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

/*
    Copyright (C) 1999 Torben Weis <[e-mail]>

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

    This library is distributed in the hope that it will be useful,
    but WITHOUT ANY WARRANTY; without even the implied warranty of
    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
    Library General Public License for more details.

    You should have received a copy of the GNU Library General Public License
    along with this library; see the file COPYING.LIB.  If not, write to
    the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
    Boston, MA 02110-1301, USA.
*/

/*
 * The currently active RFC for URL/URIs is RFC3986
 * Previous (and now deprecated) RFCs are RFC1738 and RFC2396
 */

#include "kurl.h"

// KDE_QT_ONLY is first used for dcop/client (e.g. marshalling)
#ifndef KDE_QT_ONLY
#include <kdebug.h>
#include <kglobal.h>
#include <kidna.h>
#include <kprotocolinfo.h>
#endif

#include <stdio.h>
#include <assert.h>
#include <ctype.h>
#include <stdlib.h>
#include <unistd.h>

#include <qurl.h>
#include <qdir.h>
#include <qstringlist.h>
#include <qregexp.h>
#include <qstylesheet.h>
#include <qmap.h>
#include <qtextcodec.h>
#include <qmutex.h>

#ifdef Q_WS_WIN
# define KURL_ROOTDIR_PATH "C:/"
#else
# define KURL_ROOTDIR_PATH "/"
#endif

static const QString fileProt = "file";

static QTextCodec * codecForHint( int encoding_hint /* not 0 ! */ )
{
    return QTextCodec::codecForMib( encoding_hint );
}

// encoding_offset:
// 0 encode both @ and /
// 1 encode @ but not /
// 2 encode neither @ or /
static QString encode( const QString& segment, int encoding_offset, int encoding_hint, bool isRawURI = false )
{
  const char *encode_string = "/@<>#\"&?={}|^~[]\'`\\:+%";
  encode_string += encoding_offset;

  QCString local;
  if (encoding_hint==0)
    local = segment.local8Bit();
  else
  {
      QTextCodec * textCodec = codecForHint( encoding_hint );
      if (!textCodec)
          local = segment.local8Bit();
      else
          local = textCodec->fromUnicode( segment );
  }

  int old_length = isRawURI ? local.size() - 1 : local.length();

  if ( !old_length )
    return segment.isNull() ? QString::null : QString(""); // differentiate null and empty

  // a worst case approximation
  QChar *new_segment = new QChar[ old_length * 3 + 1 ];
  int new_length = 0;

  for ( int i = 0; i < old_length; i++ )
  {
    // 'unsave' and 'reserved' characters
    // according to RFC 1738,
    // 2.2. URL Character Encoding Issues (pp. 3-4)
    // WABA: Added non-ascii
    unsigned char character = local[i];
    if ( (character <= 32) || (character >= 127) ||
         strchr(encode_string, character) )
    {
      new_segment[ new_length++ ] = '%';

      unsigned int c = character / 16;
      c += (c > 9) ? ('A' - 10) : '0';
      new_segment[ new_length++ ] = c;

      c = character % 16;
      c += (c > 9) ? ('A' - 10) : '0';
      new_segment[ new_length++ ] = c;

    }
    else
      new_segment[ new_length++ ] = local[i];
  }

  QString result = QString(new_segment, new_length);
  delete [] new_segment;
  return result;
}

static QString encodeHost( const QString& segment, bool encode_slash, int encoding_hint )
{
  // Hostnames are encoded differently
  // we use the IDNA transformation instead

  // Note: when merging qt-addon, use QResolver::domainToAscii here
#ifndef KDE_QT_ONLY
  Q_UNUSED( encode_slash );
  Q_UNUSED( encoding_hint );
  QString host = KIDNA::toAscii(segment);
  if (host.isEmpty())
     return segment;
  return host;
#else
  return encode(segment, encode_slash ? 0 : 1, encoding_hint);
#endif
}

static int hex2int( unsigned int _char )
{
  if ( _char >= 'A' && _char <='F')
    return _char - 'A' + 10;
  if ( _char >= 'a' && _char <='f')
    return _char - 'a' + 10;
  if ( _char >= '0' && _char <='9')
    return _char - '0';
  return -1;
}

// WABA: The result of lazy_encode isn't usable for a URL which
// needs to satisfies RFC requirements. However, the following
// operation will make it usable again:
//      encode(decode(...))
//
// As a result one can see that url.prettyURL() does not result in
// a RFC compliant URL but that the following sequence does:
//      KURL(url.prettyURL()).url()


static QString lazy_encode( const QString& segment, bool encodeAt=true )
{
  int old_length = segment.length();

  if ( !old_length )
    return QString::null;

  // a worst case approximation
  QChar *new_segment = new QChar[ old_length * 3 + 1 ];
  int new_length = 0;

  for ( int i = 0; i < old_length; i++ )
  {
    unsigned int character = segment[i].unicode(); // Don't use latin1()
                                                   // It returns 0 for non-latin1 values
    // Small set of really ambiguous chars
    if ((character < 32) ||  // Low ASCII
        ((character == '%') && // The escape character itself
           (i+2 < old_length) && // But only if part of a valid escape sequence!
          (hex2int(segment[i+1].unicode())!= -1) &&
          (hex2int(segment[i+2].unicode())!= -1)) ||
        (character == '?') || // Start of query delimiter
        ((character == '@') && encodeAt) || // Username delimiter
        (character == '#') || // Start of reference delimiter
        ((character == 32) && (i+1 == old_length))) // A trailing space
    {
      new_segment[ new_length++ ] = '%';

      unsigned int c = character / 16;
      c += (c > 9) ? ('A' - 10) : '0';
      new_segment[ new_length++ ] = c;

      c = character % 16;
      c += (c > 9) ? ('A' - 10) : '0';
      new_segment[ new_length++ ] = c;
    }
    else
    new_segment[ new_length++ ] = segment[i];
  }

  QString result = QString(new_segment, new_length);
  delete [] new_segment;
  return result;
}

static void decode( const QString& segment, QString &decoded, QString &encoded, int encoding_hint=0, bool updateDecoded = true, bool isRawURI = false )
{
  decoded = QString::null;
  encoded = segment;

  int old_length = segment.length();
  if ( !old_length )
    return;

  QTextCodec *textCodec = 0;
  if (encoding_hint)
      textCodec = codecForHint( encoding_hint );

  if (!textCodec)
      textCodec = QTextCodec::codecForLocale();

  QCString csegment = textCodec->fromUnicode(segment);
  // Check if everything went ok
  if (textCodec->toUnicode(csegment) != segment)
  {
      // Uh oh
      textCodec = codecForHint( 106 ); // Fall back to utf-8
      csegment = textCodec->fromUnicode(segment);
  }
  old_length = csegment.length();

  int new_length = 0;
  int new_length2 = 0;

  // make a copy of the old one
  char *new_segment = new char[ old_length + 1 ];
  QChar *new_usegment = new QChar[ old_length * 3 + 1 ];

  int i = 0;
  while( i < old_length )
  {
    bool bReencode = false;
    unsigned char character = csegment[ i++ ];
    if ((character <= ' ') || (character > 127))
       bReencode = true;

    new_usegment [ new_length2++ ] = character;
    if (character == '%' )
    {
      int a = i+1 < old_length ? hex2int( csegment[i] ) : -1;
      int b = i+1 < old_length ? hex2int( csegment[i+1] ) : -1;
      if ((a == -1) || (b == -1)) // Only replace if sequence is valid
      {
         // Contains stray %, make sure to re-encode!
         bReencode = true;
      }
      else
      {
         // Valid %xx sequence
         character = a * 16 + b; // Replace with value of %dd
         if (!isRawURI && !character && updateDecoded)
            break; // Stop at %00

         new_usegment [ new_length2++ ] = (unsigned char) csegment[i++];
         new_usegment [ new_length2++ ] = (unsigned char) csegment[i++];
      }
    }
    if (bReencode)
    {
      new_length2--;
      new_usegment [ new_length2++ ] = '%';

      unsigned int c = character / 16;
      c += (c > 9) ? ('A' - 10) : '0';
      new_usegment[ new_length2++ ] = c;

      c = character % 16;
      c += (c > 9) ? ('A' - 10) : '0';
      new_usegment[ new_length2++ ] = c;
    }

    new_segment [ new_length++ ] = character;
  }
  new_segment [ new_length ] = 0;

  encoded = QString( new_usegment, new_length2);

  // Encoding specified
  if (updateDecoded)
  {
     decoded = textCodec->toUnicode( new_segment );
     if ( isRawURI ) {
        int length = qstrlen( new_segment );
        while ( length < new_length ) {
            decoded += QChar::null;
            length += 1;
            decoded += textCodec->toUnicode( new_segment + length );
            length += qstrlen( new_segment + length );
        }
     }

     QCString validate = textCodec->fromUnicode(decoded);

     if (strcmp(validate.data(), new_segment) != 0)
     {
        decoded = QString::fromLocal8Bit(new_segment, new_length);
     }
  }

  delete [] new_segment;
  delete [] new_usegment;
}

static QString decode(const QString &segment, int encoding_hint = 0, bool isRawURI = false)
{
  QString result;
  QString tmp;
  decode(segment, result, tmp, encoding_hint, true, isRawURI);
  return result;
}

static QString cleanpath(const QString &_path, bool cleanDirSeparator, bool decodeDots)
{
  if (_path.isEmpty()) return QString::null;

  if (QDir::isRelativePath(_path))
     return _path; // Don't mangle mailto-style URLs

  QString path = _path;

  int len = path.length();

  if (decodeDots)
  {
#ifndef KDE_QT_ONLY
     static const QString &encodedDot = KGlobal::staticQString("%2e");
#else
     QString encodedDot("%2e");
#endif
     if (path.find(encodedDot, 0, false) != -1)
     {
#ifndef KDE_QT_ONLY
        static const QString &encodedDOT = KGlobal::staticQString("%2E"); // Uppercase!
#else
        QString encodedDOT("%2E");
#endif
        path.replace(encodedDot, ".");
        path.replace(encodedDOT, ".");
        len = path.length();
     }
  }

  bool slash = (len && path[len-1] == '/') ||
               (len > 1 && path[len-2] == '/' && path[len-1] == '.');

  // The following code cleans up directory path much like
  // QDir::cleanDirPath() except it can be made to ignore multiple
  // directory separators by setting the flag to false.  That fixes
  // bug# 15044, mail.altavista.com and other similar brain-dead server
  // implementations that do not follow what has been specified in
  // RFC 2396!! (dA)
  QString result;
  int cdUp, orig_pos, pos;

  cdUp = 0;
  pos = orig_pos = len;
  while ( pos && (pos = path.findRev('/',--pos)) != -1 )
  {
    len = orig_pos - pos - 1;
    if ( len == 2 && path[pos+1] == '.' && path[pos+2] == '.' )
      cdUp++;
    else
    {
      // Ignore any occurrences of '.'
      // This includes entries that simply do not make sense like /..../
      if ( (len || !cleanDirSeparator) &&
           (len != 1 || path[pos+1] != '.' ) )
      {
          if ( !cdUp )
              result.prepend(path.mid(pos, len+1));
          else
              cdUp--;
      }
    }
    orig_pos = pos;
  }

#ifdef Q_WS_WIN // prepend drive letter if exists (js)
  if (orig_pos >= 2 && isalpha(path[0].latin1()) && path[1]==':') {
    result.prepend(QString(path[0])+":");
  }
#endif

  if ( result.isEmpty() )
    result = KURL_ROOTDIR_PATH;
  else if ( slash && result[result.length()-1] != '/' )
       result.append('/');

  return result;
}

bool KURL::isRelativeURL(const QString &_url)
{
  int len = _url.length();
  if (!len) return true; // Very short relative URL.
  const QChar *str = _url.unicode();

  // Absolute URL must start with alpha-character
  if (!isalpha(str[0].latin1()))
     return true; // Relative URL

  for(int i = 1; i < len; i++)
  {
     char c = str[i].latin1(); // Note: non-latin1 chars return 0!
     if (c == ':')
        return false; // Absolute URL

     // Protocol part may only contain alpha, digit, + or -
     if (!isalpha(c) && !isdigit(c) && (c != '+') && (c != '-'))
        return true; // Relative URL
  }
  // URL did not contain ':'
  return true; // Relative URL
}

KURL::List::List(const KURL &url)
{
    append( url );
}

KURL::List::List(const QStringList &list)
{
  for (QStringList::ConstIterator it = list.begin();
       it != list.end();
       it++)
    {
      append( KURL(*it) );
    }
}

QStringList KURL::List::toStringList() const
{
  QStringList lst;
   for( KURL::List::ConstIterator it = begin();
        it != end();
        it++)
   {
      lst.append( (*it).url() );
   }
   return lst;
}


KURL::KURL()
{
  reset();
}

KURL::~KURL()
{
}


KURL::KURL( const QString &url, int encoding_hint )
{
  reset();
  parse( url, encoding_hint );
}

KURL::KURL( const char * url, int encoding_hint )
{
  reset();
  parse( QString::fromLatin1(url), encoding_hint );
}

KURL::KURL( const QCString& url, int encoding_hint )
{
  reset();
  parse( QString::fromLatin1(url), encoding_hint );
}

KURL::KURL( const KURL& _u )
{
  *this = _u;
}

QDataStream & operator<< (QDataStream & s, const KURL & a)
{
  QString QueryForWire=a.m_strQuery_encoded;
  if (!a.m_strQuery_encoded.isNull())
    QueryForWire.prepend("?");

    s << a.m_strProtocol << a.m_strUser << a.m_strPass << a.m_strHost
      << a.m_strPath << a.m_strPath_encoded << QueryForWire << a.m_strRef_encoded
      << Q_INT8(a.m_bIsMalformed ? 1 : 0) << a.m_iPort;
    return s;
}

QDataStream & operator>> (QDataStream & s, KURL & a)
{
    Q_INT8 malf;
    QString QueryFromWire;
    s >> a.m_strProtocol >> a.m_strUser >> a.m_strPass >> a.m_strHost
      >> a.m_strPath >> a.m_strPath_encoded >> QueryFromWire >> a.m_strRef_encoded
      >> malf >> a.m_iPort;
    a.m_bIsMalformed = (malf != 0);

    if ( QueryFromWire.isNull() )
      a.m_strQuery_encoded = QString::null;
    else if ( QueryFromWire.length() == 1 ) // empty query
      a.m_strQuery_encoded = "";
    else
      a.m_strQuery_encoded = QueryFromWire.mid(1);

    a.m_iUriMode = KURL::uriModeForProtocol( a.m_strProtocol );

    return s;
}

#ifndef QT_NO_NETWORKPROTOCOL
KURL::KURL( const QUrl &u )
{
  *this = u;
}
#endif

KURL::KURL( const KURL& _u, const QString& _rel_url, int encoding_hint )
{
  if (_u.hasSubURL()) // Operate on the last suburl, not the first
  {
    KURL::List lst = split( _u );
    KURL u(lst.last(), _rel_url, encoding_hint);
    lst.remove( lst.last() );
    lst.append( u );
    *this = join( lst );
    return;
  }
  // WORKAROUND THE RFC 1606 LOOPHOLE THAT ALLOWS
  // http:/index.html AS A VALID SYNTAX FOR RELATIVE
  // URLS. ( RFC 2396 section 5.2 item # 3 )
  QString rUrl = _rel_url;
  int len = _u.m_strProtocol.length();
  if ( !_u.m_strHost.isEmpty() && !rUrl.isEmpty() &&
       rUrl.find( _u.m_strProtocol, 0, false ) == 0 &&
       rUrl[len] == ':' && (rUrl[len+1] != '/' ||
       (rUrl[len+1] == '/' && rUrl[len+2] != '/')) )
  {
    rUrl.remove( 0, rUrl.find( ':' ) + 1 );
  }

  if ( rUrl.isEmpty() )
  {
    *this = _u;
  }
  else if ( rUrl[0] == '#' )
  {
    *this = _u;
    m_strRef_encoded = rUrl.mid(1);
    if ( m_strRef_encoded.isNull() )
        m_strRef_encoded = ""; // we know there was an (empty) html ref, we saw the '#'
  }
  else if ( isRelativeURL( rUrl) )
  {
    *this = _u;
    m_strQuery_encoded = QString::null;
    m_strRef_encoded = QString::null;
    if ( rUrl[0] == '/')
    {
        if ((rUrl.length() > 1) && (rUrl[1] == '/'))
        {
           m_strHost = QString::null;
           // File protocol returns file:/// without host, strip // from rUrl
           if (_u.m_strProtocol == fileProt)
              rUrl.remove(0, 2);
        }
        m_strPath = QString::null;
        m_strPath_encoded = QString::null;
    }
    else if ( rUrl[0] != '?' )
    {
       int pos = m_strPath.findRev( '/' );
       if (pos >= 0)
          m_strPath.truncate(pos);
       m_strPath += '/';
       if (!m_strPath_encoded.isEmpty())
       {
          pos = m_strPath_encoded.findRev( '/' );
          if (pos >= 0)
             m_strPath_encoded.truncate(pos);