Method flagInconsistentData

src/csvtable.cpp:1026–1062

table_index_t CsvTable::flagInconsistentData(table_index_t column) {
	std::vector<enum CellContentType> types;
	std::map<enum CellContentType, table_index_t> type_distribution;
	table_index_t ROWS = getNumberRows();

	// Guess type of each cell
	for( table_index_t r = 0; r < ROWS; ++r ) {
		enum CellContentType my_type = guessContentType(getCell(r, column));
		types.push_back(my_type);
		auto it = type_distribution.find(my_type);
		if(  it != type_distribution.end() ) {
			++it->second;
		} else {
			type_distribution.emplace(my_type, 1);
		}
	}

	// Find most popular type
	enum CellContentType best_guess = CellContentType::CELL_ANY;
	table_index_t num_cells_with_best_guess = 0;
	for( auto it = type_distribution.begin(); it != type_distribution.end(); ++it ) {
		if( it->second > num_cells_with_best_guess ) {
			best_guess = it->first;
			num_cells_with_best_guess = it->second;
		}
	}

	// Flag rows that are different than the best guess
	table_index_t count_flagged_rows = 0;
	for( table_index_t r = 0; r < ROWS; ++r ) {
		if( types.at(r) != best_guess ) {
			flagRow(r, true);
			++count_flagged_rows;
		}
	}
	return count_flagged_rows;
}


enum CsvTable::CellContentType CsvTable::guessContentType(std::string content) {