pocketpy.h 34 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929
  1. #pragma once
  2. #include "ceval.h"
  3. #include "compiler.h"
  4. #include "repl.h"
  5. #include "iter.h"
  6. #include "cffi.h"
  7. #include "io.h"
  8. #include "_generated.h"
  9. namespace pkpy {
  10. CodeObject_ VM::compile(Str source, Str filename, CompileMode mode) {
  11. Compiler compiler(this, source.c_str(), filename, mode);
  12. try{
  13. return compiler.compile();
  14. }catch(Exception& e){
  15. // std::cout << e.summary() << std::endl;
  16. _error(e);
  17. return nullptr;
  18. }
  19. }
  20. #define BIND_NUM_ARITH_OPT(name, op) \
  21. _vm->_bind_methods<1>({"int","float"}, #name, [](VM* vm, Args& args){ \
  22. if(is_both_int(args[0], args[1])){ \
  23. return VAR(_CAST(i64, args[0]) op _CAST(i64, args[1])); \
  24. }else{ \
  25. return VAR(vm->num_to_float(args[0]) op vm->num_to_float(args[1])); \
  26. } \
  27. });
  28. #define BIND_NUM_LOGICAL_OPT(name, op, is_eq) \
  29. _vm->_bind_methods<1>({"int","float"}, #name, [](VM* vm, Args& args){ \
  30. if(!is_both_int_or_float(args[0], args[1])){ \
  31. if constexpr(is_eq) return VAR(args[0] op args[1]); \
  32. vm->TypeError("unsupported operand type(s) for " #op ); \
  33. } \
  34. if(is_both_int(args[0], args[1])) \
  35. return VAR(_CAST(i64, args[0]) op _CAST(i64, args[1])); \
  36. return VAR(vm->num_to_float(args[0]) op vm->num_to_float(args[1])); \
  37. });
  38. void init_builtins(VM* _vm) {
  39. BIND_NUM_ARITH_OPT(__add__, +)
  40. BIND_NUM_ARITH_OPT(__sub__, -)
  41. BIND_NUM_ARITH_OPT(__mul__, *)
  42. BIND_NUM_LOGICAL_OPT(__lt__, <, false)
  43. BIND_NUM_LOGICAL_OPT(__le__, <=, false)
  44. BIND_NUM_LOGICAL_OPT(__gt__, >, false)
  45. BIND_NUM_LOGICAL_OPT(__ge__, >=, false)
  46. BIND_NUM_LOGICAL_OPT(__eq__, ==, true)
  47. BIND_NUM_LOGICAL_OPT(__ne__, !=, true)
  48. #undef BIND_NUM_ARITH_OPT
  49. #undef BIND_NUM_LOGICAL_OPT
  50. _vm->bind_builtin_func<1>("__sys_stdout_write", [](VM* vm, Args& args) {
  51. (*vm->_stdout) << CAST(Str&, args[0]);
  52. return vm->None;
  53. });
  54. _vm->bind_builtin_func<0>("super", [](VM* vm, Args& args) {
  55. const PyVar* self = vm->top_frame()->f_locals().try_get(m_self);
  56. if(self == nullptr) vm->TypeError("super() can only be called in a class");
  57. return vm->new_object(vm->tp_super, *self);
  58. });
  59. _vm->bind_builtin_func<1>("id", [](VM* vm, Args& args) {
  60. const PyVar& obj = args[0];
  61. if(obj.is_tagged()) return VAR((i64)0);
  62. return VAR(obj.bits);
  63. });
  64. _vm->bind_builtin_func<1>("eval", [](VM* vm, Args& args) {
  65. CodeObject_ code = vm->compile(CAST(Str&, args[0]), "<eval>", EVAL_MODE);
  66. return vm->_exec(code, vm->top_frame()->_module, vm->top_frame()->_locals);
  67. });
  68. _vm->bind_builtin_func<1>("exec", [](VM* vm, Args& args) {
  69. CodeObject_ code = vm->compile(CAST(Str&, args[0]), "<exec>", EXEC_MODE);
  70. vm->_exec(code, vm->top_frame()->_module, vm->top_frame()->_locals);
  71. return vm->None;
  72. });
  73. _vm->bind_builtin_func<-1>("exit", [](VM* vm, Args& args) {
  74. if(args.size() == 0) std::exit(0);
  75. else if(args.size() == 1) std::exit(CAST(int, args[0]));
  76. else vm->TypeError("exit() takes at most 1 argument");
  77. return vm->None;
  78. });
  79. _vm->bind_builtin_func<1>("repr", CPP_LAMBDA(vm->asRepr(args[0])));
  80. _vm->bind_builtin_func<1>("len", CPP_LAMBDA(vm->call(args[0], __len__, no_arg())));
  81. _vm->bind_builtin_func<1>("hash", [](VM* vm, Args& args){
  82. i64 value = vm->hash(args[0]);
  83. if(((value << 2) >> 2) != value) value >>= 2;
  84. return VAR(value);
  85. });
  86. _vm->bind_builtin_func<1>("chr", [](VM* vm, Args& args) {
  87. i64 i = CAST(i64, args[0]);
  88. if (i < 0 || i > 128) vm->ValueError("chr() arg not in range(128)");
  89. return VAR(std::string(1, (char)i));
  90. });
  91. _vm->bind_builtin_func<1>("ord", [](VM* vm, Args& args) {
  92. const Str& s = CAST(Str&, args[0]);
  93. if (s.size() != 1) vm->TypeError("ord() expected an ASCII character");
  94. return VAR((i64)(s.c_str()[0]));
  95. });
  96. _vm->bind_builtin_func<2>("hasattr", [](VM* vm, Args& args) {
  97. return VAR(vm->getattr(args[0], CAST(Str&, args[1]), false) != nullptr);
  98. });
  99. _vm->bind_builtin_func<3>("setattr", [](VM* vm, Args& args) {
  100. vm->setattr(args[0], CAST(Str&, args[1]), args[2]);
  101. return vm->None;
  102. });
  103. _vm->bind_builtin_func<2>("getattr", [](VM* vm, Args& args) {
  104. const Str& name = CAST(Str&, args[1]);
  105. return vm->getattr(args[0], name);
  106. });
  107. _vm->bind_builtin_func<1>("hex", [](VM* vm, Args& args) {
  108. std::stringstream ss;
  109. ss << std::hex << CAST(i64, args[0]);
  110. return VAR("0x" + ss.str());
  111. });
  112. _vm->bind_builtin_func<1>("iter", [](VM* vm, Args& args) {
  113. return vm->asIter(args[0]);
  114. });
  115. _vm->bind_builtin_func<1>("dir", [](VM* vm, Args& args) {
  116. std::set<StrName> names;
  117. if(args[0]->is_attr_valid()){
  118. std::vector<StrName> keys = args[0]->attr().keys();
  119. names.insert(keys.begin(), keys.end());
  120. }
  121. const NameDict& t_attr = vm->_t(args[0])->attr();
  122. std::vector<StrName> keys = t_attr.keys();
  123. names.insert(keys.begin(), keys.end());
  124. List ret;
  125. for (StrName name : names) ret.push_back(VAR(name.str()));
  126. return VAR(std::move(ret));
  127. });
  128. _vm->bind_method<0>("object", "__repr__", [](VM* vm, Args& args) {
  129. PyVar self = args[0];
  130. std::uintptr_t addr = self.is_tagged() ? 0 : (uintptr_t)self.get();
  131. StrStream ss;
  132. ss << std::hex << addr;
  133. Str s = "<" + OBJ_NAME(vm->_t(self)) + " object at 0x" + ss.str() + ">";
  134. return VAR(s);
  135. });
  136. _vm->bind_method<1>("object", "__eq__", CPP_LAMBDA(VAR(args[0] == args[1])));
  137. _vm->bind_method<1>("object", "__ne__", CPP_LAMBDA(VAR(args[0] != args[1])));
  138. _vm->bind_static_method<1>("type", "__new__", CPP_LAMBDA(vm->_t(args[0])));
  139. _vm->bind_method<0>("type", "__repr__", CPP_LAMBDA(VAR("<class '" + OBJ_GET(Str, args[0]->attr(__name__)) + "'>")));
  140. _vm->bind_static_method<-1>("range", "__new__", [](VM* vm, Args& args) {
  141. Range r;
  142. switch (args.size()) {
  143. case 1: r.stop = CAST(i64, args[0]); break;
  144. case 2: r.start = CAST(i64, args[0]); r.stop = CAST(i64, args[1]); break;
  145. case 3: r.start = CAST(i64, args[0]); r.stop = CAST(i64, args[1]); r.step = CAST(i64, args[2]); break;
  146. default: vm->TypeError("expected 1-3 arguments, but got " + std::to_string(args.size()));
  147. }
  148. return VAR(r);
  149. });
  150. _vm->bind_method<0>("range", "__iter__", CPP_LAMBDA(
  151. vm->PyIter(RangeIter(vm, args[0]))
  152. ));
  153. _vm->bind_method<0>("NoneType", "__repr__", CPP_LAMBDA(VAR("None")));
  154. _vm->bind_method<0>("NoneType", "__json__", CPP_LAMBDA(VAR("null")));
  155. _vm->_bind_methods<1>({"int", "float"}, "__truediv__", [](VM* vm, Args& args) {
  156. f64 rhs = vm->num_to_float(args[1]);
  157. if (rhs == 0) vm->ZeroDivisionError();
  158. return VAR(vm->num_to_float(args[0]) / rhs);
  159. });
  160. _vm->_bind_methods<1>({"int", "float"}, "__pow__", [](VM* vm, Args& args) {
  161. if(is_both_int(args[0], args[1])){
  162. i64 lhs = _CAST(i64, args[0]);
  163. i64 rhs = _CAST(i64, args[1]);
  164. bool flag = false;
  165. if(rhs < 0) {flag = true; rhs = -rhs;}
  166. i64 ret = 1;
  167. while(rhs){
  168. if(rhs & 1) ret *= lhs;
  169. lhs *= lhs;
  170. rhs >>= 1;
  171. }
  172. if(flag) return VAR((f64)(1.0 / ret));
  173. return VAR(ret);
  174. }else{
  175. return VAR((f64)std::pow(vm->num_to_float(args[0]), vm->num_to_float(args[1])));
  176. }
  177. });
  178. /************ PyInt ************/
  179. _vm->bind_static_method<1>("int", "__new__", [](VM* vm, Args& args) {
  180. if (is_type(args[0], vm->tp_int)) return args[0];
  181. if (is_type(args[0], vm->tp_float)) return VAR((i64)CAST(f64, args[0]));
  182. if (is_type(args[0], vm->tp_bool)) return VAR(_CAST(bool, args[0]) ? 1 : 0);
  183. if (is_type(args[0], vm->tp_str)) {
  184. const Str& s = CAST(Str&, args[0]);
  185. try{
  186. size_t parsed = 0;
  187. i64 val = S_TO_INT(s, &parsed, 10);
  188. if(parsed != s.size()) throw std::invalid_argument("<?>");
  189. return VAR(val);
  190. }catch(std::invalid_argument&){
  191. vm->ValueError("invalid literal for int(): " + s.escape(true));
  192. }
  193. }
  194. vm->TypeError("int() argument must be a int, float, bool or str");
  195. return vm->None;
  196. });
  197. _vm->bind_method<1>("int", "__floordiv__", [](VM* vm, Args& args) {
  198. i64 rhs = CAST(i64, args[1]);
  199. if(rhs == 0) vm->ZeroDivisionError();
  200. return VAR(CAST(i64, args[0]) / rhs);
  201. });
  202. _vm->bind_method<1>("int", "__mod__", [](VM* vm, Args& args) {
  203. i64 rhs = CAST(i64, args[1]);
  204. if(rhs == 0) vm->ZeroDivisionError();
  205. return VAR(CAST(i64, args[0]) % rhs);
  206. });
  207. _vm->bind_method<0>("int", "__repr__", CPP_LAMBDA(VAR(std::to_string(CAST(i64, args[0])))));
  208. _vm->bind_method<0>("int", "__json__", CPP_LAMBDA(VAR(std::to_string(CAST(i64, args[0])))));
  209. #define INT_BITWISE_OP(name,op) \
  210. _vm->bind_method<1>("int", #name, CPP_LAMBDA(VAR(CAST(i64, args[0]) op CAST(i64, args[1]))));
  211. INT_BITWISE_OP(__lshift__, <<)
  212. INT_BITWISE_OP(__rshift__, >>)
  213. INT_BITWISE_OP(__and__, &)
  214. INT_BITWISE_OP(__or__, |)
  215. INT_BITWISE_OP(__xor__, ^)
  216. #undef INT_BITWISE_OP
  217. /************ PyFloat ************/
  218. _vm->bind_static_method<1>("float", "__new__", [](VM* vm, Args& args) {
  219. if (is_type(args[0], vm->tp_int)) return VAR((f64)CAST(i64, args[0]));
  220. if (is_type(args[0], vm->tp_float)) return args[0];
  221. if (is_type(args[0], vm->tp_bool)) return VAR(_CAST(bool, args[0]) ? 1.0 : 0.0);
  222. if (is_type(args[0], vm->tp_str)) {
  223. const Str& s = CAST(Str&, args[0]);
  224. if(s == "inf") return VAR(INFINITY);
  225. if(s == "-inf") return VAR(-INFINITY);
  226. try{
  227. f64 val = S_TO_FLOAT(s);
  228. return VAR(val);
  229. }catch(std::invalid_argument&){
  230. vm->ValueError("invalid literal for float(): '" + s + "'");
  231. }
  232. }
  233. vm->TypeError("float() argument must be a int, float, bool or str");
  234. return vm->None;
  235. });
  236. _vm->bind_method<0>("float", "__repr__", [](VM* vm, Args& args) {
  237. f64 val = CAST(f64, args[0]);
  238. if(std::isinf(val) || std::isnan(val)) return VAR(std::to_string(val));
  239. StrStream ss;
  240. ss << std::setprecision(std::numeric_limits<f64>::max_digits10-1-2) << val;
  241. std::string s = ss.str();
  242. if(std::all_of(s.begin()+1, s.end(), isdigit)) s += ".0";
  243. return VAR(s);
  244. });
  245. _vm->bind_method<0>("float", "__json__", [](VM* vm, Args& args) {
  246. f64 val = CAST(f64, args[0]);
  247. if(std::isinf(val) || std::isnan(val)) vm->ValueError("cannot jsonify 'nan' or 'inf'");
  248. return VAR(std::to_string(val));
  249. });
  250. /************ PyString ************/
  251. _vm->bind_static_method<1>("str", "__new__", CPP_LAMBDA(vm->asStr(args[0])));
  252. _vm->bind_method<1>("str", "__add__", [](VM* vm, Args& args) {
  253. const Str& lhs = CAST(Str&, args[0]);
  254. const Str& rhs = CAST(Str&, args[1]);
  255. return VAR(lhs + rhs);
  256. });
  257. _vm->bind_method<0>("str", "__len__", [](VM* vm, Args& args) {
  258. const Str& self = CAST(Str&, args[0]);
  259. return VAR(self.u8_length());
  260. });
  261. _vm->bind_method<1>("str", "__contains__", [](VM* vm, Args& args) {
  262. const Str& self = CAST(Str&, args[0]);
  263. const Str& other = CAST(Str&, args[1]);
  264. return VAR(self.find(other) != Str::npos);
  265. });
  266. _vm->bind_method<0>("str", "__str__", CPP_LAMBDA(args[0]));
  267. _vm->bind_method<0>("str", "__iter__", CPP_LAMBDA(vm->PyIter(StringIter(vm, args[0]))));
  268. _vm->bind_method<0>("str", "__repr__", [](VM* vm, Args& args) {
  269. const Str& _self = CAST(Str&, args[0]);
  270. return VAR(_self.escape(true));
  271. });
  272. _vm->bind_method<0>("str", "__json__", [](VM* vm, Args& args) {
  273. const Str& self = CAST(Str&, args[0]);
  274. return VAR(self.escape(false));
  275. });
  276. _vm->bind_method<1>("str", "__eq__", [](VM* vm, Args& args) {
  277. if(is_type(args[0], vm->tp_str) && is_type(args[1], vm->tp_str))
  278. return VAR(CAST(Str&, args[0]) == CAST(Str&, args[1]));
  279. return VAR(args[0] == args[1]);
  280. });
  281. _vm->bind_method<1>("str", "__ne__", [](VM* vm, Args& args) {
  282. if(is_type(args[0], vm->tp_str) && is_type(args[1], vm->tp_str))
  283. return VAR(CAST(Str&, args[0]) != CAST(Str&, args[1]));
  284. return VAR(args[0] != args[1]);
  285. });
  286. _vm->bind_method<1>("str", "__getitem__", [](VM* vm, Args& args) {
  287. const Str& self (CAST(Str&, args[0]));
  288. if(is_type(args[1], vm->tp_slice)){
  289. Slice s = _CAST(Slice, args[1]);
  290. s.normalize(self.u8_length());
  291. return VAR(self.u8_substr(s.start, s.stop));
  292. }
  293. int index = CAST(int, args[1]);
  294. index = vm->normalized_index(index, self.u8_length());
  295. return VAR(self.u8_getitem(index));
  296. });
  297. _vm->bind_method<1>("str", "__gt__", [](VM* vm, Args& args) {
  298. const Str& self (CAST(Str&, args[0]));
  299. const Str& obj (CAST(Str&, args[1]));
  300. return VAR(self > obj);
  301. });
  302. _vm->bind_method<1>("str", "__lt__", [](VM* vm, Args& args) {
  303. const Str& self (CAST(Str&, args[0]));
  304. const Str& obj (CAST(Str&, args[1]));
  305. return VAR(self < obj);
  306. });
  307. _vm->bind_method<2>("str", "replace", [](VM* vm, Args& args) {
  308. const Str& _self = CAST(Str&, args[0]);
  309. const Str& _old = CAST(Str&, args[1]);
  310. const Str& _new = CAST(Str&, args[2]);
  311. Str _copy = _self;
  312. size_t pos = 0;
  313. while ((pos = _copy.find(_old, pos)) != std::string::npos) {
  314. _copy.replace(pos, _old.length(), _new);
  315. pos += _new.length();
  316. }
  317. return VAR(_copy);
  318. });
  319. _vm->bind_method<1>("str", "startswith", [](VM* vm, Args& args) {
  320. const Str& self = CAST(Str&, args[0]);
  321. const Str& prefix = CAST(Str&, args[1]);
  322. return VAR(self.find(prefix) == 0);
  323. });
  324. _vm->bind_method<1>("str", "endswith", [](VM* vm, Args& args) {
  325. const Str& self = CAST(Str&, args[0]);
  326. const Str& suffix = CAST(Str&, args[1]);
  327. return VAR(self.rfind(suffix) == self.length() - suffix.length());
  328. });
  329. _vm->bind_method<1>("str", "join", [](VM* vm, Args& args) {
  330. const Str& self = CAST(Str&, args[0]);
  331. StrStream ss;
  332. PyVar obj = vm->asList(args[1]);
  333. const List& list = CAST(List&, obj);
  334. for (int i = 0; i < list.size(); ++i) {
  335. if (i > 0) ss << self;
  336. ss << CAST(Str&, list[i]);
  337. }
  338. return VAR(ss.str());
  339. });
  340. /************ PyList ************/
  341. _vm->bind_method<1>("list", "append", [](VM* vm, Args& args) {
  342. List& self = CAST(List&, args[0]);
  343. self.push_back(args[1]);
  344. return vm->None;
  345. });
  346. _vm->bind_method<0>("list", "reverse", [](VM* vm, Args& args) {
  347. List& self = CAST(List&, args[0]);
  348. std::reverse(self.begin(), self.end());
  349. return vm->None;
  350. });
  351. _vm->bind_method<1>("list", "__mul__", [](VM* vm, Args& args) {
  352. const List& self = CAST(List&, args[0]);
  353. int n = CAST(int, args[1]);
  354. List result;
  355. result.reserve(self.size() * n);
  356. for(int i = 0; i < n; i++) result.insert(result.end(), self.begin(), self.end());
  357. return VAR(std::move(result));
  358. });
  359. _vm->bind_method<2>("list", "insert", [](VM* vm, Args& args) {
  360. List& self = CAST(List&, args[0]);
  361. int index = CAST(int, args[1]);
  362. if(index < 0) index += self.size();
  363. if(index < 0) index = 0;
  364. if(index > self.size()) index = self.size();
  365. self.insert(self.begin() + index, args[2]);
  366. return vm->None;
  367. });
  368. _vm->bind_method<0>("list", "clear", [](VM* vm, Args& args) {
  369. CAST(List&, args[0]).clear();
  370. return vm->None;
  371. });
  372. _vm->bind_method<0>("list", "copy", CPP_LAMBDA(VAR(CAST(List, args[0]))));
  373. _vm->bind_method<1>("list", "__add__", [](VM* vm, Args& args) {
  374. const List& self = CAST(List&, args[0]);
  375. const List& obj = CAST(List&, args[1]);
  376. List new_list = self;
  377. new_list.insert(new_list.end(), obj.begin(), obj.end());
  378. return VAR(new_list);
  379. });
  380. _vm->bind_method<0>("list", "__len__", [](VM* vm, Args& args) {
  381. const List& self = CAST(List&, args[0]);
  382. return VAR(self.size());
  383. });
  384. _vm->bind_method<0>("list", "__iter__", [](VM* vm, Args& args) {
  385. return vm->PyIter(ArrayIter<List>(vm, args[0]));
  386. });
  387. _vm->bind_method<1>("list", "__getitem__", [](VM* vm, Args& args) {
  388. const List& self = CAST(List&, args[0]);
  389. if(is_type(args[1], vm->tp_slice)){
  390. Slice s = _CAST(Slice, args[1]);
  391. s.normalize(self.size());
  392. List new_list;
  393. for(size_t i = s.start; i < s.stop; i++) new_list.push_back(self[i]);
  394. return VAR(std::move(new_list));
  395. }
  396. int index = CAST(int, args[1]);
  397. index = vm->normalized_index(index, self.size());
  398. return self[index];
  399. });
  400. _vm->bind_method<2>("list", "__setitem__", [](VM* vm, Args& args) {
  401. List& self = CAST(List&, args[0]);
  402. int index = CAST(int, args[1]);
  403. index = vm->normalized_index(index, self.size());
  404. self[index] = args[2];
  405. return vm->None;
  406. });
  407. _vm->bind_method<1>("list", "__delitem__", [](VM* vm, Args& args) {
  408. List& self = CAST(List&, args[0]);
  409. int index = CAST(int, args[1]);
  410. index = vm->normalized_index(index, self.size());
  411. self.erase(self.begin() + index);
  412. return vm->None;
  413. });
  414. /************ PyTuple ************/
  415. _vm->bind_static_method<1>("tuple", "__new__", [](VM* vm, Args& args) {
  416. List list = CAST(List, vm->asList(args[0]));
  417. return VAR(std::move(list));
  418. });
  419. _vm->bind_method<0>("tuple", "__iter__", [](VM* vm, Args& args) {
  420. return vm->PyIter(ArrayIter<Args>(vm, args[0]));
  421. });
  422. _vm->bind_method<1>("tuple", "__getitem__", [](VM* vm, Args& args) {
  423. const Tuple& self = CAST(Tuple&, args[0]);
  424. if(is_type(args[1], vm->tp_slice)){
  425. Slice s = _CAST(Slice, args[1]);
  426. s.normalize(self.size());
  427. List new_list;
  428. for(size_t i = s.start; i < s.stop; i++) new_list.push_back(self[i]);
  429. return VAR(std::move(new_list));
  430. }
  431. int index = CAST(int, args[1]);
  432. index = vm->normalized_index(index, self.size());
  433. return self[index];
  434. });
  435. _vm->bind_method<0>("tuple", "__len__", [](VM* vm, Args& args) {
  436. const Tuple& self = CAST(Tuple&, args[0]);
  437. return VAR(self.size());
  438. });
  439. /************ PyBool ************/
  440. _vm->bind_static_method<1>("bool", "__new__", CPP_LAMBDA(vm->asBool(args[0])));
  441. _vm->bind_method<0>("bool", "__repr__", [](VM* vm, Args& args) {
  442. bool val = CAST(bool, args[0]);
  443. return VAR(val ? "True" : "False");
  444. });
  445. _vm->bind_method<0>("bool", "__json__", [](VM* vm, Args& args) {
  446. bool val = CAST(bool, args[0]);
  447. return VAR(val ? "true" : "false");
  448. });
  449. _vm->bind_method<1>("bool", "__xor__", [](VM* vm, Args& args) {
  450. bool self = CAST(bool, args[0]);
  451. bool other = CAST(bool, args[1]);
  452. return VAR(self ^ other);
  453. });
  454. _vm->bind_method<0>("ellipsis", "__repr__", CPP_LAMBDA(VAR("Ellipsis")));
  455. }
  456. #ifdef _WIN32
  457. #define __EXPORT __declspec(dllexport)
  458. #elif __APPLE__
  459. #define __EXPORT __attribute__((visibility("default"))) __attribute__((used))
  460. #elif __EMSCRIPTEN__
  461. #include <emscripten.h>
  462. #define __EXPORT EMSCRIPTEN_KEEPALIVE
  463. #else
  464. #define __EXPORT
  465. #endif
  466. void add_module_time(VM* vm){
  467. PyVar mod = vm->new_module("time");
  468. vm->bind_func<0>(mod, "time", [](VM* vm, Args& args) {
  469. auto now = std::chrono::high_resolution_clock::now();
  470. return VAR(std::chrono::duration_cast<std::chrono::microseconds>(now.time_since_epoch()).count() / 1000000.0);
  471. });
  472. }
  473. void add_module_sys(VM* vm){
  474. PyVar mod = vm->new_module("sys");
  475. vm->setattr(mod, "version", VAR(PK_VERSION));
  476. vm->bind_func<1>(mod, "getrefcount", CPP_LAMBDA(VAR(args[0].use_count())));
  477. vm->bind_func<0>(mod, "getrecursionlimit", CPP_LAMBDA(VAR(vm->recursionlimit)));
  478. vm->bind_func<1>(mod, "setrecursionlimit", [](VM* vm, Args& args) {
  479. vm->recursionlimit = CAST(int, args[0]);
  480. return vm->None;
  481. });
  482. }
  483. void add_module_json(VM* vm){
  484. PyVar mod = vm->new_module("json");
  485. vm->bind_func<1>(mod, "loads", [](VM* vm, Args& args) {
  486. const Str& expr = CAST(Str&, args[0]);
  487. CodeObject_ code = vm->compile(expr, "<json>", JSON_MODE);
  488. return vm->_exec(code, vm->top_frame()->_module, vm->top_frame()->_locals);
  489. });
  490. vm->bind_func<1>(mod, "dumps", CPP_LAMBDA(vm->call(args[0], __json__)));
  491. }
  492. void add_module_math(VM* vm){
  493. PyVar mod = vm->new_module("math");
  494. vm->setattr(mod, "pi", VAR(3.1415926535897932384));
  495. vm->setattr(mod, "e" , VAR(2.7182818284590452354));
  496. vm->bind_func<1>(mod, "log", CPP_LAMBDA(VAR(std::log(vm->num_to_float(args[0])))));
  497. vm->bind_func<1>(mod, "log10", CPP_LAMBDA(VAR(std::log10(vm->num_to_float(args[0])))));
  498. vm->bind_func<1>(mod, "log2", CPP_LAMBDA(VAR(std::log2(vm->num_to_float(args[0])))));
  499. vm->bind_func<1>(mod, "sin", CPP_LAMBDA(VAR(std::sin(vm->num_to_float(args[0])))));
  500. vm->bind_func<1>(mod, "cos", CPP_LAMBDA(VAR(std::cos(vm->num_to_float(args[0])))));
  501. vm->bind_func<1>(mod, "tan", CPP_LAMBDA(VAR(std::tan(vm->num_to_float(args[0])))));
  502. vm->bind_func<1>(mod, "isnan", CPP_LAMBDA(VAR(std::isnan(vm->num_to_float(args[0])))));
  503. vm->bind_func<1>(mod, "isinf", CPP_LAMBDA(VAR(std::isinf(vm->num_to_float(args[0])))));
  504. vm->bind_func<1>(mod, "fabs", CPP_LAMBDA(VAR(std::fabs(vm->num_to_float(args[0])))));
  505. vm->bind_func<1>(mod, "floor", CPP_LAMBDA(VAR((i64)std::floor(vm->num_to_float(args[0])))));
  506. vm->bind_func<1>(mod, "ceil", CPP_LAMBDA(VAR((i64)std::ceil(vm->num_to_float(args[0])))));
  507. vm->bind_func<1>(mod, "sqrt", CPP_LAMBDA(VAR(std::sqrt(vm->num_to_float(args[0])))));
  508. }
  509. void add_module_dis(VM* vm){
  510. PyVar mod = vm->new_module("dis");
  511. vm->bind_func<1>(mod, "dis", [](VM* vm, Args& args) {
  512. PyVar f = args[0];
  513. if(is_type(f, vm->tp_bound_method)) f = CAST(BoundMethod, args[0]).method;
  514. CodeObject_ code = CAST(Function, f).code;
  515. (*vm->_stdout) << vm->disassemble(code);
  516. return vm->None;
  517. });
  518. }
  519. struct ReMatch {
  520. PY_CLASS(ReMatch, re, Match)
  521. i64 start;
  522. i64 end;
  523. std::smatch m;
  524. ReMatch(i64 start, i64 end, std::smatch m) : start(start), end(end), m(m) {}
  525. static void _register(VM* vm, PyVar mod, PyVar type){
  526. vm->bind_method<-1>(type, "__init__", CPP_NOT_IMPLEMENTED());
  527. vm->bind_method<0>(type, "start", CPP_LAMBDA(VAR(CAST(ReMatch&, args[0]).start)));
  528. vm->bind_method<0>(type, "end", CPP_LAMBDA(VAR(CAST(ReMatch&, args[0]).end)));
  529. vm->bind_method<0>(type, "span", [](VM* vm, Args& args) {
  530. auto& self = CAST(ReMatch&, args[0]);
  531. return VAR(two_args(VAR(self.start), VAR(self.end)));
  532. });
  533. vm->bind_method<1>(type, "group", [](VM* vm, Args& args) {
  534. auto& self = CAST(ReMatch&, args[0]);
  535. int index = CAST(int, args[1]);
  536. index = vm->normalized_index(index, self.m.size());
  537. return VAR(self.m[index].str());
  538. });
  539. }
  540. };
  541. PyVar _regex_search(const Str& pattern, const Str& string, bool fromStart, VM* vm){
  542. std::regex re(pattern);
  543. std::smatch m;
  544. if(std::regex_search(string, m, re)){
  545. if(fromStart && m.position() != 0) return vm->None;
  546. i64 start = string._to_u8_index(m.position());
  547. i64 end = string._to_u8_index(m.position() + m.length());
  548. return VAR_T(ReMatch, start, end, m);
  549. }
  550. return vm->None;
  551. };
  552. void add_module_re(VM* vm){
  553. PyVar mod = vm->new_module("re");
  554. ReMatch::register_class(vm, mod);
  555. vm->bind_func<2>(mod, "match", [](VM* vm, Args& args) {
  556. const Str& pattern = CAST(Str&, args[0]);
  557. const Str& string = CAST(Str&, args[1]);
  558. return _regex_search(pattern, string, true, vm);
  559. });
  560. vm->bind_func<2>(mod, "search", [](VM* vm, Args& args) {
  561. const Str& pattern = CAST(Str&, args[0]);
  562. const Str& string = CAST(Str&, args[1]);
  563. return _regex_search(pattern, string, false, vm);
  564. });
  565. vm->bind_func<3>(mod, "sub", [](VM* vm, Args& args) {
  566. const Str& pattern = CAST(Str&, args[0]);
  567. const Str& repl = CAST(Str&, args[1]);
  568. const Str& string = CAST(Str&, args[2]);
  569. std::regex re(pattern);
  570. return VAR(std::regex_replace(string, re, repl));
  571. });
  572. vm->bind_func<2>(mod, "split", [](VM* vm, Args& args) {
  573. const Str& pattern = CAST(Str&, args[0]);
  574. const Str& string = CAST(Str&, args[1]);
  575. std::regex re(pattern);
  576. std::sregex_token_iterator it(string.begin(), string.end(), re, -1);
  577. std::sregex_token_iterator end;
  578. List vec;
  579. for(; it != end; ++it){
  580. vec.push_back(VAR(it->str()));
  581. }
  582. return VAR(vec);
  583. });
  584. }
  585. void add_module_random(VM* vm){
  586. PyVar mod = vm->new_module("random");
  587. std::srand(std::time(0));
  588. vm->bind_func<1>(mod, "seed", [](VM* vm, Args& args) {
  589. std::srand((unsigned int)CAST(i64, args[0]));
  590. return vm->None;
  591. });
  592. vm->bind_func<0>(mod, "random", CPP_LAMBDA(VAR(std::rand() / (f64)RAND_MAX)));
  593. vm->bind_func<2>(mod, "randint", [](VM* vm, Args& args) {
  594. i64 a = CAST(i64, args[0]);
  595. i64 b = CAST(i64, args[1]);
  596. if(a > b) std::swap(a, b);
  597. return VAR(a + std::rand() % (b - a + 1));
  598. });
  599. vm->bind_func<2>(mod, "uniform", [](VM* vm, Args& args) {
  600. f64 a = CAST(f64, args[0]);
  601. f64 b = CAST(f64, args[1]);
  602. if(a > b) std::swap(a, b);
  603. return VAR(a + (b - a) * std::rand() / (f64)RAND_MAX);
  604. });
  605. CodeObject_ code = vm->compile(kPythonLibs["random"], "random.py", EXEC_MODE);
  606. vm->_exec(code, mod);
  607. }
  608. void VM::post_init(){
  609. init_builtins(this);
  610. add_module_sys(this);
  611. add_module_time(this);
  612. add_module_json(this);
  613. add_module_math(this);
  614. add_module_re(this);
  615. add_module_dis(this);
  616. add_module_random(this);
  617. add_module_io(this);
  618. add_module_os(this);
  619. add_module_c(this);
  620. _lazy_modules["functools"] = kPythonLibs["functools"];
  621. _lazy_modules["collections"] = kPythonLibs["collections"];
  622. _lazy_modules["heapq"] = kPythonLibs["heapq"];
  623. CodeObject_ code = compile(kPythonLibs["builtins"], "<builtins>", EXEC_MODE);
  624. this->_exec(code, this->builtins);
  625. code = compile(kPythonLibs["dict"], "<builtins>", EXEC_MODE);
  626. this->_exec(code, this->builtins);
  627. code = compile(kPythonLibs["set"], "<builtins>", EXEC_MODE);
  628. this->_exec(code, this->builtins);
  629. }
  630. } // namespace pkpy
  631. /*************************GLOBAL NAMESPACE*************************/
  632. class PkExportedBase{
  633. public:
  634. virtual ~PkExportedBase() = default;
  635. virtual void* get() = 0;
  636. };
  637. static std::vector<PkExportedBase*> _pk_lookup_table;
  638. template<typename T>
  639. class PkExported : public PkExportedBase{
  640. T* _ptr;
  641. public:
  642. template<typename... Args>
  643. PkExported(Args&&... args) {
  644. _ptr = new T(std::forward<Args>(args)...);
  645. _pk_lookup_table.push_back(this);
  646. }
  647. ~PkExported() override { delete _ptr; }
  648. void* get() override { return _ptr; }
  649. operator T*() { return _ptr; }
  650. };
  651. #define PKPY_ALLOCATE(T, ...) *(new PkExported<T>(__VA_ARGS__))
  652. extern "C" {
  653. __EXPORT
  654. /// Delete a pointer allocated by `pkpy_xxx_xxx`.
  655. /// It can be `VM*`, `REPL*`, `char*`, etc.
  656. ///
  657. /// !!!
  658. /// If the pointer is not allocated by `pkpy_xxx_xxx`, the behavior is undefined.
  659. /// !!!
  660. void pkpy_delete(void* p){
  661. for(int i = 0; i < _pk_lookup_table.size(); i++){
  662. if(_pk_lookup_table[i]->get() == p){
  663. delete _pk_lookup_table[i];
  664. _pk_lookup_table.erase(_pk_lookup_table.begin() + i);
  665. return;
  666. }
  667. }
  668. free(p);
  669. }
  670. __EXPORT
  671. /// Run a given source on a virtual machine.
  672. void pkpy_vm_exec(pkpy::VM* vm, const char* source){
  673. vm->exec(source, "main.py", pkpy::EXEC_MODE);
  674. }
  675. __EXPORT
  676. /// Get a global variable of a virtual machine.
  677. ///
  678. /// Return `__repr__` of the result.
  679. /// If the variable is not found, return `nullptr`.
  680. char* pkpy_vm_get_global(pkpy::VM* vm, const char* name){
  681. pkpy::PyVar* val = vm->_main->attr().try_get(name);
  682. if(val == nullptr) return nullptr;
  683. try{
  684. pkpy::Str repr = pkpy::CAST(pkpy::Str, vm->asRepr(*val));
  685. return strdup(repr.c_str());
  686. }catch(...){
  687. return nullptr;
  688. }
  689. }
  690. __EXPORT
  691. /// Evaluate an expression.
  692. ///
  693. /// Return `__repr__` of the result.
  694. /// If there is any error, return `nullptr`.
  695. char* pkpy_vm_eval(pkpy::VM* vm, const char* source){
  696. pkpy::PyVarOrNull ret = vm->exec(source, "<eval>", pkpy::EVAL_MODE);
  697. if(ret == nullptr) return nullptr;
  698. try{
  699. pkpy::Str repr = pkpy::CAST(pkpy::Str, vm->asRepr(ret));
  700. return strdup(repr.c_str());
  701. }catch(...){
  702. return nullptr;
  703. }
  704. }
  705. __EXPORT
  706. /// Create a REPL, using the given virtual machine as the backend.
  707. pkpy::REPL* pkpy_new_repl(pkpy::VM* vm){
  708. return PKPY_ALLOCATE(pkpy::REPL, vm);
  709. }
  710. __EXPORT
  711. /// Input a source line to an interactive console. Return true if need more lines.
  712. bool pkpy_repl_input(pkpy::REPL* r, const char* line){
  713. return r->input(line);
  714. }
  715. __EXPORT
  716. /// Add a source module into a virtual machine.
  717. void pkpy_vm_add_module(pkpy::VM* vm, const char* name, const char* source){
  718. vm->_lazy_modules[name] = source;
  719. }
  720. __EXPORT
  721. /// Create a virtual machine.
  722. pkpy::VM* pkpy_new_vm(bool use_stdio){
  723. return PKPY_ALLOCATE(pkpy::VM, use_stdio);
  724. }
  725. __EXPORT
  726. /// Read the standard output and standard error as string of a virtual machine.
  727. /// The `vm->use_stdio` should be `false`.
  728. /// After this operation, both stream will be cleared.
  729. ///
  730. /// Return a json representing the result.
  731. char* pkpy_vm_read_output(pkpy::VM* vm){
  732. if(vm->use_stdio) return nullptr;
  733. pkpy::StrStream* s_out = (pkpy::StrStream*)(vm->_stdout);
  734. pkpy::StrStream* s_err = (pkpy::StrStream*)(vm->_stderr);
  735. pkpy::Str _stdout = s_out->str();
  736. pkpy::Str _stderr = s_err->str();
  737. pkpy::StrStream ss;
  738. ss << '{' << "\"stdout\": " << _stdout.escape(false);
  739. ss << ", " << "\"stderr\": " << _stderr.escape(false) << '}';
  740. s_out->str(""); s_err->str("");
  741. return strdup(ss.str().c_str());
  742. }
  743. typedef i64 (*f_int_t)(char*);
  744. typedef f64 (*f_float_t)(char*);
  745. typedef bool (*f_bool_t)(char*);
  746. typedef char* (*f_str_t)(char*);
  747. typedef void (*f_None_t)(char*);
  748. static f_int_t f_int = nullptr;
  749. static f_float_t f_float = nullptr;
  750. static f_bool_t f_bool = nullptr;
  751. static f_str_t f_str = nullptr;
  752. static f_None_t f_None = nullptr;
  753. __EXPORT
  754. /// Setup the callback functions.
  755. void pkpy_setup_callbacks(f_int_t _f_int, f_float_t _f_float, f_bool_t _f_bool, f_str_t _f_str, f_None_t _f_None){
  756. f_int = _f_int;
  757. f_float = _f_float;
  758. f_bool = _f_bool;
  759. f_str = _f_str;
  760. f_None = _f_None;
  761. }
  762. __EXPORT
  763. /// Bind a function to a virtual machine.
  764. char* pkpy_vm_bind(pkpy::VM* vm, const char* mod, const char* name, int ret_code){
  765. if(!f_int || !f_float || !f_bool || !f_str || !f_None) return nullptr;
  766. static int kGlobalBindId = 0;
  767. for(int i=0; mod[i]; i++) if(mod[i] == ' ') return nullptr;
  768. for(int i=0; name[i]; i++) if(name[i] == ' ') return nullptr;
  769. std::string f_header = std::string(mod) + '.' + name + '#' + std::to_string(kGlobalBindId++);
  770. pkpy::PyVar obj = vm->_modules.contains(mod) ? vm->_modules[mod] : vm->new_module(mod);
  771. vm->bind_func<-1>(obj, name, [ret_code, f_header](pkpy::VM* vm, const pkpy::Args& args){
  772. pkpy::StrStream ss;
  773. ss << f_header;
  774. for(int i=0; i<args.size(); i++){
  775. ss << ' ';
  776. pkpy::PyVar x = vm->call(args[i], pkpy::__json__);
  777. ss << pkpy::CAST(pkpy::Str&, x);
  778. }
  779. char* packet = strdup(ss.str().c_str());
  780. switch(ret_code){
  781. case 'i': return VAR(f_int(packet));
  782. case 'f': return VAR(f_float(packet));
  783. case 'b': return VAR(f_bool(packet));
  784. case 's': {
  785. char* p = f_str(packet);
  786. if(p == nullptr) return vm->None;
  787. return VAR(p); // no need to free(p)
  788. }
  789. case 'N': f_None(packet); return vm->None;
  790. }
  791. free(packet);
  792. UNREACHABLE();
  793. return vm->None;
  794. });
  795. return strdup(f_header.c_str());
  796. }
  797. }