string DownloadLocalCopy(String tenantId, String jobId)
{
Logger.DebugFormat("Downloaded {0}", _inputFileName);
if (IsUnzippedHtmlFile())
{
return(_inputFileName);
}
var workingFolder = Path.GetDirectoryName(_inputFileName);
ZipFile.ExtractToDirectory(_inputFileName, workingFolder);
Logger.DebugFormat("Extracted zip to {0}", workingFolder);
var htmlFile = Path.ChangeExtension(_inputFileName, "html");
if (File.Exists(htmlFile))
{
Logger.DebugFormat("Html file is {0}", htmlFile);
return(htmlFile);
}
htmlFile = Path.ChangeExtension(_inputFileName, "htm");
if (File.Exists(htmlFile))
{
Logger.DebugFormat("Html file is {0}", htmlFile);
return(htmlFile);
}
var msg = string.Format("Html file not found for {0}!", jobId);
Logger.Error(msg);
throw new Exception(msg);
}
protected async override Task <ProcessResult> OnPolling(PollerJobParameters parameters, string workingFolder)
{
string pathToFile = await DownloadBlob(parameters.TenantId, parameters.JobId, parameters.FileName, workingFolder).ConfigureAwait(false);
String fileName = Path.Combine(Path.GetDirectoryName(pathToFile), parameters.All[JobKeys.FileName]);
Logger.DebugFormat("Move blob id {0} to real filename {1}", pathToFile, fileName);
if (File.Exists(fileName))
{
File.Delete(fileName);
}
File.Copy(pathToFile, fileName);
var converter = new HtmlToPdfConverterFromDiskFile(fileName, base.JobsHostConfiguration)
{
Logger = Logger
};
var pdfConvertedFileName = converter.Run(parameters.TenantId, parameters.JobId);
await AddFormatToDocumentFromFile(
parameters.TenantId,
parameters.JobId,
new DocumentFormat(DocumentFormats.Pdf),
pdfConvertedFileName,
new Dictionary <string, object>()).ConfigureAwait(false);
return(ProcessResult.Ok);
}
public void Verify_sanitize_of_single_mhtml_file()
{
var tempFile = Path.ChangeExtension(Path.GetTempFileName(), ".mht");
File.Copy(TestConfig.PathToMht, tempFile);
string mhtml = File.ReadAllText(tempFile);
MHTMLParser parser = new MHTMLParser(mhtml)
{
OutputDirectory = Path.GetDirectoryName(tempFile),
DecodeImageData = true
};
var outFile = Path.ChangeExtension(tempFile, ".html");
File.WriteAllText(outFile, parser.getHTMLText());
_sanitizer = new SafeHtmlConverter(outFile)
{
Logger = NullLogger.Instance
};
var result = _sanitizer.Run("jobtest");
Assert.That(File.Exists(result), "Output pdf file not created");
File.Delete(result);
}
public LibreOfficeUnoConversion(JobsHostConfiguration config)
{
//Needed by UNO SDK5.
//http://stackoverflow.com/questions/31856025/bootstrap-uno-api-libreoffice-exception
//look at comments of funbit. We need to set UNO_PATH and soffice should be in the PATH
//of the system.
var sofficePath = config.GetPathToLibreOffice();
var unoPath = Path.GetDirectoryName(sofficePath);
Environment.SetEnvironmentVariable("UNO_PATH", unoPath, EnvironmentVariableTarget.Process);
Environment.SetEnvironmentVariable("PATH", Environment.GetEnvironmentVariable("PATH") + @";" + unoPath, EnvironmentVariableTarget.Process);
_config = config;
Logger = NullLogger.Instance;
}
string DownloadLocalCopy(String jobId)
{
Logger.DebugFormat("Downloaded {0}", _inputFileName);
if (IsUnzippedHtmlFile())
{
return(_inputFileName);
}
var workingFolder = Path.GetDirectoryName(_inputFileName);
ZipFile.ExtractToDirectory(_inputFileName, workingFolder);
Logger.DebugFormat("Extracted zip to {0}", workingFolder);
var originalFileWithoutExtension = Path.GetFileNameWithoutExtension(_inputFileName);
var extractedFiles = Directory.GetFiles(workingFolder, "*.htm*");
String htmlFile = null;
if (extractedFiles.Length == 1)
{
htmlFile = extractedFiles[0];
}
else if (extractedFiles.Length > 1)
{
htmlFile = extractedFiles
.FirstOrDefault(f => Path.GetFileNameWithoutExtension(f).TrimEnd('.') == originalFileWithoutExtension) ??
extractedFiles[0];
}
if (htmlFile != null)
{
Logger.Debug($"Extracted html from {_inputFileName} is {htmlFile}");
return(htmlFile);
}
var msg = $"Html file not found for {jobId} {_inputFileName}";
Logger.Error(msg);
throw new Exception(msg);
}
private void BuildContainer(JobsHostConfiguration config)
{
_container = new WindsorContainer();
ContainerAccessor.Instance = _container;
_container.Register(Component.For <JobsHostConfiguration>().Instance(config));
_container.Kernel.Resolver.AddSubResolver(new CollectionResolver(_container.Kernel, true));
_container.Kernel.Resolver.AddSubResolver(new ArrayResolver(_container.Kernel, true));
_container.AddFacility <LoggingFacility>(config.CreateLoggingFacility);
_container.AddFacility <StartableFacility>();
_container.AddFacility <TypedFactoryFacility>();
_container.Register(
Component.For <IClientPasswordSet>()
.ImplementedBy <EnvironmentVariableClientPasswordSet>(),
//Register from this application
Classes.FromAssemblyInThisApplication()
.BasedOn <IPollerJob>()
.WithServiceFirstInterface(),
Classes.FromAssemblyInThisApplication()
.BasedOn <IPollerTest>()
.WithServiceFirstInterface(),
//Rgister from dll that contains Jobs in name.
Classes.FromAssemblyInDirectory(new AssemblyFilter(Path.GetDirectoryName(Assembly.GetExecutingAssembly().Location), "*.Jobs.*.*"))
.BasedOn <IPollerJob>()
.WithServiceFirstInterface(),
Classes.FromAssemblyInDirectory(new AssemblyFilter(Path.GetDirectoryName(Assembly.GetExecutingAssembly().Location), "*.Jobs.*.*"))
.BasedOn <IPollerTest>()
.WithServiceFirstInterface()
);
_container.Install(
FromAssembly.InDirectory(
new AssemblyFilter(Path.GetDirectoryName(Assembly.GetExecutingAssembly().Location), "*.Jobs.*.*")));
_logger = _container.Resolve <ILogger>();
}
public string Run(string sourceFile, string outType)
{
Logger.DebugFormat("DIRECT SOFFICE.EXE CONVERSION: Starting conversion of blobId {0} to {1}", sourceFile, outType);
string pathToLibreOffice = _config.GetPathToLibreOffice();
var outputFile = Path.ChangeExtension(sourceFile, outType);
string arguments = string.Format("--headless -convert-to {2} -outdir \"{0}\" \"{1}\" ",
Path.GetDirectoryName(sourceFile),
sourceFile,
outType
);
var psi = new ProcessStartInfo(pathToLibreOffice, arguments)
{
UseShellExecute = false,
RedirectStandardError = true,
RedirectStandardOutput = true,
CreateNoWindow = true,
WindowStyle = ProcessWindowStyle.Minimized
};
Logger.DebugFormat("Command: {0} {1}", pathToLibreOffice, arguments);
using (var p = Process.Start(psi))
{
Logger.Debug("Process started");
p.WaitForExit();
Logger.Debug("Process ended");
}
if (!File.Exists(outputFile))
{
throw new Exception("Conversion failed");
}
return(outputFile);
}
public async Task <Boolean> Run(
String pathToFile,
CreatePdfImageTaskParams createPdfImageTaskParams,
Func <int, Stream, Task <Boolean> > pageWriter)
{
String tempFileName = null;
if (Passwords.Count > 0)
{
tempFileName =
Path.Combine(Path.GetDirectoryName(pathToFile),
Path.GetFileNameWithoutExtension(pathToFile) + "_decrypted.pdf");
if (Decryptor.DecryptFile(pathToFile, tempFileName, Passwords))
{
pathToFile = tempFileName;
}
}
using (var sourceStream = File.OpenRead(pathToFile))
{
var settings = new MagickReadSettings
{
Density = new PointD(createPdfImageTaskParams.Dpi, createPdfImageTaskParams.Dpi)
};
settings.FrameIndex = 0; // First page
settings.FrameCount = 1; // Number of pages
MagickFormat imageFormat = TranslateFormat(createPdfImageTaskParams.Format);
Logger.DebugFormat("Image format is {0}", imageFormat.ToString());
using (var images = new MagickImageCollection())
{
bool done = false;
if (!_firstDone)
{
lock (LockForInitializationIssue)
{
if (!_firstDone)
{
images.Read(sourceStream, settings);
done = true;
}
}
}
if (!done)
{
images.Read(sourceStream, settings);
}
var lastImage =
Math.Min(createPdfImageTaskParams.FromPage - 1 + createPdfImageTaskParams.Pages, images.Count) -
1;
for (int page = createPdfImageTaskParams.FromPage - 1; page <= lastImage; page++)
{
var image = images[page];
image.Format = imageFormat;
using (var ms = new MemoryStream())
{
image.Write(ms);
ms.Seek(0L, SeekOrigin.Begin);
await pageWriter(page + 1, ms).ConfigureAwait(false);
}
}
}
}
if (!String.IsNullOrEmpty(tempFileName) && File.Exists(tempFileName))
{
File.Delete(tempFileName);
}
return(true);
}
