private void readFileContent(string path)
{
TextExtractor extractor = new TextExtractor(path);
string text = extractor.ExtractText();
Console.WriteLine(text);
}
// read a .doc file
static List <string> ReadDocFile(string file)
{
TextExtractor extractor = new TextExtractor(file);
string text = extractor.ExtractText(); //The string 'text' is now loaded with the text from the Word Document
return(AddTextToList(text));
}
public static void ExecuteWordExtraction(string sourceFileName, string destFileName)
{
TextExtractor extractor = new TextExtractor(sourceFileName);
string wordText = extractor.ExtractText();
FileOperators.FileWrite(destFileName, wordText);
}
private static bool TryExtractText(string file, out string fileText)
{
fileText = null;
bool success = false;
if (file.EndsWith(".docx") || file.EndsWith(".doc"))
{
try
{
TextExtractor t = new TextExtractor(file);
fileText = t.ExtractText();
success = true;
}
catch { }
}
else if (file.EndsWith(".txt"))
{
try
{
fileText = File.ReadAllText(file);
success = true;
}
catch { }
}
return(success);
}
private void btnGozat_Click(object sender, EventArgs e)
{
//Dosyayı seç
OpenFileDialog openDialog = new OpenFileDialog();
openDialog.Filter = "Documents (*.txt, *.pdf, *.docx)|*.txt; *.docx; *.pdf;";
openDialog.Title = "Select Document";
if (openDialog.ShowDialog() == DialogResult.Cancel)
{
return;
}
string ext = Path.GetExtension(openDialog.FileName);
if (ext == ".txt") //Eğer kullanıcı txt uzantılı bir dosya seçmiş ise
{
FileStream fStr;
Encoding objEncoding = Encoding.Default;
try
{
fStr = new FileStream(openDialog.FileName, FileMode.Open, FileAccess.Read);
StreamReader sr = new StreamReader(fStr, objEncoding);
txtKelime.Text = sr.ReadToEnd();
sr.Close();
}
catch (Exception exception)
{
MessageBox.Show("Error opening file", exception.ToString());
}
}
else if (ext == ".docx") //Eğer kullanıcı docx uzantılı bir dosya seçmiş ise
{
try
{
TextExtractor extractor = new TextExtractor(openDialog.FileName);
txtKelime.Text = extractor.ExtractText().Replace("\n", " ").Replace(" ", " ").Replace(" ", " ");
}
catch (Exception exception)
{
MessageBox.Show("Error opening file", exception.ToString());
}
}
else if (ext == ".pdf") //Eğer kullanıcı pdf uzantılı bir dosya seçmiş ise
{
try
{
PdfOku pdfOku = new PdfOku();
txtKelime.Text = pdfOku.getPdfResult(openDialog.FileName);
}
catch (Exception exception)
{
MessageBox.Show("Error opening file", exception.ToString());
}
}
else //Eğer kullanıcı farklı bir dosya seçmişse(ki mümkün değil)
{
MessageBox.Show("Geçerli dosya seçiniz");
}
}
public static bool ReadFileCompateText(string path, string s)
{
TextExtractor extractor = new TextExtractor(path);
//The string 'text' is now loaded with the text from the Word Document
string text = extractor.ExtractText();
return(Comparer.CheckTextIfMatch(s, text));
}
//string filePath;
public string gettext(string filePath)
{
string fileData = "";
Code7248.word_reader.TextExtractor extractor = new TextExtractor(filePath);
fileData = extractor.ExtractText();
return(fileData);
}
/// <summary>
/// Uses the buffer class to read the documents in the .docx document
/// An instance of the textextractor class is created to read the documents in the path
/// The extracttext method is used to get the text and store in the
/// </summary>
/// <param name="path"></param>
/// <returns>DocxText</returns>
string ReadDocx(string path)
{
StringBuilder bufferText = new StringBuilder();
string DocxText = bufferText.ToString();
TextExtractor textextractor = new TextExtractor(path);
string text = textextractor.ExtractText();
bufferText.Append(text);
Console.WriteLine(text);
return(DocxText);
}
public void wordReader(string path)
{
try
{
TextExtractor extractor = new TextExtractor(path);
fileContent.Append(extractor.ExtractText());
}
catch (Exception ex)
{
fileContent.Append("");
}
}
public static bool FindTextInDocFile(string fileFullPath, string text)
{
try
{
var extractor = new TextExtractor(fileFullPath);
return(extractor.ExtractText().IndexOf(text, StringComparison.OrdinalIgnoreCase) >= 0);
}
catch (Exception)
{
return(false);
}
}
// Заморочился с возможностью открывать текстовые файлы для зашифровки содержимого (без сохранения обратно)
private void tsmiOpen_Click(object sender, EventArgs e)
{
OpenFileDialog openFile = new OpenFileDialog();
// Фильтруем расширения для отображения текстовых файлов
openFile.Filter = "Все текстовые файлы|*.txt;*rtf;*.doc;*docx;|Файлы .txt|*.txt|Файлы .rtf|*.rtf|Файлы .doc|*.doc|Файлы .docx|*.docx";
if (openFile.ShowDialog() == DialogResult.OK)
{
string type = Path.GetExtension(openFile.FileName); // Извлекаем расширение
switch (type)
{
case ".txt":
try
{
rtbIn.Text = File.ReadAllText(openFile.FileName, Encoding.Default);
}
catch
{
MessageBox.Show("Ошибка загрузки");
}
break;
case ".rtf":
try
{
rtbIn.LoadFile(openFile.FileName, RichTextBoxStreamType.RichText);
}
catch
{
MessageBox.Show("Ошибка загрузки");
}
break;
case ".doc": // Работа с Interop.Word заставила задуматься, но вот чудо...
case ".docx": // ...случайно нашёл в сети библиотеку Code7248.word_reader.dll
try
{
TextExtractor extractor = new TextExtractor(openFile.FileName);
string contents = extractor.ExtractText();
rtbIn.Text = contents;
}
catch
{
MessageBox.Show("Ошибка загрузки");
}
break;
}
}
}
/// <summary>
/// Reads DOC and DOCX file types and extracts the words in each file
/// Requires: The file path is in doc or docx format only
/// </summary>
/// <param name="filenameWithPath">path of DOC or DOCX document including filename</param>
/// <exception cref="PlatformNotSupportedException">Thrown when the file to read is not of supported
/// doc format.
/// </exception>
/// <returns>
/// A Dictionary where the Key contains the filename and the Value contains the entire wordlist
/// </returns>
internal static Dictionary <string, List <string> > readDocFiles(string filenameWithPath)
{
Contract.Requires <PlatformNotSupportedException>(System.IO.Path.GetExtension(filenameWithPath).Equals(".doc") ||
System.IO.Path.GetExtension(filenameWithPath).Equals(".docx"));
List <string> result = new List <string>();
Dictionary <string, List <string> > listresult = new Dictionary <string, List <string> >();
TextExtractor extractor = new TextExtractor(filenameWithPath);
string temp = extractor.ExtractText().Trim();
result.AddRange(temp.Split(new string[] { "\t\r\n", " " }, StringSplitOptions.RemoveEmptyEntries));
listresult.Add(filenameWithPath, result);
return(listresult);
}
public DocumentVersionReturnModel Create(DocumentVersion documentVersion)
{
TextExtractor extractor = new TextExtractor(documentVersion.filePath);
return(new DocumentVersionReturnModel
{
Id = documentVersion.Id,
Text = extractor.ExtractText(),
//FilePath = documentVersion.filePath,
DocumentId = documentVersion.DocumentId,
ModifiedBy = documentVersion.ModifiedBy,
CreationDate = documentVersion.CreationDate,
VersionNumber = documentVersion.VersionNumber,
});
}
/*
* Reads .doc files
* Useage patterns from (code7248, 2012)
* @param string file (filepath)
* @return string[]
*/
static private string[] ReadDoc(string file)
{
string[] words = null;//sets new array
try
{
TextExtractor te = new TextExtractor(file); //readers text
string line = te.ExtractText(); //sets string to read line
words = SetArray(line, file); //sets array to line read
}
catch (Exception err)
{
MessageBox.Show("Error: " + err);
}
return(words);//returns array to read() method
}
private void SearchForDocFiles(string p_sPath)
{
FileInfo[] Files;
try
{
dir = new DirectoryInfo(@p_sPath);
Files = dir.GetFiles("*.doc");
}
catch
{
InvalidDirectory dirForm = new InvalidDirectory();
dirForm.Show();
return;
}
foreach (FileInfo file in Files)
{
string fileName = dir.ToString() + file.ToString();
try
{
TextExtractor extractor = new TextExtractor(fileName);
string text = extractor.ExtractText();
string textToLower = text.ToLower();
if (StringContains(textToLower, searchWords))
{
resultListBox.Items.Add(file.ToString());
CVdirs.Add(fileName);
}
}
catch
{
Console.WriteLine("could not open file: " + fileName);
}
}
}
private void button3_Click(object sender, EventArgs e)
{
timer1.Stop();
button5.Text = "P L A Y";
OpenFileDialog dlg = new OpenFileDialog();
string filepath;
dlg.Filter = "Text files(*.txt)|*.txt|Doc files(*.doc)|*.doc|Docx files(*.docx)|*.docx|All files(*.*)|*.*";
if (dlg.ShowDialog() == DialogResult.OK)
{
filepath = dlg.FileName.ToString();
string strText = string.Empty;
try
{
string ext = System.IO.Path.GetExtension(dlg.FileName);
if (ext == ".doc" || ext == ".docx")
{
Code7248.word_reader.TextExtractor extractor = new TextExtractor(dlg.FileName);
string contents = extractor.ExtractText();
MyText.Text = contents;
}
else
{
MyText.Text = File.ReadAllText(dlg.FileName);
}
MyText.Text = MyText.Text.Replace("\t", " ");
MyText.Text = MyText.Text.Replace(Environment.NewLine, "\n");
if (mychar == '\n')
{
MyText.Text = MyText.Text.Replace("\n", " ");
}
string nolstrTextine = MyText.Text.Replace("\n", System.Convert.ToString(mychar));
i = 0;
if (mychar == '\n')
{
nolstrTextine = nolstrTextine.Replace(".", ". " + mychar);
}
words = nolstrTextine.Split(new char[] { mychar }, StringSplitOptions.RemoveEmptyEntries); ;
progressBar1.Maximum = words.Length;
timer1.Stop();
button5.Text = "P L A Y";
}
catch (Exception ex)
{
MessageBox.Show(ex.Message);
}
}
}
