public async Task <JsonResult> ImageAnalyse(string data, bool flag)
{
try
{
AnalyseImage Ai = new AnalyseImage();
await Ai.ImageAnalyse(data, flag);
if (Ai.Erorr == "") //converting all object array to Json and returning the Json
{
return(Json(new { Brand = Ai.Brandarray, Tag = Ai.Tagarray, Object = Ai.Objectarray }));
}
return(Json(new { Erorr = Ai.Erorr }));
}
catch (Exception e)// handling runtime errors and returning error as Json
{
return(Json(new { Erorr = e.Message }));
}
}
public static async Task <HttpResponseMessage> Run(
[HttpTrigger(AuthorizationLevel.Anonymous, "post", Route = null)] HttpRequestMessage req,
TraceWriter log)
{
log.Info("Starting process of uploading a Wolf image");
if (!req.Content.IsMimeMultipartContent())
{
log.Error("Please pass a name in the request body");
return(req.CreateResponse(HttpStatusCode.UnsupportedMediaType,
"Please pass a name in the request body"));
}
// Confirm the JWT is valid
var tokenResult = await SecurityJWT.ValidateTokenAsync(req.Headers.Authorization);
if (tokenResult == null)
{
log.Error("Invalid JWT");
return(req.CreateResponse(HttpStatusCode.Unauthorized));
}
log.Info("JWT is all good");
// Extract the username from the token
var userName = tokenResult.FindFirst(SecurityJWT.Predicate).Value;
// Get file info from the request
var provider = new MultipartFormDataStreamProvider(Path.GetTempPath());
await req.Content.ReadAsMultipartAsync(provider);
if (provider.FileData.Count != 1)
{
return(req.CreateErrorResponse(HttpStatusCode.BadRequest, "Please only submit one file"));
}
// Parse the connection string and return a reference to the storage account
log.Info("Uploading image to blob storage");
var blockBlob = await AzureStorageHelper.UploadWolfImageAsync(provider.FileData[0]);
// Run the cognitive services vision API to confirm whether or not it's a wolf. This can take a few seconds
log.Info("Submitting request for image analysis");
var analysisResponse = await AnalyseImage.MakeAnalysisRequest(blockBlob.Uri.ToString(), log);
// Write info about the image to table storage
log.Info("Writing info in table storage");
var id = Path.GetFileNameWithoutExtension(blockBlob.Uri.ToString());
var entity = new WolfImageEntity(analysisResponse.Response.IsWolf, id)
{
UserName = userName,
Id = id,
CognitiveJson = analysisResponse.CognitiveJson,
WolfConfidence = analysisResponse.Response.WolfConfidence,
ImageDescription = analysisResponse.Response.ImageDescription,
ImageDescriptionConfidence = analysisResponse.Response.ImageDescriptionConfidence,
IsAdult = analysisResponse.Response.IsAdult,
IsRacy = analysisResponse.Response.IsRacy,
OriginalFileUrl = blockBlob.Uri.ToString()
};
await AzureStorageHelper.InsertPictureInfo(entity);
if (!analysisResponse.Response.IsWolf || analysisResponse.Response.IsAdult ||
analysisResponse.Response.IsRacy)
{
// Please note:
// ADULT = are things kids shouldn't see, such as violence
// RACY = something like a Victoria Secrets (lingerie) model scores very high. Trust me. I tested. Extensively.
log.Info("Image was not a wolf");
}
else
{
log.Info("Image was a wolf");
// It's a wolf image so submit a message in the queue so that we can resize the image with Cognitive Services.
// The main reason for queuing this, instead of using blob triggers, is that I want a steady rate of
// processing as I have a free account and therefore limited resource. I don't mind if it takes a bit longer.
await AzureStorageHelper.QueueImageResize(blockBlob.Uri.ToString());
}
// Even though I return a 200 it's possible that the image is not a wolf. Will leave it to the UI layer to
// break the news that their dog picture won't make it to the list
return(req.CreateResponse(HttpStatusCode.OK, JsonConvert.SerializeObject(analysisResponse.Response)));
}
