DigitalData.MessagingService/src/DigitalData.EmailProfiler.Infrastructure/Services/DevExpressPdfProcessingService.cs

using DevExpress.Pdf;
using DigitalData.EmailProfiler.Application.Common.Interfaces;
using DigitalData.EmailProfiler.Domain.Exceptions;

namespace DigitalData.EmailProfiler.Infrastructure.Services;

/// <summary>
/// PDF processing service using DevExpress.Pdf.
/// Implements PDF validation and embedded file extraction using streams.
/// </summary>
public class DevExpressPdfProcessingService : IPdfProcessingService
{
    public Task<bool> ValidatePdfAsync(Stream pdfStream, CancellationToken cancellationToken = default)
    {
        ArgumentNullException.ThrowIfNull(pdfStream);

        if (!pdfStream.CanRead)
            throw new ArgumentException("Stream must be readable.", nameof(pdfStream));

        if (!pdfStream.CanSeek)
            throw new ArgumentException("Stream must be seekable.", nameof(pdfStream));

        if (pdfStream.Position != 0)
            pdfStream.Position = 0;

        using var processor = new PdfDocumentProcessor();
        processor.LoadDocument(pdfStream);
        
        return Task.FromResult(true);
    }

    public async Task<IEnumerable<string>> ExtractEmbeddedFilesAsync(
        Stream pdfStream,
        string outputDirectory,
        CancellationToken cancellationToken = default)
    {
        ArgumentNullException.ThrowIfNull(pdfStream);
        ArgumentException.ThrowIfNullOrWhiteSpace(outputDirectory);

        if (!pdfStream.CanRead)
            throw new ArgumentException("Stream must be readable.", nameof(pdfStream));

        if (!pdfStream.CanSeek)
            throw new ArgumentException("Stream must be seekable.", nameof(pdfStream));

        if (pdfStream.Position != 0)
            pdfStream.Position = 0;

        if (!Directory.Exists(outputDirectory))
            Directory.CreateDirectory(outputDirectory);

        using var processor = new PdfDocumentProcessor();
        processor.LoadDocument(pdfStream);

        var extractedFiles = new List<string>();
        var attachments = processor.Document.FileAttachments;

        if (attachments == null || !attachments.Any())
            return extractedFiles;

        foreach (var attachment in attachments)
        {
            var fileName = attachment.FileName ?? $"attachment_{Guid.NewGuid()}.dat";
            var outputPath = Path.Combine(outputDirectory, fileName);

            var fileData = attachment.Data;
            if (fileData == null || fileData.Length == 0)
                continue;

            await File.WriteAllBytesAsync(outputPath, fileData, cancellationToken);
            extractedFiles.Add(outputPath);
        }

        return extractedFiles;
    }

    public Task<int> GetPageCountAsync(Stream pdfStream, CancellationToken cancellationToken = default)
    {
        ArgumentNullException.ThrowIfNull(pdfStream);

        if (!pdfStream.CanRead)
            throw new ArgumentException("Stream must be readable.", nameof(pdfStream));

        if (!pdfStream.CanSeek)
            throw new ArgumentException("Stream must be seekable.", nameof(pdfStream));

        if (pdfStream.Position != 0)
            pdfStream.Position = 0;

        using var processor = new PdfDocumentProcessor();
        processor.LoadDocument(pdfStream);

        return Task.FromResult(processor.Document.Pages.Count);
    }
}