zzy/smcc
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat add func call and rewrite codes

Browse Source
This commit is contained in:
ZZY
2025-03-07 12:29:53 +08:00
parent 09299e339c
commit 95bf44eb3f
37 changed files with 3369 additions and 1063 deletions
Download Patch File Download Diff File Expand all files Collapse all files

290
ccompiler/frontend/parser/ast/ast.c
View File

@ -14,9 +14,9 @@ void init_ast_node(struct ASTNode* node) {
}
}
struct ASTNode* find_ast_node(struct ASTNode* node, enum ASTType type) {
// struct ASTNode* find_ast_node(struct ASTNode* node, ast_type_t type) {
}
// }
#include <stdio.h>
static void pnt_depth(int depth) {
@ -25,149 +25,149 @@ static void pnt_depth(int depth) {
}
}
void pnt_ast(struct ASTNode* node, int depth) {
if (!node) return;
pnt_depth(depth);
switch (node->type) {
case NT_ROOT:
for (int i = 0; i < node->root.child_size; i++) {
pnt_ast(node->root.children[i], depth);
}
return;
// void pnt_ast(struct ASTNode* node, int depth) {
// if (!node) return;
// pnt_depth(depth);
// switch (node->type) {
// case NT_ROOT:
// for (int i = 0; i < node->root.child_size; i++) {
// pnt_ast(node->root.children[i], depth);
// }
// return;
case NT_ADD : printf("+ \n"); break; // (expr) + (expr)
case NT_SUB : printf("- \n"); break; // (expr) - (expr)
case NT_MUL : printf("* \n"); break; // (expr) * (expr)
case NT_DIV : printf("/ \n"); break; // (expr) / (expr)
case NT_MOD : printf("%%\n"); break; // (expr) % (expr)
case NT_AND : printf("& \n"); break; // (expr) & (expr)
case NT_OR : printf("| \n"); break; // (expr) | (expr)
case NT_XOR : printf("^ \n"); break; // (expr) ^ (expr)
case NT_L_SH : printf("<<\n"); break; // (expr) << (expr)
case NT_R_SH : printf(">>\n"); break; // (expr) >> (expr)
case NT_EQ : printf("==\n"); break; // (expr) == (expr)
case NT_NEQ : printf("!=\n"); break; // (expr) != (expr)
case NT_LE : printf("<=\n"); break; // (expr) <= (expr)
case NT_GE : printf(">=\n"); break; // (expr) >= (expr)
case NT_LT : printf("< \n"); break; // (expr) < (expr)
case NT_GT : printf("> \n"); break; // (expr) > (expr)
case NT_AND_AND : printf("&&\n"); break; // (expr) && (expr)
case NT_OR_OR : printf("||\n"); break; // (expr) || (expr)
case NT_NOT : printf("! \n"); break; // ! (expr)
case NT_BIT_NOT : printf("~ \n"); break; // ~ (expr)
case NT_COMMA : printf(", \n"); break; // expr, expr 逗号运算符
case NT_ASSIGN : printf("= \n"); break; // (expr) = (expr)
// case NT_COND : // (expr) ? (expr) : (expr)
// case NT_ADD : printf("+ \n"); break; // (expr) + (expr)
// case NT_SUB : printf("- \n"); break; // (expr) - (expr)
// case NT_MUL : printf("* \n"); break; // (expr) * (expr)
// case NT_DIV : printf("/ \n"); break; // (expr) / (expr)
// case NT_MOD : printf("%%\n"); break; // (expr) % (expr)
// case NT_AND : printf("& \n"); break; // (expr) & (expr)
// case NT_OR : printf("| \n"); break; // (expr) | (expr)
// case NT_XOR : printf("^ \n"); break; // (expr) ^ (expr)
// case NT_L_SH : printf("<<\n"); break; // (expr) << (expr)
// case NT_R_SH : printf(">>\n"); break; // (expr) >> (expr)
// case NT_EQ : printf("==\n"); break; // (expr) == (expr)
// case NT_NEQ : printf("!=\n"); break; // (expr) != (expr)
// case NT_LE : printf("<=\n"); break; // (expr) <= (expr)
// case NT_GE : printf(">=\n"); break; // (expr) >= (expr)
// case NT_LT : printf("< \n"); break; // (expr) < (expr)
// case NT_GT : printf("> \n"); break; // (expr) > (expr)
// case NT_AND_AND : printf("&&\n"); break; // (expr) && (expr)
// case NT_OR_OR : printf("||\n"); break; // (expr) || (expr)
// case NT_NOT : printf("! \n"); break; // ! (expr)
// case NT_BIT_NOT : printf("~ \n"); break; // ~ (expr)
// case NT_COMMA : printf(", \n"); break; // expr, expr 逗号运算符
// case NT_ASSIGN : printf("= \n"); break; // (expr) = (expr)
// // case NT_COND : // (expr) ? (expr) : (expr)
case NT_STMT_EMPTY : // ;
printf(";\n");
break;
case NT_STMT_IF : // if (cond) { ... } [else {...}]
printf("if");
pnt_ast(node->if_stmt.cond, depth+1);
pnt_ast(node->if_stmt.if_stmt, depth+1);
if (node->if_stmt.else_stmt) {
pnt_depth(depth);
printf("else");
pnt_ast(node->if_stmt.else_stmt, depth+1);
}
break;
case NT_STMT_WHILE : // while (cond) { ... }
printf("while\n");
pnt_ast(node->while_stmt.cond, depth+1);
pnt_ast(node->while_stmt.body, depth+1);
break;
case NT_STMT_DOWHILE : // do {...} while (cond)
printf("do-while\n");
pnt_ast(node->do_while_stmt.body, depth+1);
pnt_ast(node->do_while_stmt.cond, depth+1);
break;
case NT_STMT_FOR : // for (init; cond; iter) {...}
printf("for\n");
if (node->for_stmt.init)
pnt_ast(node->for_stmt.init, depth+1);
if (node->for_stmt.cond)
pnt_ast(node->for_stmt.cond, depth+1);
if (node->for_stmt.iter)
pnt_ast(node->for_stmt.iter, depth+1);
pnt_ast(node->for_stmt.body, depth+1);
break;
case NT_STMT_SWITCH : // switch (expr) { case ... }
case NT_STMT_BREAK : // break;
case NT_STMT_CONTINUE : // continue;
case NT_STMT_GOTO : // goto label;
case NT_STMT_CASE : // case const_expr:
case NT_STMT_DEFAULT : // default:
case NT_STMT_LABEL : // label:
break;
case NT_STMT_BLOCK : // { ... }
printf("{\n");
for (int i = 0; i < node->block.child_size; i++) {
pnt_ast(node->block.children[i], depth+1);
}
pnt_depth(depth);
printf("}\n");
break;
case NT_STMT_RETURN : // return expr;
printf("return");
if (node->return_stmt.expr_stmt) {
printf(" ");
pnt_ast(node->return_stmt.expr_stmt, depth+1);
} else {
printf("\n");
}
break;
case NT_STMT_EXPR : // expr;
printf("stmt\n");
pnt_ast(node->expr_stmt.expr_stmt, depth);
pnt_depth(depth);
printf(";\n");
break;
case NT_DECL_VAR : // type name; or type name = expr;
printf("decl_val\n");
break;
case NT_DECL_FUNC: // type func_name(param_list);
printf("decl func %s\n", node->func.name->syms.tok.constant.str);
break;
case NT_FUNC : // type func_name(param_list) {...}
printf("def func %s\n", node->func.name->syms.tok.constant.str);
// pnt_ast(node->child.func.params, depth);
pnt_ast(node->func.body, depth);
// pnt_ast(node->child.func.ret, depth);
break;
case NT_PARAM : // 函数形参
printf("param\n");
case NT_ARG_LIST : // 实参列表（需要与NT_CALL配合）
printf("arg_list\n");
case NT_TERM_CALL : // func (expr)
printf("call\n");
break;
case NT_TERM_IDENT:
printf("%s\n", node->syms.tok.constant.str);
break;
case NT_TERM_VAL : // Terminal Symbols like constant, identifier, keyword
struct Token * tok = &node->syms.tok;
switch (tok->type) {
case TOKEN_CHAR_LITERAL:
printf("%c\n", tok->constant.ch);
break;
case TOKEN_INT_LITERAL:
printf("%d\n", tok->constant.i);
break;
case TOKEN_STRING_LITERAL:
printf("%s\n", tok->constant.str);
break;
default:
printf("unknown term val\n");
break;
}
default:
break;
}
// case NT_STMT_EMPTY : // ;
// printf(";\n");
// break;
// case NT_STMT_IF : // if (cond) { ... } [else {...}]
// printf("if");
// pnt_ast(node->if_stmt.cond, depth+1);
// pnt_ast(node->if_stmt.if_stmt, depth+1);
// if (node->if_stmt.else_stmt) {
// pnt_depth(depth);
// printf("else");
// pnt_ast(node->if_stmt.else_stmt, depth+1);
// }
// break;
// case NT_STMT_WHILE : // while (cond) { ... }
// printf("while\n");
// pnt_ast(node->while_stmt.cond, depth+1);
// pnt_ast(node->while_stmt.body, depth+1);
// break;
// case NT_STMT_DOWHILE : // do {...} while (cond)
// printf("do-while\n");
// pnt_ast(node->do_while_stmt.body, depth+1);
// pnt_ast(node->do_while_stmt.cond, depth+1);
// break;
// case NT_STMT_FOR : // for (init; cond; iter) {...}
// printf("for\n");
// if (node->for_stmt.init)
// pnt_ast(node->for_stmt.init, depth+1);
// if (node->for_stmt.cond)
// pnt_ast(node->for_stmt.cond, depth+1);
// if (node->for_stmt.iter)
// pnt_ast(node->for_stmt.iter, depth+1);
// pnt_ast(node->for_stmt.body, depth+1);
// break;
// case NT_STMT_SWITCH : // switch (expr) { case ... }
// case NT_STMT_BREAK : // break;
// case NT_STMT_CONTINUE : // continue;
// case NT_STMT_GOTO : // goto label;
// case NT_STMT_CASE : // case const_expr:
// case NT_STMT_DEFAULT : // default:
// case NT_STMT_LABEL : // label:
// break;
// case NT_STMT_BLOCK : // { ... }
// printf("{\n");
// for (int i = 0; i < node->block.child_size; i++) {
// pnt_ast(node->block.children[i], depth+1);
// }
// pnt_depth(depth);
// printf("}\n");
// break;
// case NT_STMT_RETURN : // return expr;
// printf("return");
// if (node->return_stmt.expr_stmt) {
// printf(" ");
// pnt_ast(node->return_stmt.expr_stmt, depth+1);
// } else {
// printf("\n");
// }
// break;
// case NT_STMT_EXPR : // expr;
// printf("stmt\n");
// pnt_ast(node->expr_stmt.expr_stmt, depth);
// pnt_depth(depth);
// printf(";\n");
// break;
// case NT_DECL_VAR : // type name; or type name = expr;
// printf("decl_val\n");
// break;
// case NT_DECL_FUNC: // type func_name(param_list);
// printf("decl func %s\n", node->func.name->syms.tok.val.str);
// break;
// case NT_FUNC : // type func_name(param_list) {...}
// printf("def func %s\n", node->func.name->syms.tok.val.str);
// // pnt_ast(node->child.func.params, depth);
// pnt_ast(node->func.body, depth);
// // pnt_ast(node->child.func.ret, depth);
// break;
// case NT_PARAM : // 函数形参
// printf("param\n");
// case NT_ARG_LIST : // 实参列表（需要与NT_CALL配合）
// printf("arg_list\n");
// case NT_TERM_CALL : // func (expr)
// printf("call\n");
// break;
// case NT_TERM_IDENT:
// printf("%s\n", node->syms.tok.val.str);
// break;
// case NT_TERM_VAL : // Terminal Symbols like constant, identifier, keyword
// tok_t * tok = &node->syms.tok;
// switch (tok->type) {
// case TOKEN_CHAR_LITERAL:
// printf("%c\n", tok->val.ch);
// break;
// case TOKEN_INT_LITERAL:
// printf("%d\n", tok->val.i);
// break;
// case TOKEN_STRING_LITERAL:
// printf("%s\n", tok->val.str);
// break;
// default:
// printf("unknown term val\n");
// break;
// }
// default:
// break;
// }
// 通用子节点递归处理
if (node->type <= NT_ASSIGN) { // 表达式类统一处理子节点
if (node->expr.left) pnt_ast(node->expr.left, depth+1);
if (node->expr.right) pnt_ast(node->expr.right, depth + 1);
}
}
// // 通用子节点递归处理
// if (node->type <= NT_ASSIGN) { // 表达式类统一处理子节点
// if (node->expr.left) pnt_ast(node->expr.left, depth+1);
// if (node->expr.right) pnt_ast(node->expr.right, depth + 1);
// }
// }

56
ccompiler/frontend/parser/ast/ast.h
View File

@ -3,9 +3,10 @@
#include "../../frontend.h"
#include "../../lexer/lexer.h"
#include "../../../../libcore/vector.h"
#include "../type.h"
enum ASTType {
typedef enum {
NT_INIT,
NT_ROOT, // global scope in root node
NT_ADD, // (expr) + (expr)
@ -75,31 +76,28 @@ enum ASTType {
NT_TERM_VAL,
NT_TERM_IDENT,
NT_TERM_TYPE,
};
} ast_type_t;
struct ASTNode {
enum ASTType type;
typedef struct ASTNode {
ast_type_t type;
union {
void *children[6];
struct {
struct ASTNode** children;
int child_size;
vector_header(children, struct ASTNode*);
} root;
struct {
struct ASTNode** children; // array of children
int child_size;
vector_header(children, struct ASTNode*);
} block;
struct {
struct ASTNode* decl_node;
struct Token tok;
tok_t tok;
} syms;
struct {
struct ASTNode *arr;
int size;
vector_header(params, struct ASTNode*);
} params;
struct {
const char* name;
struct ASTNode* name;
struct ASTNode* params;
struct ASTNode* func_decl;
} call;
@ -113,13 +111,12 @@ struct ASTNode {
struct ASTNode *ret;
struct ASTNode *name;
struct ASTNode *params; // array of params
void* data;
} func_decl;
struct ASTNode *def;
} decl_func;
struct {
struct ASTNode *ret;
struct ASTNode *name;
struct ASTNode *params; // array of params
struct ASTNode *decl;
struct ASTNode *body; // optional
void* data;
} func;
struct {
struct ASTNode *left;
@ -165,27 +162,26 @@ struct ASTNode {
struct ASTNode *expr_stmt;
} expr_stmt;
};
};
} ast_node_t;
struct ASTNode* new_ast_node(void);
void init_ast_node(struct ASTNode* node);
void pnt_ast(struct ASTNode* node, int depth);
struct Parser;
typedef struct ASTNode* (*parse_func_t) (struct Parser*);
typedef struct parser parser_t;
typedef struct ASTNode* (*parse_func_t) (parser_t*);
void parse_prog(struct Parser* parser);
struct ASTNode* parse_block(struct Parser* parser);
struct ASTNode* parse_stmt(struct Parser* parser);
struct ASTNode* parse_expr(struct Parser* parser);
struct ASTNode* parse_func(struct Parser* parser);
struct ASTNode* parse_decl(struct Parser* parser);
void parse_prog(parser_t* parser);
ast_node_t* parse_decl(parser_t* parser);
ast_node_t* parse_block(parser_t* parser);
ast_node_t* parse_stmt(parser_t* parser);
ast_node_t* parse_expr(parser_t* parser);
struct ASTNode* parse_ident(struct Parser* parser);
struct ASTNode* parse_type(struct Parser* parser);
ast_node_t* parse_type(parser_t* parser);
int peek_decl(struct Parser* parser);
ast_node_t* new_ast_ident_node(tok_t* tok);
ast_node_t* expect_pop_ident(tok_buf_t* tokbuf);
struct ASTNode* parser_ident_without_pop(struct Parser* parser);
int peek_decl(tok_buf_t* tokbuf);
#endif

59
ccompiler/frontend/parser/ast/block.c
View File

@ -1,48 +1,49 @@
#include "../parser.h"
#include "ast.h"
#include "../parser.h"
#include "../symtab/symtab.h"
#ifndef BLOCK_MAX_NODE
#define BLOCK_MAX_NODE (1024)
#endif
struct ASTNode* parse_block(struct Parser* parser) {
symtab_enter_scope(parser->symtab);
// parse_decl(parser); // decl_var
enum TokenType ttype;
struct ASTNode* node = new_ast_node();
ast_node_t* new_ast_node_block() {
ast_node_t* node = new_ast_node();
node->type = NT_BLOCK;
flushpeektok(parser);
ttype = peektoktype(parser);
if (ttype != TOKEN_L_BRACE) {
error("block need '{' start");
}
poptok(parser);
vector_init(node->block.children);
return node;
}
node->block.children = malloc(sizeof(struct ASTNode*) * BLOCK_MAX_NODE);
struct ASTNode* child = NULL;
ast_node_t* parse_block(parser_t* parser) {
symtab_enter_scope(parser->symtab);
tok_buf_t *tokbuf = &parser->tokbuf;
flush_peek_tok(tokbuf);
tok_type_t ttype;
ast_node_t* node = new_ast_node_block();
expect_pop_tok(tokbuf, TOKEN_L_BRACE);
ast_node_t* child = NULL;
while (1) {
if (peek_decl(parser) == 1) {
if (peek_decl(tokbuf)) {
child = parse_decl(parser);
goto ADD_CHILD;
vector_push(node->block.children, child);
continue;
}
flushpeektok(parser);
ttype = peektoktype(parser);
flush_peek_tok(tokbuf);
ttype = peek_tok_type(tokbuf);
switch (ttype) {
case TOKEN_R_BRACE:
poptok(parser);
goto END;
default:
child = parse_stmt(parser);
goto ADD_CHILD;
break;
case TOKEN_R_BRACE: {
pop_tok(tokbuf);
goto END;
}
default: {
child = parse_stmt(parser);
vector_push(node->block.children, child);
break;
}
}
continue;
ADD_CHILD:
node->block.children[node->block.child_size++] = child;
}
END:
symtab_leave_scope(parser->symtab);

56
ccompiler/frontend/parser/ast/decl.c
View File

@ -6,9 +6,9 @@
* 0 false
* 1 true
*/
int peek_decl(struct Parser* parser) {
flushpeektok(parser);
switch (peektoktype(parser)) {
int peek_decl(tok_buf_t* tokbuf) {
flush_peek_tok(tokbuf);
switch (peek_tok_type(tokbuf)) {
case TOKEN_STATIC:
case TOKEN_EXTERN:
case TOKEN_REGISTER:
@ -16,10 +16,10 @@ int peek_decl(struct Parser* parser) {
error("not impliment");
break;
default:
flushpeektok(parser);
flush_peek_tok(tokbuf);
}
switch (peektoktype(parser)) {
switch (peek_tok_type(tokbuf)) {
case TOKEN_VOID:
case TOKEN_CHAR:
case TOKEN_SHORT:
@ -27,60 +27,62 @@ int peek_decl(struct Parser* parser) {
case TOKEN_LONG:
case TOKEN_FLOAT:
case TOKEN_DOUBLE:
// FIXME Ptr
return 1;
default:
flushpeektok(parser);
flush_peek_tok(tokbuf);
}
return 0;
}
struct ASTNode* parse_decl_val(struct Parser* parser) {
flushpeektok(parser);
// parse_type
enum TokenType ttype;
struct ASTNode* node;
ast_node_t* parse_decl_val(parser_t* parser) {
tok_buf_t* tokbuf = &parser->tokbuf;
tok_type_t ttype;
flush_peek_tok(tokbuf);
struct ASTNode* type_node = parse_type(parser);
struct ASTNode* name_node = parser_ident_without_pop(parser);
ast_node_t* node;
ast_node_t* type_node = parse_type(parser);
flush_peek_tok(tokbuf);
ast_node_t* name_node = new_ast_ident_node(peek_tok(tokbuf));
node = new_ast_node();
node->decl_val.type = type_node;
node->decl_val.name = name_node;
node->type = NT_DECL_VAR;
symtab_add_symbol(parser->symtab, name_node->syms.tok.constant.str, node);
symtab_add_symbol(parser->symtab, name_node->syms.tok.val.str, node, 0);
ttype = peektoktype(parser);
ttype = peek_tok_type(tokbuf);
if (ttype == TOKEN_ASSIGN) {
node->decl_val.expr_stmt = parse_stmt(parser);
if (node->decl_val.expr_stmt->type != NT_STMT_EXPR) {
error("parser_decl_val want stmt_expr");
}
} else if (ttype == TOKEN_SEMICOLON) {
poptok(parser);
expecttok(parser, TOKEN_SEMICOLON);
pop_tok(tokbuf);
expect_pop_tok(tokbuf, TOKEN_SEMICOLON);
} else {
error("parser_decl_val syntax error");
}
return node;
}
// 类型解析入口改进
struct ASTNode* parse_decl(struct Parser* parser) {
flushpeektok(parser);
int idx;
enum TokenType ttype;
struct ASTNode* node;
ast_node_t* parse_decl(parser_t* parser) {
tok_buf_t* tokbuf = &parser->tokbuf;
flush_peek_tok(tokbuf);
tok_type_t ttype;
ast_node_t* node;
if (peek_decl(parser) == 0) {
if (peek_decl(tokbuf) == 0) {
error("syntax error expect decl_val TYPE");
}
if (peektoktype(parser) != TOKEN_IDENT) {
if (peek_tok_type(tokbuf) != TOKEN_IDENT) {
error("syntax error expect decl_val IDENT");
}
ttype = peektoktype(parser);
ttype = peek_tok_type(tokbuf);
switch (ttype) {
case TOKEN_L_PAREN: // (
node = parse_func(parser);
return NULL;
break;
case TOKEN_ASSIGN:
case TOKEN_SEMICOLON:

218
ccompiler/frontend/parser/ast/expr.c
View File

@ -33,14 +33,18 @@ enum ParseType {
PREFIX_PARSER,
};
static struct ASTNode *parse_subexpression(struct Parser* parser, enum Precedence prec);
static ast_node_t *parse_subexpression(tok_buf_t* tokbuf, symtab_t *symtab, enum Precedence prec);
#define NEXT(prec) parse_subexpression(tokbuf, symtab, prec)
static struct ASTNode* gen_node2(struct ASTNode* left, struct ASTNode* right,
enum ASTType type) {
struct ASTNode* node = new_ast_node();
static ast_node_t* gen_node2(ast_node_t* left, ast_node_t* right,
ast_type_t type) {
ast_node_t* node = new_ast_node();
node->type = type;
node->expr.left = left;
node->expr.right = right;
return node;
// FIXME
// switch (type) {
// case NT_ADD : printf("+ \n"); break; // (expr) + (expr)
// case NT_SUB : printf("- \n"); break; // (expr) - (expr)
@ -68,154 +72,157 @@ static struct ASTNode* gen_node2(struct ASTNode* left, struct ASTNode* right,
// }
}
static struct ASTNode* parse_comma(struct Parser* parser, struct ASTNode* left) {
struct ASTNode* node = new_ast_node();
static ast_node_t* parse_comma(tok_buf_t* tokbuf, symtab_t *symtab, ast_node_t* left) {
ast_node_t* node = new_ast_node();
node->type = NT_COMMA;
node->expr.left = left;
node->expr.right = parse_subexpression(parser, PREC_EXPRESSION);
node->expr.right = NEXT(PREC_EXPRESSION);
return node;
}
static struct ASTNode* parse_assign(struct Parser* parser, struct ASTNode* left) {
flushpeektok(parser);
enum TokenType ttype = peektoktype(parser);
poptok(parser);
struct ASTNode* node = new_ast_node();
static ast_node_t* parse_assign(tok_buf_t* tokbuf, symtab_t *symtab, ast_node_t* left) {
flush_peek_tok(tokbuf);
tok_type_t ttype = peek_tok_type(tokbuf);
pop_tok(tokbuf);
ast_node_t* node = new_ast_node();
node->type = NT_ASSIGN;
// saved left
node->expr.left = left;
enum Precedence next = PREC_ASSIGNMENT + 1;
switch (ttype) {
case TOKEN_ASSIGN :
left = parse_subexpression(parser, next);
left = NEXT(next);
break;
case TOKEN_ASSIGN_ADD :
left = gen_node2(left, parse_subexpression(parser, next), NT_ADD);
left = gen_node2(left, NEXT(next), NT_ADD);
break;
case TOKEN_ASSIGN_SUB :
left = gen_node2(left, parse_subexpression(parser, next), NT_SUB);
left = gen_node2(left, NEXT(next), NT_SUB);
break;
case TOKEN_ASSIGN_MUL :
left = gen_node2(left, parse_subexpression(parser, next), NT_MUL);
left = gen_node2(left, NEXT(next), NT_MUL);
break;
case TOKEN_ASSIGN_DIV :
left = gen_node2(left, parse_subexpression(parser, next), NT_DIV);
left = gen_node2(left, NEXT(next), NT_DIV);
break;
case TOKEN_ASSIGN_MOD :
left = gen_node2(left, parse_subexpression(parser, next), NT_MOD);
left = gen_node2(left, NEXT(next), NT_MOD);
break;
case TOKEN_ASSIGN_L_SH :
left = gen_node2(left, parse_subexpression(parser, next), NT_L_SH);
left = gen_node2(left, NEXT(next), NT_L_SH);
break;
case TOKEN_ASSIGN_R_SH :
left = gen_node2(left, parse_subexpression(parser, next), NT_R_SH);
left = gen_node2(left, NEXT(next), NT_R_SH);
break;
case TOKEN_ASSIGN_AND :
left = gen_node2(left, parse_subexpression(parser, next), NT_AND);
left = gen_node2(left, NEXT(next), NT_AND);
break;
case TOKEN_ASSIGN_OR :
left = gen_node2(left, parse_subexpression(parser, next), NT_OR);
left = gen_node2(left, NEXT(next), NT_OR);
break;
case TOKEN_ASSIGN_XOR :
left = gen_node2(left, parse_subexpression(parser, next), NT_XOR);
left = gen_node2(left, NEXT(next), NT_XOR);
break;
default:
error("unsupported operator");
break;
}
node->expr.right = left;
return node;
}
static struct ASTNode* parse_cmp(struct Parser* parser, struct ASTNode* left) {
flushpeektok(parser);
enum TokenType ttype = peektoktype(parser);
poptok(parser);
struct ASTNode* node = new_ast_node();
static ast_node_t* parse_cmp(tok_buf_t* tokbuf, symtab_t *symtab, ast_node_t* left) {
flush_peek_tok(tokbuf);
tok_type_t ttype = peek_tok_type(tokbuf);
pop_tok(tokbuf);
ast_node_t* node = new_ast_node();
// saved left
node->expr.left = left;
switch (ttype) {
case TOKEN_EQ:
node->type = NT_EQ;
node->expr.right = parse_subexpression(parser, PREC_EQUALITY);
node->expr.right = NEXT(PREC_EQUALITY);
break;
case TOKEN_NEQ:
node->type = NT_NEQ;
node->expr.right = parse_subexpression(parser, PREC_EQUALITY);
node->expr.right = NEXT(PREC_EQUALITY);
break;
case TOKEN_LT:
node->type = NT_LT;
node->expr.right = parse_subexpression(parser, PREC_RELATIONAL);
node->expr.right = NEXT(PREC_RELATIONAL);
break;
case TOKEN_GT:
node->type = NT_GT;
node->expr.right = parse_subexpression(parser, PREC_RELATIONAL);
node->expr.right = NEXT(PREC_RELATIONAL);
break;
case TOKEN_LE:
node->type = NT_LE;
node->expr.right = parse_subexpression(parser, PREC_RELATIONAL);
node->expr.right = NEXT(PREC_RELATIONAL);
break;
case TOKEN_GE:
node->type = NT_GE;
node->expr.right = parse_subexpression(parser, PREC_RELATIONAL);
node->expr.right = NEXT(PREC_RELATIONAL);
break;
default:
error("invalid operator");
}
return node;
}
static struct ASTNode* parse_cal(struct Parser* parser, struct ASTNode* left) {
flushpeektok(parser);
enum TokenType ttype = peektoktype(parser);
poptok(parser);
struct ASTNode* node = new_ast_node();
static ast_node_t* parse_cal(tok_buf_t* tokbuf, symtab_t *symtab, ast_node_t* left) {
flush_peek_tok(tokbuf);
tok_type_t ttype = peek_tok_type(tokbuf);
pop_tok(tokbuf);
ast_node_t* node = new_ast_node();
node->expr.left = left;
switch (ttype) {
case TOKEN_OR_OR:
node->type = NT_OR_OR;
node->expr.right = parse_subexpression(parser, PREC_LOGICAL_OR);
node->expr.right = NEXT(PREC_LOGICAL_OR);
break;
case TOKEN_AND_AND:
node->type = NT_AND_AND;
node->expr.right = parse_subexpression(parser, PREC_LOGICAL_AND);
node->expr.right = NEXT(PREC_LOGICAL_AND);
break;
case TOKEN_OR:
node->type = NT_OR;
node->expr.right = parse_subexpression(parser, PREC_OR);
node->expr.right = NEXT(PREC_OR);
break;
case TOKEN_XOR:
node->type = NT_XOR;
node->expr.right = parse_subexpression(parser, PREC_XOR);
node->expr.right = NEXT(PREC_XOR);
break;
case TOKEN_AND:
node->type = NT_AND;
node->expr.right = parse_subexpression(parser, PREC_AND);
node->expr.right = NEXT(PREC_AND);
break;
case TOKEN_L_SH:
node->type = NT_L_SH;
node->expr.right = parse_subexpression(parser, PREC_SHIFT);
node->expr.right = NEXT(PREC_SHIFT);
break;
case TOKEN_R_SH:
node->type = NT_R_SH;
node->expr.right = parse_subexpression(parser, PREC_SHIFT);
node->expr.right = NEXT(PREC_SHIFT);
break;
case TOKEN_ADD:
node->type = NT_ADD;
node->expr.right = parse_subexpression(parser, PREC_ADDITIVE);
node->expr.right = NEXT(PREC_ADDITIVE);
break;
case TOKEN_SUB:
node->type = NT_SUB;
node->expr.right = parse_subexpression(parser, PREC_ADDITIVE);
node->expr.right = NEXT(PREC_ADDITIVE);
break;
case TOKEN_MUL:
node->type = NT_MUL;
node->expr.right = parse_subexpression(parser, PREC_MULTIPLICATIVE);
node->expr.right = NEXT(PREC_MULTIPLICATIVE);
break;
case TOKEN_DIV:
node->type = NT_DIV;
node->expr.right = parse_subexpression(parser, PREC_MULTIPLICATIVE);
node->expr.right = NEXT(PREC_MULTIPLICATIVE);
break;
case TOKEN_MOD:
node->type = NT_MOD;
node->expr.right = parse_subexpression(parser, PREC_MULTIPLICATIVE);
node->expr.right = NEXT(PREC_MULTIPLICATIVE);
break;
default:
break;
@ -223,44 +230,50 @@ static struct ASTNode* parse_cal(struct Parser* parser, struct ASTNode* left) {
return node;
}
// 新增函数调用解析
static struct ASTNode* parse_call(struct Parser* parser, struct ASTNode* ident) {
struct ASTNode* node = new_ast_node();
static ast_node_t* parse_call(tok_buf_t* tokbuf, symtab_t *symtab, ast_node_t* ident) {
ast_node_t* node = new_ast_node();
node->type = NT_TERM_CALL;
poptok(parser); // 跳过 '('
node->call.name = ident;
node->call.params = new_ast_node();
vector_init(node->call.params->params.params);
pop_tok(tokbuf); // 跳过 '('
enum TokenType ttype;
// 解析参数列表
while ((ttype = peektoktype(parser)) != TOKEN_R_PAREN) {
// add_arg(node, parse_expr(parser));
if (ttype == TOKEN_COMMA) poptok(parser);
else poptok(parser);
tok_type_t ttype;
while (1) {
flush_peek_tok(tokbuf);
ttype = peek_tok_type(tokbuf);
if (ttype == TOKEN_R_PAREN) {
break;
}
ast_node_t* param = NEXT(PREC_EXPRESSION);
vector_push(node->call.params->params.params, param);
flush_peek_tok(tokbuf);
ttype = peek_tok_type(tokbuf);
if (ttype == TOKEN_COMMA) pop_tok(tokbuf);
}
poptok(parser); // 跳过 ')'
pop_tok(tokbuf); // 跳过 ')'
char* name = ident->syms.tok.constant.str;
void* sym = symtab_lookup_symbol(parser->symtab, name);
if (sym == NULL) {
const char* name = ident->syms.tok.val.str;
ast_node_t* sym = symtab_lookup_symbol(symtab, name);
// TODO check func is match
if (sym == NULL || sym->type != NT_DECL_FUNC) {
error("function not decl %s", name);
}
node->call.name = name;
node->call.params = NULL;
node->call.name = ident;
node->call.func_decl = sym;
return node;
}
static struct ASTNode* parse_paren(struct Parser* parser, struct ASTNode* left) {
flushpeektok(parser);
enum TokenType ttype;
expecttok(parser, TOKEN_L_PAREN);
left = parse_subexpression(parser, PREC_EXPRESSION);
flushpeektok(parser);
expecttok(parser, TOKEN_R_PAREN);
static ast_node_t* parse_paren(tok_buf_t* tokbuf, symtab_t *symtab, ast_node_t* left) {
flush_peek_tok(tokbuf);
expect_pop_tok(tokbuf, TOKEN_L_PAREN);
left = NEXT(PREC_EXPRESSION);
flush_peek_tok(tokbuf);
expect_pop_tok(tokbuf, TOKEN_R_PAREN);
return left;
}
typedef struct ASTNode* (*parse_expr_fun_t)(struct Parser*, struct ASTNode*);
typedef ast_node_t* (*parse_expr_fun_t)(tok_buf_t*, symtab_t* , ast_node_t*);
static struct expr_prec_table_t {
parse_expr_fun_t parser;
enum Precedence prec;
@ -309,11 +322,11 @@ static struct expr_prec_table_t {
[TOKEN_L_PAREN] = {parse_paren, PREC_POSTFIX, INFIX_PARSER},
};
static struct ASTNode *parse_primary_expression(struct Parser* parser) {
flushpeektok(parser);
static ast_node_t *parse_primary_expression(tok_buf_t* tokbuf, symtab_t *symtab) {
flush_peek_tok(tokbuf);
struct Token* tok = peektok(parser);
struct ASTNode *node = new_ast_node();
tok_t* tok = peek_tok(tokbuf);
ast_node_t *node = new_ast_node();
node->type = NT_TERM_VAL;
node->syms.tok = *tok;
@ -330,34 +343,35 @@ static struct ASTNode *parse_primary_expression(struct Parser* parser) {
case TOKEN_STRING_LITERAL:
// node->data.data_type = TYPE_POINTER;
case TOKEN_IDENT:
node = parse_ident(parser);
if (peektoktype(parser) == TOKEN_L_PAREN) {
node = parse_call(parser, node);
node = expect_pop_ident(tokbuf);
tok_type_t ttype = peek_tok_type(tokbuf);
if (ttype == TOKEN_L_PAREN) {
node = parse_call(tokbuf, symtab, node);
} else {
void *sym = symtab_lookup_symbol(parser->symtab, tok->constant.str);
void *sym = symtab_lookup_symbol(symtab, tok->val.str);
if (sym == NULL) {
error("undefined symbol but use %s", tok->constant.str);
error("undefined symbol but use %s", tok->val.str);
}
node->type = NT_TERM_IDENT;
node->syms.decl_node = sym;
goto END;
}
goto END;
default:
return NULL;
}
poptok(parser);
pop_tok(tokbuf);
END:
return node;
}
static struct ASTNode *parse_subexpression(struct Parser* parser, enum Precedence prec) {
enum TokenType ttype;
struct expr_prec_table_t* work;
struct ASTNode* left;
static ast_node_t *parse_subexpression(tok_buf_t* tokbuf, symtab_t *symtab, enum Precedence prec) {
tok_type_t ttype;
struct expr_prec_table_t* work;
ast_node_t* left;
while (1) {
flushpeektok(parser);
ttype = peektoktype(parser);
flush_peek_tok(tokbuf);
ttype = peek_tok_type(tokbuf);
work = &expr_table[ttype];
// FIXME
if (ttype == TOKEN_SEMICOLON || ttype == TOKEN_R_PAREN) {
@ -365,16 +379,16 @@ static struct ASTNode *parse_subexpression(struct Parser* parser, enum Precedenc
}
if (work == NULL || work->parser == NULL || work->ptype == PREFIX_PARSER) {
if (work->parser != NULL) {
left = work->parser(parser, NULL);
left = work->parser(tokbuf, symtab, NULL);
} else {
left = parse_primary_expression(parser);
left = parse_primary_expression(tokbuf, symtab);
}
} else if (work->ptype == INFIX_PARSER) {
if (work->parser == NULL)
break;
if (work->prec <= prec)
break;
left = work->parser(parser, left);
left = work->parser(tokbuf, symtab, left);
}
// assert(left != NULL);
}
@ -382,9 +396,11 @@ static struct ASTNode *parse_subexpression(struct Parser* parser, enum Precedenc
return left;
}
struct ASTNode* parse_expr(struct Parser* parser) {
flushpeektok(parser);
enum TokenType ttype = peektoktype(parser);
ast_node_t* parse_expr(parser_t* parser) {
tok_buf_t* tokbuf = &(parser->tokbuf);
symtab_t *symtab = parser->symtab;
flush_peek_tok(tokbuf);
tok_type_t ttype = peek_tok_type(tokbuf);
switch (ttype) {
case TOKEN_NOT:
case TOKEN_AND:
@ -401,9 +417,9 @@ struct ASTNode* parse_expr(struct Parser* parser) {
case TOKEN_SUB_SUB:
case TOKEN_SIZEOF:
case TOKEN_IDENT:
return parse_subexpression(parser, PREC_EXPRESSION);
return NEXT(PREC_EXPRESSION);
default:
error("Want expr but not got %s", get_token_name(ttype));
error("Want expr but not got %s", get_tok_name(ttype));
break;
}
}

152
ccompiler/frontend/parser/ast/func.c
View File

@ -6,34 +6,21 @@
#define FUNC_PARAM_CACHE_SIZE 32 // 合理初始值，可覆盖99%常见情况
#endif
struct FuncParamCache {
struct Token tokens[FUNC_PARAM_CACHE_SIZE];
int read_pos; // 当前读取位置
int write_pos; // 写入位置
int depth; // 当前缓存深度
};
static enum TokenType peekcachetype(struct FuncParamCache* cache) {
return cache->tokens[cache->read_pos++].type;
}
// TODO 语义分析压入符号表
static void parse_params(struct Parser* parser, struct FuncParamCache* cache, struct ASTNode* node) {
// = peekcachetype(cache);
enum TokenType ttype;
// if (ttype != TOKEN_L_PAREN) {
// error("function expected '('\n");
// }
struct ASTNode *params = new_ast_node();
node->func.params = params;
int params_size = 0;
static void parse_params(parser_t* parser, tok_buf_t* cache, ast_node_t* node) {
tok_type_t ttype;
ast_node_t *params = new_ast_node();
node->decl_func.params = params;
vector_init(params->params.params);
while ((ttype = peekcachetype(cache)) != TOKEN_R_PAREN) {
int depth = 1;
while (depth) {
ttype = peek_tok_type(cache);
switch (ttype) {
case TOKEN_COMMA:
break;
case TOKEN_ELLIPSIS:
ttype = peekcachetype(cache);
ttype = peek_tok_type(cache);
if (ttype != TOKEN_R_PAREN) {
error("... must be a last parameter list (expect ')')");
}
@ -41,9 +28,29 @@ static void parse_params(struct Parser* parser, struct FuncParamCache* cache, st
error("not implement");
break;
case TOKEN_IDENT:
params->children[params_size++] = NULL;
// TODO 静态数组
flush_peek_tok(cache);
ast_node_t* id_node = new_ast_ident_node(peek_tok(cache));
ast_node_t* node = new_ast_node();
node->type = NT_DECL_VAR;
node->decl_val.name = id_node;
// TODO typing sys
node->decl_val.type = NULL;
node->decl_val.expr_stmt = NULL;
node->decl_val.data = NULL;
vector_push(params->params.params, node);
symtab_add_symbol(parser->symtab, id_node->syms.tok.val.str, node, 0);
break;
case TOKEN_L_PAREN: {
depth++;
break;
}
case TOKEN_R_PAREN: {
depth--;
break;
}
default:
break;
// TODO 使用cache的类型解析
// parse_type(parser);
// TODO type parse
@ -51,39 +58,42 @@ static void parse_params(struct Parser* parser, struct FuncParamCache* cache, st
// ttype = peekcachetype(cache);
// if (ttype != TOKEN_IDENT) {
// node->node_type = NT_DECL_FUNC;
// flushpeektok(parser);
// flush_peek_tok(tokbuf);
// continue;
// }
// error("function expected ')' or ','\n");
}
pop_tok(cache);
}
}
enum ASTType check_is_func_decl(struct Parser* parser, struct FuncParamCache* cache) {
cache->depth = 1;
cache->read_pos = 0;
cache->write_pos = 0;
ast_type_t check_is_func_decl(tok_buf_t* tokbuf, tok_buf_t* cache) {
expect_pop_tok(tokbuf, TOKEN_L_PAREN);
int depth = 1;
while (cache->depth) {
struct Token* tok = peektok(parser);
poptok(parser);
if (cache->write_pos >= FUNC_PARAM_CACHE_SIZE - 1) {
while (depth) {
tok_t* tok = peek_tok(tokbuf);
pop_tok(tokbuf);
if (cache->size >= cache->cap - 1) {
error("function parameter list too long");
}
cache->tokens[cache->write_pos++] = *tok;
cache->buf[cache->size++] = *tok;
switch (tok->type) {
case TOKEN_L_PAREN:
cache->depth++;
depth++;
break;
case TOKEN_R_PAREN:
cache->depth--;
depth--;
break;
default:
break;
}
}
cache->end = cache->size;
switch (peektoktype(parser)) {
switch (peek_tok_type(tokbuf)) {
case TOKEN_SEMICOLON:
poptok(parser);
pop_tok(tokbuf);
return NT_DECL_FUNC;
case TOKEN_L_BRACE:
return NT_FUNC;
@ -93,28 +103,66 @@ enum ASTType check_is_func_decl(struct Parser* parser, struct FuncParamCache* ca
}
}
struct ASTNode* parse_func(struct Parser* parser) {
struct ASTNode* ret_type = parse_type(parser);
struct ASTNode* func_name = parse_ident(parser);
static ast_node_t* new_ast_node_funcdecl(ast_node_t* ret, ast_node_t* name) {
ast_node_t* node = new_ast_node();
node->type = NT_DECL_FUNC;
node->decl_func.ret = ret;
node->decl_func.name = name;
node->decl_func.def = NULL;
return node;
}
struct ASTNode* node = new_ast_node();
node->func.ret = ret_type;
node->func.name = func_name;
void parse_func(parser_t* parser) {
tok_buf_t* tokbuf = &(parser->tokbuf);
flush_peek_tok(tokbuf);
ast_node_t* ret_node = parse_type(parser);
ast_node_t* name_node = expect_pop_ident(tokbuf);
const char* func_name = name_node->syms.tok.val.str;
ast_node_t* decl = new_ast_node_funcdecl(ret_node, name_node);
flushpeektok(parser);
expecttok(parser, TOKEN_L_PAREN);
struct FuncParamCache cache;
node->type = check_is_func_decl(parser, &cache);
tok_buf_t cache;
init_tokbuf(&cache, NULL, NULL);
cache.cap = FUNC_PARAM_CACHE_SIZE;
tok_t buf[FUNC_PARAM_CACHE_SIZE];
cache.buf = buf;
ast_type_t type = check_is_func_decl(&(parser->tokbuf), &cache);
symtab_add_symbol(parser->symtab, func_name->syms.tok.constant.str, node);
if (node->type == NT_DECL_FUNC) {
return node;
ast_node_t* prev = symtab_add_symbol(parser->symtab, func_name, decl, 1);
if (prev != NULL) {
if (prev->type != NT_DECL_FUNC) {
error("the symbol duplicate old is %d, new is func", prev->type);
}
// TODO check redeclare func is match
if (type == NT_FUNC) {
// TODO Free decl;
free(decl);
decl = prev;
goto FUNC;
}
return;
}
vector_push(parser->root->root.children, decl);
if (type == NT_DECL_FUNC) {
return;
}
FUNC:
// 该data临时用于判断是否重复定义
if (decl->decl_func.def != NULL) {
error("redefinition of function %s", func_name);
}
ast_node_t* node = new_ast_node();
node->type = NT_FUNC;
node->func.decl = decl;
node->func.data = NULL;
decl->decl_func.def = node;
symtab_enter_scope(parser->symtab);
parse_params(parser, &cache, node);
parse_params(parser, &cache, decl);
node->func.body = parse_block(parser);
symtab_leave_scope(parser->symtab);
return node;
vector_push(parser->root->root.children, node);
}

23
ccompiler/frontend/parser/ast/program.c
View File

@ -5,25 +5,30 @@
#define PROG_MAX_NODE_SIZE (1024 * 4)
#endif
void parse_prog(struct Parser* parser) {
void parse_func(parser_t* parser);
void parse_prog(parser_t* parser) {
/**
* Program := (Declaration | Definition)*
* same as
* Program := Declaration* Definition*
*/
int child_size = 0;
tok_buf_t *tokbuf = &(parser->tokbuf);
parser->root = new_ast_node();
struct ASTNode* node;
parser->root->root.children = xmalloc(sizeof(struct ASTNode*) * PROG_MAX_NODE_SIZE);
ast_node_t* node;
parser->root->type = NT_ROOT;
vector_init(parser->root->root.children);
while (1) {
flushpeektok(parser);
if (peektoktype(parser) == TOKEN_EOF) {
flush_peek_tok(tokbuf);
if (peek_tok_type(tokbuf) == TOKEN_EOF) {
break;
}
node = parse_decl(parser);
parser->root->root.children[child_size++] = node;
if (node == NULL) {
parse_func(parser);
} else {
vector_push(parser->root->root.children, node);
}
}
parser->root->type = NT_ROOT;
parser->root->root.child_size = child_size;
return;
}

114
ccompiler/frontend/parser/ast/stmt.c
View File

@ -1,27 +1,28 @@
#include "../parser.h"
#include "ast.h"
struct ASTNode* parse_stmt(struct Parser* parser) {
flushpeektok(parser);
enum TokenType ttype = peektoktype(parser);
struct ASTNode* node = new_ast_node();
ast_node_t* parse_stmt(parser_t* parser) {
tok_buf_t* tokbuf = &parser->tokbuf;
flush_peek_tok(tokbuf);
tok_type_t ttype = peek_tok_type(tokbuf);
ast_node_t* node = new_ast_node();
switch (ttype) {
case TOKEN_IF: {
/**
* if (exp) stmt
* if (exp) stmt else stmt
*/
poptok(parser);
pop_tok(tokbuf);
expecttok(parser, TOKEN_L_PAREN);
expect_pop_tok(tokbuf, TOKEN_L_PAREN);
node->if_stmt.cond = parse_expr(parser);
flushpeektok(parser);
expecttok(parser, TOKEN_R_PAREN);
flush_peek_tok(tokbuf);
expect_pop_tok(tokbuf, TOKEN_R_PAREN);
node->if_stmt.if_stmt = parse_stmt(parser);
ttype = peektoktype(parser);
ttype = peek_tok_type(tokbuf);
if (ttype == TOKEN_ELSE) {
poptok(parser);
pop_tok(tokbuf);
node->if_stmt.else_stmt = parse_stmt(parser);
} else {
node->if_stmt.else_stmt = NULL;
@ -33,11 +34,11 @@ struct ASTNode* parse_stmt(struct Parser* parser) {
/**
* switch (exp) stmt
*/
poptok(parser);
pop_tok(tokbuf);
expecttok(parser, TOKEN_L_PAREN);
expect_pop_tok(tokbuf, TOKEN_L_PAREN);
node->switch_stmt.cond = parse_expr(parser);
expecttok(parser, TOKEN_R_PAREN);
expect_pop_tok(tokbuf, TOKEN_R_PAREN);
node->switch_stmt.body = parse_stmt(parser);
node->type = NT_STMT_SWITCH;
@ -47,11 +48,11 @@ struct ASTNode* parse_stmt(struct Parser* parser) {
/**
* while (exp) stmt
*/
poptok(parser);
pop_tok(tokbuf);
expecttok(parser, TOKEN_L_PAREN);
expect_pop_tok(tokbuf, TOKEN_L_PAREN);
node->while_stmt.cond = parse_expr(parser);
expecttok(parser, TOKEN_R_PAREN);
expect_pop_tok(tokbuf, TOKEN_R_PAREN);
node->while_stmt.body = parse_stmt(parser);
node->type = NT_STMT_WHILE;
@ -61,16 +62,16 @@ struct ASTNode* parse_stmt(struct Parser* parser) {
/**
* do stmt while (exp)
*/
poptok(parser);
pop_tok(tokbuf);
node->do_while_stmt.body = parse_stmt(parser);
ttype = peektoktype(parser);
ttype = peek_tok_type(tokbuf);
if (ttype != TOKEN_WHILE) {
error("expected while after do");
}
poptok(parser);
expecttok(parser, TOKEN_L_PAREN);
pop_tok(tokbuf);
expect_pop_tok(tokbuf, TOKEN_L_PAREN);
node->do_while_stmt.cond = parse_expr(parser);
expecttok(parser, TOKEN_R_PAREN);
expect_pop_tok(tokbuf, TOKEN_R_PAREN);
node->type = NT_STMT_DOWHILE;
break;
}
@ -79,36 +80,36 @@ struct ASTNode* parse_stmt(struct Parser* parser) {
* for (init; [cond]; [iter]) stmt
*/
// node->children.stmt.for_stmt.init
poptok(parser);
ttype = peektoktype(parser);
pop_tok(tokbuf);
ttype = peek_tok_type(tokbuf);
if (ttype != TOKEN_L_PAREN) {
error("expected ( after for");
}
poptok(parser);
pop_tok(tokbuf);
// init expr or init decl_var
// TODO need add this feature
node->for_stmt.init = parse_expr(parser);
expecttok(parser, TOKEN_SEMICOLON);
expect_pop_tok(tokbuf, TOKEN_SEMICOLON);
// cond expr or null
ttype = peektoktype(parser);
ttype = peek_tok_type(tokbuf);
if (ttype != TOKEN_SEMICOLON) {
node->for_stmt.cond = parse_expr(parser);
expecttok(parser, TOKEN_SEMICOLON);
expect_pop_tok(tokbuf, TOKEN_SEMICOLON);
} else {
node->for_stmt.cond = NULL;
poptok(parser);
pop_tok(tokbuf);
}
// iter expr or null
ttype = peektoktype(parser);
ttype = peek_tok_type(tokbuf);
if (ttype != TOKEN_R_PAREN) {
node->for_stmt.iter = parse_expr(parser);
expecttok(parser, TOKEN_R_PAREN);
expect_pop_tok(tokbuf, TOKEN_R_PAREN);
} else {
node->for_stmt.iter = NULL;
poptok(parser);
pop_tok(tokbuf);
}
node->for_stmt.body = parse_stmt(parser);
@ -120,8 +121,8 @@ struct ASTNode* parse_stmt(struct Parser* parser) {
* break ;
*/
// TODO check 导致外围 for、while 或 do-while 循环或 switch 语句终止。
poptok(parser);
expecttok(parser, TOKEN_SEMICOLON);
pop_tok(tokbuf);
expect_pop_tok(tokbuf, TOKEN_SEMICOLON);
node->type = NT_STMT_BREAK;
break;
@ -131,8 +132,8 @@ struct ASTNode* parse_stmt(struct Parser* parser) {
* continue ;
*/
// TODO check 导致跳过整个 for、 while 或 do-while 循环体的剩余部分。
poptok(parser);
expecttok(parser, TOKEN_SEMICOLON);
pop_tok(tokbuf);
expect_pop_tok(tokbuf, TOKEN_SEMICOLON);
node->type = NT_STMT_CONTINUE;
break;
@ -142,16 +143,16 @@ struct ASTNode* parse_stmt(struct Parser* parser) {
* return [exp] ;
*/
// TODO 终止当前函数并返回指定值给调用方函数。
poptok(parser);
ttype = peektoktype(parser);
pop_tok(tokbuf);
ttype = peek_tok_type(tokbuf);
if (ttype != TOKEN_SEMICOLON) {
node->return_stmt.expr_stmt = parse_expr(parser);
flushpeektok(parser);
expecttok(parser, TOKEN_SEMICOLON);
flush_peek_tok(tokbuf);
expect_pop_tok(tokbuf, TOKEN_SEMICOLON);
} else {
node->return_stmt.expr_stmt = NULL;
pop_tok(tokbuf);
}
poptok(parser);
node->type = NT_STMT_RETURN;
break;
}
@ -161,15 +162,15 @@ struct ASTNode* parse_stmt(struct Parser* parser) {
*/
// TODO check label 将控制无条件转移到所欲位置。
//在无法用约定的构造将控制转移到所欲位置时使用。
poptok(parser);
pop_tok(tokbuf);
// find symbol table
ttype = peektoktype(parser);
ttype = peek_tok_type(tokbuf);
if (ttype != TOKEN_IDENT) {
error("expect identifier after goto");
}
expecttok(parser, TOKEN_SEMICOLON);
expect_pop_tok(tokbuf, TOKEN_SEMICOLON);
// TODO filling label
node->goto_stmt.label = parse_ident(parser);
node->goto_stmt.label = expect_pop_ident(tokbuf);
node->type = NT_STMT_GOTO;
break;
}
@ -181,7 +182,7 @@ struct ASTNode* parse_stmt(struct Parser* parser) {
* if () ;
* for () ;
*/
poptok(parser);
pop_tok(tokbuf);
node->type = NT_STMT_EMPTY;
break;
}
@ -193,30 +194,30 @@ struct ASTNode* parse_stmt(struct Parser* parser) {
node->type = NT_STMT_BLOCK;
break;
}
case TOKEN_IDENT: {
case TOKEN_IDENT: {
// TODO label goto
if (peektoktype(parser) != TOKEN_COLON) {
if (peek_tok_type(tokbuf) != TOKEN_COLON) {
goto EXP;
}
node->label_stmt.label = parse_ident(parser);
expecttok(parser, TOKEN_COLON);
node->label_stmt.label = expect_pop_ident(tokbuf);
expect_pop_tok(tokbuf, TOKEN_COLON);
node->type = NT_STMT_LABEL;
break;
}
case TOKEN_CASE: {
// TODO label switch
poptok(parser);
pop_tok(tokbuf);
error("unimplemented switch label");
node->label_stmt.label = parse_expr(parser);
// TODO 该表达式为const int
expecttok(parser, TOKEN_COLON);
expect_pop_tok(tokbuf, TOKEN_COLON);
node->type = NT_STMT_CASE;
break;
}
case TOKEN_DEFAULT: {
// TODO label switch default
poptok(parser);
expecttok(parser, TOKEN_COLON);
pop_tok(tokbuf);
expect_pop_tok(tokbuf, TOKEN_COLON);
node->type = NT_STMT_DEFAULT;
break;
}
@ -226,15 +227,16 @@ struct ASTNode* parse_stmt(struct Parser* parser) {
*/
EXP:
node->expr_stmt.expr_stmt = parse_expr(parser);
flushpeektok(parser);
ttype = peektoktype(parser);
flush_peek_tok(tokbuf);
ttype = peek_tok_type(tokbuf);
if (ttype != TOKEN_SEMICOLON) {
error("exp must end with \";\"");
}
poptok(parser);
pop_tok(tokbuf);
node->type = NT_STMT_EXPR;
break;
}
}
return node;
}

167
ccompiler/frontend/parser/ast/term.c
View File

@ -2,162 +2,30 @@
#include "../type.h"
#include "ast.h"
// /* 状态跳转表定义 */
// typedef void (*StateHandler)(struct Parser*, struct ASTNode**);
// enum TypeParseState {
// TPS_BASE_TYPE, // 解析基础类型 (int/char等)
// TPS_QUALIFIER, // 解析限定符 (const/volatile)
// TPS_POINTER, // 解析指针 (*)
// TPS_ARRAY, // 解析数组维度 ([n])
// TPS_FUNC_PARAMS, // 解析函数参数列表
// TPS_END,
// };
// ;
// /* 状态处理函数前置声明 */
// static void handle_base_type(struct Parser*, struct ASTNode**);
// static void handle_qualifier(struct Parser*, struct ASTNode**);
// static void handle_pointer(struct Parser*, struct ASTNode**);
// static void handle_array(struct Parser*, struct ASTNode**);
// static void handle_func_params(struct Parser*, struct ASTNode**);
// static void handle_error(struct Parser*, struct ASTNode**);
// /* 状态跳转表（核心优化部分） */
// static const struct StateTransition {
// enum TokenType tok; // 触发token
// StateHandler handler; // 处理函数
// enum TypeParseState next_state; // 下一个状态
// } state_table[][8] = {
// [TPS_QUALIFIER] = {
// {TOKEN_CONST, handle_qualifier, TPS_QUALIFIER},
// {TOKEN_VOLATILE, handle_qualifier, TPS_QUALIFIER},
// {TOKEN_VOID, handle_base_type, TPS_POINTER},
// {TOKEN_CHAR, handle_base_type, TPS_POINTER},
// {TOKEN_INT, handle_base_type, TPS_POINTER},
// {TOKEN_EOF, handle_error, TPS_QUALIFIER},
// /* 其他token默认处理 */
// {0, NULL, TPS_BASE_TYPE}
// },
// [TPS_BASE_TYPE] = {
// {TOKEN_MUL, handle_pointer, TPS_POINTER},
// {TOKEN_L_BRACKET, handle_array, TPS_ARRAY},
// {TOKEN_L_PAREN, handle_func_params,TPS_FUNC_PARAMS},
// {TOKEN_EOF, NULL, TPS_END},
// {0, NULL, TPS_POINTER}
// },
// [TPS_POINTER] = {
// {TOKEN_MUL, handle_pointer, TPS_POINTER},
// {TOKEN_L_BRACKET, handle_array, TPS_ARRAY},
// {TOKEN_L_PAREN, handle_func_params,TPS_FUNC_PARAMS},
// {0, NULL, TPS_END}
// },
// [TPS_ARRAY] = {
// {TOKEN_L_BRACKET, handle_array, TPS_ARRAY},
// {TOKEN_L_PAREN, handle_func_params,TPS_FUNC_PARAMS},
// {0, NULL, TPS_END}
// },
// [TPS_FUNC_PARAMS] = {
// {0, NULL, TPS_END}
// }
// };
// /* 新的类型解析函数 */
// struct ASTNode* parse_type(struct Parser* p) {
// struct ASTNode* type_root = NULL;
// struct ASTNode** current = &type_root;
// enum TypeParseState state = TPS_QUALIFIER;
// while (state != TPS_END) {
// enum TokenType t = peektoktype(p);
// const struct StateTransition* trans = state_table[state];
// // 查找匹配的转换规则
// while (trans->tok != 0 && trans->tok != t) {
// trans++;
// }
// if (trans->handler) {
// trans->handler(p, current);
// } else if (trans->tok == 0) { // 默认规则
// state = trans->next_state;
// continue;
// } else {
// error("syntax error type parse error");
// }
// state = trans->next_state;
// }
// return type_root;
// }
// /* 具体状态处理函数实现 */
// static void handle_qualifier(struct Parser* p, struct ASTNode** current) {
// struct ASTNode* node = new_ast_node();
// node->node_type = NT_TYPE_QUAL;
// node->data.data_type = poptok(p).type;
// if (*current) {
// (*current)->child.decl.type = node;
// } else {
// *current = node;
// }
// }
// static void handle_base_type(struct Parser* p, struct ASTNode** current) {
// struct ASTNode* node = new_ast_node();
// node->node_type = NT_TYPE_BASE;
// node->data.data_type = poptok(p).type;
// // 链接到当前节点链的末端
// while (*current && (*current)->child.decl.type) {
// current = &(*current)->child.decl.type;
// }
// if (*current) {
// (*current)->child.decl.type = node;
// } else {
// *current = node;
// }
// }
// static void handle_pointer(struct Parser* p, struct ASTNode** current) {
// poptok(p); // 吃掉*
// struct ASTNode* node = new_ast_node();
// node->node_type = NT_TYPE_PTR;
// // 插入到当前节点之前
// node->child.decl.type = *current;
// *current = node;
// }
// /* 其他处理函数类似实现... */
struct ASTNode* parser_ident_without_pop(struct Parser* parser) {
flushpeektok(parser);
struct Token* tok = peektok(parser);
ast_node_t* new_ast_ident_node(tok_t* tok) {
if (tok->type != TOKEN_IDENT) {
error("syntax error: want identifier but got %d", tok->type);
}
struct ASTNode* node = new_ast_node();
ast_node_t* node = new_ast_node();
node->type = NT_TERM_IDENT;
node->syms.tok = *tok;
node->syms.decl_node = NULL;
return node;
}
struct ASTNode* parse_ident(struct Parser* parser) {
struct ASTNode* node = parser_ident_without_pop(parser);
poptok(parser);
ast_node_t* expect_pop_ident(tok_buf_t* tokbuf) {
flush_peek_tok(tokbuf);
tok_t* tok = peek_tok(tokbuf);
ast_node_t* node = new_ast_ident_node(tok);
pop_tok(tokbuf);
return node;
}
struct ASTNode* parse_type(struct Parser* parser) {
flushpeektok(parser);
enum TokenType ttype = peektoktype(parser);
enum DataType dtype;
ast_node_t* parse_type(parser_t* parser) {
tok_buf_t* tokbuf = &parser->tokbuf;
flush_peek_tok(tokbuf);
tok_type_t ttype = peek_tok_type(tokbuf);
data_type_t dtype;
switch(ttype) {
case TOKEN_VOID: dtype = TYPE_VOID; break;
case TOKEN_CHAR: dtype = TYPE_CHAR; break;
@ -170,13 +38,14 @@ struct ASTNode* parse_type(struct Parser* parser) {
error("无效的类型说明符");
}
struct ASTNode* node = new_ast_node();
ast_node_t* node = new_ast_node();
node->type = NT_TERM_TYPE;
// node->data.data_type = dtype;
poptok(parser);
// TODO added by disable warning, will add typing system
dtype += 1;
pop_tok(tokbuf);
if (peektoktype(parser) == TOKEN_MUL) {
poptok(parser);
if (peek_tok_type(tokbuf) == TOKEN_MUL) {
pop_tok(tokbuf);
}
return node;
}

242
ccompiler/frontend/parser/ast/type.c
View File

@ -1,136 +1,136 @@
#include "../parser.h"
#include "../type.h"
// #include "../parser.h"
// #include "../type.h"
enum TypeParseState {
TPS_BASE_TYPE, // 解析基础类型 (int/char等)
TPS_QUALIFIER, // 解析限定符 (const/volatile)
TPS_POINTER, // 解析指针 (*)
TPS_ARRAY, // 解析数组维度 ([n])
TPS_FUNC_PARAMS // 解析函数参数列表
};
// enum TypeParseState {
// TPS_BASE_TYPE, // 解析基础类型 (int/char等)
// TPS_QUALIFIER, // 解析限定符 (const/volatile)
// TPS_POINTER, // 解析指针 (*)
// TPS_ARRAY, // 解析数组维度 ([n])
// TPS_FUNC_PARAMS // 解析函数参数列表
// };
struct ASTNode* parse_type(struct Parser* p) {
struct ASTNode* type_root = new_ast_node();
struct ASTNode* current = type_root;
current->type = NT_TYPE_BASE;
// ast_node_t* parse_type(parser_t* p) {
// ast_node_t* type_root = new_ast_node();
// ast_node_t* current = type_root;
// current->type = NT_TYPE_BASE;
enum TypeParseState state = TPS_QUALIFIER;
int pointer_level = 0;
// enum TypeParseState state = TPS_QUALIFIER;
// int pointer_level = 0;
while (1) {
enum TokenType t = peektoktype(p);
// while (1) {
// tok_type_t t = peektoktype(p);
switch (state) {
// 基础类型解析 (int, char等)
case TPS_BASE_TYPE:
if (is_base_type(t)) {
// current->data.data_type = token_to_datatype(t);
poptok(p);
state = TPS_POINTER;
} else {
error("Expected type specifier");
}
break;
// switch (state) {
// // 基础类型解析 (int, char等)
// case TPS_BASE_TYPE:
// if (is_base_type(t)) {
// // current->data.data_type = token_to_datatype(t);
// pop_tok(p);
// state = TPS_POINTER;
// } else {
// error("Expected type specifier");
// }
// break;
// 类型限定符 (const/volatile)
case TPS_QUALIFIER:
if (t == TOKEN_CONST || t == TOKEN_VOLATILE) {
struct ASTNode* qual_node = new_ast_node();
qual_node->type = NT_TYPE_QUAL;
qual_node->data.data_type = t; // 复用data_type字段存储限定符
current->child.decl.type = qual_node;
current = qual_node;
poptok(p);
} else {
state = TPS_BASE_TYPE;
}
break;
// // 类型限定符 (const/volatile)
// case TPS_QUALIFIER:
// if (t == TOKEN_CONST || t == TOKEN_VOLATILE) {
// ast_node_t* qual_node = new_ast_node();
// qual_node->type = NT_TYPE_QUAL;
// qual_node->data.data_type = t; // 复用data_type字段存储限定符
// current->child.decl.type = qual_node;
// current = qual_node;
// pop_tok(p);
// } else {
// state = TPS_BASE_TYPE;
// }
// break;
// 指针解析 (*)
case TPS_POINTER:
if (t == TOKEN_MUL) {
struct ASTNode* ptr_node = new_ast_node();
ptr_node->type = NT_TYPE_PTR;
current->child.decl.type = ptr_node;
current = ptr_node;
pointer_level++;
poptok(p);
} else {
state = TPS_ARRAY;
}
break;
// // 指针解析 (*)
// case TPS_POINTER:
// if (t == TOKEN_MUL) {
// ast_node_t* ptr_node = new_ast_node();
// ptr_node->type = NT_TYPE_PTR;
// current->child.decl.type = ptr_node;
// current = ptr_node;
// pointer_level++;
// pop_tok(p);
// } else {
// state = TPS_ARRAY;
// }
// break;
// 数组维度 ([n])
case TPS_ARRAY:
if (t == TOKEN_L_BRACKET) {
poptok(p); // 吃掉[
struct ASTNode* arr_node = new_ast_node();
arr_node->type = NT_TYPE_ARRAY;
// // 数组维度 ([n])
// case TPS_ARRAY:
// if (t == TOKEN_L_BRACKET) {
// pop_tok(p); // 吃掉[
// ast_node_t* arr_node = new_ast_node();
// arr_node->type = NT_TYPE_ARRAY;
// 解析数组大小（仅语法检查）
if (peektoktype(p) != TOKEN_R_BRACKET) {
parse_expr(p); // 不计算实际值
}
// // 解析数组大小（仅语法检查）
// if (peektoktype(p) != TOKEN_R_BRACKET) {
// parse_expr(p); // 不计算实际值
// }
expecttok(p, TOKEN_R_BRACKET);
current->child.decl.type = arr_node;
current = arr_node;
} else {
state = TPS_FUNC_PARAMS;
}
break;
// expecttok(p, TOKEN_R_BRACKET);
// current->child.decl.type = arr_node;
// current = arr_node;
// } else {
// state = TPS_FUNC_PARAMS;
// }
// break;
// 函数参数列表
case TPS_FUNC_PARAMS:
if (t == TOKEN_L_PAREN) {
struct ASTNode* func_node = new_ast_node();
func_node->type = NT_TYPE_FUNC;
current->child.decl.type = func_node;
// // 函数参数列表
// case TPS_FUNC_PARAMS:
// if (t == TOKEN_L_PAREN) {
// ast_node_t* func_node = new_ast_node();
// func_node->type = NT_TYPE_FUNC;
// current->child.decl.type = func_node;
// 解析参数列表（仅结构，不验证类型）
parse_param_list(p, func_node);
current = func_node;
} else {
return type_root; // 类型解析结束
}
break;
}
}
}
// 判断是否是基础类型
static int is_base_type(enum TokenType t) {
return t >= TOKEN_VOID && t <= TOKEN_DOUBLE;
}
// // 转换token到数据类型（简化版）
// static enum DataType token_to_datatype(enum TokenType t) {
// static enum DataType map[] = {
// [TOKEN_VOID] = DT_VOID,
// [TOKEN_CHAR] = DT_CHAR,
// [TOKEN_INT] = DT_INT,
// // ...其他类型映射
// };
// return map[t];
// // 解析参数列表（仅结构，不验证类型）
// parse_param_list(p, func_node);
// current = func_node;
// } else {
// return type_root; // 类型解析结束
// }
// break;
// }
// }
// }
// 解析参数列表（轻量级）
static void parse_param_list(struct Parser* p, struct ASTNode* func) {
expecttok(p, TOKEN_L_PAREN);
while (peektoktype(p) != TOKEN_R_PAREN) {
struct ASTNode* param = parse_type(p); // 递归解析类型
// 允许可选参数名（仅语法检查）
if (peektoktype(p) == TOKEN_IDENT) {
poptok(p); // 吃掉参数名
}
if (peektoktype(p) == TOKEN_COMMA) {
poptok(p);
}
}
expecttok(p, TOKEN_R_PAREN);
}
// // 判断是否是基础类型
// static int is_base_type(tok_type_t t) {
// return t >= TOKEN_VOID && t <= TOKEN_DOUBLE;
// }
// // // 转换token到数据类型（简化版）
// // static enum DataType token_to_datatype(tok_type_t t) {
// // static enum DataType map[] = {
// // [TOKEN_VOID] = DT_VOID,
// // [TOKEN_CHAR] = DT_CHAR,
// // [TOKEN_INT] = DT_INT,
// // // ...其他类型映射
// // };
// // return map[t];
// // }
// // 解析参数列表（轻量级）
// static void parse_param_list(parser_t* p, ast_node_t* func) {
// expecttok(p, TOKEN_L_PAREN);
// while (peektoktype(p) != TOKEN_R_PAREN) {
// ast_node_t* param = parse_type(p); // 递归解析类型
// // 允许可选参数名（仅语法检查）
// if (peektoktype(p) == TOKEN_IDENT) {
// pop_tok(p); // 吃掉参数名
// }
// if (peektoktype(p) == TOKEN_COMMA) {
// pop_tok(p);
// }
// }
// expecttok(p, TOKEN_R_PAREN);
// }