mirror of
https://github.com/curl/curl.git
synced 2026-07-24 01:47:20 +03:00
strparse: speed up the hex parser somewhat
Around 2.3x speed-up parsing many large hexadecimal numbers. The decimal and octal parser get marginally faster. Still very readable, compact and easy to follow code. Tweaks - combine the max and the overflow check, gains 3ns/num (use a separate check outside of the loop instead for max < base) - one less indirection in the pointer, gains 3ns/num - using the table lookup for hex nums, gains 5ns/num - unfold the num_digit() macro, gains 3s/num - use the hexasciitable unconditionally, gains 2ns/num - use post-increment pointer in the table lookup, gains 1ns/num - improved valid_digit() using the table for the hex case, gains 26 ns/num - use "max char" in valid_digit(), gains 3ns/num Behavior changes: - no longer returns STRE_TOO_BIG - only STRE_OVERFLOW - does not move the char ** on error, which is probably better Updated and extended test 1664 (significantly). Closes #16374
This commit is contained in:
parent
3fd1dfc829
commit
ad700a0917
6 changed files with 310 additions and 56 deletions
|
|
@ -995,7 +995,7 @@ static CURLcode cf_socket_ctx_init(struct cf_socket_ctx *ctx,
|
|||
p = getenv("CURL_DBG_SOCK_RMAX");
|
||||
if(p) {
|
||||
curl_off_t l;
|
||||
if(!Curl_str_number(&p, &l, SIZE_T_MAX))
|
||||
if(!Curl_str_number(&p, &l, CURL_OFF_T_MAX))
|
||||
ctx->recv_max = (size_t)l;
|
||||
}
|
||||
}
|
||||
|
|
|
|||
|
|
@ -882,14 +882,14 @@ CURLcode Curl_conn_send(struct Curl_easy *data, int sockindex,
|
|||
DEBUGASSERT(data->conn);
|
||||
conn = data->conn;
|
||||
#ifdef DEBUGBUILD
|
||||
{
|
||||
if(write_len) {
|
||||
/* Allow debug builds to override this logic to force short sends
|
||||
*/
|
||||
const char *p = getenv("CURL_SMALLSENDS");
|
||||
if(p) {
|
||||
curl_off_t altsize;
|
||||
if(!Curl_str_number(&p, &altsize, SIZE_T_MAX))
|
||||
write_len = CURLMIN(write_len, (size_t)altsize);
|
||||
if(!Curl_str_number(&p, &altsize, write_len))
|
||||
write_len = (size_t)altsize;
|
||||
}
|
||||
}
|
||||
#endif
|
||||
|
|
|
|||
|
|
@ -195,11 +195,13 @@ static CURLcode xfer_send(struct Curl_easy *data,
|
|||
/* Allow debug builds to override this logic to force short initial
|
||||
sends */
|
||||
size_t body_len = blen - hds_len;
|
||||
const char *p = getenv("CURL_SMALLREQSEND");
|
||||
if(p) {
|
||||
curl_off_t body_small;
|
||||
if(!Curl_str_number(&p, &body_small, body_len))
|
||||
blen = hds_len + (size_t)body_small;
|
||||
if(body_len) {
|
||||
const char *p = getenv("CURL_SMALLREQSEND");
|
||||
if(p) {
|
||||
curl_off_t body_small;
|
||||
if(!Curl_str_number(&p, &body_small, body_len))
|
||||
blen = hds_len + (size_t)body_small;
|
||||
}
|
||||
}
|
||||
}
|
||||
#endif
|
||||
|
|
|
|||
|
|
@ -104,40 +104,57 @@ int Curl_str_singlespace(const char **linep)
|
|||
return Curl_str_single(linep, ' ');
|
||||
}
|
||||
|
||||
/* given an ASCII hexadecimal character, return the value */
|
||||
#define HEXDIGIT2NUM(x) \
|
||||
(((x) > '9') ? Curl_raw_tolower(x) - 'a' + 10 : x - '0')
|
||||
|
||||
/* given an ASCII character and a given base, return TRUE if valid */
|
||||
#define valid_digit(digit, base) \
|
||||
(((base == 10) && ISDIGIT(digit)) || \
|
||||
((base == 16) && ISXDIGIT(digit)) || \
|
||||
((base == 8) && ISODIGIT(digit)))
|
||||
|
||||
/* given an ASCII character and a given base, return the value */
|
||||
#define num_digit(digit, base) \
|
||||
((base != 16) ? digit - '0' : HEXDIGIT2NUM(digit))
|
||||
/* given an ASCII character and max ascii, return TRUE if valid */
|
||||
#define valid_digit(x,m) \
|
||||
(((x) >= '0') && ((x) <= m) && hexasciitable[(x)-'0'])
|
||||
|
||||
/* no support for 0x prefix nor leading spaces */
|
||||
static int str_num_base(const char **linep, curl_off_t *nump, curl_off_t max,
|
||||
int base) /* 8, 10 or 16, nothing else */
|
||||
{
|
||||
/* We use 16 for the zero index (and the necessary bitwise AND in the loop)
|
||||
to be able to have a non-zero value there to make valid_digit() able to
|
||||
use the info */
|
||||
static const unsigned char hexasciitable[] = {
|
||||
16, 1, 2, 3, 4, 5, 6, 7, 8, 9, /* 0x30: 0 - 9 */
|
||||
0, 0, 0, 0, 0, 0, 0,
|
||||
10, 11, 12, 13, 14, 15, /* 0x41: A - F */
|
||||
0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
|
||||
0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
|
||||
0, 0, 0, 0, 0, 0,
|
||||
10, 11, 12, 13, 14, 15 /* 0x61: a - f */
|
||||
};
|
||||
|
||||
curl_off_t num = 0;
|
||||
const char *p;
|
||||
int m = (base == 10) ? '9' : /* the largest digit possible */
|
||||
(base == 16) ? 'f' : '7';
|
||||
DEBUGASSERT(linep && *linep && nump);
|
||||
DEBUGASSERT((base == 8) || (base == 10) || (base == 16));
|
||||
DEBUGASSERT(max >= 0); /* mostly to catch SIZE_T_MAX, which is too large */
|
||||
*nump = 0;
|
||||
if(!valid_digit(**linep, base))
|
||||
p = *linep;
|
||||
if(!valid_digit(*p, m))
|
||||
return STRE_NO_NUM;
|
||||
do {
|
||||
int n = num_digit(**linep, base);
|
||||
if(num > ((CURL_OFF_T_MAX - n) / base))
|
||||
return STRE_OVERFLOW;
|
||||
num = num * base + n;
|
||||
if(num > max)
|
||||
return STRE_BIG; /** too big */
|
||||
(*linep)++;
|
||||
} while(valid_digit(**linep, base));
|
||||
if(max < base) {
|
||||
/* special-case low max scenario because check needs to be different */
|
||||
do {
|
||||
int n = hexasciitable[*p++ - '0'] & 0x0f;
|
||||
num = num * base + n;
|
||||
if(num > max)
|
||||
return STRE_OVERFLOW;
|
||||
} while(valid_digit(*p, m));
|
||||
}
|
||||
else {
|
||||
do {
|
||||
int n = hexasciitable[*p++ - '0'] & 0x0f;
|
||||
if(num > ((max - n) / base))
|
||||
return STRE_OVERFLOW;
|
||||
num = num * base + n;
|
||||
} while(valid_digit(*p, m));
|
||||
}
|
||||
*nump = num;
|
||||
*linep = p;
|
||||
return STRE_OK;
|
||||
}
|
||||
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue