public List <string> GetLines(DetectDocumentTextResponse result)
{
var lines = new List <string>();
result.Blocks.FindAll(block => block.BlockType == "LINE").ForEach(block => lines.Add(block.Text));
return(lines);
}
public void Print(DetectDocumentTextResponse response)
{
if (response != null)
{
this.Print(response.Blocks);
}
}
/// <summary>
/// Unmarshaller the response from the service to the response class.
/// </summary>
/// <param name="context"></param>
/// <returns></returns>
public override AmazonWebServiceResponse Unmarshall(JsonUnmarshallerContext context)
{
DetectDocumentTextResponse response = new DetectDocumentTextResponse();
context.Read();
int targetDepth = context.CurrentDepth;
while (context.ReadAtDepth(targetDepth))
{
if (context.TestExpression("Blocks", targetDepth))
{
var unmarshaller = new ListUnmarshaller <Block, BlockUnmarshaller>(BlockUnmarshaller.Instance);
response.Blocks = unmarshaller.Unmarshall(context);
continue;
}
if (context.TestExpression("DocumentMetadata", targetDepth))
{
var unmarshaller = DocumentMetadataUnmarshaller.Instance;
response.DocumentMetadata = unmarshaller.Unmarshall(context);
continue;
}
}
return(response);
}
public async static void GetTextFromS3(IAmazonTextract textractClient, string bucketName, string keyName)
{
List <Block> result = null;
DetectDocumentTextRequest detectTextRequest = new DetectDocumentTextRequest()
{
Document = new Document {
S3Object = new S3Object {
Bucket = bucketName,
Name = keyName
}
}
};
try
{
Task <DetectDocumentTextResponse> detectTask = textractClient.DetectDocumentTextAsync(detectTextRequest);
DetectDocumentTextResponse detectTextResponse = await detectTask;
result = detectTextResponse.Blocks;
//PrintBlockDetails(result);
}
catch (AmazonTextractException textractException)
{
Console.WriteLine(textractException.Message, textractException.InnerException);
}
}
public async Task <List <Block> > GetTextFromStream(IAmazonTextract textractClient, MemoryStream stream)
{
List <Block> result = null;
DetectDocumentTextRequest detectTextRequest = new DetectDocumentTextRequest()
{
Document = new Document
{
Bytes = stream
}
};
try
{
Task <DetectDocumentTextResponse> detectTask = textractClient.DetectDocumentTextAsync(detectTextRequest);
DetectDocumentTextResponse detectTextResponse = await detectTask;
result = detectTextResponse.Blocks;
//PrintBlockDetails(result);
}
catch (AmazonTextractException textractException)
{
Console.WriteLine(textractException.Message, textractException.InnerException);
}
return(result);
}
public async Task <DetectDocumentTextResponse> DetectTextS3(string bucketName, string key)
{
var result = new DetectDocumentTextResponse();
var s3Object = new S3Object {
Bucket = bucketName,
Name = key
};
var request = new DetectDocumentTextRequest();
request.Document = new Document {
S3Object = s3Object
};
return(await this.textract.DetectDocumentTextAsync(request));
}
public async Task <DetectDocumentTextResponse> DetectTextLocal(string localPath)
{
var result = new DetectDocumentTextResponse();
if (File.Exists(localPath))
{
var request = new DetectDocumentTextRequest();
request.Document = new Document {
Bytes = new MemoryStream(File.ReadAllBytes(localPath))
};
return(await this.textract.DetectDocumentTextAsync(request));
}
Console.WriteLine("File: " + localPath + " doesn't exist");
return(result);
}
public override async Task RunCommand(object sender)
{
var engine = (IAutomationEngineInstance)sender;
var vAccessKey = (string)await v_AccessKey.EvaluateCode(engine);
var vSecretKey = (string)await v_SecretKey.EvaluateCode(engine);
var ocrRequest = new DetectDocumentTextRequest();
MemoryStream memStream = new MemoryStream();
if (v_ImageType == "Filepath")
{
string vFilePath = (string)await v_FilePath.EvaluateCode(engine);
FileStream image = File.OpenRead(vFilePath);
image.CopyTo(memStream);
}
else
{
Bitmap vBitmap = (Bitmap)await v_Bitmap.EvaluateCode(engine);
vBitmap.Save(memStream, ImageFormat.Jpeg);
}
ocrRequest.Document = new Document();
ocrRequest.Document.Bytes = memStream;
AmazonTextractConfig config = new AmazonTextractConfig();
config.RegionEndpoint = Amazon.RegionEndpoint.GetBySystemName(v_AWSRegion);
AmazonTextractClient client = new AmazonTextractClient(vAccessKey, vSecretKey, config);
DetectDocumentTextResponse results = client.DetectDocumentText(ocrRequest);
string readText = "";
if (results.HttpStatusCode == System.Net.HttpStatusCode.OK)
{
foreach (Block block in results.Blocks)
{
if (block.BlockType.Value == "LINE")
{
readText += block.Text + "\n";
}
}
}
else
{
throw new Exception("Call to AWS textract failed");
}
readText.SetVariableValue(engine, v_OutputUserVariableName);
}
