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_r(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((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((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.
2074 #ifdef USE_PURE_BISON
2076 #pragma segment Perl_yylex_r
2079 Perl_yylex_r(pTHX_ YYSTYPE *lvalp, int *lcharp)
2084 yylval_pointer[yyactlevel] = lvalp;
2085 yychar_pointer[yyactlevel] = lcharp;
2087 if (yyactlevel >= YYMAXLEVEL)
2088 Perl_croak(aTHX_ "panic: YYMAXLEVEL");
2090 r = Perl_yylex(aTHX);
2099 #pragma segment Perl_yylex
2103 #ifdef USE_PURE_BISON
2104 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
2117 /* check if there's an identifier for us to look at */
2118 if (PL_pending_ident) {
2119 /* pit holds the identifier we read and pending_ident is reset */
2120 char pit = PL_pending_ident;
2121 PL_pending_ident = 0;
2123 /* if we're in a my(), we can't allow dynamics here.
2124 $foo'bar has already been turned into $foo::bar, so
2125 just check for colons.
2127 if it's a legal name, the OP is a PADANY.
2130 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
2131 if (strchr(PL_tokenbuf,':'))
2132 yyerror(Perl_form(aTHX_ "No package name allowed for "
2133 "variable %s in \"our\"",
2135 tmp = pad_allocmy(PL_tokenbuf);
2138 if (strchr(PL_tokenbuf,':'))
2139 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
2141 yylval.opval = newOP(OP_PADANY, 0);
2142 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
2148 build the ops for accesses to a my() variable.
2150 Deny my($a) or my($b) in a sort block, *if* $a or $b is
2151 then used in a comparison. This catches most, but not
2152 all cases. For instance, it catches
2153 sort { my($a); $a <=> $b }
2155 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2156 (although why you'd do that is anyone's guess).
2159 if (!strchr(PL_tokenbuf,':')) {
2161 /* Check for single character per-thread SVs */
2162 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2163 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2164 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2166 yylval.opval = newOP(OP_THREADSV, 0);
2167 yylval.opval->op_targ = tmp;
2170 #endif /* USE_THREADS */
2171 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2172 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2173 /* might be an "our" variable" */
2174 if (SvFLAGS(namesv) & SVpad_OUR) {
2175 /* build ops for a bareword */
2176 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2177 sv_catpvn(sym, "::", 2);
2178 sv_catpv(sym, PL_tokenbuf+1);
2179 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2180 yylval.opval->op_private = OPpCONST_ENTERED;
2181 gv_fetchpv(SvPVX(sym),
2183 ? (GV_ADDMULTI | GV_ADDINEVAL)
2186 ((PL_tokenbuf[0] == '$') ? SVt_PV
2187 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2192 /* if it's a sort block and they're naming $a or $b */
2193 if (PL_last_lop_op == OP_SORT &&
2194 PL_tokenbuf[0] == '$' &&
2195 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2198 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2199 d < PL_bufend && *d != '\n';
2202 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2203 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2209 yylval.opval = newOP(OP_PADANY, 0);
2210 yylval.opval->op_targ = tmp;
2216 Whine if they've said @foo in a doublequoted string,
2217 and @foo isn't a variable we can find in the symbol
2220 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2221 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2222 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2223 && ckWARN(WARN_AMBIGUOUS))
2225 /* Downgraded from fatal to warning 20000522 mjd */
2226 Perl_warner(aTHX_ WARN_AMBIGUOUS,
2227 "Possible unintended interpolation of %s in string",
2232 /* build ops for a bareword */
2233 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2234 yylval.opval->op_private = OPpCONST_ENTERED;
2235 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2236 ((PL_tokenbuf[0] == '$') ? SVt_PV
2237 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2242 /* no identifier pending identification */
2244 switch (PL_lex_state) {
2246 case LEX_NORMAL: /* Some compilers will produce faster */
2247 case LEX_INTERPNORMAL: /* code if we comment these out. */
2251 /* when we've already built the next token, just pull it out of the queue */
2254 yylval = PL_nextval[PL_nexttoke];
2256 PL_lex_state = PL_lex_defer;
2257 PL_expect = PL_lex_expect;
2258 PL_lex_defer = LEX_NORMAL;
2260 return(PL_nexttype[PL_nexttoke]);
2262 /* interpolated case modifiers like \L \U, including \Q and \E.
2263 when we get here, PL_bufptr is at the \
2265 case LEX_INTERPCASEMOD:
2267 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2268 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2270 /* handle \E or end of string */
2271 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2275 if (PL_lex_casemods) {
2276 oldmod = PL_lex_casestack[--PL_lex_casemods];
2277 PL_lex_casestack[PL_lex_casemods] = '\0';
2279 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2281 PL_lex_state = LEX_INTERPCONCAT;
2285 if (PL_bufptr != PL_bufend)
2287 PL_lex_state = LEX_INTERPCONCAT;
2292 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2293 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2294 if (strchr("LU", *s) &&
2295 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2297 PL_lex_casestack[--PL_lex_casemods] = '\0';
2300 if (PL_lex_casemods > 10) {
2301 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2302 if (newlb != PL_lex_casestack) {
2304 PL_lex_casestack = newlb;
2307 PL_lex_casestack[PL_lex_casemods++] = *s;
2308 PL_lex_casestack[PL_lex_casemods] = '\0';
2309 PL_lex_state = LEX_INTERPCONCAT;
2310 PL_nextval[PL_nexttoke].ival = 0;
2313 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2315 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2317 PL_nextval[PL_nexttoke].ival = OP_LC;
2319 PL_nextval[PL_nexttoke].ival = OP_UC;
2321 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2323 Perl_croak(aTHX_ "panic: yylex");
2326 if (PL_lex_starts) {
2335 case LEX_INTERPPUSH:
2336 return sublex_push();
2338 case LEX_INTERPSTART:
2339 if (PL_bufptr == PL_bufend)
2340 return sublex_done();
2342 PL_lex_dojoin = (*PL_bufptr == '@');
2343 PL_lex_state = LEX_INTERPNORMAL;
2344 if (PL_lex_dojoin) {
2345 PL_nextval[PL_nexttoke].ival = 0;
2348 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2349 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2350 force_next(PRIVATEREF);
2352 force_ident("\"", '$');
2353 #endif /* USE_THREADS */
2354 PL_nextval[PL_nexttoke].ival = 0;
2356 PL_nextval[PL_nexttoke].ival = 0;
2358 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2361 if (PL_lex_starts++) {
2367 case LEX_INTERPENDMAYBE:
2368 if (intuit_more(PL_bufptr)) {
2369 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2375 if (PL_lex_dojoin) {
2376 PL_lex_dojoin = FALSE;
2377 PL_lex_state = LEX_INTERPCONCAT;
2380 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2381 && SvEVALED(PL_lex_repl))
2383 if (PL_bufptr != PL_bufend)
2384 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2385 PL_lex_repl = Nullsv;
2388 case LEX_INTERPCONCAT:
2390 if (PL_lex_brackets)
2391 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2393 if (PL_bufptr == PL_bufend)
2394 return sublex_done();
2396 if (SvIVX(PL_linestr) == '\'') {
2397 SV *sv = newSVsv(PL_linestr);
2400 else if ( PL_hints & HINT_NEW_RE )
2401 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2402 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2406 s = scan_const(PL_bufptr);
2408 PL_lex_state = LEX_INTERPCASEMOD;
2410 PL_lex_state = LEX_INTERPSTART;
2413 if (s != PL_bufptr) {
2414 PL_nextval[PL_nexttoke] = yylval;
2417 if (PL_lex_starts++)
2427 PL_lex_state = LEX_NORMAL;
2428 s = scan_formline(PL_bufptr);
2429 if (!PL_lex_formbrack)
2435 PL_oldoldbufptr = PL_oldbufptr;
2438 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2439 exp_name[PL_expect], s);
2445 if (isIDFIRST_lazy_if(s,UTF))
2447 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2450 goto fake_eof; /* emulate EOF on ^D or ^Z */
2455 if (PL_lex_brackets)
2456 yyerror("Missing right curly or square bracket");
2459 if (s++ < PL_bufend)
2460 goto retry; /* ignore stray nulls */
2463 if (!PL_in_eval && !PL_preambled) {
2464 PL_preambled = TRUE;
2465 sv_setpv(PL_linestr,incl_perldb());
2466 if (SvCUR(PL_linestr))
2467 sv_catpv(PL_linestr,";");
2469 while(AvFILLp(PL_preambleav) >= 0) {
2470 SV *tmpsv = av_shift(PL_preambleav);
2471 sv_catsv(PL_linestr, tmpsv);
2472 sv_catpv(PL_linestr, ";");
2475 sv_free((SV*)PL_preambleav);
2476 PL_preambleav = NULL;
2478 if (PL_minus_n || PL_minus_p) {
2479 sv_catpv(PL_linestr, "LINE: while (<>) {");
2481 sv_catpv(PL_linestr,"chomp;");
2483 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2485 GvIMPORTED_AV_on(gv);
2487 if (strchr("/'\"", *PL_splitstr)
2488 && strchr(PL_splitstr + 1, *PL_splitstr))
2489 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2492 s = "'~#\200\1'"; /* surely one char is unused...*/
2493 while (s[1] && strchr(PL_splitstr, *s)) s++;
2495 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2496 "q" + (delim == '\''), delim);
2497 for (s = PL_splitstr; *s; s++) {
2499 sv_catpvn(PL_linestr, "\\", 1);
2500 sv_catpvn(PL_linestr, s, 1);
2502 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2506 sv_catpv(PL_linestr,"@F=split(' ');");
2509 sv_catpv(PL_linestr, "\n");
2510 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2511 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2512 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2513 SV *sv = NEWSV(85,0);
2515 sv_upgrade(sv, SVt_PVMG);
2516 sv_setsv(sv,PL_linestr);
2517 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2523 bof = PL_rsfp && (PerlIO_tell(PL_rsfp) == 0); /* *Before* read! */
2524 s = filter_gets(PL_linestr, PL_rsfp, 0);
2528 if (PL_preprocess && !PL_in_eval)
2529 (void)PerlProc_pclose(PL_rsfp);
2530 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2531 PerlIO_clearerr(PL_rsfp);
2533 (void)PerlIO_close(PL_rsfp);
2535 PL_doextract = FALSE;
2537 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2538 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2539 sv_catpv(PL_linestr,";}");
2540 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2541 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2542 PL_minus_n = PL_minus_p = 0;
2545 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2546 sv_setpv(PL_linestr,"");
2547 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2549 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2550 s = swallow_bom((U8*)s);
2553 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2554 PL_doextract = FALSE;
2556 /* Incest with pod. */
2557 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2558 sv_setpv(PL_linestr, "");
2559 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2560 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2561 PL_doextract = FALSE;
2565 } while (PL_doextract);
2566 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2567 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2568 SV *sv = NEWSV(85,0);
2570 sv_upgrade(sv, SVt_PVMG);
2571 sv_setsv(sv,PL_linestr);
2572 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2574 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2575 if (CopLINE(PL_curcop) == 1) {
2576 while (s < PL_bufend && isSPACE(*s))
2578 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2582 if (*s == '#' && *(s+1) == '!')
2584 #ifdef ALTERNATE_SHEBANG
2586 static char as[] = ALTERNATE_SHEBANG;
2587 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2588 d = s + (sizeof(as) - 1);
2590 #endif /* ALTERNATE_SHEBANG */
2599 while (*d && !isSPACE(*d))
2603 #ifdef ARG_ZERO_IS_SCRIPT
2604 if (ipathend > ipath) {
2606 * HP-UX (at least) sets argv[0] to the script name,
2607 * which makes $^X incorrect. And Digital UNIX and Linux,
2608 * at least, set argv[0] to the basename of the Perl
2609 * interpreter. So, having found "#!", we'll set it right.
2611 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2612 assert(SvPOK(x) || SvGMAGICAL(x));
2613 if (sv_eq(x, CopFILESV(PL_curcop))) {
2614 sv_setpvn(x, ipath, ipathend - ipath);
2617 TAINT_NOT; /* $^X is always tainted, but that's OK */
2619 #endif /* ARG_ZERO_IS_SCRIPT */
2624 d = instr(s,"perl -");
2626 d = instr(s,"perl");
2628 /* avoid getting into infinite loops when shebang
2629 * line contains "Perl" rather than "perl" */
2631 for (d = ipathend-4; d >= ipath; --d) {
2632 if ((*d == 'p' || *d == 'P')
2633 && !ibcmp(d, "perl", 4))
2643 #ifdef ALTERNATE_SHEBANG
2645 * If the ALTERNATE_SHEBANG on this system starts with a
2646 * character that can be part of a Perl expression, then if
2647 * we see it but not "perl", we're probably looking at the
2648 * start of Perl code, not a request to hand off to some
2649 * other interpreter. Similarly, if "perl" is there, but
2650 * not in the first 'word' of the line, we assume the line
2651 * contains the start of the Perl program.
2653 if (d && *s != '#') {
2655 while (*c && !strchr("; \t\r\n\f\v#", *c))
2658 d = Nullch; /* "perl" not in first word; ignore */
2660 *s = '#'; /* Don't try to parse shebang line */
2662 #endif /* ALTERNATE_SHEBANG */
2663 #ifndef MACOS_TRADITIONAL
2668 !instr(s,"indir") &&
2669 instr(PL_origargv[0],"perl"))
2675 while (s < PL_bufend && isSPACE(*s))
2677 if (s < PL_bufend) {
2678 Newz(899,newargv,PL_origargc+3,char*);
2680 while (s < PL_bufend && !isSPACE(*s))
2683 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2686 newargv = PL_origargv;
2688 PerlProc_execv(ipath, newargv);
2689 Perl_croak(aTHX_ "Can't exec %s", ipath);
2693 U32 oldpdb = PL_perldb;
2694 bool oldn = PL_minus_n;
2695 bool oldp = PL_minus_p;
2697 while (*d && !isSPACE(*d)) d++;
2698 while (SPACE_OR_TAB(*d)) d++;
2702 if (*d == 'M' || *d == 'm') {
2704 while (*d && !isSPACE(*d)) d++;
2705 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2708 d = moreswitches(d);
2710 if ((PERLDB_LINE && !oldpdb) ||
2711 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2712 /* if we have already added "LINE: while (<>) {",
2713 we must not do it again */
2715 sv_setpv(PL_linestr, "");
2716 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2717 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2718 PL_preambled = FALSE;
2720 (void)gv_fetchfile(PL_origfilename);
2727 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2729 PL_lex_state = LEX_FORMLINE;
2734 #ifdef PERL_STRICT_CR
2735 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2737 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2739 case ' ': case '\t': case '\f': case 013:
2740 #ifdef MACOS_TRADITIONAL
2747 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2748 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2749 /* handle eval qq[#line 1 "foo"\n ...] */
2750 CopLINE_dec(PL_curcop);
2754 while (s < d && *s != '\n')
2759 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2761 PL_lex_state = LEX_FORMLINE;
2771 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2776 while (s < PL_bufend && SPACE_OR_TAB(*s))
2779 if (strnEQ(s,"=>",2)) {
2780 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2781 OPERATOR('-'); /* unary minus */
2783 PL_last_uni = PL_oldbufptr;
2784 PL_last_lop_op = OP_FTEREAD; /* good enough */
2786 case 'r': FTST(OP_FTEREAD);
2787 case 'w': FTST(OP_FTEWRITE);
2788 case 'x': FTST(OP_FTEEXEC);
2789 case 'o': FTST(OP_FTEOWNED);
2790 case 'R': FTST(OP_FTRREAD);
2791 case 'W': FTST(OP_FTRWRITE);
2792 case 'X': FTST(OP_FTREXEC);
2793 case 'O': FTST(OP_FTROWNED);
2794 case 'e': FTST(OP_FTIS);
2795 case 'z': FTST(OP_FTZERO);
2796 case 's': FTST(OP_FTSIZE);
2797 case 'f': FTST(OP_FTFILE);
2798 case 'd': FTST(OP_FTDIR);
2799 case 'l': FTST(OP_FTLINK);
2800 case 'p': FTST(OP_FTPIPE);
2801 case 'S': FTST(OP_FTSOCK);
2802 case 'u': FTST(OP_FTSUID);
2803 case 'g': FTST(OP_FTSGID);
2804 case 'k': FTST(OP_FTSVTX);
2805 case 'b': FTST(OP_FTBLK);
2806 case 'c': FTST(OP_FTCHR);
2807 case 't': FTST(OP_FTTTY);
2808 case 'T': FTST(OP_FTTEXT);
2809 case 'B': FTST(OP_FTBINARY);
2810 case 'M': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTMTIME);
2811 case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME);
2812 case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME);
2814 Perl_croak(aTHX_ "Unrecognized file test: -%c", (int)tmp);
2821 if (PL_expect == XOPERATOR)
2826 else if (*s == '>') {
2829 if (isIDFIRST_lazy_if(s,UTF)) {
2830 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2838 if (PL_expect == XOPERATOR)
2841 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2843 OPERATOR('-'); /* unary minus */
2850 if (PL_expect == XOPERATOR)
2855 if (PL_expect == XOPERATOR)
2858 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2864 if (PL_expect != XOPERATOR) {
2865 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2866 PL_expect = XOPERATOR;
2867 force_ident(PL_tokenbuf, '*');
2880 if (PL_expect == XOPERATOR) {
2884 PL_tokenbuf[0] = '%';
2885 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2886 if (!PL_tokenbuf[1]) {
2888 yyerror("Final % should be \\% or %name");
2891 PL_pending_ident = '%';
2910 switch (PL_expect) {
2913 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2915 PL_bufptr = s; /* update in case we back off */
2921 PL_expect = XTERMBLOCK;
2925 while (isIDFIRST_lazy_if(s,UTF)) {
2926 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2927 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2928 if (tmp < 0) tmp = -tmp;
2943 d = scan_str(d,TRUE,TRUE);
2946 SvREFCNT_dec(PL_lex_stuff);
2947 PL_lex_stuff = Nullsv;
2949 /* MUST advance bufptr here to avoid bogus
2950 "at end of line" context messages from yyerror().
2952 PL_bufptr = s + len;
2953 yyerror("Unterminated attribute parameter in attribute list");
2956 return 0; /* EOF indicator */
2960 SV *sv = newSVpvn(s, len);
2961 sv_catsv(sv, PL_lex_stuff);
2962 attrs = append_elem(OP_LIST, attrs,
2963 newSVOP(OP_CONST, 0, sv));
2964 SvREFCNT_dec(PL_lex_stuff);
2965 PL_lex_stuff = Nullsv;
2968 attrs = append_elem(OP_LIST, attrs,
2969 newSVOP(OP_CONST, 0,
2973 if (*s == ':' && s[1] != ':')
2976 break; /* require real whitespace or :'s */
2978 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
2979 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
2980 char q = ((*s == '\'') ? '"' : '\'');
2981 /* If here for an expression, and parsed no attrs, back off. */
2982 if (tmp == '=' && !attrs) {
2986 /* MUST advance bufptr here to avoid bogus "at end of line"
2987 context messages from yyerror().
2991 yyerror("Unterminated attribute list");
2993 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
3001 PL_nextval[PL_nexttoke].opval = attrs;
3009 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3010 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3026 if (PL_lex_brackets <= 0)
3027 yyerror("Unmatched right square bracket");
3030 if (PL_lex_state == LEX_INTERPNORMAL) {
3031 if (PL_lex_brackets == 0) {
3032 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3033 PL_lex_state = LEX_INTERPEND;
3040 if (PL_lex_brackets > 100) {
3041 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
3042 if (newlb != PL_lex_brackstack) {
3044 PL_lex_brackstack = newlb;
3047 switch (PL_expect) {
3049 if (PL_lex_formbrack) {
3053 if (PL_oldoldbufptr == PL_last_lop)
3054 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3056 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3057 OPERATOR(HASHBRACK);
3059 while (s < PL_bufend && SPACE_OR_TAB(*s))
3062 PL_tokenbuf[0] = '\0';
3063 if (d < PL_bufend && *d == '-') {
3064 PL_tokenbuf[0] = '-';
3066 while (d < PL_bufend && SPACE_OR_TAB(*d))
3069 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3070 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3072 while (d < PL_bufend && SPACE_OR_TAB(*d))
3075 char minus = (PL_tokenbuf[0] == '-');
3076 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3084 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3089 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3094 if (PL_oldoldbufptr == PL_last_lop)
3095 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3097 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3100 OPERATOR(HASHBRACK);
3101 /* This hack serves to disambiguate a pair of curlies
3102 * as being a block or an anon hash. Normally, expectation
3103 * determines that, but in cases where we're not in a
3104 * position to expect anything in particular (like inside
3105 * eval"") we have to resolve the ambiguity. This code
3106 * covers the case where the first term in the curlies is a
3107 * quoted string. Most other cases need to be explicitly
3108 * disambiguated by prepending a `+' before the opening
3109 * curly in order to force resolution as an anon hash.
3111 * XXX should probably propagate the outer expectation
3112 * into eval"" to rely less on this hack, but that could
3113 * potentially break current behavior of eval"".
3117 if (*s == '\'' || *s == '"' || *s == '`') {
3118 /* common case: get past first string, handling escapes */
3119 for (t++; t < PL_bufend && *t != *s;)
3120 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3124 else if (*s == 'q') {
3127 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3131 char open, close, term;
3134 while (t < PL_bufend && isSPACE(*t))
3138 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3142 for (t++; t < PL_bufend; t++) {
3143 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3145 else if (*t == open)
3149 for (t++; t < PL_bufend; t++) {
3150 if (*t == '\\' && t+1 < PL_bufend)
3152 else if (*t == close && --brackets <= 0)
3154 else if (*t == open)
3160 else if (isALNUM_lazy_if(t,UTF)) {
3162 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3165 while (t < PL_bufend && isSPACE(*t))
3167 /* if comma follows first term, call it an anon hash */
3168 /* XXX it could be a comma expression with loop modifiers */
3169 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3170 || (*t == '=' && t[1] == '>')))
3171 OPERATOR(HASHBRACK);
3172 if (PL_expect == XREF)
3175 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3181 yylval.ival = CopLINE(PL_curcop);
3182 if (isSPACE(*s) || *s == '#')
3183 PL_copline = NOLINE; /* invalidate current command line number */
3188 if (PL_lex_brackets <= 0)
3189 yyerror("Unmatched right curly bracket");
3191 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3192 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3193 PL_lex_formbrack = 0;
3194 if (PL_lex_state == LEX_INTERPNORMAL) {
3195 if (PL_lex_brackets == 0) {
3196 if (PL_expect & XFAKEBRACK) {
3197 PL_expect &= XENUMMASK;
3198 PL_lex_state = LEX_INTERPEND;
3200 return yylex(); /* ignore fake brackets */
3202 if (*s == '-' && s[1] == '>')
3203 PL_lex_state = LEX_INTERPENDMAYBE;
3204 else if (*s != '[' && *s != '{')
3205 PL_lex_state = LEX_INTERPEND;
3208 if (PL_expect & XFAKEBRACK) {
3209 PL_expect &= XENUMMASK;
3211 return yylex(); /* ignore fake brackets */
3221 if (PL_expect == XOPERATOR) {
3222 if (ckWARN(WARN_SEMICOLON)
3223 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3225 CopLINE_dec(PL_curcop);
3226 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3227 CopLINE_inc(PL_curcop);
3232 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3234 PL_expect = XOPERATOR;
3235 force_ident(PL_tokenbuf, '&');
3239 yylval.ival = (OPpENTERSUB_AMPER<<8);
3258 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3259 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3261 if (PL_expect == XSTATE && isALPHA(tmp) &&
3262 (s == PL_linestart+1 || s[-2] == '\n') )
3264 if (PL_in_eval && !PL_rsfp) {
3269 if (strnEQ(s,"=cut",4)) {
3283 PL_doextract = TRUE;
3286 if (PL_lex_brackets < PL_lex_formbrack) {
3288 #ifdef PERL_STRICT_CR
3289 for (t = s; SPACE_OR_TAB(*t); t++) ;
3291 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3293 if (*t == '\n' || *t == '#') {
3311 if (PL_expect != XOPERATOR) {
3312 if (s[1] != '<' && !strchr(s,'>'))
3315 s = scan_heredoc(s);
3317 s = scan_inputsymbol(s);
3318 TERM(sublex_start());
3323 SHop(OP_LEFT_SHIFT);
3337 SHop(OP_RIGHT_SHIFT);
3346 if (PL_expect == XOPERATOR) {
3347 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3350 return ','; /* grandfather non-comma-format format */
3354 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3355 PL_tokenbuf[0] = '@';
3356 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3357 sizeof PL_tokenbuf - 1, FALSE);
3358 if (PL_expect == XOPERATOR)
3359 no_op("Array length", s);
3360 if (!PL_tokenbuf[1])
3362 PL_expect = XOPERATOR;
3363 PL_pending_ident = '#';
3367 PL_tokenbuf[0] = '$';
3368 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3369 sizeof PL_tokenbuf - 1, FALSE);
3370 if (PL_expect == XOPERATOR)
3372 if (!PL_tokenbuf[1]) {
3374 yyerror("Final $ should be \\$ or $name");
3378 /* This kludge not intended to be bulletproof. */
3379 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3380 yylval.opval = newSVOP(OP_CONST, 0,
3381 newSViv(PL_compiling.cop_arybase));
3382 yylval.opval->op_private = OPpCONST_ARYBASE;
3388 if (PL_lex_state == LEX_NORMAL)
3391 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3394 PL_tokenbuf[0] = '@';
3395 if (ckWARN(WARN_SYNTAX)) {
3397 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3400 PL_bufptr = skipspace(PL_bufptr);
3401 while (t < PL_bufend && *t != ']')
3403 Perl_warner(aTHX_ WARN_SYNTAX,
3404 "Multidimensional syntax %.*s not supported",
3405 (t - PL_bufptr) + 1, PL_bufptr);
3409 else if (*s == '{') {
3410 PL_tokenbuf[0] = '%';
3411 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3412 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3414 char tmpbuf[sizeof PL_tokenbuf];
3416 for (t++; isSPACE(*t); t++) ;
3417 if (isIDFIRST_lazy_if(t,UTF)) {
3418 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3419 for (; isSPACE(*t); t++) ;
3420 if (*t == ';' && get_cv(tmpbuf, FALSE))
3421 Perl_warner(aTHX_ WARN_SYNTAX,
3422 "You need to quote \"%s\"", tmpbuf);
3428 PL_expect = XOPERATOR;
3429 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3430 bool islop = (PL_last_lop == PL_oldoldbufptr);
3431 if (!islop || PL_last_lop_op == OP_GREPSTART)
3432 PL_expect = XOPERATOR;
3433 else if (strchr("$@\"'`q", *s))
3434 PL_expect = XTERM; /* e.g. print $fh "foo" */
3435 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3436 PL_expect = XTERM; /* e.g. print $fh &sub */
3437 else if (isIDFIRST_lazy_if(s,UTF)) {
3438 char tmpbuf[sizeof PL_tokenbuf];
3439 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3440 if ((tmp = keyword(tmpbuf, len))) {
3441 /* binary operators exclude handle interpretations */
3453 PL_expect = XTERM; /* e.g. print $fh length() */
3458 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3459 if (gv && GvCVu(gv))
3460 PL_expect = XTERM; /* e.g. print $fh subr() */
3463 else if (isDIGIT(*s))
3464 PL_expect = XTERM; /* e.g. print $fh 3 */
3465 else if (*s == '.' && isDIGIT(s[1]))
3466 PL_expect = XTERM; /* e.g. print $fh .3 */
3467 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3468 PL_expect = XTERM; /* e.g. print $fh -1 */
3469 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3470 PL_expect = XTERM; /* print $fh <<"EOF" */
3472 PL_pending_ident = '$';
3476 if (PL_expect == XOPERATOR)
3478 PL_tokenbuf[0] = '@';
3479 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3480 if (!PL_tokenbuf[1]) {
3482 yyerror("Final @ should be \\@ or @name");
3485 if (PL_lex_state == LEX_NORMAL)
3487 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3489 PL_tokenbuf[0] = '%';
3491 /* Warn about @ where they meant $. */
3492 if (ckWARN(WARN_SYNTAX)) {
3493 if (*s == '[' || *s == '{') {
3495 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3497 if (*t == '}' || *t == ']') {
3499 PL_bufptr = skipspace(PL_bufptr);
3500 Perl_warner(aTHX_ WARN_SYNTAX,
3501 "Scalar value %.*s better written as $%.*s",
3502 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3507 PL_pending_ident = '@';
3510 case '/': /* may either be division or pattern */
3511 case '?': /* may either be conditional or pattern */
3512 if (PL_expect != XOPERATOR) {
3513 /* Disable warning on "study /blah/" */
3514 if (PL_oldoldbufptr == PL_last_uni
3515 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3516 || memNE(PL_last_uni, "study", 5)
3517 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3519 s = scan_pat(s,OP_MATCH);
3520 TERM(sublex_start());
3528 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3529 #ifdef PERL_STRICT_CR
3532 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3534 && (s == PL_linestart || s[-1] == '\n') )
3536 PL_lex_formbrack = 0;
3540 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3546 yylval.ival = OPf_SPECIAL;
3552 if (PL_expect != XOPERATOR)
3557 case '0': case '1': case '2': case '3': case '4':
3558 case '5': case '6': case '7': case '8': case '9':
3559 s = scan_num(s, &yylval);
3560 if (PL_expect == XOPERATOR)
3565 s = scan_str(s,FALSE,FALSE);
3566 if (PL_expect == XOPERATOR) {
3567 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3570 return ','; /* grandfather non-comma-format format */
3576 missingterm((char*)0);
3577 yylval.ival = OP_CONST;
3578 TERM(sublex_start());
3581 s = scan_str(s,FALSE,FALSE);
3582 if (PL_expect == XOPERATOR) {
3583 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3586 return ','; /* grandfather non-comma-format format */
3592 missingterm((char*)0);
3593 yylval.ival = OP_CONST;
3594 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3595 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3596 yylval.ival = OP_STRINGIFY;
3600 TERM(sublex_start());
3603 s = scan_str(s,FALSE,FALSE);
3604 if (PL_expect == XOPERATOR)
3605 no_op("Backticks",s);
3607 missingterm((char*)0);
3608 yylval.ival = OP_BACKTICK;
3610 TERM(sublex_start());
3614 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3615 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3617 if (PL_expect == XOPERATOR)
3618 no_op("Backslash",s);
3622 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3626 while (isDIGIT(*start) || *start == '_')
3628 if (*start == '.' && isDIGIT(start[1])) {
3629 s = scan_num(s, &yylval);
3632 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3633 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF)) {
3637 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3640 s = scan_num(s, &yylval);
3647 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3686 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3688 /* Some keywords can be followed by any delimiter, including ':' */
3689 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3690 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3691 (PL_tokenbuf[0] == 'q' &&
3692 strchr("qwxr", PL_tokenbuf[1])))));
3694 /* x::* is just a word, unless x is "CORE" */
3695 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3699 while (d < PL_bufend && isSPACE(*d))
3700 d++; /* no comments skipped here, or s### is misparsed */
3702 /* Is this a label? */
3703 if (!tmp && PL_expect == XSTATE
3704 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3706 yylval.pval = savepv(PL_tokenbuf);
3711 /* Check for keywords */
3712 tmp = keyword(PL_tokenbuf, len);
3714 /* Is this a word before a => operator? */
3715 if (*d == '=' && d[1] == '>') {
3717 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3718 yylval.opval->op_private = OPpCONST_BARE;
3722 if (tmp < 0) { /* second-class keyword? */
3723 GV *ogv = Nullgv; /* override (winner) */
3724 GV *hgv = Nullgv; /* hidden (loser) */
3725 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3727 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3730 if (GvIMPORTED_CV(gv))
3732 else if (! CvMETHOD(cv))
3736 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3737 (gv = *gvp) != (GV*)&PL_sv_undef &&
3738 GvCVu(gv) && GvIMPORTED_CV(gv))
3744 tmp = 0; /* overridden by import or by GLOBAL */
3747 && -tmp==KEY_lock /* XXX generalizable kludge */
3749 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3751 tmp = 0; /* any sub overrides "weak" keyword */
3753 else { /* no override */
3757 if (ckWARN(WARN_AMBIGUOUS) && hgv
3758 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3759 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3760 "Ambiguous call resolved as CORE::%s(), %s",
3761 GvENAME(hgv), "qualify as such or use &");
3768 default: /* not a keyword */
3771 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3773 /* Get the rest if it looks like a package qualifier */
3775 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3777 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3780 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3781 *s == '\'' ? "'" : "::");
3785 if (PL_expect == XOPERATOR) {
3786 if (PL_bufptr == PL_linestart) {
3787 CopLINE_dec(PL_curcop);
3788 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3789 CopLINE_inc(PL_curcop);
3792 no_op("Bareword",s);
3795 /* Look for a subroutine with this name in current package,
3796 unless name is "Foo::", in which case Foo is a bearword
3797 (and a package name). */
3800 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3802 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3803 Perl_warner(aTHX_ WARN_BAREWORD,
3804 "Bareword \"%s\" refers to nonexistent package",
3807 PL_tokenbuf[len] = '\0';
3814 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3817 /* if we saw a global override before, get the right name */
3820 sv = newSVpvn("CORE::GLOBAL::",14);
3821 sv_catpv(sv,PL_tokenbuf);
3824 sv = newSVpv(PL_tokenbuf,0);
3826 /* Presume this is going to be a bareword of some sort. */
3829 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3830 yylval.opval->op_private = OPpCONST_BARE;
3832 /* And if "Foo::", then that's what it certainly is. */
3837 /* See if it's the indirect object for a list operator. */
3839 if (PL_oldoldbufptr &&
3840 PL_oldoldbufptr < PL_bufptr &&
3841 (PL_oldoldbufptr == PL_last_lop
3842 || PL_oldoldbufptr == PL_last_uni) &&
3843 /* NO SKIPSPACE BEFORE HERE! */
3844 (PL_expect == XREF ||
3845 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3847 bool immediate_paren = *s == '(';
3849 /* (Now we can afford to cross potential line boundary.) */
3852 /* Two barewords in a row may indicate method call. */
3854 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3857 /* If not a declared subroutine, it's an indirect object. */
3858 /* (But it's an indir obj regardless for sort.) */
3860 if ((PL_last_lop_op == OP_SORT ||
3861 (!immediate_paren && (!gv || !GvCVu(gv)))) &&
3862 (PL_last_lop_op != OP_MAPSTART &&
3863 PL_last_lop_op != OP_GREPSTART))
3865 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3871 PL_expect = XOPERATOR;
3874 /* Is this a word before a => operator? */
3875 if (*s == '=' && s[1] == '>') {
3877 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
3881 /* If followed by a paren, it's certainly a subroutine. */
3884 if (gv && GvCVu(gv)) {
3885 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
3886 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3891 PL_nextval[PL_nexttoke].opval = yylval.opval;
3892 PL_expect = XOPERATOR;
3898 /* If followed by var or block, call it a method (unless sub) */
3900 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3901 PL_last_lop = PL_oldbufptr;
3902 PL_last_lop_op = OP_METHOD;
3906 /* If followed by a bareword, see if it looks like indir obj. */
3908 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3911 /* Not a method, so call it a subroutine (if defined) */
3913 if (gv && GvCVu(gv)) {
3915 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3916 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3917 "Ambiguous use of -%s resolved as -&%s()",
3918 PL_tokenbuf, PL_tokenbuf);
3919 /* Check for a constant sub */
3921 if ((sv = cv_const_sv(cv))) {
3923 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3924 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3925 yylval.opval->op_private = 0;
3929 /* Resolve to GV now. */
3930 op_free(yylval.opval);
3931 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3932 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3933 PL_last_lop = PL_oldbufptr;
3934 PL_last_lop_op = OP_ENTERSUB;
3935 /* Is there a prototype? */
3938 char *proto = SvPV((SV*)cv, len);
3941 if (strEQ(proto, "$"))
3943 if (*proto == '&' && *s == '{') {
3944 sv_setpv(PL_subname,"__ANON__");
3948 PL_nextval[PL_nexttoke].opval = yylval.opval;
3954 /* Call it a bare word */
3956 if (PL_hints & HINT_STRICT_SUBS)
3957 yylval.opval->op_private |= OPpCONST_STRICT;
3960 if (ckWARN(WARN_RESERVED)) {
3961 if (lastchar != '-') {
3962 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
3964 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
3971 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
3972 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3973 "Operator or semicolon missing before %c%s",
3974 lastchar, PL_tokenbuf);
3975 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3976 "Ambiguous use of %c resolved as operator %c",
3977 lastchar, lastchar);
3983 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3984 newSVpv(CopFILE(PL_curcop),0));
3988 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3989 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
3992 case KEY___PACKAGE__:
3993 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3995 ? newSVsv(PL_curstname)
4004 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4005 char *pname = "main";
4006 if (PL_tokenbuf[2] == 'D')
4007 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
4008 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4011 GvIOp(gv) = newIO();
4012 IoIFP(GvIOp(gv)) = PL_rsfp;
4013 #if defined(HAS_FCNTL) && defined(F_SETFD)
4015 int fd = PerlIO_fileno(PL_rsfp);
4016 fcntl(fd,F_SETFD,fd >= 3);
4019 /* Mark this internal pseudo-handle as clean */
4020 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4022 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4023 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4024 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4026 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4027 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4028 /* if the script was opened in binmode, we need to revert
4029 * it to text mode for compatibility; but only iff it has CRs
4030 * XXX this is a questionable hack at best. */
4031 if (PL_bufend-PL_bufptr > 2
4032 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4035 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4036 loc = PerlIO_tell(PL_rsfp);
4037 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4039 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4040 #if defined(__BORLANDC__)
4041 /* XXX see note in do_binmode() */
4042 ((FILE*)PL_rsfp)->flags |= _F_BIN;
4045 PerlIO_seek(PL_rsfp, loc, 0);
4060 if (PL_expect == XSTATE) {
4067 if (*s == ':' && s[1] == ':') {
4070 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4071 if (!(tmp = keyword(PL_tokenbuf, len)))
4072 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4086 LOP(OP_ACCEPT,XTERM);
4092 LOP(OP_ATAN2,XTERM);
4098 LOP(OP_BINMODE,XTERM);
4101 LOP(OP_BLESS,XTERM);
4110 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4127 if (!PL_cryptseen) {
4128 PL_cryptseen = TRUE;
4132 LOP(OP_CRYPT,XTERM);
4135 if (ckWARN(WARN_CHMOD)) {
4136 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4137 if (*d != '0' && isDIGIT(*d))
4138 Perl_warner(aTHX_ WARN_CHMOD,
4139 "chmod() mode argument is missing initial 0");
4141 LOP(OP_CHMOD,XTERM);
4144 LOP(OP_CHOWN,XTERM);
4147 LOP(OP_CONNECT,XTERM);
4163 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4167 PL_hints |= HINT_BLOCK_SCOPE;
4177 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4178 LOP(OP_DBMOPEN,XTERM);
4184 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4191 yylval.ival = CopLINE(PL_curcop);
4205 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4206 UNIBRACK(OP_ENTEREVAL);
4221 case KEY_endhostent:
4227 case KEY_endservent:
4230 case KEY_endprotoent:
4241 yylval.ival = CopLINE(PL_curcop);
4243 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4245 if ((PL_bufend - p) >= 3 &&
4246 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4248 else if ((PL_bufend - p) >= 4 &&
4249 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4252 if (isIDFIRST_lazy_if(p,UTF)) {
4253 p = scan_ident(p, PL_bufend,
4254 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4258 Perl_croak(aTHX_ "Missing $ on loop variable");
4263 LOP(OP_FORMLINE,XTERM);
4269 LOP(OP_FCNTL,XTERM);
4275 LOP(OP_FLOCK,XTERM);
4284 LOP(OP_GREPSTART, XREF);
4287 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4302 case KEY_getpriority:
4303 LOP(OP_GETPRIORITY,XTERM);
4305 case KEY_getprotobyname:
4308 case KEY_getprotobynumber:
4309 LOP(OP_GPBYNUMBER,XTERM);
4311 case KEY_getprotoent:
4323 case KEY_getpeername:
4324 UNI(OP_GETPEERNAME);
4326 case KEY_gethostbyname:
4329 case KEY_gethostbyaddr:
4330 LOP(OP_GHBYADDR,XTERM);
4332 case KEY_gethostent:
4335 case KEY_getnetbyname:
4338 case KEY_getnetbyaddr:
4339 LOP(OP_GNBYADDR,XTERM);
4344 case KEY_getservbyname:
4345 LOP(OP_GSBYNAME,XTERM);
4347 case KEY_getservbyport:
4348 LOP(OP_GSBYPORT,XTERM);
4350 case KEY_getservent:
4353 case KEY_getsockname:
4354 UNI(OP_GETSOCKNAME);
4356 case KEY_getsockopt:
4357 LOP(OP_GSOCKOPT,XTERM);
4379 yylval.ival = CopLINE(PL_curcop);
4383 LOP(OP_INDEX,XTERM);
4389 LOP(OP_IOCTL,XTERM);
4401 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4433 LOP(OP_LISTEN,XTERM);
4442 s = scan_pat(s,OP_MATCH);
4443 TERM(sublex_start());
4446 LOP(OP_MAPSTART, XREF);
4449 LOP(OP_MKDIR,XTERM);
4452 LOP(OP_MSGCTL,XTERM);
4455 LOP(OP_MSGGET,XTERM);
4458 LOP(OP_MSGRCV,XTERM);
4461 LOP(OP_MSGSND,XTERM);
4467 if (isIDFIRST_lazy_if(s,UTF)) {
4468 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4469 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4471 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4472 if (!PL_in_my_stash) {
4475 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4483 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4490 if (PL_expect != XSTATE)
4491 yyerror("\"no\" not allowed in expression");
4492 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4493 s = force_version(s);
4498 if (*s == '(' || (s = skipspace(s), *s == '('))
4505 if (isIDFIRST_lazy_if(s,UTF)) {
4507 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4509 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4510 Perl_warner(aTHX_ WARN_PRECEDENCE,
4511 "Precedence problem: open %.*s should be open(%.*s)",
4517 yylval.ival = OP_OR;
4527 LOP(OP_OPEN_DIR,XTERM);
4530 checkcomma(s,PL_tokenbuf,"filehandle");
4534 checkcomma(s,PL_tokenbuf,"filehandle");
4553 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4557 LOP(OP_PIPE_OP,XTERM);
4560 s = scan_str(s,FALSE,FALSE);
4562 missingterm((char*)0);
4563 yylval.ival = OP_CONST;
4564 TERM(sublex_start());
4570 s = scan_str(s,FALSE,FALSE);
4572 missingterm((char*)0);
4574 if (SvCUR(PL_lex_stuff)) {
4577 d = SvPV_force(PL_lex_stuff, len);
4579 for (; isSPACE(*d) && len; --len, ++d) ;
4582 if (!warned && ckWARN(WARN_QW)) {
4583 for (; !isSPACE(*d) && len; --len, ++d) {
4585 Perl_warner(aTHX_ WARN_QW,
4586 "Possible attempt to separate words with commas");
4589 else if (*d == '#') {
4590 Perl_warner(aTHX_ WARN_QW,
4591 "Possible attempt to put comments in qw() list");
4597 for (; !isSPACE(*d) && len; --len, ++d) ;
4599 words = append_elem(OP_LIST, words,
4600 newSVOP(OP_CONST, 0, tokeq(newSVpvn(b, d-b))));
4604 PL_nextval[PL_nexttoke].opval = words;
4609 SvREFCNT_dec(PL_lex_stuff);
4610 PL_lex_stuff = Nullsv;
4615 s = scan_str(s,FALSE,FALSE);
4617 missingterm((char*)0);
4618 yylval.ival = OP_STRINGIFY;
4619 if (SvIVX(PL_lex_stuff) == '\'')
4620 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4621 TERM(sublex_start());
4624 s = scan_pat(s,OP_QR);
4625 TERM(sublex_start());
4628 s = scan_str(s,FALSE,FALSE);
4630 missingterm((char*)0);
4631 yylval.ival = OP_BACKTICK;
4633 TERM(sublex_start());
4640 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4641 s = force_version(s);
4644 *PL_tokenbuf = '\0';
4645 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4646 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4647 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4649 yyerror("<> should be quotes");
4657 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4661 LOP(OP_RENAME,XTERM);
4670 LOP(OP_RINDEX,XTERM);
4693 LOP(OP_REVERSE,XTERM);
4704 TERM(sublex_start());
4706 TOKEN(1); /* force error */
4715 LOP(OP_SELECT,XTERM);
4721 LOP(OP_SEMCTL,XTERM);
4724 LOP(OP_SEMGET,XTERM);
4727 LOP(OP_SEMOP,XTERM);
4733 LOP(OP_SETPGRP,XTERM);
4735 case KEY_setpriority:
4736 LOP(OP_SETPRIORITY,XTERM);
4738 case KEY_sethostent:
4744 case KEY_setservent:
4747 case KEY_setprotoent:
4757 LOP(OP_SEEKDIR,XTERM);
4759 case KEY_setsockopt:
4760 LOP(OP_SSOCKOPT,XTERM);
4766 LOP(OP_SHMCTL,XTERM);
4769 LOP(OP_SHMGET,XTERM);
4772 LOP(OP_SHMREAD,XTERM);
4775 LOP(OP_SHMWRITE,XTERM);
4778 LOP(OP_SHUTDOWN,XTERM);
4787 LOP(OP_SOCKET,XTERM);
4789 case KEY_socketpair:
4790 LOP(OP_SOCKPAIR,XTERM);
4793 checkcomma(s,PL_tokenbuf,"subroutine name");
4795 if (*s == ';' || *s == ')') /* probably a close */
4796 Perl_croak(aTHX_ "sort is now a reserved word");
4798 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4802 LOP(OP_SPLIT,XTERM);
4805 LOP(OP_SPRINTF,XTERM);
4808 LOP(OP_SPLICE,XTERM);
4823 LOP(OP_SUBSTR,XTERM);
4829 char tmpbuf[sizeof PL_tokenbuf];
4831 expectation attrful;
4832 bool have_name, have_proto;
4837 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4838 (*s == ':' && s[1] == ':'))
4841 attrful = XATTRBLOCK;
4842 /* remember buffer pos'n for later force_word */
4843 tboffset = s - PL_oldbufptr;
4844 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4845 if (strchr(tmpbuf, ':'))
4846 sv_setpv(PL_subname, tmpbuf);
4848 sv_setsv(PL_subname,PL_curstname);
4849 sv_catpvn(PL_subname,"::",2);
4850 sv_catpvn(PL_subname,tmpbuf,len);
4857 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4858 PL_expect = XTERMBLOCK;
4859 attrful = XATTRTERM;
4860 sv_setpv(PL_subname,"?");
4864 if (key == KEY_format) {
4866 PL_lex_formbrack = PL_lex_brackets + 1;
4868 (void) force_word(PL_oldbufptr + tboffset, WORD,
4873 /* Look for a prototype */
4877 s = scan_str(s,FALSE,FALSE);
4880 SvREFCNT_dec(PL_lex_stuff);
4881 PL_lex_stuff = Nullsv;
4882 Perl_croak(aTHX_ "Prototype not terminated");
4885 d = SvPVX(PL_lex_stuff);
4887 for (p = d; *p; ++p) {
4892 SvCUR(PL_lex_stuff) = tmp;
4900 if (*s == ':' && s[1] != ':')
4901 PL_expect = attrful;
4904 PL_nextval[PL_nexttoke].opval =
4905 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4906 PL_lex_stuff = Nullsv;
4910 sv_setpv(PL_subname,"__ANON__");
4913 (void) force_word(PL_oldbufptr + tboffset, WORD,
4922 LOP(OP_SYSTEM,XREF);
4925 LOP(OP_SYMLINK,XTERM);
4928 LOP(OP_SYSCALL,XTERM);
4931 LOP(OP_SYSOPEN,XTERM);
4934 LOP(OP_SYSSEEK,XTERM);
4937 LOP(OP_SYSREAD,XTERM);
4940 LOP(OP_SYSWRITE,XTERM);
4944 TERM(sublex_start());
4965 LOP(OP_TRUNCATE,XTERM);
4977 yylval.ival = CopLINE(PL_curcop);
4981 yylval.ival = CopLINE(PL_curcop);
4985 LOP(OP_UNLINK,XTERM);
4991 LOP(OP_UNPACK,XTERM);
4994 LOP(OP_UTIME,XTERM);
4997 if (ckWARN(WARN_UMASK)) {
4998 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4999 if (*d != '0' && isDIGIT(*d))
5000 Perl_warner(aTHX_ WARN_UMASK,
5001 "umask: argument is missing initial 0");
5006 LOP(OP_UNSHIFT,XTERM);
5009 if (PL_expect != XSTATE)
5010 yyerror("\"use\" not allowed in expression");
5012 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
5013 s = force_version(s);
5014 if (*s == ';' || (s = skipspace(s), *s == ';')) {
5015 PL_nextval[PL_nexttoke].opval = Nullop;
5020 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5021 s = force_version(s);
5033 yylval.ival = CopLINE(PL_curcop);
5037 PL_hints |= HINT_BLOCK_SCOPE;
5044 LOP(OP_WAITPID,XTERM);
5052 static char ctl_l[2];
5054 if (ctl_l[0] == '\0')
5055 ctl_l[0] = toCTRL('L');
5056 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5059 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5064 if (PL_expect == XOPERATOR)
5070 yylval.ival = OP_XOR;
5075 TERM(sublex_start());
5080 #pragma segment Main
5084 Perl_keyword(pTHX_ register char *d, I32 len)
5089 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5090 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5091 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5092 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5093 if (strEQ(d,"__END__")) return KEY___END__;
5097 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5102 if (strEQ(d,"and")) return -KEY_and;
5103 if (strEQ(d,"abs")) return -KEY_abs;
5106 if (strEQ(d,"alarm")) return -KEY_alarm;
5107 if (strEQ(d,"atan2")) return -KEY_atan2;
5110 if (strEQ(d,"accept")) return -KEY_accept;
5115 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5118 if (strEQ(d,"bless")) return -KEY_bless;
5119 if (strEQ(d,"bind")) return -KEY_bind;
5120 if (strEQ(d,"binmode")) return -KEY_binmode;
5123 if (strEQ(d,"CORE")) return -KEY_CORE;
5124 if (strEQ(d,"CHECK")) return KEY_CHECK;
5129 if (strEQ(d,"cmp")) return -KEY_cmp;
5130 if (strEQ(d,"chr")) return -KEY_chr;
5131 if (strEQ(d,"cos")) return -KEY_cos;
5134 if (strEQ(d,"chop")) return -KEY_chop;
5137 if (strEQ(d,"close")) return -KEY_close;
5138 if (strEQ(d,"chdir")) return -KEY_chdir;
5139 if (strEQ(d,"chomp")) return -KEY_chomp;
5140 if (strEQ(d,"chmod")) return -KEY_chmod;
5141 if (strEQ(d,"chown")) return -KEY_chown;
5142 if (strEQ(d,"crypt")) return -KEY_crypt;
5145 if (strEQ(d,"chroot")) return -KEY_chroot;
5146 if (strEQ(d,"caller")) return -KEY_caller;
5149 if (strEQ(d,"connect")) return -KEY_connect;
5152 if (strEQ(d,"closedir")) return -KEY_closedir;
5153 if (strEQ(d,"continue")) return -KEY_continue;
5158 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5163 if (strEQ(d,"do")) return KEY_do;
5166 if (strEQ(d,"die")) return -KEY_die;
5169 if (strEQ(d,"dump")) return -KEY_dump;
5172 if (strEQ(d,"delete")) return KEY_delete;
5175 if (strEQ(d,"defined")) return KEY_defined;
5176 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5179 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5184 if (strEQ(d,"END")) return KEY_END;
5189 if (strEQ(d,"eq")) return -KEY_eq;
5192 if (strEQ(d,"eof")) return -KEY_eof;
5193 if (strEQ(d,"exp")) return -KEY_exp;
5196 if (strEQ(d,"else")) return KEY_else;
5197 if (strEQ(d,"exit")) return -KEY_exit;
5198 if (strEQ(d,"eval")) return KEY_eval;
5199 if (strEQ(d,"exec")) return -KEY_exec;
5200 if (strEQ(d,"each")) return KEY_each;
5203 if (strEQ(d,"elsif")) return KEY_elsif;
5206 if (strEQ(d,"exists")) return KEY_exists;
5207 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5210 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5211 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5214 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5217 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5218 if (strEQ(d,"endservent")) return -KEY_endservent;
5221 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5228 if (strEQ(d,"for")) return KEY_for;
5231 if (strEQ(d,"fork")) return -KEY_fork;
5234 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5235 if (strEQ(d,"flock")) return -KEY_flock;
5238 if (strEQ(d,"format")) return KEY_format;
5239 if (strEQ(d,"fileno")) return -KEY_fileno;
5242 if (strEQ(d,"foreach")) return KEY_foreach;
5245 if (strEQ(d,"formline")) return -KEY_formline;
5250 if (strnEQ(d,"get",3)) {
5255 if (strEQ(d,"ppid")) return -KEY_getppid;
5256 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5259 if (strEQ(d,"pwent")) return -KEY_getpwent;
5260 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5261 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5264 if (strEQ(d,"peername")) return -KEY_getpeername;
5265 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5266 if (strEQ(d,"priority")) return -KEY_getpriority;
5269 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5272 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5276 else if (*d == 'h') {
5277 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5278 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5279 if (strEQ(d,"hostent")) return -KEY_gethostent;
5281 else if (*d == 'n') {
5282 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5283 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5284 if (strEQ(d,"netent")) return -KEY_getnetent;
5286 else if (*d == 's') {
5287 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5288 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5289 if (strEQ(d,"servent")) return -KEY_getservent;
5290 if (strEQ(d,"sockname")) return -KEY_getsockname;
5291 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5293 else if (*d == 'g') {
5294 if (strEQ(d,"grent")) return -KEY_getgrent;
5295 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5296 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5298 else if (*d == 'l') {
5299 if (strEQ(d,"login")) return -KEY_getlogin;
5301 else if (strEQ(d,"c")) return -KEY_getc;
5306 if (strEQ(d,"gt")) return -KEY_gt;
5307 if (strEQ(d,"ge")) return -KEY_ge;
5310 if (strEQ(d,"grep")) return KEY_grep;
5311 if (strEQ(d,"goto")) return KEY_goto;
5312 if (strEQ(d,"glob")) return KEY_glob;
5315 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5320 if (strEQ(d,"hex")) return -KEY_hex;
5323 if (strEQ(d,"INIT")) return KEY_INIT;
5328 if (strEQ(d,"if")) return KEY_if;
5331 if (strEQ(d,"int")) return -KEY_int;
5334 if (strEQ(d,"index")) return -KEY_index;
5335 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5340 if (strEQ(d,"join")) return -KEY_join;
5344 if (strEQ(d,"keys")) return KEY_keys;
5345 if (strEQ(d,"kill")) return -KEY_kill;
5351 if (strEQ(d,"lt")) return -KEY_lt;
5352 if (strEQ(d,"le")) return -KEY_le;
5353 if (strEQ(d,"lc")) return -KEY_lc;
5356 if (strEQ(d,"log")) return -KEY_log;
5359 if (strEQ(d,"last")) return KEY_last;
5360 if (strEQ(d,"link")) return -KEY_link;
5361 if (strEQ(d,"lock")) return -KEY_lock;
5364 if (strEQ(d,"local")) return KEY_local;
5365 if (strEQ(d,"lstat")) return -KEY_lstat;
5368 if (strEQ(d,"length")) return -KEY_length;
5369 if (strEQ(d,"listen")) return -KEY_listen;
5372 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5375 if (strEQ(d,"localtime")) return -KEY_localtime;
5381 case 1: return KEY_m;
5383 if (strEQ(d,"my")) return KEY_my;
5386 if (strEQ(d,"map")) return KEY_map;
5389 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5392 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5393 if (strEQ(d,"msgget")) return -KEY_msgget;
5394 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5395 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5400 if (strEQ(d,"next")) return KEY_next;
5401 if (strEQ(d,"ne")) return -KEY_ne;
5402 if (strEQ(d,"not")) return -KEY_not;
5403 if (strEQ(d,"no")) return KEY_no;
5408 if (strEQ(d,"or")) return -KEY_or;
5411 if (strEQ(d,"ord")) return -KEY_ord;
5412 if (strEQ(d,"oct")) return -KEY_oct;
5413 if (strEQ(d,"our")) return KEY_our;
5416 if (strEQ(d,"open")) return -KEY_open;
5419 if (strEQ(d,"opendir")) return -KEY_opendir;
5426 if (strEQ(d,"pop")) return KEY_pop;
5427 if (strEQ(d,"pos")) return KEY_pos;
5430 if (strEQ(d,"push")) return KEY_push;
5431 if (strEQ(d,"pack")) return -KEY_pack;
5432 if (strEQ(d,"pipe")) return -KEY_pipe;
5435 if (strEQ(d,"print")) return KEY_print;
5438 if (strEQ(d,"printf")) return KEY_printf;
5441 if (strEQ(d,"package")) return KEY_package;
5444 if (strEQ(d,"prototype")) return KEY_prototype;
5449 if (strEQ(d,"q")) return KEY_q;
5450 if (strEQ(d,"qr")) return KEY_qr;
5451 if (strEQ(d,"qq")) return KEY_qq;
5452 if (strEQ(d,"qw")) return KEY_qw;
5453 if (strEQ(d,"qx")) return KEY_qx;
5455 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5460 if (strEQ(d,"ref")) return -KEY_ref;
5463 if (strEQ(d,"read")) return -KEY_read;
5464 if (strEQ(d,"rand")) return -KEY_rand;
5465 if (strEQ(d,"recv")) return -KEY_recv;
5466 if (strEQ(d,"redo")) return KEY_redo;
5469 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5470 if (strEQ(d,"reset")) return -KEY_reset;
5473 if (strEQ(d,"return")) return KEY_return;
5474 if (strEQ(d,"rename")) return -KEY_rename;
5475 if (strEQ(d,"rindex")) return -KEY_rindex;
5478 if (strEQ(d,"require")) return -KEY_require;
5479 if (strEQ(d,"reverse")) return -KEY_reverse;
5480 if (strEQ(d,"readdir")) return -KEY_readdir;
5483 if (strEQ(d,"readlink")) return -KEY_readlink;
5484 if (strEQ(d,"readline")) return -KEY_readline;
5485 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5488 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5494 case 0: return KEY_s;
5496 if (strEQ(d,"scalar")) return KEY_scalar;
5501 if (strEQ(d,"seek")) return -KEY_seek;
5502 if (strEQ(d,"send")) return -KEY_send;
5505 if (strEQ(d,"semop")) return -KEY_semop;
5508 if (strEQ(d,"select")) return -KEY_select;
5509 if (strEQ(d,"semctl")) return -KEY_semctl;
5510 if (strEQ(d,"semget")) return -KEY_semget;
5513 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5514 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5517 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5518 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5521 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5524 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5525 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5526 if (strEQ(d,"setservent")) return -KEY_setservent;
5529 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5530 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5537 if (strEQ(d,"shift")) return KEY_shift;
5540 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5541 if (strEQ(d,"shmget")) return -KEY_shmget;
5544 if (strEQ(d,"shmread")) return -KEY_shmread;
5547 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5548 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5553 if (strEQ(d,"sin")) return -KEY_sin;
5556 if (strEQ(d,"sleep")) return -KEY_sleep;
5559 if (strEQ(d,"sort")) return KEY_sort;
5560 if (strEQ(d,"socket")) return -KEY_socket;
5561 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5564 if (strEQ(d,"split")) return KEY_split;
5565 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5566 if (strEQ(d,"splice")) return KEY_splice;
5569 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5572 if (strEQ(d,"srand")) return -KEY_srand;
5575 if (strEQ(d,"stat")) return -KEY_stat;
5576 if (strEQ(d,"study")) return KEY_study;
5579 if (strEQ(d,"substr")) return -KEY_substr;
5580 if (strEQ(d,"sub")) return KEY_sub;
5585 if (strEQ(d,"system")) return -KEY_system;
5588 if (strEQ(d,"symlink")) return -KEY_symlink;
5589 if (strEQ(d,"syscall")) return -KEY_syscall;
5590 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5591 if (strEQ(d,"sysread")) return -KEY_sysread;
5592 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5595 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5604 if (strEQ(d,"tr")) return KEY_tr;
5607 if (strEQ(d,"tie")) return KEY_tie;
5610 if (strEQ(d,"tell")) return -KEY_tell;
5611 if (strEQ(d,"tied")) return KEY_tied;
5612 if (strEQ(d,"time")) return -KEY_time;
5615 if (strEQ(d,"times")) return -KEY_times;
5618 if (strEQ(d,"telldir")) return -KEY_telldir;
5621 if (strEQ(d,"truncate")) return -KEY_truncate;
5628 if (strEQ(d,"uc")) return -KEY_uc;
5631 if (strEQ(d,"use")) return KEY_use;
5634 if (strEQ(d,"undef")) return KEY_undef;
5635 if (strEQ(d,"until")) return KEY_until;
5636 if (strEQ(d,"untie")) return KEY_untie;
5637 if (strEQ(d,"utime")) return -KEY_utime;
5638 if (strEQ(d,"umask")) return -KEY_umask;
5641 if (strEQ(d,"unless")) return KEY_unless;
5642 if (strEQ(d,"unpack")) return -KEY_unpack;
5643 if (strEQ(d,"unlink")) return -KEY_unlink;
5646 if (strEQ(d,"unshift")) return KEY_unshift;
5647 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5652 if (strEQ(d,"values")) return -KEY_values;
5653 if (strEQ(d,"vec")) return -KEY_vec;
5658 if (strEQ(d,"warn")) return -KEY_warn;
5659 if (strEQ(d,"wait")) return -KEY_wait;
5662 if (strEQ(d,"while")) return KEY_while;
5663 if (strEQ(d,"write")) return -KEY_write;
5666 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5669 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5674 if (len == 1) return -KEY_x;
5675 if (strEQ(d,"xor")) return -KEY_xor;
5678 if (len == 1) return KEY_y;
5687 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5691 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5692 dTHR; /* only for ckWARN */
5693 if (ckWARN(WARN_SYNTAX)) {
5695 for (w = s+2; *w && level; w++) {
5702 for (; *w && isSPACE(*w); w++) ;
5703 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5704 Perl_warner(aTHX_ WARN_SYNTAX,
5705 "%s (...) interpreted as function",name);
5708 while (s < PL_bufend && isSPACE(*s))
5712 while (s < PL_bufend && isSPACE(*s))
5714 if (isIDFIRST_lazy_if(s,UTF)) {
5716 while (isALNUM_lazy_if(s,UTF))
5718 while (s < PL_bufend && isSPACE(*s))
5723 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5727 Perl_croak(aTHX_ "No comma allowed after %s", what);
5732 /* Either returns sv, or mortalizes sv and returns a new SV*.
5733 Best used as sv=new_constant(..., sv, ...).
5734 If s, pv are NULL, calls subroutine with one argument,
5735 and type is used with error messages only. */
5738 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5742 HV *table = GvHV(PL_hintgv); /* ^H */
5746 const char *why1, *why2, *why3;
5748 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
5751 why1 = "%^H is not consistent";
5752 why2 = strEQ(key,"charnames")
5753 ? " (missing \"use charnames ...\"?)"
5757 msg = Perl_newSVpvf(aTHX_ "constant(%s): %s%s%s",
5758 (type ? type: "undef"), why1, why2, why3);
5759 yyerror(SvPVX(msg));
5763 cvp = hv_fetch(table, key, strlen(key), FALSE);
5764 if (!cvp || !SvOK(*cvp)) {
5767 why3 = "} is not defined";
5770 sv_2mortal(sv); /* Parent created it permanently */
5773 pv = sv_2mortal(newSVpvn(s, len));
5775 typesv = sv_2mortal(newSVpv(type, 0));
5777 typesv = &PL_sv_undef;
5779 PUSHSTACKi(PERLSI_OVERLOAD);
5791 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5795 /* Check the eval first */
5796 if (!PL_in_eval && SvTRUE(ERRSV)) {
5798 sv_catpv(ERRSV, "Propagated");
5799 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5801 res = SvREFCNT_inc(sv);
5805 (void)SvREFCNT_inc(res);
5814 why1 = "Call to &{$^H{";
5816 why3 = "}} did not return a defined value";
5825 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5827 register char *d = dest;
5828 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5831 Perl_croak(aTHX_ ident_too_long);
5832 if (isALNUM(*s)) /* UTF handled below */
5834 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
5839 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5843 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5844 char *t = s + UTF8SKIP(s);
5845 while (*t & 0x80 && is_utf8_mark((U8*)t))
5847 if (d + (t - s) > e)
5848 Perl_croak(aTHX_ ident_too_long);
5849 Copy(s, d, t - s, char);
5862 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5872 e = d + destlen - 3; /* two-character token, ending NUL */
5874 while (isDIGIT(*s)) {
5876 Perl_croak(aTHX_ ident_too_long);
5883 Perl_croak(aTHX_ ident_too_long);
5884 if (isALNUM(*s)) /* UTF handled below */
5886 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
5891 else if (*s == ':' && s[1] == ':') {
5895 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5896 char *t = s + UTF8SKIP(s);
5897 while (*t & 0x80 && is_utf8_mark((U8*)t))
5899 if (d + (t - s) > e)
5900 Perl_croak(aTHX_ ident_too_long);
5901 Copy(s, d, t - s, char);
5912 if (PL_lex_state != LEX_NORMAL)
5913 PL_lex_state = LEX_INTERPENDMAYBE;
5916 if (*s == '$' && s[1] &&
5917 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
5930 if (*d == '^' && *s && isCONTROLVAR(*s)) {
5935 if (isSPACE(s[-1])) {
5938 if (!SPACE_OR_TAB(ch)) {
5944 if (isIDFIRST_lazy_if(d,UTF)) {
5948 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
5950 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
5953 Copy(s, d, e - s, char);
5958 while ((isALNUM(*s) || *s == ':') && d < e)
5961 Perl_croak(aTHX_ ident_too_long);
5964 while (s < send && SPACE_OR_TAB(*s)) s++;
5965 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
5966 dTHR; /* only for ckWARN */
5967 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
5968 const char *brack = *s == '[' ? "[...]" : "{...}";
5969 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5970 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
5971 funny, dest, brack, funny, dest, brack);
5974 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
5978 /* Handle extended ${^Foo} variables
5979 * 1999-02-27 mjd-perl-patch@plover.com */
5980 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
5984 while (isALNUM(*s) && d < e) {
5988 Perl_croak(aTHX_ ident_too_long);
5993 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
5994 PL_lex_state = LEX_INTERPEND;
5997 if (PL_lex_state == LEX_NORMAL) {
5998 dTHR; /* only for ckWARN */
5999 if (ckWARN(WARN_AMBIGUOUS) &&
6000 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
6002 Perl_warner(aTHX_ WARN_AMBIGUOUS,
6003 "Ambiguous use of %c{%s} resolved to %c%s",
6004 funny, dest, funny, dest);
6009 s = bracket; /* let the parser handle it */
6013 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
6014 PL_lex_state = LEX_INTERPEND;
6019 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
6024 *pmfl |= PMf_GLOBAL;
6026 *pmfl |= PMf_CONTINUE;
6030 *pmfl |= PMf_MULTILINE;
6032 *pmfl |= PMf_SINGLELINE;
6034 *pmfl |= PMf_EXTENDED;
6038 S_scan_pat(pTHX_ char *start, I32 type)
6043 s = scan_str(start,FALSE,FALSE);
6046 SvREFCNT_dec(PL_lex_stuff);
6047 PL_lex_stuff = Nullsv;
6048 Perl_croak(aTHX_ "Search pattern not terminated");
6051 pm = (PMOP*)newPMOP(type, 0);
6052 if (PL_multi_open == '?')
6053 pm->op_pmflags |= PMf_ONCE;
6055 while (*s && strchr("iomsx", *s))
6056 pmflag(&pm->op_pmflags,*s++);
6059 while (*s && strchr("iogcmsx", *s))
6060 pmflag(&pm->op_pmflags,*s++);
6062 pm->op_pmpermflags = pm->op_pmflags;
6064 PL_lex_op = (OP*)pm;
6065 yylval.ival = OP_MATCH;
6070 S_scan_subst(pTHX_ char *start)
6077 yylval.ival = OP_NULL;
6079 s = scan_str(start,FALSE,FALSE);
6083 SvREFCNT_dec(PL_lex_stuff);
6084 PL_lex_stuff = Nullsv;
6085 Perl_croak(aTHX_ "Substitution pattern not terminated");
6088 if (s[-1] == PL_multi_open)
6091 first_start = PL_multi_start;
6092 s = scan_str(s,FALSE,FALSE);
6095 SvREFCNT_dec(PL_lex_stuff);
6096 PL_lex_stuff = Nullsv;
6098 SvREFCNT_dec(PL_lex_repl);
6099 PL_lex_repl = Nullsv;
6100 Perl_croak(aTHX_ "Substitution replacement not terminated");
6102 PL_multi_start = first_start; /* so whole substitution is taken together */
6104 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6110 else if (strchr("iogcmsx", *s))
6111 pmflag(&pm->op_pmflags,*s++);
6118 PL_sublex_info.super_bufptr = s;
6119 PL_sublex_info.super_bufend = PL_bufend;
6121 pm->op_pmflags |= PMf_EVAL;
6122 repl = newSVpvn("",0);
6124 sv_catpv(repl, es ? "eval " : "do ");
6125 sv_catpvn(repl, "{ ", 2);
6126 sv_catsv(repl, PL_lex_repl);
6127 sv_catpvn(repl, " };", 2);
6129 SvREFCNT_dec(PL_lex_repl);
6133 pm->op_pmpermflags = pm->op_pmflags;
6134 PL_lex_op = (OP*)pm;
6135 yylval.ival = OP_SUBST;
6140 S_scan_trans(pTHX_ char *start)
6151 yylval.ival = OP_NULL;
6153 s = scan_str(start,FALSE,FALSE);
6156 SvREFCNT_dec(PL_lex_stuff);
6157 PL_lex_stuff = Nullsv;
6158 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6160 if (s[-1] == PL_multi_open)
6163 s = scan_str(s,FALSE,FALSE);
6166 SvREFCNT_dec(PL_lex_stuff);
6167 PL_lex_stuff = Nullsv;
6169 SvREFCNT_dec(PL_lex_repl);
6170 PL_lex_repl = Nullsv;
6171 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6174 New(803,tbl,256,short);
6175 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6177 complement = del = squash = 0;
6178 while (strchr("cds", *s)) {
6180 complement = OPpTRANS_COMPLEMENT;
6182 del = OPpTRANS_DELETE;
6184 squash = OPpTRANS_SQUASH;
6187 o->op_private = del|squash|complement;
6190 yylval.ival = OP_TRANS;
6195 S_scan_heredoc(pTHX_ register char *s)
6199 I32 op_type = OP_SCALAR;
6206 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6210 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6213 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6214 if (*peek && strchr("`'\"",*peek)) {
6217 s = delimcpy(d, e, s, PL_bufend, term, &len);
6227 if (!isALNUM_lazy_if(s,UTF))
6228 deprecate("bare << to mean <<\"\"");
6229 for (; isALNUM_lazy_if(s,UTF); s++) {
6234 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6235 Perl_croak(aTHX_ "Delimiter for here document is too long");
6238 len = d - PL_tokenbuf;
6239 #ifndef PERL_STRICT_CR
6240 d = strchr(s, '\r');
6244 while (s < PL_bufend) {
6250 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6259 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6264 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6265 herewas = newSVpvn(s,PL_bufend-s);
6267 s--, herewas = newSVpvn(s,d-s);
6268 s += SvCUR(herewas);
6270 tmpstr = NEWSV(87,79);
6271 sv_upgrade(tmpstr, SVt_PVIV);
6276 else if (term == '`') {
6277 op_type = OP_BACKTICK;
6278 SvIVX(tmpstr) = '\\';
6282 PL_multi_start = CopLINE(PL_curcop);
6283 PL_multi_open = PL_multi_close = '<';
6284 term = *PL_tokenbuf;
6285 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6286 char *bufptr = PL_sublex_info.super_bufptr;
6287 char *bufend = PL_sublex_info.super_bufend;
6288 char *olds = s - SvCUR(herewas);
6289 s = strchr(bufptr, '\n');
6293 while (s < bufend &&
6294 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6296 CopLINE_inc(PL_curcop);
6299 CopLINE_set(PL_curcop, PL_multi_start);
6300 missingterm(PL_tokenbuf);
6302 sv_setpvn(herewas,bufptr,d-bufptr+1);
6303 sv_setpvn(tmpstr,d+1,s-d);
6305 sv_catpvn(herewas,s,bufend-s);
6306 (void)strcpy(bufptr,SvPVX(herewas));
6313 while (s < PL_bufend &&
6314 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6316 CopLINE_inc(PL_curcop);
6318 if (s >= PL_bufend) {
6319 CopLINE_set(PL_curcop, PL_multi_start);
6320 missingterm(PL_tokenbuf);
6322 sv_setpvn(tmpstr,d+1,s-d);
6324 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6326 sv_catpvn(herewas,s,PL_bufend-s);
6327 sv_setsv(PL_linestr,herewas);
6328 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6329 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6332 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6333 while (s >= PL_bufend) { /* multiple line string? */
6335 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6336 CopLINE_set(PL_curcop, PL_multi_start);
6337 missingterm(PL_tokenbuf);
6339 CopLINE_inc(PL_curcop);
6340 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6341 #ifndef PERL_STRICT_CR
6342 if (PL_bufend - PL_linestart >= 2) {
6343 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6344 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6346 PL_bufend[-2] = '\n';
6348 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6350 else if (PL_bufend[-1] == '\r')
6351 PL_bufend[-1] = '\n';
6353 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6354 PL_bufend[-1] = '\n';
6356 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6357 SV *sv = NEWSV(88,0);
6359 sv_upgrade(sv, SVt_PVMG);
6360 sv_setsv(sv,PL_linestr);
6361 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6363 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6366 sv_catsv(PL_linestr,herewas);
6367 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6371 sv_catsv(tmpstr,PL_linestr);
6376 PL_multi_end = CopLINE(PL_curcop);
6377 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6378 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6379 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6381 SvREFCNT_dec(herewas);
6382 PL_lex_stuff = tmpstr;
6383 yylval.ival = op_type;
6388 takes: current position in input buffer
6389 returns: new position in input buffer
6390 side-effects: yylval and lex_op are set.
6395 <FH> read from filehandle
6396 <pkg::FH> read from package qualified filehandle
6397 <pkg'FH> read from package qualified filehandle
6398 <$fh> read from filehandle in $fh
6404 S_scan_inputsymbol(pTHX_ char *start)
6406 register char *s = start; /* current position in buffer */
6412 d = PL_tokenbuf; /* start of temp holding space */
6413 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6414 end = strchr(s, '\n');
6417 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6419 /* die if we didn't have space for the contents of the <>,
6420 or if it didn't end, or if we see a newline
6423 if (len >= sizeof PL_tokenbuf)
6424 Perl_croak(aTHX_ "Excessively long <> operator");
6426 Perl_croak(aTHX_ "Unterminated <> operator");
6431 Remember, only scalar variables are interpreted as filehandles by
6432 this code. Anything more complex (e.g., <$fh{$num}>) will be
6433 treated as a glob() call.
6434 This code makes use of the fact that except for the $ at the front,
6435 a scalar variable and a filehandle look the same.
6437 if (*d == '$' && d[1]) d++;
6439 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6440 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6443 /* If we've tried to read what we allow filehandles to look like, and
6444 there's still text left, then it must be a glob() and not a getline.
6445 Use scan_str to pull out the stuff between the <> and treat it
6446 as nothing more than a string.
6449 if (d - PL_tokenbuf != len) {
6450 yylval.ival = OP_GLOB;
6452 s = scan_str(start,FALSE,FALSE);
6454 Perl_croak(aTHX_ "Glob not terminated");
6458 /* we're in a filehandle read situation */
6461 /* turn <> into <ARGV> */
6463 (void)strcpy(d,"ARGV");
6465 /* if <$fh>, create the ops to turn the variable into a
6471 /* try to find it in the pad for this block, otherwise find
6472 add symbol table ops
6474 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6475 OP *o = newOP(OP_PADSV, 0);
6477 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6480 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6481 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6482 newUNOP(OP_RV2SV, 0,
6483 newGVOP(OP_GV, 0, gv)));
6485 PL_lex_op->op_flags |= OPf_SPECIAL;
6486 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6487 yylval.ival = OP_NULL;
6490 /* If it's none of the above, it must be a literal filehandle
6491 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6493 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6494 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6495 yylval.ival = OP_NULL;
6504 takes: start position in buffer
6505 keep_quoted preserve \ on the embedded delimiter(s)
6506 keep_delims preserve the delimiters around the string
6507 returns: position to continue reading from buffer
6508 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6509 updates the read buffer.
6511 This subroutine pulls a string out of the input. It is called for:
6512 q single quotes q(literal text)
6513 ' single quotes 'literal text'
6514 qq double quotes qq(interpolate $here please)
6515 " double quotes "interpolate $here please"
6516 qx backticks qx(/bin/ls -l)
6517 ` backticks `/bin/ls -l`
6518 qw quote words @EXPORT_OK = qw( func() $spam )
6519 m// regexp match m/this/
6520 s/// regexp substitute s/this/that/
6521 tr/// string transliterate tr/this/that/
6522 y/// string transliterate y/this/that/
6523 ($*@) sub prototypes sub foo ($)
6524 (stuff) sub attr parameters sub foo : attr(stuff)
6525 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6527 In most of these cases (all but <>, patterns and transliterate)
6528 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6529 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6530 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6533 It skips whitespace before the string starts, and treats the first
6534 character as the delimiter. If the delimiter is one of ([{< then
6535 the corresponding "close" character )]}> is used as the closing
6536 delimiter. It allows quoting of delimiters, and if the string has
6537 balanced delimiters ([{<>}]) it allows nesting.
6539 The lexer always reads these strings into lex_stuff, except in the
6540 case of the operators which take *two* arguments (s/// and tr///)
6541 when it checks to see if lex_stuff is full (presumably with the 1st
6542 arg to s or tr) and if so puts the string into lex_repl.
6547 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6550 SV *sv; /* scalar value: string */
6551 char *tmps; /* temp string, used for delimiter matching */
6552 register char *s = start; /* current position in the buffer */
6553 register char term; /* terminating character */
6554 register char *to; /* current position in the sv's data */
6555 I32 brackets = 1; /* bracket nesting level */
6556 bool has_utf = FALSE; /* is there any utf8 content? */
6558 /* skip space before the delimiter */
6562 /* mark where we are, in case we need to report errors */
6565 /* after skipping whitespace, the next character is the terminator */
6567 if ((term & 0x80) && UTF)
6570 /* mark where we are */
6571 PL_multi_start = CopLINE(PL_curcop);
6572 PL_multi_open = term;
6574 /* find corresponding closing delimiter */
6575 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6577 PL_multi_close = term;
6579 /* create a new SV to hold the contents. 87 is leak category, I'm
6580 assuming. 79 is the SV's initial length. What a random number. */
6582 sv_upgrade(sv, SVt_PVIV);
6584 (void)SvPOK_only(sv); /* validate pointer */
6586 /* move past delimiter and try to read a complete string */
6588 sv_catpvn(sv, s, 1);
6591 /* extend sv if need be */
6592 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6593 /* set 'to' to the next character in the sv's string */
6594 to = SvPVX(sv)+SvCUR(sv);
6596 /* if open delimiter is the close delimiter read unbridle */
6597 if (PL_multi_open == PL_multi_close) {
6598 for (; s < PL_bufend; s++,to++) {
6599 /* embedded newlines increment the current line number */
6600 if (*s == '\n' && !PL_rsfp)
6601 CopLINE_inc(PL_curcop);
6602 /* handle quoted delimiters */
6603 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6604 if (!keep_quoted && s[1] == term)
6606 /* any other quotes are simply copied straight through */
6610 /* terminate when run out of buffer (the for() condition), or
6611 have found the terminator */
6612 else if (*s == term)
6614 else if (!has_utf && (*s & 0x80) && UTF)
6620 /* if the terminator isn't the same as the start character (e.g.,
6621 matched brackets), we have to allow more in the quoting, and
6622 be prepared for nested brackets.
6625 /* read until we run out of string, or we find the terminator */
6626 for (; s < PL_bufend; s++,to++) {
6627 /* embedded newlines increment the line count */
6628 if (*s == '\n' && !PL_rsfp)
6629 CopLINE_inc(PL_curcop);
6630 /* backslashes can escape the open or closing characters */
6631 if (*s == '\\' && s+1 < PL_bufend) {
6633 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6638 /* allow nested opens and closes */
6639 else if (*s == PL_multi_close && --brackets <= 0)
6641 else if (*s == PL_multi_open)
6643 else if (!has_utf && (*s & 0x80) && UTF)
6648 /* terminate the copied string and update the sv's end-of-string */
6650 SvCUR_set(sv, to - SvPVX(sv));
6653 * this next chunk reads more into the buffer if we're not done yet
6657 break; /* handle case where we are done yet :-) */
6659 #ifndef PERL_STRICT_CR
6660 if (to - SvPVX(sv) >= 2) {
6661 if ((to[-2] == '\r' && to[-1] == '\n') ||
6662 (to[-2] == '\n' && to[-1] == '\r'))
6666 SvCUR_set(sv, to - SvPVX(sv));
6668 else if (to[-1] == '\r')
6671 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6675 /* if we're out of file, or a read fails, bail and reset the current
6676 line marker so we can report where the unterminated string began
6679 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6681 CopLINE_set(PL_curcop, PL_multi_start);
6684 /* we read a line, so increment our line counter */
6685 CopLINE_inc(PL_curcop);
6687 /* update debugger info */
6688 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6689 SV *sv = NEWSV(88,0);
6691 sv_upgrade(sv, SVt_PVMG);
6692 sv_setsv(sv,PL_linestr);
6693 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6696 /* having changed the buffer, we must update PL_bufend */
6697 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6700 /* at this point, we have successfully read the delimited string */
6703 sv_catpvn(sv, s, 1);
6706 PL_multi_end = CopLINE(PL_curcop);
6709 /* if we allocated too much space, give some back */
6710 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6711 SvLEN_set(sv, SvCUR(sv) + 1);
6712 Renew(SvPVX(sv), SvLEN(sv), char);
6715 /* decide whether this is the first or second quoted string we've read
6728 takes: pointer to position in buffer
6729 returns: pointer to new position in buffer
6730 side-effects: builds ops for the constant in yylval.op
6732 Read a number in any of the formats that Perl accepts:
6734 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6735 [\d_]+(\.[\d_]*)?[Ee](\d+)
6737 Underbars (_) are allowed in decimal numbers. If -w is on,
6738 underbars before a decimal point must be at three digit intervals.
6740 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6743 If it reads a number without a decimal point or an exponent, it will
6744 try converting the number to an integer and see if it can do so
6745 without loss of precision.
6749 Perl_scan_num(pTHX_ char *start, YYSTYPE* lvalp)
6751 register char *s = start; /* current position in buffer */
6752 register char *d; /* destination in temp buffer */
6753 register char *e; /* end of temp buffer */
6754 NV nv; /* number read, as a double */
6755 SV *sv = Nullsv; /* place to put the converted number */
6756 bool floatit; /* boolean: int or float? */
6757 char *lastub = 0; /* position of last underbar */
6758 static char number_too_long[] = "Number too long";
6760 /* We use the first character to decide what type of number this is */
6764 Perl_croak(aTHX_ "panic: scan_num");
6766 /* if it starts with a 0, it could be an octal number, a decimal in
6767 0.13 disguise, or a hexadecimal number, or a binary number. */
6771 u holds the "number so far"
6772 shift the power of 2 of the base
6773 (hex == 4, octal == 3, binary == 1)
6774 overflowed was the number more than we can hold?
6776 Shift is used when we add a digit. It also serves as an "are
6777 we in octal/hex/binary?" indicator to disallow hex characters
6784 bool overflowed = FALSE;
6785 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6786 static char* bases[5] = { "", "binary", "", "octal",
6788 static char* Bases[5] = { "", "Binary", "", "Octal",
6790 static char *maxima[5] = { "",
6791 "0b11111111111111111111111111111111",
6795 char *base, *Base, *max;
6801 } else if (s[1] == 'b') {
6805 /* check for a decimal in disguise */
6806 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6808 /* so it must be octal */
6812 base = bases[shift];
6813 Base = Bases[shift];
6814 max = maxima[shift];
6816 /* read the rest of the number */
6818 /* x is used in the overflow test,
6819 b is the digit we're adding on. */
6824 /* if we don't mention it, we're done */
6833 /* 8 and 9 are not octal */
6836 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6840 case '2': case '3': case '4':
6841 case '5': case '6': case '7':
6843 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6847 b = *s++ & 15; /* ASCII digit -> value of digit */
6851 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6852 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6853 /* make sure they said 0x */
6858 /* Prepare to put the digit we have onto the end
6859 of the number so far. We check for overflows.
6864 x = u << shift; /* make room for the digit */
6866 if ((x >> shift) != u
6867 && !(PL_hints & HINT_NEW_BINARY)) {
6871 if (ckWARN_d(WARN_OVERFLOW))
6872 Perl_warner(aTHX_ WARN_OVERFLOW,
6873 "Integer overflow in %s number",
6876 u = x | b; /* add the digit to the end */
6879 n *= nvshift[shift];
6880 /* If an NV has not enough bits in its
6881 * mantissa to represent an UV this summing of
6882 * small low-order numbers is a waste of time
6883 * (because the NV cannot preserve the
6884 * low-order bits anyway): we could just
6885 * remember when did we overflow and in the
6886 * end just multiply n by the right
6894 /* if we get here, we had success: make a scalar value from
6901 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6902 Perl_warner(aTHX_ WARN_PORTABLE,
6903 "%s number > %s non-portable",
6910 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6911 Perl_warner(aTHX_ WARN_PORTABLE,
6912 "%s number > %s non-portable",
6917 if (PL_hints & HINT_NEW_BINARY)
6918 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6923 handle decimal numbers.
6924 we're also sent here when we read a 0 as the first digit
6926 case '1': case '2': case '3': case '4': case '5':
6927 case '6': case '7': case '8': case '9': case '.':
6930 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
6933 /* read next group of digits and _ and copy into d */
6934 while (isDIGIT(*s) || *s == '_') {
6935 /* skip underscores, checking for misplaced ones
6939 dTHR; /* only for ckWARN */
6940 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
6941 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6945 /* check for end of fixed-length buffer */
6947 Perl_croak(aTHX_ number_too_long);
6948 /* if we're ok, copy the character */
6953 /* final misplaced underbar check */
6954 if (lastub && s - lastub != 3) {
6956 if (ckWARN(WARN_SYNTAX))
6957 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6960 /* read a decimal portion if there is one. avoid
6961 3..5 being interpreted as the number 3. followed
6964 if (*s == '.' && s[1] != '.') {
6968 /* copy, ignoring underbars, until we run out of
6969 digits. Note: no misplaced underbar checks!
6971 for (; isDIGIT(*s) || *s == '_'; s++) {
6972 /* fixed length buffer check */
6974 Perl_croak(aTHX_ number_too_long);
6978 if (*s == '.' && isDIGIT(s[1])) {
6979 /* oops, it's really a v-string, but without the "v" */
6985 /* read exponent part, if present */
6986 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
6990 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
6991 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
6993 /* allow positive or negative exponent */
6994 if (*s == '+' || *s == '-')
6997 /* read digits of exponent (no underbars :-) */
6998 while (isDIGIT(*s)) {
7000 Perl_croak(aTHX_ number_too_long);
7005 /* terminate the string */
7008 /* make an sv from the string */
7011 #if defined(Strtol) && defined(Strtoul)
7014 strtol/strtoll sets errno to ERANGE if the number is too big
7015 for an integer. We try to do an integer conversion first
7016 if no characters indicating "float" have been found.
7023 if (*PL_tokenbuf == '-')
7024 iv = Strtol(PL_tokenbuf, (char**)NULL, 10);
7026 uv = Strtoul(PL_tokenbuf, (char**)NULL, 10);
7028 floatit = TRUE; /* Probably just too large. */
7029 else if (*PL_tokenbuf == '-')
7031 else if (uv <= IV_MAX)
7032 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7037 nv = Atof(PL_tokenbuf);
7042 No working strtou?ll?.
7044 Unfortunately atol() doesn't do range checks (returning
7045 LONG_MIN/LONG_MAX, and setting errno to ERANGE on overflows)
7046 everywhere [1], so we cannot use use atol() (or atoll()).
7047 If we could, they would be used, as Atol(), very much like
7048 Strtol() and Strtoul() are used above.
7050 [1] XXX Configure test needed to check for atol()
7051 (and atoll()) overflow behaviour XXX
7055 We need to do this the hard way. */
7057 nv = Atof(PL_tokenbuf);
7059 /* See if we can make do with an integer value without loss of
7060 precision. We use U_V to cast to a UV, because some
7061 compilers have issues. Then we try casting it back and see
7062 if it was the same [1]. We only do this if we know we
7063 specifically read an integer. If floatit is true, then we
7064 don't need to do the conversion at all.
7066 [1] Note that this is lossy if our NVs cannot preserve our
7067 UVs. There are metaconfig defines NV_PRESERVES_UV (a boolean)
7068 and NV_PRESERVES_UV_BITS (a number), but in general we really
7069 do hope all such potentially lossy platforms have strtou?ll?
7070 to do a lossless IV/UV conversion.
7072 Maybe could do some tricks with DBL_DIG, LDBL_DIG and
7073 DBL_MANT_DIG and LDBL_MANT_DIG (these are already available
7074 as NV_DIG and NV_MANT_DIG)?
7080 if (!floatit && (NV)uv == nv) {
7082 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7090 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7091 (PL_hints & HINT_NEW_INTEGER) )
7092 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7093 (floatit ? "float" : "integer"),
7097 /* if it starts with a v, it could be a v-string */
7103 while (isDIGIT(*pos) || *pos == '_')
7105 if (!isALPHA(*pos)) {
7107 U8 tmpbuf[UTF8_MAXLEN];
7110 s++; /* get past 'v' */
7113 sv_setpvn(sv, "", 0);
7116 if (*s == '0' && isDIGIT(s[1]))
7117 yyerror("Octal number in vector unsupported");
7120 /* this is atoi() that tolerates underscores */
7123 while (--end >= s) {
7128 rev += (*end - '0') * mult;
7130 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
7131 Perl_warner(aTHX_ WARN_OVERFLOW,
7132 "Integer overflow in decimal number");
7135 tmpend = uv_to_utf8(tmpbuf, rev);
7136 utf8 = utf8 || rev > 127;
7137 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
7138 if (*pos == '.' && isDIGIT(pos[1]))
7144 while (isDIGIT(*pos) || *pos == '_')
7152 sv_utf8_downgrade(sv, TRUE);
7159 /* make the op for the constant and return */
7162 lvalp->opval = newSVOP(OP_CONST, 0, sv);
7164 lvalp->opval = Nullop;
7170 S_scan_formline(pTHX_ register char *s)
7175 SV *stuff = newSVpvn("",0);
7176 bool needargs = FALSE;
7179 if (*s == '.' || *s == /*{*/'}') {
7181 #ifdef PERL_STRICT_CR
7182 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7184 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7186 if (*t == '\n' || t == PL_bufend)
7189 if (PL_in_eval && !PL_rsfp) {
7190 eol = strchr(s,'\n');
7195 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7197 for (t = s; t < eol; t++) {
7198 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7200 goto enough; /* ~~ must be first line in formline */
7202 if (*t == '@' || *t == '^')
7205 sv_catpvn(stuff, s, eol-s);
7206 #ifndef PERL_STRICT_CR
7207 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7208 char *end = SvPVX(stuff) + SvCUR(stuff);
7217 s = filter_gets(PL_linestr, PL_rsfp, 0);
7218 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7219 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7222 yyerror("Format not terminated");
7232 PL_lex_state = LEX_NORMAL;
7233 PL_nextval[PL_nexttoke].ival = 0;
7237 PL_lex_state = LEX_FORMLINE;
7238 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7240 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7244 SvREFCNT_dec(stuff);
7245 PL_lex_formbrack = 0;
7256 PL_cshlen = strlen(PL_cshname);
7261 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7264 I32 oldsavestack_ix = PL_savestack_ix;
7265 CV* outsidecv = PL_compcv;
7269 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7271 SAVEI32(PL_subline);
7272 save_item(PL_subname);
7275 SAVESPTR(PL_comppad_name);
7276 SAVESPTR(PL_compcv);
7277 SAVEI32(PL_comppad_name_fill);
7278 SAVEI32(PL_min_intro_pending);
7279 SAVEI32(PL_max_intro_pending);
7280 SAVEI32(PL_pad_reset_pending);
7282 PL_compcv = (CV*)NEWSV(1104,0);
7283 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7284 CvFLAGS(PL_compcv) |= flags;
7286 PL_comppad = newAV();
7287 av_push(PL_comppad, Nullsv);
7288 PL_curpad = AvARRAY(PL_comppad);
7289 PL_comppad_name = newAV();
7290 PL_comppad_name_fill = 0;
7291 PL_min_intro_pending = 0;
7293 PL_subline = CopLINE(PL_curcop);
7295 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7296 PL_curpad[0] = (SV*)newAV();
7297 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7298 #endif /* USE_THREADS */
7300 comppadlist = newAV();
7301 AvREAL_off(comppadlist);
7302 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7303 av_store(comppadlist, 1, (SV*)PL_comppad);
7305 CvPADLIST(PL_compcv) = comppadlist;
7306 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7308 CvOWNER(PL_compcv) = 0;
7309 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7310 MUTEX_INIT(CvMUTEXP(PL_compcv));
7311 #endif /* USE_THREADS */
7313 return oldsavestack_ix;
7317 Perl_yywarn(pTHX_ char *s)
7320 PL_in_eval |= EVAL_WARNONLY;
7322 PL_in_eval &= ~EVAL_WARNONLY;
7327 Perl_yyerror(pTHX_ char *s)
7331 char *context = NULL;
7335 if (!yychar || (yychar == ';' && !PL_rsfp))
7337 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7338 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7339 while (isSPACE(*PL_oldoldbufptr))
7341 context = PL_oldoldbufptr;
7342 contlen = PL_bufptr - PL_oldoldbufptr;
7344 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7345 PL_oldbufptr != PL_bufptr) {
7346 while (isSPACE(*PL_oldbufptr))
7348 context = PL_oldbufptr;
7349 contlen = PL_bufptr - PL_oldbufptr;
7351 else if (yychar > 255)
7352 where = "next token ???";
7353 #ifdef USE_PURE_BISON
7354 /* GNU Bison sets the value -2 */
7355 else if (yychar == -2) {
7357 else if ((yychar & 127) == 127) {
7359 if (PL_lex_state == LEX_NORMAL ||
7360 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7361 where = "at end of line";
7362 else if (PL_lex_inpat)
7363 where = "within pattern";
7365 where = "within string";
7368 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7370 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7371 else if (isPRINT_LC(yychar))
7372 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7374 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7375 where = SvPVX(where_sv);
7377 msg = sv_2mortal(newSVpv(s, 0));
7378 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7379 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7381 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7383 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7384 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7385 Perl_sv_catpvf(aTHX_ msg,
7386 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7387 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7390 if (PL_in_eval & EVAL_WARNONLY)
7391 Perl_warn(aTHX_ "%"SVf, msg);
7394 if (PL_error_count >= 10) {
7395 if (PL_in_eval && SvCUR(ERRSV))
7396 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
7397 ERRSV, CopFILE(PL_curcop));
7399 Perl_croak(aTHX_ "%s has too many errors.\n",
7400 CopFILE(PL_curcop));
7403 PL_in_my_stash = Nullhv;
7408 S_swallow_bom(pTHX_ U8 *s)
7411 slen = SvCUR(PL_linestr);
7415 /* UTF-16 little-endian */
7416 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
7417 Perl_croak(aTHX_ "Unsupported script encoding");
7418 #ifndef PERL_NO_UTF16_FILTER
7419 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-LE script encoding\n"));
7421 if (PL_bufend > (char*)s) {
7425 filter_add(utf16rev_textfilter, NULL);
7426 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7427 PL_bufend = (char*)utf16_to_utf8_reversed(s, news,
7428 PL_bufend - (char*)s - 1,
7430 Copy(news, s, newlen, U8);
7431 SvCUR_set(PL_linestr, newlen);
7432 PL_bufend = SvPVX(PL_linestr) + newlen;
7433 news[newlen++] = '\0';
7437 Perl_croak(aTHX_ "Unsupported script encoding");
7442 if (s[1] == 0xFF) { /* UTF-16 big-endian */
7443 #ifndef PERL_NO_UTF16_FILTER
7444 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding\n"));
7446 if (PL_bufend > (char *)s) {
7450 filter_add(utf16_textfilter, NULL);
7451 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7452 PL_bufend = (char*)utf16_to_utf8(s, news,
7453 PL_bufend - (char*)s,
7455 Copy(news, s, newlen, U8);
7456 SvCUR_set(PL_linestr, newlen);
7457 PL_bufend = SvPVX(PL_linestr) + newlen;
7458 news[newlen++] = '\0';
7462 Perl_croak(aTHX_ "Unsupported script encoding");
7467 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
7468 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-8 script encoding\n"));
7473 if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */
7474 s[2] == 0xFE && s[3] == 0xFF)
7476 Perl_croak(aTHX_ "Unsupported script encoding");
7488 * Restore a source filter.
7492 restore_rsfp(pTHXo_ void *f)
7494 PerlIO *fp = (PerlIO*)f;
7496 if (PL_rsfp == PerlIO_stdin())
7497 PerlIO_clearerr(PL_rsfp);
7498 else if (PL_rsfp && (PL_rsfp != fp))
7499 PerlIO_close(PL_rsfp);
7503 #ifndef PERL_NO_UTF16_FILTER
7505 utf16_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7507 I32 count = FILTER_READ(idx+1, sv, maxlen);
7512 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7513 if (!*SvPV_nolen(sv))
7514 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7517 tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7518 sv_usepvn(sv, (char*)tmps, tend - tmps);
7524 utf16rev_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7526 I32 count = FILTER_READ(idx+1, sv, maxlen);
7531 if (!*SvPV_nolen(sv))
7532 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7535 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7536 tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7537 sv_usepvn(sv, (char*)tmps, tend - tmps);