Skip to content

Commit effe6d8

Browse files
IhorNehrutsaIhorNehrutsa
committed
py/objint.c: Code review of int_from_bytes().
Support signed param: result = int.int_from_bytes(bytearray(), order='big'|'little', signed=False|True) Signed-off-by: Ihor Nehrutsa <[email protected]>
1 parent 8987b39 commit effe6d8

File tree

5 files changed

+151
-3
lines changed

5 files changed

+151
-3
lines changed

‎ports/esp32/mpconfigport.h

+2-2
Original file line numberDiff line numberDiff line change
@@ -61,8 +61,8 @@
6161
#define MICROPY_ENABLE_GC (1)
6262
#define MICROPY_STACK_CHECK_MARGIN (1024)
6363
#define MICROPY_ENABLE_EMERGENCY_EXCEPTION_BUF (1)
64-
#define MICROPY_LONGINT_IMPL (MICROPY_LONGINT_IMPL_MPZ)
65-
#define MICROPY_ERROR_REPORTING (MICROPY_ERROR_REPORTING_NORMAL)
64+
#define MICROPY_LONGINT_IMPL (MICROPY_LONGINT_IMPL_LONGLONG) // (MICROPY_LONGINT_IMPL_MPZ)
65+
#define MICROPY_ERROR_REPORTING (MICROPY_ERROR_REPORTING_NORMAL+1)
6666
#define MICROPY_WARNINGS (1)
6767
#define MICROPY_FLOAT_IMPL (MICROPY_FLOAT_IMPL_FLOAT)
6868
#define MICROPY_STREAMS_POSIX_API (1)

‎py/objint.c

+100
Original file line numberDiff line numberDiff line change
@@ -39,6 +39,9 @@
3939
#include <math.h>
4040
#endif
4141

42+
#define debug_printf(...) mp_printf(&mp_plat_print, __VA_ARGS__); mp_printf(&mp_plat_print, "\n"); // mp_printf(&mp_plat_print, " | func:%s line:%d at %s\n", __FUNCTION__, __LINE__, __FILE__);
43+
#define _debug_printf(...) mp_printf(&mp_plat_print, __VA_ARGS__);
44+
4245
// This dispatcher function is expected to be independent of the implementation of long int
4346
static mp_obj_t mp_obj_int_make_new(const mp_obj_type_t *type_in, size_t n_args, size_t n_kw, const mp_obj_t *args) {
4447
(void)type_in;
@@ -387,6 +390,7 @@ mp_obj_t mp_obj_int_binary_op_extra_cases(mp_binary_op_t op, mp_obj_t lhs_in, mp
387390
return MP_OBJ_NULL; // op not supported
388391
}
389392

393+
/*
390394
// this is a classmethod
391395
static mp_obj_t int_from_bytes(size_t n_args, const mp_obj_t *args) {
392396
// TODO: Support signed param (assumes signed=False at the moment)
@@ -416,6 +420,102 @@ static mp_obj_t int_from_bytes(size_t n_args, const mp_obj_t *args) {
416420
}
417421
return mp_obj_new_int_from_uint(value);
418422
}
423+
*/
424+
425+
int calc_size(mp_longint_impl_t x) {
426+
int size = 0;
427+
if (x < 0) {
428+
x = -x;
429+
}
430+
do {
431+
x >>= 8;
432+
size++;
433+
} while (x);
434+
return size;
435+
}
436+
437+
void *reverce_memcpy(void *dest, const void *src, size_t len) {
438+
char *d = (char *)dest + len - 1;
439+
const char *s = src;
440+
while (len--) {
441+
*d-- = *s++;
442+
}
443+
return dest;
444+
}
445+
446+
// this is a classmethod
447+
// result = int.int_from_bytes(bytearray(), order='big', signed=False)
448+
static mp_obj_t int_from_bytes(size_t n_args, const mp_obj_t *args) {
449+
mp_buffer_info_t bufinfo;
450+
mp_get_buffer_raise(args[1], &bufinfo, MP_BUFFER_READ);
451+
bool big_endian = (n_args < 3) || (args[2] != MP_OBJ_NEW_QSTR(MP_QSTR_little));
452+
bool signd = (n_args > 3) && mp_obj_is_true(args[3]);
453+
454+
// debug_printf("mp_obj_is_bool(args[3])=%d", mp_obj_is_bool(args[3]));
455+
// debug_printf("n_args=%d, MP_SMALL_INT_MAX=%d, MP_SMALL_INT_MIN=%d", n_args, MP_SMALL_INT_MAX, MP_SMALL_INT_MIN);
456+
#if MICROPY_LONGINT_IMPL != MICROPY_LONGINT_IMPL_NONE
457+
if ((1ULL << (bufinfo.len * 8 - 2)) > (MP_SMALL_INT_MAX + 1)) {
458+
debug_printf("//1 Result will overflow a small-int so construct a big-int");
459+
debug_printf("big_endian:%d signed:%d bufinfo.len:%d", big_endian, signd, bufinfo.len);
460+
return mp_obj_int_from_bytes_impl(big_endian, signd, bufinfo.len, bufinfo.buf);
461+
}
462+
#endif
463+
464+
union {
465+
mp_int_t ival;
466+
mp_uint_t uval;
467+
byte buf[sizeof(mp_uint_t)];
468+
} result = {0};
469+
470+
if (bufinfo.len > sizeof(result)) {
471+
mp_raise_msg(&mp_type_OverflowError, MP_ERROR_TEXT("small-int too large"));
472+
bufinfo.len = sizeof(result);
473+
}
474+
475+
if (big_endian) {
476+
reverce_memcpy(&result.uval, bufinfo.buf, bufinfo.len);
477+
} else { // little endian
478+
memcpy(&result.uval, bufinfo.buf, bufinfo.len);
479+
}
480+
if (signd) {
481+
if ((sizeof(result) > bufinfo.len) && (result.buf[bufinfo.len - 1] & 0x80)) {
482+
// Sign propagation
483+
// x = 2
484+
// x.to_bytes(1, 'little', True) -> b'\x02'
485+
// x.to_bytes(4, 'little', True) -> b'\x02\x00\x00\x00'
486+
// x = -2
487+
// x.to_bytes(1, 'little', True) -> b'\xFE'
488+
// x.to_bytes(4, 'little', True) -> b'\xFE\xFF\xFF\xFF'
489+
debug_printf("result=0x%08X", result.uval);
490+
for (int i = 0; i < sizeof(result); i++) {
491+
_debug_printf("\\%02X", result.buf[i]);
492+
}
493+
memset(result.buf + bufinfo.len, 0xFF, sizeof(result) - bufinfo.len);
494+
debug_printf("\nresult=0x%08X", result.uval);
495+
for (int i = 0; i < sizeof(result); i++) {
496+
_debug_printf("\\%02X", result.buf[i]);
497+
}
498+
debug_printf("");
499+
}
500+
#if MICROPY_LONGINT_IMPL != MICROPY_LONGINT_IMPL_NONE
501+
// debug_printf("big_endian:%d signed:%d bufinfo.len:%d sizeof(result.ival):%d result.ival:%ld", big_endian, signd, bufinfo.len, sizeof(result.ival), result.ival);
502+
if (!MP_SMALL_INT_FITS(result.ival)) {
503+
debug_printf("//2 Result will overflow a small-int so construct a big-int");
504+
return mp_obj_int_from_bytes_impl(big_endian, signd, bufinfo.len, bufinfo.buf);
505+
}
506+
#endif
507+
return mp_obj_new_int(result.ival);
508+
} else {
509+
#if MICROPY_LONGINT_IMPL != MICROPY_LONGINT_IMPL_NONE
510+
// debug_printf("big_endian:%d signed:%d bufinfo.len:%d sizeof(result.uval):%d result.uval:%ld", big_endian, signd, bufinfo.len, sizeof(result.uval), result.uval);
511+
if (!MP_SMALL_UINT_FITS(result.uval)) {
512+
debug_printf("//3 Result will overflow a small-int so construct a big-int");
513+
return mp_obj_int_from_bytes_impl(big_endian, signd, bufinfo.len, bufinfo.buf);
514+
}
515+
#endif
516+
return mp_obj_new_int_from_uint(result.uval);
517+
}
518+
}
419519

420520
static MP_DEFINE_CONST_FUN_OBJ_VAR_BETWEEN(int_from_bytes_fun_obj, 2, 4, int_from_bytes);
421521
static MP_DEFINE_CONST_CLASSMETHOD_OBJ(int_from_bytes_obj, MP_ROM_PTR(&int_from_bytes_fun_obj));

‎py/objint.h

+3-1
Original file line numberDiff line numberDiff line change
@@ -54,13 +54,15 @@ char *mp_obj_int_formatted(char **buf, size_t *buf_size, size_t *fmt_size, mp_co
5454
char *mp_obj_int_formatted_impl(char **buf, size_t *buf_size, size_t *fmt_size, mp_const_obj_t self_in,
5555
int base, const char *prefix, char base_char, char comma);
5656
mp_int_t mp_obj_int_hash(mp_obj_t self_in);
57-
mp_obj_t mp_obj_int_from_bytes_impl(bool big_endian, size_t len, const byte *buf);
57+
mp_obj_t mp_obj_int_from_bytes_impl(bool big_endian, bool signd, size_t len, const byte *buf);
5858
// Returns true if 'self_in' fit into 'len' bytes of 'buf' without overflowing, 'buf' is truncated otherwise.
5959
bool mp_obj_int_to_bytes_impl(mp_obj_t self_in, bool big_endian, size_t len, byte *buf);
6060
int mp_obj_int_sign(mp_obj_t self_in);
6161
mp_obj_t mp_obj_int_unary_op(mp_unary_op_t op, mp_obj_t o_in);
6262
mp_obj_t mp_obj_int_binary_op(mp_binary_op_t op, mp_obj_t lhs_in, mp_obj_t rhs_in);
6363
mp_obj_t mp_obj_int_binary_op_extra_cases(mp_binary_op_t op, mp_obj_t lhs_in, mp_obj_t rhs_in);
6464
mp_obj_t mp_obj_int_pow3(mp_obj_t base, mp_obj_t exponent, mp_obj_t modulus);
65+
void *reverce_memcpy(void *dest, const void *src, size_t len);
66+
int calc_size(mp_longint_impl_t x);
6567

6668
#endif // MICROPY_INCLUDED_PY_OBJINT_H

‎py/objint_longlong.c

+45
Original file line numberDiff line numberDiff line change
@@ -43,6 +43,7 @@
4343
const mp_obj_int_t mp_sys_maxsize_obj = {{&mp_type_int}, MP_SSIZE_MAX};
4444
#endif
4545

46+
/*
4647
mp_obj_t mp_obj_int_from_bytes_impl(bool big_endian, size_t len, const byte *buf) {
4748
int delta = 1;
4849
if (!big_endian) {
@@ -56,6 +57,50 @@ mp_obj_t mp_obj_int_from_bytes_impl(bool big_endian, size_t len, const byte *buf
5657
}
5758
return mp_obj_new_int_from_ll(value);
5859
}
60+
*/
61+
#define debug_printf(...) mp_printf(&mp_plat_print, __VA_ARGS__); mp_printf(&mp_plat_print, "\n"); // mp_printf(&mp_plat_print, " | func:%s line:%d at %s\n", __FUNCTION__, __LINE__, __FILE__);
62+
#define _debug_printf(...) mp_printf(&mp_plat_print, __VA_ARGS__);
63+
64+
mp_obj_t mp_obj_int_from_bytes_impl(bool big_endian, bool signd, size_t len, const byte *buf) {
65+
union {
66+
mp_longint_impl_t val;
67+
byte buf[sizeof(mp_longint_impl_t)];
68+
} result = {0};
69+
70+
if (len > sizeof(result)) {
71+
len = sizeof(result);
72+
mp_raise_msg(&mp_type_OverflowError, MP_ERROR_TEXT("long int overflow"));
73+
}
74+
if (big_endian) {
75+
reverce_memcpy(&result, buf, len);
76+
} else { // little endian
77+
memcpy(&result, buf, len);
78+
}
79+
if (signd) {
80+
if ((sizeof(result) > len) && (result.buf[len - 1] & 0x80)) {
81+
// Sign propagation
82+
// x = 2
83+
// x.to_bytes(1, 'little', True) -> b'\x02'
84+
// x.to_bytes(4, 'little', True) -> b'\x02\x00\x00\x00'
85+
// x = -2
86+
// x.to_bytes(1, 'little', True) -> b'\xFE'
87+
// x.to_bytes(4, 'little', True) -> b'\xFE\xFF\xFF\xFF'
88+
debug_printf("result=0x%08X", result.val);
89+
for(int i = 0; i < sizeof(result); i++) {
90+
_debug_printf("\\%02X", result.buf[i]);
91+
}
92+
memset(result.buf + len, 0xFF, sizeof(result) - len);
93+
debug_printf("\nresult=0x%08X", result.val);
94+
for(int i = 0; i < sizeof(result); i++) {
95+
_debug_printf("\\%02X", result.buf[i]);
96+
}
97+
debug_printf("");
98+
}
99+
return mp_obj_new_int_from_ll(result.val);
100+
} else {
101+
return mp_obj_new_int_from_ull(result.val);
102+
}
103+
}
59104

60105
bool mp_obj_int_to_bytes_impl(mp_obj_t self_in, bool big_endian, size_t len, byte *buf) {
61106
assert(mp_obj_is_exact_type(self_in, &mp_type_int));

‎py/smallint.h

+1
Original file line numberDiff line numberDiff line change
@@ -60,6 +60,7 @@
6060
#endif
6161

6262
#define MP_SMALL_INT_MAX ((mp_int_t)(~(MP_SMALL_INT_MIN)))
63+
#define MP_SMALL_UINT_FITS(n) (((n) & ~MP_SMALL_INT_POSITIVE_MASK) == 0)
6364

6465
// https://stackoverflow.com/a/4589384/1976323
6566
// Number of bits in inttype_MAX, or in any (1<<k)-1 where 0 <= k < 2040

0 commit comments

Comments
 (0)