static void
EGfree (void *compiled_pattern)
{
struct compiled_regex *cregex = (struct compiled_regex *) compiled_pattern;
dfafree (&cregex->dfa);
free (cregex->patterns);
free (cregex->ckwset.trans);
free (cregex);
}
void
release_regex (struct regex *regex)
{
if (regex->dfa)
{
dfafree (regex->dfa);
free (regex->dfa);
regex->dfa = NULL;
}
regfree (®ex->pattern);
free (regex);
}
void
refree(Regexp *rp)
{
free(rp->pat.buffer);
free(rp->pat.fastmap);
if (rp->regs.start)
free(rp->regs.start);
if (rp->regs.end)
free(rp->regs.end);
if (rp->dfa)
dfafree(&(rp->dfareg));
free(rp);
}
void dispose_fast_regex( struct fast_regex * fre_t )
{
struct _fregex * fre = (struct _fregex *)( fre_t->data ) ;
if( HAS_REGEX( fre_t->options ) )
{
fre->regex.translate = NULL ;
regfree( &fre->regex ) ;
}
if( HAS_DFA( fre_t->options ) )
dfafree( &fre->dfa ) ;
if( fre->kwset )
kwsfree( fre->kwset ) ;
}
void
refree(Regexp *rp)
{
/*
* This isn't malloced, don't let regfree free it.
* (This is strictly necessary only for the old
* version of regex, but it's a good idea to keep it
* here in case regex internals change in the future.)
*/
rp->pat.translate = NULL;
regfree(& rp->pat);
if (rp->regs.start)
free(rp->regs.start);
if (rp->regs.end)
free(rp->regs.end);
if (rp->dfa)
dfafree(&(rp->dfareg));
free(rp);
}
int
match_regex (struct regex *regex, char *buf, size_t buflen,
size_t buf_start_offset, struct re_registers *regarray,
int regsize)
{
int ret;
static struct regex *regex_last;
/* printf ("Matching from %d/%d\n", buf_start_offset, buflen); */
/* Keep track of the last regexp matched. */
if (!regex)
{
regex = regex_last;
if (!regex_last)
bad_prog (_(NO_REGEX));
}
else
regex_last = regex;
/* gnulib's re_search uses signed-int as length */
if (buflen >= INT_MAX)
panic (_("regex input buffer length larger than INT_MAX"));
if (regex->pattern.no_sub && regsize)
{
/* Re-compiling an existing regex, free the previously allocated
structures. */
if (regex->dfa)
{
dfafree (regex->dfa);
free (regex->dfa);
regex->dfa = NULL;
}
regfree (®ex->pattern);
compile_regex_1 (regex, regsize);
}
regex->pattern.regs_allocated = REGS_REALLOCATE;
/* Optimized handling for '^' and '$' patterns */
if (regex->begline || regex->endline)
{
size_t offset;
if (regex->endline)
{
const char *p = NULL;
if (regex->flags & REG_NEWLINE)
p = memchr (buf + buf_start_offset, buffer_delimiter,
buflen - buf_start_offset);
offset = p ? p - buf : buflen;
}
else if (buf_start_offset == 0)
/* begline anchor, starting at beginning of the buffer. */
offset = 0;
else if (!(regex->flags & REG_NEWLINE))
/* begline anchor, starting in the middle of the text buffer,
and multiline regex is not specified - will never match.
Example: seq 2 | sed 'N;s/^/X/g' */
return 0;
else if (buf[buf_start_offset - 1] == buffer_delimiter)
/* begline anchor, starting in the middle of the text buffer,
with multiline match, and the current character
is the line delimiter - start here.
Example: seq 2 | sed 'N;s/^/X/mg' */
offset = buf_start_offset;
else
{
/* begline anchor, starting in the middle of the search buffer,
all previous optimizions didn't work: search
for the next line delimiter character in the buffer,
and start from there if found. */
const char *p = memchr (buf + buf_start_offset, buffer_delimiter,
buflen - buf_start_offset);
if (p == NULL)
return 0;
offset = p - buf + 1;
}
if (regsize)
{
size_t i;
if (!regarray->start)
{
regarray->start = XCALLOC (1, regoff_t);
regarray->end = XCALLOC (1, regoff_t);
regarray->num_regs = 1;
}
regarray->start[0] = offset;
regarray->end[0] = offset;
for (i = 1 ; i < regarray->num_regs; ++i)
regarray->start[i] = regarray->end[i] = -1;
}
return 1;
}
if (buf_start_offset == 0)
{
struct dfa *superset = dfasuperset (regex->dfa);
if (superset && !dfaexec (superset, buf, buf + buflen, true, NULL, NULL))
return 0;
if ((!regsize && (regex->flags & REG_NEWLINE))
|| (!superset && dfaisfast (regex->dfa)))
{
bool backref = false;
if (!dfaexec (regex->dfa, buf, buf + buflen, true, NULL, &backref))
return 0;
if (!regsize && (regex->flags & REG_NEWLINE) && !backref)
return 1;
}
}
/* If the buffer delimiter is not newline character, we cannot use
newline_anchor flag of regex. So do it line-by-line, and add offset
value to results. */
if ((regex->flags & REG_NEWLINE) && buffer_delimiter != '\n')
{
const char *beg, *end;
const char *start;
beg = buf;
if (buf_start_offset > 0)
{
const char *eol = memrchr (buf, buffer_delimiter, buf_start_offset);
if (eol != NULL)
beg = eol + 1;
}
start = buf + buf_start_offset;
for (;;)
{
end = memchr (beg, buffer_delimiter, buf + buflen - beg);
if (end == NULL)
end = buf + buflen;
ret = re_search (®ex->pattern, beg, end - beg,
start - beg, end - start,
regsize ? regarray : NULL);
if (ret > -1)
{
size_t i;
ret += beg - buf;
if (regsize)
{
for (i = 0; i < regarray->num_regs; ++i)
{
if (regarray->start[i] > -1)
regarray->start[i] += beg - buf;
if (regarray->end[i] > -1)
regarray->end[i] += beg - buf;
}
}
break;
}
if (end == buf + buflen)
break;
beg = start = end + 1;
}
}
else
ret = re_search (®ex->pattern, buf, buflen, buf_start_offset,
buflen - buf_start_offset,
regsize ? regarray : NULL);
return (ret > -1);
}
int compile_fast_regex( fast_regex_t * fre_t, const char * sre, size_t len )
{
struct _fregex * fre = (struct _fregex *)( fre_t->data ) ;
int errval ;
char * errstr ;
int sublen ;
char * substr ;
errval = 0 ;
/* fprintf( stderr, "compile_fast_regex( %p[%p] , %s )\n", fre_t, fre->kwset, sre ) ; */
if( fast_regex_subsys_error == NULL )
fast_regex_subsys_error = fprintf_error ;
if( fre_t->options & FRE_STRING_SEARCH )
{
if( fre_t->options & FRE_NO_KWSET )
{
return -1 ;
}
/*
* straight string match
* 'sre' represents a series of newline separated strings to search for
*/
while( sre )
{
substr = strchr( sre, '\n' ) ;
if( substr == NULL )
sublen = len ;
else
sublen = (substr - sre) ;
errstr = kwsincr( fre->kwset, sre, sublen );
if( errstr )
{
errval = (*fast_regex_subsys_error)( "kwset" , errstr ) ;
if( errval )
return errval ;
}
else
fre->num_exact_kws ++ ;
len -= (sublen+1) ;
sre = (substr) ? (substr + 1) : NULL ;
}
errstr = kwsprep( fre->kwset );
if( errstr )
{
errval = (*fast_regex_subsys_error)( "kwset", errstr ) ;
if( errval )
return errval ;
}
return 0 ;
}
if( HAS_DFA(fre_t->options) || HAS_KWSET(fre_t->options) )
{
dfasyntax( RE_SYNTAX_GREP | RE_HAT_LISTS_NOT_NEWLINE, ( fre_t->options & FRE_CASE_INSENSITIVE ) ) ;
dfacomp( sre, len, &(fre->dfa), 1 ) ;
if( HAS_KWSET(fre_t->options) )
{
errval = build_kws_from_dfa( fre ) ;
if( fre_t->options & FRE_NO_DFA )
{ /* We used the DFA only to get the keywords */
dfafree( &(fre->dfa) );
}
}
else
{
fre->kwset = 0 ;
}
}
if( errval )
return errval ;
if( HAS_REGEX(fre_t->options) )
{
re_set_syntax(RE_SYNTAX_GREP | RE_HAT_LISTS_NOT_NEWLINE);
errstr = re_compile_pattern( sre, len, &(fre->regex) ) ;
if( errstr )
{
errval = (*fast_regex_subsys_error)( "re", errstr ) ;
if( errval )
return errval ;
}
}
return errval ;
}
