3 * Copyright (c) 1991-2000, Larry Wall
5 * You may distribute under the terms of either the GNU General Public
6 * License or the Artistic License, as specified in the README file.
11 * "It all comes from here, the stench and the peril." --Frodo
15 * This file is the lexer for Perl. It's closely linked to the
18 * The main routine is yylex(), which returns the next token.
22 #define PERL_IN_TOKE_C
25 #define yychar PL_yychar
26 #define yylval PL_yylval
28 static char ident_too_long[] = "Identifier too long";
30 static void restore_rsfp(pTHXo_ void *f);
32 #define XFAKEBRACK 128
35 /*#define UTF (SvUTF8(PL_linestr) && !(PL_hints & HINT_BYTE))*/
36 #define UTF (PL_hints & HINT_UTF8)
38 /* In variables name $^X, these are the legal values for X.
39 * 1999-02-27 mjd-perl-patch@plover.com */
40 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
42 /* On MacOS, respect nonbreaking spaces */
43 #ifdef MACOS_TRADITIONAL
44 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
46 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
49 /* LEX_* are values for PL_lex_state, the state of the lexer.
50 * They are arranged oddly so that the guard on the switch statement
51 * can get by with a single comparison (if the compiler is smart enough).
54 /* #define LEX_NOTPARSING 11 is done in perl.h. */
57 #define LEX_INTERPNORMAL 9
58 #define LEX_INTERPCASEMOD 8
59 #define LEX_INTERPPUSH 7
60 #define LEX_INTERPSTART 6
61 #define LEX_INTERPEND 5
62 #define LEX_INTERPENDMAYBE 4
63 #define LEX_INTERPCONCAT 3
64 #define LEX_INTERPCONST 2
65 #define LEX_FORMLINE 1
66 #define LEX_KNOWNEXT 0
68 /* XXX If this causes problems, set i_unistd=undef in the hint file. */
70 # include <unistd.h> /* Needed for execv() */
79 YYSTYPE* yylval_pointer = NULL;
80 int* yychar_pointer = NULL;
83 # define yylval (*yylval_pointer)
84 # define yychar (*yychar_pointer)
85 # define PERL_YYLEX_PARAM yylval_pointer,yychar_pointer
87 # define yylex() Perl_yylex(aTHX_ yylval_pointer, yychar_pointer)
92 /* CLINE is a macro that ensures PL_copline has a sane value */
97 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
100 * Convenience functions to return different tokens and prime the
101 * lexer for the next token. They all take an argument.
103 * TOKEN : generic token (used for '(', DOLSHARP, etc)
104 * OPERATOR : generic operator
105 * AOPERATOR : assignment operator
106 * PREBLOCK : beginning the block after an if, while, foreach, ...
107 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
108 * PREREF : *EXPR where EXPR is not a simple identifier
109 * TERM : expression term
110 * LOOPX : loop exiting command (goto, last, dump, etc)
111 * FTST : file test operator
112 * FUN0 : zero-argument function
113 * FUN1 : not used, except for not, which isn't a UNIOP
114 * BOop : bitwise or or xor
116 * SHop : shift operator
117 * PWop : power operator
118 * PMop : pattern-matching operator
119 * Aop : addition-level operator
120 * Mop : multiplication-level operator
121 * Eop : equality-testing operator
122 * Rop : relational operator <= != gt
124 * Also see LOP and lop() below.
127 #define TOKEN(retval) return (PL_bufptr = s,(int)retval)
128 #define OPERATOR(retval) return (PL_expect = XTERM,PL_bufptr = s,(int)retval)
129 #define AOPERATOR(retval) return ao((PL_expect = XTERM,PL_bufptr = s,(int)retval))
130 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
131 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
132 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s,(int)retval)
133 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR,PL_bufptr = s,(int)retval)
134 #define LOOPX(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
135 #define FTST(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)UNIOP)
136 #define FUN0(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
137 #define FUN1(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
138 #define BOop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
139 #define BAop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
140 #define SHop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
141 #define PWop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
142 #define PMop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
143 #define Aop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
144 #define Mop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
145 #define Eop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
146 #define Rop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)RELOP)
148 /* This bit of chicanery makes a unary function followed by
149 * a parenthesis into a function with one argument, highest precedence.
151 #define UNI(f) return(yylval.ival = f, \
154 PL_last_uni = PL_oldbufptr, \
155 PL_last_lop_op = f, \
156 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
158 #define UNIBRACK(f) return(yylval.ival = f, \
160 PL_last_uni = PL_oldbufptr, \
161 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
163 /* grandfather return to old style */
164 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
169 * This subroutine detects &&= and ||= and turns an ANDAND or OROR
170 * into an OP_ANDASSIGN or OP_ORASSIGN
174 S_ao(pTHX_ int toketype)
176 if (*PL_bufptr == '=') {
178 if (toketype == ANDAND)
179 yylval.ival = OP_ANDASSIGN;
180 else if (toketype == OROR)
181 yylval.ival = OP_ORASSIGN;
189 * When Perl expects an operator and finds something else, no_op
190 * prints the warning. It always prints "<something> found where
191 * operator expected. It prints "Missing semicolon on previous line?"
192 * if the surprise occurs at the start of the line. "do you need to
193 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
194 * where the compiler doesn't know if foo is a method call or a function.
195 * It prints "Missing operator before end of line" if there's nothing
196 * after the missing operator, or "... before <...>" if there is something
197 * after the missing operator.
201 S_no_op(pTHX_ char *what, char *s)
203 char *oldbp = PL_bufptr;
204 bool is_first = (PL_oldbufptr == PL_linestart);
212 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
214 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
215 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
217 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
218 if (t < PL_bufptr && isSPACE(*t))
219 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
220 t - PL_oldoldbufptr, PL_oldoldbufptr);
223 Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
229 * Complain about missing quote/regexp/heredoc terminator.
230 * If it's called with (char *)NULL then it cauterizes the line buffer.
231 * If we're in a delimited string and the delimiter is a control
232 * character, it's reformatted into a two-char sequence like ^C.
237 S_missingterm(pTHX_ char *s)
242 char *nl = strrchr(s,'\n');
248 iscntrl(PL_multi_close)
250 PL_multi_close < 32 || PL_multi_close == 127
254 tmpbuf[1] = toCTRL(PL_multi_close);
260 *tmpbuf = PL_multi_close;
264 q = strchr(s,'"') ? '\'' : '"';
265 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
273 Perl_deprecate(pTHX_ char *s)
276 if (ckWARN(WARN_DEPRECATED))
277 Perl_warner(aTHX_ WARN_DEPRECATED, "Use of %s is deprecated", s);
282 * Deprecate a comma-less variable list.
288 deprecate("comma-less variable list");
292 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
293 * utf16-to-utf8-reversed.
296 #ifdef PERL_CR_FILTER
300 register char *s = SvPVX(sv);
301 register char *e = s + SvCUR(sv);
302 /* outer loop optimized to do nothing if there are no CR-LFs */
304 if (*s++ == '\r' && *s == '\n') {
305 /* hit a CR-LF, need to copy the rest */
306 register char *d = s - 1;
309 if (*s == '\r' && s[1] == '\n')
320 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
322 I32 count = FILTER_READ(idx+1, sv, maxlen);
323 if (count > 0 && !maxlen)
331 S_utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
333 I32 count = FILTER_READ(idx+1, sv, maxlen);
337 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
338 tend = utf16_to_utf8((U16*)SvPVX(sv), tmps, SvCUR(sv));
339 sv_usepvn(sv, (char*)tmps, tend - tmps);
345 S_utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
347 I32 count = FILTER_READ(idx+1, sv, maxlen);
351 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
352 tend = utf16_to_utf8_reversed((U16*)SvPVX(sv), tmps, SvCUR(sv));
353 sv_usepvn(sv, (char*)tmps, tend - tmps);
361 * Initialize variables. Uses the Perl save_stack to save its state (for
362 * recursive calls to the parser).
366 Perl_lex_start(pTHX_ SV *line)
372 SAVEI32(PL_lex_dojoin);
373 SAVEI32(PL_lex_brackets);
374 SAVEI32(PL_lex_casemods);
375 SAVEI32(PL_lex_starts);
376 SAVEI32(PL_lex_state);
377 SAVEVPTR(PL_lex_inpat);
378 SAVEI32(PL_lex_inwhat);
379 if (PL_lex_state == LEX_KNOWNEXT) {
380 I32 toke = PL_nexttoke;
381 while (--toke >= 0) {
382 SAVEI32(PL_nexttype[toke]);
383 SAVEVPTR(PL_nextval[toke]);
385 SAVEI32(PL_nexttoke);
388 SAVECOPLINE(PL_curcop);
391 SAVEPPTR(PL_oldbufptr);
392 SAVEPPTR(PL_oldoldbufptr);
393 SAVEPPTR(PL_linestart);
394 SAVESPTR(PL_linestr);
395 SAVEPPTR(PL_lex_brackstack);
396 SAVEPPTR(PL_lex_casestack);
397 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
398 SAVESPTR(PL_lex_stuff);
399 SAVEI32(PL_lex_defer);
400 SAVEI32(PL_sublex_info.sub_inwhat);
401 SAVESPTR(PL_lex_repl);
403 SAVEINT(PL_lex_expect);
405 PL_lex_state = LEX_NORMAL;
409 New(899, PL_lex_brackstack, 120, char);
410 New(899, PL_lex_casestack, 12, char);
411 SAVEFREEPV(PL_lex_brackstack);
412 SAVEFREEPV(PL_lex_casestack);
414 *PL_lex_casestack = '\0';
417 PL_lex_stuff = Nullsv;
418 PL_lex_repl = Nullsv;
421 PL_sublex_info.sub_inwhat = 0;
423 if (SvREADONLY(PL_linestr))
424 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
425 s = SvPV(PL_linestr, len);
426 if (len && s[len-1] != ';') {
427 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
428 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
429 sv_catpvn(PL_linestr, "\n;", 2);
431 SvTEMP_off(PL_linestr);
432 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
433 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
435 PL_rs = newSVpvn("\n", 1);
441 * Finalizer for lexing operations. Must be called when the parser is
442 * done with the lexer.
448 PL_doextract = FALSE;
453 * This subroutine has nothing to do with tilting, whether at windmills
454 * or pinball tables. Its name is short for "increment line". It
455 * increments the current line number in CopLINE(PL_curcop) and checks
456 * to see whether the line starts with a comment of the form
457 * # line 500 "foo.pm"
458 * If so, it sets the current line number and file to the values in the comment.
462 S_incline(pTHX_ char *s)
470 CopLINE_inc(PL_curcop);
473 while (SPACE_OR_TAB(*s)) s++;
474 if (strnEQ(s, "line", 4))
478 if (*s == ' ' || *s == '\t')
482 while (SPACE_OR_TAB(*s)) s++;
488 while (SPACE_OR_TAB(*s))
490 if (*s == '"' && (t = strchr(s+1, '"'))) {
495 for (t = s; !isSPACE(*t); t++) ;
498 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
500 if (*e != '\n' && *e != '\0')
501 return; /* false alarm */
507 Safefree(CopFILE(PL_curcop));
509 SvREFCNT_dec(CopFILEGV(PL_curcop));
511 CopFILE_set(PL_curcop, s);
514 CopLINE_set(PL_curcop, atoi(n)-1);
519 * Called to gobble the appropriate amount and type of whitespace.
520 * Skips comments as well.
524 S_skipspace(pTHX_ register char *s)
527 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
528 while (s < PL_bufend && SPACE_OR_TAB(*s))
534 SSize_t oldprevlen, oldoldprevlen;
535 SSize_t oldloplen, oldunilen;
536 while (s < PL_bufend && isSPACE(*s)) {
537 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
542 if (s < PL_bufend && *s == '#') {
543 while (s < PL_bufend && *s != '\n')
547 if (PL_in_eval && !PL_rsfp) {
554 /* only continue to recharge the buffer if we're at the end
555 * of the buffer, we're not reading from a source filter, and
556 * we're in normal lexing mode
558 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
559 PL_lex_state == LEX_FORMLINE)
562 /* try to recharge the buffer */
563 if ((s = filter_gets(PL_linestr, PL_rsfp,
564 (prevlen = SvCUR(PL_linestr)))) == Nullch)
566 /* end of file. Add on the -p or -n magic */
567 if (PL_minus_n || PL_minus_p) {
568 sv_setpv(PL_linestr,PL_minus_p ?
569 ";}continue{print or die qq(-p destination: $!\\n)" :
571 sv_catpv(PL_linestr,";}");
572 PL_minus_n = PL_minus_p = 0;
575 sv_setpv(PL_linestr,";");
577 /* reset variables for next time we lex */
578 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
580 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
582 /* Close the filehandle. Could be from -P preprocessor,
583 * STDIN, or a regular file. If we were reading code from
584 * STDIN (because the commandline held no -e or filename)
585 * then we don't close it, we reset it so the code can
586 * read from STDIN too.
589 if (PL_preprocess && !PL_in_eval)
590 (void)PerlProc_pclose(PL_rsfp);
591 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
592 PerlIO_clearerr(PL_rsfp);
594 (void)PerlIO_close(PL_rsfp);
599 /* not at end of file, so we only read another line */
600 /* make corresponding updates to old pointers, for yyerror() */
601 oldprevlen = PL_oldbufptr - PL_bufend;
602 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
604 oldunilen = PL_last_uni - PL_bufend;
606 oldloplen = PL_last_lop - PL_bufend;
607 PL_linestart = PL_bufptr = s + prevlen;
608 PL_bufend = s + SvCUR(PL_linestr);
610 PL_oldbufptr = s + oldprevlen;
611 PL_oldoldbufptr = s + oldoldprevlen;
613 PL_last_uni = s + oldunilen;
615 PL_last_lop = s + oldloplen;
618 /* debugger active and we're not compiling the debugger code,
619 * so store the line into the debugger's array of lines
621 if (PERLDB_LINE && PL_curstash != PL_debstash) {
622 SV *sv = NEWSV(85,0);
624 sv_upgrade(sv, SVt_PVMG);
625 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
626 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
633 * Check the unary operators to ensure there's no ambiguity in how they're
634 * used. An ambiguous piece of code would be:
636 * This doesn't mean rand() + 5. Because rand() is a unary operator,
637 * the +5 is its argument.
647 if (PL_oldoldbufptr != PL_last_uni)
649 while (isSPACE(*PL_last_uni))
651 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
652 if ((t = strchr(s, '(')) && t < PL_bufptr)
654 if (ckWARN_d(WARN_AMBIGUOUS)){
657 Perl_warner(aTHX_ WARN_AMBIGUOUS,
658 "Warning: Use of \"%s\" without parens is ambiguous",
664 /* workaround to replace the UNI() macro with a function. Only the
665 * hints/uts.sh file mentions this. Other comments elsewhere in the
666 * source indicate Microport Unix might need it too.
672 #define UNI(f) return uni(f,s)
675 S_uni(pTHX_ I32 f, char *s)
680 PL_last_uni = PL_oldbufptr;
691 #endif /* CRIPPLED_CC */
694 * LOP : macro to build a list operator. Its behaviour has been replaced
695 * with a subroutine, S_lop() for which LOP is just another name.
698 #define LOP(f,x) return lop(f,x,s)
702 * Build a list operator (or something that might be one). The rules:
703 * - if we have a next token, then it's a list operator [why?]
704 * - if the next thing is an opening paren, then it's a function
705 * - else it's a list operator
709 S_lop(pTHX_ I32 f, int x, char *s)
716 PL_last_lop = PL_oldbufptr;
731 * When the lexer realizes it knows the next token (for instance,
732 * it is reordering tokens for the parser) then it can call S_force_next
733 * to know what token to return the next time the lexer is called. Caller
734 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
735 * handles the token correctly.
739 S_force_next(pTHX_ I32 type)
741 PL_nexttype[PL_nexttoke] = type;
743 if (PL_lex_state != LEX_KNOWNEXT) {
744 PL_lex_defer = PL_lex_state;
745 PL_lex_expect = PL_expect;
746 PL_lex_state = LEX_KNOWNEXT;
752 * When the lexer knows the next thing is a word (for instance, it has
753 * just seen -> and it knows that the next char is a word char, then
754 * it calls S_force_word to stick the next word into the PL_next lookahead.
757 * char *start : buffer position (must be within PL_linestr)
758 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
759 * int check_keyword : if true, Perl checks to make sure the word isn't
760 * a keyword (do this if the word is a label, e.g. goto FOO)
761 * int allow_pack : if true, : characters will also be allowed (require,
763 * int allow_initial_tick : used by the "sub" lexer only.
767 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
772 start = skipspace(start);
774 if (isIDFIRST_lazy_if(s,UTF) ||
775 (allow_pack && *s == ':') ||
776 (allow_initial_tick && *s == '\'') )
778 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
779 if (check_keyword && keyword(PL_tokenbuf, len))
781 if (token == METHOD) {
786 PL_expect = XOPERATOR;
789 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
790 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
798 * Called when the lexer wants $foo *foo &foo etc, but the program
799 * text only contains the "foo" portion. The first argument is a pointer
800 * to the "foo", and the second argument is the type symbol to prefix.
801 * Forces the next token to be a "WORD".
802 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
806 S_force_ident(pTHX_ register char *s, int kind)
809 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
810 PL_nextval[PL_nexttoke].opval = o;
813 dTHR; /* just for in_eval */
814 o->op_private = OPpCONST_ENTERED;
815 /* XXX see note in pp_entereval() for why we forgo typo
816 warnings if the symbol must be introduced in an eval.
818 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
819 kind == '$' ? SVt_PV :
820 kind == '@' ? SVt_PVAV :
821 kind == '%' ? SVt_PVHV :
829 Perl_str_to_version(pTHX_ SV *sv)
834 char *start = SvPVx(sv,len);
835 bool utf = SvUTF8(sv) ? TRUE : FALSE;
836 char *end = start + len;
837 while (start < end) {
841 n = utf8_to_uv((U8*)start, &skip);
846 retval += ((NV)n)/nshift;
855 * Forces the next token to be a version number.
859 S_force_version(pTHX_ char *s)
861 OP *version = Nullop;
870 for (; isDIGIT(*d) || *d == '_' || *d == '.'; d++);
871 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
874 version = yylval.opval;
875 ver = cSVOPx(version)->op_sv;
876 if (SvPOK(ver) && !SvNIOK(ver)) {
877 (void)SvUPGRADE(ver, SVt_PVNV);
878 SvNVX(ver) = str_to_version(ver);
879 SvNOK_on(ver); /* hint that it is a version */
884 /* NOTE: The parser sees the package name and the VERSION swapped */
885 PL_nextval[PL_nexttoke].opval = version;
893 * Tokenize a quoted string passed in as an SV. It finds the next
894 * chunk, up to end of string or a backslash. It may make a new
895 * SV containing that chunk (if HINT_NEW_STRING is on). It also
900 S_tokeq(pTHX_ SV *sv)
911 s = SvPV_force(sv, len);
912 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
915 while (s < send && *s != '\\')
920 if ( PL_hints & HINT_NEW_STRING )
921 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
924 if (s + 1 < send && (s[1] == '\\'))
925 s++; /* all that, just for this */
930 SvCUR_set(sv, d - SvPVX(sv));
932 if ( PL_hints & HINT_NEW_STRING )
933 return new_constant(NULL, 0, "q", sv, pv, "q");
938 * Now come three functions related to double-quote context,
939 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
940 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
941 * interact with PL_lex_state, and create fake ( ... ) argument lists
942 * to handle functions and concatenation.
943 * They assume that whoever calls them will be setting up a fake
944 * join call, because each subthing puts a ',' after it. This lets
947 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
949 * (I'm not sure whether the spurious commas at the end of lcfirst's
950 * arguments and join's arguments are created or not).
955 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
957 * Pattern matching will set PL_lex_op to the pattern-matching op to
958 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
960 * OP_CONST and OP_READLINE are easy--just make the new op and return.
962 * Everything else becomes a FUNC.
964 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
965 * had an OP_CONST or OP_READLINE). This just sets us up for a
966 * call to S_sublex_push().
972 register I32 op_type = yylval.ival;
974 if (op_type == OP_NULL) {
975 yylval.opval = PL_lex_op;
979 if (op_type == OP_CONST || op_type == OP_READLINE) {
980 SV *sv = tokeq(PL_lex_stuff);
982 if (SvTYPE(sv) == SVt_PVIV) {
983 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
989 nsv = newSVpvn(p, len);
995 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
996 PL_lex_stuff = Nullsv;
1000 PL_sublex_info.super_state = PL_lex_state;
1001 PL_sublex_info.sub_inwhat = op_type;
1002 PL_sublex_info.sub_op = PL_lex_op;
1003 PL_lex_state = LEX_INTERPPUSH;
1007 yylval.opval = PL_lex_op;
1017 * Create a new scope to save the lexing state. The scope will be
1018 * ended in S_sublex_done. Returns a '(', starting the function arguments
1019 * to the uc, lc, etc. found before.
1020 * Sets PL_lex_state to LEX_INTERPCONCAT.
1029 PL_lex_state = PL_sublex_info.super_state;
1030 SAVEI32(PL_lex_dojoin);
1031 SAVEI32(PL_lex_brackets);
1032 SAVEI32(PL_lex_casemods);
1033 SAVEI32(PL_lex_starts);
1034 SAVEI32(PL_lex_state);
1035 SAVEVPTR(PL_lex_inpat);
1036 SAVEI32(PL_lex_inwhat);
1037 SAVECOPLINE(PL_curcop);
1038 SAVEPPTR(PL_bufptr);
1039 SAVEPPTR(PL_oldbufptr);
1040 SAVEPPTR(PL_oldoldbufptr);
1041 SAVEPPTR(PL_linestart);
1042 SAVESPTR(PL_linestr);
1043 SAVEPPTR(PL_lex_brackstack);
1044 SAVEPPTR(PL_lex_casestack);
1046 PL_linestr = PL_lex_stuff;
1047 PL_lex_stuff = Nullsv;
1049 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1050 = SvPVX(PL_linestr);
1051 PL_bufend += SvCUR(PL_linestr);
1052 SAVEFREESV(PL_linestr);
1054 PL_lex_dojoin = FALSE;
1055 PL_lex_brackets = 0;
1056 New(899, PL_lex_brackstack, 120, char);
1057 New(899, PL_lex_casestack, 12, char);
1058 SAVEFREEPV(PL_lex_brackstack);
1059 SAVEFREEPV(PL_lex_casestack);
1060 PL_lex_casemods = 0;
1061 *PL_lex_casestack = '\0';
1063 PL_lex_state = LEX_INTERPCONCAT;
1064 CopLINE_set(PL_curcop, PL_multi_start);
1066 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1067 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1068 PL_lex_inpat = PL_sublex_info.sub_op;
1070 PL_lex_inpat = Nullop;
1077 * Restores lexer state after a S_sublex_push.
1083 if (!PL_lex_starts++) {
1084 PL_expect = XOPERATOR;
1085 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn("",0));
1089 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1090 PL_lex_state = LEX_INTERPCASEMOD;
1094 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1095 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1096 PL_linestr = PL_lex_repl;
1098 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1099 PL_bufend += SvCUR(PL_linestr);
1100 SAVEFREESV(PL_linestr);
1101 PL_lex_dojoin = FALSE;
1102 PL_lex_brackets = 0;
1103 PL_lex_casemods = 0;
1104 *PL_lex_casestack = '\0';
1106 if (SvEVALED(PL_lex_repl)) {
1107 PL_lex_state = LEX_INTERPNORMAL;
1109 /* we don't clear PL_lex_repl here, so that we can check later
1110 whether this is an evalled subst; that means we rely on the
1111 logic to ensure sublex_done() is called again only via the
1112 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1115 PL_lex_state = LEX_INTERPCONCAT;
1116 PL_lex_repl = Nullsv;
1122 PL_bufend = SvPVX(PL_linestr);
1123 PL_bufend += SvCUR(PL_linestr);
1124 PL_expect = XOPERATOR;
1125 PL_sublex_info.sub_inwhat = 0;
1133 Extracts a pattern, double-quoted string, or transliteration. This
1136 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1137 processing a pattern (PL_lex_inpat is true), a transliteration
1138 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1140 Returns a pointer to the character scanned up to. Iff this is
1141 advanced from the start pointer supplied (ie if anything was
1142 successfully parsed), will leave an OP for the substring scanned
1143 in yylval. Caller must intuit reason for not parsing further
1144 by looking at the next characters herself.
1148 double-quoted style: \r and \n
1149 regexp special ones: \D \s
1151 backrefs: \1 (deprecated in substitution replacements)
1152 case and quoting: \U \Q \E
1153 stops on @ and $, but not for $ as tail anchor
1155 In transliterations:
1156 characters are VERY literal, except for - not at the start or end
1157 of the string, which indicates a range. scan_const expands the
1158 range to the full set of intermediate characters.
1160 In double-quoted strings:
1162 double-quoted style: \r and \n
1164 backrefs: \1 (deprecated)
1165 case and quoting: \U \Q \E
1168 scan_const does *not* construct ops to handle interpolated strings.
1169 It stops processing as soon as it finds an embedded $ or @ variable
1170 and leaves it to the caller to work out what's going on.
1172 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1174 $ in pattern could be $foo or could be tail anchor. Assumption:
1175 it's a tail anchor if $ is the last thing in the string, or if it's
1176 followed by one of ")| \n\t"
1178 \1 (backreferences) are turned into $1
1180 The structure of the code is
1181 while (there's a character to process) {
1182 handle transliteration ranges
1183 skip regexp comments
1184 skip # initiated comments in //x patterns
1185 check for embedded @foo
1186 check for embedded scalars
1188 leave intact backslashes from leave (below)
1189 deprecate \1 in strings and sub replacements
1190 handle string-changing backslashes \l \U \Q \E, etc.
1191 switch (what was escaped) {
1192 handle - in a transliteration (becomes a literal -)
1193 handle \132 octal characters
1194 handle 0x15 hex characters
1195 handle \cV (control V)
1196 handle printf backslashes (\f, \r, \n, etc)
1198 } (end if backslash)
1199 } (end while character to read)
1204 S_scan_const(pTHX_ char *start)
1206 register char *send = PL_bufend; /* end of the constant */
1207 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1208 register char *s = start; /* start of the constant */
1209 register char *d = SvPVX(sv); /* destination for copies */
1210 bool dorange = FALSE; /* are we in a translit range? */
1211 bool has_utf = FALSE; /* embedded \x{} */
1215 I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1216 ? (PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1218 I32 thisutf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1219 ? (PL_sublex_info.sub_op->op_private & (PL_lex_repl ?
1220 OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF))
1222 const char *leaveit = /* set of acceptably-backslashed characters */
1224 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1227 while (s < send || dorange) {
1228 /* get transliterations out of the way (they're most literal) */
1229 if (PL_lex_inwhat == OP_TRANS) {
1230 /* expand a range A-Z to the full set of characters. AIE! */
1232 I32 i; /* current expanded character */
1233 I32 min; /* first character in range */
1234 I32 max; /* last character in range */
1236 i = d - SvPVX(sv); /* remember current offset */
1237 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1238 d = SvPVX(sv) + i; /* refresh d after realloc */
1239 d -= 2; /* eat the first char and the - */
1241 min = (U8)*d; /* first char in range */
1242 max = (U8)d[1]; /* last char in range */
1245 if ((isLOWER(min) && isLOWER(max)) ||
1246 (isUPPER(min) && isUPPER(max))) {
1248 for (i = min; i <= max; i++)
1252 for (i = min; i <= max; i++)
1259 for (i = min; i <= max; i++)
1262 /* mark the range as done, and continue */
1267 /* range begins (ignore - as first or last char) */
1268 else if (*s == '-' && s+1 < send && s != start) {
1270 *d++ = (char)0xff; /* use illegal utf8 byte--see pmtrans */
1279 /* if we get here, we're not doing a transliteration */
1281 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1282 except for the last char, which will be done separately. */
1283 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1285 while (s < send && *s != ')')
1288 else if (s[2] == '{' /* This should match regcomp.c */
1289 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1292 char *regparse = s + (s[2] == '{' ? 3 : 4);
1295 while (count && (c = *regparse)) {
1296 if (c == '\\' && regparse[1])
1304 if (*regparse != ')') {
1305 regparse--; /* Leave one char for continuation. */
1306 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1308 while (s < regparse)
1313 /* likewise skip #-initiated comments in //x patterns */
1314 else if (*s == '#' && PL_lex_inpat &&
1315 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1316 while (s+1 < send && *s != '\n')
1320 /* check for embedded arrays (@foo, @:foo, @'foo, @{foo}, @$foo) */
1321 else if (*s == '@' && s[1]
1322 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$", s[1])))
1325 /* check for embedded scalars. only stop if we're sure it's a
1328 else if (*s == '$') {
1329 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1331 if (s + 1 < send && !strchr("()| \n\t", s[1]))
1332 break; /* in regexp, $ might be tail anchor */
1335 /* (now in tr/// code again) */
1337 if (*s & 0x80 && thisutf) {
1338 (void)utf8_to_uv((U8*)s, &len);
1340 /* illegal UTF8, make it valid */
1341 char *old_pvx = SvPVX(sv);
1342 /* need space for one extra char (NOTE: SvCUR() not set here) */
1343 d = SvGROW(sv, SvLEN(sv) + 1) + (d - old_pvx);
1344 d = (char*)uv_to_utf8((U8*)d, (U8)*s++);
1355 if (*s == '\\' && s+1 < send) {
1358 /* some backslashes we leave behind */
1359 if (*leaveit && *s && strchr(leaveit, *s)) {
1365 /* deprecate \1 in strings and substitution replacements */
1366 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1367 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1369 dTHR; /* only for ckWARN */
1370 if (ckWARN(WARN_SYNTAX))
1371 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1376 /* string-change backslash escapes */
1377 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1382 /* if we get here, it's either a quoted -, or a digit */
1385 /* quoted - in transliterations */
1387 if (PL_lex_inwhat == OP_TRANS) {
1395 if (ckWARN(WARN_MISC) && isALNUM(*s))
1396 Perl_warner(aTHX_ WARN_MISC,
1397 "Unrecognized escape \\%c passed through",
1399 /* default action is to copy the quoted character */
1404 /* \132 indicates an octal constant */
1405 case '0': case '1': case '2': case '3':
1406 case '4': case '5': case '6': case '7':
1407 len = 0; /* disallow underscores */
1408 uv = (UV)scan_oct(s, 3, &len);
1410 goto NUM_ESCAPE_INSERT;
1412 /* \x24 indicates a hex constant */
1416 char* e = strchr(s, '}');
1418 yyerror("Missing right brace on \\x{}");
1421 len = 1; /* allow underscores */
1422 uv = (UV)scan_hex(s + 1, e - s - 1, &len);
1426 len = 0; /* disallow underscores */
1427 uv = (UV)scan_hex(s, 2, &len);
1432 /* Insert oct or hex escaped character.
1433 * There will always enough room in sv since such escapes will
1434 * be longer than any utf8 sequence they can end up as
1437 if (!thisutf && !has_utf && uv > 255) {
1438 /* might need to recode whatever we have accumulated so far
1439 * if it contains any hibit chars
1443 for (c = SvPVX(sv); c < d; c++) {
1448 char *old_pvx = SvPVX(sv);
1450 d = SvGROW(sv, SvCUR(sv) + hicount + 1) + (d - old_pvx);
1459 uv_to_utf8((U8*)dst, (U8)*src--);
1469 if (thisutf || uv > 255) {
1470 d = (char*)uv_to_utf8((U8*)d, uv);
1482 /* \N{latin small letter a} is a named character */
1486 char* e = strchr(s, '}');
1492 yyerror("Missing right brace on \\N{}");
1496 res = newSVpvn(s + 1, e - s - 1);
1497 res = new_constant( Nullch, 0, "charnames",
1498 res, Nullsv, "\\N{...}" );
1499 str = SvPV(res,len);
1500 if (!has_utf && SvUTF8(res)) {
1501 char *ostart = SvPVX(sv);
1502 SvCUR_set(sv, d - ostart);
1504 sv_utf8_upgrade(sv);
1505 d = SvPVX(sv) + SvCUR(sv);
1508 if (len > e - s + 4) {
1509 char *odest = SvPVX(sv);
1511 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1512 d = SvPVX(sv) + (d - odest);
1514 Copy(str, d, len, char);
1521 yyerror("Missing braces on \\N{}");
1524 /* \c is a control character */
1539 /* printf-style backslashes, formfeeds, newlines, etc */
1557 *d++ = '\047'; /* CP 1047 */
1560 *d++ = '\057'; /* CP 1047 */
1574 } /* end if (backslash) */
1577 } /* while loop to process each character */
1579 /* terminate the string and set up the sv */
1581 SvCUR_set(sv, d - SvPVX(sv));
1586 /* shrink the sv if we allocated more than we used */
1587 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1588 SvLEN_set(sv, SvCUR(sv) + 1);
1589 Renew(SvPVX(sv), SvLEN(sv), char);
1592 /* return the substring (via yylval) only if we parsed anything */
1593 if (s > PL_bufptr) {
1594 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1595 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1597 ( PL_lex_inwhat == OP_TRANS
1599 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1602 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1609 * Returns TRUE if there's more to the expression (e.g., a subscript),
1612 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1614 * ->[ and ->{ return TRUE
1615 * { and [ outside a pattern are always subscripts, so return TRUE
1616 * if we're outside a pattern and it's not { or [, then return FALSE
1617 * if we're in a pattern and the first char is a {
1618 * {4,5} (any digits around the comma) returns FALSE
1619 * if we're in a pattern and the first char is a [
1621 * [SOMETHING] has a funky algorithm to decide whether it's a
1622 * character class or not. It has to deal with things like
1623 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1624 * anything else returns TRUE
1627 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1630 S_intuit_more(pTHX_ register char *s)
1632 if (PL_lex_brackets)
1634 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1636 if (*s != '{' && *s != '[')
1641 /* In a pattern, so maybe we have {n,m}. */
1658 /* On the other hand, maybe we have a character class */
1661 if (*s == ']' || *s == '^')
1664 /* this is terrifying, and it works */
1665 int weight = 2; /* let's weigh the evidence */
1667 unsigned char un_char = 255, last_un_char;
1668 char *send = strchr(s,']');
1669 char tmpbuf[sizeof PL_tokenbuf * 4];
1671 if (!send) /* has to be an expression */
1674 Zero(seen,256,char);
1677 else if (isDIGIT(*s)) {
1679 if (isDIGIT(s[1]) && s[2] == ']')
1685 for (; s < send; s++) {
1686 last_un_char = un_char;
1687 un_char = (unsigned char)*s;
1692 weight -= seen[un_char] * 10;
1693 if (isALNUM_lazy_if(s+1,UTF)) {
1694 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1695 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1700 else if (*s == '$' && s[1] &&
1701 strchr("[#!%*<>()-=",s[1])) {
1702 if (/*{*/ strchr("])} =",s[2]))
1711 if (strchr("wds]",s[1]))
1713 else if (seen['\''] || seen['"'])
1715 else if (strchr("rnftbxcav",s[1]))
1717 else if (isDIGIT(s[1])) {
1719 while (s[1] && isDIGIT(s[1]))
1729 if (strchr("aA01! ",last_un_char))
1731 if (strchr("zZ79~",s[1]))
1733 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1734 weight -= 5; /* cope with negative subscript */
1737 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1738 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1743 if (keyword(tmpbuf, d - tmpbuf))
1746 if (un_char == last_un_char + 1)
1748 weight -= seen[un_char];
1753 if (weight >= 0) /* probably a character class */
1763 * Does all the checking to disambiguate
1765 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1766 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1768 * First argument is the stuff after the first token, e.g. "bar".
1770 * Not a method if bar is a filehandle.
1771 * Not a method if foo is a subroutine prototyped to take a filehandle.
1772 * Not a method if it's really "Foo $bar"
1773 * Method if it's "foo $bar"
1774 * Not a method if it's really "print foo $bar"
1775 * Method if it's really "foo package::" (interpreted as package->foo)
1776 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1777 * Not a method if bar is a filehandle or package, but is quoted with
1782 S_intuit_method(pTHX_ char *start, GV *gv)
1784 char *s = start + (*start == '$');
1785 char tmpbuf[sizeof PL_tokenbuf];
1793 if ((cv = GvCVu(gv))) {
1794 char *proto = SvPVX(cv);
1804 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1805 /* start is the beginning of the possible filehandle/object,
1806 * and s is the end of it
1807 * tmpbuf is a copy of it
1810 if (*start == '$') {
1811 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1816 return *s == '(' ? FUNCMETH : METHOD;
1818 if (!keyword(tmpbuf, len)) {
1819 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1824 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1825 if (indirgv && GvCVu(indirgv))
1827 /* filehandle or package name makes it a method */
1828 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1830 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1831 return 0; /* no assumptions -- "=>" quotes bearword */
1833 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1834 newSVpvn(tmpbuf,len));
1835 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1839 return *s == '(' ? FUNCMETH : METHOD;
1847 * Return a string of Perl code to load the debugger. If PERL5DB
1848 * is set, it will return the contents of that, otherwise a
1849 * compile-time require of perl5db.pl.
1856 char *pdb = PerlEnv_getenv("PERL5DB");
1860 SETERRNO(0,SS$_NORMAL);
1861 return "BEGIN { require 'perl5db.pl' }";
1867 /* Encoded script support. filter_add() effectively inserts a
1868 * 'pre-processing' function into the current source input stream.
1869 * Note that the filter function only applies to the current source file
1870 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1872 * The datasv parameter (which may be NULL) can be used to pass
1873 * private data to this instance of the filter. The filter function
1874 * can recover the SV using the FILTER_DATA macro and use it to
1875 * store private buffers and state information.
1877 * The supplied datasv parameter is upgraded to a PVIO type
1878 * and the IoDIRP field is used to store the function pointer,
1879 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1880 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1881 * private use must be set using malloc'd pointers.
1885 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1890 if (!PL_rsfp_filters)
1891 PL_rsfp_filters = newAV();
1893 datasv = NEWSV(255,0);
1894 if (!SvUPGRADE(datasv, SVt_PVIO))
1895 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1896 IoDIRP(datasv) = (DIR*)funcp; /* stash funcp into spare field */
1897 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1898 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1899 funcp, SvPV_nolen(datasv)));
1900 av_unshift(PL_rsfp_filters, 1);
1901 av_store(PL_rsfp_filters, 0, datasv) ;
1906 /* Delete most recently added instance of this filter function. */
1908 Perl_filter_del(pTHX_ filter_t funcp)
1911 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1912 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1914 /* if filter is on top of stack (usual case) just pop it off */
1915 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1916 if (IoDIRP(datasv) == (DIR*)funcp) {
1917 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1918 IoDIRP(datasv) = (DIR*)NULL;
1919 sv_free(av_pop(PL_rsfp_filters));
1923 /* we need to search for the correct entry and clear it */
1924 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1928 /* Invoke the n'th filter function for the current rsfp. */
1930 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1933 /* 0 = read one text line */
1938 if (!PL_rsfp_filters)
1940 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1941 /* Provide a default input filter to make life easy. */
1942 /* Note that we append to the line. This is handy. */
1943 DEBUG_P(PerlIO_printf(Perl_debug_log,
1944 "filter_read %d: from rsfp\n", idx));
1948 int old_len = SvCUR(buf_sv) ;
1950 /* ensure buf_sv is large enough */
1951 SvGROW(buf_sv, old_len + maxlen) ;
1952 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1953 if (PerlIO_error(PL_rsfp))
1954 return -1; /* error */
1956 return 0 ; /* end of file */
1958 SvCUR_set(buf_sv, old_len + len) ;
1961 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1962 if (PerlIO_error(PL_rsfp))
1963 return -1; /* error */
1965 return 0 ; /* end of file */
1968 return SvCUR(buf_sv);
1970 /* Skip this filter slot if filter has been deleted */
1971 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1972 DEBUG_P(PerlIO_printf(Perl_debug_log,
1973 "filter_read %d: skipped (filter deleted)\n",
1975 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1977 /* Get function pointer hidden within datasv */
1978 funcp = (filter_t)IoDIRP(datasv);
1979 DEBUG_P(PerlIO_printf(Perl_debug_log,
1980 "filter_read %d: via function %p (%s)\n",
1981 idx, funcp, SvPV_nolen(datasv)));
1982 /* Call function. The function is expected to */
1983 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1984 /* Return: <0:error, =0:eof, >0:not eof */
1985 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1989 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1991 #ifdef PERL_CR_FILTER
1992 if (!PL_rsfp_filters) {
1993 filter_add(S_cr_textfilter,NULL);
1996 if (PL_rsfp_filters) {
1999 SvCUR_set(sv, 0); /* start with empty line */
2000 if (FILTER_READ(0, sv, 0) > 0)
2001 return ( SvPVX(sv) ) ;
2006 return (sv_gets(sv, fp, append));
2011 static char* exp_name[] =
2012 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2013 "ATTRTERM", "TERMBLOCK"
2020 Works out what to call the token just pulled out of the input
2021 stream. The yacc parser takes care of taking the ops we return and
2022 stitching them into a tree.
2028 if read an identifier
2029 if we're in a my declaration
2030 croak if they tried to say my($foo::bar)
2031 build the ops for a my() declaration
2032 if it's an access to a my() variable
2033 are we in a sort block?
2034 croak if my($a); $a <=> $b
2035 build ops for access to a my() variable
2036 if in a dq string, and they've said @foo and we can't find @foo
2038 build ops for a bareword
2039 if we already built the token before, use it.
2043 #pragma segment Perl_yylex
2046 #ifdef USE_PURE_BISON
2047 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
2060 #ifdef USE_PURE_BISON
2061 yylval_pointer = lvalp;
2062 yychar_pointer = lcharp;
2065 /* check if there's an identifier for us to look at */
2066 if (PL_pending_ident) {
2067 /* pit holds the identifier we read and pending_ident is reset */
2068 char pit = PL_pending_ident;
2069 PL_pending_ident = 0;
2071 /* if we're in a my(), we can't allow dynamics here.
2072 $foo'bar has already been turned into $foo::bar, so
2073 just check for colons.
2075 if it's a legal name, the OP is a PADANY.
2078 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
2079 if (strchr(PL_tokenbuf,':'))
2080 yyerror(Perl_form(aTHX_ "No package name allowed for "
2081 "variable %s in \"our\"",
2083 tmp = pad_allocmy(PL_tokenbuf);
2086 if (strchr(PL_tokenbuf,':'))
2087 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
2089 yylval.opval = newOP(OP_PADANY, 0);
2090 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
2096 build the ops for accesses to a my() variable.
2098 Deny my($a) or my($b) in a sort block, *if* $a or $b is
2099 then used in a comparison. This catches most, but not
2100 all cases. For instance, it catches
2101 sort { my($a); $a <=> $b }
2103 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2104 (although why you'd do that is anyone's guess).
2107 if (!strchr(PL_tokenbuf,':')) {
2109 /* Check for single character per-thread SVs */
2110 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2111 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2112 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2114 yylval.opval = newOP(OP_THREADSV, 0);
2115 yylval.opval->op_targ = tmp;
2118 #endif /* USE_THREADS */
2119 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2120 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2121 /* might be an "our" variable" */
2122 if (SvFLAGS(namesv) & SVpad_OUR) {
2123 /* build ops for a bareword */
2124 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2125 sv_catpvn(sym, "::", 2);
2126 sv_catpv(sym, PL_tokenbuf+1);
2127 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2128 yylval.opval->op_private = OPpCONST_ENTERED;
2129 gv_fetchpv(SvPVX(sym),
2131 ? (GV_ADDMULTI | GV_ADDINEVAL)
2134 ((PL_tokenbuf[0] == '$') ? SVt_PV
2135 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2140 /* if it's a sort block and they're naming $a or $b */
2141 if (PL_last_lop_op == OP_SORT &&
2142 PL_tokenbuf[0] == '$' &&
2143 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2146 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2147 d < PL_bufend && *d != '\n';
2150 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2151 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2157 yylval.opval = newOP(OP_PADANY, 0);
2158 yylval.opval->op_targ = tmp;
2164 Whine if they've said @foo in a doublequoted string,
2165 and @foo isn't a variable we can find in the symbol
2168 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2169 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2170 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2171 && ckWARN(WARN_AMBIGUOUS))
2173 /* Downgraded from fatal to warning 20000522 mjd */
2174 Perl_warner(aTHX_ WARN_AMBIGUOUS,
2175 "Possible unintended interpolation of %s in string",
2180 /* build ops for a bareword */
2181 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2182 yylval.opval->op_private = OPpCONST_ENTERED;
2183 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2184 ((PL_tokenbuf[0] == '$') ? SVt_PV
2185 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2190 /* no identifier pending identification */
2192 switch (PL_lex_state) {
2194 case LEX_NORMAL: /* Some compilers will produce faster */
2195 case LEX_INTERPNORMAL: /* code if we comment these out. */
2199 /* when we've already built the next token, just pull it out of the queue */
2202 yylval = PL_nextval[PL_nexttoke];
2204 PL_lex_state = PL_lex_defer;
2205 PL_expect = PL_lex_expect;
2206 PL_lex_defer = LEX_NORMAL;
2208 return(PL_nexttype[PL_nexttoke]);
2210 /* interpolated case modifiers like \L \U, including \Q and \E.
2211 when we get here, PL_bufptr is at the \
2213 case LEX_INTERPCASEMOD:
2215 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2216 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2218 /* handle \E or end of string */
2219 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2223 if (PL_lex_casemods) {
2224 oldmod = PL_lex_casestack[--PL_lex_casemods];
2225 PL_lex_casestack[PL_lex_casemods] = '\0';
2227 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2229 PL_lex_state = LEX_INTERPCONCAT;
2233 if (PL_bufptr != PL_bufend)
2235 PL_lex_state = LEX_INTERPCONCAT;
2240 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2241 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2242 if (strchr("LU", *s) &&
2243 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2245 PL_lex_casestack[--PL_lex_casemods] = '\0';
2248 if (PL_lex_casemods > 10) {
2249 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2250 if (newlb != PL_lex_casestack) {
2252 PL_lex_casestack = newlb;
2255 PL_lex_casestack[PL_lex_casemods++] = *s;
2256 PL_lex_casestack[PL_lex_casemods] = '\0';
2257 PL_lex_state = LEX_INTERPCONCAT;
2258 PL_nextval[PL_nexttoke].ival = 0;
2261 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2263 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2265 PL_nextval[PL_nexttoke].ival = OP_LC;
2267 PL_nextval[PL_nexttoke].ival = OP_UC;
2269 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2271 Perl_croak(aTHX_ "panic: yylex");
2274 if (PL_lex_starts) {
2283 case LEX_INTERPPUSH:
2284 return sublex_push();
2286 case LEX_INTERPSTART:
2287 if (PL_bufptr == PL_bufend)
2288 return sublex_done();
2290 PL_lex_dojoin = (*PL_bufptr == '@');
2291 PL_lex_state = LEX_INTERPNORMAL;
2292 if (PL_lex_dojoin) {
2293 PL_nextval[PL_nexttoke].ival = 0;
2296 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2297 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2298 force_next(PRIVATEREF);
2300 force_ident("\"", '$');
2301 #endif /* USE_THREADS */
2302 PL_nextval[PL_nexttoke].ival = 0;
2304 PL_nextval[PL_nexttoke].ival = 0;
2306 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2309 if (PL_lex_starts++) {
2315 case LEX_INTERPENDMAYBE:
2316 if (intuit_more(PL_bufptr)) {
2317 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2323 if (PL_lex_dojoin) {
2324 PL_lex_dojoin = FALSE;
2325 PL_lex_state = LEX_INTERPCONCAT;
2328 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2329 && SvEVALED(PL_lex_repl))
2331 if (PL_bufptr != PL_bufend)
2332 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2333 PL_lex_repl = Nullsv;
2336 case LEX_INTERPCONCAT:
2338 if (PL_lex_brackets)
2339 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2341 if (PL_bufptr == PL_bufend)
2342 return sublex_done();
2344 if (SvIVX(PL_linestr) == '\'') {
2345 SV *sv = newSVsv(PL_linestr);
2348 else if ( PL_hints & HINT_NEW_RE )
2349 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2350 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2354 s = scan_const(PL_bufptr);
2356 PL_lex_state = LEX_INTERPCASEMOD;
2358 PL_lex_state = LEX_INTERPSTART;
2361 if (s != PL_bufptr) {
2362 PL_nextval[PL_nexttoke] = yylval;
2365 if (PL_lex_starts++)
2375 PL_lex_state = LEX_NORMAL;
2376 s = scan_formline(PL_bufptr);
2377 if (!PL_lex_formbrack)
2383 PL_oldoldbufptr = PL_oldbufptr;
2386 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2387 exp_name[PL_expect], s);
2393 if (isIDFIRST_lazy_if(s,UTF))
2395 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2398 goto fake_eof; /* emulate EOF on ^D or ^Z */
2403 if (PL_lex_brackets)
2404 yyerror("Missing right curly or square bracket");
2407 if (s++ < PL_bufend)
2408 goto retry; /* ignore stray nulls */
2411 if (!PL_in_eval && !PL_preambled) {
2412 PL_preambled = TRUE;
2413 sv_setpv(PL_linestr,incl_perldb());
2414 if (SvCUR(PL_linestr))
2415 sv_catpv(PL_linestr,";");
2417 while(AvFILLp(PL_preambleav) >= 0) {
2418 SV *tmpsv = av_shift(PL_preambleav);
2419 sv_catsv(PL_linestr, tmpsv);
2420 sv_catpv(PL_linestr, ";");
2423 sv_free((SV*)PL_preambleav);
2424 PL_preambleav = NULL;
2426 if (PL_minus_n || PL_minus_p) {
2427 sv_catpv(PL_linestr, "LINE: while (<>) {");
2429 sv_catpv(PL_linestr,"chomp;");
2431 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2433 GvIMPORTED_AV_on(gv);
2435 if (strchr("/'\"", *PL_splitstr)
2436 && strchr(PL_splitstr + 1, *PL_splitstr))
2437 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2440 s = "'~#\200\1'"; /* surely one char is unused...*/
2441 while (s[1] && strchr(PL_splitstr, *s)) s++;
2443 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2444 "q" + (delim == '\''), delim);
2445 for (s = PL_splitstr; *s; s++) {
2447 sv_catpvn(PL_linestr, "\\", 1);
2448 sv_catpvn(PL_linestr, s, 1);
2450 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2454 sv_catpv(PL_linestr,"@F=split(' ');");
2457 sv_catpv(PL_linestr, "\n");
2458 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2459 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2460 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2461 SV *sv = NEWSV(85,0);
2463 sv_upgrade(sv, SVt_PVMG);
2464 sv_setsv(sv,PL_linestr);
2465 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2470 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2473 if (PL_preprocess && !PL_in_eval)
2474 (void)PerlProc_pclose(PL_rsfp);
2475 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2476 PerlIO_clearerr(PL_rsfp);
2478 (void)PerlIO_close(PL_rsfp);
2480 PL_doextract = FALSE;
2482 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2483 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2484 sv_catpv(PL_linestr,";}");
2485 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2486 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2487 PL_minus_n = PL_minus_p = 0;
2490 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2491 sv_setpv(PL_linestr,"");
2492 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2495 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2496 PL_doextract = FALSE;
2498 /* Incest with pod. */
2499 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2500 sv_setpv(PL_linestr, "");
2501 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2502 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2503 PL_doextract = FALSE;
2507 } while (PL_doextract);
2508 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2509 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2510 SV *sv = NEWSV(85,0);
2512 sv_upgrade(sv, SVt_PVMG);
2513 sv_setsv(sv,PL_linestr);
2514 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2516 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2517 if (CopLINE(PL_curcop) == 1) {
2518 while (s < PL_bufend && isSPACE(*s))
2520 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2524 if (*s == '#' && *(s+1) == '!')
2526 #ifdef ALTERNATE_SHEBANG
2528 static char as[] = ALTERNATE_SHEBANG;
2529 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2530 d = s + (sizeof(as) - 1);
2532 #endif /* ALTERNATE_SHEBANG */
2541 while (*d && !isSPACE(*d))
2545 #ifdef ARG_ZERO_IS_SCRIPT
2546 if (ipathend > ipath) {
2548 * HP-UX (at least) sets argv[0] to the script name,
2549 * which makes $^X incorrect. And Digital UNIX and Linux,
2550 * at least, set argv[0] to the basename of the Perl
2551 * interpreter. So, having found "#!", we'll set it right.
2553 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2554 assert(SvPOK(x) || SvGMAGICAL(x));
2555 if (sv_eq(x, CopFILESV(PL_curcop))) {
2556 sv_setpvn(x, ipath, ipathend - ipath);
2559 TAINT_NOT; /* $^X is always tainted, but that's OK */
2561 #endif /* ARG_ZERO_IS_SCRIPT */
2566 d = instr(s,"perl -");
2568 d = instr(s,"perl");
2570 /* avoid getting into infinite loops when shebang
2571 * line contains "Perl" rather than "perl" */
2573 for (d = ipathend-4; d >= ipath; --d) {
2574 if ((*d == 'p' || *d == 'P')
2575 && !ibcmp(d, "perl", 4))
2585 #ifdef ALTERNATE_SHEBANG
2587 * If the ALTERNATE_SHEBANG on this system starts with a
2588 * character that can be part of a Perl expression, then if
2589 * we see it but not "perl", we're probably looking at the
2590 * start of Perl code, not a request to hand off to some
2591 * other interpreter. Similarly, if "perl" is there, but
2592 * not in the first 'word' of the line, we assume the line
2593 * contains the start of the Perl program.
2595 if (d && *s != '#') {
2597 while (*c && !strchr("; \t\r\n\f\v#", *c))
2600 d = Nullch; /* "perl" not in first word; ignore */
2602 *s = '#'; /* Don't try to parse shebang line */
2604 #endif /* ALTERNATE_SHEBANG */
2605 #ifndef MACOS_TRADITIONAL
2610 !instr(s,"indir") &&
2611 instr(PL_origargv[0],"perl"))
2617 while (s < PL_bufend && isSPACE(*s))
2619 if (s < PL_bufend) {
2620 Newz(899,newargv,PL_origargc+3,char*);
2622 while (s < PL_bufend && !isSPACE(*s))
2625 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2628 newargv = PL_origargv;
2630 PerlProc_execv(ipath, newargv);
2631 Perl_croak(aTHX_ "Can't exec %s", ipath);
2635 U32 oldpdb = PL_perldb;
2636 bool oldn = PL_minus_n;
2637 bool oldp = PL_minus_p;
2639 while (*d && !isSPACE(*d)) d++;
2640 while (SPACE_OR_TAB(*d)) d++;
2644 if (*d == 'M' || *d == 'm') {
2646 while (*d && !isSPACE(*d)) d++;
2647 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2650 d = moreswitches(d);
2652 if ((PERLDB_LINE && !oldpdb) ||
2653 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2654 /* if we have already added "LINE: while (<>) {",
2655 we must not do it again */
2657 sv_setpv(PL_linestr, "");
2658 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2659 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2660 PL_preambled = FALSE;
2662 (void)gv_fetchfile(PL_origfilename);
2669 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2671 PL_lex_state = LEX_FORMLINE;
2676 #ifdef PERL_STRICT_CR
2677 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2679 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2681 case ' ': case '\t': case '\f': case 013:
2682 #ifdef MACOS_TRADITIONAL
2689 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2690 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2691 /* handle eval qq[#line 1 "foo"\n ...] */
2692 CopLINE_dec(PL_curcop);
2696 while (s < d && *s != '\n')
2701 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2703 PL_lex_state = LEX_FORMLINE;
2713 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2718 while (s < PL_bufend && SPACE_OR_TAB(*s))
2721 if (strnEQ(s,"=>",2)) {
2722 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2723 OPERATOR('-'); /* unary minus */
2725 PL_last_uni = PL_oldbufptr;
2726 PL_last_lop_op = OP_FTEREAD; /* good enough */
2728 case 'r': FTST(OP_FTEREAD);
2729 case 'w': FTST(OP_FTEWRITE);
2730 case 'x': FTST(OP_FTEEXEC);
2731 case 'o': FTST(OP_FTEOWNED);
2732 case 'R': FTST(OP_FTRREAD);
2733 case 'W': FTST(OP_FTRWRITE);
2734 case 'X': FTST(OP_FTREXEC);
2735 case 'O': FTST(OP_FTROWNED);
2736 case 'e': FTST(OP_FTIS);
2737 case 'z': FTST(OP_FTZERO);
2738 case 's': FTST(OP_FTSIZE);
2739 case 'f': FTST(OP_FTFILE);
2740 case 'd': FTST(OP_FTDIR);
2741 case 'l': FTST(OP_FTLINK);
2742 case 'p': FTST(OP_FTPIPE);
2743 case 'S': FTST(OP_FTSOCK);
2744 case 'u': FTST(OP_FTSUID);
2745 case 'g': FTST(OP_FTSGID);
2746 case 'k': FTST(OP_FTSVTX);
2747 case 'b': FTST(OP_FTBLK);
2748 case 'c': FTST(OP_FTCHR);
2749 case 't': FTST(OP_FTTTY);
2750 case 'T': FTST(OP_FTTEXT);
2751 case 'B': FTST(OP_FTBINARY);
2752 case 'M': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTMTIME);
2753 case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME);
2754 case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME);
2756 Perl_croak(aTHX_ "Unrecognized file test: -%c", (int)tmp);
2763 if (PL_expect == XOPERATOR)
2768 else if (*s == '>') {
2771 if (isIDFIRST_lazy_if(s,UTF)) {
2772 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2780 if (PL_expect == XOPERATOR)
2783 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2785 OPERATOR('-'); /* unary minus */
2792 if (PL_expect == XOPERATOR)
2797 if (PL_expect == XOPERATOR)
2800 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2806 if (PL_expect != XOPERATOR) {
2807 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2808 PL_expect = XOPERATOR;
2809 force_ident(PL_tokenbuf, '*');
2822 if (PL_expect == XOPERATOR) {
2826 PL_tokenbuf[0] = '%';
2827 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2828 if (!PL_tokenbuf[1]) {
2830 yyerror("Final % should be \\% or %name");
2833 PL_pending_ident = '%';
2852 switch (PL_expect) {
2855 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2857 PL_bufptr = s; /* update in case we back off */
2863 PL_expect = XTERMBLOCK;
2867 while (isIDFIRST_lazy_if(s,UTF)) {
2868 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2869 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2870 if (tmp < 0) tmp = -tmp;
2885 d = scan_str(d,TRUE,TRUE);
2888 SvREFCNT_dec(PL_lex_stuff);
2889 PL_lex_stuff = Nullsv;
2891 /* MUST advance bufptr here to avoid bogus
2892 "at end of line" context messages from yyerror().
2894 PL_bufptr = s + len;
2895 yyerror("Unterminated attribute parameter in attribute list");
2898 return 0; /* EOF indicator */
2902 SV *sv = newSVpvn(s, len);
2903 sv_catsv(sv, PL_lex_stuff);
2904 attrs = append_elem(OP_LIST, attrs,
2905 newSVOP(OP_CONST, 0, sv));
2906 SvREFCNT_dec(PL_lex_stuff);
2907 PL_lex_stuff = Nullsv;
2910 attrs = append_elem(OP_LIST, attrs,
2911 newSVOP(OP_CONST, 0,
2915 if (*s == ':' && s[1] != ':')
2918 break; /* require real whitespace or :'s */
2920 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
2921 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
2922 char q = ((*s == '\'') ? '"' : '\'');
2923 /* If here for an expression, and parsed no attrs, back off. */
2924 if (tmp == '=' && !attrs) {
2928 /* MUST advance bufptr here to avoid bogus "at end of line"
2929 context messages from yyerror().
2933 yyerror("Unterminated attribute list");
2935 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
2943 PL_nextval[PL_nexttoke].opval = attrs;
2951 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
2952 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
2968 if (PL_lex_brackets <= 0)
2969 yyerror("Unmatched right square bracket");
2972 if (PL_lex_state == LEX_INTERPNORMAL) {
2973 if (PL_lex_brackets == 0) {
2974 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
2975 PL_lex_state = LEX_INTERPEND;
2982 if (PL_lex_brackets > 100) {
2983 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
2984 if (newlb != PL_lex_brackstack) {
2986 PL_lex_brackstack = newlb;
2989 switch (PL_expect) {
2991 if (PL_lex_formbrack) {
2995 if (PL_oldoldbufptr == PL_last_lop)
2996 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2998 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2999 OPERATOR(HASHBRACK);
3001 while (s < PL_bufend && SPACE_OR_TAB(*s))
3004 PL_tokenbuf[0] = '\0';
3005 if (d < PL_bufend && *d == '-') {
3006 PL_tokenbuf[0] = '-';
3008 while (d < PL_bufend && SPACE_OR_TAB(*d))
3011 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3012 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3014 while (d < PL_bufend && SPACE_OR_TAB(*d))
3017 char minus = (PL_tokenbuf[0] == '-');
3018 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3026 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3031 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3036 if (PL_oldoldbufptr == PL_last_lop)
3037 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3039 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3042 OPERATOR(HASHBRACK);
3043 /* This hack serves to disambiguate a pair of curlies
3044 * as being a block or an anon hash. Normally, expectation
3045 * determines that, but in cases where we're not in a
3046 * position to expect anything in particular (like inside
3047 * eval"") we have to resolve the ambiguity. This code
3048 * covers the case where the first term in the curlies is a
3049 * quoted string. Most other cases need to be explicitly
3050 * disambiguated by prepending a `+' before the opening
3051 * curly in order to force resolution as an anon hash.
3053 * XXX should probably propagate the outer expectation
3054 * into eval"" to rely less on this hack, but that could
3055 * potentially break current behavior of eval"".
3059 if (*s == '\'' || *s == '"' || *s == '`') {
3060 /* common case: get past first string, handling escapes */
3061 for (t++; t < PL_bufend && *t != *s;)
3062 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3066 else if (*s == 'q') {
3069 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3073 char open, close, term;
3076 while (t < PL_bufend && isSPACE(*t))
3080 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3084 for (t++; t < PL_bufend; t++) {
3085 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3087 else if (*t == open)
3091 for (t++; t < PL_bufend; t++) {
3092 if (*t == '\\' && t+1 < PL_bufend)
3094 else if (*t == close && --brackets <= 0)
3096 else if (*t == open)
3102 else if (isALNUM_lazy_if(t,UTF)) {
3104 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3107 while (t < PL_bufend && isSPACE(*t))
3109 /* if comma follows first term, call it an anon hash */
3110 /* XXX it could be a comma expression with loop modifiers */
3111 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3112 || (*t == '=' && t[1] == '>')))
3113 OPERATOR(HASHBRACK);
3114 if (PL_expect == XREF)
3117 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3123 yylval.ival = CopLINE(PL_curcop);
3124 if (isSPACE(*s) || *s == '#')
3125 PL_copline = NOLINE; /* invalidate current command line number */
3130 if (PL_lex_brackets <= 0)
3131 yyerror("Unmatched right curly bracket");
3133 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3134 if (PL_lex_brackets < PL_lex_formbrack)
3135 PL_lex_formbrack = 0;
3136 if (PL_lex_state == LEX_INTERPNORMAL) {
3137 if (PL_lex_brackets == 0) {
3138 if (PL_expect & XFAKEBRACK) {
3139 PL_expect &= XENUMMASK;
3140 PL_lex_state = LEX_INTERPEND;
3142 return yylex(); /* ignore fake brackets */
3144 if (*s == '-' && s[1] == '>')
3145 PL_lex_state = LEX_INTERPENDMAYBE;
3146 else if (*s != '[' && *s != '{')
3147 PL_lex_state = LEX_INTERPEND;
3150 if (PL_expect & XFAKEBRACK) {
3151 PL_expect &= XENUMMASK;
3153 return yylex(); /* ignore fake brackets */
3163 if (PL_expect == XOPERATOR) {
3164 if (ckWARN(WARN_SEMICOLON)
3165 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3167 CopLINE_dec(PL_curcop);
3168 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3169 CopLINE_inc(PL_curcop);
3174 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3176 PL_expect = XOPERATOR;
3177 force_ident(PL_tokenbuf, '&');
3181 yylval.ival = (OPpENTERSUB_AMPER<<8);
3200 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3201 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3203 if (PL_expect == XSTATE && isALPHA(tmp) &&
3204 (s == PL_linestart+1 || s[-2] == '\n') )
3206 if (PL_in_eval && !PL_rsfp) {
3211 if (strnEQ(s,"=cut",4)) {
3225 PL_doextract = TRUE;
3228 if (PL_lex_brackets < PL_lex_formbrack) {
3230 #ifdef PERL_STRICT_CR
3231 for (t = s; SPACE_OR_TAB(*t); t++) ;
3233 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3235 if (*t == '\n' || *t == '#') {
3253 if (PL_expect != XOPERATOR) {
3254 if (s[1] != '<' && !strchr(s,'>'))
3257 s = scan_heredoc(s);
3259 s = scan_inputsymbol(s);
3260 TERM(sublex_start());
3265 SHop(OP_LEFT_SHIFT);
3279 SHop(OP_RIGHT_SHIFT);
3288 if (PL_expect == XOPERATOR) {
3289 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3292 return ','; /* grandfather non-comma-format format */
3296 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3297 PL_tokenbuf[0] = '@';
3298 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3299 sizeof PL_tokenbuf - 1, FALSE);
3300 if (PL_expect == XOPERATOR)
3301 no_op("Array length", s);
3302 if (!PL_tokenbuf[1])
3304 PL_expect = XOPERATOR;
3305 PL_pending_ident = '#';
3309 PL_tokenbuf[0] = '$';
3310 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3311 sizeof PL_tokenbuf - 1, FALSE);
3312 if (PL_expect == XOPERATOR)
3314 if (!PL_tokenbuf[1]) {
3316 yyerror("Final $ should be \\$ or $name");
3320 /* This kludge not intended to be bulletproof. */
3321 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3322 yylval.opval = newSVOP(OP_CONST, 0,
3323 newSViv(PL_compiling.cop_arybase));
3324 yylval.opval->op_private = OPpCONST_ARYBASE;
3330 if (PL_lex_state == LEX_NORMAL)
3333 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3336 PL_tokenbuf[0] = '@';
3337 if (ckWARN(WARN_SYNTAX)) {
3339 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3342 PL_bufptr = skipspace(PL_bufptr);
3343 while (t < PL_bufend && *t != ']')
3345 Perl_warner(aTHX_ WARN_SYNTAX,
3346 "Multidimensional syntax %.*s not supported",
3347 (t - PL_bufptr) + 1, PL_bufptr);
3351 else if (*s == '{') {
3352 PL_tokenbuf[0] = '%';
3353 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3354 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3356 char tmpbuf[sizeof PL_tokenbuf];
3358 for (t++; isSPACE(*t); t++) ;
3359 if (isIDFIRST_lazy_if(t,UTF)) {
3360 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3361 for (; isSPACE(*t); t++) ;
3362 if (*t == ';' && get_cv(tmpbuf, FALSE))
3363 Perl_warner(aTHX_ WARN_SYNTAX,
3364 "You need to quote \"%s\"", tmpbuf);
3370 PL_expect = XOPERATOR;
3371 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3372 bool islop = (PL_last_lop == PL_oldoldbufptr);
3373 if (!islop || PL_last_lop_op == OP_GREPSTART)
3374 PL_expect = XOPERATOR;
3375 else if (strchr("$@\"'`q", *s))
3376 PL_expect = XTERM; /* e.g. print $fh "foo" */
3377 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3378 PL_expect = XTERM; /* e.g. print $fh &sub */
3379 else if (isIDFIRST_lazy_if(s,UTF)) {
3380 char tmpbuf[sizeof PL_tokenbuf];
3381 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3382 if ((tmp = keyword(tmpbuf, len))) {
3383 /* binary operators exclude handle interpretations */
3395 PL_expect = XTERM; /* e.g. print $fh length() */
3400 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3401 if (gv && GvCVu(gv))
3402 PL_expect = XTERM; /* e.g. print $fh subr() */
3405 else if (isDIGIT(*s))
3406 PL_expect = XTERM; /* e.g. print $fh 3 */
3407 else if (*s == '.' && isDIGIT(s[1]))
3408 PL_expect = XTERM; /* e.g. print $fh .3 */
3409 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3410 PL_expect = XTERM; /* e.g. print $fh -1 */
3411 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3412 PL_expect = XTERM; /* print $fh <<"EOF" */
3414 PL_pending_ident = '$';
3418 if (PL_expect == XOPERATOR)
3420 PL_tokenbuf[0] = '@';
3421 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3422 if (!PL_tokenbuf[1]) {
3424 yyerror("Final @ should be \\@ or @name");
3427 if (PL_lex_state == LEX_NORMAL)
3429 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3431 PL_tokenbuf[0] = '%';
3433 /* Warn about @ where they meant $. */
3434 if (ckWARN(WARN_SYNTAX)) {
3435 if (*s == '[' || *s == '{') {
3437 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3439 if (*t == '}' || *t == ']') {
3441 PL_bufptr = skipspace(PL_bufptr);
3442 Perl_warner(aTHX_ WARN_SYNTAX,
3443 "Scalar value %.*s better written as $%.*s",
3444 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3449 PL_pending_ident = '@';
3452 case '/': /* may either be division or pattern */
3453 case '?': /* may either be conditional or pattern */
3454 if (PL_expect != XOPERATOR) {
3455 /* Disable warning on "study /blah/" */
3456 if (PL_oldoldbufptr == PL_last_uni
3457 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3458 || memNE(PL_last_uni, "study", 5)
3459 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3461 s = scan_pat(s,OP_MATCH);
3462 TERM(sublex_start());
3470 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3471 #ifdef PERL_STRICT_CR
3474 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3476 && (s == PL_linestart || s[-1] == '\n') )
3478 PL_lex_formbrack = 0;
3482 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3488 yylval.ival = OPf_SPECIAL;
3494 if (PL_expect != XOPERATOR)
3499 case '0': case '1': case '2': case '3': case '4':
3500 case '5': case '6': case '7': case '8': case '9':
3502 if (PL_expect == XOPERATOR)
3507 s = scan_str(s,FALSE,FALSE);
3508 if (PL_expect == XOPERATOR) {
3509 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3512 return ','; /* grandfather non-comma-format format */
3518 missingterm((char*)0);
3519 yylval.ival = OP_CONST;
3520 TERM(sublex_start());
3523 s = scan_str(s,FALSE,FALSE);
3524 if (PL_expect == XOPERATOR) {
3525 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3528 return ','; /* grandfather non-comma-format format */
3534 missingterm((char*)0);
3535 yylval.ival = OP_CONST;
3536 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3537 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3538 yylval.ival = OP_STRINGIFY;
3542 TERM(sublex_start());
3545 s = scan_str(s,FALSE,FALSE);
3546 if (PL_expect == XOPERATOR)
3547 no_op("Backticks",s);
3549 missingterm((char*)0);
3550 yylval.ival = OP_BACKTICK;
3552 TERM(sublex_start());
3556 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3557 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3559 if (PL_expect == XOPERATOR)
3560 no_op("Backslash",s);
3564 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3568 while (isDIGIT(*start) || *start == '_')
3570 if (*start == '.' && isDIGIT(start[1])) {
3574 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3575 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF)) {
3579 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3589 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3628 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3630 /* Some keywords can be followed by any delimiter, including ':' */
3631 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3632 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3633 (PL_tokenbuf[0] == 'q' &&
3634 strchr("qwxr", PL_tokenbuf[1])))));
3636 /* x::* is just a word, unless x is "CORE" */
3637 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3641 while (d < PL_bufend && isSPACE(*d))
3642 d++; /* no comments skipped here, or s### is misparsed */
3644 /* Is this a label? */
3645 if (!tmp && PL_expect == XSTATE
3646 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3648 yylval.pval = savepv(PL_tokenbuf);
3653 /* Check for keywords */
3654 tmp = keyword(PL_tokenbuf, len);
3656 /* Is this a word before a => operator? */
3657 if (*d == '=' && d[1] == '>') {
3659 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3660 yylval.opval->op_private = OPpCONST_BARE;
3664 if (tmp < 0) { /* second-class keyword? */
3665 GV *ogv = Nullgv; /* override (winner) */
3666 GV *hgv = Nullgv; /* hidden (loser) */
3667 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3669 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3672 if (GvIMPORTED_CV(gv))
3674 else if (! CvMETHOD(cv))
3678 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3679 (gv = *gvp) != (GV*)&PL_sv_undef &&
3680 GvCVu(gv) && GvIMPORTED_CV(gv))
3686 tmp = 0; /* overridden by import or by GLOBAL */
3689 && -tmp==KEY_lock /* XXX generalizable kludge */
3691 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3693 tmp = 0; /* any sub overrides "weak" keyword */
3695 else { /* no override */
3699 if (ckWARN(WARN_AMBIGUOUS) && hgv
3700 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3701 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3702 "Ambiguous call resolved as CORE::%s(), %s",
3703 GvENAME(hgv), "qualify as such or use &");
3710 default: /* not a keyword */
3713 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3715 /* Get the rest if it looks like a package qualifier */
3717 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3719 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3722 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3723 *s == '\'' ? "'" : "::");
3727 if (PL_expect == XOPERATOR) {
3728 if (PL_bufptr == PL_linestart) {
3729 CopLINE_dec(PL_curcop);
3730 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3731 CopLINE_inc(PL_curcop);
3734 no_op("Bareword",s);
3737 /* Look for a subroutine with this name in current package,
3738 unless name is "Foo::", in which case Foo is a bearword
3739 (and a package name). */
3742 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3744 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3745 Perl_warner(aTHX_ WARN_BAREWORD,
3746 "Bareword \"%s\" refers to nonexistent package",
3749 PL_tokenbuf[len] = '\0';
3756 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3759 /* if we saw a global override before, get the right name */
3762 sv = newSVpvn("CORE::GLOBAL::",14);
3763 sv_catpv(sv,PL_tokenbuf);
3766 sv = newSVpv(PL_tokenbuf,0);
3768 /* Presume this is going to be a bareword of some sort. */
3771 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3772 yylval.opval->op_private = OPpCONST_BARE;
3774 /* And if "Foo::", then that's what it certainly is. */
3779 /* See if it's the indirect object for a list operator. */
3781 if (PL_oldoldbufptr &&
3782 PL_oldoldbufptr < PL_bufptr &&
3783 (PL_oldoldbufptr == PL_last_lop
3784 || PL_oldoldbufptr == PL_last_uni) &&
3785 /* NO SKIPSPACE BEFORE HERE! */
3786 (PL_expect == XREF ||
3787 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3789 bool immediate_paren = *s == '(';
3791 /* (Now we can afford to cross potential line boundary.) */
3794 /* Two barewords in a row may indicate method call. */
3796 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3799 /* If not a declared subroutine, it's an indirect object. */
3800 /* (But it's an indir obj regardless for sort.) */
3802 if ((PL_last_lop_op == OP_SORT ||
3803 (!immediate_paren && (!gv || !GvCVu(gv)))) &&
3804 (PL_last_lop_op != OP_MAPSTART &&
3805 PL_last_lop_op != OP_GREPSTART))
3807 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3813 PL_expect = XOPERATOR;
3816 /* Is this a word before a => operator? */
3817 if (*s == '=' && s[1] == '>') {
3819 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
3823 /* If followed by a paren, it's certainly a subroutine. */
3826 if (gv && GvCVu(gv)) {
3827 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
3828 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3833 PL_nextval[PL_nexttoke].opval = yylval.opval;
3834 PL_expect = XOPERATOR;
3840 /* If followed by var or block, call it a method (unless sub) */
3842 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3843 PL_last_lop = PL_oldbufptr;
3844 PL_last_lop_op = OP_METHOD;
3848 /* If followed by a bareword, see if it looks like indir obj. */
3850 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3853 /* Not a method, so call it a subroutine (if defined) */
3855 if (gv && GvCVu(gv)) {
3857 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3858 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3859 "Ambiguous use of -%s resolved as -&%s()",
3860 PL_tokenbuf, PL_tokenbuf);
3861 /* Check for a constant sub */
3863 if ((sv = cv_const_sv(cv))) {
3865 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3866 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3867 yylval.opval->op_private = 0;
3871 /* Resolve to GV now. */
3872 op_free(yylval.opval);
3873 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3874 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3875 PL_last_lop = PL_oldbufptr;
3876 PL_last_lop_op = OP_ENTERSUB;
3877 /* Is there a prototype? */
3880 char *proto = SvPV((SV*)cv, len);
3883 if (strEQ(proto, "$"))
3885 if (*proto == '&' && *s == '{') {
3886 sv_setpv(PL_subname,"__ANON__");
3890 PL_nextval[PL_nexttoke].opval = yylval.opval;
3896 /* Call it a bare word */
3898 if (PL_hints & HINT_STRICT_SUBS)
3899 yylval.opval->op_private |= OPpCONST_STRICT;
3902 if (ckWARN(WARN_RESERVED)) {
3903 if (lastchar != '-') {
3904 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
3906 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
3913 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
3914 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3915 "Operator or semicolon missing before %c%s",
3916 lastchar, PL_tokenbuf);
3917 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3918 "Ambiguous use of %c resolved as operator %c",
3919 lastchar, lastchar);
3925 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3926 newSVpv(CopFILE(PL_curcop),0));
3930 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3931 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
3934 case KEY___PACKAGE__:
3935 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3937 ? newSVsv(PL_curstname)
3946 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
3947 char *pname = "main";
3948 if (PL_tokenbuf[2] == 'D')
3949 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
3950 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
3953 GvIOp(gv) = newIO();
3954 IoIFP(GvIOp(gv)) = PL_rsfp;
3955 #if defined(HAS_FCNTL) && defined(F_SETFD)
3957 int fd = PerlIO_fileno(PL_rsfp);
3958 fcntl(fd,F_SETFD,fd >= 3);
3961 /* Mark this internal pseudo-handle as clean */
3962 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
3964 IoTYPE(GvIOp(gv)) = '|';
3965 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
3966 IoTYPE(GvIOp(gv)) = '-';
3968 IoTYPE(GvIOp(gv)) = '<';
3969 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
3970 /* if the script was opened in binmode, we need to revert
3971 * it to text mode for compatibility; but only iff it has CRs
3972 * XXX this is a questionable hack at best. */
3973 if (PL_bufend-PL_bufptr > 2
3974 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
3977 if (IoTYPE(GvIOp(gv)) == '<') {
3978 loc = PerlIO_tell(PL_rsfp);
3979 (void)PerlIO_seek(PL_rsfp, 0L, 0);
3981 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
3982 #if defined(__BORLANDC__)
3983 /* XXX see note in do_binmode() */
3984 ((FILE*)PL_rsfp)->flags |= _F_BIN;
3987 PerlIO_seek(PL_rsfp, loc, 0);
4002 if (PL_expect == XSTATE) {
4009 if (*s == ':' && s[1] == ':') {
4012 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4013 if (!(tmp = keyword(PL_tokenbuf, len)))
4014 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4028 LOP(OP_ACCEPT,XTERM);
4034 LOP(OP_ATAN2,XTERM);
4040 LOP(OP_BINMODE,XTERM);
4043 LOP(OP_BLESS,XTERM);
4052 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4069 if (!PL_cryptseen) {
4070 PL_cryptseen = TRUE;
4074 LOP(OP_CRYPT,XTERM);
4077 if (ckWARN(WARN_CHMOD)) {
4078 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4079 if (*d != '0' && isDIGIT(*d))
4080 Perl_warner(aTHX_ WARN_CHMOD,
4081 "chmod() mode argument is missing initial 0");
4083 LOP(OP_CHMOD,XTERM);
4086 LOP(OP_CHOWN,XTERM);
4089 LOP(OP_CONNECT,XTERM);
4105 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4109 PL_hints |= HINT_BLOCK_SCOPE;
4119 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4120 LOP(OP_DBMOPEN,XTERM);
4126 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4133 yylval.ival = CopLINE(PL_curcop);
4147 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4148 UNIBRACK(OP_ENTEREVAL);
4163 case KEY_endhostent:
4169 case KEY_endservent:
4172 case KEY_endprotoent:
4183 yylval.ival = CopLINE(PL_curcop);
4185 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4187 if ((PL_bufend - p) >= 3 &&
4188 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4190 else if ((PL_bufend - p) >= 4 &&
4191 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4194 if (isIDFIRST_lazy_if(p,UTF)) {
4195 p = scan_ident(p, PL_bufend,
4196 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4200 Perl_croak(aTHX_ "Missing $ on loop variable");
4205 LOP(OP_FORMLINE,XTERM);
4211 LOP(OP_FCNTL,XTERM);
4217 LOP(OP_FLOCK,XTERM);
4226 LOP(OP_GREPSTART, XREF);
4229 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4244 case KEY_getpriority:
4245 LOP(OP_GETPRIORITY,XTERM);
4247 case KEY_getprotobyname:
4250 case KEY_getprotobynumber:
4251 LOP(OP_GPBYNUMBER,XTERM);
4253 case KEY_getprotoent:
4265 case KEY_getpeername:
4266 UNI(OP_GETPEERNAME);
4268 case KEY_gethostbyname:
4271 case KEY_gethostbyaddr:
4272 LOP(OP_GHBYADDR,XTERM);
4274 case KEY_gethostent:
4277 case KEY_getnetbyname:
4280 case KEY_getnetbyaddr:
4281 LOP(OP_GNBYADDR,XTERM);
4286 case KEY_getservbyname:
4287 LOP(OP_GSBYNAME,XTERM);
4289 case KEY_getservbyport:
4290 LOP(OP_GSBYPORT,XTERM);
4292 case KEY_getservent:
4295 case KEY_getsockname:
4296 UNI(OP_GETSOCKNAME);
4298 case KEY_getsockopt:
4299 LOP(OP_GSOCKOPT,XTERM);
4321 yylval.ival = CopLINE(PL_curcop);
4325 LOP(OP_INDEX,XTERM);
4331 LOP(OP_IOCTL,XTERM);
4343 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4375 LOP(OP_LISTEN,XTERM);
4384 s = scan_pat(s,OP_MATCH);
4385 TERM(sublex_start());
4388 LOP(OP_MAPSTART, XREF);
4391 LOP(OP_MKDIR,XTERM);
4394 LOP(OP_MSGCTL,XTERM);
4397 LOP(OP_MSGGET,XTERM);
4400 LOP(OP_MSGRCV,XTERM);
4403 LOP(OP_MSGSND,XTERM);
4409 if (isIDFIRST_lazy_if(s,UTF)) {
4410 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4411 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4413 PL_in_my_stash = gv_stashpv(PL_tokenbuf, FALSE);
4414 if (!PL_in_my_stash) {
4417 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4425 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4432 if (PL_expect != XSTATE)
4433 yyerror("\"no\" not allowed in expression");
4434 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4435 s = force_version(s);
4440 if (*s == '(' || (s = skipspace(s), *s == '('))
4447 if (isIDFIRST_lazy_if(s,UTF)) {
4449 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4451 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4452 Perl_warner(aTHX_ WARN_PRECEDENCE,
4453 "Precedence problem: open %.*s should be open(%.*s)",
4459 yylval.ival = OP_OR;
4469 LOP(OP_OPEN_DIR,XTERM);
4472 checkcomma(s,PL_tokenbuf,"filehandle");
4476 checkcomma(s,PL_tokenbuf,"filehandle");
4495 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4499 LOP(OP_PIPE_OP,XTERM);
4502 s = scan_str(s,FALSE,FALSE);
4504 missingterm((char*)0);
4505 yylval.ival = OP_CONST;
4506 TERM(sublex_start());
4512 s = scan_str(s,FALSE,FALSE);
4514 missingterm((char*)0);
4516 if (SvCUR(PL_lex_stuff)) {
4519 d = SvPV_force(PL_lex_stuff, len);
4521 for (; isSPACE(*d) && len; --len, ++d) ;
4524 if (!warned && ckWARN(WARN_QW)) {
4525 for (; !isSPACE(*d) && len; --len, ++d) {
4527 Perl_warner(aTHX_ WARN_QW,
4528 "Possible attempt to separate words with commas");
4531 else if (*d == '#') {
4532 Perl_warner(aTHX_ WARN_QW,
4533 "Possible attempt to put comments in qw() list");
4539 for (; !isSPACE(*d) && len; --len, ++d) ;
4541 words = append_elem(OP_LIST, words,
4542 newSVOP(OP_CONST, 0, tokeq(newSVpvn(b, d-b))));
4546 PL_nextval[PL_nexttoke].opval = words;
4551 SvREFCNT_dec(PL_lex_stuff);
4552 PL_lex_stuff = Nullsv;
4557 s = scan_str(s,FALSE,FALSE);
4559 missingterm((char*)0);
4560 yylval.ival = OP_STRINGIFY;
4561 if (SvIVX(PL_lex_stuff) == '\'')
4562 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4563 TERM(sublex_start());
4566 s = scan_pat(s,OP_QR);
4567 TERM(sublex_start());
4570 s = scan_str(s,FALSE,FALSE);
4572 missingterm((char*)0);
4573 yylval.ival = OP_BACKTICK;
4575 TERM(sublex_start());
4582 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4583 s = force_version(s);
4586 *PL_tokenbuf = '\0';
4587 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4588 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4589 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4591 yyerror("<> should be quotes");
4599 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4603 LOP(OP_RENAME,XTERM);
4612 LOP(OP_RINDEX,XTERM);
4635 LOP(OP_REVERSE,XTERM);
4646 TERM(sublex_start());
4648 TOKEN(1); /* force error */
4657 LOP(OP_SELECT,XTERM);
4663 LOP(OP_SEMCTL,XTERM);
4666 LOP(OP_SEMGET,XTERM);
4669 LOP(OP_SEMOP,XTERM);
4675 LOP(OP_SETPGRP,XTERM);
4677 case KEY_setpriority:
4678 LOP(OP_SETPRIORITY,XTERM);
4680 case KEY_sethostent:
4686 case KEY_setservent:
4689 case KEY_setprotoent:
4699 LOP(OP_SEEKDIR,XTERM);
4701 case KEY_setsockopt:
4702 LOP(OP_SSOCKOPT,XTERM);
4708 LOP(OP_SHMCTL,XTERM);
4711 LOP(OP_SHMGET,XTERM);
4714 LOP(OP_SHMREAD,XTERM);
4717 LOP(OP_SHMWRITE,XTERM);
4720 LOP(OP_SHUTDOWN,XTERM);
4729 LOP(OP_SOCKET,XTERM);
4731 case KEY_socketpair:
4732 LOP(OP_SOCKPAIR,XTERM);
4735 checkcomma(s,PL_tokenbuf,"subroutine name");
4737 if (*s == ';' || *s == ')') /* probably a close */
4738 Perl_croak(aTHX_ "sort is now a reserved word");
4740 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4744 LOP(OP_SPLIT,XTERM);
4747 LOP(OP_SPRINTF,XTERM);
4750 LOP(OP_SPLICE,XTERM);
4765 LOP(OP_SUBSTR,XTERM);
4771 char tmpbuf[sizeof PL_tokenbuf];
4773 expectation attrful;
4774 bool have_name, have_proto;
4779 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4780 (*s == ':' && s[1] == ':'))
4783 attrful = XATTRBLOCK;
4784 /* remember buffer pos'n for later force_word */
4785 tboffset = s - PL_oldbufptr;
4786 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4787 if (strchr(tmpbuf, ':'))
4788 sv_setpv(PL_subname, tmpbuf);
4790 sv_setsv(PL_subname,PL_curstname);
4791 sv_catpvn(PL_subname,"::",2);
4792 sv_catpvn(PL_subname,tmpbuf,len);
4799 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4800 PL_expect = XTERMBLOCK;
4801 attrful = XATTRTERM;
4802 sv_setpv(PL_subname,"?");
4806 if (key == KEY_format) {
4808 PL_lex_formbrack = PL_lex_brackets + 1;
4810 (void) force_word(PL_oldbufptr + tboffset, WORD,
4815 /* Look for a prototype */
4819 s = scan_str(s,FALSE,FALSE);
4822 SvREFCNT_dec(PL_lex_stuff);
4823 PL_lex_stuff = Nullsv;
4824 Perl_croak(aTHX_ "Prototype not terminated");
4827 d = SvPVX(PL_lex_stuff);
4829 for (p = d; *p; ++p) {
4834 SvCUR(PL_lex_stuff) = tmp;
4842 if (*s == ':' && s[1] != ':')
4843 PL_expect = attrful;
4846 PL_nextval[PL_nexttoke].opval =
4847 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4848 PL_lex_stuff = Nullsv;
4852 sv_setpv(PL_subname,"__ANON__");
4855 (void) force_word(PL_oldbufptr + tboffset, WORD,
4864 LOP(OP_SYSTEM,XREF);
4867 LOP(OP_SYMLINK,XTERM);
4870 LOP(OP_SYSCALL,XTERM);
4873 LOP(OP_SYSOPEN,XTERM);
4876 LOP(OP_SYSSEEK,XTERM);
4879 LOP(OP_SYSREAD,XTERM);
4882 LOP(OP_SYSWRITE,XTERM);
4886 TERM(sublex_start());
4907 LOP(OP_TRUNCATE,XTERM);
4919 yylval.ival = CopLINE(PL_curcop);
4923 yylval.ival = CopLINE(PL_curcop);
4927 LOP(OP_UNLINK,XTERM);
4933 LOP(OP_UNPACK,XTERM);
4936 LOP(OP_UTIME,XTERM);
4939 if (ckWARN(WARN_UMASK)) {
4940 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4941 if (*d != '0' && isDIGIT(*d))
4942 Perl_warner(aTHX_ WARN_UMASK,
4943 "umask: argument is missing initial 0");
4948 LOP(OP_UNSHIFT,XTERM);
4951 if (PL_expect != XSTATE)
4952 yyerror("\"use\" not allowed in expression");
4954 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4955 s = force_version(s);
4956 if (*s == ';' || (s = skipspace(s), *s == ';')) {
4957 PL_nextval[PL_nexttoke].opval = Nullop;
4962 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4963 s = force_version(s);
4975 yylval.ival = CopLINE(PL_curcop);
4979 PL_hints |= HINT_BLOCK_SCOPE;
4986 LOP(OP_WAITPID,XTERM);
4994 static char ctl_l[2];
4996 if (ctl_l[0] == '\0')
4997 ctl_l[0] = toCTRL('L');
4998 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5001 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5006 if (PL_expect == XOPERATOR)
5012 yylval.ival = OP_XOR;
5017 TERM(sublex_start());
5022 #pragma segment Main
5026 Perl_keyword(pTHX_ register char *d, I32 len)
5031 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5032 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5033 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5034 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5035 if (strEQ(d,"__END__")) return KEY___END__;
5039 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5044 if (strEQ(d,"and")) return -KEY_and;
5045 if (strEQ(d,"abs")) return -KEY_abs;
5048 if (strEQ(d,"alarm")) return -KEY_alarm;
5049 if (strEQ(d,"atan2")) return -KEY_atan2;
5052 if (strEQ(d,"accept")) return -KEY_accept;
5057 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5060 if (strEQ(d,"bless")) return -KEY_bless;
5061 if (strEQ(d,"bind")) return -KEY_bind;
5062 if (strEQ(d,"binmode")) return -KEY_binmode;
5065 if (strEQ(d,"CORE")) return -KEY_CORE;
5066 if (strEQ(d,"CHECK")) return KEY_CHECK;
5071 if (strEQ(d,"cmp")) return -KEY_cmp;
5072 if (strEQ(d,"chr")) return -KEY_chr;
5073 if (strEQ(d,"cos")) return -KEY_cos;
5076 if (strEQ(d,"chop")) return KEY_chop;
5079 if (strEQ(d,"close")) return -KEY_close;
5080 if (strEQ(d,"chdir")) return -KEY_chdir;
5081 if (strEQ(d,"chomp")) return KEY_chomp;
5082 if (strEQ(d,"chmod")) return -KEY_chmod;
5083 if (strEQ(d,"chown")) return -KEY_chown;
5084 if (strEQ(d,"crypt")) return -KEY_crypt;
5087 if (strEQ(d,"chroot")) return -KEY_chroot;
5088 if (strEQ(d,"caller")) return -KEY_caller;
5091 if (strEQ(d,"connect")) return -KEY_connect;
5094 if (strEQ(d,"closedir")) return -KEY_closedir;
5095 if (strEQ(d,"continue")) return -KEY_continue;
5100 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5105 if (strEQ(d,"do")) return KEY_do;
5108 if (strEQ(d,"die")) return -KEY_die;
5111 if (strEQ(d,"dump")) return -KEY_dump;
5114 if (strEQ(d,"delete")) return KEY_delete;
5117 if (strEQ(d,"defined")) return KEY_defined;
5118 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5121 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5126 if (strEQ(d,"EQ")) { deprecate(d); return -KEY_eq;}
5127 if (strEQ(d,"END")) return KEY_END;
5132 if (strEQ(d,"eq")) return -KEY_eq;
5135 if (strEQ(d,"eof")) return -KEY_eof;
5136 if (strEQ(d,"exp")) return -KEY_exp;
5139 if (strEQ(d,"else")) return KEY_else;
5140 if (strEQ(d,"exit")) return -KEY_exit;
5141 if (strEQ(d,"eval")) return KEY_eval;
5142 if (strEQ(d,"exec")) return -KEY_exec;
5143 if (strEQ(d,"each")) return KEY_each;
5146 if (strEQ(d,"elsif")) return KEY_elsif;
5149 if (strEQ(d,"exists")) return KEY_exists;
5150 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5153 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5154 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5157 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5160 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5161 if (strEQ(d,"endservent")) return -KEY_endservent;
5164 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5171 if (strEQ(d,"for")) return KEY_for;
5174 if (strEQ(d,"fork")) return -KEY_fork;
5177 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5178 if (strEQ(d,"flock")) return -KEY_flock;
5181 if (strEQ(d,"format")) return KEY_format;
5182 if (strEQ(d,"fileno")) return -KEY_fileno;
5185 if (strEQ(d,"foreach")) return KEY_foreach;
5188 if (strEQ(d,"formline")) return -KEY_formline;
5194 if (strEQ(d,"GT")) { deprecate(d); return -KEY_gt;}
5195 if (strEQ(d,"GE")) { deprecate(d); return -KEY_ge;}
5199 if (strnEQ(d,"get",3)) {
5204 if (strEQ(d,"ppid")) return -KEY_getppid;
5205 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5208 if (strEQ(d,"pwent")) return -KEY_getpwent;
5209 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5210 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5213 if (strEQ(d,"peername")) return -KEY_getpeername;
5214 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5215 if (strEQ(d,"priority")) return -KEY_getpriority;
5218 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5221 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5225 else if (*d == 'h') {
5226 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5227 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5228 if (strEQ(d,"hostent")) return -KEY_gethostent;
5230 else if (*d == 'n') {
5231 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5232 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5233 if (strEQ(d,"netent")) return -KEY_getnetent;
5235 else if (*d == 's') {
5236 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5237 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5238 if (strEQ(d,"servent")) return -KEY_getservent;
5239 if (strEQ(d,"sockname")) return -KEY_getsockname;
5240 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5242 else if (*d == 'g') {
5243 if (strEQ(d,"grent")) return -KEY_getgrent;
5244 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5245 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5247 else if (*d == 'l') {
5248 if (strEQ(d,"login")) return -KEY_getlogin;
5250 else if (strEQ(d,"c")) return -KEY_getc;
5255 if (strEQ(d,"gt")) return -KEY_gt;
5256 if (strEQ(d,"ge")) return -KEY_ge;
5259 if (strEQ(d,"grep")) return KEY_grep;
5260 if (strEQ(d,"goto")) return KEY_goto;
5261 if (strEQ(d,"glob")) return KEY_glob;
5264 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5269 if (strEQ(d,"hex")) return -KEY_hex;
5272 if (strEQ(d,"INIT")) return KEY_INIT;
5277 if (strEQ(d,"if")) return KEY_if;
5280 if (strEQ(d,"int")) return -KEY_int;
5283 if (strEQ(d,"index")) return -KEY_index;
5284 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5289 if (strEQ(d,"join")) return -KEY_join;
5293 if (strEQ(d,"keys")) return KEY_keys;
5294 if (strEQ(d,"kill")) return -KEY_kill;
5299 if (strEQ(d,"LT")) { deprecate(d); return -KEY_lt;}
5300 if (strEQ(d,"LE")) { deprecate(d); return -KEY_le;}
5306 if (strEQ(d,"lt")) return -KEY_lt;
5307 if (strEQ(d,"le")) return -KEY_le;
5308 if (strEQ(d,"lc")) return -KEY_lc;
5311 if (strEQ(d,"log")) return -KEY_log;
5314 if (strEQ(d,"last")) return KEY_last;
5315 if (strEQ(d,"link")) return -KEY_link;
5316 if (strEQ(d,"lock")) return -KEY_lock;
5319 if (strEQ(d,"local")) return KEY_local;
5320 if (strEQ(d,"lstat")) return -KEY_lstat;
5323 if (strEQ(d,"length")) return -KEY_length;
5324 if (strEQ(d,"listen")) return -KEY_listen;
5327 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5330 if (strEQ(d,"localtime")) return -KEY_localtime;
5336 case 1: return KEY_m;
5338 if (strEQ(d,"my")) return KEY_my;
5341 if (strEQ(d,"map")) return KEY_map;
5344 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5347 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5348 if (strEQ(d,"msgget")) return -KEY_msgget;
5349 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5350 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5355 if (strEQ(d,"NE")) { deprecate(d); return -KEY_ne;}
5358 if (strEQ(d,"next")) return KEY_next;
5359 if (strEQ(d,"ne")) return -KEY_ne;
5360 if (strEQ(d,"not")) return -KEY_not;
5361 if (strEQ(d,"no")) return KEY_no;
5366 if (strEQ(d,"or")) return -KEY_or;
5369 if (strEQ(d,"ord")) return -KEY_ord;
5370 if (strEQ(d,"oct")) return -KEY_oct;
5371 if (strEQ(d,"our")) return KEY_our;
5374 if (strEQ(d,"open")) return -KEY_open;
5377 if (strEQ(d,"opendir")) return -KEY_opendir;
5384 if (strEQ(d,"pop")) return KEY_pop;
5385 if (strEQ(d,"pos")) return KEY_pos;
5388 if (strEQ(d,"push")) return KEY_push;
5389 if (strEQ(d,"pack")) return -KEY_pack;
5390 if (strEQ(d,"pipe")) return -KEY_pipe;
5393 if (strEQ(d,"print")) return KEY_print;
5396 if (strEQ(d,"printf")) return KEY_printf;
5399 if (strEQ(d,"package")) return KEY_package;
5402 if (strEQ(d,"prototype")) return KEY_prototype;
5407 if (strEQ(d,"q")) return KEY_q;
5408 if (strEQ(d,"qr")) return KEY_qr;
5409 if (strEQ(d,"qq")) return KEY_qq;
5410 if (strEQ(d,"qw")) return KEY_qw;
5411 if (strEQ(d,"qx")) return KEY_qx;
5413 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5418 if (strEQ(d,"ref")) return -KEY_ref;
5421 if (strEQ(d,"read")) return -KEY_read;
5422 if (strEQ(d,"rand")) return -KEY_rand;
5423 if (strEQ(d,"recv")) return -KEY_recv;
5424 if (strEQ(d,"redo")) return KEY_redo;
5427 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5428 if (strEQ(d,"reset")) return -KEY_reset;
5431 if (strEQ(d,"return")) return KEY_return;
5432 if (strEQ(d,"rename")) return -KEY_rename;
5433 if (strEQ(d,"rindex")) return -KEY_rindex;
5436 if (strEQ(d,"require")) return -KEY_require;
5437 if (strEQ(d,"reverse")) return -KEY_reverse;
5438 if (strEQ(d,"readdir")) return -KEY_readdir;
5441 if (strEQ(d,"readlink")) return -KEY_readlink;
5442 if (strEQ(d,"readline")) return -KEY_readline;
5443 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5446 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5452 case 0: return KEY_s;
5454 if (strEQ(d,"scalar")) return KEY_scalar;
5459 if (strEQ(d,"seek")) return -KEY_seek;
5460 if (strEQ(d,"send")) return -KEY_send;
5463 if (strEQ(d,"semop")) return -KEY_semop;
5466 if (strEQ(d,"select")) return -KEY_select;
5467 if (strEQ(d,"semctl")) return -KEY_semctl;
5468 if (strEQ(d,"semget")) return -KEY_semget;
5471 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5472 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5475 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5476 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5479 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5482 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5483 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5484 if (strEQ(d,"setservent")) return -KEY_setservent;
5487 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5488 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5495 if (strEQ(d,"shift")) return KEY_shift;
5498 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5499 if (strEQ(d,"shmget")) return -KEY_shmget;
5502 if (strEQ(d,"shmread")) return -KEY_shmread;
5505 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5506 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5511 if (strEQ(d,"sin")) return -KEY_sin;
5514 if (strEQ(d,"sleep")) return -KEY_sleep;
5517 if (strEQ(d,"sort")) return KEY_sort;
5518 if (strEQ(d,"socket")) return -KEY_socket;
5519 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5522 if (strEQ(d,"split")) return KEY_split;
5523 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5524 if (strEQ(d,"splice")) return KEY_splice;
5527 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5530 if (strEQ(d,"srand")) return -KEY_srand;
5533 if (strEQ(d,"stat")) return -KEY_stat;
5534 if (strEQ(d,"study")) return KEY_study;
5537 if (strEQ(d,"substr")) return -KEY_substr;
5538 if (strEQ(d,"sub")) return KEY_sub;
5543 if (strEQ(d,"system")) return -KEY_system;
5546 if (strEQ(d,"symlink")) return -KEY_symlink;
5547 if (strEQ(d,"syscall")) return -KEY_syscall;
5548 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5549 if (strEQ(d,"sysread")) return -KEY_sysread;
5550 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5553 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5562 if (strEQ(d,"tr")) return KEY_tr;
5565 if (strEQ(d,"tie")) return KEY_tie;
5568 if (strEQ(d,"tell")) return -KEY_tell;
5569 if (strEQ(d,"tied")) return KEY_tied;
5570 if (strEQ(d,"time")) return -KEY_time;
5573 if (strEQ(d,"times")) return -KEY_times;
5576 if (strEQ(d,"telldir")) return -KEY_telldir;
5579 if (strEQ(d,"truncate")) return -KEY_truncate;
5586 if (strEQ(d,"uc")) return -KEY_uc;
5589 if (strEQ(d,"use")) return KEY_use;
5592 if (strEQ(d,"undef")) return KEY_undef;
5593 if (strEQ(d,"until")) return KEY_until;
5594 if (strEQ(d,"untie")) return KEY_untie;
5595 if (strEQ(d,"utime")) return -KEY_utime;
5596 if (strEQ(d,"umask")) return -KEY_umask;
5599 if (strEQ(d,"unless")) return KEY_unless;
5600 if (strEQ(d,"unpack")) return -KEY_unpack;
5601 if (strEQ(d,"unlink")) return -KEY_unlink;
5604 if (strEQ(d,"unshift")) return KEY_unshift;
5605 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5610 if (strEQ(d,"values")) return -KEY_values;
5611 if (strEQ(d,"vec")) return -KEY_vec;
5616 if (strEQ(d,"warn")) return -KEY_warn;
5617 if (strEQ(d,"wait")) return -KEY_wait;
5620 if (strEQ(d,"while")) return KEY_while;
5621 if (strEQ(d,"write")) return -KEY_write;
5624 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5627 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5632 if (len == 1) return -KEY_x;
5633 if (strEQ(d,"xor")) return -KEY_xor;
5636 if (len == 1) return KEY_y;
5645 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5649 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5650 dTHR; /* only for ckWARN */
5651 if (ckWARN(WARN_SYNTAX)) {
5653 for (w = s+2; *w && level; w++) {
5660 for (; *w && isSPACE(*w); w++) ;
5661 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5662 Perl_warner(aTHX_ WARN_SYNTAX,
5663 "%s (...) interpreted as function",name);
5666 while (s < PL_bufend && isSPACE(*s))
5670 while (s < PL_bufend && isSPACE(*s))
5672 if (isIDFIRST_lazy_if(s,UTF)) {
5674 while (isALNUM_lazy_if(s,UTF))
5676 while (s < PL_bufend && isSPACE(*s))
5681 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5685 Perl_croak(aTHX_ "No comma allowed after %s", what);
5690 /* Either returns sv, or mortalizes sv and returns a new SV*.
5691 Best used as sv=new_constant(..., sv, ...).
5692 If s, pv are NULL, calls subroutine with one argument,
5693 and type is used with error messages only. */
5696 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5700 HV *table = GvHV(PL_hintgv); /* ^H */
5704 const char *why1, *why2, *why3;
5706 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
5709 why1 = "%^H is not consistent";
5710 why2 = strEQ(key,"charnames")
5711 ? " (missing \"use charnames ...\"?)"
5715 msg = Perl_newSVpvf(aTHX_ "constant(%s): %s%s%s",
5716 (type ? type: "undef"), why1, why2, why3);
5717 yyerror(SvPVX(msg));
5721 cvp = hv_fetch(table, key, strlen(key), FALSE);
5722 if (!cvp || !SvOK(*cvp)) {
5725 why3 = "} is not defined";
5728 sv_2mortal(sv); /* Parent created it permanently */
5731 pv = sv_2mortal(newSVpvn(s, len));
5733 typesv = sv_2mortal(newSVpv(type, 0));
5735 typesv = &PL_sv_undef;
5737 PUSHSTACKi(PERLSI_OVERLOAD);
5750 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5754 /* Check the eval first */
5755 if (!PL_in_eval && SvTRUE(ERRSV)) {
5757 sv_catpv(ERRSV, "Propagated");
5758 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5760 res = SvREFCNT_inc(sv);
5764 (void)SvREFCNT_inc(res);
5773 why1 = "Call to &{$^H{";
5775 why3 = "}} did not return a defined value";
5784 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5786 register char *d = dest;
5787 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5790 Perl_croak(aTHX_ ident_too_long);
5791 if (isALNUM(*s)) /* UTF handled below */
5793 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
5798 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5802 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5803 char *t = s + UTF8SKIP(s);
5804 while (*t & 0x80 && is_utf8_mark((U8*)t))
5806 if (d + (t - s) > e)
5807 Perl_croak(aTHX_ ident_too_long);
5808 Copy(s, d, t - s, char);
5821 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5831 e = d + destlen - 3; /* two-character token, ending NUL */
5833 while (isDIGIT(*s)) {
5835 Perl_croak(aTHX_ ident_too_long);
5842 Perl_croak(aTHX_ ident_too_long);
5843 if (isALNUM(*s)) /* UTF handled below */
5845 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
5850 else if (*s == ':' && s[1] == ':') {
5854 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5855 char *t = s + UTF8SKIP(s);
5856 while (*t & 0x80 && is_utf8_mark((U8*)t))
5858 if (d + (t - s) > e)
5859 Perl_croak(aTHX_ ident_too_long);
5860 Copy(s, d, t - s, char);
5871 if (PL_lex_state != LEX_NORMAL)
5872 PL_lex_state = LEX_INTERPENDMAYBE;
5875 if (*s == '$' && s[1] &&
5876 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
5889 if (*d == '^' && *s && isCONTROLVAR(*s)) {
5894 if (isSPACE(s[-1])) {
5897 if (!SPACE_OR_TAB(ch)) {
5903 if (isIDFIRST_lazy_if(d,UTF)) {
5907 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
5909 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
5912 Copy(s, d, e - s, char);
5917 while ((isALNUM(*s) || *s == ':') && d < e)
5920 Perl_croak(aTHX_ ident_too_long);
5923 while (s < send && SPACE_OR_TAB(*s)) s++;
5924 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
5925 dTHR; /* only for ckWARN */
5926 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
5927 const char *brack = *s == '[' ? "[...]" : "{...}";
5928 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5929 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
5930 funny, dest, brack, funny, dest, brack);
5933 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
5937 /* Handle extended ${^Foo} variables
5938 * 1999-02-27 mjd-perl-patch@plover.com */
5939 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
5943 while (isALNUM(*s) && d < e) {
5947 Perl_croak(aTHX_ ident_too_long);
5952 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
5953 PL_lex_state = LEX_INTERPEND;
5956 if (PL_lex_state == LEX_NORMAL) {
5957 dTHR; /* only for ckWARN */
5958 if (ckWARN(WARN_AMBIGUOUS) &&
5959 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
5961 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5962 "Ambiguous use of %c{%s} resolved to %c%s",
5963 funny, dest, funny, dest);
5968 s = bracket; /* let the parser handle it */
5972 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
5973 PL_lex_state = LEX_INTERPEND;
5978 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
5983 *pmfl |= PMf_GLOBAL;
5985 *pmfl |= PMf_CONTINUE;
5989 *pmfl |= PMf_MULTILINE;
5991 *pmfl |= PMf_SINGLELINE;
5993 *pmfl |= PMf_EXTENDED;
5997 S_scan_pat(pTHX_ char *start, I32 type)
6002 s = scan_str(start,FALSE,FALSE);
6005 SvREFCNT_dec(PL_lex_stuff);
6006 PL_lex_stuff = Nullsv;
6007 Perl_croak(aTHX_ "Search pattern not terminated");
6010 pm = (PMOP*)newPMOP(type, 0);
6011 if (PL_multi_open == '?')
6012 pm->op_pmflags |= PMf_ONCE;
6014 while (*s && strchr("iomsx", *s))
6015 pmflag(&pm->op_pmflags,*s++);
6018 while (*s && strchr("iogcmsx", *s))
6019 pmflag(&pm->op_pmflags,*s++);
6021 pm->op_pmpermflags = pm->op_pmflags;
6023 PL_lex_op = (OP*)pm;
6024 yylval.ival = OP_MATCH;
6029 S_scan_subst(pTHX_ char *start)
6036 yylval.ival = OP_NULL;
6038 s = scan_str(start,FALSE,FALSE);
6042 SvREFCNT_dec(PL_lex_stuff);
6043 PL_lex_stuff = Nullsv;
6044 Perl_croak(aTHX_ "Substitution pattern not terminated");
6047 if (s[-1] == PL_multi_open)
6050 first_start = PL_multi_start;
6051 s = scan_str(s,FALSE,FALSE);
6054 SvREFCNT_dec(PL_lex_stuff);
6055 PL_lex_stuff = Nullsv;
6057 SvREFCNT_dec(PL_lex_repl);
6058 PL_lex_repl = Nullsv;
6059 Perl_croak(aTHX_ "Substitution replacement not terminated");
6061 PL_multi_start = first_start; /* so whole substitution is taken together */
6063 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6069 else if (strchr("iogcmsx", *s))
6070 pmflag(&pm->op_pmflags,*s++);
6077 PL_sublex_info.super_bufptr = s;
6078 PL_sublex_info.super_bufend = PL_bufend;
6080 pm->op_pmflags |= PMf_EVAL;
6081 repl = newSVpvn("",0);
6083 sv_catpv(repl, es ? "eval " : "do ");
6084 sv_catpvn(repl, "{ ", 2);
6085 sv_catsv(repl, PL_lex_repl);
6086 sv_catpvn(repl, " };", 2);
6088 SvREFCNT_dec(PL_lex_repl);
6092 pm->op_pmpermflags = pm->op_pmflags;
6093 PL_lex_op = (OP*)pm;
6094 yylval.ival = OP_SUBST;
6099 S_scan_trans(pTHX_ char *start)
6110 yylval.ival = OP_NULL;
6112 s = scan_str(start,FALSE,FALSE);
6115 SvREFCNT_dec(PL_lex_stuff);
6116 PL_lex_stuff = Nullsv;
6117 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6119 if (s[-1] == PL_multi_open)
6122 s = scan_str(s,FALSE,FALSE);
6125 SvREFCNT_dec(PL_lex_stuff);
6126 PL_lex_stuff = Nullsv;
6128 SvREFCNT_dec(PL_lex_repl);
6129 PL_lex_repl = Nullsv;
6130 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6134 o = newSVOP(OP_TRANS, 0, 0);
6135 utf8 = OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF;
6138 New(803,tbl,256,short);
6139 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6143 complement = del = squash = 0;
6144 while (strchr("cdsCU", *s)) {
6146 complement = OPpTRANS_COMPLEMENT;
6148 del = OPpTRANS_DELETE;
6150 squash = OPpTRANS_SQUASH;
6155 utf8 &= ~OPpTRANS_FROM_UTF;
6157 utf8 |= OPpTRANS_FROM_UTF;
6161 utf8 &= ~OPpTRANS_TO_UTF;
6163 utf8 |= OPpTRANS_TO_UTF;
6166 Perl_croak(aTHX_ "Too many /C and /U options");
6171 o->op_private = del|squash|complement|utf8;
6174 yylval.ival = OP_TRANS;
6179 S_scan_heredoc(pTHX_ register char *s)
6183 I32 op_type = OP_SCALAR;
6190 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6194 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6197 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6198 if (*peek && strchr("`'\"",*peek)) {
6201 s = delimcpy(d, e, s, PL_bufend, term, &len);
6211 if (!isALNUM_lazy_if(s,UTF))
6212 deprecate("bare << to mean <<\"\"");
6213 for (; isALNUM_lazy_if(s,UTF); s++) {
6218 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6219 Perl_croak(aTHX_ "Delimiter for here document is too long");
6222 len = d - PL_tokenbuf;
6223 #ifndef PERL_STRICT_CR
6224 d = strchr(s, '\r');
6228 while (s < PL_bufend) {
6234 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6243 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6248 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6249 herewas = newSVpvn(s,PL_bufend-s);
6251 s--, herewas = newSVpvn(s,d-s);
6252 s += SvCUR(herewas);
6254 tmpstr = NEWSV(87,79);
6255 sv_upgrade(tmpstr, SVt_PVIV);
6260 else if (term == '`') {
6261 op_type = OP_BACKTICK;
6262 SvIVX(tmpstr) = '\\';
6266 PL_multi_start = CopLINE(PL_curcop);
6267 PL_multi_open = PL_multi_close = '<';
6268 term = *PL_tokenbuf;
6269 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6270 char *bufptr = PL_sublex_info.super_bufptr;
6271 char *bufend = PL_sublex_info.super_bufend;
6272 char *olds = s - SvCUR(herewas);
6273 s = strchr(bufptr, '\n');
6277 while (s < bufend &&
6278 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6280 CopLINE_inc(PL_curcop);
6283 CopLINE_set(PL_curcop, PL_multi_start);
6284 missingterm(PL_tokenbuf);
6286 sv_setpvn(herewas,bufptr,d-bufptr+1);
6287 sv_setpvn(tmpstr,d+1,s-d);
6289 sv_catpvn(herewas,s,bufend-s);
6290 (void)strcpy(bufptr,SvPVX(herewas));
6297 while (s < PL_bufend &&
6298 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6300 CopLINE_inc(PL_curcop);
6302 if (s >= PL_bufend) {
6303 CopLINE_set(PL_curcop, PL_multi_start);
6304 missingterm(PL_tokenbuf);
6306 sv_setpvn(tmpstr,d+1,s-d);
6308 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6310 sv_catpvn(herewas,s,PL_bufend-s);
6311 sv_setsv(PL_linestr,herewas);
6312 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6313 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6316 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6317 while (s >= PL_bufend) { /* multiple line string? */
6319 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6320 CopLINE_set(PL_curcop, PL_multi_start);
6321 missingterm(PL_tokenbuf);
6323 CopLINE_inc(PL_curcop);
6324 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6325 #ifndef PERL_STRICT_CR
6326 if (PL_bufend - PL_linestart >= 2) {
6327 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6328 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6330 PL_bufend[-2] = '\n';
6332 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6334 else if (PL_bufend[-1] == '\r')
6335 PL_bufend[-1] = '\n';
6337 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6338 PL_bufend[-1] = '\n';
6340 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6341 SV *sv = NEWSV(88,0);
6343 sv_upgrade(sv, SVt_PVMG);
6344 sv_setsv(sv,PL_linestr);
6345 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6347 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6350 sv_catsv(PL_linestr,herewas);
6351 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6355 sv_catsv(tmpstr,PL_linestr);
6360 PL_multi_end = CopLINE(PL_curcop);
6361 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6362 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6363 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6365 SvREFCNT_dec(herewas);
6366 PL_lex_stuff = tmpstr;
6367 yylval.ival = op_type;
6372 takes: current position in input buffer
6373 returns: new position in input buffer
6374 side-effects: yylval and lex_op are set.
6379 <FH> read from filehandle
6380 <pkg::FH> read from package qualified filehandle
6381 <pkg'FH> read from package qualified filehandle
6382 <$fh> read from filehandle in $fh
6388 S_scan_inputsymbol(pTHX_ char *start)
6390 register char *s = start; /* current position in buffer */
6396 d = PL_tokenbuf; /* start of temp holding space */
6397 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6398 end = strchr(s, '\n');
6401 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6403 /* die if we didn't have space for the contents of the <>,
6404 or if it didn't end, or if we see a newline
6407 if (len >= sizeof PL_tokenbuf)
6408 Perl_croak(aTHX_ "Excessively long <> operator");
6410 Perl_croak(aTHX_ "Unterminated <> operator");
6415 Remember, only scalar variables are interpreted as filehandles by
6416 this code. Anything more complex (e.g., <$fh{$num}>) will be
6417 treated as a glob() call.
6418 This code makes use of the fact that except for the $ at the front,
6419 a scalar variable and a filehandle look the same.
6421 if (*d == '$' && d[1]) d++;
6423 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6424 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6427 /* If we've tried to read what we allow filehandles to look like, and
6428 there's still text left, then it must be a glob() and not a getline.
6429 Use scan_str to pull out the stuff between the <> and treat it
6430 as nothing more than a string.
6433 if (d - PL_tokenbuf != len) {
6434 yylval.ival = OP_GLOB;
6436 s = scan_str(start,FALSE,FALSE);
6438 Perl_croak(aTHX_ "Glob not terminated");
6442 /* we're in a filehandle read situation */
6445 /* turn <> into <ARGV> */
6447 (void)strcpy(d,"ARGV");
6449 /* if <$fh>, create the ops to turn the variable into a
6455 /* try to find it in the pad for this block, otherwise find
6456 add symbol table ops
6458 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6459 OP *o = newOP(OP_PADSV, 0);
6461 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6464 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6465 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6466 newUNOP(OP_RV2SV, 0,
6467 newGVOP(OP_GV, 0, gv)));
6469 PL_lex_op->op_flags |= OPf_SPECIAL;
6470 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6471 yylval.ival = OP_NULL;
6474 /* If it's none of the above, it must be a literal filehandle
6475 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6477 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6478 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6479 yylval.ival = OP_NULL;
6488 takes: start position in buffer
6489 keep_quoted preserve \ on the embedded delimiter(s)
6490 keep_delims preserve the delimiters around the string
6491 returns: position to continue reading from buffer
6492 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6493 updates the read buffer.
6495 This subroutine pulls a string out of the input. It is called for:
6496 q single quotes q(literal text)
6497 ' single quotes 'literal text'
6498 qq double quotes qq(interpolate $here please)
6499 " double quotes "interpolate $here please"
6500 qx backticks qx(/bin/ls -l)
6501 ` backticks `/bin/ls -l`
6502 qw quote words @EXPORT_OK = qw( func() $spam )
6503 m// regexp match m/this/
6504 s/// regexp substitute s/this/that/
6505 tr/// string transliterate tr/this/that/
6506 y/// string transliterate y/this/that/
6507 ($*@) sub prototypes sub foo ($)
6508 (stuff) sub attr parameters sub foo : attr(stuff)
6509 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6511 In most of these cases (all but <>, patterns and transliterate)
6512 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6513 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6514 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6517 It skips whitespace before the string starts, and treats the first
6518 character as the delimiter. If the delimiter is one of ([{< then
6519 the corresponding "close" character )]}> is used as the closing
6520 delimiter. It allows quoting of delimiters, and if the string has
6521 balanced delimiters ([{<>}]) it allows nesting.
6523 The lexer always reads these strings into lex_stuff, except in the
6524 case of the operators which take *two* arguments (s/// and tr///)
6525 when it checks to see if lex_stuff is full (presumably with the 1st
6526 arg to s or tr) and if so puts the string into lex_repl.
6531 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6534 SV *sv; /* scalar value: string */
6535 char *tmps; /* temp string, used for delimiter matching */
6536 register char *s = start; /* current position in the buffer */
6537 register char term; /* terminating character */
6538 register char *to; /* current position in the sv's data */
6539 I32 brackets = 1; /* bracket nesting level */
6540 bool has_utf = FALSE; /* is there any utf8 content? */
6542 /* skip space before the delimiter */
6546 /* mark where we are, in case we need to report errors */
6549 /* after skipping whitespace, the next character is the terminator */
6551 if ((term & 0x80) && UTF)
6554 /* mark where we are */
6555 PL_multi_start = CopLINE(PL_curcop);
6556 PL_multi_open = term;
6558 /* find corresponding closing delimiter */
6559 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6561 PL_multi_close = term;
6563 /* create a new SV to hold the contents. 87 is leak category, I'm
6564 assuming. 79 is the SV's initial length. What a random number. */
6566 sv_upgrade(sv, SVt_PVIV);
6568 (void)SvPOK_only(sv); /* validate pointer */
6570 /* move past delimiter and try to read a complete string */
6572 sv_catpvn(sv, s, 1);
6575 /* extend sv if need be */
6576 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6577 /* set 'to' to the next character in the sv's string */
6578 to = SvPVX(sv)+SvCUR(sv);
6580 /* if open delimiter is the close delimiter read unbridle */
6581 if (PL_multi_open == PL_multi_close) {
6582 for (; s < PL_bufend; s++,to++) {
6583 /* embedded newlines increment the current line number */
6584 if (*s == '\n' && !PL_rsfp)
6585 CopLINE_inc(PL_curcop);
6586 /* handle quoted delimiters */
6587 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6588 if (!keep_quoted && s[1] == term)
6590 /* any other quotes are simply copied straight through */
6594 /* terminate when run out of buffer (the for() condition), or
6595 have found the terminator */
6596 else if (*s == term)
6598 else if (!has_utf && (*s & 0x80) && UTF)
6604 /* if the terminator isn't the same as the start character (e.g.,
6605 matched brackets), we have to allow more in the quoting, and
6606 be prepared for nested brackets.
6609 /* read until we run out of string, or we find the terminator */
6610 for (; s < PL_bufend; s++,to++) {
6611 /* embedded newlines increment the line count */
6612 if (*s == '\n' && !PL_rsfp)
6613 CopLINE_inc(PL_curcop);
6614 /* backslashes can escape the open or closing characters */
6615 if (*s == '\\' && s+1 < PL_bufend) {
6617 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6622 /* allow nested opens and closes */
6623 else if (*s == PL_multi_close && --brackets <= 0)
6625 else if (*s == PL_multi_open)
6627 else if (!has_utf && (*s & 0x80) && UTF)
6632 /* terminate the copied string and update the sv's end-of-string */
6634 SvCUR_set(sv, to - SvPVX(sv));
6637 * this next chunk reads more into the buffer if we're not done yet
6641 break; /* handle case where we are done yet :-) */
6643 #ifndef PERL_STRICT_CR
6644 if (to - SvPVX(sv) >= 2) {
6645 if ((to[-2] == '\r' && to[-1] == '\n') ||
6646 (to[-2] == '\n' && to[-1] == '\r'))
6650 SvCUR_set(sv, to - SvPVX(sv));
6652 else if (to[-1] == '\r')
6655 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6659 /* if we're out of file, or a read fails, bail and reset the current
6660 line marker so we can report where the unterminated string began
6663 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6665 CopLINE_set(PL_curcop, PL_multi_start);
6668 /* we read a line, so increment our line counter */
6669 CopLINE_inc(PL_curcop);
6671 /* update debugger info */
6672 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6673 SV *sv = NEWSV(88,0);
6675 sv_upgrade(sv, SVt_PVMG);
6676 sv_setsv(sv,PL_linestr);
6677 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6680 /* having changed the buffer, we must update PL_bufend */
6681 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6684 /* at this point, we have successfully read the delimited string */
6687 sv_catpvn(sv, s, 1);
6690 PL_multi_end = CopLINE(PL_curcop);
6693 /* if we allocated too much space, give some back */
6694 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6695 SvLEN_set(sv, SvCUR(sv) + 1);
6696 Renew(SvPVX(sv), SvLEN(sv), char);
6699 /* decide whether this is the first or second quoted string we've read
6712 takes: pointer to position in buffer
6713 returns: pointer to new position in buffer
6714 side-effects: builds ops for the constant in yylval.op
6716 Read a number in any of the formats that Perl accepts:
6718 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6719 [\d_]+(\.[\d_]*)?[Ee](\d+)
6721 Underbars (_) are allowed in decimal numbers. If -w is on,
6722 underbars before a decimal point must be at three digit intervals.
6724 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6727 If it reads a number without a decimal point or an exponent, it will
6728 try converting the number to an integer and see if it can do so
6729 without loss of precision.
6733 Perl_scan_num(pTHX_ char *start)
6735 register char *s = start; /* current position in buffer */
6736 register char *d; /* destination in temp buffer */
6737 register char *e; /* end of temp buffer */
6738 NV nv; /* number read, as a double */
6739 SV *sv = Nullsv; /* place to put the converted number */
6740 bool floatit; /* boolean: int or float? */
6741 char *lastub = 0; /* position of last underbar */
6742 static char number_too_long[] = "Number too long";
6744 /* We use the first character to decide what type of number this is */
6748 Perl_croak(aTHX_ "panic: scan_num");
6750 /* if it starts with a 0, it could be an octal number, a decimal in
6751 0.13 disguise, or a hexadecimal number, or a binary number. */
6755 u holds the "number so far"
6756 shift the power of 2 of the base
6757 (hex == 4, octal == 3, binary == 1)
6758 overflowed was the number more than we can hold?
6760 Shift is used when we add a digit. It also serves as an "are
6761 we in octal/hex/binary?" indicator to disallow hex characters
6768 bool overflowed = FALSE;
6769 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6770 static char* bases[5] = { "", "binary", "", "octal",
6772 static char* Bases[5] = { "", "Binary", "", "Octal",
6774 static char *maxima[5] = { "",
6775 "0b11111111111111111111111111111111",
6779 char *base, *Base, *max;
6785 } else if (s[1] == 'b') {
6789 /* check for a decimal in disguise */
6790 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6792 /* so it must be octal */
6796 base = bases[shift];
6797 Base = Bases[shift];
6798 max = maxima[shift];
6800 /* read the rest of the number */
6802 /* x is used in the overflow test,
6803 b is the digit we're adding on. */
6808 /* if we don't mention it, we're done */
6817 /* 8 and 9 are not octal */
6820 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6824 case '2': case '3': case '4':
6825 case '5': case '6': case '7':
6827 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6831 b = *s++ & 15; /* ASCII digit -> value of digit */
6835 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6836 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6837 /* make sure they said 0x */
6842 /* Prepare to put the digit we have onto the end
6843 of the number so far. We check for overflows.
6848 x = u << shift; /* make room for the digit */
6850 if ((x >> shift) != u
6851 && !(PL_hints & HINT_NEW_BINARY)) {
6855 if (ckWARN_d(WARN_OVERFLOW))
6856 Perl_warner(aTHX_ WARN_OVERFLOW,
6857 "Integer overflow in %s number",
6860 u = x | b; /* add the digit to the end */
6863 n *= nvshift[shift];
6864 /* If an NV has not enough bits in its
6865 * mantissa to represent an UV this summing of
6866 * small low-order numbers is a waste of time
6867 * (because the NV cannot preserve the
6868 * low-order bits anyway): we could just
6869 * remember when did we overflow and in the
6870 * end just multiply n by the right
6878 /* if we get here, we had success: make a scalar value from
6885 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6886 Perl_warner(aTHX_ WARN_PORTABLE,
6887 "%s number > %s non-portable",
6894 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6895 Perl_warner(aTHX_ WARN_PORTABLE,
6896 "%s number > %s non-portable",
6901 if (PL_hints & HINT_NEW_BINARY)
6902 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6907 handle decimal numbers.
6908 we're also sent here when we read a 0 as the first digit
6910 case '1': case '2': case '3': case '4': case '5':
6911 case '6': case '7': case '8': case '9': case '.':
6914 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
6917 /* read next group of digits and _ and copy into d */
6918 while (isDIGIT(*s) || *s == '_') {
6919 /* skip underscores, checking for misplaced ones
6923 dTHR; /* only for ckWARN */
6924 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
6925 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6929 /* check for end of fixed-length buffer */
6931 Perl_croak(aTHX_ number_too_long);
6932 /* if we're ok, copy the character */
6937 /* final misplaced underbar check */
6938 if (lastub && s - lastub != 3) {
6940 if (ckWARN(WARN_SYNTAX))
6941 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6944 /* read a decimal portion if there is one. avoid
6945 3..5 being interpreted as the number 3. followed
6948 if (*s == '.' && s[1] != '.') {
6952 /* copy, ignoring underbars, until we run out of
6953 digits. Note: no misplaced underbar checks!
6955 for (; isDIGIT(*s) || *s == '_'; s++) {
6956 /* fixed length buffer check */
6958 Perl_croak(aTHX_ number_too_long);
6962 if (*s == '.' && isDIGIT(s[1])) {
6963 /* oops, it's really a v-string, but without the "v" */
6969 /* read exponent part, if present */
6970 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
6974 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
6975 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
6977 /* allow positive or negative exponent */
6978 if (*s == '+' || *s == '-')
6981 /* read digits of exponent (no underbars :-) */
6982 while (isDIGIT(*s)) {
6984 Perl_croak(aTHX_ number_too_long);
6989 /* terminate the string */
6992 /* make an sv from the string */
6995 #if defined(Strtol) && defined(Strtoul)
6998 strtol/strtoll sets errno to ERANGE if the number is too big
6999 for an integer. We try to do an integer conversion first
7000 if no characters indicating "float" have been found.
7007 if (*PL_tokenbuf == '-')
7008 iv = Strtol(PL_tokenbuf, (char**)NULL, 10);
7010 uv = Strtoul(PL_tokenbuf, (char**)NULL, 10);
7012 floatit = TRUE; /* Probably just too large. */
7013 else if (*PL_tokenbuf == '-')
7015 else if (uv <= IV_MAX)
7016 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7021 nv = Atof(PL_tokenbuf);
7026 No working strtou?ll?.
7028 Unfortunately atol() doesn't do range checks (returning
7029 LONG_MIN/LONG_MAX, and setting errno to ERANGE on overflows)
7030 everywhere [1], so we cannot use use atol() (or atoll()).
7031 If we could, they would be used, as Atol(), very much like
7032 Strtol() and Strtoul() are used above.
7034 [1] XXX Configure test needed to check for atol()
7035 (and atoll()) overflow behaviour XXX
7039 We need to do this the hard way. */
7041 nv = Atof(PL_tokenbuf);
7043 /* See if we can make do with an integer value without loss of
7044 precision. We use U_V to cast to a UV, because some
7045 compilers have issues. Then we try casting it back and see
7046 if it was the same [1]. We only do this if we know we
7047 specifically read an integer. If floatit is true, then we
7048 don't need to do the conversion at all.
7050 [1] Note that this is lossy if our NVs cannot preserve our
7051 UVs. There are metaconfig defines NV_PRESERVES_UV (a boolean)
7052 and NV_PRESERVES_UV_BITS (a number), but in general we really
7053 do hope all such potentially lossy platforms have strtou?ll?
7054 to do a lossless IV/UV conversion.
7056 Maybe could do some tricks with DBL_DIG, LDBL_DIG and
7057 DBL_MANT_DIG and LDBL_MANT_DIG (these are already available
7058 as NV_DIG and NV_MANT_DIG)?
7064 if (!floatit && (NV)uv == nv) {
7066 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7074 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7075 (PL_hints & HINT_NEW_INTEGER) )
7076 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7077 (floatit ? "float" : "integer"),
7081 /* if it starts with a v, it could be a v-string */
7087 while (isDIGIT(*pos) || *pos == '_')
7089 if (!isALPHA(*pos)) {
7091 U8 tmpbuf[UTF8_MAXLEN];
7094 s++; /* get past 'v' */
7097 sv_setpvn(sv, "", 0);
7100 if (*s == '0' && isDIGIT(s[1]))
7101 yyerror("Octal number in vector unsupported");
7104 /* this is atoi() that tolerates underscores */
7107 while (--end >= s) {
7112 rev += (*end - '0') * mult;
7114 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
7115 Perl_warner(aTHX_ WARN_OVERFLOW,
7116 "Integer overflow in decimal number");
7119 tmpend = uv_to_utf8(tmpbuf, rev);
7120 utf8 = utf8 || rev > 127;
7121 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
7122 if (*pos == '.' && isDIGIT(pos[1]))
7128 while (isDIGIT(*pos) || *pos == '_')
7136 sv_utf8_downgrade(sv, TRUE);
7143 /* make the op for the constant and return */
7146 yylval.opval = newSVOP(OP_CONST, 0, sv);
7148 yylval.opval = Nullop;
7154 S_scan_formline(pTHX_ register char *s)
7159 SV *stuff = newSVpvn("",0);
7160 bool needargs = FALSE;
7163 if (*s == '.' || *s == '}') {
7165 #ifdef PERL_STRICT_CR
7166 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7168 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7170 if (*t == '\n' || t == PL_bufend)
7173 if (PL_in_eval && !PL_rsfp) {
7174 eol = strchr(s,'\n');
7179 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7181 for (t = s; t < eol; t++) {
7182 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7184 goto enough; /* ~~ must be first line in formline */
7186 if (*t == '@' || *t == '^')
7189 sv_catpvn(stuff, s, eol-s);
7190 #ifndef PERL_STRICT_CR
7191 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7192 char *end = SvPVX(stuff) + SvCUR(stuff);
7201 s = filter_gets(PL_linestr, PL_rsfp, 0);
7202 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7203 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7206 yyerror("Format not terminated");
7216 PL_lex_state = LEX_NORMAL;
7217 PL_nextval[PL_nexttoke].ival = 0;
7221 PL_lex_state = LEX_FORMLINE;
7222 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7224 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7228 SvREFCNT_dec(stuff);
7229 PL_lex_formbrack = 0;
7240 PL_cshlen = strlen(PL_cshname);
7245 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7248 I32 oldsavestack_ix = PL_savestack_ix;
7249 CV* outsidecv = PL_compcv;
7253 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7255 SAVEI32(PL_subline);
7256 save_item(PL_subname);
7259 SAVESPTR(PL_comppad_name);
7260 SAVESPTR(PL_compcv);
7261 SAVEI32(PL_comppad_name_fill);
7262 SAVEI32(PL_min_intro_pending);
7263 SAVEI32(PL_max_intro_pending);
7264 SAVEI32(PL_pad_reset_pending);
7266 PL_compcv = (CV*)NEWSV(1104,0);
7267 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7268 CvFLAGS(PL_compcv) |= flags;
7270 PL_comppad = newAV();
7271 av_push(PL_comppad, Nullsv);
7272 PL_curpad = AvARRAY(PL_comppad);
7273 PL_comppad_name = newAV();
7274 PL_comppad_name_fill = 0;
7275 PL_min_intro_pending = 0;
7277 PL_subline = CopLINE(PL_curcop);
7279 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7280 PL_curpad[0] = (SV*)newAV();
7281 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7282 #endif /* USE_THREADS */
7284 comppadlist = newAV();
7285 AvREAL_off(comppadlist);
7286 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7287 av_store(comppadlist, 1, (SV*)PL_comppad);
7289 CvPADLIST(PL_compcv) = comppadlist;
7290 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7292 CvOWNER(PL_compcv) = 0;
7293 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7294 MUTEX_INIT(CvMUTEXP(PL_compcv));
7295 #endif /* USE_THREADS */
7297 return oldsavestack_ix;
7301 Perl_yywarn(pTHX_ char *s)
7304 PL_in_eval |= EVAL_WARNONLY;
7306 PL_in_eval &= ~EVAL_WARNONLY;
7311 Perl_yyerror(pTHX_ char *s)
7315 char *context = NULL;
7319 if (!yychar || (yychar == ';' && !PL_rsfp))
7321 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7322 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7323 while (isSPACE(*PL_oldoldbufptr))
7325 context = PL_oldoldbufptr;
7326 contlen = PL_bufptr - PL_oldoldbufptr;
7328 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7329 PL_oldbufptr != PL_bufptr) {
7330 while (isSPACE(*PL_oldbufptr))
7332 context = PL_oldbufptr;
7333 contlen = PL_bufptr - PL_oldbufptr;
7335 else if (yychar > 255)
7336 where = "next token ???";
7337 #ifdef USE_PURE_BISON
7338 /* GNU Bison sets the value -2 */
7339 else if (yychar == -2) {
7341 else if ((yychar & 127) == 127) {
7343 if (PL_lex_state == LEX_NORMAL ||
7344 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7345 where = "at end of line";
7346 else if (PL_lex_inpat)
7347 where = "within pattern";
7349 where = "within string";
7352 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7354 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7355 else if (isPRINT_LC(yychar))
7356 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7358 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7359 where = SvPVX(where_sv);
7361 msg = sv_2mortal(newSVpv(s, 0));
7362 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7363 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7365 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7367 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7368 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7369 Perl_sv_catpvf(aTHX_ msg,
7370 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7371 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7374 if (PL_in_eval & EVAL_WARNONLY)
7375 Perl_warn(aTHX_ "%"SVf, msg);
7378 if (PL_error_count >= 10) {
7379 if (PL_in_eval && SvCUR(ERRSV))
7380 Perl_croak(aTHX_ "%_%s has too many errors.\n",
7381 ERRSV, CopFILE(PL_curcop));
7383 Perl_croak(aTHX_ "%s has too many errors.\n",
7384 CopFILE(PL_curcop));
7387 PL_in_my_stash = Nullhv;
7398 * Restore a source filter.
7402 restore_rsfp(pTHXo_ void *f)
7404 PerlIO *fp = (PerlIO*)f;
7406 if (PL_rsfp == PerlIO_stdin())
7407 PerlIO_clearerr(PL_rsfp);
7408 else if (PL_rsfp && (PL_rsfp != fp))
7409 PerlIO_close(PL_rsfp);