public ProcessDocumentTest()
{
_fileManager = A.Fake <FileManager>();
_extractionManager = A.Fake <ExtractionManager>();
_callback = A.Fake <Callback>();
_processDocument = new ProcessDocument(_fileManager, _extractionManager, _callback);
_newFileToProcess = new NewFileToProcess {
Url = string.Empty
};
}
public void Must_return_ok_after_enqueue()
{
var newItem = new NewFileToProcess {
Url = "Url", CallbackUrl = "Url", AccessKey = EnvironmentVariables.AccessKey, DocumentIdentifier = "XPTO"
};
var result = _controller.Post(newItem) as OkResult;
Assert.Equal(OkStatusCode, result.StatusCode);
}
public IActionResult Post(NewFileToProcess dto)
{
if (!ModelState.IsValid)
{
return(BadRequest("Missing parameters"));
}
_backgroundJobs.Enqueue(() => _processDocument.Process(dto));
return(Ok());
}
public void Must_enqueue_new_item()
{
var newItem = new NewFileToProcess {
Url = "Url", CallbackUrl = "Url", AccessKey = EnvironmentVariables.AccessKey, DocumentIdentifier = "XPTO"
};
_controller.Post(newItem);
_backgroundJobClient.Verify(x => x.Create(
It.Is <Job>(job => job.Method.Name == "Process" && job.Args[0] == newItem),
It.IsAny <EnqueuedState>()));
}
public bool Process()
{
var allFiles = Directory.GetFiles(path, filter);
if (allFiles.Length == 0)
{
return(false);
}
foreach (var filePath in allFiles)
{
NewFileToProcess?.Invoke(filePath);
}
return(true);
}
private async Task <DocumentResultResponse> ExtractOcr(NewFileToProcess newItem)
{
try
{
var filePath = await _fileManager.Download(newItem.Url).ConfigureAwait(false);
var text = await $"pdftotext {filePath} -".Bash().ConfigureAwait(false);
var numberOfPages = _fileManager.GetNumberOfPages(filePath);
var documentResult = new DocumentResultResponse(numberOfPages, newItem.DocumentIdentifier, newItem.AccessKey);
await RunInParallel(filePath, numberOfPages, documentResult).ConfigureAwait(false);
Console.WriteLine($"Finishing document with URL {newItem.Url}");
_fileManager.Delete(filePath);
documentResult.Success = true;
return(documentResult);
}
catch (Exception ex) {
return(new DocumentResultResponse(ex.Message, newItem.DocumentIdentifier, newItem.AccessKey));
}
}
private void ProcessFile(string filePath)
{
NewFileToProcess?.Invoke(filePath);
}
public async Task Process(NewFileToProcess newItem)
{
var result = await ExtractOcr(newItem).ConfigureAwait(false);
await _callback.Send(result, newItem.CallbackUrl).ConfigureAwait(false);
}
