data_type.hpp

Go to the documentation of this file.

 
#pragma once
#include <cmath>
#include <cctype>
#include <string>
#include <cassert>
 
#include "common.hpp"
 
namespace csv {
    enum class DataType {
        UNKNOWN = -1,
        CSV_NULL,   
        CSV_STRING, 
        CSV_INT8,   
        CSV_INT16,  
        CSV_INT32,  
        CSV_INT64,  
        CSV_BIGINT, 
        CSV_DOUBLE  
    };
 
    static_assert(DataType::CSV_STRING < DataType::CSV_INT8, "String type should come before numeric types.");
    static_assert(DataType::CSV_INT8 < DataType::CSV_INT64, "Smaller integer types should come before larger integer types.");
    static_assert(DataType::CSV_INT64 < DataType::CSV_DOUBLE, "Integer types should come before floating point value types.");
 
    namespace internals {
        template<typename T>
        CSV_CONST CONSTEXPR_14
        long double pow10(const T& n) noexcept {
            static_assert(std::is_integral<T>::value, "pow10 only supports integral exponents");
            
            long double multiplicand = n > 0 ? 10 : 0.1,
                ret = 1;
 
            // Make all numbers positive
            T iterations = n > 0 ? n : -n;
            
            for (T i = 0; i < iterations; i++) {
                ret *= multiplicand;
            }
 
            return ret;
        }
 
        template<>
        CSV_CONST CONSTEXPR_14
        long double pow10(const unsigned& n) noexcept {
            long double multiplicand = n > 0 ? 10 : 0.1,
                ret = 1;
 
            for (unsigned i = 0; i < n; i++) {
                ret *= multiplicand;
            }
 
            return ret;
        }
 
#ifndef DOXYGEN_SHOULD_SKIP_THIS
        constexpr DataType int_type_arr[8] = {
            DataType::CSV_INT8,  // 1
            DataType::CSV_INT16, // 2
            DataType::UNKNOWN,
            DataType::CSV_INT32, // 4
            DataType::UNKNOWN,
            DataType::UNKNOWN,
            DataType::UNKNOWN,
            DataType::CSV_INT64  // 8
        };
 
        template<typename T>
        inline DataType type_num() {
            static_assert(std::is_integral<T>::value, "T should be an integral type.");
            static_assert(sizeof(T) <= 8, "Byte size must be no greater than 8.");
            return int_type_arr[sizeof(T) - 1];
        }
 
        template<> inline DataType type_num<float>() { return DataType::CSV_DOUBLE; }
        template<> inline DataType type_num<double>() { return DataType::CSV_DOUBLE; }
        template<> inline DataType type_num<long double>() { return DataType::CSV_DOUBLE; }
        template<> inline DataType type_num<std::nullptr_t>() { return DataType::CSV_NULL; }
        template<> inline DataType type_num<std::string>() { return DataType::CSV_STRING; }
 
        CONSTEXPR_14 DataType data_type(csv::string_view in, long double* const out = nullptr, 
            const char decimalsymbol = '.');
#endif
 
        template<size_t Bytes>
        CONSTEXPR_14 long double get_int_max() {
            static_assert(Bytes == 1 || Bytes == 2 || Bytes == 4 || Bytes == 8,
                "Bytes must be a power of 2 below 8.");
 
            CSV_MSVC_PUSH_DISABLE(4127)
            IF_CONSTEXPR (sizeof(signed char) == Bytes) {
                return (long double)std::numeric_limits<signed char>::max();
            }
            else IF_CONSTEXPR (sizeof(short) == Bytes) {
                return (long double)std::numeric_limits<short>::max();
            }
            else IF_CONSTEXPR (sizeof(int) == Bytes) {
                return (long double)std::numeric_limits<int>::max();
            }
            else IF_CONSTEXPR (sizeof(long int) == Bytes) {
                return (long double)std::numeric_limits<long int>::max();
            }
            else {
                return (long double)std::numeric_limits<long long int>::max();
            }
            CSV_MSVC_POP
        }
 
        template<size_t Bytes>
        CONSTEXPR_14 long double get_uint_max() {
            static_assert(Bytes == 1 || Bytes == 2 || Bytes == 4 || Bytes == 8,
                "Bytes must be a power of 2 below 8.");
 
            CSV_MSVC_PUSH_DISABLE(4127)
            IF_CONSTEXPR(sizeof(unsigned char) == Bytes) {
                return (long double)std::numeric_limits<unsigned char>::max();
            }
            else IF_CONSTEXPR(sizeof(unsigned short) == Bytes) {
                return (long double)std::numeric_limits<unsigned short>::max();
            }
            else IF_CONSTEXPR(sizeof(unsigned int) == Bytes) {
                return (long double)std::numeric_limits<unsigned int>::max();
            }
            else IF_CONSTEXPR(sizeof(unsigned long int) == Bytes) {
                return (long double)std::numeric_limits<unsigned long int>::max();
            }
            else {
                return (long double)std::numeric_limits<unsigned long long int>::max();
            }
            CSV_MSVC_POP
        }
 
        CONSTEXPR_VALUE_14 long double CSV_INT8_MAX = get_int_max<1>();
 
        CONSTEXPR_VALUE_14 long double CSV_INT16_MAX = get_int_max<2>();
 
        CONSTEXPR_VALUE_14 long double CSV_INT32_MAX = get_int_max<4>();
 
        CONSTEXPR_VALUE_14 long double CSV_INT64_MAX = get_int_max<8>();
 
        CONSTEXPR_VALUE_14 long double CSV_UINT8_MAX = get_uint_max<1>();
 
        CONSTEXPR_VALUE_14 long double CSV_UINT16_MAX = get_uint_max<2>();
 
        CONSTEXPR_VALUE_14 long double CSV_UINT32_MAX = get_uint_max<4>();
 
        CONSTEXPR_VALUE_14 long double CSV_UINT64_MAX = get_uint_max<8>();
 
        CSV_PRIVATE CONSTEXPR_14
        DataType _process_potential_exponential(
            csv::string_view exponential_part,
            const long double& coeff,
            long double * const out) {
            long double exponent = 0;
            auto result = data_type(exponential_part, &exponent);
 
            // Exponents in scientific notation should not be decimal numbers
            if (result >= DataType::CSV_INT8 && result < DataType::CSV_DOUBLE) {
                if (out) *out = coeff * pow10(static_cast<long long>(exponent));
                return DataType::CSV_DOUBLE;
            }
 
            return DataType::CSV_STRING;
        }
 
        CSV_PRIVATE CSV_PURE CONSTEXPR_14
        DataType _determine_integral_type(const long double& number) noexcept {
            // We can assume number is always non-negative
            assert(number >= 0);
 
            if (number <= internals::CSV_INT8_MAX)
                return DataType::CSV_INT8;
            else if (number <= internals::CSV_INT16_MAX)
                return DataType::CSV_INT16;
            else if (number <= internals::CSV_INT32_MAX)
                return DataType::CSV_INT32;
            else if (number <= internals::CSV_INT64_MAX)
                return DataType::CSV_INT64;
            else // Conversion to long long will cause an overflow
                return DataType::CSV_BIGINT;
        }
 
        CONSTEXPR_14
        DataType data_type(csv::string_view in, long double* const out, const char decimalSymbol) {
            // Empty string --> NULL
            if (in.size() == 0)
                return DataType::CSV_NULL;
 
            bool ws_allowed = true,
                dot_allowed = true,
                digit_allowed = true,
                is_negative = false,
                has_digit = false,
                prob_float = false;
 
            unsigned places_after_decimal = 0;
            long double integral_part = 0,
                decimal_part = 0;
 
            for (size_t i = 0, ilen = in.size(); i < ilen; i++) {
                const char& current = in[i];
 
                switch (current) {
                case ' ':
                    if (!ws_allowed) {
                        if (isdigit(in[i - 1])) {
                            digit_allowed = false;
                            ws_allowed = true;
                        }
                        else {
                            // Ex: '510 123 4567'
                            return DataType::CSV_STRING;
                        }
                    }
                    break;
                case '+':
                    if (!ws_allowed) {
                        return DataType::CSV_STRING;
                    }
 
                    break;
                case '-':
                    if (!ws_allowed) {
                        // Ex: '510-123-4567'
                        return DataType::CSV_STRING;
                    }
 
                    is_negative = true;
                    break;
                // case decimalSymbol: not allowed because decimalSymbol is not a literal,
                // it is handled in the default block
                case 'e':
                case 'E':
                    // Process scientific notation
                    if (prob_float || (i && i + 1 < ilen && isdigit(in[i - 1]))) {
                        size_t exponent_start_idx = i + 1;
                        prob_float = true;
 
                        // Strip out plus sign
                        if (in[i + 1] == '+') {
                            exponent_start_idx++;
                        }
 
                        return _process_potential_exponential(
                            in.substr(exponent_start_idx),
                            is_negative ? -(integral_part + decimal_part) : integral_part + decimal_part,
                            out
                        );
                    }
 
                    return DataType::CSV_STRING;
                    break;
                default:
                    short digit = static_cast<short>(current - '0');
                    if (digit >= 0 && digit <= 9) {
                        // Process digit
                        has_digit = true;
 
                        if (!digit_allowed)
                            return DataType::CSV_STRING;
                        else if (ws_allowed) // Ex: '510 456'
                            ws_allowed = false;
 
                        // Build current number
                        if (prob_float)
                            decimal_part += digit / pow10(++places_after_decimal);
                        else
                            integral_part = (integral_part * 10) + digit;
                    }
                    // case decimalSymbol: not allowed because decimalSymbol is not a literal. 
                    else if (dot_allowed && current == decimalSymbol) {
                        dot_allowed = false;
                        prob_float = true;
                    }
                    else {
                        return DataType::CSV_STRING;
                    }
                }
            }
 
            // No non-numeric/non-whitespace characters found
            if (has_digit) {
                long double number = integral_part + decimal_part;
                if (out) {
                    *out = is_negative ? -number : number;
                }
 
                return prob_float ? DataType::CSV_DOUBLE : _determine_integral_type(number);
            }
 
            // Just whitespace
            return DataType::CSV_NULL;
        }
    }
}