public void ExtractText(string html_string,
AppendTextCallback append_text_cb,
AddPropertyCallback add_prop_cb,
AppendSpaceCallback append_white_cb,
AppendSpaceCallback append_break_cb,
HotCallback hot_up_cb,
HotCallback hot_down_cb)
{
AppendText = append_text_cb;
AppendWord = append_text_cb;
AddProperty = add_prop_cb;
AppendWhiteSpace = append_white_cb;
AppendStructuralBreak = append_break_cb;
HotUp = hot_up_cb;
HotDown = hot_down_cb;
HtmlDocument doc = new HtmlDocument();
doc.ReportNode += HandleNodeEvent;
doc.StreamMode = true;
try {
doc.LoadHtml(html_string);
} catch (Exception e) {
Log.Debug(e, "Exception while filtering html string [{0}]", html_string);
}
}
public FilterHtml(bool register_filter)
{
if (register_filter)
{
base.SetVersion(version);
SnippetMode = true;
SetFileType("document");
AppendText = new AppendTextCallback(base.AppendText);
AppendWord = new AppendTextCallback(base.AppendWord);
AddProperty = new AddPropertyCallback(base.AddProperty);
AppendWhiteSpace = new AppendSpaceCallback(base.AppendWhiteSpace);
AppendStructuralBreak = new AppendSpaceCallback(base.AppendStructuralBreak);
HotUp = new HotCallback(base.HotUp);
HotDown = new HotCallback(base.HotDown);
#if ENABLE_RDF_ADAPTER
AddLink = new AddLinkCallback(base.AddLink);
#endif
}
ignore_level = 0;
building_text = false;
builder = new StringBuilder();
}
public FilterHtml (bool register_filter)
{
if (register_filter) {
base.SetVersion (version);
SnippetMode = true;
SetFileType ("document");
AppendText = new AppendTextCallback (base.AppendText);
AppendWord = new AppendTextCallback (base.AppendWord);
AddProperty = new AddPropertyCallback (base.AddProperty);
AppendWhiteSpace = new AppendSpaceCallback (base.AppendWhiteSpace);
AppendStructuralBreak = new AppendSpaceCallback (base.AppendStructuralBreak);
HotUp = new HotCallback (base.HotUp);
HotDown = new HotCallback (base.HotDown);
#if ENABLE_RDF_ADAPTER
AddLink = new AddLinkCallback (base.AddLink);
#endif
}
ignore_level = 0;
building_text = false;
builder = new StringBuilder ();
}
public void ExtractText (string html_string,
AppendTextCallback append_text_cb,
AddPropertyCallback add_prop_cb,
AppendSpaceCallback append_white_cb,
AppendSpaceCallback append_break_cb,
HotCallback hot_up_cb,
HotCallback hot_down_cb)
{
AppendText = append_text_cb;
AppendWord = append_text_cb;
AddProperty = add_prop_cb;
AppendWhiteSpace = append_white_cb;
AppendStructuralBreak = append_break_cb;
HotUp = hot_up_cb;
HotDown = hot_down_cb;
HtmlDocument doc = new HtmlDocument ();
doc.ReportNode += HandleNodeEvent;
doc.StreamMode = true;
try {
doc.LoadHtml (html_string);
} catch (Exception e) {
Log.Debug (e, "Exception while filtering html string [{0}]", html_string);
}
}