soentitycatalog.cxx

SP是一个基于GNU C++编译器

// Copyright (c) 1994, 1995, 1996 James Clark
// See the file COPYING for copying permission.

#ifdef __GNUG__
#pragma implementation
#endif

#include "splib.h"
#include "CharsetInfo.h"
#include "MessageArg.h"
#include "CatalogMessages.h"
#include "SOEntityCatalog.h"
#include "EntityDecl.h"
#include "EntityCatalog.h"
#include "Message.h"
#include "StringC.h"
#include "types.h"
#include "HashTable.h"
#include "InputSource.h"
#include "Boolean.h"
#include "SubstTable.h"
#include "CatalogEntry.h"
#include "Vector.h"
#include "StorageManager.h"
#include "macros.h"

#ifdef SP_NAMESPACE
namespace SP_NAMESPACE {
#endif

class CatalogParser;
class SOEntityCatalog;

class SOCatalogManagerImpl : public SOCatalogManager {
public:
  SOCatalogManagerImpl(const Vector<StringC> &sysids,
		       size_t nSysidsMustExist,
		       const CharsetInfo *sysidCharset,
		       const CharsetInfo *catalogCharset,
		       Boolean useDocCatalog);
  ConstPtr<EntityCatalog> makeCatalog(StringC &systemId,
				      const CharsetInfo &charset,
				      ExtendEntityManager *,
				      Messenger &) const;
  Boolean mapCatalog(ParsedSystemId &systemId,
		     ExtendEntityManager *em,
		     Messenger &mgr) const;
private:
  void addCatalogsForDocument(CatalogParser &parser,
			      StringC &sysid,
			      SOEntityCatalog *,
			      const CharsetInfo &charset,
			      Messenger &mgr) const;
  size_t nSystemCatalogsMustExist_;
  Vector<StringC> systemCatalogs_;
  const CharsetInfo *sysidCharset_;
  const CharsetInfo *catalogCharset_;
  Boolean useDocCatalog_;
};

class SOEntityCatalog : public EntityCatalog {
public:
  SOEntityCatalog(Ptr<ExtendEntityManager> em);
  typedef EntityDecl::DeclType DeclType;
  Boolean document(const CharsetInfo &, Messenger &, StringC &) const;
  Boolean sgmlDecl(const CharsetInfo &, Messenger &, StringC &) const;
  Boolean lookup(const EntityDecl &entity,
		 const Syntax &,
		 const CharsetInfo &,
		 Messenger &,
		 StringC &) const;
  Boolean lookupPublic(const StringC &,
		       const CharsetInfo &,
		       Messenger &,
		       StringC &) const;
  Boolean lookupChar(const StringC &,
                     const CharsetInfo &,
		     Messenger &,
		     UnivChar &) const;
  void addPublicId(StringC &publicId, StringC &systemId, const Location &,
		   Boolean override);
  void addDelegate(StringC &prefix, StringC &systemId, const Location &,
		   Boolean override);
  void addSystemId(StringC &systemId, StringC &replSystemId, const Location &);
  void addName(StringC &name, DeclType, StringC &systemId, const Location &,
	       Boolean override);
  void setSgmlDecl(StringC &str, const Location &loc);
  void setDocument(StringC &str, const Location &loc);
  void setBase(const Location &loc);
  void endCatalog();
  const Ptr<ExtendEntityManager> &entityManager() {
    return em_;
  }
private:
  SOEntityCatalog(const SOEntityCatalog &);	// undefined
  void operator=(const SOEntityCatalog &);	// undefined

  Boolean expandCatalogSystemId(const StringC &str,
				const Location &loc,
				size_t baseNumber,
				Boolean isNdata,
				const CharsetInfo &charset,
				const StringC *lookupPublicId,
				Messenger &mgr,
				StringC &result) const;
  const CatalogEntry *
    findBestPublicEntry(const StringC &publicId, Boolean overrideOnly,
			const CharsetInfo &charset, Boolean &delegated) const;

  class Table {
  public:
    Table();
    const CatalogEntry *lookup(const StringC &, Boolean overrideOnly) const;
    const CatalogEntry *lookup(const StringC &key,
			       const SubstTable<Char> &substTable,
			       Boolean overrideOnly) const;
    void insert(const StringC &, const CatalogEntry &, Boolean override);
    size_t count() const;
  private:
    Table(const Table &);	// undefined
    void operator=(const Table &); // undefined
    // These are entries that are applicable when an explicit system id
    // was specified in the external identifier.
    HashTable<StringC,CatalogEntry> overrideEntries_;
    // This specifies the entries that should substitute for the
    // overrideEntries_ when an explicit system identifier was not specified.
    HashTable<StringC,CatalogEntry> normalEntries_;
  };

  Table publicIds_;
  Table delegates_;
  HashTable<StringC,CatalogEntry> systemIds_;
  Table names_[5];
  size_t catalogNumber_;
  Boolean haveSgmlDecl_;
  StringC sgmlDecl_;
  Location sgmlDeclLoc_;
  size_t sgmlDeclBaseNumber_;
  StringC document_;
  Boolean haveDocument_;
  Location documentLoc_;
  size_t documentBaseNumber_;
  Boolean haveCurrentBase_;
  Vector<Location> base_;
  Ptr<ExtendEntityManager> em_;
};

class CatalogParser : private Messenger {
public:
  CatalogParser(const CharsetInfo &);
  void parseCatalog(const StringC &sysid,
		    Boolean mustExist,
		    const CharsetInfo &sysidCharset,
		    const CharsetInfo &catalogCharset,
		    InputSourceOrigin *origin,
		    SOEntityCatalog *catalog,
		    Messenger &mgr);
public:
  // Since it's a return type, it has to be public to keep some 
  // (broken) compilers happy.
  enum Param {
    eofParam,
    literalParam,
    nameParam,
    percentParam
  };
private:
  enum {
    data,
    eof,
    nul,
    lit,
    lita,
    minus,
    s,
    min				// other minimum data characters
  };
  enum { minimumLiteral = 01 };

  Messenger &messenger() { return *this; }
  void dispatchMessage(Message &);
  void dispatchMessage(const Message &);
  void initMessage(Message &);
  void parsePublic();
  void parseDelegate();
  void parseDtddecl();
  void parseSystem();
  void parseNameMap(EntityDecl::DeclType declType);
  void parseOverride();
  Param parseParam(unsigned flags = 0);
  Boolean parseArg();
  void parseLiteral(Char delim, unsigned flags);
  void parseName();
  void skipComment();
  void upcase(StringC &);
  Boolean inLoop(const Location &loc);
  Boolean isMinimumData(Xchar c) {
    int cat = categoryTable_[c];
    return (cat == min || (cat == s && c != tab_)
	    || cat == minus || cat == lita);
  }
  Xchar get() { return in_->get(messenger()); }
  void unget() { in_->ungetToken(); }
  Messenger *mgr_;
  InputSource *in_;
  SOEntityCatalog *catalog_;
  StringC param_;
  Location paramLoc_;
  Char minus_;
  Char tab_;
  Char rs_;
  Char re_;
  Char space_;
  StringC publicKey_;
  StringC systemKey_;
  StringC entityKey_;
  StringC doctypeKey_;
  StringC linktypeKey_;
  StringC notationKey_;
  StringC overrideKey_;
  StringC sgmlDeclKey_;
  StringC documentKey_;
  StringC catalogKey_;
  StringC yesKey_;
  StringC noKey_;
  StringC baseKey_;
  StringC delegateKey_;
  StringC dtddeclKey_;
  StringC sgmlKey_;
  XcharMap<unsigned char> categoryTable_;
  SubstTable<Char> substTable_;
  Boolean override_;
};

ExtendEntityManager::CatalogManager *
SOCatalogManager::make(const Vector<StringC> &sysids,
		       size_t nSysidsMustExist,
		       const CharsetInfo *sysidCharset,
		       const CharsetInfo *catalogCharset,
		       Boolean useDocCatalog)
{
  return new SOCatalogManagerImpl(sysids,
				  nSysidsMustExist,
				  sysidCharset,
				  catalogCharset,
				  useDocCatalog);
}
		       
SOCatalogManagerImpl::SOCatalogManagerImpl(const Vector<StringC> &systemCatalogs,
					   size_t nSystemCatalogsMustExist,
					   const CharsetInfo *sysidCharset,
					   const CharsetInfo *catalogCharset,
					   Boolean useDocCatalog)
: systemCatalogs_(systemCatalogs),
  nSystemCatalogsMustExist_(nSystemCatalogsMustExist),
  sysidCharset_(sysidCharset),
  catalogCharset_(catalogCharset),
  useDocCatalog_(useDocCatalog)
{
}

Boolean SOCatalogManagerImpl::mapCatalog(ParsedSystemId &systemId,
					 ExtendEntityManager *em,
					 Messenger &mgr) const
{
  Vector<ParsedSystemId::Map> maps;
  systemId.maps.swap(maps);
  while (maps.size() > 0) {
    StringC catalogSystemId;
    systemId.unparse(*sysidCharset_, 0, catalogSystemId);
    SOEntityCatalog *catalog = new SOEntityCatalog(em);
    ConstPtr<EntityCatalog> deleter(catalog);
    CatalogParser parser(*catalogCharset_);
    parser.parseCatalog(catalogSystemId, 1, *sysidCharset_, *catalogCharset_,
			InputSourceOrigin::make(), catalog, mgr);
    // FIXME do catalog caching here
    StringC s;
    if (maps.back().type == ParsedSystemId::Map::catalogDocument) {
      if (!catalog->document(*sysidCharset_, mgr, s)) {
	mgr.message(CatalogMessages::noDocumentEntry,
		    StringMessageArg(catalogSystemId));
	return 0;
      }
    }
    else {
      ASSERT(maps.back().type == ParsedSystemId::Map::catalogPublic);
      if (!catalog->lookupPublic(maps.back().publicId, *sysidCharset_, mgr,
				 s)) {
	mgr.message(CatalogMessages::noPublicEntry,
		    StringMessageArg(maps.back().publicId),
		    StringMessageArg(catalogSystemId));
	return 0;
      }
    }
    ParsedSystemId tem;
    if (!em->parseSystemId(s, *sysidCharset_, 0, 0, mgr, tem))
      return 0;
    systemId = tem;
    maps.resize(maps.size() - 1);
    for (size_t i = 0; i < systemId.maps.size(); i++)
      maps.push_back(systemId.maps[i]);
    systemId.maps.clear();
  }
  return 1;
}

ConstPtr<EntityCatalog>
SOCatalogManagerImpl::makeCatalog(StringC &systemId,
				  const CharsetInfo &charset,
				  ExtendEntityManager *em,
				  Messenger &mgr) const
{
  SOEntityCatalog *entityCatalog = new SOEntityCatalog(em);
  CatalogParser parser(*catalogCharset_);
  size_t i;
  for (i = 0; i < nSystemCatalogsMustExist_; i++)
    parser.parseCatalog(systemCatalogs_[i], 1,
			*sysidCharset_, *catalogCharset_,
			InputSourceOrigin::make(), entityCatalog,
			mgr);
  if (useDocCatalog_)
    addCatalogsForDocument(parser, systemId, entityCatalog, charset, mgr);
  for (i = nSystemCatalogsMustExist_; i < systemCatalogs_.size(); i++)
    parser.parseCatalog(systemCatalogs_[i], 0,
			*sysidCharset_, *catalogCharset_,
			InputSourceOrigin::make(), entityCatalog,
			mgr);

  return entityCatalog;
}


void SOCatalogManagerImpl::addCatalogsForDocument(CatalogParser &parser,
						  StringC &sysid,
						  SOEntityCatalog *impl,
						  const CharsetInfo &charset,
						  Messenger &mgr) const
{
  ParsedSystemId v;
  if (!impl->entityManager()->parseSystemId(sysid, charset, 0, 0, mgr, v))
    return;
  if (v.maps.size() > 0) {
    if (v.maps[0].type == ParsedSystemId::Map::catalogDocument) {
      v.maps.erase(v.maps.begin(), v.maps.begin() + 1);
      StringC tem;
      v.unparse(charset, 0, tem);
      parser.parseCatalog(tem, 1, charset, *catalogCharset_,
			  InputSourceOrigin::make(), impl, mgr);
      if (!impl->document(charset, mgr, sysid)) {
	mgr.message(CatalogMessages::noDocumentEntry, StringMessageArg(tem));
	sysid.resize(0);
      }
    }
    return;
  }
  Vector<StringC> catalogs;
  size_t i;
  for (i = 0; i < v.size(); i++)
    if (v[i].storageManager->inheritable()) {
      ParsedSystemId catalogId;
      catalogId.resize(1);
      StorageObjectSpec &spec = catalogId.back();
      spec.storageManager = v[i].storageManager;
      spec.codingSystemType = v[i].codingSystemType;
      spec.codingSystemName = v[i].codingSystemName;
      spec.specId = spec.storageManager->idCharset()->execToDesc("catalog");
      spec.storageManager->resolveRelative(v[i].specId, spec.specId, 0);
      spec.baseId = v[i].baseId;
      spec.records = v[i].records;
      StringC tem;
      catalogId.unparse(charset, 0, tem);
      for (size_t j = 0; j < catalogs.size(); j++)
	if (tem == catalogs[j]) {
	  tem.resize(0);
	  break;
	}
      if (tem.size() > 0) {
	catalogs.resize(catalogs.size() + 1);
	tem.swap(catalogs.back());
      }
    }
  for (i = 0; i < catalogs.size(); i++)
    parser.parseCatalog(catalogs[i], 0, charset,
			*catalogCharset_, InputSourceOrigin::make(), impl,
			mgr);
}