csv_row.hpp

Go to the documentation of this file.

 
#pragma once
#include <chrono>
#include <cstddef>
#include <cstdint>
#include <cmath>
#include <iterator>
#include <memory> // For CSVField
#include <limits> // For CSVField
#include <unordered_set>
#include <string>
#include <sstream>
#include <vector>
 
#include "common.hpp"
#ifdef CSV_HAS_CXX17
#include <optional>
#endif
#ifdef CSV_HAS_CXX23
#if defined(__has_include)
#if __has_include(<expected>)
#include <expected>
#ifdef __cpp_lib_expected
#define CSV_HAS_STD_EXPECTED
#endif
#endif
#endif
#endif
#include "csv_exceptions.hpp"
#ifdef CSV_HAS_CXX20
#include <ranges>
#endif
#include "data_type.hpp"
#include "../external/classify_scalar.hpp"
#include "json_converter.hpp"
#include "raw_csv_data.hpp"
 
namespace csv {
    namespace internals {
        template<typename RowSink, typename ParsePolicy, typename FieldPolicy, typename RowPolicy>
        class CSVParserCore;
        struct CSVRowRowPolicy;
        namespace parser {
            class CSVParserDriverBase;
        }
        namespace speculative {
            struct CSVRowFragment;
        }
 
        namespace data_frame { struct RowViewAccessor; }
 
        static const std::string ERROR_NAN = "Not a number.";
        static const std::string ERROR_OVERFLOW = "Overflow error.";
        static const std::string ERROR_FLOAT_TO_INT =
            "Attempted to convert a floating point value to an integral type.";
        static const std::string ERROR_NEG_TO_UNSIGNED = "Negative numbers cannot be converted to unsigned types.";
    
        // Applied when CSVRow exposes a field view with configured whitespace trimming.
        csv::string_view get_trimmed(csv::string_view sv, const WhitespaceMap& ws_flags) noexcept;
    }
 
    class CSVField {
    public:
        constexpr explicit CSVField(csv::string_view _sv) noexcept
            : sv(_sv.data() ? _sv : csv::string_view("")) {}
 
        CSVField(csv::string_view _sv, const internals::CSVFieldScalar& scalar) noexcept
            : sv(_sv.data() ? _sv : csv::string_view("")),
              type_(scalar.type) {
            switch (scalar.type) {
            case DataType::CSV_INT8:
            case DataType::CSV_INT16:
            case DataType::CSV_INT32:
            case DataType::CSV_INT64:
                this->value_.integer = scalar.integer;
                break;
            case DataType::CSV_DOUBLE:
            case DataType::CSV_BIGINT:
                this->value_.floating = scalar.floating;
                break;
            case DataType::CSV_TIMESTAMP:
                this->value_.timestamp = scalar.timestamp;
                break;
            case DataType::CSV_BOOL:
                this->value_.boolean = scalar.boolean;
                break;
            default:
                break;
            }
        }
 
        operator csv::string_view() const noexcept {
            return this->sv;
        }
 
        operator std::string() const {
            return std::string(this->sv);
        }
 
#if CSV_EXCEPTIONS_ENABLED
        template<typename T = std::string> T get() const {
            T out{};
            CSVConversionError err;
            try_get(out, err);
            if (err != CSVConversionError::None) throw std::runtime_error(csv_conversion_error_message(err));
            return out;
        }
#endif
 
#ifdef CSV_HAS_STD_EXPECTED
        template<typename T = std::string>
        std::expected<T, CSVConversionError> as() const {
            T out{};
            CSVConversionError err;
            try_get(out, err);
            return (err != CSVConversionError::None)
                ? std::expected<T, CSVConversionError>(std::unexpected(err))
                : std::expected<T, CSVConversionError>(out);
        }
#endif
 
        template<typename T = std::string>
        bool try_get(T& out) const noexcept {
            CSVConversionError error;
            return try_get(out, error);
        }
 
        template<typename T>
        bool try_get(T& out, CSVConversionError& error) const noexcept {
            error = check_convert(out);
            return error == CSVConversionError::None;
        }
 
#ifdef CSV_HAS_CXX17
        template<typename T>
        operator std::optional<T>() const {
            T out{};
            return try_get(out) ? std::optional<T>(out) : std::nullopt;
        }
#endif
 
        template<typename T = long long>
        bool try_parse_hex(T& parsedValue) const {
            static_assert(std::is_integral<T>::value,
                "try_parse_hex only works with integral types (int, long, long long, etc.)");
 
            return classify_scalar::parse_hex(this->sv.data(), this->sv.data() + this->sv.size(), parsedValue);
        }
 
        bool try_parse_decimal(long double& dVal, const char decimalSymbol = '.');
 
        bool try_parse_timestamp(std::uint64_t& out) const noexcept;
 
#ifdef DOXYGEN_SHOULD_SKIP_THIS
        template<typename T>
        bool try_parse_timestamp(T& out) const noexcept;
#else
        template<typename T>
        internals::enable_if_t<
            std::is_integral<T>::value && std::is_unsigned<T>::value && !std::is_same<T, bool>::value
            && (sizeof(T) >= sizeof(std::uint64_t)),
            bool
        >
        try_parse_timestamp(T& out) const noexcept {
            std::uint64_t milliseconds = 0;
            if (!this->try_parse_timestamp(milliseconds))
                return false;
 
            out = static_cast<T>(milliseconds);
            return true;
        }
#endif
 
        template<typename Rep, typename Period>
        bool try_parse_timestamp(std::chrono::duration<Rep, Period>& out) const noexcept {
            std::uint64_t milliseconds = 0;
            if (!this->try_parse_timestamp(milliseconds))
                return false;
 
            out = std::chrono::duration_cast<std::chrono::duration<Rep, Period>>(
                std::chrono::milliseconds(milliseconds));
            return true;
        }
 
        template<typename Duration>
        bool try_parse_timestamp(std::chrono::time_point<std::chrono::system_clock, Duration>& out) const noexcept {
            std::uint64_t milliseconds = 0;
            if (!this->try_parse_timestamp(milliseconds))
                return false;
 
            out = std::chrono::time_point<std::chrono::system_clock, Duration>(
                std::chrono::duration_cast<Duration>(std::chrono::milliseconds(milliseconds)));
            return true;
        }
 
        template<typename T>
        inline bool operator==(T other) const noexcept
        {
            static_assert(std::is_arithmetic<T>::value,
                "T should be a numeric value.");
 
            this->get_value();
            if (this->type_ < DataType::CSV_INT8 || this->type_ > DataType::CSV_DOUBLE || this->type_ == DataType::CSV_BIGINT) {
                return false;
            }
 
            return internals::is_equal(this->numeric_value_as_long_double(), static_cast<long double>(other), 0.000001L);
        }
 
        CONSTEXPR csv::string_view get_sv() const noexcept { return this->sv; }
 
        inline bool is_null() const noexcept { return type() == DataType::CSV_NULL; }
 
        inline bool is_str() const noexcept { return type() == DataType::CSV_STRING; }
 
        inline bool is_num() const noexcept {
            return type() >= DataType::CSV_INT8 && type() <= DataType::CSV_DOUBLE;
        }
 
        inline bool is_int() const noexcept {
            return (type() >= DataType::CSV_INT8) && (type() <= DataType::CSV_INT64);
        }
 
        inline bool is_float() const noexcept { return type() == DataType::CSV_DOUBLE; }
 
        inline bool is_bool() const noexcept { return type() == DataType::CSV_BOOL; }
 
        inline bool is_timestamp() const noexcept { return type() == DataType::CSV_TIMESTAMP; }
 
        inline DataType type() const noexcept {
            this->get_value();
            return type_;
        }
 
    private:
        // GCC emits a psABI note for by-value APIs involving unions with long double.
        // This is a workaround to keep the logs clean without sacrificing the benefits of a union on other compilers.
        // Give only GCC users the struct tax so normal builds and strict CI logs stay quiet.
#if defined(__GNUC__) && !defined(__clang__)
        struct FieldValue {
            constexpr FieldValue() noexcept
                : integer(0), floating(0), timestamp(0), boolean(false) {}
 
            std::int64_t integer;
            long double floating;
            std::uint64_t timestamp;
            bool boolean;
        };
#else
        union FieldValue {
            constexpr FieldValue() noexcept : floating(0) {}
 
            std::int64_t integer;
            long double floating;
            std::uint64_t timestamp;
            bool boolean;
        };
#endif
 
        struct FieldValueOutput {
            FieldValue& value;
 
            template<classify_scalar::ScalarKind Kind>
            typename std::enable_if<
                Kind == classify_scalar::scalar_int8
                || Kind == classify_scalar::scalar_int16
                || Kind == classify_scalar::scalar_int32
                || Kind == classify_scalar::scalar_int64,
                void>::type set(std::int64_t parsed) const noexcept {
                value.integer = parsed;
            }
 
            template<classify_scalar::ScalarKind Kind>
            typename std::enable_if<Kind == classify_scalar::scalar_float, void>::type set(long double parsed) const noexcept {
                value.floating = parsed;
            }
 
            template<classify_scalar::ScalarKind Kind>
            typename std::enable_if<Kind == classify_scalar::scalar_bool, void>::type set(bool parsed) const noexcept {
                value.boolean = parsed;
            }
 
            template<classify_scalar::ScalarKind Kind>
            typename std::enable_if<Kind == classify_scalar::scalar_timestamp, void>::type set(std::uint64_t parsed) const noexcept {
                value.timestamp = parsed;
            }
        };
 
        mutable FieldValue value_; 
        csv::string_view sv = ""; 
        mutable DataType type_ = DataType::UNKNOWN; 
        bool stores_integral() const noexcept {
            return type_ >= DataType::CSV_INT8 && type_ <= DataType::CSV_INT64;
        }
 
        long double numeric_value_as_long_double() const noexcept {
            return stores_integral()
                ? static_cast<long double>(value_.integer)
                : value_.floating;
        }
 
        void cache_parsed_value(DataType parsed_type, long double parsed_value) noexcept {
            type_ = parsed_type;
 
            if (parsed_type >= DataType::CSV_INT8 && parsed_type <= DataType::CSV_INT64) {
                value_.integer = static_cast<std::int64_t>(parsed_value);
            }
            else if (parsed_type == DataType::CSV_DOUBLE || parsed_type == DataType::CSV_BIGINT) {
                value_.floating = parsed_value;
            }
        }
 
        CSVConversionError check_convert(std::string& out) const noexcept {
            out.assign(this->sv.data(), this->sv.size());
            return CSVConversionError::None;
        }
 
        CSVConversionError check_convert(csv::string_view& out) const noexcept {
            out = this->sv;
            return CSVConversionError::None;
        }
 
        CSVConversionError check_convert(long double& out) const noexcept {
            if (!is_num()) return CSVConversionError::NotANumber;
            out = this->numeric_value_as_long_double();
            return CSVConversionError::None;
        }
 
        CSVConversionError check_convert(bool& out) const noexcept {
            if (this->type() != DataType::CSV_BOOL)
                return CSVConversionError::NotANumber;
 
            out = this->value_.boolean;
            return CSVConversionError::None;
        }
 
        template<typename Rep, typename Period>
        CSVConversionError check_convert(std::chrono::duration<Rep, Period>& out) const noexcept {
            if (this->type() != DataType::CSV_TIMESTAMP)
                return CSVConversionError::NotANumber;
 
            out = std::chrono::duration_cast<std::chrono::duration<Rep, Period>>(
                std::chrono::milliseconds(this->value_.timestamp));
            return CSVConversionError::None;
        }
 
        template<typename Duration>
        CSVConversionError check_convert(std::chrono::time_point<std::chrono::system_clock, Duration>& out) const noexcept {
            if (this->type() != DataType::CSV_TIMESTAMP)
                return CSVConversionError::NotANumber;
 
            out = std::chrono::time_point<std::chrono::system_clock, Duration>(
                std::chrono::duration_cast<Duration>(std::chrono::milliseconds(this->value_.timestamp)));
            return CSVConversionError::None;
        }
 
        template<typename T>
        CSVConversionError check_convert(T& out) const noexcept {
            IF_CONSTEXPR(std::is_arithmetic<T>::value) {
                if (!this->is_num())
                    return CSVConversionError::NotANumber;
                if (this->type_ == DataType::CSV_BIGINT)
                    return CSVConversionError::Overflow;
            }
 
            IF_CONSTEXPR(std::is_integral<T>::value) {
                if (this->is_float())
                    return CSVConversionError::FloatToInt;
 
                IF_CONSTEXPR(std::is_unsigned<T>::value) {
                    if (this->numeric_value_as_long_double() < 0)
                        return CSVConversionError::NegativeToUnsigned;
                }
            }
 
            IF_CONSTEXPR(!std::is_floating_point<T>::value) {
                const long double value = this->numeric_value_as_long_double();
                if (value < static_cast<long double>(std::numeric_limits<T>::min())
                    || value > static_cast<long double>(std::numeric_limits<T>::max())) {
                    return CSVConversionError::Overflow;
                }
            }
 
            out = this->stores_integral()
                ? static_cast<T>(this->value_.integer)
                : static_cast<T>(this->value_.floating);
            return CSVConversionError::None;
        }
 
        inline void get_value() const noexcept {
            if ((int)type_ < 0) {
                if (this->sv.empty()) {
                    type_ = DataType::CSV_NULL;
                    return;
                }
 
                const char* first = this->sv.data();
                const char* last = first + this->sv.size();
                typedef classify_scalar::policy_pack<
                    classify_scalar::builtin_numeric_policy<'.', false>,
                    classify_scalar::builtin_timestamp_policy,
                    classify_scalar::builtin_bool_policy
                > csv_field_policy_pack;
 
                type_ = classify_scalar::classify_scalar<
                    DataType,
                    true>(first, last, FieldValueOutput{ this->value_ }, csv_field_policy_pack());
            }
        }
    };
 
    class CSVRow {
    public:
        template<typename RowSink, typename ParsePolicy, typename FieldPolicy, typename RowPolicy>
        friend class internals::CSVParserCore;
        friend struct internals::CSVRowRowPolicy;
        friend struct internals::data_frame::RowViewAccessor;
        friend internals::parser::CSVParserDriverBase;
        friend struct internals::speculative::CSVRowFragment;
 
        CSVRow() = default;
        
        CSVRow(internals::RawCSVDataPtr _data) : data(_data) {}
        CSVRow(internals::RawCSVDataPtr _data, size_t _data_start, size_t _field_bounds)
            : data(_data), data_start(_data_start), fields_start(_field_bounds) {}
        CSVRow(internals::RawCSVDataPtr _data, size_t _data_start, size_t _field_bounds, size_t _row_length)
            : data(_data), data_start(_data_start), fields_start(_field_bounds), row_length(_row_length) {}
 
        CONSTEXPR bool empty() const noexcept { return this->size() == 0; }
 
        CONSTEXPR size_t size() const noexcept { return row_length; }
 
        size_t byte_offset() const noexcept {
            return this->data ? this->data->source_start + this->data_start : 0;
        }
 
 
        CSVField try_field(size_t index, CSVError& error) const;
        CSVField try_field(csv::string_view name, CSVError& error) const;
 
#if CSV_EXCEPTIONS_ENABLED
        CSVField operator[](size_t n) const;
        CSVField operator[](csv::string_view col_name) const;
#endif
#if !CSV_EXCEPTIONS_ENABLED
        inline std::string to_json() const {
            const auto* converter = this->get_json_converter();
            return converter ? converter->row_to_json(this->size(),
                [this](size_t i) { return this->get_field(i); }) : "{}";
        }
        inline std::string to_json_array() const {
            const auto* converter = this->get_json_converter();
            return converter ? converter->row_to_json_array(this->size(),
                [this](size_t i) { return this->get_field(i); }) : "[]";
        }
#endif
        inline std::string try_to_json(const std::vector<std::string>& subset, CSVError& error) const {
            error.clear();
            const auto* converter = this->get_json_converter();
            return converter ? converter->try_row_to_json(this->size(),
                [this](size_t i) { return this->get_field(i); }, subset, error) : "{}";
        }
        inline std::string try_to_json_array(const std::vector<std::string>& subset, CSVError& error) const {
            error.clear();
            const auto* converter = this->get_json_converter();
            return converter ? converter->try_row_to_json_array(this->size(),
                [this](size_t i) { return this->get_field(i); }, subset, error) : "[]";
        }
#if CSV_EXCEPTIONS_ENABLED
        inline std::string to_json(const std::vector<std::string>& subset = {}) const {
            CSVError error;
            auto out = try_to_json(subset, error);
            internals::throw_error(error);
            return out;
        }
        inline std::string to_json_array(const std::vector<std::string>& subset = {}) const {
            CSVError error;
            auto out = try_to_json_array(subset, error);
            internals::throw_error(error);
            return out;
        }
#endif
 
        const std::vector<std::string>& get_col_names() const {
            return this->data->col_names->get_col_names();
        }
 
        internals::ConstColNamesPtr col_names_ptr() const noexcept {
            return this->data->col_names;
        }
 
#if CSV_EXCEPTIONS_ENABLED
        std::unordered_map<std::string, std::string> to_unordered_map() const;
#endif
 
#if CSV_EXCEPTIONS_ENABLED
        std::unordered_map<std::string, std::string> to_unordered_map(
            const std::vector<std::string>& subset
        ) const;
#endif
 
        #ifdef CSV_HAS_CXX20
        auto to_sv_range() const {
            return std::views::iota(size_t{0}, this->size())
                | std::views::transform([this](size_t i) { return this->get_field(i); });
        }
        #endif
 
        operator std::vector<std::string>() const;
 
        csv::string_view raw_str() const noexcept;
 
        class iterator {
        public:
            class pointer {
            public:
                explicit pointer(const CSVField& field) noexcept;
                CSVField* operator->() noexcept;
                const CSVField* operator->() const noexcept;
 
            private:
                CSVField field_;
            };
 
#ifndef DOXYGEN_SHOULD_SKIP_THIS
            using value_type = CSVField;
            using difference_type = std::ptrdiff_t;
            using reference = CSVField;
            using iterator_category = std::input_iterator_tag;
#ifdef CSV_HAS_CXX20
            using iterator_concept = std::random_access_iterator_tag;
#endif
#endif
            iterator() = default;
            iterator(const CSVRow* row, difference_type index) noexcept;
 
            reference operator*() const;
            pointer operator->() const;
            reference operator[](difference_type n) const;
 
            iterator operator++(int) noexcept;
            iterator& operator++() noexcept;
            iterator operator--(int) noexcept;
            iterator& operator--() noexcept;
            iterator operator+(difference_type n) const noexcept;
            iterator operator-(difference_type n) const noexcept;
            iterator& operator+=(difference_type n) noexcept;
            iterator& operator-=(difference_type n) noexcept;
            difference_type operator-(const iterator& other) const noexcept;
            friend iterator operator+(difference_type n, const iterator& it) noexcept;
 
            bool operator==(const iterator& other) const noexcept;
            bool operator!=(const iterator& other) const noexcept;
            bool operator<(const iterator& other) const noexcept;
            bool operator>(const iterator& other) const noexcept;
            bool operator<=(const iterator& other) const noexcept;
            bool operator>=(const iterator& other) const noexcept;
 
        private:
            internals::RawCSVDataPtr data_;
            size_t data_start_ = 0;
            size_t fields_start_ = 0;
            size_t row_length_ = 0;
            difference_type index_ = 0;
        };
 
        class reverse_iterator {
        public:
#ifndef DOXYGEN_SHOULD_SKIP_THIS
            using value_type = iterator::value_type;
            using difference_type = iterator::difference_type;
            using pointer = iterator::pointer;
            using reference = iterator::reference;
            using iterator_category = std::input_iterator_tag;
#ifdef CSV_HAS_CXX20
            using iterator_concept = std::random_access_iterator_tag;
#endif
#endif
            reverse_iterator() = default;
            explicit reverse_iterator(iterator base) noexcept;
 
            reference operator*() const;
            pointer operator->() const;
            reference operator[](difference_type n) const;
 
            reverse_iterator operator++(int) noexcept;
            reverse_iterator& operator++() noexcept;
            reverse_iterator operator--(int) noexcept;
            reverse_iterator& operator--() noexcept;
            reverse_iterator operator+(difference_type n) const noexcept;
            reverse_iterator operator-(difference_type n) const noexcept;
            reverse_iterator& operator+=(difference_type n) noexcept;
            reverse_iterator& operator-=(difference_type n) noexcept;
            difference_type operator-(const reverse_iterator& other) const noexcept;
            friend reverse_iterator operator+(difference_type n, const reverse_iterator& it) noexcept;
 
            bool operator==(const reverse_iterator& other) const noexcept;
            bool operator!=(const reverse_iterator& other) const noexcept;
            bool operator<(const reverse_iterator& other) const noexcept;
            bool operator>(const reverse_iterator& other) const noexcept;
            bool operator<=(const reverse_iterator& other) const noexcept;
            bool operator>=(const reverse_iterator& other) const noexcept;
 
            iterator base() const noexcept;
 
        private:
            iterator base_;
        };
 
        std::unordered_map<std::string, std::string> try_to_unordered_map(CSVError& error) const;
        std::unordered_map<std::string, std::string> try_to_unordered_map(
            CSVError& error, const std::vector<std::string>& subset) const;
 
        iterator begin() const noexcept;
        iterator end() const noexcept;
        reverse_iterator rbegin() const noexcept;
        reverse_iterator rend() const noexcept;
 
    private:
        std::unordered_map<std::string, std::string> to_unordered_map_impl(
            const std::vector<std::string>* subset, CSVError& error) const;
 
        inline csv::string_view get_field_impl(size_t index, const internals::RawCSVDataPtr& _data) const {
#if CSV_EXCEPTIONS_ENABLED
            if (index >= this->size())
                throw std::runtime_error(internals::CSV_ERROR_INDEX_OUT_OF_BOUNDS);
#else
            CSV_DEBUG_ASSERT(index < this->size());
#endif
 
            const size_t field_index = this->fields_start + index;
            const auto field = _data->fields[field_index];
            csv::string_view field_str;
            if (field.has_realized_storage()) {
                field_str = _data->quote_arena.view(field.start, field.length);
            }
            else {
                field_str = csv::string_view(_data->data).substr(this->data_start + field.start, field.length);
            }
 
            if (_data->has_ws_trimming) {
                field_str = internals::get_trimmed(field_str, _data->ws_flags);
            }
 
            return field_str;
        }
 
        CSVField make_field(size_t index, const internals::RawCSVDataPtr& _data) const;
 
        csv::string_view get_field(size_t index) const;
 
        csv::string_view get_field_safe(size_t index, internals::RawCSVDataPtr _data) const;
 
        const internals::JsonConverter* get_json_converter() const {
            if (this->data.get() == nullptr) {
                return nullptr;
            }
 
            return &this->data->json_converter.get_or_create([this]() {
                const std::vector<std::string> columns = this->data->col_names
                    ? this->data->col_names->get_col_names()
                    : std::vector<std::string>();
                return std::make_shared<internals::JsonConverter>(columns);
            });
        }
 
        internals::RawCSVDataPtr data;
 
        size_t data_start = 0;
 
        size_t fields_start = 0;
 
        size_t row_length = 0;
 
        size_t data_end = (std::numeric_limits<size_t>::max)();
    };
 
#ifdef _MSC_VER
#pragma region CSVField::get Specializations
#endif
#if CSV_EXCEPTIONS_ENABLED
    template<>
    inline std::string CSVField::get<std::string>() const {
        return std::string(this->sv);
    }
 
    template<>
    CONSTEXPR_14 csv::string_view CSVField::get<csv::string_view>() const {
        return this->sv;
    }
 
    template<>
    inline long double CSVField::get<long double>() const {
        if (!is_num())
            throw std::runtime_error(internals::ERROR_NAN);
 
        return this->numeric_value_as_long_double();
    }
 
#endif
 
    // Preserve the existing constexpr view-retrieval API.
    template<>
    CONSTEXPR_14 bool CSVField::try_get<csv::string_view>(csv::string_view& out) const noexcept {
        out = this->sv;
        return true;
    }
#ifdef _MSC_VER
#pragma endregion CSVField::get Specializations
#endif
 
    template<>
    CONSTEXPR bool CSVField::operator==(const char * other) const noexcept
    {
        return this->sv == other;
    }
 
    template<>
    CONSTEXPR bool CSVField::operator==(csv::string_view other) const noexcept
    {
        return this->sv == other;
    }
}