Overload GetColumn with the Ability to Insert a Value

[decomplosion]

I have a file where we are using record types  

//Record 0 - EREAD,Size,Resource,Endpoint,Offset,Port,Value
//Record 1 - EWRITE,Size,Resource,Endpoint,Offset,Port,Value
//Record 2 - WAIT,Wait Type,Request Wait (ns),Actual Wait (ns),Timestamp (ns)
//Record 3 - VISET,Resource,Vrange,Irange,Mode,Connect,Connection Mode,Site,Value,Timestamp (ns)
//Record 4 - VISET_ADD,Resource,Connection Mode,Pin List Size,Pin List,Elapsed Time (ns)
//Record 5 - INIT_BUS_SEQUENCE,Bus Sequence Type,Timestamp (ns)
//Record 6 - EXT_BUS_STATE,Endpoint,Bus A Relay State,Bus B Relay State
//Record 7 - CHANNEL_STATE,Endpoint,Channel State
//Record 8 - CLOSURE_TAG,Close Type,Board ID,Elapsed Time (ns)
//Record 9 - GENERIC_LOG_BYTES,Size,Generic Data
//Record 10 - CEBEREAD64,Resource,Endpoint,Offset,Port,Value
//Record 11 - CEBEWRITE64,Resource,Endpoint,Offset,Port,Value
//Record 12 - FORCE,Resource,Force,Site,Hi Level,Lo Level,Timestamp (ns)
//Record 13 - DISCONNECT,Resource,Site,Timestamp (ns)
//Record 255 - FILE_HEADER,Version,MST Version,Record Type Count,Record Count,Processed Records,Wraparound Count,Timing Factor,Initial Timestamp (ns),Local Time
255,00.00.008,2025.256.1000.0,14,11442,11442,0,2600,11759956162279144,6/30/2025 8:28:05 PM
9,22,Start External Connect
2,WAIT,0,0,22776037187000

As the record types can have different column widths, the current implementation of the tool does not allow for me to use GetColumn as it will simply throw an error when the number of columns within a row do not match the column index indicated.

template<typename T>
   std::vector<T> GetColumn(const size_t pColumnIdx) const
   {
     const size_t dataColumnIdx = GetDataColumnIndex(pColumnIdx);
     std::vector<T> column;
     Converter<T> converter(mConverterParams);
     for (auto itRow = mData.begin(); itRow != mData.end(); ++itRow)
     {
       if (std::distance(mData.begin(), itRow) > mLabelParams.mColumnNameIdx)
       {
         if (dataColumnIdx < itRow->size())
         {
           T val;
           converter.ToVal(itRow->at(dataColumnIdx), val);
           column.push_back(val);
         }
         else
         {
           const std::string errStr = "requested column index " +
             std::to_string(pColumnIdx) + " >= " +
             std::to_string(itRow->size() - GetDataColumnIndex(0)) +
             " (number of columns on row index " +
             std::to_string(std::distance(mData.begin(), itRow) -
                            (mLabelParams.mColumnNameIdx + 1)) + ")";
           throw std::out_of_range(errStr);
         }
       }
     }
     return column;
   }

I was thinking it would be nice if I could choose a value to fill in the column for a row that does not have data.

template<typename T>
   std::vector<T> GetColumn(const size_t pColumnIdx, std::string defaultStr) const
   {
     const size_t dataColumnIdx = GetDataColumnIndex(pColumnIdx);
     std::vector<T> column;
     Converter<T> converter(mConverterParams);
     for (auto itRow = mData.begin(); itRow != mData.end(); ++itRow)
     {
       if (std::distance(mData.begin(), itRow) > mLabelParams.mColumnNameIdx)
       {
         if (dataColumnIdx < itRow->size())
         {
           T val;
           converter.ToVal(itRow->at(dataColumnIdx), val);
           column.push_back(val);
         }
         else
         {
           T val;
           converter.ToVal(defaultStr, val);
           column.push_back(val);
         }
       }
     }
     return column;
   }

I would think that I am not the only one who might want to use rapidcsv in this way.