feat: basic parsing of example.rx is ready

.vscode/settings.json

@@ -1,5 +1,6 @@
 {
     "files.associations": {
-        "print": "cpp"
+        "print": "cpp",
+        "cctype": "cpp"
     }
 }

example.rx

@@ -1,7 +1,7 @@
-extern putchar;
+extern putchar
 
 fn main() {
-    local a = 34;
-    local b = 35;
-    putchar(a + b);
+    local a = 34
+    local b = 35
+    putchar(a + b)
 }

include/ast.hpp

@@ -1,14 +1,21 @@
 #pragma once
 #include <cstring>
+#include <cassert>
+#include <vector>
 #include "lexer.hpp"
 
-enum NodeType
+enum class NodeType
 {
     Expression = 0,
+    // TODO: abstract "Literal" node type
+    IntLiteral,
     Extern,
     FnDecl,
     FnCall,
+    Variable,
     VarDecl,
+    Compound,
+    Program,
     COUNT_NODES,
 };
 
@@ -20,14 +27,23 @@ public:
 };
 
 #define NODE_TYPE(x) \
-    NodeType GetType() const override { return (x); }
+    NodeType GetType() const override { return NodeType::x; }
                      
 
 class ExpressionNode : public Node
 {
 public:
-    ExpressionNode()
-        : m_left(nullptr), m_right(nullptr) {}
+    enum class Operator
+    {
+        Plus = 0,
+        Minus,
+        Divide,
+        Multiply,
+        COUNT_OPERATORS,
+    };
+public:
+    ExpressionNode(Node* left, Node* right, Operator op)
+        : m_left(left), m_right(right), m_op(op) {}
     ~ExpressionNode() override {
         delete m_left;
         delete m_right;
@@ -37,9 +53,26 @@ public:
 public:
     Node* left() const { return m_left; }
     Node* right() const { return m_right; }
+    Operator op() const { return m_op; }
 private:
     Node* m_left;
     Node* m_right;
+    Operator m_op;
+};
+
+// TODO: Maybe just LiteralNode with double or int literals support
+class IntLiteralNode : public Node
+{
+public:
+    IntLiteralNode(long value)
+        : m_int_value(value) {}
+    ~IntLiteralNode() override {}
+
+    NODE_TYPE(IntLiteral)
+public:
+    long integer() const { return m_int_value; }
+private:
+    long m_int_value;
 };
 
 class ExternNode : public Node
@@ -91,6 +124,20 @@ private:
     Node* m_arg;
 };
 
+class VariableNode : public Node
+{
+public:
+    VariableNode(char* name)
+        : m_name(name) {}
+    ~VariableNode() override {
+        delete m_name;
+    }
+
+    NODE_TYPE(Variable)
+private:
+    char* m_name;
+};
+
 class VarDeclNode : public Node
 {
 public:
@@ -107,47 +154,239 @@ private:
     Node* m_value;
 };
 
+class CompoundNode : public Node
+{
+public:
+    CompoundNode() = default;
+
+    NODE_TYPE(Compound)
+
+    // --- Iteration ---
+    auto begin() { return m_nodes.begin(); }
+    auto end()   { return m_nodes.end();   }
+    auto begin() const { return m_nodes.begin(); }
+    auto end()   const { return m_nodes.end();   }
+
+    // --- Access by index ---
+    Node* operator[](size_t i) { return m_nodes[i]; }
+    const Node* operator[](size_t i) const { return m_nodes[i]; }
+
+    Node* at(size_t i) { return m_nodes.at(i); }
+    const Node* at(size_t i) const { return m_nodes.at(i); }
+
+    // --- Modifiers ---
+    void addNode(Node* n) { m_nodes.push_back(n); }
+    void removeNode(size_t idx) { m_nodes.erase(m_nodes.begin() + idx); }
+
+    // If you want full expose for iteration but not modification
+    const std::vector<Node*>& nodes() const { return m_nodes; }
+
+    size_t size() const { return m_nodes.size(); }
+    bool empty() const { return m_nodes.empty(); }
+private:
+    std::vector<Node*> m_nodes;
+};
+
+class ProgramNode : public Node
+{
+public:
+    ProgramNode() = default;
+
+    NODE_TYPE(Program)
+public:
+    void PushFunction(FnDeclNode* fn)
+    {
+        m_funcs.push_back(fn);
+    }
+
+    void PushExtern(ExternNode* extrn)
+    {
+        m_externs.push_back(extrn);
+    }
+private:
+    std::vector<FnDeclNode*> m_funcs;
+    std::vector<ExternNode*> m_externs;
+};
+
+
 class AstParser
 {
 public:
     AstParser(Lexer* lexer)
         : m_lexer(lexer) {}
 public:
+    ExternNode* ParseExtern()
+    {
+        m_lexer->NextExpect(TokenType::Id);
+        return new ExternNode(m_lexer->token().string);
+    }
+
+    FnDeclNode* ParseFnDecl()
+    {
+        // Function Declaration
+        m_lexer->NextExpect(TokenType::Id);
+        char *name = strdup(m_lexer->token().string);
+        m_lexer->NextExpect('(');
+        // TODO: parse parameters
+        m_lexer->NextExpect(')');
+        m_lexer->NextExpect('{');
+        auto compound = new CompoundNode();
+        while (m_lexer->seek_token()->token != '}')
+        {
+            compound->addNode(ParseStatement());
+        }
+        m_lexer->NextExpect('}');
+        return new FnDeclNode(name, compound);
+    }
+
+    FnCallNode* ParseFnCall(char* name)
+    {
+        // m_lexer->NextExpect(TokenType::Id);
+        // char* name = strdup(m_lexer->token().string);
+        m_lexer->NextExpect('(');
+        Node* arg = ParseExpression();
+        m_lexer->NextExpect(')');
+        return new FnCallNode(name, arg);
+    }
+
+    Node* ParseFactor()
+    {
+        auto token = m_lexer->seek_token();
+
+        switch (token->token)
+        {
+        case TokenType::IntLiteral: // integer
+        {
+            m_lexer->NextExpect(TokenType::IntLiteral);
+            auto node = new IntLiteralNode(m_lexer->token().int_number);
+            return node;
+        }
+        case TokenType::Id: // variable name or function call
+        {
+            m_lexer->NextExpect(TokenType::Id);
+            char *name = strdup(m_lexer->token().string);
+            token = m_lexer->seek_token();
+            if (token->token == '(')
+            {
+                return ParseFnCall(name);
+            }
+            return new VariableNode(name);
+        }
+        default:
+            fprintf(stderr, "%s:%d:%d: ERROR: unexpected token while parsing %ld\n", m_lexer->filename(), token->line_number, token->offset_start, token->token);
+            Exit(1);
+            break;
+        }
+
+        assert(0 && "unreachable");
+    }
+
+    Node* ParseTerm()
+    {
+        auto t = ParseFactor();
+
+        for (auto op = m_lexer->seek_token(); is_one_of(op->token, '/', '*'); op = m_lexer->seek_token())
+        {
+            m_lexer->NextToken();
+            ExpressionNode::Operator eop;
+            assert((int)ExpressionNode::Operator::COUNT_OPERATORS == 4 && "some operators may not be handled");
+            switch((char)op->token)
+            {
+            case '/':
+                eop = ExpressionNode::Operator::Divide;
+                break;
+            case '*':
+                eop = ExpressionNode::Operator::Multiply;
+                break;
+            default:
+                assert(false && "should be unreachable");
+                break;
+            }
+            auto expr = new ExpressionNode(t, ParseTerm(), eop);
+            t = expr;
+        }
+
+        return t;
+    }
+
+    Node* ParseExpression()
+    {
+        auto t = ParseTerm();
+
+        for (auto op = m_lexer->seek_token(); is_one_of(op->token, '+', '-'); op = m_lexer->seek_token())
+        {
+            m_lexer->NextToken();
+            ExpressionNode::Operator eop;
+            assert((int)ExpressionNode::Operator::COUNT_OPERATORS == 4 && "some operators may not be handled");
+            switch((char)op->token)
+            {
+            case '+':
+                eop = ExpressionNode::Operator::Plus;
+                break;
+            case '-':
+                eop = ExpressionNode::Operator::Minus;
+                break;
+            default:
+                assert(false && "should be unreachable");
+                break;
+            }
+            auto expr = new ExpressionNode(t, ParseTerm(), eop);
+            t = expr;
+        }
+
+        return t;
+    }
+
+    VarDeclNode* ParseVarDecl()
+    {
+        m_lexer->NextExpect(TokenType::Local);
+        m_lexer->NextExpect(TokenType::Id);
+        char *name = strdup(m_lexer->token().string);
+        m_lexer->NextExpect('=');
+        Node* value = ParseExpression();
+        return new VarDeclNode(name, value);
+    }
+
     Node* ParseStatement()
     {
+        auto token = m_lexer->seek_token();
+        // TODO: proper error handling
+        assert(token != nullptr && "next token should be available");
+        switch(token->token)
+        {
+        case TokenType::Local: return ParseVarDecl();
+        default: return ParseExpression();
+        }
+
+        assert(0 && "unreachable");
+        return nullptr;
+    }
+
+    ProgramNode* Parse()
+    {
+        auto program = new ProgramNode;
+
         while (m_lexer->NextToken())
         {
             auto token = m_lexer->token();
             switch(token.token)
             {
-            case Id:
-            {
-                if (strcmp(token.string, "extern") == 0)
-                {
-                    // Extern
-                    m_lexer->Eat(Id);
-                    return new ExternNode(m_lexer->token().string);
-                }
-                else if (strcmp(token.string, "fn") == 0)
-                {
-                    // Function Declaration
-                    m_lexer->Eat(Id);
-                    char* name = strdup(m_lexer->token().string);
-                    m_lexer->Eat('(');
-                    // TODO: parse parameters
-                    m_lexer->Eat(')');
-                    m_lexer->Eat('{');
-                    // TODO: parse function body
-                    m_lexer->Eat('}');
-                    return new FnDeclNode(name, nullptr);
-                }
+            case TokenType::Extern: program->PushExtern(ParseExtern()); break;
+            case TokenType::Fn: program->PushFunction(ParseFnDecl()); break;
+            default: {
+                fprintf(stderr, "%s:%d:%d: ERROR: unexpected token while parsing %ld\n", m_lexer->filename(), token.line_number, token.offset_start, token.token);
+                Exit(1);
+                break;
             }
-            break;
             }
         }
 
-        // TODO: report parse error
-        return nullptr;
+        return program;
+    }
+private:
+    void Exit(int status)
+    {
+        std::exit(status);
     }
 private:
     Lexer* m_lexer;

include/lexer.hpp

@@ -1,18 +1,51 @@
 #pragma once
 #include <iostream>
+#include <cstring>
 #include <cctype>
 
 #include "string.hpp"
 
-enum TokenType
+enum class TokenType
 {
     Eof = 256,
     Id,
     IntLiteral,
 
+    Extern,
+    Fn,
+    Local,
+
     Unknown,
 };
 
+template <typename... Ts>
+inline bool is_one_of(TokenType t, Ts... ts) {
+    return ((t == ts) || ...);
+}
+
+// operator== TokenType == char
+inline bool operator==(TokenType t, char c)
+{
+    return static_cast<char>(t) == c;
+}
+
+// operator== char == TokenType
+inline bool operator==(char c, TokenType t)
+{
+    return t == c;   // reuse the function above
+}
+
+// operator!=
+inline bool operator!=(TokenType t, char c)
+{
+    return !(t == c);
+}
+
+inline bool operator!=(char c, TokenType t)
+{
+    return !(t == c);
+}
+
 struct Token
 {
     TokenType token;
@@ -26,13 +59,32 @@ public:
     Token(TokenType t) : token(t) {}
     Token(TokenType t, long lnumber, long soffset, long eoffset)
         : token(t), line_number(lnumber), offset_start(soffset), offset_end(eoffset) {}
-    Token() : token(Unknown) {}
+    Token() : token(TokenType::Unknown) {}
 };
 
 class Lexer
 {
 public:
-    const Token& token() { return m_token; }
+    const Token& token() const { return m_token; }
+    const Token* seek_token()
+    {
+        auto s = m_token;
+        auto p = m_pos;
+        auto l = m_line;
+        auto lnl = m_last_newline;
+        if (!NextToken())
+        {
+            return nullptr;
+        }
+        auto seeked = m_token;
+        m_token = s;
+        m_pos = p;
+        m_line = l;
+        m_last_newline = lnl;
+        return new Token(seeked);
+    }
+
+    const char* filename() const { return m_filename; }
 public:
     Lexer(char* filename, StringView code)
         : m_filename(filename), m_code(code) {}
@@ -74,8 +126,24 @@ public:
                 s.PushChar(m_code.data[m_pos++]);
             }
             s.PushChar('\0');
-            m_token = Token(Id, m_line, offset_start, m_pos - m_last_newline);
+            m_token = Token(TokenType::Id, m_line, offset_start, m_pos - m_last_newline);
             m_token.string = s.data;
+
+            if (strcmp("extern", m_token.string) == 0)
+            {
+                m_token.token = TokenType::Extern;
+            }
+
+            if (strcmp("fn", m_token.string) == 0)
+            {
+                m_token.token = TokenType::Fn;
+            }
+
+            if (strcmp("local", m_token.string) == 0)
+            {
+                m_token.token = TokenType::Local;
+            }
+
             return true;
         }
 
@@ -91,7 +159,7 @@ public:
                 s.PushChar(m_code.data[m_pos++]);
             }
             s.PushChar('\0');
-            m_token = Token(IntLiteral, m_line, offset_start, m_pos - m_last_newline);
+            m_token = Token(TokenType::IntLiteral, m_line, offset_start, m_pos - m_last_newline);
             m_token.int_number = std::strtol(s.data, nullptr, hex ? 16 : 10);
             m_token.string = s.data;
             return true;
@@ -101,23 +169,23 @@ public:
         return true;
     }
 
-    void Eat(TokenType expected)
+    void NextExpect(TokenType expected)
     {
         if (!NextToken())
         {
-            fprintf(stderr, "%s:%d:%d: ERROR: expected %ld, but got EOF", m_filename, 0, 0, expected);
+            fprintf(stderr, "%s:%d:%d: ERROR: expected %ld, but got EOF\n", m_filename, 0, 0, expected);
             Exit(1);
         }
         if (token().token != expected)
         {
-            fprintf(stderr, "%s:%d:%d: ERROR: expected %ld, but got %ld", m_filename, token().line_number, token().offset_start, expected);
+            fprintf(stderr, "%s:%d:%d: ERROR: expected %ld, but got %ld\n", m_filename, token().line_number, token().offset_start, expected, token().token);
             Exit(1);
         }
     }
 
-    void Eat(char expected)
+    void NextExpect(char expected)
     {
-        Eat((TokenType)expected);
+        NextExpect((TokenType)expected);
     }
 private:
     void Exit(int status)

src/main.cpp

@@ -8,9 +8,9 @@ void dump_tokens(const char* filename, Lexer* lexer)
 {
     while (lexer->NextToken()) {
         std::print("{}:{}:{}: ", filename, lexer->token().line_number, lexer->token().offset_start);
-        if (lexer->token().token == Id)
+        if (lexer->token().token == TokenType::Id)
             std::println("id = {}", lexer->token().string);
-        else if (lexer->token().token == IntLiteral)
+        else if (lexer->token().token == TokenType::IntLiteral)
             std::println("int = {}", lexer->token().int_number);
         else 
             std::println("token = {}", (char)lexer->token().token);
@@ -48,8 +48,7 @@ int main(int argc, char** argv)
 
     AstParser parser(&lexer);
 
-    auto node = parser.ParseStatement();
-    ExternNode* extrn = reinterpret_cast<ExternNode*>(node);
+    auto program = parser.Parse();
 
     return 0;
 }