private void onContentUpdate(NewsPair p)
{
if (p == CurrentNewsPair)
{
OnNewsContentUpdate?.Invoke();
}
}
public void LoadPreviousNewsPair()
{
if (currentNewsPairIndex > 0 && currentNewsPairIndex < currentActivePairs.Count())
{
CurrentNewsPair = currentActivePairs.ElementAt(--currentNewsPairIndex);
}
OnPreviousNewsPair?.Invoke(currentNewsPairIndex > 0);
}
private string getOutputString(NewsPair pair)
{
string[] tmp = new string[pair.OriginalContent.Length];
for (int i = 0; i < tmp.Length; i++)
{
tmp[i] = pair.OriginalContent[i];
}
tmp[config.RelationLabelIndex] = pair.RelationLabel;
return(string.Join(config.SeparatorWrite, tmp));
}
private void writeNewsPairToFile()
{
if (currentActivePairs.Count() == backLimit)
{
NewsPair val = currentActivePairs.ElementAt(0);
handler.WriteToOutputFile(val);
currentNewsPairIndex--;
currentActivePairs.RemoveAt(0);
}
}
private void WriteToUncertainOutputFile(NewsPair pair)
{
if (uncertainOutputFileWriter == null)
{
uncertainOutputFileWriter = new StreamWriter(new FileStream(labelWork.UncertainOutputFileName, FileMode.Append, FileAccess.Write), Encoding.UTF8);
}
string val = getUncertainOutputString(pair);
uncertainOutputFileWriter.WriteLine(val);
uncertainOutputFileWriter.Flush();
}
public void WriteToOutputFile(NewsPair pair)
{
string val = getOutputString(pair);
outputFileWriter.WriteLine(val);
outputFileWriter.Flush();
if (pair.LabelUncertain)
{
WriteToUncertainOutputFile(pair);
}
}
private string getUncertainOutputString(NewsPair pair)
{
string[] tmp = new string[pair.OriginalContent.Length + 1];
for (int i = 0; i < pair.OriginalContent.Length; i++)
{
tmp[i] = pair.OriginalContent[i];
}
tmp[config.RelationLabelIndex] = pair.RelationLabel;
tmp[pair.OriginalContent.Length] = pair.LineNum.ToString();
return(string.Join(config.SeparatorWrite, tmp));
}
private bool addNextNewsPairToList()
{
NewsPair next = handler.ParseNextLine();
if (next == null)
{
return(true);
}
currentActivePairs.Add(next);
return(false);
}
private void startContentReadWorker(NewsPair pair)
{
if (pair == null)
{
return;
}
contentReadWorker = new BackgroundWorker();
contentReadWorker.WorkerReportsProgress = false;
contentReadWorker.WorkerSupportsCancellation = true;
contentReadWorker.DoWork += worker_DoWork;
contentReadWorker.RunWorkerCompleted += worker_RunWorkerCompleted;
contentReadWorker.RunWorkerAsync(pair);
}
public void InitActiveNewsPairList()
{
int i = 0;
while (i++ < backLimit && !addNextNewsPairToList())
{
;
}
if (currentActivePairs.Count() > 0)
{
CurrentNewsPair = currentActivePairs.ElementAt(currentNewsPairIndex);
OnNextNewsPair?.Invoke(currentNewsPairIndex > 0);
}
else
{
OnEndOfFileReached?.Invoke(currentActivePairs.Count() > 0);
}
}
public void DestroyNewsFileHandler()
{
if (handler == null)
{
return;
}
int end = isAtEOF ? currentActivePairs.Count() : Math.Min(currentActivePairs.Count() - 1, currentNewsPairIndex);
for (int i = 0; i < end; i++)
{
handler.WriteToOutputFile(currentActivePairs.ElementAt(i));
}
handler.Destroy();
currentActivePairs.Clear();
currentNewsPairIndex = 0;
handler = null;
CurrentNewsPair = null;
OnNewsFileHandlerDestroy?.Invoke();
}
public void LoadNextNewsPair()
{
isAtEOF = false;
if (currentNewsPairIndex >= currentActivePairs.Count() - 1)
{
writeNewsPairToFile();
isAtEOF = addNextNewsPairToList();
}
if (!isAtEOF)
{
currentNewsPairIndex++;
CurrentNewsPair = currentActivePairs.ElementAt(currentNewsPairIndex);
OnNextNewsPair?.Invoke(currentNewsPairIndex > 0);
}
else
{
OnEndOfFileReached?.Invoke(currentActivePairs.Count() > 0);
}
}
public NewsPair ParseNextLine()
{
string line = inputFileReader.ReadLine();
if (line == null)
{
return(null);
}
CurrentLineNum++;
string[] split = line.Split(config.SeparatorRead, StringSplitOptions.None);
NewsPair rv = null;
try
{
if (split.Length >= config.NumElementFullSourceFile)
{
rv = new NewsPair(CurrentLineNum, split, split[config.News1TitleIndex],
split[config.News1ContentIndex],
split[config.News2TitleIndex],
split[config.News2ContentIndex],
(int)double.Parse(split[config.CommonTitleWordsIndex])
);
}
else if (split.Length >= config.NumElementSimplifiedSourceFile)
{
rv = new NewsPair(CurrentLineNum, split, split[config.News1TitleIndex],
LOADING,
split[config.News2TitleIndex],
LOADING,
(int)double.Parse(split[config.CommonTitleWordsIndex])
);
contentReadQ.Enqueue(rv);
notifyContentReadWorker();
}
}
catch (Exception e)
{
Console.WriteLine(e.StackTrace);
}
return(rv);
}
private void worker_DoWork(object sender, DoWorkEventArgs e)
{
BackgroundWorker worker = sender as BackgroundWorker;
NewsPair pair = e.Argument as NewsPair;
if ((worker.CancellationPending == true))
{
e.Cancel = true;
return;
}
string[] rv = new string[config.NumElementNewsContentFile];
for (int i = 0; i < rv.Length; i++)
{
rv[i] = "(无内容)";
}
lock (sharedWorkerResourceLock)
{
if (newsFileReader == null)
{
setNewsContentForPair(pair, rv);
return;
}
string line = newsFileReader.ReadLine();
if (line == null)
{
setNewsContentForPair(pair, rv);
return;
}
CurrentContentFileLineNum++;
string[] split = line.Split(config.SeparatorRead, StringSplitOptions.None);
if (split.Length != config.NumElementNewsContentFile)
{
setNewsContentForPair(pair, rv);
return;
}
setNewsContentForPair(pair, split);
}
e.Result = pair;
}
private void setNewsContentForPair(NewsPair pair, string[] contentArray)
{
pair.News1Content = contentArray[config.News1ContentIndexInContentFile];
pair.News2Content = contentArray[config.News2ContentIndexInContentFile];
}
