void mp_parse_node_show(mp_parse_node_t pn, int indent) {
for (int i = 0; i < indent; i++) {
printf(" ");
}
if (MP_PARSE_NODE_IS_NULL(pn)) {
printf("NULL\n");
} else if (MP_PARSE_NODE_IS_LEAF(pn)) {
int arg = MP_PARSE_NODE_LEAF_ARG(pn);
switch (MP_PARSE_NODE_LEAF_KIND(pn)) {
case MP_PARSE_NODE_ID: printf("id(%s)\n", qstr_str(arg)); break;
case MP_PARSE_NODE_SMALL_INT: printf("int(%d)\n", arg); break;
case MP_PARSE_NODE_INTEGER: printf("int(%s)\n", qstr_str(arg)); break;
case MP_PARSE_NODE_DECIMAL: printf("dec(%s)\n", qstr_str(arg)); break;
case MP_PARSE_NODE_STRING: printf("str(%s)\n", qstr_str(arg)); break;
case MP_PARSE_NODE_BYTES: printf("bytes(%s)\n", qstr_str(arg)); break;
case MP_PARSE_NODE_TOKEN: printf("tok(%d)\n", arg); break;
default: assert(0);
}
} else {
mp_parse_node_struct_t *pns2 = (mp_parse_node_struct_t*)pn;
int n = pns2->kind_num_nodes >> 8;
#ifdef USE_RULE_NAME
printf("%s(%d) (n=%d)\n", rules[MP_PARSE_NODE_STRUCT_KIND(pns2)]->rule_name, MP_PARSE_NODE_STRUCT_KIND(pns2), n);
#else
printf("rule(%u) (n=%d)\n", (uint)MP_PARSE_NODE_STRUCT_KIND(pns2), n);
#endif
for (int i = 0; i < n; i++) {
mp_parse_node_show(pns2->nodes[i], indent + 2);
}
}
}
STATIC void push_result_rule(parser_t *parser, size_t src_line, const rule_t *rule, size_t num_args) {
// optimise away parenthesis around an expression if possible
if (rule->rule_id == RULE_atom_paren) {
// there should be just 1 arg for this rule
mp_parse_node_t pn = peek_result(parser, 0);
if (MP_PARSE_NODE_IS_NULL(pn)) {
// need to keep parenthesis for ()
} else if (MP_PARSE_NODE_IS_STRUCT_KIND(pn, RULE_testlist_comp)) {
// need to keep parenthesis for (a, b, ...)
} else {
// parenthesis around a single expression, so it's just the expression
return;
}
}
#if MICROPY_COMP_CONST_FOLDING
if (fold_logical_constants(parser, rule, &num_args)) {
// we folded this rule so return straight away
return;
}
if (fold_constants(parser, rule, num_args)) {
// we folded this rule so return straight away
return;
}
#endif
mp_parse_node_struct_t *pn = parser_alloc(parser, sizeof(mp_parse_node_struct_t) + sizeof(mp_parse_node_t) * num_args);
pn->source_line = src_line;
pn->kind_num_nodes = (rule->rule_id & 0xff) | (num_args << 8);
for (size_t i = num_args; i > 0; i--) {
pn->nodes[i - 1] = pop_result(parser);
}
push_result_node(parser, (mp_parse_node_t)pn);
}
void mp_parse_node_print(mp_parse_node_t pn, size_t indent) {
if (MP_PARSE_NODE_IS_STRUCT(pn)) {
printf("[% 4d] ", (int)((mp_parse_node_struct_t*)pn)->source_line);
} else {
printf(" ");
}
for (size_t i = 0; i < indent; i++) {
printf(" ");
}
if (MP_PARSE_NODE_IS_NULL(pn)) {
printf("NULL\n");
} else if (MP_PARSE_NODE_IS_SMALL_INT(pn)) {
mp_int_t arg = MP_PARSE_NODE_LEAF_SMALL_INT(pn);
printf("int(" INT_FMT ")\n", arg);
} else if (MP_PARSE_NODE_IS_LEAF(pn)) {
uintptr_t arg = MP_PARSE_NODE_LEAF_ARG(pn);
switch (MP_PARSE_NODE_LEAF_KIND(pn)) {
case MP_PARSE_NODE_ID: printf("id(%s)\n", qstr_str(arg)); break;
case MP_PARSE_NODE_STRING: printf("str(%s)\n", qstr_str(arg)); break;
case MP_PARSE_NODE_BYTES: printf("bytes(%s)\n", qstr_str(arg)); break;
case MP_PARSE_NODE_TOKEN: printf("tok(%u)\n", (uint)arg); break;
default: assert(0);
}
} else {
// node must be a mp_parse_node_struct_t
mp_parse_node_struct_t *pns = (mp_parse_node_struct_t*)pn;
if (MP_PARSE_NODE_STRUCT_KIND(pns) == RULE_string) {
printf("literal str(%.*s)\n", (int)pns->nodes[1], (char*)pns->nodes[0]);
} else if (MP_PARSE_NODE_STRUCT_KIND(pns) == RULE_bytes) {
printf("literal bytes(%.*s)\n", (int)pns->nodes[1], (char*)pns->nodes[0]);
} else if (MP_PARSE_NODE_STRUCT_KIND(pns) == RULE_const_object) {
#if MICROPY_OBJ_REPR == MICROPY_OBJ_REPR_D
printf("literal const(%016llx)\n", (uint64_t)pns->nodes[0] | ((uint64_t)pns->nodes[1] << 32));
#else
printf("literal const(%p)\n", (mp_obj_t)pns->nodes[0]);
#endif
} else {
size_t n = MP_PARSE_NODE_STRUCT_NUM_NODES(pns);
#ifdef USE_RULE_NAME
printf("%s(%u) (n=%u)\n", rules[MP_PARSE_NODE_STRUCT_KIND(pns)]->rule_name, (uint)MP_PARSE_NODE_STRUCT_KIND(pns), (uint)n);
#else
printf("rule(%u) (n=%u)\n", (uint)MP_PARSE_NODE_STRUCT_KIND(pns), (uint)n);
#endif
for (size_t i = 0; i < n; i++) {
mp_parse_node_print(pns->nodes[i], indent + 2);
}
}
}
}
int mp_parse_node_extract_list(mp_parse_node_t *pn, size_t pn_kind, mp_parse_node_t **nodes) {
if (MP_PARSE_NODE_IS_NULL(*pn)) {
*nodes = NULL;
return 0;
} else if (MP_PARSE_NODE_IS_LEAF(*pn)) {
*nodes = pn;
return 1;
} else {
mp_parse_node_struct_t *pns = (mp_parse_node_struct_t*)(*pn);
if (MP_PARSE_NODE_STRUCT_KIND(pns) != pn_kind) {
*nodes = pn;
return 1;
} else {
*nodes = pns->nodes;
return MP_PARSE_NODE_STRUCT_NUM_NODES(pns);
}
}
}
void mp_parse_node_print(mp_parse_node_t pn, int indent) {
if (MP_PARSE_NODE_IS_STRUCT(pn)) {
printf("[% 4d] ", (int)((mp_parse_node_struct_t*)pn)->source_line);
} else {
printf(" ");
}
for (int i = 0; i < indent; i++) {
printf(" ");
}
if (MP_PARSE_NODE_IS_NULL(pn)) {
printf("NULL\n");
} else if (MP_PARSE_NODE_IS_SMALL_INT(pn)) {
machine_int_t arg = MP_PARSE_NODE_LEAF_SMALL_INT(pn);
printf("int(" INT_FMT ")\n", arg);
} else if (MP_PARSE_NODE_IS_LEAF(pn)) {
machine_uint_t arg = MP_PARSE_NODE_LEAF_ARG(pn);
switch (MP_PARSE_NODE_LEAF_KIND(pn)) {
case MP_PARSE_NODE_ID: printf("id(%s)\n", qstr_str(arg)); break;
case MP_PARSE_NODE_INTEGER: printf("int(%s)\n", qstr_str(arg)); break;
case MP_PARSE_NODE_DECIMAL: printf("dec(%s)\n", qstr_str(arg)); break;
case MP_PARSE_NODE_STRING: printf("str(%s)\n", qstr_str(arg)); break;
case MP_PARSE_NODE_BYTES: printf("bytes(%s)\n", qstr_str(arg)); break;
case MP_PARSE_NODE_TOKEN: printf("tok(" INT_FMT ")\n", arg); break;
default: assert(0);
}
} else {
// node must be a mp_parse_node_struct_t
mp_parse_node_struct_t *pns = (mp_parse_node_struct_t*)pn;
if (MP_PARSE_NODE_STRUCT_KIND(pns) == RULE_string) {
printf("literal str(%.*s)\n", (int)pns->nodes[1], (char*)pns->nodes[0]);
} else {
uint n = MP_PARSE_NODE_STRUCT_NUM_NODES(pns);
#ifdef USE_RULE_NAME
printf("%s(%d) (n=%d)\n", rules[MP_PARSE_NODE_STRUCT_KIND(pns)]->rule_name, MP_PARSE_NODE_STRUCT_KIND(pns), n);
#else
printf("rule(%u) (n=%d)\n", (uint)MP_PARSE_NODE_STRUCT_KIND(pns), n);
#endif
for (uint i = 0; i < n; i++) {
mp_parse_node_print(pns->nodes[i], indent + 2);
}
}
}
}
STATIC bool fold_constants(parser_t *parser, const rule_t *rule, size_t num_args) {
// this code does folding of arbitrary integer expressions, eg 1 + 2 * 3 + 4
// it does not do partial folding, eg 1 + 2 + x -> 3 + x
mp_obj_t arg0;
if (rule->rule_id == RULE_expr
|| rule->rule_id == RULE_xor_expr
|| rule->rule_id == RULE_and_expr) {
// folding for binary ops: | ^ &
mp_parse_node_t pn = peek_result(parser, num_args - 1);
if (!mp_parse_node_get_int_maybe(pn, &arg0)) {
return false;
}
mp_binary_op_t op;
if (rule->rule_id == RULE_expr) {
op = MP_BINARY_OP_OR;
} else if (rule->rule_id == RULE_xor_expr) {
op = MP_BINARY_OP_XOR;
} else {
op = MP_BINARY_OP_AND;
}
for (ssize_t i = num_args - 2; i >= 0; --i) {
pn = peek_result(parser, i);
mp_obj_t arg1;
if (!mp_parse_node_get_int_maybe(pn, &arg1)) {
return false;
}
arg0 = mp_binary_op(op, arg0, arg1);
}
} else if (rule->rule_id == RULE_shift_expr
|| rule->rule_id == RULE_arith_expr
|| rule->rule_id == RULE_term) {
// folding for binary ops: << >> + - * / % //
mp_parse_node_t pn = peek_result(parser, num_args - 1);
if (!mp_parse_node_get_int_maybe(pn, &arg0)) {
return false;
}
for (ssize_t i = num_args - 2; i >= 1; i -= 2) {
pn = peek_result(parser, i - 1);
mp_obj_t arg1;
if (!mp_parse_node_get_int_maybe(pn, &arg1)) {
return false;
}
mp_token_kind_t tok = MP_PARSE_NODE_LEAF_ARG(peek_result(parser, i));
static const uint8_t token_to_op[] = {
MP_BINARY_OP_ADD,
MP_BINARY_OP_SUBTRACT,
MP_BINARY_OP_MULTIPLY,
255,//MP_BINARY_OP_POWER,
255,//MP_BINARY_OP_TRUE_DIVIDE,
MP_BINARY_OP_FLOOR_DIVIDE,
MP_BINARY_OP_MODULO,
255,//MP_BINARY_OP_LESS
MP_BINARY_OP_LSHIFT,
255,//MP_BINARY_OP_MORE
MP_BINARY_OP_RSHIFT,
};
mp_binary_op_t op = token_to_op[tok - MP_TOKEN_OP_PLUS];
if (op == (mp_binary_op_t)255) {
return false;
}
int rhs_sign = mp_obj_int_sign(arg1);
if (op <= MP_BINARY_OP_RSHIFT) {
// << and >> can't have negative rhs
if (rhs_sign < 0) {
return false;
}
} else if (op >= MP_BINARY_OP_FLOOR_DIVIDE) {
// % and // can't have zero rhs
if (rhs_sign == 0) {
return false;
}
}
arg0 = mp_binary_op(op, arg0, arg1);
}
} else if (rule->rule_id == RULE_factor_2) {
// folding for unary ops: + - ~
mp_parse_node_t pn = peek_result(parser, 0);
if (!mp_parse_node_get_int_maybe(pn, &arg0)) {
return false;
}
mp_token_kind_t tok = MP_PARSE_NODE_LEAF_ARG(peek_result(parser, 1));
mp_unary_op_t op;
if (tok == MP_TOKEN_OP_PLUS) {
op = MP_UNARY_OP_POSITIVE;
} else if (tok == MP_TOKEN_OP_MINUS) {
op = MP_UNARY_OP_NEGATIVE;
} else {
assert(tok == MP_TOKEN_OP_TILDE); // should be
op = MP_UNARY_OP_INVERT;
}
arg0 = mp_unary_op(op, arg0);
#if MICROPY_COMP_CONST
} else if (rule->rule_id == RULE_expr_stmt) {
mp_parse_node_t pn1 = peek_result(parser, 0);
if (!MP_PARSE_NODE_IS_NULL(pn1)
&& !(MP_PARSE_NODE_IS_STRUCT_KIND(pn1, RULE_expr_stmt_augassign)
|| MP_PARSE_NODE_IS_STRUCT_KIND(pn1, RULE_expr_stmt_assign_list))) {
// this node is of the form <x> = <y>
mp_parse_node_t pn0 = peek_result(parser, 1);
if (MP_PARSE_NODE_IS_ID(pn0)
&& MP_PARSE_NODE_IS_STRUCT_KIND(pn1, RULE_atom_expr_normal)
&& MP_PARSE_NODE_IS_ID(((mp_parse_node_struct_t*)pn1)->nodes[0])
&& MP_PARSE_NODE_LEAF_ARG(((mp_parse_node_struct_t*)pn1)->nodes[0]) == MP_QSTR_const
&& MP_PARSE_NODE_IS_STRUCT_KIND(((mp_parse_node_struct_t*)pn1)->nodes[1], RULE_trailer_paren)
) {
// code to assign dynamic constants: id = const(value)
// get the id
qstr id = MP_PARSE_NODE_LEAF_ARG(pn0);
// get the value
mp_parse_node_t pn_value = ((mp_parse_node_struct_t*)((mp_parse_node_struct_t*)pn1)->nodes[1])->nodes[0];
mp_obj_t value;
if (!mp_parse_node_get_int_maybe(pn_value, &value)) {
mp_obj_t exc = mp_obj_new_exception_msg(&mp_type_SyntaxError,
"constant must be an integer");
mp_obj_exception_add_traceback(exc, parser->lexer->source_name,
((mp_parse_node_struct_t*)pn1)->source_line, MP_QSTR_NULL);
nlr_raise(exc);
}
// store the value in the table of dynamic constants
mp_map_elem_t *elem = mp_map_lookup(&parser->consts, MP_OBJ_NEW_QSTR(id), MP_MAP_LOOKUP_ADD_IF_NOT_FOUND);
assert(elem->value == MP_OBJ_NULL);
elem->value = value;
// If the constant starts with an underscore then treat it as a private
// variable and don't emit any code to store the value to the id.
if (qstr_str(id)[0] == '_') {
pop_result(parser); // pop const(value)
pop_result(parser); // pop id
push_result_rule(parser, 0, rules[RULE_pass_stmt], 0); // replace with "pass"
return true;
}
// replace const(value) with value
pop_result(parser);
push_result_node(parser, pn_value);
// finished folding this assignment, but we still want it to be part of the tree
return false;
}
}
return false;
#endif
#if MICROPY_COMP_MODULE_CONST
} else if (rule->rule_id == RULE_atom_expr_normal) {
mp_parse_node_t pn0 = peek_result(parser, 1);
mp_parse_node_t pn1 = peek_result(parser, 0);
if (!(MP_PARSE_NODE_IS_ID(pn0)
&& MP_PARSE_NODE_IS_STRUCT_KIND(pn1, RULE_trailer_period))) {
return false;
}
// id1.id2
// look it up in constant table, see if it can be replaced with an integer
mp_parse_node_struct_t *pns1 = (mp_parse_node_struct_t*)pn1;
assert(MP_PARSE_NODE_IS_ID(pns1->nodes[0]));
qstr q_base = MP_PARSE_NODE_LEAF_ARG(pn0);
qstr q_attr = MP_PARSE_NODE_LEAF_ARG(pns1->nodes[0]);
mp_map_elem_t *elem = mp_map_lookup((mp_map_t*)&mp_constants_map, MP_OBJ_NEW_QSTR(q_base), MP_MAP_LOOKUP);
if (elem == NULL) {
return false;
}
mp_obj_t dest[2];
mp_load_method_maybe(elem->value, q_attr, dest);
if (!(dest[0] != MP_OBJ_NULL && MP_OBJ_IS_INT(dest[0]) && dest[1] == MP_OBJ_NULL)) {
return false;
}
arg0 = dest[0];
#endif
} else {
return false;
}
// success folding this rule
for (size_t i = num_args; i > 0; i--) {
pop_result(parser);
}
if (MP_OBJ_IS_SMALL_INT(arg0)) {
push_result_node(parser, mp_parse_node_new_small_int(MP_OBJ_SMALL_INT_VALUE(arg0)));
} else {
// TODO reuse memory for parse node struct?
push_result_node(parser, make_node_const_object(parser, 0, arg0));
}
return true;
}