private async Task SearchForPlagiarismLocally(string fileName)
{
richTextBox1.Text = string.Empty;
textBox2.Text = string.Empty;
var simplifiedText = TextDocumentManager.SimplifiedTextFromFile(fileName);
var plagiarismInLocalDB = await PlagiarismInLocalDBFinder.Find(simplifiedText);
richTextBox1.Text += simplifiedText + Environment.NewLine;
richTextBox1.Text += $"Words count: {plagiarismInLocalDB.WordCount}" + Environment.NewLine;
textBox2.Text += "Водность текста: " + plagiarismInLocalDB.Vodnost + " % " + Environment.NewLine;
textBox2.Text += "Тошнотность текста: " + plagiarismInLocalDB.Toshnotnost + " % " + Environment.NewLine;
foreach (var kvp in plagiarismInLocalDB.PlagiarismResult)
{
textBox2.Text += "DocumentID: " + kvp.Key + "\tWord Indexes: ";
foreach (var item in kvp.Value)
{
textBox2.Text += item + ", ";
}
textBox2.Text += Environment.NewLine;
}
}
private async Task SearchForPlagiarismInWeb(string fileName)
{
richTextBox1.Text = string.Empty;
textBox2.Text = string.Empty;
var simplifiedText = TextDocumentManager.SimplifiedTextFromFile(fileName);
var plagiarismInWeb = await PlagiarismInWebFinder.Find(simplifiedText);
richTextBox1.Text += simplifiedText + Environment.NewLine;
richTextBox1.Text += $"Words count: {plagiarismInWeb.WordCount}" + Environment.NewLine;
ShowUrlToWordIndexes(plagiarismInWeb.OrderedUrls, plagiarismInWeb.OrderedUrlToWordsIndexes);
_webResults = plagiarismInWeb.OrderedWebResults;
for (int i = 0; i < _webResults.Count; i++)
{
textBox2.Text += _webResults[i].Url;
textBox2.Text += Environment.NewLine;
textBox2.Text += String.Format("{0:P2} text found", _webResults[i].CharactersPercentage);
textBox2.Text += Environment.NewLine;
foreach (var commonTextPart in _webResults[i].CommonTextParts)
{
textBox2.Text += commonTextPart;
textBox2.Text += Environment.NewLine;
}
textBox2.Text += Environment.NewLine;
}
}
/// <summary>
/// The given workspace represents the original workspace. A callback is invoked every time this provider decides to recompute the workspace.
/// </summary>
public IncrementalWorkspaceProvider(FrontEndHostController controller, PathTable pathTable, Workspace workspace, TextDocumentManager documentManager, TestContext?testContext)
{
m_controller = controller;
m_workspace = workspace;
m_documentManager = documentManager;
m_testContext = testContext;
documentManager.Changed += DocumentChanged;
m_pathTable = pathTable;
m_timer = new Timer(_ => RecomputeWorkspaceForLatestChangedDocuments(), state: null, dueTime: Timeout.Infinite, period: Timeout.Infinite);
}
public async Task <ActionResult> Main(HttpPostedFileBase file)
{
if (file?.ContentLength > 0 && (Path.GetFileName(file.FileName).Split('.')[1] == "doc" || Path.GetFileName(file.FileName).Split('.')[1] == "docx"))
{
var types = file.ContentType;
var fileName = Path.GetFileName(file.FileName);
var path = Path.Combine(Server.MapPath("~/App_Data/uploads"), fileName);
file.SaveAs(path);
var initialText = TextDocumentManager.TextFromFile(path);
TextManager.PrepareText(initialText, out string[] initialWords, out Dictionary <int, string> initialDocIndexToSimplifiedWord, out int[] initialDocIndexes, out string[] simplifiedWords, out int wordCount);
var plagiarismInWebSearch = PlagiarismInWebFinder.FindAsync(initialWords, initialDocIndexToSimplifiedWord, initialDocIndexes, simplifiedWords, Server.MapPath("~/App_Data/uploads"));
var plagiarismInLocalDBSearch = PlagiarismInLocalDBFinder.FindAsync(initialWords, initialDocIndexToSimplifiedWord, initialDocIndexes, simplifiedWords);
await Task.WhenAll(plagiarismInWebSearch, plagiarismInLocalDBSearch);
var plagiarismInWeb = plagiarismInWebSearch.Result;
var plagiarismInLocalDB = plagiarismInLocalDBSearch.Result;
var DBPlagiarizedIndexes = plagiarismInLocalDB.InitialWordIndexToSourceIds.Keys.ToList();
DBPlagiarizedIndexes.AddRange(plagiarismInWeb.InitialWordIndexToSourceIds.Keys.ToList());
var allPlagiarismHtmlText = TextManager.ComposeHtmlText(initialWords, DBPlagiarizedIndexes);
var(vodnost, toshnotnost) = TextAnalyzer.Analyze(simplifiedWords);
PlagiarismResult plagiarism = new PlagiarismResult
{
InitialWords = initialWords,
SimplifiedWords = simplifiedWords,
WordCount = wordCount,
Vodnost = vodnost,
Toshnotnost = toshnotnost,
PlagiarismWeb = plagiarismInWeb,
PlagiarismDB = plagiarismInLocalDB,
AllPlagiarismHtmlText = allPlagiarismHtmlText,
DebugLogs = plagiarismInWeb.DebugLogs + plagiarismInLocalDB.DebugLogs
};
ViewBag.DebugLogs = plagiarism.DebugLogs;
foreach (KeyValuePair <int, HashSet <int> > lists in plagiarismInLocalDB.SourceIdToSourceWordsIndexes)
{
Session["Doc" + lists.Key] = lists.Value;
}
foreach (KeyValuePair <string, HashSet <int> > lists in plagiarismInWeb.SourceIdToSourceWordsIndexes)
{
Session[lists.Key] = lists.Value;
}
return(View("Main", plagiarism));
}
return(RedirectToAction("Main"));
}
static void Main(string[] args)
{
//string nameFile = "Пар 3-01 Інтеграл Ньютона-Лейбніца";
//string test = TextDocumentManager.TextFromFile(@"C:\Users\alex1\Desktop\" + nameFile + ".doc");
//AddDocument(nameFile, test);
//test = TextManager.SimplifyText(test).Replace("\r\n", " ");
//AddWords(Logic.IndexingForDB(TextManager.WordsFromText(test).ToArray()), nameFile);
string[] files = new string[5];
files[0] = "Пар 3-01 Інтеграл Ньютона-Лейбніца.doc";
files[1] = "LOL.docx";
files[2] = "GPK.docx";
files[3] = "Середа оцінювання адаптивних інтерфейсів користувачів.docx";
files[4] = "Документ для теста.docx";
// string fileName = "Пар 3-01 Інтеграл Ньютона-Лейбніца";
for (int i = 0; i < 5; i++)
{
string initialText = TextDocumentManager.TextFromFile(@"C:\Users\alex1\Desktop\" + files[i]);
AddDocument(files[i], initialText);
TextManager.PrepareText(initialText, out string[] initialWords, out Dictionary <int, string> initialDocIndexToSimplifiedWord, out int[] initialDocIndexes, out string[] simplifiedWords, out int wordCount);
AddWords(Logic.IndexingForDB(initialDocIndexes, simplifiedWords), files[i]);
}
//string str = "";
//List<String> example = new List<string> { "index", "data", "are", "the", "of" };
//foreach (KeyValuePair<int, List<List<int>>> pair in GetDocuments(example))
//{
// Console.WriteLine("Документ - " + pair.Key);
// foreach (List<int> positions in pair.Value)
// {
// foreach (int pos in positions)
// {
// str += Convert.ToString(pos) + " ";
// }
// Console.WriteLine(" Позицiї - " + str);
// str = "";
// }
//}
Console.WriteLine("Ready");
Console.ReadKey();
}
private void CompareTextButton_Click(object sender, EventArgs e)
{
Stopwatch s = new Stopwatch();
s.Start();
var oldText = TextDocumentManager.SimplifiedTextFromFile(oldTextFile);
var newText = TextDocumentManager.SimplifiedTextFromFile(newTextFile);
s.Stop();
textBox2.Text += $"\tReading: {s.Elapsed}" + Environment.NewLine;
s.Restart();
var diffResuls = TextComparer.CompareByWords(oldText, newText);
//CompareLines(oldText, newText);
richTextBox1.ShowDifferences(diffResuls);
s.Stop();
textBox2.Text += $"\tComparing: {s.Elapsed}" + Environment.NewLine;
}