public Parser(InputSource source)
{
TagSoup.Parser parser = new TagSoup.Parser();
parser.SetFeature(TagSoup.Parser.NAMESPACES_FEATURE, true);
StringBuilder title = new StringBuilder(), body = new StringBuilder();
DefaultHandler handler = new DefaultHandlerAnonymousHelper(this, title, body);
parser.ContentHandler = handler;
parser.ErrorHandler = handler;
parser.Parse(source);
// the javacc-based parser trimmed title (which should be done for HTML in all cases):
this.title = title.ToString().Trim();
// assign body text
this.body = body.ToString();
}
public void Run()
{
try
{
Sax.IXMLReader reader = new TagSoup.Parser(); //XMLReaderFactory.createXMLReader();
reader.ContentHandler = this;
reader.ErrorHandler = this;
while (!stopped)
{
Stream localFileIS = outerInstance.@is;
if (localFileIS != null)
{ // null means fileIS was closed on us
try
{
// To work around a bug in XERCES (XERCESJ-1257), we assume the XML is always UTF8, so we simply provide reader.
reader.Parse(new InputSource(IOUtils.GetDecodingReader(localFileIS, Encoding.UTF8)));
}
catch (Exception ioe) when(ioe.IsIOException())
{
lock (outerInstance)
{
if (localFileIS != outerInstance.@is)
{
// fileIS was closed on us, so, just fall through
}
else
{
// Exception is real
throw; // LUCENENET: CA2200: Rethrow to preserve stack details (https://docs.microsoft.com/en-us/visualstudio/code-quality/ca2200-rethrow-to-preserve-stack-details)
}
}
}
}
lock (this)
{
if (stopped || !outerInstance.m_forever)
{
nmde = new NoMoreDataException();
Monitor.Pulse(this); //notify();
return;
}
else if (localFileIS == outerInstance.@is)
{
// If file is not already re-opened then re-open it now
outerInstance.@is = outerInstance.OpenInputStream();
}
}
}
}
catch (SAXException sae)
{
throw RuntimeException.Create(sae);
}
catch (Exception ioe) when(ioe.IsIOException())
{
throw RuntimeException.Create(ioe);
}
finally
{
lock (this)
{
threadDone = true;
Monitor.Pulse(this); //Notify();
}
}
}
public void Run()
{
try
{
Sax.IXMLReader reader = new TagSoup.Parser(); //XMLReaderFactory.createXMLReader();
reader.ContentHandler = this;
reader.ErrorHandler = this;
while (!stopped)
{
Stream localFileIS = outerInstance.@is;
if (localFileIS != null)
{ // null means fileIS was closed on us
try
{
// To work around a bug in XERCES (XERCESJ-1257), we assume the XML is always UTF8, so we simply provide reader.
reader.Parse(new InputSource(IOUtils.GetDecodingReader(localFileIS, Encoding.UTF8)));
}
catch (IOException ioe)
{
lock (outerInstance)
{
if (localFileIS != outerInstance.@is)
{
// fileIS was closed on us, so, just fall through
}
else
{
// Exception is real
throw ioe;
}
}
}
}
lock (this)
{
if (stopped || !outerInstance.m_forever)
{
nmde = new NoMoreDataException();
Monitor.Pulse(this); //notify();
return;
}
else if (localFileIS == outerInstance.@is)
{
// If file is not already re-opened then re-open it now
outerInstance.@is = outerInstance.OpenInputStream();
}
}
}
}
catch (SAXException sae)
{
throw new Exception(sae.ToString(), sae);
}
catch (IOException ioe)
{
throw new Exception(ioe.ToString(), ioe);
}
finally
{
lock (this)
{
threadDone = true;
Monitor.Pulse(this); //Notify();
}
}
}
