bool Parser::parseProlog( Node *node )
{
parseXMLDeclaration( node );
while( parseMisc( node ) ) {}
parseDocType( node );
while( parseMisc( node ) ) {}
return true;
}
Node *Parser::parseDocument()
{
readChar( false );
Node *result = new Node;
result->setName( "" );
result->setNodeType( Document );
if( !parseProlog( result ) )
reportError( "Missing prolog" );
if( !parseElement( result ) )
reportError( "Missing content" );
while( parseMisc( result ) ) {}
return result;
}
UXMLElement *
UXMLParser::parse(const UnicodeString &src, UErrorCode &status) {
if(U_FAILURE(status)) {
return NULL;
}
UXMLElement *root = NULL;
fPos = 0; // TODO use just a local pos variable and pass it into functions
// where necessary?
// set all matchers to work on the input string
mXMLDecl.reset(src);
mXMLComment.reset(src);
mXMLSP.reset(src);
mXMLDoctype.reset(src);
mXMLPI.reset(src);
mXMLElemStart.reset(src);
mXMLElemEnd.reset(src);
mXMLElemEmpty.reset(src);
mXMLCharData.reset(src);
mAttrValue.reset(src);
mAttrNormalizer.reset(src);
mNewLineNormalizer.reset(src);
mAmps.reset(src);
// Consume the XML Declaration, if present.
if (mXMLDecl.lookingAt(fPos, status)) {
fPos = mXMLDecl.end(status);
}
// Consume "misc" [XML production 27] appearing before DocType
parseMisc(status);
// Consume a DocType declaration, if present.
if (mXMLDoctype.lookingAt(fPos, status)) {
fPos = mXMLDoctype.end(status);
}
// Consume additional "misc" [XML production 27] appearing after the DocType
parseMisc(status);
// Get the root element
if (mXMLElemEmpty.lookingAt(fPos, status)) {
// Root is an empty element (no nested elements or content)
root = createElement(mXMLElemEmpty, status);
fPos = mXMLElemEmpty.end(status);
} else {
if (mXMLElemStart.lookingAt(fPos, status) == FALSE) {
error("Root Element expected", status);
goto errorExit;
}
root = createElement(mXMLElemStart, status);
UXMLElement *el = root;
//
// This is the loop that consumes the root element of the document,
// including all nested content. Nested elements are handled by
// explicit pushes/pops of the element stack; there is no recursion
// in the control flow of this code.
// "el" always refers to the current element, the one to which content
// is being added. It is above the top of the element stack.
for (;;) {
// Nested Element Start
if (mXMLElemStart.lookingAt(fPos, status)) {
UXMLElement *t = createElement(mXMLElemStart, status);
el->fChildren.addElement(t, status);
t->fParent = el;
fElementStack.push(el, status);
el = t;
continue;
}
// Text Content. String is concatenated onto the current node's content,
// but only if it contains something other than spaces.
UnicodeString s = scanContent(status);
if (s.length() > 0) {
mXMLSP.reset(s);
if (mXMLSP.matches(status) == FALSE) {
// This chunk of text contains something other than just
// white space. Make a child node for it.
replaceCharRefs(s, status);
el->fChildren.addElement(s.clone(), status);
}
mXMLSP.reset(src); // The matchers need to stay set to the main input string.
continue;
}
// Comments. Discard.
if (mXMLComment.lookingAt(fPos, status)) {
fPos = mXMLComment.end(status);
continue;
}
// PIs. Discard.
if (mXMLPI.lookingAt(fPos, status)) {
fPos = mXMLPI.end(status);
continue;
}
// Element End
if (mXMLElemEnd.lookingAt(fPos, status)) {
fPos = mXMLElemEnd.end(0, status);
const UnicodeString name = mXMLElemEnd.group(1, status);
if (name != *el->fName) {
error("Element start / end tag mismatch", status);
goto errorExit;
}
if (fElementStack.empty()) {
// Close of the root element. We're done with the doc.
el = NULL;
break;
}
el = (UXMLElement *)fElementStack.pop();
continue;
}
// Empty Element. Stored as a child of the current element, but not stacked.
if (mXMLElemEmpty.lookingAt(fPos, status)) {
UXMLElement *t = createElement(mXMLElemEmpty, status);
el->fChildren.addElement(t, status);
continue;
}
// Hit something within the document that doesn't match anything.
// It's an error.
error("Unrecognized markup", status);
break;
}
if (el != NULL || !fElementStack.empty()) {
// We bailed out early, for some reason.
error("Root element not closed.", status);
goto errorExit;
}
}
// Root Element parse is complete.
// Consume the annoying xml "Misc" that can appear at the end of the doc.
parseMisc(status);
// We should have reached the end of the input
if (fPos != src.length()) {
error("Extra content at the end of the document", status);
goto errorExit;
}
// Success!
return root;
errorExit:
delete root;
return NULL;
}
