fix ast_util print for newly added node kind

basic expr parsing (not complex like compound, lambda)

README.md

@@ -22,7 +22,7 @@
 
 ## Syntax Spec
 
-```spec
+```c
 program     := defn*
 
 defn        := VAL type ID ;
@@ -55,3 +55,56 @@ lambda      := param_list compound
 
 compound    := LCURLY (stmt)* expr? RCURLY
 ```
+
+### AST Spec
+
+```c
+NODE_PROGRAM:
+    token: PROGRAM
+    children: NODE_DEFN*
+
+NODE_DEFN:
+    token: VAL
+    children: NODE_TYPE, TOKEN_ID, (NODE_EXPR)?
+    children_count: 2 | 3
+
+NODE_TYPE_SIMPLE:
+    token: ID  
+    children: NODE_ID NODE_TYPE_STAR*
+    children_count: 1+
+NODE_TYPE_COMPLEX:
+    token: COMPLEX_TYPE
+    children: NODE_TYPE_PARAM NODE_TYPE_OUT
+NODE_TYPE_PARAM
+    token: TYPE_PARAM
+    children: (NODE_TYPE | NODE_TYPE_COMPLEX)*
+    children_count: 0+
+NODE_TYPE_OUT
+    token: TYPE_OUT
+    children: (NODE_TYPE | NODE_TYPE_COMPLEX)?
+
+NODE_EXPR:
+    token: EXPR
+    children: (atom)+
+// atom definition
+NODE_NUM:
+    token: NUM
+NODE_STR:
+    token: STR
+NODE_LAMBDA:
+    token: LAMBDA
+    children: NODE_PARAM_LIST NODE_COMPOUND
+NODE_COMPOUND:
+    token: COMPOUND
+    children: (NODE_STMT)* (NODE_EXPR)?
+    children_count: 0+
+    
+
+NODE_PARAM_LIST:
+    token: PARAM_LIST
+    children: NODE_PARAM*
+NODE_PARAM:
+    token: PARAM
+    children: NODE_TYPE, TOKEN_ID
+
+```

include/ast_util.h

@@ -0,0 +1,61 @@
+#pragma once
+
+#include <stddef.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+#include "globals.h"
+
+void ast_node_add_child(ASTNode *parent, ASTNode *child);
+
+void ast_node_free(ASTNode *node);
+
+void ast_node_print(ASTNode *node, int depth);
+
+/*
+    NODE SPECIFIC FUNCTIONS
+*/
+
+ASTNode *ast_node_program();
+
+ASTNode *ast_node_defn(Token tok_val, ASTNode *type, ASTNode *id, ASTNode *expr);
+
+ASTNode *ast_node_type_simple(Token tok_id);
+
+ASTNode *ast_node_type_complex(Token tok_bracket, ASTNode *type_param, ASTNode *type_out);
+
+ASTNode *ast_node_type_param();
+
+ASTNode *ast_node_type_out();
+
+ASTNode *ast_node_type_star(Token tok_star);
+
+ASTNode *ast_node_type_void();
+
+ASTNode *ast_node_id(Token id);
+
+ASTNode *ast_node_num(Token num);
+
+ASTNode *ast_node_str(Token str);
+
+ASTNode *ast_node_star(Token star);
+
+ASTNode *ast_node_andref(Token andref);
+
+ASTNode *ast_node_expr(Token caller);
+
+ASTNode *ast_node_param_list(Token tok_lparen);
+
+ASTNode *ast_node_param(Token id, ASTNode *type);
+
+ASTNode *ast_node_compound(Token tok_lcurly);
+
+ASTNode *ast_node_lambda(ASTNode *param_list, ASTNode *body);
+
+ASTNode *ast_node_stmt_return(Token tok_return, ASTNode *expr);
+
+ASTNode *ast_node_stmt_expr(ASTNode *expr);
+
+ASTNode *ast_node_stmt_if(Token tok_if, ASTNode *condition, ASTNode *then_branch, ASTNode *else_branch);
+
+ASTNode *ast_node_stmt_set(Token tok_dollar, ASTNode *id, ASTNode *expr);

include/globals.h

@@ -31,7 +31,13 @@ typedef enum {
     ELSE,
 
     EOF_TOKEN,
-    ERROR
+    ERROR,
+    // for parser use
+
+    PARSER_USE,
+
+    VOID,
+
 } TokenType;
 
 typedef struct {
@@ -48,8 +54,43 @@ typedef struct {
 /**
  * AST Node Definitions
  */
+
+typedef enum {
+    NODE_PROGRAM,
+    NODE_DEFN,
+
+    NODE_TYPE_SIMPLE,
+    NODE_TYPE_COMPLEX,
+    NODE_TYPE_PARAM,
+    NODE_TYPE_OUT,
+    NODE_TYPE_STAR,
+
+    NODE_EXPR,
+
+    NODE_ID,
+    NODE_NUM,
+    NODE_STR,
+    NODE_STAR,
+    NODE_ANDREF,
+
+    NODE_PARAM_LIST,
+    NODE_PARAM,
+
+    NODE_LAMBDA,
+    NODE_COMPOUND,
+
+    NODE_STMT_RETURN,
+    NODE_STMT_EXPR,
+    NODE_STMT_SET,
+    NODE_STMT_IF,
+
+} NodeKind;
+
 typedef struct ASTNode {
+    NodeKind kind;
     Token token;
+
     struct ASTNode **children;
-    size_t child_count;
+    size_t count;
+    size_t capacity;
 } ASTNode;

include/parse.h

@@ -1,8 +1,10 @@
 #pragma once
 #include <stdint.h>
+#include <stdbool.h>
 
 #include "globals.h"
 #include "lex.h"
+#include "ast_util.h"
 
 typedef struct Parser {
     Lexer *lexer;
@@ -17,10 +19,6 @@ Parser *parser_new(Lexer *lexer);
 
 void parser_free(Parser *parser);
 
-static void parser_next(Parser *parser);
-
-static void parser_expect(Parser *parser, TokenType type);
-
 /*
     PARSER PARSE FUNCTIONS
 */
@@ -30,6 +28,10 @@ ASTNode *parser_parse_defn(Parser *parser);
 
 ASTNode *parser_parse_type(Parser *parser);
 
+ASTNode *parser_parse_type_simple(Parser *parser);
+
+ASTNode *parser_parse_type_complex(Parser *parser);
+
 ASTNode *parser_parse_expr(Parser *parser);
 
 ASTNode *parser_parse_atom(Parser *parser);

src/ast_util.c

@@ -0,0 +1,280 @@
+#include "ast_util.h"
+#include "util.h"
+
+static ASTNode *ast_node_new(NodeKind kind, Token token) {
+    ASTNode *node = malloc(sizeof(ASTNode));
+    node->kind = kind;
+    node->token = token;
+    node->children = NULL;
+    node->count = 0;
+    node->capacity = 0;
+    return node;
+}
+
+void ast_node_add_child(ASTNode *parent, ASTNode *child) {
+    if (parent->count >= parent->capacity) {
+        size_t new_capacity = parent->capacity == 0 ? 4 : parent->capacity * 2;
+        parent->children = realloc(parent->children, new_capacity * sizeof(ASTNode *));
+        parent->capacity = new_capacity;
+    }
+    parent->children[parent->count++] = child;
+}
+
+void ast_node_free(ASTNode *node) {
+    for (size_t i = 0; i < node->count; i++) {
+        ast_node_free(node->children[i]);
+    }
+    free(node->children);
+    free(node);
+}
+
+void ast_node_print(ASTNode *node, int depth) {
+    for (int i = 0; i < depth; i++) {
+        printf("  ");
+    }
+    printf("NodeKind: ");
+
+    switch (node->kind) {
+        case NODE_PROGRAM:
+            printf("NODE_PROGRAM");
+            break;
+        case NODE_DEFN:
+            printf("NODE_DEFN");
+            break;
+        case NODE_TYPE_SIMPLE:
+            printf("NODE_TYPE_SIMPLE");
+            break;
+        case NODE_TYPE_COMPLEX:
+            printf("NODE_TYPE_COMPLEX");
+            break;
+        case NODE_TYPE_PARAM:
+            printf("NODE_TYPE_PARAM");
+            break;
+        case NODE_TYPE_OUT:
+            printf("NODE_TYPE_OUT");
+            break;
+        case NODE_TYPE_STAR:
+            printf("NODE_TYPE_STAR");
+            break;
+        case NODE_EXPR:
+            printf("NODE_EXPR");
+            break;
+        case NODE_ID:
+            printf("NODE_ID");
+            break;
+        case NODE_NUM:
+            printf("NODE_NUM");
+            break;
+        case NODE_STR:
+            printf("NODE_STR");
+            break;
+        case NODE_STAR:
+            printf("NODE_STAR");
+            break;
+        case NODE_ANDREF:
+            printf("NODE_ANDREF");
+            break;
+        case NODE_PARAM_LIST:
+            printf("NODE_PARAM_LIST");
+            break;
+        case NODE_PARAM:
+            printf("NODE_PARAM");
+            break;
+        case NODE_LAMBDA:
+            printf("NODE_LAMBDA");
+            break;
+        case NODE_COMPOUND:
+            printf("NODE_COMPOUND");
+            break;
+        case NODE_STMT_RETURN:
+            printf("NODE_STMT_RETURN");
+            break;
+        case NODE_STMT_EXPR:
+            printf("NODE_STMT_EXPR");
+            break;
+        case NODE_STMT_SET:
+            printf("NODE_STMT_SET");
+            break;
+        case NODE_STMT_IF:
+            printf("NODE_STMT_IF");
+            break;
+    }
+
+    printf(", Token: ");
+    print_token(node->token);
+
+    printf("\n");
+    for (size_t i = 0; i < node->count; i++) {
+        ast_node_print(node->children[i], depth + 1);
+    }
+}
+
+/*
+    NODE SPECIFIC FUNCTIONS
+*/
+
+ASTNode *ast_node_program() {
+    Token tok = (Token) {
+            .type = EOF_TOKEN,
+            .line = 0,
+            .data = {0}};
+    ASTNode *node =
+            ast_node_new(NODE_PROGRAM, tok);
+
+    return node;
+}
+
+ASTNode *ast_node_defn(
+        Token tok_val /*VAL token*/, ASTNode *type, ASTNode *id, ASTNode *expr) {
+    ASTNode *node =
+            ast_node_new(NODE_DEFN, tok_val);
+    ast_node_add_child(node, type);
+    ast_node_add_child(node, id);
+
+    if (expr != NULL) {
+        ast_node_add_child(node, expr);
+    }
+    return node;
+}
+
+ASTNode *ast_node_type_simple(Token tok_id) {
+    ASTNode *node =
+            ast_node_new(NODE_TYPE_SIMPLE, tok_id);
+    return node;
+}
+
+ASTNode *ast_node_type_complex(
+        Token tok_bracket, ASTNode *type_param, ASTNode *type_out) {
+    ASTNode *node = ast_node_new(NODE_TYPE_COMPLEX, tok_bracket);
+    type_param->token = tok_bracket;
+    type_out->token = tok_bracket;
+    ast_node_add_child(node, type_param);
+    ast_node_add_child(node, type_out);
+    return node;
+}
+
+ASTNode *ast_node_type_param() {
+    ASTNode *node =
+            ast_node_new(NODE_TYPE_PARAM, (Token) {0});
+    return node;
+}
+
+ASTNode *ast_node_type_out() {
+    ASTNode *node =
+            ast_node_new(NODE_TYPE_OUT, (Token) {0});
+    return node;
+}
+
+ASTNode *ast_node_type_star(Token tok_star) {
+    ASTNode *node =
+            ast_node_new(NODE_TYPE_STAR, tok_star);
+    return node;
+}
+
+ASTNode *ast_node_type_void() {
+    ASTNode *node =
+            ast_node_new(NODE_TYPE_SIMPLE, (Token) {.type = VOID, .line = 0, .data = {0}});
+    return node;
+}
+
+ASTNode *ast_node_id(Token id) {
+    ASTNode *node =
+            ast_node_new(NODE_ID, id);
+    return node;
+}
+
+ASTNode *ast_node_num(Token num) {
+    ASTNode *node =
+            ast_node_new(NODE_NUM, num);
+    return node;
+}
+
+ASTNode *ast_node_str(Token str) {
+    ASTNode *node =
+            ast_node_new(NODE_STR, str);
+    return node;
+}
+
+ASTNode *ast_node_star(Token star) {
+    ASTNode *node =
+            ast_node_new(NODE_STAR, star);
+    return node;
+}
+
+ASTNode *ast_node_andref(Token andref) {
+    ASTNode *node =
+            ast_node_new(NODE_ANDREF, andref);
+    return node;
+}
+
+ASTNode *ast_node_expr(Token caller) {
+    ASTNode *node =
+            ast_node_new(NODE_EXPR, caller);
+    return node;
+}
+
+ASTNode *ast_node_param_list(Token tok_lparen) {
+    ASTNode *node =
+            ast_node_new(NODE_PARAM_LIST, tok_lparen);
+    return node;
+}
+
+ASTNode *ast_node_param(Token id, ASTNode *type) {
+    ASTNode *node =
+            ast_node_new(NODE_PARAM, id);
+    ast_node_add_child(node, type);
+    return node;
+}
+
+ASTNode *ast_node_compound(Token tok_lcurly) {
+    ASTNode *node =
+            ast_node_new(NODE_COMPOUND, tok_lcurly);
+    return node;
+}
+
+ASTNode *ast_node_lambda(ASTNode *param_list, ASTNode *body) {
+    ASTNode *node =
+            ast_node_new(NODE_LAMBDA, param_list->token);
+    ast_node_add_child(node, param_list);
+    ast_node_add_child(node, body);
+    return node;
+}
+
+ASTNode *ast_node_stmt_return(Token tok_return, ASTNode *expr) {
+    ASTNode *node =
+            ast_node_new(NODE_STMT_RETURN, tok_return);
+    if (expr != NULL) {
+        ast_node_add_child(node, expr);
+    }
+    return node;
+}
+
+ASTNode *ast_node_stmt_expr(ASTNode *expr) {
+    ASTNode *node =
+            ast_node_new(NODE_STMT_EXPR, expr->token);
+    ast_node_add_child(node, expr);
+    return node;
+}
+
+ASTNode *ast_node_stmt_if(
+        Token tok_if,
+        ASTNode *cond,        /* compound */
+        ASTNode *then_branch, /* compound */
+        ASTNode *else_branch /* compound */) {
+    ASTNode *node =
+            ast_node_new(NODE_STMT_IF, tok_if);
+    ast_node_add_child(node, cond);
+    ast_node_add_child(node, then_branch);
+    if (else_branch != NULL) {
+        ast_node_add_child(node, else_branch);
+    }
+    return node;
+}
+
+ASTNode *ast_node_stmt_set(Token dollar, ASTNode *id, ASTNode *expr) {
+    ASTNode *node =
+            ast_node_new(NODE_STMT_SET, dollar);
+    ast_node_add_child(node, id);
+    ast_node_add_child(node, expr);
+    return node;
+}

src/main.c

@@ -1,11 +1,25 @@
 #include "globals.h"
 #include "lex.h"
 #include "util.h"
+#include "parse.h"
 
-int main() {
+#define SCAN 0
+#define PARSE 1
+
+int main(int argc, char **argv) {
     Lexer *lexer = lexer_new();
+    
+    FILE * f;
 
-    FILE *f = fopen("test.cval", "r");
+    if (argc > 2) {
+        printf("Usage: %s [source_file]\n", argv[0]);
+        return 1;
+    }
+    else if (argc == 2) {
+        f = fopen(argv[1], "r");
+    } else {
+        f = fopen("test.cval", "r");
+    }
 
     if (f == NULL) {
         perror("Failed to open file");
@@ -14,14 +28,32 @@ int main() {
 
     lexer_set_source(lexer, f);
 
+    #if SCAN
     Token tok;
 
     do {
         tok = lexer_next_token(lexer);
         print_token(tok);
+        printf("\n");
     } while(tok.type != EOF_TOKEN && tok.type != ERROR);
+    #endif
+
+    #if PARSE
+    Parser *parser = parser_new(lexer);
+
+    ASTNode *ast_root = parser_parse_program(parser);
+
+    if (parser->flag_error) {
+        printf("Parsing failed due to errors.\n");
+    } else {
+        printf("Parsing succeeded.\n");
+        ast_node_print(ast_root, 0);
+    }
+    #endif
+
+
 
     fclose(f);
 
     return 0;
-}
+}

src/parse.c

@@ -2,6 +2,10 @@
 
 #include <stdlib.h>
 
+static void parser_next(Parser *parser);
+
+static bool parser_expect(Parser *parser, TokenType type);
+
 Parser *parser_new(Lexer *lexer) {
     Parser *parser = malloc(sizeof(Parser));
     if (parser == NULL) {
@@ -25,14 +29,218 @@ static void parser_next(Parser *parser) {
     parser->peek = lexer_next_token(parser->lexer);
 }
 
-static void parser_expect(Parser *parser, TokenType type) {
+static bool parser_expect(Parser *parser, TokenType type) {
     if (parser->current.type == type) {
         parser_next(parser);
+        return true;
     } else {
         parser->flag_error = 1;
+        return false;
     }
 }
 
 /*
     IMPL. PARSER PARSE FUNCTIONS 
-*/
+*/
+
+ASTNode *parser_parse_program(Parser *parser) {
+    ASTNode *root = ast_node_program();
+
+    while (parser->current.type == VAL) {
+        ASTNode *defn_node = parser_parse_defn(parser);
+
+        if (defn_node == NULL) {
+            ast_node_free(root);
+            parser->flag_error = 1;
+            return NULL;
+        }
+        ast_node_add_child(root, defn_node);
+    }
+
+    return root;
+}
+
+ASTNode *parser_parse_defn(Parser *parser) {
+    Token val_tok = parser->current;
+    if (!parser_expect(parser, VAL)) {// must start with VAL
+        return NULL;
+    }
+
+    ASTNode *type_node = parser_parse_type(parser);
+    if (type_node == NULL) {
+        return NULL;
+    }
+    if (parser->current.type != ID) {
+        parser->flag_error = 1;
+        return NULL;
+    }
+    Token id_tok = parser->current;
+    ASTNode *id_node = ast_node_id(id_tok);
+    parser_next(parser);
+
+    ASTNode *expr_node = NULL;
+    if (parser->current.type != SEMI) {
+        expr_node = parser_parse_expr(parser);
+        if (expr_node == NULL) {
+            parser->flag_error = 1;
+            return NULL;
+        }
+    }
+    if (!parser_expect(parser, SEMI)) {
+        return NULL;
+    }
+
+    ASTNode *defn_node = ast_node_defn(val_tok, type_node, id_node, expr_node);
+    return defn_node;
+}
+
+/*
+    PARSING TYPE
+*/
+
+ASTNode *parser_parse_type(Parser *parser) {
+    ASTNode *type_node = NULL;
+    if (parser->current.type == ID) {
+        type_node = parser_parse_type_simple(parser);
+    } else if (parser->current.type == LBRACK) {
+        type_node = parser_parse_type_complex(parser);
+    } else {
+        parser->flag_error = 1;
+        return NULL;
+    }
+    return type_node;
+}
+
+ASTNode *parser_parse_type_simple(Parser *parser) {
+    Token token = parser->current;
+    if (!parser_expect(parser, ID)) {
+        return NULL;
+    }
+    ASTNode *type_node = ast_node_type_simple(token);
+    while (parser->current.type == STAR) {
+        Token star_tok = parser->current;
+        parser_next(parser);
+
+        ASTNode *type_star = ast_node_type_star(star_tok);
+        ast_node_add_child(type_node, type_star);
+    }
+
+    return type_node;
+}
+
+ASTNode *parser_parse_type_complex(Parser *parser) {
+    Token tok = parser->current;
+    parser_next(parser);
+
+    ASTNode *types[256];
+    ASTNode *type_ret = NULL;
+    size_t cnt = 0;
+
+    ASTNode *ret = NULL;
+
+    while (parser->current.type != RBRACK && parser->current.type != ARROW) {
+        ASTNode *type = parser_parse_type(parser);
+        if (type == NULL) {
+            return NULL;
+        }
+
+        types[cnt++] = type;
+    }
+
+    if (parser->current.type == ARROW) {
+
+        parser_next(parser);
+
+        if (parser->current.type != RBRACK) {
+            type_ret = parser_parse_type(parser);
+            if (type_ret == NULL) return NULL;
+        } else {
+            type_ret = ast_node_type_void(parser);
+        }
+
+        ASTNode *type_param = ast_node_type_param();
+        for (size_t i = 0; i < cnt; i++) {
+            ast_node_add_child(type_param, types[i]);
+        }
+
+        ASTNode *type_out = ast_node_type_out();
+        ast_node_add_child(type_out, type_ret);
+        ret = ast_node_type_complex(tok, type_param, type_out);
+    } else if (parser->current.type == RBRACK) {
+
+        if (cnt >= 2) {
+            parser->flag_error = 1;// too many args
+            return NULL;
+        }
+        if (cnt == 1) {
+            ret = types[0];
+        } else {
+            ret = ast_node_type_void(parser);
+        }
+    } else {
+        parser->flag_error = 1;
+        return NULL;
+    }
+    parser_next(parser);
+    return ret;
+}
+
+/*
+    PARSING EXPR
+*/
+
+ASTNode *parser_parse_expr(Parser *parser) {
+    ASTNode *atoms[256];
+    size_t cnt = 1;
+    ASTNode *atom_head = parser_parse_atom(parser);
+    if (atom_head == NULL) {
+        parser->flag_error = 1;
+        return NULL;
+    }
+    atoms[0] = atom_head;
+
+    ASTNode *atom;
+    while (true) {
+        atom = parser_parse_atom(parser);
+        if (atom == NULL) {
+            break;
+        }
+        atoms[cnt] = atom;
+        cnt++;
+    }
+
+    ASTNode *node = ast_node_expr(atom_head->token);
+    for (size_t i = 0; i < cnt; i++) {
+        ast_node_add_child(node, atoms[i]);
+    }
+
+    return node;
+}
+
+ASTNode *parser_parse_atom(Parser *parser) {
+    if (parser->current.type == NUM) {
+        Token num_tok = parser->current;
+        parser_next(parser);
+        return ast_node_num(num_tok);
+    } else if (parser->current.type == ID) {
+        Token id_tok = parser->current;
+        parser_next(parser);
+        return ast_node_id(id_tok);
+    } else if (parser->current.type == STRING_LITERAL) {
+        Token str_tok = parser->current;
+        parser_next(parser);
+        return ast_node_str(str_tok);
+    } else if (parser->current.type == STAR) {
+        Token star_tok = parser->current;
+        parser_next(parser);
+        return ast_node_star(star_tok);
+    } else if (parser->current.type == ANDREF) {
+        Token andref_tok = parser->current;
+        parser_next(parser);
+        return ast_node_andref(andref_tok);
+    }
+    
+    else {
+        return NULL;
+    }
+}

src/util.c

@@ -103,5 +103,4 @@ void print_token(Token tok) {
     if (tok.type == ID || tok.type == NUM || tok.type == STRING_LITERAL) {
         printf("Data: %s", tok.data.string);
     }
-    printf("\n");
 }

test2.cval

@@ -0,0 +1,9 @@
+val [->int] main;
+
+val char* line;
+
+val size_t line get_line_ptr stdin line;
+
+val char first *line;
+
+val [int [int->size_t]->size_t] functor;