CodeDweller/mishmash.hpp

// mishmash.hpp//
// Copyright (C) 2019-2020 MicroNeil Research Corporation.
//
// This software is released under the MIT license. See LICENSE.TXT.
//
// Mishamash is a non-cryptographic hash optimized for short strings.

#pragma once
#include <string>
#include <vector>

namespace codedweller {

namespace masher {

    struct PrimeSet {
        uint32_t primes[256];
        uint32_t select(size_t n) const { return primes[n & 0xff]; }
    };



    const PrimeSet Primes {
        2781111551, 4115761697, 2188512439, 3902941499, 4078728127, 2919449173, 3950407729, 4081733563,
        4061783399, 2728829231, 3557004557, 3778858159, 2397085727, 3808921189, 3159484147, 4084759639,
        2938297727, 3839623429, 2441361773, 2471079053, 4270995293, 3336967261, 2717746523, 3306593077,
        2846319523, 2704420699, 2772002447, 2664263209, 3452941517, 3422898851, 2502691043, 3785866613,
        3164194337, 3348481009, 3448873073, 2340666883, 3742418039, 2298597811, 3061650257, 4286896921,
        2816281231, 3814879819, 3245870963, 2718113173, 3051925081, 2527351147, 2737897993, 3567396421,
        3422709179, 3542178331, 3740214493, 3720784621, 2821606727, 3764267237, 3923068349, 3880303043,
        3816527917, 3032326837, 4006470463, 4165427173, 2511539731, 2993685649, 2268550831, 2899250059,
        2432655553, 3959959639, 4142838449, 4100124029, 2962655827, 2648245423, 4117297801, 3331059481,
        2287716593, 4115533181, 2411037073, 3273978871, 2395052183, 2776949353, 3871801559, 2948305931,
        2257427881, 4288176181, 2624064151, 3144150523, 4282472071, 2206613407, 2960685401, 2761304627,
        3187632967, 2447380627, 2781789473, 3657272401, 3603710699, 4042745507, 4087095841, 2354409557,
        2239000559, 4164370321, 3636939347, 3622573529, 2202257819, 2792364713, 2803311809, 4006899277,
        2835975767, 2951997809, 3185753917, 2417651723, 2631379747, 2175388331, 4277708983, 3547799261,
        3093003779, 3653208029, 4084095199, 3161669417, 2698224923, 2746404013, 3209829191, 3024118079,
        3799471327, 3012753409, 2653321639, 2777700139, 3794895031, 3075822329, 2774150557, 3135465059,
        2294247001, 3557999149, 3189799397, 4080431539, 3397276529, 3931450153, 3518901437, 3135266501,
        3275906693, 4143264221, 3439131923, 2462856037, 2441331799, 2640219781, 3504419027, 3721296989,
        3434920081, 2852097271, 3199425679, 3785517557, 2984788193, 3678788947, 2308371491, 3354456173,
        4148931503, 3239885063, 3894052103, 3930187787, 3521437319, 3203332001, 2894779337, 4197047957,
        3080099353, 2304447583, 3731998477, 2733978179, 2235339121, 2490436339, 2147689987, 3276115639,
        3923204273, 3402325321, 3082785493, 3420863399, 3203562407, 2292225781, 3461192447, 2589852949,
        2772559931, 3667144931, 2163126419, 3412216831, 3699715037, 3619099453, 3875532569, 3047225267,
        3780456941, 2512094297, 3527794547, 3130190737, 3013107347, 3691645577, 3877430399, 4270767517,
        2797247227, 3976455169, 3541991143, 3737016677, 3801675821, 2233908737, 3210598117, 4200595549,
        2772203789, 2415614503, 2467043143, 2611954561, 2791210277, 2599276769, 4034287091, 3521719193,
        3060219497, 2799601709, 2178939557, 3800717749, 3426533269, 3519387097, 2861815447, 2884852603,
        2814915449, 3359572921, 3013763303, 3565679363, 3491538893, 2588490083, 3441048649, 2501343701,
        4191701149, 2251607773, 3543948791, 3511969763, 2966565907, 3640359409, 3640516129, 2586869203,
        2328651011, 4037640461, 3601356797, 4260950119, 3198526459, 3478546597, 3009442531, 3594323353,
        3205141207, 3592422991, 3357971081, 3217249313, 2903534827, 3716714951, 3388800011, 3045269987,
        2325616213, 2688355151, 3022034467, 4221197753, 2426944493, 3090262709, 3471570467, 2437347167
    };
    
    uint64_t engine(
        const unsigned char* buffer, size_t length,
        uint64_t accumulator = 0, 
        const PrimeSet& primes = Primes
    ) noexcept;
}

uint32_t mishmash(const unsigned char* buffer, size_t length) noexcept;
uint32_t mishmash(const std::string &s) noexcept;
uint32_t mishmash(const std::vector<unsigned char>& v) noexcept;

} // End namespace codedweller