PERL_ARGS_ASSERT_PRINTBUF;
- GCC_DIAG_IGNORE(-Wformat-nonliteral); /* fmt checked by caller */
+ GCC_DIAG_IGNORE_STMT(-Wformat-nonliteral); /* fmt checked by caller */
PerlIO_printf(Perl_debug_log, fmt, pv_display(tmp, s, strlen(s), 0, 60));
- GCC_DIAG_RESTORE;
+ GCC_DIAG_RESTORE_STMT;
SvREFCNT_dec(tmp);
}
SvCUR(PL_parser->linestr) + len+highhalf);
PL_parser->bufend += len+highhalf;
for (p = pv; p != e; p++) {
- U8 c = (U8)*p;
- if (! UTF8_IS_INVARIANT(c)) {
- *bufptr++ = UTF8_TWO_BYTE_HI(c);
- *bufptr++ = UTF8_TWO_BYTE_LO(c);
- } else {
- *bufptr++ = (char)c;
- }
+ append_utf8_from_native_byte(*p, (U8 **) &bufptr);
}
}
} else {
s = PL_last_uni;
while (isWORDCHAR_lazy_if_safe(s, PL_bufend, UTF) || *s == '-')
s += UTF ? UTF8SKIP(s) : 1;
- if (memchr(s, '(', PL_bufptr - s))
+ if (s < PL_bufptr && memchr(s, '(', PL_bufptr - s))
return;
Perl_ck_warner_d(aTHX_ packWARN(WARN_AMBIGUOUS),
* S_postderef
*
* This subroutine handles postfix deref syntax after the arrow has already
- * been emitted. @* $* etc. are emitted as two separate token right here.
+ * been emitted. @* $* etc. are emitted as two separate tokens right here.
* @[ @{ %[ %{ *{ are emitted also as two tokens, but this function emits
* only the first, leaving yylex to find the next.
*/
S_newSV_maybe_utf8(pTHX_ const char *const start, STRLEN len)
{
SV * const sv = newSVpvn_utf8(start, len,
- !IN_BYTES
- && UTF
- && !is_utf8_invariant_string((const U8*)start, len)
- && is_utf8_string((const U8*)start, len));
+ ! IN_BYTES
+ && UTF
+ && is_utf8_non_invariant_string((const U8*)start, len));
return sv;
}
PL_parser->lex_super_state = PL_lex_state;
PL_parser->lex_sub_inwhat = (U16)op_type;
PL_parser->lex_sub_op = PL_lex_op;
+ PL_parser->sub_no_recover = FALSE;
+ PL_parser->sub_error_count = PL_error_count;
PL_lex_state = LEX_INTERPPUSH;
PL_expect = XTERM;
else {
const line_t l = CopLINE(PL_curcop);
LEAVE;
+ if (PL_parser->sub_error_count != PL_error_count) {
+ const char * const name = OutCopFILE(PL_curcop);
+ if (PL_parser->sub_no_recover) {
+ const char * msg = "";
+ if (PL_in_eval) {
+ SV *errsv = ERRSV;
+ if (SvCUR(ERRSV)) {
+ msg = Perl_form(aTHX_ "%" SVf, SVfARG(errsv));
+ }
+ }
+ abort_execution(msg, name);
+ NOT_REACHED;
+ }
+ }
if (PL_multi_close == '<')
PL_parser->herelines += l - PL_multi_end;
PL_bufend = SvPVX(PL_linestr);
should we have to convert to
UTF-8) */
SV *res; /* result from charnames */
- STRLEN offset_to_max; /* The offset in the output to where the range
- high-end character is temporarily placed */
+ STRLEN offset_to_max = 0; /* The offset in the output to where the range
+ high-end character is temporarily placed */
/* Does something require special handling in tr/// ? This avoids extra
* work in a less likely case. As such, khw didn't feel it was worth
return TRUE;
if (*s != '{' && *s != '[')
return FALSE;
+ PL_parser->sub_no_recover = TRUE;
if (!PL_lex_inpat)
return TRUE;
Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
{
filter_t funcp;
+ I32 ret;
SV *datasv = NULL;
/* This API is bad. It should have been using unsigned int for maxlen.
Not sure if we want to change the API, but if not we should sanity
/* Call function. The function is expected to */
/* call "FILTER_READ(idx+1, buf_sv)" first. */
/* Return: <0:error, =0:eof, >0:not eof */
- return (*funcp)(aTHX_ idx, buf_sv, correct_length);
+ ENTER;
+ save_scalar(PL_errgv);
+ ret = (*funcp)(aTHX_ idx, buf_sv, correct_length);
+ LEAVE;
+ return ret;
}
STATIC char *
/* read var name, including sigil, into PL_tokenbuf */
PL_tokenbuf[0] = sigil;
parse_ident(&s, &dest, dest + sizeof(PL_tokenbuf) - 1,
- 0, cBOOL(UTF), FALSE);
+ 0, cBOOL(UTF), FALSE, FALSE);
*dest = '\0';
assert(PL_tokenbuf[1]); /* we have a variable name */
}
}
if (PL_expect == XSTATE && s[1] == '.' && s[2] == '.') {
s += 3;
- TERM(YADAYADA);
+ OPERATOR(YADAYADA);
}
if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
char tmp = *s++;
int pkgname = 0;
const char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
bool safebw;
+ bool no_op_error = FALSE;
+ if (PL_expect == XOPERATOR) {
+ if (PL_bufptr == PL_linestart) {
+ CopLINE_dec(PL_curcop);
+ Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), "%s", PL_warn_nosemi);
+ CopLINE_inc(PL_curcop);
+ }
+ else
+ /* We want to call no_op with s pointing after the
+ bareword, so defer it. But we want it to come
+ before the Bad name croak. */
+ no_op_error = TRUE;
+ }
/* Get the rest if it looks like a package qualifier */
STRLEN morelen;
s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
TRUE, &morelen);
+ if (no_op_error) {
+ no_op("Bareword",s);
+ no_op_error = FALSE;
+ }
if (!morelen)
Perl_croak(aTHX_ "Bad name after %" UTF8f "%s",
UTF8fARG(UTF, len, PL_tokenbuf),
pkgname = 1;
}
- if (PL_expect == XOPERATOR) {
- if (PL_bufptr == PL_linestart) {
- CopLINE_dec(PL_curcop);
- Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), "%s", PL_warn_nosemi);
- CopLINE_inc(PL_curcop);
- }
- else
+ if (no_op_error)
no_op("Bareword",s);
- }
/* See if the name is "Foo::",
in which case Foo is a bareword
if (!*d && !gv_stashpv(PL_tokenbuf, UTF ? SVf_UTF8 : 0))
{
/* PL_warn_reserved is constant */
- GCC_DIAG_IGNORE(-Wformat-nonliteral);
+ GCC_DIAG_IGNORE_STMT(-Wformat-nonliteral);
Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
PL_tokenbuf);
- GCC_DIAG_RESTORE;
+ GCC_DIAG_RESTORE_STMT;
}
}
}
if (!GvIO(gv))
GvIOp(gv) = newIO();
IoIFP(GvIOp(gv)) = PL_rsfp;
-#if defined(HAS_FCNTL) && defined(F_SETFD) && defined(FD_CLOEXEC)
- {
- const int fd = PerlIO_fileno(PL_rsfp);
- if (fd >= 3) {
- fcntl(fd,F_SETFD, FD_CLOEXEC);
- }
- }
-#endif
/* Mark this internal pseudo-handle as clean */
IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
if ((PerlIO*)PL_rsfp == PerlIO_stdin())
DEBUG_T({ PerlIO_printf(Perl_debug_log,
"### Pending identifier '%s'\n", PL_tokenbuf); });
+ assert(tokenbuf_len >= 2);
/* if we're in a my(), we can't allow dynamics here.
$foo'bar has already been turned into $foo::bar, so
if (has_colon) {
/* "my" variable %s can't be in a package */
/* PL_no_myglob is constant */
- GCC_DIAG_IGNORE(-Wformat-nonliteral);
+ GCC_DIAG_IGNORE_STMT(-Wformat-nonliteral);
yyerror_pv(Perl_form(aTHX_ PL_no_myglob,
PL_in_my == KEY_my ? "my" : "state",
*PL_tokenbuf == '&' ? "subroutin" : "variabl",
PL_tokenbuf),
UTF ? SVf_UTF8 : 0);
- GCC_DIAG_RESTORE;
+ GCC_DIAG_RESTORE_STMT;
}
if (PL_in_my == KEY_sigvar) {
PERL_STATIC_INLINE void
S_parse_ident(pTHX_ char **s, char **d, char * const e, int allow_package,
- bool is_utf8, bool check_dollar)
+ bool is_utf8, bool check_dollar, bool tick_warn)
{
+ int saw_tick = 0;
+ const char *olds = *s;
PERL_ARGS_ASSERT_PARSE_IDENT;
while (*s < PL_bufend) {
*(*d)++ = ':';
*(*d)++ = ':';
(*s)++;
+ saw_tick++;
}
else if (allow_package && **s == ':' && (*s)[1] == ':'
/* Disallow things like Foo::$bar. For the curious, this is
else
break;
}
+ if (UNLIKELY(tick_warn && saw_tick && PL_lex_state == LEX_INTERPNORMAL
+ && !PL_lex_brackets && ckWARN(WARN_SYNTAX))) {
+ char *d;
+ char *d2;
+ Newx(d, *s - olds + saw_tick + 2, char); /* +2 for $# */
+ d2 = d;
+ SAVEFREEPV(d);
+ Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
+ "Old package separator used in string");
+ if (olds[-1] == '#')
+ *d2++ = olds[-2];
+ *d2++ = olds[-1];
+ while (olds < *s) {
+ if (*olds == '\'') {
+ *d2++ = '\\';
+ *d2++ = *olds++;
+ }
+ else
+ *d2++ = *olds++;
+ }
+ Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
+ "\t(Did you mean \"%" UTF8f "\" instead?)\n",
+ UTF8fARG(is_utf8, d2-d, d));
+ }
return;
}
PERL_ARGS_ASSERT_SCAN_WORD;
- parse_ident(&s, &d, e, allow_package, is_utf8, TRUE);
+ parse_ident(&s, &d, e, allow_package, is_utf8, TRUE, FALSE);
*d = '\0';
*slp = d - dest;
return s;
}
}
else { /* See if it is a "normal" identifier */
- parse_ident(&s, &d, e, 1, is_utf8, FALSE);
+ parse_ident(&s, &d, e, 1, is_utf8, FALSE, TRUE);
}
*d = '\0';
d = dest;
(the later check for } being at the expected point will trap
cases where this doesn't pan out.) */
d += is_utf8 ? UTF8SKIP(d) : 1;
- parse_ident(&s, &d, e, 1, is_utf8, TRUE);
+ parse_ident(&s, &d, e, 1, is_utf8, TRUE, TRUE);
*d = '\0';
}
else { /* caret word: ${^Foo} ${^CAPTURE[0]} */
CopLINE_set(PL_curcop, orig_copline);
PL_parser->herelines = herelines;
*dest = '\0';
+ PL_parser->sub_no_recover = TRUE;
}
}
else if ( PL_lex_state == LEX_INTERPNORMAL
/* No whitespace or all! */
if (backup == s || *backup == '\n') {
- Newxz(indent, indent_len + 1, char);
+ Newx(indent, indent_len + 1, char);
memcpy(indent, backup + 1, indent_len);
+ indent[indent_len] = 0;
s--; /* before our delimiter */
PL_parser->herelines--; /* this line doesn't count */
break;
/* All whitespace or none! */
if (backup == found || SPACE_OR_TAB(*backup)) {
- Newxz(indent, indent_len + 1, char);
+ Newx(indent, indent_len + 1, char);
memcpy(indent, backup, indent_len);
+ indent[indent_len] = 0;
SvREFCNT_dec(PL_linestr);
PL_linestr = linestr_save;
PL_linestart = SvPVX(linestr_save);
I32 brackets = 1; /* bracket nesting level */
bool has_utf8 = FALSE; /* is there any utf8 content? */
IV termcode; /* terminating char. code */
- U8 termstr[UTF8_MAXBYTES]; /* terminating string */
+ U8 termstr[UTF8_MAXBYTES+1]; /* terminating string */
STRLEN termlen; /* length of terminating string */
line_t herelines;
digit:
just_zero = FALSE;
if (!overflowed) {
+ assert(shift >= 0);
x = u << shift; /* make room for the digit */
total_bits += shift;
NV nv_mult = 1.0;
#endif
bool accumulate = TRUE;
- for (h++; (isXDIGIT(*h) || *h == '_'); h++) {
+ U8 b;
+ int lim = 1 << shift;
+ for (h++; ((isXDIGIT(*h) && (b = XDIGIT_VALUE(*h)) < lim) ||
+ *h == '_'); h++) {
if (isXDIGIT(*h)) {
- U8 b = XDIGIT_VALUE(*h);
significant_bits += shift;
#ifdef HEXFP_UQUAD
if (accumulate) {
if (significant_bits < NV_MANT_DIG) {
/* We are in the long "run" of xdigits,
* accumulate the full four bits. */
+ assert(shift >= 0);
hexfp_uquad <<= shift;
hexfp_uquad |= b;
hexfp_frac_bits += shift;
- } else {
+ } else if (significant_bits - shift < NV_MANT_DIG) {
/* We are at a hexdigit either at,
* or straddling, the edge of mantissa.
* We will try grabbing as many as
significant_bits - NV_MANT_DIG;
if (tail <= 0)
tail += shift;
+ assert(tail >= 0);
hexfp_uquad <<= tail;
+ assert((shift - tail) >= 0);
hexfp_uquad |= b >> (shift - tail);
hexfp_frac_bits += tail;
}
#else /* HEXFP_NV */
if (accumulate) {
- nv_mult /= 16.0;
+ nv_mult /= nvshift[shift];
if (nv_mult > 0.0)
hexfp_nv += b * nv_mult;
else
floatit = TRUE;
}
if (floatit) {
- STORE_LC_NUMERIC_UNDERLYING_SET_STANDARD();
/* terminate the string */
*d = '\0';
if (UNLIKELY(hexfp)) {
} else {
nv = Atof(PL_tokenbuf);
}
- RESTORE_LC_NUMERIC_UNDERLYING();
sv = newSVnv(nv);
}
return KEYWORD_PLUGIN_DECLINE;
}
+/*
+=for apidoc Amx|void|wrap_keyword_plugin|Perl_keyword_plugin_t new_plugin|Perl_keyword_plugin_t *old_plugin_p
+
+Puts a C function into the chain of keyword plugins. This is the
+preferred way to manipulate the L</PL_keyword_plugin> variable.
+C<new_plugin> is a pointer to the C function that is to be added to the
+keyword plugin chain, and C<old_plugin_p> points to the storage location
+where a pointer to the next function in the chain will be stored. The
+value of C<new_plugin> is written into the L</PL_keyword_plugin> variable,
+while the value previously stored there is written to C<*old_plugin_p>.
+
+L</PL_keyword_plugin> is global to an entire process, and a module wishing
+to hook keyword parsing may find itself invoked more than once per
+process, typically in different threads. To handle that situation, this
+function is idempotent. The location C<*old_plugin_p> must initially
+(once per process) contain a null pointer. A C variable of static
+duration (declared at file scope, typically also marked C<static> to give
+it internal linkage) will be implicitly initialised appropriately, if it
+does not have an explicit initialiser. This function will only actually
+modify the plugin chain if it finds C<*old_plugin_p> to be null. This
+function is also thread safe on the small scale. It uses appropriate
+locking to avoid race conditions in accessing L</PL_keyword_plugin>.
+
+When this function is called, the function referenced by C<new_plugin>
+must be ready to be called, except for C<*old_plugin_p> being unfilled.
+In a threading situation, C<new_plugin> may be called immediately, even
+before this function has returned. C<*old_plugin_p> will always be
+appropriately set before C<new_plugin> is called. If C<new_plugin>
+decides not to do anything special with the identifier that it is given
+(which is the usual case for most calls to a keyword plugin), it must
+chain the plugin function referenced by C<*old_plugin_p>.
+
+Taken all together, XS code to install a keyword plugin should typically
+look something like this:
+
+ static Perl_keyword_plugin_t next_keyword_plugin;
+ static OP *my_keyword_plugin(pTHX_
+ char *keyword_plugin, STRLEN keyword_len, OP **op_ptr)
+ {
+ if (memEQs(keyword_ptr, keyword_len,
+ "my_new_keyword")) {
+ ...
+ } else {
+ return next_keyword_plugin(aTHX_
+ keyword_ptr, keyword_len, op_ptr);
+ }
+ }
+ BOOT:
+ wrap_keyword_plugin(my_keyword_plugin,
+ &next_keyword_plugin);
+
+Direct access to L</PL_keyword_plugin> should be avoided.
+
+=cut
+*/
+
+void
+Perl_wrap_keyword_plugin(pTHX_
+ Perl_keyword_plugin_t new_plugin, Perl_keyword_plugin_t *old_plugin_p)
+{
+ dVAR;
+
+ PERL_UNUSED_CONTEXT;
+ PERL_ARGS_ASSERT_WRAP_KEYWORD_PLUGIN;
+ if (*old_plugin_p) return;
+ KEYWORD_PLUGIN_MUTEX_LOCK;
+ if (!*old_plugin_p) {
+ *old_plugin_p = PL_keyword_plugin;
+ PL_keyword_plugin = new_plugin;
+ }
+ KEYWORD_PLUGIN_MUTEX_UNLOCK;
+}
+
#define parse_recdescent(g,p) S_parse_recdescent(aTHX_ g,p)
static void
S_parse_recdescent(pTHX_ int gramtype, I32 fakeeof)