public void testDetector3()
{
Detector detect = DetectorFactory.Create();
detect.Append("d e");
Assert.AreEqual(detect.Detect(), "en");
}
public void testDetector4()
{
Detector detect = DetectorFactory.Create();
detect.Append("\u3042\u3042\u3042\u3042a");
Assert.AreEqual(detect.Detect(), "ja");
}
public void testDetector2()
{
Detector detect = DetectorFactory.Create();
detect.Append("b d");
Assert.AreEqual(detect.Detect(), "fr");
}
/// <summary>
/// Language detection test for each file (--detectlang option)
/// <pre>
/// usage: --detectlang -d [profile directory] -a [alpha] -s [seed] [test file(s)]
/// </pre>
public void DetectLang()
{
if (LoadProfile())
{
return;
}
foreach (string filename in arglist)
{
StreamReader strm = null;
try
{
strm = new StreamReader(File.OpenRead(filename));
Detector detector = DetectorFactory.Create(GetDouble("alpha", DEFAULT_ALPHA));
if (HasOpt("--debug"))
{
detector.SetVerbose();
}
detector.Append(strm);
var probs = string.Join(" ", detector.GetProbabilities().Select((lang) => lang.ToString()));
Console.WriteLine("{0}: {1}", filename, probs);
}
catch (IOException e)
{
Debug.WriteLine(e);
}
catch (LangDetectException e)
{
Debug.WriteLine(e);
}
finally
{
try
{
if (strm != null)
{
strm.Close();
}
}
catch (IOException e) { }
}
}
}
/// <summary>
/// Batch Test of Language Detection (--batchtest option)
/// <pre>
/// usage: --batchtest -d [profile directory] -a [alpha] -s [seed] [test data(s)]
/// </pre>
/// The format of test data(s):
/// <pre>
/// [correct language name]\t[text body for test]\n
/// </pre>
public void BatchTest()
{
if (LoadProfile())
{
return;
}
Dictionary <string, List <string> > result = new Dictionary <string, List <string> >();
foreach (string filename in arglist)
{
StreamReader strm = null;
try
{
strm = new StreamReader(File.OpenRead(filename));
while (!strm.EndOfStream)
{
string line = strm.ReadLine();
int idx = line.IndexOf('\t');
if (idx <= 0)
{
continue;
}
string correctLang = line.Substring(0, idx);
string text = line.Substring(idx + 1);
Detector detector = DetectorFactory.Create(GetDouble("alpha", DEFAULT_ALPHA));
detector.Append(text);
string lang = "";
try
{
lang = detector.Detect();
}
catch (Exception e)
{
Debug.WriteLine(e);
}
if (!result.ContainsKey(correctLang))
{
result[correctLang] = new List <string>();
}
result[correctLang].Add(lang);
if (HasOpt("--debug"))
{
Console.WriteLine(correctLang + "," + lang + "," + (text.Length > 100 ? text.Substring(0, 100) : text));
}
}
}
catch (IOException e)
{
Debug.WriteLine(e);
}
catch (LangDetectException e)
{
Debug.WriteLine(e);
}
finally
{
try
{
if (strm != null)
{
strm.Close();
}
}
catch (IOException e) { }
}
List <string> langlist = new List <string>(result.Keys);
langlist.Sort();
int totalCount = 0, totalCorrect = 0;
foreach (string lang in langlist)
{
Dictionary <string, int> resultCount = new Dictionary <string, int>();
int count = 0;
List <string> list = result[lang];
foreach (string detectedLang in list)
{
++count;
if (resultCount.ContainsKey(detectedLang))
{
resultCount[detectedLang] = resultCount[detectedLang] + 1;
}
else
{
resultCount[detectedLang] = 1;
}
}
int correct = resultCount.ContainsKey(lang) ? resultCount[lang] : 0;
double rate = correct / (double)count;
Console.WriteLine(string.Format("%s (%d/%d=%.2f): %s", lang, correct, count, rate, resultCount));
totalCorrect += correct;
totalCount += count;
}
Console.WriteLine(string.Format("total: %d/%d = %.3f", totalCorrect, totalCount, totalCorrect / (double)totalCount));
}
}
