Add ZUGFeRD detection and extraction functionality
Introduced `ZugferdController` to handle ZUGFeRD-related operations, including detection and extraction of ZUGFeRD XML from PDFs via file upload or Base64-encoded payloads. Integrated `MediatR` for query/command handling. Added `ZugferdSettings` for configurable file names and patterns, and updated `appsettings.json` and `Program.cs` to support this configuration. Implemented `HasZugferdQuery` and `ExtractZugferdCommand` with their respective handlers and validators. Added DTOs (`ZugferdCheckResult`, `ZugferdExtractionResult`) for operation results. Included `ZUGFeRD-Example.pdf` for testing and integrated `Serilog.Ui.Core.Extensions` for logging.
This commit is contained in:
@@ -0,0 +1,29 @@
|
||||
namespace DocumentOperator.Application.Common.Configuration;
|
||||
|
||||
/// <summary>
|
||||
/// Configuration settings for ZUGFeRD/Factur-X/XRechnung detection
|
||||
/// </summary>
|
||||
public class ZugferdSettings
|
||||
{
|
||||
/// <summary>
|
||||
/// Exact ZUGFeRD/Factur-X/XRechnung file names to check (case-insensitive)
|
||||
/// </summary>
|
||||
public List<string> ZugferdFileNames { get; set; } = new()
|
||||
{
|
||||
"factur-x.xml",
|
||||
"zugferd-invoice.xml",
|
||||
"ZUGFeRD-invoice.xml",
|
||||
"xrechnung.xml"
|
||||
};
|
||||
|
||||
/// <summary>
|
||||
/// Partial filename patterns for ZUGFeRD detection (case-insensitive)
|
||||
/// </summary>
|
||||
public List<string> ZugferdFileNamePatterns { get; set; } = new()
|
||||
{
|
||||
"factur",
|
||||
"zugferd",
|
||||
"xrechnung",
|
||||
"peppol"
|
||||
};
|
||||
}
|
||||
@@ -0,0 +1,27 @@
|
||||
namespace DocumentOperator.Application.Common.DTOs;
|
||||
|
||||
/// <summary>
|
||||
/// DTO for ZUGFeRD check result
|
||||
/// </summary>
|
||||
public record ZugferdCheckResult
|
||||
{
|
||||
/// <summary>
|
||||
/// Indicates whether the PDF contains ZUGFeRD XML attachment
|
||||
/// </summary>
|
||||
public bool HasZugferd { get; init; }
|
||||
|
||||
/// <summary>
|
||||
/// ZUGFeRD XML file name (e.g., "factur-x.xml")
|
||||
/// </summary>
|
||||
public string? ZugferdFileName { get; init; }
|
||||
|
||||
/// <summary>
|
||||
/// ZUGFeRD XML file size in bytes
|
||||
/// </summary>
|
||||
public long? ZugferdFileSize { get; init; }
|
||||
|
||||
/// <summary>
|
||||
/// MIME type of the ZUGFeRD XML file
|
||||
/// </summary>
|
||||
public string? ZugferdMimeType { get; init; }
|
||||
}
|
||||
@@ -0,0 +1,113 @@
|
||||
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;
|
||||
|
||||
/// <summary>
|
||||
/// Command to extract ZUGFeRD XML from a PDF document
|
||||
/// </summary>
|
||||
public record ExtractZugferdCommand : IRequest<ZugferdExtractionResult>
|
||||
{
|
||||
/// <summary>
|
||||
/// PDF document stream. Must be positioned at the beginning (Position = 0).
|
||||
/// </summary>
|
||||
public required Stream PdfStream { get; init; }
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// Handler for ExtractZugferdCommand.
|
||||
/// Extracts ZUGFeRD XML from PDF and returns XML content
|
||||
/// </summary>
|
||||
public class ExtractZugferdCommandHandler(
|
||||
IPdfProcessor pdfProcessor,
|
||||
IOptions<ZugferdSettings> settings)
|
||||
: IRequestHandler<ExtractZugferdCommand, ZugferdExtractionResult>
|
||||
{
|
||||
private readonly ZugferdSettings _settings = settings.Value;
|
||||
|
||||
public async Task<ZugferdExtractionResult> 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
|
||||
};
|
||||
}
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// Validator for ExtractZugferdCommand.
|
||||
/// </summary>
|
||||
public class ExtractZugferdCommandValidator : AbstractValidator<ExtractZugferdCommand>
|
||||
{
|
||||
public ExtractZugferdCommandValidator()
|
||||
{
|
||||
RuleFor(x => x.PdfStream)
|
||||
.NotNull()
|
||||
.WithMessage("PDF stream is required");
|
||||
}
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// Result DTO for ZUGFeRD extraction
|
||||
/// </summary>
|
||||
public record ZugferdExtractionResult
|
||||
{
|
||||
/// <summary>
|
||||
/// ZUGFeRD XML file name
|
||||
/// </summary>
|
||||
public required string FileName { get; init; }
|
||||
|
||||
/// <summary>
|
||||
/// ZUGFeRD XML content as string
|
||||
/// </summary>
|
||||
public required string XmlContent { get; init; }
|
||||
|
||||
/// <summary>
|
||||
/// File size in bytes
|
||||
/// </summary>
|
||||
public long FileSize { get; init; }
|
||||
}
|
||||
@@ -0,0 +1,62 @@
|
||||
using DocumentOperator.Application.Common.Configuration;
|
||||
using DocumentOperator.Application.Common.DTOs;
|
||||
using DocumentOperator.Application.Common.Interfaces;
|
||||
using MediatR;
|
||||
using Microsoft.Extensions.Options;
|
||||
|
||||
namespace DocumentOperator.Application.HasZugferd.Queries;
|
||||
|
||||
/// <summary>
|
||||
/// Query for checking if PDF contains ZUGFeRD XML attachment
|
||||
/// </summary>
|
||||
public record HasZugferdQuery : IRequest<ZugferdCheckResult>
|
||||
{
|
||||
/// <summary>
|
||||
/// PDF as stream (caller is responsible for disposal)
|
||||
/// </summary>
|
||||
public required Stream PdfStream { get; init; }
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// Handler for HasZugferdQuery
|
||||
/// Checks if PDF contains ZUGFeRD/Factur-X XML attachment
|
||||
/// </summary>
|
||||
public class HasZugferdQueryHandler(
|
||||
IPdfProcessor pdfProcessor,
|
||||
IOptions<ZugferdSettings> settings)
|
||||
: IRequestHandler<HasZugferdQuery, ZugferdCheckResult>
|
||||
{
|
||||
private readonly ZugferdSettings _settings = settings.Value;
|
||||
|
||||
/// <summary>
|
||||
/// Checks if PDF contains ZUGFeRD XML and returns metadata
|
||||
/// </summary>
|
||||
public async Task<ZugferdCheckResult> Handle(HasZugferdQuery request, CancellationToken cancellationToken)
|
||||
{
|
||||
// Get all attachments
|
||||
var attachmentInfo = await pdfProcessor.CheckAttachmentsAsync(request.PdfStream);
|
||||
|
||||
// Check for ZUGFeRD/Factur-X XML files 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)
|
||||
{
|
||||
return new ZugferdCheckResult
|
||||
{
|
||||
HasZugferd = true,
|
||||
ZugferdFileName = zugferdAttachment.FileName,
|
||||
ZugferdFileSize = zugferdAttachment.SizeBytes,
|
||||
ZugferdMimeType = zugferdAttachment.MimeType
|
||||
};
|
||||
}
|
||||
|
||||
return new ZugferdCheckResult
|
||||
{
|
||||
HasZugferd = false
|
||||
};
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,16 @@
|
||||
using FluentValidation;
|
||||
|
||||
namespace DocumentOperator.Application.HasZugferd.Queries;
|
||||
|
||||
/// <summary>
|
||||
/// Validator for HasZugferdQuery
|
||||
/// </summary>
|
||||
public class HasZugferdQueryValidator : AbstractValidator<HasZugferdQuery>
|
||||
{
|
||||
public HasZugferdQueryValidator()
|
||||
{
|
||||
RuleFor(x => x.PdfStream)
|
||||
.NotNull()
|
||||
.WithMessage("PDF stream is required");
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user