improve number parsing speed

This commit is contained in:
blueloveTH 2023-10-11 14:37:43 +08:00
parent 332b4100a5
commit b26dbf08cc
5 changed files with 49 additions and 74 deletions

View File

@ -74,18 +74,12 @@ template <>
struct NumberTraits<4> {
using int_t = int32_t;
using float_t = float;
template<typename... Args>
static float_t stof(Args&&... args) { return std::stof(std::forward<Args>(args)...); }
};
template <>
struct NumberTraits<8> {
using int_t = int64_t;
using float_t = double;
template<typename... Args>
static float_t stof(Args&&... args) { return std::stod(std::forward<Args>(args)...); }
};
using Number = NumberTraits<sizeof(void*)>;

View File

@ -138,6 +138,6 @@ struct Lexer {
std::vector<Token> run();
};
bool parse_int(std::string_view text, i64* out, int base=10);
bool parse_int(std::string_view text, i64* out, int base);
} // namespace pkpy

View File

@ -6,6 +6,16 @@ namespace pkpy{
const uint32_t 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};
const uint32_t kLoRangeB[] = {170,186,443,451,660,1514,1522,1599,1610,1647,1747,1749,1775,1788,1791,1808,1839,1957,1969,2026,2069,2136,2154,2228,2237,2361,2365,2384,2401,2432,2444,2448,2472,2480,2482,2489,2493,2510,2525,2529,2545,2556,2570,2576,2600,2608,2611,2614,2617,2652,2654,2676,2701,2705,2728,2736,2739,2745,2749,2768,2785,2809,2828,2832,2856,2864,2867,2873,2877,2909,2913,2929,2947,2954,2960,2965,2970,2972,2975,2980,2986,3001,3024,3084,3088,3112,3129,3133,3162,3169,3200,3212,3216,3240,3251,3257,3261,3294,3297,3314,3340,3344,3386,3389,3406,3414,3425,3455,3478,3505,3515,3517,3526,3632,3635,3653,3714,3716,3722,3747,3749,3760,3763,3773,3780,3807,3840,3911,3948,3980,4138,4159,4181,4189,4193,4198,4208,4225,4238,4680,4685,4694,4696,4701,4744,4749,4784,4789,4798,4800,4805,4822,4880,4885,4954,5007,5740,5759,5786,5866,5880,5900,5905,5937,5969,5996,6000,6067,6108,6210,6264,6276,6312,6314,6389,6430,6509,6516,6571,6601,6678,6740,6963,6987,7072,7087,7141,7203,7247,7287,7404,7411,7414,7418,8504,11623,11670,11686,11694,11702,11710,11718,11726,11734,11742,12294,12348,12438,12447,12538,12543,12591,12686,12730,12799,19893,40943,40980,42124,42231,42507,42527,42539,42606,42725,42895,42999,43009,43013,43018,43042,43123,43187,43255,43259,43262,43301,43334,43388,43442,43492,43503,43518,43560,43586,43595,43631,43638,43642,43695,43697,43702,43709,43712,43714,43740,43754,43762,43782,43790,43798,43814,43822,44002,55203,55238,55291,64109,64217,64285,64296,64310,64316,64318,64321,64324,64433,64829,64911,64967,65019,65140,65276,65391,65437,65470,65479,65487,65495,65500,65547,65574,65594,65597,65613,65629,65786,66204,66256,66335,66368,66377,66421,66461,66499,66511,66717,66855,66915,67382,67413,67431,67589,67592,67637,67640,67644,67669,67702,67742,67826,67829,67861,67897,68023,68031,68096,68115,68119,68149,68220,68252,68295,68324,68405,68437,68466,68497,68680,68899,69404,69415,69445,69622,69687,69807,69864,69926,69956,70002,70006,70066,70084,70106,70108,70161,70187,70278,70280,70285,70301,70312,70366,70412,70416,70440,70448,70451,70457,70461,70480,70497,70708,70730,70751,70831,70853,70855,71086,71131,71215,71236,71338,71352,71450,71723,71935,72103,72144,72161,72163,72192,72242,72250,72272,72329,72349,72440,72712,72750,72768,72847,72966,72969,73008,73030,73061,73064,73097,73112,73458,74649,75075,78894,83526,92728,92766,92909,92975,93047,93071,94026,94032,100343,101106,110878,110930,110951,111355,113770,113788,113800,113817,123180,123214,123627,125124,126467,126495,126498,126500,126503,126514,126519,126521,126523,126530,126535,126537,126539,126543,126546,126548,126551,126553,126555,126557,126559,126562,126564,126570,126578,126583,126588,126590,126601,126619,126627,126633,126651,173782,177972,178205,183969,191456,195101};
std::set<char> kValidChars = {
'0','1','2','3','4','5','6','7','8','9',
// a-f
'a','b','c','d','e','f',
// A-Z
'A','B','C','D','E','F',
// other valid chars
'.', 'L', 'x', 'b', 'o',
};
static bool is_unicode_Lo_char(uint32_t c) {
// open a hole for carrot
if(c == U'🥕') return true;
@ -260,57 +270,39 @@ static bool is_unicode_Lo_char(uint32_t c) {
}
void Lexer::eat_number() {
PK_LOCAL_STATIC const std::regex pattern("^(0[xob])?[0-9a-fA-F]+(\\.[0-9]+)?(L)?");
std::smatch m;
const char* i = token_start;
while(*i != '\n' && *i != '\0') i++;
std::string s = std::string(token_start, i);
while(kValidChars.count(*i)) i++;
std::string_view text(token_start, i - token_start);
curr_char = i;
bool ok = std::regex_search(s, m, pattern);
PK_ASSERT(ok);
// here is m.length()-1, since the first char was eaten by lex_token()
for(int j=0; j<m.length()-1; j++) eatchar();
if(m[3].matched){
if(i[-1] == 'L'){
add_token(TK("@long"));
return;
}
if(m[1].matched && m[2].matched){
SyntaxError("binary/hex/octal literal should not contain a dot");
// try integer
i64 int_out;
bool ok = parse_int(text, &int_out, -1);
if(ok){
add_token(TK("@num"), int_out);
return;
}
// try float
double float_out;
char* p_end;
try{
float_out = std::strtod(text.begin(), &p_end);
}catch(...){
SyntaxError("invalid number literal");
}
if(p_end == text.end()){
add_token(TK("@num"), (f64)float_out);
return;
}
int base = 10;
if (m[1].matched) {
char tag = m[1].first[1];
switch (tag) {
case 'x': base = 16; break;
case 'o': base = 8; break;
case 'b': base = 2; break;
default: FATAL_ERROR();
}
}
if (m[2].matched) {
// float point number
f64 out;
size_t size;
try{
out = Number::stof(m[0], &size);
PK_ASSERT((int)size == (int)m[0].length());
}catch(...){
SyntaxError("invalid number literal");
}
add_token(TK("@num"), out);
} else {
std::string_view text(&m[0].first[0], m[0].length());
i64 out;
bool ok = parse_int(text, &out, base);
if(!ok){
SyntaxError("invalid number literal for base " + std::to_string(base));
}
add_token(TK("@num"), out);
}
SyntaxError("invalid number literal");
}
bool Lexer::lex_one_token() {
@ -490,6 +482,13 @@ bool parse_int(std::string_view text, i64* out, int base){
return t[0] == prefix[0] && t[1] == prefix[1];
};
if(base == -1){
if(f_startswith_2(text, "0b")) base = 2;
else if(f_startswith_2(text, "0o")) base = 8;
else if(f_startswith_2(text, "0x")) base = 16;
else base = 10;
}
if(base == 10){
// 10-base 12334
if(text.length() == 0) return false;

View File

@ -498,12 +498,16 @@ void init_builtins(VM* _vm) {
const Str& s = CAST(Str&, args[1]);
if(s == "inf") return VAR(INFINITY);
if(s == "-inf") return VAR(-INFINITY);
double float_out;
char* p_end;
try{
f64 val = Number::stof(s.str());
return VAR(val);
float_out = std::strtod(s.begin(), &p_end);
PK_ASSERT(p_end == s.end());
}catch(...){
vm->ValueError("invalid literal for float(): " + s.escape());
}
return VAR(float_out);
}
vm->TypeError("invalid arguments for float()");
return vm->None;

View File

@ -346,28 +346,6 @@ assert type(10//11) is int
# test int.__mod__:
assert type(11%2) is int
# /************ float ************/
# 136: 433: _vm->bind_constructor<2>("float", [](VM* vm, ArgsView args) {
# 10: 434: if (is_type(args[1], vm->tp_int)) return VAR((f64)CAST(i64, args[1]));
# 9: 435: if (is_type(args[1], vm->tp_float)) return args[1];
# 3: 436: if (is_type(args[1], vm->tp_bool)) return VAR(_CAST(bool, args[1]) ? 1.0 : 0.0);
# 3: 437: if (is_type(args[1], vm->tp_str)) {
# 3: 438: const Str& s = CAST(Str&, args[1]);
# 3: 439: if(s == "inf") return VAR(INFINITY);
# 2: 440: if(s == "-inf") return VAR(-INFINITY);
# -: 441: try{
# 2: 442: f64 val = Number::stof(s.str());
# 2: 443: return VAR(val);
# #####: 444: }catch(...){
# #####: 445: vm->ValueError("invalid literal for float(): " + s.escape());
# #####: 446: }
# #####: 447: }
# #####: 448: vm->TypeError("float() argument must be a int, float, bool or str");
# #####: 449: return vm->None;
# 10: 450: });
# test float:
try:
float('asad')
print('未能拦截错误, 在测试 float')