Bug#16233: XML: ExtractValue() fails with special characters

ExtractValue didn't understand tag and attribute names
consisting of "tricky" national letters (e.g. latin accenter letters).
It happened because XPath lex parser recognized only basic
latin letter a..z ad a part of an identifier.

Fixed to recognize all letters by means of new "full ctype" which
was added recently.

mysql-test/r/xml.result

@@ -615,3 +615,26 @@ select extractValue('<e>1</e>','last()');
 ERROR HY000: XPATH syntax error: ''
 select extractValue('<e><a>1</a></e>','/e/');
 ERROR HY000: XPATH syntax error: ''
+set names utf8;
+select extractValue('<Ñ><r>r</r></Ñ>','/Ñ/r');
+extractValue('<Ñ><r>r</r></Ñ>','/Ñ/r')
+r
+select extractValue('<r><Ñ>Ñ</Ñ></r>','/r/Ñ');
+extractValue('<r><Ñ>Ñ</Ñ></r>','/r/Ñ')
+Ñ
+select extractValue('<Ñ r="r"/>','/Ñ/@r');
+extractValue('<Ñ r="r"/>','/Ñ/@r')
+r
+select extractValue('<r Ñ="Ñ"/>','/r/@Ñ');
+extractValue('<r Ñ="Ñ"/>','/r/@Ñ')
+Ñ
+DROP PROCEDURE IF EXISTS p2;
+CREATE PROCEDURE p2 ()
+BEGIN
+DECLARE p LONGTEXT CHARACTER SET UTF8 DEFAULT '<Ñ><r>A</r></Ñ>';
+SELECT EXTRACTVALUE(p,'/Ñ/r');
+END//
+CALL p2();
+EXTRACTVALUE(p,'/Ñ/r')
+A
+DROP PROCEDURE p2;

mysql-test/t/xml.test

@@ -295,3 +295,23 @@ select extractValue('<e>1</e>','last()');
 --error 1105
 select extractValue('<e><a>1</a></e>','/e/');
 
+#
+# Bug#16233: XML: ExtractValue() fails with special characters
+#
+set names utf8;
+select extractValue('<Ñ><r>r</r></Ñ>','/Ñ/r');
+select extractValue('<r><Ñ>Ñ</Ñ></r>','/r/Ñ');
+select extractValue('<Ñ r="r"/>','/Ñ/@r');
+select extractValue('<r Ñ="Ñ"/>','/r/@Ñ');
+--disable_warnings
+DROP PROCEDURE IF EXISTS p2;
+--enable_warnings
+DELIMITER //;
+CREATE PROCEDURE p2 ()
+BEGIN
+ DECLARE p LONGTEXT CHARACTER SET UTF8 DEFAULT '<Ñ><r>A</r></Ñ>';
+ SELECT EXTRACTVALUE(p,'/Ñ/r');
+END//
+DELIMITER ;//
+CALL p2();
+DROP PROCEDURE p2;

sql/item_xmlfunc.cc

@@ -1304,30 +1304,6 @@ my_xpath_init(MY_XPATH *xpath)
 }
 
 
-/*
-  Some ctype-alike helper functions. Note, we cannot
-  reuse cs->ident_map[], because in Xpath, unlike in SQL, 
-  dash character is a valid identifier part.
-*/
-static int
-my_xident_beg(int c)
-{
-  return (((c) >= 'a' && (c) <= 'z') ||
-          ((c) >= 'A' && (c) <= 'Z') ||
-          ((c) == '_'));
-}
-
-
-static int
-my_xident_body(int c)
-{
-  return (((c) >= 'a' && (c) <= 'z') ||
-          ((c) >= 'A' && (c) <= 'Z') ||
-          ((c) >= '0' && (c) <= '9') ||
-          ((c)=='-') || ((c) == '_'));
-}
-
-
 static int
 my_xdigit(int c)
 {
@@ -1350,7 +1326,7 @@ static void
 my_xpath_lex_scan(MY_XPATH *xpath,
                   MY_XPATH_LEX *lex, const char *beg, const char *end)
 {
-  int ch;
+  int ch, ctype, length;
   for ( ; beg < end && *beg == ' ' ; beg++); // skip leading spaces
   lex->beg= beg;
   
@@ -1360,20 +1336,20 @@ my_xpath_lex_scan(MY_XPATH *xpath,
     lex->term= MY_XPATH_LEX_EOF; // end of line reached
     return;
   }
-  ch= *beg++;
 
-  if (ch > 0 && ch < 128 && simpletok[ch])
+  // Check ident, or a function call, or a keyword
+  if ((length= xpath->cs->cset->ctype(xpath->cs, &ctype,
+                                      (const uchar*) beg,
+                                      (const uchar*) end)) > 0 &&
+      ((ctype & (_MY_L | _MY_U)) || *beg == '_'))
   {
-    // a token consisting of one character found
-    lex->end= beg;
-    lex->term= ch;
-    return;
-  }
-  
-  if (my_xident_beg(ch)) // ident, or a function call, or a keyword
-  {
-    // scan until the end of the identifier
-    for ( ; beg < end && my_xident_body(*beg); beg++);
+    // scan untill the end of the idenfitier
+    for (beg+= length; 
+         (length= xpath->cs->cset->ctype(xpath->cs, &ctype,
+                                         (const uchar*) beg,
+                                         (const uchar*) end)) > 0 &&
+         ((ctype & (_MY_L | _MY_U | _MY_NMR)) || *beg == '_' || *beg == '-') ;
+         beg+= length) /* no op */;
     lex->end= beg;
 
     // check if a function call
@@ -1388,6 +1364,18 @@ my_xpath_lex_scan(MY_XPATH *xpath,
     return;
   }
 
+
+  ch= *beg++;
+  
+  if (ch > 0 && ch < 128 && simpletok[ch])
+  {
+    // a token consisting of one character found
+    lex->end= beg;
+    lex->term= ch;
+    return;
+  }
+
+
   if (my_xdigit(ch)) // a sequence of digits
   {
     for ( ; beg < end && my_xdigit(*beg) ; beg++);

strings/ctype-simple.c

@@ -1362,7 +1362,7 @@ int my_mb_ctype_8bit(CHARSET_INFO *cs, int *ctype,
     *ctype= 0;
     return MY_CS_TOOSMALL;
   }
-  *ctype= cs->ctype[*s];
+  *ctype= cs->ctype[*s + 1];
   return 1;
 }