public void DocumentExtraction_ShouldReturnResult(DocumentExtractionPreset preset) =>
new DocumentExtraction
{
Document = _testData.GetFilePath(),
TimeoutMS = (int)TimeSpan.FromMinutes(10).TotalMilliseconds,
Preset = preset,
}
.Invoke()
.Should()
.StartWith("Our Google properties revenues");
public async Task <string> ExtractDocumentAsync(string filePath, DocumentExtractionPreset preset, CancellationToken cancellationToken)
{
var config = new JObject
{
{ "preset_config", preset.ToString("F").ToLower() }
};
var docExtraction =
new DocumentExtraction(_indicoClientLegacy)
{
JsonConfig = config,
};
var job = await docExtraction.Exec(filePath, cancellationToken);
return(job.Id);
}
public async Task GetExtractionResult_ShouldReturnDocument(DocumentExtractionPreset preset)
{
// Arrange
var jobId = await _ocrClient.ExtractDocumentAsync(_dataHelper.Files().GetSampleFilePath(), preset);
var jobResult = await _jobAwaiter.WaitReadyAsync <ExtractionJobResult>(jobId, TimeSpan.FromMilliseconds(300));
if (preset != DocumentExtractionPreset.OnDocument)
{
// Act
var extractionResult = await _ocrClient.GetExtractionResultAsync(jobResult.Url);
// Assert
extractionResult.Should().NotBeNullOrEmpty();
}
else
{
// Act
var extractionResult = await _ocrClient.GetExtractionResultAsync <JArray>(jobResult.Url);
// Assert
extractionResult.Count.Should().Be(1);
}
}
public async Task ExtractDocument_ShouldReturnJobNumber(DocumentExtractionPreset preset) => (await _ocrClient.ExtractDocumentAsync(_dataHelper.Files().GetSampleFilePath(), preset, default))