bool JsonParser::JsonStringParser::advance(Char ch)
{
switch (_state)
{
case state_0:
if (ch == '\\')
_state = state_esc;
else if (ch == '"')
return true;
else
_str += ch;
break;
case state_esc:
_state = state_0;
if (ch == '"' || ch == '\\' || ch == '/')
_str += ch;
else if (ch == 'b')
_str += '\b';
else if (ch == 'f')
_str += '\f';
else if (ch == 'n')
_str += '\n';
else if (ch == 'r')
_str += '\r';
else if (ch == 't')
_str += '\t';
else if (ch == 'u')
{
_value = 0;
_count = 4;
_state = state_hex;
}
else
_jsonParser->doThrow(std::string("invalid character '") + ch.narrow() + "' in string");
break;
case state_hex:
if (ch >= '0' && ch <= '9')
_value = (_value << 4) | (ch.value() - '0');
else if (ch >= 'a' && ch <= 'f')
_value = (_value << 4) | (ch.value() - 'a' + 10);
else if (ch >= 'A' && ch <= 'F')
_value = (_value << 4) | (ch.value() - 'A' + 10);
else
_jsonParser->doThrow(std::string("invalid character '") + ch.narrow() + "' in hex sequence");
if (--_count == 0)
{
_str += Char(static_cast<wchar_t>(_value));
_state = state_0;
}
break;
}
return false;
}
int JsonParser::advance(Char ch)
{
int ret;
if (ch == '\n')
++_lineNo;
try
{
switch (_state)
{
case state_0:
if (ch == '{')
{
_state = state_object;
_deserializer->setCategory(SerializationInfo::Object);
}
else if (ch == '[')
{
_state = state_array;
_deserializer->setCategory(SerializationInfo::Array);
}
else if (ch == '"')
{
_state = state_string;
_deserializer->setCategory(SerializationInfo::Value);
}
else if ((ch >= '0' && ch <= '9') || ch == '+' || ch == '-')
{
_token = ch;
_state = state_number;
_deserializer->setCategory(SerializationInfo::Value);
}
else if (ch == '/')
{
_nextState = _state;
_state = state_comment0;
}
else if (!std::isspace(ch.value()))
{
_token = ch;
_state = state_token;
}
break;
case state_object:
if (ch == '"')
{
_state = state_object_name;
_stringParser.clear();
}
else if (ch == '}')
return 1;
else if (ch == '/')
{
_nextState = _state;
_state = state_comment0;
}
else if (std::isalpha(ch.value()))
{
_token = ch;
_state = state_object_plainname;
}
else if (!std::isspace(ch.value()))
throwInvalidCharacter(ch);
break;
case state_object_plainname:
if (std::isalnum(ch.value()))
_token += ch;
else if (std::isspace(ch.value()))
{
_stringParser.str(_token);
_state = state_object_after_name;
}
else if (ch == ':')
{
_stringParser.str(_token);
if (_next == 0)
_next = new JsonParser();
log_debug("begin object member " << _stringParser.str());
_deserializer->beginMember(Utf8Codec::encode(_stringParser.str()),
std::string(), SerializationInfo::Void);
_next->begin(*_deserializer);
_stringParser.clear();
_state = state_object_value;
}
else
throwInvalidCharacter(ch);
break;
case state_object_name:
if (_stringParser.advance(ch))
_state = state_object_after_name;
break;
case state_object_after_name:
if (ch == ':')
{
if (_next == 0)
_next = new JsonParser();
log_debug("begin object member " << _stringParser.str());
_deserializer->beginMember(Utf8Codec::encode(_stringParser.str()),
std::string(), SerializationInfo::Void);
_next->begin(*_deserializer);
_stringParser.clear();
_state = state_object_value;
}
else if (ch == '/')
{
_nextState = _state;
_state = state_comment0;
}
else if (!std::isspace(ch.value()))
throwInvalidCharacter(ch);
break;
case state_object_value:
ret = _next->advance(ch);
if (ret != 0)
{
log_debug("leave member");
_deserializer->leaveMember();
_state = state_object_e;
}
if (ret != -1)
break;
case state_object_e:
if (ch == ',')
_state = state_object_next_member;
else if (ch == '}')
return 1;
else if (ch == '/')
{
_nextState = _state;
_state = state_comment0;
}
else if (!std::isspace(ch.value()))
throwInvalidCharacter(ch);
break;
case state_object_next_member:
if (ch == '"')
{
_state = state_object_name;
_stringParser.clear();
}
else if (ch == '/')
{
_nextState = _state;
_state = state_comment0;
}
else if (std::isalpha(ch.value()))
{
_token = ch;
_state = state_object_plainname;
}
else if (!std::isspace(ch.value()))
throwInvalidCharacter(ch);
break;
case state_array:
if (ch == ']')
{
return 1;
}
else if (ch == '/')
{
_nextState = _state;
_state = state_comment0;
}
else if (!std::isspace(ch.value()))
{
if (_next == 0)
_next = new JsonParser();
log_debug("begin array member");
_deserializer->beginMember(std::string(),
std::string(), SerializationInfo::Void);
_next->begin(*_deserializer);
_next->advance(ch);
_state = state_array_value;
}
break;
case state_array_value:
ret = _next->advance(ch);
if (ret != 0)
_state = state_array_e;
if (ret != -1)
break;
case state_array_e:
if (ch == ']')
{
log_debug("leave member");
_deserializer->leaveMember();
return 1;
}
else if (ch == ',')
{
log_debug("leave member");
_deserializer->leaveMember();
log_debug("begin array member");
_deserializer->beginMember(std::string(),
std::string(), SerializationInfo::Void);
_next->begin(*_deserializer);
_state = state_array_value;
}
else if (ch == '/')
{
_nextState = _state;
_state = state_comment0;
}
else if (!std::isspace(ch.value()))
throwInvalidCharacter(ch);
break;
case state_string:
if (_stringParser.advance(ch))
{
log_debug("set string value \"" << _stringParser.str() << '"');
_deserializer->setValue(_stringParser.str());
_deserializer->setTypeName("string");
_stringParser.clear();
_state = state_end;
return 1;
}
break;
case state_number:
if (std::isspace(ch.value()))
{
log_debug("set int value \"" << _token << '"');
_deserializer->setValue(_token);
_deserializer->setTypeName("int");
_token.clear();
return 1;
}
else if (ch == '.' || ch == 'e' || ch == 'E')
{
_token += ch;
_state = state_float;
}
else if (ch >= '0' && ch <= '9')
{
_token += ch;
}
else
{
log_debug("set int value \"" << _token << '"');
_deserializer->setValue(_token);
_deserializer->setTypeName("int");
_token.clear();
return -1;
}
break;
case state_float:
if (std::isspace(ch.value()))
{
log_debug("set double value \"" << _token << '"');
_deserializer->setValue(_token);
_deserializer->setTypeName("double");
_token.clear();
return 1;
}
else if ((ch >= '0' && ch <= '9') || ch == '+' || ch == '-'
|| ch == '.' || ch == 'e' || ch == 'E')
_token += ch;
else
{
log_debug("set double value \"" << _token << '"');
_deserializer->setValue(_token);
_deserializer->setTypeName("double");
_token.clear();
return -1;
}
break;
case state_token:
if (std::isalpha(ch.value()))
_token += Char(std::tolower(ch));
else
{
if (_token == "true" || _token == "false")
{
log_debug("set bool value \"" << _token << '"');
_deserializer->setValue(_token);
_deserializer->setTypeName("bool");
_token.clear();
}
else if (_token == "null")
{
log_debug("set null value \"" << _token << '"');
_deserializer->setTypeName("null");
_deserializer->setNull();
_token.clear();
}
return -1;
}
break;
case state_comment0:
if (ch == '/')
_state = state_commentline;
else if (ch == '*')
_state = state_comment;
else
throwInvalidCharacter(ch);
break;
case state_commentline:
if (ch == '\n')
_state = _nextState;
break;
case state_comment:
if (ch == '*')
_state = state_comment_e;
break;
case state_comment_e:
if (ch == '/')
_state = _nextState;
else if (ch != '*')
_state = state_comment;
break;
case state_end:
if (ch == '/')
{
_nextState = _state;
_state = state_comment0;
}
else if (!std::isspace(ch.value()))
doThrow(std::string("unexpected character '") + ch.narrow() + "\' after end");
break;
}
}
catch (JsonParserError& e)
{
e._lineNo = _lineNo;
throw;
}
return 0;
}
void JsonParser::throwInvalidCharacter(Char ch)
{
doThrow((std::string("invalid character '") + ch.narrow() + '\''));
}
void CsvParser::advance(Char ch)
{
if (ch == L'\n')
++_lineNo;
switch (_state)
{
case state_detectDelim:
if (isalnum(ch) || ch == L'_' || ch == L' ')
{
_titles.back() += ch.narrow();
}
else if (ch == L'\n' || ch == L'\r')
{
log_debug("title=\"" << _titles.back() << '"');
_noColumns = 1;
_state = (ch == L'\r' ? state_cr : state_rowstart);
}
else if (ch == L'\'' || ch == L'"')
{
_quote = ch;
_state = state_detectDelim_q;
}
else
{
_delimiter = ch;
log_debug("delimiter=" << _delimiter.narrow());
log_debug("title=\"" << _titles.back() << '"');
_titles.push_back(std::string());
_state = state_title;
}
break;
case state_detectDelim_q:
if (ch == _quote)
{
_state = state_detectDelim_postq;
}
else
{
_titles.back() += ch.narrow();
}
break;
case state_detectDelim_postq:
if (isalnum(ch) || ch == L'_' || ch == L'\'' || ch == L'"' || ch == L' ')
{
std::ostringstream msg;
msg << "invalid character '" << ch.narrow() << "' within csv title of column " << _titles.size();
SerializationError::doThrow(msg.str());
}
else if (ch == L'\n' || ch == L'\r')
{
log_debug("title=\"" << _titles.back() << '"');
_noColumns = 1;
_state = (ch == L'\r' ? state_cr : state_rowstart);
}
else
{
_delimiter = ch;
log_debug("delimiter=" << _delimiter.narrow());
log_debug("title=\"" << _titles.back() << '"');
_titles.push_back(std::string());
_state = state_title;
}
break;
case state_title:
if (ch == L'\n' || ch == L'\r')
{
log_debug("title=\"" << _titles.back() << '"');
_state = (ch == L'\r' ? state_cr : state_rowstart);
_noColumns = _titles.size();
}
else if (ch == _delimiter)
{
log_debug("title=\"" << _titles.back() << '"');
_titles.push_back(std::string());
}
else if (ch == L'\'' || ch == L'\"')
{
if (_titles.back().empty())
{
_quote = ch;
_state = state_qtitle;
}
else
{
std::ostringstream msg;
msg << "unexpected quote character within csv title of column " << _titles.size();
SerializationError::doThrow(msg.str());
}
}
else
{
_titles.back() += ch.narrow();
}
break;
case state_qtitle:
if (ch == _quote)
{
_state = state_qtitlep;
}
else
{
_titles.back() += ch.narrow();
}
break;
case state_qtitlep:
if (ch == L'\n' || ch == L'\r')
{
log_debug("title=\"" << _titles.back() << '"');
_state = (ch == L'\r' ? state_cr : state_rowstart);
_noColumns = _titles.size();
}
else if (ch == _delimiter)
{
log_debug("title=\"" << _titles.back() << '"');
_titles.push_back(std::string());
_state = state_title;
}
else
{
std::ostringstream msg;
msg << "invalid character '" << ch.narrow() << "' within csv title of column " << _titles.size();
SerializationError::doThrow(msg.str());
}
break;
case state_cr:
_state = state_rowstart;
if (ch == L'\n')
{
break;
}
// fallthrough
case state_rowstart:
_column = 0;
log_debug("new row");
_deserializer->beginMember(std::string(),
std::string(), SerializationInfo::Array);
_state = state_datastart;
// no break
case state_datastart:
log_debug("member \""
<< (_column < _titles.size() ? _titles[_column] : std::string()) << '"');
_deserializer->beginMember(
_column < _titles.size() ? _titles[_column] : std::string(),
std::string(), SerializationInfo::Value);
if (ch == L'\n' || ch == L'\r')
{
_deserializer->leaveMember();
checkNoColumns(_column, _noColumns, _lineNo);
_deserializer->leaveMember();
_state = (ch == L'\r' ? state_cr : state_rowstart);
}
else if (ch == L'"' || ch == L'\'')
{
_quote = ch;
_state = state_qdata;
}
else if (ch == _delimiter)
{
++_column;
_deserializer->leaveMember();
}
else
{
_value += ch;
_state = state_data;
}
break;
case state_data0:
if (ch == L'"' || ch == L'\'')
{
_quote = ch;
_state = state_qdata;
break;
}
case state_data:
if (ch == L'\n' || ch == L'\r')
{
log_debug("value \"" << _value << '"');
_deserializer->setValue(_value);
_value.clear();
checkNoColumns(_column, _noColumns, _lineNo);
_deserializer->leaveMember(); // leave data item
_deserializer->leaveMember(); // leave row
_state = (ch == L'\r' ? state_cr : state_rowstart);
}
else if (ch == _delimiter)
{
log_debug("value \"" << _value << '"');
_deserializer->setValue(_value);
_value.clear();
_deserializer->leaveMember(); // leave data item
++_column;
log_debug("member \""
<< (_column < _titles.size() ? _titles[_column] : std::string()) << '"');
_deserializer->beginMember(
_column < _titles.size() ? _titles[_column] : std::string(),
std::string(), SerializationInfo::Value);
_state = state_data0;
}
else
{
_value += ch;
}
break;
case state_qdata:
if (ch == _quote)
{
log_debug("value \"" << _value << '"');
_deserializer->setValue(_value);
_value.clear();
_deserializer->leaveMember(); // leave data item
_state = state_qdata_end;
}
else
{
_value += ch;
}
break;
case state_qdata_end:
if (ch == L'\n' || ch == L'\r')
{
checkNoColumns(_column, _noColumns, _lineNo);
_deserializer->leaveMember(); // leave row
_state = (ch == L'\r' ? state_cr : state_rowstart);
}
else if (ch == _delimiter)
{
++_column;
log_debug("member \""
<< (_column < _titles.size() ? _titles[_column] : std::string()) << '"');
_deserializer->beginMember(
_column < _titles.size() ? _titles[_column] : std::string(),
std::string(), SerializationInfo::Value);
_state = state_data0;
}
else
{
_value = _quote + _value + ch;
_state = state_data;
}
break;
}
//log_debug("ch=" << ch.narrow() << " _state=" << _state);
}