Zxc200611/pocketpy
1
0
Fork 0
mirror of https://github.com/pocketpy/pocketpy synced 2025-10-20 03:20:18 +00:00
Code Issues Packages Projects Releases Wiki Activity

some fix

Browse Source
This commit is contained in:
blueloveTH 2024-07-02 15:01:04 +08:00
parent ab0f07bbd7
commit 8319cb2ad4
7 changed files with 305 additions and 124 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
include/pocketpy/common/str.h
View File

@ -29,8 +29,10 @@ typedef struct c11_bytes{
} c11_bytes;
int c11_sv__cmp(c11_sv self, c11_sv other);
int c11_sv__cmp2(c11_sv self, const char* other, int size);
int c11_sv__cmp3(c11_sv self, const char* other);
int c11_sv__cmp2(c11_sv self, const char* other);
bool c11__streq(const char* a, const char* b);
bool c11__sveq(c11_sv a, const char* b);
c11_string* c11_string__new(const char* data);
c11_string* c11_string__new2(const char* data, int size);
@ -66,6 +68,14 @@ int c11__byte_index_to_unicode(const char* data, int n);
bool c11__is_unicode_Lo_char(int c);
int c11__u8_header(unsigned char c, bool suppress);
typedef enum IntParsingResult{
IntParsing_SUCCESS,
IntParsing_FAILURE,
IntParsing_OVERFLOW,
} IntParsingResult;
IntParsingResult c11__parse_uint(c11_sv text, int64_t* out, int base);
#ifdef __cplusplus
}
#endif

8
include/pocketpy/compiler/lexer.h
View File

@ -87,14 +87,6 @@ enum Precedence {
PREC_HIGHEST,
};
typedef enum IntParsingResult{
IntParsing_SUCCESS,
IntParsing_FAILURE,
IntParsing_OVERFLOW,
} IntParsingResult;
IntParsingResult parse_uint(c11_sv text, int64_t* out, int base);
typedef struct Error Error;
typedef c11_array pk_TokenArray;

4
include/pocketpy/pocketpy.h
View File

@ -311,6 +311,10 @@ py_GlobalRef py_tpobject(py_Type type);
/// Get the type name.
const char* py_tpname(py_Type type);
/// Check if the object is an instance of the given type.
/// Re
bool py_checktype(const py_Ref self, py_Type type);
/// Python favored string formatting.
/// %d: int
/// %i: py_i64 (int64_t)

120
src/common/str.c
View File

@ -8,7 +8,6 @@
#include <ctype.h>
#include <stdio.h>
c11_string* c11_string__new(const char* data) { return c11_string__new2(data, strlen(data)); }
c11_string* c11_string__new2(const char* data, int size) {
@ -35,13 +34,9 @@ c11_string* c11_string__copy(c11_string* self) {
return retval;
}
void c11_string__delete(c11_string* self) {
free(self);
}
void c11_string__delete(c11_string* self) { free(self); }
c11_sv c11_string__sv(c11_string* self) {
return (c11_sv){self->data, self->size};
}
c11_sv c11_string__sv(c11_string* self) { return (c11_sv){self->data, self->size}; }
c11_string* c11_string__replace(c11_string* self, char old, char new_) {
c11_string* retval = c11_string__copy(self);
@ -119,9 +114,7 @@ void c11_sv__upper(c11_sv sv, c11_vector* buf) {
}
}
c11_sv c11_sv__slice(c11_sv sv, int start) {
return c11_sv__slice2(sv, start, sv.size);
}
c11_sv c11_sv__slice(c11_sv sv, int start) { return c11_sv__slice2(sv, start, sv.size); }
c11_sv c11_sv__slice2(c11_sv sv, int start, int stop) {
if(stop < start) stop = start;
@ -233,19 +226,25 @@ int c11__byte_index_to_unicode(const char* data, int n) {
//////////////
int c11_sv__cmp(c11_sv self, c11_sv other) {
return c11_sv__cmp2(self, other.data, other.size);
int res = strncmp(self.data, other.data, PK_MIN(self.size, other.size));
if(res != 0) return res;
return self.size - other.size;
}
int c11_sv__cmp2(c11_sv self, const char* other, int size) {
int c11_sv__cmp2(c11_sv self, const char* other) {
int size = strlen(other);
int res = strncmp(self.data, other, PK_MIN(self.size, size));
if(res != 0) return res;
return self.size - size;
}
int c11_sv__cmp3(c11_sv self, const char* other) {
return c11_sv__cmp2(self, other, strlen(other));
}
bool c11__streq(const char* a, const char* b) { return strcmp(a, b) == 0; }
bool c11__sveq(c11_sv a, const char* b) {
int size = strlen(b);
if(a.size != size) return false;
return memcmp(a.data, b, size) == 0;
}
// clang-format off
static const int kLoRangeA[] = {170,186,443,448,660,1488,1519,1568,1601,1646,1649,1749,1774,1786,1791,1808,1810,1869,1969,1994,2048,2112,2144,2208,2230,2308,2365,2384,2392,2418,2437,2447,2451,2474,2482,2486,2493,2510,2524,2527,2544,2556,2565,2575,2579,2602,2610,2613,2616,2649,2654,2674,2693,2703,2707,2730,2738,2741,2749,2768,2784,2809,2821,2831,2835,2858,2866,2869,2877,2908,2911,2929,2947,2949,2958,2962,2969,2972,2974,2979,2984,2990,3024,3077,3086,3090,3114,3133,3160,3168,3200,3205,3214,3218,3242,3253,3261,3294,3296,3313,3333,3342,3346,3389,3406,3412,3423,3450,3461,3482,3507,3517,3520,3585,3634,3648,3713,3716,3718,3724,3749,3751,3762,3773,3776,3804,3840,3904,3913,3976,4096,4159,4176,4186,4193,4197,4206,4213,4238,4352,4682,4688,4696,4698,4704,4746,4752,4786,4792,4800,4802,4808,4824,4882,4888,4992,5121,5743,5761,5792,5873,5888,5902,5920,5952,5984,5998,6016,6108,6176,6212,6272,6279,6314,6320,6400,6480,6512,6528,6576,6656,6688,6917,6981,7043,7086,7098,7168,7245,7258,7401,7406,7413,7418,8501,11568,11648,11680,11688,11696,11704,11712,11720,11728,11736,12294,12348,12353,12447,12449,12543,12549,12593,12704,12784,13312,19968,40960,40982,42192,42240,42512,42538,42606,42656,42895,42999,43003,43011,43015,43020,43072,43138,43250,43259,43261,43274,43312,43360,43396,43488,43495,43514,43520,43584,43588,43616,43633,43642,43646,43697,43701,43705,43712,43714,43739,43744,43762,43777,43785,43793,43808,43816,43968,44032,55216,55243,63744,64112,64285,64287,64298,64312,64318,64320,64323,64326,64467,64848,64914,65008,65136,65142,65382,65393,65440,65474,65482,65490,65498,65536,65549,65576,65596,65599,65616,65664,66176,66208,66304,66349,66370,66384,66432,66464,66504,66640,66816,66864,67072,67392,67424,67584,67592,67594,67639,67644,67647,67680,67712,67808,67828,67840,67872,67968,68030,68096,68112,68117,68121,68192,68224,68288,68297,68352,68416,68448,68480,68608,68864,69376,69415,69424,69600,69635,69763,69840,69891,69956,69968,70006,70019,70081,70106,70108,70144,70163,70272,70280,70282,70287,70303,70320,70405,70415,70419,70442,70450,70453,70461,70480,70493,70656,70727,70751,70784,70852,70855,71040,71128,71168,71236,71296,71352,71424,71680,71935,72096,72106,72161,72163,72192,72203,72250,72272,72284,72349,72384,72704,72714,72768,72818,72960,72968,72971,73030,73056,73063,73066,73112,73440,73728,74880,77824,82944,92160,92736,92880,92928,93027,93053,93952,94032,94208,100352,110592,110928,110948,110960,113664,113776,113792,113808,123136,123214,123584,124928,126464,126469,126497,126500,126503,126505,126516,126521,126523,126530,126535,126537,126539,126541,126545,126548,126551,126553,126555,126557,126559,126561,126564,126567,126572,126580,126585,126590,126592,126603,126625,126629,126635,131072,173824,177984,178208,183984,194560};
@ -274,3 +273,94 @@ int c11__u8_header(unsigned char c, bool suppress) {
if(!suppress) PK_FATAL_ERROR("invalid utf8 char\n")
return 0;
}
IntParsingResult c11__parse_uint(c11_sv text, int64_t* out, int base) {
*out = 0;
c11_sv prefix = {.data = text.data, .size = PK_MIN(2, text.size)};
if(base == -1) {
if(c11__sveq(prefix, "0b"))
base = 2;
else if(c11__sveq(prefix, "0o"))
base = 8;
else if(c11__sveq(prefix, "0x"))
base = 16;
else
base = 10;
}
if(base == 10) {
// 10-base 12334
if(text.size == 0) return IntParsing_FAILURE;
for(int i = 0; i < text.size; i++) {
char c = text.data[i];
if(c >= '0' && c <= '9') {
*out = (*out * 10) + (c - '0');
} else {
return IntParsing_FAILURE;
}
}
// "9223372036854775807".__len__() == 19
if(text.size > 19) return IntParsing_OVERFLOW;
return IntParsing_SUCCESS;
} else if(base == 2) {
// 2-base 0b101010
if(c11__sveq(prefix, "0b")) {
// text.remove_prefix(2);
text = (c11_sv){text.data + 2, text.size - 2};
}
if(text.size == 0) return IntParsing_FAILURE;
for(int i = 0; i < text.size; i++) {
char c = text.data[i];
if(c == '0' || c == '1') {
*out = (*out << 1) | (c - '0');
} else {
return IntParsing_FAILURE;
}
}
// "111111111111111111111111111111111111111111111111111111111111111".__len__() == 63
if(text.size > 63) return IntParsing_OVERFLOW;
return IntParsing_SUCCESS;
} else if(base == 8) {
// 8-base 0o123
if(c11__sveq(prefix, "0o")) {
// text.remove_prefix(2);
text = (c11_sv){text.data + 2, text.size - 2};
}
if(text.size == 0) return IntParsing_FAILURE;
for(int i = 0; i < text.size; i++) {
char c = text.data[i];
if(c >= '0' && c <= '7') {
*out = (*out << 3) | (c - '0');
} else {
return IntParsing_FAILURE;
}
}
// "777777777777777777777".__len__() == 21
if(text.size > 21) return IntParsing_OVERFLOW;
return IntParsing_SUCCESS;
} else if(base == 16) {
// 16-base 0x123
if(c11__sveq(prefix, "0x")) {
// text.remove_prefix(2);
text = (c11_sv){text.data + 2, text.size - 2};
}
if(text.size == 0) return IntParsing_FAILURE;
for(int i = 0; i < text.size; i++) {
char c = text.data[i];
if(c >= '0' && c <= '9') {
*out = (*out << 4) | (c - '0');
} else if(c >= 'a' && c <= 'f') {
*out = (*out << 4) | (c - 'a' + 10);
} else if(c >= 'A' && c <= 'F') {
*out = (*out << 4) | (c - 'A' + 10);
} else {
return IntParsing_FAILURE;
}
}
// "7fffffffffffffff".__len__() == 16
if(text.size > 16) return IntParsing_OVERFLOW;
return IntParsing_SUCCESS;
}
return IntParsing_FAILURE;
}

105
src/compiler/lexer.c
View File

@ -246,11 +246,11 @@ static Error* eat_name(pk_Lexer* self){
c11_sv name = {self->token_start, length};
if(self->src->mode == JSON_MODE) {
if(c11_sv__cmp3(name, "true") == 0) {
if(c11__sveq(name, "true")) {
add_token(self, TK_TRUE);
} else if(c11_sv__cmp3(name, "false") == 0) {
} else if(c11__sveq(name, "false")) {
add_token(self, TK_FALSE);
} else if(c11_sv__cmp3(name, "null") == 0) {
} else if(c11__sveq(name, "null")) {
add_token(self, TK_NONE);
} else {
return SyntaxError("invalid JSON token");
@ -260,12 +260,12 @@ static Error* eat_name(pk_Lexer* self){
const char** KW_BEGIN = pk_TokenSymbols + TK_FALSE;
int KW_COUNT = TK__COUNT__ - TK_FALSE;
#define less(a, b) (c11_sv__cmp3(b, a) > 0)
#define less(a, b) (c11_sv__cmp2(b, a) > 0)
int out;
c11__lower_bound(const char*, KW_BEGIN, KW_COUNT, name, less, &out);
#undef less
if(out != KW_COUNT && c11_sv__cmp3(name, KW_BEGIN[out]) == 0) {
if(out != KW_COUNT && c11__sveq(name, KW_BEGIN[out])) {
add_token(self, (TokenIndex)(out + TK_FALSE));
} else {
add_token(self, TK_ID);
@ -372,7 +372,7 @@ static Error* eat_number(pk_Lexer* self){
}
// try integer
TokenValue value = {.index = TokenValue_I64};
switch(parse_uint(text, &value._i64, -1)) {
switch(c11__parse_uint(text, &value._i64, -1)) {
case IntParsing_SUCCESS:
add_token_with_value(self, TK_NUM, value);
return NULL;
@ -554,7 +554,7 @@ static Error* from_precompiled(pk_Lexer* self) {
deserializer.curr += 5; // skip "pkpy:"
c11_sv version = TokenDeserializer__read_string(&deserializer, '\n');
if(c11_sv__cmp3(version, PK_VERSION) != 0) {
if(c11_sv__cmp2(version, PK_VERSION) != 0) {
return SyntaxError("precompiled version mismatch");
}
if(TokenDeserializer__read_uint(&deserializer, '\n') != (int64_t)self->src->mode){
@ -618,97 +618,6 @@ static Error* from_precompiled(pk_Lexer* self) {
return NULL;
}
IntParsingResult parse_uint(c11_sv text, int64_t* out, int base) {
*out = 0;
c11_sv prefix = {.data = text.data, .size = PK_MIN(2, text.size)};
if(base == -1) {
if(c11_sv__cmp3(prefix, "0b") == 0)
base = 2;
else if(c11_sv__cmp3(prefix, "0o") == 0)
base = 8;
else if(c11_sv__cmp3(prefix, "0x") == 0)
base = 16;
else
base = 10;
}
if(base == 10) {
// 10-base 12334
if(text.size == 0) return IntParsing_FAILURE;
for(int i = 0; i < text.size; i++) {
char c = text.data[i];
if(c >= '0' && c <= '9') {
*out = (*out * 10) + (c - '0');
} else {
return IntParsing_FAILURE;
}
}
// "9223372036854775807".__len__() == 19
if(text.size > 19) return IntParsing_OVERFLOW;
return IntParsing_SUCCESS;
} else if(base == 2) {
// 2-base 0b101010
if(c11_sv__cmp3(prefix, "0b") == 0) {
// text.remove_prefix(2);
text = (c11_sv){text.data + 2, text.size - 2};
}
if(text.size == 0) return IntParsing_FAILURE;
for(int i = 0; i < text.size; i++) {
char c = text.data[i];
if(c == '0' || c == '1') {
*out = (*out << 1) | (c - '0');
} else {
return IntParsing_FAILURE;
}
}
// "111111111111111111111111111111111111111111111111111111111111111".__len__() == 63
if(text.size > 63) return IntParsing_OVERFLOW;
return IntParsing_SUCCESS;
} else if(base == 8) {
// 8-base 0o123
if(c11_sv__cmp3(prefix, "0o") == 0) {
// text.remove_prefix(2);
text = (c11_sv){text.data + 2, text.size - 2};
}
if(text.size == 0) return IntParsing_FAILURE;
for(int i = 0; i < text.size; i++) {
char c = text.data[i];
if(c >= '0' && c <= '7') {
*out = (*out << 3) | (c - '0');
} else {
return IntParsing_FAILURE;
}
}
// "777777777777777777777".__len__() == 21
if(text.size > 21) return IntParsing_OVERFLOW;
return IntParsing_SUCCESS;
} else if(base == 16) {
// 16-base 0x123
if(c11_sv__cmp3(prefix, "0x") == 0) {
// text.remove_prefix(2);
text = (c11_sv){text.data + 2, text.size - 2};
}
if(text.size == 0) return IntParsing_FAILURE;
for(int i = 0; i < text.size; i++) {
char c = text.data[i];
if(c >= '0' && c <= '9') {
*out = (*out << 4) | (c - '0');
} else if(c >= 'a' && c <= 'f') {
*out = (*out << 4) | (c - 'a' + 10);
} else if(c >= 'A' && c <= 'F') {
*out = (*out << 4) | (c - 'A' + 10);
} else {
return IntParsing_FAILURE;
}
}
// "7fffffffffffffff".__len__() == 16
if(text.size > 16) return IntParsing_OVERFLOW;
return IntParsing_SUCCESS;
}
return IntParsing_FAILURE;
}
Error* pk_Lexer__process(pk_SourceData_ src, pk_TokenArray* out_tokens){
pk_Lexer lexer;
pk_Lexer__ctor(&lexer, src);

170
src/interpreter/py_number.c
View File

@ -1,4 +1,5 @@
#include "pocketpy/interpreter/vm.h"
#include "pocketpy/pocketpy.h"
#include <math.h>
@ -43,6 +44,8 @@ DEF_NUM_BINARY_OP(__ge__, >=, py_newbool, py_newbool)
#undef DEF_NUM_BINARY_OP
static bool ValueError(const char* fmt, ...) { return false; }
static bool _py_int__neg__(int argc, py_Ref argv) {
py_checkargc(1);
int64_t val = py_toint(&argv[0]);
@ -182,6 +185,161 @@ DEF_INT_BITWISE_OP(__rshift__, >>)
#undef DEF_INT_BITWISE_OP
static bool _py_int__repr__(int argc, py_Ref argv) {
py_checkargc(1);
int64_t val = py_toint(&argv[0]);
char buf[32];
int size = snprintf(buf, sizeof(buf), "%lld", (long long)val);
py_newstrn(py_retval(), buf, size);
return true;
}
static bool _py_float__repr__(int argc, py_Ref argv) {
py_checkargc(1);
double val = py_tofloat(&argv[0]);
char buf[32];
int size = snprintf(buf, sizeof(buf), "%f", val);
py_newstrn(py_retval(), buf, size);
return true;
}
union c11_8bytes {
py_i64 _i64;
py_f64 _f64;
union {
uint32_t upper;
uint32_t lower;
} bits;
};
static py_i64 c11_8bytes__hash(union c11_8bytes u) {
// https://stackoverflow.com/questions/664014/what-integer-hash-function-are-good-that-accepts-an-integer-hash-key
const uint32_t C = 2654435761;
u.bits.upper *= C;
u.bits.lower *= C;
return u._i64;
}
static bool _py_int__hash__(int argc, py_Ref argv) {
py_checkargc(1);
int64_t val = py_toint(&argv[0]);
union c11_8bytes u = {._i64 = val};
py_newint(py_retval(), c11_8bytes__hash(u));
return true;
}
static bool _py_float__hash__(int argc, py_Ref argv) {
py_checkargc(1);
double val = py_tofloat(&argv[0]);
union c11_8bytes u = {._f64 = val};
py_newint(py_retval(), c11_8bytes__hash(u));
return true;
}
static bool _py_int__new__(int argc, py_Ref argv) {
if(argc == 1 + 0) {
// int() == 0
py_newint(py_retval(), 0);
return true;
}
// 1 arg
if(argc == 1 + 1) {
switch(argv[1].type) {
case tp_float: {
// int(1.1) == 1
py_newint(py_retval(), (int64_t)py_tofloat(&argv[1]));
return true;
}
case tp_int: {
// int(1) == 1
*py_retval() = argv[1];
return true;
}
case tp_bool: {
// int(True) == 1
py_newint(py_retval(), (int64_t)py_tobool(&argv[1]));
return true;
}
case tp_str: break; // leave to the next block
default: return TypeError("invalid arguments for int()");
}
}
// 2+ args -> error
if(argc > 1 + 2) return TypeError("int() takes at most 2 arguments");
// 1 or 2 args with str
int base = 10;
if(argc == 1 + 2) {
if(!py_checktype(py_arg(2), tp_int)) return false;
base = py_toint(py_arg(2));
}
if(!py_checktype(py_arg(1), tp_str)) return false;
int size;
const char* data = py_tostrn(py_arg(1), &size);
bool negative = false;
if(size && (data[0] == '+' || data[0] == '-')) {
negative = data[0] == '-';
data++;
size--;
}
py_i64 val;
if(c11__parse_uint((c11_sv){data, size}, &val, base) != IntParsing_SUCCESS) {
return ValueError("invalid literal for int() with base %d: %q", base, data);
}
py_newint(py_retval(), negative ? -val : val);
return true;
}
static bool _py_float__new__(int argc, py_Ref argv) {
if(argc == 1 + 0) {
// float() == 0.0
py_newfloat(py_retval(), 0.0);
return true;
}
if(argc > 1 + 1) return TypeError("float() takes at most 1 argument");
// 1 arg
switch(argv[1].type) {
case tp_int: {
// float(1) == 1.0
py_newfloat(py_retval(), py_toint(&argv[1]));
return true;
}
case tp_float: {
// float(1.1) == 1.1
*py_retval() = argv[1];
return true;
}
case tp_bool: {
// float(True) == 1.0
py_newfloat(py_retval(), py_tobool(&argv[1]));
return true;
}
case tp_str: break; // leave to the next block
default: return TypeError("invalid arguments for float()");
}
// str to float
int size;
const char* data = py_tostrn(py_arg(1), &size);
if(c11__streq(data, "inf")){
py_newfloat(py_retval(), INFINITY);
return true;
}
if(c11__streq(data, "-inf")){
py_newfloat(py_retval(), -INFINITY);
return true;
}
char* p_end;
py_f64 float_out = strtod(data, &p_end);
if(p_end != data + size){
return ValueError("invalid literal for float(): %q", data);
}
py_newfloat(py_retval(), float_out);
return true;
}
void pk_VM__init_builtins(pk_VM* self) {
/****** tp_int & tp_float ******/
py_bindmagic(tp_int, __add__, _py_int__add__);
@ -208,7 +366,17 @@ void pk_VM__init_builtins(pk_VM* self) {
py_bindmagic(tp_int, __neg__, _py_int__neg__);
py_bindmagic(tp_float, __neg__, _py_float__neg__);
// TODO: __repr__, __new__, __hash__
// __repr__
py_bindmagic(tp_int, __repr__, _py_int__repr__);
py_bindmagic(tp_float, __repr__, _py_float__repr__);
// __hash__
py_bindmagic(tp_int, __hash__, _py_int__hash__);
py_bindmagic(tp_float, __hash__, _py_float__hash__);
// __new__
py_bindmagic(tp_int, __new__, _py_int__new__);
py_bindmagic(tp_float, __new__, _py_float__new__);
// __truediv__
py_bindmagic(tp_int, __truediv__, _py_int__truediv__);

8
src/public/cast.c
View File

@ -40,3 +40,11 @@ void* py_touserdata(const py_Ref self) {
}
bool py_istype(const py_Ref self, py_Type type) { return self->type == type; }
bool py_checktype(const py_Ref self, py_Type type) {
if(self->type != type) {
// py_raise(PyExc_TypeError, "expected %s, got %s", py_typename(type), py_typename(self->type));
return false;
}
return true;
}