using DocumentOperator.Application.Common.Configuration;
using DocumentOperator.Application.Common.Interfaces;
using DocumentOperator.Domain.Common.Exceptions;
using FluentValidation;
using MediatR;
using Microsoft.Extensions.Options;
namespace DocumentOperator.Application.ExtractZugferd;
///
/// Command to extract ZUGFeRD XML from a PDF document
///
public record ExtractZugferdCommand : IRequest
{
///
/// PDF document stream. Must be positioned at the beginning (Position = 0).
///
public required Stream PdfStream { get; init; }
}
///
/// Handler for ExtractZugferdCommand.
/// Extracts ZUGFeRD XML from PDF and returns XML content
///
public class ExtractZugferdCommandHandler(
IPdfProcessor pdfProcessor,
IOptions settings)
: IRequestHandler
{
private readonly ZugferdSettings _settings = settings.Value;
public async Task Handle(ExtractZugferdCommand request, CancellationToken cancellationToken)
{
// Get all attachments
var attachmentInfo = await pdfProcessor.CheckAttachmentsAsync(request.PdfStream);
// Find ZUGFeRD XML file using configured names and patterns
var zugferdAttachment = attachmentInfo.Attachments.FirstOrDefault(a =>
_settings.ZugferdFileNames.Any(name =>
a.FileName.Equals(name, StringComparison.OrdinalIgnoreCase)) ||
_settings.ZugferdFileNamePatterns.Any(pattern =>
a.FileName.Contains(pattern, StringComparison.OrdinalIgnoreCase)));
if (zugferdAttachment == null)
{
throw new NotFoundException("ZUGFeRD XML not found in PDF attachments");
}
// Reset stream position for extraction
request.PdfStream.Position = 0;
// Extract all attachments as ZIP
byte[] zipBytes = await pdfProcessor.ExtractAttachmentsAsync(request.PdfStream);
// Find ZUGFeRD XML in ZIP
using var zipStream = new MemoryStream(zipBytes);
using var zipArchive = new System.IO.Compression.ZipArchive(zipStream, System.IO.Compression.ZipArchiveMode.Read);
var zugferdEntry = zipArchive.Entries.FirstOrDefault(e =>
e.Name.Equals(zugferdAttachment.FileName, StringComparison.OrdinalIgnoreCase));
if (zugferdEntry == null)
{
throw new NotFoundException($"ZUGFeRD XML '{zugferdAttachment.FileName}' not found in extracted attachments");
}
// Read XML content
using var entryStream = zugferdEntry.Open();
using var reader = new StreamReader(entryStream);
string xmlContent = await reader.ReadToEndAsync(cancellationToken);
return new ZugferdExtractionResult
{
FileName = zugferdAttachment.FileName,
XmlContent = xmlContent,
FileSize = zugferdAttachment.SizeBytes
};
}
}
///
/// Validator for ExtractZugferdCommand.
///
public class ExtractZugferdCommandValidator : AbstractValidator
{
public ExtractZugferdCommandValidator()
{
RuleFor(x => x.PdfStream)
.NotNull()
.WithMessage("PDF stream is required");
}
}
///
/// Result DTO for ZUGFeRD extraction
///
public record ZugferdExtractionResult
{
///
/// ZUGFeRD XML file name
///
public required string FileName { get; init; }
///
/// ZUGFeRD XML content as string
///
public required string XmlContent { get; init; }
///
/// File size in bytes
///
public long FileSize { get; init; }
}