private static void CheckOnCriteria(double scienceFile, double funFile, double poemFile, double file, ParserFileInfo fileClass)
{
var sumCriteria = Math.Abs(scienceFile - file) +
Math.Abs(funFile - file) +
Math.Abs(poemFile - file);
CheckFile data = new CheckFile();
if (sumCriteria == 0)
{
sumCriteria = 1;
}
data.ScienceClass = (sumCriteria - Math.Abs(scienceFile - file)) / sumCriteria * 100;
data.FunClass = (sumCriteria - Math.Abs(funFile - file)) / sumCriteria * 100;
data.PoemClass = (sumCriteria - Math.Abs(poemFile - file)) / sumCriteria * 100;
fileClass.checkData.Add(data);
}
public void LoadFile(string Type, bool directory = false)
{
OpenFileDialog openFileDialog = new OpenFileDialog();
openFileDialog.Title = "Выберите имя";
string path = string.Empty;
if (Type != "Проверка")
{
foreach (var item in _info)
{
if (Type == item.Type)
{
_info.Remove(item);
break;
}
}
}
else
{
int count = 0;
foreach (var item in _info)
{
switch (item.Type)
{
case "Эталон Научной":
count++;
break;
case "Эталон художественного":
count++;
break;
case "Эталон стихотворения":
count++;
break;
}
}
if (count != 3)
{
MessageBox.Show("Загрузите все 3 эталона");
return;
}
}
if (openFileDialog.ShowDialog() == true)
{
path = openFileDialog.FileName;
}
string[] files = { path };
if (directory)
{
files = Directory.GetFiles(Path.GetDirectoryName(path));
}
if (string.IsNullOrEmpty(path))
{
return;
}
foreach (var file in files)
{
var fileInfo = new ParserFileInfo();
fileInfo.Type = Type;
fileInfo.FileName = file.Split('\\').Last();
if (Type == "Проверка")
{
fileInfo.NeedCheck = true;
}
LoadFile(file, fileInfo);
fileInfo.CorrelationFirstCriteria = Math.Round(FirstCriteriaWork(fileInfo.CriteriaFirst), 4);
if (double.IsNaN(fileInfo.CorrelationFirstCriteria))
{
fileInfo.CorrelationFirstCriteria = 0;
}
fileInfo.CorrelationSecondCriteria = Math.Round(FirstCriteriaWork(fileInfo.CriteriaSecond), 4);
if (double.IsNaN(fileInfo.CorrelationSecondCriteria))
{
fileInfo.CorrelationSecondCriteria = 0;
}
Infos.Add(fileInfo);
}
}
private static void LoadFile(string path, ParserFileInfo fileInfo)
{
var text = File.ReadAllText(path, Encoding.UTF8);
for (int i = 0; i < 10; i++)
{
text = text.Replace("\r\n\r\n\r\n", "\r\n\r\n");
}
var pattern = "\r\n\r\n";
var paragraphSplit = System.Text.RegularExpressions.Regex.Split(text, pattern);
var paragraphWords = 0;
fileInfo.WordCount = text.Split(' ').Where(q => q.Length > 2).ToArray().Length;
fileInfo.AverageWordLength = fileInfo.WordCount / text.Length;
Dictionary <string, int> wordsLanguages = new Dictionary <string, int>();
foreach (var word in text.Split(' ').Where(q => q.Length > 2).ToArray())
{
var language = LanguagetText(word);
if (wordsLanguages.ContainsKey(language))
{
wordsLanguages[language]++;
}
else
{
wordsLanguages.Add(language, 1);
}
}
var mainLanguage = wordsLanguages.First(q => q.Value == wordsLanguages.Max(l => l.Value));
fileInfo.ForeignWordsCount = wordsLanguages.Where(q => q.Key != mainLanguage.Key).Sum(q => q.Value);
foreach (var paragparh in paragraphSplit)
{
var line = paragparh.Split(' ');
paragraphWords = line.Where(q => q.Length > 2).ToArray().Length;
var sentece = paragparh.Split(new char[] { '.', '!', '?' });
var sentenceWords = paragraphWords / (double)sentece.Length;
fileInfo.CriteriaFirst.Add(new CriteriaData()
{
Words = paragraphWords,
SenteceWord = sentenceWords,
});
var newLineCount = System.Text.RegularExpressions.Regex.Split(paragparh, "\r\n");
foreach (var newLine in newLineCount)
{
var words = newLine.Split(' ');
var LineSenteces = newLine.Split(new char[] { '.', '!', '?' });
fileInfo.CriteriaSecond.Add(new CriteriaData()
{
Words = words.Where(q => q.Length > 2).ToArray().Length,
SenteceWord = LineSenteces.Length,
});
}
}
fileInfo.ParagraphWordsCount = fileInfo.WordCount / paragraphSplit.Length;
for (int i = 0; i < text.Length; i++)
{
var letter = text[i];
if (letter == '.' || letter == '!' || letter == '?')
{
fileInfo.SentenceCount++;
}
if (letter == '\n')
{
if (i + 2 < text.Length && text[i + 2] == '\n')
{
fileInfo.ParagraphCount++;
while (i + 2 < text.Length && text[i + 2] == '\n')
{
i += 2;
}
}
else
{
fileInfo.LineBreaksCount++;
}
}
if (letter == ',')
{
fileInfo.CommasCount++;
}
if (letter == '–' && text[i - 2] == '\n')
{
fileInfo.DirectSpeechCount++;
}
if (letter == '\"' || letter == '«')
{
fileInfo.QuotesCount++;
}
}
}
