public static async Task<IEnumerable<string>> ExtractTemplateFromPdfAsync(FileInfo pdfFile, OcrTemplate ocrTemplate)
{
FileInfo pngFile = ConvertPdfToPng(pdfFile);
List<string> results = new List<string>();
foreach (var region in ocrTemplate.Regions)
{
var xml = await CallOcrWebServiceAsync(pngFile, region.TopLeftX, region.TopLeftY, region.BottomRightX, region.BottomRightY);
results.Add(xml);
}
return results;
}
public void TestInitialize()
{
_template = new OcrTemplate
{
CourtId = 1,
DocumentType = "Case Management Statement",
OcrTemplateId = 1
};
var caseNumberRegion = new OcrRegion(1561, 957, 2291, 1158, OcrRegionName.CaseNumber);
var plaintiffRegion = new OcrRegion(150, 830, 1640, 1010, OcrRegionName.Plaintiff);
var attorneyRegion = new OcrRegion(150, 200, 1640, 590, OcrRegionName.Attorney);
var hearingDateRegion = new OcrRegion(150, 1220, 2291, 1350, OcrRegionName.HearingDate);
_template.Regions.Add(caseNumberRegion);
_template.Regions.Add(plaintiffRegion);
_template.Regions.Add(attorneyRegion);
_template.Regions.Add(hearingDateRegion);
}
static void Main(string[] args)
{
var template = new OcrTemplate
{
CourtId = 1,
DocumentType = "Case Management Statement",
OcrTemplateId = 1
};
var caseNumberRegion = new OcrRegion(1561, 957, 2291, 1158, OcrRegionName.CaseNumber);
var plaintiffRegion = new OcrRegion(150, 830, 1640, 1010, OcrRegionName.Plaintiff);
var attorneyRegion = new OcrRegion(150, 200, 1640, 590, OcrRegionName.Attorney);
var hearingDateRegion = new OcrRegion(150, 1220, 2291, 1350, OcrRegionName.HearingDate);
template.Regions.Add(caseNumberRegion);
template.Regions.Add(plaintiffRegion);
template.Regions.Add(attorneyRegion);
template.Regions.Add(hearingDateRegion);
FileInfo pdfFile = new FileInfo("OLPDFS\\43159219.pdf");
var pngFile = CloudOcrService.ConvertPdfToPng(pdfFile);
var results = CloudOcrService.ExtractTemplateFromPdfAsync(pdfFile, template);
foreach (var str in results.Result)
{
}
Console.WriteLine(results.Result.Select(s => s));
foreach (var region in template.Regions)
{
Console.WriteLine(region.RegionName.ToString() + ": " + CloudOcrService.ExtractRegionFromPng(pngFile, region));
}
Console.ReadLine();
}
public static async Task <IEnumerable <string> > ExtractTemplateFromPdfAsync(FileInfo pdfFile, OcrTemplate ocrTemplate)
{
FileInfo pngFile = ConvertPdfToPng(pdfFile);
List <string> results = new List <string>();
foreach (var region in ocrTemplate.Regions)
{
var xml = await CallOcrWebServiceAsync(pngFile, region.TopLeftX, region.TopLeftY, region.BottomRightX, region.BottomRightY);
results.Add(xml);
}
return(results);
}
