extendentitymanager.cxx

SP是一个基于GNU C++编译器

  size_t decode(Char *, const char *, size_t, const char **);
private:
  Owner<Decoder> sub_;
  ConstPtr<CharMapResource<Unsigned32> > map_;
};

MappingDecoder::MappingDecoder(Decoder *sub,
			       const ConstPtr<CharMapResource<Unsigned32> > &map)
: Decoder(sub->minBytesPerChar()), sub_(sub), map_(map)
{
}

size_t MappingDecoder::decode(Char *to, const char *s,
			      size_t slen, const char **rest)
{
  size_t n = sub_->decode(to, s, slen, rest);
  const CharMap<Unsigned32> &map = *map_;
  for (size_t i = 0; i < n; i++) {
    Unsigned32 d = map[to[i]];
    if (d & (unsigned(1) << 31))
      to[i] = (d & ~(unsigned(1) << 31));
    else
      to[i] += d;
  }
  return n;
}

Boolean MappingDecoder::convertOffset(unsigned long &offset) const
{
  return sub_->convertOffset(offset);
}
  
ExternalInputSource::ExternalInputSource(ParsedSystemId &parsedSysid,
					 const CharsetInfo &systemCharset,
					 const CharsetInfo &docCharset,
					 Boolean internalCharsetIsDocCharset,
					 Char replacementChar,
					 InputSourceOrigin *origin,
					 unsigned flags)
: InputSource(origin, 0, 0),
  mayRewind_((flags & EntityManager::mayRewind) != 0),
  mayNotExist_((flags & ExtendEntityManager::mayNotExist) != 0),
  sov_(parsedSysid.size()),
  internalCharsetIsDocCharset_(internalCharsetIsDocCharset),
  // hack
  maySetDocCharset_((flags & EntityManager::maySetDocCharset) != 0),
  replacementChar_(replacementChar)
{
  for (size_t i = 0; i < parsedSysid.size(); i++) {
    if (parsedSysid[i].codingSystemType
        != (internalCharsetIsDocCharset
	    ? StorageObjectSpec::bctf
	    : StorageObjectSpec::encoding)
	&& parsedSysid[i].codingSystemType != StorageObjectSpec::special) {
      map_ = new CharMapResource<Unsigned32>;
      buildMap(systemCharset, docCharset);
      break;
    }
  }
  for (size_t i = 0; i < sov_.size(); i++)
    sov_[i] = 0;
  init();
  info_ = new ExternalInfoImpl(parsedSysid);
  origin->setExternalInfo(info_);
}

void ExternalInputSource::setDocCharset(const CharsetInfo &docCharset,
					const CharsetInfo &systemCharset)
{
  if (!map_.isNull())
    buildMap(systemCharset, docCharset);
  willNotSetDocCharset();
}

void ExternalInputSource::willNotSetDocCharset()
{
  maySetDocCharset_ = 0;
}

void ExternalInputSource::buildMap(const CharsetInfo &systemCharset,
				   const CharsetInfo &docCharset)
{
  CharMap<Unsigned32> &map = *map_;
  // FIXME How should invalidChar be chosen when internalCharsetIsDocCharset_?
  Char invalidChar
    = internalCharsetIsDocCharset_ ? 0 : replacementChar_;
  map.setAll((Unsigned32(1) << 31) | invalidChar);
  if (internalCharsetIsDocCharset_)
    buildMap1(systemCharset, docCharset);
  else
    buildMap1(docCharset, systemCharset);
}

void ExternalInputSource::buildMap1(const CharsetInfo &fromCharset,
				    const CharsetInfo &toCharset)
{
  UnivCharsetDescIter iter(fromCharset.desc());
  for (;;) {
    WideChar descMin, descMax;
    UnivChar univMin;
    if (!iter.next(descMin, descMax, univMin))
      break;
    if (descMin > charMax)
      break;
    if (descMax > charMax)
      descMax = charMax;
    WideChar totalCount = 1 + (descMax - descMin);
    do {
      WideChar count;
      WideChar toMin;
      ISet<WideChar> set;
      int nMap = toCharset.univToDesc(univMin, toMin, set, count);
      if (count > totalCount)
	count = totalCount;
      if (nMap && toMin <= charMax) {
	Char toMax;
	if (count - 1 > charMax - toMin)
	  toMax = charMax;
	else
	  toMax = toMin + (count - 1);
	map_->setRange(descMin, descMin + (toMax - toMin), Char(toMin - descMin));
      }
      descMin += count;
      univMin += count;
      totalCount -= count;
    } while (totalCount > 0);
  }
}

void ExternalInputSource::init()
{
  so_ = 0;
  buf_ = 0;
  bufSize_ = 0;
  bufLim_ = 0;
  bufLimOffset_ = 0;
  insertRS_ = true;
  soIndex_ = 0;
  leftOver_ = 0;
  nLeftOver_ = 0;  
}

ExternalInputSource::~ExternalInputSource()
{
  if (buf_)
    delete [] buf_;
}

Boolean ExternalInputSource::rewind(Messenger &mgr)
{
  reset(0, 0);
  if (buf_)
    delete [] buf_;
  // reset makes a new EntityOrigin
  ParsedSystemId parsedSysid(info_->parsedSystemId());
  ExternalInfoImpl *oldInfo = info_;
  info_ = new ExternalInfoImpl(parsedSysid);
  so_ = 0;
  for (size_t i = 0; i < soIndex_; i++) {
    if (sov_[i] && !sov_[i]->rewind(mgr))
      return 0;
    StringC tem;
    oldInfo->getId(i, tem);
    info_->setId(i, tem);
  }
  inputSourceOrigin()->setExternalInfo(info_);
  init();
  return 1;
}

void ExternalInputSource::willNotRewind()
{
  for (size_t i = 0; i < sov_.size(); i++)
    if (sov_[i])
      sov_[i]->willNotRewind();
  mayRewind_ = 0;
}


// Round up N so that it is a power of TO.
// TO must be a power of 2.

inline
size_t roundUp(size_t n, size_t to)
{
  return (n + (to - 1)) & ~(to - 1);
}

inline
void ExternalInputSource::noteRSAt(const Char *p)
{
  info_->noteRS(bufLimOffset_ - (bufLim_ - p));
}

inline
void ExternalInputSource::noteRS()
{
  noteRSAt(cur());
}

Xchar ExternalInputSource::fill(Messenger &mgr)
{
  ASSERT(cur() == end());
  while (end() >= bufLim_) {
    // need more data
    while (so_ == 0) {
      if (soIndex_ >= sov_.size())
	return eE;
      if (soIndex_ > 0)
	info_->noteStorageObjectEnd(bufLimOffset_ - (bufLim_ - end()));
      const StorageObjectSpec &spec = info_->spec(soIndex_);
      if (!sov_[soIndex_]) {
	StringC id;
	if (mayNotExist_) {
	  NullMessenger nullMgr;
	  sov_[soIndex_]
	    = spec.storageManager->makeStorageObject(spec.specId, spec.baseId,
						     spec.search,
						     mayRewind_, nullMgr, id);
	}
	else
	  sov_[soIndex_]
	    = spec.storageManager->makeStorageObject(spec.specId, spec.baseId,
						     spec.search,
						     mayRewind_, mgr, id);
	info_->setId(soIndex_, id);
      }
      so_ = sov_[soIndex_].pointer();
      if (so_) {
	decoder_ = spec.codingSystem->makeDecoder();
	if (spec.codingSystemType != StorageObjectSpec::special
	    && spec.codingSystemType != (internalCharsetIsDocCharset_ 
					 ? StorageObjectSpec::bctf
					 : StorageObjectSpec::encoding)) {
	  decoder_ = new MappingDecoder(decoder_, map_);
	  if (maySetDocCharset_) {
	    sov_[soIndex_] = new UnbufferingStorageObject(sov_[soIndex_].extract(), &maySetDocCharset_);
	    so_ = sov_[soIndex_].pointer();
	  }
	}
	info_->setDecoder(soIndex_, decoder_);
	zapEof_ = spec.zapEof;
	switch (spec.records) {
	case StorageObjectSpec::asis:
	  recordType_ = asis;
	  insertRS_ = false;
	  break;
	case StorageObjectSpec::cr:
	  recordType_ = cr;
	  break;
	case StorageObjectSpec::lf:
	  recordType_ = lf;
	  break;
	case StorageObjectSpec::crlf:
	  recordType_ = crlf;
	  break;
	case StorageObjectSpec::find:
	  recordType_ = unknown;
	  break;
	default:
	  CANNOT_HAPPEN();
	}
	soIndex_++;
	readSize_ = so_->getBlockSize();
	nLeftOver_ = 0;
	break;
      }
      else
	setAccessError();
      soIndex_++;
    }

    size_t keepSize = end() - start();
    const size_t align = sizeof(int)/sizeof(Char);
    size_t readSizeChars = (readSize_ + (sizeof(Char) - 1))/sizeof(Char);
    readSizeChars = roundUp(readSizeChars, align);
    size_t neededSize;		// in Chars
    size_t startOffset;
    // compute neededSize and readSize
    unsigned minBytesPerChar = decoder_->minBytesPerChar();
    if (nLeftOver_ == 0 && minBytesPerChar >= sizeof(Char)) {
      // In this case we want to do decoding in place.
      // FIXME It might be a win on some systems (Irix?) to arrange that the
      // read buffer is on a page boundary.

      if (keepSize >= size_t(-1)/sizeof(Char) - (align - 1) - insertRS_)
	abort();			// FIXME throw an exception
      
      // Now size_t(-1)/sizeof(Char) - (align - 1) - insertRS_ - keepSize > 0
      if (readSizeChars
	  > size_t(-1)/sizeof(Char) - (align - 1) - insertRS_ - keepSize)
	abort();
      neededSize = roundUp(readSizeChars + keepSize + insertRS_, align);
      startOffset = ((neededSize > bufSize_ ? neededSize : bufSize_)
		     - readSizeChars - insertRS_ - keepSize);
    }
    else {
      // Needs to be room for everything before decoding.
      neededSize = (keepSize + insertRS_ + readSizeChars
		    + (nLeftOver_ + sizeof(Char) - 1)/sizeof(Char));
      // Also must be room for everything after decoding.
      size_t neededSize2
	= (keepSize + insertRS_
	   // all the converted characters
	   + (nLeftOver_ + readSize_)/minBytesPerChar
	   // enough Chars to contain left over bytes
	   + ((readSize_ % minBytesPerChar + sizeof(Char) - 1)
	      / sizeof(Char)));
      if (neededSize2 > neededSize)
	neededSize = neededSize2;
      neededSize = roundUp(neededSize, align);
      if (neededSize > size_t(-1)/sizeof(Char))
	abort();
      startOffset = 0;
    }
    if (bufSize_ < neededSize)
      reallocateBuffer(neededSize);
    Char *newStart = buf_ + startOffset;
    if (newStart != start() && keepSize > 0)
      memmove(newStart, start(), keepSize*sizeof(Char));
    char *bytesStart = (char *)(buf_ + bufSize_ - readSizeChars) - nLeftOver_;
    if (nLeftOver_ > 0 && leftOver_ != bytesStart)
      memmove(bytesStart, leftOver_, nLeftOver_);
    moveStart(newStart);
    bufLim_ = end();

    size_t nread;
    if (so_->read((char *)(buf_ + bufSize_ - readSizeChars), readSize_,
		  mgr, nread)) {
      if (nread > 0) {
	const char *bytesEnd = bytesStart + nLeftOver_ + nread;
	size_t nChars = decoder_->decode((Char *)end() + insertRS_,
					 bytesStart,
					 nLeftOver_ + nread
					 - (zapEof_ && bytesEnd[-1] == EOFCHAR),
					 &leftOver_);
	nLeftOver_ = bytesEnd - leftOver_;
	if (nChars > 0) {
	  if (insertRS_) {
	    noteRS();
	    *(Char *)end() = RS;
	    advanceEnd(end() + 1);
	    insertRS_ = false;
	    bufLim_ += 1;
	    bufLimOffset_ += 1;
	  }
	  bufLim_ += nChars;
	  bufLimOffset_ += nChars;
	  break;
	}
      }
    }
    else
      so_ = 0;
  }
  ASSERT(end() < bufLim_);
  if (insertRS_) {
    noteRS();
    insertChar(RS);
    insertRS_ = false;
    bufLimOffset_ += 1;
  }
  switch (recordType_) {
  case unknown:
    {
      const Char *e = findNextCrOrLf(end(), bufLim_);
      if (e) {
	if (*e == '\n') {
	  recordType_ = lf;
	  info_->noteInsertedRSs();
	  *(Char *)e = RE;
	  advanceEnd(e + 1);
	  insertRS_ = true;
	}
	else {
	  if (e + 1 < bufLim_) {
	    if (e[1] == '\n') {
	      recordType_ = crlf;
	      advanceEnd(e + 1);
	      if (e + 2 == bufLim_) {
		bufLim_--;
		bufLimOffset_--;
		insertRS_ = true;
	      }
	    }
	    else {
	      advanceEnd(e + 1);
	      recordType_ = cr;
	      info_->noteInsertedRSs();
	      insertRS_ = true;
	    }
	  }
	  else {
	    recordType_ = crUnknown;
	    advanceEnd(e + 1);
	  }
	}
      }
      else
	advanceEnd(bufLim_);
    }
    break;
  case crUnknown:
    {
      if (*cur() == '\n') {
	noteRS();
	advanceEnd(cur() + 1);
	recordType_ = crlf;
      }
      else {
	advanceEnd(cur() + 1);
	insertRS_ = true;
	recordType_ = cr;
	info_->noteInsertedRSs();
      }
    }
    break;
  case lf:
    {
      Char *e = (Char *)findNextLf(end(), bufLim_);
      if (e) {
	advanceEnd(e + 1);
	*e = RE;
	insertRS_ = true;
      }
      else
	advanceEnd(bufLim_);
    }
    break;
  case cr:
    {
      const Char *e = findNextCr(end(), bufLim_);
      if (e) {
	advanceEnd(e + 1);
	insertRS_ = true;
      }
      else
	advanceEnd(bufLim_);
    }
    break;
  case crlf:
    {
      const Char *e = end();
      for (;;) {
	e = findNextLf(e, bufLim_);
	if (!e) {
	  advanceEnd(bufLim_);
	  break;
	}
	// Need to delete final RS if not followed by anything.
	if (e + 1 == bufLim_) {
	  bufLim_--;
	  bufLimOffset_--;
	  advanceEnd(e);
	  insertRS_ = true;
	  if (cur() == end())
	    return fill(mgr);
	  break;
	}
	noteRSAt(e);
	e++;
      }
    }
    break;
  case asis:
    advanceEnd(bufLim_);
    break;
  default:
    CANNOT_HAPPEN();
  }
  ASSERT(cur() < end());
  return nextChar();
}

const Char *ExternalInputSource::findNextCr(const Char *start,
					    const Char *end)
{
  for (; start < end; start++)
    if (*start == '\r')
      return start;
  return 0;
}

const Char *ExternalInputSource::findNextLf(const Char *start,
					    const Char *end)
{
  for (; start < end; start++)
    if (*start == '\n')
      return start;
  return 0;
}

const Char *ExternalInputSource::findNextCrOrLf(const Char *start,
						const Char *end)
{
  for (; start < end; start++)
    if (*start == '\n' || *start == '\r')
      return start;
  return 0;
}

void ExternalInputSource::pushCharRef(Char ch, const NamedCharRef &ref)
{
  ASSERT(cur() == start());
  noteCharRef(startIndex() + (cur() - start()), ref);
  insertChar(ch);
}

void ExternalInputSource::insertChar(Char ch)
{
  if (start() > buf_) {
    if (cur() > start())
      memmove((Char *)start() - 1, start(), (cur() - start())*sizeof(Char));
    moveLeft();
    *(Char *)cur() = ch;
  }
  else {
    // must have start == buf
    if (buf_ + (bufSize_ - (nLeftOver_ + sizeof(Char) - 1)/sizeof(Char))
	== bufLim_) {
      if (bufSize_ == size_t(-1))
	abort();		// FIXME throw an exception
      reallocateBuffer(bufSize_ + 1);
    }
    else if (nLeftOver_ > 0 && ((char *)(bufLim_ + 1) > leftOver_)) {
      char *s = (char *)(buf_ + bufSize_) - nLeftOver_;
      memmove(s, leftOver_, nLeftOver_);
      leftOver_ = s;
    }
    if (cur() < bufLim_)
      memmove((Char *)cur() + 1, cur(), (bufLim_ - cur())*sizeof(Char));
    *(Char *)cur() = ch;
    advanceEnd(end() + 1);
    bufLim_ += 1;
  }
}

void ExternalInputSource::reallocateBuffer(size_t newSize)