objstr: split(None): Fix whitespace properly.

This commit is contained in:
Paul Sokolovsky 2015-04-12 00:17:16 +03:00
parent 6eb7530083
commit 8b7faa31e1
2 changed files with 5 additions and 4 deletions

View File

@ -454,7 +454,6 @@ STATIC mp_obj_t str_join(mp_obj_t self_in, mp_obj_t arg) {
return mp_obj_new_str_from_vstr(self_type, &vstr); return mp_obj_new_str_from_vstr(self_type, &vstr);
} }
#define is_ws(c) ((c) == ' ' || (c) == '\t')
enum {SPLIT = 0, KEEP = 1, SPLITLINES = 2}; enum {SPLIT = 0, KEEP = 1, SPLITLINES = 2};
STATIC inline mp_obj_t str_split_internal(mp_uint_t n_args, const mp_obj_t *args, int type) { STATIC inline mp_obj_t str_split_internal(mp_uint_t n_args, const mp_obj_t *args, int type) {
@ -476,15 +475,15 @@ STATIC inline mp_obj_t str_split_internal(mp_uint_t n_args, const mp_obj_t *args
// sep not given, so separate on whitespace // sep not given, so separate on whitespace
// Initial whitespace is not counted as split, so we pre-do it // Initial whitespace is not counted as split, so we pre-do it
while (s < top && is_ws(*s)) s++; while (s < top && unichar_isspace(*s)) s++;
while (s < top && splits != 0) { while (s < top && splits != 0) {
const byte *start = s; const byte *start = s;
while (s < top && !is_ws(*s)) s++; while (s < top && !unichar_isspace(*s)) s++;
mp_obj_list_append(res, mp_obj_new_str_of_type(self_type, start, s - start)); mp_obj_list_append(res, mp_obj_new_str_of_type(self_type, start, s - start));
if (s >= top) { if (s >= top) {
break; break;
} }
while (s < top && is_ws(*s)) s++; while (s < top && unichar_isspace(*s)) s++;
if (splits > 0) { if (splits > 0) {
splits--; splits--;
} }

View File

@ -6,6 +6,8 @@ print(" a b ".split(None, 2))
print(" a b c ".split(None, 1)) print(" a b c ".split(None, 1))
print(" a b c ".split(None, 0)) print(" a b c ".split(None, 0))
print(" a b c ".split(None, -1)) print(" a b c ".split(None, -1))
print("foo\n\t\x07\v\nbar".split())
print("foo\nbar\n".split())
# empty separator should fail # empty separator should fail
try: try: