pocketpy.h 36 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962
  1. #pragma once
  2. #include "ceval.h"
  3. #include "compiler.h"
  4. #include "repl.h"
  5. #include "iter.h"
  6. #define CPP_LAMBDA(x) ([](VM* vm, pkpy::Args& args) { return x; })
  7. #define CPP_NOT_IMPLEMENTED() ([](VM* vm, pkpy::Args& args) { vm->NotImplementedError(); return vm->None; })
  8. CodeObject_ VM::compile(Str source, Str filename, CompileMode mode) {
  9. Compiler compiler(this, source.c_str(), filename, mode);
  10. try{
  11. return compiler.compile();
  12. }catch(pkpy::Exception& e){
  13. _error(e);
  14. return nullptr;
  15. }
  16. }
  17. #define BIND_NUM_ARITH_OPT(name, op) \
  18. _vm->_bind_methods<1>({"int","float"}, #name, [](VM* vm, pkpy::Args& args){ \
  19. if(args[0]->is_type(vm->tp_int) && args[1]->is_type(vm->tp_int)){ \
  20. return vm->PyInt(vm->PyInt_AS_C(args[0]) op vm->PyInt_AS_C(args[1])); \
  21. }else{ \
  22. return vm->PyFloat(vm->num_to_float(args[0]) op vm->num_to_float(args[1])); \
  23. } \
  24. });
  25. #define BIND_NUM_LOGICAL_OPT(name, op, is_eq) \
  26. _vm->_bind_methods<1>({"int","float"}, #name, [](VM* vm, pkpy::Args& args){ \
  27. bool _0 = args[0]->is_type(vm->tp_int) || args[0]->is_type(vm->tp_float); \
  28. bool _1 = args[1]->is_type(vm->tp_int) || args[1]->is_type(vm->tp_float); \
  29. if(!_0 || !_1){ \
  30. if constexpr(is_eq) return vm->PyBool(args[0].get() op args[1].get()); \
  31. vm->TypeError("unsupported operand type(s) for " #op ); \
  32. } \
  33. return vm->PyBool(vm->num_to_float(args[0]) op vm->num_to_float(args[1])); \
  34. });
  35. void init_builtins(VM* _vm) {
  36. BIND_NUM_ARITH_OPT(__add__, +)
  37. BIND_NUM_ARITH_OPT(__sub__, -)
  38. BIND_NUM_ARITH_OPT(__mul__, *)
  39. BIND_NUM_LOGICAL_OPT(__lt__, <, false)
  40. BIND_NUM_LOGICAL_OPT(__le__, <=, false)
  41. BIND_NUM_LOGICAL_OPT(__gt__, >, false)
  42. BIND_NUM_LOGICAL_OPT(__ge__, >=, false)
  43. BIND_NUM_LOGICAL_OPT(__eq__, ==, true)
  44. BIND_NUM_LOGICAL_OPT(__ne__, !=, true)
  45. #undef BIND_NUM_ARITH_OPT
  46. #undef BIND_NUM_LOGICAL_OPT
  47. _vm->bind_builtin_func<1>("__sys_stdout_write", [](VM* vm, pkpy::Args& args) {
  48. (*vm->_stdout) << vm->PyStr_AS_C(args[0]);
  49. return vm->None;
  50. });
  51. _vm->bind_builtin_func<0>("super", [](VM* vm, pkpy::Args& args) {
  52. const PyVar* self = vm->top_frame()->f_locals().try_get(m_self);
  53. if(self == nullptr) vm->TypeError("super() can only be called in a class");
  54. return vm->new_object(vm->tp_super, *self);
  55. });
  56. _vm->bind_builtin_func<1>("eval", [](VM* vm, pkpy::Args& args) {
  57. CodeObject_ code = vm->compile(vm->PyStr_AS_C(args[0]), "<eval>", EVAL_MODE);
  58. return vm->_exec(code, vm->top_frame()->_module, vm->top_frame()->_locals);
  59. });
  60. _vm->bind_builtin_func<1>("exec", [](VM* vm, pkpy::Args& args) {
  61. CodeObject_ code = vm->compile(vm->PyStr_AS_C(args[0]), "<exec>", EXEC_MODE);
  62. vm->_exec(code, vm->top_frame()->_module, vm->top_frame()->_locals);
  63. return vm->None;
  64. });
  65. _vm->bind_builtin_func<-1>("exit", [](VM* vm, pkpy::Args& args) {
  66. if(args.size() == 0) std::exit(0);
  67. else if(args.size() == 1) std::exit((int)vm->PyInt_AS_C(args[0]));
  68. else vm->TypeError("exit() takes at most 1 argument");
  69. return vm->None;
  70. });
  71. _vm->bind_builtin_func<1>("repr", CPP_LAMBDA(vm->asRepr(args[0])));
  72. _vm->bind_builtin_func<1>("hash", CPP_LAMBDA(vm->PyInt(vm->hash(args[0]))));
  73. _vm->bind_builtin_func<1>("len", CPP_LAMBDA(vm->call(args[0], __len__, pkpy::no_arg())));
  74. _vm->bind_builtin_func<1>("chr", [](VM* vm, pkpy::Args& args) {
  75. i64 i = vm->PyInt_AS_C(args[0]);
  76. if (i < 0 || i > 128) vm->ValueError("chr() arg not in range(128)");
  77. return vm->PyStr(std::string(1, (char)i));
  78. });
  79. _vm->bind_builtin_func<1>("ord", [](VM* vm, pkpy::Args& args) {
  80. Str s = vm->PyStr_AS_C(args[0]);
  81. if (s.size() != 1) vm->TypeError("ord() expected an ASCII character");
  82. return vm->PyInt((i64)(s.c_str()[0]));
  83. });
  84. _vm->bind_builtin_func<2>("hasattr", [](VM* vm, pkpy::Args& args) {
  85. return vm->PyBool(vm->getattr(args[0], vm->PyStr_AS_C(args[1]), false) != nullptr);
  86. });
  87. _vm->bind_builtin_func<3>("setattr", [](VM* vm, pkpy::Args& args) {
  88. vm->setattr(args[0], vm->PyStr_AS_C(args[1]), args[2]);
  89. return vm->None;
  90. });
  91. _vm->bind_builtin_func<2>("getattr", [](VM* vm, pkpy::Args& args) {
  92. Str name = vm->PyStr_AS_C(args[1]);
  93. return vm->getattr(args[0], name);
  94. });
  95. _vm->bind_builtin_func<1>("hex", [](VM* vm, pkpy::Args& args) {
  96. std::stringstream ss;
  97. ss << std::hex << vm->PyInt_AS_C(args[0]);
  98. return vm->PyStr("0x" + ss.str());
  99. });
  100. _vm->bind_builtin_func<1>("dir", [](VM* vm, pkpy::Args& args) {
  101. std::vector<Str> names;
  102. if(args[0]->is_attr_valid()){
  103. for (auto& [k, _] : args[0]->attr()) names.push_back(k);
  104. }
  105. for (auto& [k, _] : vm->_t(args[0])->attr()) {
  106. if (std::find(names.begin(), names.end(), k) == names.end()) names.push_back(k);
  107. }
  108. pkpy::List ret;
  109. for (const auto& name : names) ret.push_back(vm->PyStr(name));
  110. return vm->PyList(std::move(ret));
  111. });
  112. _vm->bind_method<0>("object", "__repr__", [](VM* vm, pkpy::Args& args) {
  113. PyVar self = args[0];
  114. StrStream ss;
  115. ss << std::hex << (uintptr_t)self.get();
  116. Str s = "<" + OBJ_NAME(vm->_t(self)) + " object at 0x" + ss.str() + ">";
  117. return vm->PyStr(s);
  118. });
  119. _vm->bind_method<1>("object", "__eq__", CPP_LAMBDA(vm->PyBool(args[0] == args[1])));
  120. _vm->bind_method<1>("object", "__ne__", CPP_LAMBDA(vm->PyBool(args[0] != args[1])));
  121. _vm->bind_static_method<1>("type", "__new__", CPP_LAMBDA(vm->_t(args[0])));
  122. _vm->bind_static_method<-1>("range", "__new__", [](VM* vm, pkpy::Args& args) {
  123. pkpy::Range r;
  124. switch (args.size()) {
  125. case 1: r.stop = vm->PyInt_AS_C(args[0]); break;
  126. case 2: r.start = vm->PyInt_AS_C(args[0]); r.stop = vm->PyInt_AS_C(args[1]); break;
  127. case 3: r.start = vm->PyInt_AS_C(args[0]); r.stop = vm->PyInt_AS_C(args[1]); r.step = vm->PyInt_AS_C(args[2]); break;
  128. default: vm->TypeError("expected 1-3 arguments, but got " + std::to_string(args.size()));
  129. }
  130. return vm->PyRange(r);
  131. });
  132. _vm->bind_method<0>("range", "__iter__", CPP_LAMBDA(
  133. vm->PyIter(pkpy::make_shared<BaseIter, RangeIter>(vm, args[0]))
  134. ));
  135. _vm->bind_method<0>("NoneType", "__repr__", CPP_LAMBDA(vm->PyStr("None")));
  136. _vm->bind_method<0>("NoneType", "__json__", CPP_LAMBDA(vm->PyStr("null")));
  137. _vm->_bind_methods<1>({"int", "float"}, "__truediv__", [](VM* vm, pkpy::Args& args) {
  138. f64 rhs = vm->num_to_float(args[1]);
  139. if (rhs == 0) vm->ZeroDivisionError();
  140. return vm->PyFloat(vm->num_to_float(args[0]) / rhs);
  141. });
  142. _vm->_bind_methods<1>({"int", "float"}, "__pow__", [](VM* vm, pkpy::Args& args) {
  143. if(args[0]->is_type(vm->tp_int) && args[1]->is_type(vm->tp_int)){
  144. return vm->PyInt((i64)round(pow(vm->PyInt_AS_C(args[0]), vm->PyInt_AS_C(args[1]))));
  145. }else{
  146. return vm->PyFloat((f64)pow(vm->num_to_float(args[0]), vm->num_to_float(args[1])));
  147. }
  148. });
  149. /************ PyInt ************/
  150. _vm->bind_static_method<1>("int", "__new__", [](VM* vm, pkpy::Args& args) {
  151. if (args[0]->is_type(vm->tp_int)) return args[0];
  152. if (args[0]->is_type(vm->tp_float)) return vm->PyInt((i64)vm->PyFloat_AS_C(args[0]));
  153. if (args[0]->is_type(vm->tp_bool)) return vm->PyInt(vm->PyBool_AS_C(args[0]) ? 1 : 0);
  154. if (args[0]->is_type(vm->tp_str)) {
  155. const Str& s = vm->PyStr_AS_C(args[0]);
  156. try{
  157. size_t parsed = 0;
  158. i64 val = std::stoll(s, &parsed, 10);
  159. if(parsed != s.size()) throw std::invalid_argument("");
  160. return vm->PyInt(val);
  161. }catch(std::invalid_argument&){
  162. vm->ValueError("invalid literal for int(): '" + s + "'");
  163. }
  164. }
  165. vm->TypeError("int() argument must be a int, float, bool or str");
  166. return vm->None;
  167. });
  168. _vm->bind_method<1>("int", "__floordiv__", [](VM* vm, pkpy::Args& args) {
  169. i64 rhs = vm->PyInt_AS_C(args[1]);
  170. if(rhs == 0) vm->ZeroDivisionError();
  171. return vm->PyInt(vm->PyInt_AS_C(args[0]) / rhs);
  172. });
  173. _vm->bind_method<1>("int", "__mod__", [](VM* vm, pkpy::Args& args) {
  174. i64 rhs = vm->PyInt_AS_C(args[1]);
  175. if(rhs == 0) vm->ZeroDivisionError();
  176. return vm->PyInt(vm->PyInt_AS_C(args[0]) % rhs);
  177. });
  178. _vm->bind_method<0>("int", "__repr__", [](VM* vm, pkpy::Args& args) {
  179. return vm->PyStr(std::to_string(vm->PyInt_AS_C(args[0])));
  180. });
  181. _vm->bind_method<0>("int", "__json__", [](VM* vm, pkpy::Args& args) {
  182. return vm->PyStr(std::to_string(vm->PyInt_AS_C(args[0])));
  183. });
  184. #define __INT_BITWISE_OP(name,op) \
  185. _vm->bind_method<1>("int", #name, [](VM* vm, pkpy::Args& args) { \
  186. return vm->PyInt(vm->PyInt_AS_C(args[0]) op vm->PyInt_AS_C(args[1])); \
  187. });
  188. __INT_BITWISE_OP(__lshift__, <<)
  189. __INT_BITWISE_OP(__rshift__, >>)
  190. __INT_BITWISE_OP(__and__, &)
  191. __INT_BITWISE_OP(__or__, |)
  192. __INT_BITWISE_OP(__xor__, ^)
  193. #undef __INT_BITWISE_OP
  194. /************ PyFloat ************/
  195. _vm->bind_static_method<1>("float", "__new__", [](VM* vm, pkpy::Args& args) {
  196. if (args[0]->is_type(vm->tp_int)) return vm->PyFloat((f64)vm->PyInt_AS_C(args[0]));
  197. if (args[0]->is_type(vm->tp_float)) return args[0];
  198. if (args[0]->is_type(vm->tp_bool)) return vm->PyFloat(vm->PyBool_AS_C(args[0]) ? 1.0 : 0.0);
  199. if (args[0]->is_type(vm->tp_str)) {
  200. const Str& s = vm->PyStr_AS_C(args[0]);
  201. if(s == "inf") return vm->PyFloat(INFINITY);
  202. if(s == "-inf") return vm->PyFloat(-INFINITY);
  203. try{
  204. f64 val = std::stod(s);
  205. return vm->PyFloat(val);
  206. }catch(std::invalid_argument&){
  207. vm->ValueError("invalid literal for float(): '" + s + "'");
  208. }
  209. }
  210. vm->TypeError("float() argument must be a int, float, bool or str");
  211. return vm->None;
  212. });
  213. _vm->bind_method<0>("float", "__repr__", [](VM* vm, pkpy::Args& args) {
  214. f64 val = vm->PyFloat_AS_C(args[0]);
  215. if(std::isinf(val) || std::isnan(val)) return vm->PyStr(std::to_string(val));
  216. StrStream ss;
  217. ss << std::setprecision(std::numeric_limits<f64>::max_digits10-1) << val;
  218. std::string s = ss.str();
  219. if(std::all_of(s.begin()+1, s.end(), isdigit)) s += ".0";
  220. return vm->PyStr(s);
  221. });
  222. _vm->bind_method<0>("float", "__json__", [](VM* vm, pkpy::Args& args) {
  223. f64 val = vm->PyFloat_AS_C(args[0]);
  224. if(std::isinf(val) || std::isnan(val)) vm->ValueError("cannot jsonify 'nan' or 'inf'");
  225. return vm->PyStr(std::to_string(val));
  226. });
  227. /************ PyString ************/
  228. _vm->bind_static_method<1>("str", "__new__", CPP_LAMBDA(vm->asStr(args[0])));
  229. _vm->bind_method<1>("str", "__add__", [](VM* vm, pkpy::Args& args) {
  230. const Str& lhs = vm->PyStr_AS_C(args[0]);
  231. const Str& rhs = vm->PyStr_AS_C(args[1]);
  232. return vm->PyStr(lhs + rhs);
  233. });
  234. _vm->bind_method<0>("str", "__len__", [](VM* vm, pkpy::Args& args) {
  235. const Str& self = vm->PyStr_AS_C(args[0]);
  236. return vm->PyInt(self.u8_length());
  237. });
  238. _vm->bind_method<1>("str", "__contains__", [](VM* vm, pkpy::Args& args) {
  239. const Str& self = vm->PyStr_AS_C(args[0]);
  240. const Str& other = vm->PyStr_AS_C(args[1]);
  241. return vm->PyBool(self.find(other) != Str::npos);
  242. });
  243. _vm->bind_method<0>("str", "__str__", CPP_LAMBDA(args[0]));
  244. _vm->bind_method<0>("str", "__iter__", CPP_LAMBDA(
  245. vm->PyIter(pkpy::make_shared<BaseIter, StringIter>(vm, args[0]))
  246. ));
  247. _vm->bind_method<0>("str", "__repr__", [](VM* vm, pkpy::Args& args) {
  248. const Str& _self = vm->PyStr_AS_C(args[0]);
  249. return vm->PyStr(_self.escape(true));
  250. });
  251. _vm->bind_method<0>("str", "__json__", [](VM* vm, pkpy::Args& args) {
  252. const Str& _self = vm->PyStr_AS_C(args[0]);
  253. return vm->PyStr(_self.escape(false));
  254. });
  255. _vm->bind_method<1>("str", "__eq__", [](VM* vm, pkpy::Args& args) {
  256. if(args[0]->is_type(vm->tp_str) && args[1]->is_type(vm->tp_str))
  257. return vm->PyBool(vm->PyStr_AS_C(args[0]) == vm->PyStr_AS_C(args[1]));
  258. return vm->PyBool(args[0] == args[1]);
  259. });
  260. _vm->bind_method<1>("str", "__ne__", [](VM* vm, pkpy::Args& args) {
  261. if(args[0]->is_type(vm->tp_str) && args[1]->is_type(vm->tp_str))
  262. return vm->PyBool(vm->PyStr_AS_C(args[0]) != vm->PyStr_AS_C(args[1]));
  263. return vm->PyBool(args[0] != args[1]);
  264. });
  265. _vm->bind_method<1>("str", "__getitem__", [](VM* vm, pkpy::Args& args) {
  266. const Str& _self (vm->PyStr_AS_C(args[0]));
  267. if(args[1]->is_type(vm->tp_slice)){
  268. pkpy::Slice s = vm->PySlice_AS_C(args[1]);
  269. s.normalize(_self.u8_length());
  270. return vm->PyStr(_self.u8_substr(s.start, s.stop));
  271. }
  272. int _index = (int)vm->PyInt_AS_C(args[1]);
  273. _index = vm->normalized_index(_index, _self.u8_length());
  274. return vm->PyStr(_self.u8_getitem(_index));
  275. });
  276. _vm->bind_method<1>("str", "__gt__", [](VM* vm, pkpy::Args& args) {
  277. const Str& _self (vm->PyStr_AS_C(args[0]));
  278. const Str& _obj (vm->PyStr_AS_C(args[1]));
  279. return vm->PyBool(_self > _obj);
  280. });
  281. _vm->bind_method<1>("str", "__lt__", [](VM* vm, pkpy::Args& args) {
  282. const Str& _self (vm->PyStr_AS_C(args[0]));
  283. const Str& _obj (vm->PyStr_AS_C(args[1]));
  284. return vm->PyBool(_self < _obj);
  285. });
  286. _vm->bind_method<2>("str", "replace", [](VM* vm, pkpy::Args& args) {
  287. const Str& _self = vm->PyStr_AS_C(args[0]);
  288. const Str& _old = vm->PyStr_AS_C(args[1]);
  289. const Str& _new = vm->PyStr_AS_C(args[2]);
  290. Str _copy = _self;
  291. // replace all occurences of _old with _new in _copy
  292. size_t pos = 0;
  293. while ((pos = _copy.find(_old, pos)) != std::string::npos) {
  294. _copy.replace(pos, _old.length(), _new);
  295. pos += _new.length();
  296. }
  297. return vm->PyStr(_copy);
  298. });
  299. _vm->bind_method<1>("str", "startswith", [](VM* vm, pkpy::Args& args) {
  300. const Str& _self = vm->PyStr_AS_C(args[0]);
  301. const Str& _prefix = vm->PyStr_AS_C(args[1]);
  302. return vm->PyBool(_self.find(_prefix) == 0);
  303. });
  304. _vm->bind_method<1>("str", "endswith", [](VM* vm, pkpy::Args& args) {
  305. const Str& _self = vm->PyStr_AS_C(args[0]);
  306. const Str& _suffix = vm->PyStr_AS_C(args[1]);
  307. return vm->PyBool(_self.rfind(_suffix) == _self.length() - _suffix.length());
  308. });
  309. _vm->bind_method<1>("str", "join", [](VM* vm, pkpy::Args& args) {
  310. const Str& self = vm->PyStr_AS_C(args[0]);
  311. StrStream ss;
  312. PyVar obj = vm->asList(args[1]);
  313. const pkpy::List& list = vm->PyList_AS_C(obj);
  314. for (int i = 0; i < list.size(); ++i) {
  315. if (i > 0) ss << self;
  316. ss << vm->PyStr_AS_C(list[i]);
  317. }
  318. return vm->PyStr(ss.str());
  319. });
  320. /************ PyList ************/
  321. _vm->bind_method<1>("list", "append", [](VM* vm, pkpy::Args& args) {
  322. pkpy::List& self = vm->PyList_AS_C(args[0]);
  323. self.push_back(args[1]);
  324. return vm->None;
  325. });
  326. _vm->bind_method<0>("list", "reverse", [](VM* vm, pkpy::Args& args) {
  327. pkpy::List& self = vm->PyList_AS_C(args[0]);
  328. std::reverse(self.begin(), self.end());
  329. return vm->None;
  330. });
  331. _vm->bind_method<1>("list", "__mul__", [](VM* vm, pkpy::Args& args) {
  332. const pkpy::List& self = vm->PyList_AS_C(args[0]);
  333. int n = (int)vm->PyInt_AS_C(args[1]);
  334. pkpy::List result;
  335. result.reserve(self.size() * n);
  336. for(int i = 0; i < n; i++) result.insert(result.end(), self.begin(), self.end());
  337. return vm->PyList(std::move(result));
  338. });
  339. _vm->bind_method<2>("list", "insert", [](VM* vm, pkpy::Args& args) {
  340. pkpy::List& _self = vm->PyList_AS_C(args[0]);
  341. int index = (int)vm->PyInt_AS_C(args[1]);
  342. if(index < 0) index += _self.size();
  343. if(index < 0) index = 0;
  344. if(index > _self.size()) index = _self.size();
  345. _self.insert(_self.begin() + index, args[2]);
  346. return vm->None;
  347. });
  348. _vm->bind_method<0>("list", "clear", [](VM* vm, pkpy::Args& args) {
  349. vm->PyList_AS_C(args[0]).clear();
  350. return vm->None;
  351. });
  352. _vm->bind_method<0>("list", "copy", CPP_LAMBDA(vm->PyList(vm->PyList_AS_C(args[0]))));
  353. _vm->bind_method<1>("list", "__add__", [](VM* vm, pkpy::Args& args) {
  354. const pkpy::List& self = vm->PyList_AS_C(args[0]);
  355. const pkpy::List& obj = vm->PyList_AS_C(args[1]);
  356. pkpy::List new_list = self;
  357. new_list.insert(new_list.end(), obj.begin(), obj.end());
  358. return vm->PyList(new_list);
  359. });
  360. _vm->bind_method<0>("list", "__len__", [](VM* vm, pkpy::Args& args) {
  361. const pkpy::List& self = vm->PyList_AS_C(args[0]);
  362. return vm->PyInt(self.size());
  363. });
  364. _vm->bind_method<0>("list", "__iter__", [](VM* vm, pkpy::Args& args) {
  365. return vm->PyIter(pkpy::make_shared<BaseIter, ArrayIter<pkpy::List>>(vm, args[0]));
  366. });
  367. _vm->bind_method<1>("list", "__getitem__", [](VM* vm, pkpy::Args& args) {
  368. const pkpy::List& self = vm->PyList_AS_C(args[0]);
  369. if(args[1]->is_type(vm->tp_slice)){
  370. pkpy::Slice s = vm->PySlice_AS_C(args[1]);
  371. s.normalize(self.size());
  372. pkpy::List new_list;
  373. for(size_t i = s.start; i < s.stop; i++) new_list.push_back(self[i]);
  374. return vm->PyList(std::move(new_list));
  375. }
  376. int index = (int)vm->PyInt_AS_C(args[1]);
  377. index = vm->normalized_index(index, self.size());
  378. return self[index];
  379. });
  380. _vm->bind_method<2>("list", "__setitem__", [](VM* vm, pkpy::Args& args) {
  381. pkpy::List& self = vm->PyList_AS_C(args[0]);
  382. int index = (int)vm->PyInt_AS_C(args[1]);
  383. index = vm->normalized_index(index, self.size());
  384. self[index] = args[2];
  385. return vm->None;
  386. });
  387. _vm->bind_method<1>("list", "__delitem__", [](VM* vm, pkpy::Args& args) {
  388. pkpy::List& self = vm->PyList_AS_C(args[0]);
  389. int index = (int)vm->PyInt_AS_C(args[1]);
  390. index = vm->normalized_index(index, self.size());
  391. self.erase(self.begin() + index);
  392. return vm->None;
  393. });
  394. /************ PyTuple ************/
  395. _vm->bind_static_method<1>("tuple", "__new__", [](VM* vm, pkpy::Args& args) {
  396. pkpy::List list = vm->PyList_AS_C(vm->asList(args[0]));
  397. return vm->PyTuple(std::move(list));
  398. });
  399. _vm->bind_method<0>("tuple", "__iter__", [](VM* vm, pkpy::Args& args) {
  400. return vm->PyIter(pkpy::make_shared<BaseIter, ArrayIter<pkpy::Args>>(vm, args[0]));
  401. });
  402. _vm->bind_method<1>("tuple", "__getitem__", [](VM* vm, pkpy::Args& args) {
  403. const pkpy::Tuple& self = vm->PyTuple_AS_C(args[0]);
  404. if(args[1]->is_type(vm->tp_slice)){
  405. pkpy::Slice s = vm->PySlice_AS_C(args[1]);
  406. s.normalize(self.size());
  407. pkpy::List new_list;
  408. for(size_t i = s.start; i < s.stop; i++) new_list.push_back(self[i]);
  409. return vm->PyTuple(std::move(new_list));
  410. }
  411. int index = (int)vm->PyInt_AS_C(args[1]);
  412. index = vm->normalized_index(index, self.size());
  413. return self[index];
  414. });
  415. _vm->bind_method<0>("tuple", "__len__", [](VM* vm, pkpy::Args& args) {
  416. const pkpy::Tuple& self = vm->PyTuple_AS_C(args[0]);
  417. return vm->PyInt(self.size());
  418. });
  419. /************ PyBool ************/
  420. _vm->bind_static_method<1>("bool", "__new__", CPP_LAMBDA(vm->asBool(args[0])));
  421. _vm->bind_method<0>("bool", "__repr__", [](VM* vm, pkpy::Args& args) {
  422. bool val = vm->PyBool_AS_C(args[0]);
  423. return vm->PyStr(val ? "True" : "False");
  424. });
  425. _vm->bind_method<0>("bool", "__json__", [](VM* vm, pkpy::Args& args) {
  426. bool val = vm->PyBool_AS_C(args[0]);
  427. return vm->PyStr(val ? "true" : "false");
  428. });
  429. _vm->bind_method<1>("bool", "__xor__", [](VM* vm, pkpy::Args& args) {
  430. bool self = vm->PyBool_AS_C(args[0]);
  431. bool other = vm->PyBool_AS_C(args[1]);
  432. return vm->PyBool(self ^ other);
  433. });
  434. _vm->bind_method<0>("ellipsis", "__repr__", CPP_LAMBDA(vm->PyStr("Ellipsis")));
  435. }
  436. #include "builtins.h"
  437. #ifdef _WIN32
  438. #define __EXPORT __declspec(dllexport)
  439. #elif __APPLE__
  440. #define __EXPORT __attribute__((visibility("default"))) __attribute__((used))
  441. #elif __EMSCRIPTEN__
  442. #include <emscripten.h>
  443. #define __EXPORT EMSCRIPTEN_KEEPALIVE
  444. #else
  445. #define __EXPORT
  446. #endif
  447. void add_module_time(VM* vm){
  448. PyVar mod = vm->new_module("time");
  449. vm->bind_func<0>(mod, "time", [](VM* vm, pkpy::Args& args) {
  450. auto now = std::chrono::high_resolution_clock::now();
  451. return vm->PyFloat(std::chrono::duration_cast<std::chrono::microseconds>(now.time_since_epoch()).count() / 1000000.0);
  452. });
  453. }
  454. void add_module_sys(VM* vm){
  455. PyVar mod = vm->new_module("sys");
  456. vm->setattr(mod, "version", vm->PyStr(PK_VERSION));
  457. vm->bind_func<1>(mod, "getrefcount", CPP_LAMBDA(vm->PyInt(args[0].use_count())));
  458. vm->bind_func<0>(mod, "getrecursionlimit", CPP_LAMBDA(vm->PyInt(vm->recursionlimit)));
  459. vm->bind_func<1>(mod, "setrecursionlimit", [](VM* vm, pkpy::Args& args) {
  460. vm->recursionlimit = (int)vm->PyInt_AS_C(args[0]);
  461. return vm->None;
  462. });
  463. }
  464. void add_module_json(VM* vm){
  465. PyVar mod = vm->new_module("json");
  466. vm->bind_func<1>(mod, "loads", [](VM* vm, pkpy::Args& args) {
  467. const Str& expr = vm->PyStr_AS_C(args[0]);
  468. CodeObject_ code = vm->compile(expr, "<json>", JSON_MODE);
  469. return vm->_exec(code, vm->top_frame()->_module, vm->top_frame()->_locals);
  470. });
  471. vm->bind_func<1>(mod, "dumps", CPP_LAMBDA(vm->call(args[0], __json__)));
  472. }
  473. void add_module_math(VM* vm){
  474. PyVar mod = vm->new_module("math");
  475. vm->setattr(mod, "pi", vm->PyFloat(3.1415926535897932384));
  476. vm->setattr(mod, "e" , vm->PyFloat(2.7182818284590452354));
  477. vm->bind_func<1>(mod, "log", CPP_LAMBDA(vm->PyFloat(std::log(vm->num_to_float(args[0])))));
  478. vm->bind_func<1>(mod, "log10", CPP_LAMBDA(vm->PyFloat(std::log10(vm->num_to_float(args[0])))));
  479. vm->bind_func<1>(mod, "log2", CPP_LAMBDA(vm->PyFloat(std::log2(vm->num_to_float(args[0])))));
  480. vm->bind_func<1>(mod, "sin", CPP_LAMBDA(vm->PyFloat(std::sin(vm->num_to_float(args[0])))));
  481. vm->bind_func<1>(mod, "cos", CPP_LAMBDA(vm->PyFloat(std::cos(vm->num_to_float(args[0])))));
  482. vm->bind_func<1>(mod, "tan", CPP_LAMBDA(vm->PyFloat(std::tan(vm->num_to_float(args[0])))));
  483. vm->bind_func<1>(mod, "isnan", CPP_LAMBDA(vm->PyBool(std::isnan(vm->num_to_float(args[0])))));
  484. vm->bind_func<1>(mod, "isinf", CPP_LAMBDA(vm->PyBool(std::isinf(vm->num_to_float(args[0])))));
  485. vm->bind_func<1>(mod, "fabs", CPP_LAMBDA(vm->PyFloat(std::fabs(vm->num_to_float(args[0])))));
  486. vm->bind_func<1>(mod, "floor", CPP_LAMBDA(vm->PyInt((i64)std::floor(vm->num_to_float(args[0])))));
  487. vm->bind_func<1>(mod, "ceil", CPP_LAMBDA(vm->PyInt((i64)std::ceil(vm->num_to_float(args[0])))));
  488. vm->bind_func<1>(mod, "sqrt", CPP_LAMBDA(vm->PyFloat(std::sqrt(vm->num_to_float(args[0])))));
  489. }
  490. void add_module_dis(VM* vm){
  491. PyVar mod = vm->new_module("dis");
  492. vm->bind_func<1>(mod, "dis", [](VM* vm, pkpy::Args& args) {
  493. CodeObject_ code = vm->PyFunction_AS_C(args[0])->code;
  494. (*vm->_stdout) << vm->disassemble(code);
  495. return vm->None;
  496. });
  497. }
  498. #include <fstream>
  499. struct FileIO {
  500. PY_CLASS(io, FileIO)
  501. Str file;
  502. Str mode;
  503. std::fstream _fs;
  504. FileIO(VM* vm, Str file, Str mode): file(file), mode(mode) {
  505. if(mode == "rt" || mode == "r"){
  506. _fs.open(file, std::ios::in);
  507. }else if(mode == "wt" || mode == "w"){
  508. _fs.open(file, std::ios::out);
  509. }else if(mode == "at" || mode == "a"){
  510. _fs.open(file, std::ios::app);
  511. }
  512. if(!_fs.is_open()) vm->IOError(strerror(errno));
  513. }
  514. static void _register(VM* vm, PyVar mod, PyVar type){
  515. vm->bind_static_method<2>(type, "__new__", [](VM* vm, pkpy::Args& args){
  516. return vm->new_object<FileIO>(
  517. vm, vm->PyStr_AS_C(args[0]), vm->PyStr_AS_C(args[1])
  518. );
  519. });
  520. vm->bind_method<0>(type, "read", [](VM* vm, pkpy::Args& args){
  521. FileIO& io = vm->py_cast<FileIO>(args[0]);
  522. std::string buffer;
  523. io._fs >> buffer;
  524. return vm->PyStr(buffer);
  525. });
  526. vm->bind_method<1>(type, "write", [](VM* vm, pkpy::Args& args){
  527. FileIO& io = vm->py_cast<FileIO>(args[0]);
  528. io._fs << vm->PyStr_AS_C(args[1]);
  529. return vm->None;
  530. });
  531. vm->bind_method<0>(type, "close", [](VM* vm, pkpy::Args& args){
  532. FileIO& io = vm->py_cast<FileIO>(args[0]);
  533. io._fs.close();
  534. return vm->None;
  535. });
  536. vm->bind_method<0>(type, "__exit__", [](VM* vm, pkpy::Args& args){
  537. FileIO& io = vm->py_cast<FileIO>(args[0]);
  538. io._fs.close();
  539. return vm->None;
  540. });
  541. vm->bind_method<0>(type, "__enter__", CPP_LAMBDA(vm->None));
  542. }
  543. };
  544. void add_module_io(VM* vm){
  545. PyVar mod = vm->new_module("io");
  546. PyVar type = vm->register_class<FileIO>(mod);
  547. vm->bind_builtin_func<2>("open", [type](VM* vm, const pkpy::Args& args){
  548. return vm->call(type, args);
  549. });
  550. }
  551. void add_module_os(VM* vm){}
  552. struct ReMatch {
  553. PY_CLASS(re, Match)
  554. i64 start;
  555. i64 end;
  556. std::smatch m;
  557. ReMatch(i64 start, i64 end, std::smatch m) : start(start), end(end), m(m) {}
  558. static void _register(VM* vm, PyVar mod, PyVar type){
  559. vm->bind_method<-1>(type, "__init__", CPP_NOT_IMPLEMENTED());
  560. vm->bind_method<0>(type, "start", CPP_LAMBDA(vm->PyInt(vm->py_cast<ReMatch>(args[0]).start)));
  561. vm->bind_method<0>(type, "end", CPP_LAMBDA(vm->PyInt(vm->py_cast<ReMatch>(args[0]).end)));
  562. vm->bind_method<0>(type, "span", [](VM* vm, pkpy::Args& args) {
  563. auto& self = vm->py_cast<ReMatch>(args[0]);
  564. return vm->PyTuple({ vm->PyInt(self.start), vm->PyInt(self.end) });
  565. });
  566. vm->bind_method<1>(type, "group", [](VM* vm, pkpy::Args& args) {
  567. auto& self = vm->py_cast<ReMatch>(args[0]);
  568. int index = (int)vm->PyInt_AS_C(args[1]);
  569. index = vm->normalized_index(index, self.m.size());
  570. return vm->PyStr(self.m[index].str());
  571. });
  572. }
  573. };
  574. PyVar _regex_search(const Str& pattern, const Str& string, bool fromStart, VM* vm){
  575. std::regex re(pattern);
  576. std::smatch m;
  577. if(std::regex_search(string, m, re)){
  578. if(fromStart && m.position() != 0) return vm->None;
  579. i64 start = string._to_u8_index(m.position());
  580. i64 end = string._to_u8_index(m.position() + m.length());
  581. return vm->new_object<ReMatch>(start, end, m);
  582. }
  583. return vm->None;
  584. };
  585. void add_module_re(VM* vm){
  586. PyVar mod = vm->new_module("re");
  587. vm->register_class<ReMatch>(mod);
  588. vm->bind_func<2>(mod, "match", [](VM* vm, pkpy::Args& args) {
  589. const Str& pattern = vm->PyStr_AS_C(args[0]);
  590. const Str& string = vm->PyStr_AS_C(args[1]);
  591. return _regex_search(pattern, string, true, vm);
  592. });
  593. vm->bind_func<2>(mod, "search", [](VM* vm, pkpy::Args& args) {
  594. const Str& pattern = vm->PyStr_AS_C(args[0]);
  595. const Str& string = vm->PyStr_AS_C(args[1]);
  596. return _regex_search(pattern, string, false, vm);
  597. });
  598. vm->bind_func<3>(mod, "sub", [](VM* vm, pkpy::Args& args) {
  599. const Str& pattern = vm->PyStr_AS_C(args[0]);
  600. const Str& repl = vm->PyStr_AS_C(args[1]);
  601. const Str& string = vm->PyStr_AS_C(args[2]);
  602. std::regex re(pattern);
  603. return vm->PyStr(std::regex_replace(string, re, repl));
  604. });
  605. vm->bind_func<2>(mod, "split", [](VM* vm, pkpy::Args& args) {
  606. const Str& pattern = vm->PyStr_AS_C(args[0]);
  607. const Str& string = vm->PyStr_AS_C(args[1]);
  608. std::regex re(pattern);
  609. std::sregex_token_iterator it(string.begin(), string.end(), re, -1);
  610. std::sregex_token_iterator end;
  611. pkpy::List vec;
  612. for(; it != end; ++it){
  613. vec.push_back(vm->PyStr(it->str()));
  614. }
  615. return vm->PyList(vec);
  616. });
  617. }
  618. void add_module_random(VM* vm){
  619. PyVar mod = vm->new_module("random");
  620. std::srand(std::time(0));
  621. vm->bind_func<1>(mod, "seed", [](VM* vm, pkpy::Args& args) {
  622. std::srand((unsigned int)vm->PyInt_AS_C(args[0]));
  623. return vm->None;
  624. });
  625. vm->bind_func<0>(mod, "random", CPP_LAMBDA(vm->PyFloat((f64)std::rand() / RAND_MAX)));
  626. vm->bind_func<2>(mod, "randint", [](VM* vm, pkpy::Args& args) {
  627. i64 a = vm->PyInt_AS_C(args[0]);
  628. i64 b = vm->PyInt_AS_C(args[1]);
  629. if(a > b) std::swap(a, b);
  630. return vm->PyInt(a + std::rand() % (b - a + 1));
  631. });
  632. vm->bind_func<2>(mod, "uniform", [](VM* vm, pkpy::Args& args) {
  633. f64 a = vm->PyFloat_AS_C(args[0]);
  634. f64 b = vm->PyFloat_AS_C(args[1]);
  635. if(a > b) std::swap(a, b);
  636. return vm->PyFloat(a + (b - a) * std::rand() / RAND_MAX);
  637. });
  638. CodeObject_ code = vm->compile(kRandomCode, "random.py", EXEC_MODE);
  639. vm->_exec(code, mod, pkpy::make_shared<pkpy::NameDict>());
  640. }
  641. class _PkExported{
  642. public:
  643. virtual ~_PkExported() = default;
  644. virtual void* get() = 0;
  645. };
  646. static std::vector<_PkExported*> _pk_lookup_table;
  647. template<typename T>
  648. class PkExported : public _PkExported{
  649. T* _ptr;
  650. public:
  651. template<typename... Args>
  652. PkExported(Args&&... args) {
  653. _ptr = new T(std::forward<Args>(args)...);
  654. _pk_lookup_table.push_back(this);
  655. }
  656. ~PkExported() override { delete _ptr; }
  657. void* get() override { return _ptr; }
  658. operator T*() { return _ptr; }
  659. };
  660. #define PKPY_ALLOCATE(T, ...) *(new PkExported<T>(__VA_ARGS__))
  661. extern "C" {
  662. __EXPORT
  663. /// Delete a pointer allocated by `pkpy_xxx_xxx`.
  664. /// It can be `VM*`, `REPL*`, `char*`, etc.
  665. ///
  666. /// !!!
  667. /// If the pointer is not allocated by `pkpy_xxx_xxx`, the behavior is undefined.
  668. /// !!!
  669. void pkpy_delete(void* p){
  670. for(int i = 0; i < _pk_lookup_table.size(); i++){
  671. if(_pk_lookup_table[i]->get() == p){
  672. delete _pk_lookup_table[i];
  673. _pk_lookup_table.erase(_pk_lookup_table.begin() + i);
  674. return;
  675. }
  676. }
  677. free(p);
  678. }
  679. __EXPORT
  680. /// Run a given source on a virtual machine.
  681. void pkpy_vm_exec(VM* vm, const char* source){
  682. vm->exec(source, "main.py", EXEC_MODE);
  683. }
  684. __EXPORT
  685. /// Get a global variable of a virtual machine.
  686. ///
  687. /// Return `__repr__` of the result.
  688. /// If the variable is not found, return `nullptr`.
  689. char* pkpy_vm_get_global(VM* vm, const char* name){
  690. auto it = vm->_main->attr().find(name);
  691. if(it == vm->_main->attr().end()) return nullptr;
  692. try{
  693. Str _repr = vm->PyStr_AS_C(vm->asRepr(it->second));
  694. return strdup(_repr.c_str());
  695. }catch(...){
  696. return nullptr;
  697. }
  698. }
  699. __EXPORT
  700. /// Evaluate an expression.
  701. ///
  702. /// Return `__repr__` of the result.
  703. /// If there is any error, return `nullptr`.
  704. char* pkpy_vm_eval(VM* vm, const char* source){
  705. PyVarOrNull ret = vm->exec(source, "<eval>", EVAL_MODE);
  706. if(ret == nullptr) return nullptr;
  707. try{
  708. Str _repr = vm->PyStr_AS_C(vm->asRepr(ret));
  709. return strdup(_repr.c_str());
  710. }catch(...){
  711. return nullptr;
  712. }
  713. }
  714. __EXPORT
  715. /// Create a REPL, using the given virtual machine as the backend.
  716. REPL* pkpy_new_repl(VM* vm){
  717. return PKPY_ALLOCATE(REPL, vm);
  718. }
  719. __EXPORT
  720. /// Input a source line to an interactive console. Return true if need more lines.
  721. bool pkpy_repl_input(REPL* r, const char* line){
  722. return r->input(line);
  723. }
  724. __EXPORT
  725. /// Add a source module into a virtual machine.
  726. void pkpy_vm_add_module(VM* vm, const char* name, const char* source){
  727. vm->_lazy_modules[name] = source;
  728. }
  729. __EXPORT
  730. /// Create a virtual machine.
  731. VM* pkpy_new_vm(bool use_stdio){
  732. VM* vm = PKPY_ALLOCATE(VM, use_stdio);
  733. init_builtins(vm);
  734. add_module_sys(vm);
  735. add_module_time(vm);
  736. add_module_json(vm);
  737. add_module_math(vm);
  738. add_module_re(vm);
  739. add_module_dis(vm);
  740. add_module_random(vm);
  741. add_module_io(vm);
  742. add_module_os(vm);
  743. CodeObject_ code = vm->compile(kBuiltinsCode, "<builtins>", EXEC_MODE);
  744. vm->_exec(code, vm->builtins, pkpy::make_shared<pkpy::NameDict>());
  745. return vm;
  746. }
  747. __EXPORT
  748. /// Read the standard output and standard error as string of a virtual machine.
  749. /// The `vm->use_stdio` should be `false`.
  750. /// After this operation, both stream will be cleared.
  751. ///
  752. /// Return a json representing the result.
  753. char* pkpy_vm_read_output(VM* vm){
  754. if(vm->use_stdio) return nullptr;
  755. StrStream* s_out = (StrStream*)(vm->_stdout);
  756. StrStream* s_err = (StrStream*)(vm->_stderr);
  757. Str _stdout = s_out->str();
  758. Str _stderr = s_err->str();
  759. StrStream ss;
  760. ss << '{' << "\"stdout\": " << _stdout.escape(false);
  761. ss << ", " << "\"stderr\": " << _stderr.escape(false) << '}';
  762. s_out->str(""); s_err->str("");
  763. return strdup(ss.str().c_str());
  764. }
  765. typedef i64 (*f_int_t)(char*);
  766. typedef f64 (*f_float_t)(char*);
  767. typedef bool (*f_bool_t)(char*);
  768. typedef char* (*f_str_t)(char*);
  769. typedef void (*f_None_t)(char*);
  770. static f_int_t f_int = nullptr;
  771. static f_float_t f_float = nullptr;
  772. static f_bool_t f_bool = nullptr;
  773. static f_str_t f_str = nullptr;
  774. static f_None_t f_None = nullptr;
  775. __EXPORT
  776. /// Setup the callback functions.
  777. 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){
  778. ::f_int = f_int;
  779. ::f_float = f_float;
  780. ::f_bool = f_bool;
  781. ::f_str = f_str;
  782. ::f_None = f_None;
  783. }
  784. __EXPORT
  785. /// Bind a function to a virtual machine.
  786. char* pkpy_vm_bind(VM* vm, const char* mod, const char* name, int ret_code){
  787. if(!f_int || !f_float || !f_bool || !f_str || !f_None) return nullptr;
  788. static int kGlobalBindId = 0;
  789. for(int i=0; mod[i]; i++) if(mod[i] == ' ') return nullptr;
  790. for(int i=0; name[i]; i++) if(name[i] == ' ') return nullptr;
  791. std::string f_header = std::string(mod) + '.' + name + '#' + std::to_string(kGlobalBindId++);
  792. PyVar obj = vm->_modules.contains(mod) ? vm->_modules[mod] : vm->new_module(mod);
  793. vm->bind_func<-1>(obj, name, [ret_code, f_header](VM* vm, const pkpy::Args& args){
  794. StrStream ss;
  795. ss << f_header;
  796. for(int i=0; i<args.size(); i++){
  797. ss << ' ';
  798. PyVar x = vm->call(args[i], __json__);
  799. ss << vm->PyStr_AS_C(x);
  800. }
  801. char* packet = strdup(ss.str().c_str());
  802. switch(ret_code){
  803. case 'i': return vm->PyInt(f_int(packet));
  804. case 'f': return vm->PyFloat(f_float(packet));
  805. case 'b': return vm->PyBool(f_bool(packet));
  806. case 's': {
  807. char* p = f_str(packet);
  808. if(p == nullptr) return vm->None;
  809. return vm->PyStr(p); // no need to free(p)
  810. }
  811. case 'N': f_None(packet); return vm->None;
  812. }
  813. free(packet);
  814. UNREACHABLE();
  815. return vm->None;
  816. });
  817. return strdup(f_header.c_str());
  818. }
  819. }