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);
31 #ifndef PERL_NO_UTF16_FILTER
32 static I32 utf16_textfilter(pTHXo_ int idx, SV *sv, int maxlen);
33 static I32 utf16rev_textfilter(pTHXo_ int idx, SV *sv, int maxlen);
36 #define XFAKEBRACK 128
39 /*#define UTF (SvUTF8(PL_linestr) && !(PL_hints & HINT_BYTE))*/
40 #define UTF (PL_hints & HINT_UTF8)
42 /* In variables name $^X, these are the legal values for X.
43 * 1999-02-27 mjd-perl-patch@plover.com */
44 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
46 /* On MacOS, respect nonbreaking spaces */
47 #ifdef MACOS_TRADITIONAL
48 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
50 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
53 /* LEX_* are values for PL_lex_state, the state of the lexer.
54 * They are arranged oddly so that the guard on the switch statement
55 * can get by with a single comparison (if the compiler is smart enough).
58 /* #define LEX_NOTPARSING 11 is done in perl.h. */
61 #define LEX_INTERPNORMAL 9
62 #define LEX_INTERPCASEMOD 8
63 #define LEX_INTERPPUSH 7
64 #define LEX_INTERPSTART 6
65 #define LEX_INTERPEND 5
66 #define LEX_INTERPENDMAYBE 4
67 #define LEX_INTERPCONCAT 3
68 #define LEX_INTERPCONST 2
69 #define LEX_FORMLINE 1
70 #define LEX_KNOWNEXT 0
72 /* XXX If this causes problems, set i_unistd=undef in the hint file. */
74 # include <unistd.h> /* Needed for execv() */
84 #define YYMAXLEVEL 100
86 YYSTYPE* yylval_pointer[YYMAXLEVEL];
87 int* yychar_pointer[YYMAXLEVEL];
91 # define yylval (*yylval_pointer[yyactlevel])
92 # define yychar (*yychar_pointer[yyactlevel])
93 # define PERL_YYLEX_PARAM yylval_pointer[yyactlevel],yychar_pointer[yyactlevel]
95 # define yylex() Perl_yylex(aTHX_ yylval_pointer[yyactlevel],yychar_pointer[yyactlevel])
100 /* CLINE is a macro that ensures PL_copline has a sane value */
105 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
108 * Convenience functions to return different tokens and prime the
109 * lexer for the next token. They all take an argument.
111 * TOKEN : generic token (used for '(', DOLSHARP, etc)
112 * OPERATOR : generic operator
113 * AOPERATOR : assignment operator
114 * PREBLOCK : beginning the block after an if, while, foreach, ...
115 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
116 * PREREF : *EXPR where EXPR is not a simple identifier
117 * TERM : expression term
118 * LOOPX : loop exiting command (goto, last, dump, etc)
119 * FTST : file test operator
120 * FUN0 : zero-argument function
121 * FUN1 : not used, except for not, which isn't a UNIOP
122 * BOop : bitwise or or xor
124 * SHop : shift operator
125 * PWop : power operator
126 * PMop : pattern-matching operator
127 * Aop : addition-level operator
128 * Mop : multiplication-level operator
129 * Eop : equality-testing operator
130 * Rop : relational operator <= != gt
132 * Also see LOP and lop() below.
135 #define TOKEN(retval) return (PL_bufptr = s,(int)retval)
136 #define OPERATOR(retval) return (PL_expect = XTERM,PL_bufptr = s,(int)retval)
137 #define AOPERATOR(retval) return ao((PL_expect = XTERM,PL_bufptr = s,(int)retval))
138 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
139 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
140 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s,(int)retval)
141 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR,PL_bufptr = s,(int)retval)
142 #define LOOPX(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
143 #define FTST(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)UNIOP)
144 #define FUN0(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
145 #define FUN1(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
146 #define BOop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
147 #define BAop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
148 #define SHop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
149 #define PWop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
150 #define PMop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
151 #define Aop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
152 #define Mop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
153 #define Eop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
154 #define Rop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)RELOP)
156 /* This bit of chicanery makes a unary function followed by
157 * a parenthesis into a function with one argument, highest precedence.
159 #define UNI(f) return(yylval.ival = f, \
162 PL_last_uni = PL_oldbufptr, \
163 PL_last_lop_op = f, \
164 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
166 #define UNIBRACK(f) return(yylval.ival = f, \
168 PL_last_uni = PL_oldbufptr, \
169 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
171 /* grandfather return to old style */
172 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
177 * This subroutine detects &&= and ||= and turns an ANDAND or OROR
178 * into an OP_ANDASSIGN or OP_ORASSIGN
182 S_ao(pTHX_ int toketype)
184 if (*PL_bufptr == '=') {
186 if (toketype == ANDAND)
187 yylval.ival = OP_ANDASSIGN;
188 else if (toketype == OROR)
189 yylval.ival = OP_ORASSIGN;
197 * When Perl expects an operator and finds something else, no_op
198 * prints the warning. It always prints "<something> found where
199 * operator expected. It prints "Missing semicolon on previous line?"
200 * if the surprise occurs at the start of the line. "do you need to
201 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
202 * where the compiler doesn't know if foo is a method call or a function.
203 * It prints "Missing operator before end of line" if there's nothing
204 * after the missing operator, or "... before <...>" if there is something
205 * after the missing operator.
209 S_no_op(pTHX_ char *what, char *s)
211 char *oldbp = PL_bufptr;
212 bool is_first = (PL_oldbufptr == PL_linestart);
218 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
220 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
221 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
223 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
224 if (t < PL_bufptr && isSPACE(*t))
225 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
226 t - PL_oldoldbufptr, PL_oldoldbufptr);
230 Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
237 * Complain about missing quote/regexp/heredoc terminator.
238 * If it's called with (char *)NULL then it cauterizes the line buffer.
239 * If we're in a delimited string and the delimiter is a control
240 * character, it's reformatted into a two-char sequence like ^C.
245 S_missingterm(pTHX_ char *s)
250 char *nl = strrchr(s,'\n');
256 iscntrl(PL_multi_close)
258 PL_multi_close < 32 || PL_multi_close == 127
262 tmpbuf[1] = toCTRL(PL_multi_close);
268 *tmpbuf = PL_multi_close;
272 q = strchr(s,'"') ? '\'' : '"';
273 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
281 Perl_deprecate(pTHX_ char *s)
284 if (ckWARN(WARN_DEPRECATED))
285 Perl_warner(aTHX_ WARN_DEPRECATED, "Use of %s is deprecated", s);
290 * Deprecate a comma-less variable list.
296 deprecate("comma-less variable list");
300 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
301 * utf16-to-utf8-reversed.
304 #ifdef PERL_CR_FILTER
308 register char *s = SvPVX(sv);
309 register char *e = s + SvCUR(sv);
310 /* outer loop optimized to do nothing if there are no CR-LFs */
312 if (*s++ == '\r' && *s == '\n') {
313 /* hit a CR-LF, need to copy the rest */
314 register char *d = s - 1;
317 if (*s == '\r' && s[1] == '\n')
328 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
330 I32 count = FILTER_READ(idx+1, sv, maxlen);
331 if (count > 0 && !maxlen)
339 * Initialize variables. Uses the Perl save_stack to save its state (for
340 * recursive calls to the parser).
344 Perl_lex_start(pTHX_ SV *line)
350 SAVEI32(PL_lex_dojoin);
351 SAVEI32(PL_lex_brackets);
352 SAVEI32(PL_lex_casemods);
353 SAVEI32(PL_lex_starts);
354 SAVEI32(PL_lex_state);
355 SAVEVPTR(PL_lex_inpat);
356 SAVEI32(PL_lex_inwhat);
357 if (PL_lex_state == LEX_KNOWNEXT) {
358 I32 toke = PL_nexttoke;
359 while (--toke >= 0) {
360 SAVEI32(PL_nexttype[toke]);
361 SAVEVPTR(PL_nextval[toke]);
363 SAVEI32(PL_nexttoke);
365 SAVECOPLINE(PL_curcop);
368 SAVEPPTR(PL_oldbufptr);
369 SAVEPPTR(PL_oldoldbufptr);
370 SAVEPPTR(PL_linestart);
371 SAVESPTR(PL_linestr);
372 SAVEPPTR(PL_lex_brackstack);
373 SAVEPPTR(PL_lex_casestack);
374 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
375 SAVESPTR(PL_lex_stuff);
376 SAVEI32(PL_lex_defer);
377 SAVEI32(PL_sublex_info.sub_inwhat);
378 SAVESPTR(PL_lex_repl);
380 SAVEINT(PL_lex_expect);
382 PL_lex_state = LEX_NORMAL;
386 New(899, PL_lex_brackstack, 120, char);
387 New(899, PL_lex_casestack, 12, char);
388 SAVEFREEPV(PL_lex_brackstack);
389 SAVEFREEPV(PL_lex_casestack);
391 *PL_lex_casestack = '\0';
394 PL_lex_stuff = Nullsv;
395 PL_lex_repl = Nullsv;
399 PL_sublex_info.sub_inwhat = 0;
401 if (SvREADONLY(PL_linestr))
402 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
403 s = SvPV(PL_linestr, len);
404 if (len && s[len-1] != ';') {
405 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
406 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
407 sv_catpvn(PL_linestr, "\n;", 2);
409 SvTEMP_off(PL_linestr);
410 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
411 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
413 PL_rs = newSVpvn("\n", 1);
419 * Finalizer for lexing operations. Must be called when the parser is
420 * done with the lexer.
426 PL_doextract = FALSE;
431 * This subroutine has nothing to do with tilting, whether at windmills
432 * or pinball tables. Its name is short for "increment line". It
433 * increments the current line number in CopLINE(PL_curcop) and checks
434 * to see whether the line starts with a comment of the form
435 * # line 500 "foo.pm"
436 * If so, it sets the current line number and file to the values in the comment.
440 S_incline(pTHX_ char *s)
448 CopLINE_inc(PL_curcop);
451 while (SPACE_OR_TAB(*s)) s++;
452 if (strnEQ(s, "line", 4))
456 if (*s == ' ' || *s == '\t')
460 while (SPACE_OR_TAB(*s)) s++;
466 while (SPACE_OR_TAB(*s))
468 if (*s == '"' && (t = strchr(s+1, '"'))) {
473 for (t = s; !isSPACE(*t); t++) ;
476 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
478 if (*e != '\n' && *e != '\0')
479 return; /* false alarm */
485 Safefree(CopFILE(PL_curcop));
487 SvREFCNT_dec(CopFILEGV(PL_curcop));
489 CopFILE_set(PL_curcop, s);
492 CopLINE_set(PL_curcop, atoi(n)-1);
497 * Called to gobble the appropriate amount and type of whitespace.
498 * Skips comments as well.
502 S_skipspace(pTHX_ register char *s)
505 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
506 while (s < PL_bufend && SPACE_OR_TAB(*s))
512 SSize_t oldprevlen, oldoldprevlen;
513 SSize_t oldloplen, oldunilen;
514 while (s < PL_bufend && isSPACE(*s)) {
515 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
520 if (s < PL_bufend && *s == '#') {
521 while (s < PL_bufend && *s != '\n')
525 if (PL_in_eval && !PL_rsfp) {
532 /* only continue to recharge the buffer if we're at the end
533 * of the buffer, we're not reading from a source filter, and
534 * we're in normal lexing mode
536 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
537 PL_lex_state == LEX_FORMLINE)
540 /* try to recharge the buffer */
541 if ((s = filter_gets(PL_linestr, PL_rsfp,
542 (prevlen = SvCUR(PL_linestr)))) == Nullch)
544 /* end of file. Add on the -p or -n magic */
545 if (PL_minus_n || PL_minus_p) {
546 sv_setpv(PL_linestr,PL_minus_p ?
547 ";}continue{print or die qq(-p destination: $!\\n)" :
549 sv_catpv(PL_linestr,";}");
550 PL_minus_n = PL_minus_p = 0;
553 sv_setpv(PL_linestr,";");
555 /* reset variables for next time we lex */
556 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
558 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
560 /* Close the filehandle. Could be from -P preprocessor,
561 * STDIN, or a regular file. If we were reading code from
562 * STDIN (because the commandline held no -e or filename)
563 * then we don't close it, we reset it so the code can
564 * read from STDIN too.
567 if (PL_preprocess && !PL_in_eval)
568 (void)PerlProc_pclose(PL_rsfp);
569 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
570 PerlIO_clearerr(PL_rsfp);
572 (void)PerlIO_close(PL_rsfp);
577 /* not at end of file, so we only read another line */
578 /* make corresponding updates to old pointers, for yyerror() */
579 oldprevlen = PL_oldbufptr - PL_bufend;
580 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
582 oldunilen = PL_last_uni - PL_bufend;
584 oldloplen = PL_last_lop - PL_bufend;
585 PL_linestart = PL_bufptr = s + prevlen;
586 PL_bufend = s + SvCUR(PL_linestr);
588 PL_oldbufptr = s + oldprevlen;
589 PL_oldoldbufptr = s + oldoldprevlen;
591 PL_last_uni = s + oldunilen;
593 PL_last_lop = s + oldloplen;
596 /* debugger active and we're not compiling the debugger code,
597 * so store the line into the debugger's array of lines
599 if (PERLDB_LINE && PL_curstash != PL_debstash) {
600 SV *sv = NEWSV(85,0);
602 sv_upgrade(sv, SVt_PVMG);
603 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
604 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
611 * Check the unary operators to ensure there's no ambiguity in how they're
612 * used. An ambiguous piece of code would be:
614 * This doesn't mean rand() + 5. Because rand() is a unary operator,
615 * the +5 is its argument.
625 if (PL_oldoldbufptr != PL_last_uni)
627 while (isSPACE(*PL_last_uni))
629 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
630 if ((t = strchr(s, '(')) && t < PL_bufptr)
632 if (ckWARN_d(WARN_AMBIGUOUS)){
635 Perl_warner(aTHX_ WARN_AMBIGUOUS,
636 "Warning: Use of \"%s\" without parens is ambiguous",
642 /* workaround to replace the UNI() macro with a function. Only the
643 * hints/uts.sh file mentions this. Other comments elsewhere in the
644 * source indicate Microport Unix might need it too.
650 #define UNI(f) return uni(f,s)
653 S_uni(pTHX_ I32 f, char *s)
658 PL_last_uni = PL_oldbufptr;
669 #endif /* CRIPPLED_CC */
672 * LOP : macro to build a list operator. Its behaviour has been replaced
673 * with a subroutine, S_lop() for which LOP is just another name.
676 #define LOP(f,x) return lop(f,x,s)
680 * Build a list operator (or something that might be one). The rules:
681 * - if we have a next token, then it's a list operator [why?]
682 * - if the next thing is an opening paren, then it's a function
683 * - else it's a list operator
687 S_lop(pTHX_ I32 f, int x, char *s)
694 PL_last_lop = PL_oldbufptr;
709 * When the lexer realizes it knows the next token (for instance,
710 * it is reordering tokens for the parser) then it can call S_force_next
711 * to know what token to return the next time the lexer is called. Caller
712 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
713 * handles the token correctly.
717 S_force_next(pTHX_ I32 type)
719 PL_nexttype[PL_nexttoke] = type;
721 if (PL_lex_state != LEX_KNOWNEXT) {
722 PL_lex_defer = PL_lex_state;
723 PL_lex_expect = PL_expect;
724 PL_lex_state = LEX_KNOWNEXT;
730 * When the lexer knows the next thing is a word (for instance, it has
731 * just seen -> and it knows that the next char is a word char, then
732 * it calls S_force_word to stick the next word into the PL_next lookahead.
735 * char *start : buffer position (must be within PL_linestr)
736 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
737 * int check_keyword : if true, Perl checks to make sure the word isn't
738 * a keyword (do this if the word is a label, e.g. goto FOO)
739 * int allow_pack : if true, : characters will also be allowed (require,
741 * int allow_initial_tick : used by the "sub" lexer only.
745 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
750 start = skipspace(start);
752 if (isIDFIRST_lazy_if(s,UTF) ||
753 (allow_pack && *s == ':') ||
754 (allow_initial_tick && *s == '\'') )
756 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
757 if (check_keyword && keyword(PL_tokenbuf, len))
759 if (token == METHOD) {
764 PL_expect = XOPERATOR;
767 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
768 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
776 * Called when the lexer wants $foo *foo &foo etc, but the program
777 * text only contains the "foo" portion. The first argument is a pointer
778 * to the "foo", and the second argument is the type symbol to prefix.
779 * Forces the next token to be a "WORD".
780 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
784 S_force_ident(pTHX_ register char *s, int kind)
787 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
788 PL_nextval[PL_nexttoke].opval = o;
791 dTHR; /* just for in_eval */
792 o->op_private = OPpCONST_ENTERED;
793 /* XXX see note in pp_entereval() for why we forgo typo
794 warnings if the symbol must be introduced in an eval.
796 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
797 kind == '$' ? SVt_PV :
798 kind == '@' ? SVt_PVAV :
799 kind == '%' ? SVt_PVHV :
807 Perl_str_to_version(pTHX_ SV *sv)
812 char *start = SvPVx(sv,len);
813 bool utf = SvUTF8(sv) ? TRUE : FALSE;
814 char *end = start + len;
815 while (start < end) {
819 n = utf8_to_uv_chk((U8*)start, len, &skip, 0);
824 retval += ((NV)n)/nshift;
833 * Forces the next token to be a version number.
837 S_force_version(pTHX_ char *s)
839 OP *version = Nullop;
848 for (; isDIGIT(*d) || *d == '_' || *d == '.'; d++);
849 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
851 s = scan_num(s, &yylval);
852 version = yylval.opval;
853 ver = cSVOPx(version)->op_sv;
854 if (SvPOK(ver) && !SvNIOK(ver)) {
855 (void)SvUPGRADE(ver, SVt_PVNV);
856 SvNVX(ver) = str_to_version(ver);
857 SvNOK_on(ver); /* hint that it is a version */
862 /* NOTE: The parser sees the package name and the VERSION swapped */
863 PL_nextval[PL_nexttoke].opval = version;
871 * Tokenize a quoted string passed in as an SV. It finds the next
872 * chunk, up to end of string or a backslash. It may make a new
873 * SV containing that chunk (if HINT_NEW_STRING is on). It also
878 S_tokeq(pTHX_ SV *sv)
889 s = SvPV_force(sv, len);
890 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
893 while (s < send && *s != '\\')
898 if ( PL_hints & HINT_NEW_STRING )
899 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
902 if (s + 1 < send && (s[1] == '\\'))
903 s++; /* all that, just for this */
908 SvCUR_set(sv, d - SvPVX(sv));
910 if ( PL_hints & HINT_NEW_STRING )
911 return new_constant(NULL, 0, "q", sv, pv, "q");
916 * Now come three functions related to double-quote context,
917 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
918 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
919 * interact with PL_lex_state, and create fake ( ... ) argument lists
920 * to handle functions and concatenation.
921 * They assume that whoever calls them will be setting up a fake
922 * join call, because each subthing puts a ',' after it. This lets
925 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
927 * (I'm not sure whether the spurious commas at the end of lcfirst's
928 * arguments and join's arguments are created or not).
933 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
935 * Pattern matching will set PL_lex_op to the pattern-matching op to
936 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
938 * OP_CONST and OP_READLINE are easy--just make the new op and return.
940 * Everything else becomes a FUNC.
942 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
943 * had an OP_CONST or OP_READLINE). This just sets us up for a
944 * call to S_sublex_push().
950 register I32 op_type = yylval.ival;
952 if (op_type == OP_NULL) {
953 yylval.opval = PL_lex_op;
957 if (op_type == OP_CONST || op_type == OP_READLINE) {
958 SV *sv = tokeq(PL_lex_stuff);
960 if (SvTYPE(sv) == SVt_PVIV) {
961 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
967 nsv = newSVpvn(p, len);
973 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
974 PL_lex_stuff = Nullsv;
978 PL_sublex_info.super_state = PL_lex_state;
979 PL_sublex_info.sub_inwhat = op_type;
980 PL_sublex_info.sub_op = PL_lex_op;
981 PL_lex_state = LEX_INTERPPUSH;
985 yylval.opval = PL_lex_op;
995 * Create a new scope to save the lexing state. The scope will be
996 * ended in S_sublex_done. Returns a '(', starting the function arguments
997 * to the uc, lc, etc. found before.
998 * Sets PL_lex_state to LEX_INTERPCONCAT.
1007 PL_lex_state = PL_sublex_info.super_state;
1008 SAVEI32(PL_lex_dojoin);
1009 SAVEI32(PL_lex_brackets);
1010 SAVEI32(PL_lex_casemods);
1011 SAVEI32(PL_lex_starts);
1012 SAVEI32(PL_lex_state);
1013 SAVEVPTR(PL_lex_inpat);
1014 SAVEI32(PL_lex_inwhat);
1015 SAVECOPLINE(PL_curcop);
1016 SAVEPPTR(PL_bufptr);
1017 SAVEPPTR(PL_oldbufptr);
1018 SAVEPPTR(PL_oldoldbufptr);
1019 SAVEPPTR(PL_linestart);
1020 SAVESPTR(PL_linestr);
1021 SAVEPPTR(PL_lex_brackstack);
1022 SAVEPPTR(PL_lex_casestack);
1024 PL_linestr = PL_lex_stuff;
1025 PL_lex_stuff = Nullsv;
1027 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1028 = SvPVX(PL_linestr);
1029 PL_bufend += SvCUR(PL_linestr);
1030 SAVEFREESV(PL_linestr);
1032 PL_lex_dojoin = FALSE;
1033 PL_lex_brackets = 0;
1034 New(899, PL_lex_brackstack, 120, char);
1035 New(899, PL_lex_casestack, 12, char);
1036 SAVEFREEPV(PL_lex_brackstack);
1037 SAVEFREEPV(PL_lex_casestack);
1038 PL_lex_casemods = 0;
1039 *PL_lex_casestack = '\0';
1041 PL_lex_state = LEX_INTERPCONCAT;
1042 CopLINE_set(PL_curcop, PL_multi_start);
1044 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1045 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1046 PL_lex_inpat = PL_sublex_info.sub_op;
1048 PL_lex_inpat = Nullop;
1055 * Restores lexer state after a S_sublex_push.
1061 if (!PL_lex_starts++) {
1062 PL_expect = XOPERATOR;
1063 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn("",0));
1067 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1068 PL_lex_state = LEX_INTERPCASEMOD;
1072 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1073 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1074 PL_linestr = PL_lex_repl;
1076 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1077 PL_bufend += SvCUR(PL_linestr);
1078 SAVEFREESV(PL_linestr);
1079 PL_lex_dojoin = FALSE;
1080 PL_lex_brackets = 0;
1081 PL_lex_casemods = 0;
1082 *PL_lex_casestack = '\0';
1084 if (SvEVALED(PL_lex_repl)) {
1085 PL_lex_state = LEX_INTERPNORMAL;
1087 /* we don't clear PL_lex_repl here, so that we can check later
1088 whether this is an evalled subst; that means we rely on the
1089 logic to ensure sublex_done() is called again only via the
1090 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1093 PL_lex_state = LEX_INTERPCONCAT;
1094 PL_lex_repl = Nullsv;
1100 PL_bufend = SvPVX(PL_linestr);
1101 PL_bufend += SvCUR(PL_linestr);
1102 PL_expect = XOPERATOR;
1103 PL_sublex_info.sub_inwhat = 0;
1111 Extracts a pattern, double-quoted string, or transliteration. This
1114 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1115 processing a pattern (PL_lex_inpat is true), a transliteration
1116 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1118 Returns a pointer to the character scanned up to. Iff this is
1119 advanced from the start pointer supplied (ie if anything was
1120 successfully parsed), will leave an OP for the substring scanned
1121 in yylval. Caller must intuit reason for not parsing further
1122 by looking at the next characters herself.
1126 double-quoted style: \r and \n
1127 regexp special ones: \D \s
1129 backrefs: \1 (deprecated in substitution replacements)
1130 case and quoting: \U \Q \E
1131 stops on @ and $, but not for $ as tail anchor
1133 In transliterations:
1134 characters are VERY literal, except for - not at the start or end
1135 of the string, which indicates a range. scan_const expands the
1136 range to the full set of intermediate characters.
1138 In double-quoted strings:
1140 double-quoted style: \r and \n
1142 backrefs: \1 (deprecated)
1143 case and quoting: \U \Q \E
1146 scan_const does *not* construct ops to handle interpolated strings.
1147 It stops processing as soon as it finds an embedded $ or @ variable
1148 and leaves it to the caller to work out what's going on.
1150 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1152 $ in pattern could be $foo or could be tail anchor. Assumption:
1153 it's a tail anchor if $ is the last thing in the string, or if it's
1154 followed by one of ")| \n\t"
1156 \1 (backreferences) are turned into $1
1158 The structure of the code is
1159 while (there's a character to process) {
1160 handle transliteration ranges
1161 skip regexp comments
1162 skip # initiated comments in //x patterns
1163 check for embedded @foo
1164 check for embedded scalars
1166 leave intact backslashes from leave (below)
1167 deprecate \1 in strings and sub replacements
1168 handle string-changing backslashes \l \U \Q \E, etc.
1169 switch (what was escaped) {
1170 handle - in a transliteration (becomes a literal -)
1171 handle \132 octal characters
1172 handle 0x15 hex characters
1173 handle \cV (control V)
1174 handle printf backslashes (\f, \r, \n, etc)
1176 } (end if backslash)
1177 } (end while character to read)
1182 S_scan_const(pTHX_ char *start)
1184 register char *send = PL_bufend; /* end of the constant */
1185 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1186 register char *s = start; /* start of the constant */
1187 register char *d = SvPVX(sv); /* destination for copies */
1188 bool dorange = FALSE; /* are we in a translit range? */
1189 bool didrange = FALSE; /* did we just finish a range? */
1190 bool has_utf = FALSE; /* embedded \x{} */
1193 I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1194 ? (PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1196 I32 thisutf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1197 ? (PL_sublex_info.sub_op->op_private & (PL_lex_repl ?
1198 OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF))
1200 const char *leaveit = /* set of acceptably-backslashed characters */
1202 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1205 while (s < send || dorange) {
1206 /* get transliterations out of the way (they're most literal) */
1207 if (PL_lex_inwhat == OP_TRANS) {
1208 /* expand a range A-Z to the full set of characters. AIE! */
1210 I32 i; /* current expanded character */
1211 I32 min; /* first character in range */
1212 I32 max; /* last character in range */
1214 i = d - SvPVX(sv); /* remember current offset */
1215 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1216 d = SvPVX(sv) + i; /* refresh d after realloc */
1217 d -= 2; /* eat the first char and the - */
1219 min = (U8)*d; /* first char in range */
1220 max = (U8)d[1]; /* last char in range */
1224 "Invalid [] range \"%c-%c\" in transliteration operator",
1225 (char)min, (char)max);
1229 if ((isLOWER(min) && isLOWER(max)) ||
1230 (isUPPER(min) && isUPPER(max))) {
1232 for (i = min; i <= max; i++)
1236 for (i = min; i <= max; i++)
1243 for (i = min; i <= max; i++)
1246 /* 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 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
1258 *d++ = (char)0xff; /* use illegal utf8 byte--see pmtrans */
1270 /* if we get here, we're not doing a transliteration */
1272 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1273 except for the last char, which will be done separately. */
1274 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1276 while (s < send && *s != ')')
1279 else if (s[2] == '{' /* This should match regcomp.c */
1280 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1283 char *regparse = s + (s[2] == '{' ? 3 : 4);
1286 while (count && (c = *regparse)) {
1287 if (c == '\\' && regparse[1])
1295 if (*regparse != ')') {
1296 regparse--; /* Leave one char for continuation. */
1297 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1299 while (s < regparse)
1304 /* likewise skip #-initiated comments in //x patterns */
1305 else if (*s == '#' && PL_lex_inpat &&
1306 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1307 while (s+1 < send && *s != '\n')
1311 /* check for embedded arrays
1312 (@foo, @:foo, @'foo, @{foo}, @$foo, @+, @-)
1314 else if (*s == '@' && s[1]
1315 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1])))
1318 /* check for embedded scalars. only stop if we're sure it's a
1321 else if (*s == '$') {
1322 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1324 if (s + 1 < send && !strchr("()| \n\t", s[1]))
1325 break; /* in regexp, $ might be tail anchor */
1328 /* (now in tr/// code again) */
1330 if (*s & 0x80 && thisutf) {
1334 uv = utf8_to_uv_chk((U8*)s, send - s, &len, UTF8_CHECK_ONLY);
1336 /* Illegal UTF8 (a high-bit byte), make it valid. */
1337 char *old_pvx = SvPVX(sv);
1338 /* need space for one extra char (NOTE: SvCUR() not set here) */
1339 d = SvGROW(sv, SvLEN(sv) + 1) + (d - old_pvx);
1340 d = (char*)uv_to_utf8((U8*)d, (U8)*s++);
1351 if (*s == '\\' && s+1 < send) {
1354 /* some backslashes we leave behind */
1355 if (*leaveit && *s && strchr(leaveit, *s)) {
1361 /* deprecate \1 in strings and substitution replacements */
1362 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1363 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1365 dTHR; /* only for ckWARN */
1366 if (ckWARN(WARN_SYNTAX))
1367 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1372 /* string-change backslash escapes */
1373 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1378 /* if we get here, it's either a quoted -, or a digit */
1381 /* quoted - in transliterations */
1383 if (PL_lex_inwhat == OP_TRANS) {
1391 if (ckWARN(WARN_MISC) && isALNUM(*s))
1392 Perl_warner(aTHX_ WARN_MISC,
1393 "Unrecognized escape \\%c passed through",
1395 /* default action is to copy the quoted character */
1400 /* \132 indicates an octal constant */
1401 case '0': case '1': case '2': case '3':
1402 case '4': case '5': case '6': case '7':
1404 STRLEN len = 0; /* disallow underscores */
1405 uv = (UV)scan_oct(s, 3, &len);
1408 goto NUM_ESCAPE_INSERT;
1410 /* \x24 indicates a hex constant */
1414 char* e = strchr(s, '}');
1416 yyerror("Missing right brace on \\x{}");
1420 STRLEN len = 1; /* allow underscores */
1421 uv = (UV)scan_hex(s + 1, e - s - 1, &len);
1427 STRLEN len = 0; /* disallow underscores */
1428 uv = (UV)scan_hex(s, 2, &len);
1434 /* Insert oct or hex escaped character.
1435 * There will always enough room in sv since such escapes will
1436 * be longer than any utf8 sequence they can end up as
1439 if (!thisutf && !has_utf && uv > 255) {
1440 /* might need to recode whatever we have accumulated so far
1441 * if it contains any hibit chars
1445 for (c = SvPVX(sv); c < d; c++) {
1450 char *old_pvx = SvPVX(sv);
1452 d = SvGROW(sv, SvCUR(sv) + hicount + 1) + (d - old_pvx);
1461 uv_to_utf8((U8*)dst, (U8)*src--);
1471 if (thisutf || uv > 255) {
1472 d = (char*)uv_to_utf8((U8*)d, uv);
1484 /* \N{latin small letter a} is a named character */
1488 char* e = strchr(s, '}');
1494 yyerror("Missing right brace on \\N{}");
1498 res = newSVpvn(s + 1, e - s - 1);
1499 res = new_constant( Nullch, 0, "charnames",
1500 res, Nullsv, "\\N{...}" );
1501 str = SvPV(res,len);
1502 if (!has_utf && SvUTF8(res)) {
1503 char *ostart = SvPVX(sv);
1504 SvCUR_set(sv, d - ostart);
1507 sv_utf8_upgrade(sv);
1508 /* this just broke our allocation above... */
1509 SvGROW(sv, send - start);
1510 d = SvPVX(sv) + SvCUR(sv);
1513 if (len > e - s + 4) {
1514 char *odest = SvPVX(sv);
1516 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1517 d = SvPVX(sv) + (d - odest);
1519 Copy(str, d, len, char);
1526 yyerror("Missing braces on \\N{}");
1529 /* \c is a control character */
1546 /* printf-style backslashes, formfeeds, newlines, etc */
1564 *d++ = '\047'; /* CP 1047 */
1567 *d++ = '\057'; /* CP 1047 */
1581 } /* end if (backslash) */
1584 } /* while loop to process each character */
1586 /* terminate the string and set up the sv */
1588 SvCUR_set(sv, d - SvPVX(sv));
1593 /* shrink the sv if we allocated more than we used */
1594 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1595 SvLEN_set(sv, SvCUR(sv) + 1);
1596 Renew(SvPVX(sv), SvLEN(sv), char);
1599 /* return the substring (via yylval) only if we parsed anything */
1600 if (s > PL_bufptr) {
1601 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1602 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1604 ( PL_lex_inwhat == OP_TRANS
1606 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1609 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1616 * Returns TRUE if there's more to the expression (e.g., a subscript),
1619 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1621 * ->[ and ->{ return TRUE
1622 * { and [ outside a pattern are always subscripts, so return TRUE
1623 * if we're outside a pattern and it's not { or [, then return FALSE
1624 * if we're in a pattern and the first char is a {
1625 * {4,5} (any digits around the comma) returns FALSE
1626 * if we're in a pattern and the first char is a [
1628 * [SOMETHING] has a funky algorithm to decide whether it's a
1629 * character class or not. It has to deal with things like
1630 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1631 * anything else returns TRUE
1634 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1637 S_intuit_more(pTHX_ register char *s)
1639 if (PL_lex_brackets)
1641 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1643 if (*s != '{' && *s != '[')
1648 /* In a pattern, so maybe we have {n,m}. */
1665 /* On the other hand, maybe we have a character class */
1668 if (*s == ']' || *s == '^')
1671 /* this is terrifying, and it works */
1672 int weight = 2; /* let's weigh the evidence */
1674 unsigned char un_char = 255, last_un_char;
1675 char *send = strchr(s,']');
1676 char tmpbuf[sizeof PL_tokenbuf * 4];
1678 if (!send) /* has to be an expression */
1681 Zero(seen,256,char);
1684 else if (isDIGIT(*s)) {
1686 if (isDIGIT(s[1]) && s[2] == ']')
1692 for (; s < send; s++) {
1693 last_un_char = un_char;
1694 un_char = (unsigned char)*s;
1699 weight -= seen[un_char] * 10;
1700 if (isALNUM_lazy_if(s+1,UTF)) {
1701 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1702 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1707 else if (*s == '$' && s[1] &&
1708 strchr("[#!%*<>()-=",s[1])) {
1709 if (/*{*/ strchr("])} =",s[2]))
1718 if (strchr("wds]",s[1]))
1720 else if (seen['\''] || seen['"'])
1722 else if (strchr("rnftbxcav",s[1]))
1724 else if (isDIGIT(s[1])) {
1726 while (s[1] && isDIGIT(s[1]))
1736 if (strchr("aA01! ",last_un_char))
1738 if (strchr("zZ79~",s[1]))
1740 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1741 weight -= 5; /* cope with negative subscript */
1744 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1745 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1750 if (keyword(tmpbuf, d - tmpbuf))
1753 if (un_char == last_un_char + 1)
1755 weight -= seen[un_char];
1760 if (weight >= 0) /* probably a character class */
1770 * Does all the checking to disambiguate
1772 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1773 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1775 * First argument is the stuff after the first token, e.g. "bar".
1777 * Not a method if bar is a filehandle.
1778 * Not a method if foo is a subroutine prototyped to take a filehandle.
1779 * Not a method if it's really "Foo $bar"
1780 * Method if it's "foo $bar"
1781 * Not a method if it's really "print foo $bar"
1782 * Method if it's really "foo package::" (interpreted as package->foo)
1783 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1784 * Not a method if bar is a filehandle or package, but is quoted with
1789 S_intuit_method(pTHX_ char *start, GV *gv)
1791 char *s = start + (*start == '$');
1792 char tmpbuf[sizeof PL_tokenbuf];
1800 if ((cv = GvCVu(gv))) {
1801 char *proto = SvPVX(cv);
1811 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1812 /* start is the beginning of the possible filehandle/object,
1813 * and s is the end of it
1814 * tmpbuf is a copy of it
1817 if (*start == '$') {
1818 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1823 return *s == '(' ? FUNCMETH : METHOD;
1825 if (!keyword(tmpbuf, len)) {
1826 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1831 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1832 if (indirgv && GvCVu(indirgv))
1834 /* filehandle or package name makes it a method */
1835 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1837 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1838 return 0; /* no assumptions -- "=>" quotes bearword */
1840 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1841 newSVpvn(tmpbuf,len));
1842 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1846 return *s == '(' ? FUNCMETH : METHOD;
1854 * Return a string of Perl code to load the debugger. If PERL5DB
1855 * is set, it will return the contents of that, otherwise a
1856 * compile-time require of perl5db.pl.
1863 char *pdb = PerlEnv_getenv("PERL5DB");
1867 SETERRNO(0,SS$_NORMAL);
1868 return "BEGIN { require 'perl5db.pl' }";
1874 /* Encoded script support. filter_add() effectively inserts a
1875 * 'pre-processing' function into the current source input stream.
1876 * Note that the filter function only applies to the current source file
1877 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1879 * The datasv parameter (which may be NULL) can be used to pass
1880 * private data to this instance of the filter. The filter function
1881 * can recover the SV using the FILTER_DATA macro and use it to
1882 * store private buffers and state information.
1884 * The supplied datasv parameter is upgraded to a PVIO type
1885 * and the IoDIRP/IoANY field is used to store the function pointer,
1886 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1887 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1888 * private use must be set using malloc'd pointers.
1892 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1897 if (!PL_rsfp_filters)
1898 PL_rsfp_filters = newAV();
1900 datasv = NEWSV(255,0);
1901 if (!SvUPGRADE(datasv, SVt_PVIO))
1902 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1903 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
1904 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1905 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1906 funcp, SvPV_nolen(datasv)));
1907 av_unshift(PL_rsfp_filters, 1);
1908 av_store(PL_rsfp_filters, 0, datasv) ;
1913 /* Delete most recently added instance of this filter function. */
1915 Perl_filter_del(pTHX_ filter_t funcp)
1918 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1919 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1921 /* if filter is on top of stack (usual case) just pop it off */
1922 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1923 if (IoANY(datasv) == (void *)funcp) {
1924 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1925 IoANY(datasv) = (void *)NULL;
1926 sv_free(av_pop(PL_rsfp_filters));
1930 /* we need to search for the correct entry and clear it */
1931 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1935 /* Invoke the n'th filter function for the current rsfp. */
1937 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1940 /* 0 = read one text line */
1945 if (!PL_rsfp_filters)
1947 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1948 /* Provide a default input filter to make life easy. */
1949 /* Note that we append to the line. This is handy. */
1950 DEBUG_P(PerlIO_printf(Perl_debug_log,
1951 "filter_read %d: from rsfp\n", idx));
1955 int old_len = SvCUR(buf_sv) ;
1957 /* ensure buf_sv is large enough */
1958 SvGROW(buf_sv, old_len + maxlen) ;
1959 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1960 if (PerlIO_error(PL_rsfp))
1961 return -1; /* error */
1963 return 0 ; /* end of file */
1965 SvCUR_set(buf_sv, old_len + len) ;
1968 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1969 if (PerlIO_error(PL_rsfp))
1970 return -1; /* error */
1972 return 0 ; /* end of file */
1975 return SvCUR(buf_sv);
1977 /* Skip this filter slot if filter has been deleted */
1978 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1979 DEBUG_P(PerlIO_printf(Perl_debug_log,
1980 "filter_read %d: skipped (filter deleted)\n",
1982 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1984 /* Get function pointer hidden within datasv */
1985 funcp = (filter_t)IoANY(datasv);
1986 DEBUG_P(PerlIO_printf(Perl_debug_log,
1987 "filter_read %d: via function %p (%s)\n",
1988 idx, funcp, SvPV_nolen(datasv)));
1989 /* Call function. The function is expected to */
1990 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1991 /* Return: <0:error, =0:eof, >0:not eof */
1992 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1996 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1998 #ifdef PERL_CR_FILTER
1999 if (!PL_rsfp_filters) {
2000 filter_add(S_cr_textfilter,NULL);
2003 if (PL_rsfp_filters) {
2006 SvCUR_set(sv, 0); /* start with empty line */
2007 if (FILTER_READ(0, sv, 0) > 0)
2008 return ( SvPVX(sv) ) ;
2013 return (sv_gets(sv, fp, append));
2017 S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2021 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2025 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2026 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2028 return GvHV(gv); /* Foo:: */
2031 /* use constant CLASS => 'MyClass' */
2032 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2034 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2035 pkgname = SvPV_nolen(sv);
2039 return gv_stashpv(pkgname, FALSE);
2043 static char* exp_name[] =
2044 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2045 "ATTRTERM", "TERMBLOCK"
2052 Works out what to call the token just pulled out of the input
2053 stream. The yacc parser takes care of taking the ops we return and
2054 stitching them into a tree.
2060 if read an identifier
2061 if we're in a my declaration
2062 croak if they tried to say my($foo::bar)
2063 build the ops for a my() declaration
2064 if it's an access to a my() variable
2065 are we in a sort block?
2066 croak if my($a); $a <=> $b
2067 build ops for access to a my() variable
2068 if in a dq string, and they've said @foo and we can't find @foo
2070 build ops for a bareword
2071 if we already built the token before, use it.
2075 #pragma segment Perl_yylex
2078 #ifdef USE_PURE_BISON
2079 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
2087 #ifdef USE_PURE_BISON
2088 yylval_pointer[yyactlevel] = lvalp;
2089 yychar_pointer[yyactlevel] = lcharp;
2091 if (yyactlevel >= YYMAXLEVEL)
2092 Perl_croak(aTHX_ "panic: YYMAXLEVEL");
2097 #ifdef USE_PURE_BISON
2105 S_syylex(pTHX) /* need to be separate from yylex for reentrancy */
2115 /* check if there's an identifier for us to look at */
2116 if (PL_pending_ident) {
2117 /* pit holds the identifier we read and pending_ident is reset */
2118 char pit = PL_pending_ident;
2119 PL_pending_ident = 0;
2121 /* if we're in a my(), we can't allow dynamics here.
2122 $foo'bar has already been turned into $foo::bar, so
2123 just check for colons.
2125 if it's a legal name, the OP is a PADANY.
2128 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
2129 if (strchr(PL_tokenbuf,':'))
2130 yyerror(Perl_form(aTHX_ "No package name allowed for "
2131 "variable %s in \"our\"",
2133 tmp = pad_allocmy(PL_tokenbuf);
2136 if (strchr(PL_tokenbuf,':'))
2137 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
2139 yylval.opval = newOP(OP_PADANY, 0);
2140 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
2146 build the ops for accesses to a my() variable.
2148 Deny my($a) or my($b) in a sort block, *if* $a or $b is
2149 then used in a comparison. This catches most, but not
2150 all cases. For instance, it catches
2151 sort { my($a); $a <=> $b }
2153 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2154 (although why you'd do that is anyone's guess).
2157 if (!strchr(PL_tokenbuf,':')) {
2159 /* Check for single character per-thread SVs */
2160 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2161 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2162 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2164 yylval.opval = newOP(OP_THREADSV, 0);
2165 yylval.opval->op_targ = tmp;
2168 #endif /* USE_THREADS */
2169 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2170 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2171 /* might be an "our" variable" */
2172 if (SvFLAGS(namesv) & SVpad_OUR) {
2173 /* build ops for a bareword */
2174 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2175 sv_catpvn(sym, "::", 2);
2176 sv_catpv(sym, PL_tokenbuf+1);
2177 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2178 yylval.opval->op_private = OPpCONST_ENTERED;
2179 gv_fetchpv(SvPVX(sym),
2181 ? (GV_ADDMULTI | GV_ADDINEVAL)
2184 ((PL_tokenbuf[0] == '$') ? SVt_PV
2185 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2190 /* if it's a sort block and they're naming $a or $b */
2191 if (PL_last_lop_op == OP_SORT &&
2192 PL_tokenbuf[0] == '$' &&
2193 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2196 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2197 d < PL_bufend && *d != '\n';
2200 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2201 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2207 yylval.opval = newOP(OP_PADANY, 0);
2208 yylval.opval->op_targ = tmp;
2214 Whine if they've said @foo in a doublequoted string,
2215 and @foo isn't a variable we can find in the symbol
2218 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2219 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2220 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2221 && ckWARN(WARN_AMBIGUOUS))
2223 /* Downgraded from fatal to warning 20000522 mjd */
2224 Perl_warner(aTHX_ WARN_AMBIGUOUS,
2225 "Possible unintended interpolation of %s in string",
2230 /* build ops for a bareword */
2231 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2232 yylval.opval->op_private = OPpCONST_ENTERED;
2233 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2234 ((PL_tokenbuf[0] == '$') ? SVt_PV
2235 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2240 /* no identifier pending identification */
2242 switch (PL_lex_state) {
2244 case LEX_NORMAL: /* Some compilers will produce faster */
2245 case LEX_INTERPNORMAL: /* code if we comment these out. */
2249 /* when we've already built the next token, just pull it out of the queue */
2252 yylval = PL_nextval[PL_nexttoke];
2254 PL_lex_state = PL_lex_defer;
2255 PL_expect = PL_lex_expect;
2256 PL_lex_defer = LEX_NORMAL;
2258 return(PL_nexttype[PL_nexttoke]);
2260 /* interpolated case modifiers like \L \U, including \Q and \E.
2261 when we get here, PL_bufptr is at the \
2263 case LEX_INTERPCASEMOD:
2265 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2266 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2268 /* handle \E or end of string */
2269 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2273 if (PL_lex_casemods) {
2274 oldmod = PL_lex_casestack[--PL_lex_casemods];
2275 PL_lex_casestack[PL_lex_casemods] = '\0';
2277 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2279 PL_lex_state = LEX_INTERPCONCAT;
2283 if (PL_bufptr != PL_bufend)
2285 PL_lex_state = LEX_INTERPCONCAT;
2290 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2291 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2292 if (strchr("LU", *s) &&
2293 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2295 PL_lex_casestack[--PL_lex_casemods] = '\0';
2298 if (PL_lex_casemods > 10) {
2299 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2300 if (newlb != PL_lex_casestack) {
2302 PL_lex_casestack = newlb;
2305 PL_lex_casestack[PL_lex_casemods++] = *s;
2306 PL_lex_casestack[PL_lex_casemods] = '\0';
2307 PL_lex_state = LEX_INTERPCONCAT;
2308 PL_nextval[PL_nexttoke].ival = 0;
2311 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2313 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2315 PL_nextval[PL_nexttoke].ival = OP_LC;
2317 PL_nextval[PL_nexttoke].ival = OP_UC;
2319 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2321 Perl_croak(aTHX_ "panic: yylex");
2324 if (PL_lex_starts) {
2333 case LEX_INTERPPUSH:
2334 return sublex_push();
2336 case LEX_INTERPSTART:
2337 if (PL_bufptr == PL_bufend)
2338 return sublex_done();
2340 PL_lex_dojoin = (*PL_bufptr == '@');
2341 PL_lex_state = LEX_INTERPNORMAL;
2342 if (PL_lex_dojoin) {
2343 PL_nextval[PL_nexttoke].ival = 0;
2346 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2347 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2348 force_next(PRIVATEREF);
2350 force_ident("\"", '$');
2351 #endif /* USE_THREADS */
2352 PL_nextval[PL_nexttoke].ival = 0;
2354 PL_nextval[PL_nexttoke].ival = 0;
2356 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2359 if (PL_lex_starts++) {
2365 case LEX_INTERPENDMAYBE:
2366 if (intuit_more(PL_bufptr)) {
2367 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2373 if (PL_lex_dojoin) {
2374 PL_lex_dojoin = FALSE;
2375 PL_lex_state = LEX_INTERPCONCAT;
2378 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2379 && SvEVALED(PL_lex_repl))
2381 if (PL_bufptr != PL_bufend)
2382 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2383 PL_lex_repl = Nullsv;
2386 case LEX_INTERPCONCAT:
2388 if (PL_lex_brackets)
2389 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2391 if (PL_bufptr == PL_bufend)
2392 return sublex_done();
2394 if (SvIVX(PL_linestr) == '\'') {
2395 SV *sv = newSVsv(PL_linestr);
2398 else if ( PL_hints & HINT_NEW_RE )
2399 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2400 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2404 s = scan_const(PL_bufptr);
2406 PL_lex_state = LEX_INTERPCASEMOD;
2408 PL_lex_state = LEX_INTERPSTART;
2411 if (s != PL_bufptr) {
2412 PL_nextval[PL_nexttoke] = yylval;
2415 if (PL_lex_starts++)
2425 PL_lex_state = LEX_NORMAL;
2426 s = scan_formline(PL_bufptr);
2427 if (!PL_lex_formbrack)
2433 PL_oldoldbufptr = PL_oldbufptr;
2436 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2437 exp_name[PL_expect], s);
2443 if (isIDFIRST_lazy_if(s,UTF))
2445 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2448 goto fake_eof; /* emulate EOF on ^D or ^Z */
2453 if (PL_lex_brackets)
2454 yyerror("Missing right curly or square bracket");
2457 if (s++ < PL_bufend)
2458 goto retry; /* ignore stray nulls */
2461 if (!PL_in_eval && !PL_preambled) {
2462 PL_preambled = TRUE;
2463 sv_setpv(PL_linestr,incl_perldb());
2464 if (SvCUR(PL_linestr))
2465 sv_catpv(PL_linestr,";");
2467 while(AvFILLp(PL_preambleav) >= 0) {
2468 SV *tmpsv = av_shift(PL_preambleav);
2469 sv_catsv(PL_linestr, tmpsv);
2470 sv_catpv(PL_linestr, ";");
2473 sv_free((SV*)PL_preambleav);
2474 PL_preambleav = NULL;
2476 if (PL_minus_n || PL_minus_p) {
2477 sv_catpv(PL_linestr, "LINE: while (<>) {");
2479 sv_catpv(PL_linestr,"chomp;");
2481 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2483 GvIMPORTED_AV_on(gv);
2485 if (strchr("/'\"", *PL_splitstr)
2486 && strchr(PL_splitstr + 1, *PL_splitstr))
2487 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2490 s = "'~#\200\1'"; /* surely one char is unused...*/
2491 while (s[1] && strchr(PL_splitstr, *s)) s++;
2493 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2494 "q" + (delim == '\''), delim);
2495 for (s = PL_splitstr; *s; s++) {
2497 sv_catpvn(PL_linestr, "\\", 1);
2498 sv_catpvn(PL_linestr, s, 1);
2500 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2504 sv_catpv(PL_linestr,"@F=split(' ');");
2507 sv_catpv(PL_linestr, "\n");
2508 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2509 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2510 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2511 SV *sv = NEWSV(85,0);
2513 sv_upgrade(sv, SVt_PVMG);
2514 sv_setsv(sv,PL_linestr);
2515 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2521 bof = PL_rsfp && (PerlIO_tell(PL_rsfp) == 0); /* *Before* read! */
2522 s = filter_gets(PL_linestr, PL_rsfp, 0);
2526 if (PL_preprocess && !PL_in_eval)
2527 (void)PerlProc_pclose(PL_rsfp);
2528 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2529 PerlIO_clearerr(PL_rsfp);
2531 (void)PerlIO_close(PL_rsfp);
2533 PL_doextract = FALSE;
2535 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2536 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2537 sv_catpv(PL_linestr,";}");
2538 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2539 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2540 PL_minus_n = PL_minus_p = 0;
2543 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2544 sv_setpv(PL_linestr,"");
2545 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2547 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2548 s = swallow_bom((U8*)s);
2551 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2552 PL_doextract = FALSE;
2554 /* Incest with pod. */
2555 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2556 sv_setpv(PL_linestr, "");
2557 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2558 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2559 PL_doextract = FALSE;
2563 } while (PL_doextract);
2564 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2565 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2566 SV *sv = NEWSV(85,0);
2568 sv_upgrade(sv, SVt_PVMG);
2569 sv_setsv(sv,PL_linestr);
2570 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2572 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2573 if (CopLINE(PL_curcop) == 1) {
2574 while (s < PL_bufend && isSPACE(*s))
2576 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2580 if (*s == '#' && *(s+1) == '!')
2582 #ifdef ALTERNATE_SHEBANG
2584 static char as[] = ALTERNATE_SHEBANG;
2585 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2586 d = s + (sizeof(as) - 1);
2588 #endif /* ALTERNATE_SHEBANG */
2597 while (*d && !isSPACE(*d))
2601 #ifdef ARG_ZERO_IS_SCRIPT
2602 if (ipathend > ipath) {
2604 * HP-UX (at least) sets argv[0] to the script name,
2605 * which makes $^X incorrect. And Digital UNIX and Linux,
2606 * at least, set argv[0] to the basename of the Perl
2607 * interpreter. So, having found "#!", we'll set it right.
2609 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2610 assert(SvPOK(x) || SvGMAGICAL(x));
2611 if (sv_eq(x, CopFILESV(PL_curcop))) {
2612 sv_setpvn(x, ipath, ipathend - ipath);
2615 TAINT_NOT; /* $^X is always tainted, but that's OK */
2617 #endif /* ARG_ZERO_IS_SCRIPT */
2622 d = instr(s,"perl -");
2624 d = instr(s,"perl");
2626 /* avoid getting into infinite loops when shebang
2627 * line contains "Perl" rather than "perl" */
2629 for (d = ipathend-4; d >= ipath; --d) {
2630 if ((*d == 'p' || *d == 'P')
2631 && !ibcmp(d, "perl", 4))
2641 #ifdef ALTERNATE_SHEBANG
2643 * If the ALTERNATE_SHEBANG on this system starts with a
2644 * character that can be part of a Perl expression, then if
2645 * we see it but not "perl", we're probably looking at the
2646 * start of Perl code, not a request to hand off to some
2647 * other interpreter. Similarly, if "perl" is there, but
2648 * not in the first 'word' of the line, we assume the line
2649 * contains the start of the Perl program.
2651 if (d && *s != '#') {
2653 while (*c && !strchr("; \t\r\n\f\v#", *c))
2656 d = Nullch; /* "perl" not in first word; ignore */
2658 *s = '#'; /* Don't try to parse shebang line */
2660 #endif /* ALTERNATE_SHEBANG */
2661 #ifndef MACOS_TRADITIONAL
2666 !instr(s,"indir") &&
2667 instr(PL_origargv[0],"perl"))
2673 while (s < PL_bufend && isSPACE(*s))
2675 if (s < PL_bufend) {
2676 Newz(899,newargv,PL_origargc+3,char*);
2678 while (s < PL_bufend && !isSPACE(*s))
2681 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2684 newargv = PL_origargv;
2686 PerlProc_execv(ipath, newargv);
2687 Perl_croak(aTHX_ "Can't exec %s", ipath);
2691 U32 oldpdb = PL_perldb;
2692 bool oldn = PL_minus_n;
2693 bool oldp = PL_minus_p;
2695 while (*d && !isSPACE(*d)) d++;
2696 while (SPACE_OR_TAB(*d)) d++;
2700 if (*d == 'M' || *d == 'm') {
2702 while (*d && !isSPACE(*d)) d++;
2703 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2706 d = moreswitches(d);
2708 if ((PERLDB_LINE && !oldpdb) ||
2709 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2710 /* if we have already added "LINE: while (<>) {",
2711 we must not do it again */
2713 sv_setpv(PL_linestr, "");
2714 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2715 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2716 PL_preambled = FALSE;
2718 (void)gv_fetchfile(PL_origfilename);
2725 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2727 PL_lex_state = LEX_FORMLINE;
2732 #ifdef PERL_STRICT_CR
2733 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2735 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2737 case ' ': case '\t': case '\f': case 013:
2738 #ifdef MACOS_TRADITIONAL
2745 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2746 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2747 /* handle eval qq[#line 1 "foo"\n ...] */
2748 CopLINE_dec(PL_curcop);
2752 while (s < d && *s != '\n')
2757 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2759 PL_lex_state = LEX_FORMLINE;
2769 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2774 while (s < PL_bufend && SPACE_OR_TAB(*s))
2777 if (strnEQ(s,"=>",2)) {
2778 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2779 OPERATOR('-'); /* unary minus */
2781 PL_last_uni = PL_oldbufptr;
2782 PL_last_lop_op = OP_FTEREAD; /* good enough */
2784 case 'r': FTST(OP_FTEREAD);
2785 case 'w': FTST(OP_FTEWRITE);
2786 case 'x': FTST(OP_FTEEXEC);
2787 case 'o': FTST(OP_FTEOWNED);
2788 case 'R': FTST(OP_FTRREAD);
2789 case 'W': FTST(OP_FTRWRITE);
2790 case 'X': FTST(OP_FTREXEC);
2791 case 'O': FTST(OP_FTROWNED);
2792 case 'e': FTST(OP_FTIS);
2793 case 'z': FTST(OP_FTZERO);
2794 case 's': FTST(OP_FTSIZE);
2795 case 'f': FTST(OP_FTFILE);
2796 case 'd': FTST(OP_FTDIR);
2797 case 'l': FTST(OP_FTLINK);
2798 case 'p': FTST(OP_FTPIPE);
2799 case 'S': FTST(OP_FTSOCK);
2800 case 'u': FTST(OP_FTSUID);
2801 case 'g': FTST(OP_FTSGID);
2802 case 'k': FTST(OP_FTSVTX);
2803 case 'b': FTST(OP_FTBLK);
2804 case 'c': FTST(OP_FTCHR);
2805 case 't': FTST(OP_FTTTY);
2806 case 'T': FTST(OP_FTTEXT);
2807 case 'B': FTST(OP_FTBINARY);
2808 case 'M': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTMTIME);
2809 case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME);
2810 case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME);
2812 Perl_croak(aTHX_ "Unrecognized file test: -%c", (int)tmp);
2819 if (PL_expect == XOPERATOR)
2824 else if (*s == '>') {
2827 if (isIDFIRST_lazy_if(s,UTF)) {
2828 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2836 if (PL_expect == XOPERATOR)
2839 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2841 OPERATOR('-'); /* unary minus */
2848 if (PL_expect == XOPERATOR)
2853 if (PL_expect == XOPERATOR)
2856 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2862 if (PL_expect != XOPERATOR) {
2863 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2864 PL_expect = XOPERATOR;
2865 force_ident(PL_tokenbuf, '*');
2878 if (PL_expect == XOPERATOR) {
2882 PL_tokenbuf[0] = '%';
2883 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2884 if (!PL_tokenbuf[1]) {
2886 yyerror("Final % should be \\% or %name");
2889 PL_pending_ident = '%';
2908 switch (PL_expect) {
2911 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2913 PL_bufptr = s; /* update in case we back off */
2919 PL_expect = XTERMBLOCK;
2923 while (isIDFIRST_lazy_if(s,UTF)) {
2924 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2925 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2926 if (tmp < 0) tmp = -tmp;
2941 d = scan_str(d,TRUE,TRUE);
2944 SvREFCNT_dec(PL_lex_stuff);
2945 PL_lex_stuff = Nullsv;
2947 /* MUST advance bufptr here to avoid bogus
2948 "at end of line" context messages from yyerror().
2950 PL_bufptr = s + len;
2951 yyerror("Unterminated attribute parameter in attribute list");
2954 return 0; /* EOF indicator */
2958 SV *sv = newSVpvn(s, len);
2959 sv_catsv(sv, PL_lex_stuff);
2960 attrs = append_elem(OP_LIST, attrs,
2961 newSVOP(OP_CONST, 0, sv));
2962 SvREFCNT_dec(PL_lex_stuff);
2963 PL_lex_stuff = Nullsv;
2966 attrs = append_elem(OP_LIST, attrs,
2967 newSVOP(OP_CONST, 0,
2971 if (*s == ':' && s[1] != ':')
2974 break; /* require real whitespace or :'s */
2976 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
2977 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
2978 char q = ((*s == '\'') ? '"' : '\'');
2979 /* If here for an expression, and parsed no attrs, back off. */
2980 if (tmp == '=' && !attrs) {
2984 /* MUST advance bufptr here to avoid bogus "at end of line"
2985 context messages from yyerror().
2989 yyerror("Unterminated attribute list");
2991 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
2999 PL_nextval[PL_nexttoke].opval = attrs;
3007 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3008 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3024 if (PL_lex_brackets <= 0)
3025 yyerror("Unmatched right square bracket");
3028 if (PL_lex_state == LEX_INTERPNORMAL) {
3029 if (PL_lex_brackets == 0) {
3030 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3031 PL_lex_state = LEX_INTERPEND;
3038 if (PL_lex_brackets > 100) {
3039 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
3040 if (newlb != PL_lex_brackstack) {
3042 PL_lex_brackstack = newlb;
3045 switch (PL_expect) {
3047 if (PL_lex_formbrack) {
3051 if (PL_oldoldbufptr == PL_last_lop)
3052 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3054 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3055 OPERATOR(HASHBRACK);
3057 while (s < PL_bufend && SPACE_OR_TAB(*s))
3060 PL_tokenbuf[0] = '\0';
3061 if (d < PL_bufend && *d == '-') {
3062 PL_tokenbuf[0] = '-';
3064 while (d < PL_bufend && SPACE_OR_TAB(*d))
3067 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3068 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3070 while (d < PL_bufend && SPACE_OR_TAB(*d))
3073 char minus = (PL_tokenbuf[0] == '-');
3074 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3082 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3087 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3092 if (PL_oldoldbufptr == PL_last_lop)
3093 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3095 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3098 OPERATOR(HASHBRACK);
3099 /* This hack serves to disambiguate a pair of curlies
3100 * as being a block or an anon hash. Normally, expectation
3101 * determines that, but in cases where we're not in a
3102 * position to expect anything in particular (like inside
3103 * eval"") we have to resolve the ambiguity. This code
3104 * covers the case where the first term in the curlies is a
3105 * quoted string. Most other cases need to be explicitly
3106 * disambiguated by prepending a `+' before the opening
3107 * curly in order to force resolution as an anon hash.
3109 * XXX should probably propagate the outer expectation
3110 * into eval"" to rely less on this hack, but that could
3111 * potentially break current behavior of eval"".
3115 if (*s == '\'' || *s == '"' || *s == '`') {
3116 /* common case: get past first string, handling escapes */
3117 for (t++; t < PL_bufend && *t != *s;)
3118 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3122 else if (*s == 'q') {
3125 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3129 char open, close, term;
3132 while (t < PL_bufend && isSPACE(*t))
3136 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3140 for (t++; t < PL_bufend; t++) {
3141 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3143 else if (*t == open)
3147 for (t++; t < PL_bufend; t++) {
3148 if (*t == '\\' && t+1 < PL_bufend)
3150 else if (*t == close && --brackets <= 0)
3152 else if (*t == open)
3158 else if (isALNUM_lazy_if(t,UTF)) {
3160 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3163 while (t < PL_bufend && isSPACE(*t))
3165 /* if comma follows first term, call it an anon hash */
3166 /* XXX it could be a comma expression with loop modifiers */
3167 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3168 || (*t == '=' && t[1] == '>')))
3169 OPERATOR(HASHBRACK);
3170 if (PL_expect == XREF)
3173 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3179 yylval.ival = CopLINE(PL_curcop);
3180 if (isSPACE(*s) || *s == '#')
3181 PL_copline = NOLINE; /* invalidate current command line number */
3186 if (PL_lex_brackets <= 0)
3187 yyerror("Unmatched right curly bracket");
3189 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3190 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3191 PL_lex_formbrack = 0;
3192 if (PL_lex_state == LEX_INTERPNORMAL) {
3193 if (PL_lex_brackets == 0) {
3194 if (PL_expect & XFAKEBRACK) {
3195 PL_expect &= XENUMMASK;
3196 PL_lex_state = LEX_INTERPEND;
3198 return yylex(); /* ignore fake brackets */
3200 if (*s == '-' && s[1] == '>')
3201 PL_lex_state = LEX_INTERPENDMAYBE;
3202 else if (*s != '[' && *s != '{')
3203 PL_lex_state = LEX_INTERPEND;
3206 if (PL_expect & XFAKEBRACK) {
3207 PL_expect &= XENUMMASK;
3209 return yylex(); /* ignore fake brackets */
3219 if (PL_expect == XOPERATOR) {
3220 if (ckWARN(WARN_SEMICOLON)
3221 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3223 CopLINE_dec(PL_curcop);
3224 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3225 CopLINE_inc(PL_curcop);
3230 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3232 PL_expect = XOPERATOR;
3233 force_ident(PL_tokenbuf, '&');
3237 yylval.ival = (OPpENTERSUB_AMPER<<8);
3256 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3257 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3259 if (PL_expect == XSTATE && isALPHA(tmp) &&
3260 (s == PL_linestart+1 || s[-2] == '\n') )
3262 if (PL_in_eval && !PL_rsfp) {
3267 if (strnEQ(s,"=cut",4)) {
3281 PL_doextract = TRUE;
3284 if (PL_lex_brackets < PL_lex_formbrack) {
3286 #ifdef PERL_STRICT_CR
3287 for (t = s; SPACE_OR_TAB(*t); t++) ;
3289 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3291 if (*t == '\n' || *t == '#') {
3309 if (PL_expect != XOPERATOR) {
3310 if (s[1] != '<' && !strchr(s,'>'))
3313 s = scan_heredoc(s);
3315 s = scan_inputsymbol(s);
3316 TERM(sublex_start());
3321 SHop(OP_LEFT_SHIFT);
3335 SHop(OP_RIGHT_SHIFT);
3344 if (PL_expect == XOPERATOR) {
3345 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3348 return ','; /* grandfather non-comma-format format */
3352 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3353 PL_tokenbuf[0] = '@';
3354 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3355 sizeof PL_tokenbuf - 1, FALSE);
3356 if (PL_expect == XOPERATOR)
3357 no_op("Array length", s);
3358 if (!PL_tokenbuf[1])
3360 PL_expect = XOPERATOR;
3361 PL_pending_ident = '#';
3365 PL_tokenbuf[0] = '$';
3366 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3367 sizeof PL_tokenbuf - 1, FALSE);
3368 if (PL_expect == XOPERATOR)
3370 if (!PL_tokenbuf[1]) {
3372 yyerror("Final $ should be \\$ or $name");
3376 /* This kludge not intended to be bulletproof. */
3377 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3378 yylval.opval = newSVOP(OP_CONST, 0,
3379 newSViv(PL_compiling.cop_arybase));
3380 yylval.opval->op_private = OPpCONST_ARYBASE;
3386 if (PL_lex_state == LEX_NORMAL)
3389 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3392 PL_tokenbuf[0] = '@';
3393 if (ckWARN(WARN_SYNTAX)) {
3395 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3398 PL_bufptr = skipspace(PL_bufptr);
3399 while (t < PL_bufend && *t != ']')
3401 Perl_warner(aTHX_ WARN_SYNTAX,
3402 "Multidimensional syntax %.*s not supported",
3403 (t - PL_bufptr) + 1, PL_bufptr);
3407 else if (*s == '{') {
3408 PL_tokenbuf[0] = '%';
3409 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3410 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3412 char tmpbuf[sizeof PL_tokenbuf];
3414 for (t++; isSPACE(*t); t++) ;
3415 if (isIDFIRST_lazy_if(t,UTF)) {
3416 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3417 for (; isSPACE(*t); t++) ;
3418 if (*t == ';' && get_cv(tmpbuf, FALSE))
3419 Perl_warner(aTHX_ WARN_SYNTAX,
3420 "You need to quote \"%s\"", tmpbuf);
3426 PL_expect = XOPERATOR;
3427 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3428 bool islop = (PL_last_lop == PL_oldoldbufptr);
3429 if (!islop || PL_last_lop_op == OP_GREPSTART)
3430 PL_expect = XOPERATOR;
3431 else if (strchr("$@\"'`q", *s))
3432 PL_expect = XTERM; /* e.g. print $fh "foo" */
3433 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3434 PL_expect = XTERM; /* e.g. print $fh &sub */
3435 else if (isIDFIRST_lazy_if(s,UTF)) {
3436 char tmpbuf[sizeof PL_tokenbuf];
3437 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3438 if ((tmp = keyword(tmpbuf, len))) {
3439 /* binary operators exclude handle interpretations */
3451 PL_expect = XTERM; /* e.g. print $fh length() */
3456 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3457 if (gv && GvCVu(gv))
3458 PL_expect = XTERM; /* e.g. print $fh subr() */
3461 else if (isDIGIT(*s))
3462 PL_expect = XTERM; /* e.g. print $fh 3 */
3463 else if (*s == '.' && isDIGIT(s[1]))
3464 PL_expect = XTERM; /* e.g. print $fh .3 */
3465 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3466 PL_expect = XTERM; /* e.g. print $fh -1 */
3467 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3468 PL_expect = XTERM; /* print $fh <<"EOF" */
3470 PL_pending_ident = '$';
3474 if (PL_expect == XOPERATOR)
3476 PL_tokenbuf[0] = '@';
3477 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3478 if (!PL_tokenbuf[1]) {
3480 yyerror("Final @ should be \\@ or @name");
3483 if (PL_lex_state == LEX_NORMAL)
3485 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3487 PL_tokenbuf[0] = '%';
3489 /* Warn about @ where they meant $. */
3490 if (ckWARN(WARN_SYNTAX)) {
3491 if (*s == '[' || *s == '{') {
3493 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3495 if (*t == '}' || *t == ']') {
3497 PL_bufptr = skipspace(PL_bufptr);
3498 Perl_warner(aTHX_ WARN_SYNTAX,
3499 "Scalar value %.*s better written as $%.*s",
3500 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3505 PL_pending_ident = '@';
3508 case '/': /* may either be division or pattern */
3509 case '?': /* may either be conditional or pattern */
3510 if (PL_expect != XOPERATOR) {
3511 /* Disable warning on "study /blah/" */
3512 if (PL_oldoldbufptr == PL_last_uni
3513 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3514 || memNE(PL_last_uni, "study", 5)
3515 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3517 s = scan_pat(s,OP_MATCH);
3518 TERM(sublex_start());
3526 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3527 #ifdef PERL_STRICT_CR
3530 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3532 && (s == PL_linestart || s[-1] == '\n') )
3534 PL_lex_formbrack = 0;
3538 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3544 yylval.ival = OPf_SPECIAL;
3550 if (PL_expect != XOPERATOR)
3555 case '0': case '1': case '2': case '3': case '4':
3556 case '5': case '6': case '7': case '8': case '9':
3557 s = scan_num(s, &yylval);
3558 if (PL_expect == XOPERATOR)
3563 s = scan_str(s,FALSE,FALSE);
3564 if (PL_expect == XOPERATOR) {
3565 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3568 return ','; /* grandfather non-comma-format format */
3574 missingterm((char*)0);
3575 yylval.ival = OP_CONST;
3576 TERM(sublex_start());
3579 s = scan_str(s,FALSE,FALSE);
3580 if (PL_expect == XOPERATOR) {
3581 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3584 return ','; /* grandfather non-comma-format format */
3590 missingterm((char*)0);
3591 yylval.ival = OP_CONST;
3592 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3593 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3594 yylval.ival = OP_STRINGIFY;
3598 TERM(sublex_start());
3601 s = scan_str(s,FALSE,FALSE);
3602 if (PL_expect == XOPERATOR)
3603 no_op("Backticks",s);
3605 missingterm((char*)0);
3606 yylval.ival = OP_BACKTICK;
3608 TERM(sublex_start());
3612 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3613 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3615 if (PL_expect == XOPERATOR)
3616 no_op("Backslash",s);
3620 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3624 while (isDIGIT(*start) || *start == '_')
3626 if (*start == '.' && isDIGIT(start[1])) {
3627 s = scan_num(s, &yylval);
3630 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3631 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF)) {
3635 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3638 s = scan_num(s, &yylval);
3645 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3684 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3686 /* Some keywords can be followed by any delimiter, including ':' */
3687 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3688 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3689 (PL_tokenbuf[0] == 'q' &&
3690 strchr("qwxr", PL_tokenbuf[1])))));
3692 /* x::* is just a word, unless x is "CORE" */
3693 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3697 while (d < PL_bufend && isSPACE(*d))
3698 d++; /* no comments skipped here, or s### is misparsed */
3700 /* Is this a label? */
3701 if (!tmp && PL_expect == XSTATE
3702 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3704 yylval.pval = savepv(PL_tokenbuf);
3709 /* Check for keywords */
3710 tmp = keyword(PL_tokenbuf, len);
3712 /* Is this a word before a => operator? */
3713 if (*d == '=' && d[1] == '>') {
3715 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3716 yylval.opval->op_private = OPpCONST_BARE;
3720 if (tmp < 0) { /* second-class keyword? */
3721 GV *ogv = Nullgv; /* override (winner) */
3722 GV *hgv = Nullgv; /* hidden (loser) */
3723 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3725 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3728 if (GvIMPORTED_CV(gv))
3730 else if (! CvMETHOD(cv))
3734 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3735 (gv = *gvp) != (GV*)&PL_sv_undef &&
3736 GvCVu(gv) && GvIMPORTED_CV(gv))
3742 tmp = 0; /* overridden by import or by GLOBAL */
3745 && -tmp==KEY_lock /* XXX generalizable kludge */
3747 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3749 tmp = 0; /* any sub overrides "weak" keyword */
3751 else { /* no override */
3755 if (ckWARN(WARN_AMBIGUOUS) && hgv
3756 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3757 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3758 "Ambiguous call resolved as CORE::%s(), %s",
3759 GvENAME(hgv), "qualify as such or use &");
3766 default: /* not a keyword */
3769 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3771 /* Get the rest if it looks like a package qualifier */
3773 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3775 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3778 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3779 *s == '\'' ? "'" : "::");
3783 if (PL_expect == XOPERATOR) {
3784 if (PL_bufptr == PL_linestart) {
3785 CopLINE_dec(PL_curcop);
3786 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3787 CopLINE_inc(PL_curcop);
3790 no_op("Bareword",s);
3793 /* Look for a subroutine with this name in current package,
3794 unless name is "Foo::", in which case Foo is a bearword
3795 (and a package name). */
3798 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3800 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3801 Perl_warner(aTHX_ WARN_BAREWORD,
3802 "Bareword \"%s\" refers to nonexistent package",
3805 PL_tokenbuf[len] = '\0';
3812 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3815 /* if we saw a global override before, get the right name */
3818 sv = newSVpvn("CORE::GLOBAL::",14);
3819 sv_catpv(sv,PL_tokenbuf);
3822 sv = newSVpv(PL_tokenbuf,0);
3824 /* Presume this is going to be a bareword of some sort. */
3827 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3828 yylval.opval->op_private = OPpCONST_BARE;
3830 /* And if "Foo::", then that's what it certainly is. */
3835 /* See if it's the indirect object for a list operator. */
3837 if (PL_oldoldbufptr &&
3838 PL_oldoldbufptr < PL_bufptr &&
3839 (PL_oldoldbufptr == PL_last_lop
3840 || PL_oldoldbufptr == PL_last_uni) &&
3841 /* NO SKIPSPACE BEFORE HERE! */
3842 (PL_expect == XREF ||
3843 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3845 bool immediate_paren = *s == '(';
3847 /* (Now we can afford to cross potential line boundary.) */
3850 /* Two barewords in a row may indicate method call. */
3852 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3855 /* If not a declared subroutine, it's an indirect object. */
3856 /* (But it's an indir obj regardless for sort.) */
3858 if ((PL_last_lop_op == OP_SORT ||
3859 (!immediate_paren && (!gv || !GvCVu(gv)))) &&
3860 (PL_last_lop_op != OP_MAPSTART &&
3861 PL_last_lop_op != OP_GREPSTART))
3863 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3869 PL_expect = XOPERATOR;
3872 /* Is this a word before a => operator? */
3873 if (*s == '=' && s[1] == '>') {
3875 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
3879 /* If followed by a paren, it's certainly a subroutine. */
3882 if (gv && GvCVu(gv)) {
3883 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
3884 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3889 PL_nextval[PL_nexttoke].opval = yylval.opval;
3890 PL_expect = XOPERATOR;
3896 /* If followed by var or block, call it a method (unless sub) */
3898 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3899 PL_last_lop = PL_oldbufptr;
3900 PL_last_lop_op = OP_METHOD;
3904 /* If followed by a bareword, see if it looks like indir obj. */
3906 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3909 /* Not a method, so call it a subroutine (if defined) */
3911 if (gv && GvCVu(gv)) {
3913 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3914 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3915 "Ambiguous use of -%s resolved as -&%s()",
3916 PL_tokenbuf, PL_tokenbuf);
3917 /* Check for a constant sub */
3919 if ((sv = cv_const_sv(cv))) {
3921 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3922 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3923 yylval.opval->op_private = 0;
3927 /* Resolve to GV now. */
3928 op_free(yylval.opval);
3929 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3930 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3931 PL_last_lop = PL_oldbufptr;
3932 PL_last_lop_op = OP_ENTERSUB;
3933 /* Is there a prototype? */
3936 char *proto = SvPV((SV*)cv, len);
3939 if (strEQ(proto, "$"))
3941 if (*proto == '&' && *s == '{') {
3942 sv_setpv(PL_subname,"__ANON__");
3946 PL_nextval[PL_nexttoke].opval = yylval.opval;
3952 /* Call it a bare word */
3954 if (PL_hints & HINT_STRICT_SUBS)
3955 yylval.opval->op_private |= OPpCONST_STRICT;
3958 if (ckWARN(WARN_RESERVED)) {
3959 if (lastchar != '-') {
3960 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
3962 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
3969 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
3970 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3971 "Operator or semicolon missing before %c%s",
3972 lastchar, PL_tokenbuf);
3973 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3974 "Ambiguous use of %c resolved as operator %c",
3975 lastchar, lastchar);
3981 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3982 newSVpv(CopFILE(PL_curcop),0));
3986 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3987 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
3990 case KEY___PACKAGE__:
3991 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3993 ? newSVsv(PL_curstname)
4002 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4003 char *pname = "main";
4004 if (PL_tokenbuf[2] == 'D')
4005 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
4006 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4009 GvIOp(gv) = newIO();
4010 IoIFP(GvIOp(gv)) = PL_rsfp;
4011 #if defined(HAS_FCNTL) && defined(F_SETFD)
4013 int fd = PerlIO_fileno(PL_rsfp);
4014 fcntl(fd,F_SETFD,fd >= 3);
4017 /* Mark this internal pseudo-handle as clean */
4018 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4020 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4021 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4022 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4024 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4025 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4026 /* if the script was opened in binmode, we need to revert
4027 * it to text mode for compatibility; but only iff it has CRs
4028 * XXX this is a questionable hack at best. */
4029 if (PL_bufend-PL_bufptr > 2
4030 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4033 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4034 loc = PerlIO_tell(PL_rsfp);
4035 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4037 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4038 #if defined(__BORLANDC__)
4039 /* XXX see note in do_binmode() */
4040 ((FILE*)PL_rsfp)->flags |= _F_BIN;
4043 PerlIO_seek(PL_rsfp, loc, 0);
4058 if (PL_expect == XSTATE) {
4065 if (*s == ':' && s[1] == ':') {
4068 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4069 if (!(tmp = keyword(PL_tokenbuf, len)))
4070 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4084 LOP(OP_ACCEPT,XTERM);
4090 LOP(OP_ATAN2,XTERM);
4096 LOP(OP_BINMODE,XTERM);
4099 LOP(OP_BLESS,XTERM);
4108 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4125 if (!PL_cryptseen) {
4126 PL_cryptseen = TRUE;
4130 LOP(OP_CRYPT,XTERM);
4133 if (ckWARN(WARN_CHMOD)) {
4134 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4135 if (*d != '0' && isDIGIT(*d))
4136 Perl_warner(aTHX_ WARN_CHMOD,
4137 "chmod() mode argument is missing initial 0");
4139 LOP(OP_CHMOD,XTERM);
4142 LOP(OP_CHOWN,XTERM);
4145 LOP(OP_CONNECT,XTERM);
4161 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4165 PL_hints |= HINT_BLOCK_SCOPE;
4175 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4176 LOP(OP_DBMOPEN,XTERM);
4182 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4189 yylval.ival = CopLINE(PL_curcop);
4203 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4204 UNIBRACK(OP_ENTEREVAL);
4219 case KEY_endhostent:
4225 case KEY_endservent:
4228 case KEY_endprotoent:
4239 yylval.ival = CopLINE(PL_curcop);
4241 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4243 if ((PL_bufend - p) >= 3 &&
4244 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4246 else if ((PL_bufend - p) >= 4 &&
4247 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4250 if (isIDFIRST_lazy_if(p,UTF)) {
4251 p = scan_ident(p, PL_bufend,
4252 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4256 Perl_croak(aTHX_ "Missing $ on loop variable");
4261 LOP(OP_FORMLINE,XTERM);
4267 LOP(OP_FCNTL,XTERM);
4273 LOP(OP_FLOCK,XTERM);
4282 LOP(OP_GREPSTART, XREF);
4285 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4300 case KEY_getpriority:
4301 LOP(OP_GETPRIORITY,XTERM);
4303 case KEY_getprotobyname:
4306 case KEY_getprotobynumber:
4307 LOP(OP_GPBYNUMBER,XTERM);
4309 case KEY_getprotoent:
4321 case KEY_getpeername:
4322 UNI(OP_GETPEERNAME);
4324 case KEY_gethostbyname:
4327 case KEY_gethostbyaddr:
4328 LOP(OP_GHBYADDR,XTERM);
4330 case KEY_gethostent:
4333 case KEY_getnetbyname:
4336 case KEY_getnetbyaddr:
4337 LOP(OP_GNBYADDR,XTERM);
4342 case KEY_getservbyname:
4343 LOP(OP_GSBYNAME,XTERM);
4345 case KEY_getservbyport:
4346 LOP(OP_GSBYPORT,XTERM);
4348 case KEY_getservent:
4351 case KEY_getsockname:
4352 UNI(OP_GETSOCKNAME);
4354 case KEY_getsockopt:
4355 LOP(OP_GSOCKOPT,XTERM);
4377 yylval.ival = CopLINE(PL_curcop);
4381 LOP(OP_INDEX,XTERM);
4387 LOP(OP_IOCTL,XTERM);
4399 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4431 LOP(OP_LISTEN,XTERM);
4440 s = scan_pat(s,OP_MATCH);
4441 TERM(sublex_start());
4444 LOP(OP_MAPSTART, XREF);
4447 LOP(OP_MKDIR,XTERM);
4450 LOP(OP_MSGCTL,XTERM);
4453 LOP(OP_MSGGET,XTERM);
4456 LOP(OP_MSGRCV,XTERM);
4459 LOP(OP_MSGSND,XTERM);
4465 if (isIDFIRST_lazy_if(s,UTF)) {
4466 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4467 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4469 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4470 if (!PL_in_my_stash) {
4473 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4481 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4488 if (PL_expect != XSTATE)
4489 yyerror("\"no\" not allowed in expression");
4490 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4491 s = force_version(s);
4496 if (*s == '(' || (s = skipspace(s), *s == '('))
4503 if (isIDFIRST_lazy_if(s,UTF)) {
4505 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4507 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4508 Perl_warner(aTHX_ WARN_PRECEDENCE,
4509 "Precedence problem: open %.*s should be open(%.*s)",
4515 yylval.ival = OP_OR;
4525 LOP(OP_OPEN_DIR,XTERM);
4528 checkcomma(s,PL_tokenbuf,"filehandle");
4532 checkcomma(s,PL_tokenbuf,"filehandle");
4551 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4555 LOP(OP_PIPE_OP,XTERM);
4558 s = scan_str(s,FALSE,FALSE);
4560 missingterm((char*)0);
4561 yylval.ival = OP_CONST;
4562 TERM(sublex_start());
4568 s = scan_str(s,FALSE,FALSE);
4570 missingterm((char*)0);
4572 if (SvCUR(PL_lex_stuff)) {
4575 d = SvPV_force(PL_lex_stuff, len);
4577 for (; isSPACE(*d) && len; --len, ++d) ;
4580 if (!warned && ckWARN(WARN_QW)) {
4581 for (; !isSPACE(*d) && len; --len, ++d) {
4583 Perl_warner(aTHX_ WARN_QW,
4584 "Possible attempt to separate words with commas");
4587 else if (*d == '#') {
4588 Perl_warner(aTHX_ WARN_QW,
4589 "Possible attempt to put comments in qw() list");
4595 for (; !isSPACE(*d) && len; --len, ++d) ;
4597 words = append_elem(OP_LIST, words,
4598 newSVOP(OP_CONST, 0, tokeq(newSVpvn(b, d-b))));
4602 PL_nextval[PL_nexttoke].opval = words;
4607 SvREFCNT_dec(PL_lex_stuff);
4608 PL_lex_stuff = Nullsv;
4613 s = scan_str(s,FALSE,FALSE);
4615 missingterm((char*)0);
4616 yylval.ival = OP_STRINGIFY;
4617 if (SvIVX(PL_lex_stuff) == '\'')
4618 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4619 TERM(sublex_start());
4622 s = scan_pat(s,OP_QR);
4623 TERM(sublex_start());
4626 s = scan_str(s,FALSE,FALSE);
4628 missingterm((char*)0);
4629 yylval.ival = OP_BACKTICK;
4631 TERM(sublex_start());
4638 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4639 s = force_version(s);
4642 *PL_tokenbuf = '\0';
4643 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4644 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4645 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4647 yyerror("<> should be quotes");
4655 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4659 LOP(OP_RENAME,XTERM);
4668 LOP(OP_RINDEX,XTERM);
4691 LOP(OP_REVERSE,XTERM);
4702 TERM(sublex_start());
4704 TOKEN(1); /* force error */
4713 LOP(OP_SELECT,XTERM);
4719 LOP(OP_SEMCTL,XTERM);
4722 LOP(OP_SEMGET,XTERM);
4725 LOP(OP_SEMOP,XTERM);
4731 LOP(OP_SETPGRP,XTERM);
4733 case KEY_setpriority:
4734 LOP(OP_SETPRIORITY,XTERM);
4736 case KEY_sethostent:
4742 case KEY_setservent:
4745 case KEY_setprotoent:
4755 LOP(OP_SEEKDIR,XTERM);
4757 case KEY_setsockopt:
4758 LOP(OP_SSOCKOPT,XTERM);
4764 LOP(OP_SHMCTL,XTERM);
4767 LOP(OP_SHMGET,XTERM);
4770 LOP(OP_SHMREAD,XTERM);
4773 LOP(OP_SHMWRITE,XTERM);
4776 LOP(OP_SHUTDOWN,XTERM);
4785 LOP(OP_SOCKET,XTERM);
4787 case KEY_socketpair:
4788 LOP(OP_SOCKPAIR,XTERM);
4791 checkcomma(s,PL_tokenbuf,"subroutine name");
4793 if (*s == ';' || *s == ')') /* probably a close */
4794 Perl_croak(aTHX_ "sort is now a reserved word");
4796 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4800 LOP(OP_SPLIT,XTERM);
4803 LOP(OP_SPRINTF,XTERM);
4806 LOP(OP_SPLICE,XTERM);
4821 LOP(OP_SUBSTR,XTERM);
4827 char tmpbuf[sizeof PL_tokenbuf];
4829 expectation attrful;
4830 bool have_name, have_proto;
4835 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4836 (*s == ':' && s[1] == ':'))
4839 attrful = XATTRBLOCK;
4840 /* remember buffer pos'n for later force_word */
4841 tboffset = s - PL_oldbufptr;
4842 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4843 if (strchr(tmpbuf, ':'))
4844 sv_setpv(PL_subname, tmpbuf);
4846 sv_setsv(PL_subname,PL_curstname);
4847 sv_catpvn(PL_subname,"::",2);
4848 sv_catpvn(PL_subname,tmpbuf,len);
4855 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4856 PL_expect = XTERMBLOCK;
4857 attrful = XATTRTERM;
4858 sv_setpv(PL_subname,"?");
4862 if (key == KEY_format) {
4864 PL_lex_formbrack = PL_lex_brackets + 1;
4866 (void) force_word(PL_oldbufptr + tboffset, WORD,
4871 /* Look for a prototype */
4875 s = scan_str(s,FALSE,FALSE);
4878 SvREFCNT_dec(PL_lex_stuff);
4879 PL_lex_stuff = Nullsv;
4880 Perl_croak(aTHX_ "Prototype not terminated");
4883 d = SvPVX(PL_lex_stuff);
4885 for (p = d; *p; ++p) {
4890 SvCUR(PL_lex_stuff) = tmp;
4898 if (*s == ':' && s[1] != ':')
4899 PL_expect = attrful;
4902 PL_nextval[PL_nexttoke].opval =
4903 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4904 PL_lex_stuff = Nullsv;
4908 sv_setpv(PL_subname,"__ANON__");
4911 (void) force_word(PL_oldbufptr + tboffset, WORD,
4920 LOP(OP_SYSTEM,XREF);
4923 LOP(OP_SYMLINK,XTERM);
4926 LOP(OP_SYSCALL,XTERM);
4929 LOP(OP_SYSOPEN,XTERM);
4932 LOP(OP_SYSSEEK,XTERM);
4935 LOP(OP_SYSREAD,XTERM);
4938 LOP(OP_SYSWRITE,XTERM);
4942 TERM(sublex_start());
4963 LOP(OP_TRUNCATE,XTERM);
4975 yylval.ival = CopLINE(PL_curcop);
4979 yylval.ival = CopLINE(PL_curcop);
4983 LOP(OP_UNLINK,XTERM);
4989 LOP(OP_UNPACK,XTERM);
4992 LOP(OP_UTIME,XTERM);
4995 if (ckWARN(WARN_UMASK)) {
4996 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4997 if (*d != '0' && isDIGIT(*d))
4998 Perl_warner(aTHX_ WARN_UMASK,
4999 "umask: argument is missing initial 0");
5004 LOP(OP_UNSHIFT,XTERM);
5007 if (PL_expect != XSTATE)
5008 yyerror("\"use\" not allowed in expression");
5010 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
5011 s = force_version(s);
5012 if (*s == ';' || (s = skipspace(s), *s == ';')) {
5013 PL_nextval[PL_nexttoke].opval = Nullop;
5018 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5019 s = force_version(s);
5031 yylval.ival = CopLINE(PL_curcop);
5035 PL_hints |= HINT_BLOCK_SCOPE;
5042 LOP(OP_WAITPID,XTERM);
5050 static char ctl_l[2];
5052 if (ctl_l[0] == '\0')
5053 ctl_l[0] = toCTRL('L');
5054 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5057 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5062 if (PL_expect == XOPERATOR)
5068 yylval.ival = OP_XOR;
5073 TERM(sublex_start());
5078 #pragma segment Main
5082 Perl_keyword(pTHX_ register char *d, I32 len)
5087 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5088 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5089 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5090 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5091 if (strEQ(d,"__END__")) return KEY___END__;
5095 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5100 if (strEQ(d,"and")) return -KEY_and;
5101 if (strEQ(d,"abs")) return -KEY_abs;
5104 if (strEQ(d,"alarm")) return -KEY_alarm;
5105 if (strEQ(d,"atan2")) return -KEY_atan2;
5108 if (strEQ(d,"accept")) return -KEY_accept;
5113 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5116 if (strEQ(d,"bless")) return -KEY_bless;
5117 if (strEQ(d,"bind")) return -KEY_bind;
5118 if (strEQ(d,"binmode")) return -KEY_binmode;
5121 if (strEQ(d,"CORE")) return -KEY_CORE;
5122 if (strEQ(d,"CHECK")) return KEY_CHECK;
5127 if (strEQ(d,"cmp")) return -KEY_cmp;
5128 if (strEQ(d,"chr")) return -KEY_chr;
5129 if (strEQ(d,"cos")) return -KEY_cos;
5132 if (strEQ(d,"chop")) return -KEY_chop;
5135 if (strEQ(d,"close")) return -KEY_close;
5136 if (strEQ(d,"chdir")) return -KEY_chdir;
5137 if (strEQ(d,"chomp")) return -KEY_chomp;
5138 if (strEQ(d,"chmod")) return -KEY_chmod;
5139 if (strEQ(d,"chown")) return -KEY_chown;
5140 if (strEQ(d,"crypt")) return -KEY_crypt;
5143 if (strEQ(d,"chroot")) return -KEY_chroot;
5144 if (strEQ(d,"caller")) return -KEY_caller;
5147 if (strEQ(d,"connect")) return -KEY_connect;
5150 if (strEQ(d,"closedir")) return -KEY_closedir;
5151 if (strEQ(d,"continue")) return -KEY_continue;
5156 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5161 if (strEQ(d,"do")) return KEY_do;
5164 if (strEQ(d,"die")) return -KEY_die;
5167 if (strEQ(d,"dump")) return -KEY_dump;
5170 if (strEQ(d,"delete")) return KEY_delete;
5173 if (strEQ(d,"defined")) return KEY_defined;
5174 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5177 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5182 if (strEQ(d,"END")) return KEY_END;
5187 if (strEQ(d,"eq")) return -KEY_eq;
5190 if (strEQ(d,"eof")) return -KEY_eof;
5191 if (strEQ(d,"exp")) return -KEY_exp;
5194 if (strEQ(d,"else")) return KEY_else;
5195 if (strEQ(d,"exit")) return -KEY_exit;
5196 if (strEQ(d,"eval")) return KEY_eval;
5197 if (strEQ(d,"exec")) return -KEY_exec;
5198 if (strEQ(d,"each")) return KEY_each;
5201 if (strEQ(d,"elsif")) return KEY_elsif;
5204 if (strEQ(d,"exists")) return KEY_exists;
5205 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5208 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5209 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5212 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5215 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5216 if (strEQ(d,"endservent")) return -KEY_endservent;
5219 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5226 if (strEQ(d,"for")) return KEY_for;
5229 if (strEQ(d,"fork")) return -KEY_fork;
5232 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5233 if (strEQ(d,"flock")) return -KEY_flock;
5236 if (strEQ(d,"format")) return KEY_format;
5237 if (strEQ(d,"fileno")) return -KEY_fileno;
5240 if (strEQ(d,"foreach")) return KEY_foreach;
5243 if (strEQ(d,"formline")) return -KEY_formline;
5248 if (strnEQ(d,"get",3)) {
5253 if (strEQ(d,"ppid")) return -KEY_getppid;
5254 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5257 if (strEQ(d,"pwent")) return -KEY_getpwent;
5258 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5259 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5262 if (strEQ(d,"peername")) return -KEY_getpeername;
5263 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5264 if (strEQ(d,"priority")) return -KEY_getpriority;
5267 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5270 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5274 else if (*d == 'h') {
5275 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5276 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5277 if (strEQ(d,"hostent")) return -KEY_gethostent;
5279 else if (*d == 'n') {
5280 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5281 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5282 if (strEQ(d,"netent")) return -KEY_getnetent;
5284 else if (*d == 's') {
5285 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5286 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5287 if (strEQ(d,"servent")) return -KEY_getservent;
5288 if (strEQ(d,"sockname")) return -KEY_getsockname;
5289 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5291 else if (*d == 'g') {
5292 if (strEQ(d,"grent")) return -KEY_getgrent;
5293 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5294 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5296 else if (*d == 'l') {
5297 if (strEQ(d,"login")) return -KEY_getlogin;
5299 else if (strEQ(d,"c")) return -KEY_getc;
5304 if (strEQ(d,"gt")) return -KEY_gt;
5305 if (strEQ(d,"ge")) return -KEY_ge;
5308 if (strEQ(d,"grep")) return KEY_grep;
5309 if (strEQ(d,"goto")) return KEY_goto;
5310 if (strEQ(d,"glob")) return KEY_glob;
5313 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5318 if (strEQ(d,"hex")) return -KEY_hex;
5321 if (strEQ(d,"INIT")) return KEY_INIT;
5326 if (strEQ(d,"if")) return KEY_if;
5329 if (strEQ(d,"int")) return -KEY_int;
5332 if (strEQ(d,"index")) return -KEY_index;
5333 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5338 if (strEQ(d,"join")) return -KEY_join;
5342 if (strEQ(d,"keys")) return KEY_keys;
5343 if (strEQ(d,"kill")) return -KEY_kill;
5349 if (strEQ(d,"lt")) return -KEY_lt;
5350 if (strEQ(d,"le")) return -KEY_le;
5351 if (strEQ(d,"lc")) return -KEY_lc;
5354 if (strEQ(d,"log")) return -KEY_log;
5357 if (strEQ(d,"last")) return KEY_last;
5358 if (strEQ(d,"link")) return -KEY_link;
5359 if (strEQ(d,"lock")) return -KEY_lock;
5362 if (strEQ(d,"local")) return KEY_local;
5363 if (strEQ(d,"lstat")) return -KEY_lstat;
5366 if (strEQ(d,"length")) return -KEY_length;
5367 if (strEQ(d,"listen")) return -KEY_listen;
5370 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5373 if (strEQ(d,"localtime")) return -KEY_localtime;
5379 case 1: return KEY_m;
5381 if (strEQ(d,"my")) return KEY_my;
5384 if (strEQ(d,"map")) return KEY_map;
5387 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5390 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5391 if (strEQ(d,"msgget")) return -KEY_msgget;
5392 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5393 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5398 if (strEQ(d,"next")) return KEY_next;
5399 if (strEQ(d,"ne")) return -KEY_ne;
5400 if (strEQ(d,"not")) return -KEY_not;
5401 if (strEQ(d,"no")) return KEY_no;
5406 if (strEQ(d,"or")) return -KEY_or;
5409 if (strEQ(d,"ord")) return -KEY_ord;
5410 if (strEQ(d,"oct")) return -KEY_oct;
5411 if (strEQ(d,"our")) return KEY_our;
5414 if (strEQ(d,"open")) return -KEY_open;
5417 if (strEQ(d,"opendir")) return -KEY_opendir;
5424 if (strEQ(d,"pop")) return KEY_pop;
5425 if (strEQ(d,"pos")) return KEY_pos;
5428 if (strEQ(d,"push")) return KEY_push;
5429 if (strEQ(d,"pack")) return -KEY_pack;
5430 if (strEQ(d,"pipe")) return -KEY_pipe;
5433 if (strEQ(d,"print")) return KEY_print;
5436 if (strEQ(d,"printf")) return KEY_printf;
5439 if (strEQ(d,"package")) return KEY_package;
5442 if (strEQ(d,"prototype")) return KEY_prototype;
5447 if (strEQ(d,"q")) return KEY_q;
5448 if (strEQ(d,"qr")) return KEY_qr;
5449 if (strEQ(d,"qq")) return KEY_qq;
5450 if (strEQ(d,"qw")) return KEY_qw;
5451 if (strEQ(d,"qx")) return KEY_qx;
5453 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5458 if (strEQ(d,"ref")) return -KEY_ref;
5461 if (strEQ(d,"read")) return -KEY_read;
5462 if (strEQ(d,"rand")) return -KEY_rand;
5463 if (strEQ(d,"recv")) return -KEY_recv;
5464 if (strEQ(d,"redo")) return KEY_redo;
5467 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5468 if (strEQ(d,"reset")) return -KEY_reset;
5471 if (strEQ(d,"return")) return KEY_return;
5472 if (strEQ(d,"rename")) return -KEY_rename;
5473 if (strEQ(d,"rindex")) return -KEY_rindex;
5476 if (strEQ(d,"require")) return -KEY_require;
5477 if (strEQ(d,"reverse")) return -KEY_reverse;
5478 if (strEQ(d,"readdir")) return -KEY_readdir;
5481 if (strEQ(d,"readlink")) return -KEY_readlink;
5482 if (strEQ(d,"readline")) return -KEY_readline;
5483 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5486 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5492 case 0: return KEY_s;
5494 if (strEQ(d,"scalar")) return KEY_scalar;
5499 if (strEQ(d,"seek")) return -KEY_seek;
5500 if (strEQ(d,"send")) return -KEY_send;
5503 if (strEQ(d,"semop")) return -KEY_semop;
5506 if (strEQ(d,"select")) return -KEY_select;
5507 if (strEQ(d,"semctl")) return -KEY_semctl;
5508 if (strEQ(d,"semget")) return -KEY_semget;
5511 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5512 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5515 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5516 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5519 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5522 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5523 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5524 if (strEQ(d,"setservent")) return -KEY_setservent;
5527 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5528 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5535 if (strEQ(d,"shift")) return KEY_shift;
5538 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5539 if (strEQ(d,"shmget")) return -KEY_shmget;
5542 if (strEQ(d,"shmread")) return -KEY_shmread;
5545 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5546 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5551 if (strEQ(d,"sin")) return -KEY_sin;
5554 if (strEQ(d,"sleep")) return -KEY_sleep;
5557 if (strEQ(d,"sort")) return KEY_sort;
5558 if (strEQ(d,"socket")) return -KEY_socket;
5559 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5562 if (strEQ(d,"split")) return KEY_split;
5563 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5564 if (strEQ(d,"splice")) return KEY_splice;
5567 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5570 if (strEQ(d,"srand")) return -KEY_srand;
5573 if (strEQ(d,"stat")) return -KEY_stat;
5574 if (strEQ(d,"study")) return KEY_study;
5577 if (strEQ(d,"substr")) return -KEY_substr;
5578 if (strEQ(d,"sub")) return KEY_sub;
5583 if (strEQ(d,"system")) return -KEY_system;
5586 if (strEQ(d,"symlink")) return -KEY_symlink;
5587 if (strEQ(d,"syscall")) return -KEY_syscall;
5588 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5589 if (strEQ(d,"sysread")) return -KEY_sysread;
5590 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5593 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5602 if (strEQ(d,"tr")) return KEY_tr;
5605 if (strEQ(d,"tie")) return KEY_tie;
5608 if (strEQ(d,"tell")) return -KEY_tell;
5609 if (strEQ(d,"tied")) return KEY_tied;
5610 if (strEQ(d,"time")) return -KEY_time;
5613 if (strEQ(d,"times")) return -KEY_times;
5616 if (strEQ(d,"telldir")) return -KEY_telldir;
5619 if (strEQ(d,"truncate")) return -KEY_truncate;
5626 if (strEQ(d,"uc")) return -KEY_uc;
5629 if (strEQ(d,"use")) return KEY_use;
5632 if (strEQ(d,"undef")) return KEY_undef;
5633 if (strEQ(d,"until")) return KEY_until;
5634 if (strEQ(d,"untie")) return KEY_untie;
5635 if (strEQ(d,"utime")) return -KEY_utime;
5636 if (strEQ(d,"umask")) return -KEY_umask;
5639 if (strEQ(d,"unless")) return KEY_unless;
5640 if (strEQ(d,"unpack")) return -KEY_unpack;
5641 if (strEQ(d,"unlink")) return -KEY_unlink;
5644 if (strEQ(d,"unshift")) return KEY_unshift;
5645 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5650 if (strEQ(d,"values")) return -KEY_values;
5651 if (strEQ(d,"vec")) return -KEY_vec;
5656 if (strEQ(d,"warn")) return -KEY_warn;
5657 if (strEQ(d,"wait")) return -KEY_wait;
5660 if (strEQ(d,"while")) return KEY_while;
5661 if (strEQ(d,"write")) return -KEY_write;
5664 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5667 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5672 if (len == 1) return -KEY_x;
5673 if (strEQ(d,"xor")) return -KEY_xor;
5676 if (len == 1) return KEY_y;
5685 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5689 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5690 dTHR; /* only for ckWARN */
5691 if (ckWARN(WARN_SYNTAX)) {
5693 for (w = s+2; *w && level; w++) {
5700 for (; *w && isSPACE(*w); w++) ;
5701 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5702 Perl_warner(aTHX_ WARN_SYNTAX,
5703 "%s (...) interpreted as function",name);
5706 while (s < PL_bufend && isSPACE(*s))
5710 while (s < PL_bufend && isSPACE(*s))
5712 if (isIDFIRST_lazy_if(s,UTF)) {
5714 while (isALNUM_lazy_if(s,UTF))
5716 while (s < PL_bufend && isSPACE(*s))
5721 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5725 Perl_croak(aTHX_ "No comma allowed after %s", what);
5730 /* Either returns sv, or mortalizes sv and returns a new SV*.
5731 Best used as sv=new_constant(..., sv, ...).
5732 If s, pv are NULL, calls subroutine with one argument,
5733 and type is used with error messages only. */
5736 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5740 HV *table = GvHV(PL_hintgv); /* ^H */
5744 const char *why1, *why2, *why3;
5746 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
5749 why1 = "%^H is not consistent";
5750 why2 = strEQ(key,"charnames")
5751 ? " (missing \"use charnames ...\"?)"
5755 msg = Perl_newSVpvf(aTHX_ "constant(%s): %s%s%s",
5756 (type ? type: "undef"), why1, why2, why3);
5757 yyerror(SvPVX(msg));
5761 cvp = hv_fetch(table, key, strlen(key), FALSE);
5762 if (!cvp || !SvOK(*cvp)) {
5765 why3 = "} is not defined";
5768 sv_2mortal(sv); /* Parent created it permanently */
5771 pv = sv_2mortal(newSVpvn(s, len));
5773 typesv = sv_2mortal(newSVpv(type, 0));
5775 typesv = &PL_sv_undef;
5777 PUSHSTACKi(PERLSI_OVERLOAD);
5789 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5793 /* Check the eval first */
5794 if (!PL_in_eval && SvTRUE(ERRSV)) {
5796 sv_catpv(ERRSV, "Propagated");
5797 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5799 res = SvREFCNT_inc(sv);
5803 (void)SvREFCNT_inc(res);
5812 why1 = "Call to &{$^H{";
5814 why3 = "}} did not return a defined value";
5823 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5825 register char *d = dest;
5826 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5829 Perl_croak(aTHX_ ident_too_long);
5830 if (isALNUM(*s)) /* UTF handled below */
5832 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
5837 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5841 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5842 char *t = s + UTF8SKIP(s);
5843 while (*t & 0x80 && is_utf8_mark((U8*)t))
5845 if (d + (t - s) > e)
5846 Perl_croak(aTHX_ ident_too_long);
5847 Copy(s, d, t - s, char);
5860 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5870 e = d + destlen - 3; /* two-character token, ending NUL */
5872 while (isDIGIT(*s)) {
5874 Perl_croak(aTHX_ ident_too_long);
5881 Perl_croak(aTHX_ ident_too_long);
5882 if (isALNUM(*s)) /* UTF handled below */
5884 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
5889 else if (*s == ':' && s[1] == ':') {
5893 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5894 char *t = s + UTF8SKIP(s);
5895 while (*t & 0x80 && is_utf8_mark((U8*)t))
5897 if (d + (t - s) > e)
5898 Perl_croak(aTHX_ ident_too_long);
5899 Copy(s, d, t - s, char);
5910 if (PL_lex_state != LEX_NORMAL)
5911 PL_lex_state = LEX_INTERPENDMAYBE;
5914 if (*s == '$' && s[1] &&
5915 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
5928 if (*d == '^' && *s && isCONTROLVAR(*s)) {
5933 if (isSPACE(s[-1])) {
5936 if (!SPACE_OR_TAB(ch)) {
5942 if (isIDFIRST_lazy_if(d,UTF)) {
5946 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
5948 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
5951 Copy(s, d, e - s, char);
5956 while ((isALNUM(*s) || *s == ':') && d < e)
5959 Perl_croak(aTHX_ ident_too_long);
5962 while (s < send && SPACE_OR_TAB(*s)) s++;
5963 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
5964 dTHR; /* only for ckWARN */
5965 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
5966 const char *brack = *s == '[' ? "[...]" : "{...}";
5967 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5968 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
5969 funny, dest, brack, funny, dest, brack);
5972 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
5976 /* Handle extended ${^Foo} variables
5977 * 1999-02-27 mjd-perl-patch@plover.com */
5978 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
5982 while (isALNUM(*s) && d < e) {
5986 Perl_croak(aTHX_ ident_too_long);
5991 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
5992 PL_lex_state = LEX_INTERPEND;
5995 if (PL_lex_state == LEX_NORMAL) {
5996 dTHR; /* only for ckWARN */
5997 if (ckWARN(WARN_AMBIGUOUS) &&
5998 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
6000 Perl_warner(aTHX_ WARN_AMBIGUOUS,
6001 "Ambiguous use of %c{%s} resolved to %c%s",
6002 funny, dest, funny, dest);
6007 s = bracket; /* let the parser handle it */
6011 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
6012 PL_lex_state = LEX_INTERPEND;
6017 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
6022 *pmfl |= PMf_GLOBAL;
6024 *pmfl |= PMf_CONTINUE;
6028 *pmfl |= PMf_MULTILINE;
6030 *pmfl |= PMf_SINGLELINE;
6032 *pmfl |= PMf_EXTENDED;
6036 S_scan_pat(pTHX_ char *start, I32 type)
6041 s = scan_str(start,FALSE,FALSE);
6044 SvREFCNT_dec(PL_lex_stuff);
6045 PL_lex_stuff = Nullsv;
6046 Perl_croak(aTHX_ "Search pattern not terminated");
6049 pm = (PMOP*)newPMOP(type, 0);
6050 if (PL_multi_open == '?')
6051 pm->op_pmflags |= PMf_ONCE;
6053 while (*s && strchr("iomsx", *s))
6054 pmflag(&pm->op_pmflags,*s++);
6057 while (*s && strchr("iogcmsx", *s))
6058 pmflag(&pm->op_pmflags,*s++);
6060 pm->op_pmpermflags = pm->op_pmflags;
6062 PL_lex_op = (OP*)pm;
6063 yylval.ival = OP_MATCH;
6068 S_scan_subst(pTHX_ char *start)
6075 yylval.ival = OP_NULL;
6077 s = scan_str(start,FALSE,FALSE);
6081 SvREFCNT_dec(PL_lex_stuff);
6082 PL_lex_stuff = Nullsv;
6083 Perl_croak(aTHX_ "Substitution pattern not terminated");
6086 if (s[-1] == PL_multi_open)
6089 first_start = PL_multi_start;
6090 s = scan_str(s,FALSE,FALSE);
6093 SvREFCNT_dec(PL_lex_stuff);
6094 PL_lex_stuff = Nullsv;
6096 SvREFCNT_dec(PL_lex_repl);
6097 PL_lex_repl = Nullsv;
6098 Perl_croak(aTHX_ "Substitution replacement not terminated");
6100 PL_multi_start = first_start; /* so whole substitution is taken together */
6102 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6108 else if (strchr("iogcmsx", *s))
6109 pmflag(&pm->op_pmflags,*s++);
6116 PL_sublex_info.super_bufptr = s;
6117 PL_sublex_info.super_bufend = PL_bufend;
6119 pm->op_pmflags |= PMf_EVAL;
6120 repl = newSVpvn("",0);
6122 sv_catpv(repl, es ? "eval " : "do ");
6123 sv_catpvn(repl, "{ ", 2);
6124 sv_catsv(repl, PL_lex_repl);
6125 sv_catpvn(repl, " };", 2);
6127 SvREFCNT_dec(PL_lex_repl);
6131 pm->op_pmpermflags = pm->op_pmflags;
6132 PL_lex_op = (OP*)pm;
6133 yylval.ival = OP_SUBST;
6138 S_scan_trans(pTHX_ char *start)
6149 yylval.ival = OP_NULL;
6151 s = scan_str(start,FALSE,FALSE);
6154 SvREFCNT_dec(PL_lex_stuff);
6155 PL_lex_stuff = Nullsv;
6156 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6158 if (s[-1] == PL_multi_open)
6161 s = scan_str(s,FALSE,FALSE);
6164 SvREFCNT_dec(PL_lex_stuff);
6165 PL_lex_stuff = Nullsv;
6167 SvREFCNT_dec(PL_lex_repl);
6168 PL_lex_repl = Nullsv;
6169 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6172 New(803,tbl,256,short);
6173 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6175 complement = del = squash = 0;
6176 while (strchr("cds", *s)) {
6178 complement = OPpTRANS_COMPLEMENT;
6180 del = OPpTRANS_DELETE;
6182 squash = OPpTRANS_SQUASH;
6185 o->op_private = del|squash|complement;
6188 yylval.ival = OP_TRANS;
6193 S_scan_heredoc(pTHX_ register char *s)
6197 I32 op_type = OP_SCALAR;
6204 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6208 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6211 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6212 if (*peek && strchr("`'\"",*peek)) {
6215 s = delimcpy(d, e, s, PL_bufend, term, &len);
6225 if (!isALNUM_lazy_if(s,UTF))
6226 deprecate("bare << to mean <<\"\"");
6227 for (; isALNUM_lazy_if(s,UTF); s++) {
6232 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6233 Perl_croak(aTHX_ "Delimiter for here document is too long");
6236 len = d - PL_tokenbuf;
6237 #ifndef PERL_STRICT_CR
6238 d = strchr(s, '\r');
6242 while (s < PL_bufend) {
6248 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6257 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6262 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6263 herewas = newSVpvn(s,PL_bufend-s);
6265 s--, herewas = newSVpvn(s,d-s);
6266 s += SvCUR(herewas);
6268 tmpstr = NEWSV(87,79);
6269 sv_upgrade(tmpstr, SVt_PVIV);
6274 else if (term == '`') {
6275 op_type = OP_BACKTICK;
6276 SvIVX(tmpstr) = '\\';
6280 PL_multi_start = CopLINE(PL_curcop);
6281 PL_multi_open = PL_multi_close = '<';
6282 term = *PL_tokenbuf;
6283 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6284 char *bufptr = PL_sublex_info.super_bufptr;
6285 char *bufend = PL_sublex_info.super_bufend;
6286 char *olds = s - SvCUR(herewas);
6287 s = strchr(bufptr, '\n');
6291 while (s < bufend &&
6292 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6294 CopLINE_inc(PL_curcop);
6297 CopLINE_set(PL_curcop, PL_multi_start);
6298 missingterm(PL_tokenbuf);
6300 sv_setpvn(herewas,bufptr,d-bufptr+1);
6301 sv_setpvn(tmpstr,d+1,s-d);
6303 sv_catpvn(herewas,s,bufend-s);
6304 (void)strcpy(bufptr,SvPVX(herewas));
6311 while (s < PL_bufend &&
6312 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6314 CopLINE_inc(PL_curcop);
6316 if (s >= PL_bufend) {
6317 CopLINE_set(PL_curcop, PL_multi_start);
6318 missingterm(PL_tokenbuf);
6320 sv_setpvn(tmpstr,d+1,s-d);
6322 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6324 sv_catpvn(herewas,s,PL_bufend-s);
6325 sv_setsv(PL_linestr,herewas);
6326 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6327 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6330 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6331 while (s >= PL_bufend) { /* multiple line string? */
6333 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6334 CopLINE_set(PL_curcop, PL_multi_start);
6335 missingterm(PL_tokenbuf);
6337 CopLINE_inc(PL_curcop);
6338 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6339 #ifndef PERL_STRICT_CR
6340 if (PL_bufend - PL_linestart >= 2) {
6341 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6342 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6344 PL_bufend[-2] = '\n';
6346 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6348 else if (PL_bufend[-1] == '\r')
6349 PL_bufend[-1] = '\n';
6351 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6352 PL_bufend[-1] = '\n';
6354 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6355 SV *sv = NEWSV(88,0);
6357 sv_upgrade(sv, SVt_PVMG);
6358 sv_setsv(sv,PL_linestr);
6359 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6361 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6364 sv_catsv(PL_linestr,herewas);
6365 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6369 sv_catsv(tmpstr,PL_linestr);
6374 PL_multi_end = CopLINE(PL_curcop);
6375 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6376 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6377 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6379 SvREFCNT_dec(herewas);
6380 PL_lex_stuff = tmpstr;
6381 yylval.ival = op_type;
6386 takes: current position in input buffer
6387 returns: new position in input buffer
6388 side-effects: yylval and lex_op are set.
6393 <FH> read from filehandle
6394 <pkg::FH> read from package qualified filehandle
6395 <pkg'FH> read from package qualified filehandle
6396 <$fh> read from filehandle in $fh
6402 S_scan_inputsymbol(pTHX_ char *start)
6404 register char *s = start; /* current position in buffer */
6410 d = PL_tokenbuf; /* start of temp holding space */
6411 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6412 end = strchr(s, '\n');
6415 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6417 /* die if we didn't have space for the contents of the <>,
6418 or if it didn't end, or if we see a newline
6421 if (len >= sizeof PL_tokenbuf)
6422 Perl_croak(aTHX_ "Excessively long <> operator");
6424 Perl_croak(aTHX_ "Unterminated <> operator");
6429 Remember, only scalar variables are interpreted as filehandles by
6430 this code. Anything more complex (e.g., <$fh{$num}>) will be
6431 treated as a glob() call.
6432 This code makes use of the fact that except for the $ at the front,
6433 a scalar variable and a filehandle look the same.
6435 if (*d == '$' && d[1]) d++;
6437 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6438 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6441 /* If we've tried to read what we allow filehandles to look like, and
6442 there's still text left, then it must be a glob() and not a getline.
6443 Use scan_str to pull out the stuff between the <> and treat it
6444 as nothing more than a string.
6447 if (d - PL_tokenbuf != len) {
6448 yylval.ival = OP_GLOB;
6450 s = scan_str(start,FALSE,FALSE);
6452 Perl_croak(aTHX_ "Glob not terminated");
6456 /* we're in a filehandle read situation */
6459 /* turn <> into <ARGV> */
6461 (void)strcpy(d,"ARGV");
6463 /* if <$fh>, create the ops to turn the variable into a
6469 /* try to find it in the pad for this block, otherwise find
6470 add symbol table ops
6472 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6473 OP *o = newOP(OP_PADSV, 0);
6475 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6478 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6479 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6480 newUNOP(OP_RV2SV, 0,
6481 newGVOP(OP_GV, 0, gv)));
6483 PL_lex_op->op_flags |= OPf_SPECIAL;
6484 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6485 yylval.ival = OP_NULL;
6488 /* If it's none of the above, it must be a literal filehandle
6489 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6491 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6492 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6493 yylval.ival = OP_NULL;
6502 takes: start position in buffer
6503 keep_quoted preserve \ on the embedded delimiter(s)
6504 keep_delims preserve the delimiters around the string
6505 returns: position to continue reading from buffer
6506 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6507 updates the read buffer.
6509 This subroutine pulls a string out of the input. It is called for:
6510 q single quotes q(literal text)
6511 ' single quotes 'literal text'
6512 qq double quotes qq(interpolate $here please)
6513 " double quotes "interpolate $here please"
6514 qx backticks qx(/bin/ls -l)
6515 ` backticks `/bin/ls -l`
6516 qw quote words @EXPORT_OK = qw( func() $spam )
6517 m// regexp match m/this/
6518 s/// regexp substitute s/this/that/
6519 tr/// string transliterate tr/this/that/
6520 y/// string transliterate y/this/that/
6521 ($*@) sub prototypes sub foo ($)
6522 (stuff) sub attr parameters sub foo : attr(stuff)
6523 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6525 In most of these cases (all but <>, patterns and transliterate)
6526 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6527 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6528 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6531 It skips whitespace before the string starts, and treats the first
6532 character as the delimiter. If the delimiter is one of ([{< then
6533 the corresponding "close" character )]}> is used as the closing
6534 delimiter. It allows quoting of delimiters, and if the string has
6535 balanced delimiters ([{<>}]) it allows nesting.
6537 The lexer always reads these strings into lex_stuff, except in the
6538 case of the operators which take *two* arguments (s/// and tr///)
6539 when it checks to see if lex_stuff is full (presumably with the 1st
6540 arg to s or tr) and if so puts the string into lex_repl.
6545 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6548 SV *sv; /* scalar value: string */
6549 char *tmps; /* temp string, used for delimiter matching */
6550 register char *s = start; /* current position in the buffer */
6551 register char term; /* terminating character */
6552 register char *to; /* current position in the sv's data */
6553 I32 brackets = 1; /* bracket nesting level */
6554 bool has_utf = FALSE; /* is there any utf8 content? */
6556 /* skip space before the delimiter */
6560 /* mark where we are, in case we need to report errors */
6563 /* after skipping whitespace, the next character is the terminator */
6565 if ((term & 0x80) && UTF)
6568 /* mark where we are */
6569 PL_multi_start = CopLINE(PL_curcop);
6570 PL_multi_open = term;
6572 /* find corresponding closing delimiter */
6573 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6575 PL_multi_close = term;
6577 /* create a new SV to hold the contents. 87 is leak category, I'm
6578 assuming. 79 is the SV's initial length. What a random number. */
6580 sv_upgrade(sv, SVt_PVIV);
6582 (void)SvPOK_only(sv); /* validate pointer */
6584 /* move past delimiter and try to read a complete string */
6586 sv_catpvn(sv, s, 1);
6589 /* extend sv if need be */
6590 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6591 /* set 'to' to the next character in the sv's string */
6592 to = SvPVX(sv)+SvCUR(sv);
6594 /* if open delimiter is the close delimiter read unbridle */
6595 if (PL_multi_open == PL_multi_close) {
6596 for (; s < PL_bufend; s++,to++) {
6597 /* embedded newlines increment the current line number */
6598 if (*s == '\n' && !PL_rsfp)
6599 CopLINE_inc(PL_curcop);
6600 /* handle quoted delimiters */
6601 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6602 if (!keep_quoted && s[1] == term)
6604 /* any other quotes are simply copied straight through */
6608 /* terminate when run out of buffer (the for() condition), or
6609 have found the terminator */
6610 else if (*s == term)
6612 else if (!has_utf && (*s & 0x80) && UTF)
6618 /* if the terminator isn't the same as the start character (e.g.,
6619 matched brackets), we have to allow more in the quoting, and
6620 be prepared for nested brackets.
6623 /* read until we run out of string, or we find the terminator */
6624 for (; s < PL_bufend; s++,to++) {
6625 /* embedded newlines increment the line count */
6626 if (*s == '\n' && !PL_rsfp)
6627 CopLINE_inc(PL_curcop);
6628 /* backslashes can escape the open or closing characters */
6629 if (*s == '\\' && s+1 < PL_bufend) {
6631 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6636 /* allow nested opens and closes */
6637 else if (*s == PL_multi_close && --brackets <= 0)
6639 else if (*s == PL_multi_open)
6641 else if (!has_utf && (*s & 0x80) && UTF)
6646 /* terminate the copied string and update the sv's end-of-string */
6648 SvCUR_set(sv, to - SvPVX(sv));
6651 * this next chunk reads more into the buffer if we're not done yet
6655 break; /* handle case where we are done yet :-) */
6657 #ifndef PERL_STRICT_CR
6658 if (to - SvPVX(sv) >= 2) {
6659 if ((to[-2] == '\r' && to[-1] == '\n') ||
6660 (to[-2] == '\n' && to[-1] == '\r'))
6664 SvCUR_set(sv, to - SvPVX(sv));
6666 else if (to[-1] == '\r')
6669 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6673 /* if we're out of file, or a read fails, bail and reset the current
6674 line marker so we can report where the unterminated string began
6677 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6679 CopLINE_set(PL_curcop, PL_multi_start);
6682 /* we read a line, so increment our line counter */
6683 CopLINE_inc(PL_curcop);
6685 /* update debugger info */
6686 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6687 SV *sv = NEWSV(88,0);
6689 sv_upgrade(sv, SVt_PVMG);
6690 sv_setsv(sv,PL_linestr);
6691 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6694 /* having changed the buffer, we must update PL_bufend */
6695 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6698 /* at this point, we have successfully read the delimited string */
6701 sv_catpvn(sv, s, 1);
6704 PL_multi_end = CopLINE(PL_curcop);
6707 /* if we allocated too much space, give some back */
6708 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6709 SvLEN_set(sv, SvCUR(sv) + 1);
6710 Renew(SvPVX(sv), SvLEN(sv), char);
6713 /* decide whether this is the first or second quoted string we've read
6726 takes: pointer to position in buffer
6727 returns: pointer to new position in buffer
6728 side-effects: builds ops for the constant in yylval.op
6730 Read a number in any of the formats that Perl accepts:
6732 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6733 [\d_]+(\.[\d_]*)?[Ee](\d+)
6735 Underbars (_) are allowed in decimal numbers. If -w is on,
6736 underbars before a decimal point must be at three digit intervals.
6738 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6741 If it reads a number without a decimal point or an exponent, it will
6742 try converting the number to an integer and see if it can do so
6743 without loss of precision.
6747 Perl_scan_num(pTHX_ char *start, YYSTYPE* lvalp)
6749 register char *s = start; /* current position in buffer */
6750 register char *d; /* destination in temp buffer */
6751 register char *e; /* end of temp buffer */
6752 NV nv; /* number read, as a double */
6753 SV *sv = Nullsv; /* place to put the converted number */
6754 bool floatit; /* boolean: int or float? */
6755 char *lastub = 0; /* position of last underbar */
6756 static char number_too_long[] = "Number too long";
6758 /* We use the first character to decide what type of number this is */
6762 Perl_croak(aTHX_ "panic: scan_num");
6764 /* if it starts with a 0, it could be an octal number, a decimal in
6765 0.13 disguise, or a hexadecimal number, or a binary number. */
6769 u holds the "number so far"
6770 shift the power of 2 of the base
6771 (hex == 4, octal == 3, binary == 1)
6772 overflowed was the number more than we can hold?
6774 Shift is used when we add a digit. It also serves as an "are
6775 we in octal/hex/binary?" indicator to disallow hex characters
6782 bool overflowed = FALSE;
6783 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6784 static char* bases[5] = { "", "binary", "", "octal",
6786 static char* Bases[5] = { "", "Binary", "", "Octal",
6788 static char *maxima[5] = { "",
6789 "0b11111111111111111111111111111111",
6793 char *base, *Base, *max;
6799 } else if (s[1] == 'b') {
6803 /* check for a decimal in disguise */
6804 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6806 /* so it must be octal */
6810 base = bases[shift];
6811 Base = Bases[shift];
6812 max = maxima[shift];
6814 /* read the rest of the number */
6816 /* x is used in the overflow test,
6817 b is the digit we're adding on. */
6822 /* if we don't mention it, we're done */
6831 /* 8 and 9 are not octal */
6834 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6838 case '2': case '3': case '4':
6839 case '5': case '6': case '7':
6841 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6845 b = *s++ & 15; /* ASCII digit -> value of digit */
6849 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6850 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6851 /* make sure they said 0x */
6856 /* Prepare to put the digit we have onto the end
6857 of the number so far. We check for overflows.
6862 x = u << shift; /* make room for the digit */
6864 if ((x >> shift) != u
6865 && !(PL_hints & HINT_NEW_BINARY)) {
6869 if (ckWARN_d(WARN_OVERFLOW))
6870 Perl_warner(aTHX_ WARN_OVERFLOW,
6871 "Integer overflow in %s number",
6874 u = x | b; /* add the digit to the end */
6877 n *= nvshift[shift];
6878 /* If an NV has not enough bits in its
6879 * mantissa to represent an UV this summing of
6880 * small low-order numbers is a waste of time
6881 * (because the NV cannot preserve the
6882 * low-order bits anyway): we could just
6883 * remember when did we overflow and in the
6884 * end just multiply n by the right
6892 /* if we get here, we had success: make a scalar value from
6899 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6900 Perl_warner(aTHX_ WARN_PORTABLE,
6901 "%s number > %s non-portable",
6908 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6909 Perl_warner(aTHX_ WARN_PORTABLE,
6910 "%s number > %s non-portable",
6915 if (PL_hints & HINT_NEW_BINARY)
6916 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6921 handle decimal numbers.
6922 we're also sent here when we read a 0 as the first digit
6924 case '1': case '2': case '3': case '4': case '5':
6925 case '6': case '7': case '8': case '9': case '.':
6928 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
6931 /* read next group of digits and _ and copy into d */
6932 while (isDIGIT(*s) || *s == '_') {
6933 /* skip underscores, checking for misplaced ones
6937 dTHR; /* only for ckWARN */
6938 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
6939 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6943 /* check for end of fixed-length buffer */
6945 Perl_croak(aTHX_ number_too_long);
6946 /* if we're ok, copy the character */
6951 /* final misplaced underbar check */
6952 if (lastub && s - lastub != 3) {
6954 if (ckWARN(WARN_SYNTAX))
6955 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6958 /* read a decimal portion if there is one. avoid
6959 3..5 being interpreted as the number 3. followed
6962 if (*s == '.' && s[1] != '.') {
6966 /* copy, ignoring underbars, until we run out of
6967 digits. Note: no misplaced underbar checks!
6969 for (; isDIGIT(*s) || *s == '_'; s++) {
6970 /* fixed length buffer check */
6972 Perl_croak(aTHX_ number_too_long);
6976 if (*s == '.' && isDIGIT(s[1])) {
6977 /* oops, it's really a v-string, but without the "v" */
6983 /* read exponent part, if present */
6984 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
6988 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
6989 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
6991 /* allow positive or negative exponent */
6992 if (*s == '+' || *s == '-')
6995 /* read digits of exponent (no underbars :-) */
6996 while (isDIGIT(*s)) {
6998 Perl_croak(aTHX_ number_too_long);
7003 /* terminate the string */
7006 /* make an sv from the string */
7009 #if defined(Strtol) && defined(Strtoul)
7012 strtol/strtoll sets errno to ERANGE if the number is too big
7013 for an integer. We try to do an integer conversion first
7014 if no characters indicating "float" have been found.
7021 if (*PL_tokenbuf == '-')
7022 iv = Strtol(PL_tokenbuf, (char**)NULL, 10);
7024 uv = Strtoul(PL_tokenbuf, (char**)NULL, 10);
7026 floatit = TRUE; /* Probably just too large. */
7027 else if (*PL_tokenbuf == '-')
7029 else if (uv <= IV_MAX)
7030 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7035 nv = Atof(PL_tokenbuf);
7040 No working strtou?ll?.
7042 Unfortunately atol() doesn't do range checks (returning
7043 LONG_MIN/LONG_MAX, and setting errno to ERANGE on overflows)
7044 everywhere [1], so we cannot use use atol() (or atoll()).
7045 If we could, they would be used, as Atol(), very much like
7046 Strtol() and Strtoul() are used above.
7048 [1] XXX Configure test needed to check for atol()
7049 (and atoll()) overflow behaviour XXX
7053 We need to do this the hard way. */
7055 nv = Atof(PL_tokenbuf);
7057 /* See if we can make do with an integer value without loss of
7058 precision. We use U_V to cast to a UV, because some
7059 compilers have issues. Then we try casting it back and see
7060 if it was the same [1]. We only do this if we know we
7061 specifically read an integer. If floatit is true, then we
7062 don't need to do the conversion at all.
7064 [1] Note that this is lossy if our NVs cannot preserve our
7065 UVs. There are metaconfig defines NV_PRESERVES_UV (a boolean)
7066 and NV_PRESERVES_UV_BITS (a number), but in general we really
7067 do hope all such potentially lossy platforms have strtou?ll?
7068 to do a lossless IV/UV conversion.
7070 Maybe could do some tricks with DBL_DIG, LDBL_DIG and
7071 DBL_MANT_DIG and LDBL_MANT_DIG (these are already available
7072 as NV_DIG and NV_MANT_DIG)?
7078 if (!floatit && (NV)uv == nv) {
7080 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7088 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7089 (PL_hints & HINT_NEW_INTEGER) )
7090 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7091 (floatit ? "float" : "integer"),
7095 /* if it starts with a v, it could be a v-string */
7101 while (isDIGIT(*pos) || *pos == '_')
7103 if (!isALPHA(*pos)) {
7105 U8 tmpbuf[UTF8_MAXLEN];
7108 s++; /* get past 'v' */
7111 sv_setpvn(sv, "", 0);
7114 if (*s == '0' && isDIGIT(s[1]))
7115 yyerror("Octal number in vector unsupported");
7118 /* this is atoi() that tolerates underscores */
7121 while (--end >= s) {
7126 rev += (*end - '0') * mult;
7128 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
7129 Perl_warner(aTHX_ WARN_OVERFLOW,
7130 "Integer overflow in decimal number");
7133 tmpend = uv_to_utf8(tmpbuf, rev);
7134 utf8 = utf8 || rev > 127;
7135 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
7136 if (*pos == '.' && isDIGIT(pos[1]))
7142 while (isDIGIT(*pos) || *pos == '_')
7150 sv_utf8_downgrade(sv, TRUE);
7157 /* make the op for the constant and return */
7160 lvalp->opval = newSVOP(OP_CONST, 0, sv);
7162 lvalp->opval = Nullop;
7168 S_scan_formline(pTHX_ register char *s)
7173 SV *stuff = newSVpvn("",0);
7174 bool needargs = FALSE;
7177 if (*s == '.' || *s == /*{*/'}') {
7179 #ifdef PERL_STRICT_CR
7180 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7182 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7184 if (*t == '\n' || t == PL_bufend)
7187 if (PL_in_eval && !PL_rsfp) {
7188 eol = strchr(s,'\n');
7193 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7195 for (t = s; t < eol; t++) {
7196 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7198 goto enough; /* ~~ must be first line in formline */
7200 if (*t == '@' || *t == '^')
7203 sv_catpvn(stuff, s, eol-s);
7204 #ifndef PERL_STRICT_CR
7205 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7206 char *end = SvPVX(stuff) + SvCUR(stuff);
7215 s = filter_gets(PL_linestr, PL_rsfp, 0);
7216 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7217 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7220 yyerror("Format not terminated");
7230 PL_lex_state = LEX_NORMAL;
7231 PL_nextval[PL_nexttoke].ival = 0;
7235 PL_lex_state = LEX_FORMLINE;
7236 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7238 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7242 SvREFCNT_dec(stuff);
7243 PL_lex_formbrack = 0;
7254 PL_cshlen = strlen(PL_cshname);
7259 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7262 I32 oldsavestack_ix = PL_savestack_ix;
7263 CV* outsidecv = PL_compcv;
7267 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7269 SAVEI32(PL_subline);
7270 save_item(PL_subname);
7273 SAVESPTR(PL_comppad_name);
7274 SAVESPTR(PL_compcv);
7275 SAVEI32(PL_comppad_name_fill);
7276 SAVEI32(PL_min_intro_pending);
7277 SAVEI32(PL_max_intro_pending);
7278 SAVEI32(PL_pad_reset_pending);
7280 PL_compcv = (CV*)NEWSV(1104,0);
7281 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7282 CvFLAGS(PL_compcv) |= flags;
7284 PL_comppad = newAV();
7285 av_push(PL_comppad, Nullsv);
7286 PL_curpad = AvARRAY(PL_comppad);
7287 PL_comppad_name = newAV();
7288 PL_comppad_name_fill = 0;
7289 PL_min_intro_pending = 0;
7291 PL_subline = CopLINE(PL_curcop);
7293 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7294 PL_curpad[0] = (SV*)newAV();
7295 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7296 #endif /* USE_THREADS */
7298 comppadlist = newAV();
7299 AvREAL_off(comppadlist);
7300 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7301 av_store(comppadlist, 1, (SV*)PL_comppad);
7303 CvPADLIST(PL_compcv) = comppadlist;
7304 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7306 CvOWNER(PL_compcv) = 0;
7307 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7308 MUTEX_INIT(CvMUTEXP(PL_compcv));
7309 #endif /* USE_THREADS */
7311 return oldsavestack_ix;
7315 Perl_yywarn(pTHX_ char *s)
7318 PL_in_eval |= EVAL_WARNONLY;
7320 PL_in_eval &= ~EVAL_WARNONLY;
7325 Perl_yyerror(pTHX_ char *s)
7329 char *context = NULL;
7333 if (!yychar || (yychar == ';' && !PL_rsfp))
7335 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7336 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7337 while (isSPACE(*PL_oldoldbufptr))
7339 context = PL_oldoldbufptr;
7340 contlen = PL_bufptr - PL_oldoldbufptr;
7342 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7343 PL_oldbufptr != PL_bufptr) {
7344 while (isSPACE(*PL_oldbufptr))
7346 context = PL_oldbufptr;
7347 contlen = PL_bufptr - PL_oldbufptr;
7349 else if (yychar > 255)
7350 where = "next token ???";
7351 #ifdef USE_PURE_BISON
7352 /* GNU Bison sets the value -2 */
7353 else if (yychar == -2) {
7355 else if ((yychar & 127) == 127) {
7357 if (PL_lex_state == LEX_NORMAL ||
7358 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7359 where = "at end of line";
7360 else if (PL_lex_inpat)
7361 where = "within pattern";
7363 where = "within string";
7366 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7368 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7369 else if (isPRINT_LC(yychar))
7370 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7372 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7373 where = SvPVX(where_sv);
7375 msg = sv_2mortal(newSVpv(s, 0));
7376 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7377 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7379 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7381 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7382 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7383 Perl_sv_catpvf(aTHX_ msg,
7384 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7385 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7388 if (PL_in_eval & EVAL_WARNONLY)
7389 Perl_warn(aTHX_ "%"SVf, msg);
7392 if (PL_error_count >= 10) {
7393 if (PL_in_eval && SvCUR(ERRSV))
7394 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
7395 ERRSV, CopFILE(PL_curcop));
7397 Perl_croak(aTHX_ "%s has too many errors.\n",
7398 CopFILE(PL_curcop));
7401 PL_in_my_stash = Nullhv;
7406 S_swallow_bom(pTHX_ U8 *s)
7409 slen = SvCUR(PL_linestr);
7413 /* UTF-16 little-endian */
7414 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
7415 Perl_croak(aTHX_ "Unsupported script encoding");
7416 #ifndef PERL_NO_UTF16_FILTER
7417 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-LE script encoding\n"));
7419 if (PL_bufend > (char*)s) {
7423 filter_add(utf16rev_textfilter, NULL);
7424 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7425 PL_bufend = (char*)utf16_to_utf8_reversed(s, news,
7426 PL_bufend - (char*)s - 1,
7428 Copy(news, s, newlen, U8);
7429 SvCUR_set(PL_linestr, newlen);
7430 PL_bufend = SvPVX(PL_linestr) + newlen;
7431 news[newlen++] = '\0';
7435 Perl_croak(aTHX_ "Unsupported script encoding");
7440 if (s[1] == 0xFF) { /* UTF-16 big-endian */
7441 #ifndef PERL_NO_UTF16_FILTER
7442 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding\n"));
7444 if (PL_bufend > (char *)s) {
7448 filter_add(utf16_textfilter, NULL);
7449 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7450 PL_bufend = (char*)utf16_to_utf8(s, news,
7451 PL_bufend - (char*)s,
7453 Copy(news, s, newlen, U8);
7454 SvCUR_set(PL_linestr, newlen);
7455 PL_bufend = SvPVX(PL_linestr) + newlen;
7456 news[newlen++] = '\0';
7460 Perl_croak(aTHX_ "Unsupported script encoding");
7465 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
7466 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-8 script encoding\n"));
7471 if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */
7472 s[2] == 0xFE && s[3] == 0xFF)
7474 Perl_croak(aTHX_ "Unsupported script encoding");
7486 * Restore a source filter.
7490 restore_rsfp(pTHXo_ void *f)
7492 PerlIO *fp = (PerlIO*)f;
7494 if (PL_rsfp == PerlIO_stdin())
7495 PerlIO_clearerr(PL_rsfp);
7496 else if (PL_rsfp && (PL_rsfp != fp))
7497 PerlIO_close(PL_rsfp);
7501 #ifndef PERL_NO_UTF16_FILTER
7503 utf16_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7505 I32 count = FILTER_READ(idx+1, sv, maxlen);
7510 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7511 if (!*SvPV_nolen(sv))
7512 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7515 tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7516 sv_usepvn(sv, (char*)tmps, tend - tmps);
7522 utf16rev_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7524 I32 count = FILTER_READ(idx+1, sv, maxlen);
7529 if (!*SvPV_nolen(sv))
7530 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7533 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7534 tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7535 sv_usepvn(sv, (char*)tmps, tend - tmps);