/**
* processes the tag.
* @param start if true we are dealing with a tag that has just been opened; if false we are closing a tag.
*/
private void ProcessTag(bool start)
{
if (start)
{
nested++;
doc.StartElement(tag, attributes);
}
else
{
nested--;
doc.EndElement(tag);
}
}
/**
* processes the tag.
* @param start if true we are dealing with a tag that has just been opened; if false we are closing a tag.
*/
private void ProcessTag(bool start)
{
if (start)
{
nested++;
doc.StartElement(tag, attributes);
}
else
{
// White spaces following new lines need to be ignored in HTML
if (newLineHandler.IsNewLineTag(tag))
{
nowhite = false;
}
nested--;
doc.EndElement(tag);
}
}
/**
* Parses the XML document firing the events to the handler.
* @param doc the document handler
* @param r the document. The encoding is already resolved. The reader is not closed
* @throws IOException on error
*/
public static void Parse(ISimpleXMLDocHandler doc, ISimpleXMLDocHandlerComment comment, TextReader reader, bool html)
{
Stack st = new Stack();
int depth = 0;
int mode = PRE;
int c = 0;
int quotec = '"';
depth = 0;
StringBuilder sb = new StringBuilder();
StringBuilder etag = new StringBuilder();
String tagName = null;
String lvalue = null;
String rvalue = null;
Hashtable attrs = null;
st = new Stack();
doc.StartDocument();
int line=1, col=0;
bool eol = false;
if (html)
mode = TEXT;
int pushBack = -1;
while (true) {
if (pushBack != -1) {
c = pushBack;
pushBack = -1;
}
else
c = reader.Read();
if (c == -1)
break;
// We need to map \r, \r\n, and \n to \n
// See XML spec section 2.11
if (c == '\n' && eol) {
eol = false;
continue;
} else if (eol) {
eol = false;
} else if (c == '\n') {
line++;
col=0;
} else if (c == '\r') {
eol = true;
c = '\n';
line++;
col=0;
} else {
col++;
}
if (mode == DONE) {
doc.EndDocument();
return;
// We are between tags collecting text.
} else if (mode == TEXT) {
if (c == '<') {
st.Push(mode);
mode = START_TAG;
if (sb.Length > 0) {
doc.Text(sb.ToString());
sb.Length = 0;
}
} else if (c == '&') {
st.Push(mode);
mode = ENTITY;
etag.Length = 0;
} else
sb.Append((char)c);
// we are processing a closing tag: e.g. </foo>
} else if (mode == CLOSE_TAG) {
if (c == '>') {
mode = PopMode(st);
tagName = sb.ToString();
if (html)
tagName = tagName.ToLower(CultureInfo.InvariantCulture);
sb.Length = 0;
depth--;
if (!html && depth==0)
mode = DONE;
doc.EndElement(tagName);
} else {
if (!char.IsWhiteSpace((char)c))
sb.Append((char)c);
}
// we are processing CDATA
} else if (mode == CDATA) {
if (c == '>'
&& sb.ToString().EndsWith("]]")) {
sb.Length = sb.Length-2;
doc.Text(sb.ToString());
sb.Length = 0;
mode = PopMode(st);
} else
sb.Append((char)c);
// we are processing a comment. We are inside
// the <!-- .... --> looking for the -->.
} else if (mode == COMMENT) {
if (c == '>'
&& sb.ToString().EndsWith("--")) {
if (comment != null) {
sb.Length = sb.Length - 2;
comment.Comment(sb.ToString());
}
sb.Length = 0;
mode = PopMode(st);
} else
sb.Append((char)c);
// We are outside the root tag element
} else if (mode == PRE) {
if (c == '<') {
mode = TEXT;
st.Push(mode);
mode = START_TAG;
}
// We are inside one of these <? ... ?>
// or one of these <!DOCTYPE ... >
} else if (mode == DOCTYPE) {
if (c == '>') {
mode = PopMode(st);
if (mode == TEXT) mode = PRE;
}
// we have just seen a < and
// are wondering what we are looking at
// <foo>, </foo>, <!-- ... --->, etc.
} else if (mode == START_TAG) {
mode = PopMode(st);
if (c == '/') {
st.Push(mode);
mode = CLOSE_TAG;
} else if (c == '?') {
mode = DOCTYPE;
} else {
st.Push(mode);
mode = OPEN_TAG;
tagName = null;
attrs = new Hashtable();
sb.Append((char)c);
}
// we are processing an entity, e.g. <, », etc.
} else if (mode == ENTITY) {
if (c == ';') {
mode = PopMode(st);
String cent = etag.ToString();
etag.Length = 0;
if (cent.StartsWith("#x")) {
try {
char ci = (char)int.Parse(cent.Substring(2), NumberStyles.AllowHexSpecifier);
sb.Append(ci);
}
catch {
sb.Append('&').Append(cent).Append(';');
}
}
else if (cent.StartsWith("#")) {
try {
char ci = (char)int.Parse(cent.Substring(1));
sb.Append(ci);
}
catch {
sb.Append('&').Append(cent).Append(';');
}
}
else {
char ce = EntitiesToUnicode.DecodeEntity(cent);
if (ce == '\0')
sb.Append('&').Append(cent).Append(';');
else
sb.Append(ce);
}
} else if ((c != '#' && (c < '0' || c > '9') && (c < 'a' || c > 'z')
&& (c < 'A' || c > 'Z')) || etag.Length >= 7) {
mode = PopMode(st);
pushBack = c;
sb.Append('&').Append(etag.ToString());
etag.Length = 0;
}
else {
etag.Append((char)c);
}
// we have just seen something like this:
// <foo a="b"/
// and are looking for the final >.
} else if (mode == SINGLE_TAG) {
if (tagName == null)
tagName = sb.ToString();
if (html)
tagName = tagName.ToLower(CultureInfo.InvariantCulture);
if (c != '>')
Exc("Expected > for tag: <"+tagName+"/>",line,col);
doc.StartElement(tagName,attrs);
doc.EndElement(tagName);
if (!html && depth==0) {
doc.EndDocument();
return;
}
sb.Length = 0;
attrs = new Hashtable();
tagName = null;
mode = PopMode(st);
// we are processing something
// like this <foo ... >. It could
// still be a <!-- ... --> or something.
} else if (mode == OPEN_TAG) {
if (c == '>') {
if (tagName == null)
tagName = sb.ToString();
if (html)
tagName = tagName.ToLower(CultureInfo.InvariantCulture);
sb.Length = 0;
depth++;
doc.StartElement(tagName,attrs);
tagName = null;
attrs = new Hashtable();
mode = PopMode(st);
} else if (c == '/') {
mode = SINGLE_TAG;
} else if (c == '-' && sb.ToString().Equals("!-")) {
mode = COMMENT;
sb.Length = 0;
} else if (c == '[' && sb.ToString().Equals("![CDATA")) {
mode = CDATA;
sb.Length = 0;
} else if (c == 'E' && sb.ToString().Equals("!DOCTYP")) {
sb.Length = 0;
mode = DOCTYPE;
} else if (char.IsWhiteSpace((char)c)) {
tagName = sb.ToString();
if (html)
tagName = tagName.ToLower(CultureInfo.InvariantCulture);
sb.Length = 0;
mode = IN_TAG;
} else {
sb.Append((char)c);
}
// We are processing the quoted right-hand side
// of an element's attribute.
} else if (mode == QUOTE) {
if (html && quotec == ' ' && c == '>') {
rvalue = sb.ToString();
sb.Length = 0;
attrs[lvalue] = rvalue;
mode = PopMode(st);
doc.StartElement(tagName,attrs);
depth++;
tagName = null;
attrs = new Hashtable();
}
else if (html && quotec == ' ' && char.IsWhiteSpace((char)c)) {
rvalue = sb.ToString();
sb.Length = 0;
attrs[lvalue] = rvalue;
mode = IN_TAG;
}
else if (html && quotec == ' ') {
sb.Append((char)c);
}
else if (c == quotec) {
rvalue = sb.ToString();
sb.Length = 0;
attrs[lvalue] = rvalue;
mode = IN_TAG;
// See section the XML spec, section 3.3.3
// on normalization processing.
} else if (" \r\n\u0009".IndexOf((char)c)>=0) {
sb.Append(' ');
} else if (c == '&') {
st.Push(mode);
mode = ENTITY;
etag.Length = 0;
} else {
sb.Append((char)c);
}
} else if (mode == ATTRIBUTE_RVALUE) {
if (c == '"' || c == '\'') {
quotec = c;
mode = QUOTE;
} else if (char.IsWhiteSpace((char)c)) {
} else if (html && c == '>') {
attrs[lvalue] = sb.ToString();
sb.Length = 0;
mode = PopMode(st);
doc.StartElement(tagName,attrs);
depth++;
tagName = null;
attrs = new Hashtable();
} else if (html) {
sb.Append((char)c);
quotec = ' ';
mode = QUOTE;
} else {
Exc("Error in attribute processing",line,col);
}
} else if (mode == ATTRIBUTE_LVALUE) {
if (char.IsWhiteSpace((char)c)) {
lvalue = sb.ToString();
if (html)
lvalue = lvalue.ToLower(CultureInfo.InvariantCulture);
sb.Length = 0;
mode = ATTRIBUTE_EQUAL;
} else if (c == '=') {
lvalue = sb.ToString();
if (html)
lvalue = lvalue.ToLower(CultureInfo.InvariantCulture);
sb.Length = 0;
mode = ATTRIBUTE_RVALUE;
} else if (html && c == '>') {
sb.Length = 0;
mode = PopMode(st);
doc.StartElement(tagName,attrs);
depth++;
tagName = null;
attrs = new Hashtable();
} else {
sb.Append((char)c);
}
} else if (mode == ATTRIBUTE_EQUAL) {
if (c == '=') {
mode = ATTRIBUTE_RVALUE;
} else if (char.IsWhiteSpace((char)c)) {
} else if (html && c == '>') {
sb.Length = 0;
mode = PopMode(st);
doc.StartElement(tagName,attrs);
depth++;
tagName = null;
attrs = new Hashtable();
} else if (html && c == '/') {
sb.Length = 0;
mode = SINGLE_TAG;
} else if (html) {
sb.Length = 0;
sb.Append((char)c);
mode = ATTRIBUTE_LVALUE;
} else {
Exc("Error in attribute processing.",line,col);
}
} else if (mode == IN_TAG) {
if (c == '>') {
mode = PopMode(st);
doc.StartElement(tagName,attrs);
depth++;
tagName = null;
attrs = new Hashtable();
} else if (c == '/') {
mode = SINGLE_TAG;
} else if (char.IsWhiteSpace((char)c)) {
} else {
mode = ATTRIBUTE_LVALUE;
sb.Append((char)c);
}
}
}
if (html || mode == DONE) {
if (html && mode == TEXT)
doc.Text(sb.ToString());
doc.EndDocument();
}
else
Exc("missing end tag",line,col);
}