/*
* Executes matching of the precompiled pattern on the input string.
* Returns REG_OK or REG_NOMATCH depending on if we find a match or not.
*/
int
tre_match_fast(const fastmatch_t *fg, const void *data, size_t len,
tre_str_type_t type, int nmatch, regmatch_t pmatch[], int eflags)
{
unsigned int shift, u = 0, v = 0;
ssize_t j = 0;
int ret = REG_NOMATCH;
int mismatch;
const char *str_byte = data;
const void *startptr = NULL;
const tre_char_t *str_wide = data;
/* Calculate length if unspecified. */
if (len == (size_t)-1)
switch (type)
{
case STR_WIDE:
len = tre_strlen(str_wide);
break;
default:
len = strlen(str_byte);
break;
}
/* Shortcut for empty pattern */
if (fg->matchall)
{
if (!fg->nosub && nmatch >= 1)
{
pmatch[0].rm_so = 0;
pmatch[0].rm_eo = len;
}
if (fg->bol && fg->eol)
return (len == 0) ? REG_OK : REG_NOMATCH;
else
return REG_OK;
}
/* No point in going farther if we do not have enough data. */
switch (type)
{
case STR_WIDE:
if (len < fg->wlen)
return ret;
shift = fg->wlen;
break;
default:
if (len < fg->len)
return ret;
shift = fg->len;
}
/*
* REG_NOTBOL means not anchoring ^ to the beginning of the line, so we
* can shift one because there can't be a match at the beginning.
*/
if (fg->bol && (eflags & REG_NOTBOL))
j = 1;
/*
* Like above, we cannot have a match at the very end when anchoring to
* the end and REG_NOTEOL is specified.
*/
if (fg->eol && (eflags & REG_NOTEOL))
len--;
if (fg->reversed)
j = len - (type == STR_WIDE ? fg->wlen : fg->len);
/* Only try once at the beginning or ending of the line. */
if ((fg->bol || fg->eol) && !fg->newline && !(eflags & REG_NOTBOL) &&
!(eflags & REG_NOTEOL))
{
/* Simple text comparison. */
if (!((fg->bol && fg->eol) &&
(type == STR_WIDE ? (len != fg->wlen) : (len != fg->len))))
{
/* Determine where in data to start search at. */
j = fg->eol ? len - (type == STR_WIDE ? fg->wlen : fg->len) : 0;
SKIP_CHARS(j);
mismatch = fastcmp(fg, startptr, type);
if (mismatch == REG_OK)
{
if (fg->word && !IS_ON_WORD_BOUNDARY)
return ret;
if (!fg->nosub && nmatch >= 1)
{
pmatch[0].rm_so = j;
pmatch[0].rm_eo = j + (type == STR_WIDE ? fg->wlen : fg->len);
}
return REG_OK;
}
}
}
else
{
/* Quick Search / Turbo Boyer-Moore algorithm. */
do
{
SKIP_CHARS(j);
mismatch = fastcmp(fg, startptr, type);
if (mismatch == REG_OK)
{
if (fg->word)
CHECK_WORD_BOUNDARY;
if (fg->bol)
CHECK_BOL_ANCHOR;
if (fg->eol)
CHECK_EOL_ANCHOR;
if (!fg->nosub && nmatch >= 1)
{
pmatch[0].rm_so = j;
pmatch[0].rm_eo = j + ((type == STR_WIDE) ? fg->wlen : fg->len);
}
return REG_OK;
}
else if (mismatch > 0)
return mismatch;
mismatch = -mismatch - 1;
SHIFT;
} while (!IS_OUT_OF_BOUNDS);
}
return ret;
}
int
tre_fastwcomp(fastmatch_t *preg, const wchar_t *regex, int cflags)
{
return tre_fastwncomp(preg, regex, regex ? tre_strlen(regex) : 0, cflags);
}