public void CanExtractUTF8TextContents()
{
var sampleTextFileContent = "This is just some sample text that could be in a file.";
var extractor = new TextFileContentExtractor();
var extractedContent = extractor.ExtractTextContents(Encoding.UTF8.GetBytes(sampleTextFileContent));
Assert.Equal(sampleTextFileContent, extractedContent);
}
public async Task ExecutesFullIntegrationPass()
{
var jobStatusStore = new JobStatusStore(Configuration.StorageConnectionString, Configuration.JobStatusContainerName);
var blobReader = new AzureBlobReader(Configuration.StorageConnectionString, Configuration.FileUploadContainerName);
var extractor = new TextFileContentExtractor();
var searchIndex = new AzureSearchIndex(Configuration.SearchServiceName, Configuration.SearchAdminKey);
var docScorer = new TextDocumentScorer(searchIndex);
var workflow = new ParsingWorkflow(jobStatusStore, blobReader, extractor, searchIndex, docScorer);
var blobId = Guid.NewGuid().ToString().Replace("-", String.Empty);
var jobId = Guid.NewGuid().ToString().Replace("-", String.Empty);
var blobUri = String.Format("{0}/{1}", jobId, blobId);
var blobDetails = new BlobDetails();
blobDetails.ContainerName = Configuration.FileUploadContainerName;
blobDetails.FullBlobPath = blobUri;
blobDetails.DocumentId = blobId;
blobDetails.JobId = jobId;
var job = new JobStatus();
job.OriginalFileName = "not-real-file.txt";
job.IsComplete = false;
job.JobStartTime = DateTime.UtcNow;
job.JobId = jobId;
await jobStatusStore.UpdateStatusAsync(job);
await createSampleBlob(blobUri);
await workflow.ExecuteAsync(blobDetails);
job = await jobStatusStore.ReadStatusAsync(jobId);
var categoryCount = job.Categories.Length;
Assert.Equal(1, categoryCount);
Assert.Equal("Heavy Hitter", job.Categories[0]);
}
