StrangeTools/PdfExtractor/Program.cs

using System.Diagnostics;
using UglyToad.PdfPig;

if (args.Length != 2)
{
    Console.WriteLine("Err: 2 args required, <inputDir> <outputDir>");
    return 1;
}

var inputDir = args[0];
var outputDir = args[1];

if (!Directory.Exists(inputDir))
{
    Console.WriteLine("Err: inputDir dose not exist");
    return 1;
}

if (!Directory.Exists(outputDir))
{
    Console.WriteLine("Err: outputDir dose not exist");
    return 1;
}

Process.GetCurrentProcess().PriorityClass = ProcessPriorityClass.Idle;


var fullInputDir = Path.GetFullPath(inputDir);
var fullOutputDir = Path.GetFullPath(outputDir);

var filePathsToProcess = Directory.GetFiles(fullInputDir);

var signatureJpg = new byte[] { 0xFF, 0xD8, 0xFF };

void ProcessFunc(string inputFilePath)
{
    var outputFileDir = Path.Combine(fullOutputDir, Path.GetFileNameWithoutExtension(inputFilePath));

    if (Directory.Exists(outputFileDir) == false)
    {
        lock (Console.Out) Console.WriteLine("Create dir: " + outputFileDir);
        Directory.CreateDirectory(outputFileDir);
    }

    var pdfBytes = File.ReadAllBytes(inputFilePath);
    using var document = PdfDocument.Open(pdfBytes);
    var pages = document.GetPages().ToArray();
    for (var pageIndex = 0; pageIndex < pages.Length; pageIndex++)
    {
        var page = pages[pageIndex];

        var images = page.GetImages().ToArray();
        for (var imageIndex = 0; imageIndex < images.Length; imageIndex++)
        {
            var image = images[imageIndex];
            var outputFileName = $"P{pageIndex + 1:0000}_{imageIndex + 1:000}";
            if (image.TryGetPng(out var bytes))
            {
                var finalOutputPath = Path.Combine(outputFileDir, outputFileName + ".png");
                if (File.Exists(finalOutputPath))
                {
                    Console.WriteLine($"SKIP {finalOutputPath}");
                    continue;
                }

                File.WriteAllBytes(finalOutputPath, bytes);
                Console.WriteLine($"SAVE {finalOutputPath}");
            }
            else if (image.TryGetBytes(out var irlBytes))
            {
                var finalOutputPath = Path.Combine(outputFileDir, outputFileName + ".bin");
                if (File.Exists(finalOutputPath))
                {
                    Console.WriteLine($"SKIP {finalOutputPath}");
                    continue;
                }

                File.WriteAllBytes(finalOutputPath, bytes);
                Console.WriteLine($"SAVE {finalOutputPath}");
            }
            else
            {
                bytes = image.RawBytes.ToArray();
                var span = (ReadOnlySpan<byte>)bytes;
                var extName = ".bin";

                if (span.Length > signatureJpg.Length && span.StartsWith(signatureJpg))
                {
                    extName = ".jpg";
                }
                else
                {
                    int bp = 0;
                }

                var finalOutputPath = Path.Combine(outputFileDir, outputFileName + extName);
                if (File.Exists(finalOutputPath))
                {
                    Console.WriteLine($"SKIP {finalOutputPath}");
                    continue;
                }

                File.WriteAllBytes(finalOutputPath, bytes);
                Console.WriteLine($"SAVE {finalOutputPath}");
            }
        }
    }
}

Parallel.ForEach(filePathsToProcess, ProcessFunc);

// for debug
//foreach (var s in filePathsToProcess) ProcessFunc(s);

return 0;