RolemasterDB/src/RolemasterDb.ImportTool/PdfXmlExtractor.cs

using System.Diagnostics;

namespace RolemasterDb.ImportTool;

public sealed class PdfXmlExtractor
{
    public const int XmlAlignedRenderDpi = 108;

    public async Task ExtractAsync(string pdfPath, string outputPath, CancellationToken cancellationToken = default)
    {
        Directory.CreateDirectory(Path.GetDirectoryName(outputPath)!);

        var startInfo = new ProcessStartInfo
        {
            FileName = "pdftohtml",
            RedirectStandardError = true,
            RedirectStandardOutput = true,
            UseShellExecute = false,
            CreateNoWindow = true
        };

        startInfo.ArgumentList.Add("-xml");
        startInfo.ArgumentList.Add("-i");
        startInfo.ArgumentList.Add("-noframes");
        startInfo.ArgumentList.Add(pdfPath);
        startInfo.ArgumentList.Add(outputPath);

        using var process = new Process { StartInfo = startInfo };
        process.Start();
        await process.WaitForExitAsync(cancellationToken);

        if (process.ExitCode != 0)
        {
            var error = await process.StandardError.ReadToEndAsync(cancellationToken);
            throw new InvalidOperationException($"pdftohtml failed for '{pdfPath}': {error}");
        }
    }

    public Task RenderPagePngAsync(
        string pdfPath,
        int pageNumber,
        string outputPath,
        CancellationToken cancellationToken = default) =>
        RenderPngAsync(pdfPath, pageNumber, outputPath, null, null, null, null, cancellationToken);

    public Task RenderCropPngAsync(
        string pdfPath,
        int pageNumber,
        int left,
        int top,
        int width,
        int height,
        string outputPath,
        CancellationToken cancellationToken = default) =>
        RenderPngAsync(pdfPath, pageNumber, outputPath, left, top, width, height, cancellationToken);

    private static async Task RenderPngAsync(
        string pdfPath,
        int pageNumber,
        string outputPath,
        int? left,
        int? top,
        int? width,
        int? height,
        CancellationToken cancellationToken)
    {
        Directory.CreateDirectory(Path.GetDirectoryName(outputPath)!);

        var startInfo = new ProcessStartInfo
        {
            FileName = "pdftoppm",
            RedirectStandardError = true,
            RedirectStandardOutput = true,
            UseShellExecute = false,
            CreateNoWindow = true
        };

        startInfo.ArgumentList.Add("-png");
        startInfo.ArgumentList.Add("-r");
        startInfo.ArgumentList.Add(XmlAlignedRenderDpi.ToString());
        startInfo.ArgumentList.Add("-f");
        startInfo.ArgumentList.Add(pageNumber.ToString());
        startInfo.ArgumentList.Add("-l");
        startInfo.ArgumentList.Add(pageNumber.ToString());
        startInfo.ArgumentList.Add("-singlefile");

        if (left.HasValue && top.HasValue && width.HasValue && height.HasValue)
        {
            startInfo.ArgumentList.Add("-x");
            startInfo.ArgumentList.Add(left.Value.ToString());
            startInfo.ArgumentList.Add("-y");
            startInfo.ArgumentList.Add(top.Value.ToString());
            startInfo.ArgumentList.Add("-W");
            startInfo.ArgumentList.Add(width.Value.ToString());
            startInfo.ArgumentList.Add("-H");
            startInfo.ArgumentList.Add(height.Value.ToString());
        }

        startInfo.ArgumentList.Add(pdfPath);
        startInfo.ArgumentList.Add(Path.Combine(Path.GetDirectoryName(outputPath)!, Path.GetFileNameWithoutExtension(outputPath)));

        using var process = new Process { StartInfo = startInfo };
        process.Start();
        await process.WaitForExitAsync(cancellationToken);

        if (process.ExitCode != 0)
        {
            var error = await process.StandardError.ReadToEndAsync(cancellationToken);
            throw new InvalidOperationException($"pdftoppm failed for '{pdfPath}': {error}");
        }

        if (!File.Exists(outputPath))
        {
            throw new InvalidOperationException($"pdftoppm completed but did not create '{outputPath}'.");
        }
    }
}