/// <summary>
/// Put bytes into the parser.
/// </summary>
/// <param name="buf">The bytes to put into the parse stream</param>
/// <param name="offset">Offset into buf to start at</param>
/// <param name="length">Number of bytes to write</param>
public void Push(byte[] buf, int offset, int length)
{
// or assert, really, but this is a little nicer.
if (length == 0)
return;
// No locking is required. Read() won't get called again
// until this method returns.
// TODO: only do this copy if we have a partial token at the
// end of parsing.
byte[] copy = new byte[length];
System.Buffer.BlockCopy(buf, offset, copy, 0, length);
m_buf.Write(copy);
byte[] b = m_buf.GetBuffer();
int off = 0;
TOK tok = TOK.END_TAG;
ContentToken ct = new ContentToken();
try
{
while (off < b.Length)
{
if (m_cdata)
tok = m_enc.tokenizeCdataSection(b, off, b.Length, ct);
else
tok = m_enc.tokenizeContent(b, off, b.Length, ct);
switch (tok)
{
case TOK.EMPTY_ELEMENT_NO_ATTS:
case TOK.EMPTY_ELEMENT_WITH_ATTS:
StartTag(b, off, ct, tok);
EndTag(b, off, ct, tok);
break;
case TOK.START_TAG_NO_ATTS:
case TOK.START_TAG_WITH_ATTS:
StartTag(b, off, ct, tok);
break;
case TOK.END_TAG:
EndTag(b, off, ct, tok);
break;
case TOK.DATA_CHARS:
case TOK.DATA_NEWLINE:
AddText(utf.GetString(b, off, ct.TokenEnd - off));
break;
case TOK.CHAR_REF:
case TOK.MAGIC_ENTITY_REF:
AddText(new string(new char[] { ct.RefChar1 }));
break;
case TOK.CHAR_PAIR_REF:
AddText(new string(new char[] {ct.RefChar1,
ct.RefChar2}));
break;
case TOK.COMMENT:
if (current != null)
{
// <!-- 4
// --> 3
int start = off + 4 * m_enc.MinBytesPerChar;
int end = ct.TokenEnd - off -
7 * m_enc.MinBytesPerChar;
string text = utf.GetString(b, start, end);
current.AddChild(new Comment(text));
}
break;
case TOK.CDATA_SECT_OPEN:
m_cdata = true;
break;
case TOK.CDATA_SECT_CLOSE:
m_cdata = false;
break;
case TOK.XML_DECL:
// thou shalt use UTF8, and XML version 1.
// i shall ignore evidence to the contrary...
// TODO: Throw an exception if these assuptions are
// wrong
break;
case TOK.ENTITY_REF:
case TOK.PI:
#if CF
throw new util.NotImplementedException("Token type not implemented: " + tok);
#else
throw new System.NotImplementedException("Token type not implemented: " + tok);
#endif
}
off = ct.TokenEnd;
}
}
catch (PartialTokenException)
{
// ignored;
}
catch (ExtensibleTokenException)
{
// ignored;
}
catch (Exception ex)
{
if (OnStreamError != null)
OnStreamError(this, ex);
}
finally
{
m_buf.Clear(off);
}
}
private string NormalizeAttributeValue(byte[] buf, int offset, int length)
{
if (length == 0)
return null;
string val = null;
BufferAggregate buffer = new BufferAggregate();
byte[] copy = new byte[length];
System.Buffer.BlockCopy(buf, offset, copy, 0, length);
buffer.Write(copy);
byte[] b = buffer.GetBuffer();
int off = 0;
TOK tok = TOK.END_TAG;
ContentToken ct = new ContentToken();
try
{
while (off < b.Length)
{
//tok = m_enc.tokenizeContent(b, off, b.Length, ct);
tok = m_enc.tokenizeAttributeValue(b, off, b.Length, ct);
switch (tok)
{
case TOK.ATTRIBUTE_VALUE_S:
case TOK.DATA_CHARS:
case TOK.DATA_NEWLINE:
val += (utf.GetString(b, off, ct.TokenEnd - off));
break;
case TOK.CHAR_REF:
case TOK.MAGIC_ENTITY_REF:
val += new string(new char[] { ct.RefChar1 });
break;
case TOK.CHAR_PAIR_REF:
val += new string(new char[] {ct.RefChar1, ct.RefChar2});
break;
case TOK.ENTITY_REF:
#if CF
throw new util.NotImplementedException("Token type not implemented: " + tok);
#else
throw new System.NotImplementedException("Token type not implemented: " + tok);
#endif
}
off = ct.TokenEnd;
}
}
catch (PartialTokenException)
{
// ignored;
}
catch (ExtensibleTokenException)
{
// ignored;
}
catch (Exception ex)
{
if (OnStreamError != null)
OnStreamError(this, ex);
}
finally
{
buffer.Clear(off);
}
return val;
}
private void StartTag(byte[] buf, int offset,
ContentToken ct, TOK tok)
{
m_Depth++;
int colon;
string name;
string prefix;
Hashtable ht = new Hashtable();
m_ns.PushScope();
// if i have attributes
if ((tok == TOK.START_TAG_WITH_ATTS) ||
(tok == TOK.EMPTY_ELEMENT_WITH_ATTS))
{
int start;
int end;
string val;
for (int i=0; i<ct.getAttributeSpecifiedCount(); i++)
{
start = ct.getAttributeNameStart(i);
end = ct.getAttributeNameEnd(i);
name = utf.GetString(buf, start, end - start);
start = ct.getAttributeValueStart(i);
end = ct.getAttributeValueEnd(i);
//val = utf.GetString(buf, start, end - start);
val = NormalizeAttributeValue(buf, start, end - start);
// <foo b='&'/>
// <foo b='&amp;'
// TODO: if val includes &, it gets double-escaped
if (name.StartsWith("xmlns:"))
{
colon = name.IndexOf(':');
prefix = name.Substring(colon+1);
m_ns.AddNamespace(prefix, val);
}
else if (name == "xmlns")
{
m_ns.AddNamespace(string.Empty, val);
}
else
{
ht.Add(name, val);
}
}
}
name = utf.GetString(buf,
offset + m_enc.MinBytesPerChar,
ct.NameEnd - offset - m_enc.MinBytesPerChar);
colon = name.IndexOf(':');
string ns = "";
prefix = null;
if (colon > 0)
{
prefix = name.Substring(0, colon);
name = name.Substring(colon + 1);
ns = m_ns.LookupNamespace(prefix);
}
else
{
ns = m_ns.DefaultNamespace;
}
Element newel = ElementFactory.GetElement(prefix, name, ns);
foreach (string attrname in ht.Keys)
{
newel.SetAttribute(attrname, (string)ht[attrname]);
}
if (m_root == null)
{
m_root = newel;
//FireOnDocumentStart(m_root);
if (OnStreamStart!=null)
OnStreamStart(this, m_root);
}
else
{
if (current != null)
current.AddChild(newel);
current = newel;
}
}
private void EndTag(byte[] buf, int offset, ContentToken ct, TOK tok)
{
m_Depth--;
m_ns.PopScope();
if (current == null)
{// end of doc
if (OnStreamEnd!=null)
OnStreamEnd(this, m_root);
// FireOnDocumentEnd();
return;
}
string name = null;
if ((tok == TOK.EMPTY_ELEMENT_WITH_ATTS) ||
(tok == TOK.EMPTY_ELEMENT_NO_ATTS))
name = utf.GetString(buf,
offset + m_enc.MinBytesPerChar,
ct.NameEnd - offset -
m_enc.MinBytesPerChar);
else
name = utf.GetString(buf,
offset + m_enc.MinBytesPerChar*2,
ct.NameEnd - offset -
m_enc.MinBytesPerChar*2);
// if (current.Name != name)
// throw new Exception("Invalid end tag: " + name +
// " != " + current.Name);
Element parent = (Element) current.Parent;
if (parent == null)
{
DoRaiseOnStreamElement(current);
//if (OnStreamElement!=null)
// OnStreamElement(this, current);
//FireOnElement(current);
}
current = parent;
}