public static async Task <PlagiarismInLocalDB> Find(string text)
{
var simplifiedText = TextManager.SimplifyText(text).Replace("\r\n", " ");
var words = TextManager.WordsFromText(simplifiedText).ToArray();
var wordCount = words.Length;
var plagiarismResult = new Dictionary <int, HashSet <(int DBDocIndex, int initialDocIndex)> >();
double vodnost = 0;
for (int i = 0; i <= words.Length - Shingle.Lenght; i++)
{
if (TextManager.StopWords.Contains(words[i]))
{
vodnost++;
}
var documentIdToWordsPositions = SQLLoader.GetDocuments(Shingle.ListFromWords(words, i));
var plagiarismForShingle = Logic.FindPlagiarism(documentIdToWordsPositions, i);
foreach (var kvp in plagiarismForShingle)
{
if (plagiarismResult.TryGetValue(kvp.Key, out HashSet <(int DBDocIndex, int initialDocIndex)> plagiarizedPositions))
{
plagiarizedPositions.UnionWith(kvp.Value);
}
public static async Task <PlagiarismInLocalDB> Find(string text)
{
var simplifiedText = TextManager.SimplifyText(text).Replace("\r\n", " ");
var words = TextManager.WordsFromText(simplifiedText).ToArray();
var wordCount = words.Length;
var plagiarismResult = new Dictionary <int, HashSet <(int DBDocIndex, int initialDocIndex)> >();
double vodnost = 0;
for (int i = 0; i <= words.Length - Shingle.Lenght; i++)
{
if (TextManager.StopWords.Contains(words[i]))
{
vodnost++;
}
var documentIdToWordsPositions = SQLLoader.GetDocuments(Shingle.ListFromWords(words, i));
//var plagiarismForShingle = Logic.FindPlagiarism(documentIdToWordsPositions, i);
//foreach (var kvp in plagiarismForShingle)
//{
// if (plagiarismResult.TryGetValue(kvp.Key, out HashSet<(int DBDocIndex, int initialDocIndex)> plagiarizedPositions))
// {
// plagiarizedPositions.UnionWith(kvp.Value);
// }
// else
// {
// plagiarismResult.Add(kvp.Key, new HashSet<(int, int)>(kvp.Value));
// }
//}
Console.WriteLine(i);
}
for (int i = words.Length - Shingle.Lenght + 1; i < words.Length; i++)
{
if (TextManager.StopWords.Contains(words[i]))
{
vodnost++;
}
}
vodnost /= Convert.ToDouble(words.Length);
double toshnotnost = (words.Length - words.Distinct().Count()) / Convert.ToDouble(words.Length);
return(new PlagiarismInLocalDB
{
Words = words,
WordCount = wordCount,
PlagiarismResult = plagiarismResult,
Vodnost = vodnost,
Toshnotnost = toshnotnost,
Text = text
});
}
public static async Task <Plagiarism <int> > FindAsync(string[] initialWords, Dictionary <int, string> initialDocIndexToSimplifiedWord, int[] initialDocIndexes, string[] simplifiedWords)
{
string debugLog = "";
var plagiarismDB = new Plagiarism <int>();
Stopwatch stopwatch = new Stopwatch();
stopwatch.Start();
var shinglesCount = initialDocIndexes.Length - Shingle.Lenght;
var tasks = new Task <Plagiarism <int> > [shinglesCount + 1];
for (int i = 0; i <= shinglesCount; i++)
{
int savei = i;
tasks[i] = Task <Plagiarism <int> > .Factory.StartNew(() =>
{
List <string> wordsList = new List <string>();
for (int j = 0; j < Shingle.Lenght; j++)
{
wordsList.Add(initialDocIndexToSimplifiedWord[initialDocIndexes[savei + j]]);
}
var documentIdToDBDocWordsPositionsForShingle = SQLLoader.GetDocuments(wordsList);
var initialDocIndexesForShingle = new List <int>();
for (int j = 0; j < Shingle.Lenght; j++)
{
initialDocIndexesForShingle.Add(initialDocIndexes[savei + j]);
}
var plagiarismDBForShingle = Plagiarism <int> .FindPlagiarism(documentIdToDBDocWordsPositionsForShingle, initialDocIndexesForShingle);
return(plagiarismDBForShingle);
});
}
Task.WaitAll(tasks);
for (int i = 0; i <= shinglesCount; i++)
{
var plagiarismDBForShingle = tasks[i].Result;
plagiarismDB.Add(plagiarismDBForShingle);
}
stopwatch.Stop();
debugLog += "DB PLAG TIME " + stopwatch.ElapsedMilliseconds + " ";
foreach (var kvp in plagiarismDB.SourceIdToInitialWordsIndexes)
{
plagiarismDB.SourceIdToInitialDocumentHtml.Add(kvp.Key, TextManager.ComposeHtmlText(initialWords, kvp.Value));
}
plagiarismDB.DebugLogs = debugLog;
return(plagiarismDB);
}