static void Main(string[] args)
{
string subscriptionKey = ConfigurationManager.AppSettings["subscriptionKey"];
if (string.IsNullOrWhiteSpace(subscriptionKey))
{
Console.ForegroundColor = ConsoleColor.DarkGreen;
Console.WriteLine("To play this sample, you should firstly get a subscription key and put it into the App.Config file.");
Console.WriteLine("If you don't have one, please access");
Console.WriteLine("http://www.projectoxford.ai/doc/general/subscription-key-mgmt");
Console.ResetColor();
Console.WriteLine();
Console.WriteLine("Please enter any key......");
Console.ReadLine();
return;
}
var vision = new VisionHelper(subscriptionKey);
string imagePathorUrl = "http://www.colins-it.co.uk/blog/wp-content/uploads/2012/08/keep-calm.jpg";
// var imagePathorUrl = GetValidImagePathorUrl();
vision.RecognizeText(imagePathorUrl);
Console.ReadLine();
}
static void Main(string[] args)
{
var searchPath = "pdf";
var outPath = "image";
// Note, this will create a new Azure Search Index for the OCR text
Console.WriteLine("Creating Azure Search index...");
AzureSearch.CreateIndex(serviceClient, indexName);
// Creating an image directory
if (Directory.Exists(outPath) == false)
{
Directory.CreateDirectory(outPath);
}
foreach (var filename in Directory.GetFiles(searchPath, "*.pdf", SearchOption.TopDirectoryOnly))
{
Console.WriteLine("Extracting images from {0} \r\n", System.IO.Path.GetFileName(filename));
var images = PdfImageExtractor.ExtractImages(filename);
Console.WriteLine("{0} images found.", images.Count);
Console.WriteLine();
var directory = System.IO.Path.GetDirectoryName(filename);
foreach (var name in images.Keys)
{
if (name.LastIndexOf(".") + 1 != name.Length)
{
images[name].Save(Path.Combine(outPath, name));
}
}
string ocrText = string.Empty;
Console.WriteLine("Extracting text from image... \r\n");
foreach (var imagefilename in Directory.GetFiles(outPath))
{
OcrResults ocr = vision.RecognizeText(imagefilename);
ocrText += vision.GetRetrieveText(ocr);
File.Delete(imagefilename);
}
Console.WriteLine("Extracting key phrases from processed text... \r\n");
KeyPhraseResult keyPhraseResult = TextExtraction.ProcessText(ocrText);
// Take the resulting orcText and upload to a new Azure Search Index
// It is highly recommended that you upload documents in batches rather
// individually like is done here
if (ocrText.Length > 0)
{
Console.WriteLine("Uploading extracted text to Azure Search...\r\n");
string fileNameOnly = System.IO.Path.GetFileName(filename);
string fileId = System.Convert.ToBase64String(System.Text.Encoding.UTF8.GetBytes(fileNameOnly));
AzureSearch.UploadDocuments(indexClient, fileId, fileNameOnly, ocrText, keyPhraseResult);
}
}
// Execute a test search
Console.WriteLine("Execute Search...");
AzureSearch.SearchDocuments(indexClient, "Azure Search");
Console.WriteLine("All done. Press any key to continue.");
Console.ReadLine();
}