| 123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107 |
- #pragma once
- #include "codeobject.h"
- #include "common.h"
- #include "expr.h"
- #include "obj.h"
- namespace pkpy{
- class Compiler;
- typedef void (Compiler::*PrattCallback)();
- struct PrattRule{
- PrattCallback prefix;
- PrattCallback infix;
- Precedence precedence;
- };
- class Compiler {
- inline static PrattRule rules[kTokenCount];
- std::unique_ptr<Lexer> lexer;
- stack<CodeEmitContext> contexts;
- VM* vm;
- bool unknown_global_scope; // for eval/exec() call
- bool used;
- // for parsing token stream
- int i = 0;
- std::vector<Token> tokens;
- const Token& prev() const{ return tokens.at(i-1); }
- const Token& curr() const{ return tokens.at(i); }
- const Token& next() const{ return tokens.at(i+1); }
- const Token& err() const{
- if(i >= tokens.size()) return prev();
- return curr();
- }
- void advance(int delta=1) { i += delta; }
- CodeEmitContext* ctx() { return &contexts.top(); }
- CompileMode mode() const{ return lexer->src->mode; }
- NameScope name_scope() const {
- auto s = contexts.size()>1 ? NAME_LOCAL : NAME_GLOBAL;
- if(unknown_global_scope && s == NAME_GLOBAL) s = NAME_GLOBAL_UNKNOWN;
- return s;
- }
- CodeObject_ push_global_context(){
- CodeObject_ co = make_sp<CodeObject>(lexer->src, lexer->src->filename);
- contexts.push(CodeEmitContext(vm, co, contexts.size()));
- return co;
- }
- FuncDecl_ push_f_context(Str name){
- FuncDecl_ decl = make_sp<FuncDecl>();
- decl->code = make_sp<CodeObject>(lexer->src, name);
- decl->nested = name_scope() == NAME_LOCAL;
- contexts.push(CodeEmitContext(vm, decl->code, contexts.size()));
- return decl;
- }
- void pop_context(){
- if(!ctx()->s_expr.empty()){
- throw std::runtime_error("!ctx()->s_expr.empty()\n" + ctx()->_log_s_expr());
- }
- // add a `return None` in the end as a guard
- // previously, we only do this if the last opcode is not a return
- // however, this is buggy...since there may be a jump to the end (out of bound) even if the last opcode is a return
- ctx()->emit(OP_LOAD_NONE, BC_NOARG, BC_KEEPLINE);
- ctx()->emit(OP_RETURN_VALUE, BC_NOARG, BC_KEEPLINE);
- // ctx()->co->optimize(vm);
- if(ctx()->co->varnames.size() > PK_MAX_CO_VARNAMES){
- SyntaxError("maximum number of local variables exceeded");
- }
- contexts.pop();
- }
- static void init_pratt_rules(){
- if(rules[TK(".")].precedence != PREC_NONE) return;
- // http://journal.stuffwithstuff.com/2011/03/19/pratt-parsers-expression-parsing-made-easy/
- #define METHOD(name) &Compiler::name
- #define NO_INFIX nullptr, PREC_NONE
- for(TokenIndex i=0; i<kTokenCount; i++) rules[i] = { nullptr, NO_INFIX };
- rules[TK(".")] = { nullptr, METHOD(exprAttrib), PREC_ATTRIB };
- rules[TK("(")] = { METHOD(exprGroup), METHOD(exprCall), PREC_CALL };
- rules[TK("[")] = { METHOD(exprList), METHOD(exprSubscr), PREC_SUBSCRIPT };
- rules[TK("{")] = { METHOD(exprMap), NO_INFIX };
- rules[TK("%")] = { nullptr, METHOD(exprBinaryOp), PREC_FACTOR };
- rules[TK("+")] = { nullptr, METHOD(exprBinaryOp), PREC_TERM };
- rules[TK("-")] = { METHOD(exprUnaryOp), METHOD(exprBinaryOp), PREC_TERM };
- rules[TK("*")] = { METHOD(exprUnaryOp), METHOD(exprBinaryOp), PREC_FACTOR };
- rules[TK("/")] = { nullptr, METHOD(exprBinaryOp), PREC_FACTOR };
- rules[TK("//")] = { nullptr, METHOD(exprBinaryOp), PREC_FACTOR };
- rules[TK("**")] = { METHOD(exprUnaryOp), METHOD(exprBinaryOp), PREC_EXPONENT };
- rules[TK(">")] = { nullptr, METHOD(exprBinaryOp), PREC_COMPARISION };
- rules[TK("<")] = { nullptr, METHOD(exprBinaryOp), PREC_COMPARISION };
- rules[TK("==")] = { nullptr, METHOD(exprBinaryOp), PREC_COMPARISION };
- rules[TK("!=")] = { nullptr, METHOD(exprBinaryOp), PREC_COMPARISION };
- rules[TK(">=")] = { nullptr, METHOD(exprBinaryOp), PREC_COMPARISION };
- rules[TK("<=")] = { nullptr, METHOD(exprBinaryOp), PREC_COMPARISION };
- rules[TK("in")] = { nullptr, METHOD(exprBinaryOp), PREC_COMPARISION };
- rules[TK("is")] = { nullptr, METHOD(exprBinaryOp), PREC_COMPARISION };
- rules[TK("<<")] = { nullptr, METHOD(exprBinaryOp), PREC_BITWISE_SHIFT };
- rules[TK(">>")] = { nullptr, METHOD(exprBinaryOp), PREC_BITWISE_SHIFT };
- rules[TK("&")] = { nullptr, METHOD(exprBinaryOp), PREC_BITWISE_AND };
- rules[TK("|")] = { nullptr, METHOD(exprBinaryOp), PREC_BITWISE_OR };
- rules[TK("^")] = { nullptr, METHOD(exprBinaryOp), PREC_BITWISE_XOR };
- rules[TK("@")] = { nullptr, METHOD(exprBinaryOp), PREC_FACTOR };
- rules[TK("if")] = { nullptr, METHOD(exprTernary), PREC_TERNARY };
- rules[TK(",")] = { nullptr, METHOD(exprTuple), PREC_TUPLE };
- rules[TK("not in")] = { nullptr, METHOD(exprBinaryOp), PREC_COMPARISION };
- rules[TK("is not")] = { nullptr, METHOD(exprBinaryOp), PREC_COMPARISION };
- rules[TK("and") ] = { nullptr, METHOD(exprAnd), PREC_LOGICAL_AND };
- rules[TK("or")] = { nullptr, METHOD(exprOr), PREC_LOGICAL_OR };
- rules[TK("not")] = { METHOD(exprNot), nullptr, PREC_LOGICAL_NOT };
- rules[TK("True")] = { METHOD(exprLiteral0), NO_INFIX };
- rules[TK("False")] = { METHOD(exprLiteral0), NO_INFIX };
- rules[TK("None")] = { METHOD(exprLiteral0), NO_INFIX };
- rules[TK("...")] = { METHOD(exprLiteral0), NO_INFIX };
- rules[TK("lambda")] = { METHOD(exprLambda), NO_INFIX };
- rules[TK("@id")] = { METHOD(exprName), NO_INFIX };
- rules[TK("@num")] = { METHOD(exprLiteral), NO_INFIX };
- rules[TK("@str")] = { METHOD(exprLiteral), NO_INFIX };
- rules[TK("@fstr")] = { METHOD(exprFString), NO_INFIX };
- rules[TK("@long")] = { METHOD(exprLong), NO_INFIX };
- #undef METHOD
- #undef NO_INFIX
- }
- bool match(TokenIndex expected) {
- if (curr().type != expected) return false;
- advance();
- return true;
- }
- void consume(TokenIndex expected) {
- if (!match(expected)){
- SyntaxError(
- fmt("expected '", TK_STR(expected), "', got '", TK_STR(curr().type), "'")
- );
- }
- }
- bool match_newlines_repl(){
- return match_newlines(mode()==REPL_MODE);
- }
- bool match_newlines(bool repl_throw=false) {
- bool consumed = false;
- if (curr().type == TK("@eol")) {
- while (curr().type == TK("@eol")) advance();
- consumed = true;
- }
- if (repl_throw && curr().type == TK("@eof")){
- throw NeedMoreLines(ctx()->is_compiling_class);
- }
- return consumed;
- }
- bool match_end_stmt() {
- if (match(TK(";"))) { match_newlines(); return true; }
- if (match_newlines() || curr().type == TK("@eof")) return true;
- if (curr().type == TK("@dedent")) return true;
- return false;
- }
- void consume_end_stmt() {
- if (!match_end_stmt()) SyntaxError("expected statement end");
- }
- /*************************************************/
- void EXPR(bool push_stack=true) {
- parse_expression(PREC_TUPLE+1, push_stack);
- }
- void EXPR_TUPLE(bool push_stack=true) {
- parse_expression(PREC_TUPLE, push_stack);
- }
- // special case for `for loop` and `comp`
- Expr_ EXPR_VARS(){
- std::vector<Expr_> items;
- do {
- consume(TK("@id"));
- items.push_back(make_expr<NameExpr>(prev().str(), name_scope()));
- } while(match(TK(",")));
- if(items.size()==1) return std::move(items[0]);
- return make_expr<TupleExpr>(std::move(items));
- }
- template <typename T, typename... Args>
- std::unique_ptr<T> make_expr(Args&&... args) {
- std::unique_ptr<T> expr = std::make_unique<T>(std::forward<Args>(args)...);
- expr->line = prev().line;
- return expr;
- }
- void exprLiteral(){
- ctx()->s_expr.push(make_expr<LiteralExpr>(prev().value));
- }
- void exprLong(){
- ctx()->s_expr.push(make_expr<LongExpr>(prev().str()));
- }
- void exprFString(){
- ctx()->s_expr.push(make_expr<FStringExpr>(std::get<Str>(prev().value)));
- }
- void exprLambda(){
- FuncDecl_ decl = push_f_context("<lambda>");
- auto e = make_expr<LambdaExpr>(decl);
- if(!match(TK(":"))){
- _compile_f_args(e->decl, false);
- consume(TK(":"));
- }
- // https://github.com/blueloveTH/pocketpy/issues/37
- parse_expression(PREC_LAMBDA + 1, false);
- ctx()->emit(OP_RETURN_VALUE, BC_NOARG, BC_KEEPLINE);
- pop_context();
- ctx()->s_expr.push(std::move(e));
- }
- void exprTuple(){
- std::vector<Expr_> items;
- items.push_back(ctx()->s_expr.popx());
- do {
- if(curr().brackets_level) match_newlines_repl();
- if(!is_expression()) break;
- EXPR();
- items.push_back(ctx()->s_expr.popx());
- if(curr().brackets_level) match_newlines_repl();
- } while(match(TK(",")));
- ctx()->s_expr.push(make_expr<TupleExpr>(
- std::move(items)
- ));
- }
- void exprOr(){
- auto e = make_expr<OrExpr>();
- e->lhs = ctx()->s_expr.popx();
- parse_expression(PREC_LOGICAL_OR + 1);
- e->rhs = ctx()->s_expr.popx();
- ctx()->s_expr.push(std::move(e));
- }
-
- void exprAnd(){
- auto e = make_expr<AndExpr>();
- e->lhs = ctx()->s_expr.popx();
- parse_expression(PREC_LOGICAL_AND + 1);
- e->rhs = ctx()->s_expr.popx();
- ctx()->s_expr.push(std::move(e));
- }
-
- void exprTernary(){
- auto e = make_expr<TernaryExpr>();
- e->true_expr = ctx()->s_expr.popx();
- // cond
- parse_expression(PREC_TERNARY + 1);
- e->cond = ctx()->s_expr.popx();
- consume(TK("else"));
- // if false
- parse_expression(PREC_TERNARY + 1);
- e->false_expr = ctx()->s_expr.popx();
- ctx()->s_expr.push(std::move(e));
- }
-
- void exprBinaryOp(){
- auto e = make_expr<BinaryExpr>();
- e->op = prev().type;
- e->lhs = ctx()->s_expr.popx();
- parse_expression(rules[e->op].precedence + 1);
- e->rhs = ctx()->s_expr.popx();
- ctx()->s_expr.push(std::move(e));
- }
- void exprNot() {
- parse_expression(PREC_LOGICAL_NOT + 1);
- ctx()->s_expr.push(make_expr<NotExpr>(ctx()->s_expr.popx()));
- }
-
- void exprUnaryOp(){
- TokenIndex op = prev().type;
- parse_expression(PREC_UNARY + 1);
- switch(op){
- case TK("-"):
- ctx()->s_expr.push(make_expr<NegatedExpr>(ctx()->s_expr.popx()));
- break;
- case TK("*"):
- ctx()->s_expr.push(make_expr<StarredExpr>(1, ctx()->s_expr.popx()));
- break;
- case TK("**"):
- ctx()->s_expr.push(make_expr<StarredExpr>(2, ctx()->s_expr.popx()));
- break;
- default: FATAL_ERROR();
- }
- }
- void exprGroup(){
- match_newlines_repl();
- EXPR_TUPLE(); // () is just for change precedence
- match_newlines_repl();
- consume(TK(")"));
- if(ctx()->s_expr.top()->is_tuple()) return;
- Expr_ g = make_expr<GroupedExpr>(ctx()->s_expr.popx());
- ctx()->s_expr.push(std::move(g));
- }
- template<typename T>
- void _consume_comp(Expr_ expr){
- static_assert(std::is_base_of<CompExpr, T>::value);
- std::unique_ptr<CompExpr> ce = make_expr<T>();
- ce->expr = std::move(expr);
- ce->vars = EXPR_VARS();
- consume(TK("in"));
- parse_expression(PREC_TERNARY + 1);
- ce->iter = ctx()->s_expr.popx();
- match_newlines_repl();
- if(match(TK("if"))){
- parse_expression(PREC_TERNARY + 1);
- ce->cond = ctx()->s_expr.popx();
- }
- ctx()->s_expr.push(std::move(ce));
- match_newlines_repl();
- }
- void exprList() {
- int line = prev().line;
- std::vector<Expr_> items;
- do {
- match_newlines_repl();
- if (curr().type == TK("]")) break;
- EXPR();
- items.push_back(ctx()->s_expr.popx());
- match_newlines_repl();
- if(items.size()==1 && match(TK("for"))){
- _consume_comp<ListCompExpr>(std::move(items[0]));
- consume(TK("]"));
- return;
- }
- match_newlines_repl();
- } while (match(TK(",")));
- consume(TK("]"));
- auto e = make_expr<ListExpr>(std::move(items));
- e->line = line; // override line
- ctx()->s_expr.push(std::move(e));
- }
- void exprMap() {
- bool parsing_dict = false; // {...} may be dict or set
- std::vector<Expr_> items;
- do {
- match_newlines_repl();
- if (curr().type == TK("}")) break;
- EXPR();
- int star_level = ctx()->s_expr.top()->star_level();
- if(star_level==2 || curr().type == TK(":")){
- parsing_dict = true;
- }
- if(parsing_dict){
- auto dict_item = make_expr<DictItemExpr>();
- if(star_level == 2){
- dict_item->key = nullptr;
- dict_item->value = ctx()->s_expr.popx();
- }else{
- consume(TK(":"));
- EXPR();
- dict_item->key = ctx()->s_expr.popx();
- dict_item->value = ctx()->s_expr.popx();
- }
- items.push_back(std::move(dict_item));
- }else{
- items.push_back(ctx()->s_expr.popx());
- }
- match_newlines_repl();
- if(items.size()==1 && match(TK("for"))){
- if(parsing_dict) _consume_comp<DictCompExpr>(std::move(items[0]));
- else _consume_comp<SetCompExpr>(std::move(items[0]));
- consume(TK("}"));
- return;
- }
- match_newlines_repl();
- } while (match(TK(",")));
- consume(TK("}"));
- if(items.size()==0 || parsing_dict){
- auto e = make_expr<DictExpr>(std::move(items));
- ctx()->s_expr.push(std::move(e));
- }else{
- auto e = make_expr<SetExpr>(std::move(items));
- ctx()->s_expr.push(std::move(e));
- }
- }
- void exprCall() {
- auto e = make_expr<CallExpr>();
- e->callable = ctx()->s_expr.popx();
- do {
- match_newlines_repl();
- if (curr().type==TK(")")) break;
- if(curr().type==TK("@id") && next().type==TK("=")) {
- consume(TK("@id"));
- Str key = prev().str();
- consume(TK("="));
- EXPR();
- e->kwargs.push_back({key, ctx()->s_expr.popx()});
- } else{
- EXPR();
- if(ctx()->s_expr.top()->star_level() == 2){
- // **kwargs
- e->kwargs.push_back({"**", ctx()->s_expr.popx()});
- }else{
- // positional argument
- if(!e->kwargs.empty()) SyntaxError("positional argument follows keyword argument");
- e->args.push_back(ctx()->s_expr.popx());
- }
- }
- match_newlines_repl();
- } while (match(TK(",")));
- consume(TK(")"));
- if(e->args.size() > 32767) SyntaxError("too many positional arguments");
- if(e->kwargs.size() > 32767) SyntaxError("too many keyword arguments");
- ctx()->s_expr.push(std::move(e));
- }
- void exprName(){
- Str name = prev().str();
- NameScope scope = name_scope();
- if(ctx()->global_names.count(name)){
- scope = NAME_GLOBAL;
- }
- ctx()->s_expr.push(make_expr<NameExpr>(name, scope));
- }
- void exprAttrib() {
- consume(TK("@id"));
- ctx()->s_expr.push(
- make_expr<AttribExpr>(ctx()->s_expr.popx(), prev().str())
- );
- }
-
- void exprSubscr() {
- auto e = make_expr<SubscrExpr>();
- e->a = ctx()->s_expr.popx();
- auto slice = make_expr<SliceExpr>();
- bool is_slice = false;
- // a[<0> <state:1> : state<3> : state<5>]
- int state = 0;
- do{
- switch(state){
- case 0:
- if(match(TK(":"))){
- is_slice=true;
- state=2;
- break;
- }
- if(match(TK("]"))) SyntaxError();
- EXPR_TUPLE();
- slice->start = ctx()->s_expr.popx();
- state=1;
- break;
- case 1:
- if(match(TK(":"))){
- is_slice=true;
- state=2;
- break;
- }
- if(match(TK("]"))) goto __SUBSCR_END;
- SyntaxError("expected ':' or ']'");
- break;
- case 2:
- if(match(TK(":"))){
- state=4;
- break;
- }
- if(match(TK("]"))) goto __SUBSCR_END;
- EXPR_TUPLE();
- slice->stop = ctx()->s_expr.popx();
- state=3;
- break;
- case 3:
- if(match(TK(":"))){
- state=4;
- break;
- }
- if(match(TK("]"))) goto __SUBSCR_END;
- SyntaxError("expected ':' or ']'");
- break;
- case 4:
- if(match(TK("]"))) goto __SUBSCR_END;
- EXPR_TUPLE();
- slice->step = ctx()->s_expr.popx();
- state=5;
- break;
- case 5: consume(TK("]")); goto __SUBSCR_END;
- }
- }while(true);
- __SUBSCR_END:
- if(is_slice){
- e->b = std::move(slice);
- }else{
- if(state != 1) FATAL_ERROR();
- e->b = std::move(slice->start);
- }
- ctx()->s_expr.push(std::move(e));
- }
- void exprLiteral0() {
- ctx()->s_expr.push(make_expr<Literal0Expr>(prev().type));
- }
- void compile_block_body() {
- consume(TK(":"));
- if(curr().type!=TK("@eol") && curr().type!=TK("@eof")){
- compile_stmt(); // inline block
- return;
- }
- if(!match_newlines(mode()==REPL_MODE)){
- SyntaxError("expected a new line after ':'");
- }
- consume(TK("@indent"));
- while (curr().type != TK("@dedent")) {
- match_newlines();
- compile_stmt();
- match_newlines();
- }
- consume(TK("@dedent"));
- }
- Str _compile_import() {
- if(name_scope() != NAME_GLOBAL) SyntaxError("import statement should be used in global scope");
- Opcode op = OP_IMPORT_NAME;
- if(match(TK("."))) op = OP_IMPORT_NAME_REL;
- consume(TK("@id"));
- Str name = prev().str();
- ctx()->emit(op, StrName(name).index, prev().line);
- return name;
- }
- // import a as b
- void compile_normal_import() {
- do {
- Str name = _compile_import();
- if (match(TK("as"))) {
- consume(TK("@id"));
- name = prev().str();
- }
- ctx()->emit(OP_STORE_GLOBAL, StrName(name).index, prev().line);
- } while (match(TK(",")));
- consume_end_stmt();
- }
- // from a import b as c, d as e
- void compile_from_import() {
- _compile_import();
- consume(TK("import"));
- if (match(TK("*"))) {
- ctx()->emit(OP_IMPORT_STAR, BC_NOARG, prev().line);
- consume_end_stmt();
- return;
- }
- do {
- ctx()->emit(OP_DUP_TOP, BC_NOARG, BC_KEEPLINE);
- consume(TK("@id"));
- Str name = prev().str();
- ctx()->emit(OP_LOAD_ATTR, StrName(name).index, prev().line);
- if (match(TK("as"))) {
- consume(TK("@id"));
- name = prev().str();
- }
- ctx()->emit(OP_STORE_GLOBAL, StrName(name).index, prev().line);
- } while (match(TK(",")));
- ctx()->emit(OP_POP_TOP, BC_NOARG, BC_KEEPLINE);
- consume_end_stmt();
- }
- bool is_expression(){
- PrattCallback prefix = rules[curr().type].prefix;
- return prefix != nullptr;
- }
- void parse_expression(int precedence, bool push_stack=true) {
- PrattCallback prefix = rules[curr().type].prefix;
- if (prefix == nullptr) SyntaxError(Str("expected an expression, got ") + TK_STR(curr().type));
- advance();
- (this->*prefix)();
- while (rules[curr().type].precedence >= precedence) {
- TokenIndex op = curr().type;
- advance();
- PrattCallback infix = rules[op].infix;
- PK_ASSERT(infix != nullptr);
- (this->*infix)();
- }
- if(!push_stack) ctx()->emit_expr();
- }
- void compile_if_stmt() {
- EXPR(false); // condition
- int patch = ctx()->emit(OP_POP_JUMP_IF_FALSE, BC_NOARG, prev().line);
- compile_block_body();
- if (match(TK("elif"))) {
- int exit_patch = ctx()->emit(OP_JUMP_ABSOLUTE, BC_NOARG, prev().line);
- ctx()->patch_jump(patch);
- compile_if_stmt();
- ctx()->patch_jump(exit_patch);
- } else if (match(TK("else"))) {
- int exit_patch = ctx()->emit(OP_JUMP_ABSOLUTE, BC_NOARG, prev().line);
- ctx()->patch_jump(patch);
- compile_block_body();
- ctx()->patch_jump(exit_patch);
- } else {
- ctx()->patch_jump(patch);
- }
- }
- void compile_while_loop() {
- ctx()->enter_block(WHILE_LOOP);
- EXPR(false); // condition
- int patch = ctx()->emit(OP_POP_JUMP_IF_FALSE, BC_NOARG, prev().line);
- compile_block_body();
- ctx()->emit(OP_LOOP_CONTINUE, BC_NOARG, BC_KEEPLINE);
- ctx()->patch_jump(patch);
- ctx()->exit_block();
- }
- void compile_for_loop() {
- Expr_ vars = EXPR_VARS();
- consume(TK("in"));
- EXPR_TUPLE(false);
- ctx()->emit(OP_GET_ITER, BC_NOARG, BC_KEEPLINE);
- ctx()->enter_block(FOR_LOOP);
- ctx()->emit(OP_FOR_ITER, BC_NOARG, BC_KEEPLINE);
- bool ok = vars->emit_store(ctx());
- if(!ok) SyntaxError(); // this error occurs in `vars` instead of this line, but...nevermind
- compile_block_body();
- ctx()->emit(OP_LOOP_CONTINUE, BC_NOARG, BC_KEEPLINE);
- ctx()->exit_block();
- }
- void compile_try_except() {
- ctx()->enter_block(TRY_EXCEPT);
- compile_block_body();
- std::vector<int> patches = {
- ctx()->emit(OP_JUMP_ABSOLUTE, BC_NOARG, BC_KEEPLINE)
- };
- ctx()->exit_block();
- do {
- consume(TK("except"));
- if(match(TK("@id"))){
- ctx()->emit(OP_EXCEPTION_MATCH, StrName(prev().str()).index, prev().line);
- }else{
- ctx()->emit(OP_LOAD_TRUE, BC_NOARG, BC_KEEPLINE);
- }
- int patch = ctx()->emit(OP_POP_JUMP_IF_FALSE, BC_NOARG, BC_KEEPLINE);
- // pop the exception on match
- ctx()->emit(OP_POP_EXCEPTION, BC_NOARG, BC_KEEPLINE);
- compile_block_body();
- patches.push_back(ctx()->emit(OP_JUMP_ABSOLUTE, BC_NOARG, BC_KEEPLINE));
- ctx()->patch_jump(patch);
- }while(curr().type == TK("except"));
- // no match, re-raise
- ctx()->emit(OP_RE_RAISE, BC_NOARG, BC_KEEPLINE);
- for (int patch : patches) ctx()->patch_jump(patch);
- }
- void compile_decorated(){
- std::vector<Expr_> decorators;
- do{
- EXPR();
- decorators.push_back(ctx()->s_expr.popx());
- if(!match_newlines_repl()) SyntaxError();
- }while(match(TK("@")));
- consume(TK("def"));
- compile_function(decorators);
- }
- bool try_compile_assignment(){
- switch (curr().type) {
- case TK("+="): case TK("-="): case TK("*="): case TK("/="): case TK("//="): case TK("%="):
- case TK("<<="): case TK(">>="): case TK("&="): case TK("|="): case TK("^="): {
- Expr* lhs_p = ctx()->s_expr.top().get();
- if(lhs_p->is_starred()) SyntaxError();
- if(ctx()->is_compiling_class) SyntaxError();
- advance();
- auto e = make_expr<BinaryExpr>();
- e->op = prev().type - 1; // -1 to remove =
- e->lhs = ctx()->s_expr.popx();
- EXPR_TUPLE();
- e->rhs = ctx()->s_expr.popx();
- if(e->is_starred()) SyntaxError();
- e->emit(ctx());
- bool ok = lhs_p->emit_store(ctx());
- if(!ok) SyntaxError();
- } return true;
- case TK("="): {
- int n = 0;
- while(match(TK("="))){
- EXPR_TUPLE();
- n += 1;
- }
- // stack size is n+1
- Expr_ val = ctx()->s_expr.popx();
- val->emit(ctx());
- for(int i=1; i<n; i++) ctx()->emit(OP_DUP_TOP, BC_NOARG, BC_KEEPLINE);
- for(int i=0; i<n; i++){
- auto e = ctx()->s_expr.popx();
- if(e->is_starred()) SyntaxError();
- bool ok = e->emit_store(ctx());
- if(!ok) SyntaxError();
- }
- } return true;
- default: return false;
- }
- }
- void compile_stmt() {
- advance();
- int kw_line = prev().line; // backup line number
- switch(prev().type){
- case TK("break"):
- if (!ctx()->is_curr_block_loop()) SyntaxError("'break' outside loop");
- ctx()->emit(OP_LOOP_BREAK, BC_NOARG, kw_line);
- consume_end_stmt();
- break;
- case TK("continue"):
- if (!ctx()->is_curr_block_loop()) SyntaxError("'continue' not properly in loop");
- ctx()->emit(OP_LOOP_CONTINUE, BC_NOARG, kw_line);
- consume_end_stmt();
- break;
- case TK("yield"):
- if (contexts.size() <= 1) SyntaxError("'yield' outside function");
- EXPR_TUPLE(false);
- // if yield present, mark the function as generator
- ctx()->co->is_generator = true;
- ctx()->emit(OP_YIELD_VALUE, BC_NOARG, kw_line);
- consume_end_stmt();
- break;
- case TK("yield from"):
- if (contexts.size() <= 1) SyntaxError("'yield from' outside function");
- EXPR_TUPLE(false);
- // if yield from present, mark the function as generator
- ctx()->co->is_generator = true;
- ctx()->emit(OP_GET_ITER, BC_NOARG, kw_line);
- ctx()->enter_block(FOR_LOOP);
- ctx()->emit(OP_FOR_ITER, BC_NOARG, BC_KEEPLINE);
- ctx()->emit(OP_YIELD_VALUE, BC_NOARG, BC_KEEPLINE);
- ctx()->emit(OP_LOOP_CONTINUE, BC_NOARG, BC_KEEPLINE);
- ctx()->exit_block();
- consume_end_stmt();
- break;
- case TK("return"):
- if (contexts.size() <= 1) SyntaxError("'return' outside function");
- if(match_end_stmt()){
- ctx()->emit(OP_LOAD_NONE, BC_NOARG, kw_line);
- }else{
- EXPR_TUPLE(false);
- consume_end_stmt();
- }
- ctx()->emit(OP_RETURN_VALUE, BC_NOARG, kw_line);
- break;
- /*************************************************/
- case TK("if"): compile_if_stmt(); break;
- case TK("while"): compile_while_loop(); break;
- case TK("for"): compile_for_loop(); break;
- case TK("import"): compile_normal_import(); break;
- case TK("from"): compile_from_import(); break;
- case TK("def"): compile_function(); break;
- case TK("@"): compile_decorated(); break;
- case TK("try"): compile_try_except(); break;
- case TK("pass"): consume_end_stmt(); break;
- /*************************************************/
- case TK("++"):{
- consume(TK("@id"));
- StrName name(prev().sv());
- switch(name_scope()){
- case NAME_LOCAL:
- ctx()->emit(OP_INC_FAST, ctx()->add_varname(name), prev().line);
- break;
- case NAME_GLOBAL:
- ctx()->emit(OP_INC_GLOBAL, name.index, prev().line);
- break;
- default: SyntaxError(); break;
- }
- consume_end_stmt();
- break;
- }
- case TK("--"):{
- consume(TK("@id"));
- StrName name(prev().sv());
- switch(name_scope()){
- case NAME_LOCAL:
- ctx()->emit(OP_DEC_FAST, ctx()->add_varname(name), prev().line);
- break;
- case NAME_GLOBAL:
- ctx()->emit(OP_DEC_GLOBAL, name.index, prev().line);
- break;
- default: SyntaxError(); break;
- }
- consume_end_stmt();
- break;
- }
- case TK("assert"):
- EXPR_TUPLE(false);
- ctx()->emit(OP_ASSERT, BC_NOARG, kw_line);
- consume_end_stmt();
- break;
- case TK("global"):
- do {
- consume(TK("@id"));
- ctx()->global_names.insert(prev().str());
- } while (match(TK(",")));
- consume_end_stmt();
- break;
- case TK("raise"): {
- consume(TK("@id"));
- int dummy_t = StrName(prev().str()).index;
- if(match(TK("(")) && !match(TK(")"))){
- EXPR(false); consume(TK(")"));
- }else{
- ctx()->emit(OP_LOAD_NONE, BC_NOARG, kw_line);
- }
- ctx()->emit(OP_RAISE, dummy_t, kw_line);
- consume_end_stmt();
- } break;
- case TK("del"): {
- EXPR_TUPLE();
- Expr_ e = ctx()->s_expr.popx();
- bool ok = e->emit_del(ctx());
- if(!ok) SyntaxError();
- consume_end_stmt();
- } break;
- case TK("with"): {
- EXPR(false);
- consume(TK("as"));
- consume(TK("@id"));
- StrName name(prev().str());
- ctx()->emit(OP_STORE_NAME, name.index, prev().line);
- ctx()->emit(OP_LOAD_NAME, name.index, prev().line);
- ctx()->emit(OP_WITH_ENTER, BC_NOARG, prev().line);
- compile_block_body();
- ctx()->emit(OP_LOAD_NAME, name.index, prev().line);
- ctx()->emit(OP_WITH_EXIT, BC_NOARG, prev().line);
- } break;
- /*************************************************/
- case TK("$label"): {
- if(mode()!=EXEC_MODE) SyntaxError("'label' is only available in EXEC_MODE");
- consume(TK("@id"));
- bool ok = ctx()->add_label(prev().str());
- if(!ok) SyntaxError("label " + prev().str().escape() + " already exists");
- consume_end_stmt();
- } break;
- case TK("$goto"):
- if(mode()!=EXEC_MODE) SyntaxError("'goto' is only available in EXEC_MODE");
- consume(TK("@id"));
- ctx()->emit(OP_GOTO, StrName(prev().str()).index, prev().line);
- consume_end_stmt();
- break;
- /*************************************************/
- // handle dangling expression or assignment
- default: {
- advance(-1); // do revert since we have pre-called advance() at the beginning
- EXPR_TUPLE();
- // eat variable's type hint
- if(match(TK(":"))) consume_type_hints();
- if(!try_compile_assignment()){
- if(!ctx()->s_expr.empty() && ctx()->s_expr.top()->is_starred()){
- SyntaxError();
- }
- ctx()->emit_expr();
- if((mode()==CELL_MODE || mode()==REPL_MODE) && name_scope()==NAME_GLOBAL){
- ctx()->emit(OP_PRINT_EXPR, BC_NOARG, BC_KEEPLINE);
- }else{
- ctx()->emit(OP_POP_TOP, BC_NOARG, BC_KEEPLINE);
- }
- }
- consume_end_stmt();
- }
- }
- }
- void consume_type_hints(){
- EXPR();
- ctx()->s_expr.pop();
- }
- void compile_class(){
- consume(TK("@id"));
- int namei = StrName(prev().str()).index;
- int super_namei = -1;
- if(match(TK("("))){
- if(match(TK("@id"))){
- super_namei = StrName(prev().str()).index;
- }
- consume(TK(")"));
- }
- if(super_namei == -1) ctx()->emit(OP_LOAD_NONE, BC_NOARG, prev().line);
- else ctx()->emit(OP_LOAD_GLOBAL, super_namei, prev().line);
- ctx()->emit(OP_BEGIN_CLASS, namei, BC_KEEPLINE);
- ctx()->is_compiling_class = true;
- compile_block_body();
- ctx()->is_compiling_class = false;
- ctx()->emit(OP_END_CLASS, BC_NOARG, BC_KEEPLINE);
- }
- void _compile_f_args(FuncDecl_ decl, bool enable_type_hints){
- int state = 0; // 0 for args, 1 for *args, 2 for k=v, 3 for **kwargs
- do {
- if(state > 3) SyntaxError();
- if(state == 3) SyntaxError("**kwargs should be the last argument");
- match_newlines();
- if(match(TK("*"))){
- if(state < 1) state = 1;
- else SyntaxError("*args should be placed before **kwargs");
- }
- else if(match(TK("**"))){
- state = 3;
- }
- consume(TK("@id"));
- StrName name = prev().str();
- // check duplicate argument name
- for(int i: decl->args){
- if(decl->code->varnames[i] == name) {
- SyntaxError("duplicate argument name");
- }
- }
- for(auto& kv: decl->kwargs){
- if(decl->code->varnames[kv.key] == name){
- SyntaxError("duplicate argument name");
- }
- }
- if(decl->starred_arg!=-1 && decl->code->varnames[decl->starred_arg] == name){
- SyntaxError("duplicate argument name");
- }
- if(decl->starred_kwarg!=-1 && decl->code->varnames[decl->starred_kwarg] == name){
- SyntaxError("duplicate argument name");
- }
- // eat type hints
- if(enable_type_hints && match(TK(":"))) consume_type_hints();
- if(state == 0 && curr().type == TK("=")) state = 2;
- int index = ctx()->add_varname(name);
- switch (state)
- {
- case 0:
- decl->args.push_back(index);
- break;
- case 1:
- decl->starred_arg = index;
- state+=1;
- break;
- case 2: {
- consume(TK("="));
- PyObject* value = read_literal();
- if(value == nullptr){
- SyntaxError(Str("default argument must be a literal"));
- }
- decl->kwargs.push_back(FuncDecl::KwArg{index, value});
- } break;
- case 3:
- decl->starred_kwarg = index;
- state+=1;
- break;
- }
- } while (match(TK(",")));
- }
- void compile_function(const std::vector<Expr_>& decorators={}){
- Str decl_name;
- consume(TK("@id"));
- decl_name = prev().str();
- FuncDecl_ decl = push_f_context(decl_name);
- consume(TK("("));
- if (!match(TK(")"))) {
- _compile_f_args(decl, true);
- consume(TK(")"));
- }
- if(match(TK("->"))) consume_type_hints();
- compile_block_body();
- pop_context();
- PyObject* docstring = nullptr;
- if(decl->code->codes.size()>=2 && decl->code->codes[0].op == OP_LOAD_CONST && decl->code->codes[1].op == OP_POP_TOP){
- PyObject* c = decl->code->consts[decl->code->codes[0].arg];
- if(is_type(c, vm->tp_str)){
- decl->code->codes[0].op = OP_NO_OP;
- decl->code->codes[1].op = OP_NO_OP;
- docstring = c;
- }
- }
- ctx()->emit(OP_LOAD_FUNCTION, ctx()->add_func_decl(decl), prev().line);
- if(docstring != nullptr){
- ctx()->emit(OP_SETUP_DOCSTRING, ctx()->add_const(docstring), prev().line);
- }
- // add decorators
- for(auto it=decorators.rbegin(); it!=decorators.rend(); ++it){
- (*it)->emit(ctx());
- ctx()->emit(OP_ROT_TWO, BC_NOARG, (*it)->line);
- ctx()->emit(OP_LOAD_NULL, BC_NOARG, BC_KEEPLINE);
- ctx()->emit(OP_ROT_TWO, BC_NOARG, BC_KEEPLINE);
- ctx()->emit(OP_CALL, 1, (*it)->line);
- }
- if(!ctx()->is_compiling_class){
- auto e = make_expr<NameExpr>(decl_name, name_scope());
- e->emit_store(ctx());
- }else{
- int index = StrName(decl_name).index;
- ctx()->emit(OP_STORE_CLASS_ATTR, index, prev().line);
- }
- }
- PyObject* to_object(const TokenValue& value){
- PyObject* obj = nullptr;
- if(std::holds_alternative<i64>(value)){
- obj = VAR(std::get<i64>(value));
- }
- if(std::holds_alternative<f64>(value)){
- obj = VAR(std::get<f64>(value));
- }
- if(std::holds_alternative<Str>(value)){
- obj = VAR(std::get<Str>(value));
- }
- if(obj == nullptr) FATAL_ERROR();
- return obj;
- }
- PyObject* read_literal(){
- advance();
- switch(prev().type){
- case TK("-"): {
- consume(TK("@num"));
- PyObject* val = to_object(prev().value);
- return vm->py_negate(val);
- }
- case TK("@num"): return to_object(prev().value);
- case TK("@str"): return to_object(prev().value);
- case TK("True"): return VAR(true);
- case TK("False"): return VAR(false);
- case TK("None"): return vm->None;
- case TK("..."): return vm->Ellipsis;
- default: break;
- }
- return nullptr;
- }
- void SyntaxError(Str msg){ lexer->throw_err("SyntaxError", msg, err().line, err().start); }
- void SyntaxError(){ lexer->throw_err("SyntaxError", "invalid syntax", err().line, err().start); }
- void IndentationError(Str msg){ lexer->throw_err("IndentationError", msg, err().line, err().start); }
- public:
- Compiler(VM* vm, const Str& source, const Str& filename, CompileMode mode, bool unknown_global_scope=false){
- this->vm = vm;
- this->used = false;
- this->unknown_global_scope = unknown_global_scope;
- this->lexer = std::make_unique<Lexer>(
- make_sp<SourceData>(source, filename, mode)
- );
- init_pratt_rules();
- }
- CodeObject_ compile(){
- if(used) FATAL_ERROR();
- used = true;
- tokens = lexer->run();
- // if(lexer->src->filename == "<stdin>"){
- // for(auto& t: tokens) std::cout << t.info() << std::endl;
- // }
- CodeObject_ code = push_global_context();
- advance(); // skip @sof, so prev() is always valid
- match_newlines(); // skip possible leading '\n'
- if(mode()==EVAL_MODE) {
- EXPR_TUPLE(false);
- consume(TK("@eof"));
- ctx()->emit(OP_RETURN_VALUE, BC_NOARG, BC_KEEPLINE);
- pop_context();
- return code;
- }else if(mode()==JSON_MODE){
- EXPR();
- Expr_ e = ctx()->s_expr.popx();
- if(!e->is_json_object()) SyntaxError("expect a JSON object, literal or array");
- consume(TK("@eof"));
- e->emit(ctx());
- ctx()->emit(OP_RETURN_VALUE, BC_NOARG, BC_KEEPLINE);
- pop_context();
- return code;
- }
- while (!match(TK("@eof"))) {
- if (match(TK("class"))) {
- compile_class();
- } else {
- compile_stmt();
- }
- match_newlines();
- }
- pop_context();
- return code;
- }
- };
- #undef BC_NOARG
- #undef BC_KEEPLINE
- } // namespace pkpy
|