neeraj 2003/11/17 23:34:36
Modified: java/src/org/apache/xerces/dom DOMConfigurationImpl.java
DOMNormalizer.java
Log:
Now applications can set the value of well-formedness feature to either true or fals
when normalizing document.
Revision Changes Path
1.14 +16 -15
xml-xerces/java/src/org/apache/xerces/dom/DOMConfigurationImpl.java
Index: DOMConfigurationImpl.java
===================================================================
RCS file:
/home/cvs/xml-xerces/java/src/org/apache/xerces/dom/DOMConfigurationImpl.java,v
retrieving revision 1.13
retrieving revision 1.14
diff -u -r1.13 -r1.14
--- DOMConfigurationImpl.java 17 Nov 2003 13:48:40 -0000 1.13
+++ DOMConfigurationImpl.java 18 Nov 2003 07:34:36 -0000 1.14
@@ -188,7 +188,7 @@
protected final static short COMMENTS = 0x1<<5;
protected final static short VALIDATE = 0x1<<6;
protected final static short PSVI = 0x1<<7;
-
+ protected final static short WELLFORMED = 0x1<<8;
// components
/** Symbol table. */
@@ -290,6 +290,7 @@
features |= COMMENTS;
features |= CDATA;
features |= SPLITCDATA;
+ features |= WELLFORMED;
if (symbolTable == null) {
symbolTable = new SymbolTable();
@@ -569,23 +570,22 @@
}
else if (name.equals(Constants.DOM_ENTITIES)) {
features = (short) (state ? features | ENTITIES : features &
~ENTITIES);
-
}
else if (name.equals(Constants.DOM_SPLIT_CDATA)) {
features = (short) (state ? features | SPLITCDATA : features &
~SPLITCDATA);
-
}
else if (name.equals(Constants.DOM_VALIDATE)) {
features = (short) (state ? features | VALIDATE : features &
~VALIDATE);
-
}
+ else if (name.equals(Constants.DOM_WELLFORMED)) {
+ features = (short) (state ? features | WELLFORMED : features &
~WELLFORMED );
+ }
else if (name.equals(Constants.DOM_INFOSET)
|| name.equals(Constants.DOM_NORMALIZE_CHARACTERS)
|| name.equals(Constants.DOM_CANONICAL_FORM)
|| name.equals(Constants.DOM_VALIDATE_IF_SCHEMA)
|| name.equals(Constants.DOM_CHECK_CHAR_NORMALIZATION)
- //REVISIT: we need to support true value
- || name.equals(Constants.DOM_WELLFORMED)) {
+ ) {
if (state) { // true is not supported
String msg =
DOMMessageFormatter.formatMessage(
@@ -809,13 +809,14 @@
else if (name.equals(Constants.DOM_VALIDATE)) {
return (features & VALIDATE) != 0 ? Boolean.TRUE :
Boolean.FALSE;
}
- else if (name.equals(Constants.DOM_INFOSET)
+ else if (name.equals(Constants.DOM_WELLFORMED)) {
+ return (features & WELLFORMED) != 0 ? Boolean.TRUE :
Boolean.FALSE;
+ }
+ else if ( name.equals(Constants.DOM_INFOSET)
|| name.equals(Constants.DOM_NORMALIZE_CHARACTERS)
|| name.equals(Constants.DOM_CANONICAL_FORM)
|| name.equals(Constants.DOM_VALIDATE_IF_SCHEMA)
- || name.equals(Constants.DOM_CHECK_CHAR_NORMALIZATION)
- //REVISIT: currently its set to false
- || name.equals(Constants.DOM_WELLFORMED)
+ || name.equals(Constants.DOM_CHECK_CHAR_NORMALIZATION)
) {
return Boolean.FALSE;
}
@@ -898,7 +899,9 @@
|| name.equals(Constants.DOM_ENTITIES)
|| name.equals(Constants.DOM_SPLIT_CDATA)
|| name.equals(Constants.DOM_NAMESPACES)
- || name.equals(Constants.DOM_VALIDATE)) {
+ || name.equals(Constants.DOM_VALIDATE)
+ || name.equals(Constants.DOM_WELLFORMED)
+ ) {
return true ;
}//features whose parameter value can not be set to 'true'
else if (
@@ -906,9 +909,7 @@
|| name.equals(Constants.DOM_NORMALIZE_CHARACTERS)
|| name.equals(Constants.DOM_CANONICAL_FORM)
|| name.equals(Constants.DOM_VALIDATE_IF_SCHEMA)
- || name.equals(Constants.DOM_CHECK_CHAR_NORMALIZATION)
- //REVISIT: we need to support true value
- || name.equals(Constants.DOM_WELLFORMED)
+ || name.equals(Constants.DOM_CHECK_CHAR_NORMALIZATION)
) {
return (value.equals(Boolean.TRUE)) ? false : true;
}//features whose parameter value can not be set to 'false'
1.43 +64 -73 xml-xerces/java/src/org/apache/xerces/dom/DOMNormalizer.java
Index: DOMNormalizer.java
===================================================================
RCS file: /home/cvs/xml-xerces/java/src/org/apache/xerces/dom/DOMNormalizer.java,v
retrieving revision 1.42
retrieving revision 1.43
diff -u -r1.42 -r1.43
--- DOMNormalizer.java 17 Nov 2003 15:26:01 -0000 1.42
+++ DOMNormalizer.java 18 Nov 2003 07:34:36 -0000 1.43
@@ -314,8 +314,11 @@
if (DEBUG_ND) {
System.out.println("==>normalizeNode:{element}
"+node.getNodeName());
}
- //do the name check only when version of the document was changed.
- if(fDocument.isXMLVersionChanged()){
+
+ //do the name check only when version of the document was changed &
+ //application has set the value of well-formed features to true
+ if ( ((fConfiguration.features & DOMConfigurationImpl.WELLFORMED)
!= 0) &&
+ fDocument.isXMLVersionChanged()){
//take care of namespaces
if(fNamespaceValidation){
//checkQName does checking based on the version of the
document
@@ -402,10 +405,11 @@
Attr attr = (Attr)attributes.item(i);
//removeDefault(attr, attributes);
attr.normalize();
- //REVISIT: As of right now we are doing checks only if
the XML version changed at any moment
- //but it is possible that bad XML characters enter into
DOM when created in memory -- so we should
- //still be doing these checks when document is loaded
or modified in memory
- if(fDocument.isXMLVersionChanged()){
+
+ //do the name check only when version of the document
was changed &
+ //application has set the value of well-formed features
to true
+ if ( ((fConfiguration.features &
DOMConfigurationImpl.WELLFORMED) != 0) &&
+ fDocument.isXMLVersionChanged()){
fDocument.isXMLName(node.getNodeName() ,
fDocument.isXML11Version());
}
// XML 1.0 attribute value normalization
@@ -485,15 +489,13 @@
}
}//if comment node need not be removed
else {
- //REVISIT: As of right now we are doing checks only if the XML
version changed at any moment
- //but it is possible that bad XML characters enter into DOM
when created in memory -- so we should
- //still be doing these checks when document is loaded or
modified in memory
- if(fDocument.isXMLVersionChanged()){
+
+ //do the well-formed valid character check when application has
set the value of well-formed features to true
+ if ( ((fConfiguration.features &
DOMConfigurationImpl.WELLFORMED) != 0)){
String commentdata = ((Comment)node).getData();
//check comments for invalid xml chracter as per the version
//of the document
- checkInValidXMLCharacters(commentdata,
fDocument.isXML11Version());
-
+ checkInValidXMLCharacters(commentdata,
fDocument.isXML11Version());
}
}//end-else if comment node is not to be removed.
}
@@ -501,8 +503,11 @@
if (DEBUG_ND) {
System.out.println("==>normalizeNode:{entityRef}
"+node.getNodeName());
}
- //do the name check only when version of the document was changed.
- if(fDocument.isXMLVersionChanged()){
+
+ //do the name check only when version of the document was changed &
+ //application has set the value of well-formed features to true
+ if ( ((fConfiguration.features & DOMConfigurationImpl.WELLFORMED)
!= 0) &&
+ fDocument.isXMLVersionChanged()){
//REVISIT: checkQName takes care of the version of the document
//but isXMLName doesn't.... why its so ?
fDocument.isXMLName(node.getNodeName() ,
fDocument.isXML11Version());
@@ -551,13 +556,10 @@
//or presence of ']]>' in CDATA shouldnot affect the following
checks
//we should be checking for presence of valid XML characters
- //REVISIT: As of right now we are doing checks only if the XML
version changed at any moment
- //but it is possible that bad XML characters enter into DOM when
created in memory -- so we should
- //still be doing these checks when document is loaded or modified
in memory
-
- if(fDocument.isXMLVersionChanged()){
- String cdatavalue = node.getNodeValue() ;
- checkInValidXMLCharacters(cdatavalue,
fDocument.isXML11Version());
+
+ //do the well-formed valid character check when application has set
the value of well-formed features to true
+ if ( ((fConfiguration.features & DOMConfigurationImpl.WELLFORMED)
!= 0) ){
+ checkInValidXMLCharacters(node.getNodeValue(),
fDocument.isXML11Version());
}
if ((fConfiguration.features & DOMConfigurationImpl.CDATA) == 0) {
@@ -621,13 +623,9 @@
((Text)node).appendData(next.getNodeValue());
node.getParentNode().removeChild( next );
- //check the text values for valid xml character as per document
version...
-
- //REVISIT: As of right now we are doing checks only if the XML
version changed at any moment
- //but it is possible that bad XML characters enter into DOM
when created in memory -- so we should
- //still be doing these checks when document is loaded or
modified in memory
-
- if(fDocument.isXMLVersionChanged()){
+ //check the text values for valid xml character as per document
version
+ //when application has set the value of well-formed features to true
+ if ( ((fConfiguration.features & DOMConfigurationImpl.WELLFORMED)
!= 0) ){
checkInValidXMLCharacters(node.getNodeValue(),
fDocument.isXML11Version());
}
@@ -637,13 +635,9 @@
// If kid is empty, remove it
node.getParentNode().removeChild( node );
} else {
- //check the text values for valid xml character as per document
version...
-
- //REVISIT: As of right now we are doing checks only if the XML
version changed at any moment
- //but it is possible that bad XML characters enter into DOM
when created in memory -- so we should
- //still be doing these checks when document is loaded or
modified in memory
-
- if(fDocument.isXMLVersionChanged()){
+ //check the text values for valid xml character as per document
version...
+ //do the name check when application has set the value of
well-formed features to true
+ if ( ((fConfiguration.features &
DOMConfigurationImpl.WELLFORMED) != 0) ){
checkInValidXMLCharacters(node.getNodeValue(),
fDocument.isXML11Version());
}
@@ -683,38 +677,35 @@
break;
}
case org.w3c.dom.Node.PROCESSING_INSTRUCTION_NODE: {
- //REVISIT: As of right now we are doing checks only if the XML version
changed at any moment
- //but it is possible that bad XML characters enter into DOM when
created in memory -- so we should
- //still be doing these checks when document is loaded or modified in
memory
- if(!fDocument.isXMLVersionChanged()){
- break ;
+ //do the well-formed valid PI target name , data check when application
has set the value of well-formed feature to true
+ if((fConfiguration.features & DOMConfigurationImpl.WELLFORMED) != 0 ){
+ ProcessingInstruction pinode = (ProcessingInstruction)node ;
+
+ String target = pinode.getTarget();
+ //1.check PI target name
+ if(fDocument.isXML11Version()){
+
+ if(!XML11Char.isXML11ValidName(target)){
+ //REVISIT: As per DOM it is error but as per XML spec.
it is fatal error
+ reportDOMError("Invalid Character in node name",
+ DOMError.SEVERITY_FATAL_ERROR, node,
"wf-invalid-character-in-node-name");
+ }
+ }
+ else{
+ if(!XMLChar.isValidName(target)){
+ //REVISIT: As per DOM it is error but as per XML spec.
it is fatal error
+ reportDOMError("Invalid Character in node name",
+ DOMError.SEVERITY_FATAL_ERROR, node,
"wf-invalid-character-in-node-name");
+ }
+ }
+
+ //2. check PI data
+ //processing isntruction data may have certain characters
+ //which may not be valid XML character
+
+ checkInValidXMLCharacters(pinode.getData(),
fDocument.isXML11Version());
}
- //processing isntruction data may have certain characters
- //which may not be valid XML character
- ProcessingInstruction pinode = (ProcessingInstruction)node ;
-
- String target = pinode.getTarget();
- //1.check PI target name
- if(fDocument.isXML11Version()){
-
- if(!XML11Char.isXML11ValidName(target)){
- //REVISIT: As per DOM it is error but as per XML spec. it
is fatal error
- reportDOMError("Invalid Character in node name",
- DOMError.SEVERITY_FATAL_ERROR, node,
"wf-invalid-character-in-node-name");
- }
- }
- else{
- if(!XMLChar.isValidName(target)){
- //REVISIT: As per DOM it is error but as per XML spec. it
is fatal error
- reportDOMError("Invalid Character in node name",
- DOMError.SEVERITY_FATAL_ERROR, node,
"wf-invalid-character-in-node-name");
- }
- }
-
- //2. check PI data
- checkInValidXMLCharacters(pinode.getData(), fDocument.isXML11Version());
-
}//end case Node.PROCESSING_INSTRUCTION_NODE
}//end of switch
@@ -762,11 +753,11 @@
// Record all valid local declarations
for (int k=0; k < attributes.getLength(); k++) {
Attr attr = (Attr)attributes.getItem(k);
-
- //REVISIT: As of right now we are doing checks only if the XML
version changed at any moment
- //but it is possible that bad XML characters enter into DOM when
created in memory -- so we should
- //still be doing these checks when document is loaded or modified
in memory
- if(fDocument.isXMLVersionChanged()){
+
+ //do the name check only when version of the document was changed &
+ //application has set the value of well-formed features to true
+ if ( ((fConfiguration.features & DOMConfigurationImpl.WELLFORMED)
!= 0) &&
+ fDocument.isXMLVersionChanged()){
//checkQName does checking based on the version of the document
fDocument.checkQName(attr.getPrefix() , attr.getLocalName()) ;
}
@@ -1014,7 +1005,7 @@
int i = 0 ;
while(i < datalength){
if(XML11Char.isXML11Invalid(dataarray[i++])){
- String msg = "Invalid XML Character " +
Integer.toString(dataarray[i-1], 16) ;
+ String msg = "Invalid XML Character " + "'" +
Integer.toString(dataarray[i-1], 16) + "'" + "in the DOM." ;
//REVISIT: As per DOM it is error but as per XML spec. it is
fatal error
reportDOMError(msg,
DOMError.SEVERITY_FATAL_ERROR, null,
"wf-invalid-character");
@@ -1027,7 +1018,7 @@
int i = 0 ;
while(i < datalength){
if( XMLChar.isInvalid(dataarray[i++]) ){
- String msg = "Invalid XML Character " +
Integer.toString(dataarray[i-1], 16) ;
+ String msg = "Invalid XML Character " + "'" +
Integer.toString(dataarray[i-1], 16) + "'" + "in the DOM." ;
//REVISIT: As per DOM it is error but as per XML spec. it is
fatal error
reportDOMError(msg,
DOMError.SEVERITY_FATAL_ERROR, null,
"wf-invalid-character");
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
