refactor importer

include/pocketpy/opcodes.h

@@ -106,8 +106,7 @@ OPCODE(UNARY_INVERT)
 OPCODE(GET_ITER)
 OPCODE(FOR_ITER)
 /**************************/
-OPCODE(IMPORT_NAME)
-OPCODE(IMPORT_NAME_REL)
+OPCODE(IMPORT_PATH)
 OPCODE(POP_IMPORT_STAR)
 /**************************/
 OPCODE(UNPACK_SEQUENCE)

include/pocketpy/str.h

@@ -72,8 +72,9 @@ struct Str{
     Str upper() const;
     Str escape(bool single_quote=true) const;
     int index(const Str& sub, int start=0) const;
+    Str replace(char old, char new_) const;
     Str replace(const Str& old, const Str& new_, int count=-1) const;
-    std::vector<std::string_view> split(const Str& sep) const;
+    std::vector<std::string_view> split(const Str& sep, bool remove_empty) const;
 
     /*************unicode*************/
     int _unicode_index_to_byte(int i) const;
@@ -186,6 +187,7 @@ const StrName __enter__ = StrName::get("__enter__");
 const StrName __exit__ = StrName::get("__exit__");
 const StrName __name__ = StrName::get("__name__");
 const StrName __all__ = StrName::get("__all__");
+const StrName __package__ = StrName::get("__package__");
 
 const StrName pk_id_add = StrName::get("add");
 const StrName pk_id_set = StrName::get("set");

include/pocketpy/vm.h

@@ -368,6 +368,7 @@ public:
     void UnboundLocalError(StrName name){ _error("UnboundLocalError", fmt("local variable ", name.escape() + " referenced before assignment")); }
     void KeyError(PyObject* obj){ _error("KeyError", PK_OBJ_GET(Str, py_repr(obj))); }
     void BinaryOptError(const char* op) { TypeError(fmt("unsupported operand type(s) for ", op)); }
+    void ImportError(const Str& msg){ _error("ImportError", msg); }
 
     void AttributeError(PyObject* obj, StrName name){
         // OBJ_NAME calls getattr, which may lead to a infinite recursion
@@ -409,31 +410,21 @@ public:
     }
 
     struct ImportContext{
-        // 0: normal; 1: __init__.py; 2: relative
-        std::vector<std::pair<StrName, int>> pending;
-
+        std::vector<StrName> pending;
         struct Temp{
-            VM* vm;
+            ImportContext* ctx;
             StrName name;
-
-            Temp(VM* vm, StrName name, int type): vm(vm), name(name){
-                ImportContext* ctx = &vm->_import_context;
-                ctx->pending.emplace_back(name, type);
-            }
-
-            ~Temp(){
-                ImportContext* ctx = &vm->_import_context;
-                ctx->pending.pop_back();
+            Temp(ImportContext* ctx, StrName name) : ctx(ctx), name(name){
+                ctx->pending.push_back(name);
             }
+            ~Temp(){ ctx->pending.pop_back(); }
         };
 
-        Temp temp(VM* vm, StrName name, int type){
-            return Temp(vm, name, type);
-        }
+        Temp scope(StrName name){ return {this, name}; }
     };
 
     ImportContext _import_context;
-    PyObject* py_import(Str path, bool relative=false);
+    PyObject* py_import(Str path, PyObject* _module);
     ~VM();
 
 #if PK_DEBUG_CEVAL_STEP
@@ -447,7 +438,7 @@ public:
     bool py_bool(PyObject* obj);
     i64 py_hash(PyObject* obj);
     PyObject* py_list(PyObject*);
-    PyObject* new_module(StrName name);
+    PyObject* new_module(Str name, Str package="");
     Str disassemble(CodeObject_ co);
     void init_builtin_types();
     PyObject* getattr(PyObject* obj, StrName name, bool throw_err=true);

src/ceval.cpp

@@ -590,13 +590,9 @@ __NEXT_STEP:;
         }
         DISPATCH();
     /*****************************************/
-    TARGET(IMPORT_NAME)
+    TARGET(IMPORT_PATH)
         _0 = co_consts[byte.arg];
-        PUSH(py_import(CAST(Str&, _0)));
-        DISPATCH();
-    TARGET(IMPORT_NAME_REL)
-        _0 = co_consts[byte.arg];
-        PUSH(py_import(CAST(Str&, _0), true));
+        PUSH(py_import(CAST(Str&, _0), frame->_module));
         DISPATCH();
     TARGET(POP_IMPORT_STAR) {
         _0 = POPX();        // pop the module
@@ -606,7 +602,7 @@ __NEXT_STEP:;
                 _name = StrName::get(CAST(Str&, key).sv());
                 PyObject* value = _0->attr().try_get(_name);
                 if(value == nullptr){
-                    _error("ImportError", fmt("cannot import name ", _name.escape()));
+                    ImportError(fmt("cannot import name ", _name.escape()));
                 }else{
                     frame->f_globals().set(_name, value);
                 }

src/compiler.cpp

@@ -479,7 +479,7 @@ __SUBSCR_END:
         do {
             consume(TK("@id"));
             Str name = prev().str();
-            ctx()->emit(OP_IMPORT_NAME, ctx()->add_const(VAR(name)), prev().line);
+            ctx()->emit(OP_IMPORT_PATH, ctx()->add_const(VAR(name)), prev().line);
             if (match(TK("as"))) {
                 consume(TK("@id"));
                 name = prev().str();
@@ -493,37 +493,45 @@ __SUBSCR_END:
     // from a.b import c [as d]
     // from . import a [as b]
     // from .a import b [as c]
+    // from ..a import b [as c]
     // from .a.b import c [as d]
     // from xxx import *
     void Compiler::compile_from_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;
-        std::vector<Str> parts;
+        int dots = 0;
 
-        if(op == OP_IMPORT_NAME_REL){
+        while(true){
+            switch(curr().type){
+                case TK("."): dots++; break;
+                case TK("..."): dots+=3; break;
+                default: goto __EAT_DOTS_END;
+            }
+            advance();
+        }
+__EAT_DOTS_END:
+        std::stringstream ss;
+        for(int i=0; i<dots; i++) ss << '.';
+
+        if(dots > 0){
+            // @id is optional if dots > 0
             if(match(TK("@id"))){
-                parts.push_back(prev().str());
+                ss << prev().str();
                 while (match(TK("."))) {
                     consume(TK("@id"));
-                    parts.push_back(prev().str());
+                    ss << "." << prev().str();
                 }
             }
         }else{
+            // @id is required if dots == 0
             consume(TK("@id"));
-            parts.push_back(prev().str());
+            ss << prev().str();
             while (match(TK("."))) {
                 consume(TK("@id"));
-                parts.push_back(prev().str());
+                ss << "." << prev().str();
             }
         }
 
-        FastStrStream ss;
-        for (int i=0; i<parts.size(); i++) {
-            if(i > 0) ss << ".";
-            ss << parts[i];
-        }
-        ctx()->emit(op, ctx()->add_const(VAR(ss.str())), prev().line);
+        ctx()->emit(OP_IMPORT_PATH, ctx()->add_const(VAR(ss.str())), prev().line);
         consume(TK("import"));
 
         if (match(TK("*"))) {
@@ -537,7 +545,6 @@ __SUBSCR_END:
             ctx()->emit(OP_DUP_TOP, BC_NOARG, BC_KEEPLINE);
             consume(TK("@id"));
             Str name = prev().str();
-            // module's __getattr__ should be customized or use a new opcode...
             ctx()->emit(OP_LOAD_ATTR, StrName(name).index, prev().line);
             if (match(TK("as"))) {
                 consume(TK("@id"));

src/pocketpy.cpp

@@ -171,18 +171,18 @@ void init_builtins(VM* _vm) {
             if(ext == ".so" || ext == ".dll" || ext == ".dylib"){
                 dylib_entry_t entry = load_dylib(name.c_str());
                 if(!entry){
-                    vm->_error("ImportError", "cannot load dynamic library: " + name.escape());
+                    vm->ImportError("cannot load dynamic library: " + name.escape());
                 }
                 vm->_c.s_view.push(ArgsView(vm->s_data.end(), vm->s_data.end()));
                 const char* name = entry(vm, PK_VERSION);
                 vm->_c.s_view.pop();
                 if(name == nullptr){
-                    vm->_error("ImportError", "module initialization failed: " + Str(name).escape());
+                    vm->ImportError("module initialization failed: " + Str(name).escape());
                 }
                 return vm->_modules[name];
             }
         }
-        return vm->py_import(name);
+        return vm->py_import(name, vm->top_frame()->_module);
     });
 
     _vm->bind_builtin_func<2>("divmod", [](VM* vm, ArgsView args) {
@@ -560,7 +560,7 @@ void init_builtins(VM* _vm) {
 
     _vm->bind(_vm->_t(_vm->tp_str), "split(self, sep=' ')", [](VM* vm, ArgsView args) {
         const Str& self = _CAST(Str&, args[0]);
-        std::vector<std::string_view> parts = self.split(CAST(Str&, args[1]));
+        std::vector<std::string_view> parts = self.split(CAST(Str&, args[1]), false);
         List ret(parts.size());
         for(int i=0; i<parts.size(); i++) ret[i] = VAR(Str(parts[i]));
         return VAR(std::move(ret));
@@ -1225,6 +1225,16 @@ void init_builtins(VM* _vm) {
         }
         return vm->True;
     });
+
+    _vm->bind__repr__(_vm->tp_module, [](VM* vm, PyObject* obj) {
+        const Str& package = CAST(Str&, obj->attr(__package__));
+        Str name = CAST(Str&, obj->attr(__name__));
+        if(!package.empty()){
+            name = package + "." + name;
+        }
+        return VAR(fmt("<module ", name.escape(), ">"));
+    });
+
     /************ property ************/
     _vm->bind_constructor<-1>("property", [](VM* vm, ArgsView args) {
         if(args.size() == 1+1){

src/str.cpp

@@ -254,6 +254,14 @@ int utf8len(unsigned char c, bool suppress){
         return p - data;
     }
 
+    Str Str::replace(char old, char new_) const{
+        Str copied = *this;
+        for(int i=0; i<copied.size; i++){
+            if(copied.data[i] == old) copied.data[i] = new_;
+        }
+        return copied;
+    }
+
     Str Str::replace(const Str& old, const Str& new_, int count) const {
         std::stringstream ss;
         int start = 0;
@@ -308,16 +316,19 @@ int utf8len(unsigned char c, bool suppress){
         return _byte_index_to_unicode(size);
     }
 
-    std::vector<std::string_view> Str::split(const Str& sep) const{
+    std::vector<std::string_view> Str::split(const Str& sep, bool remove_empty) const{
         std::vector<std::string_view> result;
+        std::string_view tmp;
         int start = 0;
         while(true){
             int i = index(sep, start);
             if(i == -1) break;
-            result.push_back(sv().substr(start, i - start));
+            tmp = sv().substr(start, i - start);
+            if(!remove_empty || !tmp.empty()) result.push_back(tmp);
             start = i + sep.size;
         }
-        result.push_back(sv().substr(start, size - start));
+        tmp = sv().substr(start, size - start);
+        if(!remove_empty || !tmp.empty()) result.push_back(tmp);
         return result;
     }
 

src/vm.cpp

@@ -215,57 +215,73 @@ namespace pkpy{
         return call_method(obj, __next__);
     }
 
-    PyObject* VM::py_import(Str name, bool relative){
-        // path is '.' separated
-        Str filename;
-        int type;
-        if(relative){
-            ImportContext* ctx = &_import_context;
-            type = 2;
-            for(auto it=ctx->pending.rbegin(); it!=ctx->pending.rend(); ++it){
-                if(it->second == 2) continue;
-                if(it->second == 1){
-                    filename = fmt(it->first, kPlatformSep, name, ".py");
-                    name = fmt(it->first, '.', name).c_str();
-                    break;
-                }
+    PyObject* VM::py_import(Str path, PyObject* _module){
+        if(path.empty()) vm->ValueError("empty module name");
+
+        auto f_join = [](const std::vector<std::string_view>& cpnts){
+            std::stringstream ss;
+            for(int i=0; i<cpnts.size(); i++){
+                if(i != 0) ss << ".";
+                ss << cpnts[i];
             }
-            if(filename.length() == 0) _error("ImportError", "relative import outside of package");
-        }else{
-            type = 0;
-            filename = fmt(name, ".py");
-        }
-        for(auto& [k, v]: _import_context.pending){
-            if(k == name){
-                vm->_error("ImportError", fmt("circular import ", name.escape()));
+            return Str(ss.str());
+        };
+
+        if(path[0] == '.'){
+            Str _mod_name = CAST(Str&, _module->attr(__name__));
+            Str _mod_package = CAST(Str&, _module->attr(__package__));
+            // get _module's fullname
+            if(!_mod_package.empty()) _mod_name = _mod_package + "." + _mod_name;
+            // convert relative path to absolute path
+            std::vector<std::string_view> cpnts = _mod_name.split(".", true);
+            int prefix = 0;     // how many dots in the prefix
+            for(int i=0; i<path.length(); i++){
+                if(path[i] == '.') prefix++;
+                else break;
             }
+            if(prefix > cpnts.size()) ImportError("attempted relative import beyond top-level package");
+            path = path.substr(prefix);     // remove prefix
+            for(int i=1; i<prefix; i++) cpnts.pop_back();
+            cpnts.push_back(path.sv());
+            path = f_join(cpnts);
         }
+
+        StrName name(path);     // path to StrName
+
+        // check circular import
+        for(StrName pending_name: _import_context.pending){
+            if(pending_name == name) ImportError(fmt("circular import ", name.escape()));
+        }
+
         PyObject* ext_mod = _modules.try_get(name);
-        if(ext_mod == nullptr){
-            Str source;
-            auto it = _lazy_modules.find(name);
-            if(it == _lazy_modules.end()){
-                Bytes b = _import_handler(filename);
-                if(!relative && !b){
-                    filename = fmt(name, kPlatformSep, "__init__.py");
-                    b = _import_handler(filename);
-                    if(b) type = 1;
-                }
-                if(!b) _error("ImportError", fmt("module ", name.escape(), " not found"));
-                source = Str(b.str());
-            }else{
-                source = it->second;
-                _lazy_modules.erase(it);
+        if(ext_mod != nullptr) return ext_mod;
+
+        // try import
+        Str filename = path.replace('.', kPlatformSep) + ".py";
+        Str source;
+        auto it = _lazy_modules.find(name);
+        if(it == _lazy_modules.end()){
+            Bytes b = _import_handler(filename);
+            if(!b){
+                filename = path.replace('.', kPlatformSep).str() + kPlatformSep + "__init__.py";
+                b = _import_handler(filename);
             }
-            auto _ = _import_context.temp(this, name, type);
-            CodeObject_ code = compile(source, filename, EXEC_MODE);
-            PyObject* new_mod = new_module(name);
-            _exec(code, new_mod);
-            new_mod->attr()._try_perfect_rehash();
-            return new_mod;
+            if(!b) ImportError(fmt("module ", path.escape(), " not found"));
+            source = Str(b.str());
         }else{
-            return ext_mod;
+            source = it->second;
+            _lazy_modules.erase(it);
         }
+        auto _ = _import_context.scope(name);
+        CodeObject_ code = compile(source, filename, EXEC_MODE);
+
+        auto all_cpnts = path.split(".", true);
+        Str name_cpnt = all_cpnts.back();
+        all_cpnts.pop_back();
+        PyObject* new_mod = new_module(name_cpnt, f_join(all_cpnts));
+        _exec(code, new_mod);
+        new_mod->attr()._try_perfect_rehash();
+        return new_mod;
     }
 
     VM::~VM() {
@@ -471,12 +487,15 @@ PyObject* VM::format(Str spec, PyObject* obj){
     return VAR(ret);
 }
 
-PyObject* VM::new_module(StrName name) {
+PyObject* VM::new_module(Str name, Str package) {
     PyObject* obj = heap._new<DummyModule>(tp_module);
-    obj->attr().set("__name__", VAR(name.sv()));
+    obj->attr().set(__name__, VAR(name));
+    obj->attr().set(__package__, VAR(package));
     // we do not allow override in order to avoid memory leak
     // it is because Module objects are not garbage collected
     if(_modules.contains(name)) throw std::runtime_error("module already exists");
+    // convert to fullname and set it into _modules
+    if(!package.empty()) name = package + "." + name;
     _modules.set(name, obj);
     return obj;
 }
@@ -484,14 +503,14 @@ PyObject* VM::new_module(StrName name) {
 static std::string _opcode_argstr(VM* vm, Bytecode byte, const CodeObject* co){
     std::string argStr = byte.arg == -1 ? "" : std::to_string(byte.arg);
     switch(byte.op){
-        case OP_LOAD_CONST: case OP_FORMAT_STRING:
+        case OP_LOAD_CONST: case OP_FORMAT_STRING: case OP_IMPORT_PATH:
             if(vm != nullptr){
                 argStr += fmt(" (", CAST(Str, vm->py_repr(co->consts[byte.arg])), ")");
             }
             break;
         case OP_LOAD_NAME: case OP_LOAD_GLOBAL: case OP_LOAD_NONLOCAL: case OP_STORE_GLOBAL:
         case OP_LOAD_ATTR: case OP_LOAD_METHOD: case OP_STORE_ATTR: case OP_DELETE_ATTR:
-        case OP_IMPORT_NAME: case OP_BEGIN_CLASS: case OP_RAISE:
+        case OP_BEGIN_CLASS: case OP_RAISE:
         case OP_DELETE_GLOBAL: case OP_INC_GLOBAL: case OP_DEC_GLOBAL: case OP_STORE_CLASS_ATTR:
             argStr += fmt(" (", StrName(byte.arg).sv(), ")");
             break;
@@ -884,10 +903,6 @@ PyObject* VM::vectorcall(int ARGC, int KWARGC, bool op_call){
     return nullptr;
 }
 
-
-
-
-
 // https://docs.python.org/3/howto/descriptor.html#invocation-from-an-instance
 PyObject* VM::getattr(PyObject* obj, StrName name, bool throw_err){
     PyObject* objtype;
@@ -919,6 +934,11 @@ PyObject* VM::getattr(PyObject* obj, StrName name, bool throw_err){
         }
         return cls_var;
     }
+    
+    if(is_non_tagged_type(obj, tp_module)){
+        // try import and cache it!
+    }
+
     if(throw_err) AttributeError(obj, name);
     return nullptr;
 }