Commit 94f68300 authored by Damien George's avatar Damien George
Browse files

py: Make str.replace do 2 passes over the string.

parent 1aa1c511
......@@ -413,61 +413,82 @@ mp_obj_t str_replace(uint n_args, const mp_obj_t *args) {
assert(MP_OBJ_IS_STR(args[1]));
assert(MP_OBJ_IS_STR(args[2]));
int max_rep = 0;
machine_int_t max_rep = 0;
if (n_args == 4) {
assert(MP_OBJ_IS_SMALL_INT(args[3]));
max_rep = MP_OBJ_SMALL_INT_VALUE(args[3]);
if (max_rep == 0) {
return(args[0]);
}
else if (max_rep < 0) {
return args[0];
} else if (max_rep < 0) {
max_rep = 0;
}
}
//if max_rep is still 0 by this point we will need to do all possible replacements
// if max_rep is still 0 by this point we will need to do all possible replacements
GET_STR_DATA_LEN(args[0], str, str_len);
GET_STR_DATA_LEN(args[1], old, old_len);
GET_STR_DATA_LEN(args[2], new, new_len);
// old won't exist in str if it's longer, so nothing to replace
if (old_len > str_len) {
return(args[0]);
}
return args[0];
}
// data for the replaced string
byte *data = NULL;
mp_obj_t replaced_str = MP_OBJ_NULL;
// do 2 passes over the string:
// first pass computes the required length of the replaced string
// second pass does the replacements
for (;;) {
machine_uint_t replaced_str_index = 0;
machine_uint_t num_replacements_done = 0;
const byte *old_occurrence;
const byte *offset_ptr = str;
machine_uint_t offset_num = 0;
while ((old_occurrence = find_subbytes(offset_ptr, str_len - offset_num, old, old_len)) != NULL) {
// copy from just after end of last occurrence of to-be-replaced string to right before start of next occurrence
if (data != NULL) {
memcpy(data + replaced_str_index, offset_ptr, old_occurrence - offset_ptr);
}
replaced_str_index += old_occurrence - offset_ptr;
// copy the replacement string
if (data != NULL) {
memcpy(data + replaced_str_index, new, new_len);
}
replaced_str_index += new_len;
offset_ptr = old_occurrence + old_len;
offset_num = offset_ptr - str;
size_t required_len = 0;
const byte *old_occurrence;
const byte *offset_ptr = str;
size_t offset_num = 0;
while((old_occurrence = find_subbytes(offset_ptr, str_len - offset_num, old, old_len)) != NULL){
required_len += old_occurrence - offset_ptr;
required_len += new_len;
offset_ptr = old_occurrence + old_len;
offset_num = offset_ptr - str;
}
required_len += str_len - offset_num;
byte *data;
mp_obj_t replaced_str = mp_obj_str_builder_start(mp_obj_get_type(args[0]), required_len, &data);
size_t replaced_str_index = 0;
int replacements_done = 0;
offset_ptr = str;
offset_num = 0;
while((old_occurrence = find_subbytes(offset_ptr, str_len - offset_num, old, old_len)) != NULL){
//copy from just after end of last occurrence of to-be-replaced string to right before start of next occurrence
memcpy(data + replaced_str_index, offset_ptr, old_occurrence - offset_ptr);
replaced_str_index += old_occurrence - offset_ptr;
//copy the replacement string
memcpy(data + replaced_str_index, new, new_len);
replaced_str_index += new_len;
offset_ptr = old_occurrence + old_len;
offset_num = offset_ptr - str;
replacements_done++;
if (max_rep != 0 && replacements_done == max_rep){
break;
}
num_replacements_done++;
if (max_rep != 0 && num_replacements_done == max_rep){
break;
}
}
// copy from just after end of last occurrence of to-be-replaced string to end of old string
if (data != NULL) {
memcpy(data + replaced_str_index, offset_ptr, str_len - offset_num);
}
replaced_str_index += str_len - offset_num;
if (data == NULL) {
// first pass
if (num_replacements_done == 0) {
// no substr found, return original string
return args[0];
} else {
// substr found, allocate new string
replaced_str = mp_obj_str_builder_start(mp_obj_get_type(args[0]), replaced_str_index, &data);
}
} else {
// second pass, we are done
break;
}
}
//copy from just after end of last occurrence of to-be-replaced string to end of old string
memcpy(data + replaced_str_index, offset_ptr, str_len - offset_num);
return mp_obj_str_builder_end(replaced_str);
}
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment