UV c;
/* Need to check this, otherwise 128..255 won't match */
- c = utf8_to_uv(s, send - s, &ulen, 0);
+ c = utf8n_to_uvchr(s, send - s, &ulen, 0);
if (c < 0x100 && (ch = tbl[c]) >= 0) {
matches++;
- d = uv_to_utf8(d, ch);
+ d = uvchr_to_utf8(d, ch);
s += ulen;
}
else { /* No match -> copy */
I32 matches = 0;
STRLEN len;
short *tbl;
+ I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT;
tbl = (short*)cPVOP->op_pv;
if (!tbl)
while (s < send) {
UV c;
STRLEN ulen;
- c = utf8_to_uv(s, send - s, &ulen, 0);
- if (c < 0x100 && tbl[c] >= 0)
+ c = utf8n_to_uvchr(s, send - s, &ulen, 0);
+ if (c < 0x100) {
+ if (tbl[c] >= 0)
+ matches++;
+ } else if (complement)
matches++;
s += ulen;
}
I32 isutf8;
I32 matches = 0;
I32 grows = PL_op->op_private & OPpTRANS_GROWS;
- STRLEN len;
+ I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT;
+ I32 del = PL_op->op_private & OPpTRANS_DELETE;
+ STRLEN len, rlen;
short *tbl;
I32 ch;
else
d = s;
dstart = d;
+ if (complement && !del)
+ rlen = tbl[0x100];
+
+#ifdef MACOS_TRADITIONAL
+#define comp CoMP /* "comp" is a keyword in some compilers ... */
+#endif
if (PL_op->op_private & OPpTRANS_SQUASH) {
- U8* p = send;
UV pch = 0xfeedface;
while (s < send) {
STRLEN len;
- UV comp = utf8_to_uv_simple(s, &len);
+ UV comp = utf8_to_uvchr(s, &len);
- if (comp > 0xff) { /* always unmapped */
- Copy(s, d, len, U8);
- d += len;
+ if (comp > 0xff) {
+ if (!complement) {
+ Copy(s, d, len, U8);
+ d += len;
+ }
+ else {
+ matches++;
+ if (!del) {
+ ch = (comp - 0x100 < rlen) ?
+ tbl[comp+1] : tbl[0x100+rlen];
+ if (ch != pch) {
+ d = uvchr_to_utf8(d, ch);
+ pch = ch;
+ }
+ s += len;
+ continue;
+ }
+ }
}
else if ((ch = tbl[comp]) >= 0) {
matches++;
if (ch != pch) {
- d = uv_to_utf8(d, ch);
+ d = uvchr_to_utf8(d, ch);
pch = ch;
}
s += len;
else {
while (s < send) {
STRLEN len;
- UV comp = utf8_to_uv_simple(s, &len);
- if (comp > 0xff) { /* always unmapped */
- Copy(s, d, len, U8);
- d += len;
+ UV comp = utf8_to_uvchr(s, &len);
+ if (comp > 0xff) {
+ if (!complement) {
+ Copy(s, d, len, U8);
+ d += len;
+ }
+ else {
+ matches++;
+ if (!del) {
+ if (comp - 0x100 < rlen)
+ d = uvchr_to_utf8(d, tbl[comp+1]);
+ else
+ d = uvchr_to_utf8(d, tbl[0x100+rlen]);
+ }
+ }
}
else if ((ch = tbl[comp]) >= 0) {
- d = uv_to_utf8(d, ch);
+ d = uvchr_to_utf8(d, ch);
matches++;
}
else if (ch == -1) { /* -1 is unmapped character */
if ((uv = swash_fetch(rv, s)) < none) {
s += UTF8SKIP(s);
matches++;
- d = uv_to_utf8(d, uv);
+ d = uvchr_to_utf8(d, uv);
}
else if (uv == none) {
int i = UTF8SKIP(s);
int i = UTF8SKIP(s);
s += i;
matches++;
- d = uv_to_utf8(d, final);
+ d = uvchr_to_utf8(d, final);
}
else
s += UTF8SKIP(s);
}
SvSETMAGIC(sv);
SvUTF8_on(sv);
+ /* Downgrading just 'cos it will is suspect - NI-S */
if (!isutf8 && !(PL_hints & HINT_UTF8))
sv_utf8_downgrade(sv, TRUE);
UV puv = 0xfeedface;
while (s < send) {
uv = swash_fetch(rv, s);
-
+
if (d > dend) {
STRLEN clen = d - dstart;
STRLEN nlen = dend - dstart + len + UTF8_MAXLEN;
}
if (uv < none) {
matches++;
+ s += UTF8SKIP(s);
if (uv != puv) {
- d = uv_to_utf8(d, uv);
+ d = uvchr_to_utf8(d, uv);
puv = uv;
}
- s += UTF8SKIP(s);
continue;
}
else if (uv == none) { /* "none" is unmapped character */
}
else if (uv == extra && !del) {
matches++;
+ s += UTF8SKIP(s);
if (uv != puv) {
- d = uv_to_utf8(d, final);
+ d = uvchr_to_utf8(d, final);
puv = final;
}
- s += UTF8SKIP(s);
continue;
}
matches++; /* "none+1" is delete character */
}
if (uv < none) {
matches++;
- d = uv_to_utf8(d, uv);
s += UTF8SKIP(s);
+ d = uvchr_to_utf8(d, uv);
continue;
}
else if (uv == none) { /* "none" is unmapped character */
}
else if (uv == extra && !del) {
matches++;
- d = uv_to_utf8(d, final);
s += UTF8SKIP(s);
+ d = uvchr_to_utf8(d, final);
continue;
}
matches++; /* "none+1" is delete character */
s = send - 1;
while (s > start && UTF8_IS_CONTINUATION(*s))
s--;
- if (utf8_to_uv_simple((U8*)s, 0)) {
+ if (utf8_to_uvchr((U8*)s, 0)) {
sv_setpvn(astr, s, send - s);
*s = '\0';
SvCUR_set(sv, s - start);
switch (optype) {
case OP_BIT_AND:
while (lulen && rulen) {
- luc = utf8_to_uv((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
+ luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
lc += ulen;
lulen -= ulen;
- ruc = utf8_to_uv((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
+ ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
rc += ulen;
rulen -= ulen;
duc = luc & ruc;
- dc = (char*)uv_to_utf8((U8*)dc, duc);
+ dc = (char*)uvchr_to_utf8((U8*)dc, duc);
}
if (sv == left || sv == right)
(void)sv_usepvn(sv, dcsave, needlen);
break;
case OP_BIT_XOR:
while (lulen && rulen) {
- luc = utf8_to_uv((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
+ luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
lc += ulen;
lulen -= ulen;
- ruc = utf8_to_uv((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
+ ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
rc += ulen;
rulen -= ulen;
duc = luc ^ ruc;
- dc = (char*)uv_to_utf8((U8*)dc, duc);
+ dc = (char*)uvchr_to_utf8((U8*)dc, duc);
}
goto mop_up_utf;
case OP_BIT_OR:
while (lulen && rulen) {
- luc = utf8_to_uv((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
+ luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
lc += ulen;
lulen -= ulen;
- ruc = utf8_to_uv((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
+ ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
rc += ulen;
rulen -= ulen;
duc = luc | ruc;
- dc = (char*)uv_to_utf8((U8*)dc, duc);
+ dc = (char*)uvchr_to_utf8((U8*)dc, duc);
}
mop_up_utf:
if (sv == left || sv == right)
OP *
Perl_do_kv(pTHX)
{
- djSP;
+ dSP;
HV *hv = (HV*)POPs;
HV *keys;
register HE *entry;