py/binary.c

/*
 * This file is part of the MicroPython project, http://micropython.org/
 *
 * The MIT License (MIT)
 *
 * Copyright (c) 2014-2017 Paul Sokolovsky
 * Copyright (c) 2014-2019 Damien P. George
 *
 * Permission is hereby granted, free of charge, to any person obtaining a copy
 * of this software and associated documentation files (the "Software"), to deal
 * in the Software without restriction, including without limitation the rights
 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 * copies of the Software, and to permit persons to whom the Software is
 * furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in
 * all copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
 * THE SOFTWARE.
 */

#include <stdint.h>
#include <stdlib.h>
#include <stddef.h>
#include <string.h>
#include <assert.h>

#include "py/binary.h"
#include "py/smallint.h"
#include "py/objint.h"
#include "py/runtime.h"

// Helpers to work with binary-encoded data

#ifndef alignof
#define alignof(type) offsetof(struct { char c; type t; }, t)
#endif

size_t mp_binary_get_size(char struct_type, char val_type, size_t *palign) {
    size_t size = 0;
    int align = 1;
    switch (struct_type) {
        case '<':
        case '>':
            switch (val_type) {
                case 'b':
                case 'B':
                    size = 1;
                    break;
                case 'h':
                case 'H':
                    size = 2;
                    break;
                case 'i':
                case 'I':
                    size = 4;
                    break;
                case 'l':
                case 'L':
                    size = 4;
                    break;
                case 'q':
                case 'Q':
                    size = 8;
                    break;
                case 'P':
                case 'O':
                case 'S':
                    size = sizeof(void *);
                    break;
                case 'e':
                    size = 2;
                    break;
                case 'f':
                    size = 4;
                    break;
                case 'd':
                    size = 8;
                    break;
            }
            break;
        case '@': {
            // TODO:
            // The simplest heuristic for alignment is to align by value
            // size, but that doesn't work for "bigger than int" types,
            // for example, long long may very well have long alignment
            // So, we introduce separate alignment handling, but having
            // formal support for that is different from actually supporting
            // particular (or any) ABI.
            switch (val_type) {
                case BYTEARRAY_TYPECODE:
                case 'b':
                case 'B':
                    align = size = 1;
                    break;
                case 'h':
                case 'H':
                    align = alignof(short);
                    size = sizeof(short);
                    break;
                case 'i':
                case 'I':
                    align = alignof(int);
                    size = sizeof(int);
                    break;
                case 'l':
                case 'L':
                    align = alignof(long);
                    size = sizeof(long);
                    break;
                case 'q':
                case 'Q':
                    align = alignof(long long);
                    size = sizeof(long long);
                    break;
                case 'P':
                case 'O':
                case 'S':
                    align = alignof(void *);
                    size = sizeof(void *);
                    break;
                case 'e':
                    align = 2;
                    size = 2;
                    break;
                case 'f':
                    align = alignof(float);
                    size = sizeof(float);
                    break;
                case 'd':
                    align = alignof(double);
                    size = sizeof(double);
                    break;
            }
        }
    }

    if (size == 0) {
        mp_raise_ValueError(MP_ERROR_TEXT("bad typecode"));
    }

    if (palign != NULL) {
        *palign = align;
    }
    return size;
}

#if MICROPY_PY_BUILTINS_FLOAT && MICROPY_FLOAT_USE_NATIVE_FLT16

static inline float mp_decode_half_float(uint16_t hf) {
    union {
        uint16_t i;
        _Float16 f;
    } fpu = { .i = hf };
    return fpu.f;
}

static inline uint16_t mp_encode_half_float(float x) {
    union {
        uint16_t i;
        _Float16 f;
    } fp_sp = { .f = (_Float16)x };
    return fp_sp.i;
}

#elif MICROPY_PY_BUILTINS_FLOAT

static float mp_decode_half_float(uint16_t hf) {
    union {
        uint32_t i;
        float f;
    } fpu;

    uint16_t m = hf & 0x3ff;
    int e = (hf >> 10) & 0x1f;
    if (e == 0x1f) {
        // Half-float is infinity.
        e = 0xff;
    } else if (e) {
        // Half-float is normal.
        e += 127 - 15;
    } else if (m) {
        // Half-float is subnormal, make it normal.
        e = 127 - 15;
        while (!(m & 0x400)) {
            m <<= 1;
            --e;
        }
        m -= 0x400;
        ++e;
    }

    fpu.i = ((hf & 0x8000) << 16) | (e << 23) | (m << 13);
    return fpu.f;
}

static uint16_t mp_encode_half_float(float x) {
    union {
        uint32_t i;
        float f;
    } fpu = { .f = x };

    uint16_t m = (fpu.i >> 13) & 0x3ff;
    if (fpu.i & (1 << 12)) {
        // Round up.
        ++m;
    }
    int e = (fpu.i >> 23) & 0xff;

    if (e == 0xff) {
        // Infinity.
        e = 0x1f;
    } else if (e != 0) {
        e -= 127 - 15;
        if (e < 0) {
            // Underflow: denormalized, or zero.
            if (e >= -11) {
                m = (m | 0x400) >> -e;
                if (m & 1) {
                    m = (m >> 1) + 1;
                } else {
                    m >>= 1;
                }
            } else {
                m = 0;
            }
            e = 0;
        } else if (e > 0x3f) {
            // Overflow: infinity.
            e = 0x1f;
            m = 0;
        }
    }

    uint16_t bits = ((fpu.i >> 16) & 0x8000) | (e << 10) | m;
    return bits;
}

#endif

mp_obj_t mp_binary_get_val_array(char typecode, void *p, size_t index) {
    mp_int_t val = 0;
    switch (typecode) {
        case 'b':
            val = ((signed char *)p)[index];
            break;
        case BYTEARRAY_TYPECODE:
        case 'B':
            val = ((unsigned char *)p)[index];
            break;
        case 'h':
            val = ((short *)p)[index];
            break;
        case 'H':
            val = ((unsigned short *)p)[index];
            break;
        case 'i':
            return mp_obj_new_int(((int *)p)[index]);
        case 'I':
            return mp_obj_new_int_from_uint(((unsigned int *)p)[index]);
        case 'l':
            return mp_obj_new_int(((long *)p)[index]);
        case 'L':
            return mp_obj_new_int_from_uint(((unsigned long *)p)[index]);
        #if MICROPY_LONGINT_IMPL != MICROPY_LONGINT_IMPL_NONE
        case 'q':
            return mp_obj_new_int_from_ll(((long long *)p)[index]);
        case 'Q':
            return mp_obj_new_int_from_ull(((unsigned long long *)p)[index]);
        #endif
        #if MICROPY_PY_BUILTINS_FLOAT
        case 'f':
            return mp_obj_new_float_from_f(((float *)p)[index]);
        case 'd':
            return mp_obj_new_float_from_d(((double *)p)[index]);
        #endif
        // Extension to CPython: array of objects
        case 'O':
            return ((mp_obj_t *)p)[index];
        // Extension to CPython: array of pointers
        case 'P':
            return mp_obj_new_int((mp_int_t)(uintptr_t)((void **)p)[index]);
    }
    return MP_OBJ_NEW_SMALL_INT(val);
}

// The long long type is guaranteed to hold at least 64 bits, and size is at
// most 8 (for q and Q), so we will always be able to parse the given data
// and fit it into a long long.
long long mp_binary_get_int(size_t size, bool is_signed, bool big_endian, const byte *src) {
    int delta;
    if (!big_endian) {
        delta = -1;
        src += size - 1;
    } else {
        delta = 1;
    }

    unsigned long long val = 0;
    if (is_signed && *src & 0x80) {
        val = -1;
    }
    for (uint i = 0; i < size; i++) {
        val <<= 8;
        val |= *src;
        src += delta;
    }

    return val;
}

#define is_signed(typecode) (typecode > 'Z')
mp_obj_t mp_binary_get_val(char struct_type, char val_type, byte *p_base, byte **ptr) {
    byte *p = *ptr;
    size_t align;

    size_t size = mp_binary_get_size(struct_type, val_type, &align);
    if (struct_type == '@') {
        // Align p relative to p_base
        p = p_base + (uintptr_t)MP_ALIGN(p - p_base, align);
        #if MP_ENDIANNESS_LITTLE
        struct_type = '<';
        #else
        struct_type = '>';
        #endif
    }
    *ptr = p + size;

    long long val = mp_binary_get_int(size, is_signed(val_type), (struct_type == '>'), p);

    if (val_type == 'O') {
        return (mp_obj_t)(mp_uint_t)val;
    } else if (val_type == 'S') {
        const char *s_val = (const char *)(uintptr_t)(mp_uint_t)val;
        return mp_obj_new_str_from_cstr(s_val);
    #if MICROPY_PY_BUILTINS_FLOAT
    } else if (val_type == 'e') {
        return mp_obj_new_float_from_f(mp_decode_half_float(val));
    } else if (val_type == 'f') {
        union {
            uint32_t i;
            float f;
        } fpu = {val};
        return mp_obj_new_float_from_f(fpu.f);
    } else if (val_type == 'd') {
        union {
            uint64_t i;
            double f;
        } fpu = {val};
        return mp_obj_new_float_from_d(fpu.f);
    #endif
    } else if (is_signed(val_type)) {
        if ((long long)MP_SMALL_INT_MIN <= val && val <= (long long)MP_SMALL_INT_MAX) {
            return mp_obj_new_int((mp_int_t)val);
        } else {
            return mp_obj_new_int_from_ll(val);
        }
    } else {
        if ((unsigned long long)val <= (unsigned long long)MP_SMALL_INT_MAX) {
            return mp_obj_new_int_from_uint((mp_uint_t)val);
        } else {
            return mp_obj_new_int_from_ull(val);
        }
    }
}

void mp_binary_set_int(size_t val_sz, bool big_endian, byte *dest, mp_uint_t val) {
    if (MP_ENDIANNESS_LITTLE && !big_endian) {
        memcpy(dest, &val, val_sz);
    } else if (MP_ENDIANNESS_BIG && big_endian) {
        // only copy the least-significant val_sz bytes
        memcpy(dest, (byte *)&val + sizeof(mp_uint_t) - val_sz, val_sz);
    } else {
        const byte *src;
        if (MP_ENDIANNESS_LITTLE) {
            src = (const byte *)&val + val_sz;
        } else {
            src = (const byte *)&val + sizeof(mp_uint_t);
        }
        while (val_sz--) {
            *dest++ = *--src;
        }
    }
}

void mp_binary_set_val(char struct_type, char val_type, mp_obj_t val_in, byte *p_base, byte **ptr) {
    byte *p = *ptr;
    size_t align;

    size_t size = mp_binary_get_size(struct_type, val_type, &align);
    if (struct_type == '@') {
        // Align p relative to p_base
        p = p_base + (uintptr_t)MP_ALIGN(p - p_base, align);
        if (MP_ENDIANNESS_LITTLE) {
            struct_type = '<';
        } else {
            struct_type = '>';
        }
    }
    *ptr = p + size;

    mp_uint_t val;
    switch (val_type) {
        case 'O':
            val = (mp_uint_t)val_in;
            break;
        #if MICROPY_PY_BUILTINS_FLOAT
        case 'e':
            val = mp_encode_half_float(mp_obj_get_float_to_f(val_in));
            break;
        case 'f': {
            union {
                uint32_t i;
                float f;
            } fp_sp;
            fp_sp.f = mp_obj_get_float_to_f(val_in);
            val = fp_sp.i;
            break;
        }
        case 'd': {
            union {
                uint64_t i64;
                uint32_t i32[2];
                double f;
            } fp_dp;
            fp_dp.f = mp_obj_get_float_to_d(val_in);
            if (MP_BYTES_PER_OBJ_WORD == 8) {
                val = fp_dp.i64;
            } else {
                int be = struct_type == '>';
                mp_binary_set_int(sizeof(uint32_t), be, p, fp_dp.i32[MP_ENDIANNESS_BIG ^ be]);
                p += sizeof(uint32_t);
                val = fp_dp.i32[MP_ENDIANNESS_LITTLE ^ be];
            }
            break;
        }
        #endif
        default:
            #if MICROPY_LONGINT_IMPL != MICROPY_LONGINT_IMPL_NONE
            if (mp_obj_is_exact_type(val_in, &mp_type_int)) {
                mp_obj_int_to_bytes_impl(val_in, struct_type == '>', size, p);
                return;
            }
            #endif

            val = mp_obj_get_int(val_in);
            // zero/sign extend if needed
            if (MP_BYTES_PER_OBJ_WORD < 8 && size > sizeof(val)) {
                int c = (mp_int_t)val < 0 ? 0xff : 0x00;
                memset(p, c, size);
                if (struct_type == '>') {
                    p += size - sizeof(val);
                }
            }
            break;
    }

    mp_binary_set_int(MIN((size_t)size, sizeof(val)), struct_type == '>', p, val);
}

void mp_binary_set_val_array(char typecode, void *p, size_t index, mp_obj_t val_in) {
    switch (typecode) {
        #if MICROPY_PY_BUILTINS_FLOAT
        case 'f':
            ((float *)p)[index] = mp_obj_get_float_to_f(val_in);
            break;
        case 'd':
            ((double *)p)[index] = mp_obj_get_float_to_d(val_in);
            break;
        #endif
        // Extension to CPython: array of objects
        case 'O':
            ((mp_obj_t *)p)[index] = val_in;
            break;
        default:
            #if MICROPY_LONGINT_IMPL != MICROPY_LONGINT_IMPL_NONE
            if (mp_obj_is_exact_type(val_in, &mp_type_int)) {
                size_t size = mp_binary_get_size('@', typecode, NULL);
                mp_obj_int_to_bytes_impl(val_in, MP_ENDIANNESS_BIG,
                    size, (uint8_t *)p + index * size);
                return;
            }
            #endif
            mp_binary_set_val_array_from_int(typecode, p, index, mp_obj_get_int(val_in));
    }
}

void mp_binary_set_val_array_from_int(char typecode, void *p, size_t index, mp_int_t val) {
    switch (typecode) {
        case 'b':
            ((signed char *)p)[index] = val;
            break;
        case BYTEARRAY_TYPECODE:
        case 'B':
            ((unsigned char *)p)[index] = val;
            break;
        case 'h':
            ((short *)p)[index] = val;
            break;
        case 'H':
            ((unsigned short *)p)[index] = val;
            break;
        case 'i':
            ((int *)p)[index] = val;
            break;
        case 'I':
            ((unsigned int *)p)[index] = val;
            break;
        case 'l':
            ((long *)p)[index] = val;
            break;
        case 'L':
            ((unsigned long *)p)[index] = val;
            break;
        #if MICROPY_LONGINT_IMPL != MICROPY_LONGINT_IMPL_NONE
        case 'q':
            ((long long *)p)[index] = val;
            break;
        case 'Q':
            ((unsigned long long *)p)[index] = val;
            break;
        #endif
        #if MICROPY_PY_BUILTINS_FLOAT
        case 'f':
            ((float *)p)[index] = (float)val;
            break;
        case 'd':
            ((double *)p)[index] = (double)val;
            break;
        #endif
        // Extension to CPython: array of pointers
        case 'P':
            ((void **)p)[index] = (void *)(uintptr_t)val;
            break;
    }
}