public async Task <IActionResult> GetDocumentStatus(string mortgageApplicationId, string documentId, CancellationToken cancellationToken)
{
var blob = await StorageHelper.GetBlob(_storageAccountName, "rawdocuments", $"{mortgageApplicationId.ToLower()}/{documentId}");
if (!(await blob.ExistsAsync()))
{
return(NotFound());
}
MortgageApplicationDocument doc = new MortgageApplicationDocument();
doc.PopuplateFromBlobProperties(blob);
var finalBlob = await StorageHelper.GetBlob(_storageAccountName, "parseddocuments", $"final/{mortgageApplicationId.ToLower()}/{documentId}");
if (await finalBlob.ExistsAsync())
{
doc.Status = MortgageApplicationStatus.Processed;
doc.HasParsedResults = true;
}
var markedupBlob = await StorageHelper.GetBlob(_storageAccountName, "parseddocuments", $"markedup/{mortgageApplicationId.ToLower()}/{documentId}");
if (await markedupBlob.ExistsAsync())
{
doc.HasMarkedupResults = true;
}
return(Ok(doc));
}
public static async Task Run([BlobTrigger("rawdocuments/{name}", Connection = "Documents")] CloudBlockBlob fileBlob,
[Blob("parseddocuments", Connection = "Documents")] CloudBlobContainer parsedDirectory,
ILogger log, Microsoft.Azure.WebJobs.ExecutionContext context,
CancellationToken cancellationToken)
{
try
{
if (fileBlob.Metadata.ContainsKey("DocumentType"))
{
//skip already processed documents
return;
}
var config = new ConfigurationBuilder()
.SetBasePath(context.FunctionAppDirectory)
.AddJsonFile("local.settings.json", optional: true, reloadOnChange: true)
.AddEnvironmentVariables()
.Build();
string predictionKey = config["PredictionKey"];
string predictionProjectId = config["PredictionProjectId"];
string cognativeKey = config["CognativeKey"];
double minimumPrecision = double.Parse(config["MinimalPrecision"]);
string computerVisionServiceHostName = config["ComputerVisionServiceHostName"];
string customVisionServiceHostName = config["CustomVisionServiceHostName"];
string documentURLWithSAS = GetBlobSasUri(fileBlob);
string tagName = await AnalyseImageAsync(documentURLWithSAS, predictionKey, predictionProjectId, minimumPrecision, customVisionServiceHostName, cancellationToken);
if (!tagName.IsEmpty())
{
MortgageApplicationDocument mortgageApplicationDocument = new MortgageApplicationDocument();
mortgageApplicationDocument.PopuplateFromBlobProperties(fileBlob);
mortgageApplicationDocument.DateParsed = DateTimeOffset.UtcNow;
var parsedResults = await ParseDocumentTextAsync(documentURLWithSAS, cognativeKey, computerVisionServiceHostName, mortgageApplicationDocument.FileName, cancellationToken);
string parsedFileName = $"{tagName}/{fileBlob.Name.Trim('/')}";
CloudBlockBlob blob = parsedDirectory.GetBlockBlobReference(parsedFileName);
mortgageApplicationDocument.SetBlobProperties(blob);
blob.Properties.ContentType = "application/json";
await blob.UploadTextAsync(JsonConvert.SerializeObject(parsedResults));
}
else
{
tagName = "N/A";
}
fileBlob.Metadata.Add("DocumentType", tagName);
await fileBlob.SetMetadataAsync();
}
catch (Exception ex)
{
log.LogError($"Identity Document failed: {ex.ToString()}");
}
}
public static async Task Run([BlobTrigger("parseddocuments/w2/{name}", Connection = "Documents")] CloudBlockBlob parsedBlob,
[Blob("parseddocuments", Connection = "Documents")] CloudBlobContainer blobDirectory,
[Blob("rawdocuments/{name}", FileAccess.Read)] Stream originalImage,
[Blob("parseddocuments/markedup/{name}", FileAccess.ReadWrite)] CloudBlockBlob markedUpImage,
ILogger log, Microsoft.Azure.WebJobs.ExecutionContext context, CancellationToken cancellationToken)
{
try
{
MortgageApplicationDocument mortgageApplicationDocument = new MortgageApplicationDocument();
mortgageApplicationDocument.PopuplateFromBlobProperties(parsedBlob);
mortgageApplicationDocument.HasParsedResults = true;
mortgageApplicationDocument.DocumentType = "w2";
string jsonContents = await parsedBlob.DownloadTextAsync();
ProcessedDocument document = JsonConvert.DeserializeObject <ProcessedDocument>(jsonContents);
CloudBlockBlob finalBlob = blobDirectory.GetBlockBlobReference(parsedBlob.Name.Replace("w2/", "final/"));
mortgageApplicationDocument.Status = MortgageApplicationStatus.Processed;
mortgageApplicationDocument.DateProcessed = DateTimeOffset.UtcNow;
mortgageApplicationDocument.SetBlobProperties(finalBlob);
FindDataLine(document, "SSN", "Social Security Number", mortgageApplicationDocument, "a. Employee's social security number", "social security number", "Employee's soc. sec. number", "Employee's social security number");
FindMoneyLine(document, "Wages", "Total Wages", mortgageApplicationDocument, "Wages");
FindMoneyLine(document, "FedTax", "Federal Tax Witholding", mortgageApplicationDocument, "2. Federal income tax", "Federal income tax withheld");
FindDataLine(document, "CopyType", "Copy Type", mortgageApplicationDocument, "Copy");
FindMoneyLine(document, "SocialSecurityWage", "Social Security Wage", mortgageApplicationDocument, "Social security wages");
FindMoneyLine(document, "SocialSecurityTax", "Social Security Tax", mortgageApplicationDocument, "Social security tax withheld");
FindMoneyLine(document, "MedicareWage", "Medicare Wage", mortgageApplicationDocument, "Medicare wages and tips");
FindMoneyLine(document, "MedicareTax", "Medicare Tax", mortgageApplicationDocument, "Medicare tax withheld");
FindMoneyLine(document, "SocialSecurityTips", "Social Security Tips", mortgageApplicationDocument, "Social security tips");
FindDataLine(document, "EmployerDetails", "Employer Details", mortgageApplicationDocument, "Employer's name, address, and ZIP code");
FindDataLine(document, "EmployeeDetails", "Employee Details", mortgageApplicationDocument, "Employee's first name and initial", "Last name", "Employee's name, address, and ZIP code", "Employee's name (first, middle indial, last)");
await finalBlob.UploadTextAsync(JsonConvert.SerializeObject(mortgageApplicationDocument));
await MarkupService.HighlightHits(document, originalImage, markedUpImage);
}
catch (Exception ex)
{
log.LogError(ex, "W2 Processor Failed");
}
}
public async Task <ActionResult <UploadDocumentResult> > UploadDocument(IFormFile file, string mortgageApplicationId, CancellationToken cancellationToken)
{
UploadDocumentResult result = new UploadDocumentResult();
if (mortgageApplicationId.IsEmpty())
{
result.Message = "Mortgage Application Id not specified";
return(result);
}
if (file == null || file.FileName.IsEmpty())
{
result.Message = "No Files Uploaded";
return(result);
}
MortgageApplicationDocument fileResult = new MortgageApplicationDocument();
fileResult.FileName = ContentDispositionHeaderValue.Parse(file.ContentDisposition).FileName.Trim('"');
fileResult.DocumentId = Guid.NewGuid().ToString("n");
fileResult.MortgageApplicationId = mortgageApplicationId;
fileResult.DateUploaded = DateTimeOffset.UtcNow;
var blob = await StorageHelper.GetBlob(_storageAccountName, "rawdocuments", $"{mortgageApplicationId.ToLower()}/{fileResult.DocumentId}");
blob.Metadata.Add("MortgageApplicationId", mortgageApplicationId);
blob.Metadata.Add("FileName", fileResult.FileName);
blob.Metadata.Add("CreateTime", fileResult.DateUploaded.ToString());
blob.Metadata.Add("DocumentId", fileResult.DocumentId);
blob.Properties.ContentType = file.ContentType;
using (var fileStream = file.OpenReadStream())
{
await blob.UploadFromStreamAsync(fileStream, null, _retryPolicy, null, cancellationToken);
result.FilesUploaded.Add(fileResult);
}
result.Succeeded = true;
return(result);
}
public async Task <ActionResult <MortgageApplication> > GetMortgageApplication(string mortgageApplicationId, CancellationToken cancellationToken)
{
var blobs = await StorageHelper.GetBlobs(_storageAccountName, "rawdocuments", $"{mortgageApplicationId.ToLower()}/");
MortgageApplication mortgageApplication = new MortgageApplication()
{
MortgageApplicationId = mortgageApplicationId, Documents = new List <MortgageApplicationDocument>()
};
if (blobs.SafeAny())
{
foreach (var blob in blobs)
{
MortgageApplicationDocument doc = new MortgageApplicationDocument();
doc.PopuplateFromBlobProperties(blob);
string documentId = doc.DocumentId;
CloudBlockBlob finalBlob = await StorageHelper.GetBlob(_storageAccountName, "parseddocuments", $"final/{doc.MortgageApplicationId}/{documentId}");
MortgageApplicationDocument finalDoc = Newtonsoft.Json.JsonConvert.DeserializeObject <MortgageApplicationDocument>(await finalBlob.DownloadTextAsync());
mortgageApplication.Documents.Add(finalDoc);
var markedupBlob = await StorageHelper.GetBlob(_storageAccountName, "parseddocuments", $"markedup/{mortgageApplicationId.ToLower()}/{documentId}");
finalDoc.HasParsedResults = true;
if (await markedupBlob.ExistsAsync())
{
finalDoc.HasMarkedupResults = true;
}
}
}
else
{
return(NotFound());
}
return(Ok(mortgageApplication));
}
private static DataField FindDataLine(ProcessedDocument document, string name, string title, MortgageApplicationDocument mortgageApplication, params string[] titlesToFind)
{
ProcessedLine foundLine = null;
foreach (var titleToFind in titlesToFind)
{
string cleanTitle = CLEANUP.Replace(titleToFind, "");
foundLine = document.Lines.Where(l => CLEANUP.Replace(l.Text, "").IndexOf(cleanTitle, StringComparison.OrdinalIgnoreCase) > -1).FirstOrDefault();
if (foundLine != null)
{
break;
}
}
var dataLine = foundLine?.BoundingBox.FindClosestsBelow(document.Lines) ?? null;
if (dataLine == null)
{
return(null);
}
DataField dataField = new DataField()
{
FieldName = name,
FieldTitle = title,
LabelBox = foundLine.BoundingBox,
ValueBox = dataLine.BoundingBox,
Value = dataLine.Text
};
if (mortgageApplication.DataFields == null)
{
mortgageApplication.DataFields = new List <DataField>();
}
mortgageApplication.DataFields.Add(dataField);
return(dataField);
}
private static DataField FindMoneyLine(ProcessedDocument document, string name, string title, MortgageApplicationDocument mortgageApplication, params string[] titlesToFind)
{
DataField dataField = FindDataLine(document, name, title, mortgageApplication, titlesToFind);
if (dataField != null)
{
double tmpVal;
if (double.TryParse(CLEANUPMONEY.Replace(dataField.Value, ""), out tmpVal))
{
dataField.Value = tmpVal.ToString("c");
}
}
return(dataField);
}