From 881078403e9edf63f3145f2a484a58d5a74f74f1 Mon Sep 17 00:00:00 2001 From: Paul Sokolovsky Date: Sat, 26 Apr 2014 06:20:08 +0300 Subject: [PATCH] objstr: Implement .lstrip() & .rstrip(). Share code with .strip(). TODO: optimize .rstrip(). --- py/objstr.c | 30 ++++++++++++++++++++++++++++-- py/qstrdefs.h | 2 ++ tests/basics/string_strip.py | 6 ++++++ 3 files changed, 36 insertions(+), 2 deletions(-) diff --git a/py/objstr.c b/py/objstr.c index b0d5cba6f1..b5504da1b5 100644 --- a/py/objstr.c +++ b/py/objstr.c @@ -493,7 +493,9 @@ STATIC mp_obj_t str_startswith(mp_obj_t self_in, mp_obj_t arg) { return MP_BOOL(memcmp(str, prefix, prefix_len) == 0); } -STATIC mp_obj_t str_strip(uint n_args, const mp_obj_t *args) { +enum { LSTRIP, RSTRIP, STRIP }; + +STATIC mp_obj_t str_uni_strip(int type, uint n_args, const mp_obj_t *args) { assert(1 <= n_args && n_args <= 2); assert(MP_OBJ_IS_STR(args[0])); @@ -516,13 +518,18 @@ STATIC mp_obj_t str_strip(uint n_args, const mp_obj_t *args) { machine_uint_t first_good_char_pos = 0; bool first_good_char_pos_set = false; machine_uint_t last_good_char_pos = 0; + // TODO: For RSPLIT, scan from end for (machine_uint_t i = 0; i < orig_str_len; i++) { if (find_subbytes(chars_to_del, chars_to_del_len, &orig_str[i], 1, 1) == NULL) { - last_good_char_pos = i; if (!first_good_char_pos_set) { first_good_char_pos = i; + if (type == LSTRIP) { + last_good_char_pos = orig_str_len - 1; + break; + } first_good_char_pos_set = true; } + last_good_char_pos = i; } } @@ -532,11 +539,26 @@ STATIC mp_obj_t str_strip(uint n_args, const mp_obj_t *args) { } assert(last_good_char_pos >= first_good_char_pos); + if (type == RSTRIP) { + first_good_char_pos = 0; + } //+1 to accomodate the last character machine_uint_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); } +STATIC mp_obj_t str_strip(uint n_args, const mp_obj_t *args) { + return str_uni_strip(STRIP, n_args, args); +} + +STATIC mp_obj_t str_lstrip(uint n_args, const mp_obj_t *args) { + return str_uni_strip(LSTRIP, n_args, args); +} + +STATIC mp_obj_t str_rstrip(uint n_args, const mp_obj_t *args) { + return str_uni_strip(RSTRIP, n_args, args); +} + // Takes an int arg, but only parses unsigned numbers, and only changes // *num if at least one digit was parsed. static int str_to_int(const char *str, int *num) { @@ -1354,6 +1376,8 @@ STATIC MP_DEFINE_CONST_FUN_OBJ_2(str_join_obj, str_join); STATIC MP_DEFINE_CONST_FUN_OBJ_VAR_BETWEEN(str_split_obj, 1, 3, str_split); STATIC MP_DEFINE_CONST_FUN_OBJ_2(str_startswith_obj, str_startswith); STATIC MP_DEFINE_CONST_FUN_OBJ_VAR_BETWEEN(str_strip_obj, 1, 2, str_strip); +STATIC MP_DEFINE_CONST_FUN_OBJ_VAR_BETWEEN(str_lstrip_obj, 1, 2, str_lstrip); +STATIC MP_DEFINE_CONST_FUN_OBJ_VAR_BETWEEN(str_rstrip_obj, 1, 2, str_rstrip); STATIC MP_DEFINE_CONST_FUN_OBJ_VAR(str_format_obj, 1, mp_obj_str_format); STATIC MP_DEFINE_CONST_FUN_OBJ_VAR_BETWEEN(str_replace_obj, 3, 4, str_replace); STATIC MP_DEFINE_CONST_FUN_OBJ_VAR_BETWEEN(str_count_obj, 2, 4, str_count); @@ -1373,6 +1397,8 @@ STATIC const mp_map_elem_t str_locals_dict_table[] = { { MP_OBJ_NEW_QSTR(MP_QSTR_split), (mp_obj_t)&str_split_obj }, { MP_OBJ_NEW_QSTR(MP_QSTR_startswith), (mp_obj_t)&str_startswith_obj }, { MP_OBJ_NEW_QSTR(MP_QSTR_strip), (mp_obj_t)&str_strip_obj }, + { MP_OBJ_NEW_QSTR(MP_QSTR_lstrip), (mp_obj_t)&str_lstrip_obj }, + { MP_OBJ_NEW_QSTR(MP_QSTR_rstrip), (mp_obj_t)&str_rstrip_obj }, { MP_OBJ_NEW_QSTR(MP_QSTR_format), (mp_obj_t)&str_format_obj }, { MP_OBJ_NEW_QSTR(MP_QSTR_replace), (mp_obj_t)&str_replace_obj }, { MP_OBJ_NEW_QSTR(MP_QSTR_count), (mp_obj_t)&str_count_obj }, diff --git a/py/qstrdefs.h b/py/qstrdefs.h index 784bf59d23..bcbbcf98e9 100644 --- a/py/qstrdefs.h +++ b/py/qstrdefs.h @@ -180,6 +180,8 @@ Q(pop) Q(sort) Q(join) Q(strip) +Q(lstrip) +Q(rstrip) Q(format) Q(key) Q(reverse) diff --git a/tests/basics/string_strip.py b/tests/basics/string_strip.py index 518dfd66ec..8e03eff93a 100644 --- a/tests/basics/string_strip.py +++ b/tests/basics/string_strip.py @@ -4,3 +4,9 @@ print(" T E S T".strip()) print("abcabc".strip("ce")) print("aaa".strip("b")) print("abc efg ".strip("g a")) + +print(' spacious '.lstrip()) +print('www.example.com'.lstrip('cmowz.')) + +print(' spacious '.rstrip()) +print('mississippi'.rstrip('ipz'))