public static void GenerateTripletFlowNt2(string fileName, long limit, sema2012m.Engine engine, sema2012m.LogLine turlog)
{
StreamReader sr = new StreamReader(fileName);
// Начало транзакции ввода
engine.InitAdapterBuffers();
//engine.IncludeXElement(xelement);
XElement xelement = null;
string current_subject = null;
long cnt = 0;
string line = null;
bool skip = false;
while ((line = sr.ReadLine()) != null)
{
if (cnt > limit) break; cnt++;
if (cnt % 1000000 == 0) turlog("Загружено: " + cnt + " линий.");
if (string.IsNullOrWhiteSpace(line) || line[0] == '@') continue;
string subject;
int position = GetEntityOrLiteral(line, 0, out subject);
if (subject == null) continue;
string predicate;
position = GetEntityOrLiteral(line, position, out predicate);
if (predicate == null) continue;
string obj;
bool isObjectProperty = line[position] == '\"' ? false : true;
position = GetEntityOrLiteral(line, position, out obj);
if (obj == null) continue;
if (xelement == null || subject != current_subject)
{
if (!skip && xelement != null) engine.IncludeXElement(xelement);
xelement = new XElement(sema2012m.ONames.rdfdescription, new XAttribute(sema2012m.ONames.rdfabout, subject));
current_subject = subject;
}
if (xelement.Elements().Count() > 100) continue;
if (isObjectProperty)
{
xelement.Add(new XElement(sema2012m.ONames.GetXName(predicate), new XAttribute(sema2012m.ONames.rdfresource, obj)));
}
else // DatatypeProperty
{
xelement.Add(new XElement(sema2012m.ONames.GetXName(predicate), obj)); // пока без языкового спецификатора
}
}
if (!skip && xelement != null) engine.IncludeXElement(xelement);
// Конец транзакции ввода
engine.FlushAdapterBuffers();
}
