public void TestParseResultsBySentences()
{
PtTagset.TagDelimiter = '_';
TagsetBase[][] testOutList;
var testResult = PtTagset.TryParse(_largeTestFile, out testOutList);
Assert.IsTrue(testResult);
Assert.IsNotNull(testOutList);
Assert.AreNotEqual(0, testOutList.Length);
Assert.IsNotNull(testOutList[0]);
Assert.AreNotEqual(0, testOutList[0].Length);
Console.WriteLine("Number of sentences '{0}'", testOutList.Length);
for (var i = 0; i < testOutList.Length; i++)
{
Console.WriteLine("Sentence '{0}' length is '{1}'", i, testOutList[i].Length);
}
Assert.IsNotNull(testOutList[1]);
Assert.AreNotEqual(0, testOutList[1].Length);
Assert.IsInstanceOf(typeof(Determiner), testOutList[1][0]);
Assert.IsInstanceOf(typeof(Adjective), testOutList[1][1]);
var secondtolastValue = testOutList[36][0].Value;
var lastValue = testOutList[36][1].Value;
Console.WriteLine(secondtolastValue);
Console.WriteLine(lastValue);
}
public void TestParseResults()
{
const string TEST_INPUT = "This_DT is_VBZ a_DT sample_NN sentence_NN";
TagsetBase[] testOutList = null;
var testResult = PtTagset.TryParse(TEST_INPUT, out testOutList);
for (var i = 0; i < testOutList.Length; i++)
{
Console.WriteLine("{0,-8}{1}", i, testOutList[i].Value);
}
Assert.IsTrue(testResult);
Assert.AreEqual(5, testOutList.Count());
Assert.IsInstanceOf(typeof(Determiner), testOutList[0]);
Assert.AreEqual("This", testOutList[0].Value);
Assert.IsInstanceOf(typeof(Verb3rdPersonSingularPresent), testOutList[1]);
Assert.AreEqual("is", testOutList[1].Value);
Assert.IsInstanceOf(typeof(Determiner), testOutList[2]);
Assert.AreEqual("a", testOutList[2].Value);
Assert.IsInstanceOf(typeof(NounSingularOrMass), testOutList[3]);
Assert.AreEqual("sample", testOutList[3].Value);
Assert.IsInstanceOf(typeof(NounSingularOrMass), testOutList[4]);
Assert.AreEqual("sentence", testOutList[4].Value);
}
public void TestAllCodesRepresented()
{
foreach (var allCode in AllCodes)
{
Console.WriteLine(allCode);
Assert.IsNotNull(PtTagset.GetTagset(allCode));
}
}
public void TestTryParseJaggedArray()
{
TagsetBase[][] tagsOut;
var testInput = @"When_WRB in_IN the_DT Course_NNP of_IN human_JJ events_NNS ,_, it_PRP becomes_VBZ necessary_JJ for_IN one_CD people_NNS to_TO dissolve_VB the_DT political_JJ bands_NNS which_WDT have_VBP connected_VBN them_PRP with_IN another_DT ,_, and_CC to_TO assume_VB among_IN the_DT powers_NNS of_IN the_DT earth_NN ,_, the_DT separate_JJ and_CC equal_JJ station_NN to_TO which_WDT the_DT Laws_NNPS of_IN Nature_NNP and_CC of_IN Nature_NNP 's_POS God_NNP entitle_VB them_PRP ,_, a_DT decent_JJ respect_NN to_TO the_DT opinions_NNS of_IN mankind_NN requires_VBZ that_IN they_PRP should_MD declare_VB the_DT causes_NNS which_WDT impel_VBP them_PRP to_TO the_DT separation_NN ._.";
var testResult = PtTagset.TryParse(testInput, out tagsOut);
Assert.IsTrue(testResult);
Assert.IsNotNull(tagsOut);
Assert.AreNotEqual(0, tagsOut.Length);
}
public static TagsetBase[] ToPosTagsets(this string commonEnText)
{
var taggedText = ToTaggedString(commonEnText);
if (string.IsNullOrWhiteSpace(taggedText))
{
return(null);
}
TagsetBase[] tagsOut;
if (!PtTagset.TryParse(taggedText, out tagsOut))
{
return(null);
}
return(tagsOut);
}
