public static async Task <HttpResponseMessage> Run([HttpTrigger(AuthorizationLevel.Function, "get", "post", Route = "Recognition/ExtractText")] HttpRequestMessage req, TraceWriter log, ExecutionContext context)
{
try
{
if (!DemoConfiguration.UnlockSupport(log))
{
return(GenerateErrorMessage(ApiError.LicenseNotSet, req));
}
var leadParameterObject = ParseLeadWebRequestParameters(req);
if (!leadParameterObject.Successful)
{
return(GenerateErrorMessage(ApiError.InvalidRequest, req));
}
var imageReturn = await GetImageStreamAsync(leadParameterObject.LeadWebRequest.fileUrl, req, DemoConfiguration.MaxUrlMbs);
if (!imageReturn.Successful)
{
return(GenerateErrorMessage(imageReturn.ErrorType.Value, req));
}
using (imageReturn.Stream)
{
LoadDocumentOptions options = new LoadDocumentOptions()
{
FirstPageNumber = leadParameterObject.LeadWebRequest.FirstPage,
LastPageNumber = leadParameterObject.LeadWebRequest.LastPage
};
RecognitionEngine recognitionEngine = new RecognitionEngine
{
WorkingDirectory = Path.GetTempPath(),
OcrEngine = GetOcrEngine()
};
var documentPageText = recognitionEngine.ExtractText(imageReturn.Stream, options);
List <ExtractTextData> PageDataList = new List <ExtractTextData>();
int currentPage = options.FirstPageNumber;
foreach (var page in documentPageText)
{
for (int i = 0; i < page.Words.Count; i++)
{
var word = page.Words[i];
word.Bounds = word.Bounds.ToLeadRect().ToLeadRectD();
page.Words[i] = word;
}
for (int i = 0; i < page.Characters.Count; i++)
{
var character = page.Characters[i];
character.Bounds = character.Bounds.ToLeadRect().ToLeadRectD();
page.Characters[i] = character;
}
ExtractTextData pageData = new ExtractTextData
{
PageNumber = currentPage,
PageText = page.Text,
Words = page.Words,
Characters = page.Characters
};
PageDataList.Add(pageData);
currentPage++;
}
using (var ms = new MemoryStream())
{
using (TextWriter tw = new StreamWriter(ms))
{
tw.Write(JsonConvert.SerializeObject(PageDataList));
tw.Flush();
ms.Position = 0;
Guid id = Guid.NewGuid();
string baseName = $"ExtractText-{id}.json";
var blobUri = UploadFileToBlobStorage(ms, baseName);
var returnRequest = req.CreateResponse(HttpStatusCode.OK);
returnRequest.Content = new StringContent(JsonConvert.SerializeObject(blobUri));
return(returnRequest);
}
}
}
}
catch (Exception ex)
{
log.Error($"API Error occurred for request: {context.InvocationId} \n Details: {JsonConvert.SerializeObject(ex)}");
return(GenerateErrorMessage(ApiError.InternalServerError, req));
}
}
private async Task <HttpResponseMessage> ParseText([FromUri] LeadWebRequest request, bool additionalInfo)
{
try
{
AuthenticateRequest();
if (!VerifyCommonParameters(request))
{
throw new MalformedRequestException();
}
using (var stream = await GetImageStream(request.fileUrl))
{
int lastPage = request.LastPage;
ValidateFile(stream, ref lastPage);
LoadDocumentOptions options = new LoadDocumentOptions()
{
FirstPageNumber = request.FirstPage,
LastPageNumber = lastPage
};
RecognitionEngine recognitionEngine = new RecognitionEngine();
recognitionEngine.OcrEngine = ocrEngine;
recognitionEngine.WorkingDirectory = Path.GetTempPath();
var documentPageText = recognitionEngine.ExtractText(stream, options);
List <ExtractTextData> PageDataList = new List <ExtractTextData>();
int currentPage = options.FirstPageNumber;
if (!additionalInfo)
{
foreach (var page in documentPageText)
{
for (int i = 0; i < page.Words.Count; i++)
{
var word = page.Words[i];
word.Bounds = word.Bounds.ToLeadRect().ToLeadRectD();
page.Words[i] = word;
}
ExtractTextData pageData = new ExtractTextData
{
PageNumber = currentPage,
PageText = page.Text,
Words = page.Words.Select(w => new { w.Value, w.Bounds }).ToList()
};
PageDataList.Add(pageData);
currentPage++;
}
}
else
{
foreach (var page in documentPageText)
{
for (int i = 0; i < page.Words.Count; i++)
{
var word = page.Words[i];
word.Bounds = word.Bounds.ToLeadRect().ToLeadRectD();
page.Words[i] = word;
}
for (int i = 0; i < page.Characters.Count; i++)
{
var character = page.Characters[i];
character.Bounds = character.Bounds.ToLeadRect().ToLeadRectD();
page.Characters[i] = character;
}
ExtractTextData pageData = new ExtractTextData
{
PageNumber = currentPage,
PageText = page.Text,
Words = page.Words,
Characters = page.Characters
};
PageDataList.Add(pageData);
currentPage++;
}
}
using (var ms = new MemoryStream())
{
using (TextWriter tw = new StreamWriter(ms))
{
tw.Write(JsonConvert.SerializeObject(PageDataList));
tw.Flush();
ms.Position = 0;
Guid id = Guid.NewGuid();
string baseName = $"ExtractText-{id}.json";
string urlPath = $"{Url.Request.RequestUri.GetComponents(UriComponents.SchemeAndServer, UriFormat.Unescaped)}/{new DirectoryInfo(DemoConfiguration.OutputFileDirectory).Name}/{baseName}";
string serverFilePath = $"{DemoConfiguration.OutputFileDirectory}{baseName}";
SaveToDisk(ms, serverFilePath);
return(new HttpResponseMessage(HttpStatusCode.OK)
{
Content = new StringContent(urlPath)
});
}
}
}
}
catch (Exception e)
{
return(GenerateExceptionMessage(e));
}
}