/src/boost/boost/json/basic_parser_impl.hpp

Source
//
// Copyright (c) 2019 Vinnie Falco (vinnie.falco@gmail.com)
// Copyright (c) 2020 Krystian Stasiowski (sdkrystian@gmail.com)
//
// Distributed under the Boost Software License, Version 1.0. (See accompanying
// file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
//
// Official repository: https://github.com/boostorg/json
//

#ifndef BOOST_JSON_BASIC_PARSER_IMPL_HPP
#define BOOST_JSON_BASIC_PARSER_IMPL_HPP

#include <boost/json/detail/config.hpp>
#include <boost/json/detail/literals.hpp>
#include <boost/json/basic_parser.hpp>
#include <boost/json/error.hpp>
#include <boost/json/detail/buffer.hpp>
#include <boost/json/detail/charconv/from_chars.hpp>
#include <boost/json/detail/sse2.hpp>
#include <boost/mp11/algorithm.hpp>
#include <boost/mp11/integral.hpp>
#include <cmath>
#include <limits>
#include <cstring>

#ifdef _MSC_VER
#pragma warning(push)
#pragma warning(disable: 4702) // unreachable code
#pragma warning(disable: 4127) // conditional expression is constant
#endif

/*  This file must be manually included to get the
    function template definitions for basic_parser.
*/

/*  Reference:

    https://www.json.org/

    RFC 7159: The JavaScript Object Notation (JSON) Data Interchange Format
    https://tools.ietf.org/html/rfc7159

    https://ampl.com/netlib/fp/dtoa.c
*/

#ifndef BOOST_JSON_DOCS

namespace boost {
namespace json {
namespace detail {

inline
double
pow10(int exp) noexcept
{
    static double const tab[618] = {
                        1e-308, 1e-307, 1e-306, 1e-305, 1e-304, 1e-303, 1e-302, 1e-301,

        1e-300, 1e-299, 1e-298, 1e-297, 1e-296, 1e-295, 1e-294, 1e-293, 1e-292, 1e-291,
        1e-290, 1e-289, 1e-288, 1e-287, 1e-286, 1e-285, 1e-284, 1e-283, 1e-282, 1e-281,
        1e-280, 1e-279, 1e-278, 1e-277, 1e-276, 1e-275, 1e-274, 1e-273, 1e-272, 1e-271,
        1e-270, 1e-269, 1e-268, 1e-267, 1e-266, 1e-265, 1e-264, 1e-263, 1e-262, 1e-261,
        1e-260, 1e-259, 1e-258, 1e-257, 1e-256, 1e-255, 1e-254, 1e-253, 1e-252, 1e-251,
        1e-250, 1e-249, 1e-248, 1e-247, 1e-246, 1e-245, 1e-244, 1e-243, 1e-242, 1e-241,
        1e-240, 1e-239, 1e-238, 1e-237, 1e-236, 1e-235, 1e-234, 1e-233, 1e-232, 1e-231,
        1e-230, 1e-229, 1e-228, 1e-227, 1e-226, 1e-225, 1e-224, 1e-223, 1e-222, 1e-221,
        1e-220, 1e-219, 1e-218, 1e-217, 1e-216, 1e-215, 1e-214, 1e-213, 1e-212, 1e-211,
        1e-210, 1e-209, 1e-208, 1e-207, 1e-206, 1e-205, 1e-204, 1e-203, 1e-202, 1e-201,

        1e-200, 1e-199, 1e-198, 1e-197, 1e-196, 1e-195, 1e-194, 1e-193, 1e-192, 1e-191,
        1e-190, 1e-189, 1e-188, 1e-187, 1e-186, 1e-185, 1e-184, 1e-183, 1e-182, 1e-181,
        1e-180, 1e-179, 1e-178, 1e-177, 1e-176, 1e-175, 1e-174, 1e-173, 1e-172, 1e-171,
        1e-170, 1e-169, 1e-168, 1e-167, 1e-166, 1e-165, 1e-164, 1e-163, 1e-162, 1e-161,
        1e-160, 1e-159, 1e-158, 1e-157, 1e-156, 1e-155, 1e-154, 1e-153, 1e-152, 1e-151,
        1e-150, 1e-149, 1e-148, 1e-147, 1e-146, 1e-145, 1e-144, 1e-143, 1e-142, 1e-141,
        1e-140, 1e-139, 1e-138, 1e-137, 1e-136, 1e-135, 1e-134, 1e-133, 1e-132, 1e-131,
        1e-130, 1e-129, 1e-128, 1e-127, 1e-126, 1e-125, 1e-124, 1e-123, 1e-122, 1e-121,
        1e-120, 1e-119, 1e-118, 1e-117, 1e-116, 1e-115, 1e-114, 1e-113, 1e-112, 1e-111,
        1e-110, 1e-109, 1e-108, 1e-107, 1e-106, 1e-105, 1e-104, 1e-103, 1e-102, 1e-101,

        1e-100, 1e-099, 1e-098, 1e-097, 1e-096, 1e-095, 1e-094, 1e-093, 1e-092, 1e-091,
        1e-090, 1e-089, 1e-088, 1e-087, 1e-086, 1e-085, 1e-084, 1e-083, 1e-082, 1e-081,
        1e-080, 1e-079, 1e-078, 1e-077, 1e-076, 1e-075, 1e-074, 1e-073, 1e-072, 1e-071,
        1e-070, 1e-069, 1e-068, 1e-067, 1e-066, 1e-065, 1e-064, 1e-063, 1e-062, 1e-061,
        1e-060, 1e-059, 1e-058, 1e-057, 1e-056, 1e-055, 1e-054, 1e-053, 1e-052, 1e-051,
        1e-050, 1e-049, 1e-048, 1e-047, 1e-046, 1e-045, 1e-044, 1e-043, 1e-042, 1e-041,
        1e-040, 1e-039, 1e-038, 1e-037, 1e-036, 1e-035, 1e-034, 1e-033, 1e-032, 1e-031,
        1e-030, 1e-029, 1e-028, 1e-027, 1e-026, 1e-025, 1e-024, 1e-023, 1e-022, 1e-021,
        1e-020, 1e-019, 1e-018, 1e-017, 1e-016, 1e-015, 1e-014, 1e-013, 1e-012, 1e-011,
        1e-010, 1e-009, 1e-008, 1e-007, 1e-006, 1e-005, 1e-004, 1e-003, 1e-002, 1e-001,

        1e+000, 1e+001, 1e+002, 1e+003, 1e+004, 1e+005, 1e+006, 1e+007, 1e+008, 1e+009,
        1e+010, 1e+011, 1e+012, 1e+013, 1e+014, 1e+015, 1e+016, 1e+017, 1e+018, 1e+019,
        1e+020, 1e+021, 1e+022, 1e+023, 1e+024, 1e+025, 1e+026, 1e+027, 1e+028, 1e+029,
        1e+030, 1e+031, 1e+032, 1e+033, 1e+034, 1e+035, 1e+036, 1e+037, 1e+038, 1e+039,
        1e+040, 1e+041, 1e+042, 1e+043, 1e+044, 1e+045, 1e+046, 1e+047, 1e+048, 1e+049,
        1e+050, 1e+051, 1e+052, 1e+053, 1e+054, 1e+055, 1e+056, 1e+057, 1e+058, 1e+059,
        1e+060, 1e+061, 1e+062, 1e+063, 1e+064, 1e+065, 1e+066, 1e+067, 1e+068, 1e+069,
        1e+070, 1e+071, 1e+072, 1e+073, 1e+074, 1e+075, 1e+076, 1e+077, 1e+078, 1e+079,
        1e+080, 1e+081, 1e+082, 1e+083, 1e+084, 1e+085, 1e+086, 1e+087, 1e+088, 1e+089,
        1e+090, 1e+091, 1e+092, 1e+093, 1e+094, 1e+095, 1e+096, 1e+097, 1e+098, 1e+099,

        1e+100, 1e+101, 1e+102, 1e+103, 1e+104, 1e+105, 1e+106, 1e+107, 1e+108, 1e+109,
        1e+110, 1e+111, 1e+112, 1e+113, 1e+114, 1e+115, 1e+116, 1e+117, 1e+118, 1e+119,
        1e+120, 1e+121, 1e+122, 1e+123, 1e+124, 1e+125, 1e+126, 1e+127, 1e+128, 1e+129,
        1e+130, 1e+131, 1e+132, 1e+133, 1e+134, 1e+135, 1e+136, 1e+137, 1e+138, 1e+139,
        1e+140, 1e+141, 1e+142, 1e+143, 1e+144, 1e+145, 1e+146, 1e+147, 1e+148, 1e+149,
        1e+150, 1e+151, 1e+152, 1e+153, 1e+154, 1e+155, 1e+156, 1e+157, 1e+158, 1e+159,
        1e+160, 1e+161, 1e+162, 1e+163, 1e+164, 1e+165, 1e+166, 1e+167, 1e+168, 1e+169,
        1e+170, 1e+171, 1e+172, 1e+173, 1e+174, 1e+175, 1e+176, 1e+177, 1e+178, 1e+179,
        1e+180, 1e+181, 1e+182, 1e+183, 1e+184, 1e+185, 1e+186, 1e+187, 1e+188, 1e+189,
        1e+190, 1e+191, 1e+192, 1e+193, 1e+194, 1e+195, 1e+196, 1e+197, 1e+198, 1e+199,

        1e+200, 1e+201, 1e+202, 1e+203, 1e+204, 1e+205, 1e+206, 1e+207, 1e+208, 1e+209,
        1e+210, 1e+211, 1e+212, 1e+213, 1e+214, 1e+215, 1e+216, 1e+217, 1e+218, 1e+219,
        1e+220, 1e+221, 1e+222, 1e+223, 1e+224, 1e+225, 1e+226, 1e+227, 1e+228, 1e+229,
        1e+230, 1e+231, 1e+232, 1e+233, 1e+234, 1e+235, 1e+236, 1e+237, 1e+238, 1e+239,
        1e+240, 1e+241, 1e+242, 1e+243, 1e+244, 1e+245, 1e+246, 1e+247, 1e+248, 1e+249,
        1e+250, 1e+251, 1e+252, 1e+253, 1e+254, 1e+255, 1e+256, 1e+257, 1e+258, 1e+259,
        1e+260, 1e+261, 1e+262, 1e+263, 1e+264, 1e+265, 1e+266, 1e+267, 1e+268, 1e+269,
        1e+270, 1e+271, 1e+272, 1e+273, 1e+274, 1e+275, 1e+276, 1e+277, 1e+278, 1e+279,
        1e+280, 1e+281, 1e+282, 1e+283, 1e+284, 1e+285, 1e+286, 1e+287, 1e+288, 1e+289,
        1e+290, 1e+291, 1e+292, 1e+293, 1e+294, 1e+295, 1e+296, 1e+297, 1e+298, 1e+299,

        1e+300, 1e+301, 1e+302, 1e+303, 1e+304, 1e+305, 1e+306, 1e+307, 1e+308 };

    if( exp > 308 )
    {
        return std::numeric_limits<double>::infinity();
    }
    else if( exp < -308 )
    {
        // due to the way pow10 is used by dec_to_float,
        // we can afford to return 0.0 here
        return 0.0;
    }
    else
    {
        exp += 308;
        BOOST_ASSERT(exp >= 0 && exp < 618);
        return tab[exp];
    }
}

inline
double
dec_to_float(
    std::uint64_t m,
    std::int32_t e,
    bool neg) noexcept
{
    // convert to double explicitly to silence warnings
    double x = static_cast<double>(m);
    if(neg)
        x = -x;

    if(e < -305)
    {
        x *= 1e-305 ;
        e += 305;
    }

    if(e >= -22 && e < 0)
        return x / pow10(-e);

    return x * pow10(e);
}

inline
bool
is_control(char c) noexcept
{
    return static_cast<unsigned char>(c) < 32;
}

inline
int
hex_digit(unsigned char c) noexcept
{
    // by Peter Dimov
    if( c >= '0' && c <= '9' )
        return c - '0';
    c &= ~0x20;
    if( c >= 'A' && c <= 'F' )
        return 10 + c - 'A';
    return -1;
}

} // detail

//----------------------------------------------------------

template< class Handler >
template< bool StackEmpty_, char First_ >
struct basic_parser<Handler>::
parse_number_helper
{
    basic_parser* parser;
    char const* p;

    template< std::size_t N >
    char const*
    operator()( mp11::mp_size_t<N> ) const
    {
        return parser->parse_number(
            p,
            std::integral_constant<bool, StackEmpty_>(),
            std::integral_constant<char, First_>(),
            std::integral_constant<
                number_precision, static_cast<number_precision>(N)>() );
    }
};

//----------------------------------------------------------

template<class Handler>
void
basic_parser<Handler>::
reserve()
{
    if(BOOST_JSON_LIKELY(
        ! st_.empty()))
        return;
    // Reserve the largest stack we need,
    // to avoid reallocation during suspend.
    st_.reserve(
        sizeof(state) + // document parsing state
        (sizeof(state) +
            sizeof(std::size_t)) * depth() + // array and object state + size
        sizeof(state) + // value parsing state
        sizeof(std::size_t) + // string size
        sizeof(state)); // comment state
}

//----------------------------------------------------------
//
// The sentinel value is returned by parse functions
// to indicate that the parser failed, or suspended.
// this is used as it is distinct from all valid values
// for data in write

template<class Handler>
const char*
basic_parser<Handler>::
sentinel()
{
    // the "+1" ensures that the returned pointer is unique even if
    // the given input buffer borders on this object
    return reinterpret_cast<
        const char*>(this) + 1;
}

template<class Handler>
bool
basic_parser<Handler>::
incomplete(
    const detail::const_stream_wrapper& cs)
{
    return cs.begin() == sentinel();
}

//----------------------------------------------------------
//
// These functions are declared with the BOOST_NOINLINE
// attribute to avoid polluting the parsers hot-path.
// They return the canary value to indicate suspension
// or failure.

template<class Handler>
const char*
basic_parser<Handler>::
suspend_or_fail(state st)
{
    if(BOOST_JSON_LIKELY(
        ! ec_ && more_))
    {
        // suspend
        reserve();
        st_.push_unchecked(st);
    }
    return sentinel();
}

template<class Handler>
const char*
basic_parser<Handler>::
suspend_or_fail(
    state st,
    std::size_t n)
{
    if(BOOST_JSON_LIKELY(
        ! ec_ && more_))
    {
        // suspend
        reserve();
        st_.push_unchecked(n);
        st_.push_unchecked(st);
    }
    return sentinel();
}


template<class Handler>
const char*
basic_parser<Handler>::
fail(const char* p) noexcept
{
    BOOST_ASSERT( p != sentinel() );
    end_ = p;
    return sentinel();
}

template<class Handler>
const char*
basic_parser<Handler>::
fail(
    const char* p,
    error ev,
    source_location const* loc) noexcept
{
    BOOST_ASSERT( p != sentinel() );
    end_ = p;
    ec_.assign(ev, loc);
    return sentinel();
}

template<class Handler>
const char*
basic_parser<Handler>::
maybe_suspend(
    const char* p,
    state st)
{
    if( p != sentinel() )
        end_ = p;
    if(BOOST_JSON_LIKELY(more_))
    {
        // suspend
        reserve();
        st_.push_unchecked(st);
    }
    return sentinel();
}

template<class Handler>
const char*
basic_parser<Handler>::
maybe_suspend(
    const char* p,
    state st,
    std::size_t n)
{
    BOOST_ASSERT( p != sentinel() );
    end_ = p;
    if(BOOST_JSON_LIKELY(more_))
    {
        // suspend
        reserve();
        st_.push_unchecked(n);
        st_.push_unchecked(st);
    }
    return sentinel();
}

template<class Handler>
const char*
basic_parser<Handler>::
maybe_suspend(
    const char* p,
    state st,
    const number& num)
{
    BOOST_ASSERT( p != sentinel() );
    end_ = p;
    if(BOOST_JSON_LIKELY(more_))
    {
        // suspend
        num_ = num;
        reserve();
        st_.push_unchecked(st);;
    }
    return sentinel();
}

template<class Handler>
const char*
basic_parser<Handler>::
suspend(
    const char* p,
    state st)
{
    BOOST_ASSERT( p != sentinel() );
    end_ = p;
    // suspend
    reserve();
    st_.push_unchecked(st);
    return sentinel();
}

template<class Handler>
const char*
basic_parser<Handler>::
suspend(
    const char* p,
    state st,
    const number& num)
{
    BOOST_ASSERT( p != sentinel() );
    end_ = p;
    // suspend
    num_ = num;
    reserve();
    st_.push_unchecked(st);
    return sentinel();
}

template<class Handler>
template<
    bool StackEmpty_/*,
    bool Terminal_*/>
const char*
basic_parser<Handler>::
parse_comment(const char* p,
    std::integral_constant<bool, StackEmpty_> stack_empty,
    /*std::integral_constant<bool, Terminal_>*/ bool terminal)
{
    detail::const_stream_wrapper cs(p, end_);
    const char* start = cs.begin();
    std::size_t remain;
    if(! stack_empty && ! st_.empty())
    {
        state st;
        st_.pop(st);
        switch(st)
        {
            default: BOOST_JSON_UNREACHABLE();
            case state::com1: goto do_com1;
            case state::com2: goto do_com2;
            case state::com3: goto do_com3;
            case state::com4: goto do_com4;
        }
    }
    BOOST_ASSERT(*cs == '/');
    ++cs;
do_com1:
    if(BOOST_JSON_UNLIKELY(! cs))
        return maybe_suspend(cs.begin(), state::com1);
    switch(*cs)
    {
    default:
        {
            BOOST_STATIC_CONSTEXPR source_location loc
                = BOOST_CURRENT_LOCATION;
            return fail(cs.begin(), error::syntax, &loc);
        }
    case '/':
        ++cs;
do_com2:
        // KRYSTIAN TODO: this is a mess, we have to fix this
        remain = cs.remain();
        cs = remain ? static_cast<const char*>(
            std::memchr(cs.begin(), '\n', remain)) : sentinel();
        if(! cs.begin())
            cs = sentinel();
        if(BOOST_JSON_UNLIKELY(incomplete(cs)))
        {
            // if the doc does not terminate
            // with a newline, treat it as the
            // end of the comment
            if(terminal && ! more_)
            {
                if(BOOST_JSON_UNLIKELY(! h_.on_comment(
                    {start, cs.remain(start)}, ec_)))
                    return fail(cs.end());
                return cs.end();
            }
            if(BOOST_JSON_UNLIKELY(! h_.on_comment_part(
                {start, cs.remain(start)}, ec_)))
                return fail(cs.end());
            if(terminal)
                return suspend(cs.end(), state::com2);
            return maybe_suspend(cs.end(), state::com2);
        }
        break;
    case '*':
        do
        {
            ++cs;
do_com3:
            // KRYSTIAN TODO: this is a mess, we have to fix this
            remain = cs.remain();
            cs = remain ? static_cast<const char*>(
                std::memchr(cs.begin(), '*', remain)) : sentinel();
            if(! cs.begin())
                cs = sentinel();
            // stopped inside a c comment
            if(BOOST_JSON_UNLIKELY(incomplete(cs)))
            {
                if(BOOST_JSON_UNLIKELY(! h_.on_comment_part(
                    {start, cs.remain(start)}, ec_)))
                    return fail(cs.end());
                return maybe_suspend(cs.end(), state::com3);
            }
            // found a asterisk, check if the next char is a slash
            ++cs;
do_com4:
            if(BOOST_JSON_UNLIKELY(! cs))
            {
                if(BOOST_JSON_UNLIKELY(! h_.on_comment_part(
                    {start, cs.used(start)}, ec_)))
                    return fail(cs.begin());
                return maybe_suspend(cs.begin(), state::com4);
            }
        }
        while(*cs != '/');
    }
    ++cs;
    if(BOOST_JSON_UNLIKELY(! h_.on_comment(
        {start, cs.used(start)}, ec_)))
        return fail(cs.begin());
    return cs.begin();
}

template<class Handler>
template<bool StackEmpty_>
const char*
basic_parser<Handler>::
parse_document(const char* p,
    std::integral_constant<bool, StackEmpty_> stack_empty)
{
    detail::const_stream_wrapper cs(p, end_);
    if(! stack_empty && ! st_.empty())
    {
        state st;
        st_.peek(st);
        switch(st)
        {
        default: goto do_doc2;
        case state::doc1:
                 st_.pop(st);
                 goto do_doc1;
        case state::doc3:
                 st_.pop(st);
                 goto do_doc3;
        case state::com1: case state::com2:
        case state::com3: case state::com4:
                 goto do_doc4;
        }
    }
do_doc1:
    cs = detail::count_whitespace(cs.begin(), cs.end());
    if(BOOST_JSON_UNLIKELY(! cs))
        return maybe_suspend(cs.begin(), state::doc1);
do_doc2:
    switch(+opt_.allow_comments |
        (opt_.allow_trailing_commas << 1) |
        (opt_.allow_invalid_utf8 << 2))
    {
    // no extensions
    default:
        cs = parse_value(cs.begin(), stack_empty, std::false_type(), std::false_type(), std::false_type(), opt_.allow_invalid_utf16);
        break;
    // comments
    case 1:
        cs = parse_value(cs.begin(), stack_empty, std::true_type(), std::false_type(), std::false_type(), opt_.allow_invalid_utf16);
        break;
    // trailing
    case 2:
        cs = parse_value(cs.begin(), stack_empty, std::false_type(), std::true_type(), std::false_type(), opt_.allow_invalid_utf16);
        break;
    // comments & trailing
    case 3:
        cs = parse_value(cs.begin(), stack_empty, std::true_type(), std::true_type(), std::false_type(), opt_.allow_invalid_utf16);
        break;
    // skip validation
    case 4:
        cs = parse_value(cs.begin(), stack_empty, std::false_type(), std::false_type(), std::true_type(), opt_.allow_invalid_utf16);
        break;
    // comments & skip validation
    case 5:
        cs = parse_value(cs.begin(), stack_empty, std::true_type(), std::false_type(), std::true_type(), opt_.allow_invalid_utf16);
        break;
    // trailing & skip validation
    case 6:
        cs = parse_value(cs.begin(), stack_empty, std::false_type(), std::true_type(), std::true_type(), opt_.allow_invalid_utf16);
        break;
    // comments & trailing & skip validation
    case 7:
        cs = parse_value(cs.begin(), stack_empty, std::true_type(), std::true_type(), std::true_type(), opt_.allow_invalid_utf16);
        break;
    }
    if(BOOST_JSON_UNLIKELY(incomplete(cs)))
        // the appropriate state has already been pushed into stack
        return sentinel();
do_doc3:
    cs = detail::count_whitespace(cs.begin(), cs.end());
    if(BOOST_JSON_UNLIKELY(! cs))
    {
        if(more_)
            return suspend(cs.begin(), state::doc3);
    }
    else if(opt_.allow_comments && *cs == '/')
    {
do_doc4:
        cs = parse_comment(cs.begin(), stack_empty, std::true_type());
        if(BOOST_JSON_UNLIKELY(incomplete(cs)))
            return sentinel();
        goto do_doc3;
    }
    return cs.begin();
}

template<class Handler>
template<
    bool StackEmpty_,
    bool AllowComments_/*,
    bool AllowTrailing_,
    bool AllowBadUTF8_*/>
const char*
basic_parser<Handler>::
parse_value(const char* p,
    std::integral_constant<bool, StackEmpty_> stack_empty,
    std::integral_constant<bool, AllowComments_> allow_comments,
    /*std::integral_constant<bool, AllowTrailing_>*/ bool allow_trailing,
    /*std::integral_constant<bool, AllowBadUTF8_>*/ bool allow_bad_utf8,
    bool allow_bad_utf16)
{
    if(stack_empty || st_.empty())
    {
loop:
        switch(*p)
        {
        case '0':
            return mp11::mp_with_index<3>(
                static_cast<unsigned char>(opt_.numbers),
                parse_number_helper<true, '0'>{ this, p });
        case '-':
            return mp11::mp_with_index<3>(
                static_cast<unsigned char>(opt_.numbers),
                parse_number_helper<true, '-'>{ this, p });
        case '1': case '2': case '3':
        case '4': case '5': case '6':
        case '7': case '8': case '9':
            return mp11::mp_with_index<3>(
                static_cast<unsigned char>(opt_.numbers),
                parse_number_helper<true, '+'>{ this, p });
        case 'n':
            return parse_literal( p, detail::literals_c<detail::literals::null>() );
        case 't':
            return parse_literal( p, detail::literals_c<detail::literals::true_>() );
        case 'f':
            return parse_literal( p, detail::literals_c<detail::literals::false_>() );
        case 'I':
            if( !opt_.allow_infinity_and_nan )
            {
                BOOST_STATIC_CONSTEXPR source_location loc
                    = BOOST_CURRENT_LOCATION;
                return fail(p, error::syntax, &loc);
            }
            return parse_literal( p, detail::literals_c<detail::literals::infinity>() );
        case 'N':
            if( !opt_.allow_infinity_and_nan )
            {
                BOOST_STATIC_CONSTEXPR source_location loc
                    = BOOST_CURRENT_LOCATION;
                return fail(p, error::syntax, &loc);
            }
            return parse_literal(p, detail::literals_c<detail::literals::nan>() );
        case '"':
            return parse_string(p, std::true_type(), std::false_type(), allow_bad_utf8, allow_bad_utf16);
        case '[':
            return parse_array(p, std::true_type(), allow_comments, allow_trailing, allow_bad_utf8, allow_bad_utf16);
        case '{':
            return parse_object(p, std::true_type(), allow_comments, allow_trailing, allow_bad_utf8, allow_bad_utf16);
        case '/':
            if(! allow_comments)
            {
                BOOST_STATIC_CONSTEXPR source_location loc
                    = BOOST_CURRENT_LOCATION;
                return fail(p, error::syntax, &loc);
            }
            p = parse_comment(p, stack_empty, std::false_type());
            // KRYSTIAN NOTE: incomplete takes const_stream, we either
            // can add an overload, change the existing one to take a pointer,
            // or just leave it as is
            if(BOOST_JSON_UNLIKELY(p == sentinel()))
                return maybe_suspend(p, state::val2);
            BOOST_FALLTHROUGH;
        case ' ':
        case '\t':
        case '\n':
        case '\r':
            p = detail::count_whitespace(p, end_);
            if(BOOST_JSON_UNLIKELY(p == end_))
                return maybe_suspend(p, state::val1);
            goto loop;
        default:
            {
                BOOST_STATIC_CONSTEXPR source_location loc
                    = BOOST_CURRENT_LOCATION;
                return fail(p, error::syntax, &loc);
            }
        }
    }
    return resume_value(p, allow_comments, allow_trailing, allow_bad_utf8, allow_bad_utf16);
}

template<class Handler>
template<
    bool AllowComments_/*,
    bool AllowTrailing_,
    bool AllowBadUTF8_*/>
const char*
basic_parser<Handler>::
resume_value(const char* p,
    std::integral_constant<bool, AllowComments_> allow_comments,
    /*std::integral_constant<bool, AllowTrailing_>*/ bool allow_trailing,
    /*std::integral_constant<bool, AllowBadUTF8_>*/ bool allow_bad_utf8,
    bool allow_bad_utf16)
{
    state st;
    st_.peek(st);
    switch(st)
    {
    default: BOOST_JSON_UNREACHABLE();
    case state::lit1:
        return parse_literal(p,  detail::literals_c<detail::literals::resume>() );

    case state::str1: case state::str2:
    case state::str8:
        return parse_string(p, std::false_type(), std::false_type(), allow_bad_utf8, allow_bad_utf16);

    case state::arr1: case state::arr2:
    case state::arr3: case state::arr4:
    case state::arr5: case state::arr6:
        return parse_array(p, std::false_type(), allow_comments, allow_trailing, allow_bad_utf8, allow_bad_utf16);

    case state::obj1: case state::obj2:
    case state::obj3: case state::obj4:
    case state::obj5: case state::obj6:
    case state::obj7: case state::obj8:
    case state::obj9: case state::obj10:
    case state::obj11:
        return parse_object(p, std::false_type(), allow_comments, allow_trailing, allow_bad_utf8, allow_bad_utf16);

    case state::num1: case state::num2:
    case state::num3: case state::num4:
    case state::num5: case state::num6:
    case state::num7: case state::num8:
    case state::exp1: case state::exp2:
    case state::exp3:
        return mp11::mp_with_index<3>(
            static_cast<unsigned char>(opt_.numbers),
            parse_number_helper<false, 0>{ this, p });

    // KRYSTIAN NOTE: these are special cases
    case state::val1:
    {
        st_.pop(st);
        BOOST_ASSERT(st_.empty());
        p = detail::count_whitespace(p, end_);
        if(BOOST_JSON_UNLIKELY(p == end_))
            return maybe_suspend(p, state::val1);
        return parse_value(p, std::true_type(), allow_comments, allow_trailing, allow_bad_utf8, allow_bad_utf16);
    }

    case state::val2:
    {
        st_.pop(st);
        p = parse_comment(p, std::false_type(), std::false_type());
        if(BOOST_JSON_UNLIKELY(p == sentinel()))
            return maybe_suspend(p, state::val2);
        if(BOOST_JSON_UNLIKELY( p == end_ ))
            return maybe_suspend(p, state::val3);
        BOOST_ASSERT(st_.empty());
        return parse_value(p, std::true_type(), std::true_type(), allow_trailing, allow_bad_utf8, allow_bad_utf16);
    }

    case state::val3:
    {
        st_.pop(st);
        return parse_value(p, std::true_type(), std::true_type(), allow_trailing, allow_bad_utf8, allow_bad_utf16);
    }
    }
}

template<class Handler>
template<class Literal>
const char*
basic_parser<Handler>::
parse_literal(const char* p, Literal)
{
    using L = detail::literals;

    std::size_t cur_lit;
    std::size_t offset;

    detail::const_stream_wrapper cs(p, end_);
    BOOST_IF_CONSTEXPR( Literal::value != L::resume )
    {
        constexpr std::size_t index = literal_index(Literal::value);
        constexpr char const* literal = detail::literal_strings[index];
        constexpr std::size_t sz = detail::literal_sizes[index];

        if(BOOST_JSON_LIKELY( cs.remain() >= sz ))
        {
            int const cmp = std::memcmp(cs.begin(), literal, sz);
            if( cmp != 0 )
            {
                BOOST_STATIC_CONSTEXPR source_location loc = BOOST_CURRENT_LOCATION;
                return fail(cs.begin(), error::syntax, &loc);
            }

            BOOST_IF_CONSTEXPR( Literal::value == L::null )
            {
                if(BOOST_JSON_UNLIKELY(
                    ! h_.on_null(ec_)))
                    return fail(cs.begin());
            }
            else BOOST_IF_CONSTEXPR( Literal::value == L::true_ )
            {
                if(BOOST_JSON_UNLIKELY(
                    ! h_.on_bool(true, ec_)))
                    return fail(cs.begin());
            }
            else BOOST_IF_CONSTEXPR( Literal::value == L::false_ )
            {
                if(BOOST_JSON_UNLIKELY(
                    ! h_.on_bool(false, ec_)))
                    return fail(cs.begin());
            }
            else BOOST_IF_CONSTEXPR( Literal::value == L::infinity )
            {
                if(BOOST_JSON_UNLIKELY(
                    ! h_.on_double(
                        std::numeric_limits<double>::infinity(),
                        string_view(literal, sz),
                        ec_)))
                    return fail(cs.begin());
            }
            else BOOST_IF_CONSTEXPR( Literal::value == L::neg_infinity )
            {
                if(BOOST_JSON_UNLIKELY(
                    ! h_.on_double(
                        -std::numeric_limits<double>::infinity(),
                        string_view(literal, sz),
                        ec_)))
                    return fail(cs.begin());
            }
            else BOOST_IF_CONSTEXPR( Literal::value == L::nan )
            {
                if(BOOST_JSON_UNLIKELY(
                    ! h_.on_double(
                        std::numeric_limits<double>::quiet_NaN(),
                        string_view(literal, sz),
                        ec_)))
                    return fail(cs.begin());
            }
            else
            {
                BOOST_JSON_UNREACHABLE();
            }

            cs += sz;
            return cs.begin();
        }

        offset = 0;
        cur_lit = index;
    }
    else
    {
        state st;
        st_.pop(st);
        BOOST_ASSERT( st == state::lit1 );

        cur_lit = cur_lit_;
        offset = lit_offset_;
    }

    std::size_t const lit_size = detail::literal_sizes[cur_lit];
    std::size_t const size = (std::min)( lit_size - offset, cs.remain() );
    int cmp = 0;
    if(BOOST_JSON_LIKELY( cs.begin() ))
        cmp = std::memcmp(
            cs.begin(), detail::literal_strings[cur_lit] + offset, size );
    if( cmp != 0 )
    {
        BOOST_STATIC_CONSTEXPR source_location loc = BOOST_CURRENT_LOCATION;
        return fail(cs.begin(), error::syntax, &loc);
    }

    if(BOOST_JSON_UNLIKELY( offset + size < lit_size ))
    {
        BOOST_ASSERT( cur_lit < 256 );
        cur_lit_ = static_cast<unsigned char>( cur_lit );
        BOOST_ASSERT( offset + size < 256 );
        lit_offset_ = static_cast<unsigned char>( offset + size );
        return maybe_suspend(cs.begin() + size, state::lit1);
    }

    switch( static_cast<L>(cur_lit) )
    {
    case L::null:
        if(BOOST_JSON_UNLIKELY(
            ! h_.on_null(ec_)))
            return fail(cs.begin());
        break;
    case L::true_:
        if(BOOST_JSON_UNLIKELY(
            ! h_.on_bool(true, ec_)))
            return fail(cs.begin());
        break;
    case L::false_:
        if(BOOST_JSON_UNLIKELY(
            ! h_.on_bool(false, ec_)))
            return fail(cs.begin());
        break;
    case L::infinity:
        if(BOOST_JSON_UNLIKELY(
            ! h_.on_double(
                std::numeric_limits<double>::infinity(),
                string_view(
                    detail::literal_strings[ literal_index(L::infinity) ],
                    detail::literal_sizes[ literal_index(L::infinity) ]),
                ec_)))
            return fail(cs.begin());
        break;
    case L::neg_infinity:
        if(BOOST_JSON_UNLIKELY(
            ! h_.on_double(
                -std::numeric_limits<double>::infinity(),
                string_view(
                    detail::literal_strings[ literal_index(L::neg_infinity) ],
                    detail::literal_sizes[ literal_index(L::neg_infinity) ]),
                ec_)))
            return fail(cs.begin());
        break;
    case L::nan:
        if(BOOST_JSON_UNLIKELY(
            ! h_.on_double(
                std::numeric_limits<double>::quiet_NaN(),
                string_view(
                    detail::literal_strings[ literal_index(L::nan) ],
                    detail::literal_sizes[ literal_index(L::nan) ]),
                ec_)))
            return fail(cs.begin());
        break;
    default: BOOST_JSON_UNREACHABLE();
    }

    cs += size;
    return cs.begin();
}

//----------------------------------------------------------

template<class Handler>
template<bool StackEmpty_, bool IsKey_>
const char*
basic_parser<Handler>::
parse_string(const char* p,
    std::integral_constant<bool, StackEmpty_> stack_empty,
    std::integral_constant<bool, IsKey_> is_key,
    bool allow_bad_utf8,
    bool allow_bad_utf16)
{
    detail::const_stream_wrapper cs(p, end_);
    std::size_t total;
    char const* start;
    std::size_t size;
    if(! stack_empty && ! st_.empty())
    {
        state st;
        st_.pop(st);
        st_.pop(total);
        switch(st)
        {
        default: BOOST_JSON_UNREACHABLE();
        case state::str2: goto do_str2;
        case state::str8: goto do_str8;
        case state::str1: break;
        }
    }
    else
    {
        BOOST_ASSERT(*cs == '\x22'); // '"'
        ++cs;
        total = 0;
    }

do_str1:
    start = cs.begin();
    cs = allow_bad_utf8?
        detail::count_valid<true>(cs.begin(), cs.end()):
        detail::count_valid<false>(cs.begin(), cs.end());
    size = cs.used(start);
    if(is_key)
    {
        BOOST_ASSERT(total <= Handler::max_key_size);
        if(BOOST_JSON_UNLIKELY(size >
            Handler::max_key_size - total))
        {
            BOOST_STATIC_CONSTEXPR source_location loc
                = BOOST_CURRENT_LOCATION;
            return fail(cs.begin(), error::key_too_large, &loc);
        }
    }
    else
    {
        BOOST_ASSERT(total <= Handler::max_string_size);
        if(BOOST_JSON_UNLIKELY(size >
            Handler::max_string_size - total))
        {
            BOOST_STATIC_CONSTEXPR source_location loc
                = BOOST_CURRENT_LOCATION;
            return fail(cs.begin(), error::string_too_large, &loc);
        }
    }
    total += size;
    if(BOOST_JSON_UNLIKELY(! cs))
    {
        // call handler if the string isn't empty
        if(BOOST_JSON_LIKELY(size))
        {
            {
                bool r = is_key?
                    h_.on_key_part( {start, size}, total, ec_ ):
                    h_.on_string_part( {start, size}, total, ec_ );

                if(BOOST_JSON_UNLIKELY(!r))
                {
                    return fail(cs.begin());
                }
            }
        }
        return maybe_suspend(cs.begin(), state::str1, total);
    }
    // at this point all valid characters have been skipped, so any remaining
    // if there are any more characters, they are either escaped, or incomplete
    // utf8, or invalid utf8
    if(BOOST_JSON_UNLIKELY(*cs != '\x22')) // '"'
    {
        // sequence is invalid or incomplete
        if((*cs & 0x80) && !allow_bad_utf8)
        {
            seq_.save(cs.begin(), cs.remain());
            if(BOOST_JSON_UNLIKELY(seq_.complete()))
            {
                BOOST_STATIC_CONSTEXPR source_location loc
                    = BOOST_CURRENT_LOCATION;
                return fail(cs.begin(), error::syntax, &loc);
            }
            if(BOOST_JSON_LIKELY(size))
            {
                bool const r = is_key?
                    h_.on_key_part( {start, size}, total, ec_ ):
                    h_.on_string_part( {start, size}, total, ec_ );
                if(BOOST_JSON_UNLIKELY( !r ))
                    return fail( cs.begin() );
            }
            return maybe_suspend(cs.end(), state::str8, total);
        }
        else if(BOOST_JSON_LIKELY(*cs == '\\'))
        {
            // flush unescaped run from input
            if(BOOST_JSON_LIKELY(size))
            {
                bool const r = is_key?
                    h_.on_key_part( {start, size}, total, ec_ ):
                    h_.on_string_part( {start, size}, total, ec_ );
                if(BOOST_JSON_UNLIKELY( !r ))
                    return fail( cs.begin() );
            }
do_str2:
            cs = parse_escaped(cs.begin(), total, stack_empty, is_key, allow_bad_utf16);
            if(BOOST_JSON_UNLIKELY( incomplete(cs) ))
                return suspend_or_fail(state::str2, total);

            goto do_str1;
        }
        // illegal control
        BOOST_STATIC_CONSTEXPR source_location loc = BOOST_CURRENT_LOCATION;
        return fail(cs.begin(), error::syntax, &loc);
    }

    {
        bool r = is_key?
            h_.on_key( {start, size}, total, ec_ ):
            h_.on_string( {start, size}, total, ec_ );

        if(BOOST_JSON_UNLIKELY(!r))
        {
            return fail(cs.begin());
        }
    }

    ++cs;
    return cs.begin();

do_str8:
    uint8_t needed = seq_.needed();
    if(BOOST_JSON_UNLIKELY( !seq_.append(cs.begin(), cs.remain()) ))
        return maybe_suspend(cs.end(), state::str8, total);
    if(BOOST_JSON_UNLIKELY( !seq_.valid() ))
    {
        BOOST_STATIC_CONSTEXPR source_location loc = BOOST_CURRENT_LOCATION;
        return fail(cs.begin(), error::syntax, &loc);
    }
    {
        std::size_t const n = seq_.length();
        bool r;
        if(is_key)
        {
            BOOST_ASSERT(total <= Handler::max_key_size);
            if(BOOST_JSON_UNLIKELY(n >
                Handler::max_key_size - total))
            {
                BOOST_STATIC_CONSTEXPR source_location loc
                    = BOOST_CURRENT_LOCATION;
                return fail(cs.begin(), error::key_too_large, &loc);
            }
            total += n;
            r = h_.on_key_part( {seq_.data(), n}, total, ec_ );
        }
        else
        {
            BOOST_ASSERT(total <= Handler::max_string_size);
            if(BOOST_JSON_UNLIKELY(n >
                Handler::max_string_size - total))
            {
                BOOST_STATIC_CONSTEXPR source_location loc
                    = BOOST_CURRENT_LOCATION;
                return fail(cs.begin(), error::string_too_large, &loc);
            }
            total += n;
            r = h_.on_string_part( {seq_.data(), n}, total, ec_ );
        }
        if(BOOST_JSON_UNLIKELY( !r ))
            return fail( cs.begin() );
    }
    cs += needed;
    goto do_str1;
}

template<class Handler>
template<bool StackEmpty_>
const char*
basic_parser<Handler>::
parse_escaped(
    const char* p,
    std::size_t& total,
    std::integral_constant<bool, StackEmpty_> stack_empty,
    bool is_key,
    bool allow_bad_utf16)
{
    constexpr unsigned urc = 0xFFFD; // Unicode replacement character
    auto const ev_too_large = is_key?
        error::key_too_large : error::string_too_large;
    auto const max_size = is_key?
        Handler::max_key_size : Handler::max_string_size;
    int digit;

    //---------------------------------------------------------------
    //
    // To handle escapes, a local temporary buffer accumulates
    // the unescaped result. The algorithm attempts to fill the
    // buffer to capacity before invoking the handler.
    // In some cases the temporary buffer needs to be flushed
    // before it is full:
    // * When the closing double quote is seen
    // * When there in no more input (and more is expected later)
    // A goal of the algorithm is to call the handler as few times
    // as possible. Thus, when the first escape is encountered,
    // the algorithm attempts to fill the temporary buffer first.
    //
    detail::buffer<BOOST_JSON_STACK_BUFFER_SIZE> temp;

    // Unescaped JSON is never larger than its escaped version.
    // To efficiently process only what will fit in the temporary buffer,
    // the size of the input stream is temporarily "clipped" to the size
    // of the temporary buffer.
    // handle escaped character
    detail::clipped_const_stream cs(p, end_);
    cs.clip(temp.max_size());

    if(! stack_empty && ! st_.empty())
    {
        state st;
        st_.pop(st);
        switch(st)
        {
        default: BOOST_JSON_UNREACHABLE();
        case state::str3: goto do_str3;
        case state::str4: goto do_str4;
        case state::str5: goto do_str5;
        case state::str6: goto do_str6;
        case state::str7: goto do_str7;
        case state::sur1: goto do_sur1;
        case state::sur2: goto do_sur2;
        case state::sur3: goto do_sur3;
        case state::sur4: goto do_sur4;
        case state::sur5: goto do_sur5;
        case state::sur6: goto do_sur6;
        }
    }

    while(true)
    {
        BOOST_ASSERT( temp.capacity() );
        BOOST_ASSERT(*cs == '\\');
        ++cs;
do_str3:
        if(BOOST_JSON_UNLIKELY(! cs))
        {
            if(BOOST_JSON_LIKELY(! temp.empty()))
            {
                BOOST_ASSERT(total <= max_size);
                if(BOOST_JSON_UNLIKELY(
                    temp.size() > max_size - total))
                {
                    BOOST_STATIC_CONSTEXPR source_location loc
                        = BOOST_CURRENT_LOCATION;
                    return fail(cs.begin(), ev_too_large, &loc);
                }
                total += temp.size();
                {
                    bool r = is_key
                        ? h_.on_key_part(temp.get(), total, ec_)
                        : h_.on_string_part(temp.get(), total, ec_);

                    if(BOOST_JSON_UNLIKELY(!r))
                    {
                        return fail(cs.begin());
                    }
                }
                temp.clear();
            }
            cs.clip(temp.max_size());
            if(BOOST_JSON_UNLIKELY(! cs))
                return maybe_suspend(cs.begin(), state::str3);
        }
        switch(*cs)
        {
        default:
            {
                BOOST_STATIC_CONSTEXPR source_location loc
                    = BOOST_CURRENT_LOCATION;
                return fail(cs.begin(), error::syntax, &loc);
            }
        case '\x22': // '"'
            temp.push_back('\x22');
            ++cs;
            break;
        case '\\':
            temp.push_back('\\');
            ++cs;
            break;
        case '/':
            temp.push_back('/');
            ++cs;
            break;
        case 'b':
            temp.push_back('\x08');
            ++cs;
            break;
        case 'f':
            temp.push_back('\x0c');
            ++cs;
            break;
        case 'n':
            temp.push_back('\x0a');
            ++cs;
            break;
        case 'r':
            temp.push_back('\x0d');
            ++cs;
            break;
        case 't':
            temp.push_back('\x09');
            ++cs;
            break;
        case 'u':
            // utf16 escape
            //
            // fast path only when the buffer
            // is large enough for 2 surrogates
            if(BOOST_JSON_LIKELY(cs.remain() > 10))
            {
                // KRYSTIAN TODO: this could be done
                // with fewer instructions
                digit = detail::load_little_endian<4>(
                    cs.begin() + 1);
                int d4 = detail::hex_digit(static_cast<
                    unsigned char>(digit >> 24));
                int d3 = detail::hex_digit(static_cast<
                    unsigned char>(digit >> 16));
                int d2 = detail::hex_digit(static_cast<
                    unsigned char>(digit >> 8));
                int d1 = detail::hex_digit(static_cast<
                    unsigned char>(digit));
                if(BOOST_JSON_UNLIKELY(
                    (d1 | d2 | d3 | d4) == -1))
                {
                    if(d1 != -1)
                        ++cs;
                    if(d2 != -1)
                        ++cs;
                    if(d3 != -1)
                        ++cs;
                    BOOST_STATIC_CONSTEXPR source_location loc
                        = BOOST_CURRENT_LOCATION;
                    return fail(cs.begin(), error::expected_hex_digit, &loc);
                }
                // 32 bit unicode scalar value
                unsigned u1 =
                    (d1 << 12) + (d2 << 8) +
                    (d3 << 4) + d4;
                // valid unicode scalar values are
                // [0, D7FF] and [E000, 10FFFF]
                // values within this range are valid utf-8
                // code points and invalid leading surrogates.
                if(BOOST_JSON_LIKELY(
                    u1 < 0xd800 || u1 > 0xdfff))
                {
                    cs += 5;
                    temp.append_utf8(u1);
                    break;
                }
                if(BOOST_JSON_UNLIKELY(u1 > 0xdbff))
                {
                    // If it's an illegal leading surrogate and
                    // the parser does not allow it, return an error.
                    if(!allow_bad_utf16)
                    {
                        BOOST_STATIC_CONSTEXPR source_location loc
                            = BOOST_CURRENT_LOCATION;
                        return fail(cs.begin(), error::illegal_leading_surrogate,
                            &loc);
                    }
                    // Otherwise, append the Unicode replacement character
                    else
                    {
                        cs += 5;
                        temp.append_utf8(urc);
                        break;
                    }
                }
                cs += 5;
                // KRYSTIAN TODO: this can be a two byte load
                // and a single comparison. We lose error information,
                // but it's faster.
                if(BOOST_JSON_UNLIKELY(*cs != '\\'))
                {
                    // If the next character is not a backslash and
                    // the parser does not allow it, return a syntax error.
                    if(!allow_bad_utf16)
                    {
                        BOOST_STATIC_CONSTEXPR source_location loc
                            = BOOST_CURRENT_LOCATION;
                        return fail(cs.begin(), error::syntax, &loc);
                    }
                    // Otherwise, append the Unicode replacement character since
                    // the first code point is a valid leading surrogate
                    else
                    {
                        temp.append_utf8(urc);
                        break;
                    }
                }
                ++cs;
                if(BOOST_JSON_UNLIKELY(*cs != 'u'))
                {
                    if (!allow_bad_utf16)
                    {
                        BOOST_STATIC_CONSTEXPR source_location loc
                            = BOOST_CURRENT_LOCATION;
                        return fail(cs.begin(), error::syntax, &loc);
                    }
                    // Otherwise, append the Unicode replacement character since
                    // the first code point is a valid leading surrogate
                    else
                    {
                        temp.append_utf8(urc);
                        goto do_str3;
                    }
                }
                ++cs;
                digit = detail::load_little_endian<4>(cs.begin());
                d4 = detail::hex_digit(static_cast<
                    unsigned char>(digit >> 24));
                d3 = detail::hex_digit(static_cast<
                    unsigned char>(digit >> 16));
                d2 = detail::hex_digit(static_cast<
                    unsigned char>(digit >> 8));
                d1 = detail::hex_digit(static_cast<
                    unsigned char>(digit));
                if(BOOST_JSON_UNLIKELY(
                    (d1 | d2 | d3 | d4) == -1))
                {
                    if(d1 != -1)
                        ++cs;
                    if(d2 != -1)
                        ++cs;
                    if(d3 != -1)
                        ++cs;
                    BOOST_STATIC_CONSTEXPR source_location loc
                        = BOOST_CURRENT_LOCATION;
                    return fail(cs.begin(), error::expected_hex_digit, &loc);
                }
                unsigned u2 =
                    (d1 << 12) + (d2 << 8) +
                    (d3 << 4) + d4;
                // Check if the second code point is a valid trailing surrogate.
                // Valid trailing surrogates are [DC00, DFFF]
                if(BOOST_JSON_UNLIKELY(
                    u2 < 0xdc00 || u2 > 0xdfff))
                {
                    // If not valid and the parser does not allow it, return an error.
                    if(!allow_bad_utf16)
                    {
                        BOOST_STATIC_CONSTEXPR source_location loc
                            = BOOST_CURRENT_LOCATION;
                        return fail(cs.begin(), error::illegal_trailing_surrogate,
                            &loc);
                    }
                    // Append the replacement character for the
                    // first leading surrogate.
                    cs += 4;
                    temp.append_utf8(urc);
                    // Check if the second code point is a
                    // valid unicode scalar value (invalid leading
                    // or trailing surrogate)
                    if (u2 < 0xd800 || u2 > 0xdbff)
                    {
                        temp.append_utf8(u2);
                        break;
                    }
                    // If it is a valid leading surrogate
                    else
                    {
                        u1_ = u2;
                        goto do_sur1;
                    }
                }
                cs += 4;
                // Calculate the Unicode code point from the surrogate pair and
                // append the UTF-8 representation.
                unsigned cp =
                    ((u1 - 0xd800) << 10) +
                    ((u2 - 0xdc00)) +
                        0x10000;
                // utf-16 surrogate pair
                temp.append_utf8(cp);
                break;
            }
            // flush
            if(BOOST_JSON_LIKELY(! temp.empty()))
            {
                BOOST_ASSERT(total <= max_size);
                if(BOOST_JSON_UNLIKELY(
                    temp.size() > max_size - total))
                {
                    BOOST_STATIC_CONSTEXPR source_location loc
                        = BOOST_CURRENT_LOCATION;
                    return fail(cs.begin(), ev_too_large, &loc);
                }
                total += temp.size();
                {
                    bool r = is_key
                        ? h_.on_key_part(temp.get(), total, ec_)
                        : h_.on_string_part(temp.get(), total, ec_);

                    if(BOOST_JSON_UNLIKELY(!r))
                    {
                        return fail(cs.begin());
                    }
                }
                temp.clear();
                cs.clip(temp.max_size());
            }
            ++cs;
            // utf-16 escape
    do_str4:
            if(BOOST_JSON_UNLIKELY(! cs))
                return maybe_suspend(cs.begin(), state::str4);
            digit = detail::hex_digit(*cs);
            if(BOOST_JSON_UNLIKELY(digit == -1))
            {
                BOOST_STATIC_CONSTEXPR source_location loc
                    = BOOST_CURRENT_LOCATION;
                return fail(cs.begin(), error::expected_hex_digit, &loc);
            }
            ++cs;
            u1_ = digit << 12;
    do_str5:
            if(BOOST_JSON_UNLIKELY(! cs))
                return maybe_suspend(cs.begin(), state::str5);
            digit = detail::hex_digit(*cs);
            if(BOOST_JSON_UNLIKELY(digit == -1))
            {
                BOOST_STATIC_CONSTEXPR source_location loc
                    = BOOST_CURRENT_LOCATION;
                return fail(cs.begin(), error::expected_hex_digit, &loc);
            }
            ++cs;
            u1_ += digit << 8;
    do_str6:
            if(BOOST_JSON_UNLIKELY(! cs))
                return maybe_suspend(cs.begin(), state::str6);
            digit = detail::hex_digit(*cs);
            if(BOOST_JSON_UNLIKELY(digit == -1))
            {
                BOOST_STATIC_CONSTEXPR source_location loc
                    = BOOST_CURRENT_LOCATION;
                return fail(cs.begin(), error::expected_hex_digit, &loc);
            }
            ++cs;
            u1_ += digit << 4;
    do_str7:
            if(BOOST_JSON_UNLIKELY(! cs))
                return maybe_suspend(cs.begin(), state::str7);
            digit = detail::hex_digit(*cs);
            if(BOOST_JSON_UNLIKELY(digit == -1))
            {
                BOOST_STATIC_CONSTEXPR source_location loc
                    = BOOST_CURRENT_LOCATION;
                return fail(cs.begin(), error::expected_hex_digit, &loc);
            }
            ++cs;
            u1_ += digit;
            if(BOOST_JSON_LIKELY(
                u1_ < 0xd800 || u1_ > 0xdfff))
            {
                BOOST_ASSERT(temp.empty());
                // utf-8 codepoint
                temp.append_utf8(u1_);
                break;
            }
            if(BOOST_JSON_UNLIKELY(u1_ > 0xdbff))
            {
                // If it's an illegal leading surrogate and
                // the parser does not allow it, return an error.
                if(!allow_bad_utf16)
                {
                    BOOST_STATIC_CONSTEXPR source_location loc
                        = BOOST_CURRENT_LOCATION;
                    return fail(cs.begin(), error::illegal_leading_surrogate, &loc);
                }
                // Otherwise, append the Unicode replacement character
                else
                {
                    BOOST_ASSERT(temp.empty());
                    temp.append_utf8(urc);
                    break;
                }
            }
    do_sur1:
            if(BOOST_JSON_UNLIKELY(! cs))
                return maybe_suspend(cs.begin(), state::sur1);
            if(BOOST_JSON_UNLIKELY(*cs != '\\'))
            {
                // If the next character is not a backslash and
                // the parser does not allow it, return a syntax error.
                if(!allow_bad_utf16)
                {
                    BOOST_STATIC_CONSTEXPR source_location loc
                        = BOOST_CURRENT_LOCATION;
                    return fail(cs.begin(), error::syntax, &loc);
                }
                // Otherwise, append the Unicode replacement character since
                // the first code point is a valid leading surrogate
                else
                {
                    temp.append_utf8(urc);
                    break;
                }
            }
            ++cs;
    do_sur2:
            if(BOOST_JSON_UNLIKELY(! cs))
                return maybe_suspend(cs.begin(), state::sur2);
            if(BOOST_JSON_UNLIKELY(*cs != 'u'))
            {
                if (!allow_bad_utf16)
                {
                    BOOST_STATIC_CONSTEXPR source_location loc
                        = BOOST_CURRENT_LOCATION;
                    return fail(cs.begin(), error::syntax, &loc);
                }
                // Otherwise, append the Unicode replacement character since
                // the first code point is a valid leading surrogate
                else
                {
                    temp.append_utf8(urc);
                    goto do_str3;
                }
            }
            ++cs;
    do_sur3:
            if(BOOST_JSON_UNLIKELY(! cs))
                return maybe_suspend(cs.begin(), state::sur3);
            digit = detail::hex_digit(*cs);
            if(BOOST_JSON_UNLIKELY(digit == -1))
            {
                BOOST_STATIC_CONSTEXPR source_location loc
                    = BOOST_CURRENT_LOCATION;
                return fail(cs.begin(), error::expected_hex_digit, &loc);
            }
            ++cs;
            u2_ = digit << 12;
    do_sur4:
            if(BOOST_JSON_UNLIKELY(! cs))
                return maybe_suspend(cs.begin(), state::sur4);
            digit = detail::hex_digit(*cs);
            if(BOOST_JSON_UNLIKELY(digit == -1))
            {
                BOOST_STATIC_CONSTEXPR source_location loc
                    = BOOST_CURRENT_LOCATION;
                return fail(cs.begin(), error::expected_hex_digit, &loc);
            }
            ++cs;
            u2_ += digit << 8;
    do_sur5:
            if(BOOST_JSON_UNLIKELY(! cs))
                return maybe_suspend(cs.begin(), state::sur5);
            digit = detail::hex_digit(*cs);
            if(BOOST_JSON_UNLIKELY(digit == -1))
            {
                BOOST_STATIC_CONSTEXPR source_location loc
                    = BOOST_CURRENT_LOCATION;
                return fail(cs.begin(), error::expected_hex_digit, &loc);
            }
            ++cs;
            u2_ += digit << 4;
    do_sur6:
            if(BOOST_JSON_UNLIKELY(! cs))
                return maybe_suspend(cs.begin(), state::sur6);
            digit = detail::hex_digit(*cs);
            if(BOOST_JSON_UNLIKELY(digit == -1))
            {
                BOOST_STATIC_CONSTEXPR source_location loc
                    = BOOST_CURRENT_LOCATION;
                return fail(cs.begin(), error::expected_hex_digit, &loc);
            }
            ++cs;
            u2_ += digit;
            // Check if the second code point is a valid trailing surrogate.
            // Valid trailing surrogates are [DC00, DFFF]
            if(BOOST_JSON_UNLIKELY(
                u2_ < 0xdc00 || u2_ > 0xdfff))
            {
                // If not valid and the parser does not allow it, return an error.
                if(!allow_bad_utf16)
                {
                    BOOST_STATIC_CONSTEXPR source_location loc
                        = BOOST_CURRENT_LOCATION;
                    return fail(cs.begin(), error::illegal_trailing_surrogate, &loc);
                }
                // Append the replacement character for the
                // first leading surrogate.
                temp.append_utf8(urc);
                // Check if the second code point is a
                // valid unicode scalar value (invalid leading
                // or trailing surrogate)
                if (u2_ < 0xd800 || u2_ > 0xdbff)
                {
                    temp.append_utf8(u2_);
                    break;
                }
                // If it is a valid leading surrogate
                else
                {
                    u1_ = u2_;
                    goto do_sur1;
                }
            }
            // Calculate the Unicode code point from the surrogate pair and
            // append the UTF-8 representation.
            unsigned cp =
                ((u1_ - 0xd800) << 10) +
                ((u2_ - 0xdc00)) +
                    0x10000;
            // utf-16 surrogate pair
            temp.append_utf8(cp);
        }

        // flush
        if(BOOST_JSON_UNLIKELY( !cs ) || *cs != '\\')
            break;
    }

    if(BOOST_JSON_LIKELY( temp.size() ))
    {
        BOOST_ASSERT(total <= max_size);
        if(BOOST_JSON_UNLIKELY( temp.size() > max_size - total ))
        {
            BOOST_STATIC_CONSTEXPR source_location loc
                = BOOST_CURRENT_LOCATION;
            return fail(cs.begin(), ev_too_large, &loc);
        }

        total += temp.size();
        bool const r = is_key
            ? h_.on_key_part(temp.get(), total, ec_)
            : h_.on_string_part(temp.get(), total, ec_);
        if(BOOST_JSON_UNLIKELY( !r ))
            return fail( cs.begin() );
    }

    return cs.begin();
}

//----------------------------------------------------------

template<class Handler>
template<
    bool StackEmpty_,
    bool AllowComments_/*,
    bool AllowTrailing_,
    bool AllowBadUTF8_*/>
const char*
basic_parser<Handler>::
parse_object(const char* p,
    std::integral_constant<bool, StackEmpty_> stack_empty,
    std::integral_constant<bool, AllowComments_> allow_comments,
    /*std::integral_constant<bool, AllowTrailing_>*/ bool allow_trailing,
    /*std::integral_constant<bool, AllowBadUTF8_>*/ bool allow_bad_utf8,
    bool allow_bad_utf16)
{
    detail::const_stream_wrapper cs(p, end_);
    std::size_t size;
    if(! stack_empty && ! st_.empty())
    {
        // resume
        state st;
        st_.pop(st);
        st_.pop(size);
        switch(st)
        {
        default: BOOST_JSON_UNREACHABLE();
        case state::obj1: goto do_obj1;
        case state::obj2: goto do_obj2;
        case state::obj3: goto do_obj3;
        case state::obj4: goto do_obj4;
        case state::obj5: goto do_obj5;
        case state::obj6: goto do_obj6;
        case state::obj7: goto do_obj7;
        case state::obj8: goto do_obj8;
        case state::obj9: goto do_obj9;
        case state::obj10: goto do_obj10;
        case state::obj11: goto do_obj11;
        }
    }
    BOOST_ASSERT(*cs == '{');
    size = 0;
    if(BOOST_JSON_UNLIKELY(! depth_))
    {
        BOOST_STATIC_CONSTEXPR source_location loc = BOOST_CURRENT_LOCATION;
        return fail(cs.begin(), error::too_deep, &loc);
    }
    --depth_;
    if(BOOST_JSON_UNLIKELY(
        ! h_.on_object_begin(ec_)))
        return fail(cs.begin());
    ++cs;
    // object:
    //     '{' *ws '}'
    //     '{' *ws string *ws ':' *ws value *ws *[ ',' *ws string *ws ':' *ws value *ws ] '}'
do_obj1:
    cs = detail::count_whitespace(cs.begin(), cs.end());
    if(BOOST_JSON_UNLIKELY(! cs))
        return maybe_suspend(cs.begin(), state::obj1, size);
    if(BOOST_JSON_LIKELY(*cs != '}'))
    {
        if(BOOST_JSON_UNLIKELY(*cs != '\x22'))
        {
            if(allow_comments && *cs == '/')
            {
do_obj2:
                cs = parse_comment(cs.begin(), stack_empty, std::false_type());
                if(BOOST_JSON_UNLIKELY(incomplete(cs)))
                    return suspend_or_fail(state::obj2, size);
                goto do_obj1;
            }
            BOOST_STATIC_CONSTEXPR source_location loc
                = BOOST_CURRENT_LOCATION;
            return fail(cs.begin(), error::syntax, &loc);
        }
loop:
        if(BOOST_JSON_UNLIKELY(++size >
            Handler::max_object_size))
        {
            BOOST_STATIC_CONSTEXPR source_location loc
                = BOOST_CURRENT_LOCATION;
            return fail(cs.begin(), error::object_too_large, &loc);
        }
do_obj3:
        cs = parse_string(cs.begin(), stack_empty, std::true_type(), allow_bad_utf8, allow_bad_utf16);
        if(BOOST_JSON_UNLIKELY(incomplete(cs)))
            return suspend_or_fail(state::obj3, size);
do_obj4:
        cs = detail::count_whitespace(cs.begin(), cs.end());
        if(BOOST_JSON_UNLIKELY(! cs))
            return maybe_suspend(cs.begin(), state::obj4, size);
        if(BOOST_JSON_UNLIKELY(*cs != ':'))
        {
            if(allow_comments && *cs == '/')
            {
do_obj5:
                cs = parse_comment(cs.begin(), stack_empty, std::false_type());
                if(BOOST_JSON_UNLIKELY(incomplete(cs)))
                    return suspend_or_fail(state::obj5, size);
                goto do_obj4;
            }
            BOOST_STATIC_CONSTEXPR source_location loc
                = BOOST_CURRENT_LOCATION;
            return fail(cs.begin(), error::syntax, &loc);
        }
        ++cs;
do_obj6:
        cs = detail::count_whitespace(cs.begin(), cs.end());
        if(BOOST_JSON_UNLIKELY(! cs))
            return maybe_suspend(cs.begin(), state::obj6, size);
do_obj7:
        cs = parse_value(cs.begin(), stack_empty, allow_comments, allow_trailing, allow_bad_utf8, allow_bad_utf16);
        if(BOOST_JSON_UNLIKELY(incomplete(cs)))
            return suspend_or_fail(state::obj7, size);
do_obj8:
        cs = detail::count_whitespace(cs.begin(), cs.end());
        if(BOOST_JSON_UNLIKELY(! cs))
            return maybe_suspend(cs.begin(), state::obj8, size);
        if(BOOST_JSON_LIKELY(*cs == ','))
        {
            ++cs;
do_obj9:
            cs = detail::count_whitespace(cs.begin(), cs.end());
            if(BOOST_JSON_UNLIKELY(! cs))
                return maybe_suspend(cs.begin(), state::obj9, size);

            // loop for next element
            if(BOOST_JSON_LIKELY(*cs == '\x22'))
                goto loop;
            if(! allow_trailing || *cs != '}')
            {
                if(allow_comments && *cs == '/')
                {
do_obj10:
                    cs = parse_comment(cs.begin(), stack_empty, std::false_type());
                    if(BOOST_JSON_UNLIKELY(incomplete(cs)))
                        return suspend_or_fail(state::obj10, size);
                    goto do_obj9;
                }
                BOOST_STATIC_CONSTEXPR source_location loc
                    = BOOST_CURRENT_LOCATION;
                return fail(cs.begin(), error::syntax, &loc);
            }
        }
        else if(BOOST_JSON_UNLIKELY(*cs != '}'))
        {
            if(allow_comments && *cs == '/')
            {
do_obj11:
                cs = parse_comment(cs.begin(), stack_empty, std::false_type());
                if(BOOST_JSON_UNLIKELY(incomplete(cs)))
                    return suspend_or_fail(state::obj11, size);
                goto do_obj8;
            }
            BOOST_STATIC_CONSTEXPR source_location loc
                = BOOST_CURRENT_LOCATION;
            return fail(cs.begin(), error::syntax, &loc);
        }
        // got closing brace, fall through
    }
    if(BOOST_JSON_UNLIKELY(
        ! h_.on_object_end(size, ec_)))
        return fail(cs.begin());
    ++depth_;
    ++cs;
    return cs.begin();
}

//----------------------------------------------------------

template<class Handler>
template<
    bool StackEmpty_,
    bool AllowComments_/*,
    bool AllowTrailing_,
    bool AllowBadUTF8_*/>
const char*
basic_parser<Handler>::
parse_array(const char* p,
    std::integral_constant<bool, StackEmpty_> stack_empty,
    std::integral_constant<bool, AllowComments_> allow_comments,
    /*std::integral_constant<bool, AllowTrailing_>*/ bool allow_trailing,
    /*std::integral_constant<bool, AllowBadUTF8_>*/ bool allow_bad_utf8,
    bool allow_bad_utf16)
{
    detail::const_stream_wrapper cs(p, end_);
    std::size_t size;
    if(! stack_empty && ! st_.empty())
    {
        // resume
        state st;
        st_.pop(st);
        st_.pop(size);
        switch(st)
        {
        default: BOOST_JSON_UNREACHABLE();
        case state::arr1: goto do_arr1;
        case state::arr2: goto do_arr2;
        case state::arr3: goto do_arr3;
        case state::arr4: goto do_arr4;
        case state::arr5: goto do_arr5;
        case state::arr6: goto do_arr6;
        }
    }
    BOOST_ASSERT(*cs == '[');
    size = 0;
    if(BOOST_JSON_UNLIKELY(! depth_))
    {
        BOOST_STATIC_CONSTEXPR source_location loc = BOOST_CURRENT_LOCATION;
        return fail(cs.begin(), error::too_deep, &loc);
    }
    --depth_;
    if(BOOST_JSON_UNLIKELY(
        ! h_.on_array_begin(ec_)))
        return fail(cs.begin());
    ++cs;
    // array:
    //     '[' *ws ']'
    //     '[' *ws value *ws *[ ',' *ws value *ws ] ']'
do_arr1:
    cs = detail::count_whitespace(cs.begin(), cs.end());
    if(BOOST_JSON_UNLIKELY(! cs))
        return maybe_suspend(cs.begin(), state::arr1, size);
    if(BOOST_JSON_LIKELY(*cs != ']'))
    {
loop:
        if(allow_comments && *cs == '/')
        {
do_arr2:
            cs = parse_comment(cs.begin(), stack_empty, std::false_type());
            if(BOOST_JSON_UNLIKELY(incomplete(cs)))
                return suspend_or_fail(state::arr2, size);
            goto do_arr1;
        }
        if(BOOST_JSON_UNLIKELY(++size >
            Handler::max_array_size))
        {
            BOOST_STATIC_CONSTEXPR source_location loc
                = BOOST_CURRENT_LOCATION;
            return fail(cs.begin(), error::array_too_large, &loc);
        }
do_arr3:
        // array is not empty, value required
        cs = parse_value(cs.begin(), stack_empty, allow_comments, allow_trailing, allow_bad_utf8, allow_bad_utf16);
        if(BOOST_JSON_UNLIKELY(incomplete(cs)))
            return suspend_or_fail(state::arr3, size);
do_arr4:
        cs = detail::count_whitespace(cs.begin(), cs.end());
        if(BOOST_JSON_UNLIKELY(! cs))
            return maybe_suspend(cs.begin(), state::arr4, size);
        if(BOOST_JSON_LIKELY(*cs == ','))
        {
            ++cs;
do_arr5:
            cs = detail::count_whitespace(cs.begin(), cs.end());
            if(BOOST_JSON_UNLIKELY(! cs))
                return maybe_suspend(cs.begin(), state::arr5, size);
            // loop for next element
            if(! allow_trailing || *cs != ']')
                goto loop;
        }
        else if(BOOST_JSON_UNLIKELY(*cs != ']'))
        {
            if(allow_comments && *cs == '/')
            {
do_arr6:
                cs = parse_comment(cs.begin(), stack_empty, std::false_type());
                if(BOOST_JSON_UNLIKELY(incomplete(cs)))
                    return suspend_or_fail(state::arr6, size);
                goto do_arr4;
            }
            BOOST_STATIC_CONSTEXPR source_location loc
                = BOOST_CURRENT_LOCATION;
            return fail(cs.begin(), error::syntax, &loc);
        }
        // got closing bracket; fall through
    }
    if(BOOST_JSON_UNLIKELY(
        ! h_.on_array_end(size, ec_)))
        return fail(cs.begin());
    ++depth_;
    ++cs;
    return cs.begin();
}

//----------------------------------------------------------

template<class Handler>
template<bool StackEmpty_, char First_, number_precision Numbers_>
const char*
basic_parser<Handler>::
parse_number(const char* p,
    std::integral_constant<bool, StackEmpty_> stack_empty,
    std::integral_constant<char, First_> first,
    std::integral_constant<number_precision, Numbers_> mode)
{
    constexpr bool precise_parsing = mode == number_precision::precise;
    constexpr bool no_parsing = mode == number_precision::none;

    // only one of these will be true if we are not resuming
    // if negative then !zero_first && !nonzero_first
    // if zero_first then !nonzero_first && !negative
    // if nonzero_first then !zero_first && !negative
    bool const negative = first == '-';
    bool const zero_first = first == '0';
    bool const nonzero_first = first == '+';
    detail::const_stream_wrapper cs(p, end_);
    number num;
    const char* begin = cs.begin();
    if(stack_empty || st_.empty())
    {
        num.bias = 0;
        num.exp = 0;
        num.frac = false;
        num_buf_.clear();

        //----------------------------------
        //
        // '-'
        // leading minus sign
        //
        BOOST_ASSERT(cs);
        if(negative)
            ++cs;

        num.neg = negative;
        num.frac = false;
        num.exp = 0;
        num.bias = 0;

        // fast path
        if( cs.remain() >= 16 + 1 + 16 ) // digits . digits
        {
            int n1;

            if( nonzero_first ||
                (negative && *cs != '0') )
            {
                n1 = detail::count_digits( cs.begin() );
                BOOST_ASSERT(n1 >= 0 && n1 <= 16);

                if( negative && n1 == 0 && opt_.allow_infinity_and_nan )
                {
                    return parse_literal(
                        p - 1,
                        detail::literals_c<detail::literals::neg_infinity>());
                }

                if( ! nonzero_first && n1 == 0 )
                {
                    // digit required
                    BOOST_STATIC_CONSTEXPR source_location loc
                        = BOOST_CURRENT_LOCATION;
                    return fail(cs.begin(), error::syntax, &loc);
                }

                BOOST_IF_CONSTEXPR( !no_parsing )
                    num.mant = detail::parse_unsigned( 0, cs.begin(), n1 );
                else
                    num.mant = 0;

                cs += n1;

                // integer or floating-point with
                // >= 16 leading digits
                if( n1 == 16 )
                {
                    goto do_num2;
                }
            }
            else
            {
                // 0. floating-point or 0e integer
                num.mant = 0;
                n1 = 0;
                ++cs;
            }

            {
                const char c = *cs;
                if(c != '.')
                {
                    if((c | 32) == 'e')
                    {
                        ++cs;
                        goto do_exp1;
                    }
                    BOOST_IF_CONSTEXPR( negative && !no_parsing )
                        num.mant = ~num.mant + 1;
                    goto finish_signed;
                }
            }

            // floating-point number

            ++cs;

            int n2 = detail::count_digits( cs.begin() );
            BOOST_ASSERT(n2 >= 0 && n2 <= 16);

            if( n2 == 0 )
            {
                // digit required
                BOOST_STATIC_CONSTEXPR source_location loc
                    = BOOST_CURRENT_LOCATION;
                return fail(cs.begin(), error::syntax, &loc);
            }

            // floating-point mantissa overflow
            if( n1 + n2 >= 19 )
            {
                goto do_num7;
            }

            BOOST_IF_CONSTEXPR( !no_parsing )
                num.mant = detail::parse_unsigned( num.mant, cs.begin(), n2 );

            BOOST_ASSERT(num.bias == 0);

            num.bias -= n2;

            cs += n2;

            char ch = *cs;

            if( (ch | 32) == 'e' )
            {
                ++cs;
                goto do_exp1;
            }
            else if( ch >= '0' && ch <= '9' )
            {
                goto do_num8;
            }

            goto finish_dub;
        }
    }
    else
    {
        num = num_;
        state st;
        st_.pop(st);
        switch(st)
        {
        default: BOOST_JSON_UNREACHABLE();
        case state::num1: goto do_num1;
        case state::num2: goto do_num2;
        case state::num3: goto do_num3;
        case state::num4: goto do_num4;
        case state::num5: goto do_num5;
        case state::num6: goto do_num6;
        case state::num7: goto do_num7;
        case state::num8: goto do_num8;
        case state::exp1: goto do_exp1;
        case state::exp2: goto do_exp2;
        case state::exp3: goto do_exp3;
        }
    }

    //----------------------------------
    //
    // DIGIT
    // first digit
    //
do_num1:
    if(zero_first || nonzero_first ||
        BOOST_JSON_LIKELY(cs))
    {
        char const c = *cs;
        if(zero_first)
        {
            ++cs;
            num.mant = 0;
            goto do_num6;
        }
        else if(nonzero_first || BOOST_JSON_LIKELY(
            c >= '1' && c <= '9'))
        {
            ++cs;
            num.mant = c - '0';
        }
        else if(BOOST_JSON_UNLIKELY(
            c == '0'))
        {
            ++cs;
            num.mant = 0;
            goto do_num6;
        }
        else if( (negative || num.neg) && opt_.allow_infinity_and_nan )
        {
            st_.push(state::lit1);
            cur_lit_ = literal_index(detail::literals::neg_infinity);
            lit_offset_ = 1;
            return parse_literal(
                cs.begin(), detail::literals_c<detail::literals::resume>() );
        }
        else
        {
            BOOST_STATIC_CONSTEXPR source_location loc
                = BOOST_CURRENT_LOCATION;
            return fail(cs.begin(), error::syntax, &loc);
        }
    }
    else
    {
        if(BOOST_JSON_UNLIKELY(
            ! h_.on_number_part(
                {begin, cs.used(begin)}, ec_)))
            return fail(cs.begin());

        BOOST_IF_CONSTEXPR( precise_parsing )
            num_buf_.append( begin, cs.used(begin) );
        return maybe_suspend(
            cs.begin(), state::num1, num);
    }

    //----------------------------------
    //
    // 1*DIGIT
    // significant digits left of decimal
    //
do_num2:
    if(negative || (!stack_empty && num.neg))
    {
        for(;;)
        {
            if(BOOST_JSON_UNLIKELY(! cs))
            {
                if(BOOST_JSON_UNLIKELY(more_))
                {
                    if(BOOST_JSON_UNLIKELY(
                        ! h_.on_number_part(
                            {begin, cs.used(begin)}, ec_)))
                        return fail(cs.begin());

                    BOOST_IF_CONSTEXPR( precise_parsing )
                        num_buf_.append( begin, cs.used(begin) );
                    return suspend(cs.begin(), state::num2, num);
                }
                goto finish_int;
            }
            char const c = *cs;
            if(BOOST_JSON_LIKELY(
                c >= '0' && c <= '9'))
            {
                ++cs;
                //              9223372036854775808 INT64_MIN
                if( num.mant  > 922337203685477580 || (
                    num.mant == 922337203685477580 && c > '8'))
                    break;
                BOOST_IF_CONSTEXPR( !no_parsing )
                    num.mant = 10 * num.mant + ( c - '0' );
                continue;
            }
            goto do_num6; // [.eE]
        }
    }
    else
    {
        for(;;)
        {
            if(BOOST_JSON_UNLIKELY(! cs))
            {
                if(BOOST_JSON_UNLIKELY(more_))
                {
                    if(BOOST_JSON_UNLIKELY(
                        ! h_.on_number_part(
                            {begin, cs.used(begin)}, ec_)))
                        return fail(cs.begin());

                    BOOST_IF_CONSTEXPR( precise_parsing )
                        num_buf_.append( begin, cs.used(begin) );
                    return suspend(cs.begin(), state::num2, num);
                }
                goto finish_int;
            }
            char const c = *cs;
            if(BOOST_JSON_LIKELY(
                c >= '0' && c <= '9'))
            {
                ++cs;
                //              18446744073709551615 UINT64_MAX
                if( num.mant  > 1844674407370955161 || (
                    num.mant == 1844674407370955161 && c > '5'))
                    break;
                BOOST_IF_CONSTEXPR( !no_parsing )
                    num.mant = 10 * num.mant + ( c - '0' );
            }
            else
            {
                goto do_num6; // [.eE]
            }
        }
    }
    ++num.bias;

    //----------------------------------
    //
    // 1*DIGIT
    // non-significant digits left of decimal
    //
do_num3:
    for(;;)
    {
        if(BOOST_JSON_UNLIKELY(! cs))
        {
            if(BOOST_JSON_UNLIKELY(more_))
            {
                if(BOOST_JSON_UNLIKELY(
                    ! h_.on_number_part(
                        {begin, cs.used(begin)}, ec_)))
                    return fail(cs.begin());

                BOOST_IF_CONSTEXPR( precise_parsing )
                    num_buf_.append( begin, cs.used(begin) );
                return suspend(cs.begin(), state::num3, num);
            }
            goto finish_dub;
        }
        char const c = *cs;
        if(BOOST_JSON_UNLIKELY(
            c >= '0' && c <= '9'))
        {
            if(BOOST_JSON_UNLIKELY( num.bias + 1 == INT_MAX ))
            {
                BOOST_STATIC_CONSTEXPR source_location loc
                    = BOOST_CURRENT_LOCATION;
                return fail(cs.begin(), error::exponent_overflow, &loc);
            }
            ++cs;
            ++num.bias;
        }
        else if(BOOST_JSON_LIKELY(
            c == '.'))
        {
            ++cs;
            break;
        }
        else if((c | 32) == 'e')
        {
            ++cs;
            goto do_exp1;
        }
        else
        {
            goto finish_dub;
        }
    }

    //----------------------------------
    //
    // DIGIT
    // first non-significant digit
    // to the right of decimal
    //
do_num4:
    {
        if(BOOST_JSON_UNLIKELY(! cs))
        {
            if(BOOST_JSON_UNLIKELY(
                ! h_.on_number_part(
                    {begin, cs.used(begin)}, ec_)))
                return fail(cs.begin());

            BOOST_IF_CONSTEXPR( precise_parsing )
                num_buf_.append( begin, cs.used(begin) );
            return maybe_suspend(
                cs.begin(), state::num4, num);
        }
        char const c = *cs;
        if(BOOST_JSON_LIKELY(
            //static_cast<unsigned char>(c - '0') < 10))
            c >= '0' && c <= '9'))
        {
            ++cs;
        }
        else
        {
            // digit required
            BOOST_STATIC_CONSTEXPR source_location loc
                = BOOST_CURRENT_LOCATION;
            return fail(cs.begin(), error::syntax, &loc);
        }
    }

    //----------------------------------
    //
    // 1*DIGIT
    // non-significant digits
    // to the right of decimal
    //
do_num5:
    for(;;)
    {
        if(BOOST_JSON_UNLIKELY(! cs))
        {
            if(BOOST_JSON_UNLIKELY(more_))
            {
                if(BOOST_JSON_UNLIKELY(
                    ! h_.on_number_part(
                        {begin, cs.used(begin)}, ec_)))
                    return fail(cs.begin());

                BOOST_IF_CONSTEXPR( precise_parsing )
                    num_buf_.append( begin, cs.used(begin) );
                return suspend(cs.begin(), state::num5, num);
            }
            goto finish_dub;
        }
        char const c = *cs;
        if(BOOST_JSON_LIKELY(
            c >= '0' && c <= '9'))
        {
            ++cs;
        }
        else if((c | 32) == 'e')
        {
            ++cs;
            goto do_exp1;
        }
        else
        {
            goto finish_dub;
        }
    }

    //----------------------------------
    //
    // [.eE]
    //
do_num6:
    {
        if(BOOST_JSON_UNLIKELY(! cs))
        {
            if(BOOST_JSON_UNLIKELY(more_))
            {
                if(BOOST_JSON_UNLIKELY(
                    ! h_.on_number_part(
                        {begin, cs.used(begin)}, ec_)))
                    return fail(cs.begin());

                BOOST_IF_CONSTEXPR( precise_parsing )
                    num_buf_.append( begin, cs.used(begin) );
                return suspend(cs.begin(), state::num6, num);
            }
            goto finish_int;
        }
        char const c = *cs;
        if(BOOST_JSON_LIKELY(
            c == '.'))
        {
            ++cs;
        }
        else if((c | 32) == 'e')
        {
            ++cs;
            goto do_exp1;
        }
        else
        {
            goto finish_int;
        }
    }

    //----------------------------------
    //
    // DIGIT
    // first significant digit
    // to the right of decimal
    //
do_num7:
    {
        if(BOOST_JSON_UNLIKELY(! cs))
        {
            if(BOOST_JSON_UNLIKELY(more_))
            {
                if(BOOST_JSON_UNLIKELY(
                    ! h_.on_number_part(
                        {begin, cs.used(begin)}, ec_)))
                    return fail(cs.begin());

                BOOST_IF_CONSTEXPR( precise_parsing )
                    num_buf_.append( begin, cs.used(begin) );
                return suspend(cs.begin(), state::num7, num);
            }
            // digit required
            BOOST_STATIC_CONSTEXPR source_location loc
                = BOOST_CURRENT_LOCATION;
            return fail(cs.begin(), error::syntax, &loc);
        }
        char const c = *cs;
        if(BOOST_JSON_UNLIKELY(
            c < '0' || c > '9'))
        {
            // digit required
            BOOST_STATIC_CONSTEXPR source_location loc
                = BOOST_CURRENT_LOCATION;
            return fail(cs.begin(), error::syntax, &loc);
        }
    }

    //----------------------------------
    //
    // 1*DIGIT
    // significant digits
    // to the right of decimal
    //
do_num8:
    for(;;)
    {
        if(BOOST_JSON_UNLIKELY(! cs))
        {
            if(BOOST_JSON_UNLIKELY(more_))
            {
                if(BOOST_JSON_UNLIKELY(
                    ! h_.on_number_part(
                        {begin, cs.used(begin)}, ec_)))
                    return fail(cs.begin());

                BOOST_IF_CONSTEXPR( precise_parsing )
                    num_buf_.append( begin, cs.used(begin) );
                return suspend(cs.begin(), state::num8, num);
            }
            goto finish_dub;
        }
        char const c = *cs;
        if(BOOST_JSON_LIKELY(
            c >= '0' && c <= '9'))
        {
            ++cs;
            if(!no_parsing && BOOST_JSON_LIKELY(
                num.mant <= 9007199254740991)) // 2^53-1
            {
                if(BOOST_JSON_UNLIKELY( num.bias - 1 == INT_MIN ))
                {
                    BOOST_STATIC_CONSTEXPR source_location loc
                        = BOOST_CURRENT_LOCATION;
                    return fail(cs.begin(), error::exponent_overflow, &loc);
                }
                --num.bias;
                num.mant = 10 * num.mant + ( c - '0' );
            }
            else
            {
                goto do_num5;
            }
        }
        else if((c | 32) == 'e')
        {
            ++cs;
            goto do_exp1;
        }
        else
        {
            goto finish_dub;
        }
    }

    //----------------------------------
    //
    // *[+-]
    //
do_exp1:
    if(BOOST_JSON_UNLIKELY(! cs))
    {
        if(BOOST_JSON_UNLIKELY(
            ! h_.on_number_part(
                {begin, cs.used(begin)}, ec_)))
            return fail(cs.begin());

        BOOST_IF_CONSTEXPR( precise_parsing )
            num_buf_.append( begin, cs.used(begin) );
        return maybe_suspend(
            cs.begin(), state::exp1, num);
    }
    if(*cs == '+')
    {
        ++cs;
    }
    else if(*cs == '-')
    {
        ++cs;
        num.frac = true;
    }

    //----------------------------------
    //
    // DIGIT
    // first digit of the exponent
    //
do_exp2:
    {
        if(BOOST_JSON_UNLIKELY(! cs))
        {
            if(BOOST_JSON_UNLIKELY(more_))
            {
                if(BOOST_JSON_UNLIKELY(
                    ! h_.on_number_part(
                        {begin, cs.used(begin)}, ec_)))
                    return fail(cs.begin());

                BOOST_IF_CONSTEXPR( precise_parsing )
                    num_buf_.append( begin, cs.used(begin) );
                return suspend(cs.begin(), state::exp2, num);
            }
            // digit required
            BOOST_STATIC_CONSTEXPR source_location loc
                = BOOST_CURRENT_LOCATION;
            return fail(cs.begin(), error::syntax, &loc);
        }
        char const c = *cs;
        if(BOOST_JSON_UNLIKELY(
            c < '0' || c > '9'))
        {
            // digit required
            BOOST_STATIC_CONSTEXPR source_location loc
                = BOOST_CURRENT_LOCATION;
            return fail(cs.begin(), error::syntax, &loc);
        }
        ++cs;
        num.exp = c - '0';
    }

    //----------------------------------
    //
    // 1*DIGIT
    // subsequent digits in the exponent
    //
do_exp3:
    for(;;)
    {
        if(BOOST_JSON_UNLIKELY(! cs))
        {
            if(BOOST_JSON_UNLIKELY(more_))
            {
                if(BOOST_JSON_UNLIKELY(
                    ! h_.on_number_part(
                        {begin, cs.used(begin)}, ec_)))
                    return fail(cs.begin());

                BOOST_IF_CONSTEXPR( precise_parsing )
                    num_buf_.append( begin, cs.used(begin) );
                return suspend(cs.begin(), state::exp3, num);
            }
        }
        else
        {
            char const c = *cs;
            if(BOOST_JSON_LIKELY( c >= '0' && c <= '9' ))
            {
                if(BOOST_JSON_UNLIKELY(
                //              2147483647 INT_MAX
                    num.exp  >  214748364 ||
                    (num.exp == 214748364 && c > '7')
                ))
                    num.exp = INT_MAX;
                else BOOST_IF_CONSTEXPR( !no_parsing )
                    num.exp = 10 * num.exp + ( c - '0' );

                ++cs;
                continue;
            }
        }
        BOOST_ASSERT(num.exp >= 0);
        if ( num.frac )
        {
            if(BOOST_JSON_UNLIKELY( num.bias < (INT_MIN + num.exp) ))
            {
                // if exponent overflowed, bias is a very large negative
                // number, and mantissa isn't zero, then we cannot parse the
                // number correctly
                if(BOOST_JSON_UNLIKELY(
                    (num.exp == INT_MAX) &&
                    (num.bias < 0) &&
                    (num.exp + num.bias < 308) &&
                    num.mant ))
                {
                    BOOST_STATIC_CONSTEXPR source_location loc
                        = BOOST_CURRENT_LOCATION;
                    return fail(cs.begin(), error::exponent_overflow, &loc);
                }

                num.bias = 0;
                num.exp = INT_MAX;
            }
        }
        else if (BOOST_JSON_UNLIKELY( num.bias > (INT_MAX - num.exp) ))
        {
            // if exponent overflowed, bias is a very large positive number,
            // and mantissa isn't zero, then we cannot parse the
            // number correctly
            if(BOOST_JSON_UNLIKELY(
                (num.exp == INT_MAX) &&
                (num.bias > 0) &&
                (num.exp - num.bias < 308) &&
                num.mant ))
            {
                BOOST_STATIC_CONSTEXPR source_location loc
                    = BOOST_CURRENT_LOCATION;
                return fail(cs.begin(), error::exponent_overflow, &loc);
            }

            num.bias = 0;
            num.exp = INT_MAX;
        }
        goto finish_dub;
    }

finish_int:
    if(negative || (!stack_empty && num.neg))
    {
        if(BOOST_JSON_UNLIKELY(
            ! h_.on_int64(static_cast<
                int64_t>(~num.mant + 1), {begin, cs.used(begin)}, ec_)))
            return fail(cs.begin());
        return cs.begin();
    }
    if(num.mant <= INT64_MAX)
    {
finish_signed:
        if(BOOST_JSON_UNLIKELY(
            ! h_.on_int64(static_cast<
                int64_t>(num.mant), {begin, cs.used(begin)}, ec_)))
            return fail(cs.begin());
        return cs.begin();
    }
    if(BOOST_JSON_UNLIKELY(
        ! h_.on_uint64(num.mant, {begin, cs.used(begin)}, ec_)))
        return fail(cs.begin());
    return cs.begin();
finish_dub:
    double d;
    std::size_t const size = cs.used(begin);
    BOOST_ASSERT( !num_buf_.size() || precise_parsing );
    BOOST_IF_CONSTEXPR( precise_parsing )
    {
        char const* data = begin;
        std::size_t full_size = size;
         // if we previously suspended or if the current input ends with the
         // number, we need to copy the current part of the number to the
         // temporary buffer
        if(BOOST_JSON_UNLIKELY( num_buf_.size() ))
        {
            data = num_buf_.append( begin, size );
            full_size = num_buf_.size();
        }
        auto const err = detail::charconv::from_chars(
            data, data + full_size, d );
        BOOST_ASSERT( err.ec != std::errc::invalid_argument );
        BOOST_ASSERT( err.ptr == data + full_size );
        (void)err;
    }
    else BOOST_IF_CONSTEXPR( no_parsing )
        d = 0;
    else
        d = detail::dec_to_float(
            num.mant,
            num.bias + (num.frac ?
                -num.exp : num.exp),
            num.neg);
    if(BOOST_JSON_UNLIKELY(
        ! h_.on_double(d, {begin, size}, ec_)))
        return fail(cs.begin());
    return cs.begin();
}

//----------------------------------------------------------

template<class Handler>
template<class... Args>
basic_parser<Handler>::
basic_parser(
    parse_options const& opt,
    Args&&... args)
    : h_(std::forward<Args>(args)...)
    , opt_(opt)
{
}

//----------------------------------------------------------

template<class Handler>
void
basic_parser<Handler>::
reset() noexcept
{
    ec_ = {};
    st_.clear();
    more_ = true;
    done_ = false;
    clean_ = true;
    num_buf_.clear();
}

template<class Handler>
void
basic_parser<Handler>::
fail(system::error_code ec) noexcept
{
    if(! ec)
    {
        // assign an arbitrary
        // error code to prevent UB
        BOOST_JSON_FAIL(ec_, error::incomplete);
    }
    else
    {
        ec_ = ec;
    }
    done_ = false;
}

//----------------------------------------------------------

template<class Handler>
std::size_t
basic_parser<Handler>::
write_some(
    bool more,
    char const* data,
    std::size_t size,
    system::error_code& ec)
{
    // see if we exited via exception
    // on the last call to write_some
    if(! clean_)
    {
        // prevent UB
        if(! ec_)
        {
            BOOST_JSON_FAIL(ec_, error::exception);
        }
    }
    if(ec_)
    {
        // error is sticky
        ec = ec_;
        return 0;
    }
    clean_ = false;
    more_ = more;
    end_ = data + size;
    const char* p;
    if(BOOST_JSON_LIKELY(st_.empty()))
    {
        // first time
        depth_ = opt_.max_depth;
        if(BOOST_JSON_UNLIKELY(
            ! h_.on_document_begin(ec_)))
        {
            ec = ec_;
            return 0;
        }
        p = parse_document(data, std::true_type());
    }
    else
    {
        p = parse_document(data, std::false_type());
    }

    if(BOOST_JSON_LIKELY(p != sentinel()))
    {
        BOOST_ASSERT(! ec_);
        if(! done_)
        {
            done_ = true;
            h_.on_document_end(ec_);
        }
    }
    else
    {
        if(! ec_)
        {
            if(! more_)
            {
                BOOST_JSON_FAIL(ec_, error::incomplete);
            }
            else if(! st_.empty())
            {
                // consume as much trailing whitespace in
                // the JSON document as possible, but still
                // consider the parse complete
                state st;
                st_.peek(st);
                if( st == state::doc3 &&
                    ! done_)
                {
                    done_ = true;
                    h_.on_document_end(ec_);
                }
            }
        }
        p = end_;
    }
    ec = ec_;
    clean_ = true;
    return p - data;
}

template<class Handler>
std::size_t
basic_parser<Handler>::
write_some(
    bool more,
    char const* data,
    std::size_t size,
    std::error_code& ec)
{
    system::error_code jec;
    std::size_t const result = write_some(more, data, size, jec);
    ec = jec;
    return result;
}

#endif

} // namespace json
} // namespace boost

#ifdef _MSC_VER
#pragma warning(pop)
#endif

#endif

Coverage Report

Created: 2026-09-13 07:10