}
void
-Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg)
+Perl_do_sprintf(pTHX_ SV *sv, SSize_t len, SV **sarg)
{
STRLEN patlen;
const char * const pat = SvPV_const(*sarg, patlen);
bool do_taint = FALSE;
PERL_ARGS_ASSERT_DO_SPRINTF;
+ assert(len >= 1);
if (SvTAINTED(*sarg))
TAINT_PROPER(
SvUTF8_off(sv);
if (DO_UTF8(*sarg))
SvUTF8_on(sv);
- sv_vsetpvfn(sv, pat, patlen, NULL, sarg + 1, len - 1, &do_taint);
+ sv_vsetpvfn(sv, pat, patlen, NULL, sarg + 1, (Size_t)(len - 1), &do_taint);
SvSETMAGIC(sv);
if (do_taint)
SvTAINTED_on(sv);
STRLEN lensave;
const char *lsave;
const char *rsave;
- bool left_utf;
- bool right_utf;
- bool do_warn_above_ff = ckWARN_d(WARN_DEPRECATED);
STRLEN needlen = 0;
+ bool result_needs_to_be_utf8 = FALSE;
+ bool left_utf8 = FALSE;
+ bool right_utf8 = FALSE;
+ U8 * left_non_downgraded = NULL;
+ U8 * right_non_downgraded = NULL;
+ Size_t left_non_downgraded_len = 0;
+ Size_t right_non_downgraded_len = 0;
+ char * non_downgraded = NULL;
+ Size_t non_downgraded_len = 0;
PERL_ARGS_ASSERT_DO_VOP;
if (sv != left || (optype != OP_BIT_AND && !SvOK(sv)))
SvPVCLEAR(sv); /* avoid undef warning on |= and ^= */
if (sv == left) {
- lsave = lc = SvPV_force_nomg(left, leftlen);
+ lc = SvPV_force_nomg(left, leftlen);
}
else {
- lsave = lc = SvPV_nomg_const(left, leftlen);
+ lc = SvPV_nomg_const(left, leftlen);
SvPV_force_nomg_nolen(sv);
}
- rsave = rc = SvPV_nomg_const(right, rightlen);
+ rc = SvPV_nomg_const(right, rightlen);
/* This needs to come after SvPV to ensure that string overloading has
fired off. */
- left_utf = DO_UTF8(left);
- right_utf = DO_UTF8(right);
-
- if (left_utf && !right_utf) {
- /* Avoid triggering overloading again by using temporaries.
- Maybe there should be a variant of sv_utf8_upgrade that takes pvn
- */
- right = newSVpvn_flags(rsave, rightlen, SVs_TEMP);
- sv_utf8_upgrade(right);
- rsave = rc = SvPV_nomg_const(right, rightlen);
- right_utf = TRUE;
+ /* Create downgraded temporaries of any UTF-8 encoded operands */
+ if (DO_UTF8(left)) {
+ const U8 * save_lc = (U8 *) lc;
+
+ left_utf8 = TRUE;
+ result_needs_to_be_utf8 = TRUE;
+
+ left_non_downgraded_len = leftlen;
+ lc = (char *) bytes_from_utf8_loc((const U8 *) lc, &leftlen,
+ &left_utf8,
+ (const U8 **) &left_non_downgraded);
+ /* Calculate the number of trailing unconvertible bytes. This quantity
+ * is the original length minus the length of the converted portion. */
+ left_non_downgraded_len -= left_non_downgraded - save_lc;
+ SAVEFREEPV(lc);
}
- else if (!left_utf && right_utf) {
- left = newSVpvn_flags(lsave, leftlen, SVs_TEMP);
- sv_utf8_upgrade(left);
- lsave = lc = SvPV_nomg_const(left, leftlen);
- left_utf = TRUE;
+ if (DO_UTF8(right)) {
+ const U8 * save_rc = (U8 *) rc;
+
+ right_utf8 = TRUE;
+ result_needs_to_be_utf8 = TRUE;
+
+ right_non_downgraded_len = rightlen;
+ rc = (char *) bytes_from_utf8_loc((const U8 *) rc, &rightlen,
+ &right_utf8,
+ (const U8 **) &right_non_downgraded);
+ right_non_downgraded_len -= right_non_downgraded - save_rc;
+ SAVEFREEPV(rc);
+ }
+
+ /* We set 'len' to the length that the operation actually operates on. The
+ * dangling part of the longer operand doesn't actually participate in the
+ * operation. What happens is that we pretend that the shorter operand has
+ * been extended to the right by enough imaginary zeros to match the length
+ * of the longer one. But we know in advance the result of the operation
+ * on zeros without having to do it. In the case of '&', the result is
+ * zero, and the dangling portion is simply discarded. For '|' and '^', the
+ * result is the same as the other operand, so the dangling part is just
+ * appended to the final result, unchanged. We currently accept above-FF
+ * code points in the dangling portion, as that's how it has long worked,
+ * and code depends on it staying that way. But it is now fatal for
+ * above-FF to appear in the portion that does get operated on. Hence, any
+ * above-FF must come only in the longer operand, and only in its dangling
+ * portion. That means that at least one of the operands has to be
+ * entirely non-UTF-8, and the length of that operand has to be before the
+ * first above-FF in the other */
+ if (left_utf8) {
+ if (right_utf8 || rightlen > leftlen) {
+ Perl_croak(aTHX_ fatal_above_ff_msg, PL_op_desc[optype]);
+ }
+ len = rightlen;
+ }
+ else if (right_utf8) {
+ if (leftlen > rightlen) {
+ Perl_croak(aTHX_ fatal_above_ff_msg, PL_op_desc[optype]);
+ }
+ len = leftlen;
+ }
+ else { /* Neither is UTF-8 */
+ len = leftlen < rightlen ? leftlen : rightlen;
}
- len = leftlen < rightlen ? leftlen : rightlen;
lensave = len;
+ lsave = lc;
+ rsave = rc;
+
SvCUR_set(sv, len);
(void)SvPOK_only(sv);
- if ((left_utf || right_utf) && (sv == left || sv == right)) {
- needlen = optype == OP_BIT_AND ? len : leftlen + rightlen;
- Newxz(dc, needlen + 1, char);
- }
- else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
+ if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
dc = SvPV_force_nomg_nolen(sv);
if (SvLEN(sv) < len + 1) {
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 {
needlen = optype == OP_BIT_AND
sv_usepvn_flags(sv, dc, needlen, SV_HAS_TRAILING_NUL);
dc = SvPVX(sv); /* sv_usepvn() calls Renew() */
}
- if (left_utf || right_utf) {
- char *dcorig = dc;
- char *dcsave = NULL;
- STRLEN lulen = leftlen;
- STRLEN rulen = rightlen;
- switch (optype) {
- case OP_BIT_AND:
- while (lulen && rulen) {
- UV duc, luc, ruc;
- STRLEN ulen;
- luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
- lc += ulen;
- lulen -= ulen;
- ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
- rc += ulen;
- rulen -= ulen;
- duc = luc & ruc;
- dc = (char*)uvchr_to_utf8((U8*)dc, duc);
- if (do_warn_above_ff && (luc > 0xff || ruc > 0xff)) {
- Perl_warner(aTHX_ packWARN(WARN_DEPRECATED),
- deprecated_above_ff_msg, PL_op_desc[optype]);
- /* Warn only once per operation */
- do_warn_above_ff = FALSE;
- }
- }
- if (sv == left || sv == right)
- (void)sv_usepvn(sv, dcorig, needlen);
- SvCUR_set(sv, dc - dcorig);
- *SvEND(sv) = 0;
- break;
- case OP_BIT_XOR:
- while (lulen && rulen) {
- UV duc, luc, ruc;
- STRLEN ulen;
- luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
- lc += ulen;
- lulen -= ulen;
- ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
- rc += ulen;
- rulen -= ulen;
- duc = luc ^ ruc;
- dc = (char*)uvchr_to_utf8((U8*)dc, duc);
- if (do_warn_above_ff && (luc > 0xff || ruc > 0xff)) {
- Perl_warner(aTHX_ packWARN(WARN_DEPRECATED),
- deprecated_above_ff_msg, PL_op_desc[optype]);
- do_warn_above_ff = FALSE;
- }
- }
- goto mop_up_utf;
- case OP_BIT_OR:
- while (lulen && rulen) {
- UV duc, luc, ruc;
- STRLEN ulen;
- luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
- lc += ulen;
- lulen -= ulen;
- ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
- rc += ulen;
- rulen -= ulen;
- duc = luc | ruc;
- dc = (char*)uvchr_to_utf8((U8*)dc, duc);
- if (do_warn_above_ff && (luc > 0xff || ruc > 0xff)) {
- Perl_warner(aTHX_ packWARN(WARN_DEPRECATED),
- deprecated_above_ff_msg, PL_op_desc[optype]);
- do_warn_above_ff = FALSE;
- }
- }
- mop_up_utf:
- if (rulen)
- dcsave = savepvn(rc, rulen);
- else if (lulen)
- dcsave = savepvn(lc, lulen);
- if (sv == left || sv == right)
- (void)sv_usepvn(sv, dcorig, needlen); /* uses Renew(); defaults to nomg */
- SvCUR_set(sv, dc - dcorig);
- if (rulen)
- sv_catpvn_nomg(sv, dcsave, rulen);
- else if (lulen)
- sv_catpvn_nomg(sv, dcsave, lulen);
- else
- *SvEND(sv) = '\0';
- Safefree(dcsave);
- break;
- default:
- if (sv == left || sv == right)
- Safefree(dcorig);
- Perl_croak(aTHX_ "panic: do_vop called for op %u (%s)",
- (unsigned)optype, PL_op_name[optype]);
- }
- SvUTF8_on(sv);
- goto finish;
- }
- else
#ifdef LIBERAL
if (len >= sizeof(long)*4 &&
!((unsigned long)dc % sizeof(long)) &&
len = remainder;
}
#endif
- {
- switch (optype) {
- case OP_BIT_AND:
- while (len--)
- *dc++ = *lc++ & *rc++;
- *dc = '\0';
- break;
- case OP_BIT_XOR:
- while (len--)
- *dc++ = *lc++ ^ *rc++;
- goto mop_up;
- case OP_BIT_OR:
- while (len--)
- *dc++ = *lc++ | *rc++;
- mop_up:
- len = lensave;
- if (rightlen > len) {
- if (dc == rc)
- SvCUR(sv) = rightlen;
- else
- sv_catpvn_nomg(sv, rsave + len, rightlen - len);
- }
- else if (leftlen > len) {
- if (dc == lc)
- SvCUR(sv) = leftlen;
- else
- sv_catpvn_nomg(sv, lsave + len, leftlen - len);
- }
- *SvEND(sv) = '\0';
+ switch (optype) {
+ case OP_BIT_AND:
+ while (len--)
+ *dc++ = *lc++ & *rc++;
+ *dc = '\0';
+ break;
+ case OP_BIT_XOR:
+ while (len--)
+ *dc++ = *lc++ ^ *rc++;
+ goto mop_up;
+ case OP_BIT_OR:
+ while (len--)
+ *dc++ = *lc++ | *rc++;
+ mop_up:
+ len = lensave;
+ if (rightlen > len) {
+ if (dc == rc)
+ SvCUR(sv) = rightlen;
+ else
+ sv_catpvn_nomg(sv, rsave + len, rightlen - len);
+ }
+ else if (leftlen > len) {
+ if (dc == lc)
+ SvCUR(sv) = leftlen;
+ else
+ sv_catpvn_nomg(sv, lsave + len, leftlen - len);
+ }
+ *SvEND(sv) = '\0';
+
+ /* If there is trailing stuff that couldn't be converted from UTF-8, it
+ * is appended as-is for the ^ and | operators. This preserves
+ * backwards compatibility */
+ if (right_non_downgraded) {
+ non_downgraded = (char *) right_non_downgraded;
+ non_downgraded_len = right_non_downgraded_len;
+ }
+ else if (left_non_downgraded) {
+ non_downgraded = (char *) left_non_downgraded;
+ non_downgraded_len = left_non_downgraded_len;
+ }
- break;
- }
+ break;
+ }
+
+ if (result_needs_to_be_utf8) {
+ sv_utf8_upgrade_nomg(sv);
+
+ /* Append any trailing UTF-8 as-is. */
+ if (non_downgraded) {
+ sv_catpvn_nomg(sv, non_downgraded, non_downgraded_len);
+ }
}
- finish:
+
SvTAINT(sv);
}
-/* used for: pp_keys(), pp_values() */
+/* Perl_do_kv() may be:
+ * * called directly as the pp function for pp_keys() and pp_values();
+ * * It may also be called directly when the op is OP_AVHVSWITCH, to
+ * implement CORE::keys(), CORE::values().
+ *
+ * In all cases it expects an HV on the stack and returns a list of keys,
+ * values, or key-value pairs, depending on PL_op.
+ */
OP *
Perl_do_kv(pTHX)
{
dSP;
HV * const keys = MUTABLE_HV(POPs);
- HE *entry;
- SSize_t extend_size;
const U8 gimme = GIMME_V;
- const I32 dokv = (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV);
- /* op_type is OP_RKEYS/OP_RVALUES if pp_rkeys delegated to here */
- const I32 dokeys = dokv || (PL_op->op_type == OP_KEYS)
- || ( PL_op->op_type == OP_AVHVSWITCH
- && (PL_op->op_private & 3) + OP_EACH == OP_KEYS );
- const I32 dovalues = dokv || (PL_op->op_type == OP_VALUES)
- || ( PL_op->op_type == OP_AVHVSWITCH
- && (PL_op->op_private & 3) + OP_EACH == OP_VALUES );
+
+ const I32 dokeys = (PL_op->op_type == OP_KEYS)
+ || ( PL_op->op_type == OP_AVHVSWITCH
+ && (PL_op->op_private & OPpAVHVSWITCH_MASK)
+ + OP_EACH == OP_KEYS);
+
+ const I32 dovalues = (PL_op->op_type == OP_VALUES)
+ || ( PL_op->op_type == OP_AVHVSWITCH
+ && (PL_op->op_private & OPpAVHVSWITCH_MASK)
+ + OP_EACH == OP_VALUES);
+
+ assert( PL_op->op_type == OP_KEYS
+ || PL_op->op_type == OP_VALUES
+ || PL_op->op_type == OP_AVHVSWITCH);
+
+ assert(!( PL_op->op_type == OP_VALUES
+ && (PL_op->op_private & OPpMAYBE_LVSUB)));
(void)hv_iterinit(keys); /* always reset iterator regardless */
IV i;
dTARGET;
+ /* note that in 'scalar(keys %h)' the OP_KEYS is usually
+ * optimised away and the action is performed directly by the
+ * padhv or rv2hv op. We now only get here via OP_AVHVSWITCH
+ * and \&CORE::keys
+ */
if (! SvTIED_mg((const SV *)keys, PERL_MAGIC_tied) ) {
i = HvUSEDKEYS(keys);
}
Perl_croak(aTHX_ "Can't modify keys in list assignment");
}
- /* 2*HvUSEDKEYS() should never be big enough to truncate or wrap */
- assert(HvUSEDKEYS(keys) <= (SSize_t_MAX >> 1));
- extend_size = (SSize_t)HvUSEDKEYS(keys) * (dokeys + dovalues);
- EXTEND(SP, extend_size);
-
- while ((entry = hv_iternext(keys))) {
- if (dokeys) {
- SV* const sv = hv_iterkeysv(entry);
- XPUSHs(sv);
- }
- if (dovalues) {
- SV *const sv = hv_iterval(keys,entry);
- XPUSHs(sv);
- }
- }
- RETURN;
+ PUTBACK;
+ hv_pushkv(keys, (dokeys | (dovalues << 1)));
+ return NORMAL;
}
/*
https://perl5.git.perl.org / perl5.git / blobdiff