/***************************************************************************
                          mimemessage.cpp  -  description
                             -------------------
    begin                : Sat Mar 8 2003
    copyright            : (C) 2003 by Mike K. Bennett
    email                : mkb137b@hotmail.com
 ***************************************************************************/

/***************************************************************************
 *                                                                         *
 *   This program is free software; you can redistribute it and/or modify  *
 *   it under the terms of the GNU General Public License as published by  *
 *   the Free Software Foundation; either version 2 of the License, or     *
 *   (at your option) any later version.                                   *
 *                                                                         *
 ***************************************************************************/

#include "mimemessage.h"

#include "../kmessdebug.h"

#include <QByteArray>
#include <QRegExp>
#include <QTextCodec>

#include <KCharsets>
#include <KCodecs>
#include <KGlobal>



/*
   The decode...() functions and getCodecByName() come from KMail
   and are therefore (C) KMail developers.
*/


// The constructor
MimeMessage::MimeMessage()
{
}



// The constructor that parses a message
MimeMessage::MimeMessage(const QString &message)
{
#ifdef KMESSDEBUG_MIMEMESSAGE
  kmDebug() << "Parsing the sub message";
#endif
  parseMessage( message );
}



// The constructor that parses a message
MimeMessage::MimeMessage(const QByteArray &message)
{
  // we get the header without the binary data that comes after
  int nullPos = message.indexOf('\0');
  int endMime = message.indexOf("\r\n\r\n");
  QString messageHeader( QString::fromUtf8(message.data(), (nullPos == -1) ? endMime : nullPos ) );

  // Regexps to parse the header
  QRegExp rx1("Content-Type: ([A-Za-z0-9$!*/\\-]*)");
  QRegExp rx2("Message-ID: ([^\r\n]+)");

  // Get the Content-Type header
  QString contentType;
  if( rx1.indexIn( messageHeader ) != -1 )
  {
    contentType = rx1.cap(1);
  }

  // Some messages should be preserved as binary, so take care of that.
  // - p2p messages have a 48 byte header.
  // - multi-packet messages can be split between a double byte utf-8 char.
  bool isP2P         = ( contentType == "application/x-msnmsgrp2p" );
  bool isMultiPacket = ( rx2.indexIn( messageHeader ) != -1 );
  if( isP2P || isMultiPacket )
  {
#ifdef KMESSDEBUG_MIMEMESSAGE
    if( isP2P )
    {
      kmDebug() << "Parsing the message, extracting binary p2p body.";
    }
    else if( isMultiPacket )
    {
      kmDebug() << "Parsing the message, preserving multi-packet body as binary.";
    }
#endif

    int bodyStart = endMime + 4; // 2 newlines
    int bodySize  = message.size() - bodyStart;

    // Extract the Mime fields from the message
    QString mimeData( QString::fromUtf8( message.data(), bodyStart ) );
    parseMessage( mimeData );

    // Extract the binary data from the message.
    // Copy to our own managed copy.
    binaryBody_ = QByteArray( message.data() + bodyStart, bodySize );
  }
  else
  {
#ifdef KMESSDEBUG_MIMEMESSAGE
    kmDebug() << "Parsing the message.";
#endif

    // This is a normal plain-text message.
    // The fromUtf8 call is required to convert Unicode characters properly! (like Chinese)
    parseMessage( QString::fromUtf8( message.data(), message.size()) );
  }
}



// The copy constructor
MimeMessage::MimeMessage(const MimeMessage& other)
{
#ifdef KMESSDEBUG_MIMEMESSAGE
  kmDebug() << "copy constructor";
#endif

  // Get the body of the other message
  fields_     = other.fields_;
  values_     = other.values_;
  body_       = other.body_;
  binaryBody_ = other.binaryBody_;

  // QValueList is implicitly shared, so the assign operation
  // does not create another copy. It uses "copy on write".

  // QByteArray is explicitly shared, but it's not a problem here,
  // since we don't modify the array data of binaryBody_.

#ifdef KMESSTEST
  // Make sure the ascii-zero characters are copied correctly:
  KMESS_ASSERT( binaryBody_.size() == other.binaryBody_.size() );
  KMESS_ASSERT( fields_.size()     == other.fields_.size() );
#endif
}



// The destructor
MimeMessage::~MimeMessage()
{
}



// Add a field to the message
void MimeMessage::addField(const QString& field, const QString& value)
{
  fields_ << field;
  values_ << value;
}


// Change a field, or add it
00162 void MimeMessage::setField(const QString& field, const QString& value)
{
  int index = fields_.indexOf( field );
  if(-1 == index)
  {
    addField( field, value );
  }
  else
  {
    values_[ index ] = value;
  }
}



// Clear all fields in MimeMessage for re-use the object
void MimeMessage::clearFields()
{
  body_ = "";
  fields_.clear();
  values_.clear();
  binaryBody_ = "";
}



// decodes MIME strings like =?iso...=...?= ...
QString MimeMessage::decodeRFC2047String(const QByteArray &aStr) const
{
  QString result;
  QByteArray charset;
  char *pos, *beg, *end, *mid=0;
  QByteArray str, cstr, LWSP_buffer;
  char encoding='Q', ch;
  bool valid, lastWasEncodedWord=FALSE;
  const int maxLen=200;
  int i;

  if( ! aStr.contains("=?") )
  {
    //QString str( messageCodec->toUnicode(aStr) );
    QString str( QString::fromUtf8( aStr ) );
    if (str.indexOf('\n') == -1) return str;
    QString str2((QChar*)0, str.length());
    int i = 0;
    while (i < str.length())
    {
      if (str[i] == '\n')
      {
        str2 += ' ';
        i += 2;
      } else {
        str2 += str[i];
        i++;
      }
    }
    return str2;
  }

  for (pos = const_cast<QByteArray&>( aStr ).data(); *pos; pos++)
  {
    // line unfolding
    if ( pos[0] == '\r' && pos[1] == '\n' ) {
      pos++;
      continue;
    }
    if ( pos[0] == '\n' )
      continue;
    // collect LWSP after encoded-words,
    // because we might need to throw it out
    // (when the next word is an encoded-word)
    if ( lastWasEncodedWord && ( pos[0] == ' ' || pos[0] == '\t' ) ) {
      LWSP_buffer += pos[0];
      continue;
    }
    // verbatimly copy normal text
    if (pos[0]!='=' || pos[1]!='?')
    {
      result += LWSP_buffer + pos[0];
      LWSP_buffer = 0;
      lastWasEncodedWord = FALSE;
      continue;
    }
    // found possible encoded-word
    beg = pos+2;
    end = beg;
    valid = TRUE;
    // parse charset name
    charset = "";
    for (i=2,pos+=2; i<maxLen && (*pos!='?'&&(*pos==' '||ispunct(*pos)||isalnum(*pos))); ++i)
    {
      charset += *pos;
      pos++;
    }
    if (*pos!='?' || i<4 || i>=maxLen) valid = FALSE;
    else
    {
      // get encoding and check delimiting question marks
      encoding = (char) toupper(pos[1]);
      if (pos[2]!='?' || (encoding!='Q' && encoding!='B'))
        valid = FALSE;
      pos+=3;
      i+=3;
    }
    if (valid)
    {
      mid = pos;
      // search for end of encoded part
      while (i<maxLen && *pos && !(*pos=='?' && *(pos+1)=='='))
      {
        ++i;
        ++pos;
      }
      end = pos+2;//end now points to the first char after the encoded string
      if (i>=maxLen || !*pos) valid = FALSE;
    }
    if (valid)
    {
      // valid encoding: decode and throw away separating LWSP
      ch = *pos;
      *pos = '\0';
      str = QByteArray( mid ).left((int)(mid - pos - 1));
      if (encoding == 'Q')
      {
        // decode quoted printable text
        for (i=str.length()-1; i>=0; i--)
          if (str[i]=='_') str[i]=' ';
        cstr = KCodecs::quotedPrintableDecode(str);
      }
      else
      {
        // decode base64 text
        cstr = QByteArray::fromBase64( str );
      }
      QTextCodec *codec = getCodecByName(charset);
      if (!codec)
      {
        result += QString::fromUtf8( cstr );
      }
      else
      {
        result += codec->toUnicode(cstr);
      }

      lastWasEncodedWord = TRUE;

      *pos = ch;
      pos = end -1;
    }
    else
    {
      // invalid encoding, keep separating LWSP.
      //result += "=?";
      //pos = beg -1; // because pos gets increased shortly afterwards
      pos = beg - 2;
      result += LWSP_buffer;
      result += *pos++;
      result += *pos;
      lastWasEncodedWord = FALSE;
    }
    LWSP_buffer = 0;
  }
  return result;
}



// Return the body of the message
const QString& MimeMessage::getBody() const
{
  return body_;
}



// Return the P2P data of the message
const QByteArray& MimeMessage::getBinaryBody() const
{
  return binaryBody_;
}



// Finds a QTextCodec by name
QTextCodec* MimeMessage::getCodecByName(const QByteArray& codecName )
{
  if ( codecName.isEmpty() )
  {
    return 0;
  }

  return KGlobal::charsets()->codecForName( codecName.toLower() );
}



// Return the field and value at the given index
void MimeMessage::getFieldAndValue(QString& field, QString& value, const int index) const
{
  if ( index < fields_.count() )
  {
    field = fields_[index];
    value = values_[index];
  }
  else
  {
    field = QString::null;
    value = QString::null;
  }
}



// Return the message fields as a big string
QString MimeMessage::getFields() const
{
  QString message( "" );

  // Get the fields and values
  for( int i = 0; i < fields_.count(); i++ )
  {
    message += fields_[i] + ": " + values_[i] + "\r\n";
  }

  return message;
}



// Return the entire message as a big string
QByteArray MimeMessage::getMessage() const
{
  // So much easier with Qt4!
  QString message = QString( getFields() + "\r\n" + body_);

  // for messages with an empty body must use two \r\n
  // sequences. This fixes typing notifications not showing in WLM 2009.
  if( body_.isEmpty() && binaryBody_.isEmpty() )
  {
    message += "\r\n";
  }

  QByteArray textPart = message.toUtf8();

  return textPart + binaryBody_;
}



// The total number of fields
uint MimeMessage::getNoFields() const
{
  return fields_.count();
}



// Get one parameter of a value that has multiple parameters
QString MimeMessage::getSubValue(const QString& field, const QString& subField) const
{
  QString value, parameter;
  int     left, right;
  // Get the value referred to by the field
  value = getValue( field );
  if ( !value.isNull() )
  {
    // If the subfield isn't specified, then get whatever is at the start of the message until the
    //  first semicolon or the end of the line
    if ( subField.isNull() )
    {
      left = 0;
    }
    else
    {
      // The left of the parameter is "subField=", the right is the next semicolon or the end of the line
      left = value.indexOf( subField + "=" );
      if ( left >= 0 )
      {
        left += subField.length() + 1;
      }
    }
    if ( left >= 0 )
    {
      right = value.indexOf( ";", left );
      if ( right < 0 )
      {
        right = value.length();
      }
      // Get the parameter
      parameter = value.mid( left, ( right - left ) );
      return parameter;
    }
  }
  return QString::null;
}



// Get a value given a field
QString MimeMessage::getValue(const QString& field ) const
{
  // Search the fields for a match
  for( int i = 0; i < fields_.count(); i++ )
  {
    if ( fields_[i] == field )
    {
      return values_[i];
    }
  }
  kmWarning() << "This message contained no field \"" << field << "\".";
  return QString();
}



// Test whether a given field exists in the message header
bool MimeMessage::hasField(const QString& field) const
{
  // Search the fields for a match
  for( int i = 0; i < fields_.count(); i++ )
  {
    if ( fields_[i] == field )
    {
      return true;
    }
  }

  return false;
}


// Parse the message into type, body, and fields and values
void MimeMessage::parseMessage(const QString& message)
{
  QString      head, field, value;
  QStringList  lines;

  // Split the message into head and body
  // Split the head into its various lines
  splitMessage( head, body_, message );
  splitHead( lines, head );

  // Split all the lines into field and value
  for( int i = 0; i < lines.count(); i++ )
  {
    if( lines[i].startsWith("\t") )
    {
#ifdef KMESSTEST
      KMESS_ASSERT( fields_.last() == "Received" );
#endif
      // Addition to parse email headers, with Received: headers that continue at
      // the next line
      values_[ values_.count() - 1 ] += "\r\n" + lines[i].mid(2);
    }
    else
    {
      splitLine( field, value, lines[i] );
      if ( !field.isEmpty() )
      {
        // Add the fields and values to the lists
        addField( field, value );
      }
    }
  }

#ifdef KMESSTEST
  KMESS_ASSERT( fields_.count() == values_.count() );
#endif
}



// Print the contents of the message to kDebug (for debugging purposes)
void MimeMessage::print() const
{
  kmDebug() << "Printing MIME message, " << fields_.count() << " fields.";
  // Get the fields and values
  for( int i = 0; i < fields_.count(); i++ )
  {
    kmDebug() << "Field: \"" << fields_[i] << "\" value: \"" << values_[i] << "\"";
  }

  if( binaryBody_.isEmpty() )
  {
    kmDebug() << "Body:" << endl << body_;
  }
  else
  {
    kmDebug() << "Body: (" << binaryBody_.size() << " bytes of binary data)";
  }
}



// Set the message body
void MimeMessage::setBody(const QString& body)
{
  body_ = body;
}



// Set the P2P data of the message
void MimeMessage::setBinaryBody(const QByteArray &header, const QByteArray &body, const char footer[4])
{
  int headerSize = header.size();
  int bodySize   = body.size();
  int footerPos  = headerSize + bodySize;

  // Footer needs to be wrapped in a QByteArray
  // or replace( start, end, char* ) will reduce the size of the footer (Qt 4.3.3)
  QByteArray footerWrapper( footer, 4 );

  // So much easier with Qt4!
  binaryBody_.resize( footerPos + 4 );
  binaryBody_.replace( 0, headerSize, header );
  binaryBody_.replace( headerSize, bodySize, body );
  binaryBody_.replace( footerPos, 4, footerWrapper );

#ifdef KMESSTEST
  KMESS_ASSERT( binaryBody_.size() >= 52 ); // 48 bytes header, 4 bytes footer.
#endif
}



// Set the P2P data of the message
void MimeMessage::setBinaryBody(const QByteArray &data)
{
  binaryBody_ = data;
}



// Split a line between field and value
void MimeMessage::splitLine(QString& field, QString& value, const QString& line) const
{
  if ( line.isEmpty() )
  {
    return;
  }

  int index;
  index = line.indexOf(":");
  if ( index >= 0 )
  {
    field = line.left( index );
    if ( ( index + 1 ) < (int)line.length() )
    {
      value = line.right( line.length() - index - 2 );
    }
    else
    {
      value = QString::null;
    }
  }
  else
  {
    kmWarning() << "Couldn't split line '" << line << "'";
  }
}



// Split the message head into components and store it in the string list
void MimeMessage::splitHead(QStringList& stringList, const QString& head) const
{
  // adam (16 Oct 10) - split on \n only
  // the OfflineIM service is doing something funky and only
  // terminating lines with \n. If we split on \n, then trim() the string,
  // we'll strip out any \r's as well.
  stringList = head.split("\n");
  for( int i = 0; i < stringList.count(); i++ )
  {
    // special case for email headers (apparently?)
    if ( stringList[i].startsWith( "\t" ) ) continue;

    stringList[i] = stringList[i].trimmed();
  }
}



// Split a message into head and body
void MimeMessage::splitMessage(QString& head, QString& body, const QString& message) const
{
  int index;
  int count;
  // The message is split into head and body at "\r\n\r\n"
  index = message.indexOf( "\r\n\r\n" );
  count = 4;

  // Offline IM MIME messages don't use \r\n anymore?
  // wtf? (adam - 16 Oct 10)
  if ( index < 0 )
  {
    index = message.indexOf("\n\n");
    count = 2;
  }

  if ( index < 0 )
  {
    head = message;
    body = "";
  }
  else
  {
    head = message.left( index + ( count / 2 ) ); // Keep a "\r\n" at the end
    body = message.right( message.length() - index - count );
  }
}