soentitycatalog.cxx

SP是一个基于GNU C++编译器

SOEntityCatalog::SOEntityCatalog(Ptr<ExtendEntityManager> em)
: em_(em), catalogNumber_(0), haveSgmlDecl_(0), haveDocument_(0),
  haveCurrentBase_(0)
{
}

void SOEntityCatalog::endCatalog()
{
  catalogNumber_++;
  haveCurrentBase_ = 0;
}

Boolean SOEntityCatalog::expandCatalogSystemId(const StringC &str,
					       const Location &loc,
					       size_t baseNumber,
					       Boolean isNdata,
					       const CharsetInfo &charset,
					       const StringC *lookupPublicId,
					       Messenger &mgr,
					       StringC &result) const
{
  return em_->expandSystemId(str,
			     (baseNumber ? base_[baseNumber - 1] : loc),
			     isNdata,
			     charset,
			     lookupPublicId,
			     mgr,
			     result);
}

Boolean SOEntityCatalog::lookup(const EntityDecl &entity,
				const Syntax &syntax,
				const CharsetInfo &charset,
				Messenger &mgr,
				StringC &result) const
{
  const CatalogEntry *entry = 0;
  const CatalogEntry *delegatedEntry = 0;
  if (entity.systemIdPointer())
    entry = systemIds_.lookup(*entity.systemIdPointer());
  if (entity.publicIdPointer()) {
    const CatalogEntry *publicEntry;
    Boolean delegated;
    publicEntry = findBestPublicEntry(*entity.publicIdPointer(),
				      entity.systemIdPointer() != 0,
				      charset,
				      delegated);
    if (publicEntry && delegated)
      delegatedEntry = publicEntry;
    // match for system id has priority over match for public id in same
    // catalog
    if (publicEntry
	&& (!entry || publicEntry->catalogNumber < entry->catalogNumber))
      entry = publicEntry;
  }
  if (entity.name().size() > 0
      && (!entry || entry->catalogNumber > 0)) {
    const CatalogEntry *entityEntry; 
    int tableIndex = (entity.declType() >= EntityDecl::parameterEntity
		      ? int(entity.declType()) - 1
		      : int(entity.declType()));
    StringC name(entity.name());
    Boolean subst;
    switch (entity.declType()) {
    case EntityDecl::parameterEntity:
      {
	StringC tem(name);
	name = syntax.peroDelim();
	name += tem;
      }
      // fall through
    case EntityDecl::generalEntity:
      subst = syntax.namecaseEntity();
      break;
    default:
      subst = syntax.namecaseGeneral();
      break;
    }
    if (!subst)
      entityEntry = names_[tableIndex].lookup(name,
					      entity.systemIdPointer() != 0);
    else
      entityEntry = names_[tableIndex].lookup(name,
					      syntax.upperSubstTable(),
					      entity.systemIdPointer() != 0);
    // match for public id has priority over match for entity in same
    // catalog
    if (entityEntry
	&& (!entry || entityEntry->catalogNumber < entry->catalogNumber))
      entry = entityEntry;
  }
  if (entry)
    return expandCatalogSystemId(entry->to,
				 entry->loc,
				 entry->baseNumber,
				 entity.dataType() == EntityDecl::ndata,
				 charset,
				 entry == delegatedEntry
				 ? entity.publicIdPointer()
				 : 0,
				 mgr,
				 result);
  if (entity.systemIdPointer())
    return em_->expandSystemId(*entity.systemIdPointer(),
			       entity.defLocation(),
			       entity.dataType() == EntityDecl::ndata,
			       charset,
			       0,
			       mgr,
			       result);
  return 0;
}

Boolean SOEntityCatalog::lookupPublic(const StringC &publicId,
				      const CharsetInfo &charset,
				      Messenger &mgr,
				      StringC &result) const
{
  Boolean delegated;
  const CatalogEntry *entry = findBestPublicEntry(publicId, 0, charset,
						  delegated);
  return (entry
	  && expandCatalogSystemId(entry->to, entry->loc, entry->baseNumber,
				   0, charset, delegated ? &publicId : 0,
				   mgr, result));
				 
}

Boolean SOEntityCatalog::lookupChar(const StringC &name,
				    const CharsetInfo &charset,
				    Messenger &mgr,
				    UnivChar &result) const
{
  Boolean delegated;
  const CatalogEntry *entry = findBestPublicEntry(name, 0, charset,
						  delegated);
  if (!entry)
    return 0;
  if (delegated)
    return 0;  // FIXME
  const StringC &number = entry->to;
  if (number.size() == 0)
    return 0;
  UnivChar n = 0;
  for (size_t i = 0; i < number.size(); i++) {
    int d = charset.digitWeight(number[i]);
    if (d < 0)
      return 0;
    if (n <= UnivChar(-1)/10 && (n *= 10) <= UnivChar(-1) - d)
      n += d;
  }
  result = n;
  return 1;
}

const CatalogEntry *
SOEntityCatalog::findBestPublicEntry(const StringC &publicId,
				     Boolean overrideOnly,
				     const CharsetInfo &charset,
				     Boolean &delegated) const
{
  Char slash = charset.execToDesc('/');
  Char colon = charset.execToDesc(':');
  const CatalogEntry *bestEntry = 0;
  for (size_t i = 0; i <= publicId.size(); i++) {
    if ((i + 1 < publicId.size()
	 && (publicId[i] == slash || publicId[i] == colon)
	 && publicId[i + 1] == publicId[i])
	|| (i >= 2
	    && (publicId[i - 1] == slash || publicId[i - 1] == colon)
	    && publicId[i - 2] == publicId[i - 1])) {
      StringC tem(publicId.data(), i);
      const CatalogEntry *entry = delegates_.lookup(tem, overrideOnly);
      if (entry
	  && (!bestEntry
	      || entry->catalogNumber <= bestEntry->catalogNumber)) {
	bestEntry = entry;
	delegated = 1;
      }
    }
  }
  const CatalogEntry *entry = publicIds_.lookup(publicId, overrideOnly);
  if (entry
      && (!bestEntry || entry->catalogNumber <= bestEntry->catalogNumber)) {
    bestEntry = entry;
    delegated = 0;
  }
  return bestEntry;
}

Boolean SOEntityCatalog::sgmlDecl(const CharsetInfo &charset,
				  Messenger &mgr,
				  StringC &result) const

{
  return haveSgmlDecl_ && expandCatalogSystemId(sgmlDecl_, sgmlDeclLoc_,
						sgmlDeclBaseNumber_,
						0, charset, 0, mgr, result);
}

Boolean SOEntityCatalog::document(const CharsetInfo &charset,
				  Messenger &mgr,
				  StringC &result) const

{
  return haveDocument_ && expandCatalogSystemId(document_, documentLoc_,
						documentBaseNumber_,
						0, charset, 0, mgr, result);
}

void SOEntityCatalog::addPublicId(StringC &publicId, StringC &systemId,
				  const Location &loc, Boolean override)
{
  CatalogEntry entry;
  entry.loc = loc;
  entry.catalogNumber = catalogNumber_;
  entry.baseNumber = haveCurrentBase_ ? base_.size() : 0;
  systemId.swap(entry.to);
  publicIds_.insert(publicId, entry, override);
}

void SOEntityCatalog::addDelegate(StringC &prefix, StringC &systemId,
				  const Location &loc, Boolean override)
{
  CatalogEntry entry;
  entry.loc = loc;
  entry.catalogNumber = catalogNumber_;
  entry.baseNumber = haveCurrentBase_ ? base_.size() : 0;
  systemId.swap(entry.to);
  delegates_.insert(prefix, entry, override);
}

void SOEntityCatalog::addSystemId(StringC &systemId, StringC &toSystemId,
				  const Location &loc)
{
  CatalogEntry entry;
  entry.loc = loc;
  entry.catalogNumber = catalogNumber_;
  entry.baseNumber = haveCurrentBase_ ? base_.size() : 0;
  toSystemId.swap(entry.to);
  systemIds_.insert(systemId, entry, false);
}

void SOEntityCatalog::addName(StringC &name, DeclType declType,
			      StringC &systemId, const Location &loc,
			      Boolean override)
{
  CatalogEntry entry;
  entry.loc = loc;
  entry.catalogNumber = catalogNumber_;
  entry.baseNumber = haveCurrentBase_ ? base_.size() : 0;
  int tableIndex = (declType >= EntityDecl::parameterEntity
		    ? int(declType) - 1
		    : int(declType));
  entry.serial = names_[tableIndex].count();
  systemId.swap(entry.to);
  names_[tableIndex].insert(name, entry, override);
}

void SOEntityCatalog::setSgmlDecl(StringC &str, const Location &loc)
{
  if (!haveSgmlDecl_) {
    haveSgmlDecl_ = true;
    str.swap(sgmlDecl_);
    sgmlDeclLoc_ = loc;
    sgmlDeclBaseNumber_ = haveCurrentBase_ ? base_.size() : 0;

  }
}

void SOEntityCatalog::setDocument(StringC &str, const Location &loc)
{
  if (!haveDocument_) {
    haveDocument_ = true;
    str.swap(document_);
    documentLoc_ = loc;
    documentBaseNumber_ = haveCurrentBase_ ? base_.size() : 0;
  }
}

void SOEntityCatalog::setBase(const Location &loc)
{
  if (loc.origin().isNull())
    haveCurrentBase_ = 0;
  else {
    haveCurrentBase_ = 1;
    base_.push_back(loc);
  }
}

SOEntityCatalog::Table::Table()
{
}

void SOEntityCatalog::Table::insert(const StringC &key,
				    const CatalogEntry &entry,
				    Boolean override)
{
  if (override)
    overrideEntries_.insert(key, entry, false);
  else {
    const CatalogEntry *e = overrideEntries_.lookup(key);
    if (!e)
      normalEntries_.insert(key, entry, false);
  }
}

const CatalogEntry *SOEntityCatalog::Table::lookup(const StringC &key,
						   Boolean overrideOnly) const
{
  if (!overrideOnly) {
    const CatalogEntry *e = normalEntries_.lookup(key);
    if (e)
      return e;
  }
  return overrideEntries_.lookup(key);
}

const CatalogEntry *
SOEntityCatalog::Table::lookup(const StringC &name,
			       const SubstTable<Char> &substTable,
			       Boolean overrideOnly) const
{
  HashTableIter<StringC,CatalogEntry> iter1(overrideEntries_);
  HashTableIter<StringC,CatalogEntry> iter2(normalEntries_);
  HashTableIter<StringC,CatalogEntry> *iters[2];
  int nIter = 0;
  iters[nIter++] = &iter1;
  if (!overrideOnly)
    iters[nIter++] = &iter2;
  const CatalogEntry *entry = 0;
  for (int i = 0; i < nIter; i++) {
    HashTableIter<StringC,CatalogEntry> &iter = *iters[i];
    const StringC *key;
    const CatalogEntry *value;
    StringC buffer;
    while (iter.next(key, value)) {
      buffer = *key;
      for (size_t j = 0; j < buffer.size(); j++)
	substTable.subst(buffer[j]);
      if (buffer == name) {
	if (!entry || value->serial < entry->serial)
	  entry = value;
      }
    }
  }
  return entry;
}

size_t SOEntityCatalog::Table::count() const
{
  return normalEntries_.count() + overrideEntries_.count();
}

CatalogParser::CatalogParser(const CharsetInfo &charset)
: categoryTable_(data),
  entityKey_(charset.execToDesc("ENTITY")),
  publicKey_(charset.execToDesc("PUBLIC")),
  systemKey_(charset.execToDesc("SYSTEM")),
  doctypeKey_(charset.execToDesc("DOCTYPE")),
  linktypeKey_(charset.execToDesc("LINKTYPE")),
  notationKey_(charset.execToDesc("NOTATION")),
  overrideKey_(charset.execToDesc("OVERRIDE")),
  sgmlDeclKey_(charset.execToDesc("SGMLDECL")),
  documentKey_(charset.execToDesc("DOCUMENT")),
  catalogKey_(charset.execToDesc("CATALOG")),
  yesKey_(charset.execToDesc("YES")),
  noKey_(charset.execToDesc("NO")),
  baseKey_(charset.execToDesc("BASE")),
  delegateKey_(charset.execToDesc("DELEGATE")),
  dtddeclKey_(charset.execToDesc("DTDDECL")),
  sgmlKey_(charset.execToDesc("SGML"))
{
  static const char lcletters[] = "abcdefghijklmnopqrstuvwxyz";
  static const char ucletters[] = "ABCDEFGHIJKLMNOPQRSTUVWXYZ";
  // minimum data other than lcletter, ucletter
  static const char minChars[] = "0123456789-.'()+,/:=?";
  static const char wwwMinChars[] = { 33, 35, 36, 37, 42, 59, 64, 95, 0 };
  static const char sChars[] = " \n\r\t";
  categoryTable_.setChar(0, nul);
  const char *p;
  const char *q;
  for (p = lcletters, q = ucletters; *p; p++, q++) {
    Char lc = charset.execToDesc(*p);
    Char uc = charset.execToDesc(*q);