#endif
#endif
+#define HALF_UTF8_UPGRADE(start,end) \
+ STMT_START { \
+ if ((start)<(end)) { \
+ U8* NeWsTr; \
+ STRLEN LeN = (end) - (start); \
+ NeWsTr = bytes_to_utf8(start, &LeN); \
+ Safefree(start); \
+ (start) = NeWsTr; \
+ (end) = (start) + LeN; \
+ } \
+ } STMT_END
+
STATIC I32
-S_do_trans_simple(pTHX_ SV *sv) /* SPC - OK */
+S_do_trans_simple(pTHX_ SV *sv)
{
dTHR;
U8 *s;
+ U8 *d;
U8 *send;
+ U8 *dstart;
I32 matches = 0;
- I32 hasutf = SvUTF8(sv);
+ I32 sutf = SvUTF8(sv);
STRLEN len;
short *tbl;
I32 ch;
s = (U8*)SvPV(sv, len);
send = s + len;
- while (s < send) {
- if (hasutf && *s & 0x80)
- s+=UTF8SKIP(s); /* Given that we're here because tbl is !UTF8...*/
- else {
+ /* First, take care of non-UTF8 input strings, because they're easy */
+ if (!sutf) {
+ while (s < send) {
if ((ch = tbl[*s]) >= 0) {
- matches++;
- *s = ch;
+ matches++;
+ *s++ = ch;
}
- s++;
+ else
+ s++;
+ }
+ SvSETMAGIC(sv);
+ return matches;
+ }
+
+ /* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */
+ Newz(0, d, len*2+1, U8);
+ dstart = d;
+ while (s < send) {
+ STRLEN ulen;
+ short c;
+
+ ulen = 1;
+ /* Need to check this, otherwise 128..255 won't match */
+ c = utf8_to_uv(s, send - s, &ulen, 0);
+ if (c < 0x100 && (ch = tbl[(short)c]) >= 0) {
+ matches++;
+ if (ch < 0x80)
+ *d++ = ch;
+ else
+ d = uv_to_utf8(d,ch);
+ s += ulen;
+ }
+ else { /* No match -> copy */
+ while (ulen--)
+ *d++ = *s++;
}
}
+ *d = '\0';
+ sv_setpvn(sv, (const char*)dstart, d - dstart);
+ Safefree(dstart);
+ SvUTF8_on(sv);
SvSETMAGIC(sv);
-
return matches;
}
while (s < send) {
if (hasutf && *s & 0x80)
- s+=UTF8SKIP(s);
+ s += UTF8SKIP(s);
else {
- if (tbl[*s] >= 0)
+ UV c;
+ STRLEN ulen;
+ ulen = 1;
+ if (hasutf)
+ c = utf8_to_uv(s, send - s, &ulen, 0);
+ else
+ c = *s;
+ if (c < 0x100 && tbl[c] >= 0)
matches++;
- s++;
+ s += ulen;
}
}
}
STATIC I32
-S_do_trans_complex(pTHX_ SV *sv)/* SPC - OK */
+S_do_trans_complex(pTHX_ SV *sv)/* SPC - NOT OK */
{
dTHR;
U8 *s;
while (s < send) {
if (hasutf && *s & 0x80)
- s+=UTF8SKIP(s);
+ s += UTF8SKIP(s);
else {
if ((ch = tbl[*s]) >= 0) {
*d = ch;
else
p = d++;
}
- else if (ch == -1) /* -1 is unmapped character */
+ else if (ch == -1) /* -1 is unmapped character */
*d++ = *s; /* -2 is delete character */
s++;
}
else {
while (s < send) {
if (hasutf && *s & 0x80)
- s+=UTF8SKIP(s);
+ s += UTF8SKIP(s);
else {
if ((ch = tbl[*s]) >= 0) {
*d = ch;
matches++;
d++;
}
- else if (ch == -1) /* -1 is unmapped character */
+ else if (ch == -1) /* -1 is unmapped character */
*d++ = *s; /* -2 is delete character */
s++;
}
}
}
- matches += send - d; /* account for disappeared chars */
+ matches += send - d; /* account for disappeared chars */
*d = '\0';
SvCUR_set(sv, d - (U8*)SvPVX(sv));
SvSETMAGIC(sv);
UV extra = none + 1;
UV final;
UV uv;
- I32 isutf;
+ I32 isutf;
I32 howmany;
isutf = SvUTF8(sv);
if ((uv = swash_fetch(rv, s)) < none) {
s += UTF8SKIP(s);
matches++;
- if (uv & 0x80 && !isutf) {
- /* Sneaky-upgrade dstart...d */
- U8* new;
- STRLEN len;
- len = dstart - d;
- new = bytes_to_utf8(dstart, &len);
- Copy(new,dstart,len,U8*);
- d = dstart + len;
- isutf++;
- }
+ if ((uv & 0x80) && !isutf++)
+ HALF_UTF8_UPGRADE(dstart,d);
d = uv_to_utf8(d, uv);
}
else if (uv == none) {
int i;
- i = UTF8SKIP(s);
- if (i > 1 && !isutf) {
- U8* new;
- STRLEN len;
- len = dstart - d;
- new = bytes_to_utf8(dstart, &len);
- Copy(new,dstart,len,U8*);
- d = dstart + len;
- isutf++;
- }
+ i = UTF8SKIP(s);
+ if (i > 1 && !isutf++)
+ HALF_UTF8_UPGRADE(dstart,d);
while(i--)
- *d++ = *s++;
+ *d++ = *s++;
}
else if (uv == extra) {
int i;
- i = UTF8SKIP(s);
+ i = UTF8SKIP(s);
s += i;
matches++;
- if (i > 1 && !isutf) {
- U8* new;
- STRLEN len;
- len = dstart - d;
- new = bytes_to_utf8(dstart, &len);
- Copy(new,dstart,len,U8*);
- d = dstart + len;
- isutf++;
- }
+ if (i > 1 && !isutf++)
+ HALF_UTF8_UPGRADE(dstart,d);
d = uv_to_utf8(d, final);
}
else
s += UTF8SKIP(s);
}
*d = '\0';
- SvPV_set(sv, dstart);
- SvCUR_set(sv, d - dstart);
+ sv_setpvn(sv, (const char*)dstart, d - dstart);
SvSETMAGIC(sv);
if (isutf)
SvUTF8_on(sv);
U8 *d;
I32 matches = 0;
I32 squash = PL_op->op_private & OPpTRANS_SQUASH;
- I32 from_utf = PL_op->op_private & OPpTRANS_FROM_UTF;
- I32 to_utf = PL_op->op_private & OPpTRANS_TO_UTF;
I32 del = PL_op->op_private & OPpTRANS_DELETE;
SV* rv = (SV*)cSVOP->op_sv;
HV* hv = (HV*)SvRV(rv);
UV uv;
STRLEN len;
U8 *dst;
+ I32 isutf = SvUTF8(sv);
s = (U8*)SvPV(sv, len);
send = s + len;
if (svp)
final = SvUV(*svp);
- if (PL_op->op_private & OPpTRANS_GROWS) {
- I32 bits = 16;
-
- svp = hv_fetch(hv, "BITS", 4, FALSE);
- if (svp)
- bits = (I32)SvIV(*svp);
-
- Newz(801, d, len * (bits >> 3) + 1, U8);
+ Newz(0, d, len*2+1, U8);
dst = d;
- }
- else {
- d = s;
- dst = 0;
- }
if (squash) {
UV puv = 0xfeedface;
while (s < send) {
- if (from_utf) {
+ if (SvUTF8(sv))
uv = swash_fetch(rv, s);
- }
else {
U8 tmpbuf[2];
uv = *s++;
}
uv = swash_fetch(rv, tmpbuf);
}
+
if (uv < none) {
matches++;
if (uv != puv) {
- if (uv >= 0x80 && to_utf)
- d = uv_to_utf8(d, uv);
- else
- *d++ = (U8)uv;
+ if ((uv & 0x80) && !isutf++)
+ HALF_UTF8_UPGRADE(dst,d);
+ d = uv_to_utf8(d, uv);
puv = uv;
}
- if (from_utf)
- s += UTF8SKIP(s);
+ s += UTF8SKIP(s);
continue;
}
else if (uv == none) { /* "none" is unmapped character */
- if (from_utf) {
- if (*s < 0x80)
- *d++ = *s++;
- else if (to_utf) {
- int i;
- for (i = UTF8SKIP(s); i; --i)
- *d++ = *s++;
- }
- else {
- I32 ulen;
- *d++ = (U8)utf8_to_uv(s, &ulen);
- s += ulen;
- }
- }
- else { /* must be to_utf only */
- d = uv_to_utf8(d, s[-1]);
- }
+ STRLEN ulen;
+ *d++ = (U8)utf8_to_uv(s, send - s, &ulen, 0);
+ s += ulen;
puv = 0xfeedface;
continue;
}
else if (uv == extra && !del) {
matches++;
if (uv != puv) {
- if (final >= 0x80 && to_utf)
- d = uv_to_utf8(d, final);
- else
- *d++ = (U8)final;
+ d = uv_to_utf8(d, final);
puv = final;
}
- if (from_utf)
- s += UTF8SKIP(s);
+ s += UTF8SKIP(s);
continue;
}
- matches++; /* "none+1" is delete character */
- if (from_utf)
- s += UTF8SKIP(s);
+ matches++; /* "none+1" is delete character */
+ s += UTF8SKIP(s);
}
}
else {
while (s < send) {
- if (from_utf) {
+ if (SvUTF8(sv))
uv = swash_fetch(rv, s);
- }
else {
U8 tmpbuf[2];
uv = *s++;
}
if (uv < none) {
matches++;
- if (uv >= 0x80 && to_utf)
- d = uv_to_utf8(d, uv);
- else
- *d++ = (U8)uv;
- if (from_utf)
- s += UTF8SKIP(s);
+ d = uv_to_utf8(d, uv);
+ s += UTF8SKIP(s);
continue;
}
else if (uv == none) { /* "none" is unmapped character */
- if (from_utf) {
- if (*s < 0x80)
- *d++ = *s++;
- else if (to_utf) {
- int i;
- for (i = UTF8SKIP(s); i; --i)
- *d++ = *s++;
- }
- else {
- I32 ulen;
- *d++ = (U8)utf8_to_uv(s, &ulen);
- s += ulen;
- }
- }
- else { /* must be to_utf only */
- d = uv_to_utf8(d, s[-1]);
- }
+ STRLEN ulen;
+ *d++ = (U8)utf8_to_uv(s, send - s, &ulen, 0);
+ s += ulen;
continue;
}
else if (uv == extra && !del) {
matches++;
- if (final >= 0x80 && to_utf)
- d = uv_to_utf8(d, final);
- else
- *d++ = (U8)final;
- if (from_utf)
- s += UTF8SKIP(s);
+ d = uv_to_utf8(d, final);
+ s += UTF8SKIP(s);
continue;
}
- matches++; /* "none+1" is delete character */
- if (from_utf)
- s += UTF8SKIP(s);
+ matches++; /* "none+1" is delete character */
+ s += UTF8SKIP(s);
}
}
if (dst)
{
dTHR;
STRLEN len;
- I32 hasutf = (PL_op->op_private &
+ I32 hasutf = (PL_op->op_private &
(OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF));
- PL_op->op_private &= ~hasutf;
if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL))
Perl_croak(aTHX_ PL_no_modify);
DEBUG_t( Perl_deb(aTHX_ "2.TBL\n"));
- switch (PL_op->op_private & 63) {
+ switch (PL_op->op_private & ~hasutf & 63) {
case 0:
- if (hasutf)
- return do_trans_simple_utf8(sv);
- else
- return do_trans_simple(sv);
+ if (hasutf)
+ return do_trans_simple_utf8(sv);
+ else
+ return do_trans_simple(sv);
case OPpTRANS_IDENTICAL:
- if (hasutf)
- return do_trans_count_utf8(sv);
- else
- return do_trans_count(sv);
+ if (hasutf)
+ return do_trans_count_utf8(sv);
+ else
+ return do_trans_count(sv);
default:
- if (hasutf)
+ if (hasutf)
return do_trans_complex_utf8(sv);
else
return do_trans_complex(sv);
(void)SvUPGRADE(sv, SVt_PV);
if (SvLEN(sv) < len + items) { /* current length is way too short */
while (items-- > 0) {
- if (*mark && !SvGMAGICAL(*mark) && SvOK(*mark)) {
+ if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) {
SvPV(*mark, tmplen);
len += tmplen;
}
SvTAINTED_on(sv);
}
-/* XXX SvUTF8 support missing! */
+/* currently converts input to bytes if possible, but doesn't sweat failure */
UV
Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size)
{
if (offset < 0)
return retnum;
- if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
+ if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
Perl_croak(aTHX_ "Illegal number of bits in vec");
+
+ if (SvUTF8(sv))
+ (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE);
+
offset *= size; /* turn into bit offset */
len = (offset + size + 7) / 8; /* required number of bytes */
if (len > srclen) {
((UV) s[offset + 4] << 24) +
((UV) s[offset + 5] << 16);
else
- retnum =
+ retnum =
((UV) s[offset ] << 56) +
((UV) s[offset + 1] << 48) +
((UV) s[offset + 2] << 40) +
return retnum;
}
-/* XXX SvUTF8 support missing! */
+/* currently converts input to bytes if possible but doesn't sweat failures,
+ * although it does ensure that the string it clobbers is not marked as
+ * utf8-valid any more
+ */
void
Perl_do_vecset(pTHX_ SV *sv)
{
if (!targ)
return;
s = (unsigned char*)SvPV_force(targ, targlen);
+ if (SvUTF8(targ)) {
+ /* This is handled by the SvPOK_only below...
+ if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE))
+ SvUTF8_off(targ);
+ */
+ (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE);
+ }
+
(void)SvPOK_only(targ);
lval = SvUV(sv);
offset = LvTARGOFF(sv);
+ if (offset < 0)
+ Perl_croak(aTHX_ "Assigning to negative offset in vec");
size = LvTARGLEN(sv);
- if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
+ if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
Perl_croak(aTHX_ "Illegal number of bits in vec");
-
+
offset *= size; /* turn into bit offset */
len = (offset + size + 7) / 8; /* required number of bytes */
if (len > targlen) {
(void)memzero((char *)(s + targlen), len - targlen + 1);
SvCUR_set(targ, len);
}
-
+
if (size < 8) {
mask = (1 << size) - 1;
size = offset & 7;
STRLEN len;
char *s;
dTHR;
-
+
if (SvTYPE(sv) == SVt_PVAV) {
register I32 i;
I32 max;
nope:
SvSETMAGIC(sv);
return count;
-}
+}
void
Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right)
char *rsave;
bool left_utf = DO_UTF8(left);
bool right_utf = DO_UTF8(right);
+ I32 needlen;
if (left_utf && !right_utf)
sv_utf8_upgrade(right);
rsave = rc = SvPV(right, rightlen);
len = leftlen < rightlen ? leftlen : rightlen;
lensave = len;
- if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
+ if ((left_utf || right_utf) && (sv == left || sv == right)) {
+ needlen = optype == OP_BIT_AND ? len : leftlen + rightlen;
+ Newz(801, dc, needlen + 1, char);
+ }
+ else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
STRLEN n_a;
dc = SvPV_force(sv, n_a);
if (SvCUR(sv) < len) {
dc = SvGROW(sv, len + 1);
(void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1);
}
+ if (optype != OP_BIT_AND && (left_utf || right_utf))
+ dc = SvGROW(sv, leftlen + rightlen + 1);
}
else {
- I32 needlen = ((optype == OP_BIT_AND)
- ? len : (leftlen > rightlen ? leftlen : rightlen));
+ needlen = ((optype == OP_BIT_AND)
+ ? len : (leftlen > rightlen ? leftlen : rightlen));
Newz(801, dc, needlen + 1, char);
(void)sv_usepvn(sv, dc, needlen);
dc = SvPVX(sv); /* sv_usepvn() calls Renew() */
(void)SvPOK_only(sv);
if (left_utf || right_utf) {
UV duc, luc, ruc;
+ char *dcsave = dc;
STRLEN lulen = leftlen;
STRLEN rulen = rightlen;
- STRLEN dulen = 0;
- I32 ulen;
-
- if (optype != OP_BIT_AND)
- dc = SvGROW(sv, leftlen+rightlen+1);
+ STRLEN ulen;
switch (optype) {
case OP_BIT_AND:
while (lulen && rulen) {
- luc = utf8_to_uv((U8*)lc, &ulen);
+ luc = utf8_to_uv((U8*)lc, lulen, &ulen, 0);
lc += ulen;
lulen -= ulen;
- ruc = utf8_to_uv((U8*)rc, &ulen);
+ ruc = utf8_to_uv((U8*)rc, rulen, &ulen, 0);
rc += ulen;
rulen -= ulen;
duc = luc & ruc;
dc = (char*)uv_to_utf8((U8*)dc, duc);
}
- dulen = dc - SvPVX(sv);
- SvCUR_set(sv, dulen);
+ if (sv == left || sv == right)
+ (void)sv_usepvn(sv, dcsave, needlen);
+ SvCUR_set(sv, dc - dcsave);
break;
case OP_BIT_XOR:
while (lulen && rulen) {
- luc = utf8_to_uv((U8*)lc, &ulen);
+ luc = utf8_to_uv((U8*)lc, lulen, &ulen, 0);
lc += ulen;
lulen -= ulen;
- ruc = utf8_to_uv((U8*)rc, &ulen);
+ ruc = utf8_to_uv((U8*)rc, rulen, &ulen, 0);
rc += ulen;
rulen -= ulen;
duc = luc ^ ruc;
goto mop_up_utf;
case OP_BIT_OR:
while (lulen && rulen) {
- luc = utf8_to_uv((U8*)lc, &ulen);
+ luc = utf8_to_uv((U8*)lc, lulen, &ulen, 0);
lc += ulen;
lulen -= ulen;
- ruc = utf8_to_uv((U8*)rc, &ulen);
+ ruc = utf8_to_uv((U8*)rc, rulen, &ulen, 0);
rc += ulen;
rulen -= ulen;
duc = luc | ruc;
dc = (char*)uv_to_utf8((U8*)dc, duc);
}
mop_up_utf:
- dulen = dc - SvPVX(sv);
- SvCUR_set(sv, dulen);
+ if (sv == left || sv == right)
+ (void)sv_usepvn(sv, dcsave, needlen);
+ SvCUR_set(sv, dc - dcsave);
if (rulen)
sv_catpvn(sv, rc, rulen);
else if (lulen)
I32 dokeys = (PL_op->op_type == OP_KEYS);
I32 dovalues = (PL_op->op_type == OP_VALUES);
I32 realhv = (SvTYPE(hv) == SVt_PVHV);
-
- if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV)
+
+ if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV)
dokeys = dovalues = TRUE;
if (!hv) {
projects / p5sagit/p5-mst-13.2.git / blobdiff