parser.cpp

ACE编程的一本经典BIBLE的源代码,喜欢网络编程的别错过

                        this->nested_namespace_--;
                      }
                      }
                    return 0;
                  }
                default:              // a new nested element?
                  this->parse_element (0 ACEXML_ENV_ARG_PARAMETER);
                  ACEXML_CHECK_RETURN (-1);
                  break;
              }
            break;
          case '&':
            if (this->peek () == '#')
              {
                ACEXML_Char buf[7];
                size_t len = 0;
                do
                  {
                    len = sizeof (buf);
                    if (this->parse_char_reference (buf, len) != 0)
                      {
                        // [WFC: Legal Character]
                        this->fatal_error (ACE_TEXT ("Invalid CharRef")
                                           ACEXML_ENV_ARG_PARAMETER);
                        ACEXML_CHECK_RETURN (-1);
                      }
                  } while (buf[0] == '&' && this->peek() == '#');
                  for (size_t j = 0; j < len; ++j)
                    this->obstack_.grow (buf[j]);
                cdata_length += len;
              }
            else
              {
                this->ref_state_ = ACEXML_ParserInt::IN_CONTENT;
                int length = this->parse_entity_reference(ACEXML_ENV_SINGLE_ARG_PARAMETER);
                ACEXML_CHECK_RETURN (-1);
                if (length == 1)
                  cdata_length++;
              }
            break;
          case '\x20': case '\x0D': case '\x0A': case '\x09':
//             if (this->validate_)
//               {
//                 // Flush out any non-whitespace characters
//                 if (cdata_length != 0)
//                   {
//                     cdata = this->obstack_.freeze ();
//                     this->content_handler_->characters(cdata, 0, cdata_length
//                                                        ACEXML_ENV_ARG_PARAMETER);
//                     ACEXML_CHECK_RETURN (-1);
//                     this->obstack_.unwind (cdata);
//                     cdata_length = 0;
//                   }
//                 ++cdata_length;
//                 this->obstack_.grow (ch);
//                 while (1)
//                   {
//                     ch = this->peek();
//                     if (ch == '\x20' || ch == '\x0D' || ch == '\x0A' ||
//                         ch == '\x09')
//                       {
//                         ch = this->get();
//                         this->obstack_.grow (ch);
//                         continue;
//                       }
//                     break;
//                   }
//                 cdata = this->obstack_.freeze ();
//                 this->content_handler_->ignorableWhitespace (cdata, 0,
//                                                              cdata_length
//                                                              ACEXML_ENV_ARG_PARAMETER);
//                 ACEXML_CHECK_RETURN (-1);
//                 this->obstack_.unwind (cdata);
//                 cdata_length = 0;
//                 break;
//               }
            // Fall thru...
          default:
            ++cdata_length;
            this->obstack_.grow (ch);
        }
    }
  ACE_NOTREACHED (return 0;)
}


int
ACEXML_Parser::parse_cdata (ACEXML_ENV_SINGLE_ARG_DECL)
  ACE_THROW_SPEC ((ACEXML_SAXException))
{
  if (this->parse_token (ACE_TEXT ("[CDATA[")) < 0)
    {
      this->fatal_error(ACE_TEXT ("Expecting '[CDATA[' at beginning of CDATA ")
                        ACE_TEXT ("section")
                        ACEXML_ENV_ARG_PARAMETER);
      ACEXML_CHECK_RETURN (-1);
    }

  ACEXML_Char ch;
  int datalen = 0;
  ACEXML_Char *cdata = 0;
  while (1)
    {
      ch = this->get ();
      // Anything goes except the sequence "]]>".
      if (ch == ']' && this->peek() == ']')
        {
          ACEXML_Char temp = ch;
          ch = this->get();
          if (ch == ']' && this->peek() == '>')
            {
              ch = this->get();
              cdata = this->obstack_.freeze ();
              this->content_handler_->characters (cdata, 0, datalen
                                                  ACEXML_ENV_ARG_PARAMETER);
              ACEXML_CHECK_RETURN (-1);
              this->obstack_.unwind(cdata);
              return 0;
            }
          this->obstack_.grow (temp);
          ++datalen;
        }
      this->obstack_.grow (ch);
      ++datalen;
    };
  ACE_NOTREACHED (return -1);
}


int
ACEXML_Parser::parse_entity_decl (ACEXML_ENV_SINGLE_ARG_DECL)
  ACE_THROW_SPEC ((ACEXML_SAXException))
{
  ACEXML_Char nextch = 0;

  if ((this->parse_token (ACE_TEXT ("NTITY")) < 0) ||
      this->skip_whitespace_count (&nextch) == 0)
    {
      this->fatal_error (ACE_TEXT ("Expecting keyword ENTITY followed by a ")
                         ACE_TEXT ("space") ACEXML_ENV_ARG_PARAMETER);
      ACEXML_CHECK_RETURN (-1);
    }

  int is_GEDecl = 1;
  if (nextch == '%')            // This is a PEDecl.
    {
      is_GEDecl = 0;
      this->get ();             // consume the '%'
      if (this->skip_whitespace_count (&nextch) == 0)
        {
          this->fatal_error (ACE_TEXT ("Expecting space between % and ")
                             ACE_TEXT ("entity name")
                             ACEXML_ENV_ARG_PARAMETER);
          ACEXML_CHECK_RETURN (-1);
        }
    }

  ACEXML_Char *entity_name = this->parse_name ();
  if (entity_name == 0)
    {
      this->fatal_error (ACE_TEXT ("Invalid entity name")
                         ACEXML_ENV_ARG_PARAMETER);
      ACEXML_CHECK_RETURN (-1);
    }

  if (this->skip_whitespace_count (&nextch) == 0)
    {
      this->fatal_error (ACE_TEXT ("Expecting space between entity name and ")
                         ACE_TEXT ("entityDef")
                         ACEXML_ENV_ARG_PARAMETER);
      ACEXML_CHECK_RETURN (-1);
    }
  int retval = 0;
  if (nextch == '\'' || nextch == '"')
    {
      ACEXML_Char *entity_value = 0;
      if (this->parse_entity_value (entity_value
                                    ACEXML_ENV_ARG_PARAMETER) != 0)
        {
          this->fatal_error(ACE_TEXT ("Invalid EntityValue")
                            ACEXML_ENV_ARG_PARAMETER);
          ACEXML_CHECK_RETURN (-1);
        }
      if (is_GEDecl)
        retval = this->internal_GE_.add_entity (entity_name,
                                                entity_value);
      else
        retval = this->internal_PE_.add_entity (entity_name,
                                                entity_value);
      if (retval < 0)
        {
          this->fatal_error (ACE_TEXT ("Internal Parser Error in adding")
                             ACE_TEXT ("Entity to map")
                             ACEXML_ENV_ARG_PARAMETER);
          ACEXML_CHECK_RETURN (-1);
        }
      else if (retval == 1)
        {
          this->warning (ACE_TEXT ("Duplicate entity found")
                         ACEXML_ENV_ARG_PARAMETER);
          ACEXML_CHECK_RETURN (-1);
        }
    }
  else
    {
      ACEXML_Char *systemid, *publicid;

      this->parse_external_id (publicid, systemid
                               ACEXML_ENV_ARG_PARAMETER);
      ACEXML_CHECK_RETURN (-1);
      if (systemid == 0)
        {
          this->fatal_error(ACE_TEXT ("Invalid SystemLiteral")
                            ACEXML_ENV_ARG_PARAMETER);
          ACEXML_CHECK_RETURN (-1);
        }
      this->skip_whitespace_count (&nextch);
      if (nextch == 'N')        // NDATA section followed
        {
          if (is_GEDecl == 0)
            {
              this->fatal_error(ACE_TEXT ("Invalid NDataDecl in PEDef")
                                ACEXML_ENV_ARG_PARAMETER);
              ACEXML_CHECK_RETURN (-1);
            }

          if ((this->parse_token (ACE_TEXT ("NDATA")) < 0) ||
              this->skip_whitespace_count (&nextch) == 0)
            {
              this->fatal_error(ACE_TEXT ("Expecting keyword NDATA followed ")
                                ACE_TEXT ("by a space") ACEXML_ENV_ARG_PARAMETER);
              ACEXML_CHECK_RETURN (-1);
            }

          ACEXML_Char *ndata = this->parse_name ();
          if (this->validate_) // [VC: Notation Declared]
            {
              if (!this->notations_.resolve_entity (ndata))
                {
                  this->fatal_error (ACE_TEXT ("Undeclared Notation name")
                                     ACEXML_ENV_ARG_PARAMETER);
                  ACEXML_CHECK_RETURN (-1);
                }
              this->dtd_handler_->unparsedEntityDecl(entity_name, publicid,
                                                     systemid, ndata
                                                     ACEXML_ENV_ARG_PARAMETER);
              ACEXML_CHECK_RETURN (-1);
            }
        }
      else
        {
          if (is_GEDecl)
            retval = this->external_GE_.add_entity (entity_name,
                                                    systemid);
          else
            retval = this->external_PE_.add_entity (entity_name,
                                                    systemid);
          if (retval < 0)
            {
              this->fatal_error(ACE_TEXT ("Internal Parser Error")
                                ACEXML_ENV_ARG_PARAMETER);
              ACEXML_CHECK_RETURN (-1);
            }
          else if (retval == 1)
            this->warning(ACE_TEXT ("Duplicate external entity")
                          ACEXML_ENV_ARG_PARAMETER);
          if (is_GEDecl)
            retval = this->external_GE_.add_entity (entity_name,
                                                    publicid);
          else
            retval = this->external_PE_.add_entity (entity_name,
                                                    publicid);
          if (retval < 0)
            {
              this->fatal_error(ACE_TEXT ("Internal Parser Error")
                                ACEXML_ENV_ARG_PARAMETER);
              ACEXML_CHECK_RETURN (-1);
            }
          else if (retval == 1)
            this->warning (ACE_TEXT ("Duplicate entity definition")
                           ACEXML_ENV_ARG_PARAMETER);
        }
    }

  // End of ENTITY definition
  if (this->skip_whitespace() != '>')
    {
      this->fatal_error(ACE_TEXT ("Expecting '>' at end of entityDef")
                        ACEXML_ENV_ARG_PARAMETER);
      ACEXML_CHECK_RETURN (-1);
    }
  return 0;
}

int
ACEXML_Parser::parse_attlist_decl (ACEXML_ENV_SINGLE_ARG_DECL)
  ACE_THROW_SPEC ((ACEXML_SAXException))
{
  if (this->parse_token (ACE_TEXT ("ATTLIST")) < 0)
    {
      this->fatal_error(ACE_TEXT ("Expecting keyword 'ATTLIST'")
                        ACEXML_ENV_ARG_PARAMETER);
      ACEXML_CHECK_RETURN (-1);
    }
  int count = check_for_PE_reference (ACEXML_ENV_SINGLE_ARG_PARAMETER);
  ACEXML_CHECK_RETURN (-1);
  if (!count)
    {
      this->fatal_error(ACE_TEXT ("Expecting space between ATTLIST and ")
                        ACE_TEXT ("element name") ACEXML_ENV_ARG_PARAMETER);
      ACEXML_CHECK_RETURN (-1);
    }

  ACEXML_Char *element_name = this->parse_name ();
  if (element_name == 0)
    {
      this->fatal_error(ACE_TEXT ("Invalid element Name in attlistDecl")
                        ACEXML_ENV_ARG_PARAMETER);
      ACEXML_CHECK_RETURN (-1);
    }
  ACEXML_Char fwd = 0;
  count = this->skip_whitespace_count (&fwd);
  // Parse AttDef*
  while (fwd != '>')
    {
      if (!this->check_for_PE_reference (ACEXML_ENV_SINGLE_ARG_PARAMETER)
          && !count)
        this->fatal_error(ACE_TEXT ("Expecting space between element ")
                          ACE_TEXT ("name and AttDef")
                          ACEXML_ENV_ARG_PARAMETER);
      ACEXML_CHECK_RETURN (-1);
      this->skip_whitespace_count (&fwd);
      if (fwd == '>')
        break;

      count = this->check_for_PE_reference (ACEXML_ENV_SINGLE_ARG_PARAMETER);
      ACEXML_CHECK_RETURN (-1);

      this->parse_attname (ACEXML_ENV_SINGLE_ARG_PARAMETER);
      ACEXML_CHECK_RETURN (-1);

      count = this->check_for_PE_reference (ACEXML_ENV_SINGLE_ARG_PARAMETER);
      ACEXML_CHECK_RETURN (-1);
      if (!count)
        {
          this->fatal_error(ACE_TEXT ("Expecting space between AttName and ")
                            ACE_TEXT ("AttType") ACEXML_ENV_ARG_PARAMETER);
          ACEXML_CHECK_RETURN (-1);
        }
      this->parse_atttype (ACEXML_ENV_SINGLE_ARG_PARAMETER);
      ACEXML_CHECK_RETURN (-1);

      count = this->check_for_PE_reference (ACEXML_ENV_SINGLE_ARG_PARAMETER);
      ACEXML_CHECK_RETURN (-1);
      if (!count)
        {
          this->fatal_error(ACE_TEXT ("Expecting space between AttType and")
                            ACE_TEXT (" DefaultDecl")
                            ACEXML_ENV_ARG_PARAMETER);
          ACEXML_CHECK_RETURN (-1);
        }
      this->parse_defaultdecl (ACEXML_ENV_SINGLE_ARG_PARAMETER);