在RTF中提取嵌入式图像对象

时间:2013-02-08 19:08:35

标签: c# .net image rtf

我有rtf个包含嵌入对象(图像)的文档。我需要将其提取为Image对象(或任何其他可用格式)。我已经检查了这个CodeProject article,但默认的应用程序没有正确呈现它们(它们呈现'默认图像'图像,而不是图像本身),所以我继续前进。

This is a screen of the RTF in WordPad

以下是RTF代码的示例(由于大小,我不得不缩短它):

{\rtf1\ansi\deff0{\fonttbl{\f0\fnil\fcharset0 MS Sans Serif;}}
\viewkind4\uc1\pard\lang1033\f0\fs18{\object\objemb{\*\objclass Package}\objw855\objh810{\*\objdata 
01050000
02000000
08000000
5061636b61676500
00000000
00000000
1f900000
02007369675f5f2e6a706700433a5c55736572735c726563657074696f6e5c4465736b746f705c
5369676e6174757265735c7369675f5f2e6a7067000000030034000000433a5c55736572735c52
45434550547e315c417070446174615c4c6f63616c5c54656d705c7369675f5f20283132292e6a
706700c18e0000ffd8ffe000104a46494600010101004800470000ffdb00430001010101010101
010101010101010101010101010101010101010101010101010101010101010101010101010101
010101010101010101010101010101010101ffdb00430101010101010101010101010101010101
010101010101010101010101010101010101010101010101010101010101010101010101010101
010101010101010101ffc0001108012c03e803012200021101031101ffc4001f00010002030002
0301000000000000000000090a07080b050602030401ffc4003f10000006030001040201030301
04070900000203040506010708090a11121314152116172223314118192532591a24576598d6d8
2933384651788497b7ffc4001a010101000301010000000000000000000000030204050106ffc4
002b11010003010100020103030402030000000002030401051112130614211522230731415124
32536162ffda000c03010002110311003f00bfc000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000000000000000000000
...
005c0072006500630065007000740069006f006e005c004400650073006b0074006f0070005c00
5300690067006e006100740075007200650073005c007300690067005f005f002e006a00700067
00
01050000
00000000
}{\result{\pict\wmetafile8\picw2010\pich1905\picwgoal855\pichgoal810 
0100090000033b0700000200210600000000050000000b0200000000050000000c02350038001c
000000fb02f4ff000000000000900100000001000000005365676f65205549000e0a52104c2308
00dd1900d894ef758001f3758d0e664a040000002d010000050000000902000000000500000001
02ffffff00a5000000410bc600880020002000000000002000200000000c002800000020000000
400000000100010000000000000100000000000000000000000000000000000000000000ffffff
...
0021001c001c000000fb021000070000000000bc02000000000102022253797374656d00008d0e
664a00000a0022008a0100000000ffffffff8cdd1900040000002d010100030000000000
}}}\par
}

3 个答案:

答案 0 :(得分:15)

这是一段可以从RTF流中提取所有对象('Package'类对象)的代码:

    public static void ExtractPackageObjects(string filePath)
    {
        using (StreamReader sr = new StreamReader(filePath))
        {
            RtfReader reader = new RtfReader(sr);
            IEnumerator<RtfObject> enumerator = reader.Read().GetEnumerator();
            while(enumerator.MoveNext())
            {
                if (enumerator.Current.Text == "object")
                {
                    if (RtfReader.MoveToNextControlWord(enumerator, "objclass"))
                    {
                        string className = RtfReader.GetNextText(enumerator);
                        if (className == "Package")
                        {
                            if (RtfReader.MoveToNextControlWord(enumerator, "objdata"))
                            {
                                byte[] data = RtfReader.GetNextTextAsByteArray(enumerator);
                                using (MemoryStream packageData = new MemoryStream())
                                {
                                    RtfReader.ExtractObjectData(new MemoryStream(data), packageData);
                                    packageData.Position = 0;
                                    PackagedObject po = PackagedObject.Extract(packageData);
                                    File.WriteAllBytes(po.DisplayName, po.Data);
                                }
                            }
                        }
                    }
                }
            }
        }
    }

以下是此代码使用的实用程序类。有一个简单的基于流的RTF解析器,允许访问有趣的控制字。

还有一个实用程序可从序列化的Object Packager实例中提取数据。 Object Packager是一个差不多20年前OLE1.0的事情,序列化的二进制格式没有记录(据我所知),但这是可以理解的。

这适用于您提供的示例,但您可能需要调整其中的内容。

public class RtfReader
{
    public RtfReader(TextReader reader)
    {
        if (reader == null)
            throw new ArgumentNullException("reader");

        Reader = reader;
    }

    public TextReader Reader { get; private set; }

    public IEnumerable<RtfObject> Read()
    {
        StringBuilder controlWord = new StringBuilder();
        StringBuilder text = new StringBuilder();
        Stack<RtfParseState> stack = new Stack<RtfParseState>();
        RtfParseState state = RtfParseState.Group;

        do
        {
            int i = Reader.Read();
            if (i < 0)
            {
                if (!string.IsNullOrWhiteSpace(controlWord.ToString()))
                    yield return new RtfControlWord(controlWord.ToString());

                if (!string.IsNullOrWhiteSpace(text.ToString()))
                    yield return new RtfText(text.ToString());

                yield break;
            }

            char c = (char)i;

            // noise chars
            if ((c == '\r') ||
                (c == '\n'))
                continue;

            switch (state)
            {
                case RtfParseState.Group:
                    if (c == '{')
                    {
                        stack.Push(state);
                        break;
                    }

                    if (c == '\\')
                    {
                        state = RtfParseState.ControlWord;
                        break;
                    }
                    break;

                case RtfParseState.ControlWord:
                    if (c == '\\')
                    {
                        // another controlWord
                        if (!string.IsNullOrWhiteSpace(controlWord.ToString()))
                        {
                            yield return new RtfControlWord(controlWord.ToString());
                            controlWord.Clear();
                        }
                        break;
                    }

                    if (c == '{')
                    {
                        // a new group
                        state = RtfParseState.Group;
                        if (!string.IsNullOrWhiteSpace(controlWord.ToString()))
                        {
                            yield return new RtfControlWord(controlWord.ToString());
                            controlWord.Clear();
                        }
                        break;
                    }

                    if (c == '}')
                    {
                        // close group
                        state = stack.Count > 0 ? stack.Pop() : RtfParseState.Group;
                        if (!string.IsNullOrWhiteSpace(controlWord.ToString()))
                        {
                            yield return new RtfControlWord(controlWord.ToString());
                            controlWord.Clear();
                        }
                        break;
                    }

                    if (!Char.IsLetterOrDigit(c))
                    {
                        state = RtfParseState.Text;
                        text.Append(c);
                        if (!string.IsNullOrWhiteSpace(controlWord.ToString()))
                        {
                            yield return new RtfControlWord(controlWord.ToString());
                            controlWord.Clear();
                        }
                        break;
                    }

                    controlWord.Append(c);
                    break;

                case RtfParseState.Text:
                    if (c == '\\')
                    {
                        state = RtfParseState.EscapedText;
                        break;
                    }

                    if (c == '{')
                    {
                        if (!string.IsNullOrWhiteSpace(text.ToString()))
                        {
                            yield return new RtfText(text.ToString());
                            text.Clear();
                        }

                        // a new group
                        state = RtfParseState.Group;
                        break;
                    }

                    if (c == '}')
                    {
                        if (!string.IsNullOrWhiteSpace(text.ToString()))
                        {
                            yield return new RtfText(text.ToString());
                            text.Clear();
                        }

                        // close group
                        state = stack.Count > 0 ? stack.Pop() : RtfParseState.Group;
                        break;
                    }
                    text.Append(c);
                    break;

                case RtfParseState.EscapedText:
                    if ((c == '\\') || (c == '}') || (c == '{'))
                    {
                        state = RtfParseState.Text;
                        text.Append(c);
                        break;
                    }

                    // ansi character escape
                    if (c == '\'')
                    {
                        text.Append(FromHexa((char)Reader.Read(), (char)Reader.Read()));
                        break;
                    }

                    if (!string.IsNullOrWhiteSpace(text.ToString()))
                    {
                        yield return new RtfText(text.ToString());
                        text.Clear();
                    }

                    // in fact, it's a normal controlWord
                    controlWord.Append(c);
                    state = RtfParseState.ControlWord;
                    break;
            }
        }
        while (true);
    }

    public static bool MoveToNextControlWord(IEnumerator<RtfObject> enumerator, string word)
    {
        if (enumerator == null)
            throw new ArgumentNullException("enumerator");

        while (enumerator.MoveNext())
        {
            if (enumerator.Current.Text == word)
                return true;
        }
        return false;
    }

    public static string GetNextText(IEnumerator<RtfObject> enumerator)
    {
        if (enumerator == null)
            throw new ArgumentNullException("enumerator");

        while (enumerator.MoveNext())
        {
            RtfText text = enumerator.Current as RtfText;
            if (text != null)
                return text.Text;
        }
        return null;
    }

    public static byte[] GetNextTextAsByteArray(IEnumerator<RtfObject> enumerator)
    {
        if (enumerator == null)
            throw new ArgumentNullException("enumerator");

        while (enumerator.MoveNext())
        {
            RtfText text = enumerator.Current as RtfText;
            if (text != null)
            {
                List<byte> bytes = new List<byte>();
                for (int i = 0; i < text.Text.Length; i += 2)
                {
                    bytes.Add((byte)FromHexa(text.Text[i], text.Text[i + 1]));
                }
                return bytes.ToArray();
            }
        }
        return null;
    }

    // Extracts an EmbeddedObject/ObjectHeader from a stream
    // see [MS -OLEDS]: Object Linking and Embedding (OLE) Data Structures for more information
    // chapter 2.2: OLE1.0 Format Structures 
    public static void ExtractObjectData(Stream inputStream, Stream outputStream)
    {
        if (inputStream == null)
            throw new ArgumentNullException("inputStream");

        if (outputStream == null)
            throw new ArgumentNullException("outputStream");

        BinaryReader reader = new BinaryReader(inputStream);
        reader.ReadInt32(); // OLEVersion
        int formatId = reader.ReadInt32(); // FormatID
        if (formatId != 2) // see 2.2.4 Object Header. 2 means EmbeddedObject
            throw new NotSupportedException();

        ReadLengthPrefixedAnsiString(reader); // className
        ReadLengthPrefixedAnsiString(reader); // topicName
        ReadLengthPrefixedAnsiString(reader); // itemName

        int nativeDataSize = reader.ReadInt32();
        byte[] bytes = reader.ReadBytes(nativeDataSize);
        outputStream.Write(bytes, 0, bytes.Length);
    }

    // see chapter 2.1.4 LengthPrefixedAnsiString
    private static string ReadLengthPrefixedAnsiString(BinaryReader reader)
    {
        int length = reader.ReadInt32();
        if (length == 0)
            return string.Empty;

        byte[] bytes = reader.ReadBytes(length);
        return Encoding.Default.GetString(bytes, 0, length - 1);
    }

    private enum RtfParseState
    {
        ControlWord,
        Text,
        EscapedText,
        Group
    }

    private static char FromHexa(char hi, char lo)
    {
        return (char)byte.Parse(hi.ToString() + lo, NumberStyles.HexNumber);
    }
}

// Utility class to parse an OLE1.0 OLEOBJECT
public class PackagedObject
{
    private PackagedObject()
    {
    }

    public string DisplayName { get; private set; }
    public string IconFilePath { get; private set; }
    public int IconIndex { get; private set; }
    public string FilePath { get; private set; }
    public byte[] Data { get; private set; }

    private static string ReadAnsiString(BinaryReader reader)
    {
        StringBuilder sb = new StringBuilder();
        do
        {
            byte b = reader.ReadByte();
            if (b == 0)
                return sb.ToString();

            sb.Append((char)b);
        }
        while (true);
    }

    public static PackagedObject Extract(Stream inputStream)
    {
        if (inputStream == null)
            throw new ArgumentNullException("inputStream");

        BinaryReader reader = new BinaryReader(inputStream);
        reader.ReadUInt16(); // sig
        PackagedObject po = new PackagedObject();
        po.DisplayName = ReadAnsiString(reader);
        po.IconFilePath = ReadAnsiString(reader);
        po.IconIndex = reader.ReadUInt16();
        int type = reader.ReadUInt16();
        if (type != 3) // 3 is file, 1 is link
            throw new NotSupportedException();

        reader.ReadInt32(); // nextsize
        po.FilePath = ReadAnsiString(reader);
        int dataSize = reader.ReadInt32();
        po.Data = reader.ReadBytes(dataSize);
        // note after that, there may be unicode + long path info
        return po;
    }
}

public class RtfObject
{
    public RtfObject(string text)
    {
        if (text == null)
            throw new ArgumentNullException("text");

        Text = text.Trim();
    }

    public string Text { get; private set; }
}

public class RtfText : RtfObject
{
    public RtfText(string text)
        : base(text)
    {
    }
}

public class RtfControlWord : RtfObject
{
    public RtfControlWord(string name)
        : base(name)
    {
    }
}

答案 1 :(得分:5)

好的,这应该适合你。为了演示我的解决方案,我创建了一个带有PictureBox的WinForms项目,其绘图事件处理程序映射到以下函数:

 private void rtfImage_Paint(object sender, PaintEventArgs e)
    {
        string rtfStr = System.IO.File.ReadAllText("MySampleFile.rtf");
        string imageDataHex = ExtractImgHex(rtfStr);
        byte[] imageBuffer = ToBinary(imageDataHex);
        Image image;
        using (MemoryStream stream = new MemoryStream(imageBuffer))
        {
            image = Image.FromStream(stream);
        }
        Rectangle rect = new Rectangle(0, 0, 100, 100);
        e.Graphics.DrawImage(image, rect);                        
    }

此代码依赖于System.Drawing.Image.FromStream()方法以及两个“辅助”函数:

字符串提取器:

    string ExtractImgHex(string s)
    {
        // I'm sure you could use regex here, but this works.
        // This assumes one picture per file; loops required otherwise
        int pictTagIdx = s.IndexOf("{\\pict\\");
        int startIndex = s.IndexOf(" ", pictTagIdx)+1;
        int endIndex = s.IndexOf("}", startIndex);
        return s.Substring(startIndex, endIndex - startIndex);
    }

...和二进制转换器:

    public static byte[] ToBinary(string imageDataHex)
    {
        //this function taken entirely from:
        // http://www.codeproject.com/Articles/27431/Writing-Your-Own-RTF-Converter
        if (imageDataHex == null)
        {
            throw new ArgumentNullException("imageDataHex");
        }

        int hexDigits = imageDataHex.Length;
        int dataSize = hexDigits / 2;
        byte[] imageDataBinary = new byte[dataSize];

        StringBuilder hex = new StringBuilder(2);

        int dataPos = 0;
        for (int i = 0; i < hexDigits; i++)
        {
            char c = imageDataHex[i];
            if (char.IsWhiteSpace(c))
            {
                continue;
            }
            hex.Append(imageDataHex[i]);
            if (hex.Length == 2)
            {
                imageDataBinary[dataPos] = byte.Parse(hex.ToString(), System.Globalization.NumberStyles.HexNumber);
                dataPos++;
                hex.Remove(0, 2);
            }
        }
        return imageDataBinary;
    }

答案 2 :(得分:1)

下面的代码可以提取所有类型的嵌入式对象。包括带有原始文件名的图片/文档/邮件等。并将它们保存在本地路径中。

string MyDir = @"E:\temp\";
Document doc = new Document(MyDir + "Requirement#4.rtf");

NodeCollection nodeColl = doc.GetChildNodes(NodeType.Shape, true);
foreach (var node in nodeColl)
{
    Shape shape1 = (Shape)node;
    if (shape1.OleFormat != null)
    {
        shape1.OleFormat.Save(MyDir + shape1.OleFormat.SuggestedFileName + shape1.OleFormat.SuggestedExtension);
    }
}