使用 iText 7，导出 Flate 编码图像的正确方法是什么？

Question

我正在尝试使用 iText 7.19 版创建代码以导出 PDF 中的图像。我在使用 Flate 编码图像时遇到了一些问题。我用作示例的 Microsoft 免费书籍中的所有 Flate 编码图像（请参阅Moving to Microsoft Visual Studio 2010）总是呈粉红色，这取决于我如何尝试复制它们可能会失真的字节。

如果我尝试一次复制所有图像字节（请参阅下面代码中的 SaveFlateEncodedImage2 方法），它们会像这样扭曲：

如果我尝试逐行复制它们（请参阅下面代码中的 SaveFlateEncodedImage 方法），它们就像这个一样是粉红色的

这是我用来导出它们的代码：

using iText.Kernel;
using iText.Kernel.Pdf;
using iText.Kernel.Pdf.Filters;
using System;
using System.Drawing;
using System.Drawing.Imaging;
using System.IO;
using System.Runtime.InteropServices;

namespace ITextPdfStuff
{
    public class MyPdfImageExtractor
    {
        private readonly string _pdfFileName;

        public MyPdfImageExtractor(string pdfFileName)
        {
            _pdfFileName = pdfFileName;
        }

        public void ExtractToDirectory(string directoryName)
        {
            using (var reader = new PdfReader(_pdfFileName))
            {
                // Avoid iText.Kernel.Crypto.BadPasswordException: https://stackoverflow.com/a/48065052/97803
                reader.SetUnethicalReading(true);

                using (var pdfDoc = new PdfDocument(reader))
                {
                    ExtractImagesOnAllPages(pdfDoc, directoryName);
                }
            }
        }


        private void ExtractImagesOnAllPages(PdfDocument pdfDoc, string directoryName)
        {
            Console.WriteLine($"Number of pdf {pdfDoc.GetNumberOfPdfObjects()} objects");

            // Extract objects https://itextpdf.com/en/resources/examples/itext-7/extracting-objects-pdf
            for (int objNumber = 1; objNumber <= pdfDoc.GetNumberOfPdfObjects(); objNumber++)
            {

                PdfObject currentObject = pdfDoc.GetPdfObject(objNumber);

                if (currentObject != null && currentObject.IsStream())
                {
                    try
                    {                 
                        ExtractImagesOneImage(currentObject as PdfStream, Path.Combine(directoryName, $"image{objNumber}.png"));
                    }
                    catch (Exception ex)
                    {
                        Console.WriteLine($"Object number {objNumber} is NOT an image! -- error: {ex.Message}");
                    }
                }
            }
        }

        private void ExtractImagesOneImage(PdfStream someStream, string fileName)
        {
            var pdfDict = (PdfDictionary)someStream;
            string subType = pdfDict.Get(PdfName.Subtype)?.ToString() ?? string.Empty;

            bool isImage = subType == "/Image";

            if (isImage == false)
                return;

            bool decoded = false;


            string filter = pdfDict.Get(PdfName.Filter).ToString();

            if (filter == "/FlateDecode")
            {
                SaveFlateEncodedImage(fileName, pdfDict, someStream.GetBytes(false));
            }
            else
            {
                byte[] imgData;

                try
                {
                    imgData = someStream.GetBytes(decoded);
                }
                catch (PdfException ex)
                {
                    imgData = someStream.GetBytes(!decoded);
                }

                SaveNormalImage(fileName, imgData);
            }

        }

        private void SaveNormalImage(string fileName, byte[] imgData)
        {
            using (var memStream = new System.IO.MemoryStream(imgData))
            using (var image = System.Drawing.Image.FromStream(memStream))
            {
                image.Save(fileName, ImageFormat.Png);
                Console.WriteLine($"{Path.GetFileName(fileName)}");
            }
        }

        private void SaveFlateEncodedImage(string fileName, PdfDictionary pdfDict, byte[] imgData)
        {
            int width = int.Parse(pdfDict.Get(PdfName.Width).ToString());
            int height = int.Parse(pdfDict.Get(PdfName.Height).ToString());
            int bpp = int.Parse(pdfDict.Get(PdfName.BitsPerComponent).ToString());

            // Example that helped: https://stackoverflow.com/a/8517377/97803
            PixelFormat pixelFormat;
            switch (bpp)
            {
                case 1:
                    pixelFormat = PixelFormat.Format1bppIndexed;
                    break;
                case 8:
                    pixelFormat = PixelFormat.Format8bppIndexed;
                    break;
                case 24:
                    pixelFormat = PixelFormat.Format24bppRgb;
                    break;
                default:
                    throw new Exception("Unknown pixel format " + bpp);
            }

            // .NET docs https://api.itextpdf.com/iText7/dotnet/7.1.9/classi_text_1_1_kernel_1_1_pdf_1_1_filters_1_1_flate_decode_strict_filter.html
            // Java docs have more detail: https://api.itextpdf.com/iText7/java/7.1.7/com/itextpdf/kernel/pdf/filters/FlateDecodeFilter.html
            imgData = FlateDecodeStrictFilter.FlateDecode(imgData, true);
            //  byte[] streamBytes = FlateDecodeStrictFilter.DecodePredictor(imgData, pdfDict);

            // Copy the image one row at a time
            using (var bmp = new Bitmap(width, height, pixelFormat))
            {
                BitmapData bmpData = bmp.LockBits(new Rectangle(0, 0, width, height), ImageLockMode.WriteOnly, pixelFormat);

                int length = (int)Math.Ceiling(width * bpp / 8.0);
                for (int i = 0; i < height; i++)
                {
                    int offset = i * length;
                    int scanOffset = i * bmpData.Stride;
                    Marshal.Copy(imgData, offset, new IntPtr(bmpData.Scan0.ToInt64() + scanOffset), length);
                }

                bmp.UnlockBits(bmpData);
                bmp.Save(fileName, ImageFormat.Png);
            }

            Console.WriteLine($"FlateDecode! {Path.GetFileName(fileName)}");
        }


        /// <summary>This method distorts the image badly</summary>
        private void SaveFlateEncodedImage2(string fileName, PdfDictionary pdfDict, byte[] imgData)
        {
            int width = int.Parse(pdfDict.Get(PdfName.Width).ToString());
            int height = int.Parse(pdfDict.Get(PdfName.Height).ToString());
            int bpp = int.Parse(pdfDict.Get(PdfName.BitsPerComponent).ToString());

            // Example that helped: https://stackoverflow.com/a/8517377/97803
            PixelFormat pixelFormat;
            switch (bpp)
            {
                case 1:
                    pixelFormat = PixelFormat.Format1bppIndexed;
                    break;
                case 8:
                    pixelFormat = PixelFormat.Format8bppIndexed;
                    break;
                case 24:
                    pixelFormat = PixelFormat.Format24bppRgb;
                    break;
                default:
                    throw new Exception("Unknown pixel format " + bpp);
            }

            // .NET docs https://api.itextpdf.com/iText7/dotnet/7.1.9/classi_text_1_1_kernel_1_1_pdf_1_1_filters_1_1_flate_decode_strict_filter.html
            // Java docs have more detail: https://api.itextpdf.com/iText7/java/7.1.7/com/itextpdf/kernel/pdf/filters/FlateDecodeFilter.html
            imgData = FlateDecodeStrictFilter.FlateDecode(imgData, true);
            // byte[] streamBytes = FlateDecodeStrictFilter.DecodePredictor(imgData, pdfDict);

            // Copy the entire image in one go
            using (var bmp = new Bitmap(width, height, pixelFormat))
            {
                BitmapData bmpData = bmp.LockBits(new Rectangle(0, 0, width, height), ImageLockMode.WriteOnly, pixelFormat);
                Marshal.Copy(imgData, 0, bmpData.Scan0, imgData.Length);
                bmp.UnlockBits(bmpData);
                bmp.Save(fileName, ImageFormat.Png);
            }

            Console.WriteLine($"FlateDecode! {Path.GetFileName(fileName)}");
        }
    }
}

可以在 .NET Core 控制台应用程序中像这样实例化和调用代码：

  string existingFileName = @"c:\temp\ReallyLongBook1.pdf";
  var imageExtractor = new MyPdfImageExtractor(existingFileName);
  imageExtractor.ExtractToDirectory(@"c:\temp\images");

我正在通过此代码运行以下免费的 Microsoft 书籍： Moving to Microsoft Visual Studio 2010

有问题的图片在第 10 页，是黑白的（不是粉红色的）。

我不是 PDF 专家，几天来我一直在敲打这段代码，现在挑选了一些示例来尝试将其拼凑在一起。任何能让我超越粉红色图像的帮助，将不胜感激。

--------2020 年 2 月 4 日更新-----

这是 MKL 建议更改后的修订版本。他的更改提取了比我更多的图像，并生成了我上面提到的书中出现的外观正确的图像：

using iText.Kernel.Pdf;
using iText.Kernel.Pdf.Canvas.Parser;
using iText.Kernel.Pdf.Canvas.Parser.Data;
using iText.Kernel.Pdf.Canvas.Parser.Listener;
using iText.Kernel.Pdf.Xobject;
using System;
using System.Collections.Generic;
using System.IO;

namespace ITextPdfStuff
{
    public class MyPdfImageExtractor
    {
        private readonly string _pdfFileName;

        public MyPdfImageExtractor(string pdfFileName)
        {
            _pdfFileName = pdfFileName;
        }

        public void ExtractToDirectory(string directoryName)
        {
            using (var reader = new PdfReader(_pdfFileName))
            {
                // Avoid iText.Kernel.Crypto.BadPasswordException: https://stackoverflow.com/a/48065052/97803
                reader.SetUnethicalReading(true);

                using (var pdfDoc = new PdfDocument(reader))
                {
                    ExtractImagesOnAllPages(pdfDoc, directoryName);
                }
            }
        }

        private void ExtractImagesOnAllPages(PdfDocument pdfDoc, string directoryName)
        {
            Console.WriteLine($"Number of pdf {pdfDoc.GetNumberOfPdfObjects()} objects");

            IEventListener strategy = new ImageRenderListener(Path.Combine(directoryName, @"image{0}.{1}"));
            PdfCanvasProcessor parser = new PdfCanvasProcessor(strategy);
            for (var i = 1; i <= pdfDoc.GetNumberOfPages(); i++)
            {
                parser.ProcessPageContent(pdfDoc.GetPage(i));
            }
        }
    }


    public class ImageRenderListener : IEventListener
    {
        public ImageRenderListener(string format)
        {
            this.format = format;
        }

        public void EventOccurred(IEventData data, EventType type)
        {
            if (data is ImageRenderInfo imageData)
            {
                try
                {
                    PdfImageXObject imageObject = imageData.GetImage();
                    if (imageObject == null)
                    {
                        Console.WriteLine("Image could not be read.");
                    }
                    else
                    {
                        File.WriteAllBytes(string.Format(format, index++, imageObject.IdentifyImageFileExtension()), imageObject.GetImageBytes());
                    }
                }
                catch (Exception ex)
                {
                    Console.WriteLine("Image could not be read: {0}.", ex.Message);
                }
            }
        }

        public ICollection<EventType> GetSupportedEvents()
        {
            return null;
        }

        string format;
        int index = 0;
    }
}

Answer 1

PDF 在内部支持一种非常灵活的位图图像格式，尤其是在涉及不同颜色空间方面。

iText 在其解析 API 中支持导出其中的子集，本质上是可以轻松导出为常规 JPEG 或 PNG 的图像子集。

因此，首先尝试使用 iText 解析 API 进行导出是有意义的。你可以这样做：

Directory.CreateDirectory(@"extract\");
using (PdfReader reader = new PdfReader(@"Moving to Microsoft Visual Studio 2010 ebook.pdf"))
using (PdfDocument pdfDocument = new PdfDocument(reader))
{
    IEventListener strategy = new ImageRenderListener(@"extract\Moving to Microsoft Visual Studio 2010 ebook-i7-{0}.{1}");
    PdfCanvasProcessor parser = new PdfCanvasProcessor(strategy);
    for (var i = 1; i <= pdfDocument.GetNumberOfPages(); i++)
    {
        parser.ProcessPageContent(pdfDocument.GetPage(i));
    }
}

使用助手类ImageRenderListener:

public class ImageRenderListener : IEventListener
{
    public ImageRenderListener(string format)
    {
        this.format = format;
    }

    public void EventOccurred(IEventData data, EventType type)
    {
        if (data is ImageRenderInfo imageData)
        {
            try
            {
                PdfImageXObject imageObject = imageData.GetImage();
                if (imageObject == null)
                {
                    Console.WriteLine("Image could not be read.");
                }
                else
                {
                    File.WriteAllBytes(string.Format(format, index++, imageObject.IdentifyImageFileExtension()), imageObject.GetImageBytes());
                }
            }
            catch (Exception ex)
            {
                Console.WriteLine("Image could not be read: {0}.", ex.Message);
            }
        }
    }

    public ICollection<EventType> GetSupportedEvents()
    {
        return null;
    }

    string format;
    int index = 0;
}

对于您的示例文档，它成功导出了近 400 张图片，其中包括您上面的示例图片：

但也有不到 30 张图像无法导出，在标准输出中您会发现“无法读取图像：不支持颜色空间 /DeviceN..”