PCRE_EXP_DEFN int PCRE_CALL_CONVENTION
pcre32_get_stringnumber(const pcre32 *code, PCRE_SPTR32 stringname)
#endif
{
int rc;
int entrysize;
int top, bot;
pcre_uchar *nametable;
#ifdef COMPILE_PCRE8
if ((rc = pcre_fullinfo(code, NULL, PCRE_INFO_NAMECOUNT, &top)) != 0)
return rc;
if (top <= 0) return PCRE_ERROR_NOSUBSTRING;
if ((rc = pcre_fullinfo(code, NULL, PCRE_INFO_NAMEENTRYSIZE, &entrysize)) != 0)
return rc;
if ((rc = pcre_fullinfo(code, NULL, PCRE_INFO_NAMETABLE, &nametable)) != 0)
return rc;
#endif
#ifdef COMPILE_PCRE16
if ((rc = pcre16_fullinfo(code, NULL, PCRE_INFO_NAMECOUNT, &top)) != 0)
return rc;
if (top <= 0) return PCRE_ERROR_NOSUBSTRING;
if ((rc = pcre16_fullinfo(code, NULL, PCRE_INFO_NAMEENTRYSIZE, &entrysize)) != 0)
return rc;
if ((rc = pcre16_fullinfo(code, NULL, PCRE_INFO_NAMETABLE, &nametable)) != 0)
return rc;
#endif
#ifdef COMPILE_PCRE32
if ((rc = pcre32_fullinfo(code, NULL, PCRE_INFO_NAMECOUNT, &top)) != 0)
return rc;
if (top <= 0) return PCRE_ERROR_NOSUBSTRING;
if ((rc = pcre32_fullinfo(code, NULL, PCRE_INFO_NAMEENTRYSIZE, &entrysize)) != 0)
return rc;
if ((rc = pcre32_fullinfo(code, NULL, PCRE_INFO_NAMETABLE, &nametable)) != 0)
return rc;
#endif
bot = 0;
while (top > bot)
{
int mid = (top + bot) / 2;
pcre_uchar *entry = nametable + entrysize*mid;
int c = STRCMP_UC_UC((pcre_uchar *)stringname,
(pcre_uchar *)(entry + IMM2_SIZE));
if (c == 0) return GET2(entry, 0);
if (c > 0) bot = mid + 1; else top = mid;
}
return PCRE_ERROR_NOSUBSTRING;
}
int regcomp(regex_t *preg, MCStringRef pattern, int cflags)
{
const char *errorptr;
int erroffset;
int options = 0;
if ((cflags & REG_ICASE) != 0)
options |= PCRE_CASELESS;
if ((cflags & REG_NEWLINE) != 0)
options |= PCRE_MULTILINE;
// AL-2014-08-20: [[ Bug 13186 ]] Ensure pattern string doesn't get permanently converted to UTF-16
MCAutoStringRef t_temp;
/* UNCHECKED */ MCStringMutableCopy(pattern, &t_temp);
// SN-2014-01-14: [[ libpcre update ]]
preg->re_pcre = pcre16_compile((PCRE_SPTR16)MCStringGetCharPtr(*t_temp), options, &errorptr, &erroffset, NULL);
preg->re_erroffset = erroffset;
if (preg->re_pcre == NULL)
return eint[erroffset];
// [[ libpcre udpate ]] SN-2014-01-10: pcre_info() is deprecated, must be replaced with pcre_fullinfo()
return pcre16_fullinfo((const pcre16 *)preg->re_pcre, NULL, PCRE_INFO_CAPTURECOUNT, &preg->re_nsub);
}
static const unsigned char *tables(int mode)
{
/* The purpose of this function to allow valgrind
for reporting invalid reads and writes. */
static unsigned char *tables_copy;
const char *errorptr;
int erroroffset;
const unsigned char *default_tables;
#ifdef SUPPORT_PCRE8
pcre *regex;
char null_str[1] = { 0 };
#else
pcre16 *regex;
PCRE_UCHAR16 null_str[1] = { 0 };
#endif
if (mode) {
if (tables_copy)
free(tables_copy);
tables_copy = NULL;
return NULL;
}
if (tables_copy)
return tables_copy;
default_tables = NULL;
#ifdef SUPPORT_PCRE8
regex = pcre_compile(null_str, 0, &errorptr, &erroroffset, NULL);
if (regex) {
pcre_fullinfo(regex, NULL, PCRE_INFO_DEFAULT_TABLES, &default_tables);
pcre_free(regex);
}
#else
regex = pcre16_compile(null_str, 0, &errorptr, &erroroffset, NULL);
if (regex) {
pcre16_fullinfo(regex, NULL, PCRE_INFO_DEFAULT_TABLES, &default_tables);
pcre16_free(regex);
}
#endif
/* Shouldn't ever happen. */
if (!default_tables)
return NULL;
/* Unfortunately this value cannot get from pcre_fullinfo.
Since this is a test program, this is acceptable at the moment. */
tables_copy = (unsigned char *)malloc(1088);
if (!tables_copy)
return NULL;
memcpy(tables_copy, default_tables, 1088);
return tables_copy;
}
HL_PRIM ereg *regexp_regexp_new_options( vbyte *str, vbyte *opts ) {
ereg *r;
const char *error;
int err_offset;
int errorcode;
pcre16 *p;
uchar *o = (uchar*)opts;
int options = 0;
while( *o ) {
switch( *o++ ) {
case 'i':
options |= PCRE_CASELESS;
break;
case 's':
options |= PCRE_DOTALL;
break;
case 'm':
options |= PCRE_MULTILINE;
break;
case 'u':
options |= PCRE_UTF8;
break;
case 'g':
options |= PCRE_UNGREEDY;
break;
default:
return NULL;
}
}
p = pcre16_compile2((PCRE_SPTR16)str,options,&errorcode,&error,&err_offset,NULL);
if( p == NULL ) {
hl_buffer *b = hl_alloc_buffer();
hl_buffer_str(b,USTR("Regexp compilation error : "));
hl_buffer_cstr(b,error);
hl_buffer_str(b,USTR(" in "));
hl_buffer_str(b,(uchar*)str);
hl_error_msg(USTR("%s"),hl_buffer_content(b,NULL));
}
r = (ereg*)hl_gc_alloc_finalizer(sizeof(ereg));
r->finalize = regexp_finalize;
r->p = p;
r->nmatches = 0;
r->matched = 0;
pcre16_fullinfo(p,NULL,PCRE_INFO_CAPTURECOUNT,&r->nmatches);
r->nmatches++;
r->matches = (int*)malloc(sizeof(int) * 3 * r->nmatches);
limit.flags = PCRE_EXTRA_MATCH_LIMIT_RECURSION;
limit.match_limit_recursion = 3500; // adapted based on Windows 1MB stack size
return r;
}
void create16(pcre16 *inR, String inExpr, int inFlags)
{
rUtf8 = 0;
rUtf16 = inR;
expr = inExpr;
HX_OBJ_WB_GET(this, expr.raw_ref());
flags = inFlags;
nmatchs = 0;
pcre16_fullinfo(rUtf16,NULL,PCRE_INFO_CAPTURECOUNT,&nmatchs);
nmatchs++;
matchs = (int*)malloc(sizeof(int) * 3 * nmatchs);
_hx_set_finalizer(this, finalize);
}
PCRE_EXP_DEFN int PCRE_CALL_CONVENTION
pcre16_get_stringtable_entries(const pcre16 *code, PCRE_SPTR16 stringname,
PCRE_UCHAR16 **firstptr, PCRE_UCHAR16 **lastptr)
#endif
{
int rc;
int entrysize;
int top, bot;
pcre_uchar *nametable, *lastentry;
#ifdef COMPILE_PCRE8
if ((rc = pcre_fullinfo(code, NULL, PCRE_INFO_NAMECOUNT, &top)) != 0)
return rc;
if (top <= 0) return PCRE_ERROR_NOSUBSTRING;
if ((rc = pcre_fullinfo(code, NULL, PCRE_INFO_NAMEENTRYSIZE, &entrysize)) != 0)
return rc;
if ((rc = pcre_fullinfo(code, NULL, PCRE_INFO_NAMETABLE, &nametable)) != 0)
return rc;
#endif
#ifdef COMPILE_PCRE16
if ((rc = pcre16_fullinfo(code, NULL, PCRE_INFO_NAMECOUNT, &top)) != 0)
return rc;
if (top <= 0) return PCRE_ERROR_NOSUBSTRING;
if ((rc = pcre16_fullinfo(code, NULL, PCRE_INFO_NAMEENTRYSIZE, &entrysize)) != 0)
return rc;
if ((rc = pcre16_fullinfo(code, NULL, PCRE_INFO_NAMETABLE, &nametable)) != 0)
return rc;
#endif
lastentry = nametable + entrysize * (top - 1);
bot = 0;
while (top > bot)
{
int mid = (top + bot) / 2;
pcre_uchar *entry = nametable + entrysize*mid;
int c = STRCMP_UC_UC((pcre_uchar *)stringname,
(pcre_uchar *)(entry + IMM2_SIZE));
if (c == 0)
{
pcre_uchar *first = entry;
pcre_uchar *last = entry;
while (first > nametable)
{
if (STRCMP_UC_UC((pcre_uchar *)stringname,
(pcre_uchar *)(first - entrysize + IMM2_SIZE)) != 0) break;
first -= entrysize;
}
while (last < lastentry)
{
if (STRCMP_UC_UC((pcre_uchar *)stringname,
(pcre_uchar *)(last + entrysize + IMM2_SIZE)) != 0) break;
last += entrysize;
}
#ifdef COMPILE_PCRE8
*firstptr = (char *)first;
*lastptr = (char *)last;
#else
*firstptr = (PCRE_UCHAR16 *)first;
*lastptr = (PCRE_UCHAR16 *)last;
#endif
return entrysize;
}
if (c > 0) bot = mid + 1; else top = mid;
}
return PCRE_ERROR_NOSUBSTRING;
}
ejsval
_ejs_regexp_replace(ejsval str, ejsval search_re, ejsval replace)
{
EJSRegExp* re = (EJSRegExp*)EJSVAL_TO_OBJECT(search_re);
pcre16_extra extra;
memset (&extra, 0, sizeof(extra));
pcre16* code = (pcre16*)re->compiled_pattern;
int capture_count;
pcre16_fullinfo (code, NULL, PCRE_INFO_CAPTURECOUNT, &capture_count);
int ovec_count = 3 * (1 + capture_count);
int* ovec = malloc(sizeof(int) * ovec_count);
int cur_off = 0;
do {
EJSPrimString *flat_str = _ejs_string_flatten (str);
jschar *chars_str = flat_str->data.flat;
int rv = pcre16_exec(code, &extra,
chars_str, flat_str->length, cur_off,
PCRE_NO_UTF16_CHECK, ovec, ovec_count);
if (rv < 0)
break;
ejsval replaceval;
if (EJSVAL_IS_FUNCTION(replace)) {
ejsval substr_match = _ejs_string_new_substring (str, ovec[0], ovec[1] - ovec[0]);
ejsval capture = _ejs_string_new_substring (str, ovec[2], ovec[3] - ovec[2]);
_ejs_log ("substring match is %s\n", ucs2_to_utf8(_ejs_string_flatten(substr_match)->data.flat));
_ejs_log ("capture is %s\n", ucs2_to_utf8(_ejs_string_flatten(capture)->data.flat));
int argc = 3;
ejsval args[3];
args[0] = substr_match;
args[1] = capture;
args[2] = _ejs_undefined;
replaceval = ToString(_ejs_invoke_closure (replace, _ejs_undefined, argc, args));
}
else {
replaceval = ToString(replace);
}
if (ovec[0] == 0) {
// we matched from the beginning of the string, so nothing from there to prepend
str = _ejs_string_concat (replaceval, _ejs_string_new_substring (str, ovec[1], flat_str->length - ovec[1]));
}
else {
str = _ejs_string_concatv (_ejs_string_new_substring (str, 0, ovec[0]),
replaceval,
_ejs_string_new_substring (str, ovec[1], flat_str->length - ovec[1]),
_ejs_null);
}
cur_off = ovec[1];
// if the RegExp object was created without a 'g' flag, only replace the first match
if (!re->global)
break;
} while (EJS_TRUE);
free (ovec);
return str;
}
