Extraire un object image incorporé au format RTF

Voici un morceau de code capable d’extraire tous les objects (objects de classe ‘Package’) d’un stream RTF:

  public static void ExtractPackageObjects(ssortingng filePath) { using (StreamReader sr = new StreamReader(filePath)) { RtfReader reader = new RtfReader(sr); IEnumerator enumerator = reader.Read().GetEnumerator(); while(enumerator.MoveNext()) { if (enumerator.Current.Text == "object") { if (RtfReader.MoveToNextControlWord(enumerator, "objclass")) { ssortingng 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); } } } } } } } } 

Et voici les classes utilitaires utilisées par ce code. Il existe un parsingur RTF simple basé sur le stream qui permet d’obtenir les mots de contrôle intéressants.

Il existe également un utilitaire permettant d’extraire des données d’une instance sérialisée Object Packager. Object Packager est une chose OLE1.0 datant d’ il y a presque 20 ans et le format binary sérialisé n’est pas documenté (à ma connaissance), mais c’est compréhensible.

Cela fonctionne bien sur votre échantillon fourni, mais vous devrez peut-être adapter les choses.

 public class RtfReader { public RtfReader(TextReader reader) { if (reader == null) throw new ArgumentNullException("reader"); Reader = reader; } public TextReader Reader { get; private set; } public IEnumerable Read() { SsortingngBuilder controlWord = new SsortingngBuilder(); SsortingngBuilder text = new SsortingngBuilder(); Stack stack = new Stack(); 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 (!ssortingng.IsNullOrWhiteSpace(controlWord.ToSsortingng())) { yield return new RtfControlWord(controlWord.ToSsortingng()); controlWord.Clear(); } break; } if (!Char.IsLetterOrDigit(c)) { state = RtfParseState.Text; text.Append(c); if (!ssortingng.IsNullOrWhiteSpace(controlWord.ToSsortingng())) { yield return new RtfControlWord(controlWord.ToSsortingng()); controlWord.Clear(); } break; } controlWord.Append(c); break; case RtfParseState.Text: if (c == '\\') { state = RtfParseState.EscapedText; break; } if (c == '{') { if (!ssortingng.IsNullOrWhiteSpace(text.ToSsortingng())) { yield return new RtfText(text.ToSsortingng()); text.Clear(); } // a new group state = RtfParseState.Group; break; } if (c == '}') { if (!ssortingng.IsNullOrWhiteSpace(text.ToSsortingng())) { yield return new RtfText(text.ToSsortingng()); 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 (!ssortingng.IsNullOrWhiteSpace(text.ToSsortingng())) { yield return new RtfText(text.ToSsortingng()); text.Clear(); } // in fact, it's a normal controlWord controlWord.Append(c); state = RtfParseState.ControlWord; break; } } while (true); } public static bool MoveToNextControlWord(IEnumerator enumerator, ssortingng word) { if (enumerator == null) throw new ArgumentNullException("enumerator"); while (enumerator.MoveNext()) { if (enumerator.Current.Text == word) return true; } return false; } public static ssortingng GetNextText(IEnumerator 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 enumerator) { if (enumerator == null) throw new ArgumentNullException("enumerator"); while (enumerator.MoveNext()) { RtfText text = enumerator.Current as RtfText; if (text != null) { List bytes = new List(); 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) { } } 

OK, ça devrait marcher pour vous. Pour illustrer ma solution, j’ai créé un projet WinForms avec un PictureBox dont le gestionnaire d’événements de peinture a été mappé à la fonction suivante:

  private void rtfImage_Paint(object sender, PaintEventArgs e) { ssortingng rtfStr = System.IO.File.ReadAllText("MySampleFile.rtf"); ssortingng 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); } 

Ce code s’appuie sur la méthode System.Drawing.Image.FromStream () , avec deux fonctions “helper”:

Un extracteur de chaîne:

  ssortingng ExtractImgHex(ssortingng 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.Subssortingng(startIndex, endIndex - startIndex); } 

… et un convertisseur binary:

  public static byte[] ToBinary(ssortingng 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]; SsortingngBuilder hex = new SsortingngBuilder(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; }