Fixed problem with sort order of some russian letters in the mchar module. PGPRO-1509

Author: vbwagner

contrib/mchar/expected/mm.out

@@ -803,3 +803,44 @@ SELECT * FROM a, c WHERE mvarchar255 = mchar2;
 
 DROP TABLE a;
 DROP TABLE c;
+select * from (values ('е'::mchar),('ё'),('еа'),('еб'),('ее'),('еж'),('ёа'),('ёб'),('ёё'),('ёж'),('ёе'),('её')) z order by 1;
+ column1 
+---------
+ е
+ еа
+ еб
+ ее
+ её
+ еж
+ ё
+ ёа
+ ёб
+ ёе
+ ёё
+ ёж
+(12 rows)
+
+select mvarchar_icase_cmp('ё', 'е');
+ mvarchar_icase_cmp 
+--------------------
+                  1
+(1 row)
+
+select mvarchar_icase_cmp('Ё', 'Е');
+ mvarchar_icase_cmp 
+--------------------
+                  1
+(1 row)
+
+select mvarchar_icase_cmp('й', 'и');
+ mvarchar_icase_cmp 
+--------------------
+                  1
+(1 row)
+
+select mvarchar_icase_cmp('Й', 'И');
+ mvarchar_icase_cmp 
+--------------------
+                  1
+(1 row)
+

contrib/mchar/mchar_recode.c

@@ -31,6 +31,7 @@ createUObjs() {
 			elog(ERROR,"ICU ucol_open returns %d (%s)", err,  u_errorName(err));
 		}
 
+		/* UCOL_PRIMARY doesn't distinguish И & Й, Е & Ё */
 		ucol_setStrength( colCaseInsensitive, UCOL_SECONDARY );
 	}
 
@@ -114,29 +115,49 @@ FillWhiteSpace( UChar *dst, int n ) {
 int 
 UCharCaseCompare(UChar * a, int alen, UChar *b, int blen) {
 	int len = Min(alen, blen);
-	int res;
+	int i, res;
 
 	createUObjs();
 
-	res = (int)ucol_strcoll( colCaseInsensitive,
-							  a, len,
-							  b, len);
-	if ( res == 0 && alen != blen )
+	/*
+	 * Preventing any influence of following characters to
+	 * current one, try
+	 * select * from (values ('е'::mchar),('ё'),('еа'),('еб'),('ее'),('еж'),
+	 *						 ('ёа'),('ёб'),('ёё'),('ёж'),('ёе'),('её')) z
+	 * order by 1;
+	 */
+	for (i=0; i<len; i++)
+	{
+		res = (int)ucol_strcoll(colCaseInsensitive,
+								a + i, 1,
+								b + i, 1);
+		if (res)
+			return res;
+	}
+
+	if (alen != blen)
 		return (alen > blen) ? 1 : - 1;
-	return res;
+	return 0;
 }
 
 int 
 UCharCompare(UChar * a, int alen, UChar *b, int blen) {
 	int len = Min(alen, blen);
-	int res;
+	int i, res;
 
 	createUObjs();
 
-	res =  (int)ucol_strcoll( colCaseSensitive,
-							  a, len,
-							  b, len);
-	if ( res == 0 && alen != blen )
+	/* see above */
+	for (i=0; i<len; i++)
+	{
+		res = (int)ucol_strcoll(colCaseSensitive,
+								a + i, 1,
+								b + i, 1);
+		if (res)
+			return res;
+	}
+
+	if (alen != blen)
 		return (alen > blen) ? 1 : - 1;
-	return res;
+	return 0;
 }

contrib/mchar/sql/mm.sql

@@ -183,3 +183,8 @@ SELECT * FROM a, c WHERE mvarchar255 = mchar2;
 DROP TABLE a;
 DROP TABLE c;
 
+select * from (values ('е'::mchar),('ё'),('еа'),('еб'),('ее'),('еж'),('ёа'),('ёб'),('ёё'),('ёж'),('ёе'),('её')) z order by 1;
+select mvarchar_icase_cmp('ё', 'е');
+select mvarchar_icase_cmp('Ё', 'Е');
+select mvarchar_icase_cmp('й', 'и');
+select mvarchar_icase_cmp('Й', 'И');