private void RunFileCollector(string currentFolder)
{
string[] filesInCurrentFolder = Directory.GetFiles(currentFolder, "*.*", SearchOption.AllDirectories);
int countOfFiles = filesInCurrentFolder.Length;
int fileCount = 0;
foreach (string file in filesInCurrentFolder)
{
FileObject fo = new FileObject(file);
fileList.Add(fo);
try
{
switch (Path.GetExtension(file).ToLower())
{
case ".zip":
int increment = Convert.ToInt32(80.0 / Convert.ToDouble(countOfFiles));
ParseZip(fo.fullPath, fo.fullPath, increment);
break;
case ".doc":
case ".docx":
ParseDoc(ref fo, fo.fullPath);
break;
case ".pdf":
ParsePDF(ref fo, fo.fullPath);
break;
case ".xls":
case ".xlsx":
ParseXls(ref fo, fo.fullPath);
break;
case ".ppt":
case ".pptx":
ParsePpt(ref fo, fo.fullPath);
break;
}
}
catch (Exception ex)
{
fo.comment = "File parsing error: " + ex.Message;
}
fileCount++;
progress = Convert.ToInt32(80.0 * Convert.ToDouble(fileCount) / Convert.ToDouble(countOfFiles));
Program.mainWindow.updateProgress(progress);
}
}
private void ParseZip(string baseFolder, string filePath, int increment)
{
using (ZipArchive zipFile = ZipFile.OpenRead(filePath))
{
IReadOnlyCollection <ZipArchiveEntry> zippedFiles = zipFile.Entries;
string tmpUnzipFolder = tmpFolder + "\\" + Guid.NewGuid();
Directory.CreateDirectory(tmpUnzipFolder);
int noOfFiles = zippedFiles.Count;
int FileCount = 0;
foreach (ZipArchiveEntry zippedFile in zippedFiles)
{
string zippedPath = zippedFile.FullName;
if (zippedPath.Substring(zippedPath.Length - 1) != "/")
{
FileObject fo = new FileObject(baseFolder + "\\" + zippedPath.Replace('/', '\\'));
fileList.Add(fo);
string tmpFile = Path.Combine(tmpUnzipFolder, zippedFile.Name);
try
{
switch (Path.GetExtension(zippedPath).ToLower())
{
case ".zip":
zippedFile.ExtractToFile(tmpFile, true);
int zipincrement = Convert.ToInt32(Convert.ToDouble(increment) / Convert.ToDouble(noOfFiles));
ParseZip(fo.fullPath, tmpFile, zipincrement);
File.Delete(tmpFile);
break;
case ".doc":
case ".docx":
zippedFile.ExtractToFile(tmpFile, true);
ParseDoc(ref fo, tmpFile);
File.Delete(tmpFile);
break;
case ".pdf":
zippedFile.ExtractToFile(tmpFile, true);
ParsePDF(ref fo, tmpFile);
File.Delete(tmpFile);
break;
case ".xls":
case ".xlsx":
zippedFile.ExtractToFile(tmpFile, true);
ParseXls(ref fo, tmpFile);
File.Delete(tmpFile);
break;
case ".ppt":
case ".pptx":
zippedFile.ExtractToFile(tmpFile, true);
ParsePpt(ref fo, tmpFile);
File.Delete(tmpFile);
break;
}
}
catch (Exception ex)
{
fo.comment = "File parsing error: " + ex.Message;
}
FileCount++;
progress = Convert.ToInt32((increment * Convert.ToDouble(FileCount) / Convert.ToDouble(noOfFiles)));
Program.mainWindow.updateProgress(progress);
}
}
Directory.Delete(tmpUnzipFolder);
}
}
private void ParseXls(ref FileObject fo, string filePath)
{
Workbook wb = new Workbook(filePath);
Aspose.Cells.Properties.BuiltInDocumentPropertyCollection dp = wb.BuiltInDocumentProperties;
WorksheetCollection wsc = wb.Worksheets;
fo.pageCount = wsc.Count;
int NoOfImages = 0;
int NoOfEmbeddedDocs = 0;
bool isProtected = false;
foreach (Worksheet ws in wsc)
{
OleObjectCollection oles = ws.OleObjects;
if (ws.IsProtected)
{
isProtected = true;
}
foreach (OleObject ole in oles)
{
switch (ole.FileFormatType)
{
case FileFormatType.Doc:
case FileFormatType.Xlsm:
case FileFormatType.Docx:
case FileFormatType.Xlsx:
case FileFormatType.Ppt:
case FileFormatType.Pdf:
case FileFormatType.CSV:
case FileFormatType.VSD:
case FileFormatType.VSDX:
case FileFormatType.Html:
case FileFormatType.XML:
NoOfEmbeddedDocs++;
break;
case FileFormatType.BMP:
case FileFormatType.TIFF:
NoOfImages++;
break;
default:
NoOfImages++;
break;
}
}
}
fo.embeddedDocsCount = NoOfEmbeddedDocs;
fo.imageCount = NoOfImages;
fo.hasPassword = isProtected;
string tmpFolderToExtract = tmpFolder + "\\" + Guid.NewGuid();
Directory.CreateDirectory(tmpFolderToExtract);
string tmpTextFile = tmpFolderToExtract + "\\" + "tmpTextexport.txt";
byte[] workbookData = new byte[0];
TxtSaveOptions opts = new TxtSaveOptions();
opts.Separator = ' ';
for (int idx = 0; idx < wb.Worksheets.Count; idx++)
{
MemoryStream ms = new MemoryStream();
wb.Worksheets.ActiveSheetIndex = idx;
wb.Save(ms, opts);
ms.Position = 0;
byte[] sheetData = ms.ToArray();
byte[] combinedArray = new byte[workbookData.Length + sheetData.Length];
Array.Copy(workbookData, 0, combinedArray, 0, workbookData.Length);
Array.Copy(sheetData, 0, combinedArray, workbookData.Length, sheetData.Length);
workbookData = combinedArray;
}
File.WriteAllBytes(tmpTextFile, workbookData);
fo.wordCount = GetWordCount(tmpTextFile);
fo.characterCount = GetCharCount(tmpTextFile);
if (File.Exists(tmpTextFile))
{
File.Delete(tmpTextFile);
}
if (Directory.Exists(tmpFolderToExtract))
{
Directory.Delete(tmpFolderToExtract);
}
}