private static void TestSentences(JavaSDME javaMe, SharpSDME sharpME, string data) {
var jSentences = javaMe.sentPosDetect(data);
var sSentences = sharpME.SentPosDetect(data);
var jProb = javaMe.getSentenceProbabilities();
var sProb = sharpME.GetSentenceProbabilities();
Assert.AreEqual(jSentences.Length, sSentences.Length);
Assert.AreEqual(jProb.Length, sProb.Length);
for (int i = 0; i < jSentences.Length; i++) {
var a = jSentences[i].getCoveredText(data).toString(); // CharSequence -> string
var b = sSentences[i].GetCoveredText(data);
Assert.AreEqual(jSentences[i].getStart(), sSentences[i].Start);
Assert.AreEqual(jSentences[i].getEnd(), sSentences[i].End);
Assert.AreEqual(a, b);
Assert.AreEqual(jProb[i], sProb[i]);
}
}
private static void TestSentences(JavaSDME javaMe, SharpSDME sharpME, string data)
{
var jSentences = javaMe.sentPosDetect(data);
var sSentences = sharpME.SentPosDetect(data);
var jProb = javaMe.getSentenceProbabilities();
var sProb = sharpME.GetSentenceProbabilities();
Assert.AreEqual(jSentences.Length, sSentences.Length);
Assert.AreEqual(jProb.Length, sProb.Length);
for (var i = 0; i < jSentences.Length; i++)
{
var a = jSentences[i].getCoveredText(data).toString(); // CharSequence -> string
var b = sSentences[i].GetCoveredText(data);
Assert.AreEqual(jSentences[i].getStart(), sSentences[i].Start);
Assert.AreEqual(jSentences[i].getEnd(), sSentences[i].End);
Assert.AreEqual(a, b);
Assert.AreEqual(jProb[i], sProb[i]);
}
}
internal static void EvalSentences(SentenceDetectorME sentDetect) {
const string sampleSentences1 = "This is a test. There are many tests, this is the second.";
var sents = sentDetect.SentDetect(sampleSentences1);
Assert.AreEqual(sents.Length, 2);
Assert.AreEqual(sents[0], "This is a test.");
Assert.AreEqual(sents[1], "There are many tests, this is the second.");
var probs = sentDetect.GetSentenceProbabilities();
Assert.AreEqual(probs.Length, 2);
const string sampleSentences2 = "This is a test. There are many tests, this is the second";
sents = sentDetect.SentDetect(sampleSentences2);
Assert.AreEqual(sents.Length, 2);
probs = sentDetect.GetSentenceProbabilities();
Assert.AreEqual(probs.Length, 2);
Assert.AreEqual(sents[0], "This is a test.");
Assert.AreEqual(sents[1], "There are many tests, this is the second");
const string sampleSentences3 = "This is a \"test\". He said \"There are many tests, this is the second.\"";
sents = sentDetect.SentDetect(sampleSentences3);
Assert.AreEqual(sents.Length, 2);
probs = sentDetect.GetSentenceProbabilities();
Assert.AreEqual(probs.Length, 2);
Assert.AreEqual(sents[0], "This is a \"test\".");
Assert.AreEqual(sents[1], "He said \"There are many tests, this is the second.\"");
const string sampleSentences4 = "This is a \"test\". I said \"This is a test.\" Any questions?";
sents = sentDetect.SentDetect(sampleSentences4);
Assert.AreEqual(sents.Length, 3);
probs = sentDetect.GetSentenceProbabilities();
Assert.AreEqual(probs.Length, 3);
Assert.AreEqual(sents[0], "This is a \"test\".");
Assert.AreEqual(sents[1], "I said \"This is a test.\"");
Assert.AreEqual(sents[2], "Any questions?");
const string sampleSentences5 = "This is a one sentence test space at the end. ";
sents = sentDetect.SentDetect(sampleSentences5);
Assert.AreEqual(1, sentDetect.GetSentenceProbabilities().Length);
Assert.AreEqual(sents[0], "This is a one sentence test space at the end.");
const string sampleSentences6 = "This is a one sentences test with tab at the end. ";
sents = sentDetect.SentDetect(sampleSentences6);
Assert.AreEqual(sents[0], "This is a one sentences test with tab at the end.");
const string sampleSentences7 = "This is a test. With spaces between the two sentences.";
sents = sentDetect.SentDetect(sampleSentences7);
Assert.AreEqual(sents[0], "This is a test.");
Assert.AreEqual(sents[1], "With spaces between the two sentences.");
const string sampleSentences9 = "";
sents = sentDetect.SentDetect(sampleSentences9);
Assert.AreEqual(0, sents.Length);
const string sampleSentences10 = " "; // whitespaces and tabs
sents = sentDetect.SentDetect(sampleSentences10);
Assert.AreEqual(0, sents.Length);
const string sampleSentences11 = "This is test sentence without a dot at the end and spaces ";
sents = sentDetect.SentDetect(sampleSentences11);
Assert.AreEqual(sents[0], "This is test sentence without a dot at the end and spaces");
probs = sentDetect.GetSentenceProbabilities();
Assert.AreEqual(1, probs.Length);
const string sampleSentence12 = " This is a test.";
sents = sentDetect.SentDetect(sampleSentence12);
Assert.AreEqual(sents[0], "This is a test.");
const string sampleSentence13 = " This is a test";
sents = sentDetect.SentDetect(sampleSentence13);
Assert.AreEqual(sents[0], "This is a test");
// Test that sentPosDetect also works
var pos = sentDetect.SentPosDetect(sampleSentences2);
Assert.AreEqual(pos.Length, 2);
probs = sentDetect.GetSentenceProbabilities();
Assert.AreEqual(probs.Length, 2);
Assert.AreEqual(new Span(0, 15), pos[0]);
Assert.AreEqual(new Span(16, 56), pos[1]);
}
