0'') AS codepoint_hex4,
+  CONVERT(CHAR(seq USING utf32) USING {}) COLLATE {} AS c
+FROM
+  seq_0_to_65535', @@character_set_connection, @@collation_connection);
+
+SELECT COLLATION(c) FROM v_bmp LIMIT 1;
+
+SELECT HEX(codepoint) FROM v_bmp WHERE HEX(WEIGHT_STRING(c))='FFFD';
+
+SELECT
+  SUM(codepoint_hex4=HEX(WEIGHT_STRING(c))) AS count_bmp_weight_is_codepoint,
+  SUM(codepoint_hex4<>HEX(WEIGHT_STRING(c))) AS count_bmp_weight_is_not_codepoint
+FROM v_bmp;
+
+SELECT codepoint_hex4,HEX(WEIGHT_STRING(c))
+FROM v_bmp
+WHERE codepoint_hex4<>HEX(WEIGHT_STRING(c));
+
+DROP VIEW v_bmp;
diff --git a/mysql-test/include/ctype_unicode_ws_supplementary.inc b/mysql-test/include/ctype_unicode_ws_supplementary.inc
new file mode 100644
index 00000000000..82273e8fb6a
--- /dev/null
+++ b/mysql-test/include/ctype_unicode_ws_supplementary.inc
@@ -0,0 +1,19 @@
+--source include/have_utf32.inc
+--source include/have_sequence.inc
+
+EXECUTE IMMEDIATE SFORMAT('
+CREATE VIEW v_supplementary AS
+SELECT
+  seq AS codepoint,
+  CONVERT(CHAR(seq USING utf32) USING {}) COLLATE {} AS c
+FROM
+  seq_65536_to_1114111', @@character_set_connection, @@collation_connection);
+
+SELECT COLLATION(c) FROM v_supplementary LIMIT 1;
+
+SELECT
+  SUM(HEX(WEIGHT_STRING(c))<>'FFFD'),
+  SUM(HEX(WEIGHT_STRING(c))='FFFD')
+FROM v_supplementary;
+
+DROP VIEW v_supplementary;
-- 
cgit v1.2.1