public TextGenerator(WordTypes wordType = WordTypes.Word)
{
if (wordType == WordTypes.Name)
this.corpus = Corpus.DeserializeFromEmbeddedResource("names.bin");
else
this.corpus = Corpus.DeserializeFromEmbeddedResource("text.bin");
this.wordType = wordType;
}
/// <summary>
/// Initializes a new instance of the <see cref="TextGenerator"/> class.
/// Will generate text based on provided corpus. Use this to generate fake text based on different languages.
/// </summary>
/// <param name="corpus">The corpus.</param>
public TextGenerator(Corpus corpus)
{
if (corpus == null)
throw new ArgumentException("Corpus is not provided.");
this.corpus = corpus;
}
/// <summary>
/// Creates the corpus from text.
/// Text should be long enough to generate decent results (e.g. 10Kb+).
/// </summary>
/// <param name="text">The text.</param>
/// <returns></returns>
public static Corpus CreateFromText(string text)
{
Dictionary<LetterStats, LetterStats> letterData = new Dictionary<LetterStats, LetterStats>();
int charCount = 0;
// clean text
string clean = Regex.Replace(text, "[^\\p{L} \t\n]", string.Empty).ToLower();
string[] words = clean.Split(new char[] { ' ', '\n', '\t' }, StringSplitOptions.RemoveEmptyEntries);
for (int i = 0; i < words.Length; i++)
{
char prevLetter = '\0';
for (int l = 0; l < words[i].Length; l++)
{
LetterStats ls = new LetterStats(words[i][l], prevLetter, l);
if (letterData.ContainsKey(ls))
letterData[ls].IncrementCount();
else
letterData.Add(ls, ls);
prevLetter = words[i][l];
charCount++;
}
}
Corpus c = new Corpus() { CharCount = charCount, WordCount = words.Length };
c.letters.AddRange(letterData.Keys);
return c;
}
