private void LoadWordFrequencies()
{
foreach (XmlNode book in BibleXml.DocumentElement.ChildNodes)
{
var bookName = book.Attributes["n"].InnerText;
foreach (XmlNode chapter in book.ChildNodes)
{
var chapterName = chapter.Attributes["n"].InnerText;
foreach (XmlNode verse in chapter.ChildNodes)
{
TotalVerseCount++;
var verseNumber = verse.Attributes["n"].InnerText;
string text = verse.InnerText.ToUpperInvariant();
var chars = new List <char>();
for (int i = 0; i < text.Length; i++)
{
var c = text[i];
if (!char.IsLetter(c) && c != ' ' && c != '-')
{
continue;
}
// allow hyphenated words
if (i > 0 && i < text.Length - 1 && c == '-')
{
if (!char.IsLetter(text[i - 1]) || !char.IsLetter(text[i + 1]))
{
continue;
}
}
chars.Add(text[i]);
}
text = string.Concat(chars).ToUpperInvariant();
var words = text.Split(' ', StringSplitOptions.RemoveEmptyEntries);
foreach (var word in words)
{
var tempWord = word.Trim().TrimStart('-').TrimEnd('-');
if (string.IsNullOrWhiteSpace(tempWord))
{
continue;
}
TotalWordCount++;
if (WordFrequencies.ContainsKey(tempWord))
{
WordFrequencies[tempWord]++;
}
else
{
WordFrequencies[tempWord] = 1;
}
UniqueWords.Add(tempWord);
}
}
}
}
}
public static void Test(){
UniqueWords u = new UniqueWords();
string[] arr = {"apple","bat","cat","cat","tin","jungle","bat"};
}
static void Main(string[] args)
{
ApplicationContext context = new ApplicationContext();
context.UniqueWords.Load();
Console.Write("Что сделать? (1-вывести из базы, 2-ввести URL): ");
var answer = int.Parse(Console.ReadLine());
if (answer == 1)
{
Console.WriteLine("\n\nСписок слов:\n");
foreach (var word in context.UniqueWords.Local.ToList())
{
Console.WriteLine($"{word.Word} - {word.Count}");
}
}
else if (answer == 2)
{
ParseHtml parse = new ParseHtml();
Console.Write("Введите URL: ");
var url = Console.ReadLine();
Console.WriteLine();
Console.Write("Введите кол-во вхождений слова: ");
var countText = Console.ReadLine();
try
{
if (int.TryParse(countText, out int count))
{
//"https://www.simbirsoft.com/"
var streamHtml = parse.GetHtmlStream(url);
var parsedHtml = parse.ParseHtmlOfStream(streamHtml);
UniqueWords uniqueWords = new UniqueWords();
var clearArray = uniqueWords.ClearArrayWithWord(parsedHtml);
var dictionaryUniqueWord = uniqueWords.CountUniqueWord(clearArray);
foreach (var keyValue in dictionaryUniqueWord)
{
if (keyValue.Value >= count)
{
Console.WriteLine($"{keyValue.Key} - {keyValue.Value}");
var item = context.UniqueWords.Local.FirstOrDefault(x => x.Word == keyValue.Key && x.Url == url);
if (item == null)
{
var lastItem = context.UniqueWords.Local.LastOrDefault();
var lastIndex = lastItem == null ? 1 : lastItem.IdWord + 1;
context.UniqueWords.Add(new UniqueWord {
IdWord = lastIndex, Url = url, Word = keyValue.Key, Count = keyValue.Value
});
}
else
{
item.Count += keyValue.Value;
}
}
}
context.SaveChanges();
Console.WriteLine("\nНажмите на любую кнопку");
}
else
{
throw new Exception("Некорректное число");
}
}
catch (Exception ex)
{
Console.WriteLine(ex.Message);
}
}
Console.WriteLine("\nНажмите любую кнопку...");
Console.ReadKey();
Process.Start(Assembly.GetExecutingAssembly().Location);
Environment.Exit(0);
}