// when processing file
// - first move the file
// - then update the doc record to say processed
public bool ProcessPdfFile(string fileName, string uniqName, bool bExtractImages, bool bDontOverwriteExistingImages, bool bExtractText, bool bRecogniseDoc,
bool bAddToDocInfoDb, bool bAddToDocPagesDb)
{
// First check if doc details are already in db
if (!ScanDocInfoRecordExists(uniqName))
return false;
// Make a copy of the file in the archive location
string archiveFileName = ScanDocHandler.GetArchiveFileName(uniqName);
if (!Delimon.Win32.IO.File.Exists(archiveFileName))
{
string statusStr = "";
bool bResult = CopyFile(fileName, archiveFileName, ref statusStr);
if (!bResult)
{
logger.Error("Can't make archive copy {0} excp {1}", archiveFileName, statusStr);
return false;
}
}
else
{
logger.Info("Archive file already exists {0}", archiveFileName);
}
// Extract text blocks from file
ScanPages scanPages = new ScanPages(uniqName);
int totalNumPages = 0;
if (bExtractText)
{
PdfTextAndLocExtractor pdfExtractor = new PdfTextAndLocExtractor();
scanPages = pdfExtractor.ExtractDocInfo(uniqName, fileName, _scanConfig._maxPagesForText, ref totalNumPages);
}
// Extract images from file
if (bExtractImages)
{
bool procImages = (!bDontOverwriteExistingImages) | (!Delimon.Win32.IO.File.Exists(PdfRasterizer.GetFilenameOfImageOfPage(_scanConfig._docAdminImgFolderBase, uniqName, 1, false)));
if (procImages)
{
PdfRasterizer rs = new PdfRasterizer(fileName, THUMBNAIL_POINTS_PER_INCH);
try
{
List<string> imgFileNames = rs.GeneratePageFiles(uniqName, scanPages, _scanConfig._docAdminImgFolderBase, _scanConfig._maxPagesForImages, false);
}
finally
{
rs.Close();
}
}
}
// Form partial document info
DateTime fileDateTime = Delimon.Win32.IO.File.GetCreationTime(fileName);
ScanDocInfo scanDocInfo = new ScanDocInfo(uniqName, totalNumPages, scanPages.scanPagesText.Count, fileDateTime, fileName.Replace('\\', '/'), false);
// Add records to mongo databases
if (bAddToDocPagesDb)
AddScanPagesRecToMongo(scanPages);
if (bAddToDocInfoDb)
AddDocInfoRecToMongo(scanDocInfo);
// Request update to unfiled documents list
_scanDocInfoCache.RequestUnfiledListUpdate();
return true;
}
private void AddPages_DoWork(object sender, DoWorkEventArgs e)
{
BackgroundWorker worker = sender as BackgroundWorker;
try
{
_pdfRasterizer = new ScanMonitorApp.PdfRasterizer(_curFileNames[_curBackgroundLoadingFileIdx], POINTS_PER_INCH);
}
catch(Exception excp)
{
logger.Error("PDF Editor requires Ghostscript {0}", excp.Message);
MessageBox.Show("PDF Editor requires Ghostscript to be installed");
return;
}
try
{
int startNewPageNum = 1;
int startNewFileNum = 1;
int pageTotal = 0;
GetFileAndPageOfLastOutDoc(out startNewFileNum, out startNewPageNum, out pageTotal);
// Extract page images
for (int i = 0; i < _pdfRasterizer.NumPages(); i++)
{
if ((worker.CancellationPending == true))
{
e.Cancel = true;
break;
}
System.Drawing.Image pageImg = _pdfRasterizer.GetPageImage(i + 1, false);
object[] args = new object[4];
args[0] = i;
args[1] = startNewPageNum;
args[2] = startNewFileNum;
args[3] = pageImg;
this.Dispatcher.BeginInvoke((Action<int, int, int, System.Drawing.Image>)delegate(int pageIdx, int startNewPgNum, int startNewFilNum, System.Drawing.Image pagImg)
{
BitmapImage bitmap = ConvertToBitmap(pagImg);
PdfPageInfo pgInfo = new PdfPageInfo();
pgInfo.PageNum = pageIdx + 1;
pgInfo.FileIndex = _curBackgroundLoadingFileIdx;
pgInfo.ThumbBitmap = bitmap;
pgInfo.SplitAfter = false;
pgInfo.DeletePage = false;
pgInfo.PageRotation = 0;
pgInfo.ShowFileNum = (_curBackgroundLoadingFileIdx > 0);
pgInfo.NewDocPageNum = pageIdx + startNewPgNum;
pgInfo.NewDocFileNum = startNewFilNum;
_pdfPageList.Add(pgInfo);
}, args);
Thread.Sleep(50);
(sender as BackgroundWorker).ReportProgress(i * 100 / _pdfRasterizer.NumPages(), null);
}
}
catch(Exception excp)
{
logger.Error("PDF Editor AddPages_DoWork failed excp {0}", excp.Message);
}
finally
{
// Close file
_pdfRasterizer.Close();
_pdfRasterizer = null;
}
}