public void PrintDebug(GetDocumentAnalysisResponse response)
{
response.Blocks.ForEach(y => {
Console.WriteLine("<block>");
Console.WriteLine(y.Id + ":" + y.BlockType + ":" + y.Text);
if (y.BlockType == "KEY_VALUE_SET")
{
Console.WriteLine(" <KEY_VALUE_SET>");
PrintBlock(y);
Console.WriteLine(" </KEY_VALUE_SET>");
}
else if (y.BlockType == "TABLE")
{
Console.WriteLine(" <TABLE>");
PrintBlock(y);
Console.WriteLine(" </TABLE>");
}
else if (y.BlockType == "CELL")
{
Console.WriteLine(" <CELL>");
PrintBlock(y);
Console.WriteLine(" </CELL>");
}
Console.WriteLine("</block>");
});
}
// Gets the result of a PDF analysis request.
private async Task <GetDocumentAnalysisResponse> GetDocAnalysisResponse(StartDocumentAnalysisResponse response)
{
// Get jobID from start analysis response
var x = new GetDocumentAnalysisRequest();
x.JobId = response.JobId;
// Poll for analysis to finish. Can take over 15sec. to
// get results, thus the somewhat long delay.
GetDocumentAnalysisResponse res = await this.textractClient.GetDocumentAnalysisAsync(x);
int c = 0;
while (res.JobStatus != "SUCCEEDED")
{
await Task.Delay(200);
res = await this.textractClient.GetDocumentAnalysisAsync(x);
c++;
System.Diagnostics.Debug.WriteLine("Trying again.... " + res.JobStatus + " Attempt " + c);
}
return(res);
}
public TextractDocument(GetDocumentAnalysisResponse response)
{
Pages = new List <Page>();
var allPageBlocks = ParseDocumentPagesAndBlockMap(response);
Parse(allPageBlocks, 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)
{
GetDocumentAnalysisResponse response = new GetDocumentAnalysisResponse();
context.Read();
int targetDepth = context.CurrentDepth;
while (context.ReadAtDepth(targetDepth))
{
if (context.TestExpression("AnalyzeDocumentModelVersion", targetDepth))
{
var unmarshaller = StringUnmarshaller.Instance;
response.AnalyzeDocumentModelVersion = unmarshaller.Unmarshall(context);
continue;
}
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;
}
if (context.TestExpression("JobStatus", targetDepth))
{
var unmarshaller = StringUnmarshaller.Instance;
response.JobStatus = unmarshaller.Unmarshall(context);
continue;
}
if (context.TestExpression("NextToken", targetDepth))
{
var unmarshaller = StringUnmarshaller.Instance;
response.NextToken = unmarshaller.Unmarshall(context);
continue;
}
if (context.TestExpression("StatusMessage", targetDepth))
{
var unmarshaller = StringUnmarshaller.Instance;
response.StatusMessage = unmarshaller.Unmarshall(context);
continue;
}
if (context.TestExpression("Warnings", targetDepth))
{
var unmarshaller = new ListUnmarshaller <Warning, WarningUnmarshaller>(WarningUnmarshaller.Instance);
response.Warnings = unmarshaller.Unmarshall(context);
continue;
}
}
return(response);
}
public string GetRawText(GetDocumentAnalysisResponse response)
{
StringBuilder sb = new StringBuilder();
foreach (var element in response.Blocks)
{
if (element.BlockType == "LINE")
{
sb.AppendLine(element.Text);
}
}
return(sb.ToString());
}
// result can be stored in .tsv
// (which does not use , or ; as separator - which can be appeared in key/value)
public string CreateKeyValueText(GetDocumentAnalysisResponse response)
{
StringBuilder sb = new StringBuilder();
string seperator = "\t";
sb.AppendLine("Key" + seperator + "Value");
var document = new TextractDocument(response);
document.Pages.ForEach(page => {
page.Form.Fields.ForEach(f => {
sb.AppendLine(f.Key + seperator + f.Value);
});
});
return(sb.ToString());
}
private List <List <Block> > ParseDocumentPagesAndBlockMap(GetDocumentAnalysisResponse response)
{
var allPageBlocks = new List <List <Block> >();
var pageBlocks = new List <Block>();
// Inconsistancy in page number
var numberOfPages = response.DocumentMetadata.Pages;
var noOfPages = response.Blocks.Count(x => x.BlockType == BlockType.PAGE);
var noOfTable = response.Blocks.Count(x => x.BlockType == BlockType.TABLE);
foreach (var block in response.Blocks)
{
if (block.BlockType == BlockType.PAGE)
{
if (pageBlocks.Count > 0)
{
allPageBlocks.Add(pageBlocks);
}
pageBlocks = new List <Block> {
block
};
}
else
{
pageBlocks.Add(block);
}
}
if (pageBlocks != null)
{
allPageBlocks.Add(pageBlocks);
}
return(allPageBlocks);
}
