Bug#24747 XPath error with the node name "Text"

mysql-test/r/xml.result:
  Adding test case
mysql-test/t/xml.test:
  Adding test case
sql/item_xmlfunc.cc:
  Bug#24747 XPath error with the node name "Text"
  
  Problem: keywords, nodetype names and axis names
  didn't work in node name context.
  
  Fix:
  1. Changes in lexem scanner behaviour when an identifier has been read:
  - search through axis names and return MY_XPATH_AXIS_xxx only
    when identifier is followed by two semicolons. Don't check axis
    names and return MY_XPATH_IDENT of not followed by two semicolons.
  - search through nodetype names and return MY_XPATH_LEX_NODETYPE
    only when identifier is followed by left parenthesis.
    Don't check nodetype names and return MY_XPATH_LEX_IDENT when
    not followed by parenthesis
  2. Change in syntax analizer: QName scanner now accepts keywords 
     AND, OR, MOD, DIV as valid node names.

mysql-test/r/xml.result

@@ -809,3 +809,78 @@ ExtractValue(@xml, "/entry[(50<pt)]/id")
 select ExtractValue(@xml, "/entry[(50<=pt)]/id");
 ExtractValue(@xml, "/entry[(50<=pt)]/id")
 pt50
+select ExtractValue('<a><b><Text>test</Text></b></a>','/a/b/Text');
+ExtractValue('<a><b><Text>test</Text></b></a>','/a/b/Text')
+test
+select ExtractValue('<a><b><comment>test</comment></b></a>','/a/b/comment');
+ExtractValue('<a><b><comment>test</comment></b></a>','/a/b/comment')
+test
+select ExtractValue('<a><b><node>test</node></b></a>','/a/b/node');
+ExtractValue('<a><b><node>test</node></b></a>','/a/b/node')
+test
+select ExtractValue('<a><b><processing-instruction>test</processing-instruction></b></a>','/a/b/processing-instruction');
+ExtractValue('<a><b><processing-instruction>test</processing-instruction></b></a>','/a/b/processing-instruction')
+test
+select ExtractValue('<a><and>test</and></a>', '/a/and');
+ExtractValue('<a><and>test</and></a>', '/a/and')
+test
+select ExtractValue('<a><or>test</or></a>', '/a/or');
+ExtractValue('<a><or>test</or></a>', '/a/or')
+test
+select ExtractValue('<a><mod>test</mod></a>', '/a/mod');
+ExtractValue('<a><mod>test</mod></a>', '/a/mod')
+test
+select ExtractValue('<a><div>test</div></a>', '/a/div');
+ExtractValue('<a><div>test</div></a>', '/a/div')
+test
+select ExtractValue('<a><and:and>test</and:and></a>', '/a/and:and');
+ExtractValue('<a><and:and>test</and:and></a>', '/a/and:and')
+test
+select ExtractValue('<a><or:or>test</or:or></a>', '/a/or:or');
+ExtractValue('<a><or:or>test</or:or></a>', '/a/or:or')
+test
+select ExtractValue('<a><mod:mod>test</mod:mod></a>', '/a/mod:mod');
+ExtractValue('<a><mod:mod>test</mod:mod></a>', '/a/mod:mod')
+test
+select ExtractValue('<a><div:div>test</div:div></a>', '/a/div:div');
+ExtractValue('<a><div:div>test</div:div></a>', '/a/div:div')
+test
+select ExtractValue('<a><ancestor>test</ancestor></a>', '/a/ancestor');
+ExtractValue('<a><ancestor>test</ancestor></a>', '/a/ancestor')
+test
+select ExtractValue('<a><ancestor-or-self>test</ancestor-or-self></a>', '/a/ancestor-or-self');
+ExtractValue('<a><ancestor-or-self>test</ancestor-or-self></a>', '/a/ancestor-or-self')
+test
+select ExtractValue('<a><attribute>test</attribute></a>', '/a/attribute');
+ExtractValue('<a><attribute>test</attribute></a>', '/a/attribute')
+test
+select ExtractValue('<a><child>test</child></a>', '/a/child');
+ExtractValue('<a><child>test</child></a>', '/a/child')
+test
+select ExtractValue('<a><descendant>test</descendant></a>', '/a/descendant');
+ExtractValue('<a><descendant>test</descendant></a>', '/a/descendant')
+test
+select ExtractValue('<a><descendant-or-self>test</descendant-or-self></a>', '/a/descendant-or-self');
+ExtractValue('<a><descendant-or-self>test</descendant-or-self></a>', '/a/descendant-or-self')
+test
+select ExtractValue('<a><following>test</following></a>', '/a/following');
+ExtractValue('<a><following>test</following></a>', '/a/following')
+test
+select ExtractValue('<a><following-sibling>test</following-sibling></a>', '/a/following-sibling');
+ExtractValue('<a><following-sibling>test</following-sibling></a>', '/a/following-sibling')
+test
+select ExtractValue('<a><namespace>test</namespace></a>', '/a/namespace');
+ExtractValue('<a><namespace>test</namespace></a>', '/a/namespace')
+test
+select ExtractValue('<a><parent>test</parent></a>', '/a/parent');
+ExtractValue('<a><parent>test</parent></a>', '/a/parent')
+test
+select ExtractValue('<a><preceding>test</preceding></a>', '/a/preceding');
+ExtractValue('<a><preceding>test</preceding></a>', '/a/preceding')
+test
+select ExtractValue('<a><preceding-sibling>test</preceding-sibling></a>', '/a/preceding-sibling');
+ExtractValue('<a><preceding-sibling>test</preceding-sibling></a>', '/a/preceding-sibling')
+test
+select ExtractValue('<a><self>test</self></a>', '/a/self');
+ExtractValue('<a><self>test</self></a>', '/a/self')
+test

mysql-test/t/xml.test

@@ -406,3 +406,41 @@ select ExtractValue(@xml, "/entry[(50>pt)]/id");
 select ExtractValue(@xml, "/entry[(50>=pt)]/id");
 select ExtractValue(@xml, "/entry[(50<pt)]/id");
 select ExtractValue(@xml, "/entry[(50<=pt)]/id");
+
+#
+# Bug#24747 XPath error with the node name "Text"
+#
+#
+# Test nodetypes in node name context
+#
+select ExtractValue('<a><b><Text>test</Text></b></a>','/a/b/Text');
+select ExtractValue('<a><b><comment>test</comment></b></a>','/a/b/comment');
+select ExtractValue('<a><b><node>test</node></b></a>','/a/b/node');
+select ExtractValue('<a><b><processing-instruction>test</processing-instruction></b></a>','/a/b/processing-instruction');
+#
+# Test keywords in node name contexts
+#
+select ExtractValue('<a><and>test</and></a>', '/a/and');
+select ExtractValue('<a><or>test</or></a>', '/a/or');
+select ExtractValue('<a><mod>test</mod></a>', '/a/mod');
+select ExtractValue('<a><div>test</div></a>', '/a/div');
+select ExtractValue('<a><and:and>test</and:and></a>', '/a/and:and');
+select ExtractValue('<a><or:or>test</or:or></a>', '/a/or:or');
+select ExtractValue('<a><mod:mod>test</mod:mod></a>', '/a/mod:mod');
+select ExtractValue('<a><div:div>test</div:div></a>', '/a/div:div');
+#
+# Test axis names in node name context
+#
+select ExtractValue('<a><ancestor>test</ancestor></a>', '/a/ancestor');
+select ExtractValue('<a><ancestor-or-self>test</ancestor-or-self></a>', '/a/ancestor-or-self');
+select ExtractValue('<a><attribute>test</attribute></a>', '/a/attribute');
+select ExtractValue('<a><child>test</child></a>', '/a/child');
+select ExtractValue('<a><descendant>test</descendant></a>', '/a/descendant');
+select ExtractValue('<a><descendant-or-self>test</descendant-or-self></a>', '/a/descendant-or-self');
+select ExtractValue('<a><following>test</following></a>', '/a/following');
+select ExtractValue('<a><following-sibling>test</following-sibling></a>', '/a/following-sibling');
+select ExtractValue('<a><namespace>test</namespace></a>', '/a/namespace');
+select ExtractValue('<a><parent>test</parent></a>', '/a/parent');
+select ExtractValue('<a><preceding>test</preceding></a>', '/a/preceding');
+select ExtractValue('<a><preceding-sibling>test</preceding-sibling></a>', '/a/preceding-sibling');
+select ExtractValue('<a><self>test</self></a>', '/a/self');

sql/item_xmlfunc.cc

@@ -1047,12 +1047,12 @@ static struct my_xpath_keyword_names_st my_keyword_names[] =
   {MY_XPATH_LEX_OR      , "or"                     ,  2, 0 },
   {MY_XPATH_LEX_DIV     , "div"                    ,  3, 0 },
   {MY_XPATH_LEX_MOD     , "mod"                    ,  3, 0 },
+  {0,NULL,0,0}
+};  
 
-  {MY_XPATH_LEX_NODETYPE, "comment"                ,  7, 0 },
-  {MY_XPATH_LEX_NODETYPE, "text"                   ,  4, 0 },
-  {MY_XPATH_LEX_NODETYPE, "processing-instruction" ,  22,0 },
-  {MY_XPATH_LEX_NODETYPE, "node"                   ,  4, 0 },
 
+static struct my_xpath_keyword_names_st my_axis_names[]=
+{
   {MY_XPATH_LEX_AXIS,"ancestor"          , 8,MY_XPATH_AXIS_ANCESTOR          },
   {MY_XPATH_LEX_AXIS,"ancestor-or-self"  ,16,MY_XPATH_AXIS_ANCESTOR_OR_SELF  },
   {MY_XPATH_LEX_AXIS,"attribute"         , 9,MY_XPATH_AXIS_ATTRIBUTE         },
@@ -1066,7 +1066,16 @@ static struct my_xpath_keyword_names_st my_keyword_names[] =
   {MY_XPATH_LEX_AXIS,"preceding"         , 9,MY_XPATH_AXIS_PRECEDING         },
   {MY_XPATH_LEX_AXIS,"preceding-sibling" ,17,MY_XPATH_AXIS_PRECEDING_SIBLING },
   {MY_XPATH_LEX_AXIS,"self"              , 4,MY_XPATH_AXIS_SELF              },
+  {0,NULL,0,0}
+};
 
+
+static struct my_xpath_keyword_names_st my_nodetype_names[]=
+{
+  {MY_XPATH_LEX_NODETYPE, "comment"                ,  7, 0 },
+  {MY_XPATH_LEX_NODETYPE, "text"                   ,  4, 0 },
+  {MY_XPATH_LEX_NODETYPE, "processing-instruction" ,  22,0 },
+  {MY_XPATH_LEX_NODETYPE, "node"                   ,  4, 0 },
   {0,NULL,0,0}
 };
 
@@ -1081,11 +1090,14 @@ static struct my_xpath_keyword_names_st my_keyword_names[] =
     - Token type, on lookup success.
     - MY_XPATH_LEX_IDENT, on lookup failure.
 */
-static int my_xpath_keyword(MY_XPATH *x, const char *beg, const char *end)
+static int
+my_xpath_keyword(MY_XPATH *x,
+                 struct my_xpath_keyword_names_st *keyword_names,
+                 const char *beg, const char *end)
 {
   struct my_xpath_keyword_names_st *k;
   size_t length= end-beg;
-  for (k= my_keyword_names; k->name; k++)
+  for (k= keyword_names; k->name; k++)
   {
     if (length == k->length && !strncasecmp(beg, k->name, length))
     {
@@ -1371,15 +1383,32 @@ my_xpath_lex_scan(MY_XPATH *xpath,
          beg+= length) /* no op */;
     lex->end= beg;
 
-    // check if a function call
-    if (*beg == '(' && (xpath->func= my_xpath_function(lex->beg, beg)))
+    if (beg < end)
     {
+      if (*beg == '(')
+      {
+        /*
+         check if a function call, e.g.: count(/a/b)
+         or a nodetype test,       e.g.: /a/b/text()
+        */
+        if ((xpath->func= my_xpath_function(lex->beg, beg)))
           lex->term= MY_XPATH_LEX_FUNC;
+        else
+          lex->term= my_xpath_keyword(xpath, my_nodetype_names,
+                                      lex->beg, beg);
         return;
       }
-
+      // check if an axis specifier, e.g.: /a/b/child::*
+      else if (*beg == ':' && beg + 1 < end && beg[1] == ':')
+      {
+        lex->term= my_xpath_keyword(xpath, my_axis_names,
+                                    lex->beg, beg);
+        return;
+      }
+    }
     // check if a keyword
-    lex->term= my_xpath_keyword(xpath, lex->beg, beg);
+    lex->term= my_xpath_keyword(xpath, my_keyword_names,
+                                lex->beg, beg);
     return;
   }
 
@@ -2333,6 +2362,36 @@ static int my_xpath_parse_Number(MY_XPATH *xpath)
 }
 
 
+/*
+  Scan NCName.
+  
+  SYNOPSYS
+    
+    The keywords AND, OR, MOD, DIV are valid identitiers
+    when they are in identifier context:
+    
+    SELECT
+    ExtractValue('<and><or><mod><div>VALUE</div></mod></or></and>',
+                 '/and/or/mod/div')
+    ->  VALUE
+    
+  RETURN
+    1 - success
+    0 - failure
+*/
+
+static int
+my_xpath_parse_NCName(MY_XPATH *xpath)
+{
+  return
+    my_xpath_parse_term(xpath, MY_XPATH_LEX_IDENT) ||
+    my_xpath_parse_term(xpath, MY_XPATH_LEX_AND)   ||
+    my_xpath_parse_term(xpath, MY_XPATH_LEX_OR)    ||
+    my_xpath_parse_term(xpath, MY_XPATH_LEX_MOD)   ||
+    my_xpath_parse_term(xpath, MY_XPATH_LEX_DIV) ? 1 : 0;
+}
+
+
 /*
   QName grammar can be found in a separate document
   http://www.w3.org/TR/REC-xml-names/#NT-QName
@@ -2341,16 +2400,17 @@ static int my_xpath_parse_Number(MY_XPATH *xpath)
   [7] 	Prefix    ::= NCName
   [8] 	LocalPart ::= NCName
 */
+
 static int
 my_xpath_parse_QName(MY_XPATH *xpath)
 {
   const char *beg;
-  if (!my_xpath_parse_term(xpath, MY_XPATH_LEX_IDENT))
+  if (!my_xpath_parse_NCName(xpath))
     return 0;
   beg= xpath->prevtok.beg;
   if (!my_xpath_parse_term(xpath, MY_XPATH_LEX_COLON))
     return 1; /* Non qualified name */
-  if (!my_xpath_parse_term(xpath, MY_XPATH_LEX_IDENT))
+  if (!my_xpath_parse_NCName(xpath))
     return 0;
   xpath->prevtok.beg= beg;
   return 1;