/// <summary>Gets file entity using absolute path.</summary>
/// <param name="absolutePath">Absolute path.</param>
/// <exception cref="System.NotSupportedException">Specified file is not supports read or security error is detected.</exception>
/// <returns>The <see cref="File"/> entity.</returns>
private static File GetScanFileByPath(string absolutePath)
{
try
{
var fileInfo = new FileInfo(absolutePath);
var scanFile = new File
{
Filename = fileInfo.Name,
Extension = fileInfo.Extension,
Path = fileInfo.DirectoryName,
FullName = fileInfo.FullName
};
return(scanFile);
}
catch (System.Exception ex)
{
throw new System.NotSupportedException(ex.Message, ex.InnerException);
}
}
/// <summary>Scans the location of words in the StreamReader of the file.</summary>
/// <param name="scanFile">Scan file entity.</param>
/// <param name="stream">Stream reader for the text file.</param>
/// <param name="type">Type of result.</param>
/// <exception cref="FileNotFoundException">Specified pathname does not exist.</exception>
/// <exception cref="DirectoryNotFoundException">Directory cannot be found.</exception>
/// <exception cref="System.NotSupportedException">Specified file is not supports read or security error is detected.</exception>
private static ScanResult ParseFile(File scanFile, TextReader stream, TypeResult type)
{
var pattern = new Regex(@"[^\W_\d]([^\W_\d]|[-’'](?=[^\W_\d]))*([^\W_\d]|['’])?");
var wordsLocker = new object();
var compositionsLocker = new object();
var fileWords = new List <Word>();
var compositions = new List <Composition>();
var lines = new Dictionary <int, string>();
string currentLine;
var counter = 1;
while ((currentLine = stream.ReadLine()) != null)
{
lines.Add(counter, currentLine.ToLower());
counter++;
}
Parallel.ForEach(
lines,
line =>
{
var words = pattern.Matches(line.Value);
for (var i = 0; i < words.Count; i++)
{
var scanWord = GetOrCreateScanWord(wordsLocker, fileWords, scanFile, words[i].Value);
if (type == TypeResult.CompositionOfWords)
{
AddWordToCompositions(compositionsLocker, compositions, scanWord, line.Key, words[i].Index + 1);
}
}
});
return(new ScanResult {
Words = fileWords, Compositions = compositions
});
}
/// <summary>Scans the unique words in the StreamReader of the file.</summary>
/// <param name="scanFile">Scan file entity.</param>
/// <param name="stream">Stream reader for the text file.</param>
/// <exception cref="FileNotFoundException">Specified pathname does not exist.</exception>
/// <exception cref="DirectoryNotFoundException">Directory cannot be found.</exception>
/// <exception cref="System.NotSupportedException">Specified file is not supports read or security error is detected.</exception>
/// <returns>Unsorted collection of words in file.</returns>
public List <Word> ParseUnigueWordsInFile(File scanFile, StreamReader stream)
{
return(ParseFile(scanFile, stream, TypeResult.OnlyUniqueWordsInFile).Words);
}
/// <summary>Gets or creates word entity using the word string.</summary>
/// <param name="wordsLocker">Mutex for adding words.</param>
/// <param name="fileWords">Existing words to compare.</param>
/// <param name="scanFile">File containing this word.</param>
/// <param name="wordText">The word string.</param>
/// <returns>The <see cref="Word"/> entity.</returns>
private static Word GetOrCreateScanWord(object wordsLocker, ICollection <Word> fileWords, File scanFile, string wordText)
{
Word word;
lock (wordsLocker)
{
word = fileWords.FirstOrDefault(w => w.TheWord == wordText);
if (!Equals(word, default(Word)))
{
word.Count++;
return(word);
}
word = new Word {
File = scanFile, TheWord = wordText, Count = 1
};
fileWords.Add(word);
}
return(word);
}
/// <summary>Scans all the words and their positions in the StreamReader of the file.</summary>
/// <param name="scanFile">Scan file entity.</param>
/// <param name="stream">Stream reader for the text file.</param>
/// <exception cref="FileNotFoundException">Specified pathname does not exist.</exception>
/// <exception cref="DirectoryNotFoundException">Directory cannot be found.</exception>
/// <exception cref="System.NotSupportedException">Specified file is not supports read or security error is detected.</exception>
/// <returns>Unsorted collection of word compositions in file.</returns>
public List <Composition> ParseAllWordsInFile(File scanFile, StreamReader stream)
{
return(ParseFile(scanFile, stream, TypeResult.CompositionOfWords).Compositions);
}