public static void Main()
{
var generator = new TextGenerator(WordTypes.Name);
var names = generator.GenerateText(1000);
var trie = new Trie();
var words = new HashSet<string>();
names.Split(' ').ToList().ForEach(
x =>
{
words.Add(x);
trie.AddWord(x);
});
var result = new StringBuilder();
foreach (var word in words.OrderBy(x => x))
{
int occurenceCount;
trie.TryFindWord(word, out occurenceCount);
result.AppendFormat("{0} -> {1} times", word, occurenceCount).AppendLine();
}
Console.WriteLine(result);
}
public void CreateNonEnglishTextTest()
{
string s = File.ReadAllText(@"Corpuses\rus.txt", Encoding.UTF8);
Corpus c = Corpus.CreateFromText(s);
TextGenerator wc = new TextGenerator(c);
string text = wc.GenerateText(200);
text.Split(' ').Length.Should().Be(200);
text.Should().EndWith(".");
}
public void MakeWords()
{
TextGenerator wc = new TextGenerator();
List<string> wordsList = new List<string>(20);
for (int i = 0; i < 20; i++)
{
wordsList.Add(wc.GenerateWord(6));
}
string words = string.Join(", ", wordsList);
}
public void MakeNames()
{
TextGenerator wc = new TextGenerator(WordTypes.Name);
List<string> namesList = new List<string>(20);
for (int i = 0; i < 20; i++)
{
namesList.Add(wc.GenerateWord(6));
}
string names = string.Join(", ", namesList);
}
public void MakeRussianAndSpanishText()
{
string s = File.ReadAllText(@"Corpuses\rus.txt", Encoding.UTF8);
Corpus c = Corpus.CreateFromText(s);
// Corpus can be serialized for later usage; the resulting file can be embedded as a resource; see
//c.SerializeToFile(string) and Corpus.DeserializeFromEmbeddedResource() methods.
TextGenerator wc = new TextGenerator(c);
string russian = wc.GenerateText(20);
s = File.ReadAllText(@"Corpuses\spanish.txt", Encoding.UTF8);
c = Corpus.CreateFromText(s);
wc = new TextGenerator(c);
string spanish = wc.GenerateText(20);
}
public void WordCreator_ShouldNotGenerateWords_ThatHave3SameAdjacentLetters()
{
TextGenerator wc = new TextGenerator();
string text = wc.GenerateText(10000);
string[] split = text.Split(' ', '.');
split.Any(x => Regex.Match(x, "(.)\\1{2,}").Groups.Count > 1).Should().BeFalse();
}
public void WordCreator_ShouldMakeWords()
{
TextGenerator wc = new TextGenerator();
wc.GenerateWord(6).Should().NotBeNullOrEmpty().And.HaveLength(6);
}
public void WordCreator_ShouldMakeNames()
{
TextGenerator wc = new TextGenerator(WordTypes.Name);
for (int i = 0; i < 100; i++)
{
string name = wc.GenerateWord(6);
name.Should().NotBeNullOrEmpty().And.HaveLength(6);
name[0].Should().Be(char.ToUpper(name[0]));
}
}
public void WordCreator_ShouldMakeLongWords()
{
TextGenerator wc = new TextGenerator();
wc.GenerateWord(100).Should().NotBeNullOrEmpty();
string s = wc.GenerateWord(100);
}
public void WordCreatorConstructors_ShouldNotTrowExceptions()
{
TextGenerator wc = new TextGenerator();
wc = new TextGenerator(WordTypes.Name);
Corpus c = Corpus.DeserializeFromEmbeddedResource("names.bin");
wc = new TextGenerator(c);
}
public void Text_ShouldNotEndWithDoubleDots()
{
TextGenerator wc = new TextGenerator();
string text = wc.GenerateText(50000);
text.Should().NotContain("..");
}
public void TextGenerator_ShouldTakeLessThan_500ms_ToGenerate_10000Words()
{
TextGenerator wc = new TextGenerator();
Stopwatch sw = Stopwatch.StartNew();
for (int i = 0; i < 4; i++)
{
string text = wc.GenerateText(10000);
}
Debug.WriteLine("Total elapsed: " + sw.ElapsedMilliseconds / 4 + " ms");
(sw.ElapsedMilliseconds / 4).Should().BeLessOrEqualTo(500);
}
public void IWant_TextGenerator_ToTakeLessThan_500ms_ToGenerate_1000Words()
{
TextGenerator wc = new TextGenerator();
Stopwatch sw = Stopwatch.StartNew();
string text = wc.GenerateText(1000);
Debug.WriteLine("Total elapsed: " + sw.ElapsedMilliseconds + " ms");
sw.ElapsedMilliseconds.Should().BeLessOrEqualTo(500);
}
public void MakeText()
{
TextGenerator wc = new TextGenerator();
string text = wc.GenerateText(20);
}
public void CreateTextTest()
{
TextGenerator wc = new TextGenerator();
string text = wc.GenerateText(200);
text.Split(' ').Length.Should().Be(200);
text.Should().EndWith(".");
}
public void CreateText_ShouldWorkOffSmallCorpus()
{
string s = File.ReadAllText(@"Corpuses\short.txt", Encoding.UTF8);
Corpus c = Corpus.CreateFromText(s);
TextGenerator wc = new TextGenerator(c);
string text = wc.GenerateText(2000);
text.Split(' ').Length.Should().Be(2000);
text.Should().EndWith(".");
}
public void Text_ShouldNotHaveSentences_LongerThan_MaxSentenceLength()
{
TextGenerator wc = new TextGenerator();
string text = wc.GenerateText(10000);
string[] split = text.Split('.');
split.Count(x => x.Split(' ').Length > wc.MaxSentenceLength).Should().Be(0);
}
