private async Task PerformSplit()
{
try
{
Trace.Indent();
var maxWordLength = int.Parse(IFdata.GetField("maxWordLength"));
var probRatio = double.Parse(IFdata.GetField("probRatio"));
var bemsRatio = double.Parse(IFdata.GetField("bemsRatio"));
CHBsplit.Enabled = false;
string fileName = "output.txt";
var encoding = Encoding.UTF8;
using (var writer = new StreamWriter(fileName, false, encoding))
{
TXBout.Clear();
var d = new SentenceSplitter.WordIdentifiedEventHandler((word) =>
{
writer.WriteLine(word);
TXBout.Invoke(new Action(() =>
{
if (TXBout.TextLength < 10000)
{
TXBout.AppendText($"{word}\r\n");
if (TXBout.TextLength >= 10000)
{
TXBout.AppendText("......(Cut)\r\n");
}
}
}));
});
var ss = new SentenceSplitter(trie, baseDataLength);
try
{
ss.WordIdentified += d;
Trace.WriteLine("Splitting...");
var mainInputs = string.IsNullOrWhiteSpace(TXBdata.Text) ? (txbDataFileContent != null ? txbDataFileContent : data) : TXBdata.Text;
var inputs = mainInputs.Split(' ', '\r', '\n', '\t');
long cnt = 0;
foreach (var input in inputs)
{
cnt += (await ss.SplitAsync(
input,
maxWordLength,
probType,
true)).Count;
}
Trace.WriteLine($"{cnt} words identified.");
}
catch (Exception error) { TXBout.Text = error.ToString(); }
finally { ss.WordIdentified -= d; }
writer.Close();
}
}
catch (Exception error) { TXBout.Text = error.ToString(); }
finally { Trace.Unindent(); CHBsplit.CheckState = CheckState.Indeterminate; CHBsplit.Enabled = true; }
}
private async void BTNexportList_Click(object sender, EventArgs e)
{
try
{
Trace.Indent();
int threshold = int.Parse(Microsoft.VisualBasic.Interaction.InputBox("Threshold?", "", "25"));
List <Tuple <int, int> > s = new List <Tuple <int, int> >();
Trace.WriteLine("Searching...");
try
{
Trace.Indent();
using (StreamWriter writer = new StreamWriter("output.txt", false, Encoding.UTF8))
{
TXBout.Clear();
long progress = 0;
var lastUpdateTime = DateTime.Now;
await sa.ListFrequentWords(threshold, new Func <string, Task>(async(str) =>
{
++progress;
var count = sa.UpperBound(str) - sa.LowerBound(str);
await writer.WriteLineAsync($"{str},{count}");
if (TXBout.TextLength < 100000)
{
TXBout.AppendText($"{str}\t{count}\r\n");
if (TXBout.TextLength >= 100000)
{
TXBout.AppendText("......(Cut)");
}
}
if ((DateTime.Now - lastUpdateTime).TotalSeconds > 0.5)
{
Trace.WriteLine($"{progress} words listed, Ex:{str}\t{count}");
lastUpdateTime = DateTime.Now;
}
}));
}
}
finally { Trace.Unindent(); }
Trace.Write("Done");
}
catch (Exception error)
{
TXBout.Text = error.ToString();
}
finally { Trace.Unindent(); }
}
private async Task PerformSplit()
{
try
{
Trace.Indent();
CHBsplit.Enabled = false;
string fileName = "output.txt";
var encoding = Encoding.UTF8;
using (var writer = new StreamWriter(fileName, false, encoding))
{
TXBout.Clear();
var d = new SentenceSplitter.WordIdentifiedEventHandler((word) =>
{
writer.WriteLine(word);
TXBout.Invoke(new Action(() =>
{
if (TXBout.TextLength < 10000)
{
TXBout.AppendText($"{word}\r\n");
if (TXBout.TextLength >= 10000)
{
TXBout.AppendText("......(Cut)\r\n");
}
}
}));
});
try
{
ss.WordIdentified += d;
Trace.WriteLine("Splitting...");
var ans = await ss.SplitAsync(
string.IsNullOrWhiteSpace(TXBdata.Text)?(txbDataFileContent != null ? txbDataFileContent : sa.S) : TXBdata.Text,
maxWordLength,
probType,
CHBverbose.Checked);
Trace.WriteLine($"{ans.Count} words identified.");
}
catch (Exception error) { TXBout.Text = error.ToString(); }
finally { ss.WordIdentified -= d; }
writer.Close();
}
}
catch (Exception error) { TXBout.Text = error.ToString(); }
finally { Trace.Unindent(); CHBsplit.CheckState = CheckState.Indeterminate; CHBsplit.Enabled = true; }
}