Click here to Skip to main content
15,881,882 members
Articles / Programming Languages / C++14
Tip/Trick

bits.hpp: A Header File with Some Goodies for Doing Manipulation of Bits and Words

Rate me:
Please Sign up or sign in to vote.
4.53/5 (4 votes)
18 May 2021MIT3 min read 4.4K   92   11   1
Easily shift bits in memory of arbitrary length, declare integer sizes programmatically, endian conversion, and more
This is a header file I use in my GFX library to handle all the "bit twiddling" as well as some basic utilities for dealing with machine words in the form of uint32_t, int64_t, and the like.

Introduction

I'm producing this as much for me as for you, because some of it was difficult to write, and all of it has turned out to be useful to me at points - enough that I wanted to make a semi-permanent record of this code, while at the same time sharing it with you, gentle reader.

The scope of it isn't entirely "clean" because it was evolved and grown as I needed features. It handles a collection of things that are useful taken together, but don't always have a lot to do with each other, like bit shifting vs. the uintx<X> and intx<X> templates. The basic scope is that it handles bit twiddling and machine intrinsic word manipulation which covers the above.

Compiler Compatibility

This was tested with GCC and Clang using -std=C++14 or greater. It probably won't work with Microsoft's compiler, and I've made no effort to that end. Doing so is non-trivial, not so much because I'm using non-standard features or uncommon headers, but because Microsoft.

Using this Mess

First, #define HTCW_LITTLE_ENDIAN or HTCW_BIG_ENDIAN. This defaults to little endian. It indicates the byte order of the target machine.

Next, #define HTCW_MAX_WORD X where X is the number of bits. It defaults to 64 and should be a machine word size. This indicates the maximum word size the machine can handle, not necessarily the maximum it can handle natively. For example, if targeting the ATmega2560 8-bit processor, you'd use 32, since it can handle an integer up to 32 bits in size, even though it can only use native word manipulation instructions up to 16 bits.

Next, include bits.hpp after which you'll want to use the bits namespace.

Now in terms of functionality, first we have endianness() and endian_mode which indicate the byte order of the target platform.

Next up, we have the swap() functions which simply swap byte order. One of the methods modifies the data inline and swaps an arbitrary number of bytes.

Next, we have from_be() and from_le() which convert from big endian to the target machine's byte order or from little endian to the target machine's byte order.

Next, there are three helper methods which shouldn't be needed by you. They compute nearest word sizes and also contain an algorithm for creating compound word structures for an arbitrary number of bytes.

Next, there are some helpers that facilitate the templates intx<>, uintx<>, and realx<>. These templates produce a native word integer or real type that is the minimum of the specified number of bits. intx<6> will resolve to an int8_t, while uintx<9> will resolve to uint16_t. Note that there is also int_max, uint_max and real_max which give you the maximum available size integer or real.

Now we have signedx<T> and unsignedx<T>. These return a signed or unsigned version of the integer type passed in.

Now we have mask<> template, which creates a mask for the specified number of bits, aligned to the left of a word, or to the right of a word, and including not masks for each..

The next group of functions are for bit twiddling. They set and clear bits or shift bits left and right, and can do so even on non-byte boundaries, meaning for example, you can shift a 3 bit block starting 7 bits in in a larger chunk of memory to the left or right.

They typically take an offset and the number of bits. That indicates your "window" for the shift or set operation.

Note that the flexibility of these functions can make them slower than doing shifts and such yourself, but they can handle whole arrays.

Here's the code in its entirety in case you prefer to copy and paste instead of download it:

C++
#ifndef HTCW_BITS_HPP
#define HTCW_BITS_HPP
#define PACKED __attribute__((__packed__))
#define PACK // TODO: Add Microsoft pack pragmas here
#define RESTORE_PACK
#ifndef HTCW_MAX_WORD
#define HTCW_MAX_WORD 64
#endif
#include <stddef.h>
#include <stdint.h>
#include <string.h>
namespace bits {
    enum struct endian_mode {
        none = 0,
        big_endian = 1,
        little_endian = 2
    };
    // true if the platform is little endian
    constexpr static endian_mode endianness()
    {
#ifdef HTCW_BIG_ENDIAN
        return endian_mode::big_endian;
#elif defined(HTCW_LITTLE_ENDIAN) || defined(ESP_PLATFORM)
        return endian_mode::little_endian;
#else
        return endian_mode::none;
#endif
        
    }
    // swaps byte order
    constexpr static inline uint16_t swap(uint16_t value)
    {
        return (value >> 8) | (value << 8);
    }
    // swaps byte order
    constexpr static inline uint32_t swap(uint32_t value)
    {
        uint32_t tmp = ((value << 8) & 0xFF00FF00) | ((value >> 8) & 0xFF00FF);
        return (tmp << 16) | (tmp >> 16);
    }
#if HTCW_MAX_WORD >= 64
    // swaps byte order
    constexpr static inline uint64_t swap(uint64_t value) {
        value = (value & 0x00000000FFFFFFFF) << 32 | (value & 0xFFFFFFFF00000000) >> 32;
        value = (value & 0x0000FFFF0000FFFF) << 16 | (value & 0xFFFF0000FFFF0000) >> 16;
        value = (value & 0x00FF00FF00FF00FF) << 8  | (value & 0xFF00FF00FF00FF00) >> 8;
        return value;
    }
#endif

    // swaps byte order (no-op to resolve ambiguous overload)
    constexpr static inline uint8_t swap(uint8_t value) {
        return value;
    }
    template<size_t SizeBytes>
    constexpr static inline void swap_inline(void* data) {
        switch(SizeBytes) {
            case 0:
            case 1:
                return;
            case 2:
                *((uint16_t*)data)=swap(*(uint16_t*)data);
                return;
            case 4:
                *((uint32_t*)data)=swap(*(uint32_t*)data);
                return;
#if HTCW_MAX_WORD >=64            
            case 8:
                *((uint64_t*)data)=swap(*(uint64_t*)data);
                return;
#endif
        }
        for (size_t low = 0, high = SizeBytes - 1; low < high; low++, high--) {
            size_t tmp = ((uint8_t*)data)[low];
            ((uint8_t*)data)[low] = ((uint8_t*)data)[high];
            ((uint8_t*)data)[high] = tmp;
        }
    }
    constexpr static inline uint16_t from_le(uint16_t value) {
        if(endianness()==endian_mode::big_endian)
            return swap(value);
        return value;
    }
    constexpr static inline uint32_t from_le(uint32_t value) {
        if(endianness()==endian_mode::big_endian)
            return swap(value);
        return value;
    }
#if HTCW_MAX_WORD >=64
    constexpr static inline uint64_t from_le(uint64_t value) {
        if(endianness()==endian_mode::big_endian)
            return swap(value);
        return value;
    }
#endif
    constexpr static inline uint8_t from_le(uint8_t value) {
        return value;
    }

    constexpr static inline uint16_t from_be(uint16_t value) {
        if(endianness()==endian_mode::little_endian)
            return swap(value);
        return value;
    }
    constexpr static inline uint32_t from_be(uint32_t value) {
        if(endianness()==endian_mode::little_endian)
            return swap(value);
        return value;
    }
#if HTCW_MAX_WORD >=64
    constexpr static inline uint64_t from_be(uint64_t value) {
        if(endianness()==endian_mode::little_endian)
            return swap(value);
        return value;
    }
#endif
    constexpr static inline uint8_t from_be(uint8_t value) {
        return value;
    }
    constexpr size_t get_word_size(size_t size) {
#if HTCW_MAX_WORD >= 64
        if(size>64) return 0; 
        if(size>32) return 64;
#else 
        if(size>32) return 0;
#endif
        
        if(size>16) return 32;
        if(size>8) return 16;
        return 8;
    }

    constexpr size_t get_word_count(size_t size) {
    size_t result = 0;
        while(size>0) {
            size_t ws = get_word_size(size);
            if(ws==0)
            break;
            size-=ws;
            ++result;
        }
        return result;
    }
    constexpr size_t get_left_word(size_t size,size_t index) {
        size_t i = 0;
        while(size>0 && i<=index) {
            size_t ws = get_word_size(size);
            if(ws==0)
            return 0;
            if(index==i)
                return ws;
            size-=ws;
            ++i;
        }
        return 0;
    }
    constexpr size_t get_right_word(size_t size,size_t index) {
        size_t c = get_word_count(size);
        return get_left_word(size,c-index-1);
        size_t i = 0;
        while(size>0 && i<=index) {
            size_t ws = get_word_size(size);
            if(ws==0)
            return 0;
            if(index==i)
                return ws;
            size-=ws;
            ++i;
        }
        return 0;
    }
    template <size_t BitWidth> class int_helper {};
    template <> struct int_helper<8> { using type = int8_t; };
    template <> struct int_helper<16> { using type = int16_t; };
    template <> struct int_helper<32> { using type = int32_t; };
#if HTCW_MAX_WORD >=64
    template <> struct int_helper<64> { using type = int64_t; };
    template<size_t Width> using intx = typename int_helper<Width>::type;
    using int_max = typename int_helper<HTCW_MAX_WORD>::type;
#endif
    template <size_t BitWidth> class uint_helper {};
    template <> struct uint_helper<8> { using type = uint8_t; };
    template <> struct uint_helper<16> { using type = uint16_t; };
    template <> struct uint_helper<32> { using type = uint32_t; };
#if HTCW_MAX_WORD >=64
    template <> struct uint_helper<64> { using type = uint64_t; };
#endif
    template<size_t Width> using uintx = typename uint_helper<Width>::type;
    using uint_max = uint_helper<HTCW_MAX_WORD>::type;
    template <size_t BitWidth> class real_helper {};
    template <> struct real_helper<32> { using type = float; };
#if HTCW_MAX_WORD >=64
    template <> struct real_helper<64> { using type = double; };
#endif
    template<size_t Width> using realx = typename real_helper<Width>::type;
    using real_max = typename real_helper<HTCW_MAX_WORD>::type;
    template <typename T> struct signed_helper {};
    template <> struct signed_helper<float> { using type = float; };
    template <> struct signed_helper<double> { using type = double; };
    template <> struct signed_helper<int8_t> { using type = int8_t; };
    template <> struct signed_helper<uint8_t> { using type = int8_t; };
    template <> struct signed_helper<int16_t> { using type = int16_t; };
    template <> struct signed_helper<uint16_t> { using type = int16_t; };
    template <> struct signed_helper<int32_t> { using type = int32_t; };
    template <> struct signed_helper<uint32_t> { using type = int32_t; };
#if HTCW_MAX_WORD >=64
    template <> struct signed_helper<int64_t> { using type = int64_t; };
    template <> struct signed_helper<uint64_t> { using type = int64_t; };
#endif
    template<typename T> using signedx = typename signed_helper<T>::type;

    template <typename T> struct unsigned_helper {};
    template <> struct unsigned_helper<float> { using type = float; };
    template <> struct unsigned_helper<double> { using type = double; };
    template <> struct unsigned_helper<int8_t> { using type = uint8_t; };
    template <> struct unsigned_helper<uint8_t> { using type = uint8_t; };
    template <> struct unsigned_helper<int16_t> { using type = uint16_t; };
    template <> struct unsigned_helper<uint16_t> { using type = uint16_t; };
    template <> struct unsigned_helper<int32_t> { using type = uint32_t; };
    template <> struct unsigned_helper<uint32_t> { using type = uint32_t; };
#if HTCW_MAX_WORD >=64
    template <> struct unsigned_helper<int64_t> { using type = uint64_t; };
    template <> struct unsigned_helper<uint64_t> { using type = uint64_t; };
#endif
    template<typename T> using unsignedx = typename unsigned_helper<T>::type;
    
    template<size_t Width>
    struct mask {
        using int_type = uintx<get_word_size(Width)>;
    private:
        constexpr static const int_type int_mask = ~int_type(0);
    public:
        constexpr static const int_type left = int_mask<<((sizeof(int_type)*8-Width)&int_mask);
        constexpr static const int_type right = int_mask>>(sizeof(int_type)*8-Width);
        constexpr static const int_type not_left = ~left;
        constexpr static const int_type not_right = ~right;
    };

    constexpr inline static void set_bits
        (void* bits,size_t offset_bits,size_t size_bits,bool value) {
        const size_t offset_bytes = offset_bits / 8;
        const size_t offset = offset_bits % 8;
        const size_t total_size_bytes = (offset_bits+size_bits+7)/8;
        const size_t overhang = (offset_bits+size_bits) % 8;
        uint8_t* pbegin = ((uint8_t*)bits)+offset_bytes;
        uint8_t* pend = ((uint8_t*)bits)+total_size_bytes;
        uint8_t* plast = pend-(pbegin!=pend);

        const uint8_t maskL = 0!=offset?
                                    (uint8_t)((uint8_t(0xFF>>offset))):
                                    uint8_t(0xff);
        const uint8_t maskR = 0!=overhang?
                                (uint8_t)~((uint8_t(0xFF>>overhang))):
                                uint8_t(0xFF);
        if(value) {
            if(pbegin==plast) {
                const uint8_t mask = maskL & maskR;
                *pbegin|=mask;
                return;
            }
            *pbegin|=maskL;
            *plast|=maskR;
            if(pbegin+1<plast) {
                const size_t len = (plast)-(pbegin+1);
                if(0!=len && len<=total_size_bytes)
                    memset(pbegin+1,0xFF,len);
            }
        } else {
            if(pbegin==plast) {
                const uint8_t mask = maskL & maskR;
                *pbegin&=~mask;
                return;
            }
            *pbegin&=~maskL;
            *plast&=~maskR;
            
            if(pbegin+1<plast) {
                const size_t len = (plast)-(pbegin+1);
                if(0!=len && len<=total_size_bytes)
                    memset(pbegin+1,0,len);
            }        
        }
    }

    template<size_t OffsetBits,size_t SizeBits,bool Value> 
        constexpr inline static void set_bits(void* bits) {
        if(0==SizeBits)
            return;
        const size_t offset_bytes = OffsetBits / 8;
        const size_t offset = OffsetBits % 8;
        const size_t total_size_bytes = (OffsetBits+SizeBits+7)/8;
        const size_t overhang = (OffsetBits+SizeBits) % 8;
        uint8_t* pbegin = ((uint8_t*)bits)+offset_bytes;
        uint8_t* pend = ((uint8_t*)bits)+total_size_bytes;
        uint8_t* plast = pend-(pbegin!=pend);

        const uint8_t maskL = 0!=offset?
                                    (uint8_t)((uint8_t(0xFF>>offset))):
                                    uint8_t(0xff);
        const uint8_t maskR = 0!=overhang?
                                (uint8_t)~((uint8_t(0xFF>>overhang))):
                                uint8_t(0xFF);
        if(Value) {
            if(pbegin==plast) {
                const uint8_t mask = maskL & maskR;
                *pbegin|=mask;
                return;
            }
            *pbegin|=maskL;
            *plast|=maskR;
            if(pbegin+1<plast) {
                const size_t len = (plast)-(pbegin+1);
                if(0!=len && len<=total_size_bytes)
                    memset(pbegin+1,0xFF,len);
            }
        } else {
            if(pbegin==plast) {
                const uint8_t mask = maskL & maskR;
                *pbegin&=~mask;
                return;
            }
            *pbegin&=~maskL;
            *plast&=~maskR;
            
            if(pbegin+1<plast) {
                const size_t len = (plast)-(pbegin+1);
                if(0!=len && len<=total_size_bytes)
                    memset(pbegin+1,0,len);
            }        
        }
    }
constexpr inline static void set_bits
    (size_t offset_bits,size_t size_bits,void* dst,const void* src) {
    const size_t offset_bytes = offset_bits / 8;
    const size_t offset = offset_bits % 8;
    const size_t total_size_bytes = (offset_bits+size_bits+7)/8;
    const size_t overhang = (offset_bits+size_bits) % 8;
    uint8_t* pbegin = ((uint8_t*)dst)+offset_bytes;
    uint8_t* psbegin = ((uint8_t*)src)+offset_bytes;
    uint8_t* pend = ((uint8_t*)dst)+total_size_bytes;
    uint8_t* plast = pend-(pbegin!=pend);

    const uint8_t maskL = 0!=offset?
                                (uint8_t)((uint8_t(0xFF>>offset))):
                                uint8_t(0xff);
    const uint8_t maskR = 0!=overhang?
                            (uint8_t)~((uint8_t(0xFF>>overhang))):
                            uint8_t(0xFF);
    if(pbegin==plast) {
        uint8_t v = *psbegin;
        const uint8_t mask = maskL & maskR;
        v&=mask;
        *pbegin&=~mask;
        *pbegin|=v;
        return;
    }
    *pbegin&=~maskL;
    *pbegin|=((*psbegin)&maskL);
    *plast&=~maskR;
    *plast|=((*(psbegin+total_size_bytes-1))&maskR);
    if(pbegin+1<plast) {
        const size_t len = plast-(pbegin+1);
        if(0!=len&&len<=total_size_bytes) 
            memcpy(pbegin+1,psbegin+1,len);
    }
}
    template<size_t OffsetBits,size_t SizeBits> 
        constexpr inline static void set_bits(void* dst,const void* src) {
        const size_t offset_bytes = OffsetBits / 8;
        const size_t offset = OffsetBits % 8;
        const size_t total_size_bytes = (OffsetBits+SizeBits+7)/8;
        const size_t overhang = (OffsetBits+SizeBits) % 8;
        uint8_t* pbegin = ((uint8_t*)dst)+offset_bytes;
        uint8_t* psbegin = ((uint8_t*)src)+offset_bytes;
        uint8_t* pend = ((uint8_t*)dst)+total_size_bytes;
        uint8_t* plast = pend-(pbegin!=pend);
        uint8_t* pslast =psbegin+total_size_bytes-1;
        const uint8_t maskL = 0!=offset?
                                    (uint8_t)((uint8_t(0xFF>>offset))):
                                    uint8_t(0xff);
        const uint8_t maskR = 0!=overhang?
                                (uint8_t)~((uint8_t(0xFF>>overhang))):
                                uint8_t(0xFF);
        if(pbegin==plast) {
            uint8_t v = *psbegin;
            const uint8_t mask = maskL & maskR;
            v&=mask;
            *pbegin=(*pbegin&uint8_t(~mask))|v;
            return;
        }
        *pbegin=(*pbegin&~maskL)|((*psbegin)&maskL);
        
        *plast=(*plast&~maskR)|((*pslast)&maskR);
        if(pbegin+1<plast) {
            const size_t len = plast-(pbegin+1);
            if(0!=len&&len<=total_size_bytes) 
                memcpy(pbegin+1,psbegin+1,len);
        }
    }
    
    template<size_t OffsetBits,size_t SizeBits, size_t Shift> 
        constexpr inline static void shift_left(void* bits) {
        if(nullptr==bits || 0==SizeBits || 0==Shift) {
            return;
        }
        // special case if we shift all the bits out
        if(Shift>=SizeBits) {
            set_bits<OffsetBits,SizeBits,false>(bits);
            return;
        }
        uint8_t* pbegin = ((uint8_t*)bits)+(OffsetBits/8);
        const size_t offset = OffsetBits % 8;
        const size_t shift_bytes = Shift / 8;
        const size_t shift_bits = Shift % 8;
        const size_t overhang = (SizeBits+OffsetBits) % 8;
        // preserves left prior to offset
        const uint8_t left_mask = ((uint8_t)uint8_t(0xFF<<(8-offset)));
        // preserves right after overhang
        const uint8_t right_mask = 0!=overhang?uint8_t(0xFF>>overhang):0;
        uint8_t* pend = pbegin+(size_t)((OffsetBits+SizeBits+7)/8);
        uint8_t* plast = pend-1;
        uint8_t* psrc = pbegin+shift_bytes;
        uint8_t* pdst = pbegin;
        if(pbegin+1==pend) {
            // special case for a shift all within one byte
            uint8_t save_mask = left_mask|right_mask;
            uint8_t tmp = *pbegin;
            *pbegin = uint8_t(uint8_t(tmp<<shift_bits)&~save_mask)|
                    uint8_t(tmp&save_mask);
            return;
        }
        // preserve the ends so we can
        // fix them up later
        uint8_t left = *pbegin;
        uint8_t right = *(pend-1);
        
        while(pdst!=pend) {
            uint8_t src = psrc<pend?*psrc:0;
            uint8_t src2 = (psrc+1)<pend?*(psrc+1):0;
            *pdst = (src<<shift_bits)|(src2>>(8-shift_bits));
            ++psrc;
            ++pdst;
        }
        
        *pbegin=(left&left_mask)|uint8_t(*pbegin&~left_mask);
        --pend;
        *plast=uint8_t(right&right_mask)|uint8_t(*plast&uint8_t(~right_mask));
    };

constexpr static void shift_left
    (void* bits,size_t offset_bits,size_t size_bits, size_t shift) {
    if(nullptr==bits || 0==size_bits || 0==shift) {
        return;
    }
    // special case if we shift all the bits out
    if(shift>=size_bits) {
        set_bits(bits,offset_bits,size_bits,false);
        return;
    }
    uint8_t* pbegin = ((uint8_t*)bits)+(offset_bits/8);
    const size_t offset = offset_bits % 8;
    const size_t shift_bytes = shift / 8;
    const size_t shift_bits = shift % 8;
    const size_t overhang = (size_bits+offset_bits) % 8;
    // preserves left prior to offset
    const uint8_t left_mask = ((uint8_t)uint8_t(0xFF<<(8-offset)));
    // preserves right after overhang
    const uint8_t right_mask = 0!=overhang?uint8_t(0xFF>>overhang):0;
    uint8_t* pend = pbegin+(size_t)((offset_bits+size_bits+7)/8);
    uint8_t* plast = pend-1;
    uint8_t* psrc = pbegin+shift_bytes;
    uint8_t* pdst = pbegin;
    if(pbegin+1==pend) {
        // special case for a shift all within one byte
        uint8_t save_mask = left_mask|right_mask;
        uint8_t tmp = *pbegin;
        *pbegin = uint8_t(uint8_t(tmp<<shift_bits)&~save_mask)|
                uint8_t(tmp&save_mask);
        return;
    }
    // preserve the ends so we can
    // fix them up later
    uint8_t left = *pbegin;
    uint8_t right = *(pend-1);
    
    while(pdst!=pend) {
        uint8_t src = psrc<pend?*psrc:0;
        uint8_t src2 = (psrc+1)<pend?*(psrc+1):0;
        *pdst = (src<<shift_bits)|(src2>>(8-shift_bits));
        ++psrc;
        ++pdst;
    }
    
    *pbegin=(left&left_mask)|uint8_t(*pbegin&~left_mask);
    --pend;
    *plast=uint8_t(right&right_mask)|uint8_t(*plast&uint8_t(~right_mask));
};

    template<size_t OffsetBits,size_t SizeBits, size_t Shift> 
        constexpr inline static void shift_right(void* bits) {
        if(nullptr==bits || 0==SizeBits || 0==Shift) {
            return;
        }
        // special case if we shift all the bits out
        if(Shift>=SizeBits) {
            set_bits<OffsetBits,SizeBits,false>(bits);
            return;
        }
        uint8_t* pbegin = ((uint8_t*)bits)+(OffsetBits/8);
        const size_t offset = OffsetBits % 8;
        const size_t shift_bytes = Shift / 8;
        const size_t shift_bits = Shift % 8;
        const size_t overhang = (SizeBits+OffsetBits) % 8;
        // preserves left prior to offset
        const uint8_t left_mask = ((uint8_t)uint8_t(0xFF<<(8-offset)));
        // preserves right after overhang
        const uint8_t right_mask = 0!=overhang?uint8_t(0xFF>>overhang):0;
        uint8_t* pend = pbegin+(size_t)((OffsetBits+SizeBits+7)/8)-(OffsetBits/8);
        uint8_t* plast = pend-1;
        uint8_t* psrc = (pend-1)-shift_bytes;
        uint8_t* pdst = pend-1;
        if(pbegin+1==pend) {
            // special case for a shift all within one byte
            uint8_t save_mask = left_mask|right_mask;
            uint8_t tmp = *pbegin;
            *pbegin = uint8_t(uint8_t(tmp>>shift_bits)&~save_mask)|
                    uint8_t(tmp&save_mask);
            return;
        }
        // preserve the ends so we can
        // fix them up later
        uint8_t left = *pbegin;
        uint8_t right = *psrc;
        
        while(pdst>=pbegin) {
            uint8_t src = psrc>=pbegin?*psrc:0;
            uint8_t src2 = (psrc-1)>=pbegin?*(psrc-1):0;
            *pdst = (src>>shift_bits)|(src2<<(8-shift_bits));
            --psrc;
            --pdst;
        }
        
        *pbegin=uint8_t(left&left_mask)|uint8_t(*pbegin&~left_mask);
        *plast=uint8_t(right&right_mask)|uint8_t(*plast&uint8_t(~right_mask));
    };
    constexpr static void shift_right
        (void* bits,size_t offset_bits,size_t size_bits, size_t shift) {
        if(nullptr==bits || 0==size_bits || 0==shift) {
            return;
        }
        // special case if we shift all the bits out
        if(shift>=size_bits) {
            set_bits(bits,offset_bits,size_bits,false);
            return;
        }
        uint8_t* pbegin = ((uint8_t*)bits)+(offset_bits/8);
        const size_t offset = offset_bits % 8;
        const size_t shift_bytes = shift / 8;
        const size_t shift_bits = shift % 8;
        const size_t overhang = (size_bits+offset_bits) % 8;
        // preserves left prior to offset
        const uint8_t left_mask = ((uint8_t)uint8_t(0xFF<<(8-offset)));
        // preserves right after overhang
        const uint8_t right_mask = 0!=overhang?uint8_t(0xFF>>overhang):0;
        uint8_t* pend = pbegin+(size_t)((offset_bits+size_bits+7)/8)-(offset_bits/8);
        uint8_t* plast = pend-1;
        uint8_t* psrc = (pend-1)-shift_bytes;
        uint8_t* pdst = pend-1;
        if(pbegin+1==pend) {
            // special case for a shift all within one byte
            uint8_t save_mask = left_mask|right_mask;
            uint8_t tmp = *pbegin;
            *pbegin = uint8_t(uint8_t(tmp>>shift_bits)&~save_mask)|
                    uint8_t(tmp&save_mask);
            return;
        }
        // preserve the ends so we can
        // fix them up later
        uint8_t left = *pbegin;
        uint8_t right = *psrc;
        
        while(pdst>=pbegin) {
            uint8_t src = psrc>=pbegin?*psrc:0;
            uint8_t src2 = (psrc-1)>=pbegin?*(psrc-1):0;
            *pdst = (src>>shift_bits)|(src2<<(8-shift_bits));
            --psrc;
            --pdst;
        }
        
        *pbegin=uint8_t(left&left_mask)|uint8_t(*pbegin&~left_mask);
        *plast=uint8_t(right&right_mask)|uint8_t(*plast&uint8_t(~right_mask));
    };
}
#endif

History

  • 18th May, 2021 - Initial submission

License

This article, along with any associated source code and files, is licensed under The MIT License


Written By
United States United States
Just a shiny lil monster. Casts spells in C++. Mostly harmless.

Comments and Discussions

 
GeneralMy vote of 5 Pin
Vincent Radio21-May-21 4:28
professionalVincent Radio21-May-21 4:28 

General General    News News    Suggestion Suggestion    Question Question    Bug Bug    Answer Answer    Joke Joke    Praise Praise    Rant Rant    Admin Admin   

Use Ctrl+Left/Right to switch messages, Ctrl+Up/Down to switch threads, Ctrl+Shift+Left/Right to switch pages.