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);
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);
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 uv = (UV)scan_oct(s, 3, &len);
1394 goto NUM_ESCAPE_INSERT;
1396 /* \x24 indicates a hex constant */
1400 char* e = strchr(s, '}');
1402 yyerror("Missing right brace on \\x{}");
1405 uv = (UV)scan_hex(s + 1, e - s - 1, &len);
1409 uv = (UV)scan_hex(s, 2, &len);
1414 /* Insert oct or hex escaped character.
1415 * There will always enough room in sv since such escapes will
1416 * be longer than any utf8 sequence they can end up as
1419 if (!thisutf && !has_utf && uv > 255) {
1420 /* might need to recode whatever we have accumulated so far
1421 * if it contains any hibit chars
1425 for (c = SvPVX(sv); c < d; c++) {
1430 char *old_pvx = SvPVX(sv);
1432 d = SvGROW(sv, SvCUR(sv) + hicount + 1) + (d - old_pvx);
1441 uv_to_utf8((U8*)dst, (U8)*src--);
1451 if (thisutf || uv > 255) {
1452 d = (char*)uv_to_utf8((U8*)d, uv);
1464 /* \N{latin small letter a} is a named character */
1468 char* e = strchr(s, '}');
1474 yyerror("Missing right brace on \\N{}");
1478 res = newSVpvn(s + 1, e - s - 1);
1479 res = new_constant( Nullch, 0, "charnames",
1480 res, Nullsv, "\\N{...}" );
1481 str = SvPV(res,len);
1482 if (!has_utf && SvUTF8(res)) {
1483 char *ostart = SvPVX(sv);
1484 SvCUR_set(sv, d - ostart);
1486 sv_utf8_upgrade(sv);
1487 d = SvPVX(sv) + SvCUR(sv);
1490 if (len > e - s + 4) {
1491 char *odest = SvPVX(sv);
1493 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1494 d = SvPVX(sv) + (d - odest);
1496 Copy(str, d, len, char);
1503 yyerror("Missing braces on \\N{}");
1506 /* \c is a control character */
1521 /* printf-style backslashes, formfeeds, newlines, etc */
1539 *d++ = '\047'; /* CP 1047 */
1542 *d++ = '\057'; /* CP 1047 */
1556 } /* end if (backslash) */
1559 } /* while loop to process each character */
1561 /* terminate the string and set up the sv */
1563 SvCUR_set(sv, d - SvPVX(sv));
1568 /* shrink the sv if we allocated more than we used */
1569 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1570 SvLEN_set(sv, SvCUR(sv) + 1);
1571 Renew(SvPVX(sv), SvLEN(sv), char);
1574 /* return the substring (via yylval) only if we parsed anything */
1575 if (s > PL_bufptr) {
1576 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1577 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1579 ( PL_lex_inwhat == OP_TRANS
1581 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1584 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1591 * Returns TRUE if there's more to the expression (e.g., a subscript),
1594 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1596 * ->[ and ->{ return TRUE
1597 * { and [ outside a pattern are always subscripts, so return TRUE
1598 * if we're outside a pattern and it's not { or [, then return FALSE
1599 * if we're in a pattern and the first char is a {
1600 * {4,5} (any digits around the comma) returns FALSE
1601 * if we're in a pattern and the first char is a [
1603 * [SOMETHING] has a funky algorithm to decide whether it's a
1604 * character class or not. It has to deal with things like
1605 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1606 * anything else returns TRUE
1609 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1612 S_intuit_more(pTHX_ register char *s)
1614 if (PL_lex_brackets)
1616 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1618 if (*s != '{' && *s != '[')
1623 /* In a pattern, so maybe we have {n,m}. */
1640 /* On the other hand, maybe we have a character class */
1643 if (*s == ']' || *s == '^')
1646 /* this is terrifying, and it works */
1647 int weight = 2; /* let's weigh the evidence */
1649 unsigned char un_char = 255, last_un_char;
1650 char *send = strchr(s,']');
1651 char tmpbuf[sizeof PL_tokenbuf * 4];
1653 if (!send) /* has to be an expression */
1656 Zero(seen,256,char);
1659 else if (isDIGIT(*s)) {
1661 if (isDIGIT(s[1]) && s[2] == ']')
1667 for (; s < send; s++) {
1668 last_un_char = un_char;
1669 un_char = (unsigned char)*s;
1674 weight -= seen[un_char] * 10;
1675 if (isALNUM_lazy_if(s+1,UTF)) {
1676 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1677 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1682 else if (*s == '$' && s[1] &&
1683 strchr("[#!%*<>()-=",s[1])) {
1684 if (/*{*/ strchr("])} =",s[2]))
1693 if (strchr("wds]",s[1]))
1695 else if (seen['\''] || seen['"'])
1697 else if (strchr("rnftbxcav",s[1]))
1699 else if (isDIGIT(s[1])) {
1701 while (s[1] && isDIGIT(s[1]))
1711 if (strchr("aA01! ",last_un_char))
1713 if (strchr("zZ79~",s[1]))
1715 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1716 weight -= 5; /* cope with negative subscript */
1719 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1720 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1725 if (keyword(tmpbuf, d - tmpbuf))
1728 if (un_char == last_un_char + 1)
1730 weight -= seen[un_char];
1735 if (weight >= 0) /* probably a character class */
1745 * Does all the checking to disambiguate
1747 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1748 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1750 * First argument is the stuff after the first token, e.g. "bar".
1752 * Not a method if bar is a filehandle.
1753 * Not a method if foo is a subroutine prototyped to take a filehandle.
1754 * Not a method if it's really "Foo $bar"
1755 * Method if it's "foo $bar"
1756 * Not a method if it's really "print foo $bar"
1757 * Method if it's really "foo package::" (interpreted as package->foo)
1758 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1759 * Not a method if bar is a filehandle or package, but is quoted with
1764 S_intuit_method(pTHX_ char *start, GV *gv)
1766 char *s = start + (*start == '$');
1767 char tmpbuf[sizeof PL_tokenbuf];
1775 if ((cv = GvCVu(gv))) {
1776 char *proto = SvPVX(cv);
1786 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1787 /* start is the beginning of the possible filehandle/object,
1788 * and s is the end of it
1789 * tmpbuf is a copy of it
1792 if (*start == '$') {
1793 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1798 return *s == '(' ? FUNCMETH : METHOD;
1800 if (!keyword(tmpbuf, len)) {
1801 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1806 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1807 if (indirgv && GvCVu(indirgv))
1809 /* filehandle or package name makes it a method */
1810 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1812 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1813 return 0; /* no assumptions -- "=>" quotes bearword */
1815 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1816 newSVpvn(tmpbuf,len));
1817 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1821 return *s == '(' ? FUNCMETH : METHOD;
1829 * Return a string of Perl code to load the debugger. If PERL5DB
1830 * is set, it will return the contents of that, otherwise a
1831 * compile-time require of perl5db.pl.
1838 char *pdb = PerlEnv_getenv("PERL5DB");
1842 SETERRNO(0,SS$_NORMAL);
1843 return "BEGIN { require 'perl5db.pl' }";
1849 /* Encoded script support. filter_add() effectively inserts a
1850 * 'pre-processing' function into the current source input stream.
1851 * Note that the filter function only applies to the current source file
1852 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1854 * The datasv parameter (which may be NULL) can be used to pass
1855 * private data to this instance of the filter. The filter function
1856 * can recover the SV using the FILTER_DATA macro and use it to
1857 * store private buffers and state information.
1859 * The supplied datasv parameter is upgraded to a PVIO type
1860 * and the IoDIRP field is used to store the function pointer,
1861 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1862 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1863 * private use must be set using malloc'd pointers.
1867 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1872 if (!PL_rsfp_filters)
1873 PL_rsfp_filters = newAV();
1875 datasv = NEWSV(255,0);
1876 if (!SvUPGRADE(datasv, SVt_PVIO))
1877 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1878 IoDIRP(datasv) = (DIR*)funcp; /* stash funcp into spare field */
1879 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1880 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1881 funcp, SvPV_nolen(datasv)));
1882 av_unshift(PL_rsfp_filters, 1);
1883 av_store(PL_rsfp_filters, 0, datasv) ;
1888 /* Delete most recently added instance of this filter function. */
1890 Perl_filter_del(pTHX_ filter_t funcp)
1893 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1894 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1896 /* if filter is on top of stack (usual case) just pop it off */
1897 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1898 if (IoDIRP(datasv) == (DIR*)funcp) {
1899 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1900 IoDIRP(datasv) = (DIR*)NULL;
1901 sv_free(av_pop(PL_rsfp_filters));
1905 /* we need to search for the correct entry and clear it */
1906 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1910 /* Invoke the n'th filter function for the current rsfp. */
1912 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1915 /* 0 = read one text line */
1920 if (!PL_rsfp_filters)
1922 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1923 /* Provide a default input filter to make life easy. */
1924 /* Note that we append to the line. This is handy. */
1925 DEBUG_P(PerlIO_printf(Perl_debug_log,
1926 "filter_read %d: from rsfp\n", idx));
1930 int old_len = SvCUR(buf_sv) ;
1932 /* ensure buf_sv is large enough */
1933 SvGROW(buf_sv, old_len + maxlen) ;
1934 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1935 if (PerlIO_error(PL_rsfp))
1936 return -1; /* error */
1938 return 0 ; /* end of file */
1940 SvCUR_set(buf_sv, old_len + len) ;
1943 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1944 if (PerlIO_error(PL_rsfp))
1945 return -1; /* error */
1947 return 0 ; /* end of file */
1950 return SvCUR(buf_sv);
1952 /* Skip this filter slot if filter has been deleted */
1953 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1954 DEBUG_P(PerlIO_printf(Perl_debug_log,
1955 "filter_read %d: skipped (filter deleted)\n",
1957 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1959 /* Get function pointer hidden within datasv */
1960 funcp = (filter_t)IoDIRP(datasv);
1961 DEBUG_P(PerlIO_printf(Perl_debug_log,
1962 "filter_read %d: via function %p (%s)\n",
1963 idx, funcp, SvPV_nolen(datasv)));
1964 /* Call function. The function is expected to */
1965 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1966 /* Return: <0:error, =0:eof, >0:not eof */
1967 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1971 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1973 #ifdef PERL_CR_FILTER
1974 if (!PL_rsfp_filters) {
1975 filter_add(S_cr_textfilter,NULL);
1978 if (PL_rsfp_filters) {
1981 SvCUR_set(sv, 0); /* start with empty line */
1982 if (FILTER_READ(0, sv, 0) > 0)
1983 return ( SvPVX(sv) ) ;
1988 return (sv_gets(sv, fp, append));
1993 static char* exp_name[] =
1994 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
1995 "ATTRTERM", "TERMBLOCK"
2002 Works out what to call the token just pulled out of the input
2003 stream. The yacc parser takes care of taking the ops we return and
2004 stitching them into a tree.
2010 if read an identifier
2011 if we're in a my declaration
2012 croak if they tried to say my($foo::bar)
2013 build the ops for a my() declaration
2014 if it's an access to a my() variable
2015 are we in a sort block?
2016 croak if my($a); $a <=> $b
2017 build ops for access to a my() variable
2018 if in a dq string, and they've said @foo and we can't find @foo
2020 build ops for a bareword
2021 if we already built the token before, use it.
2025 #ifdef USE_PURE_BISON
2026 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
2039 #ifdef USE_PURE_BISON
2040 yylval_pointer = lvalp;
2041 yychar_pointer = lcharp;
2044 /* check if there's an identifier for us to look at */
2045 if (PL_pending_ident) {
2046 /* pit holds the identifier we read and pending_ident is reset */
2047 char pit = PL_pending_ident;
2048 PL_pending_ident = 0;
2050 /* if we're in a my(), we can't allow dynamics here.
2051 $foo'bar has already been turned into $foo::bar, so
2052 just check for colons.
2054 if it's a legal name, the OP is a PADANY.
2057 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
2058 if (strchr(PL_tokenbuf,':'))
2059 yyerror(Perl_form(aTHX_ "No package name allowed for "
2060 "variable %s in \"our\"",
2062 tmp = pad_allocmy(PL_tokenbuf);
2065 if (strchr(PL_tokenbuf,':'))
2066 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
2068 yylval.opval = newOP(OP_PADANY, 0);
2069 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
2075 build the ops for accesses to a my() variable.
2077 Deny my($a) or my($b) in a sort block, *if* $a or $b is
2078 then used in a comparison. This catches most, but not
2079 all cases. For instance, it catches
2080 sort { my($a); $a <=> $b }
2082 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2083 (although why you'd do that is anyone's guess).
2086 if (!strchr(PL_tokenbuf,':')) {
2088 /* Check for single character per-thread SVs */
2089 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2090 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2091 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2093 yylval.opval = newOP(OP_THREADSV, 0);
2094 yylval.opval->op_targ = tmp;
2097 #endif /* USE_THREADS */
2098 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2099 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2100 /* might be an "our" variable" */
2101 if (SvFLAGS(namesv) & SVpad_OUR) {
2102 /* build ops for a bareword */
2103 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2104 sv_catpvn(sym, "::", 2);
2105 sv_catpv(sym, PL_tokenbuf+1);
2106 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2107 yylval.opval->op_private = OPpCONST_ENTERED;
2108 gv_fetchpv(SvPVX(sym),
2110 ? (GV_ADDMULTI | GV_ADDINEVAL)
2113 ((PL_tokenbuf[0] == '$') ? SVt_PV
2114 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2119 /* if it's a sort block and they're naming $a or $b */
2120 if (PL_last_lop_op == OP_SORT &&
2121 PL_tokenbuf[0] == '$' &&
2122 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2125 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2126 d < PL_bufend && *d != '\n';
2129 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2130 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2136 yylval.opval = newOP(OP_PADANY, 0);
2137 yylval.opval->op_targ = tmp;
2143 Whine if they've said @foo in a doublequoted string,
2144 and @foo isn't a variable we can find in the symbol
2147 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2148 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2149 if (!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2150 yyerror(Perl_form(aTHX_ "In string, %s now must be written as \\%s",
2151 PL_tokenbuf, PL_tokenbuf));
2154 /* build ops for a bareword */
2155 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2156 yylval.opval->op_private = OPpCONST_ENTERED;
2157 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2158 ((PL_tokenbuf[0] == '$') ? SVt_PV
2159 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2164 /* no identifier pending identification */
2166 switch (PL_lex_state) {
2168 case LEX_NORMAL: /* Some compilers will produce faster */
2169 case LEX_INTERPNORMAL: /* code if we comment these out. */
2173 /* when we've already built the next token, just pull it out of the queue */
2176 yylval = PL_nextval[PL_nexttoke];
2178 PL_lex_state = PL_lex_defer;
2179 PL_expect = PL_lex_expect;
2180 PL_lex_defer = LEX_NORMAL;
2182 return(PL_nexttype[PL_nexttoke]);
2184 /* interpolated case modifiers like \L \U, including \Q and \E.
2185 when we get here, PL_bufptr is at the \
2187 case LEX_INTERPCASEMOD:
2189 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2190 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2192 /* handle \E or end of string */
2193 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2197 if (PL_lex_casemods) {
2198 oldmod = PL_lex_casestack[--PL_lex_casemods];
2199 PL_lex_casestack[PL_lex_casemods] = '\0';
2201 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2203 PL_lex_state = LEX_INTERPCONCAT;
2207 if (PL_bufptr != PL_bufend)
2209 PL_lex_state = LEX_INTERPCONCAT;
2214 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2215 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2216 if (strchr("LU", *s) &&
2217 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2219 PL_lex_casestack[--PL_lex_casemods] = '\0';
2222 if (PL_lex_casemods > 10) {
2223 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2224 if (newlb != PL_lex_casestack) {
2226 PL_lex_casestack = newlb;
2229 PL_lex_casestack[PL_lex_casemods++] = *s;
2230 PL_lex_casestack[PL_lex_casemods] = '\0';
2231 PL_lex_state = LEX_INTERPCONCAT;
2232 PL_nextval[PL_nexttoke].ival = 0;
2235 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2237 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2239 PL_nextval[PL_nexttoke].ival = OP_LC;
2241 PL_nextval[PL_nexttoke].ival = OP_UC;
2243 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2245 Perl_croak(aTHX_ "panic: yylex");
2248 if (PL_lex_starts) {
2257 case LEX_INTERPPUSH:
2258 return sublex_push();
2260 case LEX_INTERPSTART:
2261 if (PL_bufptr == PL_bufend)
2262 return sublex_done();
2264 PL_lex_dojoin = (*PL_bufptr == '@');
2265 PL_lex_state = LEX_INTERPNORMAL;
2266 if (PL_lex_dojoin) {
2267 PL_nextval[PL_nexttoke].ival = 0;
2270 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2271 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2272 force_next(PRIVATEREF);
2274 force_ident("\"", '$');
2275 #endif /* USE_THREADS */
2276 PL_nextval[PL_nexttoke].ival = 0;
2278 PL_nextval[PL_nexttoke].ival = 0;
2280 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2283 if (PL_lex_starts++) {
2289 case LEX_INTERPENDMAYBE:
2290 if (intuit_more(PL_bufptr)) {
2291 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2297 if (PL_lex_dojoin) {
2298 PL_lex_dojoin = FALSE;
2299 PL_lex_state = LEX_INTERPCONCAT;
2302 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2303 && SvEVALED(PL_lex_repl))
2305 if (PL_bufptr != PL_bufend)
2306 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2307 PL_lex_repl = Nullsv;
2310 case LEX_INTERPCONCAT:
2312 if (PL_lex_brackets)
2313 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2315 if (PL_bufptr == PL_bufend)
2316 return sublex_done();
2318 if (SvIVX(PL_linestr) == '\'') {
2319 SV *sv = newSVsv(PL_linestr);
2322 else if ( PL_hints & HINT_NEW_RE )
2323 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2324 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2328 s = scan_const(PL_bufptr);
2330 PL_lex_state = LEX_INTERPCASEMOD;
2332 PL_lex_state = LEX_INTERPSTART;
2335 if (s != PL_bufptr) {
2336 PL_nextval[PL_nexttoke] = yylval;
2339 if (PL_lex_starts++)
2349 PL_lex_state = LEX_NORMAL;
2350 s = scan_formline(PL_bufptr);
2351 if (!PL_lex_formbrack)
2357 PL_oldoldbufptr = PL_oldbufptr;
2360 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2361 exp_name[PL_expect], s);
2367 if (isIDFIRST_lazy_if(s,UTF))
2369 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2372 goto fake_eof; /* emulate EOF on ^D or ^Z */
2377 if (PL_lex_brackets)
2378 yyerror("Missing right curly or square bracket");
2381 if (s++ < PL_bufend)
2382 goto retry; /* ignore stray nulls */
2385 if (!PL_in_eval && !PL_preambled) {
2386 PL_preambled = TRUE;
2387 sv_setpv(PL_linestr,incl_perldb());
2388 if (SvCUR(PL_linestr))
2389 sv_catpv(PL_linestr,";");
2391 while(AvFILLp(PL_preambleav) >= 0) {
2392 SV *tmpsv = av_shift(PL_preambleav);
2393 sv_catsv(PL_linestr, tmpsv);
2394 sv_catpv(PL_linestr, ";");
2397 sv_free((SV*)PL_preambleav);
2398 PL_preambleav = NULL;
2400 if (PL_minus_n || PL_minus_p) {
2401 sv_catpv(PL_linestr, "LINE: while (<>) {");
2403 sv_catpv(PL_linestr,"chomp;");
2405 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2407 GvIMPORTED_AV_on(gv);
2409 if (strchr("/'\"", *PL_splitstr)
2410 && strchr(PL_splitstr + 1, *PL_splitstr))
2411 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2414 s = "'~#\200\1'"; /* surely one char is unused...*/
2415 while (s[1] && strchr(PL_splitstr, *s)) s++;
2417 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2418 "q" + (delim == '\''), delim);
2419 for (s = PL_splitstr; *s; s++) {
2421 sv_catpvn(PL_linestr, "\\", 1);
2422 sv_catpvn(PL_linestr, s, 1);
2424 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2428 sv_catpv(PL_linestr,"@F=split(' ');");
2431 sv_catpv(PL_linestr, "\n");
2432 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2433 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2434 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2435 SV *sv = NEWSV(85,0);
2437 sv_upgrade(sv, SVt_PVMG);
2438 sv_setsv(sv,PL_linestr);
2439 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2444 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2447 if (PL_preprocess && !PL_in_eval)
2448 (void)PerlProc_pclose(PL_rsfp);
2449 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2450 PerlIO_clearerr(PL_rsfp);
2452 (void)PerlIO_close(PL_rsfp);
2454 PL_doextract = FALSE;
2456 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2457 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2458 sv_catpv(PL_linestr,";}");
2459 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2460 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2461 PL_minus_n = PL_minus_p = 0;
2464 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2465 sv_setpv(PL_linestr,"");
2466 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2469 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2470 PL_doextract = FALSE;
2472 /* Incest with pod. */
2473 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2474 sv_setpv(PL_linestr, "");
2475 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2476 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2477 PL_doextract = FALSE;
2481 } while (PL_doextract);
2482 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2483 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2484 SV *sv = NEWSV(85,0);
2486 sv_upgrade(sv, SVt_PVMG);
2487 sv_setsv(sv,PL_linestr);
2488 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2490 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2491 if (CopLINE(PL_curcop) == 1) {
2492 while (s < PL_bufend && isSPACE(*s))
2494 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2498 if (*s == '#' && *(s+1) == '!')
2500 #ifdef ALTERNATE_SHEBANG
2502 static char as[] = ALTERNATE_SHEBANG;
2503 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2504 d = s + (sizeof(as) - 1);
2506 #endif /* ALTERNATE_SHEBANG */
2515 while (*d && !isSPACE(*d))
2519 #ifdef ARG_ZERO_IS_SCRIPT
2520 if (ipathend > ipath) {
2522 * HP-UX (at least) sets argv[0] to the script name,
2523 * which makes $^X incorrect. And Digital UNIX and Linux,
2524 * at least, set argv[0] to the basename of the Perl
2525 * interpreter. So, having found "#!", we'll set it right.
2527 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2528 assert(SvPOK(x) || SvGMAGICAL(x));
2529 if (sv_eq(x, CopFILESV(PL_curcop))) {
2530 sv_setpvn(x, ipath, ipathend - ipath);
2533 TAINT_NOT; /* $^X is always tainted, but that's OK */
2535 #endif /* ARG_ZERO_IS_SCRIPT */
2540 d = instr(s,"perl -");
2542 d = instr(s,"perl");
2544 /* avoid getting into infinite loops when shebang
2545 * line contains "Perl" rather than "perl" */
2547 for (d = ipathend-4; d >= ipath; --d) {
2548 if ((*d == 'p' || *d == 'P')
2549 && !ibcmp(d, "perl", 4))
2559 #ifdef ALTERNATE_SHEBANG
2561 * If the ALTERNATE_SHEBANG on this system starts with a
2562 * character that can be part of a Perl expression, then if
2563 * we see it but not "perl", we're probably looking at the
2564 * start of Perl code, not a request to hand off to some
2565 * other interpreter. Similarly, if "perl" is there, but
2566 * not in the first 'word' of the line, we assume the line
2567 * contains the start of the Perl program.
2569 if (d && *s != '#') {
2571 while (*c && !strchr("; \t\r\n\f\v#", *c))
2574 d = Nullch; /* "perl" not in first word; ignore */
2576 *s = '#'; /* Don't try to parse shebang line */
2578 #endif /* ALTERNATE_SHEBANG */
2583 !instr(s,"indir") &&
2584 instr(PL_origargv[0],"perl"))
2590 while (s < PL_bufend && isSPACE(*s))
2592 if (s < PL_bufend) {
2593 Newz(899,newargv,PL_origargc+3,char*);
2595 while (s < PL_bufend && !isSPACE(*s))
2598 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2601 newargv = PL_origargv;
2603 PerlProc_execv(ipath, newargv);
2604 Perl_croak(aTHX_ "Can't exec %s", ipath);
2607 U32 oldpdb = PL_perldb;
2608 bool oldn = PL_minus_n;
2609 bool oldp = PL_minus_p;
2611 while (*d && !isSPACE(*d)) d++;
2612 while (*d == ' ' || *d == '\t') d++;
2616 if (*d == 'M' || *d == 'm') {
2618 while (*d && !isSPACE(*d)) d++;
2619 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2622 d = moreswitches(d);
2624 if ((PERLDB_LINE && !oldpdb) ||
2625 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2626 /* if we have already added "LINE: while (<>) {",
2627 we must not do it again */
2629 sv_setpv(PL_linestr, "");
2630 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2631 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2632 PL_preambled = FALSE;
2634 (void)gv_fetchfile(PL_origfilename);
2641 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2643 PL_lex_state = LEX_FORMLINE;
2648 #ifdef PERL_STRICT_CR
2649 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2651 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2653 case ' ': case '\t': case '\f': case 013:
2658 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2659 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2660 /* handle eval qq[#line 1 "foo"\n ...] */
2661 CopLINE_dec(PL_curcop);
2665 while (s < d && *s != '\n')
2670 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2672 PL_lex_state = LEX_FORMLINE;
2682 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2687 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2690 if (strnEQ(s,"=>",2)) {
2691 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2692 OPERATOR('-'); /* unary minus */
2694 PL_last_uni = PL_oldbufptr;
2695 PL_last_lop_op = OP_FTEREAD; /* good enough */
2697 case 'r': FTST(OP_FTEREAD);
2698 case 'w': FTST(OP_FTEWRITE);
2699 case 'x': FTST(OP_FTEEXEC);
2700 case 'o': FTST(OP_FTEOWNED);
2701 case 'R': FTST(OP_FTRREAD);
2702 case 'W': FTST(OP_FTRWRITE);
2703 case 'X': FTST(OP_FTREXEC);
2704 case 'O': FTST(OP_FTROWNED);
2705 case 'e': FTST(OP_FTIS);
2706 case 'z': FTST(OP_FTZERO);
2707 case 's': FTST(OP_FTSIZE);
2708 case 'f': FTST(OP_FTFILE);
2709 case 'd': FTST(OP_FTDIR);
2710 case 'l': FTST(OP_FTLINK);
2711 case 'p': FTST(OP_FTPIPE);
2712 case 'S': FTST(OP_FTSOCK);
2713 case 'u': FTST(OP_FTSUID);
2714 case 'g': FTST(OP_FTSGID);
2715 case 'k': FTST(OP_FTSVTX);
2716 case 'b': FTST(OP_FTBLK);
2717 case 'c': FTST(OP_FTCHR);
2718 case 't': FTST(OP_FTTTY);
2719 case 'T': FTST(OP_FTTEXT);
2720 case 'B': FTST(OP_FTBINARY);
2721 case 'M': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTMTIME);
2722 case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME);
2723 case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME);
2725 Perl_croak(aTHX_ "Unrecognized file test: -%c", (int)tmp);
2732 if (PL_expect == XOPERATOR)
2737 else if (*s == '>') {
2740 if (isIDFIRST_lazy_if(s,UTF)) {
2741 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2749 if (PL_expect == XOPERATOR)
2752 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2754 OPERATOR('-'); /* unary minus */
2761 if (PL_expect == XOPERATOR)
2766 if (PL_expect == XOPERATOR)
2769 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2775 if (PL_expect != XOPERATOR) {
2776 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2777 PL_expect = XOPERATOR;
2778 force_ident(PL_tokenbuf, '*');
2791 if (PL_expect == XOPERATOR) {
2795 PL_tokenbuf[0] = '%';
2796 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2797 if (!PL_tokenbuf[1]) {
2799 yyerror("Final % should be \\% or %name");
2802 PL_pending_ident = '%';
2821 switch (PL_expect) {
2824 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2826 PL_bufptr = s; /* update in case we back off */
2832 PL_expect = XTERMBLOCK;
2836 while (isIDFIRST_lazy_if(s,UTF)) {
2837 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2838 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2839 if (tmp < 0) tmp = -tmp;
2854 d = scan_str(d,TRUE,TRUE);
2857 SvREFCNT_dec(PL_lex_stuff);
2858 PL_lex_stuff = Nullsv;
2860 /* MUST advance bufptr here to avoid bogus
2861 "at end of line" context messages from yyerror().
2863 PL_bufptr = s + len;
2864 yyerror("Unterminated attribute parameter in attribute list");
2867 return 0; /* EOF indicator */
2871 SV *sv = newSVpvn(s, len);
2872 sv_catsv(sv, PL_lex_stuff);
2873 attrs = append_elem(OP_LIST, attrs,
2874 newSVOP(OP_CONST, 0, sv));
2875 SvREFCNT_dec(PL_lex_stuff);
2876 PL_lex_stuff = Nullsv;
2879 attrs = append_elem(OP_LIST, attrs,
2880 newSVOP(OP_CONST, 0,
2884 if (*s == ':' && s[1] != ':')
2887 break; /* require real whitespace or :'s */
2889 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
2890 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
2891 char q = ((*s == '\'') ? '"' : '\'');
2892 /* If here for an expression, and parsed no attrs, back off. */
2893 if (tmp == '=' && !attrs) {
2897 /* MUST advance bufptr here to avoid bogus "at end of line"
2898 context messages from yyerror().
2902 yyerror("Unterminated attribute list");
2904 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
2912 PL_nextval[PL_nexttoke].opval = attrs;
2920 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
2921 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
2926 if (CopLINE(PL_curcop) < PL_copline)
2927 PL_copline = CopLINE(PL_curcop);
2938 if (PL_lex_brackets <= 0)
2939 yyerror("Unmatched right square bracket");
2942 if (PL_lex_state == LEX_INTERPNORMAL) {
2943 if (PL_lex_brackets == 0) {
2944 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
2945 PL_lex_state = LEX_INTERPEND;
2952 if (PL_lex_brackets > 100) {
2953 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
2954 if (newlb != PL_lex_brackstack) {
2956 PL_lex_brackstack = newlb;
2959 switch (PL_expect) {
2961 if (PL_lex_formbrack) {
2965 if (PL_oldoldbufptr == PL_last_lop)
2966 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2968 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2969 OPERATOR(HASHBRACK);
2971 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2974 PL_tokenbuf[0] = '\0';
2975 if (d < PL_bufend && *d == '-') {
2976 PL_tokenbuf[0] = '-';
2978 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2981 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
2982 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
2984 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2987 char minus = (PL_tokenbuf[0] == '-');
2988 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
2996 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3001 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3006 if (PL_oldoldbufptr == PL_last_lop)
3007 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3009 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3012 OPERATOR(HASHBRACK);
3013 /* This hack serves to disambiguate a pair of curlies
3014 * as being a block or an anon hash. Normally, expectation
3015 * determines that, but in cases where we're not in a
3016 * position to expect anything in particular (like inside
3017 * eval"") we have to resolve the ambiguity. This code
3018 * covers the case where the first term in the curlies is a
3019 * quoted string. Most other cases need to be explicitly
3020 * disambiguated by prepending a `+' before the opening
3021 * curly in order to force resolution as an anon hash.
3023 * XXX should probably propagate the outer expectation
3024 * into eval"" to rely less on this hack, but that could
3025 * potentially break current behavior of eval"".
3029 if (*s == '\'' || *s == '"' || *s == '`') {
3030 /* common case: get past first string, handling escapes */
3031 for (t++; t < PL_bufend && *t != *s;)
3032 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3036 else if (*s == 'q') {
3039 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3043 char open, close, term;
3046 while (t < PL_bufend && isSPACE(*t))
3050 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3054 for (t++; t < PL_bufend; t++) {
3055 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3057 else if (*t == open)
3061 for (t++; t < PL_bufend; t++) {
3062 if (*t == '\\' && t+1 < PL_bufend)
3064 else if (*t == close && --brackets <= 0)
3066 else if (*t == open)
3072 else if (isALNUM_lazy_if(t,UTF)) {
3074 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3077 while (t < PL_bufend && isSPACE(*t))
3079 /* if comma follows first term, call it an anon hash */
3080 /* XXX it could be a comma expression with loop modifiers */
3081 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3082 || (*t == '=' && t[1] == '>')))
3083 OPERATOR(HASHBRACK);
3084 if (PL_expect == XREF)
3087 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3093 yylval.ival = CopLINE(PL_curcop);
3094 if (isSPACE(*s) || *s == '#')
3095 PL_copline = NOLINE; /* invalidate current command line number */
3100 if (PL_lex_brackets <= 0)
3101 yyerror("Unmatched right curly bracket");
3103 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3104 if (PL_lex_brackets < PL_lex_formbrack)
3105 PL_lex_formbrack = 0;
3106 if (PL_lex_state == LEX_INTERPNORMAL) {
3107 if (PL_lex_brackets == 0) {
3108 if (PL_expect & XFAKEBRACK) {
3109 PL_expect &= XENUMMASK;
3110 PL_lex_state = LEX_INTERPEND;
3112 return yylex(); /* ignore fake brackets */
3114 if (*s == '-' && s[1] == '>')
3115 PL_lex_state = LEX_INTERPENDMAYBE;
3116 else if (*s != '[' && *s != '{')
3117 PL_lex_state = LEX_INTERPEND;
3120 if (PL_expect & XFAKEBRACK) {
3121 PL_expect &= XENUMMASK;
3123 return yylex(); /* ignore fake brackets */
3133 if (PL_expect == XOPERATOR) {
3134 if (ckWARN(WARN_SEMICOLON)
3135 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3137 CopLINE_dec(PL_curcop);
3138 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3139 CopLINE_inc(PL_curcop);
3144 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3146 PL_expect = XOPERATOR;
3147 force_ident(PL_tokenbuf, '&');
3151 yylval.ival = (OPpENTERSUB_AMPER<<8);
3170 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3171 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3173 if (PL_expect == XSTATE && isALPHA(tmp) &&
3174 (s == PL_linestart+1 || s[-2] == '\n') )
3176 if (PL_in_eval && !PL_rsfp) {
3181 if (strnEQ(s,"=cut",4)) {
3195 PL_doextract = TRUE;
3198 if (PL_lex_brackets < PL_lex_formbrack) {
3200 #ifdef PERL_STRICT_CR
3201 for (t = s; *t == ' ' || *t == '\t'; t++) ;
3203 for (t = s; *t == ' ' || *t == '\t' || *t == '\r'; t++) ;
3205 if (*t == '\n' || *t == '#') {
3223 if (PL_expect != XOPERATOR) {
3224 if (s[1] != '<' && !strchr(s,'>'))
3227 s = scan_heredoc(s);
3229 s = scan_inputsymbol(s);
3230 TERM(sublex_start());
3235 SHop(OP_LEFT_SHIFT);
3249 SHop(OP_RIGHT_SHIFT);
3258 if (PL_expect == XOPERATOR) {
3259 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3262 return ','; /* grandfather non-comma-format format */
3266 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3267 PL_tokenbuf[0] = '@';
3268 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3269 sizeof PL_tokenbuf - 1, FALSE);
3270 if (PL_expect == XOPERATOR)
3271 no_op("Array length", s);
3272 if (!PL_tokenbuf[1])
3274 PL_expect = XOPERATOR;
3275 PL_pending_ident = '#';
3279 PL_tokenbuf[0] = '$';
3280 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3281 sizeof PL_tokenbuf - 1, FALSE);
3282 if (PL_expect == XOPERATOR)
3284 if (!PL_tokenbuf[1]) {
3286 yyerror("Final $ should be \\$ or $name");
3290 /* This kludge not intended to be bulletproof. */
3291 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3292 yylval.opval = newSVOP(OP_CONST, 0,
3293 newSViv(PL_compiling.cop_arybase));
3294 yylval.opval->op_private = OPpCONST_ARYBASE;
3300 if (PL_lex_state == LEX_NORMAL)
3303 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3306 PL_tokenbuf[0] = '@';
3307 if (ckWARN(WARN_SYNTAX)) {
3309 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3312 PL_bufptr = skipspace(PL_bufptr);
3313 while (t < PL_bufend && *t != ']')
3315 Perl_warner(aTHX_ WARN_SYNTAX,
3316 "Multidimensional syntax %.*s not supported",
3317 (t - PL_bufptr) + 1, PL_bufptr);
3321 else if (*s == '{') {
3322 PL_tokenbuf[0] = '%';
3323 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3324 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3326 char tmpbuf[sizeof PL_tokenbuf];
3328 for (t++; isSPACE(*t); t++) ;
3329 if (isIDFIRST_lazy_if(t,UTF)) {
3330 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3331 for (; isSPACE(*t); t++) ;
3332 if (*t == ';' && get_cv(tmpbuf, FALSE))
3333 Perl_warner(aTHX_ WARN_SYNTAX,
3334 "You need to quote \"%s\"", tmpbuf);
3340 PL_expect = XOPERATOR;
3341 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3342 bool islop = (PL_last_lop == PL_oldoldbufptr);
3343 if (!islop || PL_last_lop_op == OP_GREPSTART)
3344 PL_expect = XOPERATOR;
3345 else if (strchr("$@\"'`q", *s))
3346 PL_expect = XTERM; /* e.g. print $fh "foo" */
3347 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3348 PL_expect = XTERM; /* e.g. print $fh &sub */
3349 else if (isIDFIRST_lazy_if(s,UTF)) {
3350 char tmpbuf[sizeof PL_tokenbuf];
3351 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3352 if ((tmp = keyword(tmpbuf, len))) {
3353 /* binary operators exclude handle interpretations */
3365 PL_expect = XTERM; /* e.g. print $fh length() */
3370 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3371 if (gv && GvCVu(gv))
3372 PL_expect = XTERM; /* e.g. print $fh subr() */
3375 else if (isDIGIT(*s))
3376 PL_expect = XTERM; /* e.g. print $fh 3 */
3377 else if (*s == '.' && isDIGIT(s[1]))
3378 PL_expect = XTERM; /* e.g. print $fh .3 */
3379 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3380 PL_expect = XTERM; /* e.g. print $fh -1 */
3381 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3382 PL_expect = XTERM; /* print $fh <<"EOF" */
3384 PL_pending_ident = '$';
3388 if (PL_expect == XOPERATOR)
3390 PL_tokenbuf[0] = '@';
3391 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3392 if (!PL_tokenbuf[1]) {
3394 yyerror("Final @ should be \\@ or @name");
3397 if (PL_lex_state == LEX_NORMAL)
3399 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3401 PL_tokenbuf[0] = '%';
3403 /* Warn about @ where they meant $. */
3404 if (ckWARN(WARN_SYNTAX)) {
3405 if (*s == '[' || *s == '{') {
3407 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3409 if (*t == '}' || *t == ']') {
3411 PL_bufptr = skipspace(PL_bufptr);
3412 Perl_warner(aTHX_ WARN_SYNTAX,
3413 "Scalar value %.*s better written as $%.*s",
3414 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3419 PL_pending_ident = '@';
3422 case '/': /* may either be division or pattern */
3423 case '?': /* may either be conditional or pattern */
3424 if (PL_expect != XOPERATOR) {
3425 /* Disable warning on "study /blah/" */
3426 if (PL_oldoldbufptr == PL_last_uni
3427 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3428 || memNE(PL_last_uni, "study", 5)
3429 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3431 s = scan_pat(s,OP_MATCH);
3432 TERM(sublex_start());
3440 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3441 #ifdef PERL_STRICT_CR
3444 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3446 && (s == PL_linestart || s[-1] == '\n') )
3448 PL_lex_formbrack = 0;
3452 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3458 yylval.ival = OPf_SPECIAL;
3464 if (PL_expect != XOPERATOR)
3469 case '0': case '1': case '2': case '3': case '4':
3470 case '5': case '6': case '7': case '8': case '9':
3472 if (PL_expect == XOPERATOR)
3477 s = scan_str(s,FALSE,FALSE);
3478 if (PL_expect == XOPERATOR) {
3479 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3482 return ','; /* grandfather non-comma-format format */
3488 missingterm((char*)0);
3489 yylval.ival = OP_CONST;
3490 TERM(sublex_start());
3493 s = scan_str(s,FALSE,FALSE);
3494 if (PL_expect == XOPERATOR) {
3495 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3498 return ','; /* grandfather non-comma-format format */
3504 missingterm((char*)0);
3505 yylval.ival = OP_CONST;
3506 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3507 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3508 yylval.ival = OP_STRINGIFY;
3512 TERM(sublex_start());
3515 s = scan_str(s,FALSE,FALSE);
3516 if (PL_expect == XOPERATOR)
3517 no_op("Backticks",s);
3519 missingterm((char*)0);
3520 yylval.ival = OP_BACKTICK;
3522 TERM(sublex_start());
3526 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3527 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3529 if (PL_expect == XOPERATOR)
3530 no_op("Backslash",s);
3534 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3538 while (isDIGIT(*start) || *start == '_')
3540 if (*start == '.' && isDIGIT(start[1])) {
3544 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3545 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF)) {
3549 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3559 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3598 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3600 /* Some keywords can be followed by any delimiter, including ':' */
3601 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3602 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3603 (PL_tokenbuf[0] == 'q' &&
3604 strchr("qwxr", PL_tokenbuf[1])))));
3606 /* x::* is just a word, unless x is "CORE" */
3607 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3611 while (d < PL_bufend && isSPACE(*d))
3612 d++; /* no comments skipped here, or s### is misparsed */
3614 /* Is this a label? */
3615 if (!tmp && PL_expect == XSTATE
3616 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3618 yylval.pval = savepv(PL_tokenbuf);
3623 /* Check for keywords */
3624 tmp = keyword(PL_tokenbuf, len);
3626 /* Is this a word before a => operator? */
3627 if (*d == '=' && d[1] == '>') {
3629 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3630 yylval.opval->op_private = OPpCONST_BARE;
3634 if (tmp < 0) { /* second-class keyword? */
3635 GV *ogv = Nullgv; /* override (winner) */
3636 GV *hgv = Nullgv; /* hidden (loser) */
3637 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3639 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3642 if (GvIMPORTED_CV(gv))
3644 else if (! CvMETHOD(cv))
3648 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3649 (gv = *gvp) != (GV*)&PL_sv_undef &&
3650 GvCVu(gv) && GvIMPORTED_CV(gv))
3656 tmp = 0; /* overridden by import or by GLOBAL */
3659 && -tmp==KEY_lock /* XXX generalizable kludge */
3661 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3663 tmp = 0; /* any sub overrides "weak" keyword */
3665 else { /* no override */
3669 if (ckWARN(WARN_AMBIGUOUS) && hgv
3670 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3671 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3672 "Ambiguous call resolved as CORE::%s(), %s",
3673 GvENAME(hgv), "qualify as such or use &");
3680 default: /* not a keyword */
3683 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3685 /* Get the rest if it looks like a package qualifier */
3687 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3689 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3692 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3693 *s == '\'' ? "'" : "::");
3697 if (PL_expect == XOPERATOR) {
3698 if (PL_bufptr == PL_linestart) {
3699 CopLINE_dec(PL_curcop);
3700 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3701 CopLINE_inc(PL_curcop);
3704 no_op("Bareword",s);
3707 /* Look for a subroutine with this name in current package,
3708 unless name is "Foo::", in which case Foo is a bearword
3709 (and a package name). */
3712 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3714 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3715 Perl_warner(aTHX_ WARN_BAREWORD,
3716 "Bareword \"%s\" refers to nonexistent package",
3719 PL_tokenbuf[len] = '\0';
3726 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3729 /* if we saw a global override before, get the right name */
3732 sv = newSVpvn("CORE::GLOBAL::",14);
3733 sv_catpv(sv,PL_tokenbuf);
3736 sv = newSVpv(PL_tokenbuf,0);
3738 /* Presume this is going to be a bareword of some sort. */
3741 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3742 yylval.opval->op_private = OPpCONST_BARE;
3744 /* And if "Foo::", then that's what it certainly is. */
3749 /* See if it's the indirect object for a list operator. */
3751 if (PL_oldoldbufptr &&
3752 PL_oldoldbufptr < PL_bufptr &&
3753 (PL_oldoldbufptr == PL_last_lop
3754 || PL_oldoldbufptr == PL_last_uni) &&
3755 /* NO SKIPSPACE BEFORE HERE! */
3756 (PL_expect == XREF ||
3757 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3759 bool immediate_paren = *s == '(';
3761 /* (Now we can afford to cross potential line boundary.) */
3764 /* Two barewords in a row may indicate method call. */
3766 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3769 /* If not a declared subroutine, it's an indirect object. */
3770 /* (But it's an indir obj regardless for sort.) */
3772 if ((PL_last_lop_op == OP_SORT ||
3773 (!immediate_paren && (!gv || !GvCVu(gv)))) &&
3774 (PL_last_lop_op != OP_MAPSTART &&
3775 PL_last_lop_op != OP_GREPSTART))
3777 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3783 PL_expect = XOPERATOR;
3786 /* Is this a word before a => operator? */
3787 if (*s == '=' && s[1] == '>') {
3789 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
3793 /* If followed by a paren, it's certainly a subroutine. */
3796 if (gv && GvCVu(gv)) {
3797 for (d = s + 1; *d == ' ' || *d == '\t'; d++) ;
3798 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3803 PL_nextval[PL_nexttoke].opval = yylval.opval;
3804 PL_expect = XOPERATOR;
3810 /* If followed by var or block, call it a method (unless sub) */
3812 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3813 PL_last_lop = PL_oldbufptr;
3814 PL_last_lop_op = OP_METHOD;
3818 /* If followed by a bareword, see if it looks like indir obj. */
3820 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3823 /* Not a method, so call it a subroutine (if defined) */
3825 if (gv && GvCVu(gv)) {
3827 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3828 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3829 "Ambiguous use of -%s resolved as -&%s()",
3830 PL_tokenbuf, PL_tokenbuf);
3831 /* Check for a constant sub */
3833 if ((sv = cv_const_sv(cv))) {
3835 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3836 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3837 yylval.opval->op_private = 0;
3841 /* Resolve to GV now. */
3842 op_free(yylval.opval);
3843 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3844 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3845 PL_last_lop = PL_oldbufptr;
3846 PL_last_lop_op = OP_ENTERSUB;
3847 /* Is there a prototype? */
3850 char *proto = SvPV((SV*)cv, len);
3853 if (strEQ(proto, "$"))
3855 if (*proto == '&' && *s == '{') {
3856 sv_setpv(PL_subname,"__ANON__");
3860 PL_nextval[PL_nexttoke].opval = yylval.opval;
3866 /* Call it a bare word */
3868 if (PL_hints & HINT_STRICT_SUBS)
3869 yylval.opval->op_private |= OPpCONST_STRICT;
3872 if (ckWARN(WARN_RESERVED)) {
3873 if (lastchar != '-') {
3874 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
3876 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
3883 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
3884 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3885 "Operator or semicolon missing before %c%s",
3886 lastchar, PL_tokenbuf);
3887 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3888 "Ambiguous use of %c resolved as operator %c",
3889 lastchar, lastchar);
3895 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3896 newSVpv(CopFILE(PL_curcop),0));
3900 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3901 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
3904 case KEY___PACKAGE__:
3905 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3907 ? newSVsv(PL_curstname)
3916 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
3917 char *pname = "main";
3918 if (PL_tokenbuf[2] == 'D')
3919 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
3920 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
3923 GvIOp(gv) = newIO();
3924 IoIFP(GvIOp(gv)) = PL_rsfp;
3925 #if defined(HAS_FCNTL) && defined(F_SETFD)
3927 int fd = PerlIO_fileno(PL_rsfp);
3928 fcntl(fd,F_SETFD,fd >= 3);
3931 /* Mark this internal pseudo-handle as clean */
3932 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
3934 IoTYPE(GvIOp(gv)) = '|';
3935 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
3936 IoTYPE(GvIOp(gv)) = '-';
3938 IoTYPE(GvIOp(gv)) = '<';
3939 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
3940 /* if the script was opened in binmode, we need to revert
3941 * it to text mode for compatibility; but only iff it has CRs
3942 * XXX this is a questionable hack at best. */
3943 if (PL_bufend-PL_bufptr > 2
3944 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
3947 if (IoTYPE(GvIOp(gv)) == '<') {
3948 loc = PerlIO_tell(PL_rsfp);
3949 (void)PerlIO_seek(PL_rsfp, 0L, 0);
3951 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
3952 #if defined(__BORLANDC__)
3953 /* XXX see note in do_binmode() */
3954 ((FILE*)PL_rsfp)->flags |= _F_BIN;
3957 PerlIO_seek(PL_rsfp, loc, 0);
3972 if (PL_expect == XSTATE) {
3979 if (*s == ':' && s[1] == ':') {
3982 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3983 if (!(tmp = keyword(PL_tokenbuf, len)))
3984 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
3998 LOP(OP_ACCEPT,XTERM);
4004 LOP(OP_ATAN2,XTERM);
4010 LOP(OP_BINMODE,XTERM);
4013 LOP(OP_BLESS,XTERM);
4022 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4039 if (!PL_cryptseen) {
4040 PL_cryptseen = TRUE;
4044 LOP(OP_CRYPT,XTERM);
4047 if (ckWARN(WARN_CHMOD)) {
4048 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4049 if (*d != '0' && isDIGIT(*d))
4050 Perl_warner(aTHX_ WARN_CHMOD,
4051 "chmod() mode argument is missing initial 0");
4053 LOP(OP_CHMOD,XTERM);
4056 LOP(OP_CHOWN,XTERM);
4059 LOP(OP_CONNECT,XTERM);
4075 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4079 PL_hints |= HINT_BLOCK_SCOPE;
4089 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4090 LOP(OP_DBMOPEN,XTERM);
4096 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4103 yylval.ival = CopLINE(PL_curcop);
4117 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4118 UNIBRACK(OP_ENTEREVAL);
4133 case KEY_endhostent:
4139 case KEY_endservent:
4142 case KEY_endprotoent:
4153 yylval.ival = CopLINE(PL_curcop);
4155 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4157 if ((PL_bufend - p) >= 3 &&
4158 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4160 else if ((PL_bufend - p) >= 4 &&
4161 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4164 if (isIDFIRST_lazy_if(p,UTF)) {
4165 p = scan_ident(p, PL_bufend,
4166 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4170 Perl_croak(aTHX_ "Missing $ on loop variable");
4175 LOP(OP_FORMLINE,XTERM);
4181 LOP(OP_FCNTL,XTERM);
4187 LOP(OP_FLOCK,XTERM);
4196 LOP(OP_GREPSTART, XREF);
4199 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4214 case KEY_getpriority:
4215 LOP(OP_GETPRIORITY,XTERM);
4217 case KEY_getprotobyname:
4220 case KEY_getprotobynumber:
4221 LOP(OP_GPBYNUMBER,XTERM);
4223 case KEY_getprotoent:
4235 case KEY_getpeername:
4236 UNI(OP_GETPEERNAME);
4238 case KEY_gethostbyname:
4241 case KEY_gethostbyaddr:
4242 LOP(OP_GHBYADDR,XTERM);
4244 case KEY_gethostent:
4247 case KEY_getnetbyname:
4250 case KEY_getnetbyaddr:
4251 LOP(OP_GNBYADDR,XTERM);
4256 case KEY_getservbyname:
4257 LOP(OP_GSBYNAME,XTERM);
4259 case KEY_getservbyport:
4260 LOP(OP_GSBYPORT,XTERM);
4262 case KEY_getservent:
4265 case KEY_getsockname:
4266 UNI(OP_GETSOCKNAME);
4268 case KEY_getsockopt:
4269 LOP(OP_GSOCKOPT,XTERM);
4291 yylval.ival = CopLINE(PL_curcop);
4295 LOP(OP_INDEX,XTERM);
4301 LOP(OP_IOCTL,XTERM);
4313 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4345 LOP(OP_LISTEN,XTERM);
4354 s = scan_pat(s,OP_MATCH);
4355 TERM(sublex_start());
4358 LOP(OP_MAPSTART, XREF);
4361 LOP(OP_MKDIR,XTERM);
4364 LOP(OP_MSGCTL,XTERM);
4367 LOP(OP_MSGGET,XTERM);
4370 LOP(OP_MSGRCV,XTERM);
4373 LOP(OP_MSGSND,XTERM);
4379 if (isIDFIRST_lazy_if(s,UTF)) {
4380 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4381 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4383 PL_in_my_stash = gv_stashpv(PL_tokenbuf, FALSE);
4384 if (!PL_in_my_stash) {
4387 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4395 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4402 if (PL_expect != XSTATE)
4403 yyerror("\"no\" not allowed in expression");
4404 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4405 s = force_version(s);
4410 if (*s == '(' || (s = skipspace(s), *s == '('))
4417 if (isIDFIRST_lazy_if(s,UTF)) {
4419 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4421 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4422 Perl_warner(aTHX_ WARN_PRECEDENCE,
4423 "Precedence problem: open %.*s should be open(%.*s)",
4429 yylval.ival = OP_OR;
4439 LOP(OP_OPEN_DIR,XTERM);
4442 checkcomma(s,PL_tokenbuf,"filehandle");
4446 checkcomma(s,PL_tokenbuf,"filehandle");
4465 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4469 LOP(OP_PIPE_OP,XTERM);
4472 s = scan_str(s,FALSE,FALSE);
4474 missingterm((char*)0);
4475 yylval.ival = OP_CONST;
4476 TERM(sublex_start());
4482 s = scan_str(s,FALSE,FALSE);
4484 missingterm((char*)0);
4486 if (SvCUR(PL_lex_stuff)) {
4489 d = SvPV_force(PL_lex_stuff, len);
4491 for (; isSPACE(*d) && len; --len, ++d) ;
4494 if (!warned && ckWARN(WARN_QW)) {
4495 for (; !isSPACE(*d) && len; --len, ++d) {
4497 Perl_warner(aTHX_ WARN_QW,
4498 "Possible attempt to separate words with commas");
4501 else if (*d == '#') {
4502 Perl_warner(aTHX_ WARN_QW,
4503 "Possible attempt to put comments in qw() list");
4509 for (; !isSPACE(*d) && len; --len, ++d) ;
4511 words = append_elem(OP_LIST, words,
4512 newSVOP(OP_CONST, 0, tokeq(newSVpvn(b, d-b))));
4516 PL_nextval[PL_nexttoke].opval = words;
4521 SvREFCNT_dec(PL_lex_stuff);
4522 PL_lex_stuff = Nullsv;
4527 s = scan_str(s,FALSE,FALSE);
4529 missingterm((char*)0);
4530 yylval.ival = OP_STRINGIFY;
4531 if (SvIVX(PL_lex_stuff) == '\'')
4532 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4533 TERM(sublex_start());
4536 s = scan_pat(s,OP_QR);
4537 TERM(sublex_start());
4540 s = scan_str(s,FALSE,FALSE);
4542 missingterm((char*)0);
4543 yylval.ival = OP_BACKTICK;
4545 TERM(sublex_start());
4552 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4553 s = force_version(s);
4556 *PL_tokenbuf = '\0';
4557 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4558 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4559 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4561 yyerror("<> should be quotes");
4569 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4573 LOP(OP_RENAME,XTERM);
4582 LOP(OP_RINDEX,XTERM);
4605 LOP(OP_REVERSE,XTERM);
4616 TERM(sublex_start());
4618 TOKEN(1); /* force error */
4627 LOP(OP_SELECT,XTERM);
4633 LOP(OP_SEMCTL,XTERM);
4636 LOP(OP_SEMGET,XTERM);
4639 LOP(OP_SEMOP,XTERM);
4645 LOP(OP_SETPGRP,XTERM);
4647 case KEY_setpriority:
4648 LOP(OP_SETPRIORITY,XTERM);
4650 case KEY_sethostent:
4656 case KEY_setservent:
4659 case KEY_setprotoent:
4669 LOP(OP_SEEKDIR,XTERM);
4671 case KEY_setsockopt:
4672 LOP(OP_SSOCKOPT,XTERM);
4678 LOP(OP_SHMCTL,XTERM);
4681 LOP(OP_SHMGET,XTERM);
4684 LOP(OP_SHMREAD,XTERM);
4687 LOP(OP_SHMWRITE,XTERM);
4690 LOP(OP_SHUTDOWN,XTERM);
4699 LOP(OP_SOCKET,XTERM);
4701 case KEY_socketpair:
4702 LOP(OP_SOCKPAIR,XTERM);
4705 checkcomma(s,PL_tokenbuf,"subroutine name");
4707 if (*s == ';' || *s == ')') /* probably a close */
4708 Perl_croak(aTHX_ "sort is now a reserved word");
4710 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4714 LOP(OP_SPLIT,XTERM);
4717 LOP(OP_SPRINTF,XTERM);
4720 LOP(OP_SPLICE,XTERM);
4735 LOP(OP_SUBSTR,XTERM);
4741 char tmpbuf[sizeof PL_tokenbuf];
4743 expectation attrful;
4744 bool have_name, have_proto;
4749 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4750 (*s == ':' && s[1] == ':'))
4753 attrful = XATTRBLOCK;
4754 /* remember buffer pos'n for later force_word */
4755 tboffset = s - PL_oldbufptr;
4756 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4757 if (strchr(tmpbuf, ':'))
4758 sv_setpv(PL_subname, tmpbuf);
4760 sv_setsv(PL_subname,PL_curstname);
4761 sv_catpvn(PL_subname,"::",2);
4762 sv_catpvn(PL_subname,tmpbuf,len);
4769 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4770 PL_expect = XTERMBLOCK;
4771 attrful = XATTRTERM;
4772 sv_setpv(PL_subname,"?");
4776 if (key == KEY_format) {
4778 PL_lex_formbrack = PL_lex_brackets + 1;
4780 (void) force_word(PL_oldbufptr + tboffset, WORD,
4785 /* Look for a prototype */
4789 s = scan_str(s,FALSE,FALSE);
4792 SvREFCNT_dec(PL_lex_stuff);
4793 PL_lex_stuff = Nullsv;
4794 Perl_croak(aTHX_ "Prototype not terminated");
4797 d = SvPVX(PL_lex_stuff);
4799 for (p = d; *p; ++p) {
4804 SvCUR(PL_lex_stuff) = tmp;
4812 if (*s == ':' && s[1] != ':')
4813 PL_expect = attrful;
4816 PL_nextval[PL_nexttoke].opval =
4817 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4818 PL_lex_stuff = Nullsv;
4822 sv_setpv(PL_subname,"__ANON__");
4825 (void) force_word(PL_oldbufptr + tboffset, WORD,
4834 LOP(OP_SYSTEM,XREF);
4837 LOP(OP_SYMLINK,XTERM);
4840 LOP(OP_SYSCALL,XTERM);
4843 LOP(OP_SYSOPEN,XTERM);
4846 LOP(OP_SYSSEEK,XTERM);
4849 LOP(OP_SYSREAD,XTERM);
4852 LOP(OP_SYSWRITE,XTERM);
4856 TERM(sublex_start());
4877 LOP(OP_TRUNCATE,XTERM);
4889 yylval.ival = CopLINE(PL_curcop);
4893 yylval.ival = CopLINE(PL_curcop);
4897 LOP(OP_UNLINK,XTERM);
4903 LOP(OP_UNPACK,XTERM);
4906 LOP(OP_UTIME,XTERM);
4909 if (ckWARN(WARN_UMASK)) {
4910 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4911 if (*d != '0' && isDIGIT(*d))
4912 Perl_warner(aTHX_ WARN_UMASK,
4913 "umask: argument is missing initial 0");
4918 LOP(OP_UNSHIFT,XTERM);
4921 if (PL_expect != XSTATE)
4922 yyerror("\"use\" not allowed in expression");
4924 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4925 s = force_version(s);
4926 if (*s == ';' || (s = skipspace(s), *s == ';')) {
4927 PL_nextval[PL_nexttoke].opval = Nullop;
4932 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4933 s = force_version(s);
4945 yylval.ival = CopLINE(PL_curcop);
4949 PL_hints |= HINT_BLOCK_SCOPE;
4956 LOP(OP_WAITPID,XTERM);
4964 static char ctl_l[2];
4966 if (ctl_l[0] == '\0')
4967 ctl_l[0] = toCTRL('L');
4968 gv_fetchpv(ctl_l,TRUE, SVt_PV);
4971 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
4976 if (PL_expect == XOPERATOR)
4982 yylval.ival = OP_XOR;
4987 TERM(sublex_start());
4993 Perl_keyword(pTHX_ register char *d, I32 len)
4998 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
4999 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5000 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5001 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5002 if (strEQ(d,"__END__")) return KEY___END__;
5006 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5011 if (strEQ(d,"and")) return -KEY_and;
5012 if (strEQ(d,"abs")) return -KEY_abs;
5015 if (strEQ(d,"alarm")) return -KEY_alarm;
5016 if (strEQ(d,"atan2")) return -KEY_atan2;
5019 if (strEQ(d,"accept")) return -KEY_accept;
5024 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5027 if (strEQ(d,"bless")) return -KEY_bless;
5028 if (strEQ(d,"bind")) return -KEY_bind;
5029 if (strEQ(d,"binmode")) return -KEY_binmode;
5032 if (strEQ(d,"CORE")) return -KEY_CORE;
5033 if (strEQ(d,"CHECK")) return KEY_CHECK;
5038 if (strEQ(d,"cmp")) return -KEY_cmp;
5039 if (strEQ(d,"chr")) return -KEY_chr;
5040 if (strEQ(d,"cos")) return -KEY_cos;
5043 if (strEQ(d,"chop")) return KEY_chop;
5046 if (strEQ(d,"close")) return -KEY_close;
5047 if (strEQ(d,"chdir")) return -KEY_chdir;
5048 if (strEQ(d,"chomp")) return KEY_chomp;
5049 if (strEQ(d,"chmod")) return -KEY_chmod;
5050 if (strEQ(d,"chown")) return -KEY_chown;
5051 if (strEQ(d,"crypt")) return -KEY_crypt;
5054 if (strEQ(d,"chroot")) return -KEY_chroot;
5055 if (strEQ(d,"caller")) return -KEY_caller;
5058 if (strEQ(d,"connect")) return -KEY_connect;
5061 if (strEQ(d,"closedir")) return -KEY_closedir;
5062 if (strEQ(d,"continue")) return -KEY_continue;
5067 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5072 if (strEQ(d,"do")) return KEY_do;
5075 if (strEQ(d,"die")) return -KEY_die;
5078 if (strEQ(d,"dump")) return -KEY_dump;
5081 if (strEQ(d,"delete")) return KEY_delete;
5084 if (strEQ(d,"defined")) return KEY_defined;
5085 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5088 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5093 if (strEQ(d,"EQ")) { deprecate(d); return -KEY_eq;}
5094 if (strEQ(d,"END")) return KEY_END;
5099 if (strEQ(d,"eq")) return -KEY_eq;
5102 if (strEQ(d,"eof")) return -KEY_eof;
5103 if (strEQ(d,"exp")) return -KEY_exp;
5106 if (strEQ(d,"else")) return KEY_else;
5107 if (strEQ(d,"exit")) return -KEY_exit;
5108 if (strEQ(d,"eval")) return KEY_eval;
5109 if (strEQ(d,"exec")) return -KEY_exec;
5110 if (strEQ(d,"each")) return KEY_each;
5113 if (strEQ(d,"elsif")) return KEY_elsif;
5116 if (strEQ(d,"exists")) return KEY_exists;
5117 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5120 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5121 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5124 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5127 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5128 if (strEQ(d,"endservent")) return -KEY_endservent;
5131 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5138 if (strEQ(d,"for")) return KEY_for;
5141 if (strEQ(d,"fork")) return -KEY_fork;
5144 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5145 if (strEQ(d,"flock")) return -KEY_flock;
5148 if (strEQ(d,"format")) return KEY_format;
5149 if (strEQ(d,"fileno")) return -KEY_fileno;
5152 if (strEQ(d,"foreach")) return KEY_foreach;
5155 if (strEQ(d,"formline")) return -KEY_formline;
5161 if (strEQ(d,"GT")) { deprecate(d); return -KEY_gt;}
5162 if (strEQ(d,"GE")) { deprecate(d); return -KEY_ge;}
5166 if (strnEQ(d,"get",3)) {
5171 if (strEQ(d,"ppid")) return -KEY_getppid;
5172 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5175 if (strEQ(d,"pwent")) return -KEY_getpwent;
5176 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5177 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5180 if (strEQ(d,"peername")) return -KEY_getpeername;
5181 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5182 if (strEQ(d,"priority")) return -KEY_getpriority;
5185 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5188 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5192 else if (*d == 'h') {
5193 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5194 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5195 if (strEQ(d,"hostent")) return -KEY_gethostent;
5197 else if (*d == 'n') {
5198 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5199 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5200 if (strEQ(d,"netent")) return -KEY_getnetent;
5202 else if (*d == 's') {
5203 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5204 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5205 if (strEQ(d,"servent")) return -KEY_getservent;
5206 if (strEQ(d,"sockname")) return -KEY_getsockname;
5207 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5209 else if (*d == 'g') {
5210 if (strEQ(d,"grent")) return -KEY_getgrent;
5211 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5212 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5214 else if (*d == 'l') {
5215 if (strEQ(d,"login")) return -KEY_getlogin;
5217 else if (strEQ(d,"c")) return -KEY_getc;
5222 if (strEQ(d,"gt")) return -KEY_gt;
5223 if (strEQ(d,"ge")) return -KEY_ge;
5226 if (strEQ(d,"grep")) return KEY_grep;
5227 if (strEQ(d,"goto")) return KEY_goto;
5228 if (strEQ(d,"glob")) return KEY_glob;
5231 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5236 if (strEQ(d,"hex")) return -KEY_hex;
5239 if (strEQ(d,"INIT")) return KEY_INIT;
5244 if (strEQ(d,"if")) return KEY_if;
5247 if (strEQ(d,"int")) return -KEY_int;
5250 if (strEQ(d,"index")) return -KEY_index;
5251 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5256 if (strEQ(d,"join")) return -KEY_join;
5260 if (strEQ(d,"keys")) return KEY_keys;
5261 if (strEQ(d,"kill")) return -KEY_kill;
5266 if (strEQ(d,"LT")) { deprecate(d); return -KEY_lt;}
5267 if (strEQ(d,"LE")) { deprecate(d); return -KEY_le;}
5273 if (strEQ(d,"lt")) return -KEY_lt;
5274 if (strEQ(d,"le")) return -KEY_le;
5275 if (strEQ(d,"lc")) return -KEY_lc;
5278 if (strEQ(d,"log")) return -KEY_log;
5281 if (strEQ(d,"last")) return KEY_last;
5282 if (strEQ(d,"link")) return -KEY_link;
5283 if (strEQ(d,"lock")) return -KEY_lock;
5286 if (strEQ(d,"local")) return KEY_local;
5287 if (strEQ(d,"lstat")) return -KEY_lstat;
5290 if (strEQ(d,"length")) return -KEY_length;
5291 if (strEQ(d,"listen")) return -KEY_listen;
5294 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5297 if (strEQ(d,"localtime")) return -KEY_localtime;
5303 case 1: return KEY_m;
5305 if (strEQ(d,"my")) return KEY_my;
5308 if (strEQ(d,"map")) return KEY_map;
5311 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5314 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5315 if (strEQ(d,"msgget")) return -KEY_msgget;
5316 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5317 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5322 if (strEQ(d,"NE")) { deprecate(d); return -KEY_ne;}
5325 if (strEQ(d,"next")) return KEY_next;
5326 if (strEQ(d,"ne")) return -KEY_ne;
5327 if (strEQ(d,"not")) return -KEY_not;
5328 if (strEQ(d,"no")) return KEY_no;
5333 if (strEQ(d,"or")) return -KEY_or;
5336 if (strEQ(d,"ord")) return -KEY_ord;
5337 if (strEQ(d,"oct")) return -KEY_oct;
5338 if (strEQ(d,"our")) return KEY_our;
5341 if (strEQ(d,"open")) return -KEY_open;
5344 if (strEQ(d,"opendir")) return -KEY_opendir;
5351 if (strEQ(d,"pop")) return KEY_pop;
5352 if (strEQ(d,"pos")) return KEY_pos;
5355 if (strEQ(d,"push")) return KEY_push;
5356 if (strEQ(d,"pack")) return -KEY_pack;
5357 if (strEQ(d,"pipe")) return -KEY_pipe;
5360 if (strEQ(d,"print")) return KEY_print;
5363 if (strEQ(d,"printf")) return KEY_printf;
5366 if (strEQ(d,"package")) return KEY_package;
5369 if (strEQ(d,"prototype")) return KEY_prototype;
5374 if (strEQ(d,"q")) return KEY_q;
5375 if (strEQ(d,"qr")) return KEY_qr;
5376 if (strEQ(d,"qq")) return KEY_qq;
5377 if (strEQ(d,"qw")) return KEY_qw;
5378 if (strEQ(d,"qx")) return KEY_qx;
5380 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5385 if (strEQ(d,"ref")) return -KEY_ref;
5388 if (strEQ(d,"read")) return -KEY_read;
5389 if (strEQ(d,"rand")) return -KEY_rand;
5390 if (strEQ(d,"recv")) return -KEY_recv;
5391 if (strEQ(d,"redo")) return KEY_redo;
5394 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5395 if (strEQ(d,"reset")) return -KEY_reset;
5398 if (strEQ(d,"return")) return KEY_return;
5399 if (strEQ(d,"rename")) return -KEY_rename;
5400 if (strEQ(d,"rindex")) return -KEY_rindex;
5403 if (strEQ(d,"require")) return -KEY_require;
5404 if (strEQ(d,"reverse")) return -KEY_reverse;
5405 if (strEQ(d,"readdir")) return -KEY_readdir;
5408 if (strEQ(d,"readlink")) return -KEY_readlink;
5409 if (strEQ(d,"readline")) return -KEY_readline;
5410 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5413 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5419 case 0: return KEY_s;
5421 if (strEQ(d,"scalar")) return KEY_scalar;
5426 if (strEQ(d,"seek")) return -KEY_seek;
5427 if (strEQ(d,"send")) return -KEY_send;
5430 if (strEQ(d,"semop")) return -KEY_semop;
5433 if (strEQ(d,"select")) return -KEY_select;
5434 if (strEQ(d,"semctl")) return -KEY_semctl;
5435 if (strEQ(d,"semget")) return -KEY_semget;
5438 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5439 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5442 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5443 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5446 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5449 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5450 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5451 if (strEQ(d,"setservent")) return -KEY_setservent;
5454 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5455 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5462 if (strEQ(d,"shift")) return KEY_shift;
5465 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5466 if (strEQ(d,"shmget")) return -KEY_shmget;
5469 if (strEQ(d,"shmread")) return -KEY_shmread;
5472 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5473 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5478 if (strEQ(d,"sin")) return -KEY_sin;
5481 if (strEQ(d,"sleep")) return -KEY_sleep;
5484 if (strEQ(d,"sort")) return KEY_sort;
5485 if (strEQ(d,"socket")) return -KEY_socket;
5486 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5489 if (strEQ(d,"split")) return KEY_split;
5490 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5491 if (strEQ(d,"splice")) return KEY_splice;
5494 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5497 if (strEQ(d,"srand")) return -KEY_srand;
5500 if (strEQ(d,"stat")) return -KEY_stat;
5501 if (strEQ(d,"study")) return KEY_study;
5504 if (strEQ(d,"substr")) return -KEY_substr;
5505 if (strEQ(d,"sub")) return KEY_sub;
5510 if (strEQ(d,"system")) return -KEY_system;
5513 if (strEQ(d,"symlink")) return -KEY_symlink;
5514 if (strEQ(d,"syscall")) return -KEY_syscall;
5515 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5516 if (strEQ(d,"sysread")) return -KEY_sysread;
5517 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5520 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5529 if (strEQ(d,"tr")) return KEY_tr;
5532 if (strEQ(d,"tie")) return KEY_tie;
5535 if (strEQ(d,"tell")) return -KEY_tell;
5536 if (strEQ(d,"tied")) return KEY_tied;
5537 if (strEQ(d,"time")) return -KEY_time;
5540 if (strEQ(d,"times")) return -KEY_times;
5543 if (strEQ(d,"telldir")) return -KEY_telldir;
5546 if (strEQ(d,"truncate")) return -KEY_truncate;
5553 if (strEQ(d,"uc")) return -KEY_uc;
5556 if (strEQ(d,"use")) return KEY_use;
5559 if (strEQ(d,"undef")) return KEY_undef;
5560 if (strEQ(d,"until")) return KEY_until;
5561 if (strEQ(d,"untie")) return KEY_untie;
5562 if (strEQ(d,"utime")) return -KEY_utime;
5563 if (strEQ(d,"umask")) return -KEY_umask;
5566 if (strEQ(d,"unless")) return KEY_unless;
5567 if (strEQ(d,"unpack")) return -KEY_unpack;
5568 if (strEQ(d,"unlink")) return -KEY_unlink;
5571 if (strEQ(d,"unshift")) return KEY_unshift;
5572 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5577 if (strEQ(d,"values")) return -KEY_values;
5578 if (strEQ(d,"vec")) return -KEY_vec;
5583 if (strEQ(d,"warn")) return -KEY_warn;
5584 if (strEQ(d,"wait")) return -KEY_wait;
5587 if (strEQ(d,"while")) return KEY_while;
5588 if (strEQ(d,"write")) return -KEY_write;
5591 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5594 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5599 if (len == 1) return -KEY_x;
5600 if (strEQ(d,"xor")) return -KEY_xor;
5603 if (len == 1) return KEY_y;
5612 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5616 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5617 dTHR; /* only for ckWARN */
5618 if (ckWARN(WARN_SYNTAX)) {
5620 for (w = s+2; *w && level; w++) {
5627 for (; *w && isSPACE(*w); w++) ;
5628 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5629 Perl_warner(aTHX_ WARN_SYNTAX,
5630 "%s (...) interpreted as function",name);
5633 while (s < PL_bufend && isSPACE(*s))
5637 while (s < PL_bufend && isSPACE(*s))
5639 if (isIDFIRST_lazy_if(s,UTF)) {
5641 while (isALNUM_lazy_if(s,UTF))
5643 while (s < PL_bufend && isSPACE(*s))
5648 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5652 Perl_croak(aTHX_ "No comma allowed after %s", what);
5657 /* Either returns sv, or mortalizes sv and returns a new SV*.
5658 Best used as sv=new_constant(..., sv, ...).
5659 If s, pv are NULL, calls subroutine with one argument,
5660 and type is used with error messages only. */
5663 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5667 HV *table = GvHV(PL_hintgv); /* ^H */
5671 const char *why1, *why2, *why3;
5673 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
5676 why1 = "%^H is not consistent";
5677 why2 = strEQ(key,"charnames")
5678 ? " (missing \"use charnames ...\"?)"
5682 msg = Perl_newSVpvf(aTHX_ "constant(%s): %s%s%s",
5683 (type ? type: "undef"), why1, why2, why3);
5684 yyerror(SvPVX(msg));
5688 cvp = hv_fetch(table, key, strlen(key), FALSE);
5689 if (!cvp || !SvOK(*cvp)) {
5692 why3 = "} is not defined";
5695 sv_2mortal(sv); /* Parent created it permanently */
5698 pv = sv_2mortal(newSVpvn(s, len));
5700 typesv = sv_2mortal(newSVpv(type, 0));
5702 typesv = &PL_sv_undef;
5704 PUSHSTACKi(PERLSI_OVERLOAD);
5717 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5721 /* Check the eval first */
5722 if (!PL_in_eval && SvTRUE(ERRSV)) {
5724 sv_catpv(ERRSV, "Propagated");
5725 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5727 res = SvREFCNT_inc(sv);
5731 (void)SvREFCNT_inc(res);
5740 why1 = "Call to &{$^H{";
5742 why3 = "}} did not return a defined value";
5751 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5753 register char *d = dest;
5754 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5757 Perl_croak(aTHX_ ident_too_long);
5758 if (isALNUM(*s)) /* UTF handled below */
5760 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
5765 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5769 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5770 char *t = s + UTF8SKIP(s);
5771 while (*t & 0x80 && is_utf8_mark((U8*)t))
5773 if (d + (t - s) > e)
5774 Perl_croak(aTHX_ ident_too_long);
5775 Copy(s, d, t - s, char);
5788 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5798 e = d + destlen - 3; /* two-character token, ending NUL */
5800 while (isDIGIT(*s)) {
5802 Perl_croak(aTHX_ ident_too_long);
5809 Perl_croak(aTHX_ ident_too_long);
5810 if (isALNUM(*s)) /* UTF handled below */
5812 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
5817 else if (*s == ':' && s[1] == ':') {
5821 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5822 char *t = s + UTF8SKIP(s);
5823 while (*t & 0x80 && is_utf8_mark((U8*)t))
5825 if (d + (t - s) > e)
5826 Perl_croak(aTHX_ ident_too_long);
5827 Copy(s, d, t - s, char);
5838 if (PL_lex_state != LEX_NORMAL)
5839 PL_lex_state = LEX_INTERPENDMAYBE;
5842 if (*s == '$' && s[1] &&
5843 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
5856 if (*d == '^' && *s && isCONTROLVAR(*s)) {
5861 if (isSPACE(s[-1])) {
5864 if (ch != ' ' && ch != '\t') {
5870 if (isIDFIRST_lazy_if(d,UTF)) {
5874 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
5876 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
5879 Copy(s, d, e - s, char);
5884 while ((isALNUM(*s) || *s == ':') && d < e)
5887 Perl_croak(aTHX_ ident_too_long);
5890 while (s < send && (*s == ' ' || *s == '\t')) s++;
5891 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
5892 dTHR; /* only for ckWARN */
5893 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
5894 const char *brack = *s == '[' ? "[...]" : "{...}";
5895 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5896 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
5897 funny, dest, brack, funny, dest, brack);
5900 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
5904 /* Handle extended ${^Foo} variables
5905 * 1999-02-27 mjd-perl-patch@plover.com */
5906 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
5910 while (isALNUM(*s) && d < e) {
5914 Perl_croak(aTHX_ ident_too_long);
5919 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
5920 PL_lex_state = LEX_INTERPEND;
5923 if (PL_lex_state == LEX_NORMAL) {
5924 dTHR; /* only for ckWARN */
5925 if (ckWARN(WARN_AMBIGUOUS) &&
5926 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
5928 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5929 "Ambiguous use of %c{%s} resolved to %c%s",
5930 funny, dest, funny, dest);
5935 s = bracket; /* let the parser handle it */
5939 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
5940 PL_lex_state = LEX_INTERPEND;
5945 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
5950 *pmfl |= PMf_GLOBAL;
5952 *pmfl |= PMf_CONTINUE;
5956 *pmfl |= PMf_MULTILINE;
5958 *pmfl |= PMf_SINGLELINE;
5960 *pmfl |= PMf_EXTENDED;
5964 S_scan_pat(pTHX_ char *start, I32 type)
5969 s = scan_str(start,FALSE,FALSE);
5972 SvREFCNT_dec(PL_lex_stuff);
5973 PL_lex_stuff = Nullsv;
5974 Perl_croak(aTHX_ "Search pattern not terminated");
5977 pm = (PMOP*)newPMOP(type, 0);
5978 if (PL_multi_open == '?')
5979 pm->op_pmflags |= PMf_ONCE;
5981 while (*s && strchr("iomsx", *s))
5982 pmflag(&pm->op_pmflags,*s++);
5985 while (*s && strchr("iogcmsx", *s))
5986 pmflag(&pm->op_pmflags,*s++);
5988 pm->op_pmpermflags = pm->op_pmflags;
5990 PL_lex_op = (OP*)pm;
5991 yylval.ival = OP_MATCH;
5996 S_scan_subst(pTHX_ char *start)
6003 yylval.ival = OP_NULL;
6005 s = scan_str(start,FALSE,FALSE);
6009 SvREFCNT_dec(PL_lex_stuff);
6010 PL_lex_stuff = Nullsv;
6011 Perl_croak(aTHX_ "Substitution pattern not terminated");
6014 if (s[-1] == PL_multi_open)
6017 first_start = PL_multi_start;
6018 s = scan_str(s,FALSE,FALSE);
6021 SvREFCNT_dec(PL_lex_stuff);
6022 PL_lex_stuff = Nullsv;
6024 SvREFCNT_dec(PL_lex_repl);
6025 PL_lex_repl = Nullsv;
6026 Perl_croak(aTHX_ "Substitution replacement not terminated");
6028 PL_multi_start = first_start; /* so whole substitution is taken together */
6030 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6036 else if (strchr("iogcmsx", *s))
6037 pmflag(&pm->op_pmflags,*s++);
6044 PL_sublex_info.super_bufptr = s;
6045 PL_sublex_info.super_bufend = PL_bufend;
6047 pm->op_pmflags |= PMf_EVAL;
6048 repl = newSVpvn("",0);
6050 sv_catpv(repl, es ? "eval " : "do ");
6051 sv_catpvn(repl, "{ ", 2);
6052 sv_catsv(repl, PL_lex_repl);
6053 sv_catpvn(repl, " };", 2);
6055 SvREFCNT_dec(PL_lex_repl);
6059 pm->op_pmpermflags = pm->op_pmflags;
6060 PL_lex_op = (OP*)pm;
6061 yylval.ival = OP_SUBST;
6066 S_scan_trans(pTHX_ char *start)
6077 yylval.ival = OP_NULL;
6079 s = scan_str(start,FALSE,FALSE);
6082 SvREFCNT_dec(PL_lex_stuff);
6083 PL_lex_stuff = Nullsv;
6084 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6086 if (s[-1] == PL_multi_open)
6089 s = scan_str(s,FALSE,FALSE);
6092 SvREFCNT_dec(PL_lex_stuff);
6093 PL_lex_stuff = Nullsv;
6095 SvREFCNT_dec(PL_lex_repl);
6096 PL_lex_repl = Nullsv;
6097 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6101 o = newSVOP(OP_TRANS, 0, 0);
6102 utf8 = OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF;
6105 New(803,tbl,256,short);
6106 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6110 complement = del = squash = 0;
6111 while (strchr("cdsCU", *s)) {
6113 complement = OPpTRANS_COMPLEMENT;
6115 del = OPpTRANS_DELETE;
6117 squash = OPpTRANS_SQUASH;
6122 utf8 &= ~OPpTRANS_FROM_UTF;
6124 utf8 |= OPpTRANS_FROM_UTF;
6128 utf8 &= ~OPpTRANS_TO_UTF;
6130 utf8 |= OPpTRANS_TO_UTF;
6133 Perl_croak(aTHX_ "Too many /C and /U options");
6138 o->op_private = del|squash|complement|utf8;
6141 yylval.ival = OP_TRANS;
6146 S_scan_heredoc(pTHX_ register char *s)
6150 I32 op_type = OP_SCALAR;
6157 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6161 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6164 for (peek = s; *peek == ' ' || *peek == '\t'; peek++) ;
6165 if (*peek && strchr("`'\"",*peek)) {
6168 s = delimcpy(d, e, s, PL_bufend, term, &len);
6178 if (!isALNUM_lazy_if(s,UTF))
6179 deprecate("bare << to mean <<\"\"");
6180 for (; isALNUM_lazy_if(s,UTF); s++) {
6185 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6186 Perl_croak(aTHX_ "Delimiter for here document is too long");
6189 len = d - PL_tokenbuf;
6190 #ifndef PERL_STRICT_CR
6191 d = strchr(s, '\r');
6195 while (s < PL_bufend) {
6201 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6210 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6215 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6216 herewas = newSVpvn(s,PL_bufend-s);
6218 s--, herewas = newSVpvn(s,d-s);
6219 s += SvCUR(herewas);
6221 tmpstr = NEWSV(87,79);
6222 sv_upgrade(tmpstr, SVt_PVIV);
6227 else if (term == '`') {
6228 op_type = OP_BACKTICK;
6229 SvIVX(tmpstr) = '\\';
6233 PL_multi_start = CopLINE(PL_curcop);
6234 PL_multi_open = PL_multi_close = '<';
6235 term = *PL_tokenbuf;
6236 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6237 char *bufptr = PL_sublex_info.super_bufptr;
6238 char *bufend = PL_sublex_info.super_bufend;
6239 char *olds = s - SvCUR(herewas);
6240 s = strchr(bufptr, '\n');
6244 while (s < bufend &&
6245 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6247 CopLINE_inc(PL_curcop);
6250 CopLINE_set(PL_curcop, PL_multi_start);
6251 missingterm(PL_tokenbuf);
6253 sv_setpvn(herewas,bufptr,d-bufptr+1);
6254 sv_setpvn(tmpstr,d+1,s-d);
6256 sv_catpvn(herewas,s,bufend-s);
6257 (void)strcpy(bufptr,SvPVX(herewas));
6264 while (s < PL_bufend &&
6265 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6267 CopLINE_inc(PL_curcop);
6269 if (s >= PL_bufend) {
6270 CopLINE_set(PL_curcop, PL_multi_start);
6271 missingterm(PL_tokenbuf);
6273 sv_setpvn(tmpstr,d+1,s-d);
6275 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6277 sv_catpvn(herewas,s,PL_bufend-s);
6278 sv_setsv(PL_linestr,herewas);
6279 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6280 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6283 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6284 while (s >= PL_bufend) { /* multiple line string? */
6286 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6287 CopLINE_set(PL_curcop, PL_multi_start);
6288 missingterm(PL_tokenbuf);
6290 CopLINE_inc(PL_curcop);
6291 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6292 #ifndef PERL_STRICT_CR
6293 if (PL_bufend - PL_linestart >= 2) {
6294 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6295 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6297 PL_bufend[-2] = '\n';
6299 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6301 else if (PL_bufend[-1] == '\r')
6302 PL_bufend[-1] = '\n';
6304 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6305 PL_bufend[-1] = '\n';
6307 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6308 SV *sv = NEWSV(88,0);
6310 sv_upgrade(sv, SVt_PVMG);
6311 sv_setsv(sv,PL_linestr);
6312 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6314 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6317 sv_catsv(PL_linestr,herewas);
6318 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6322 sv_catsv(tmpstr,PL_linestr);
6327 PL_multi_end = CopLINE(PL_curcop);
6328 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6329 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6330 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6332 SvREFCNT_dec(herewas);
6333 PL_lex_stuff = tmpstr;
6334 yylval.ival = op_type;
6339 takes: current position in input buffer
6340 returns: new position in input buffer
6341 side-effects: yylval and lex_op are set.
6346 <FH> read from filehandle
6347 <pkg::FH> read from package qualified filehandle
6348 <pkg'FH> read from package qualified filehandle
6349 <$fh> read from filehandle in $fh
6355 S_scan_inputsymbol(pTHX_ char *start)
6357 register char *s = start; /* current position in buffer */
6363 d = PL_tokenbuf; /* start of temp holding space */
6364 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6365 end = strchr(s, '\n');
6368 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6370 /* die if we didn't have space for the contents of the <>,
6371 or if it didn't end, or if we see a newline
6374 if (len >= sizeof PL_tokenbuf)
6375 Perl_croak(aTHX_ "Excessively long <> operator");
6377 Perl_croak(aTHX_ "Unterminated <> operator");
6382 Remember, only scalar variables are interpreted as filehandles by
6383 this code. Anything more complex (e.g., <$fh{$num}>) will be
6384 treated as a glob() call.
6385 This code makes use of the fact that except for the $ at the front,
6386 a scalar variable and a filehandle look the same.
6388 if (*d == '$' && d[1]) d++;
6390 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6391 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6394 /* If we've tried to read what we allow filehandles to look like, and
6395 there's still text left, then it must be a glob() and not a getline.
6396 Use scan_str to pull out the stuff between the <> and treat it
6397 as nothing more than a string.
6400 if (d - PL_tokenbuf != len) {
6401 yylval.ival = OP_GLOB;
6403 s = scan_str(start,FALSE,FALSE);
6405 Perl_croak(aTHX_ "Glob not terminated");
6409 /* we're in a filehandle read situation */
6412 /* turn <> into <ARGV> */
6414 (void)strcpy(d,"ARGV");
6416 /* if <$fh>, create the ops to turn the variable into a
6422 /* try to find it in the pad for this block, otherwise find
6423 add symbol table ops
6425 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6426 OP *o = newOP(OP_PADSV, 0);
6428 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6431 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6432 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6433 newUNOP(OP_RV2SV, 0,
6434 newGVOP(OP_GV, 0, gv)));
6436 PL_lex_op->op_flags |= OPf_SPECIAL;
6437 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6438 yylval.ival = OP_NULL;
6441 /* If it's none of the above, it must be a literal filehandle
6442 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6444 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6445 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6446 yylval.ival = OP_NULL;
6455 takes: start position in buffer
6456 keep_quoted preserve \ on the embedded delimiter(s)
6457 keep_delims preserve the delimiters around the string
6458 returns: position to continue reading from buffer
6459 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6460 updates the read buffer.
6462 This subroutine pulls a string out of the input. It is called for:
6463 q single quotes q(literal text)
6464 ' single quotes 'literal text'
6465 qq double quotes qq(interpolate $here please)
6466 " double quotes "interpolate $here please"
6467 qx backticks qx(/bin/ls -l)
6468 ` backticks `/bin/ls -l`
6469 qw quote words @EXPORT_OK = qw( func() $spam )
6470 m// regexp match m/this/
6471 s/// regexp substitute s/this/that/
6472 tr/// string transliterate tr/this/that/
6473 y/// string transliterate y/this/that/
6474 ($*@) sub prototypes sub foo ($)
6475 (stuff) sub attr parameters sub foo : attr(stuff)
6476 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6478 In most of these cases (all but <>, patterns and transliterate)
6479 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6480 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6481 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6484 It skips whitespace before the string starts, and treats the first
6485 character as the delimiter. If the delimiter is one of ([{< then
6486 the corresponding "close" character )]}> is used as the closing
6487 delimiter. It allows quoting of delimiters, and if the string has
6488 balanced delimiters ([{<>}]) it allows nesting.
6490 The lexer always reads these strings into lex_stuff, except in the
6491 case of the operators which take *two* arguments (s/// and tr///)
6492 when it checks to see if lex_stuff is full (presumably with the 1st
6493 arg to s or tr) and if so puts the string into lex_repl.
6498 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6501 SV *sv; /* scalar value: string */
6502 char *tmps; /* temp string, used for delimiter matching */
6503 register char *s = start; /* current position in the buffer */
6504 register char term; /* terminating character */
6505 register char *to; /* current position in the sv's data */
6506 I32 brackets = 1; /* bracket nesting level */
6507 bool has_utf = FALSE; /* is there any utf8 content? */
6509 /* skip space before the delimiter */
6513 /* mark where we are, in case we need to report errors */
6516 /* after skipping whitespace, the next character is the terminator */
6518 if ((term & 0x80) && UTF)
6521 /* mark where we are */
6522 PL_multi_start = CopLINE(PL_curcop);
6523 PL_multi_open = term;
6525 /* find corresponding closing delimiter */
6526 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6528 PL_multi_close = term;
6530 /* create a new SV to hold the contents. 87 is leak category, I'm
6531 assuming. 79 is the SV's initial length. What a random number. */
6533 sv_upgrade(sv, SVt_PVIV);
6535 (void)SvPOK_only(sv); /* validate pointer */
6537 /* move past delimiter and try to read a complete string */
6539 sv_catpvn(sv, s, 1);
6542 /* extend sv if need be */
6543 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6544 /* set 'to' to the next character in the sv's string */
6545 to = SvPVX(sv)+SvCUR(sv);
6547 /* if open delimiter is the close delimiter read unbridle */
6548 if (PL_multi_open == PL_multi_close) {
6549 for (; s < PL_bufend; s++,to++) {
6550 /* embedded newlines increment the current line number */
6551 if (*s == '\n' && !PL_rsfp)
6552 CopLINE_inc(PL_curcop);
6553 /* handle quoted delimiters */
6554 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6555 if (!keep_quoted && s[1] == term)
6557 /* any other quotes are simply copied straight through */
6561 /* terminate when run out of buffer (the for() condition), or
6562 have found the terminator */
6563 else if (*s == term)
6565 else if (!has_utf && (*s & 0x80) && UTF)
6571 /* if the terminator isn't the same as the start character (e.g.,
6572 matched brackets), we have to allow more in the quoting, and
6573 be prepared for nested brackets.
6576 /* read until we run out of string, or we find the terminator */
6577 for (; s < PL_bufend; s++,to++) {
6578 /* embedded newlines increment the line count */
6579 if (*s == '\n' && !PL_rsfp)
6580 CopLINE_inc(PL_curcop);
6581 /* backslashes can escape the open or closing characters */
6582 if (*s == '\\' && s+1 < PL_bufend) {
6584 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6589 /* allow nested opens and closes */
6590 else if (*s == PL_multi_close && --brackets <= 0)
6592 else if (*s == PL_multi_open)
6594 else if (!has_utf && (*s & 0x80) && UTF)
6599 /* terminate the copied string and update the sv's end-of-string */
6601 SvCUR_set(sv, to - SvPVX(sv));
6604 * this next chunk reads more into the buffer if we're not done yet
6608 break; /* handle case where we are done yet :-) */
6610 #ifndef PERL_STRICT_CR
6611 if (to - SvPVX(sv) >= 2) {
6612 if ((to[-2] == '\r' && to[-1] == '\n') ||
6613 (to[-2] == '\n' && to[-1] == '\r'))
6617 SvCUR_set(sv, to - SvPVX(sv));
6619 else if (to[-1] == '\r')
6622 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6626 /* if we're out of file, or a read fails, bail and reset the current
6627 line marker so we can report where the unterminated string began
6630 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6632 CopLINE_set(PL_curcop, PL_multi_start);
6635 /* we read a line, so increment our line counter */
6636 CopLINE_inc(PL_curcop);
6638 /* update debugger info */
6639 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6640 SV *sv = NEWSV(88,0);
6642 sv_upgrade(sv, SVt_PVMG);
6643 sv_setsv(sv,PL_linestr);
6644 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6647 /* having changed the buffer, we must update PL_bufend */
6648 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6651 /* at this point, we have successfully read the delimited string */
6654 sv_catpvn(sv, s, 1);
6657 PL_multi_end = CopLINE(PL_curcop);
6660 /* if we allocated too much space, give some back */
6661 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6662 SvLEN_set(sv, SvCUR(sv) + 1);
6663 Renew(SvPVX(sv), SvLEN(sv), char);
6666 /* decide whether this is the first or second quoted string we've read
6679 takes: pointer to position in buffer
6680 returns: pointer to new position in buffer
6681 side-effects: builds ops for the constant in yylval.op
6683 Read a number in any of the formats that Perl accepts:
6685 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6686 [\d_]+(\.[\d_]*)?[Ee](\d+)
6688 Underbars (_) are allowed in decimal numbers. If -w is on,
6689 underbars before a decimal point must be at three digit intervals.
6691 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6694 If it reads a number without a decimal point or an exponent, it will
6695 try converting the number to an integer and see if it can do so
6696 without loss of precision.
6700 Perl_scan_num(pTHX_ char *start)
6702 register char *s = start; /* current position in buffer */
6703 register char *d; /* destination in temp buffer */
6704 register char *e; /* end of temp buffer */
6705 NV value; /* number read, as a double */
6706 SV *sv = Nullsv; /* place to put the converted number */
6707 bool floatit; /* boolean: int or float? */
6708 char *lastub = 0; /* position of last underbar */
6709 static char number_too_long[] = "Number too long";
6711 /* We use the first character to decide what type of number this is */
6715 Perl_croak(aTHX_ "panic: scan_num");
6717 /* if it starts with a 0, it could be an octal number, a decimal in
6718 0.13 disguise, or a hexadecimal number, or a binary number. */
6722 u holds the "number so far"
6723 shift the power of 2 of the base
6724 (hex == 4, octal == 3, binary == 1)
6725 overflowed was the number more than we can hold?
6727 Shift is used when we add a digit. It also serves as an "are
6728 we in octal/hex/binary?" indicator to disallow hex characters
6735 bool overflowed = FALSE;
6736 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6737 static char* bases[5] = { "", "binary", "", "octal",
6739 static char* Bases[5] = { "", "Binary", "", "Octal",
6741 static char *maxima[5] = { "",
6742 "0b11111111111111111111111111111111",
6746 char *base, *Base, *max;
6752 } else if (s[1] == 'b') {
6756 /* check for a decimal in disguise */
6757 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6759 /* so it must be octal */
6763 base = bases[shift];
6764 Base = Bases[shift];
6765 max = maxima[shift];
6767 /* read the rest of the number */
6769 /* x is used in the overflow test,
6770 b is the digit we're adding on. */
6775 /* if we don't mention it, we're done */
6784 /* 8 and 9 are not octal */
6787 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6791 case '2': case '3': case '4':
6792 case '5': case '6': case '7':
6794 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6798 b = *s++ & 15; /* ASCII digit -> value of digit */
6802 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6803 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6804 /* make sure they said 0x */
6809 /* Prepare to put the digit we have onto the end
6810 of the number so far. We check for overflows.
6815 x = u << shift; /* make room for the digit */
6817 if ((x >> shift) != u
6818 && !(PL_hints & HINT_NEW_BINARY)) {
6822 if (ckWARN_d(WARN_OVERFLOW))
6823 Perl_warner(aTHX_ WARN_OVERFLOW,
6824 "Integer overflow in %s number",
6827 u = x | b; /* add the digit to the end */
6830 n *= nvshift[shift];
6831 /* If an NV has not enough bits in its
6832 * mantissa to represent an UV this summing of
6833 * small low-order numbers is a waste of time
6834 * (because the NV cannot preserve the
6835 * low-order bits anyway): we could just
6836 * remember when did we overflow and in the
6837 * end just multiply n by the right
6845 /* if we get here, we had success: make a scalar value from
6852 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6853 Perl_warner(aTHX_ WARN_PORTABLE,
6854 "%s number > %s non-portable",
6861 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6862 Perl_warner(aTHX_ WARN_PORTABLE,
6863 "%s number > %s non-portable",
6868 if (PL_hints & HINT_NEW_BINARY)
6869 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6874 handle decimal numbers.
6875 we're also sent here when we read a 0 as the first digit
6877 case '1': case '2': case '3': case '4': case '5':
6878 case '6': case '7': case '8': case '9': case '.':
6881 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
6884 /* read next group of digits and _ and copy into d */
6885 while (isDIGIT(*s) || *s == '_') {
6886 /* skip underscores, checking for misplaced ones
6890 dTHR; /* only for ckWARN */
6891 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
6892 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6896 /* check for end of fixed-length buffer */
6898 Perl_croak(aTHX_ number_too_long);
6899 /* if we're ok, copy the character */
6904 /* final misplaced underbar check */
6905 if (lastub && s - lastub != 3) {
6907 if (ckWARN(WARN_SYNTAX))
6908 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6911 /* read a decimal portion if there is one. avoid
6912 3..5 being interpreted as the number 3. followed
6915 if (*s == '.' && s[1] != '.') {
6919 /* copy, ignoring underbars, until we run out of
6920 digits. Note: no misplaced underbar checks!
6922 for (; isDIGIT(*s) || *s == '_'; s++) {
6923 /* fixed length buffer check */
6925 Perl_croak(aTHX_ number_too_long);
6929 if (*s == '.' && isDIGIT(s[1])) {
6930 /* oops, it's really a v-string, but without the "v" */
6936 /* read exponent part, if present */
6937 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
6941 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
6942 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
6944 /* allow positive or negative exponent */
6945 if (*s == '+' || *s == '-')
6948 /* read digits of exponent (no underbars :-) */
6949 while (isDIGIT(*s)) {
6951 Perl_croak(aTHX_ number_too_long);
6956 /* terminate the string */
6959 /* make an sv from the string */
6962 /* unfortunately this monster needs to be on one line or
6963 makedepend will be confused. */
6964 #if (defined(USE_64_BIT_INT) && (!defined(HAS_STRTOLL)|| !defined(HAS_STRTOULL))) || (!defined(USE_64_BIT_INT) && (!defined(HAS_STRTOL) || !defined(HAS_STRTOUL)))
6967 No working strto[u]l[l]. Since atoi() doesn't do range checks,
6968 we need to do this the hard way.
6971 value = Atof(PL_tokenbuf);
6974 See if we can make do with an integer value without loss of
6975 precision. We use I_V to cast to an int, because some
6976 compilers have issues. Then we try casting it back and see
6977 if it was the same. We only do this if we know we
6978 specifically read an integer.
6980 Note: if floatit is true, then we don't need to do the
6984 UV tryuv = U_V(value);
6985 if (!floatit && (NV)tryuv == value) {
6986 if (tryuv <= IV_MAX)
6987 sv_setiv(sv, (IV)tryuv);
6989 sv_setuv(sv, tryuv);
6992 sv_setnv(sv, value);
6996 strtol/strtoll sets errno to ERANGE if the number is too big
6997 for an integer. We try to do an integer conversion first
6998 if no characters indicating "float" have been found.
7005 if (*PL_tokenbuf == '-')
7006 iv = Strtol(PL_tokenbuf, (char**)NULL, 10);
7008 uv = Strtoul(PL_tokenbuf, (char**)NULL, 10);
7010 floatit = TRUE; /* probably just too large */
7011 else if (*PL_tokenbuf == '-')
7017 value = Atof(PL_tokenbuf);
7018 sv_setnv(sv, value);
7021 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7022 (PL_hints & HINT_NEW_INTEGER) )
7023 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7024 (floatit ? "float" : "integer"),
7028 /* if it starts with a v, it could be a v-string */
7034 while (isDIGIT(*pos) || *pos == '_')
7036 if (!isALPHA(*pos)) {
7038 U8 tmpbuf[UTF8_MAXLEN];
7041 s++; /* get past 'v' */
7044 sv_setpvn(sv, "", 0);
7047 if (*s == '0' && isDIGIT(s[1]))
7048 yyerror("Octal number in vector unsupported");
7051 /* this is atoi() that tolerates underscores */
7054 while (--end >= s) {
7059 rev += (*end - '0') * mult;
7061 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
7062 Perl_warner(aTHX_ WARN_OVERFLOW,
7063 "Integer overflow in decimal number");
7066 tmpend = uv_to_utf8(tmpbuf, rev);
7067 utf8 = utf8 || rev > 127;
7068 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
7069 if (*pos == '.' && isDIGIT(pos[1]))
7075 while (isDIGIT(*pos) || *pos == '_')
7083 sv_utf8_downgrade(sv, TRUE);
7090 /* make the op for the constant and return */
7093 yylval.opval = newSVOP(OP_CONST, 0, sv);
7095 yylval.opval = Nullop;
7101 S_scan_formline(pTHX_ register char *s)
7106 SV *stuff = newSVpvn("",0);
7107 bool needargs = FALSE;
7110 if (*s == '.' || *s == '}') {
7112 #ifdef PERL_STRICT_CR
7113 for (t = s+1;*t == ' ' || *t == '\t'; t++) ;
7115 for (t = s+1;*t == ' ' || *t == '\t' || *t == '\r'; t++) ;
7117 if (*t == '\n' || t == PL_bufend)
7120 if (PL_in_eval && !PL_rsfp) {
7121 eol = strchr(s,'\n');
7126 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7128 for (t = s; t < eol; t++) {
7129 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7131 goto enough; /* ~~ must be first line in formline */
7133 if (*t == '@' || *t == '^')
7136 sv_catpvn(stuff, s, eol-s);
7137 #ifndef PERL_STRICT_CR
7138 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7139 char *end = SvPVX(stuff) + SvCUR(stuff);
7148 s = filter_gets(PL_linestr, PL_rsfp, 0);
7149 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7150 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7153 yyerror("Format not terminated");
7163 PL_lex_state = LEX_NORMAL;
7164 PL_nextval[PL_nexttoke].ival = 0;
7168 PL_lex_state = LEX_FORMLINE;
7169 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7171 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7175 SvREFCNT_dec(stuff);
7176 PL_lex_formbrack = 0;
7187 PL_cshlen = strlen(PL_cshname);
7192 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7195 I32 oldsavestack_ix = PL_savestack_ix;
7196 CV* outsidecv = PL_compcv;
7200 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7202 SAVEI32(PL_subline);
7203 save_item(PL_subname);
7206 SAVESPTR(PL_comppad_name);
7207 SAVESPTR(PL_compcv);
7208 SAVEI32(PL_comppad_name_fill);
7209 SAVEI32(PL_min_intro_pending);
7210 SAVEI32(PL_max_intro_pending);
7211 SAVEI32(PL_pad_reset_pending);
7213 PL_compcv = (CV*)NEWSV(1104,0);
7214 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7215 CvFLAGS(PL_compcv) |= flags;
7217 PL_comppad = newAV();
7218 av_push(PL_comppad, Nullsv);
7219 PL_curpad = AvARRAY(PL_comppad);
7220 PL_comppad_name = newAV();
7221 PL_comppad_name_fill = 0;
7222 PL_min_intro_pending = 0;
7224 PL_subline = CopLINE(PL_curcop);
7226 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7227 PL_curpad[0] = (SV*)newAV();
7228 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7229 #endif /* USE_THREADS */
7231 comppadlist = newAV();
7232 AvREAL_off(comppadlist);
7233 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7234 av_store(comppadlist, 1, (SV*)PL_comppad);
7236 CvPADLIST(PL_compcv) = comppadlist;
7237 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7239 CvOWNER(PL_compcv) = 0;
7240 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7241 MUTEX_INIT(CvMUTEXP(PL_compcv));
7242 #endif /* USE_THREADS */
7244 return oldsavestack_ix;
7248 Perl_yywarn(pTHX_ char *s)
7251 PL_in_eval |= EVAL_WARNONLY;
7253 PL_in_eval &= ~EVAL_WARNONLY;
7258 Perl_yyerror(pTHX_ char *s)
7262 char *context = NULL;
7266 if (!yychar || (yychar == ';' && !PL_rsfp))
7268 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7269 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7270 while (isSPACE(*PL_oldoldbufptr))
7272 context = PL_oldoldbufptr;
7273 contlen = PL_bufptr - PL_oldoldbufptr;
7275 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7276 PL_oldbufptr != PL_bufptr) {
7277 while (isSPACE(*PL_oldbufptr))
7279 context = PL_oldbufptr;
7280 contlen = PL_bufptr - PL_oldbufptr;
7282 else if (yychar > 255)
7283 where = "next token ???";
7284 #ifdef USE_PURE_BISON
7285 /* GNU Bison sets the value -2 */
7286 else if (yychar == -2) {
7288 else if ((yychar & 127) == 127) {
7290 if (PL_lex_state == LEX_NORMAL ||
7291 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7292 where = "at end of line";
7293 else if (PL_lex_inpat)
7294 where = "within pattern";
7296 where = "within string";
7299 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7301 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7302 else if (isPRINT_LC(yychar))
7303 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7305 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7306 where = SvPVX(where_sv);
7308 msg = sv_2mortal(newSVpv(s, 0));
7309 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7310 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7312 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7314 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7315 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7316 Perl_sv_catpvf(aTHX_ msg,
7317 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7318 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7321 if (PL_in_eval & EVAL_WARNONLY)
7322 Perl_warn(aTHX_ "%"SVf, msg);
7325 if (PL_error_count >= 10) {
7326 if (PL_in_eval && SvCUR(ERRSV))
7327 Perl_croak(aTHX_ "%_%s has too many errors.\n",
7328 ERRSV, CopFILE(PL_curcop));
7330 Perl_croak(aTHX_ "%s has too many errors.\n",
7331 CopFILE(PL_curcop));
7334 PL_in_my_stash = Nullhv;
7345 * Restore a source filter.
7349 restore_rsfp(pTHXo_ void *f)
7351 PerlIO *fp = (PerlIO*)f;
7353 if (PL_rsfp == PerlIO_stdin())
7354 PerlIO_clearerr(PL_rsfp);
7355 else if (PL_rsfp && (PL_rsfp != fp))
7356 PerlIO_close(PL_rsfp);