just spent waaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaay too long working on eina_str_split/full. never again.
this function was severely broken since 1.0: it did not abide by the limit param, and it was almost comically slow. I have (hopefully) fixed both of these issues. I also fixed the bug where this function was less than 200 lines long. I'll backport this once it gets tested/reviewed/reverted SVN revision: 71871
This commit is contained in:
Mike Blumenkrantz
parent
1d1cd02899
commit
8dbcdefab9
|
|
@ -24,3 +24,4 @@ Guillaume Friloux <guillaume.friloux@asp64.com>
|
|||
Jonas M. Gastal <jgastal@profusion.mobi>
|
||||
Raphael Kubo da Costa <rakuco@freebsd.org>
|
||||
Jérôme Pinot <ngc891@gmail.com>
|
||||
Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
|
||||
|
|
|
|||
|
|
@ -295,4 +295,4 @@
|
|||
|
||||
2012-06-08 Mike Blumenkrantz
|
||||
|
||||
* Prevent negative max_token count for eina_str_split functions
|
||||
* Fixed eina_str_split_full() to behave properly and be much faster
|
||||
|
|
|
|||
|
|
@ -35,7 +35,7 @@
|
|||
|
||||
#include "eina_private.h"
|
||||
#include "eina_str.h"
|
||||
#include "eina_safety_checks.h"
|
||||
|
||||
/*============================================================================*
|
||||
* Local *
|
||||
*============================================================================*/
|
||||
|
|
@ -71,12 +71,38 @@ eina_str_split_full_helper(const char *str,
|
|||
int max_tokens,
|
||||
unsigned int *elements)
|
||||
{
|
||||
char *s, **str_array;
|
||||
char *s, *pos, **str_array;
|
||||
const char *src;
|
||||
size_t len, dlen;
|
||||
unsigned int tokens;
|
||||
unsigned int tokens = 0, x;
|
||||
const char *idx[256] = {NULL};
|
||||
|
||||
EINA_SAFETY_ON_TRUE_RETURN_VAL(max_tokens < 0, NULL);
|
||||
if (max_tokens < 0) max_tokens = 0;
|
||||
if (max_tokens == 1)
|
||||
{
|
||||
str_array = malloc(sizeof(char *) * 2);
|
||||
if (!str_array)
|
||||
{
|
||||
if (elements)
|
||||
*elements = 0;
|
||||
|
||||
return NULL;
|
||||
}
|
||||
|
||||
s = strdup(str);
|
||||
if (!s)
|
||||
{
|
||||
free(str_array);
|
||||
if (elements)
|
||||
*elements = 0;
|
||||
|
||||
return NULL;
|
||||
}
|
||||
if (elements)
|
||||
*elements = 1;
|
||||
str_array[0] = s;
|
||||
return str_array;
|
||||
}
|
||||
dlen = strlen(delim);
|
||||
if (dlen == 0)
|
||||
{
|
||||
|
|
@ -86,7 +112,6 @@ eina_str_split_full_helper(const char *str,
|
|||
return NULL;
|
||||
}
|
||||
|
||||
tokens = 0;
|
||||
src = str;
|
||||
/* count tokens and check strlen(str) */
|
||||
while (*src != '\0')
|
||||
|
|
@ -101,15 +126,18 @@ eina_str_split_full_helper(const char *str,
|
|||
if (EINA_UNLIKELY(d == d_end))
|
||||
{
|
||||
src = tmp;
|
||||
if (tokens < (sizeof(idx) / sizeof(idx[0])))
|
||||
{
|
||||
idx[tokens] = tmp;
|
||||
//printf("token %d='%s'\n", tokens + 1, idx[tokens]);
|
||||
}
|
||||
tokens++;
|
||||
if (tokens && (tokens == (unsigned int)max_tokens)) break;
|
||||
}
|
||||
else
|
||||
src++;
|
||||
}
|
||||
len = src - str;
|
||||
|
||||
if ((max_tokens > 0) && (tokens > (unsigned int)max_tokens))
|
||||
tokens = max_tokens;
|
||||
len = src - str + strlen(src);
|
||||
|
||||
str_array = malloc(sizeof(char *) * (tokens + 2));
|
||||
if (!str_array)
|
||||
|
|
@ -120,6 +148,24 @@ eina_str_split_full_helper(const char *str,
|
|||
return NULL;
|
||||
}
|
||||
|
||||
if (!tokens)
|
||||
{
|
||||
s = strdup(str);
|
||||
if (!s)
|
||||
{
|
||||
free(str_array);
|
||||
if (elements)
|
||||
*elements = 0;
|
||||
|
||||
return NULL;
|
||||
}
|
||||
str_array[0] = s;
|
||||
str_array[1] = NULL;
|
||||
if (elements)
|
||||
*elements = 2;
|
||||
return str_array;
|
||||
}
|
||||
|
||||
s = malloc(len + 1);
|
||||
if (!s)
|
||||
{
|
||||
|
|
@ -130,38 +176,118 @@ eina_str_split_full_helper(const char *str,
|
|||
return NULL;
|
||||
}
|
||||
|
||||
/* copy tokens and string */
|
||||
tokens = 0;
|
||||
str_array[0] = s;
|
||||
src = str;
|
||||
while (*src != '\0')
|
||||
|
||||
if (len == tokens * dlen)
|
||||
{
|
||||
const char *d = delim, *d_end = d + dlen;
|
||||
const char *tmp = src;
|
||||
for (; (d < d_end) && (*tmp != '\0'); d++, tmp++)
|
||||
/* someone's having a laugh somewhere */
|
||||
memset(s, 0, len + 1);
|
||||
for (x = 1; x < tokens + 1; x++)
|
||||
str_array[x] = s + x;
|
||||
str_array[x] = NULL;
|
||||
if (elements)
|
||||
*elements = x + 1;
|
||||
return str_array;
|
||||
}
|
||||
/* copy tokens and string */
|
||||
if (idx[0] - str - dlen > len)
|
||||
{
|
||||
/* FIXME: don't think this can happen but putting this here just in case */
|
||||
abort();
|
||||
}
|
||||
pos = s;
|
||||
for (x = 0; x < MIN(tokens, (sizeof(idx) / sizeof(idx[0]))); x++)
|
||||
{
|
||||
if (x + 1 < (sizeof(idx) / sizeof(idx[0])))
|
||||
{
|
||||
if (EINA_LIKELY(*d != *tmp))
|
||||
break;
|
||||
}
|
||||
if (EINA_UNLIKELY(d == d_end))
|
||||
{
|
||||
src = tmp;
|
||||
*s = '\0';
|
||||
s += dlen;
|
||||
tokens++;
|
||||
str_array[tokens] = s;
|
||||
/* first one is special */
|
||||
if (!x)
|
||||
{
|
||||
eina_strlcpy(pos, str, idx[x] - str - dlen + 1);
|
||||
str_array[x] = pos;
|
||||
//printf("str_array[%d] = '%s'\n", x, str_array[x]);
|
||||
pos += idx[x] - str - dlen + 1;
|
||||
if ((tokens == 1) && (idx[0]))
|
||||
{
|
||||
eina_strlcpy(pos, idx[x], len + 1 - (pos - s));
|
||||
x++, tokens++;
|
||||
str_array[x] = pos;
|
||||
}
|
||||
}
|
||||
/* more tokens */
|
||||
else if (idx[x + 1])
|
||||
{
|
||||
eina_strlcpy(pos, idx[x - 1], idx[x] - idx[x - 1] - dlen + 1);
|
||||
str_array[x] = pos;
|
||||
//printf("str_array[%d] = '%s'\n", x, str_array[x]);
|
||||
pos += idx[x] - idx[x - 1] - dlen + 1;
|
||||
}
|
||||
/* last token */
|
||||
else
|
||||
{
|
||||
if (max_tokens && ((unsigned int)max_tokens < tokens + 1))
|
||||
eina_strlcpy(pos, idx[x - 1], len + 1 - (pos - s));
|
||||
else
|
||||
{
|
||||
//printf("diff: %d\n", len + 1 - (pos - s));
|
||||
eina_strlcpy(pos, idx[x - 1], idx[x] - idx[x - 1] - dlen + 1);
|
||||
str_array[x] = pos;
|
||||
//printf("str_array[%d] = '%s'\n", x, str_array[x]);
|
||||
pos += idx[x] - idx[x - 1] - dlen + 1;
|
||||
x++, tokens++;
|
||||
eina_strlcpy(pos, idx[x - 1], len + 1 - (pos - s));
|
||||
}
|
||||
str_array[x] = pos;
|
||||
//printf("str_array[%d] = '%s'\n", x, str_array[x]);
|
||||
}
|
||||
}
|
||||
/* no more tokens saved after this one */
|
||||
else
|
||||
{
|
||||
*s = *src;
|
||||
s++;
|
||||
src++;
|
||||
eina_strlcpy(pos, idx[x - 1], idx[x] - idx[x - 1] - dlen + 1);
|
||||
str_array[x] = pos;
|
||||
//printf("str_array[%d] = '%s'\n", x, str_array[x]);
|
||||
pos += idx[x] - idx[x - 1] - dlen + 1;
|
||||
src = idx[x];
|
||||
x++, tokens++;
|
||||
str_array[x] = s = pos;
|
||||
break;
|
||||
}
|
||||
}
|
||||
*s = '\0';
|
||||
str_array[tokens + 1] = NULL;
|
||||
if ((x != tokens) && ((!max_tokens) || (x < tokens)))
|
||||
{
|
||||
while (*src != '\0')
|
||||
{
|
||||
const char *d = delim, *d_end = d + dlen;
|
||||
const char *tmp = src;
|
||||
for (; (d < d_end) && (*tmp != '\0'); d++, tmp++)
|
||||
{
|
||||
if (EINA_LIKELY(*d != *tmp))
|
||||
break;
|
||||
}
|
||||
if (((!max_tokens) || (((tokens == (unsigned int)max_tokens) || x < tokens - 2))) && (EINA_UNLIKELY(d == d_end)))
|
||||
{
|
||||
src = tmp;
|
||||
*s = '\0';
|
||||
s++, x++;
|
||||
//printf("str_array[%d] = '%s'\n", x, str_array[x - 1]);
|
||||
str_array[x] = s;
|
||||
}
|
||||
else
|
||||
{
|
||||
*s = *src;
|
||||
s++, src++;
|
||||
}
|
||||
}
|
||||
*s = 0;
|
||||
}
|
||||
str_array[tokens] = NULL;
|
||||
if (elements)
|
||||
*elements = (tokens + 1);
|
||||
{
|
||||
*elements = tokens;
|
||||
if ((!max_tokens) || (tokens == (unsigned int)max_tokens))
|
||||
(*elements)++;
|
||||
}
|
||||
|
||||
return str_array;
|
||||
}
|
||||
|
|
|
|||
Loading…
Reference in New Issue