public async void CanVerifyNonExistentDocument()
{
var searchIndexer = new AzureSearchIndex(Configuration.SearchServiceName, Configuration.SearchAdminKey);
var doesExist = await searchIndexer.VerifyDocumentAvailable("123", 5);
Assert.False(doesExist);
}
public async void CanIndexPlainTextDocuments()
{
var documentId1 = rng.Next(1000, 10000).ToString();
var textContent1 = "This is some special content to index again, even though you already did.";
var documentId2 = rng.Next(1000, 10000).ToString();
var textContent2 = "In other cases, you may want to look for content that isn't so special";
var searchText = "special content";
var searchIndexer = new AzureSearchIndex(Configuration.SearchServiceName, Configuration.SearchAdminKey);
try
{
await searchIndexer.IndexTextContentAsync(documentId1, textContent1);
await searchIndexer.IndexTextContentAsync(documentId2, textContent2);
await searchIndexer.VerifyDocumentAvailable(documentId1, 5);
await searchIndexer.VerifyDocumentAvailable(documentId2, 5);
var results = await searchIndexer.QuerySearchIndexAsync(searchText);
Assert.Equal(2, results.Count);
}
finally
{
await searchIndexer.RemoveTextContentAsync(documentId1);
await searchIndexer.RemoveTextContentAsync(documentId2);
}
}
public async void SupportProximitySearch(string searchText, int expectedResultCount)
{
var documentId = rng.Next(1000, 10000).ToString();
var textContent = "This is a file that may contain a keyword such as police, discrimination, or abuse.";
var searchIndexer = new AzureSearchIndex(Configuration.SearchServiceName, Configuration.SearchAdminKey);
try
{
await searchIndexer.IndexTextContentAsync(documentId, textContent);
await searchIndexer.VerifyDocumentAvailable(documentId, 5);
var results = await searchIndexer.QueryDocumentAsync(documentId, searchText);
Assert.Equal(expectedResultCount, results.Count);
}
finally
{
await searchIndexer.RemoveTextContentAsync(documentId);
}
}
public async void SupportMultipleCriteriaWithSingleQuery(string queryText, bool doesMatch)
{
var documentId = rng.Next(1000, 10000).ToString();
var textContent = "This sentence will hit on one word, but the query contains multiple keywords.";
var searchIndexer = new AzureSearchIndex(Configuration.SearchServiceName, Configuration.SearchAdminKey);
try
{
await searchIndexer.IndexTextContentAsync(documentId, textContent);
await searchIndexer.VerifyDocumentAvailable(documentId, 5);
var results = await searchIndexer.QueryDocumentAsync(documentId, queryText, true);
Assert.Equal(doesMatch, (results.Count == 1));
}
finally
{
await searchIndexer.RemoveTextContentAsync(documentId);
}
}
public async void SupportExclusionCriteria(string searchText, int expectedResultCount)
{
var documentId = rng.Next(1000, 10000).ToString();
var textContent = "This phrase will contain WHISTLEBLOWER but not the secret word.";
var searchIndexer = new AzureSearchIndex(Configuration.SearchServiceName, Configuration.SearchAdminKey);
try
{
await searchIndexer.IndexTextContentAsync(documentId, textContent);
await searchIndexer.VerifyDocumentAvailable(documentId, 5);
var results = await searchIndexer.QueryDocumentAsync(documentId, searchText);
Assert.Equal(expectedResultCount, results.Count);
}
finally
{
await searchIndexer.RemoveTextContentAsync(documentId);
}
}
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]);
}
public async void CanSearchSingleDocument()
{
var documentId = rng.Next(1000, 10000).ToString();
var textContent = "This is text for a single document, that may contain information about a police report.";
var searchText = "police report";
var searchIndexer = new AzureSearchIndex(Configuration.SearchServiceName, Configuration.SearchAdminKey);
try
{
await searchIndexer.IndexTextContentAsync(documentId, textContent);
await searchIndexer.VerifyDocumentAvailable(documentId, 5);
var results = await searchIndexer.QueryDocumentAsync(documentId, searchText);
Assert.Equal(1, results.Count);
Assert.Equal(documentId, results.Results.First().Document.DocumentId);
}
finally
{
await searchIndexer.RemoveTextContentAsync(documentId);
}
}
//[Fact]
public async Task ClearSearchIndex()
{
var searchIndex = new AzureSearchIndex(Configuration.SearchServiceName, Configuration.SearchAdminKey);
await searchIndex.ClearSearchIndex();
}