parsesd.cxx

SP是一个基于GNU C++编译器

    Boolean givenError;
    if (referencePublic(id, PublicId::CHARSET, givenError))
      found = sdParseExternalCharset(*sdBuilder.sd, baseDesc);
    else if (!givenError) {
      found = 0;
      PublicId::OwnerType ownerType;
      if (id.getOwnerType(ownerType) && ownerType == PublicId::ISO) {
	StringC sequence;
	if (id.getDesignatingSequence(sequence)) {
	  CharsetRegistry::ISORegistrationNumber number
	   = CharsetRegistry::getRegistrationNumber(sequence, sd().internalCharset());
	  if (number != CharsetRegistry::UNREGISTERED) {
	    Owner<CharsetRegistry::Iter> iter(CharsetRegistry::makeIter(number));
	    if (iter) {
	      found = 1;
	      WideChar min;
	      WideChar max;
	      UnivChar univ;
	      while (iter->next(min, max, univ))
		baseDesc.addRange(min, max, univ);
	    }
	  }
	}
      }
      if (!found)
	message(ParserMessages::unknownBaseset, StringMessageArg(id.string()));
    }
    else
      found = 0;
    if (!found)
      maybeISO646 = 0;
    decl.addSection(id);
    if (!parseSdParam(AllowedSdParams(SdParam::reservedName + Sd::rDESCSET),
		      parm))
      return 0;
    if (!parseSdParam(AllowedSdParams(SdParam::number), parm))
      return 0;
    do {
      WideChar min = parm.n;
      if (!parseSdParam(AllowedSdParams(SdParam::number), parm))
	return 0;
      Number count = parm.n;
      Number adjCount;
      if (options().warnSgmlDecl && count == 0)
	message(ParserMessages::zeroNumberOfCharacters);
      decl.rangeDeclared(min, count, multiplyDeclared);
      if (isDocument
	  && count > 0
	  && (min > charMax || count - 1 > charMax - min)) {
	message(ParserMessages::documentCharMax, NumberMessageArg(charMax));
	adjCount = min > charMax ? 0 : 1 + (charMax - min);
	maybeISO646 = 0;
      }
      else
	adjCount = count;
      if (!parseSdParam(AllowedSdParams(SdParam::number,
					SdParam::minimumLiteral,
					SdParam::reservedName + Sd::rUNUSED),
			parm))
	return 0;
      switch (parm.type) {
      case SdParam::number:
	decl.addRange(min, count, parm.n);
	if (found && adjCount > 0) {
	  ISet<WideChar> baseMissing;
	  desc.addBaseRange(baseDesc, min, min + (adjCount - 1), parm.n,
			    baseMissing);
	  if (!baseMissing.isEmpty() && options().warnSgmlDecl)
	    message(ParserMessages::basesetCharsMissing,
		    CharsetMessageArg(baseMissing));
	}
	break;
      case SdParam::reservedName + Sd::rUNUSED:
	decl.addRange(min, count);
	break;
      case SdParam::minimumLiteral:
	{
	  UnivChar c = charNameToUniv(*sdBuilder.sd, parm.literalText.string());
	  if (adjCount > 256) {
	    message(ParserMessages::tooManyCharsMinimumLiteral);
	    adjCount = 256;
	  }
	  for (Number i = 0; i < adjCount; i++)
	    desc.addRange(min + i, min + i, c);
	}
	maybeISO646 = 0;
        decl.addRange(min, count, parm.literalText.string());
        break;
      default:
        CANNOT_HAPPEN();			 
      }
      SdParam::Type follow = (isDocument
			      ? SdParam::reservedName + Sd::rCAPACITY
			      : SdParam::reservedName + Sd::rFUNCTION);
      if (!parseSdParam(AllowedSdParams(SdParam::number,
					SdParam::reservedName + Sd::rBASESET,
					follow),
			parm))
	return 0;
	
    } while (parm.type == SdParam::number);
  } while (parm.type == SdParam::reservedName + Sd::rBASESET);
  if (!multiplyDeclared.isEmpty())
    message(ParserMessages::duplicateCharNumbers,
	    CharsetMessageArg(multiplyDeclared));
  ISet<WideChar> declaredSet;
  decl.declaredSet(declaredSet);
  ISetIter<WideChar> iter(declaredSet);
  WideChar min, max, lastMax;
  if (iter.next(min, max)) {
    ISet<WideChar> holes;
    lastMax = max;
    while (iter.next(min, max)) {
      if (min - lastMax > 1)
	holes.addRange(lastMax + 1, min - 1);
      lastMax = max;
    }
    if (!holes.isEmpty())
      message(ParserMessages::codeSetHoles, CharsetMessageArg(holes));
  }
  if (!isDocument && sdBuilder.sd->scopeInstance()) {
    // If scope is INSTANCE, syntax reference character set
    // must be same as reference.
    UnivCharsetDescIter iter(desc);
    WideChar descMin, descMax;
    UnivChar univMin;
    Char nextDescMin = 0;
    while (maybeISO646) {
      if (!iter.next(descMin, descMax, univMin)) {
	if (nextDescMin != 128)
	  maybeISO646 = 0;
	break;
      }
      if (descMin != nextDescMin || univMin != descMin)
	maybeISO646 = 0;
      nextDescMin = descMax + 1;
    }
    if (!maybeISO646)
      message(ParserMessages::scopeInstanceSyntaxCharset);
  }
  return 1;
}

Boolean Parser::sdParseExternalCharset(Sd &sd, UnivCharsetDesc &desc)
{
  SdParam parm;
  for (;;) {
    if (!parseSdParam(AllowedSdParams(SdParam::number, SdParam::eE),
		     parm))
      break;
    if (parm.type == SdParam::eE)
      return 1;
    WideChar min = parm.n;
    if (!parseSdParam(AllowedSdParams(SdParam::number), parm))
      break;
    Number count = parm.n;
    if (!parseSdParam(AllowedSdParams(SdParam::number,
				      SdParam::minimumLiteral,
				      SdParam::reservedName + Sd::rUNUSED),
		     parm))
      break;
    if (parm.type == SdParam::number) {
      if (count > 0)
	desc.addRange(min, min + (count - 1), parm.n);
    }
    else if (parm.type == SdParam::minimumLiteral) {
      UnivChar c = charNameToUniv(sd, parm.literalText.string());
      if (count > 256) {
	message(ParserMessages::tooManyCharsMinimumLiteral);
	count = 256;
      }
      for (Number i = 0; i < count; i++)
	desc.addRange(min + i, min + i, c);
    }
  }
  popInputStack();
  return 0;
}

UnivChar Parser::charNameToUniv(Sd &sd, const StringC &name)
{
  UnivChar univ;
  if (entityCatalog().lookupChar(name, sd.internalCharset(), messenger(), univ))
    return univ;
  else
    return sd.nameToUniv(name);
}

Boolean Parser::sdParseCapacity(SdBuilder &sdBuilder, SdParam &parm)
{
  if (!parseSdParam(sdBuilder.www
                    ? AllowedSdParams(SdParam::reservedName + Sd::rNONE,
				      SdParam::reservedName + Sd::rPUBLIC,
                                      SdParam::reservedName + Sd::rSGMLREF)
		    : AllowedSdParams(SdParam::reservedName + Sd::rPUBLIC,
				      SdParam::reservedName + Sd::rSGMLREF),
		    parm))
    return 0;
#if _MSC_VER == 1100
  // Workaround for Visual C++ 5.0 bug
  int
#else
  Boolean
#endif
    pushed = 0;
  if (parm.type == SdParam::reservedName + Sd::rNONE)
    return parseSdParam(AllowedSdParams(SdParam::reservedName + Sd::rSCOPE),
			parm);
  if (parm.type == SdParam::reservedName + Sd::rPUBLIC) {
    if (!parseSdParam(AllowedSdParams(SdParam::minimumLiteral), parm))
      return 0;
    PublicId id;
    PublicId::TextClass textClass;
    const MessageType1 *err;
    if (!id.init(parm.literalText, sd().internalCharset(), syntax().space(), err))
      sdBuilder.addFormalError(currentLocation(),
			       *err,
			       id.string());
    else if (id.getTextClass(textClass)
	     && textClass != PublicId::CAPACITY)
      sdBuilder.addFormalError(currentLocation(),
			       ParserMessages::capacityTextClass,
			       id.string());
    const StringC &str = id.string();
    if (str != sd().execToInternal("ISO 8879-1986//CAPACITY Reference//EN")
	&& str != sd().execToInternal("ISO 8879:1986//CAPACITY Reference//EN")) {
      Boolean givenError;
      if (referencePublic(id, PublicId::CAPACITY, givenError))
	pushed = 1;
      else if (!givenError)
	message(ParserMessages::unknownCapacitySet, StringMessageArg(str));
    }
    if (!pushed)
      return parseSdParam(AllowedSdParams(SdParam::reservedName + Sd::rSCOPE),
			  parm);
  }

  PackedBoolean capacitySpecified[Sd::nCapacity];
  int i;
  for (i = 0; i < Sd::nCapacity; i++)
    capacitySpecified[i] = 0;
  if (!parseSdParam(AllowedSdParams(SdParam::capacityName), parm))
    return 0;
  do {
    Sd::Capacity capacityIndex = parm.capacityIndex;
    if (!parseSdParam(AllowedSdParams(SdParam::number), parm))
      return 0;
    
    if (!capacitySpecified[capacityIndex]) {
      sdBuilder.sd->setCapacity(capacityIndex, parm.n);
      capacitySpecified[capacityIndex] = 1;
    }
    else if (options().warnSgmlDecl)
      message(ParserMessages::duplicateCapacity,
	      StringMessageArg(sd().capacityName(i)));
    int final = pushed ? int(SdParam::eE) : SdParam::reservedName + Sd::rSCOPE;
    if (!parseSdParam(AllowedSdParams(SdParam::capacityName, final),
		      parm))
      return 0;
  } while (parm.type == SdParam::capacityName);
  Number totalcap = sdBuilder.sd->capacity(0);
  for (i = 1; i < Sd::nCapacity; i++)
    if (sdBuilder.sd->capacity(i) > totalcap)
      message(ParserMessages::capacityExceedsTotalcap,
	      StringMessageArg(sd().capacityName(i)));
  if (pushed)
    return parseSdParam(AllowedSdParams(SdParam::reservedName + Sd::rSCOPE),
			parm);
  return 1;
}

Boolean Parser::referencePublic(const PublicId &id,
				PublicId::TextClass entityType,
				Boolean &givenError)
{
  givenError = 0;
  StringC sysid;
  if (entityCatalog().lookupPublic(id.string(),
				   sd().internalCharset(),
				   messenger(),
				   sysid)) {
    Location loc = currentLocation();
    eventHandler().sgmlDeclEntity(new (eventAllocator())
				  SgmlDeclEntityEvent(id,
						      entityType,
						      sysid,
						      loc));
    Ptr<EntityOrigin> origin(EntityOrigin::make(internalAllocator(),
						ConstPtr<Entity>(0),
						loc));
    if (currentMarkup())
      currentMarkup()->addEntityStart(origin);
    InputSource *in = entityManager().open(sysid,
					   sd().docCharset(),
					   origin.pointer(),
					   0,
					   messenger());
    if (!in) {
      givenError = 1;
      return 0;
    }
    pushInput(in);
    return 1;
  }
  return 0;
}

Boolean Parser::sdParseScope(SdBuilder &sdBuilder, SdParam &parm)
{
  if (!parseSdParam(AllowedSdParams(SdParam::reservedName + Sd::rINSTANCE,
				    SdParam::reservedName + Sd::rDOCUMENT),
		    parm))
    return 0;
  if (parm.type == SdParam::reservedName + Sd::rINSTANCE)
    sdBuilder.sd->setScopeInstance();
  return 1;
}

Boolean Parser::sdParseSyntax(SdBuilder &sdBuilder, SdParam &parm)
{
  if (!parseSdParam(AllowedSdParams(SdParam::reservedName + Sd::rSYNTAX),
		    parm))
    return 0;
  if (!parseSdParam(AllowedSdParams(SdParam::reservedName + Sd::rSHUNCHAR,
				    SdParam::reservedName + Sd::rPUBLIC),
		    parm))
    return 0;

  if (parm.type == SdParam::reservedName + Sd::rPUBLIC) {
    if (!parseSdParam(AllowedSdParams(SdParam::minimumLiteral), parm))
      return 0;
    PublicId id;
    const MessageType1 *err;
    PublicId::TextClass textClass;
    if (!id.init(parm.literalText, sd().internalCharset(), syntax().space(), err))
      sdBuilder.addFormalError(currentLocation(),
			       *err,
			       id.string());
    else if (id.getTextClass(textClass)
	     && textClass != PublicId::SYNTAX)
      sdBuilder.addFormalError(currentLocation(),
			       ParserMessages::syntaxTextClass,
			       id.string());
    if (!parseSdParam(AllowedSdParams(SdParam::reservedName + Sd::rFEATURES,
				      SdParam::reservedName + Sd::rSWITCHES),
		      parm))
      return 0;
    Vector<UnivChar> charSwitches;
    if (parm.type == SdParam::reservedName + Sd::rSWITCHES) {
      if (!parseSdParam(AllowedSdParams(SdParam::number), parm))
	return 0;
      for (;;) {
	SyntaxChar c = parm.n;
	if (!parseSdParam(AllowedSdParams(SdParam::number), parm))
	  return 0;
	sdBuilder.switcher.addSwitch(c, parm.n);
	if (!parseSdParam(AllowedSdParams(SdParam::number,
					  SdParam::reservedName
					  + Sd::rFEATURES),
			  parm))
	  return 0;
	if (parm.type != SdParam::number)
	  break;
      }
    }
    const StandardSyntaxSpec *spec = lookupSyntax(id);
    if (spec) {
      if (!setStandardSyntax(*sdBuilder.syntax,
			     *spec,
			     sdBuilder.sd->internalCharset(),
			     sdBuilder.switcher,
			     sdBuilder.www))
	sdBuilder.valid = 0;
    }
    else {
      Boolean givenError;
      if (referencePublic(id, PublicId::SYNTAX, givenError)) {
	sdBuilder.externalSyntax = 1;
	SdParam parm2;
	if (!parseSdParam(AllowedSdParams(SdParam::reservedName
					  + Sd::rSHUNCHAR),
			  parm2))
	  return 0;
	if (!sdParseExplicitSyntax(sdBuilder, parm2))
	  return 0;
      }
      else {
	if (!givenError)
	  message(ParserMessages::unknownPublicSyntax,
		  StringMessageArg(id.string()));
	sdBuilder.valid = 0;
      }
    }
  }
  else {
    if (!sdParseExplicitSyntax(sdBuilder, parm))
      return 0;
  }
  if (!sdBuilder.sd->scopeInstance()) {
    // we know the significant chars now
    ISet<WideChar> invalidSgmlChar;
    sdBuilder.syntax->checkSgmlChar(*sdBuilder.sd,
				    0,
				    1,
				    invalidSgmlChar);
    if (!invalidSgmlChar.isEmpty())
      message(ParserMessages::invalidSgmlChar, CharsetMessageArg(invalidSgmlChar));
  }
  checkSyntaxNamelen(*sdBuilder.syntax);
  checkSwitchesMarkup(sdBuilder.switcher);
  return 1;
}

Boolean Parser::sdParseExplicitSyntax(SdBuilder &sdBuilder,
				      SdParam &parm)
{
  typedef Boolean (Parser::*SdParser)(SdBuilder &, SdParam &);
  static SdParser parsers[] = {
    &Parser::sdParseShunchar,
    &Parser::sdParseSyntaxCharset,
    &Parser::sdParseFunction,
    &Parser::sdParseNaming,
    &Parser::sdParseDelim,
    &Parser::sdParseNames,
    &Parser::sdParseQuantity
    };
  for (size_t i = 0; i < SIZEOF(parsers); i++)
    if (!(this->*(parsers[i]))(sdBuilder, parm))
      return 0;
  return 1;
}

const StandardSyntaxSpec *Parser::lookupSyntax(const PublicId &id)
{
  PublicId::OwnerType ownerType;
  if (!id.getOwnerType(ownerType) || ownerType != PublicId::ISO)
    return 0;
  StringC str;
  if (!id.getOwner(str))
    return 0;
  if (str != sd().execToInternal("ISO 8879:1986")
      && str != sd().execToInternal("ISO 8879-1986"))
    return 0;
  PublicId::TextClass textClass;
  if (!id.getTextClass(textClass) || textClass != PublicId::SYNTAX)
    return 0;
  if (!id.getDescription(str))
    return 0;
  if (str == sd().execToInternal("Reference"))
    return &refSyntax;
  if (str == sd().execToInternal("Core"))
    return &coreSyntax;
  return 0;