#include "spl_sema.h" void spl_scope_init(spl_scope_t *scope) { vec_init(scope->scopes); vec_push(scope->scopes, (spl_scope_node_t){0}); scope->current_scope = 0; scope->root_scope = 0; } void spl_scope_drop(spl_scope_t *scope) { vec_free(scope->scopes); scope->current_scope = 0; scope->root_scope = 0; } spl_scope_id_t spl_scope_alloc(spl_scope_t *scope) { spl_scope_node_t n; n.parent = scope->current_scope; map_init(n.symbols, MAP_HASH_STR, MAP_CMP_STR); vec_push(scope->scopes, n); return vec_size(scope->scopes) - 1; } bool spl_scope_insert(spl_scope_t *scope, spl_scope_id_t id, spl_symbol_t symbol) { if (!id || id >= vec_size(scope->scopes)) { LOG_FATAL("sema scope insert error"); return false; } spl_symbol_t old = {0}; spl_scope_node_t *node = &vec_at(scope->scopes, id); if (map_get(node->symbols, symbol.name, &old)) { // spl_type_def(sema->type, old); TODO debug node SPL_ERROR(&(spl_dbg_node_t){0}, "find same symbol %s", symbol.name); return false; } map_put(node->symbols, symbol.name, symbol); return true; } bool spl_scope_find(spl_scope_t *scope, const char *name, spl_symbol_t *out) { if (scope == NULL || name == NULL || out == NULL) { return false; } for (spl_scope_id_t id = scope->current_scope; id != 0; id = vec_at(scope->scopes, id).parent) { spl_scope_node_t *s = &vec_at(scope->scopes, id); if (map_get(s->symbols, name, out)) { Assert(out->kind != SPL_SYMBOL_KIND_ERROR); return true; } } return false; } // Parse once static void sema_collect(spl_sema_t *sema, spl_ast_node_ref_t ref) { Assert(sema != NULL && ref != 0); spl_ast_node_t *n = spl_ast_node(sema->ast, ref); spl_symbol_t symbol = {0}; Assert(n != NULL); switch (n->kind) { case SPL_AST_CONTAINER_MEMBERS: { symbol.name = "$file"; symbol.kind = SPL_SYMBOL_KIND_TYPE; symbol.node = spl_type_def_alloc(sema->type); n->resolved_def_id = symbol.node; spl_scope_id_t scope_id = spl_scope_alloc(sema->scope); sema->scope->current_scope = scope_id; sema->scope->root_scope = scope_id; spl_scope_insert(sema->scope, scope_id, symbol); vec_for(n->container_members, i) { sema_collect(sema, vec_at(n->container_members, i)); } spl_def_node_t *def = spl_type_def(sema->type, symbol.node); def->kind = SPL_DEF_AGG; break; } case SPL_AST_TYPE_DECL: { symbol.name = n->type_decl.name; symbol.kind = SPL_SYMBOL_KIND_TYPE; symbol.node = spl_type_def_alloc(sema->type); n->resolved_def_id = symbol.node; n = spl_ast_node(sema->ast, n->type_decl.type_expr); if (!n) { SPL_ERROR(&n->dbg, "type_decl must have type_expr"); return; } spl_def_node_t *def = spl_type_def(sema->type, symbol.node); switch (n->kind) { case SPL_AST_TYPE_STRUCT: case SPL_AST_TYPE_UNION: case SPL_AST_TYPE_ENUM: case SPL_AST_TYPE_SHAPE: /* TODO for shape */ def->kind = SPL_DEF_AGG; spl_scope_id_t scope_id = spl_scope_alloc(sema->scope); spl_scope_id_t old_id = sema->scope->current_scope; spl_scope_insert(sema->scope, old_id, symbol); sema->scope->current_scope = scope_id; vec_for(n->type_expr.aggregate_list, i) { sema_collect(sema, vec_at(n->type_expr.aggregate_list, i)); } sema->scope->current_scope = old_id; break; default: def->kind = SPL_DEF_SCALAR; break; } spl_scope_insert(sema->scope, sema->scope->current_scope, symbol); } break; case SPL_AST_FN_DECL: case SPL_AST_FN_DEFINE: { symbol.name = n->fn_decl.name; symbol.kind = SPL_SYMBOL_KIND_FN; symbol.node = spl_type_def_alloc(sema->type); n->resolved_def_id = symbol.node; spl_def_node_t *def = spl_type_def(sema->type, symbol.node); def->kind = SPL_DEF_FN_PARAMS; spl_scope_insert(sema->scope, sema->scope->current_scope, symbol); } break; case SPL_AST_VAR_DECL: case SPL_AST_CONST_DECL: { symbol.name = n->var_const_decl.name; symbol.kind = SPL_SYMBOL_KIND_VAR; symbol.node = spl_type_def_alloc(sema->type); n->resolved_def_id = symbol.node; spl_def_node_t *def = spl_type_def(sema->type, symbol.node); def->kind = SPL_DEF_VAR; spl_scope_insert(sema->scope, sema->scope->current_scope, symbol); } break; case SPL_AST_MEMBER_DECL: { symbol.name = n->member_decl.name; symbol.kind = SPL_SYMBOL_KIND_MEMBER; symbol.node = spl_type_def_alloc(sema->type); n->resolved_def_id = symbol.node; spl_def_node_t *def = spl_type_def(sema->type, symbol.node); def->kind = SPL_DEF_MEMBER; spl_scope_insert(sema->scope, sema->scope->current_scope, symbol); } break; default: break; } } // Parse twice static void sema_parse(spl_sema_t *sema, spl_ast_node_ref_t ref) { if (ref == 0) return; Assert(sema != NULL); spl_ast_node_t *n = spl_ast_node(sema->ast, ref); spl_symbol_t symbol = {0}; Assert(n != NULL); switch (n->kind) { case SPL_AST_NONE: UNREACHABLE(); break; case SPL_AST_CONTAINER_MEMBERS: { vec_for(n->container_members, i) { sema_parse(sema, vec_at(n->container_members, i)); } } break; case SPL_AST_FN_DECL: case SPL_AST_FN_DEFINE: { if (n->resolved_def_id == 0) { Panic("fn `%s` don't collect", n->fn_decl.name); } spl_def_node_t *def = spl_type_def(sema->type, n->resolved_def_id); def->kind = SPL_DEF_FN_PARAMS; vec_init(def->fn_params_def); spl_scope_id_t scope_id = spl_scope_alloc(sema->scope); spl_scope_id_t old_id = sema->scope->current_scope; sema->scope->current_scope = scope_id; vec_for(n->fn_decl.param_list, i) { spl_ast_node_ref_t ref = vec_at(n->fn_decl.param_list, i); sema_parse(sema, ref); def = spl_type_def(sema->type, n->resolved_def_id); spl_ast_node_t *node = spl_ast_node(sema->ast, ref); Assert(node->kind == SPL_AST_PARAM_DECL); spl_var_def_t var_def = {0}; var_def.name = node->param_decl.name; var_def.scope_id = sema->scope->current_scope; vec_push(def->fn_params_def, var_def); } vec_for(n->fn_decl.block, i) { sema_parse(sema, vec_at(n->fn_decl.block, i)); } sema->scope->current_scope = old_id; // spl_type_def_fn_build(sema->type, n->resolved_def_id, ) } break; case SPL_AST_TYPE_DECL: { if (n->type_decl.type_expr) sema_parse(sema, n->type_decl.type_expr); } break; case SPL_AST_VAR_DECL: case SPL_AST_CONST_DECL: { if (n->resolved_def_id == 0) { symbol.name = n->var_const_decl.name; symbol.kind = SPL_SYMBOL_KIND_VAR; symbol.node = spl_type_def_alloc(sema->type); n->resolved_def_id = symbol.node; spl_def_node_t *def = spl_type_def(sema->type, symbol.node); def->kind = SPL_DEF_VAR; spl_scope_insert(sema->scope, sema->scope->current_scope, symbol); } if (n->var_const_decl.type_expr) sema_parse(sema, n->var_const_decl.type_expr); if (n->var_const_decl.expr) sema_parse(sema, n->var_const_decl.expr); } break; case SPL_AST_MEMBER_DECL: { if (n->member_decl.type_expr) sema_parse(sema, n->member_decl.type_expr); } break; case SPL_AST__COMPTIME_STMT: case SPL_AST__DIRECTIVE_BLOCK: break; case SPL_AST_PARAM_DECL: { if (n->resolved_def_id == 0) { symbol.name = n->param_decl.name; symbol.kind = SPL_SYMBOL_KIND_VAR; symbol.node = spl_type_def_alloc(sema->type); n->resolved_def_id = symbol.node; spl_def_node_t *def = spl_type_def(sema->type, symbol.node); def->kind = SPL_DEF_VAR; spl_scope_insert(sema->scope, sema->scope->current_scope, symbol); } if (n->param_decl.type_expr) sema_parse(sema, n->param_decl.type_expr); } break; case SPL_AST_ATTR_ITEM: vec_for(n->attr_item.expr_list, i) { sema_parse(sema, vec_at(n->attr_item.expr_list, i)); } break; case SPL_AST_ARGG_INIT_ITEM: { if (n->aggregate_init_item.expr) sema_parse(sema, n->aggregate_init_item.expr); } break; case SPL_AST_IF_STATEMENT: { if (n->if_statement.expr) sema_parse(sema, n->if_statement.expr); vec_for(n->if_statement.if_block, i) { sema_parse(sema, vec_at(n->if_statement.if_block, i)); } vec_for(n->if_statement.else_block, i) { sema_parse(sema, vec_at(n->if_statement.else_block, i)); } } break; case SPL_AST_IFVAR_STATEMENT: { if (n->ifvar_statement.packed_expr) sema_parse(sema, n->ifvar_statement.packed_expr); vec_for(n->ifvar_statement.if_block, i) { sema_parse(sema, vec_at(n->ifvar_statement.if_block, i)); } vec_for(n->ifvar_statement.else_block, i) { sema_parse(sema, vec_at(n->ifvar_statement.else_block, i)); } } break; case SPL_AST_WHILE_STATEMENT: { if (n->while_statement.expr) sema_parse(sema, n->while_statement.expr); vec_for(n->while_statement.while_block, i) { sema_parse(sema, vec_at(n->while_statement.while_block, i)); } } break; case SPL_AST_LOOP_STATEMENT: { vec_for(n->loop_statement.loop_block, i) { sema_parse(sema, vec_at(n->loop_statement.loop_block, i)); } } break; case SPL_AST_FOR_STATEMENT: { vec_for(n->for_statement.expr_vec, i) { sema_parse(sema, vec_at(n->for_statement.expr_vec, i)); } vec_for(n->for_statement.block, i) { sema_parse(sema, vec_at(n->for_statement.block, i)); } } break; case SPL_AST_MATCH_STATEMENT: { if (n->match_statement.expr) sema_parse(sema, n->match_statement.expr); vec_for(n->match_statement.paced_exprs, i) { sema_parse(sema, vec_at(n->match_statement.paced_exprs, i)); } vec_for(n->match_statement.match_block, i) { sema_parse(sema, vec_at(n->match_statement.match_block, i)); } } break; case SPL_AST_RET_STATEMENT: { sema_parse(sema, n->ret_statement.expr); } break; case SPL_AST_BREAK_STATEMENT: case SPL_AST_CONTINUE_STATEMENT: case SPL_AST_DEFER_STATEMENT: { vec_for(n->defer_statement.block_or_statement, i) { sema_parse(sema, vec_at(n->defer_statement.block_or_statement, i)); } } break; case SPL_AST_TRY_STATEMENT: case SPL_AST_CATCH_STATEMENT: case SPL_AST_ERRDEFER_STATEMEMT: case SPL_AST_EXPR_STATEMENT: break; case SPL_AST_PACKED_EXPR: { if (n->packed_expr.expr) sema_parse(sema, n->packed_expr.expr); } break; case SPL_AST_ASSIGN_EXPR: case SPL_AST_ASSIGN_ADD_EXPR: case SPL_AST_ASSIGN_SUB_EXPR: case SPL_AST_ASSIGN_MUL_EXPR: case SPL_AST_ASSIGN_DIV_EXPR: case SPL_AST_ASSIGN_MOD_EXPR: case SPL_AST_ASSIGN_AND_EXPR: case SPL_AST_ASSIGN_OR_EXPR: case SPL_AST_ASSIGN_XOR_EXPR: case SPL_AST_ASSIGN_LSHIFT_EXPR: case SPL_AST_ASSIGN_USHIFT_EXPR: case SPL_AST_BOOL_OR_EXPR: case SPL_AST_BOOL_AND_EXPR: case SPL_AST_BIT_OR_EXPR: case SPL_AST_BIT_XOR_EXPR: case SPL_AST_BIT_AND_EXPR: case SPL_AST_CMP_EQ_EXPR: case SPL_AST_CMP_NE_EXPR: case SPL_AST_CMP_LE_EXPR: case SPL_AST_CMP_GE_EXPR: case SPL_AST_CMP_LT_EXPR: case SPL_AST_CMP_GT_EXPR: case SPL_AST_RANGE_EXPR: case SPL_AST_LSHIFT_EXPR: case SPL_AST_RSHIFT_EXPR: case SPL_AST_ADD_EXPR: case SPL_AST_SUB_EXPR: case SPL_AST_MUL_EXPR: case SPL_AST_DIV_EXPR: case SPL_AST_MOD_EXPR: if (n->op_expr.left) sema_parse(sema, n->op_expr.left); if (n->op_expr.right) sema_parse(sema, n->op_expr.right); break; case SPL_AST_MINUS_EXPR: case SPL_AST_NOT_EXPR: case SPL_AST_BIT_NOT_EXPR: case SPL_AST_ADDRESS_EXPR: if (n->prefix_expr.postfix_expr) sema_parse(sema, n->prefix_expr.postfix_expr); break; case SPL_AST_CALL_EXPR: if (n->postfix_expr.primary_expr) sema_parse(sema, n->postfix_expr.primary_expr); vec_for(n->postfix_expr.call_expr, i) { sema_parse(sema, vec_at(n->postfix_expr.call_expr, i)); } break; case SPL_AST_FIELD_EXPR: case SPL_AST_DEREF_EXPR: if (n->postfix_expr.primary_expr) sema_parse(sema, n->postfix_expr.primary_expr); break; case SPL_AST_INDEX_EXPR: if (n->postfix_expr.primary_expr) sema_parse(sema, n->postfix_expr.primary_expr); if (n->postfix_expr.index_expr) sema_parse(sema, n->postfix_expr.index_expr); break; case SPL_AST_SLICE_EXPR: if (n->postfix_expr.primary_expr) sema_parse(sema, n->postfix_expr.primary_expr); if (n->postfix_expr.slice_expr.begin) sema_parse(sema, n->postfix_expr.slice_expr.begin); if (n->postfix_expr.slice_expr.end) sema_parse(sema, n->postfix_expr.slice_expr.end); break; case SPL_AST_AS_EXPR: if (n->postfix_expr.primary_expr) sema_parse(sema, n->postfix_expr.primary_expr); if (n->postfix_expr.type_expr) sema_parse(sema, n->postfix_expr.type_expr); break; case SPL_AST_EXPR_INTEGER_LIT: case SPL_AST_EXPR_FLOAT_LIT: case SPL_AST_EXPR_CHAR_LIT: case SPL_AST_EXPR_STRING_LIT: case SPL_AST_EXPR_TRUE: case SPL_AST_EXPR_FALSE: case SPL_AST_EXPR_NULL: case SPL_AST_EXPR_UNDEFINED: break; case SPL_AST_EXPR_IDENT: { if (!spl_scope_find(sema->scope, n->primary_expr.ident, &symbol)) { SPL_ERROR(&n->dbg, "can't find ident `%s`", n->primary_expr.ident); return; } Assert(symbol.node != 0); n->resolved_def_id = symbol.node; } break; case SPL_AST_ARGGREGATE_INIT: vec_for(n->primary_expr.aggregate_init.expr, i) { sema_parse(sema, vec_at(n->primary_expr.aggregate_init.expr, i)); } break; case SPL_AST_EXPR_EXPR: if (n->primary_expr.expr) sema_parse(sema, n->primary_expr.expr); break; case SPL_AST_ARRAY_LIT: if (n->primary_expr.array_lit_expr.type_expr) sema_parse(sema, n->primary_expr.array_lit_expr.type_expr); vec_for(n->primary_expr.array_lit_expr.expr_list, i) { sema_parse(sema, vec_at(n->primary_expr.array_lit_expr.expr_list, i)); } break; case SPL_AST_BUILTIN_EXPR: vec_for(n->primary_expr.builtin_expr.expr_list, i) { sema_parse(sema, vec_at(n->primary_expr.builtin_expr.expr_list, i)); } break; case SPL_AST_BLOCK_EXPR: vec_for(n->primary_expr.block_expr, i) { sema_parse(sema, vec_at(n->primary_expr.block_expr, i)); } break; case SPL_AST_BASE_TYPE_FN: vec_for(n->type_expr.attr_list, i) { sema_parse(sema, vec_at(n->type_expr.attr_list, i)); } vec_for(n->type_expr.fn_type.param_list, i) { sema_parse(sema, vec_at(n->type_expr.fn_type.param_list, i)); } if (n->type_expr.fn_type.type_expr) sema_parse(sema, n->type_expr.fn_type.type_expr); break; case SPL_AST_BASE_TYPE_PATH: vec_for(n->type_expr.attr_list, i) { sema_parse(sema, vec_at(n->type_expr.attr_list, i)); } break; case SPL_AST_TYPE_POINTER: vec_for(n->type_expr.attr_list, i) { sema_parse(sema, vec_at(n->type_expr.attr_list, i)); } if (n->type_expr.pointer_type.pointee) sema_parse(sema, n->type_expr.pointer_type.pointee); break; case SPL_AST_TYPE_ARRAY: vec_for(n->type_expr.attr_list, i) { sema_parse(sema, vec_at(n->type_expr.attr_list, i)); } if (n->type_expr.array_type.element) sema_parse(sema, n->type_expr.array_type.element); if (n->type_expr.array_type.size) sema_parse(sema, n->type_expr.array_type.size); break; case SPL_AST_TYPE_SLICE: vec_for(n->type_expr.attr_list, i) { sema_parse(sema, vec_at(n->type_expr.attr_list, i)); } if (n->type_expr.slice_type.element) sema_parse(sema, n->type_expr.slice_type.element); break; case SPL_AST_TYPE_STRUCT: case SPL_AST_TYPE_UNION: case SPL_AST_TYPE_SHAPE: case SPL_AST_TYPE_ENUM: vec_for(n->type_expr.attr_list, i) { sema_parse(sema, vec_at(n->type_expr.attr_list, i)); } vec_for(n->type_expr.aggregate_list, i) { sema_parse(sema, vec_at(n->type_expr.aggregate_list, i)); } break; case SPL_AST_TYPE_VOID: case SPL_AST_TYPE_BOOL: case SPL_AST_TYPE_OPAQUE: case SPL_AST_TYPE_I8: case SPL_AST_TYPE_U8: case SPL_AST_TYPE_I16: case SPL_AST_TYPE_U16: case SPL_AST_TYPE_I32: case SPL_AST_TYPE_U32: case SPL_AST_TYPE_I64: case SPL_AST_TYPE_U64: case SPL_AST_TYPE_ISIZE: case SPL_AST_TYPE_USIZE: case SPL_AST_TYPE__F32: case SPL_AST_TYPE__F64: case SPL_AST_TYPE_ANY: case SPL_AST_TYPE_IDENT: break; case SPL_AST_COUNT: UNREACHABLE(); break; } } void spl_sema_init(spl_sema_t *sema, spl_ast_t *ast, spl_type_t *type, spl_scope_t *scope) { Assert(sema && ast && type && scope); sema->ast = ast; sema->type = type; sema->scope = scope; sema->error_count = 0; sema->root = (spl_symbol_t){0}; } void spl_sema_drop(spl_sema_t *sema) { (void)sema; } void spl_sema_run(spl_sema_t *sema) { if (!sema->ast || !sema->ast->root) return; spl_ast_node_t *ast_node = spl_ast_node(sema->ast, sema->ast->root); Assert(ast_node != NULL && ast_node->kind == SPL_AST_CONTAINER_MEMBERS); // 第一趟 实现顺序无关地相互引用 sema_collect(sema, sema->ast->root); // 第二趟 正式解析 sema_parse(sema, sema->ast->root); } void spl_sema_dump(spl_sema_t *sema) { printf("Scopes:\n"); vec_for(sema->scope->scopes, i) { if (i == 0) continue; spl_scope_node_t scope = vec_at(sema->scope->scopes, i); printf("scope[%zu] parent=%zu\n", i, scope.parent); map_for(scope.symbols, j) { spl_symbol_t *symbol = &unsafe_map_at(scope.symbols, j).val; printf(" %s -> def`%s`#%zu\n", unsafe_map_at(scope.symbols, j).key, symbol->name, symbol->node); } } printf("TypeTable:\n"); vec_for(sema->type->type_table, i) { printf(" id#%zu type=", i); spl_type_pure_dump(sema->type, i); printf("\n"); } printf("DefTable:\n"); vec_for(sema->type->def_table, i) { printf(" def#%zu ", i); spl_type_def_dump(sema->type, i); printf("\n"); } }