public static void FixWordTest()
{
string[] words = {"Проверка", "праверка"};
var detector = new SpellChecker(TrigramIndex.CreateFromDefaultDictionaryAnd(Program.DefaultNotStemmedQuestionList));
var fixedWords = words.Select(detector.Fix).ToArray();
Console.WriteLine(String.Join("\n",
words.Zip(fixedWords, (w1, w2) => ((w1 != w2) ? "!!!\t" : "") + w1 + "\t-\t" + w2)));
}
public static void TestCreation()
{
var detector = new SpellChecker(TrigramIndex.CreateFromDefaultDictionaryAnd(Program.DefaultNotStemmedQuestionList));
Console.WriteLine("Now we can fix:");
var words =
Program.DefaultNotStemmedQuestionList.GetAllQuestions().Take(10).SelectMany(
item => item.WholeText.SplitInWordsAndStripHTML()).ToArray();
var start = DateTime.Now;
var fixedWords = words.Select(detector.Fix).ToArray();
Console.WriteLine("For {0} sec to {1} words:\n{2}", (DateTime.Now - start).TotalSeconds, fixedWords.Length,
String.Join("\n",
words.Zip(fixedWords, (w1, w2) => ((w1 != w2) ? "!!!\t" : "") + w1 + "\t-\t" + w2)));
}
public static void ModifyTyposCorpus(QuestionList ql)
{
var detector = new SpellChecker(TrigramIndex.CreateFrom(ql));
Console.WriteLine("I am Modifying");
var start = DateTime.Now;
foreach (var question in ql.GetAllQuestions())
{
question.Text = String.Join(" ", question.Text.SplitInWordsAndStripHTML().Select(detector.Fix));
question.Title = String.Join(" ", question.Title.SplitInWordsAndStripHTML().Select(detector.Fix));
}
Console.WriteLine("Questions modified in {0}", (DateTime.Now - start).TotalSeconds);
start = DateTime.Now;
foreach (var answer in ql.GetAllAnswers())
{
answer.Text = String.Join(" ", answer.Text.SplitInWordsAndStripHTML().Select(detector.Fix));
}
Console.WriteLine("Answers modified in {0}", (DateTime.Now - start).TotalSeconds);
File.WriteAllLines(Program.QuestionsNoTyposFileName, ql.GetAllQuestions().Select(Question.FormatStringWrite));
File.WriteAllLines(Program.AnswersNoTyposFileName, ql.GetAllAnswers().Select(Answer.FormatStringWrite));
}