/// <summary>
/// Counts the words in the text files we have for the currently selected language.
/// </summary>
public void CountWords()
{
var inputDirectory = new DirectoryInfo(CommonFiles.DocsPath(_languageId));
var inputFiles = inputDirectory.GetFiles("*.txt", SearchOption.TopDirectoryOnly);
string outputFilePath = Path.Combine(CommonFiles.AllWordsSorted(_languageId));
var wordCounter = new WordCounter();
wordCounter.CountWords(inputFiles, outputFilePath, _ignoreCase);
// Reload words here?
//InitializeLanguage();
}
/// <summary>
/// Counts n-grams in the text files for the selected language.
/// </summary>
private void CountNGrams()
{
try
{
string lang = _languageId;
var indexer =
new NGramIndexer(CommonFiles.Words(lang), CommonFiles.WordCounts(lang), CommonFiles.NGrams(lang), CommonFiles.NGramCounts(lang), CommonFiles.NGramsByWord(lang), CommonFiles.WordsByNGram(lang))
{
Order = NGramOrder
};
var counter = new WordCounter();
var inputDirectory = new DirectoryInfo(CommonFiles.DocsPath(lang));
var inputFiles = inputDirectory.GetFiles("*.txt", SearchOption.TopDirectoryOnly).Take(NGramMaxFiles).ToList();
var knownWords = new HashSet<string>(_translator.GetKnownWords());
counter.CountNGrams(indexer, inputFiles, knownWords, _ignoreCase);
}
catch (IOException ex)
{
Debug.WriteLine(ex.Message + ex.StackTrace);
return;
}
}