kurl.cpp

来自「konqueror3 embedded版本, KDE环境下的当家浏览器的嵌入式版」· C++ 代码 · 共 2,357 行 · 第 1/4 页

          m_strPath_encoded += '/';
       }
    }
    else
    {
       if ( m_strPath.isEmpty() )
          m_strPath = '/';
    }
    KURL tmp( url() + rUrl, encoding_hint);
    *this = tmp;
    cleanPath(false);
  }
  else
  {
    KURL tmp( rUrl, encoding_hint);
    *this = tmp;
    // Preserve userinfo if applicable.
    if (!_u.m_strUser.isEmpty() && m_strUser.isEmpty() && (_u.m_strHost == m_strHost) && (_u.m_strProtocol == m_strProtocol))
    {
       m_strUser = _u.m_strUser;
       m_strPass = _u.m_strPass;
    }
    cleanPath(false);
  }
}

void KURL::reset()
{
  m_strProtocol = QString::null;
  m_strUser = QString::null;
  m_strPass = QString::null;
  m_strHost = QString::null;
  m_strPath = QString::null;
  m_strPath_encoded = QString::null;
  m_strQuery_encoded = QString::null;
  m_strRef_encoded = QString::null;
  m_bIsMalformed = true;
  m_iPort = 0;
  m_iUriMode = Auto;
}

bool KURL::isEmpty() const
{
  return (m_strPath.isEmpty() && m_strProtocol.isEmpty());
}

void KURL::parse( const QString& _url, int encoding_hint )
{
    if ( _url.isEmpty() || m_iUriMode == Invalid )
    {
	m_strProtocol = _url;
	m_iUriMode = Invalid;
	return;
    }

    const QChar* buf = _url.unicode();
    const QChar* orig = buf;
    uint len = _url.length();
    uint pos = 0;

    // Node 1: Accept alpha or slash
    QChar x = buf[pos++];
#ifdef Q_WS_WIN
    /* win32: accept <letter>: or <letter>:/ or <letter>:\ */
    const bool alpha = isalpha((int)x);
    if (alpha && len<2)
        goto NodeErr;
    if (alpha && buf[pos]==':' && (len==2 || (len>2 && (buf[pos+1]=='/' || buf[pos+1]=='\\'))))
#else
    if ( x == '/' )
#endif
    {
	// A slash means we immediately proceed to parse it as a file URL.
	m_iUriMode = URL;
	m_strProtocol = fileProt;
	parseURL( _url, encoding_hint );
	return;
    }
    if ( !isalpha( (int)x ) )
	goto NodeErr;

    // Node 2: Accept any amount of (alpha|digit|'+'|'-')
    // '.' is not currently accepted, because current KURL may be confused.
    // Proceed with :// :/ or :
    while( pos < len && (isalpha((int)buf[pos]) || isdigit((int)buf[pos]) ||
			 buf[pos] == '+' || buf[pos] == '-')) pos++;

    if (pos < len && buf[pos] == ':' )
    {
	m_strProtocol = QString( orig, pos ).lower();
	if ( m_iUriMode == Auto )
	    m_iUriMode = uriModeForProtocol( m_strProtocol );
	// Proceed to correct parse function.
	switch ( m_iUriMode )
	{
	case RawURI:
	    parseRawURI( _url );
	    return;
	case Mailto:
	    parseMailto( _url );
	    return;
	case URL:
	    parseURL( _url, encoding_hint );
	    return;
	default:
	    // Unknown URI mode results in an invalid URI.
	    break;
	}
    }

NodeErr:
    reset();
    m_strProtocol = _url;
    m_iUriMode = Invalid;
}

void KURL::parseRawURI( const QString& _url, int encoding_hint )
{
    uint len = _url.length();
    const QChar* buf = _url.unicode();

    uint pos = 0;

    // Accept any amount of (alpha|digit|'+'|'-')
    // '.' is not currently accepted, because current KURL may be confused.
    // Proceed with :
    while( pos < len && (isalpha((int)buf[pos]) || isdigit((int)buf[pos]) ||
			 buf[pos] == '+' || buf[pos] == '-')) pos++;

    // Note that m_strProtocol is already set here, so we just skip over the protocol.
    if (pos < len && buf[pos] == ':' )
	pos++;
    else { // can't happen, the caller checked all this already
	reset();
	m_strProtocol = _url;
	m_iUriMode = Invalid;
	return;
    }

    if ( pos == len ) // can't happen, the caller checked this already
	m_strPath = QString::null;
    else
	m_strPath = decode( QString( buf + pos, len - pos ), encoding_hint, true );

    m_bIsMalformed = false;

    return;
}

void KURL::parseMailto( const QString& _url, int encoding_hint )
{
    parseURL( _url, encoding_hint);
    if ( m_bIsMalformed )
        return;
    QRegExp mailre("(.+@)(.+)");
    if ( mailre.exactMatch( m_strPath ) )
    {
#ifndef KDE_QT_ONLY
	QString host = KIDNA::toUnicode( mailre.cap( 2 ) );
	if (host.isEmpty())
	    host = mailre.cap( 2 ).lower();
#else
	QString host = mailre.cap( 2 ).lower();
#endif
	m_strPath = mailre.cap( 1 ) + host;
  }
}

void KURL::parseURL( const QString& _url, int encoding_hint )
{
  QString port;
  bool badHostName = false;
  int start = 0;
  uint len = _url.length();
  const QChar* buf = _url.unicode();

  QChar delim;
  QString tmp;

  uint pos = 0;

  // Node 1: Accept alpha or slash
  QChar x = buf[pos++];
#ifdef Q_WS_WIN
  /* win32: accept <letter>: or <letter>:/ or <letter>:\ */
  const bool alpha = isalpha((int)x);
  if (alpha && len<2)
    goto NodeErr;
  if (alpha && buf[pos]==':' && (len==2 || (len>2 && (buf[pos+1]=='/' || buf[pos+1]=='\\'))))
#else
  if ( x == '/' )
#endif
    goto Node9;
  if ( !isalpha( (int)x ) )
    goto NodeErr;

  // Node 2: Accept any amount of (alpha|digit|'+'|'-')
  // '.' is not currently accepted, because current KURL may be confused.
  // Proceed with :// :/ or :
  while( pos < len && (isalpha((int)buf[pos]) || isdigit((int)buf[pos]) ||
          buf[pos] == '+' || buf[pos] == '-')) pos++;

  // Note that m_strProtocol is already set here, so we just skip over the protocol.
  if ( pos+2 < len && buf[pos] == ':' && buf[pos+1] == '/' && buf[pos+2] == '/' )
    {
      pos += 3;
    }
  else if (pos+1 < len && buf[pos] == ':' ) // Need to always compare length()-1 otherwise KURL passes "http:" as legal!!
    {
      pos++;
      start = pos;
      goto Node9;
    }
  else
    goto NodeErr;

  //Node 3: We need at least one character here
  if ( pos == len )
      goto NodeErr;
  start = pos;

  // Node 4: Accept any amount of characters.
  if (buf[pos] == '[')     // An IPv6 host follows.
      goto Node8;
  // Terminate on / or @ or ? or # or " or ; or <
  x = buf[pos];
  while( (x != ':') && (x != '@') && (x != '/') && (x != '?') && (x != '#') )
  {
     if ((x == '\"') || (x == ';') || (x == '<'))
        badHostName = true;
     if (++pos == len)
        break;
     x = buf[pos];
  }
  if ( pos == len )
    {
      if (badHostName)
         goto NodeErr;

      setHost(decode(QString( buf + start, pos - start ), encoding_hint));
      goto NodeOk;
    }
  if ( x == '@' )
    {
      m_strUser = decode(QString( buf + start, pos - start ), encoding_hint);
      pos++;
      goto Node7;
    }
  else if ( (x == '/') || (x == '?') || (x == '#'))
    {
      if (badHostName)
         goto NodeErr;

      setHost(decode(QString( buf + start, pos - start ), encoding_hint));
      start = pos;
      goto Node9;
    }
  else if ( x != ':' )
    goto NodeErr;
  m_strUser = decode(QString( buf + start, pos - start ), encoding_hint);
  pos++;

  // Node 5: We need at least one character
  if ( pos == len )
    goto NodeErr;
  start = pos++;

  // Node 6: Read everything until @, /, ? or #
  while( (pos < len) &&
		(buf[pos] != '@') &&
		(buf[pos] != '/') &&
		(buf[pos] != '?') &&
		(buf[pos] != '#')) pos++;
  // If we now have a '@' the ':' seperates user and password.
  // Otherwise it seperates host and port.
  if ( (pos == len) || (buf[pos] != '@') )
    {
      // Ok the : was used to separate host and port
      if (badHostName)
         goto NodeErr;
      setHost(m_strUser);
      m_strUser = QString::null;
      QString tmp( buf + start, pos - start );
      char *endptr;
      m_iPort = (unsigned short int)strtol(tmp.ascii(), &endptr, 10);
      if ((pos == len) && (strlen(endptr) == 0))
        goto NodeOk;
      // there is more after the digits
      pos -= strlen(endptr);
      if ((buf[pos] != '@') &&
          (buf[pos] != '/') &&
          (buf[pos] != '?') &&
          (buf[pos] != '#'))
        goto NodeErr;

      start = pos;
      goto Node9;
    }
  m_strPass = decode(QString( buf + start, pos - start), encoding_hint);
  pos++;

  // Node 7: We need at least one character
 Node7:
  if ( pos == len )
    goto NodeErr;

 Node8:
  if (buf[pos] == '[')
  {
    // IPv6 address
    start = ++pos; // Skip '['

    if (pos == len)
    {
       badHostName = true;
       goto NodeErr;
    }
    // Node 8a: Read everything until ] or terminate
    badHostName = false;
    x = buf[pos];
    while( (x != ']') )
    {
       if ((x == '\"') || (x == ';') || (x == '<'))
          badHostName = true;
       if (++pos == len)
       {
          badHostName = true;
          break;
       }
       x = buf[pos];
    }
    if (badHostName)
       goto NodeErr;
    setHost(decode(QString( buf + start, pos - start ), encoding_hint));
    if (pos < len) pos++; // Skip ']'
    if (pos == len)
       goto NodeOk;
  }
  else
  {
    // Non IPv6 address, with a user
    start = pos;

    // Node 8b: Read everything until / : or terminate
    badHostName = false;
    x = buf[pos];
    while( (x != ':') && (x != '@') && (x != '/') && (x != '?') && (x != '#') )
    {
       if ((x == '\"') || (x == ';') || (x == '<'))
          badHostName = true;
       if (++pos == len)
          break;
       x = buf[pos];
    }
    if (badHostName)
       goto NodeErr;
    if ( pos == len )
    {
       setHost(decode(QString( buf + start, pos - start ), encoding_hint));
       goto NodeOk;
    }
    setHost(decode(QString( buf + start, pos - start ), encoding_hint));
  }
  x = buf[pos];
  if ( x == '/' || x == '#' || x == '?' )
    {
      start = pos;
      goto Node9;
    }
  else if ( x != ':' )
    goto NodeErr;
  pos++;

  // Node 8c: Accept at least one digit
  if ( pos == len )
    goto NodeErr;
  start = pos;
  if ( !isdigit( buf[pos++] ) )
    goto NodeErr;

  // Node 8d: Accept any amount of digits
  while( pos < len && isdigit( buf[pos] ) ) pos++;
  port = QString( buf + start, pos - start );
  m_iPort = port.toUShort();
  if ( pos == len )
    goto NodeOk;
  start = pos;

 Node9: // parse path until query or reference reached

  while( pos < len && buf[pos] != '#' && buf[pos]!='?' ) pos++;

  tmp = QString( buf + start, pos - start );
  //kdDebug(126)<<" setting encoded path to:"<<tmp<<endl;
  setEncodedPath( tmp, encoding_hint );

  if ( pos == len )
      goto NodeOk;

 //Node10: // parse query or reference depending on what comes first
  delim = (buf[pos++]=='#'?'?':'#');

  start = pos;

  while(pos < len && buf[pos]!=delim ) pos++;

  tmp = QString(buf + start, pos - start);
  if (delim=='#')
      _setQuery(tmp, encoding_hint);
  else
      m_strRef_encoded = tmp;

  if (pos == len)
      goto NodeOk;

 //Node11: // feed the rest into the remaining variable
  tmp = QString( buf + pos + 1, len - pos - 1);
  if (delim == '#')
      m_strRef_encoded = tmp;
  else
      _setQuery(tmp, encoding_hint);

 NodeOk:
  //kdDebug(126)<<"parsing finished. m_strProtocol="<<m_strProtocol<<" m_strHost="<<m_strHost<<" m_strPath="<<m_strPath<<endl;
  m_bIsMalformed = false; // Valid URL

  //kdDebug()<<"Prot="<<m_strProtocol<<"\nUser="<<m_strUser<<"\nPass="<<m_strPass<<"\nHost="<<m_strHost<<"\nPath="<<m_strPath<<"\nQuery="<<m_strQuery_encoded<<"\nRef="<<m_strRef_encoded<<"\nPort="<<m_iPort<<endl;
  if (m_strProtocol.isEmpty())
  {
    m_iUriMode = URL;
    m_strProtocol = fileProt;
  }
  return;

 NodeErr:
//  kdDebug(126) << "KURL couldn't parse URL \"" << _url << "\"" << endl;
  reset();
  m_strProtocol = _url;
  m_iUriMode = Invalid;
}

KURL& KURL::operator=( const QString& _url )
{
  reset();
  parse( _url );

  return *this;
}

KURL& KURL::operator=( const char * _url )
{
  reset();
  parse( QString::fromLatin1(_url) );

  return *this;
}

#ifndef QT_NO_NETWORKPROTOCOL
KURL& KURL::operator=( const QUrl & u )
{
  m_strProtocol = u.protocol();
  m_iUriMode = Auto;
  m_strUser = u.user();
  m_strPass = u.password();
  m_strHost = u.host();
  m_strPath = u.path( false );
  m_strPath_encoded = QString::null;
  m_strQuery_encoded = u.query();
  m_strRef_encoded = u.ref();
  m_bIsMalformed = !u.isValid();
  m_iPort = u.port();

  return *this;
}
#endif

KURL& KURL::operator=( const KURL& _u )
{
  m_strProtocol = _u.m_strProtocol;
  m_strUser = _u.m_strUser;
  m_strPass = _u.m_strPass;
  m_strHost = _u.m_strHost;
  m_strPath = _u.m_strPath;
  m_strPath_encoded = _u.m_strPath_encoded;
  m_strQuery_encoded = _u.m_strQuery_encoded;
  m_strRef_encoded = _u.m_strRef_encoded;
  m_bIsMalformed = _u.m_bIsMalformed;
  m_iPort = _u.m_iPort;
  m_iUriMode = _u.m_iUriMode;

  return *this;
}

bool KURL::operator<( const KURL& _u) const
{
  int i;
  if (!_u.isValid())
  {
     if (!isValid())
     {
        i = m_strProtocol.compare(_u.m_strProtocol);
        return (i < 0);
     }
     return false;
  }
  if (!isValid())
     return true;

  i = m_strProtocol.compare(_u.m_strProtocol);
  if (i) return (i < 0);

  i = m_strHost.compare(_u.m_strHost);
  if (i) return (i < 0);

  if (m_iPort != _u.m_iPort) return (m_iPort < _u.m_iPort);

  i = m_strPath.compare(_u.m_strPath);
  if (i) return (i < 0);

  i = m_strQuery_encoded.compare(_u.m_strQuery_encoded);
  if (i) return (i < 0);

  i = m_strRef_encoded.compare(_u.m_strRef_encoded);
  if (i) return (i < 0);

  i = m_strUser.compare(_u.m_strUser);
  if (i) return (i < 0);

  i = m_strPass.compare(_u.m_strPass);
  if (i) return (i < 0);

  return false;
}

bool KURL::operator==( const KURL& _u ) const
{
  if ( !isValid() || !_u.isValid() )
    return false;

  if ( m_strProtocol == _u.m_strProtocol &&
       m_strUser == _u.m_strUser &&
       m_strPass == _u.m_strPass &&
       m_strHost == _u.m_strHost &&
       m_strPath == _u.m_strPath &&
       // The encoded path may be null, but the URLs are still equal (David)
       ( m_strPath_encoded.isNull() || _u.m_strPath_encoded.isNull() ||
         m_strPath_encoded == _u.m_strPath_encoded ) &&
       m_strQuery_encoded == _u.m_strQuery_encoded &&
       m_strRef_encoded == _u.m_strRef_encoded &&
       m_iPort == _u.m_iPort )
  {
    return true;
  }

  return false;
}

bool KURL::operator==( const QString& _u ) const
{
  KURL u( _u );
  return ( *this == u );
}

bool KURL::cmp( const KURL &u, bool ignore_trailing ) const
{
  return equals( u, ignore_trailing );
}

bool KURL::equals( const KURL &_u, bool ignore_trailing ) const
{
  if ( !isValid() || !_u.isValid() )
    return false;

  if ( ignore_trailing )
  {
    QString path1 = path(1);
    QString path2 = _u.path(1);
    if ( path1 != path2 )
      return false;

    if ( m_strProtocol == _u.m_strProtocol &&
         m_strUser == _u.m_strUser &&
         m_strPass == _u.m_strPass &&
         m_strHost == _u.m_strHost &&
         m_strQuery_encoded == _u.m_strQuery_encoded &&
         m_strRef_encoded == _u.m_strRef_encoded &&
         m_iPort == _u.m_iPort )
      return true;

    return false;