kshowmail: kshowmail/decodeRFC2047.cpp Source File

00001 /***************************************************************************
+00002                           decodeRFC2047.cpp  -  description
+00003                              -------------------
+00004     begin                : Mon Jan 28 2002
+00005     copyright            : (C) 2002 by Eggert Ehmke
+00006     email                : eggert.ehmke@berlin.de
+00007  ***************************************************************************/
+00008 
+00009 /***************************************************************************
+00010  *                                                                         *
+00011  *   This program is free software; you can redistribute it and/or modify  *
+00012  *   it under the terms of the GNU General Public License as published by  *
+00013  *   the Free Software Foundation; either version 2 of the License, or     *
+00014  *   (at your option) any later version.                                   *
+00015  *                                                                         *
+00016  ***************************************************************************/
+00017 
+00018 
+00019 #include "decodeRFC2047.h"
+00020 
+00021 /*
+00022  * These functions have been adapted from the KMail program
+00023  */
+00024 
+00025 QCString decodeQuotedPrintable(const QCString& aStr)
+00026 {
+00027   QCString bStr = aStr;
+00028   if (aStr.isNull())
+00029     bStr = "";
+00030 
+00031   DwString dwsrc(bStr.data());
+00032   DwString dwdest;
+00033 
+00034   DwDecodeQuotedPrintable(dwsrc, dwdest);
+00035   return dwdest.c_str();
+00036 }
+00037 
+00038 QCString decodeBase64(const QCString& aStr)
+00039 {
+00040   QCString bStr = aStr;
+00041   if (aStr.isNull())
+00042     bStr = "";
+00043   while (bStr.length() < 16) bStr += "=";
+00044 
+00045   DwString dwsrc(bStr.data(), bStr.length());
+00046   DwString dwdest;
+00047   QCString result;
+00048 
+00049   DwDecodeBase64(dwsrc, dwdest);
+00050   result = dwdest.c_str();
+00051   return result;
+00052 }
+00053 
+00054 QTextCodec* codecForName(const QCString& _str)
+00055 {
+00056   if (_str.isEmpty()) return NULL;
+00057   if (_str.lower() == "shift_jis" || _str.lower() == "shift-jis")
+00058     return QTextCodec::codecForName("sjis");
+00059   return QTextCodec::codecForName(_str.lower().replace(
+00060     QRegExp("windows"), "cp") );
+00061 }
+00062 
+00063 QString Codecs::decodeRFC2047(const QCString& aStr)
+00064 {
+00065   QString result;
+00066   QCString charset;
+00067   char *pos, *beg, *end, *mid;
+00068   QCString str, cstr, LWSP_buffer;
+00069   char encoding, ch;
+00070   bool valid, lastWasEncodedWord=FALSE;
+00071   const int maxLen=200;
+00072   int i;
+00073 
+00074   if (aStr.find("=?") < 0)
+00075     return QString::fromLocal8Bit(aStr).replace(QRegExp("\n[\t ]")," ");
+00076 
+00077   for (pos=aStr.data(); *pos; pos++)
+00078   {
+00079     // line unfolding
+00080     if ( pos[0] == '\r' && pos[1] == '\n' ) {
+00081       pos++;
+00082       continue;
+00083     }
+00084     if ( pos[0] == '\n' )
+00085       continue;
+00086     // collect LWSP after encoded-words,
+00087     // because we might need to throw it out
+00088     // (when the next word is an encoded-word)
+00089     if ( lastWasEncodedWord && ( pos[0] == ' ' || pos[0] == '\t' ) )
+00090     {
+00091       LWSP_buffer += pos[0];
+00092       continue;
+00093     }
+00094     // verbatimly copy normal text
+00095     if (pos[0]!='=' || pos[1]!='?')
+00096     {
+00097       result += LWSP_buffer + pos[0];
+00098       LWSP_buffer = 0;
+00099       lastWasEncodedWord = FALSE;
+00100       continue;
+00101     }
+00102     // found possible encoded-word
+00103     beg = pos+2;
+00104     end = beg;
+00105     valid = TRUE;
+00106     // parse charset name
+00107     charset = "";
+00108     for (i=2,pos+=2; i<maxLen && (*pos!='?'&&(*pos==' '||ispunct(*pos)||isalnum(*pos))); i++)
+00109     {
+00110       charset += *pos;
+00111       pos++;
+00112     }
+00113     if (*pos!='?' || i<4 || i>=maxLen) valid = FALSE;
+00114     else
+00115     {
+00116       // get encoding and check delimiting question marks
+00117       encoding = toupper(pos[1]);
+00118       if (pos[2]!='?' || (encoding!='Q' && encoding!='B'))
+00119                             valid = FALSE;
+00120       pos+=3;
+00121       i+=3;
+00122     }
+00123     if (valid)
+00124     {
+00125       mid = pos;
+00126       // search for end of encoded part
+00127       while (i<maxLen && *pos && !(*pos=='?' && *(pos+1)=='='))
+00128       {
+00129                             i++;
+00130                             pos++;
+00131       }
+00132       end = pos+2;//end now points to the first char after the encoded string
+00133       if (i>=maxLen || !*pos)
+00134        valid = FALSE;
+00135     }
+00136     if (valid)
+00137     {
+00138       // valid encoding: decode and throw away separating LWSP
+00139       ch = *pos;
+00140       *pos = '\0';
+00141       str = QCString(mid).left((int)(mid - pos - 1));
+00142       if (encoding == 'Q')
+00143       {
+00144                             // decode quoted printable text
+00145                             for (i=str.length()-1; i>=0; i--)
+00146                      if (str[i]=='_')
+00147                             str[i]=' ';
+00148                             cstr = decodeQuotedPrintable(str);
+00149       }
+00150       else
+00151       {
+00152                             // decode base64 text
+00153                             cstr = decodeBase64(str);
+00154       }
+00155       QTextCodec *codec = codecForName(charset);
+00156       if (!codec)
+00157        codec = codecForName(KGlobal::locale()->encoding());
+00158       if (codec)
+00159        result += codec->toUnicode(cstr);
+00160       else
+00161        result += QString::fromLocal8Bit(cstr);
+00162       lastWasEncodedWord = TRUE;
+00163 
+00164       *pos = ch;
+00165       pos = end -1;
+00166     }
+00167     else
+00168     {
+00169       // invalid encoding, keep separating LWSP.
+00170       //result += "=?";
+00171       //pos = beg -1; // because pos gets increased shortly afterwards
+00172       pos = beg - 2;
+00173       result += LWSP_buffer;
+00174       result += *pos++;
+00175       result += *pos;
+00176       lastWasEncodedWord = FALSE;
+00177     }
+00178     LWSP_buffer = 0;
+00179   }
+00180   return result;
+00181 }
+00182 
+