static mp_obj_t str_find(uint n_args, const mp_obj_t *args) {
assert(2 <= n_args && n_args <= 4);
assert(MP_OBJ_IS_STR(args[0]));
assert(MP_OBJ_IS_STR(args[1]));
GET_STR_DATA_LEN(args[0], haystack, haystack_len);
GET_STR_DATA_LEN(args[1], needle, needle_len);
size_t start = 0;
size_t end = haystack_len;
/* TODO use a non-exception-throwing mp_get_index */
if (n_args >= 3 && args[2] != mp_const_none) {
start = mp_get_index(&str_type, haystack_len, args[2]);
}
if (n_args >= 4 && args[3] != mp_const_none) {
end = mp_get_index(&str_type, haystack_len, args[3]);
}
const byte *p = find_subbytes(haystack + start, haystack_len - start, needle, needle_len);
if (p == NULL) {
// not found
return MP_OBJ_NEW_SMALL_INT(-1);
} else {
// found
machine_int_t pos = p - haystack;
if (pos + needle_len > end) {
pos = -1;
}
return MP_OBJ_NEW_SMALL_INT(pos);
}
}
// TODO: (Much) more variety in args
static mp_obj_t str_startswith(mp_obj_t self_in, mp_obj_t arg) {
GET_STR_DATA_LEN(self_in, str, str_len);
GET_STR_DATA_LEN(arg, prefix, prefix_len);
if (prefix_len > str_len) {
return mp_const_false;
}
return MP_BOOL(memcmp(str, prefix, prefix_len) == 0);
}
mp_obj_t str_join(mp_obj_t self_in, mp_obj_t arg) {
assert(MP_OBJ_IS_STR(self_in));
// get separation string
GET_STR_DATA_LEN(self_in, sep_str, sep_len);
// process args
uint seq_len;
mp_obj_t *seq_items;
if (MP_OBJ_IS_TYPE(arg, &tuple_type)) {
mp_obj_tuple_get(arg, &seq_len, &seq_items);
} else if (MP_OBJ_IS_TYPE(arg, &list_type)) {
mp_obj_list_get(arg, &seq_len, &seq_items);
} else {
goto bad_arg;
}
// count required length
int required_len = 0;
for (int i = 0; i < seq_len; i++) {
if (!MP_OBJ_IS_STR(seq_items[i])) {
goto bad_arg;
}
if (i > 0) {
required_len += sep_len;
}
GET_STR_LEN(seq_items[i], l);
required_len += l;
}
// make joined string
byte *data;
mp_obj_t joined_str = mp_obj_str_builder_start(required_len, &data);
for (int i = 0; i < seq_len; i++) {
if (i > 0) {
memcpy(data, sep_str, sep_len);
data += sep_len;
}
GET_STR_DATA_LEN(seq_items[i], s, l);
memcpy(data, s, l);
data += l;
}
// return joined string
return mp_obj_str_builder_end(joined_str);
bad_arg:
nlr_jump(mp_obj_new_exception_msg(MP_QSTR_TypeError, "?str.join expecting a list of str's"));
}
mp_obj_t str_format(uint n_args, const mp_obj_t *args) {
assert(MP_OBJ_IS_STR(args[0]));
GET_STR_DATA_LEN(args[0], str, len);
int arg_i = 1;
vstr_t *vstr = vstr_new();
for (const byte *top = str + len; str < top; str++) {
if (*str == '{') {
str++;
if (str < top && *str == '{') {
vstr_add_char(vstr, '{');
} else {
while (str < top && *str != '}') str++;
if (arg_i >= n_args) {
nlr_jump(mp_obj_new_exception_msg(MP_QSTR_IndexError, "tuple index out of range"));
}
// TODO: may be PRINT_REPR depending on formatting code
mp_obj_print_helper((void (*)(void*, const char*, ...))vstr_printf, vstr, args[arg_i], PRINT_STR);
arg_i++;
}
} else {
vstr_add_char(vstr, *str);
}
}
mp_obj_t s = mp_obj_new_str((byte*)vstr->buf, vstr->len, false);
vstr_free(vstr);
return s;
}
bool mp_obj_str_equal(mp_obj_t s1, mp_obj_t s2) {
if (MP_OBJ_IS_QSTR(s1) && MP_OBJ_IS_QSTR(s2)) {
return s1 == s2;
} else {
GET_STR_HASH(s1, h1);
GET_STR_HASH(s2, h2);
if (h1 != h2) {
return false;
}
GET_STR_DATA_LEN(s1, d1, l1);
GET_STR_DATA_LEN(s2, d2, l2);
if (l1 != l2) {
return false;
}
return strncmp((const char*)d1, (const char*)d2, l1) == 0;
}
}
void str_print(void (*print)(void *env, const char *fmt, ...), void *env, mp_obj_t self_in, mp_print_kind_t kind) {
GET_STR_DATA_LEN(self_in, str_data, str_len);
if (kind == PRINT_STR) {
print(env, "%.*s", str_len, str_data);
} else {
// TODO need to escape chars etc
print(env, "'%.*s'", str_len, str_data);
}
}
// only use this function if you need the str data to be zero terminated
// at the moment all strings are zero terminated to help with C ASCIIZ compatibility
const char *mp_obj_str_get_str(mp_obj_t self_in) {
if (MP_OBJ_IS_STR(self_in)) {
GET_STR_DATA_LEN(self_in, s, l);
(void)l; // len unused
return (const char*)s;
} else {
nlr_jump(mp_obj_new_exception_msg_varg(MP_QSTR_TypeError, "Can't convert '%s' object to str implicitly",
mp_obj_get_type_str(self_in)));
}
}
mp_obj_t str_strip(uint n_args, const mp_obj_t *args) {
assert(1 <= n_args && n_args <= 2);
assert(MP_OBJ_IS_STR(args[0]));
const byte *chars_to_del;
uint chars_to_del_len;
static const byte whitespace[] = " \t\n\r\v\f";
if (n_args == 1) {
chars_to_del = whitespace;
chars_to_del_len = sizeof(whitespace);
} else {
assert(MP_OBJ_IS_STR(args[1]));
GET_STR_DATA_LEN(args[1], s, l);
chars_to_del = s;
chars_to_del_len = l;
}
GET_STR_DATA_LEN(args[0], orig_str, orig_str_len);
size_t first_good_char_pos = 0;
bool first_good_char_pos_set = false;
size_t last_good_char_pos = 0;
for (size_t i = 0; i < orig_str_len; i++) {
if (!chr_in_str(chars_to_del, chars_to_del_len, orig_str[i])) {
last_good_char_pos = i;
if (!first_good_char_pos_set) {
first_good_char_pos = i;
first_good_char_pos_set = true;
}
}
}
if (first_good_char_pos == 0 && last_good_char_pos == 0) {
// string is all whitespace, return ''
return MP_OBJ_NEW_QSTR(MP_QSTR_);
}
assert(last_good_char_pos >= first_good_char_pos);
//+1 to accomodate the last character
size_t stripped_len = last_good_char_pos - first_good_char_pos + 1;
return mp_obj_new_str(orig_str + first_good_char_pos, stripped_len, false);
}
const byte *mp_obj_str_get_data(mp_obj_t self_in, uint *len) {
if (MP_OBJ_IS_STR(self_in)) {
GET_STR_DATA_LEN(self_in, s, l);
*len = l;
return s;
} else {
nlr_jump(mp_obj_new_exception_msg_varg(MP_QSTR_TypeError, "Can't convert '%s' object to str implicitly",
mp_obj_get_type_str(self_in)));
}
}
mp_obj_t str_it_iternext(mp_obj_t self_in) {
mp_obj_str_it_t *self = self_in;
GET_STR_DATA_LEN(self->str, str, len);
if (self->cur < len) {
mp_obj_t o_out = mp_obj_new_str(str + self->cur, 1, true);
self->cur += 1;
return o_out;
} else {
return mp_const_stop_iteration;
}
}
STATIC mp_obj_t uni_unary_op(mp_uint_t op, mp_obj_t self_in) {
GET_STR_DATA_LEN(self_in, str_data, str_len);
switch (op) {
case MP_UNARY_OP_BOOL:
return mp_obj_new_bool(str_len != 0);
case MP_UNARY_OP_LEN:
return MP_OBJ_NEW_SMALL_INT(unichar_charlen((const char *)str_data, str_len));
default:
return MP_OBJ_NULL; // op not supported
}
}
STATIC void uni_print(const mp_print_t *print, mp_obj_t self_in, mp_print_kind_t kind) {
GET_STR_DATA_LEN(self_in, str_data, str_len);
#if MICROPY_PY_UJSON
if (kind == PRINT_JSON) {
mp_str_print_json(print, str_data, str_len);
return;
}
#endif
if (kind == PRINT_STR) {
mp_printf(print, "%.*s", str_len, str_data);
} else {
uni_print_quoted(print, str_data, str_len);
}
}
STATIC void uni_print(void (*print)(void *env, const char *fmt, ...), void *env, mp_obj_t self_in, mp_print_kind_t kind) {
GET_STR_DATA_LEN(self_in, str_data, str_len);
#if MICROPY_PY_UJSON
if (kind == PRINT_JSON) {
mp_str_print_json(print, env, str_data, str_len);
return;
}
#endif
if (kind == PRINT_STR) {
print(env, "%.*s", str_len, str_data);
} else {
uni_print_quoted(print, env, str_data, str_len);
}
}
STATIC mp_obj_t str_subscr(mp_obj_t self_in, mp_obj_t index, mp_obj_t value) {
mp_obj_type_t *type = mp_obj_get_type(self_in);
assert(type == &mp_type_str);
GET_STR_DATA_LEN(self_in, self_data, self_len);
if (value == MP_OBJ_SENTINEL) {
// load
#if MICROPY_PY_BUILTINS_SLICE
if (MP_OBJ_IS_TYPE(index, &mp_type_slice)) {
mp_obj_t ostart, ostop, ostep;
mp_obj_slice_get(index, &ostart, &ostop, &ostep);
if (ostep != mp_const_none && ostep != MP_OBJ_NEW_SMALL_INT(1)) {
nlr_raise(mp_obj_new_exception_msg(&mp_type_NotImplementedError,
"only slices with step=1 (aka None) are supported"));
}
const byte *pstart, *pstop;
if (ostart != mp_const_none) {
pstart = str_index_to_ptr(type, self_data, self_len, ostart, true);
} else {
pstart = self_data;
}
if (ostop != mp_const_none) {
// pstop will point just after the stop character. This depends on
// the \0 at the end of the string.
pstop = str_index_to_ptr(type, self_data, self_len, ostop, true);
} else {
pstop = self_data + self_len;
}
if (pstop < pstart) {
return MP_OBJ_NEW_QSTR(MP_QSTR_);
}
return mp_obj_new_str_of_type(type, (const byte *)pstart, pstop - pstart);
}
#endif
const byte *s = str_index_to_ptr(type, self_data, self_len, index, false);
int len = 1;
if (UTF8_IS_NONASCII(*s)) {
// Count the number of 1 bits (after the first)
for (char mask = 0x40; *s & mask; mask >>= 1) {
++len;
}
}
return mp_obj_new_str((const char*)s, len, true); // This will create a one-character string
} else {
return MP_OBJ_NULL; // op not supported
STATIC mp_obj_t str_make_new(mp_obj_t type_in, mp_uint_t n_args, mp_uint_t n_kw, const mp_obj_t *args) {
#if MICROPY_CPYTHON_COMPAT
if (n_kw != 0) {
mp_arg_error_unimpl_kw();
}
#endif
switch (n_args) {
case 0:
return MP_OBJ_NEW_QSTR(MP_QSTR_);
case 1:
{
vstr_t *vstr = vstr_new();
mp_obj_print_helper((void (*)(void*, const char*, ...))vstr_printf, vstr, args[0], PRINT_STR);
mp_obj_t s = mp_obj_new_str(vstr->buf, vstr->len, false);
vstr_free(vstr);
return s;
}
case 2:
case 3:
{
// TODO: validate 2nd/3rd args
if (MP_OBJ_IS_TYPE(args[0], &mp_type_bytes)) {
GET_STR_DATA_LEN(args[0], str_data, str_len);
GET_STR_HASH(args[0], str_hash);
mp_obj_str_t *o = mp_obj_new_str_of_type(&mp_type_str, NULL, str_len);
o->data = str_data;
o->hash = str_hash;
return o;
} else {
mp_buffer_info_t bufinfo;
mp_get_buffer_raise(args[0], &bufinfo, MP_BUFFER_READ);
return mp_obj_new_str(bufinfo.buf, bufinfo.len, false);
}
}
default:
nlr_raise(mp_obj_new_exception_msg(&mp_type_TypeError, "str takes at most 3 arguments"));
}
}
static mp_obj_t str_split(uint n_args, const mp_obj_t *args) {
int splits = -1;
mp_obj_t sep = mp_const_none;
if (n_args > 1) {
sep = args[1];
if (n_args > 2) {
splits = MP_OBJ_SMALL_INT_VALUE(args[2]);
}
}
assert(sep == mp_const_none);
(void)sep; // unused; to hush compiler warning
mp_obj_t res = mp_obj_new_list(0, NULL);
GET_STR_DATA_LEN(args[0], s, len);
const byte *top = s + len;
const byte *start;
// Initial whitespace is not counted as split, so we pre-do it
while (s < top && is_ws(*s)) s++;
while (s < top && splits != 0) {
start = s;
while (s < top && !is_ws(*s)) s++;
rt_list_append(res, mp_obj_new_str(start, s - start, false));
if (s >= top) {
break;
}
while (s < top && is_ws(*s)) s++;
if (splits > 0) {
splits--;
}
}
if (s < top) {
rt_list_append(res, mp_obj_new_str(s, top - s, false));
}
return res;
}
mp_obj_t str_binary_op(int op, mp_obj_t lhs_in, mp_obj_t rhs_in) {
GET_STR_DATA_LEN(lhs_in, lhs_data, lhs_len);
switch (op) {
case RT_BINARY_OP_SUBSCR:
// TODO: need predicate to check for int-like type (bools are such for example)
// ["no", "yes"][1 == 2] is common idiom
if (MP_OBJ_IS_SMALL_INT(rhs_in)) {
uint index = mp_get_index(mp_obj_get_type(lhs_in), lhs_len, rhs_in);
return mp_obj_new_str(lhs_data + index, 1, true);
#if MICROPY_ENABLE_SLICE
} else if (MP_OBJ_IS_TYPE(rhs_in, &slice_type)) {
machine_int_t start, stop, step;
mp_obj_slice_get(rhs_in, &start, &stop, &step);
assert(step == 1);
if (start < 0) {
start = lhs_len + start;
if (start < 0) {
start = 0;
}
} else if (start > lhs_len) {
start = lhs_len;
}
if (stop <= 0) {
stop = lhs_len + stop;
// CPython returns empty string in such case
if (stop < 0) {
stop = start;
}
} else if (stop > lhs_len) {
stop = lhs_len;
}
return mp_obj_new_str(lhs_data + start, stop - start, false);
#endif
} else {
// Message doesn't match CPython, but we don't have so much bytes as they
// to spend them on verbose wording
nlr_jump(mp_obj_new_exception_msg(MP_QSTR_TypeError, "index must be int"));
}
case RT_BINARY_OP_ADD:
case RT_BINARY_OP_INPLACE_ADD:
if (MP_OBJ_IS_STR(rhs_in)) {
// add 2 strings
GET_STR_DATA_LEN(rhs_in, rhs_data, rhs_len);
int alloc_len = lhs_len + rhs_len;
/* code for making qstr
byte *q_ptr;
byte *val = qstr_build_start(alloc_len, &q_ptr);
memcpy(val, lhs_data, lhs_len);
memcpy(val + lhs_len, rhs_data, rhs_len);
return MP_OBJ_NEW_QSTR(qstr_build_end(q_ptr));
*/
// code for non-qstr
byte *data;
mp_obj_t s = mp_obj_str_builder_start(alloc_len, &data);
memcpy(data, lhs_data, lhs_len);
memcpy(data + lhs_len, rhs_data, rhs_len);
return mp_obj_str_builder_end(s);
}
break;
case RT_COMPARE_OP_IN:
case RT_COMPARE_OP_NOT_IN:
/* NOTE `a in b` is `b.__contains__(a)` */
if (MP_OBJ_IS_STR(rhs_in)) {
GET_STR_DATA_LEN(rhs_in, rhs_data, rhs_len);
return MP_BOOL((op == RT_COMPARE_OP_IN) ^ (find_subbytes(lhs_data, lhs_len, rhs_data, rhs_len) == NULL));
}
break;
case RT_BINARY_OP_MULTIPLY:
{
if (!MP_OBJ_IS_SMALL_INT(rhs_in)) {
return NULL;
}
int n = MP_OBJ_SMALL_INT_VALUE(rhs_in);
byte *data;
mp_obj_t s = mp_obj_str_builder_start(lhs_len * n, &data);
mp_seq_multiply(lhs_data, sizeof(*lhs_data), lhs_len, n, data);
return mp_obj_str_builder_end(s);
}
}
return MP_OBJ_NULL; // op not supported
}
