next steps

.vscode/settings.json

@@ -71,6 +71,18 @@
     "cinttypes": "cpp",
     "typeinfo": "cpp",
     "valarray": "cpp",
-    "variant": "cpp"
+    "variant": "cpp",
+    "int.h": "c",
+    "macros.h": "c",
+    "stdint.h": "c",
+    "type.h": "c",
+    "stdlib.h": "c",
+    "bitset": "cpp",
+    "codecvt": "cpp",
+    "forward_list": "cpp",
+    "set": "cpp",
+    "unordered_set": "cpp",
+    "source_location": "cpp",
+    "shared_mutex": "cpp"
   },
 }

compiler/include/AST.h

@@ -1,8 +1,12 @@
 #pragma once
 
 #include <string>
+#include <vector>
 
 class Expr;
+class Type;
+
+typedef std::pair<std::string, Type *> FnArg;
 
 class ASTNode
 {
@@ -12,38 +16,127 @@ class Stmt : public ASTNode
 {
 };
 
-class Decl : public Stmt
+class Module : public ASTNode
 {
 public:
-  Decl(const std::string &identifier, const Expr *value)
+  Module(const std::vector<Stmt *> &stmts) : stmts(std::move(stmts)) {}
-      : identifier(identifier), value(value) {}
+  const std::vector<Stmt *> stmts;
+};
 
-  const std::string identifier;
+class ValDecl : public Stmt
+{
+public:
+  ValDecl(const std::string &name, const Type *type, const Expr *value)
+      : name(name), type(type), value(value) {}
+
+  const std::string name;
+  const Type *type;
   const Expr *value;
 };
 
+class FnDecl : public Stmt
+{
+public:
+  FnDecl(const std::string &name, const std::vector<FnArg> &args, const Type *returnType, const Expr *body)
+      : name(name), args(std::move(args)), returnType(returnType), body(body) {}
+
+  const std::string name;
+  const std::vector<FnArg> args;
+  const Type *returnType;
+  const Expr *body;
+};
+
+class Type : public ASTNode
+{
+};
+
+class TupleType : public Type
+{
+public:
+  TupleType(const std::vector<Type *> &types) : types(types) {}
+  const std::vector<Type *> types;
+};
+
+class FunctionType : public Type
+{
+public:
+  FunctionType(const std::vector<Type *> &from, const Type *to)
+      : from(std::move(from)), to(to) {}
+
+  const std::vector<Type *> from;
+  const Type *to;
+};
+
+class NamedType : public Type
+{
+public:
+  NamedType(const std::string &name) : name(name) {}
+  const std::string name;
+};
+
 class Expr : public ASTNode
 {
 };
 
+class CallExpr : public Expr
+{
+public:
+  CallExpr(const Expr *callee, const std::vector<Expr *> &args)
+      : callee(callee), args(std::move(args)) {}
+
+  const Expr *callee;
+  const std::vector<Expr *> args;
+};
+
+class UnaryExpr : public Expr
+{
+public:
+  UnaryExpr(const std::string &op, const Expr *expr)
+      : op(op), expr(expr) {}
+
+  const std::string op;
+  const Expr *expr;
+};
+
+class BinExpr : public Expr
+{
+public:
+  BinExpr(const Expr *left, const std::string &op, const Expr *right)
+      : left(left), op(op), right(right) {}
+
+  const Expr *left;
+  const std::string op;
+  const Expr *right;
+};
+
+class PrefExpr : public Expr
+{
+public:
+  PrefExpr(const Expr *expr) : expr(expr) {}
+
+  const Expr *expr;
+};
+
+class Identifier : public Expr
+{
+public:
+  Identifier(const std::string &name) : name(name) {}
+
+  const std::string name;
+};
+
 class IntValue : public Expr
 {
 public:
-  const long value;
   IntValue(const long value) : value(value) {}
+
+  const long value;
 };
 
 class FloatValue : public Expr
 {
 public:
-  const double value;
   FloatValue(const double value) : value(value) {}
-};
 
-class Function : public Expr
+  const double value;
-{
-};
-
-class List : public Expr
-{
 };

compiler/include/ParserDriver.h

@@ -17,6 +17,7 @@ public:
 
   std::string file;
   yy::location location;
+  Module *module;
 };
 
 #define YY_DECL yy::parser::symbol_type yylex(ParserDriver &driver)

compiler/lexer.ll

@@ -4,10 +4,10 @@
 
 #define yyterminate() yy::parser::make_END(yy::location())
 
-#define _token(token) yy::parser::make_##token(driver.location)
+#define _token(token) yy::parser::make_##token(yytext, driver.location)
 %}
 
-%option noyywrap nounput noinput batch debug
+%option noyywrap nounput noinput batch
 
 digit       [0-9]
 bindigit    [0-1]
@@ -16,6 +16,8 @@ hexdigit    [0-9a-fA-F]
 
 letter      [a-zA-Z]
 
+opchar      ("+"|"-"|"*"|"/"|"%"|"!"|"&"|"$"|"§"|"|"|"="|"<"|">"|"?"|"~"|"#"|":"|"^"|"\\")
+
 whitespace  [ \n\t\r\v]+
 
 %%
@@ -27,14 +29,18 @@ whitespace  [ \n\t\r\v]+
 
 {digit}*"."{digit}+   { return yy::parser::make_FLOAT(std::strtod(yytext, NULL), driver.location); }
 
-// TODO: operator '==' won't work this way
+
-"->"    { return _token(R_ARR); }
+{opchar}+     { std::string text = yytext;
-"<-"    { return _token(L_ARR); }
+                if (text == "=") return _token(EQUALS);
+                if (text == "->") return _token(RARR);
+                return _token(OPERATOR); }
+
+","           { return _token(COMMA); }
 ";"           { return _token(SEMI); }
-"="     { return _token(EQUALS); } 
+"("           { return _token(LPAREN); }
-
+")"           { return _token(RPAREN); }
-("+"|"-"|"*"|"/"|"%"|"!"|"&"|"$"|"§"|"|"|"="|"<"|">"|"?"|"~"|"#"|":"|"^"|"\\")+  { return _token(OPERATOR); }
+"["           { return _token(LBRACKET); }
-
+"]"           { return _token(RBRACKET); }
 
 "fn"          { return _token(FN); }
 "unop"        { return _token(UNOP); }
@@ -44,7 +50,9 @@ whitespace  [ \n\t\r\v]+
 "declare"     { return _token(DECLARE); }
 
 
-{letter}({digit}|{letter})*   { return yy::parser::make_IDENTIFIER(yytext, driver.location); }
+{letter}({digit}|{letter})*   { return _token(IDENTIFIER); }
+
+<<EOF>>   { return yy::parser::make_END(driver.location); }
 
 {whitespace}                  { ; }
 

compiler/parser.yy

@@ -32,15 +32,19 @@ class ParserDriver;
 %token <long> INT "integer literal"
 %token <double> FLOAT "float literal"
 
-%token <std::string> OPERATOR "operator";
+%token <std::string>
-%token <std::string> IDENTIFIER "identifier"
+  OPERATOR "operator"
+  IDENTIFIER "identifier"
 
-%token L_ARR "<-"
+  RARR "->"
-%token R_ARR "->"
+  EQUALS "="
-%token EQUALS "="
+  COMMA ","
-%token SEMI ";"
+  SEMI ";"
+  LPAREN "("
+  RPAREN ")"
+  LBRACKET "["
+  RBRACKET "]"
 
-%token
   FN "fn"
   UNOP "unop"
   BINOP "binop"
@@ -52,47 +56,144 @@ class ParserDriver;
 %token END 0 "end of file"
 
 %type <std::vector<Stmt *>> stmts;
-%type <Stmt *> stmt;
+%type <Stmt *>
-%type <Decl *> decl;
+  stmt
-%type <Expr *> expr;
+  decl
-%type <Value *> literal;
+;
 
-%type <std::string> operator;
+%type <FnDecl *> fnDecl;
-%type <std::string> identifier;
+%type <ValDecl *> valDecl;
-%type <std::string> keyword;
+
+%type <Type *>
+  type
+  type0
+;
+
+%type <std::vector<Expr *>>
+  exprs
+  args
+;
+
+%type <Expr *>
+  expr
+  expr3
+  expr2
+  expr1
+  expr0
+  literal
+;
+
+%type <FnArg> fnArg;
+%type <std::vector<FnArg>>
+  fnArgs
+  fnArgs0
+;
+
+%type <std::vector<std::string>>
+  lambdaArgs
+  identifiers
+;
+
+%type <std::string>
+  operator
+  identifier
+  keyword
+;
+
+%start module
 
 %%
 
-module: stmts;
+module: stmts END { driver.module = new Module($1); };
 
 stmts: stmt         { $$ = std::vector<Stmt *>(); $$.push_back($1); }
      | stmts stmt   { $1.push_back($2); $$ = std::move($1); }
      ;
 
-stmt: decl  { $$ = $1; }
+stmt: decl;
+
+decl: valDecl { $$ = $1; }
+    | fnDecl  { $$ = $1; }
     ;
 
-decl: VAL identifier "=" expr ";"   { $$ = new Decl($2, $4); }
+valDecl: VAL identifier "=" expr ";"        { $$ = new ValDecl($2, nullptr, $4); }
+       | VAL identifier type "=" expr ";"   { $$ = new ValDecl($2, $3, $5); }
        ;
 
-expr: literal       { $$ = $1; }
+fnDecl: FN identifier "[" fnArgs "]" type "=" expr ";"  { $$ = new FnDecl($2, $4, $6, $8); }
       ;
 
+fnArgs: %empty    { $$ = std::vector<FnArg>(); }
+      | fnArgs0   { $$ = std::move($1); }
+      ;
+
+fnArgs0: fnArg              { $$ = std::vector<FnArg>(); $$.push_back($1); }
+       | fnArgs0 "," fnArg  { $$.push_back($3); $$ = std::move($1); }
+       ;
+
+fnArg: identifier type  { $$ = FnArg($1, $2); }
+     ;
+
+type: type0
+    ;
+
+type0: identifier     { $$ = new NamedType($1); }
+     | "(" type ")"   { $$ = $2; }   
+     ;
+
+exprs: expr             { $$ = std::vector<Expr *>(); $$.push_back($1); }
+     | exprs "," expr   { $1.push_back($3); $$ = std::move($1); }
+     ;
+
+expr: expr3 { $$ = $1; }
+    ;
+
+expr3: expr2                          { $$ = $1; }
+     | "[" lambdaArgs "]" "->" expr   { $$ = NULL; }
+     ;
+
+expr2: expr1                  { $$ = $1; }
+     | expr2 operator expr1   { $$ = new BinExpr($1, $2, $3); }
+     ;
+
+expr1: expr0            { $$ = $1; }
+     | operator expr1   { $$ = new UnaryExpr($1, $2); }
+     ;
+
+expr0: literal              { $$ = $1; }
+     | identifier           { $$ = new Identifier($1); }
+     | expr0 "[" args "]"   { $$ = new CallExpr($1, $3); }
+     | "(" expr ")"         { $$ = new PrefExpr($2); }
+     ;
+
+lambdaArgs: %empty        { $$ = std::vector<std::string>(); }
+          | identifiers   { $$ = std::move($1); }
+          ;
+
+args: %empty  { $$ = std::vector<Expr *>(); }
+    | exprs   { $$ = std::move($1); }
+    ;
+
+
 literal: INT        { $$ = new IntValue($1); }
        | FLOAT      { $$ = new FloatValue($1); }
        ;
 
-operator: L_ARR     { $$ = $1; }
+operator: "->"      { $$ = $1; }
-        | R_ARR     { $$ = $1; }
+        | "="       { $$ = $1; }
-        | EQUALS    { $$ = $1; }
         | OPERATOR  { $$ = $1; }
         ;
 
+identifiers: identifier                   { $$ = std::vector<std::string>(); $$.push_back($1); }
+           | identifiers "," identifier   { $1.push_back($3); $$ = std::move($1); }
+           ;
+
 identifier: keyword     { $$ = $1; }
           | IDENTIFIER  { $$ = $1; }
           ;
 
 keyword: FN       { $$ = $1; }
+       | VAL      { $$ = $1; }
        | UNOP     { $$ = $1; }
        | BINOP    { $$ = $1; }
        | IMPORT   { $$ = $1; }
@@ -101,6 +202,6 @@ keyword: FN       { $$ = $1; }
 
 %%
 
-void yy::parser::error (const location_type& l, const std::string& m) {
+void yy::parser::error(const location_type& l, const std::string& m) {
   std::cerr << l << ": " << m << '\n';
 }

compiler/src/main.cpp

@@ -1,5 +1,30 @@
-#include <string>
+#include "ParserDriver.h"
 
-int main()
+#include <iostream>
+#include <fstream>
+#include <sstream>
+
+std::stringstream readFile(const std::string &path)
 {
+  std::ifstream t(path);
+  std::stringstream buffer;
+  buffer << t.rdbuf();
+  return buffer;
+}
+
+int main(int argc, char *argv[])
+{
+  if (argc != 2)
+  {
+    std::cout << "Usage:" << std::endl
+              << "\t" << argv[0] << " <path/to/file>" << std::endl;
+    exit(EXIT_FAILURE);
+  }
+
+  auto buf = readFile(argv[1]);
+  auto parser = ParserDriver();
+  parser.parse(buf.str());
+
+  auto fn = (FnDecl *)parser.module->stmts.at(0);
+  std::cout << fn->name << std::endl;
 }

examples/42.plsm

@@ -0,0 +1 @@
+fn main[] Int = 42;

examples/stdlib.plsm

@@ -1,30 +1,29 @@
 type Bool = class {
-  declare unop ! -> Bool;
+  declare unop ! Bool;
 
-  declare binop &&(b Bool) -> Bool;
+  declare binop &&(b Bool) Bool;
-  declare binop ||(b Bool) -> Bool;
+  declare binop ||(b Bool) Bool;
 };
 
 type Int = class {
-  declare unop + -> Int;
+  declare unop + Int;
-  declare unop - -> Int;
+  declare unop - Int;
 
-  declare binop ==(b Int) -> Bool;
+  declare binop ==(Int) Bool;
-  declare binop !=(b Int) -> Bool;
+  declare binop !=(Int) Bool;
-  declare binop >(b Int) -> Bool;
+  declare binop >(Int) Bool;
-  declare binop <(b Int) -> Bool;
+  declare binop <(Int) Bool;
-  declare binop >=(b Int) -> Bool;
+  declare binop >=(Int) Bool;
-  declare binop <=(b Int) -> Bool;
+  declare binop <=(Int) Bool;
 
-  declare binop +(b Int) -> Int;
+  declare binop +(Int) Int;
-  declare binop -(b Int) -> Int;
+  declare binop -(Int) Int;
-  declare binop *(b Int) -> Int;
+  declare binop *(Int) Int;
-  declare binop /(b Int) -> Int;
+  declare binop /(Int) Int;
-  declare binop %(b Int) -> Int;
+  declare binop %(Int) Int;
 
-  declare fn str() -> String; 
+  declare factory (f Float);
+  declare factory ();
+
+  declare fn str() String; 
 };
-
-type String = class (data List[Char]) {
-
-}

lib/CMakeLists.txt

@@ -6,4 +6,5 @@ set(CMAKE_C_STANDARD 11)
 
 file(GLOB_RECURSE sources ${CMAKE_SOURCE_DIR}/src/*.c)
 include_directories(${CMAKE_SOURCE_DIR}/include)
-add_library(libplsm SHARED ${sources})
+add_library(plsm SHARED ${sources})
+target_link_libraries(plsm gc)

lib/include/function.h

@@ -1,6 +0,0 @@
-#pragma once
-
-typedef struct
-{
-
-} function_t;

lib/include/init.h

@@ -0,0 +1,3 @@
+#pragma once
+
+void plsm_init();

lib/include/list.h

@@ -1,16 +0,0 @@
-#pragma once
-
-#include "type.h"
-#include "value.h"
-
-typedef struct
-{
-  value_t *value;
-  list_node_t *next;
-} list_node_t;
-
-typedef struct
-{
-  type_t *type;
-  list_node_t *head;
-} list_t;

lib/include/macros.h

@@ -1,7 +1,8 @@
 #pragma once
 
-#include <stdlib.h>
 #include <stdio.h>
+#include <stdint.h>
+#include <stdlib.h>
 
 #define DIE             \
   {                     \
@@ -14,6 +15,35 @@
     DIE;                  \
   }
 
-#define MODINIT(module) void plsmmod_##module##_init()
+#define TYPE(type) plsm_##type##_t
+#define TYPESIZE(type) (sizeof(TYPE(type)) - sizeof(plsm_base_t))
 
-#define LIB(name) plsmlib_##name
+#define MAKETYPE(type) plsm_maketype_##type
+#define DEFMAKETYPE(type) plsm_type_t *MAKETYPE(type)
+
+#define MAKE(type) plsm_make_##type
+#define DEFMAKE(type) TYPE(type) MAKE(type)
+
+#define OP_ADD add
+#define OP_SUB sub
+#define OP_MUL mul
+#define OP_DIV div
+#define OP_MOD mod
+#define OP_EQ eq
+#define OP_NEQ neq
+#define OP_LAND land
+#define OP_LOR lor
+#define OP_GT gt
+#define OP_GTE gte
+#define OP_LT lt
+#define OP_LTE lte
+#define OP_NOT not
+
+#define UNOP(action, intype) plsm_unop_##action##_##intype
+#define DEFUNOP(action, restype, intype) TYPE(restype) UNOP(action, intype)(TYPE(intype) * it)
+
+#define BINOP(action, ltype, rtype) plsm_binop_##action##_##ltype##_##rtype
+#define DEFBINOP(action, restype, ltype, rtype) \
+  TYPE(restype)                                 \
+  BINOP(action, ltype, rtype)                   \
+  (TYPE(ltype) * it, TYPE(rtype) * other)

lib/include/type.h

@@ -2,13 +2,23 @@
 
 #include "macros.h"
 
-typedef struct type_t type_t;
+typedef struct plsm_type_t plsm_type_t;
 
-typedef struct type_t
+typedef struct plsm_type_t
 {
-  type_t *base;
+  plsm_type_t *base;
-  int opaque;
   char *name;
-} type_t;
+  uint32_t size;
+} plsm_type_t;
 
-type_t *LIB(makeType)(type_t *base, int opaque, char *name);
+plsm_type_t *plsm_maketype(plsm_type_t *base, char *name, uint32_t size);
+
+typedef struct plsm_base_t
+{
+  plsm_type_t *type;
+} plsm_base_t;
+
+typedef struct TYPE(Bool) TYPE(Bool);
+
+TYPE(Bool)
+plsm_instanceof(plsm_base_t *value, plsm_type_t *type);

lib/include/types/bool.h

@@ -0,0 +1,15 @@
+#pragma once
+
+#include "macros.h"
+#include "type.h"
+
+#include <stdint.h>
+
+typedef struct TYPE(Bool)
+{
+  plsm_type_t *type;
+  uint8_t value;
+} TYPE(Bool);
+
+DEFMAKE(Bool)
+(intptr_t value);

lib/include/types/int.h

@@ -0,0 +1,23 @@
+#pragma once
+
+#include "macros.h"
+#include "type.h"
+
+#include <gc.h>
+
+typedef struct TYPE(Int)
+{
+  plsm_type_t *type;
+  int64_t value;
+} TYPE(Int);
+
+DEFMAKE(Int)
+(int64_t value);
+
+DEFBINOP(BINOP_ADD, Int, Int, Int);
+DEFBINOP(BINOP_SUB, Int, Int, Int);
+DEFBINOP(BINOP_MUL, Int, Int, Int);
+DEFBINOP(BINOP_DIV, Int, Int, Int);
+DEFBINOP(BINOP_MOD, Int, Int, Int);
+
+DEFBINOP(BINOP_EQ, Bool, Int, Int);

lib/include/types/types.h

@@ -0,0 +1,4 @@
+#pragma once
+
+#include "int.h"
+#include "bool.h"

lib/include/value.h

@@ -1,9 +0,0 @@
-#pragma once
-
-#include "type.h"
-
-typedef struct
-{
-  type_t *type;
-  void *value;
-} value_t;

lib/src/function.c

@@ -1 +0,0 @@
-#include "function.h"

lib/src/init.c

@@ -0,0 +1,15 @@
+#include "init.h"
+
+#include <gc.h>
+
+void plsm_init()
+{
+  static int initialized = 0;
+
+  if (initialized)
+    return;
+
+  GC_INIT();
+
+  initialized = 1;
+}

lib/src/module.c

@@ -1,5 +0,0 @@
-#include "macros.h"
-
-MODINIT(stdlib)
-{
-}

lib/src/type.c

@@ -1,16 +1,28 @@
 #include "type.h"
 
-#include <stdlib.h>
+#include <gc.h>
+#include <stddef.h>
+#include "types/bool.h"
 
-type_t *LIB(makeType)(type_t *base, int opaque, char *name)
+plsm_type_t *plsm_maketype(plsm_type_t *base, char *name, uint32_t size)
 {
-  type_t *type = malloc(sizeof(type_t));
+  plsm_type_t *type = GC_malloc(sizeof(plsm_type_t));
-  if (!type)
+  if (type == NULL)
-    DIE_ERRNO("malloc");
+    DIE_ERRNO("GC_malloc");
 
   type->base = base;
-  type->opaque = opaque;
   type->name = name;
+  type->size = size;
 
   return type;
 }
+
+TYPE(Bool)
+plsm_instanceof(plsm_base_t *value, plsm_type_t *type)
+{
+  for (plsm_type_t *t = value->type; t != NULL; t = t->base)
+    if (t == type)
+      return MAKE(Bool)(1);
+
+  return MAKE(Bool)(0);
+}

lib/src/types.c

@@ -1,12 +0,0 @@
-#include "type.h"
-
-type_t *LIB(typeAny), *LIB(typeNum), *LIB(typeInt), *LIB(typeFloat);
-
-void LIB(initTypes)()
-{
-  LIB(typeAny) = LIB(makeType)(NULL, "Any");
-
-  LIB(typeNum) = LIB(makeType)(LIB(typeAny), "Num");
-  LIB(typeInt) = LIB(makeType)(LIB(typeNum), "Int");
-  LIB(typeFloat) = LIB(makeType)(LIB(typeNum), "Float");
-}

lib/src/types/bool.c

@@ -0,0 +1,27 @@
+#include "types/bool.h"
+
+#include <gc.h>
+
+DEFMAKE(Bool)
+(intptr_t value)
+{
+  return (TYPE(Bool)){
+      .type = NULL,
+      .value = value != 0,
+  };
+}
+
+DEFUNOP(OP_NOT, Bool, Bool)
+{
+  return MAKE(Bool)(!it->value);
+}
+
+DEFBINOP(OP_LAND, Bool, Bool, Bool)
+{
+  return MAKE(Bool)(it->value && other->value);
+}
+
+DEFBINOP(OP_LOR, Bool, Bool, Bool)
+{
+  return MAKE(Bool)(it->value || other->value);
+}

lib/src/types/int.c

@@ -0,0 +1,64 @@
+#include "types/int.h"
+
+#include <gc.h>
+#include "type.h"
+#include "types/bool.h"
+
+DEFMAKETYPE(Int)
+()
+{
+  static plsm_type_t *type = NULL;
+  if (type)
+    return type;
+
+  return (type = plsm_maketype(NULL, "Int", TYPESIZE(Int)));
+}
+
+DEFMAKE(Int)
+(int64_t value)
+{
+  return (TYPE(Int)){
+      .type = NULL,
+      .value = value,
+  };
+}
+
+DEFUNOP(OP_ADD, Int, Int)
+{
+  return *it;
+}
+
+DEFUNOP(OP_SUB, Int, Int)
+{
+  return MAKE(Int)(it->value * -1);
+}
+
+DEFBINOP(OP_ADD, Int, Int, Int)
+{
+  return MAKE(Int)(it->value + other->value);
+}
+
+DEFBINOP(OP_SUB, Int, Int, Int)
+{
+  return MAKE(Int)(it->value - other->value);
+}
+
+DEFBINOP(OP_MUL, Int, Int, Int)
+{
+  return MAKE(Int)(it->value * other->value);
+}
+
+DEFBINOP(OP_DIV, Int, Int, Int)
+{
+  return MAKE(Int)(it->value / other->value);
+}
+
+DEFBINOP(OP_MOD, Int, Int, Int)
+{
+  return MAKE(Int)(it->value % other->value);
+}
+
+DEFBINOP(OP_EQ, Bool, Int, Int)
+{
+  return MAKE(Bool)((intptr_t)(it->value == other->value));
+}