如何自动发票数据提取和验证

string[] invoiceFiles = Directory.GetFiles("./invoices", "*.pdf"); // or *.jpg, *.png

using Aspose.OCR;
List<string> errors = new List<string>();
InvoiceRecognitionSettings settings = new InvoiceRecognitionSettings();
settings.Language = Language.English;
AsposeOcr ocr = new AsposeOcr();

using (var writer = new StreamWriter("invoice_results.csv"))
{
    writer.WriteLine("File,Vendor,Date,Total,Status,Error");
    foreach (var file in invoiceFiles)
    {
        try
        {
            OcrInput input = new OcrInput(InputType.SingleImage);
            input.Add(file);
            var results = ocr.RecognizeInvoice(input, settings);
            var text = results[0].RecognitionText;
            // Example: Extract fields with regex or parsing
            string vendor = ExtractField(text, "Vendor:");
            string date = ExtractField(text, "Date:");
            string total = ExtractField(text, "Total:");
            bool valid = ValidateInvoiceData(vendor, date, total);
            writer.WriteLine($"{file},{vendor},{date},{total},{(valid ? "Valid" : "Invalid")},");
        }
        catch (Exception ex)
        {
            writer.WriteLine($"{file},,,,Error,{ex.Message}");
        }
    }
}
// Helper methods to extract and validate fields (simplified)
string ExtractField(string text, string fieldName)
{
    // Implement regex or logic to extract field from text
    return ""; // Example stub
}
bool ValidateInvoiceData(string vendor, string date, string total)
{
    // Implement checks for expected formats, totals, required fields
    return !string.IsNullOrEmpty(vendor) && !string.IsNullOrEmpty(date) && !string.IsNullOrEmpty(total);
}