Optimise string parsing when there are no escape characters

This resulted in approximately 40% reduction in instructions used in
parse_string() when parsing a 4MB JSON file, as measured in callgrind.
This commit is contained in:
Roger Light 2023-08-10 09:26:19 +01:00
parent bb27ffa152
commit e64b984ddc
2 changed files with 67 additions and 58 deletions

16
cJSON.c
View File

@ -776,6 +776,8 @@ static cJSON_bool parse_string(cJSON * const item, parse_buffer * const input_bu
const unsigned char *input_end = buffer_at_offset(input_buffer) + 1;
unsigned char *output_pointer = NULL;
unsigned char *output = NULL;
size_t skipped_bytes = 0;
size_t allocation_length = 0;
/* not a string */
if (buffer_at_offset(input_buffer)[0] != '\"')
@ -785,8 +787,6 @@ static cJSON_bool parse_string(cJSON * const item, parse_buffer * const input_bu
{
/* calculate approximate size of the output (overestimate) */
size_t allocation_length = 0;
size_t skipped_bytes = 0;
while (((size_t)(input_end - input_buffer->content) < input_buffer->length) && (*input_end != '\"'))
{
/* is escape sequence */
@ -816,8 +816,16 @@ static cJSON_bool parse_string(cJSON * const item, parse_buffer * const input_bu
}
}
/* If there are no escaped characters, we can use memcmp */
if(skipped_bytes == 0)
{
memcpy(output, input_pointer, allocation_length);
output[allocation_length-1] = '\0';
}
else
{
/* else loop through the string literal */
output_pointer = output;
/* loop through the string literal */
while (input_pointer < input_end)
{
if (*input_pointer != '\\')
@ -872,9 +880,9 @@ static cJSON_bool parse_string(cJSON * const item, parse_buffer * const input_bu
input_pointer += sequence_length;
}
}
/* zero terminate the output */
*output_pointer = '\0';
}
item->type = cJSON_String;
item->valuestring = (char*)output;

View File

@ -73,6 +73,7 @@ static void assert_not_parse_string(const char * const string)
static void parse_string_should_parse_strings(void)
{
assert_parse_string("\"\"", "");
assert_parse_string("\"Simple string\"", "Simple string");
assert_parse_string(
"\" !\\\"#$%&'()*+,-./\\/0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\\\]^_'abcdefghijklmnopqrstuvwxyz{|}~\"",
" !\"#$%&'()*+,-.//0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\]^_'abcdefghijklmnopqrstuvwxyz{|}~");