void SplitFileIntoWords(string fullName)
{
var dof = new DescriptionOfFile(fullName);
dof.Index = Globals.allFiles.Files.Count;
Globals.allFiles.Files.AddLast(dof);
var words = new Dictionary <string, int> ();
string content = File.ReadAllText(fullName);
var matches = Regex.Matches(content, @"([a-z_]+)", RegexOptions.IgnoreCase | RegexOptions.Multiline);
foreach (var match in matches)
{
string word = IndexOfWords.GetNormalizedString(match.ToString());
if (words.ContainsKey(word))
{
words [word]++;
}
else
{
words.Add(word, 1);
}
}
var matchesRus = Regex.Matches(content, @"([а-яёА-ЯЁ_а́еёио́у́ы́э́ю́я́]+)", RegexOptions.Multiline);
foreach (var match in matchesRus)
{
string word = IndexOfWords.GetNormalizedString(match.ToString());
if (words.ContainsKey(word))
{
words [word]++;
}
else
{
words.Add(word, 1);
}
}
var matchesNum = Regex.Matches(content, @"([0-9]+)", RegexOptions.Multiline);
foreach (var match in matchesNum)
{
string word = IndexOfWords.GetNormalizedString(match.ToString());
if (words.ContainsKey(word))
{
words [word]++;
}
else
{
words.Add(word, 1);
}
}
var matchesWeird = Regex.Matches(content, @"([C][#])|([C][+][+])|([F][#])|([F][*])|([A][*])|([C][\-][\-])", RegexOptions.Multiline);
foreach (var match in matchesWeird)
{
string word = IndexOfWords.GetNormalizedString(match.ToString());
if (words.ContainsKey(word))
{
words [word]++;
}
else
{
words.Add(word, 1);
}
}
foreach (var uniqueWord in words.Keys)
{
var wrappedWord = Globals.allWords.NormalizeAndAdd(uniqueWord);
wrappedWord.Occurences.Files.AddLast(dof);
}
}
public static void InitIndex()
{
Globals.allWords = new IndexOfWords();
Globals.allFiles = new CollectionOfFileDescriptions();
}
