Bug#18285: ExtractValue not returning character data within <![CDATA[]]> as expected

mysql-test/r/xml.result:
  Adding test case
mysql-test/t/xml.test:
  Adding test case
strings/xml.c:
  Adding missing support of CDATA sections.

3 files changed, 31 insertions, 0 deletions

diff --git a/mysql-test/r/xml.result b/mysql-test/r/xml.result
index 52f80000015..03fff98f1d4 100644
--- a/mysql-test/r/xml.result
+++ b/mysql-test/r/xml.result
@@ -609,6 +609,9 @@ extractvalue('<a>Jack</a>','/a[contains(../a,"j")]' collate latin1_bin)
 select extractvalue('<a>Jack</a>' collate latin1_bin,'/a[contains(../a,"j")]');
 extractvalue('<a>Jack</a>' collate latin1_bin,'/a[contains(../a,"j")]')
 
+select ExtractValue('<tag1><![CDATA[test]]></tag1>','/tag1');
+ExtractValue('<tag1><![CDATA[test]]></tag1>','/tag1')
+test
 select extractValue('<e>1</e>','position()');
 ERROR HY000: XPATH syntax error: ''
 select extractValue('<e>1</e>','last()');
diff --git a/mysql-test/t/xml.test b/mysql-test/t/xml.test
index af3ec2d827e..0dda1922b8d 100644
--- a/mysql-test/t/xml.test
+++ b/mysql-test/t/xml.test
@@ -279,6 +279,12 @@ select extractvalue('<a>Jack</a>','/a[contains(../a,"j")]' collate latin1_bin);
 select extractvalue('<a>Jack</a>' collate latin1_bin,'/a[contains(../a,"j")]');
 
 #
+# Bug#18285: ExtractValue not returning character
+# data within <![CDATA[]]> as expected
+#
+select ExtractValue('<tag1><![CDATA[test]]></tag1>','/tag1');
+
+#
 # Bug #18171 XML: ExtractValue: the XPath position()
 # function crashes the server!
 #
diff --git a/strings/xml.c b/strings/xml.c
index 705da592257..4c09e85604d 100644
--- a/strings/xml.c
+++ b/strings/xml.c
@@ -30,6 +30,7 @@
 #define MY_XML_TEXT	'T'
 #define MY_XML_QUESTION	'?'
 #define MY_XML_EXCLAM   '!'
+#define MY_XML_CDATA    'D'
 
 typedef struct xml_attr_st
 {
@@ -45,6 +46,7 @@ static const char *lex2str(int lex)
     case MY_XML_EOF:      return "EOF";
     case MY_XML_STRING:   return "STRING";
     case MY_XML_IDENT:    return "IDENT";
+    case MY_XML_CDATA:    return "CDATA";
     case MY_XML_EQ:       return "'='";
     case MY_XML_LT:       return "'<'";
     case MY_XML_GT:       return "'>'";
@@ -90,6 +92,20 @@ static int my_xml_scan(MY_XML_PARSER *p,MY_XML_ATTR *a)
     a->end=p->cur;
     lex=MY_XML_COMMENT;
   }
+  else if (!bcmp(p->cur, "<![CDATA[",9))
+  {
+    p->cur+= 9;
+    for (; p->cur < p->end - 2 ; p->cur++)
+    {
+      if (p->cur[0] == ']' && p->cur[1] == ']' && p->cur[2] == '>')
+      {
+        p->cur+= 3;
+        a->end= p->cur;
+        break;
+      }
+    }
+    lex= MY_XML_CDATA;
+  }
   else if (strchr("?=/<>!",p->cur[0]))
   {
     p->cur++;
@@ -215,7 +231,13 @@ int my_xml_parse(MY_XML_PARSER *p,const char *str, uint len)
       lex=my_xml_scan(p,&a);
       
       if (MY_XML_COMMENT == lex)
+        continue;
+      
+      if (lex == MY_XML_CDATA)
       {
+        a.beg+= 9;
+        a.end-= 3;
+        my_xml_value(p, a.beg, (uint) (a.end-a.beg));
         continue;
       }