public void FileIsCorrupt()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
extractor.SaveToFolder("TestFiles\\A corrupt compound document.doc", outputFolder);
}
public void OdpWithPassword()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
extractor.SaveToFolder("TestFiles\\An ODP document with password.odp", outputFolder);
}
public void DocWithPassword()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
extractor.SaveToFolder("TestFiles\\A DOC word document with password.doc", outputFolder);
}
public void DocxWithoutEmbeddedFiles()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
var files = extractor.SaveToFolder("TestFiles\\A DOCX word document without embedded files.docx", outputFolder);
Assert.IsTrue(files.Count == 0);
}
public void DocWithDocumentOleObjectAttachedPathBroken()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
extractor.SaveToFolder("TestFiles\\A DOC word document with document ole object attached path broken.doc", outputFolder);
}
public void PptxWithPassword()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
extractor.SaveToFolder("TestFiles\\A PPTX PowerPoint document with password.pptx", outputFolder);
}
public void XlsxWithPassword()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
extractor.SaveToFolder("TestFiles\\A XLSX excel document with password.xlsx", outputFolder);
}
public void FileTypeNotSupported()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
extractor.SaveToFolder("TestFiles\\filetypenotsupported.txt", outputFolder);
}
public void RtfWitht3EmbeddedFilesAndNoSpaceDelimiters()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
var files = extractor.SaveToFolder("TestFiles\\A RTF document with 3 embedded files and no space delimiters.rtf", outputFolder);
Assert.IsTrue(files.Count == 3);
}
public void DocWith2EmbeddedImages()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
var files = extractor.SaveToFolder("TestFiles\\A DOC word document with embedded images.doc", outputFolder);
Assert.IsTrue(files.Count == 2);
}
public void DocWithoutEmbeddedFiles()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
var files = extractor.Extract("TestFiles\\A DOC word document without embedded files.doc", outputFolder);
Assert.IsTrue(files.Count == 0);
}
public void OdsWith2EmbeddedFiles()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
var files = extractor.Extract("TestFiles\\An ODS document with 2 embedded files.ods", outputFolder);
Assert.IsTrue(files.Count == 2);
}
public void PptWithoutEmbeddedFiles()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
var files = extractor.Extract("TestFiles\\A PPT PowerPoint document without embedded files.ppt", outputFolder);
Assert.IsTrue(files.Count == 0);
}
public void XlsxWith2EmbeddedFiles()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
var files = extractor.SaveToFolder("TestFiles\\A XLSX excel document with 2 embedded files.xlsx", outputFolder);
Assert.IsTrue(files.Count == 2);
}
public void RtfWitht11EmbeddedFiles()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
var files = extractor.SaveToFolder("TestFiles\\A RTF document with 11 embedded files.rtf", outputFolder);
Assert.IsTrue(files.Count == 11);
}
public void DocxWithEmbeddedMathTypeObjectSuccessfulExtractsNothing()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
var files = extractor.SaveToFolder("TestFiles\\MathType 5 Object.docx", outputFolder);
Assert.AreEqual(0, files.Count);
}
public void PptxWith3EmbeddedFiles()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
var files = extractor.SaveToFolder("TestFiles\\A PPTX powerpoint document with 3 embedded files.pptx", outputFolder);
Assert.IsTrue(files.Count == 3);
}
public void OdpWith3EmbeddedFiles()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
var files = extractor.SaveToFolder("TestFiles\\An ODP document with 3 embedded files.odp", outputFolder);
Assert.IsTrue(files.Count == 3);
}
public void PptWithEmbeddedMsClipArtGalleryObject()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
var files = extractor.SaveToFolder("TestFiles\\A PPT PowerPoint document containing a MS ClipArt Gallery object.ppt", outputFolder);
Assert.IsTrue(files.Count == 0);
}
public void DocxWith7EmbeddedFiles()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
var files = extractor.SaveToFolder("TestFiles\\A DOCX word document with 7 embedded files.docx", outputFolder);
Assert.IsTrue(files.Count == 7);
}
public void DocWithDocumentOleObjectAttachedPathRemoved()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
var files = extractor.SaveToFolder("TestFiles\\A DOC word document with document ole object attached path removed.doc", outputFolder);
Assert.IsTrue(files.Count == 1);
Assert.AreEqual(Path.GetFileName(files[0]), "Embedded object.pdf");
}
private void SelectButton_Click(object sender, EventArgs e)
{
// Create an instance of the opeKn file dialog box.
var openFileDialog1 = new OpenFileDialog
{
// ReSharper disable once LocalizableElement
Filter = "Microsoft Office files|*.ODT;*.DOC;*.DOCM;*.DOCX;*.DOT;*.DOTM;*.DOTX;*.RTF;*.XLS;*.XLSB;*.XLSM;*.XLSX;*.XLT;" +
"*.XLTM;*.XLTX;*.XLW;*.POT;*.PPT;*.POTM;*.POTX;*.PPS;*.PPSM;*.PPSX;*.PPTM;*.PPTX",
FilterIndex = 1,
Multiselect = false
};
// Process input if the user clicked OK.
if (openFileDialog1.ShowDialog() == DialogResult.OK)
{
// Open the selected file to read.
string tempFolder = null;
try
{
tempFolder = GetTemporaryFolder();
_tempFolders.Add(tempFolder);
var extractor = new OfficeExtractor.Extractor();
var files = extractor.Extract(openFileDialog1.FileName, tempFolder);
FilesListBox.Items.Clear();
if (files == null)
{
return;
}
foreach (var file in files)
{
FilesListBox.Items.Add(file);
}
}
catch (Exception ex)
{
if (tempFolder != null && Directory.Exists(tempFolder))
{
Directory.Delete(tempFolder, true);
}
MessageBox.Show(GetInnerException(ex));
}
}
}
private void oleBackgroundWorker_DoWork(object sender, System.ComponentModel.DoWorkEventArgs e)
{
for (int i = 0; i < openFileDialog1.FileNames.Length; i++)
{
try
{
string file = openFileDialog1.FileNames[i];
string outputFileDirectory = Path.Combine(folderBrowserDialog1.SelectedPath,
Path.GetFileNameWithoutExtension(file));
if (!Directory.Exists(outputFileDirectory))
{
Directory.CreateDirectory(outputFileDirectory);
}
var extractor = new OfficeExtractor.Extractor();
var oleFiles = extractor.Extract(file, outputFileDirectory);
filesListBox.Items.Add(Path.GetFileName(file));
for (int j = 1; j <= oleFiles.Count; j++)
{
if (j != oleFiles.Count)
{
oleBackgroundWorker.ReportProgress(
(int)((i + ((decimal)j / oleFiles.Count)) / openFileDialog1.FileNames.Length * 100),
"├───" + Path.GetFileName(oleFiles[j - 1]));
}
else
{
oleBackgroundWorker.ReportProgress(
(int)((i + ((decimal)j / oleFiles.Count)) / openFileDialog1.FileNames.Length * 100),
"└───" + Path.GetFileName(oleFiles[j - 1]));
}
}
_recentOutputPath = folderBrowserDialog1.SelectedPath;
if (!Directory.EnumerateFileSystemEntries(outputFileDirectory).Any())
{
Directory.Delete(outputFileDirectory);
}
}
catch (Exception ex)
{
MessageBox.Show(ex.Message);
}
}
}
private void SelectButton_Click(object sender, EventArgs e)
{
// Create an instance of the opeKn file dialog box.
var openFileDialog1 = new OpenFileDialog
{
// ReSharper disable once LocalizableElement
Filter = "Microsoft Office files|*.ODT;*.DOC;*.DOCM;*.DOCX;*.DOT;*.DOTM;*.RTF;*.XLS;*.XLSB;*.XLSM;*.XLSX;*.XLT;" +
"*.XLTM;*.XLTX;*.XLW;*.POT;*.PPT;*.POTM;*.POTX;*.PPS;*.PPSM;*.PPSX;*.PPTM;*.PPTX",
FilterIndex = 1,
Multiselect = false
};
// Process input if the user clicked OK.
if (openFileDialog1.ShowDialog() == DialogResult.OK)
{
// Open the selected file to read.
string tempFolder = null;
try
{
tempFolder = GetTemporaryFolder();
_tempFolders.Add(tempFolder);
var extractor = new OfficeExtractor.Extractor();
var files = extractor.SaveToFolder(openFileDialog1.FileName, tempFolder);
FilesListBox.Items.Clear();
if (files == null) return;
foreach (var file in files)
FilesListBox.Items.Add(file);
}
catch (Exception ex)
{
if (tempFolder != null && Directory.Exists(tempFolder))
Directory.Delete(tempFolder, true);
MessageBox.Show(GetInnerException(ex));
}
}
}
public void XlsxWithPassword()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
extractor.SaveToFolder("TestFiles\\A XLSX excel document with password.xlsx", outputFolder);
}
public void PptxWith3EmbeddedFiles()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
var files = extractor.SaveToFolder("TestFiles\\A PPTX powerpoint document with 3 embedded files.pptx", outputFolder);
Assert.IsTrue(files.Count == 3);
}
public void PptxWithPassword()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
extractor.SaveToFolder("TestFiles\\A PPTX PowerPoint document with password.pptx", outputFolder);
}
public void OdpWith3EmbeddedFiles()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
var files = extractor.SaveToFolder("TestFiles\\An ODP document with 3 embedded files.odp", outputFolder);
Assert.IsTrue(files.Count == 3);
}
public void OdpWithPassword()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
extractor.SaveToFolder("TestFiles\\An ODP document with password.odp", outputFolder);
}
public void FileTypeNotSupported()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
extractor.SaveToFolder("TestFiles\\filetypenotsupported.txt", outputFolder);
}
public void FileIsCorrupt()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
extractor.SaveToFolder("TestFiles\\A corrupt compound document.doc", outputFolder);
}
public void DocxWithPassword()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
extractor.SaveToFolder("TestFiles\\A DOCX word document with password.docx", outputFolder);
}
public void XlsxWith2EmbeddedFiles()
{
var outputFolder = CreateTemporaryFolder();
var extractor = new OfficeExtractor.Extractor();
var files = extractor.SaveToFolder("TestFiles\\A XLSX excel document with 2 embedded files.xlsx", outputFolder);
Assert.IsTrue(files.Count == 2);
}
