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 /* LEX_* are values for PL_lex_state, the state of the lexer.
43 * They are arranged oddly so that the guard on the switch statement
44 * can get by with a single comparison (if the compiler is smart enough).
47 /* #define LEX_NOTPARSING 11 is done in perl.h. */
50 #define LEX_INTERPNORMAL 9
51 #define LEX_INTERPCASEMOD 8
52 #define LEX_INTERPPUSH 7
53 #define LEX_INTERPSTART 6
54 #define LEX_INTERPEND 5
55 #define LEX_INTERPENDMAYBE 4
56 #define LEX_INTERPCONCAT 3
57 #define LEX_INTERPCONST 2
58 #define LEX_FORMLINE 1
59 #define LEX_KNOWNEXT 0
61 /* XXX If this causes problems, set i_unistd=undef in the hint file. */
63 # include <unistd.h> /* Needed for execv() */
72 YYSTYPE* yylval_pointer = NULL;
73 int* yychar_pointer = NULL;
76 # define yylval (*yylval_pointer)
77 # define yychar (*yychar_pointer)
78 # define PERL_YYLEX_PARAM yylval_pointer,yychar_pointer
80 # define yylex() Perl_yylex(aTHX_ yylval_pointer, yychar_pointer)
85 /* CLINE is a macro that ensures PL_copline has a sane value */
90 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
93 * Convenience functions to return different tokens and prime the
94 * lexer for the next token. They all take an argument.
96 * TOKEN : generic token (used for '(', DOLSHARP, etc)
97 * OPERATOR : generic operator
98 * AOPERATOR : assignment operator
99 * PREBLOCK : beginning the block after an if, while, foreach, ...
100 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
101 * PREREF : *EXPR where EXPR is not a simple identifier
102 * TERM : expression term
103 * LOOPX : loop exiting command (goto, last, dump, etc)
104 * FTST : file test operator
105 * FUN0 : zero-argument function
106 * FUN1 : not used, except for not, which isn't a UNIOP
107 * BOop : bitwise or or xor
109 * SHop : shift operator
110 * PWop : power operator
111 * PMop : pattern-matching operator
112 * Aop : addition-level operator
113 * Mop : multiplication-level operator
114 * Eop : equality-testing operator
115 * Rop : relational operator <= != gt
117 * Also see LOP and lop() below.
120 #define TOKEN(retval) return (PL_bufptr = s,(int)retval)
121 #define OPERATOR(retval) return (PL_expect = XTERM,PL_bufptr = s,(int)retval)
122 #define AOPERATOR(retval) return ao((PL_expect = XTERM,PL_bufptr = s,(int)retval))
123 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
124 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
125 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s,(int)retval)
126 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR,PL_bufptr = s,(int)retval)
127 #define LOOPX(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
128 #define FTST(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)UNIOP)
129 #define FUN0(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
130 #define FUN1(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
131 #define BOop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
132 #define BAop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
133 #define SHop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
134 #define PWop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
135 #define PMop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
136 #define Aop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
137 #define Mop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
138 #define Eop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
139 #define Rop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)RELOP)
141 /* This bit of chicanery makes a unary function followed by
142 * a parenthesis into a function with one argument, highest precedence.
144 #define UNI(f) return(yylval.ival = f, \
147 PL_last_uni = PL_oldbufptr, \
148 PL_last_lop_op = f, \
149 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
151 #define UNIBRACK(f) return(yylval.ival = f, \
153 PL_last_uni = PL_oldbufptr, \
154 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
156 /* grandfather return to old style */
157 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
162 * This subroutine detects &&= and ||= and turns an ANDAND or OROR
163 * into an OP_ANDASSIGN or OP_ORASSIGN
167 S_ao(pTHX_ int toketype)
169 if (*PL_bufptr == '=') {
171 if (toketype == ANDAND)
172 yylval.ival = OP_ANDASSIGN;
173 else if (toketype == OROR)
174 yylval.ival = OP_ORASSIGN;
182 * When Perl expects an operator and finds something else, no_op
183 * prints the warning. It always prints "<something> found where
184 * operator expected. It prints "Missing semicolon on previous line?"
185 * if the surprise occurs at the start of the line. "do you need to
186 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
187 * where the compiler doesn't know if foo is a method call or a function.
188 * It prints "Missing operator before end of line" if there's nothing
189 * after the missing operator, or "... before <...>" if there is something
190 * after the missing operator.
194 S_no_op(pTHX_ char *what, char *s)
196 char *oldbp = PL_bufptr;
197 bool is_first = (PL_oldbufptr == PL_linestart);
205 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
207 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
208 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
210 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
211 if (t < PL_bufptr && isSPACE(*t))
212 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
213 t - PL_oldoldbufptr, PL_oldoldbufptr);
216 Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
222 * Complain about missing quote/regexp/heredoc terminator.
223 * If it's called with (char *)NULL then it cauterizes the line buffer.
224 * If we're in a delimited string and the delimiter is a control
225 * character, it's reformatted into a two-char sequence like ^C.
230 S_missingterm(pTHX_ char *s)
235 char *nl = strrchr(s,'\n');
241 iscntrl(PL_multi_close)
243 PL_multi_close < 32 || PL_multi_close == 127
247 tmpbuf[1] = toCTRL(PL_multi_close);
253 *tmpbuf = PL_multi_close;
257 q = strchr(s,'"') ? '\'' : '"';
258 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
266 Perl_deprecate(pTHX_ char *s)
269 if (ckWARN(WARN_DEPRECATED))
270 Perl_warner(aTHX_ WARN_DEPRECATED, "Use of %s is deprecated", s);
275 * Deprecate a comma-less variable list.
281 deprecate("comma-less variable list");
285 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
286 * utf16-to-utf8-reversed.
289 #ifdef PERL_CR_FILTER
293 register char *s = SvPVX(sv);
294 register char *e = s + SvCUR(sv);
295 /* outer loop optimized to do nothing if there are no CR-LFs */
297 if (*s++ == '\r' && *s == '\n') {
298 /* hit a CR-LF, need to copy the rest */
299 register char *d = s - 1;
302 if (*s == '\r' && s[1] == '\n')
313 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
315 I32 count = FILTER_READ(idx+1, sv, maxlen);
316 if (count > 0 && !maxlen)
324 S_utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
326 I32 count = FILTER_READ(idx+1, sv, maxlen);
330 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
331 tend = utf16_to_utf8((U16*)SvPVX(sv), tmps, SvCUR(sv));
332 sv_usepvn(sv, (char*)tmps, tend - tmps);
338 S_utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
340 I32 count = FILTER_READ(idx+1, sv, maxlen);
344 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
345 tend = utf16_to_utf8_reversed((U16*)SvPVX(sv), tmps, SvCUR(sv));
346 sv_usepvn(sv, (char*)tmps, tend - tmps);
354 * Initialize variables. Uses the Perl save_stack to save its state (for
355 * recursive calls to the parser).
359 Perl_lex_start(pTHX_ SV *line)
365 SAVEI32(PL_lex_dojoin);
366 SAVEI32(PL_lex_brackets);
367 SAVEI32(PL_lex_casemods);
368 SAVEI32(PL_lex_starts);
369 SAVEI32(PL_lex_state);
370 SAVEVPTR(PL_lex_inpat);
371 SAVEI32(PL_lex_inwhat);
372 if (PL_lex_state == LEX_KNOWNEXT) {
373 I32 toke = PL_nexttoke;
374 while (--toke >= 0) {
375 SAVEI32(PL_nexttype[toke]);
376 SAVEVPTR(PL_nextval[toke]);
378 SAVEI32(PL_nexttoke);
381 SAVECOPLINE(PL_curcop);
384 SAVEPPTR(PL_oldbufptr);
385 SAVEPPTR(PL_oldoldbufptr);
386 SAVEPPTR(PL_linestart);
387 SAVESPTR(PL_linestr);
388 SAVEPPTR(PL_lex_brackstack);
389 SAVEPPTR(PL_lex_casestack);
390 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
391 SAVESPTR(PL_lex_stuff);
392 SAVEI32(PL_lex_defer);
393 SAVEI32(PL_sublex_info.sub_inwhat);
394 SAVESPTR(PL_lex_repl);
396 SAVEINT(PL_lex_expect);
398 PL_lex_state = LEX_NORMAL;
402 New(899, PL_lex_brackstack, 120, char);
403 New(899, PL_lex_casestack, 12, char);
404 SAVEFREEPV(PL_lex_brackstack);
405 SAVEFREEPV(PL_lex_casestack);
407 *PL_lex_casestack = '\0';
410 PL_lex_stuff = Nullsv;
411 PL_lex_repl = Nullsv;
414 PL_sublex_info.sub_inwhat = 0;
416 if (SvREADONLY(PL_linestr))
417 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
418 s = SvPV(PL_linestr, len);
419 if (len && s[len-1] != ';') {
420 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
421 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
422 sv_catpvn(PL_linestr, "\n;", 2);
424 SvTEMP_off(PL_linestr);
425 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
426 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
428 PL_rs = newSVpvn("\n", 1);
434 * Finalizer for lexing operations. Must be called when the parser is
435 * done with the lexer.
441 PL_doextract = FALSE;
446 * This subroutine has nothing to do with tilting, whether at windmills
447 * or pinball tables. Its name is short for "increment line". It
448 * increments the current line number in CopLINE(PL_curcop) and checks
449 * to see whether the line starts with a comment of the form
450 * # line 500 "foo.pm"
451 * If so, it sets the current line number and file to the values in the comment.
455 S_incline(pTHX_ char *s)
463 CopLINE_inc(PL_curcop);
466 while (*s == ' ' || *s == '\t') s++;
467 if (strnEQ(s, "line", 4))
471 if (*s == ' ' || *s == '\t')
475 while (*s == ' ' || *s == '\t') s++;
481 while (*s == ' ' || *s == '\t')
483 if (*s == '"' && (t = strchr(s+1, '"'))) {
488 for (t = s; !isSPACE(*t); t++) ;
491 while (*e == ' ' || *e == '\t' || *e == '\r' || *e == '\f')
493 if (*e != '\n' && *e != '\0')
494 return; /* false alarm */
499 CopFILE_set(PL_curcop, s);
501 CopLINE_set(PL_curcop, atoi(n)-1);
506 * Called to gobble the appropriate amount and type of whitespace.
507 * Skips comments as well.
511 S_skipspace(pTHX_ register char *s)
514 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
515 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
521 SSize_t oldprevlen, oldoldprevlen;
522 SSize_t oldloplen, oldunilen;
523 while (s < PL_bufend && isSPACE(*s)) {
524 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
529 if (s < PL_bufend && *s == '#') {
530 while (s < PL_bufend && *s != '\n')
534 if (PL_in_eval && !PL_rsfp) {
541 /* only continue to recharge the buffer if we're at the end
542 * of the buffer, we're not reading from a source filter, and
543 * we're in normal lexing mode
545 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
546 PL_lex_state == LEX_FORMLINE)
549 /* try to recharge the buffer */
550 if ((s = filter_gets(PL_linestr, PL_rsfp,
551 (prevlen = SvCUR(PL_linestr)))) == Nullch)
553 /* end of file. Add on the -p or -n magic */
554 if (PL_minus_n || PL_minus_p) {
555 sv_setpv(PL_linestr,PL_minus_p ?
556 ";}continue{print or die qq(-p destination: $!\\n)" :
558 sv_catpv(PL_linestr,";}");
559 PL_minus_n = PL_minus_p = 0;
562 sv_setpv(PL_linestr,";");
564 /* reset variables for next time we lex */
565 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
567 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
569 /* Close the filehandle. Could be from -P preprocessor,
570 * STDIN, or a regular file. If we were reading code from
571 * STDIN (because the commandline held no -e or filename)
572 * then we don't close it, we reset it so the code can
573 * read from STDIN too.
576 if (PL_preprocess && !PL_in_eval)
577 (void)PerlProc_pclose(PL_rsfp);
578 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
579 PerlIO_clearerr(PL_rsfp);
581 (void)PerlIO_close(PL_rsfp);
586 /* not at end of file, so we only read another line */
587 /* make corresponding updates to old pointers, for yyerror() */
588 oldprevlen = PL_oldbufptr - PL_bufend;
589 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
591 oldunilen = PL_last_uni - PL_bufend;
593 oldloplen = PL_last_lop - PL_bufend;
594 PL_linestart = PL_bufptr = s + prevlen;
595 PL_bufend = s + SvCUR(PL_linestr);
597 PL_oldbufptr = s + oldprevlen;
598 PL_oldoldbufptr = s + oldoldprevlen;
600 PL_last_uni = s + oldunilen;
602 PL_last_lop = s + oldloplen;
605 /* debugger active and we're not compiling the debugger code,
606 * so store the line into the debugger's array of lines
608 if (PERLDB_LINE && PL_curstash != PL_debstash) {
609 SV *sv = NEWSV(85,0);
611 sv_upgrade(sv, SVt_PVMG);
612 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
613 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
620 * Check the unary operators to ensure there's no ambiguity in how they're
621 * used. An ambiguous piece of code would be:
623 * This doesn't mean rand() + 5. Because rand() is a unary operator,
624 * the +5 is its argument.
634 if (PL_oldoldbufptr != PL_last_uni)
636 while (isSPACE(*PL_last_uni))
638 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
639 if ((t = strchr(s, '(')) && t < PL_bufptr)
641 if (ckWARN_d(WARN_AMBIGUOUS)){
644 Perl_warner(aTHX_ WARN_AMBIGUOUS,
645 "Warning: Use of \"%s\" without parens is ambiguous",
651 /* workaround to replace the UNI() macro with a function. Only the
652 * hints/uts.sh file mentions this. Other comments elsewhere in the
653 * source indicate Microport Unix might need it too.
659 #define UNI(f) return uni(f,s)
662 S_uni(pTHX_ I32 f, char *s)
667 PL_last_uni = PL_oldbufptr;
678 #endif /* CRIPPLED_CC */
681 * LOP : macro to build a list operator. Its behaviour has been replaced
682 * with a subroutine, S_lop() for which LOP is just another name.
685 #define LOP(f,x) return lop(f,x,s)
689 * Build a list operator (or something that might be one). The rules:
690 * - if we have a next token, then it's a list operator [why?]
691 * - if the next thing is an opening paren, then it's a function
692 * - else it's a list operator
696 S_lop(pTHX_ I32 f, int x, char *s)
703 PL_last_lop = PL_oldbufptr;
718 * When the lexer realizes it knows the next token (for instance,
719 * it is reordering tokens for the parser) then it can call S_force_next
720 * to know what token to return the next time the lexer is called. Caller
721 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
722 * handles the token correctly.
726 S_force_next(pTHX_ I32 type)
728 PL_nexttype[PL_nexttoke] = type;
730 if (PL_lex_state != LEX_KNOWNEXT) {
731 PL_lex_defer = PL_lex_state;
732 PL_lex_expect = PL_expect;
733 PL_lex_state = LEX_KNOWNEXT;
739 * When the lexer knows the next thing is a word (for instance, it has
740 * just seen -> and it knows that the next char is a word char, then
741 * it calls S_force_word to stick the next word into the PL_next lookahead.
744 * char *start : buffer position (must be within PL_linestr)
745 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
746 * int check_keyword : if true, Perl checks to make sure the word isn't
747 * a keyword (do this if the word is a label, e.g. goto FOO)
748 * int allow_pack : if true, : characters will also be allowed (require,
750 * int allow_initial_tick : used by the "sub" lexer only.
754 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
759 start = skipspace(start);
761 if (isIDFIRST_lazy_if(s,UTF) ||
762 (allow_pack && *s == ':') ||
763 (allow_initial_tick && *s == '\'') )
765 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
766 if (check_keyword && keyword(PL_tokenbuf, len))
768 if (token == METHOD) {
773 PL_expect = XOPERATOR;
776 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
777 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
785 * Called when the lexer wants $foo *foo &foo etc, but the program
786 * text only contains the "foo" portion. The first argument is a pointer
787 * to the "foo", and the second argument is the type symbol to prefix.
788 * Forces the next token to be a "WORD".
789 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
793 S_force_ident(pTHX_ register char *s, int kind)
796 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
797 PL_nextval[PL_nexttoke].opval = o;
800 dTHR; /* just for in_eval */
801 o->op_private = OPpCONST_ENTERED;
802 /* XXX see note in pp_entereval() for why we forgo typo
803 warnings if the symbol must be introduced in an eval.
805 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
806 kind == '$' ? SVt_PV :
807 kind == '@' ? SVt_PVAV :
808 kind == '%' ? SVt_PVHV :
816 Perl_str_to_version(pTHX_ SV *sv)
821 char *start = SvPVx(sv,len);
822 bool utf = SvUTF8(sv) ? TRUE : FALSE;
823 char *end = start + len;
824 while (start < end) {
828 n = utf8_to_uv((U8*)start, &skip);
833 retval += ((NV)n)/nshift;
842 * Forces the next token to be a version number.
846 S_force_version(pTHX_ char *s)
848 OP *version = Nullop;
857 for (; isDIGIT(*d) || *d == '_' || *d == '.'; d++);
858 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
861 version = yylval.opval;
862 ver = cSVOPx(version)->op_sv;
863 if (SvPOK(ver) && !SvNIOK(ver)) {
864 (void)SvUPGRADE(ver, SVt_PVNV);
865 SvNVX(ver) = str_to_version(ver);
866 SvNOK_on(ver); /* hint that it is a version */
871 /* NOTE: The parser sees the package name and the VERSION swapped */
872 PL_nextval[PL_nexttoke].opval = version;
880 * Tokenize a quoted string passed in as an SV. It finds the next
881 * chunk, up to end of string or a backslash. It may make a new
882 * SV containing that chunk (if HINT_NEW_STRING is on). It also
887 S_tokeq(pTHX_ SV *sv)
898 s = SvPV_force(sv, len);
899 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
902 while (s < send && *s != '\\')
907 if ( PL_hints & HINT_NEW_STRING )
908 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
911 if (s + 1 < send && (s[1] == '\\'))
912 s++; /* all that, just for this */
917 SvCUR_set(sv, d - SvPVX(sv));
919 if ( PL_hints & HINT_NEW_STRING )
920 return new_constant(NULL, 0, "q", sv, pv, "q");
925 * Now come three functions related to double-quote context,
926 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
927 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
928 * interact with PL_lex_state, and create fake ( ... ) argument lists
929 * to handle functions and concatenation.
930 * They assume that whoever calls them will be setting up a fake
931 * join call, because each subthing puts a ',' after it. This lets
934 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
936 * (I'm not sure whether the spurious commas at the end of lcfirst's
937 * arguments and join's arguments are created or not).
942 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
944 * Pattern matching will set PL_lex_op to the pattern-matching op to
945 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
947 * OP_CONST and OP_READLINE are easy--just make the new op and return.
949 * Everything else becomes a FUNC.
951 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
952 * had an OP_CONST or OP_READLINE). This just sets us up for a
953 * call to S_sublex_push().
959 register I32 op_type = yylval.ival;
961 if (op_type == OP_NULL) {
962 yylval.opval = PL_lex_op;
966 if (op_type == OP_CONST || op_type == OP_READLINE) {
967 SV *sv = tokeq(PL_lex_stuff);
969 if (SvTYPE(sv) == SVt_PVIV) {
970 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
976 nsv = newSVpvn(p, len);
980 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
981 PL_lex_stuff = Nullsv;
985 PL_sublex_info.super_state = PL_lex_state;
986 PL_sublex_info.sub_inwhat = op_type;
987 PL_sublex_info.sub_op = PL_lex_op;
988 PL_lex_state = LEX_INTERPPUSH;
992 yylval.opval = PL_lex_op;
1002 * Create a new scope to save the lexing state. The scope will be
1003 * ended in S_sublex_done. Returns a '(', starting the function arguments
1004 * to the uc, lc, etc. found before.
1005 * Sets PL_lex_state to LEX_INTERPCONCAT.
1014 PL_lex_state = PL_sublex_info.super_state;
1015 SAVEI32(PL_lex_dojoin);
1016 SAVEI32(PL_lex_brackets);
1017 SAVEI32(PL_lex_casemods);
1018 SAVEI32(PL_lex_starts);
1019 SAVEI32(PL_lex_state);
1020 SAVEVPTR(PL_lex_inpat);
1021 SAVEI32(PL_lex_inwhat);
1022 SAVECOPLINE(PL_curcop);
1023 SAVEPPTR(PL_bufptr);
1024 SAVEPPTR(PL_oldbufptr);
1025 SAVEPPTR(PL_oldoldbufptr);
1026 SAVEPPTR(PL_linestart);
1027 SAVESPTR(PL_linestr);
1028 SAVEPPTR(PL_lex_brackstack);
1029 SAVEPPTR(PL_lex_casestack);
1031 PL_linestr = PL_lex_stuff;
1032 PL_lex_stuff = Nullsv;
1034 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1035 = SvPVX(PL_linestr);
1036 PL_bufend += SvCUR(PL_linestr);
1037 SAVEFREESV(PL_linestr);
1039 PL_lex_dojoin = FALSE;
1040 PL_lex_brackets = 0;
1041 New(899, PL_lex_brackstack, 120, char);
1042 New(899, PL_lex_casestack, 12, char);
1043 SAVEFREEPV(PL_lex_brackstack);
1044 SAVEFREEPV(PL_lex_casestack);
1045 PL_lex_casemods = 0;
1046 *PL_lex_casestack = '\0';
1048 PL_lex_state = LEX_INTERPCONCAT;
1049 CopLINE_set(PL_curcop, PL_multi_start);
1051 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1052 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1053 PL_lex_inpat = PL_sublex_info.sub_op;
1055 PL_lex_inpat = Nullop;
1062 * Restores lexer state after a S_sublex_push.
1068 if (!PL_lex_starts++) {
1069 PL_expect = XOPERATOR;
1070 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn("",0));
1074 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1075 PL_lex_state = LEX_INTERPCASEMOD;
1079 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1080 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1081 PL_linestr = PL_lex_repl;
1083 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1084 PL_bufend += SvCUR(PL_linestr);
1085 SAVEFREESV(PL_linestr);
1086 PL_lex_dojoin = FALSE;
1087 PL_lex_brackets = 0;
1088 PL_lex_casemods = 0;
1089 *PL_lex_casestack = '\0';
1091 if (SvEVALED(PL_lex_repl)) {
1092 PL_lex_state = LEX_INTERPNORMAL;
1094 /* we don't clear PL_lex_repl here, so that we can check later
1095 whether this is an evalled subst; that means we rely on the
1096 logic to ensure sublex_done() is called again only via the
1097 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1100 PL_lex_state = LEX_INTERPCONCAT;
1101 PL_lex_repl = Nullsv;
1107 PL_bufend = SvPVX(PL_linestr);
1108 PL_bufend += SvCUR(PL_linestr);
1109 PL_expect = XOPERATOR;
1110 PL_sublex_info.sub_inwhat = 0;
1118 Extracts a pattern, double-quoted string, or transliteration. This
1121 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1122 processing a pattern (PL_lex_inpat is true), a transliteration
1123 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1125 Returns a pointer to the character scanned up to. Iff this is
1126 advanced from the start pointer supplied (ie if anything was
1127 successfully parsed), will leave an OP for the substring scanned
1128 in yylval. Caller must intuit reason for not parsing further
1129 by looking at the next characters herself.
1133 double-quoted style: \r and \n
1134 regexp special ones: \D \s
1136 backrefs: \1 (deprecated in substitution replacements)
1137 case and quoting: \U \Q \E
1138 stops on @ and $, but not for $ as tail anchor
1140 In transliterations:
1141 characters are VERY literal, except for - not at the start or end
1142 of the string, which indicates a range. scan_const expands the
1143 range to the full set of intermediate characters.
1145 In double-quoted strings:
1147 double-quoted style: \r and \n
1149 backrefs: \1 (deprecated)
1150 case and quoting: \U \Q \E
1153 scan_const does *not* construct ops to handle interpolated strings.
1154 It stops processing as soon as it finds an embedded $ or @ variable
1155 and leaves it to the caller to work out what's going on.
1157 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1159 $ in pattern could be $foo or could be tail anchor. Assumption:
1160 it's a tail anchor if $ is the last thing in the string, or if it's
1161 followed by one of ")| \n\t"
1163 \1 (backreferences) are turned into $1
1165 The structure of the code is
1166 while (there's a character to process) {
1167 handle transliteration ranges
1168 skip regexp comments
1169 skip # initiated comments in //x patterns
1170 check for embedded @foo
1171 check for embedded scalars
1173 leave intact backslashes from leave (below)
1174 deprecate \1 in strings and sub replacements
1175 handle string-changing backslashes \l \U \Q \E, etc.
1176 switch (what was escaped) {
1177 handle - in a transliteration (becomes a literal -)
1178 handle \132 octal characters
1179 handle 0x15 hex characters
1180 handle \cV (control V)
1181 handle printf backslashes (\f, \r, \n, etc)
1183 } (end if backslash)
1184 } (end while character to read)
1189 S_scan_const(pTHX_ char *start)
1191 register char *send = PL_bufend; /* end of the constant */
1192 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1193 register char *s = start; /* start of the constant */
1194 register char *d = SvPVX(sv); /* destination for copies */
1195 bool dorange = FALSE; /* are we in a translit range? */
1196 bool has_utf = FALSE; /* embedded \x{} */
1200 I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1201 ? (PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1203 I32 thisutf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1204 ? (PL_sublex_info.sub_op->op_private & (PL_lex_repl ?
1205 OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF))
1207 const char *leaveit = /* set of acceptably-backslashed characters */
1209 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1212 while (s < send || dorange) {
1213 /* get transliterations out of the way (they're most literal) */
1214 if (PL_lex_inwhat == OP_TRANS) {
1215 /* expand a range A-Z to the full set of characters. AIE! */
1217 I32 i; /* current expanded character */
1218 I32 min; /* first character in range */
1219 I32 max; /* last character in range */
1221 i = d - SvPVX(sv); /* remember current offset */
1222 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1223 d = SvPVX(sv) + i; /* refresh d after realloc */
1224 d -= 2; /* eat the first char and the - */
1226 min = (U8)*d; /* first char in range */
1227 max = (U8)d[1]; /* last char in range */
1230 if ((isLOWER(min) && isLOWER(max)) ||
1231 (isUPPER(min) && isUPPER(max))) {
1233 for (i = min; i <= max; i++)
1237 for (i = min; i <= max; i++)
1244 for (i = min; i <= max; i++)
1247 /* mark the range as done, and continue */
1252 /* range begins (ignore - as first or last char) */
1253 else if (*s == '-' && s+1 < send && s != start) {
1255 *d++ = (char)0xff; /* use illegal utf8 byte--see pmtrans */
1264 /* if we get here, we're not doing a transliteration */
1266 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1267 except for the last char, which will be done separately. */
1268 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1270 while (s < send && *s != ')')
1273 else if (s[2] == '{' /* This should match regcomp.c */
1274 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1277 char *regparse = s + (s[2] == '{' ? 3 : 4);
1280 while (count && (c = *regparse)) {
1281 if (c == '\\' && regparse[1])
1289 if (*regparse != ')') {
1290 regparse--; /* Leave one char for continuation. */
1291 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1293 while (s < regparse)
1298 /* likewise skip #-initiated comments in //x patterns */
1299 else if (*s == '#' && PL_lex_inpat &&
1300 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1301 while (s+1 < send && *s != '\n')
1305 /* check for embedded arrays (@foo, @:foo, @'foo, @{foo}, @$foo) */
1306 else if (*s == '@' && s[1]
1307 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$", s[1])))
1310 /* check for embedded scalars. only stop if we're sure it's a
1313 else if (*s == '$') {
1314 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1316 if (s + 1 < send && !strchr("()| \n\t", s[1]))
1317 break; /* in regexp, $ might be tail anchor */
1320 /* (now in tr/// code again) */
1322 if (*s & 0x80 && thisutf) {
1323 (void)utf8_to_uv((U8*)s, &len);
1325 /* illegal UTF8, make it valid */
1326 char *old_pvx = SvPVX(sv);
1327 /* need space for one extra char (NOTE: SvCUR() not set here) */
1328 d = SvGROW(sv, SvLEN(sv) + 1) + (d - old_pvx);
1329 d = (char*)uv_to_utf8((U8*)d, (U8)*s++);
1340 if (*s == '\\' && s+1 < send) {
1343 /* some backslashes we leave behind */
1344 if (*leaveit && *s && strchr(leaveit, *s)) {
1350 /* deprecate \1 in strings and substitution replacements */
1351 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1352 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1354 dTHR; /* only for ckWARN */
1355 if (ckWARN(WARN_SYNTAX))
1356 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1361 /* string-change backslash escapes */
1362 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1367 /* if we get here, it's either a quoted -, or a digit */
1370 /* quoted - in transliterations */
1372 if (PL_lex_inwhat == OP_TRANS) {
1380 if (ckWARN(WARN_MISC) && isALNUM(*s))
1381 Perl_warner(aTHX_ WARN_MISC,
1382 "Unrecognized escape \\%c passed through",
1384 /* default action is to copy the quoted character */
1389 /* \132 indicates an octal constant */
1390 case '0': case '1': case '2': case '3':
1391 case '4': case '5': case '6': case '7':
1392 len = 0; /* disallow underscores */
1393 uv = (UV)scan_oct(s, 3, &len);
1395 goto NUM_ESCAPE_INSERT;
1397 /* \x24 indicates a hex constant */
1401 char* e = strchr(s, '}');
1403 yyerror("Missing right brace on \\x{}");
1406 len = 1; /* allow underscores */
1407 uv = (UV)scan_hex(s + 1, e - s - 1, &len);
1411 len = 0; /* disallow underscores */
1412 uv = (UV)scan_hex(s, 2, &len);
1417 /* Insert oct or hex escaped character.
1418 * There will always enough room in sv since such escapes will
1419 * be longer than any utf8 sequence they can end up as
1422 if (!thisutf && !has_utf && uv > 255) {
1423 /* might need to recode whatever we have accumulated so far
1424 * if it contains any hibit chars
1428 for (c = SvPVX(sv); c < d; c++) {
1433 char *old_pvx = SvPVX(sv);
1435 d = SvGROW(sv, SvCUR(sv) + hicount + 1) + (d - old_pvx);
1444 uv_to_utf8((U8*)dst, (U8)*src--);
1454 if (thisutf || uv > 255) {
1455 d = (char*)uv_to_utf8((U8*)d, uv);
1467 /* \N{latin small letter a} is a named character */
1471 char* e = strchr(s, '}');
1477 yyerror("Missing right brace on \\N{}");
1481 res = newSVpvn(s + 1, e - s - 1);
1482 res = new_constant( Nullch, 0, "charnames",
1483 res, Nullsv, "\\N{...}" );
1484 str = SvPV(res,len);
1485 if (!has_utf && SvUTF8(res)) {
1486 char *ostart = SvPVX(sv);
1487 SvCUR_set(sv, d - ostart);
1489 sv_utf8_upgrade(sv);
1490 d = SvPVX(sv) + SvCUR(sv);
1493 if (len > e - s + 4) {
1494 char *odest = SvPVX(sv);
1496 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1497 d = SvPVX(sv) + (d - odest);
1499 Copy(str, d, len, char);
1506 yyerror("Missing braces on \\N{}");
1509 /* \c is a control character */
1524 /* printf-style backslashes, formfeeds, newlines, etc */
1542 *d++ = '\047'; /* CP 1047 */
1545 *d++ = '\057'; /* CP 1047 */
1559 } /* end if (backslash) */
1562 } /* while loop to process each character */
1564 /* terminate the string and set up the sv */
1566 SvCUR_set(sv, d - SvPVX(sv));
1571 /* shrink the sv if we allocated more than we used */
1572 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1573 SvLEN_set(sv, SvCUR(sv) + 1);
1574 Renew(SvPVX(sv), SvLEN(sv), char);
1577 /* return the substring (via yylval) only if we parsed anything */
1578 if (s > PL_bufptr) {
1579 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1580 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1582 ( PL_lex_inwhat == OP_TRANS
1584 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1587 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1594 * Returns TRUE if there's more to the expression (e.g., a subscript),
1597 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1599 * ->[ and ->{ return TRUE
1600 * { and [ outside a pattern are always subscripts, so return TRUE
1601 * if we're outside a pattern and it's not { or [, then return FALSE
1602 * if we're in a pattern and the first char is a {
1603 * {4,5} (any digits around the comma) returns FALSE
1604 * if we're in a pattern and the first char is a [
1606 * [SOMETHING] has a funky algorithm to decide whether it's a
1607 * character class or not. It has to deal with things like
1608 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1609 * anything else returns TRUE
1612 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1615 S_intuit_more(pTHX_ register char *s)
1617 if (PL_lex_brackets)
1619 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1621 if (*s != '{' && *s != '[')
1626 /* In a pattern, so maybe we have {n,m}. */
1643 /* On the other hand, maybe we have a character class */
1646 if (*s == ']' || *s == '^')
1649 /* this is terrifying, and it works */
1650 int weight = 2; /* let's weigh the evidence */
1652 unsigned char un_char = 255, last_un_char;
1653 char *send = strchr(s,']');
1654 char tmpbuf[sizeof PL_tokenbuf * 4];
1656 if (!send) /* has to be an expression */
1659 Zero(seen,256,char);
1662 else if (isDIGIT(*s)) {
1664 if (isDIGIT(s[1]) && s[2] == ']')
1670 for (; s < send; s++) {
1671 last_un_char = un_char;
1672 un_char = (unsigned char)*s;
1677 weight -= seen[un_char] * 10;
1678 if (isALNUM_lazy_if(s+1,UTF)) {
1679 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1680 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1685 else if (*s == '$' && s[1] &&
1686 strchr("[#!%*<>()-=",s[1])) {
1687 if (/*{*/ strchr("])} =",s[2]))
1696 if (strchr("wds]",s[1]))
1698 else if (seen['\''] || seen['"'])
1700 else if (strchr("rnftbxcav",s[1]))
1702 else if (isDIGIT(s[1])) {
1704 while (s[1] && isDIGIT(s[1]))
1714 if (strchr("aA01! ",last_un_char))
1716 if (strchr("zZ79~",s[1]))
1718 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1719 weight -= 5; /* cope with negative subscript */
1722 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1723 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1728 if (keyword(tmpbuf, d - tmpbuf))
1731 if (un_char == last_un_char + 1)
1733 weight -= seen[un_char];
1738 if (weight >= 0) /* probably a character class */
1748 * Does all the checking to disambiguate
1750 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1751 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1753 * First argument is the stuff after the first token, e.g. "bar".
1755 * Not a method if bar is a filehandle.
1756 * Not a method if foo is a subroutine prototyped to take a filehandle.
1757 * Not a method if it's really "Foo $bar"
1758 * Method if it's "foo $bar"
1759 * Not a method if it's really "print foo $bar"
1760 * Method if it's really "foo package::" (interpreted as package->foo)
1761 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1762 * Not a method if bar is a filehandle or package, but is quoted with
1767 S_intuit_method(pTHX_ char *start, GV *gv)
1769 char *s = start + (*start == '$');
1770 char tmpbuf[sizeof PL_tokenbuf];
1778 if ((cv = GvCVu(gv))) {
1779 char *proto = SvPVX(cv);
1789 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1790 /* start is the beginning of the possible filehandle/object,
1791 * and s is the end of it
1792 * tmpbuf is a copy of it
1795 if (*start == '$') {
1796 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1801 return *s == '(' ? FUNCMETH : METHOD;
1803 if (!keyword(tmpbuf, len)) {
1804 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1809 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1810 if (indirgv && GvCVu(indirgv))
1812 /* filehandle or package name makes it a method */
1813 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1815 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1816 return 0; /* no assumptions -- "=>" quotes bearword */
1818 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1819 newSVpvn(tmpbuf,len));
1820 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1824 return *s == '(' ? FUNCMETH : METHOD;
1832 * Return a string of Perl code to load the debugger. If PERL5DB
1833 * is set, it will return the contents of that, otherwise a
1834 * compile-time require of perl5db.pl.
1841 char *pdb = PerlEnv_getenv("PERL5DB");
1845 SETERRNO(0,SS$_NORMAL);
1846 return "BEGIN { require 'perl5db.pl' }";
1852 /* Encoded script support. filter_add() effectively inserts a
1853 * 'pre-processing' function into the current source input stream.
1854 * Note that the filter function only applies to the current source file
1855 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1857 * The datasv parameter (which may be NULL) can be used to pass
1858 * private data to this instance of the filter. The filter function
1859 * can recover the SV using the FILTER_DATA macro and use it to
1860 * store private buffers and state information.
1862 * The supplied datasv parameter is upgraded to a PVIO type
1863 * and the IoDIRP field is used to store the function pointer,
1864 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1865 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1866 * private use must be set using malloc'd pointers.
1870 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1875 if (!PL_rsfp_filters)
1876 PL_rsfp_filters = newAV();
1878 datasv = NEWSV(255,0);
1879 if (!SvUPGRADE(datasv, SVt_PVIO))
1880 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1881 IoDIRP(datasv) = (DIR*)funcp; /* stash funcp into spare field */
1882 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1883 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1884 funcp, SvPV_nolen(datasv)));
1885 av_unshift(PL_rsfp_filters, 1);
1886 av_store(PL_rsfp_filters, 0, datasv) ;
1891 /* Delete most recently added instance of this filter function. */
1893 Perl_filter_del(pTHX_ filter_t funcp)
1896 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1897 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1899 /* if filter is on top of stack (usual case) just pop it off */
1900 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1901 if (IoDIRP(datasv) == (DIR*)funcp) {
1902 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1903 IoDIRP(datasv) = (DIR*)NULL;
1904 sv_free(av_pop(PL_rsfp_filters));
1908 /* we need to search for the correct entry and clear it */
1909 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1913 /* Invoke the n'th filter function for the current rsfp. */
1915 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1918 /* 0 = read one text line */
1923 if (!PL_rsfp_filters)
1925 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1926 /* Provide a default input filter to make life easy. */
1927 /* Note that we append to the line. This is handy. */
1928 DEBUG_P(PerlIO_printf(Perl_debug_log,
1929 "filter_read %d: from rsfp\n", idx));
1933 int old_len = SvCUR(buf_sv) ;
1935 /* ensure buf_sv is large enough */
1936 SvGROW(buf_sv, old_len + maxlen) ;
1937 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1938 if (PerlIO_error(PL_rsfp))
1939 return -1; /* error */
1941 return 0 ; /* end of file */
1943 SvCUR_set(buf_sv, old_len + len) ;
1946 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1947 if (PerlIO_error(PL_rsfp))
1948 return -1; /* error */
1950 return 0 ; /* end of file */
1953 return SvCUR(buf_sv);
1955 /* Skip this filter slot if filter has been deleted */
1956 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1957 DEBUG_P(PerlIO_printf(Perl_debug_log,
1958 "filter_read %d: skipped (filter deleted)\n",
1960 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1962 /* Get function pointer hidden within datasv */
1963 funcp = (filter_t)IoDIRP(datasv);
1964 DEBUG_P(PerlIO_printf(Perl_debug_log,
1965 "filter_read %d: via function %p (%s)\n",
1966 idx, funcp, SvPV_nolen(datasv)));
1967 /* Call function. The function is expected to */
1968 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1969 /* Return: <0:error, =0:eof, >0:not eof */
1970 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1974 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1976 #ifdef PERL_CR_FILTER
1977 if (!PL_rsfp_filters) {
1978 filter_add(S_cr_textfilter,NULL);
1981 if (PL_rsfp_filters) {
1984 SvCUR_set(sv, 0); /* start with empty line */
1985 if (FILTER_READ(0, sv, 0) > 0)
1986 return ( SvPVX(sv) ) ;
1991 return (sv_gets(sv, fp, append));
1996 static char* exp_name[] =
1997 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
1998 "ATTRTERM", "TERMBLOCK"
2005 Works out what to call the token just pulled out of the input
2006 stream. The yacc parser takes care of taking the ops we return and
2007 stitching them into a tree.
2013 if read an identifier
2014 if we're in a my declaration
2015 croak if they tried to say my($foo::bar)
2016 build the ops for a my() declaration
2017 if it's an access to a my() variable
2018 are we in a sort block?
2019 croak if my($a); $a <=> $b
2020 build ops for access to a my() variable
2021 if in a dq string, and they've said @foo and we can't find @foo
2023 build ops for a bareword
2024 if we already built the token before, use it.
2028 #ifdef USE_PURE_BISON
2029 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
2042 #ifdef USE_PURE_BISON
2043 yylval_pointer = lvalp;
2044 yychar_pointer = lcharp;
2047 /* check if there's an identifier for us to look at */
2048 if (PL_pending_ident) {
2049 /* pit holds the identifier we read and pending_ident is reset */
2050 char pit = PL_pending_ident;
2051 PL_pending_ident = 0;
2053 /* if we're in a my(), we can't allow dynamics here.
2054 $foo'bar has already been turned into $foo::bar, so
2055 just check for colons.
2057 if it's a legal name, the OP is a PADANY.
2060 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
2061 if (strchr(PL_tokenbuf,':'))
2062 yyerror(Perl_form(aTHX_ "No package name allowed for "
2063 "variable %s in \"our\"",
2065 tmp = pad_allocmy(PL_tokenbuf);
2068 if (strchr(PL_tokenbuf,':'))
2069 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
2071 yylval.opval = newOP(OP_PADANY, 0);
2072 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
2078 build the ops for accesses to a my() variable.
2080 Deny my($a) or my($b) in a sort block, *if* $a or $b is
2081 then used in a comparison. This catches most, but not
2082 all cases. For instance, it catches
2083 sort { my($a); $a <=> $b }
2085 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2086 (although why you'd do that is anyone's guess).
2089 if (!strchr(PL_tokenbuf,':')) {
2091 /* Check for single character per-thread SVs */
2092 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2093 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2094 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2096 yylval.opval = newOP(OP_THREADSV, 0);
2097 yylval.opval->op_targ = tmp;
2100 #endif /* USE_THREADS */
2101 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2102 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2103 /* might be an "our" variable" */
2104 if (SvFLAGS(namesv) & SVpad_OUR) {
2105 /* build ops for a bareword */
2106 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2107 sv_catpvn(sym, "::", 2);
2108 sv_catpv(sym, PL_tokenbuf+1);
2109 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2110 yylval.opval->op_private = OPpCONST_ENTERED;
2111 gv_fetchpv(SvPVX(sym),
2113 ? (GV_ADDMULTI | GV_ADDINEVAL)
2116 ((PL_tokenbuf[0] == '$') ? SVt_PV
2117 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2122 /* if it's a sort block and they're naming $a or $b */
2123 if (PL_last_lop_op == OP_SORT &&
2124 PL_tokenbuf[0] == '$' &&
2125 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2128 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2129 d < PL_bufend && *d != '\n';
2132 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2133 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2139 yylval.opval = newOP(OP_PADANY, 0);
2140 yylval.opval->op_targ = tmp;
2146 Whine if they've said @foo in a doublequoted string,
2147 and @foo isn't a variable we can find in the symbol
2150 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2151 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2152 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2153 && ckWARN(WARN_AMBIGUOUS))
2155 /* Downgraded from fatal to warning 20000522 mjd */
2156 Perl_warner(aTHX_ WARN_AMBIGUOUS,
2157 "Possible unintended interpolation of %s in string",
2162 /* build ops for a bareword */
2163 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2164 yylval.opval->op_private = OPpCONST_ENTERED;
2165 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2166 ((PL_tokenbuf[0] == '$') ? SVt_PV
2167 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2172 /* no identifier pending identification */
2174 switch (PL_lex_state) {
2176 case LEX_NORMAL: /* Some compilers will produce faster */
2177 case LEX_INTERPNORMAL: /* code if we comment these out. */
2181 /* when we've already built the next token, just pull it out of the queue */
2184 yylval = PL_nextval[PL_nexttoke];
2186 PL_lex_state = PL_lex_defer;
2187 PL_expect = PL_lex_expect;
2188 PL_lex_defer = LEX_NORMAL;
2190 return(PL_nexttype[PL_nexttoke]);
2192 /* interpolated case modifiers like \L \U, including \Q and \E.
2193 when we get here, PL_bufptr is at the \
2195 case LEX_INTERPCASEMOD:
2197 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2198 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2200 /* handle \E or end of string */
2201 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2205 if (PL_lex_casemods) {
2206 oldmod = PL_lex_casestack[--PL_lex_casemods];
2207 PL_lex_casestack[PL_lex_casemods] = '\0';
2209 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2211 PL_lex_state = LEX_INTERPCONCAT;
2215 if (PL_bufptr != PL_bufend)
2217 PL_lex_state = LEX_INTERPCONCAT;
2222 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2223 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2224 if (strchr("LU", *s) &&
2225 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2227 PL_lex_casestack[--PL_lex_casemods] = '\0';
2230 if (PL_lex_casemods > 10) {
2231 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2232 if (newlb != PL_lex_casestack) {
2234 PL_lex_casestack = newlb;
2237 PL_lex_casestack[PL_lex_casemods++] = *s;
2238 PL_lex_casestack[PL_lex_casemods] = '\0';
2239 PL_lex_state = LEX_INTERPCONCAT;
2240 PL_nextval[PL_nexttoke].ival = 0;
2243 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2245 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2247 PL_nextval[PL_nexttoke].ival = OP_LC;
2249 PL_nextval[PL_nexttoke].ival = OP_UC;
2251 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2253 Perl_croak(aTHX_ "panic: yylex");
2256 if (PL_lex_starts) {
2265 case LEX_INTERPPUSH:
2266 return sublex_push();
2268 case LEX_INTERPSTART:
2269 if (PL_bufptr == PL_bufend)
2270 return sublex_done();
2272 PL_lex_dojoin = (*PL_bufptr == '@');
2273 PL_lex_state = LEX_INTERPNORMAL;
2274 if (PL_lex_dojoin) {
2275 PL_nextval[PL_nexttoke].ival = 0;
2278 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2279 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2280 force_next(PRIVATEREF);
2282 force_ident("\"", '$');
2283 #endif /* USE_THREADS */
2284 PL_nextval[PL_nexttoke].ival = 0;
2286 PL_nextval[PL_nexttoke].ival = 0;
2288 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2291 if (PL_lex_starts++) {
2297 case LEX_INTERPENDMAYBE:
2298 if (intuit_more(PL_bufptr)) {
2299 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2305 if (PL_lex_dojoin) {
2306 PL_lex_dojoin = FALSE;
2307 PL_lex_state = LEX_INTERPCONCAT;
2310 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2311 && SvEVALED(PL_lex_repl))
2313 if (PL_bufptr != PL_bufend)
2314 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2315 PL_lex_repl = Nullsv;
2318 case LEX_INTERPCONCAT:
2320 if (PL_lex_brackets)
2321 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2323 if (PL_bufptr == PL_bufend)
2324 return sublex_done();
2326 if (SvIVX(PL_linestr) == '\'') {
2327 SV *sv = newSVsv(PL_linestr);
2330 else if ( PL_hints & HINT_NEW_RE )
2331 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2332 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2336 s = scan_const(PL_bufptr);
2338 PL_lex_state = LEX_INTERPCASEMOD;
2340 PL_lex_state = LEX_INTERPSTART;
2343 if (s != PL_bufptr) {
2344 PL_nextval[PL_nexttoke] = yylval;
2347 if (PL_lex_starts++)
2357 PL_lex_state = LEX_NORMAL;
2358 s = scan_formline(PL_bufptr);
2359 if (!PL_lex_formbrack)
2365 PL_oldoldbufptr = PL_oldbufptr;
2368 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2369 exp_name[PL_expect], s);
2375 if (isIDFIRST_lazy_if(s,UTF))
2377 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2380 goto fake_eof; /* emulate EOF on ^D or ^Z */
2385 if (PL_lex_brackets)
2386 yyerror("Missing right curly or square bracket");
2389 if (s++ < PL_bufend)
2390 goto retry; /* ignore stray nulls */
2393 if (!PL_in_eval && !PL_preambled) {
2394 PL_preambled = TRUE;
2395 sv_setpv(PL_linestr,incl_perldb());
2396 if (SvCUR(PL_linestr))
2397 sv_catpv(PL_linestr,";");
2399 while(AvFILLp(PL_preambleav) >= 0) {
2400 SV *tmpsv = av_shift(PL_preambleav);
2401 sv_catsv(PL_linestr, tmpsv);
2402 sv_catpv(PL_linestr, ";");
2405 sv_free((SV*)PL_preambleav);
2406 PL_preambleav = NULL;
2408 if (PL_minus_n || PL_minus_p) {
2409 sv_catpv(PL_linestr, "LINE: while (<>) {");
2411 sv_catpv(PL_linestr,"chomp;");
2413 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2415 GvIMPORTED_AV_on(gv);
2417 if (strchr("/'\"", *PL_splitstr)
2418 && strchr(PL_splitstr + 1, *PL_splitstr))
2419 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2422 s = "'~#\200\1'"; /* surely one char is unused...*/
2423 while (s[1] && strchr(PL_splitstr, *s)) s++;
2425 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2426 "q" + (delim == '\''), delim);
2427 for (s = PL_splitstr; *s; s++) {
2429 sv_catpvn(PL_linestr, "\\", 1);
2430 sv_catpvn(PL_linestr, s, 1);
2432 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2436 sv_catpv(PL_linestr,"@F=split(' ');");
2439 sv_catpv(PL_linestr, "\n");
2440 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2441 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2442 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2443 SV *sv = NEWSV(85,0);
2445 sv_upgrade(sv, SVt_PVMG);
2446 sv_setsv(sv,PL_linestr);
2447 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2452 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2455 if (PL_preprocess && !PL_in_eval)
2456 (void)PerlProc_pclose(PL_rsfp);
2457 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2458 PerlIO_clearerr(PL_rsfp);
2460 (void)PerlIO_close(PL_rsfp);
2462 PL_doextract = FALSE;
2464 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2465 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2466 sv_catpv(PL_linestr,";}");
2467 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2468 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2469 PL_minus_n = PL_minus_p = 0;
2472 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2473 sv_setpv(PL_linestr,"");
2474 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2477 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2478 PL_doextract = FALSE;
2480 /* Incest with pod. */
2481 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2482 sv_setpv(PL_linestr, "");
2483 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2484 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2485 PL_doextract = FALSE;
2489 } while (PL_doextract);
2490 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2491 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2492 SV *sv = NEWSV(85,0);
2494 sv_upgrade(sv, SVt_PVMG);
2495 sv_setsv(sv,PL_linestr);
2496 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2498 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2499 if (CopLINE(PL_curcop) == 1) {
2500 while (s < PL_bufend && isSPACE(*s))
2502 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2506 if (*s == '#' && *(s+1) == '!')
2508 #ifdef ALTERNATE_SHEBANG
2510 static char as[] = ALTERNATE_SHEBANG;
2511 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2512 d = s + (sizeof(as) - 1);
2514 #endif /* ALTERNATE_SHEBANG */
2523 while (*d && !isSPACE(*d))
2527 #ifdef ARG_ZERO_IS_SCRIPT
2528 if (ipathend > ipath) {
2530 * HP-UX (at least) sets argv[0] to the script name,
2531 * which makes $^X incorrect. And Digital UNIX and Linux,
2532 * at least, set argv[0] to the basename of the Perl
2533 * interpreter. So, having found "#!", we'll set it right.
2535 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2536 assert(SvPOK(x) || SvGMAGICAL(x));
2537 if (sv_eq(x, CopFILESV(PL_curcop))) {
2538 sv_setpvn(x, ipath, ipathend - ipath);
2541 TAINT_NOT; /* $^X is always tainted, but that's OK */
2543 #endif /* ARG_ZERO_IS_SCRIPT */
2548 d = instr(s,"perl -");
2550 d = instr(s,"perl");
2552 /* avoid getting into infinite loops when shebang
2553 * line contains "Perl" rather than "perl" */
2555 for (d = ipathend-4; d >= ipath; --d) {
2556 if ((*d == 'p' || *d == 'P')
2557 && !ibcmp(d, "perl", 4))
2567 #ifdef ALTERNATE_SHEBANG
2569 * If the ALTERNATE_SHEBANG on this system starts with a
2570 * character that can be part of a Perl expression, then if
2571 * we see it but not "perl", we're probably looking at the
2572 * start of Perl code, not a request to hand off to some
2573 * other interpreter. Similarly, if "perl" is there, but
2574 * not in the first 'word' of the line, we assume the line
2575 * contains the start of the Perl program.
2577 if (d && *s != '#') {
2579 while (*c && !strchr("; \t\r\n\f\v#", *c))
2582 d = Nullch; /* "perl" not in first word; ignore */
2584 *s = '#'; /* Don't try to parse shebang line */
2586 #endif /* ALTERNATE_SHEBANG */
2591 !instr(s,"indir") &&
2592 instr(PL_origargv[0],"perl"))
2598 while (s < PL_bufend && isSPACE(*s))
2600 if (s < PL_bufend) {
2601 Newz(899,newargv,PL_origargc+3,char*);
2603 while (s < PL_bufend && !isSPACE(*s))
2606 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2609 newargv = PL_origargv;
2611 PerlProc_execv(ipath, newargv);
2612 Perl_croak(aTHX_ "Can't exec %s", ipath);
2615 U32 oldpdb = PL_perldb;
2616 bool oldn = PL_minus_n;
2617 bool oldp = PL_minus_p;
2619 while (*d && !isSPACE(*d)) d++;
2620 while (*d == ' ' || *d == '\t') d++;
2624 if (*d == 'M' || *d == 'm') {
2626 while (*d && !isSPACE(*d)) d++;
2627 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2630 d = moreswitches(d);
2632 if ((PERLDB_LINE && !oldpdb) ||
2633 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2634 /* if we have already added "LINE: while (<>) {",
2635 we must not do it again */
2637 sv_setpv(PL_linestr, "");
2638 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2639 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2640 PL_preambled = FALSE;
2642 (void)gv_fetchfile(PL_origfilename);
2649 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2651 PL_lex_state = LEX_FORMLINE;
2656 #ifdef PERL_STRICT_CR
2657 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2659 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2661 case ' ': case '\t': case '\f': case 013:
2666 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2667 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2668 /* handle eval qq[#line 1 "foo"\n ...] */
2669 CopLINE_dec(PL_curcop);
2673 while (s < d && *s != '\n')
2678 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2680 PL_lex_state = LEX_FORMLINE;
2690 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2695 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2698 if (strnEQ(s,"=>",2)) {
2699 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2700 OPERATOR('-'); /* unary minus */
2702 PL_last_uni = PL_oldbufptr;
2703 PL_last_lop_op = OP_FTEREAD; /* good enough */
2705 case 'r': FTST(OP_FTEREAD);
2706 case 'w': FTST(OP_FTEWRITE);
2707 case 'x': FTST(OP_FTEEXEC);
2708 case 'o': FTST(OP_FTEOWNED);
2709 case 'R': FTST(OP_FTRREAD);
2710 case 'W': FTST(OP_FTRWRITE);
2711 case 'X': FTST(OP_FTREXEC);
2712 case 'O': FTST(OP_FTROWNED);
2713 case 'e': FTST(OP_FTIS);
2714 case 'z': FTST(OP_FTZERO);
2715 case 's': FTST(OP_FTSIZE);
2716 case 'f': FTST(OP_FTFILE);
2717 case 'd': FTST(OP_FTDIR);
2718 case 'l': FTST(OP_FTLINK);
2719 case 'p': FTST(OP_FTPIPE);
2720 case 'S': FTST(OP_FTSOCK);
2721 case 'u': FTST(OP_FTSUID);
2722 case 'g': FTST(OP_FTSGID);
2723 case 'k': FTST(OP_FTSVTX);
2724 case 'b': FTST(OP_FTBLK);
2725 case 'c': FTST(OP_FTCHR);
2726 case 't': FTST(OP_FTTTY);
2727 case 'T': FTST(OP_FTTEXT);
2728 case 'B': FTST(OP_FTBINARY);
2729 case 'M': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTMTIME);
2730 case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME);
2731 case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME);
2733 Perl_croak(aTHX_ "Unrecognized file test: -%c", (int)tmp);
2740 if (PL_expect == XOPERATOR)
2745 else if (*s == '>') {
2748 if (isIDFIRST_lazy_if(s,UTF)) {
2749 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2757 if (PL_expect == XOPERATOR)
2760 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2762 OPERATOR('-'); /* unary minus */
2769 if (PL_expect == XOPERATOR)
2774 if (PL_expect == XOPERATOR)
2777 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2783 if (PL_expect != XOPERATOR) {
2784 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2785 PL_expect = XOPERATOR;
2786 force_ident(PL_tokenbuf, '*');
2799 if (PL_expect == XOPERATOR) {
2803 PL_tokenbuf[0] = '%';
2804 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2805 if (!PL_tokenbuf[1]) {
2807 yyerror("Final % should be \\% or %name");
2810 PL_pending_ident = '%';
2829 switch (PL_expect) {
2832 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2834 PL_bufptr = s; /* update in case we back off */
2840 PL_expect = XTERMBLOCK;
2844 while (isIDFIRST_lazy_if(s,UTF)) {
2845 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2846 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2847 if (tmp < 0) tmp = -tmp;
2862 d = scan_str(d,TRUE,TRUE);
2865 SvREFCNT_dec(PL_lex_stuff);
2866 PL_lex_stuff = Nullsv;
2868 /* MUST advance bufptr here to avoid bogus
2869 "at end of line" context messages from yyerror().
2871 PL_bufptr = s + len;
2872 yyerror("Unterminated attribute parameter in attribute list");
2875 return 0; /* EOF indicator */
2879 SV *sv = newSVpvn(s, len);
2880 sv_catsv(sv, PL_lex_stuff);
2881 attrs = append_elem(OP_LIST, attrs,
2882 newSVOP(OP_CONST, 0, sv));
2883 SvREFCNT_dec(PL_lex_stuff);
2884 PL_lex_stuff = Nullsv;
2887 attrs = append_elem(OP_LIST, attrs,
2888 newSVOP(OP_CONST, 0,
2892 if (*s == ':' && s[1] != ':')
2895 break; /* require real whitespace or :'s */
2897 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
2898 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
2899 char q = ((*s == '\'') ? '"' : '\'');
2900 /* If here for an expression, and parsed no attrs, back off. */
2901 if (tmp == '=' && !attrs) {
2905 /* MUST advance bufptr here to avoid bogus "at end of line"
2906 context messages from yyerror().
2910 yyerror("Unterminated attribute list");
2912 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
2920 PL_nextval[PL_nexttoke].opval = attrs;
2928 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
2929 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
2934 if (CopLINE(PL_curcop) < PL_copline)
2935 PL_copline = CopLINE(PL_curcop);
2946 if (PL_lex_brackets <= 0)
2947 yyerror("Unmatched right square bracket");
2950 if (PL_lex_state == LEX_INTERPNORMAL) {
2951 if (PL_lex_brackets == 0) {
2952 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
2953 PL_lex_state = LEX_INTERPEND;
2960 if (PL_lex_brackets > 100) {
2961 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
2962 if (newlb != PL_lex_brackstack) {
2964 PL_lex_brackstack = newlb;
2967 switch (PL_expect) {
2969 if (PL_lex_formbrack) {
2973 if (PL_oldoldbufptr == PL_last_lop)
2974 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2976 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2977 OPERATOR(HASHBRACK);
2979 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2982 PL_tokenbuf[0] = '\0';
2983 if (d < PL_bufend && *d == '-') {
2984 PL_tokenbuf[0] = '-';
2986 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2989 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
2990 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
2992 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2995 char minus = (PL_tokenbuf[0] == '-');
2996 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3004 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3009 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3014 if (PL_oldoldbufptr == PL_last_lop)
3015 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3017 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3020 OPERATOR(HASHBRACK);
3021 /* This hack serves to disambiguate a pair of curlies
3022 * as being a block or an anon hash. Normally, expectation
3023 * determines that, but in cases where we're not in a
3024 * position to expect anything in particular (like inside
3025 * eval"") we have to resolve the ambiguity. This code
3026 * covers the case where the first term in the curlies is a
3027 * quoted string. Most other cases need to be explicitly
3028 * disambiguated by prepending a `+' before the opening
3029 * curly in order to force resolution as an anon hash.
3031 * XXX should probably propagate the outer expectation
3032 * into eval"" to rely less on this hack, but that could
3033 * potentially break current behavior of eval"".
3037 if (*s == '\'' || *s == '"' || *s == '`') {
3038 /* common case: get past first string, handling escapes */
3039 for (t++; t < PL_bufend && *t != *s;)
3040 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3044 else if (*s == 'q') {
3047 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3051 char open, close, term;
3054 while (t < PL_bufend && isSPACE(*t))
3058 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3062 for (t++; t < PL_bufend; t++) {
3063 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3065 else if (*t == open)
3069 for (t++; t < PL_bufend; t++) {
3070 if (*t == '\\' && t+1 < PL_bufend)
3072 else if (*t == close && --brackets <= 0)
3074 else if (*t == open)
3080 else if (isALNUM_lazy_if(t,UTF)) {
3082 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3085 while (t < PL_bufend && isSPACE(*t))
3087 /* if comma follows first term, call it an anon hash */
3088 /* XXX it could be a comma expression with loop modifiers */
3089 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3090 || (*t == '=' && t[1] == '>')))
3091 OPERATOR(HASHBRACK);
3092 if (PL_expect == XREF)
3095 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3101 yylval.ival = CopLINE(PL_curcop);
3102 if (isSPACE(*s) || *s == '#')
3103 PL_copline = NOLINE; /* invalidate current command line number */
3108 if (PL_lex_brackets <= 0)
3109 yyerror("Unmatched right curly bracket");
3111 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3112 if (PL_lex_brackets < PL_lex_formbrack)
3113 PL_lex_formbrack = 0;
3114 if (PL_lex_state == LEX_INTERPNORMAL) {
3115 if (PL_lex_brackets == 0) {
3116 if (PL_expect & XFAKEBRACK) {
3117 PL_expect &= XENUMMASK;
3118 PL_lex_state = LEX_INTERPEND;
3120 return yylex(); /* ignore fake brackets */
3122 if (*s == '-' && s[1] == '>')
3123 PL_lex_state = LEX_INTERPENDMAYBE;
3124 else if (*s != '[' && *s != '{')
3125 PL_lex_state = LEX_INTERPEND;
3128 if (PL_expect & XFAKEBRACK) {
3129 PL_expect &= XENUMMASK;
3131 return yylex(); /* ignore fake brackets */
3141 if (PL_expect == XOPERATOR) {
3142 if (ckWARN(WARN_SEMICOLON)
3143 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3145 CopLINE_dec(PL_curcop);
3146 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3147 CopLINE_inc(PL_curcop);
3152 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3154 PL_expect = XOPERATOR;
3155 force_ident(PL_tokenbuf, '&');
3159 yylval.ival = (OPpENTERSUB_AMPER<<8);
3178 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3179 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3181 if (PL_expect == XSTATE && isALPHA(tmp) &&
3182 (s == PL_linestart+1 || s[-2] == '\n') )
3184 if (PL_in_eval && !PL_rsfp) {
3189 if (strnEQ(s,"=cut",4)) {
3203 PL_doextract = TRUE;
3206 if (PL_lex_brackets < PL_lex_formbrack) {
3208 #ifdef PERL_STRICT_CR
3209 for (t = s; *t == ' ' || *t == '\t'; t++) ;
3211 for (t = s; *t == ' ' || *t == '\t' || *t == '\r'; t++) ;
3213 if (*t == '\n' || *t == '#') {
3231 if (PL_expect != XOPERATOR) {
3232 if (s[1] != '<' && !strchr(s,'>'))
3235 s = scan_heredoc(s);
3237 s = scan_inputsymbol(s);
3238 TERM(sublex_start());
3243 SHop(OP_LEFT_SHIFT);
3257 SHop(OP_RIGHT_SHIFT);
3266 if (PL_expect == XOPERATOR) {
3267 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3270 return ','; /* grandfather non-comma-format format */
3274 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3275 PL_tokenbuf[0] = '@';
3276 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3277 sizeof PL_tokenbuf - 1, FALSE);
3278 if (PL_expect == XOPERATOR)
3279 no_op("Array length", s);
3280 if (!PL_tokenbuf[1])
3282 PL_expect = XOPERATOR;
3283 PL_pending_ident = '#';
3287 PL_tokenbuf[0] = '$';
3288 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3289 sizeof PL_tokenbuf - 1, FALSE);
3290 if (PL_expect == XOPERATOR)
3292 if (!PL_tokenbuf[1]) {
3294 yyerror("Final $ should be \\$ or $name");
3298 /* This kludge not intended to be bulletproof. */
3299 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3300 yylval.opval = newSVOP(OP_CONST, 0,
3301 newSViv(PL_compiling.cop_arybase));
3302 yylval.opval->op_private = OPpCONST_ARYBASE;
3308 if (PL_lex_state == LEX_NORMAL)
3311 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3314 PL_tokenbuf[0] = '@';
3315 if (ckWARN(WARN_SYNTAX)) {
3317 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3320 PL_bufptr = skipspace(PL_bufptr);
3321 while (t < PL_bufend && *t != ']')
3323 Perl_warner(aTHX_ WARN_SYNTAX,
3324 "Multidimensional syntax %.*s not supported",
3325 (t - PL_bufptr) + 1, PL_bufptr);
3329 else if (*s == '{') {
3330 PL_tokenbuf[0] = '%';
3331 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3332 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3334 char tmpbuf[sizeof PL_tokenbuf];
3336 for (t++; isSPACE(*t); t++) ;
3337 if (isIDFIRST_lazy_if(t,UTF)) {
3338 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3339 for (; isSPACE(*t); t++) ;
3340 if (*t == ';' && get_cv(tmpbuf, FALSE))
3341 Perl_warner(aTHX_ WARN_SYNTAX,
3342 "You need to quote \"%s\"", tmpbuf);
3348 PL_expect = XOPERATOR;
3349 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3350 bool islop = (PL_last_lop == PL_oldoldbufptr);
3351 if (!islop || PL_last_lop_op == OP_GREPSTART)
3352 PL_expect = XOPERATOR;
3353 else if (strchr("$@\"'`q", *s))
3354 PL_expect = XTERM; /* e.g. print $fh "foo" */
3355 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3356 PL_expect = XTERM; /* e.g. print $fh &sub */
3357 else if (isIDFIRST_lazy_if(s,UTF)) {
3358 char tmpbuf[sizeof PL_tokenbuf];
3359 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3360 if ((tmp = keyword(tmpbuf, len))) {
3361 /* binary operators exclude handle interpretations */
3373 PL_expect = XTERM; /* e.g. print $fh length() */
3378 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3379 if (gv && GvCVu(gv))
3380 PL_expect = XTERM; /* e.g. print $fh subr() */
3383 else if (isDIGIT(*s))
3384 PL_expect = XTERM; /* e.g. print $fh 3 */
3385 else if (*s == '.' && isDIGIT(s[1]))
3386 PL_expect = XTERM; /* e.g. print $fh .3 */
3387 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3388 PL_expect = XTERM; /* e.g. print $fh -1 */
3389 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3390 PL_expect = XTERM; /* print $fh <<"EOF" */
3392 PL_pending_ident = '$';
3396 if (PL_expect == XOPERATOR)
3398 PL_tokenbuf[0] = '@';
3399 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3400 if (!PL_tokenbuf[1]) {
3402 yyerror("Final @ should be \\@ or @name");
3405 if (PL_lex_state == LEX_NORMAL)
3407 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3409 PL_tokenbuf[0] = '%';
3411 /* Warn about @ where they meant $. */
3412 if (ckWARN(WARN_SYNTAX)) {
3413 if (*s == '[' || *s == '{') {
3415 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3417 if (*t == '}' || *t == ']') {
3419 PL_bufptr = skipspace(PL_bufptr);
3420 Perl_warner(aTHX_ WARN_SYNTAX,
3421 "Scalar value %.*s better written as $%.*s",
3422 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3427 PL_pending_ident = '@';
3430 case '/': /* may either be division or pattern */
3431 case '?': /* may either be conditional or pattern */
3432 if (PL_expect != XOPERATOR) {
3433 /* Disable warning on "study /blah/" */
3434 if (PL_oldoldbufptr == PL_last_uni
3435 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3436 || memNE(PL_last_uni, "study", 5)
3437 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3439 s = scan_pat(s,OP_MATCH);
3440 TERM(sublex_start());
3448 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3449 #ifdef PERL_STRICT_CR
3452 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3454 && (s == PL_linestart || s[-1] == '\n') )
3456 PL_lex_formbrack = 0;
3460 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3466 yylval.ival = OPf_SPECIAL;
3472 if (PL_expect != XOPERATOR)
3477 case '0': case '1': case '2': case '3': case '4':
3478 case '5': case '6': case '7': case '8': case '9':
3480 if (PL_expect == XOPERATOR)
3485 s = scan_str(s,FALSE,FALSE);
3486 if (PL_expect == XOPERATOR) {
3487 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3490 return ','; /* grandfather non-comma-format format */
3496 missingterm((char*)0);
3497 yylval.ival = OP_CONST;
3498 TERM(sublex_start());
3501 s = scan_str(s,FALSE,FALSE);
3502 if (PL_expect == XOPERATOR) {
3503 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3506 return ','; /* grandfather non-comma-format format */
3512 missingterm((char*)0);
3513 yylval.ival = OP_CONST;
3514 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3515 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3516 yylval.ival = OP_STRINGIFY;
3520 TERM(sublex_start());
3523 s = scan_str(s,FALSE,FALSE);
3524 if (PL_expect == XOPERATOR)
3525 no_op("Backticks",s);
3527 missingterm((char*)0);
3528 yylval.ival = OP_BACKTICK;
3530 TERM(sublex_start());
3534 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3535 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3537 if (PL_expect == XOPERATOR)
3538 no_op("Backslash",s);
3542 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3546 while (isDIGIT(*start) || *start == '_')
3548 if (*start == '.' && isDIGIT(start[1])) {
3552 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3553 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF)) {
3557 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3567 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3606 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3608 /* Some keywords can be followed by any delimiter, including ':' */
3609 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3610 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3611 (PL_tokenbuf[0] == 'q' &&
3612 strchr("qwxr", PL_tokenbuf[1])))));
3614 /* x::* is just a word, unless x is "CORE" */
3615 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3619 while (d < PL_bufend && isSPACE(*d))
3620 d++; /* no comments skipped here, or s### is misparsed */
3622 /* Is this a label? */
3623 if (!tmp && PL_expect == XSTATE
3624 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3626 yylval.pval = savepv(PL_tokenbuf);
3631 /* Check for keywords */
3632 tmp = keyword(PL_tokenbuf, len);
3634 /* Is this a word before a => operator? */
3635 if (*d == '=' && d[1] == '>') {
3637 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3638 yylval.opval->op_private = OPpCONST_BARE;
3642 if (tmp < 0) { /* second-class keyword? */
3643 GV *ogv = Nullgv; /* override (winner) */
3644 GV *hgv = Nullgv; /* hidden (loser) */
3645 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3647 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3650 if (GvIMPORTED_CV(gv))
3652 else if (! CvMETHOD(cv))
3656 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3657 (gv = *gvp) != (GV*)&PL_sv_undef &&
3658 GvCVu(gv) && GvIMPORTED_CV(gv))
3664 tmp = 0; /* overridden by import or by GLOBAL */
3667 && -tmp==KEY_lock /* XXX generalizable kludge */
3669 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3671 tmp = 0; /* any sub overrides "weak" keyword */
3673 else { /* no override */
3677 if (ckWARN(WARN_AMBIGUOUS) && hgv
3678 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3679 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3680 "Ambiguous call resolved as CORE::%s(), %s",
3681 GvENAME(hgv), "qualify as such or use &");
3688 default: /* not a keyword */
3691 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3693 /* Get the rest if it looks like a package qualifier */
3695 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3697 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3700 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3701 *s == '\'' ? "'" : "::");
3705 if (PL_expect == XOPERATOR) {
3706 if (PL_bufptr == PL_linestart) {
3707 CopLINE_dec(PL_curcop);
3708 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3709 CopLINE_inc(PL_curcop);
3712 no_op("Bareword",s);
3715 /* Look for a subroutine with this name in current package,
3716 unless name is "Foo::", in which case Foo is a bearword
3717 (and a package name). */
3720 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3722 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3723 Perl_warner(aTHX_ WARN_BAREWORD,
3724 "Bareword \"%s\" refers to nonexistent package",
3727 PL_tokenbuf[len] = '\0';
3734 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3737 /* if we saw a global override before, get the right name */
3740 sv = newSVpvn("CORE::GLOBAL::",14);
3741 sv_catpv(sv,PL_tokenbuf);
3744 sv = newSVpv(PL_tokenbuf,0);
3746 /* Presume this is going to be a bareword of some sort. */
3749 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3750 yylval.opval->op_private = OPpCONST_BARE;
3752 /* And if "Foo::", then that's what it certainly is. */
3757 /* See if it's the indirect object for a list operator. */
3759 if (PL_oldoldbufptr &&
3760 PL_oldoldbufptr < PL_bufptr &&
3761 (PL_oldoldbufptr == PL_last_lop
3762 || PL_oldoldbufptr == PL_last_uni) &&
3763 /* NO SKIPSPACE BEFORE HERE! */
3764 (PL_expect == XREF ||
3765 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3767 bool immediate_paren = *s == '(';
3769 /* (Now we can afford to cross potential line boundary.) */
3772 /* Two barewords in a row may indicate method call. */
3774 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3777 /* If not a declared subroutine, it's an indirect object. */
3778 /* (But it's an indir obj regardless for sort.) */
3780 if ((PL_last_lop_op == OP_SORT ||
3781 (!immediate_paren && (!gv || !GvCVu(gv)))) &&
3782 (PL_last_lop_op != OP_MAPSTART &&
3783 PL_last_lop_op != OP_GREPSTART))
3785 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3791 PL_expect = XOPERATOR;
3794 /* Is this a word before a => operator? */
3795 if (*s == '=' && s[1] == '>') {
3797 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
3801 /* If followed by a paren, it's certainly a subroutine. */
3804 if (gv && GvCVu(gv)) {
3805 for (d = s + 1; *d == ' ' || *d == '\t'; d++) ;
3806 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3811 PL_nextval[PL_nexttoke].opval = yylval.opval;
3812 PL_expect = XOPERATOR;
3818 /* If followed by var or block, call it a method (unless sub) */
3820 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3821 PL_last_lop = PL_oldbufptr;
3822 PL_last_lop_op = OP_METHOD;
3826 /* If followed by a bareword, see if it looks like indir obj. */
3828 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3831 /* Not a method, so call it a subroutine (if defined) */
3833 if (gv && GvCVu(gv)) {
3835 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3836 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3837 "Ambiguous use of -%s resolved as -&%s()",
3838 PL_tokenbuf, PL_tokenbuf);
3839 /* Check for a constant sub */
3841 if ((sv = cv_const_sv(cv))) {
3843 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3844 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3845 yylval.opval->op_private = 0;
3849 /* Resolve to GV now. */
3850 op_free(yylval.opval);
3851 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3852 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3853 PL_last_lop = PL_oldbufptr;
3854 PL_last_lop_op = OP_ENTERSUB;
3855 /* Is there a prototype? */
3858 char *proto = SvPV((SV*)cv, len);
3861 if (strEQ(proto, "$"))
3863 if (*proto == '&' && *s == '{') {
3864 sv_setpv(PL_subname,"__ANON__");
3868 PL_nextval[PL_nexttoke].opval = yylval.opval;
3874 /* Call it a bare word */
3876 if (PL_hints & HINT_STRICT_SUBS)
3877 yylval.opval->op_private |= OPpCONST_STRICT;
3880 if (ckWARN(WARN_RESERVED)) {
3881 if (lastchar != '-') {
3882 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
3884 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
3891 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
3892 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3893 "Operator or semicolon missing before %c%s",
3894 lastchar, PL_tokenbuf);
3895 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3896 "Ambiguous use of %c resolved as operator %c",
3897 lastchar, lastchar);
3903 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3904 newSVpv(CopFILE(PL_curcop),0));
3908 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3909 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
3912 case KEY___PACKAGE__:
3913 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3915 ? newSVsv(PL_curstname)
3924 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
3925 char *pname = "main";
3926 if (PL_tokenbuf[2] == 'D')
3927 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
3928 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
3931 GvIOp(gv) = newIO();
3932 IoIFP(GvIOp(gv)) = PL_rsfp;
3933 #if defined(HAS_FCNTL) && defined(F_SETFD)
3935 int fd = PerlIO_fileno(PL_rsfp);
3936 fcntl(fd,F_SETFD,fd >= 3);
3939 /* Mark this internal pseudo-handle as clean */
3940 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
3942 IoTYPE(GvIOp(gv)) = '|';
3943 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
3944 IoTYPE(GvIOp(gv)) = '-';
3946 IoTYPE(GvIOp(gv)) = '<';
3947 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
3948 /* if the script was opened in binmode, we need to revert
3949 * it to text mode for compatibility; but only iff it has CRs
3950 * XXX this is a questionable hack at best. */
3951 if (PL_bufend-PL_bufptr > 2
3952 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
3955 if (IoTYPE(GvIOp(gv)) == '<') {
3956 loc = PerlIO_tell(PL_rsfp);
3957 (void)PerlIO_seek(PL_rsfp, 0L, 0);
3959 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
3960 #if defined(__BORLANDC__)
3961 /* XXX see note in do_binmode() */
3962 ((FILE*)PL_rsfp)->flags |= _F_BIN;
3965 PerlIO_seek(PL_rsfp, loc, 0);
3980 if (PL_expect == XSTATE) {
3987 if (*s == ':' && s[1] == ':') {
3990 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3991 if (!(tmp = keyword(PL_tokenbuf, len)))
3992 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4006 LOP(OP_ACCEPT,XTERM);
4012 LOP(OP_ATAN2,XTERM);
4018 LOP(OP_BINMODE,XTERM);
4021 LOP(OP_BLESS,XTERM);
4030 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4047 if (!PL_cryptseen) {
4048 PL_cryptseen = TRUE;
4052 LOP(OP_CRYPT,XTERM);
4055 if (ckWARN(WARN_CHMOD)) {
4056 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4057 if (*d != '0' && isDIGIT(*d))
4058 Perl_warner(aTHX_ WARN_CHMOD,
4059 "chmod() mode argument is missing initial 0");
4061 LOP(OP_CHMOD,XTERM);
4064 LOP(OP_CHOWN,XTERM);
4067 LOP(OP_CONNECT,XTERM);
4083 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4087 PL_hints |= HINT_BLOCK_SCOPE;
4097 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4098 LOP(OP_DBMOPEN,XTERM);
4104 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4111 yylval.ival = CopLINE(PL_curcop);
4125 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4126 UNIBRACK(OP_ENTEREVAL);
4141 case KEY_endhostent:
4147 case KEY_endservent:
4150 case KEY_endprotoent:
4161 yylval.ival = CopLINE(PL_curcop);
4163 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4165 if ((PL_bufend - p) >= 3 &&
4166 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4168 else if ((PL_bufend - p) >= 4 &&
4169 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4172 if (isIDFIRST_lazy_if(p,UTF)) {
4173 p = scan_ident(p, PL_bufend,
4174 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4178 Perl_croak(aTHX_ "Missing $ on loop variable");
4183 LOP(OP_FORMLINE,XTERM);
4189 LOP(OP_FCNTL,XTERM);
4195 LOP(OP_FLOCK,XTERM);
4204 LOP(OP_GREPSTART, XREF);
4207 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4222 case KEY_getpriority:
4223 LOP(OP_GETPRIORITY,XTERM);
4225 case KEY_getprotobyname:
4228 case KEY_getprotobynumber:
4229 LOP(OP_GPBYNUMBER,XTERM);
4231 case KEY_getprotoent:
4243 case KEY_getpeername:
4244 UNI(OP_GETPEERNAME);
4246 case KEY_gethostbyname:
4249 case KEY_gethostbyaddr:
4250 LOP(OP_GHBYADDR,XTERM);
4252 case KEY_gethostent:
4255 case KEY_getnetbyname:
4258 case KEY_getnetbyaddr:
4259 LOP(OP_GNBYADDR,XTERM);
4264 case KEY_getservbyname:
4265 LOP(OP_GSBYNAME,XTERM);
4267 case KEY_getservbyport:
4268 LOP(OP_GSBYPORT,XTERM);
4270 case KEY_getservent:
4273 case KEY_getsockname:
4274 UNI(OP_GETSOCKNAME);
4276 case KEY_getsockopt:
4277 LOP(OP_GSOCKOPT,XTERM);
4299 yylval.ival = CopLINE(PL_curcop);
4303 LOP(OP_INDEX,XTERM);
4309 LOP(OP_IOCTL,XTERM);
4321 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4353 LOP(OP_LISTEN,XTERM);
4362 s = scan_pat(s,OP_MATCH);
4363 TERM(sublex_start());
4366 LOP(OP_MAPSTART, XREF);
4369 LOP(OP_MKDIR,XTERM);
4372 LOP(OP_MSGCTL,XTERM);
4375 LOP(OP_MSGGET,XTERM);
4378 LOP(OP_MSGRCV,XTERM);
4381 LOP(OP_MSGSND,XTERM);
4387 if (isIDFIRST_lazy_if(s,UTF)) {
4388 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4389 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4391 PL_in_my_stash = gv_stashpv(PL_tokenbuf, FALSE);
4392 if (!PL_in_my_stash) {
4395 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4403 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4410 if (PL_expect != XSTATE)
4411 yyerror("\"no\" not allowed in expression");
4412 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4413 s = force_version(s);
4418 if (*s == '(' || (s = skipspace(s), *s == '('))
4425 if (isIDFIRST_lazy_if(s,UTF)) {
4427 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4429 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4430 Perl_warner(aTHX_ WARN_PRECEDENCE,
4431 "Precedence problem: open %.*s should be open(%.*s)",
4437 yylval.ival = OP_OR;
4447 LOP(OP_OPEN_DIR,XTERM);
4450 checkcomma(s,PL_tokenbuf,"filehandle");
4454 checkcomma(s,PL_tokenbuf,"filehandle");
4473 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4477 LOP(OP_PIPE_OP,XTERM);
4480 s = scan_str(s,FALSE,FALSE);
4482 missingterm((char*)0);
4483 yylval.ival = OP_CONST;
4484 TERM(sublex_start());
4490 s = scan_str(s,FALSE,FALSE);
4492 missingterm((char*)0);
4494 if (SvCUR(PL_lex_stuff)) {
4497 d = SvPV_force(PL_lex_stuff, len);
4499 for (; isSPACE(*d) && len; --len, ++d) ;
4502 if (!warned && ckWARN(WARN_QW)) {
4503 for (; !isSPACE(*d) && len; --len, ++d) {
4505 Perl_warner(aTHX_ WARN_QW,
4506 "Possible attempt to separate words with commas");
4509 else if (*d == '#') {
4510 Perl_warner(aTHX_ WARN_QW,
4511 "Possible attempt to put comments in qw() list");
4517 for (; !isSPACE(*d) && len; --len, ++d) ;
4519 words = append_elem(OP_LIST, words,
4520 newSVOP(OP_CONST, 0, tokeq(newSVpvn(b, d-b))));
4524 PL_nextval[PL_nexttoke].opval = words;
4529 SvREFCNT_dec(PL_lex_stuff);
4530 PL_lex_stuff = Nullsv;
4535 s = scan_str(s,FALSE,FALSE);
4537 missingterm((char*)0);
4538 yylval.ival = OP_STRINGIFY;
4539 if (SvIVX(PL_lex_stuff) == '\'')
4540 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4541 TERM(sublex_start());
4544 s = scan_pat(s,OP_QR);
4545 TERM(sublex_start());
4548 s = scan_str(s,FALSE,FALSE);
4550 missingterm((char*)0);
4551 yylval.ival = OP_BACKTICK;
4553 TERM(sublex_start());
4560 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4561 s = force_version(s);
4564 *PL_tokenbuf = '\0';
4565 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4566 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4567 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4569 yyerror("<> should be quotes");
4577 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4581 LOP(OP_RENAME,XTERM);
4590 LOP(OP_RINDEX,XTERM);
4613 LOP(OP_REVERSE,XTERM);
4624 TERM(sublex_start());
4626 TOKEN(1); /* force error */
4635 LOP(OP_SELECT,XTERM);
4641 LOP(OP_SEMCTL,XTERM);
4644 LOP(OP_SEMGET,XTERM);
4647 LOP(OP_SEMOP,XTERM);
4653 LOP(OP_SETPGRP,XTERM);
4655 case KEY_setpriority:
4656 LOP(OP_SETPRIORITY,XTERM);
4658 case KEY_sethostent:
4664 case KEY_setservent:
4667 case KEY_setprotoent:
4677 LOP(OP_SEEKDIR,XTERM);
4679 case KEY_setsockopt:
4680 LOP(OP_SSOCKOPT,XTERM);
4686 LOP(OP_SHMCTL,XTERM);
4689 LOP(OP_SHMGET,XTERM);
4692 LOP(OP_SHMREAD,XTERM);
4695 LOP(OP_SHMWRITE,XTERM);
4698 LOP(OP_SHUTDOWN,XTERM);
4707 LOP(OP_SOCKET,XTERM);
4709 case KEY_socketpair:
4710 LOP(OP_SOCKPAIR,XTERM);
4713 checkcomma(s,PL_tokenbuf,"subroutine name");
4715 if (*s == ';' || *s == ')') /* probably a close */
4716 Perl_croak(aTHX_ "sort is now a reserved word");
4718 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4722 LOP(OP_SPLIT,XTERM);
4725 LOP(OP_SPRINTF,XTERM);
4728 LOP(OP_SPLICE,XTERM);
4743 LOP(OP_SUBSTR,XTERM);
4749 char tmpbuf[sizeof PL_tokenbuf];
4751 expectation attrful;
4752 bool have_name, have_proto;
4757 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4758 (*s == ':' && s[1] == ':'))
4761 attrful = XATTRBLOCK;
4762 /* remember buffer pos'n for later force_word */
4763 tboffset = s - PL_oldbufptr;
4764 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4765 if (strchr(tmpbuf, ':'))
4766 sv_setpv(PL_subname, tmpbuf);
4768 sv_setsv(PL_subname,PL_curstname);
4769 sv_catpvn(PL_subname,"::",2);
4770 sv_catpvn(PL_subname,tmpbuf,len);
4777 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4778 PL_expect = XTERMBLOCK;
4779 attrful = XATTRTERM;
4780 sv_setpv(PL_subname,"?");
4784 if (key == KEY_format) {
4786 PL_lex_formbrack = PL_lex_brackets + 1;
4788 (void) force_word(PL_oldbufptr + tboffset, WORD,
4793 /* Look for a prototype */
4797 s = scan_str(s,FALSE,FALSE);
4800 SvREFCNT_dec(PL_lex_stuff);
4801 PL_lex_stuff = Nullsv;
4802 Perl_croak(aTHX_ "Prototype not terminated");
4805 d = SvPVX(PL_lex_stuff);
4807 for (p = d; *p; ++p) {
4812 SvCUR(PL_lex_stuff) = tmp;
4820 if (*s == ':' && s[1] != ':')
4821 PL_expect = attrful;
4824 PL_nextval[PL_nexttoke].opval =
4825 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4826 PL_lex_stuff = Nullsv;
4830 sv_setpv(PL_subname,"__ANON__");
4833 (void) force_word(PL_oldbufptr + tboffset, WORD,
4842 LOP(OP_SYSTEM,XREF);
4845 LOP(OP_SYMLINK,XTERM);
4848 LOP(OP_SYSCALL,XTERM);
4851 LOP(OP_SYSOPEN,XTERM);
4854 LOP(OP_SYSSEEK,XTERM);
4857 LOP(OP_SYSREAD,XTERM);
4860 LOP(OP_SYSWRITE,XTERM);
4864 TERM(sublex_start());
4885 LOP(OP_TRUNCATE,XTERM);
4897 yylval.ival = CopLINE(PL_curcop);
4901 yylval.ival = CopLINE(PL_curcop);
4905 LOP(OP_UNLINK,XTERM);
4911 LOP(OP_UNPACK,XTERM);
4914 LOP(OP_UTIME,XTERM);
4917 if (ckWARN(WARN_UMASK)) {
4918 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4919 if (*d != '0' && isDIGIT(*d))
4920 Perl_warner(aTHX_ WARN_UMASK,
4921 "umask: argument is missing initial 0");
4926 LOP(OP_UNSHIFT,XTERM);
4929 if (PL_expect != XSTATE)
4930 yyerror("\"use\" not allowed in expression");
4932 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4933 s = force_version(s);
4934 if (*s == ';' || (s = skipspace(s), *s == ';')) {
4935 PL_nextval[PL_nexttoke].opval = Nullop;
4940 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4941 s = force_version(s);
4953 yylval.ival = CopLINE(PL_curcop);
4957 PL_hints |= HINT_BLOCK_SCOPE;
4964 LOP(OP_WAITPID,XTERM);
4972 static char ctl_l[2];
4974 if (ctl_l[0] == '\0')
4975 ctl_l[0] = toCTRL('L');
4976 gv_fetchpv(ctl_l,TRUE, SVt_PV);
4979 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
4984 if (PL_expect == XOPERATOR)
4990 yylval.ival = OP_XOR;
4995 TERM(sublex_start());
5001 Perl_keyword(pTHX_ register char *d, I32 len)
5006 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5007 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5008 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5009 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5010 if (strEQ(d,"__END__")) return KEY___END__;
5014 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5019 if (strEQ(d,"and")) return -KEY_and;
5020 if (strEQ(d,"abs")) return -KEY_abs;
5023 if (strEQ(d,"alarm")) return -KEY_alarm;
5024 if (strEQ(d,"atan2")) return -KEY_atan2;
5027 if (strEQ(d,"accept")) return -KEY_accept;
5032 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5035 if (strEQ(d,"bless")) return -KEY_bless;
5036 if (strEQ(d,"bind")) return -KEY_bind;
5037 if (strEQ(d,"binmode")) return -KEY_binmode;
5040 if (strEQ(d,"CORE")) return -KEY_CORE;
5041 if (strEQ(d,"CHECK")) return KEY_CHECK;
5046 if (strEQ(d,"cmp")) return -KEY_cmp;
5047 if (strEQ(d,"chr")) return -KEY_chr;
5048 if (strEQ(d,"cos")) return -KEY_cos;
5051 if (strEQ(d,"chop")) return KEY_chop;
5054 if (strEQ(d,"close")) return -KEY_close;
5055 if (strEQ(d,"chdir")) return -KEY_chdir;
5056 if (strEQ(d,"chomp")) return KEY_chomp;
5057 if (strEQ(d,"chmod")) return -KEY_chmod;
5058 if (strEQ(d,"chown")) return -KEY_chown;
5059 if (strEQ(d,"crypt")) return -KEY_crypt;
5062 if (strEQ(d,"chroot")) return -KEY_chroot;
5063 if (strEQ(d,"caller")) return -KEY_caller;
5066 if (strEQ(d,"connect")) return -KEY_connect;
5069 if (strEQ(d,"closedir")) return -KEY_closedir;
5070 if (strEQ(d,"continue")) return -KEY_continue;
5075 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5080 if (strEQ(d,"do")) return KEY_do;
5083 if (strEQ(d,"die")) return -KEY_die;
5086 if (strEQ(d,"dump")) return -KEY_dump;
5089 if (strEQ(d,"delete")) return KEY_delete;
5092 if (strEQ(d,"defined")) return KEY_defined;
5093 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5096 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5101 if (strEQ(d,"EQ")) { deprecate(d); return -KEY_eq;}
5102 if (strEQ(d,"END")) return KEY_END;
5107 if (strEQ(d,"eq")) return -KEY_eq;
5110 if (strEQ(d,"eof")) return -KEY_eof;
5111 if (strEQ(d,"exp")) return -KEY_exp;
5114 if (strEQ(d,"else")) return KEY_else;
5115 if (strEQ(d,"exit")) return -KEY_exit;
5116 if (strEQ(d,"eval")) return KEY_eval;
5117 if (strEQ(d,"exec")) return -KEY_exec;
5118 if (strEQ(d,"each")) return KEY_each;
5121 if (strEQ(d,"elsif")) return KEY_elsif;
5124 if (strEQ(d,"exists")) return KEY_exists;
5125 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5128 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5129 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5132 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5135 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5136 if (strEQ(d,"endservent")) return -KEY_endservent;
5139 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5146 if (strEQ(d,"for")) return KEY_for;
5149 if (strEQ(d,"fork")) return -KEY_fork;
5152 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5153 if (strEQ(d,"flock")) return -KEY_flock;
5156 if (strEQ(d,"format")) return KEY_format;
5157 if (strEQ(d,"fileno")) return -KEY_fileno;
5160 if (strEQ(d,"foreach")) return KEY_foreach;
5163 if (strEQ(d,"formline")) return -KEY_formline;
5169 if (strEQ(d,"GT")) { deprecate(d); return -KEY_gt;}
5170 if (strEQ(d,"GE")) { deprecate(d); return -KEY_ge;}
5174 if (strnEQ(d,"get",3)) {
5179 if (strEQ(d,"ppid")) return -KEY_getppid;
5180 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5183 if (strEQ(d,"pwent")) return -KEY_getpwent;
5184 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5185 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5188 if (strEQ(d,"peername")) return -KEY_getpeername;
5189 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5190 if (strEQ(d,"priority")) return -KEY_getpriority;
5193 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5196 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5200 else if (*d == 'h') {
5201 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5202 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5203 if (strEQ(d,"hostent")) return -KEY_gethostent;
5205 else if (*d == 'n') {
5206 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5207 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5208 if (strEQ(d,"netent")) return -KEY_getnetent;
5210 else if (*d == 's') {
5211 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5212 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5213 if (strEQ(d,"servent")) return -KEY_getservent;
5214 if (strEQ(d,"sockname")) return -KEY_getsockname;
5215 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5217 else if (*d == 'g') {
5218 if (strEQ(d,"grent")) return -KEY_getgrent;
5219 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5220 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5222 else if (*d == 'l') {
5223 if (strEQ(d,"login")) return -KEY_getlogin;
5225 else if (strEQ(d,"c")) return -KEY_getc;
5230 if (strEQ(d,"gt")) return -KEY_gt;
5231 if (strEQ(d,"ge")) return -KEY_ge;
5234 if (strEQ(d,"grep")) return KEY_grep;
5235 if (strEQ(d,"goto")) return KEY_goto;
5236 if (strEQ(d,"glob")) return KEY_glob;
5239 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5244 if (strEQ(d,"hex")) return -KEY_hex;
5247 if (strEQ(d,"INIT")) return KEY_INIT;
5252 if (strEQ(d,"if")) return KEY_if;
5255 if (strEQ(d,"int")) return -KEY_int;
5258 if (strEQ(d,"index")) return -KEY_index;
5259 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5264 if (strEQ(d,"join")) return -KEY_join;
5268 if (strEQ(d,"keys")) return KEY_keys;
5269 if (strEQ(d,"kill")) return -KEY_kill;
5274 if (strEQ(d,"LT")) { deprecate(d); return -KEY_lt;}
5275 if (strEQ(d,"LE")) { deprecate(d); return -KEY_le;}
5281 if (strEQ(d,"lt")) return -KEY_lt;
5282 if (strEQ(d,"le")) return -KEY_le;
5283 if (strEQ(d,"lc")) return -KEY_lc;
5286 if (strEQ(d,"log")) return -KEY_log;
5289 if (strEQ(d,"last")) return KEY_last;
5290 if (strEQ(d,"link")) return -KEY_link;
5291 if (strEQ(d,"lock")) return -KEY_lock;
5294 if (strEQ(d,"local")) return KEY_local;
5295 if (strEQ(d,"lstat")) return -KEY_lstat;
5298 if (strEQ(d,"length")) return -KEY_length;
5299 if (strEQ(d,"listen")) return -KEY_listen;
5302 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5305 if (strEQ(d,"localtime")) return -KEY_localtime;
5311 case 1: return KEY_m;
5313 if (strEQ(d,"my")) return KEY_my;
5316 if (strEQ(d,"map")) return KEY_map;
5319 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5322 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5323 if (strEQ(d,"msgget")) return -KEY_msgget;
5324 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5325 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5330 if (strEQ(d,"NE")) { deprecate(d); return -KEY_ne;}
5333 if (strEQ(d,"next")) return KEY_next;
5334 if (strEQ(d,"ne")) return -KEY_ne;
5335 if (strEQ(d,"not")) return -KEY_not;
5336 if (strEQ(d,"no")) return KEY_no;
5341 if (strEQ(d,"or")) return -KEY_or;
5344 if (strEQ(d,"ord")) return -KEY_ord;
5345 if (strEQ(d,"oct")) return -KEY_oct;
5346 if (strEQ(d,"our")) return KEY_our;
5349 if (strEQ(d,"open")) return -KEY_open;
5352 if (strEQ(d,"opendir")) return -KEY_opendir;
5359 if (strEQ(d,"pop")) return KEY_pop;
5360 if (strEQ(d,"pos")) return KEY_pos;
5363 if (strEQ(d,"push")) return KEY_push;
5364 if (strEQ(d,"pack")) return -KEY_pack;
5365 if (strEQ(d,"pipe")) return -KEY_pipe;
5368 if (strEQ(d,"print")) return KEY_print;
5371 if (strEQ(d,"printf")) return KEY_printf;
5374 if (strEQ(d,"package")) return KEY_package;
5377 if (strEQ(d,"prototype")) return KEY_prototype;
5382 if (strEQ(d,"q")) return KEY_q;
5383 if (strEQ(d,"qr")) return KEY_qr;
5384 if (strEQ(d,"qq")) return KEY_qq;
5385 if (strEQ(d,"qw")) return KEY_qw;
5386 if (strEQ(d,"qx")) return KEY_qx;
5388 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5393 if (strEQ(d,"ref")) return -KEY_ref;
5396 if (strEQ(d,"read")) return -KEY_read;
5397 if (strEQ(d,"rand")) return -KEY_rand;
5398 if (strEQ(d,"recv")) return -KEY_recv;
5399 if (strEQ(d,"redo")) return KEY_redo;
5402 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5403 if (strEQ(d,"reset")) return -KEY_reset;
5406 if (strEQ(d,"return")) return KEY_return;
5407 if (strEQ(d,"rename")) return -KEY_rename;
5408 if (strEQ(d,"rindex")) return -KEY_rindex;
5411 if (strEQ(d,"require")) return -KEY_require;
5412 if (strEQ(d,"reverse")) return -KEY_reverse;
5413 if (strEQ(d,"readdir")) return -KEY_readdir;
5416 if (strEQ(d,"readlink")) return -KEY_readlink;
5417 if (strEQ(d,"readline")) return -KEY_readline;
5418 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5421 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5427 case 0: return KEY_s;
5429 if (strEQ(d,"scalar")) return KEY_scalar;
5434 if (strEQ(d,"seek")) return -KEY_seek;
5435 if (strEQ(d,"send")) return -KEY_send;
5438 if (strEQ(d,"semop")) return -KEY_semop;
5441 if (strEQ(d,"select")) return -KEY_select;
5442 if (strEQ(d,"semctl")) return -KEY_semctl;
5443 if (strEQ(d,"semget")) return -KEY_semget;
5446 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5447 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5450 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5451 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5454 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5457 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5458 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5459 if (strEQ(d,"setservent")) return -KEY_setservent;
5462 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5463 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5470 if (strEQ(d,"shift")) return KEY_shift;
5473 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5474 if (strEQ(d,"shmget")) return -KEY_shmget;
5477 if (strEQ(d,"shmread")) return -KEY_shmread;
5480 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5481 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5486 if (strEQ(d,"sin")) return -KEY_sin;
5489 if (strEQ(d,"sleep")) return -KEY_sleep;
5492 if (strEQ(d,"sort")) return KEY_sort;
5493 if (strEQ(d,"socket")) return -KEY_socket;
5494 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5497 if (strEQ(d,"split")) return KEY_split;
5498 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5499 if (strEQ(d,"splice")) return KEY_splice;
5502 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5505 if (strEQ(d,"srand")) return -KEY_srand;
5508 if (strEQ(d,"stat")) return -KEY_stat;
5509 if (strEQ(d,"study")) return KEY_study;
5512 if (strEQ(d,"substr")) return -KEY_substr;
5513 if (strEQ(d,"sub")) return KEY_sub;
5518 if (strEQ(d,"system")) return -KEY_system;
5521 if (strEQ(d,"symlink")) return -KEY_symlink;
5522 if (strEQ(d,"syscall")) return -KEY_syscall;
5523 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5524 if (strEQ(d,"sysread")) return -KEY_sysread;
5525 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5528 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5537 if (strEQ(d,"tr")) return KEY_tr;
5540 if (strEQ(d,"tie")) return KEY_tie;
5543 if (strEQ(d,"tell")) return -KEY_tell;
5544 if (strEQ(d,"tied")) return KEY_tied;
5545 if (strEQ(d,"time")) return -KEY_time;
5548 if (strEQ(d,"times")) return -KEY_times;
5551 if (strEQ(d,"telldir")) return -KEY_telldir;
5554 if (strEQ(d,"truncate")) return -KEY_truncate;
5561 if (strEQ(d,"uc")) return -KEY_uc;
5564 if (strEQ(d,"use")) return KEY_use;
5567 if (strEQ(d,"undef")) return KEY_undef;
5568 if (strEQ(d,"until")) return KEY_until;
5569 if (strEQ(d,"untie")) return KEY_untie;
5570 if (strEQ(d,"utime")) return -KEY_utime;
5571 if (strEQ(d,"umask")) return -KEY_umask;
5574 if (strEQ(d,"unless")) return KEY_unless;
5575 if (strEQ(d,"unpack")) return -KEY_unpack;
5576 if (strEQ(d,"unlink")) return -KEY_unlink;
5579 if (strEQ(d,"unshift")) return KEY_unshift;
5580 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5585 if (strEQ(d,"values")) return -KEY_values;
5586 if (strEQ(d,"vec")) return -KEY_vec;
5591 if (strEQ(d,"warn")) return -KEY_warn;
5592 if (strEQ(d,"wait")) return -KEY_wait;
5595 if (strEQ(d,"while")) return KEY_while;
5596 if (strEQ(d,"write")) return -KEY_write;
5599 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5602 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5607 if (len == 1) return -KEY_x;
5608 if (strEQ(d,"xor")) return -KEY_xor;
5611 if (len == 1) return KEY_y;
5620 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5624 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5625 dTHR; /* only for ckWARN */
5626 if (ckWARN(WARN_SYNTAX)) {
5628 for (w = s+2; *w && level; w++) {
5635 for (; *w && isSPACE(*w); w++) ;
5636 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5637 Perl_warner(aTHX_ WARN_SYNTAX,
5638 "%s (...) interpreted as function",name);
5641 while (s < PL_bufend && isSPACE(*s))
5645 while (s < PL_bufend && isSPACE(*s))
5647 if (isIDFIRST_lazy_if(s,UTF)) {
5649 while (isALNUM_lazy_if(s,UTF))
5651 while (s < PL_bufend && isSPACE(*s))
5656 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5660 Perl_croak(aTHX_ "No comma allowed after %s", what);
5665 /* Either returns sv, or mortalizes sv and returns a new SV*.
5666 Best used as sv=new_constant(..., sv, ...).
5667 If s, pv are NULL, calls subroutine with one argument,
5668 and type is used with error messages only. */
5671 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5675 HV *table = GvHV(PL_hintgv); /* ^H */
5679 const char *why1, *why2, *why3;
5681 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
5684 why1 = "%^H is not consistent";
5685 why2 = strEQ(key,"charnames")
5686 ? " (missing \"use charnames ...\"?)"
5690 msg = Perl_newSVpvf(aTHX_ "constant(%s): %s%s%s",
5691 (type ? type: "undef"), why1, why2, why3);
5692 yyerror(SvPVX(msg));
5696 cvp = hv_fetch(table, key, strlen(key), FALSE);
5697 if (!cvp || !SvOK(*cvp)) {
5700 why3 = "} is not defined";
5703 sv_2mortal(sv); /* Parent created it permanently */
5706 pv = sv_2mortal(newSVpvn(s, len));
5708 typesv = sv_2mortal(newSVpv(type, 0));
5710 typesv = &PL_sv_undef;
5712 PUSHSTACKi(PERLSI_OVERLOAD);
5725 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5729 /* Check the eval first */
5730 if (!PL_in_eval && SvTRUE(ERRSV)) {
5732 sv_catpv(ERRSV, "Propagated");
5733 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5735 res = SvREFCNT_inc(sv);
5739 (void)SvREFCNT_inc(res);
5748 why1 = "Call to &{$^H{";
5750 why3 = "}} did not return a defined value";
5759 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5761 register char *d = dest;
5762 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5765 Perl_croak(aTHX_ ident_too_long);
5766 if (isALNUM(*s)) /* UTF handled below */
5768 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
5773 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5777 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5778 char *t = s + UTF8SKIP(s);
5779 while (*t & 0x80 && is_utf8_mark((U8*)t))
5781 if (d + (t - s) > e)
5782 Perl_croak(aTHX_ ident_too_long);
5783 Copy(s, d, t - s, char);
5796 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5806 e = d + destlen - 3; /* two-character token, ending NUL */
5808 while (isDIGIT(*s)) {
5810 Perl_croak(aTHX_ ident_too_long);
5817 Perl_croak(aTHX_ ident_too_long);
5818 if (isALNUM(*s)) /* UTF handled below */
5820 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
5825 else if (*s == ':' && s[1] == ':') {
5829 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5830 char *t = s + UTF8SKIP(s);
5831 while (*t & 0x80 && is_utf8_mark((U8*)t))
5833 if (d + (t - s) > e)
5834 Perl_croak(aTHX_ ident_too_long);
5835 Copy(s, d, t - s, char);
5846 if (PL_lex_state != LEX_NORMAL)
5847 PL_lex_state = LEX_INTERPENDMAYBE;
5850 if (*s == '$' && s[1] &&
5851 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
5864 if (*d == '^' && *s && isCONTROLVAR(*s)) {
5869 if (isSPACE(s[-1])) {
5872 if (ch != ' ' && ch != '\t') {
5878 if (isIDFIRST_lazy_if(d,UTF)) {
5882 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
5884 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
5887 Copy(s, d, e - s, char);
5892 while ((isALNUM(*s) || *s == ':') && d < e)
5895 Perl_croak(aTHX_ ident_too_long);
5898 while (s < send && (*s == ' ' || *s == '\t')) s++;
5899 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
5900 dTHR; /* only for ckWARN */
5901 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
5902 const char *brack = *s == '[' ? "[...]" : "{...}";
5903 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5904 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
5905 funny, dest, brack, funny, dest, brack);
5908 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
5912 /* Handle extended ${^Foo} variables
5913 * 1999-02-27 mjd-perl-patch@plover.com */
5914 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
5918 while (isALNUM(*s) && d < e) {
5922 Perl_croak(aTHX_ ident_too_long);
5927 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
5928 PL_lex_state = LEX_INTERPEND;
5931 if (PL_lex_state == LEX_NORMAL) {
5932 dTHR; /* only for ckWARN */
5933 if (ckWARN(WARN_AMBIGUOUS) &&
5934 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
5936 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5937 "Ambiguous use of %c{%s} resolved to %c%s",
5938 funny, dest, funny, dest);
5943 s = bracket; /* let the parser handle it */
5947 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
5948 PL_lex_state = LEX_INTERPEND;
5953 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
5958 *pmfl |= PMf_GLOBAL;
5960 *pmfl |= PMf_CONTINUE;
5964 *pmfl |= PMf_MULTILINE;
5966 *pmfl |= PMf_SINGLELINE;
5968 *pmfl |= PMf_EXTENDED;
5972 S_scan_pat(pTHX_ char *start, I32 type)
5977 s = scan_str(start,FALSE,FALSE);
5980 SvREFCNT_dec(PL_lex_stuff);
5981 PL_lex_stuff = Nullsv;
5982 Perl_croak(aTHX_ "Search pattern not terminated");
5985 pm = (PMOP*)newPMOP(type, 0);
5986 if (PL_multi_open == '?')
5987 pm->op_pmflags |= PMf_ONCE;
5989 while (*s && strchr("iomsx", *s))
5990 pmflag(&pm->op_pmflags,*s++);
5993 while (*s && strchr("iogcmsx", *s))
5994 pmflag(&pm->op_pmflags,*s++);
5996 pm->op_pmpermflags = pm->op_pmflags;
5998 PL_lex_op = (OP*)pm;
5999 yylval.ival = OP_MATCH;
6004 S_scan_subst(pTHX_ char *start)
6011 yylval.ival = OP_NULL;
6013 s = scan_str(start,FALSE,FALSE);
6017 SvREFCNT_dec(PL_lex_stuff);
6018 PL_lex_stuff = Nullsv;
6019 Perl_croak(aTHX_ "Substitution pattern not terminated");
6022 if (s[-1] == PL_multi_open)
6025 first_start = PL_multi_start;
6026 s = scan_str(s,FALSE,FALSE);
6029 SvREFCNT_dec(PL_lex_stuff);
6030 PL_lex_stuff = Nullsv;
6032 SvREFCNT_dec(PL_lex_repl);
6033 PL_lex_repl = Nullsv;
6034 Perl_croak(aTHX_ "Substitution replacement not terminated");
6036 PL_multi_start = first_start; /* so whole substitution is taken together */
6038 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6044 else if (strchr("iogcmsx", *s))
6045 pmflag(&pm->op_pmflags,*s++);
6052 PL_sublex_info.super_bufptr = s;
6053 PL_sublex_info.super_bufend = PL_bufend;
6055 pm->op_pmflags |= PMf_EVAL;
6056 repl = newSVpvn("",0);
6058 sv_catpv(repl, es ? "eval " : "do ");
6059 sv_catpvn(repl, "{ ", 2);
6060 sv_catsv(repl, PL_lex_repl);
6061 sv_catpvn(repl, " };", 2);
6063 SvREFCNT_dec(PL_lex_repl);
6067 pm->op_pmpermflags = pm->op_pmflags;
6068 PL_lex_op = (OP*)pm;
6069 yylval.ival = OP_SUBST;
6074 S_scan_trans(pTHX_ char *start)
6085 yylval.ival = OP_NULL;
6087 s = scan_str(start,FALSE,FALSE);
6090 SvREFCNT_dec(PL_lex_stuff);
6091 PL_lex_stuff = Nullsv;
6092 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6094 if (s[-1] == PL_multi_open)
6097 s = scan_str(s,FALSE,FALSE);
6100 SvREFCNT_dec(PL_lex_stuff);
6101 PL_lex_stuff = Nullsv;
6103 SvREFCNT_dec(PL_lex_repl);
6104 PL_lex_repl = Nullsv;
6105 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6109 o = newSVOP(OP_TRANS, 0, 0);
6110 utf8 = OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF;
6113 New(803,tbl,256,short);
6114 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6118 complement = del = squash = 0;
6119 while (strchr("cdsCU", *s)) {
6121 complement = OPpTRANS_COMPLEMENT;
6123 del = OPpTRANS_DELETE;
6125 squash = OPpTRANS_SQUASH;
6130 utf8 &= ~OPpTRANS_FROM_UTF;
6132 utf8 |= OPpTRANS_FROM_UTF;
6136 utf8 &= ~OPpTRANS_TO_UTF;
6138 utf8 |= OPpTRANS_TO_UTF;
6141 Perl_croak(aTHX_ "Too many /C and /U options");
6146 o->op_private = del|squash|complement|utf8;
6149 yylval.ival = OP_TRANS;
6154 S_scan_heredoc(pTHX_ register char *s)
6158 I32 op_type = OP_SCALAR;
6165 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6169 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6172 for (peek = s; *peek == ' ' || *peek == '\t'; peek++) ;
6173 if (*peek && strchr("`'\"",*peek)) {
6176 s = delimcpy(d, e, s, PL_bufend, term, &len);
6186 if (!isALNUM_lazy_if(s,UTF))
6187 deprecate("bare << to mean <<\"\"");
6188 for (; isALNUM_lazy_if(s,UTF); s++) {
6193 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6194 Perl_croak(aTHX_ "Delimiter for here document is too long");
6197 len = d - PL_tokenbuf;
6198 #ifndef PERL_STRICT_CR
6199 d = strchr(s, '\r');
6203 while (s < PL_bufend) {
6209 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6218 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6223 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6224 herewas = newSVpvn(s,PL_bufend-s);
6226 s--, herewas = newSVpvn(s,d-s);
6227 s += SvCUR(herewas);
6229 tmpstr = NEWSV(87,79);
6230 sv_upgrade(tmpstr, SVt_PVIV);
6235 else if (term == '`') {
6236 op_type = OP_BACKTICK;
6237 SvIVX(tmpstr) = '\\';
6241 PL_multi_start = CopLINE(PL_curcop);
6242 PL_multi_open = PL_multi_close = '<';
6243 term = *PL_tokenbuf;
6244 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6245 char *bufptr = PL_sublex_info.super_bufptr;
6246 char *bufend = PL_sublex_info.super_bufend;
6247 char *olds = s - SvCUR(herewas);
6248 s = strchr(bufptr, '\n');
6252 while (s < bufend &&
6253 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6255 CopLINE_inc(PL_curcop);
6258 CopLINE_set(PL_curcop, PL_multi_start);
6259 missingterm(PL_tokenbuf);
6261 sv_setpvn(herewas,bufptr,d-bufptr+1);
6262 sv_setpvn(tmpstr,d+1,s-d);
6264 sv_catpvn(herewas,s,bufend-s);
6265 (void)strcpy(bufptr,SvPVX(herewas));
6272 while (s < PL_bufend &&
6273 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6275 CopLINE_inc(PL_curcop);
6277 if (s >= PL_bufend) {
6278 CopLINE_set(PL_curcop, PL_multi_start);
6279 missingterm(PL_tokenbuf);
6281 sv_setpvn(tmpstr,d+1,s-d);
6283 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6285 sv_catpvn(herewas,s,PL_bufend-s);
6286 sv_setsv(PL_linestr,herewas);
6287 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6288 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6291 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6292 while (s >= PL_bufend) { /* multiple line string? */
6294 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6295 CopLINE_set(PL_curcop, PL_multi_start);
6296 missingterm(PL_tokenbuf);
6298 CopLINE_inc(PL_curcop);
6299 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6300 #ifndef PERL_STRICT_CR
6301 if (PL_bufend - PL_linestart >= 2) {
6302 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6303 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6305 PL_bufend[-2] = '\n';
6307 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6309 else if (PL_bufend[-1] == '\r')
6310 PL_bufend[-1] = '\n';
6312 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6313 PL_bufend[-1] = '\n';
6315 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6316 SV *sv = NEWSV(88,0);
6318 sv_upgrade(sv, SVt_PVMG);
6319 sv_setsv(sv,PL_linestr);
6320 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6322 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6325 sv_catsv(PL_linestr,herewas);
6326 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6330 sv_catsv(tmpstr,PL_linestr);
6335 PL_multi_end = CopLINE(PL_curcop);
6336 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6337 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6338 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6340 SvREFCNT_dec(herewas);
6341 PL_lex_stuff = tmpstr;
6342 yylval.ival = op_type;
6347 takes: current position in input buffer
6348 returns: new position in input buffer
6349 side-effects: yylval and lex_op are set.
6354 <FH> read from filehandle
6355 <pkg::FH> read from package qualified filehandle
6356 <pkg'FH> read from package qualified filehandle
6357 <$fh> read from filehandle in $fh
6363 S_scan_inputsymbol(pTHX_ char *start)
6365 register char *s = start; /* current position in buffer */
6371 d = PL_tokenbuf; /* start of temp holding space */
6372 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6373 end = strchr(s, '\n');
6376 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6378 /* die if we didn't have space for the contents of the <>,
6379 or if it didn't end, or if we see a newline
6382 if (len >= sizeof PL_tokenbuf)
6383 Perl_croak(aTHX_ "Excessively long <> operator");
6385 Perl_croak(aTHX_ "Unterminated <> operator");
6390 Remember, only scalar variables are interpreted as filehandles by
6391 this code. Anything more complex (e.g., <$fh{$num}>) will be
6392 treated as a glob() call.
6393 This code makes use of the fact that except for the $ at the front,
6394 a scalar variable and a filehandle look the same.
6396 if (*d == '$' && d[1]) d++;
6398 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6399 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6402 /* If we've tried to read what we allow filehandles to look like, and
6403 there's still text left, then it must be a glob() and not a getline.
6404 Use scan_str to pull out the stuff between the <> and treat it
6405 as nothing more than a string.
6408 if (d - PL_tokenbuf != len) {
6409 yylval.ival = OP_GLOB;
6411 s = scan_str(start,FALSE,FALSE);
6413 Perl_croak(aTHX_ "Glob not terminated");
6417 /* we're in a filehandle read situation */
6420 /* turn <> into <ARGV> */
6422 (void)strcpy(d,"ARGV");
6424 /* if <$fh>, create the ops to turn the variable into a
6430 /* try to find it in the pad for this block, otherwise find
6431 add symbol table ops
6433 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6434 OP *o = newOP(OP_PADSV, 0);
6436 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6439 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6440 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6441 newUNOP(OP_RV2SV, 0,
6442 newGVOP(OP_GV, 0, gv)));
6444 PL_lex_op->op_flags |= OPf_SPECIAL;
6445 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6446 yylval.ival = OP_NULL;
6449 /* If it's none of the above, it must be a literal filehandle
6450 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6452 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6453 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6454 yylval.ival = OP_NULL;
6463 takes: start position in buffer
6464 keep_quoted preserve \ on the embedded delimiter(s)
6465 keep_delims preserve the delimiters around the string
6466 returns: position to continue reading from buffer
6467 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6468 updates the read buffer.
6470 This subroutine pulls a string out of the input. It is called for:
6471 q single quotes q(literal text)
6472 ' single quotes 'literal text'
6473 qq double quotes qq(interpolate $here please)
6474 " double quotes "interpolate $here please"
6475 qx backticks qx(/bin/ls -l)
6476 ` backticks `/bin/ls -l`
6477 qw quote words @EXPORT_OK = qw( func() $spam )
6478 m// regexp match m/this/
6479 s/// regexp substitute s/this/that/
6480 tr/// string transliterate tr/this/that/
6481 y/// string transliterate y/this/that/
6482 ($*@) sub prototypes sub foo ($)
6483 (stuff) sub attr parameters sub foo : attr(stuff)
6484 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6486 In most of these cases (all but <>, patterns and transliterate)
6487 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6488 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6489 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6492 It skips whitespace before the string starts, and treats the first
6493 character as the delimiter. If the delimiter is one of ([{< then
6494 the corresponding "close" character )]}> is used as the closing
6495 delimiter. It allows quoting of delimiters, and if the string has
6496 balanced delimiters ([{<>}]) it allows nesting.
6498 The lexer always reads these strings into lex_stuff, except in the
6499 case of the operators which take *two* arguments (s/// and tr///)
6500 when it checks to see if lex_stuff is full (presumably with the 1st
6501 arg to s or tr) and if so puts the string into lex_repl.
6506 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6509 SV *sv; /* scalar value: string */
6510 char *tmps; /* temp string, used for delimiter matching */
6511 register char *s = start; /* current position in the buffer */
6512 register char term; /* terminating character */
6513 register char *to; /* current position in the sv's data */
6514 I32 brackets = 1; /* bracket nesting level */
6515 bool has_utf = FALSE; /* is there any utf8 content? */
6517 /* skip space before the delimiter */
6521 /* mark where we are, in case we need to report errors */
6524 /* after skipping whitespace, the next character is the terminator */
6526 if ((term & 0x80) && UTF)
6529 /* mark where we are */
6530 PL_multi_start = CopLINE(PL_curcop);
6531 PL_multi_open = term;
6533 /* find corresponding closing delimiter */
6534 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6536 PL_multi_close = term;
6538 /* create a new SV to hold the contents. 87 is leak category, I'm
6539 assuming. 79 is the SV's initial length. What a random number. */
6541 sv_upgrade(sv, SVt_PVIV);
6543 (void)SvPOK_only(sv); /* validate pointer */
6545 /* move past delimiter and try to read a complete string */
6547 sv_catpvn(sv, s, 1);
6550 /* extend sv if need be */
6551 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6552 /* set 'to' to the next character in the sv's string */
6553 to = SvPVX(sv)+SvCUR(sv);
6555 /* if open delimiter is the close delimiter read unbridle */
6556 if (PL_multi_open == PL_multi_close) {
6557 for (; s < PL_bufend; s++,to++) {
6558 /* embedded newlines increment the current line number */
6559 if (*s == '\n' && !PL_rsfp)
6560 CopLINE_inc(PL_curcop);
6561 /* handle quoted delimiters */
6562 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6563 if (!keep_quoted && s[1] == term)
6565 /* any other quotes are simply copied straight through */
6569 /* terminate when run out of buffer (the for() condition), or
6570 have found the terminator */
6571 else if (*s == term)
6573 else if (!has_utf && (*s & 0x80) && UTF)
6579 /* if the terminator isn't the same as the start character (e.g.,
6580 matched brackets), we have to allow more in the quoting, and
6581 be prepared for nested brackets.
6584 /* read until we run out of string, or we find the terminator */
6585 for (; s < PL_bufend; s++,to++) {
6586 /* embedded newlines increment the line count */
6587 if (*s == '\n' && !PL_rsfp)
6588 CopLINE_inc(PL_curcop);
6589 /* backslashes can escape the open or closing characters */
6590 if (*s == '\\' && s+1 < PL_bufend) {
6592 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6597 /* allow nested opens and closes */
6598 else if (*s == PL_multi_close && --brackets <= 0)
6600 else if (*s == PL_multi_open)
6602 else if (!has_utf && (*s & 0x80) && UTF)
6607 /* terminate the copied string and update the sv's end-of-string */
6609 SvCUR_set(sv, to - SvPVX(sv));
6612 * this next chunk reads more into the buffer if we're not done yet
6616 break; /* handle case where we are done yet :-) */
6618 #ifndef PERL_STRICT_CR
6619 if (to - SvPVX(sv) >= 2) {
6620 if ((to[-2] == '\r' && to[-1] == '\n') ||
6621 (to[-2] == '\n' && to[-1] == '\r'))
6625 SvCUR_set(sv, to - SvPVX(sv));
6627 else if (to[-1] == '\r')
6630 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6634 /* if we're out of file, or a read fails, bail and reset the current
6635 line marker so we can report where the unterminated string began
6638 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6640 CopLINE_set(PL_curcop, PL_multi_start);
6643 /* we read a line, so increment our line counter */
6644 CopLINE_inc(PL_curcop);
6646 /* update debugger info */
6647 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6648 SV *sv = NEWSV(88,0);
6650 sv_upgrade(sv, SVt_PVMG);
6651 sv_setsv(sv,PL_linestr);
6652 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6655 /* having changed the buffer, we must update PL_bufend */
6656 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6659 /* at this point, we have successfully read the delimited string */
6662 sv_catpvn(sv, s, 1);
6665 PL_multi_end = CopLINE(PL_curcop);
6668 /* if we allocated too much space, give some back */
6669 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6670 SvLEN_set(sv, SvCUR(sv) + 1);
6671 Renew(SvPVX(sv), SvLEN(sv), char);
6674 /* decide whether this is the first or second quoted string we've read
6687 takes: pointer to position in buffer
6688 returns: pointer to new position in buffer
6689 side-effects: builds ops for the constant in yylval.op
6691 Read a number in any of the formats that Perl accepts:
6693 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6694 [\d_]+(\.[\d_]*)?[Ee](\d+)
6696 Underbars (_) are allowed in decimal numbers. If -w is on,
6697 underbars before a decimal point must be at three digit intervals.
6699 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6702 If it reads a number without a decimal point or an exponent, it will
6703 try converting the number to an integer and see if it can do so
6704 without loss of precision.
6708 Perl_scan_num(pTHX_ char *start)
6710 register char *s = start; /* current position in buffer */
6711 register char *d; /* destination in temp buffer */
6712 register char *e; /* end of temp buffer */
6713 NV nv; /* number read, as a double */
6714 SV *sv = Nullsv; /* place to put the converted number */
6715 bool floatit; /* boolean: int or float? */
6716 char *lastub = 0; /* position of last underbar */
6717 static char number_too_long[] = "Number too long";
6719 /* We use the first character to decide what type of number this is */
6723 Perl_croak(aTHX_ "panic: scan_num");
6725 /* if it starts with a 0, it could be an octal number, a decimal in
6726 0.13 disguise, or a hexadecimal number, or a binary number. */
6730 u holds the "number so far"
6731 shift the power of 2 of the base
6732 (hex == 4, octal == 3, binary == 1)
6733 overflowed was the number more than we can hold?
6735 Shift is used when we add a digit. It also serves as an "are
6736 we in octal/hex/binary?" indicator to disallow hex characters
6743 bool overflowed = FALSE;
6744 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6745 static char* bases[5] = { "", "binary", "", "octal",
6747 static char* Bases[5] = { "", "Binary", "", "Octal",
6749 static char *maxima[5] = { "",
6750 "0b11111111111111111111111111111111",
6754 char *base, *Base, *max;
6760 } else if (s[1] == 'b') {
6764 /* check for a decimal in disguise */
6765 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6767 /* so it must be octal */
6771 base = bases[shift];
6772 Base = Bases[shift];
6773 max = maxima[shift];
6775 /* read the rest of the number */
6777 /* x is used in the overflow test,
6778 b is the digit we're adding on. */
6783 /* if we don't mention it, we're done */
6792 /* 8 and 9 are not octal */
6795 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6799 case '2': case '3': case '4':
6800 case '5': case '6': case '7':
6802 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6806 b = *s++ & 15; /* ASCII digit -> value of digit */
6810 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6811 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6812 /* make sure they said 0x */
6817 /* Prepare to put the digit we have onto the end
6818 of the number so far. We check for overflows.
6823 x = u << shift; /* make room for the digit */
6825 if ((x >> shift) != u
6826 && !(PL_hints & HINT_NEW_BINARY)) {
6830 if (ckWARN_d(WARN_OVERFLOW))
6831 Perl_warner(aTHX_ WARN_OVERFLOW,
6832 "Integer overflow in %s number",
6835 u = x | b; /* add the digit to the end */
6838 n *= nvshift[shift];
6839 /* If an NV has not enough bits in its
6840 * mantissa to represent an UV this summing of
6841 * small low-order numbers is a waste of time
6842 * (because the NV cannot preserve the
6843 * low-order bits anyway): we could just
6844 * remember when did we overflow and in the
6845 * end just multiply n by the right
6853 /* if we get here, we had success: make a scalar value from
6860 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6861 Perl_warner(aTHX_ WARN_PORTABLE,
6862 "%s number > %s non-portable",
6869 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6870 Perl_warner(aTHX_ WARN_PORTABLE,
6871 "%s number > %s non-portable",
6876 if (PL_hints & HINT_NEW_BINARY)
6877 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6882 handle decimal numbers.
6883 we're also sent here when we read a 0 as the first digit
6885 case '1': case '2': case '3': case '4': case '5':
6886 case '6': case '7': case '8': case '9': case '.':
6889 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
6892 /* read next group of digits and _ and copy into d */
6893 while (isDIGIT(*s) || *s == '_') {
6894 /* skip underscores, checking for misplaced ones
6898 dTHR; /* only for ckWARN */
6899 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
6900 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6904 /* check for end of fixed-length buffer */
6906 Perl_croak(aTHX_ number_too_long);
6907 /* if we're ok, copy the character */
6912 /* final misplaced underbar check */
6913 if (lastub && s - lastub != 3) {
6915 if (ckWARN(WARN_SYNTAX))
6916 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6919 /* read a decimal portion if there is one. avoid
6920 3..5 being interpreted as the number 3. followed
6923 if (*s == '.' && s[1] != '.') {
6927 /* copy, ignoring underbars, until we run out of
6928 digits. Note: no misplaced underbar checks!
6930 for (; isDIGIT(*s) || *s == '_'; s++) {
6931 /* fixed length buffer check */
6933 Perl_croak(aTHX_ number_too_long);
6937 if (*s == '.' && isDIGIT(s[1])) {
6938 /* oops, it's really a v-string, but without the "v" */
6944 /* read exponent part, if present */
6945 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
6949 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
6950 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
6952 /* allow positive or negative exponent */
6953 if (*s == '+' || *s == '-')
6956 /* read digits of exponent (no underbars :-) */
6957 while (isDIGIT(*s)) {
6959 Perl_croak(aTHX_ number_too_long);
6964 /* terminate the string */
6967 /* make an sv from the string */
6970 #if defined(Strtol) && defined(Strtoul)
6973 strtol/strtoll sets errno to ERANGE if the number is too big
6974 for an integer. We try to do an integer conversion first
6975 if no characters indicating "float" have been found.
6982 if (*PL_tokenbuf == '-')
6983 iv = Strtol(PL_tokenbuf, (char**)NULL, 10);
6985 uv = Strtoul(PL_tokenbuf, (char**)NULL, 10);
6987 floatit = TRUE; /* Probably just too large. */
6988 else if (*PL_tokenbuf == '-')
6990 else if (uv <= IV_MAX)
6991 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
6996 nv = Atof(PL_tokenbuf);
7001 No working strtou?ll?.
7003 Unfortunately atol() doesn't do range checks (returning
7004 LONG_MIN/LONG_MAX, and setting errno to ERANGE on overflows)
7005 everywhere [1], so we cannot use use atol() (or atoll()).
7006 If we could, they would be used, as Atol(), very much like
7007 Strtol() and Strtoul() are used above.
7009 [1] XXX Configure test needed to check for atol()
7010 (and atoll() overflow behaviour) XXX --jhi
7012 We need to do this the hard way. */
7014 nv = Atof(PL_tokenbuf);
7016 /* See if we can make do with an integer value without loss of
7017 precision. We use U_V to cast to a UV, because some
7018 compilers have issues. Then we try casting it back and see
7019 if it was the same [1]. We only do this if we know we
7020 specifically read an integer. If floatit is true, then we
7021 don't need to do the conversion at all.
7023 [1] Note that this is lossy if our NVs cannot preserve our
7024 UVs. There is a metaconfig define, NV_PRESERVES_UV, but we
7025 really do hope all such platforms have strtou?ll? to do a
7026 lossless IV/UV conversion.
7027 XXX Configure test needed to check how many UV bits
7028 do our NVs preserve, really (the current test checks
7029 for the roundtrip of ~0) XXX --jhi
7030 Maybe do some tricks with DBL_MANT_DIG and LDBL_MANT_DIG,
7031 and DBL_DIG, LDBL_DIG (this is already available as NV_DIG)?
7035 if (!floatit && (NV)uv == nv) {
7037 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7045 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7046 (PL_hints & HINT_NEW_INTEGER) )
7047 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7048 (floatit ? "float" : "integer"),
7052 /* if it starts with a v, it could be a v-string */
7058 while (isDIGIT(*pos) || *pos == '_')
7060 if (!isALPHA(*pos)) {
7062 U8 tmpbuf[UTF8_MAXLEN];
7065 s++; /* get past 'v' */
7068 sv_setpvn(sv, "", 0);
7071 if (*s == '0' && isDIGIT(s[1]))
7072 yyerror("Octal number in vector unsupported");
7075 /* this is atoi() that tolerates underscores */
7078 while (--end >= s) {
7083 rev += (*end - '0') * mult;
7085 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
7086 Perl_warner(aTHX_ WARN_OVERFLOW,
7087 "Integer overflow in decimal number");
7090 tmpend = uv_to_utf8(tmpbuf, rev);
7091 utf8 = utf8 || rev > 127;
7092 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
7093 if (*pos == '.' && isDIGIT(pos[1]))
7099 while (isDIGIT(*pos) || *pos == '_')
7107 sv_utf8_downgrade(sv, TRUE);
7114 /* make the op for the constant and return */
7117 yylval.opval = newSVOP(OP_CONST, 0, sv);
7119 yylval.opval = Nullop;
7125 S_scan_formline(pTHX_ register char *s)
7130 SV *stuff = newSVpvn("",0);
7131 bool needargs = FALSE;
7134 if (*s == '.' || *s == '}') {
7136 #ifdef PERL_STRICT_CR
7137 for (t = s+1;*t == ' ' || *t == '\t'; t++) ;
7139 for (t = s+1;*t == ' ' || *t == '\t' || *t == '\r'; t++) ;
7141 if (*t == '\n' || t == PL_bufend)
7144 if (PL_in_eval && !PL_rsfp) {
7145 eol = strchr(s,'\n');
7150 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7152 for (t = s; t < eol; t++) {
7153 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7155 goto enough; /* ~~ must be first line in formline */
7157 if (*t == '@' || *t == '^')
7160 sv_catpvn(stuff, s, eol-s);
7161 #ifndef PERL_STRICT_CR
7162 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7163 char *end = SvPVX(stuff) + SvCUR(stuff);
7172 s = filter_gets(PL_linestr, PL_rsfp, 0);
7173 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7174 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7177 yyerror("Format not terminated");
7187 PL_lex_state = LEX_NORMAL;
7188 PL_nextval[PL_nexttoke].ival = 0;
7192 PL_lex_state = LEX_FORMLINE;
7193 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7195 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7199 SvREFCNT_dec(stuff);
7200 PL_lex_formbrack = 0;
7211 PL_cshlen = strlen(PL_cshname);
7216 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7219 I32 oldsavestack_ix = PL_savestack_ix;
7220 CV* outsidecv = PL_compcv;
7224 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7226 SAVEI32(PL_subline);
7227 save_item(PL_subname);
7230 SAVESPTR(PL_comppad_name);
7231 SAVESPTR(PL_compcv);
7232 SAVEI32(PL_comppad_name_fill);
7233 SAVEI32(PL_min_intro_pending);
7234 SAVEI32(PL_max_intro_pending);
7235 SAVEI32(PL_pad_reset_pending);
7237 PL_compcv = (CV*)NEWSV(1104,0);
7238 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7239 CvFLAGS(PL_compcv) |= flags;
7241 PL_comppad = newAV();
7242 av_push(PL_comppad, Nullsv);
7243 PL_curpad = AvARRAY(PL_comppad);
7244 PL_comppad_name = newAV();
7245 PL_comppad_name_fill = 0;
7246 PL_min_intro_pending = 0;
7248 PL_subline = CopLINE(PL_curcop);
7250 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7251 PL_curpad[0] = (SV*)newAV();
7252 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7253 #endif /* USE_THREADS */
7255 comppadlist = newAV();
7256 AvREAL_off(comppadlist);
7257 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7258 av_store(comppadlist, 1, (SV*)PL_comppad);
7260 CvPADLIST(PL_compcv) = comppadlist;
7261 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7263 CvOWNER(PL_compcv) = 0;
7264 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7265 MUTEX_INIT(CvMUTEXP(PL_compcv));
7266 #endif /* USE_THREADS */
7268 return oldsavestack_ix;
7272 Perl_yywarn(pTHX_ char *s)
7275 PL_in_eval |= EVAL_WARNONLY;
7277 PL_in_eval &= ~EVAL_WARNONLY;
7282 Perl_yyerror(pTHX_ char *s)
7286 char *context = NULL;
7290 if (!yychar || (yychar == ';' && !PL_rsfp))
7292 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7293 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7294 while (isSPACE(*PL_oldoldbufptr))
7296 context = PL_oldoldbufptr;
7297 contlen = PL_bufptr - PL_oldoldbufptr;
7299 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7300 PL_oldbufptr != PL_bufptr) {
7301 while (isSPACE(*PL_oldbufptr))
7303 context = PL_oldbufptr;
7304 contlen = PL_bufptr - PL_oldbufptr;
7306 else if (yychar > 255)
7307 where = "next token ???";
7308 #ifdef USE_PURE_BISON
7309 /* GNU Bison sets the value -2 */
7310 else if (yychar == -2) {
7312 else if ((yychar & 127) == 127) {
7314 if (PL_lex_state == LEX_NORMAL ||
7315 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7316 where = "at end of line";
7317 else if (PL_lex_inpat)
7318 where = "within pattern";
7320 where = "within string";
7323 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7325 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7326 else if (isPRINT_LC(yychar))
7327 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7329 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7330 where = SvPVX(where_sv);
7332 msg = sv_2mortal(newSVpv(s, 0));
7333 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7334 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7336 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7338 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7339 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7340 Perl_sv_catpvf(aTHX_ msg,
7341 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7342 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7345 if (PL_in_eval & EVAL_WARNONLY)
7346 Perl_warn(aTHX_ "%"SVf, msg);
7349 if (PL_error_count >= 10) {
7350 if (PL_in_eval && SvCUR(ERRSV))
7351 Perl_croak(aTHX_ "%_%s has too many errors.\n",
7352 ERRSV, CopFILE(PL_curcop));
7354 Perl_croak(aTHX_ "%s has too many errors.\n",
7355 CopFILE(PL_curcop));
7358 PL_in_my_stash = Nullhv;
7369 * Restore a source filter.
7373 restore_rsfp(pTHXo_ void *f)
7375 PerlIO *fp = (PerlIO*)f;
7377 if (PL_rsfp == PerlIO_stdin())
7378 PerlIO_clearerr(PL_rsfp);
7379 else if (PL_rsfp && (PL_rsfp != fp))
7380 PerlIO_close(PL_rsfp);