3 * Copyright (c) 1991-2000, Larry Wall
5 * You may distribute under the terms of either the GNU General Public
6 * License or the Artistic License, as specified in the README file.
11 * "It all comes from here, the stench and the peril." --Frodo
15 * This file is the lexer for Perl. It's closely linked to the
18 * The main routine is yylex(), which returns the next token.
22 #define PERL_IN_TOKE_C
25 #define yychar PL_yychar
26 #define yylval PL_yylval
28 static char ident_too_long[] = "Identifier too long";
30 static void restore_rsfp(pTHXo_ void *f);
31 #ifndef PERL_NO_UTF16_FILTER
32 static I32 utf16_textfilter(pTHXo_ int idx, SV *sv, int maxlen);
33 static I32 utf16rev_textfilter(pTHXo_ int idx, SV *sv, int maxlen);
36 #define XFAKEBRACK 128
39 /*#define UTF (SvUTF8(PL_linestr) && !(PL_hints & HINT_BYTE))*/
40 #define UTF (PL_hints & HINT_UTF8)
42 /* In variables name $^X, these are the legal values for X.
43 * 1999-02-27 mjd-perl-patch@plover.com */
44 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
46 /* On MacOS, respect nonbreaking spaces */
47 #ifdef MACOS_TRADITIONAL
48 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
50 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
53 /* LEX_* are values for PL_lex_state, the state of the lexer.
54 * They are arranged oddly so that the guard on the switch statement
55 * can get by with a single comparison (if the compiler is smart enough).
58 /* #define LEX_NOTPARSING 11 is done in perl.h. */
61 #define LEX_INTERPNORMAL 9
62 #define LEX_INTERPCASEMOD 8
63 #define LEX_INTERPPUSH 7
64 #define LEX_INTERPSTART 6
65 #define LEX_INTERPEND 5
66 #define LEX_INTERPENDMAYBE 4
67 #define LEX_INTERPCONCAT 3
68 #define LEX_INTERPCONST 2
69 #define LEX_FORMLINE 1
70 #define LEX_KNOWNEXT 0
72 /* XXX If this causes problems, set i_unistd=undef in the hint file. */
74 # include <unistd.h> /* Needed for execv() */
84 #define YYMAXLEVEL 100
86 YYSTYPE* yylval_pointer[YYMAXLEVEL];
87 int* yychar_pointer[YYMAXLEVEL];
91 # define yylval (*yylval_pointer[yyactlevel])
92 # define yychar (*yychar_pointer[yyactlevel])
93 # define PERL_YYLEX_PARAM yylval_pointer[yyactlevel],yychar_pointer[yyactlevel]
95 # define yylex() Perl_yylex(aTHX_ yylval_pointer[yyactlevel],yychar_pointer[yyactlevel])
100 /* CLINE is a macro that ensures PL_copline has a sane value */
105 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
108 * Convenience functions to return different tokens and prime the
109 * lexer for the next token. They all take an argument.
111 * TOKEN : generic token (used for '(', DOLSHARP, etc)
112 * OPERATOR : generic operator
113 * AOPERATOR : assignment operator
114 * PREBLOCK : beginning the block after an if, while, foreach, ...
115 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
116 * PREREF : *EXPR where EXPR is not a simple identifier
117 * TERM : expression term
118 * LOOPX : loop exiting command (goto, last, dump, etc)
119 * FTST : file test operator
120 * FUN0 : zero-argument function
121 * FUN1 : not used, except for not, which isn't a UNIOP
122 * BOop : bitwise or or xor
124 * SHop : shift operator
125 * PWop : power operator
126 * PMop : pattern-matching operator
127 * Aop : addition-level operator
128 * Mop : multiplication-level operator
129 * Eop : equality-testing operator
130 * Rop : relational operator <= != gt
132 * Also see LOP and lop() below.
135 #define TOKEN(retval) return (PL_bufptr = s,(int)retval)
136 #define OPERATOR(retval) return (PL_expect = XTERM,PL_bufptr = s,(int)retval)
137 #define AOPERATOR(retval) return ao((PL_expect = XTERM,PL_bufptr = s,(int)retval))
138 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
139 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
140 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s,(int)retval)
141 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR,PL_bufptr = s,(int)retval)
142 #define LOOPX(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
143 #define FTST(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)UNIOP)
144 #define FUN0(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
145 #define FUN1(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
146 #define BOop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
147 #define BAop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
148 #define SHop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
149 #define PWop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
150 #define PMop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
151 #define Aop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
152 #define Mop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
153 #define Eop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
154 #define Rop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)RELOP)
156 /* This bit of chicanery makes a unary function followed by
157 * a parenthesis into a function with one argument, highest precedence.
159 #define UNI(f) return(yylval.ival = f, \
162 PL_last_uni = PL_oldbufptr, \
163 PL_last_lop_op = f, \
164 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
166 #define UNIBRACK(f) return(yylval.ival = f, \
168 PL_last_uni = PL_oldbufptr, \
169 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
171 /* grandfather return to old style */
172 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
177 * This subroutine detects &&= and ||= and turns an ANDAND or OROR
178 * into an OP_ANDASSIGN or OP_ORASSIGN
182 S_ao(pTHX_ int toketype)
184 if (*PL_bufptr == '=') {
186 if (toketype == ANDAND)
187 yylval.ival = OP_ANDASSIGN;
188 else if (toketype == OROR)
189 yylval.ival = OP_ORASSIGN;
197 * When Perl expects an operator and finds something else, no_op
198 * prints the warning. It always prints "<something> found where
199 * operator expected. It prints "Missing semicolon on previous line?"
200 * if the surprise occurs at the start of the line. "do you need to
201 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
202 * where the compiler doesn't know if foo is a method call or a function.
203 * It prints "Missing operator before end of line" if there's nothing
204 * after the missing operator, or "... before <...>" if there is something
205 * after the missing operator.
209 S_no_op(pTHX_ char *what, char *s)
211 char *oldbp = PL_bufptr;
212 bool is_first = (PL_oldbufptr == PL_linestart);
218 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
220 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
221 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
223 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
224 if (t < PL_bufptr && isSPACE(*t))
225 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
226 t - PL_oldoldbufptr, PL_oldoldbufptr);
230 Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
237 * Complain about missing quote/regexp/heredoc terminator.
238 * If it's called with (char *)NULL then it cauterizes the line buffer.
239 * If we're in a delimited string and the delimiter is a control
240 * character, it's reformatted into a two-char sequence like ^C.
245 S_missingterm(pTHX_ char *s)
250 char *nl = strrchr(s,'\n');
256 iscntrl(PL_multi_close)
258 PL_multi_close < 32 || PL_multi_close == 127
262 tmpbuf[1] = toCTRL(PL_multi_close);
268 *tmpbuf = PL_multi_close;
272 q = strchr(s,'"') ? '\'' : '"';
273 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
281 Perl_deprecate(pTHX_ char *s)
284 if (ckWARN(WARN_DEPRECATED))
285 Perl_warner(aTHX_ WARN_DEPRECATED, "Use of %s is deprecated", s);
290 * Deprecate a comma-less variable list.
296 deprecate("comma-less variable list");
300 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
301 * utf16-to-utf8-reversed.
304 #ifdef PERL_CR_FILTER
308 register char *s = SvPVX(sv);
309 register char *e = s + SvCUR(sv);
310 /* outer loop optimized to do nothing if there are no CR-LFs */
312 if (*s++ == '\r' && *s == '\n') {
313 /* hit a CR-LF, need to copy the rest */
314 register char *d = s - 1;
317 if (*s == '\r' && s[1] == '\n')
328 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
330 I32 count = FILTER_READ(idx+1, sv, maxlen);
331 if (count > 0 && !maxlen)
339 * Initialize variables. Uses the Perl save_stack to save its state (for
340 * recursive calls to the parser).
344 Perl_lex_start(pTHX_ SV *line)
350 SAVEI32(PL_lex_dojoin);
351 SAVEI32(PL_lex_brackets);
352 SAVEI32(PL_lex_casemods);
353 SAVEI32(PL_lex_starts);
354 SAVEI32(PL_lex_state);
355 SAVEVPTR(PL_lex_inpat);
356 SAVEI32(PL_lex_inwhat);
357 if (PL_lex_state == LEX_KNOWNEXT) {
358 I32 toke = PL_nexttoke;
359 while (--toke >= 0) {
360 SAVEI32(PL_nexttype[toke]);
361 SAVEVPTR(PL_nextval[toke]);
363 SAVEI32(PL_nexttoke);
365 SAVECOPLINE(PL_curcop);
368 SAVEPPTR(PL_oldbufptr);
369 SAVEPPTR(PL_oldoldbufptr);
370 SAVEPPTR(PL_linestart);
371 SAVESPTR(PL_linestr);
372 SAVEPPTR(PL_lex_brackstack);
373 SAVEPPTR(PL_lex_casestack);
374 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
375 SAVESPTR(PL_lex_stuff);
376 SAVEI32(PL_lex_defer);
377 SAVEI32(PL_sublex_info.sub_inwhat);
378 SAVESPTR(PL_lex_repl);
380 SAVEINT(PL_lex_expect);
382 PL_lex_state = LEX_NORMAL;
386 New(899, PL_lex_brackstack, 120, char);
387 New(899, PL_lex_casestack, 12, char);
388 SAVEFREEPV(PL_lex_brackstack);
389 SAVEFREEPV(PL_lex_casestack);
391 *PL_lex_casestack = '\0';
394 PL_lex_stuff = Nullsv;
395 PL_lex_repl = Nullsv;
399 PL_sublex_info.sub_inwhat = 0;
401 if (SvREADONLY(PL_linestr))
402 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
403 s = SvPV(PL_linestr, len);
404 if (len && s[len-1] != ';') {
405 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
406 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
407 sv_catpvn(PL_linestr, "\n;", 2);
409 SvTEMP_off(PL_linestr);
410 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
411 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
413 PL_rs = newSVpvn("\n", 1);
419 * Finalizer for lexing operations. Must be called when the parser is
420 * done with the lexer.
426 PL_doextract = FALSE;
431 * This subroutine has nothing to do with tilting, whether at windmills
432 * or pinball tables. Its name is short for "increment line". It
433 * increments the current line number in CopLINE(PL_curcop) and checks
434 * to see whether the line starts with a comment of the form
435 * # line 500 "foo.pm"
436 * If so, it sets the current line number and file to the values in the comment.
440 S_incline(pTHX_ char *s)
448 CopLINE_inc(PL_curcop);
451 while (SPACE_OR_TAB(*s)) s++;
452 if (strnEQ(s, "line", 4))
456 if (*s == ' ' || *s == '\t')
460 while (SPACE_OR_TAB(*s)) s++;
466 while (SPACE_OR_TAB(*s))
468 if (*s == '"' && (t = strchr(s+1, '"'))) {
473 for (t = s; !isSPACE(*t); t++) ;
476 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
478 if (*e != '\n' && *e != '\0')
479 return; /* false alarm */
485 Safefree(CopFILE(PL_curcop));
487 SvREFCNT_dec(CopFILEGV(PL_curcop));
489 CopFILE_set(PL_curcop, s);
492 CopLINE_set(PL_curcop, atoi(n)-1);
497 * Called to gobble the appropriate amount and type of whitespace.
498 * Skips comments as well.
502 S_skipspace(pTHX_ register char *s)
505 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
506 while (s < PL_bufend && SPACE_OR_TAB(*s))
512 SSize_t oldprevlen, oldoldprevlen;
513 SSize_t oldloplen, oldunilen;
514 while (s < PL_bufend && isSPACE(*s)) {
515 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
520 if (s < PL_bufend && *s == '#') {
521 while (s < PL_bufend && *s != '\n')
525 if (PL_in_eval && !PL_rsfp) {
532 /* only continue to recharge the buffer if we're at the end
533 * of the buffer, we're not reading from a source filter, and
534 * we're in normal lexing mode
536 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
537 PL_lex_state == LEX_FORMLINE)
540 /* try to recharge the buffer */
541 if ((s = filter_gets(PL_linestr, PL_rsfp,
542 (prevlen = SvCUR(PL_linestr)))) == Nullch)
544 /* end of file. Add on the -p or -n magic */
545 if (PL_minus_n || PL_minus_p) {
546 sv_setpv(PL_linestr,PL_minus_p ?
547 ";}continue{print or die qq(-p destination: $!\\n)" :
549 sv_catpv(PL_linestr,";}");
550 PL_minus_n = PL_minus_p = 0;
553 sv_setpv(PL_linestr,";");
555 /* reset variables for next time we lex */
556 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
558 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
560 /* Close the filehandle. Could be from -P preprocessor,
561 * STDIN, or a regular file. If we were reading code from
562 * STDIN (because the commandline held no -e or filename)
563 * then we don't close it, we reset it so the code can
564 * read from STDIN too.
567 if (PL_preprocess && !PL_in_eval)
568 (void)PerlProc_pclose(PL_rsfp);
569 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
570 PerlIO_clearerr(PL_rsfp);
572 (void)PerlIO_close(PL_rsfp);
577 /* not at end of file, so we only read another line */
578 /* make corresponding updates to old pointers, for yyerror() */
579 oldprevlen = PL_oldbufptr - PL_bufend;
580 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
582 oldunilen = PL_last_uni - PL_bufend;
584 oldloplen = PL_last_lop - PL_bufend;
585 PL_linestart = PL_bufptr = s + prevlen;
586 PL_bufend = s + SvCUR(PL_linestr);
588 PL_oldbufptr = s + oldprevlen;
589 PL_oldoldbufptr = s + oldoldprevlen;
591 PL_last_uni = s + oldunilen;
593 PL_last_lop = s + oldloplen;
596 /* debugger active and we're not compiling the debugger code,
597 * so store the line into the debugger's array of lines
599 if (PERLDB_LINE && PL_curstash != PL_debstash) {
600 SV *sv = NEWSV(85,0);
602 sv_upgrade(sv, SVt_PVMG);
603 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
604 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
611 * Check the unary operators to ensure there's no ambiguity in how they're
612 * used. An ambiguous piece of code would be:
614 * This doesn't mean rand() + 5. Because rand() is a unary operator,
615 * the +5 is its argument.
625 if (PL_oldoldbufptr != PL_last_uni)
627 while (isSPACE(*PL_last_uni))
629 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
630 if ((t = strchr(s, '(')) && t < PL_bufptr)
632 if (ckWARN_d(WARN_AMBIGUOUS)){
635 Perl_warner(aTHX_ WARN_AMBIGUOUS,
636 "Warning: Use of \"%s\" without parens is ambiguous",
642 /* workaround to replace the UNI() macro with a function. Only the
643 * hints/uts.sh file mentions this. Other comments elsewhere in the
644 * source indicate Microport Unix might need it too.
650 #define UNI(f) return uni(f,s)
653 S_uni(pTHX_ I32 f, char *s)
658 PL_last_uni = PL_oldbufptr;
669 #endif /* CRIPPLED_CC */
672 * LOP : macro to build a list operator. Its behaviour has been replaced
673 * with a subroutine, S_lop() for which LOP is just another name.
676 #define LOP(f,x) return lop(f,x,s)
680 * Build a list operator (or something that might be one). The rules:
681 * - if we have a next token, then it's a list operator [why?]
682 * - if the next thing is an opening paren, then it's a function
683 * - else it's a list operator
687 S_lop(pTHX_ I32 f, int x, char *s)
694 PL_last_lop = PL_oldbufptr;
709 * When the lexer realizes it knows the next token (for instance,
710 * it is reordering tokens for the parser) then it can call S_force_next
711 * to know what token to return the next time the lexer is called. Caller
712 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
713 * handles the token correctly.
717 S_force_next(pTHX_ I32 type)
719 PL_nexttype[PL_nexttoke] = type;
721 if (PL_lex_state != LEX_KNOWNEXT) {
722 PL_lex_defer = PL_lex_state;
723 PL_lex_expect = PL_expect;
724 PL_lex_state = LEX_KNOWNEXT;
730 * When the lexer knows the next thing is a word (for instance, it has
731 * just seen -> and it knows that the next char is a word char, then
732 * it calls S_force_word to stick the next word into the PL_next lookahead.
735 * char *start : buffer position (must be within PL_linestr)
736 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
737 * int check_keyword : if true, Perl checks to make sure the word isn't
738 * a keyword (do this if the word is a label, e.g. goto FOO)
739 * int allow_pack : if true, : characters will also be allowed (require,
741 * int allow_initial_tick : used by the "sub" lexer only.
745 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
750 start = skipspace(start);
752 if (isIDFIRST_lazy_if(s,UTF) ||
753 (allow_pack && *s == ':') ||
754 (allow_initial_tick && *s == '\'') )
756 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
757 if (check_keyword && keyword(PL_tokenbuf, len))
759 if (token == METHOD) {
764 PL_expect = XOPERATOR;
767 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
768 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
776 * Called when the lexer wants $foo *foo &foo etc, but the program
777 * text only contains the "foo" portion. The first argument is a pointer
778 * to the "foo", and the second argument is the type symbol to prefix.
779 * Forces the next token to be a "WORD".
780 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
784 S_force_ident(pTHX_ register char *s, int kind)
787 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
788 PL_nextval[PL_nexttoke].opval = o;
791 dTHR; /* just for in_eval */
792 o->op_private = OPpCONST_ENTERED;
793 /* XXX see note in pp_entereval() for why we forgo typo
794 warnings if the symbol must be introduced in an eval.
796 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
797 kind == '$' ? SVt_PV :
798 kind == '@' ? SVt_PVAV :
799 kind == '%' ? SVt_PVHV :
807 Perl_str_to_version(pTHX_ SV *sv)
812 char *start = SvPVx(sv,len);
813 bool utf = SvUTF8(sv) ? TRUE : FALSE;
814 char *end = start + len;
815 while (start < end) {
819 n = utf8_to_uv_chk((U8*)start, &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 $1
1158 The structure of the code is
1159 while (there's a character to process) {
1160 handle transliteration ranges
1161 skip regexp comments
1162 skip # initiated comments in //x patterns
1163 check for embedded @foo
1164 check for embedded scalars
1166 leave intact backslashes from leave (below)
1167 deprecate \1 in strings and sub replacements
1168 handle string-changing backslashes \l \U \Q \E, etc.
1169 switch (what was escaped) {
1170 handle - in a transliteration (becomes a literal -)
1171 handle \132 octal characters
1172 handle 0x15 hex characters
1173 handle \cV (control V)
1174 handle printf backslashes (\f, \r, \n, etc)
1176 } (end if backslash)
1177 } (end while character to read)
1182 S_scan_const(pTHX_ char *start)
1184 register char *send = PL_bufend; /* end of the constant */
1185 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1186 register char *s = start; /* start of the constant */
1187 register char *d = SvPVX(sv); /* destination for copies */
1188 bool dorange = FALSE; /* are we in a translit range? */
1189 bool didrange = FALSE; /* did we just finish a range? */
1190 bool has_utf = FALSE; /* embedded \x{} */
1194 I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1195 ? (PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1197 I32 thisutf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1198 ? (PL_sublex_info.sub_op->op_private & (PL_lex_repl ?
1199 OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF))
1201 const char *leaveit = /* set of acceptably-backslashed characters */
1203 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1206 while (s < send || dorange) {
1207 /* get transliterations out of the way (they're most literal) */
1208 if (PL_lex_inwhat == OP_TRANS) {
1209 /* expand a range A-Z to the full set of characters. AIE! */
1211 I32 i; /* current expanded character */
1212 I32 min; /* first character in range */
1213 I32 max; /* last character in range */
1215 i = d - SvPVX(sv); /* remember current offset */
1216 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1217 d = SvPVX(sv) + i; /* refresh d after realloc */
1218 d -= 2; /* eat the first char and the - */
1220 min = (U8)*d; /* first char in range */
1221 max = (U8)d[1]; /* last char in range */
1225 "Invalid [] range \"%c-%c\" in transliteration operator",
1226 (char)min, (char)max);
1230 if ((isLOWER(min) && isLOWER(max)) ||
1231 (isUPPER(min) && isUPPER(max))) {
1233 for (i = min; i <= max; i++)
1237 for (i = min; i <= max; i++)
1244 for (i = min; i <= max; i++)
1247 /* mark the range as done, and continue */
1253 /* range begins (ignore - as first or last char) */
1254 else if (*s == '-' && s+1 < send && s != start) {
1256 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
1259 *d++ = (char)0xff; /* use illegal utf8 byte--see pmtrans */
1271 /* if we get here, we're not doing a transliteration */
1273 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1274 except for the last char, which will be done separately. */
1275 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1277 while (s < send && *s != ')')
1280 else if (s[2] == '{' /* This should match regcomp.c */
1281 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1284 char *regparse = s + (s[2] == '{' ? 3 : 4);
1287 while (count && (c = *regparse)) {
1288 if (c == '\\' && regparse[1])
1296 if (*regparse != ')') {
1297 regparse--; /* Leave one char for continuation. */
1298 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1300 while (s < regparse)
1305 /* likewise skip #-initiated comments in //x patterns */
1306 else if (*s == '#' && PL_lex_inpat &&
1307 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1308 while (s+1 < send && *s != '\n')
1312 /* check for embedded arrays
1313 (@foo, @:foo, @'foo, @{foo}, @$foo, @+, @-)
1315 else if (*s == '@' && s[1]
1316 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1])))
1319 /* check for embedded scalars. only stop if we're sure it's a
1322 else if (*s == '$') {
1323 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1325 if (s + 1 < send && !strchr("()| \n\t", s[1]))
1326 break; /* in regexp, $ might be tail anchor */
1329 /* (now in tr/// code again) */
1331 if (*s & 0x80 && thisutf) {
1332 (void)utf8_to_uv_chk((U8*)s, &len, 0);
1334 /* illegal UTF8, make it valid */
1335 char *old_pvx = SvPVX(sv);
1336 /* need space for one extra char (NOTE: SvCUR() not set here) */
1337 d = SvGROW(sv, SvLEN(sv) + 1) + (d - old_pvx);
1338 d = (char*)uv_to_utf8((U8*)d, (U8)*s++);
1349 if (*s == '\\' && s+1 < send) {
1352 /* some backslashes we leave behind */
1353 if (*leaveit && *s && strchr(leaveit, *s)) {
1359 /* deprecate \1 in strings and substitution replacements */
1360 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1361 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1363 dTHR; /* only for ckWARN */
1364 if (ckWARN(WARN_SYNTAX))
1365 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1370 /* string-change backslash escapes */
1371 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1376 /* if we get here, it's either a quoted -, or a digit */
1379 /* quoted - in transliterations */
1381 if (PL_lex_inwhat == OP_TRANS) {
1389 if (ckWARN(WARN_MISC) && isALNUM(*s))
1390 Perl_warner(aTHX_ WARN_MISC,
1391 "Unrecognized escape \\%c passed through",
1393 /* default action is to copy the quoted character */
1398 /* \132 indicates an octal constant */
1399 case '0': case '1': case '2': case '3':
1400 case '4': case '5': case '6': case '7':
1401 len = 0; /* disallow underscores */
1402 uv = (UV)scan_oct(s, 3, &len);
1404 goto NUM_ESCAPE_INSERT;
1406 /* \x24 indicates a hex constant */
1410 char* e = strchr(s, '}');
1412 yyerror("Missing right brace on \\x{}");
1415 len = 1; /* allow underscores */
1416 uv = (UV)scan_hex(s + 1, e - s - 1, &len);
1420 len = 0; /* disallow underscores */
1421 uv = (UV)scan_hex(s, 2, &len);
1426 /* Insert oct or hex escaped character.
1427 * There will always enough room in sv since such escapes will
1428 * be longer than any utf8 sequence they can end up as
1431 if (!thisutf && !has_utf && uv > 255) {
1432 /* might need to recode whatever we have accumulated so far
1433 * if it contains any hibit chars
1437 for (c = SvPVX(sv); c < d; c++) {
1442 char *old_pvx = SvPVX(sv);
1444 d = SvGROW(sv, SvCUR(sv) + hicount + 1) + (d - old_pvx);
1453 uv_to_utf8((U8*)dst, (U8)*src--);
1463 if (thisutf || uv > 255) {
1464 d = (char*)uv_to_utf8((U8*)d, uv);
1476 /* \N{latin small letter a} is a named character */
1480 char* e = strchr(s, '}');
1486 yyerror("Missing right brace on \\N{}");
1490 res = newSVpvn(s + 1, e - s - 1);
1491 res = new_constant( Nullch, 0, "charnames",
1492 res, Nullsv, "\\N{...}" );
1493 str = SvPV(res,len);
1494 if (!has_utf && SvUTF8(res)) {
1495 char *ostart = SvPVX(sv);
1496 SvCUR_set(sv, d - ostart);
1499 sv_utf8_upgrade(sv);
1500 /* this just broke our allocation above... */
1501 SvGROW(sv, send - start);
1502 d = SvPVX(sv) + SvCUR(sv);
1505 if (len > e - s + 4) {
1506 char *odest = SvPVX(sv);
1508 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1509 d = SvPVX(sv) + (d - odest);
1511 Copy(str, d, len, char);
1518 yyerror("Missing braces on \\N{}");
1521 /* \c is a control character */
1536 /* printf-style backslashes, formfeeds, newlines, etc */
1554 *d++ = '\047'; /* CP 1047 */
1557 *d++ = '\057'; /* CP 1047 */
1571 } /* end if (backslash) */
1574 } /* while loop to process each character */
1576 /* terminate the string and set up the sv */
1578 SvCUR_set(sv, d - SvPVX(sv));
1583 /* shrink the sv if we allocated more than we used */
1584 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1585 SvLEN_set(sv, SvCUR(sv) + 1);
1586 Renew(SvPVX(sv), SvLEN(sv), char);
1589 /* return the substring (via yylval) only if we parsed anything */
1590 if (s > PL_bufptr) {
1591 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1592 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1594 ( PL_lex_inwhat == OP_TRANS
1596 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1599 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1606 * Returns TRUE if there's more to the expression (e.g., a subscript),
1609 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1611 * ->[ and ->{ return TRUE
1612 * { and [ outside a pattern are always subscripts, so return TRUE
1613 * if we're outside a pattern and it's not { or [, then return FALSE
1614 * if we're in a pattern and the first char is a {
1615 * {4,5} (any digits around the comma) returns FALSE
1616 * if we're in a pattern and the first char is a [
1618 * [SOMETHING] has a funky algorithm to decide whether it's a
1619 * character class or not. It has to deal with things like
1620 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1621 * anything else returns TRUE
1624 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1627 S_intuit_more(pTHX_ register char *s)
1629 if (PL_lex_brackets)
1631 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1633 if (*s != '{' && *s != '[')
1638 /* In a pattern, so maybe we have {n,m}. */
1655 /* On the other hand, maybe we have a character class */
1658 if (*s == ']' || *s == '^')
1661 /* this is terrifying, and it works */
1662 int weight = 2; /* let's weigh the evidence */
1664 unsigned char un_char = 255, last_un_char;
1665 char *send = strchr(s,']');
1666 char tmpbuf[sizeof PL_tokenbuf * 4];
1668 if (!send) /* has to be an expression */
1671 Zero(seen,256,char);
1674 else if (isDIGIT(*s)) {
1676 if (isDIGIT(s[1]) && s[2] == ']')
1682 for (; s < send; s++) {
1683 last_un_char = un_char;
1684 un_char = (unsigned char)*s;
1689 weight -= seen[un_char] * 10;
1690 if (isALNUM_lazy_if(s+1,UTF)) {
1691 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1692 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1697 else if (*s == '$' && s[1] &&
1698 strchr("[#!%*<>()-=",s[1])) {
1699 if (/*{*/ strchr("])} =",s[2]))
1708 if (strchr("wds]",s[1]))
1710 else if (seen['\''] || seen['"'])
1712 else if (strchr("rnftbxcav",s[1]))
1714 else if (isDIGIT(s[1])) {
1716 while (s[1] && isDIGIT(s[1]))
1726 if (strchr("aA01! ",last_un_char))
1728 if (strchr("zZ79~",s[1]))
1730 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1731 weight -= 5; /* cope with negative subscript */
1734 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1735 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1740 if (keyword(tmpbuf, d - tmpbuf))
1743 if (un_char == last_un_char + 1)
1745 weight -= seen[un_char];
1750 if (weight >= 0) /* probably a character class */
1760 * Does all the checking to disambiguate
1762 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1763 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1765 * First argument is the stuff after the first token, e.g. "bar".
1767 * Not a method if bar is a filehandle.
1768 * Not a method if foo is a subroutine prototyped to take a filehandle.
1769 * Not a method if it's really "Foo $bar"
1770 * Method if it's "foo $bar"
1771 * Not a method if it's really "print foo $bar"
1772 * Method if it's really "foo package::" (interpreted as package->foo)
1773 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1774 * Not a method if bar is a filehandle or package, but is quoted with
1779 S_intuit_method(pTHX_ char *start, GV *gv)
1781 char *s = start + (*start == '$');
1782 char tmpbuf[sizeof PL_tokenbuf];
1790 if ((cv = GvCVu(gv))) {
1791 char *proto = SvPVX(cv);
1801 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1802 /* start is the beginning of the possible filehandle/object,
1803 * and s is the end of it
1804 * tmpbuf is a copy of it
1807 if (*start == '$') {
1808 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1813 return *s == '(' ? FUNCMETH : METHOD;
1815 if (!keyword(tmpbuf, len)) {
1816 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1821 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1822 if (indirgv && GvCVu(indirgv))
1824 /* filehandle or package name makes it a method */
1825 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1827 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1828 return 0; /* no assumptions -- "=>" quotes bearword */
1830 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1831 newSVpvn(tmpbuf,len));
1832 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1836 return *s == '(' ? FUNCMETH : METHOD;
1844 * Return a string of Perl code to load the debugger. If PERL5DB
1845 * is set, it will return the contents of that, otherwise a
1846 * compile-time require of perl5db.pl.
1853 char *pdb = PerlEnv_getenv("PERL5DB");
1857 SETERRNO(0,SS$_NORMAL);
1858 return "BEGIN { require 'perl5db.pl' }";
1864 /* Encoded script support. filter_add() effectively inserts a
1865 * 'pre-processing' function into the current source input stream.
1866 * Note that the filter function only applies to the current source file
1867 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1869 * The datasv parameter (which may be NULL) can be used to pass
1870 * private data to this instance of the filter. The filter function
1871 * can recover the SV using the FILTER_DATA macro and use it to
1872 * store private buffers and state information.
1874 * The supplied datasv parameter is upgraded to a PVIO type
1875 * and the IoDIRP/IoANY field is used to store the function pointer,
1876 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1877 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1878 * private use must be set using malloc'd pointers.
1882 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1887 if (!PL_rsfp_filters)
1888 PL_rsfp_filters = newAV();
1890 datasv = NEWSV(255,0);
1891 if (!SvUPGRADE(datasv, SVt_PVIO))
1892 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1893 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
1894 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1895 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1896 funcp, SvPV_nolen(datasv)));
1897 av_unshift(PL_rsfp_filters, 1);
1898 av_store(PL_rsfp_filters, 0, datasv) ;
1903 /* Delete most recently added instance of this filter function. */
1905 Perl_filter_del(pTHX_ filter_t funcp)
1908 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1909 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1911 /* if filter is on top of stack (usual case) just pop it off */
1912 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1913 if (IoANY(datasv) == (void *)funcp) {
1914 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1915 IoANY(datasv) = (void *)NULL;
1916 sv_free(av_pop(PL_rsfp_filters));
1920 /* we need to search for the correct entry and clear it */
1921 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1925 /* Invoke the n'th filter function for the current rsfp. */
1927 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1930 /* 0 = read one text line */
1935 if (!PL_rsfp_filters)
1937 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1938 /* Provide a default input filter to make life easy. */
1939 /* Note that we append to the line. This is handy. */
1940 DEBUG_P(PerlIO_printf(Perl_debug_log,
1941 "filter_read %d: from rsfp\n", idx));
1945 int old_len = SvCUR(buf_sv) ;
1947 /* ensure buf_sv is large enough */
1948 SvGROW(buf_sv, old_len + maxlen) ;
1949 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1950 if (PerlIO_error(PL_rsfp))
1951 return -1; /* error */
1953 return 0 ; /* end of file */
1955 SvCUR_set(buf_sv, old_len + len) ;
1958 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1959 if (PerlIO_error(PL_rsfp))
1960 return -1; /* error */
1962 return 0 ; /* end of file */
1965 return SvCUR(buf_sv);
1967 /* Skip this filter slot if filter has been deleted */
1968 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1969 DEBUG_P(PerlIO_printf(Perl_debug_log,
1970 "filter_read %d: skipped (filter deleted)\n",
1972 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1974 /* Get function pointer hidden within datasv */
1975 funcp = (filter_t)IoANY(datasv);
1976 DEBUG_P(PerlIO_printf(Perl_debug_log,
1977 "filter_read %d: via function %p (%s)\n",
1978 idx, funcp, SvPV_nolen(datasv)));
1979 /* Call function. The function is expected to */
1980 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1981 /* Return: <0:error, =0:eof, >0:not eof */
1982 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1986 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1988 #ifdef PERL_CR_FILTER
1989 if (!PL_rsfp_filters) {
1990 filter_add(S_cr_textfilter,NULL);
1993 if (PL_rsfp_filters) {
1996 SvCUR_set(sv, 0); /* start with empty line */
1997 if (FILTER_READ(0, sv, 0) > 0)
1998 return ( SvPVX(sv) ) ;
2003 return (sv_gets(sv, fp, append));
2007 S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2011 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2015 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2016 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2018 return GvHV(gv); /* Foo:: */
2021 /* use constant CLASS => 'MyClass' */
2022 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2024 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2025 pkgname = SvPV_nolen(sv);
2029 return gv_stashpv(pkgname, FALSE);
2033 static char* exp_name[] =
2034 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2035 "ATTRTERM", "TERMBLOCK"
2042 Works out what to call the token just pulled out of the input
2043 stream. The yacc parser takes care of taking the ops we return and
2044 stitching them into a tree.
2050 if read an identifier
2051 if we're in a my declaration
2052 croak if they tried to say my($foo::bar)
2053 build the ops for a my() declaration
2054 if it's an access to a my() variable
2055 are we in a sort block?
2056 croak if my($a); $a <=> $b
2057 build ops for access to a my() variable
2058 if in a dq string, and they've said @foo and we can't find @foo
2060 build ops for a bareword
2061 if we already built the token before, use it.
2065 #pragma segment Perl_yylex
2068 #ifdef USE_PURE_BISON
2069 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
2077 #ifdef USE_PURE_BISON
2078 yylval_pointer[yyactlevel] = lvalp;
2079 yychar_pointer[yyactlevel] = lcharp;
2081 if (yyactlevel >= YYMAXLEVEL)
2082 Perl_croak(aTHX_ "panic: YYMAXLEVEL");
2087 #ifdef USE_PURE_BISON
2095 S_syylex(pTHX) /* need to be separate from yylex for reentrancy */
2105 /* check if there's an identifier for us to look at */
2106 if (PL_pending_ident) {
2107 /* pit holds the identifier we read and pending_ident is reset */
2108 char pit = PL_pending_ident;
2109 PL_pending_ident = 0;
2111 /* if we're in a my(), we can't allow dynamics here.
2112 $foo'bar has already been turned into $foo::bar, so
2113 just check for colons.
2115 if it's a legal name, the OP is a PADANY.
2118 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
2119 if (strchr(PL_tokenbuf,':'))
2120 yyerror(Perl_form(aTHX_ "No package name allowed for "
2121 "variable %s in \"our\"",
2123 tmp = pad_allocmy(PL_tokenbuf);
2126 if (strchr(PL_tokenbuf,':'))
2127 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
2129 yylval.opval = newOP(OP_PADANY, 0);
2130 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
2136 build the ops for accesses to a my() variable.
2138 Deny my($a) or my($b) in a sort block, *if* $a or $b is
2139 then used in a comparison. This catches most, but not
2140 all cases. For instance, it catches
2141 sort { my($a); $a <=> $b }
2143 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2144 (although why you'd do that is anyone's guess).
2147 if (!strchr(PL_tokenbuf,':')) {
2149 /* Check for single character per-thread SVs */
2150 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2151 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2152 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2154 yylval.opval = newOP(OP_THREADSV, 0);
2155 yylval.opval->op_targ = tmp;
2158 #endif /* USE_THREADS */
2159 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2160 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2161 /* might be an "our" variable" */
2162 if (SvFLAGS(namesv) & SVpad_OUR) {
2163 /* build ops for a bareword */
2164 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2165 sv_catpvn(sym, "::", 2);
2166 sv_catpv(sym, PL_tokenbuf+1);
2167 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2168 yylval.opval->op_private = OPpCONST_ENTERED;
2169 gv_fetchpv(SvPVX(sym),
2171 ? (GV_ADDMULTI | GV_ADDINEVAL)
2174 ((PL_tokenbuf[0] == '$') ? SVt_PV
2175 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2180 /* if it's a sort block and they're naming $a or $b */
2181 if (PL_last_lop_op == OP_SORT &&
2182 PL_tokenbuf[0] == '$' &&
2183 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2186 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2187 d < PL_bufend && *d != '\n';
2190 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2191 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2197 yylval.opval = newOP(OP_PADANY, 0);
2198 yylval.opval->op_targ = tmp;
2204 Whine if they've said @foo in a doublequoted string,
2205 and @foo isn't a variable we can find in the symbol
2208 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2209 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2210 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2211 && ckWARN(WARN_AMBIGUOUS))
2213 /* Downgraded from fatal to warning 20000522 mjd */
2214 Perl_warner(aTHX_ WARN_AMBIGUOUS,
2215 "Possible unintended interpolation of %s in string",
2220 /* build ops for a bareword */
2221 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2222 yylval.opval->op_private = OPpCONST_ENTERED;
2223 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2224 ((PL_tokenbuf[0] == '$') ? SVt_PV
2225 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2230 /* no identifier pending identification */
2232 switch (PL_lex_state) {
2234 case LEX_NORMAL: /* Some compilers will produce faster */
2235 case LEX_INTERPNORMAL: /* code if we comment these out. */
2239 /* when we've already built the next token, just pull it out of the queue */
2242 yylval = PL_nextval[PL_nexttoke];
2244 PL_lex_state = PL_lex_defer;
2245 PL_expect = PL_lex_expect;
2246 PL_lex_defer = LEX_NORMAL;
2248 return(PL_nexttype[PL_nexttoke]);
2250 /* interpolated case modifiers like \L \U, including \Q and \E.
2251 when we get here, PL_bufptr is at the \
2253 case LEX_INTERPCASEMOD:
2255 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2256 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2258 /* handle \E or end of string */
2259 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2263 if (PL_lex_casemods) {
2264 oldmod = PL_lex_casestack[--PL_lex_casemods];
2265 PL_lex_casestack[PL_lex_casemods] = '\0';
2267 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2269 PL_lex_state = LEX_INTERPCONCAT;
2273 if (PL_bufptr != PL_bufend)
2275 PL_lex_state = LEX_INTERPCONCAT;
2280 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2281 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2282 if (strchr("LU", *s) &&
2283 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2285 PL_lex_casestack[--PL_lex_casemods] = '\0';
2288 if (PL_lex_casemods > 10) {
2289 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2290 if (newlb != PL_lex_casestack) {
2292 PL_lex_casestack = newlb;
2295 PL_lex_casestack[PL_lex_casemods++] = *s;
2296 PL_lex_casestack[PL_lex_casemods] = '\0';
2297 PL_lex_state = LEX_INTERPCONCAT;
2298 PL_nextval[PL_nexttoke].ival = 0;
2301 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2303 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2305 PL_nextval[PL_nexttoke].ival = OP_LC;
2307 PL_nextval[PL_nexttoke].ival = OP_UC;
2309 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2311 Perl_croak(aTHX_ "panic: yylex");
2314 if (PL_lex_starts) {
2323 case LEX_INTERPPUSH:
2324 return sublex_push();
2326 case LEX_INTERPSTART:
2327 if (PL_bufptr == PL_bufend)
2328 return sublex_done();
2330 PL_lex_dojoin = (*PL_bufptr == '@');
2331 PL_lex_state = LEX_INTERPNORMAL;
2332 if (PL_lex_dojoin) {
2333 PL_nextval[PL_nexttoke].ival = 0;
2336 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2337 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2338 force_next(PRIVATEREF);
2340 force_ident("\"", '$');
2341 #endif /* USE_THREADS */
2342 PL_nextval[PL_nexttoke].ival = 0;
2344 PL_nextval[PL_nexttoke].ival = 0;
2346 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2349 if (PL_lex_starts++) {
2355 case LEX_INTERPENDMAYBE:
2356 if (intuit_more(PL_bufptr)) {
2357 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2363 if (PL_lex_dojoin) {
2364 PL_lex_dojoin = FALSE;
2365 PL_lex_state = LEX_INTERPCONCAT;
2368 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2369 && SvEVALED(PL_lex_repl))
2371 if (PL_bufptr != PL_bufend)
2372 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2373 PL_lex_repl = Nullsv;
2376 case LEX_INTERPCONCAT:
2378 if (PL_lex_brackets)
2379 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2381 if (PL_bufptr == PL_bufend)
2382 return sublex_done();
2384 if (SvIVX(PL_linestr) == '\'') {
2385 SV *sv = newSVsv(PL_linestr);
2388 else if ( PL_hints & HINT_NEW_RE )
2389 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2390 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2394 s = scan_const(PL_bufptr);
2396 PL_lex_state = LEX_INTERPCASEMOD;
2398 PL_lex_state = LEX_INTERPSTART;
2401 if (s != PL_bufptr) {
2402 PL_nextval[PL_nexttoke] = yylval;
2405 if (PL_lex_starts++)
2415 PL_lex_state = LEX_NORMAL;
2416 s = scan_formline(PL_bufptr);
2417 if (!PL_lex_formbrack)
2423 PL_oldoldbufptr = PL_oldbufptr;
2426 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2427 exp_name[PL_expect], s);
2433 if (isIDFIRST_lazy_if(s,UTF))
2435 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2438 goto fake_eof; /* emulate EOF on ^D or ^Z */
2443 if (PL_lex_brackets)
2444 yyerror("Missing right curly or square bracket");
2447 if (s++ < PL_bufend)
2448 goto retry; /* ignore stray nulls */
2451 if (!PL_in_eval && !PL_preambled) {
2452 PL_preambled = TRUE;
2453 sv_setpv(PL_linestr,incl_perldb());
2454 if (SvCUR(PL_linestr))
2455 sv_catpv(PL_linestr,";");
2457 while(AvFILLp(PL_preambleav) >= 0) {
2458 SV *tmpsv = av_shift(PL_preambleav);
2459 sv_catsv(PL_linestr, tmpsv);
2460 sv_catpv(PL_linestr, ";");
2463 sv_free((SV*)PL_preambleav);
2464 PL_preambleav = NULL;
2466 if (PL_minus_n || PL_minus_p) {
2467 sv_catpv(PL_linestr, "LINE: while (<>) {");
2469 sv_catpv(PL_linestr,"chomp;");
2471 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2473 GvIMPORTED_AV_on(gv);
2475 if (strchr("/'\"", *PL_splitstr)
2476 && strchr(PL_splitstr + 1, *PL_splitstr))
2477 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2480 s = "'~#\200\1'"; /* surely one char is unused...*/
2481 while (s[1] && strchr(PL_splitstr, *s)) s++;
2483 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2484 "q" + (delim == '\''), delim);
2485 for (s = PL_splitstr; *s; s++) {
2487 sv_catpvn(PL_linestr, "\\", 1);
2488 sv_catpvn(PL_linestr, s, 1);
2490 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2494 sv_catpv(PL_linestr,"@F=split(' ');");
2497 sv_catpv(PL_linestr, "\n");
2498 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2499 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2500 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2501 SV *sv = NEWSV(85,0);
2503 sv_upgrade(sv, SVt_PVMG);
2504 sv_setsv(sv,PL_linestr);
2505 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2511 bof = PL_rsfp && (PerlIO_tell(PL_rsfp) == 0); /* *Before* read! */
2512 s = filter_gets(PL_linestr, PL_rsfp, 0);
2516 if (PL_preprocess && !PL_in_eval)
2517 (void)PerlProc_pclose(PL_rsfp);
2518 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2519 PerlIO_clearerr(PL_rsfp);
2521 (void)PerlIO_close(PL_rsfp);
2523 PL_doextract = FALSE;
2525 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2526 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2527 sv_catpv(PL_linestr,";}");
2528 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2529 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2530 PL_minus_n = PL_minus_p = 0;
2533 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2534 sv_setpv(PL_linestr,"");
2535 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2537 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2538 s = swallow_bom((U8*)s);
2541 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2542 PL_doextract = FALSE;
2544 /* Incest with pod. */
2545 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2546 sv_setpv(PL_linestr, "");
2547 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2548 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2549 PL_doextract = FALSE;
2553 } while (PL_doextract);
2554 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2555 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2556 SV *sv = NEWSV(85,0);
2558 sv_upgrade(sv, SVt_PVMG);
2559 sv_setsv(sv,PL_linestr);
2560 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2562 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2563 if (CopLINE(PL_curcop) == 1) {
2564 while (s < PL_bufend && isSPACE(*s))
2566 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2570 if (*s == '#' && *(s+1) == '!')
2572 #ifdef ALTERNATE_SHEBANG
2574 static char as[] = ALTERNATE_SHEBANG;
2575 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2576 d = s + (sizeof(as) - 1);
2578 #endif /* ALTERNATE_SHEBANG */
2587 while (*d && !isSPACE(*d))
2591 #ifdef ARG_ZERO_IS_SCRIPT
2592 if (ipathend > ipath) {
2594 * HP-UX (at least) sets argv[0] to the script name,
2595 * which makes $^X incorrect. And Digital UNIX and Linux,
2596 * at least, set argv[0] to the basename of the Perl
2597 * interpreter. So, having found "#!", we'll set it right.
2599 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2600 assert(SvPOK(x) || SvGMAGICAL(x));
2601 if (sv_eq(x, CopFILESV(PL_curcop))) {
2602 sv_setpvn(x, ipath, ipathend - ipath);
2605 TAINT_NOT; /* $^X is always tainted, but that's OK */
2607 #endif /* ARG_ZERO_IS_SCRIPT */
2612 d = instr(s,"perl -");
2614 d = instr(s,"perl");
2616 /* avoid getting into infinite loops when shebang
2617 * line contains "Perl" rather than "perl" */
2619 for (d = ipathend-4; d >= ipath; --d) {
2620 if ((*d == 'p' || *d == 'P')
2621 && !ibcmp(d, "perl", 4))
2631 #ifdef ALTERNATE_SHEBANG
2633 * If the ALTERNATE_SHEBANG on this system starts with a
2634 * character that can be part of a Perl expression, then if
2635 * we see it but not "perl", we're probably looking at the
2636 * start of Perl code, not a request to hand off to some
2637 * other interpreter. Similarly, if "perl" is there, but
2638 * not in the first 'word' of the line, we assume the line
2639 * contains the start of the Perl program.
2641 if (d && *s != '#') {
2643 while (*c && !strchr("; \t\r\n\f\v#", *c))
2646 d = Nullch; /* "perl" not in first word; ignore */
2648 *s = '#'; /* Don't try to parse shebang line */
2650 #endif /* ALTERNATE_SHEBANG */
2651 #ifndef MACOS_TRADITIONAL
2656 !instr(s,"indir") &&
2657 instr(PL_origargv[0],"perl"))
2663 while (s < PL_bufend && isSPACE(*s))
2665 if (s < PL_bufend) {
2666 Newz(899,newargv,PL_origargc+3,char*);
2668 while (s < PL_bufend && !isSPACE(*s))
2671 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2674 newargv = PL_origargv;
2676 PerlProc_execv(ipath, newargv);
2677 Perl_croak(aTHX_ "Can't exec %s", ipath);
2681 U32 oldpdb = PL_perldb;
2682 bool oldn = PL_minus_n;
2683 bool oldp = PL_minus_p;
2685 while (*d && !isSPACE(*d)) d++;
2686 while (SPACE_OR_TAB(*d)) d++;
2690 if (*d == 'M' || *d == 'm') {
2692 while (*d && !isSPACE(*d)) d++;
2693 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2696 d = moreswitches(d);
2698 if ((PERLDB_LINE && !oldpdb) ||
2699 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2700 /* if we have already added "LINE: while (<>) {",
2701 we must not do it again */
2703 sv_setpv(PL_linestr, "");
2704 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2705 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2706 PL_preambled = FALSE;
2708 (void)gv_fetchfile(PL_origfilename);
2715 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2717 PL_lex_state = LEX_FORMLINE;
2722 #ifdef PERL_STRICT_CR
2723 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2725 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2727 case ' ': case '\t': case '\f': case 013:
2728 #ifdef MACOS_TRADITIONAL
2735 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2736 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2737 /* handle eval qq[#line 1 "foo"\n ...] */
2738 CopLINE_dec(PL_curcop);
2742 while (s < d && *s != '\n')
2747 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2749 PL_lex_state = LEX_FORMLINE;
2759 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2764 while (s < PL_bufend && SPACE_OR_TAB(*s))
2767 if (strnEQ(s,"=>",2)) {
2768 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2769 OPERATOR('-'); /* unary minus */
2771 PL_last_uni = PL_oldbufptr;
2772 PL_last_lop_op = OP_FTEREAD; /* good enough */
2774 case 'r': FTST(OP_FTEREAD);
2775 case 'w': FTST(OP_FTEWRITE);
2776 case 'x': FTST(OP_FTEEXEC);
2777 case 'o': FTST(OP_FTEOWNED);
2778 case 'R': FTST(OP_FTRREAD);
2779 case 'W': FTST(OP_FTRWRITE);
2780 case 'X': FTST(OP_FTREXEC);
2781 case 'O': FTST(OP_FTROWNED);
2782 case 'e': FTST(OP_FTIS);
2783 case 'z': FTST(OP_FTZERO);
2784 case 's': FTST(OP_FTSIZE);
2785 case 'f': FTST(OP_FTFILE);
2786 case 'd': FTST(OP_FTDIR);
2787 case 'l': FTST(OP_FTLINK);
2788 case 'p': FTST(OP_FTPIPE);
2789 case 'S': FTST(OP_FTSOCK);
2790 case 'u': FTST(OP_FTSUID);
2791 case 'g': FTST(OP_FTSGID);
2792 case 'k': FTST(OP_FTSVTX);
2793 case 'b': FTST(OP_FTBLK);
2794 case 'c': FTST(OP_FTCHR);
2795 case 't': FTST(OP_FTTTY);
2796 case 'T': FTST(OP_FTTEXT);
2797 case 'B': FTST(OP_FTBINARY);
2798 case 'M': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTMTIME);
2799 case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME);
2800 case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME);
2802 Perl_croak(aTHX_ "Unrecognized file test: -%c", (int)tmp);
2809 if (PL_expect == XOPERATOR)
2814 else if (*s == '>') {
2817 if (isIDFIRST_lazy_if(s,UTF)) {
2818 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2826 if (PL_expect == XOPERATOR)
2829 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2831 OPERATOR('-'); /* unary minus */
2838 if (PL_expect == XOPERATOR)
2843 if (PL_expect == XOPERATOR)
2846 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2852 if (PL_expect != XOPERATOR) {
2853 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2854 PL_expect = XOPERATOR;
2855 force_ident(PL_tokenbuf, '*');
2868 if (PL_expect == XOPERATOR) {
2872 PL_tokenbuf[0] = '%';
2873 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2874 if (!PL_tokenbuf[1]) {
2876 yyerror("Final % should be \\% or %name");
2879 PL_pending_ident = '%';
2898 switch (PL_expect) {
2901 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2903 PL_bufptr = s; /* update in case we back off */
2909 PL_expect = XTERMBLOCK;
2913 while (isIDFIRST_lazy_if(s,UTF)) {
2914 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2915 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2916 if (tmp < 0) tmp = -tmp;
2931 d = scan_str(d,TRUE,TRUE);
2934 SvREFCNT_dec(PL_lex_stuff);
2935 PL_lex_stuff = Nullsv;
2937 /* MUST advance bufptr here to avoid bogus
2938 "at end of line" context messages from yyerror().
2940 PL_bufptr = s + len;
2941 yyerror("Unterminated attribute parameter in attribute list");
2944 return 0; /* EOF indicator */
2948 SV *sv = newSVpvn(s, len);
2949 sv_catsv(sv, PL_lex_stuff);
2950 attrs = append_elem(OP_LIST, attrs,
2951 newSVOP(OP_CONST, 0, sv));
2952 SvREFCNT_dec(PL_lex_stuff);
2953 PL_lex_stuff = Nullsv;
2956 attrs = append_elem(OP_LIST, attrs,
2957 newSVOP(OP_CONST, 0,
2961 if (*s == ':' && s[1] != ':')
2964 break; /* require real whitespace or :'s */
2966 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
2967 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
2968 char q = ((*s == '\'') ? '"' : '\'');
2969 /* If here for an expression, and parsed no attrs, back off. */
2970 if (tmp == '=' && !attrs) {
2974 /* MUST advance bufptr here to avoid bogus "at end of line"
2975 context messages from yyerror().
2979 yyerror("Unterminated attribute list");
2981 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
2989 PL_nextval[PL_nexttoke].opval = attrs;
2997 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
2998 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3014 if (PL_lex_brackets <= 0)
3015 yyerror("Unmatched right square bracket");
3018 if (PL_lex_state == LEX_INTERPNORMAL) {
3019 if (PL_lex_brackets == 0) {
3020 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3021 PL_lex_state = LEX_INTERPEND;
3028 if (PL_lex_brackets > 100) {
3029 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
3030 if (newlb != PL_lex_brackstack) {
3032 PL_lex_brackstack = newlb;
3035 switch (PL_expect) {
3037 if (PL_lex_formbrack) {
3041 if (PL_oldoldbufptr == PL_last_lop)
3042 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3044 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3045 OPERATOR(HASHBRACK);
3047 while (s < PL_bufend && SPACE_OR_TAB(*s))
3050 PL_tokenbuf[0] = '\0';
3051 if (d < PL_bufend && *d == '-') {
3052 PL_tokenbuf[0] = '-';
3054 while (d < PL_bufend && SPACE_OR_TAB(*d))
3057 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3058 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3060 while (d < PL_bufend && SPACE_OR_TAB(*d))
3063 char minus = (PL_tokenbuf[0] == '-');
3064 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3072 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3077 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3082 if (PL_oldoldbufptr == PL_last_lop)
3083 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3085 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3088 OPERATOR(HASHBRACK);
3089 /* This hack serves to disambiguate a pair of curlies
3090 * as being a block or an anon hash. Normally, expectation
3091 * determines that, but in cases where we're not in a
3092 * position to expect anything in particular (like inside
3093 * eval"") we have to resolve the ambiguity. This code
3094 * covers the case where the first term in the curlies is a
3095 * quoted string. Most other cases need to be explicitly
3096 * disambiguated by prepending a `+' before the opening
3097 * curly in order to force resolution as an anon hash.
3099 * XXX should probably propagate the outer expectation
3100 * into eval"" to rely less on this hack, but that could
3101 * potentially break current behavior of eval"".
3105 if (*s == '\'' || *s == '"' || *s == '`') {
3106 /* common case: get past first string, handling escapes */
3107 for (t++; t < PL_bufend && *t != *s;)
3108 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3112 else if (*s == 'q') {
3115 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3119 char open, close, term;
3122 while (t < PL_bufend && isSPACE(*t))
3126 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3130 for (t++; t < PL_bufend; t++) {
3131 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3133 else if (*t == open)
3137 for (t++; t < PL_bufend; t++) {
3138 if (*t == '\\' && t+1 < PL_bufend)
3140 else if (*t == close && --brackets <= 0)
3142 else if (*t == open)
3148 else if (isALNUM_lazy_if(t,UTF)) {
3150 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3153 while (t < PL_bufend && isSPACE(*t))
3155 /* if comma follows first term, call it an anon hash */
3156 /* XXX it could be a comma expression with loop modifiers */
3157 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3158 || (*t == '=' && t[1] == '>')))
3159 OPERATOR(HASHBRACK);
3160 if (PL_expect == XREF)
3163 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3169 yylval.ival = CopLINE(PL_curcop);
3170 if (isSPACE(*s) || *s == '#')
3171 PL_copline = NOLINE; /* invalidate current command line number */
3176 if (PL_lex_brackets <= 0)
3177 yyerror("Unmatched right curly bracket");
3179 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3180 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3181 PL_lex_formbrack = 0;
3182 if (PL_lex_state == LEX_INTERPNORMAL) {
3183 if (PL_lex_brackets == 0) {
3184 if (PL_expect & XFAKEBRACK) {
3185 PL_expect &= XENUMMASK;
3186 PL_lex_state = LEX_INTERPEND;
3188 return yylex(); /* ignore fake brackets */
3190 if (*s == '-' && s[1] == '>')
3191 PL_lex_state = LEX_INTERPENDMAYBE;
3192 else if (*s != '[' && *s != '{')
3193 PL_lex_state = LEX_INTERPEND;
3196 if (PL_expect & XFAKEBRACK) {
3197 PL_expect &= XENUMMASK;
3199 return yylex(); /* ignore fake brackets */
3209 if (PL_expect == XOPERATOR) {
3210 if (ckWARN(WARN_SEMICOLON)
3211 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3213 CopLINE_dec(PL_curcop);
3214 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3215 CopLINE_inc(PL_curcop);
3220 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3222 PL_expect = XOPERATOR;
3223 force_ident(PL_tokenbuf, '&');
3227 yylval.ival = (OPpENTERSUB_AMPER<<8);
3246 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3247 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3249 if (PL_expect == XSTATE && isALPHA(tmp) &&
3250 (s == PL_linestart+1 || s[-2] == '\n') )
3252 if (PL_in_eval && !PL_rsfp) {
3257 if (strnEQ(s,"=cut",4)) {
3271 PL_doextract = TRUE;
3274 if (PL_lex_brackets < PL_lex_formbrack) {
3276 #ifdef PERL_STRICT_CR
3277 for (t = s; SPACE_OR_TAB(*t); t++) ;
3279 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3281 if (*t == '\n' || *t == '#') {
3299 if (PL_expect != XOPERATOR) {
3300 if (s[1] != '<' && !strchr(s,'>'))
3303 s = scan_heredoc(s);
3305 s = scan_inputsymbol(s);
3306 TERM(sublex_start());
3311 SHop(OP_LEFT_SHIFT);
3325 SHop(OP_RIGHT_SHIFT);
3334 if (PL_expect == XOPERATOR) {
3335 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3338 return ','; /* grandfather non-comma-format format */
3342 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3343 PL_tokenbuf[0] = '@';
3344 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3345 sizeof PL_tokenbuf - 1, FALSE);
3346 if (PL_expect == XOPERATOR)
3347 no_op("Array length", s);
3348 if (!PL_tokenbuf[1])
3350 PL_expect = XOPERATOR;
3351 PL_pending_ident = '#';
3355 PL_tokenbuf[0] = '$';
3356 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3357 sizeof PL_tokenbuf - 1, FALSE);
3358 if (PL_expect == XOPERATOR)
3360 if (!PL_tokenbuf[1]) {
3362 yyerror("Final $ should be \\$ or $name");
3366 /* This kludge not intended to be bulletproof. */
3367 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3368 yylval.opval = newSVOP(OP_CONST, 0,
3369 newSViv(PL_compiling.cop_arybase));
3370 yylval.opval->op_private = OPpCONST_ARYBASE;
3376 if (PL_lex_state == LEX_NORMAL)
3379 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3382 PL_tokenbuf[0] = '@';
3383 if (ckWARN(WARN_SYNTAX)) {
3385 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3388 PL_bufptr = skipspace(PL_bufptr);
3389 while (t < PL_bufend && *t != ']')
3391 Perl_warner(aTHX_ WARN_SYNTAX,
3392 "Multidimensional syntax %.*s not supported",
3393 (t - PL_bufptr) + 1, PL_bufptr);
3397 else if (*s == '{') {
3398 PL_tokenbuf[0] = '%';
3399 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3400 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3402 char tmpbuf[sizeof PL_tokenbuf];
3404 for (t++; isSPACE(*t); t++) ;
3405 if (isIDFIRST_lazy_if(t,UTF)) {
3406 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3407 for (; isSPACE(*t); t++) ;
3408 if (*t == ';' && get_cv(tmpbuf, FALSE))
3409 Perl_warner(aTHX_ WARN_SYNTAX,
3410 "You need to quote \"%s\"", tmpbuf);
3416 PL_expect = XOPERATOR;
3417 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3418 bool islop = (PL_last_lop == PL_oldoldbufptr);
3419 if (!islop || PL_last_lop_op == OP_GREPSTART)
3420 PL_expect = XOPERATOR;
3421 else if (strchr("$@\"'`q", *s))
3422 PL_expect = XTERM; /* e.g. print $fh "foo" */
3423 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3424 PL_expect = XTERM; /* e.g. print $fh &sub */
3425 else if (isIDFIRST_lazy_if(s,UTF)) {
3426 char tmpbuf[sizeof PL_tokenbuf];
3427 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3428 if ((tmp = keyword(tmpbuf, len))) {
3429 /* binary operators exclude handle interpretations */
3441 PL_expect = XTERM; /* e.g. print $fh length() */
3446 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3447 if (gv && GvCVu(gv))
3448 PL_expect = XTERM; /* e.g. print $fh subr() */
3451 else if (isDIGIT(*s))
3452 PL_expect = XTERM; /* e.g. print $fh 3 */
3453 else if (*s == '.' && isDIGIT(s[1]))
3454 PL_expect = XTERM; /* e.g. print $fh .3 */
3455 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3456 PL_expect = XTERM; /* e.g. print $fh -1 */
3457 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3458 PL_expect = XTERM; /* print $fh <<"EOF" */
3460 PL_pending_ident = '$';
3464 if (PL_expect == XOPERATOR)
3466 PL_tokenbuf[0] = '@';
3467 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3468 if (!PL_tokenbuf[1]) {
3470 yyerror("Final @ should be \\@ or @name");
3473 if (PL_lex_state == LEX_NORMAL)
3475 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3477 PL_tokenbuf[0] = '%';
3479 /* Warn about @ where they meant $. */
3480 if (ckWARN(WARN_SYNTAX)) {
3481 if (*s == '[' || *s == '{') {
3483 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3485 if (*t == '}' || *t == ']') {
3487 PL_bufptr = skipspace(PL_bufptr);
3488 Perl_warner(aTHX_ WARN_SYNTAX,
3489 "Scalar value %.*s better written as $%.*s",
3490 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3495 PL_pending_ident = '@';
3498 case '/': /* may either be division or pattern */
3499 case '?': /* may either be conditional or pattern */
3500 if (PL_expect != XOPERATOR) {
3501 /* Disable warning on "study /blah/" */
3502 if (PL_oldoldbufptr == PL_last_uni
3503 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3504 || memNE(PL_last_uni, "study", 5)
3505 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3507 s = scan_pat(s,OP_MATCH);
3508 TERM(sublex_start());
3516 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3517 #ifdef PERL_STRICT_CR
3520 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3522 && (s == PL_linestart || s[-1] == '\n') )
3524 PL_lex_formbrack = 0;
3528 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3534 yylval.ival = OPf_SPECIAL;
3540 if (PL_expect != XOPERATOR)
3545 case '0': case '1': case '2': case '3': case '4':
3546 case '5': case '6': case '7': case '8': case '9':
3547 s = scan_num(s, &yylval);
3548 if (PL_expect == XOPERATOR)
3553 s = scan_str(s,FALSE,FALSE);
3554 if (PL_expect == XOPERATOR) {
3555 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3558 return ','; /* grandfather non-comma-format format */
3564 missingterm((char*)0);
3565 yylval.ival = OP_CONST;
3566 TERM(sublex_start());
3569 s = scan_str(s,FALSE,FALSE);
3570 if (PL_expect == XOPERATOR) {
3571 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3574 return ','; /* grandfather non-comma-format format */
3580 missingterm((char*)0);
3581 yylval.ival = OP_CONST;
3582 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3583 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3584 yylval.ival = OP_STRINGIFY;
3588 TERM(sublex_start());
3591 s = scan_str(s,FALSE,FALSE);
3592 if (PL_expect == XOPERATOR)
3593 no_op("Backticks",s);
3595 missingterm((char*)0);
3596 yylval.ival = OP_BACKTICK;
3598 TERM(sublex_start());
3602 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3603 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3605 if (PL_expect == XOPERATOR)
3606 no_op("Backslash",s);
3610 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3614 while (isDIGIT(*start) || *start == '_')
3616 if (*start == '.' && isDIGIT(start[1])) {
3617 s = scan_num(s, &yylval);
3620 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3621 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF)) {
3625 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3628 s = scan_num(s, &yylval);
3635 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3674 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3676 /* Some keywords can be followed by any delimiter, including ':' */
3677 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3678 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3679 (PL_tokenbuf[0] == 'q' &&
3680 strchr("qwxr", PL_tokenbuf[1])))));
3682 /* x::* is just a word, unless x is "CORE" */
3683 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3687 while (d < PL_bufend && isSPACE(*d))
3688 d++; /* no comments skipped here, or s### is misparsed */
3690 /* Is this a label? */
3691 if (!tmp && PL_expect == XSTATE
3692 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3694 yylval.pval = savepv(PL_tokenbuf);
3699 /* Check for keywords */
3700 tmp = keyword(PL_tokenbuf, len);
3702 /* Is this a word before a => operator? */
3703 if (*d == '=' && d[1] == '>') {
3705 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3706 yylval.opval->op_private = OPpCONST_BARE;
3710 if (tmp < 0) { /* second-class keyword? */
3711 GV *ogv = Nullgv; /* override (winner) */
3712 GV *hgv = Nullgv; /* hidden (loser) */
3713 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3715 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3718 if (GvIMPORTED_CV(gv))
3720 else if (! CvMETHOD(cv))
3724 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3725 (gv = *gvp) != (GV*)&PL_sv_undef &&
3726 GvCVu(gv) && GvIMPORTED_CV(gv))
3732 tmp = 0; /* overridden by import or by GLOBAL */
3735 && -tmp==KEY_lock /* XXX generalizable kludge */
3737 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3739 tmp = 0; /* any sub overrides "weak" keyword */
3741 else { /* no override */
3745 if (ckWARN(WARN_AMBIGUOUS) && hgv
3746 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3747 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3748 "Ambiguous call resolved as CORE::%s(), %s",
3749 GvENAME(hgv), "qualify as such or use &");
3756 default: /* not a keyword */
3759 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3761 /* Get the rest if it looks like a package qualifier */
3763 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3765 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3768 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3769 *s == '\'' ? "'" : "::");
3773 if (PL_expect == XOPERATOR) {
3774 if (PL_bufptr == PL_linestart) {
3775 CopLINE_dec(PL_curcop);
3776 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3777 CopLINE_inc(PL_curcop);
3780 no_op("Bareword",s);
3783 /* Look for a subroutine with this name in current package,
3784 unless name is "Foo::", in which case Foo is a bearword
3785 (and a package name). */
3788 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3790 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3791 Perl_warner(aTHX_ WARN_BAREWORD,
3792 "Bareword \"%s\" refers to nonexistent package",
3795 PL_tokenbuf[len] = '\0';
3802 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3805 /* if we saw a global override before, get the right name */
3808 sv = newSVpvn("CORE::GLOBAL::",14);
3809 sv_catpv(sv,PL_tokenbuf);
3812 sv = newSVpv(PL_tokenbuf,0);
3814 /* Presume this is going to be a bareword of some sort. */
3817 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3818 yylval.opval->op_private = OPpCONST_BARE;
3820 /* And if "Foo::", then that's what it certainly is. */
3825 /* See if it's the indirect object for a list operator. */
3827 if (PL_oldoldbufptr &&
3828 PL_oldoldbufptr < PL_bufptr &&
3829 (PL_oldoldbufptr == PL_last_lop
3830 || PL_oldoldbufptr == PL_last_uni) &&
3831 /* NO SKIPSPACE BEFORE HERE! */
3832 (PL_expect == XREF ||
3833 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3835 bool immediate_paren = *s == '(';
3837 /* (Now we can afford to cross potential line boundary.) */
3840 /* Two barewords in a row may indicate method call. */
3842 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3845 /* If not a declared subroutine, it's an indirect object. */
3846 /* (But it's an indir obj regardless for sort.) */
3848 if ((PL_last_lop_op == OP_SORT ||
3849 (!immediate_paren && (!gv || !GvCVu(gv)))) &&
3850 (PL_last_lop_op != OP_MAPSTART &&
3851 PL_last_lop_op != OP_GREPSTART))
3853 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3859 PL_expect = XOPERATOR;
3862 /* Is this a word before a => operator? */
3863 if (*s == '=' && s[1] == '>') {
3865 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
3869 /* If followed by a paren, it's certainly a subroutine. */
3872 if (gv && GvCVu(gv)) {
3873 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
3874 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3879 PL_nextval[PL_nexttoke].opval = yylval.opval;
3880 PL_expect = XOPERATOR;
3886 /* If followed by var or block, call it a method (unless sub) */
3888 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3889 PL_last_lop = PL_oldbufptr;
3890 PL_last_lop_op = OP_METHOD;
3894 /* If followed by a bareword, see if it looks like indir obj. */
3896 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3899 /* Not a method, so call it a subroutine (if defined) */
3901 if (gv && GvCVu(gv)) {
3903 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3904 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3905 "Ambiguous use of -%s resolved as -&%s()",
3906 PL_tokenbuf, PL_tokenbuf);
3907 /* Check for a constant sub */
3909 if ((sv = cv_const_sv(cv))) {
3911 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3912 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3913 yylval.opval->op_private = 0;
3917 /* Resolve to GV now. */
3918 op_free(yylval.opval);
3919 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3920 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3921 PL_last_lop = PL_oldbufptr;
3922 PL_last_lop_op = OP_ENTERSUB;
3923 /* Is there a prototype? */
3926 char *proto = SvPV((SV*)cv, len);
3929 if (strEQ(proto, "$"))
3931 if (*proto == '&' && *s == '{') {
3932 sv_setpv(PL_subname,"__ANON__");
3936 PL_nextval[PL_nexttoke].opval = yylval.opval;
3942 /* Call it a bare word */
3944 if (PL_hints & HINT_STRICT_SUBS)
3945 yylval.opval->op_private |= OPpCONST_STRICT;
3948 if (ckWARN(WARN_RESERVED)) {
3949 if (lastchar != '-') {
3950 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
3952 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
3959 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
3960 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3961 "Operator or semicolon missing before %c%s",
3962 lastchar, PL_tokenbuf);
3963 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3964 "Ambiguous use of %c resolved as operator %c",
3965 lastchar, lastchar);
3971 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3972 newSVpv(CopFILE(PL_curcop),0));
3976 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3977 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
3980 case KEY___PACKAGE__:
3981 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3983 ? newSVsv(PL_curstname)
3992 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
3993 char *pname = "main";
3994 if (PL_tokenbuf[2] == 'D')
3995 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
3996 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
3999 GvIOp(gv) = newIO();
4000 IoIFP(GvIOp(gv)) = PL_rsfp;
4001 #if defined(HAS_FCNTL) && defined(F_SETFD)
4003 int fd = PerlIO_fileno(PL_rsfp);
4004 fcntl(fd,F_SETFD,fd >= 3);
4007 /* Mark this internal pseudo-handle as clean */
4008 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4010 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4011 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4012 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4014 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4015 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4016 /* if the script was opened in binmode, we need to revert
4017 * it to text mode for compatibility; but only iff it has CRs
4018 * XXX this is a questionable hack at best. */
4019 if (PL_bufend-PL_bufptr > 2
4020 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4023 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4024 loc = PerlIO_tell(PL_rsfp);
4025 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4027 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4028 #if defined(__BORLANDC__)
4029 /* XXX see note in do_binmode() */
4030 ((FILE*)PL_rsfp)->flags |= _F_BIN;
4033 PerlIO_seek(PL_rsfp, loc, 0);
4048 if (PL_expect == XSTATE) {
4055 if (*s == ':' && s[1] == ':') {
4058 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4059 if (!(tmp = keyword(PL_tokenbuf, len)))
4060 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4074 LOP(OP_ACCEPT,XTERM);
4080 LOP(OP_ATAN2,XTERM);
4086 LOP(OP_BINMODE,XTERM);
4089 LOP(OP_BLESS,XTERM);
4098 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4115 if (!PL_cryptseen) {
4116 PL_cryptseen = TRUE;
4120 LOP(OP_CRYPT,XTERM);
4123 if (ckWARN(WARN_CHMOD)) {
4124 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4125 if (*d != '0' && isDIGIT(*d))
4126 Perl_warner(aTHX_ WARN_CHMOD,
4127 "chmod() mode argument is missing initial 0");
4129 LOP(OP_CHMOD,XTERM);
4132 LOP(OP_CHOWN,XTERM);
4135 LOP(OP_CONNECT,XTERM);
4151 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4155 PL_hints |= HINT_BLOCK_SCOPE;
4165 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4166 LOP(OP_DBMOPEN,XTERM);
4172 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4179 yylval.ival = CopLINE(PL_curcop);
4193 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4194 UNIBRACK(OP_ENTEREVAL);
4209 case KEY_endhostent:
4215 case KEY_endservent:
4218 case KEY_endprotoent:
4229 yylval.ival = CopLINE(PL_curcop);
4231 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4233 if ((PL_bufend - p) >= 3 &&
4234 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4236 else if ((PL_bufend - p) >= 4 &&
4237 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4240 if (isIDFIRST_lazy_if(p,UTF)) {
4241 p = scan_ident(p, PL_bufend,
4242 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4246 Perl_croak(aTHX_ "Missing $ on loop variable");
4251 LOP(OP_FORMLINE,XTERM);
4257 LOP(OP_FCNTL,XTERM);
4263 LOP(OP_FLOCK,XTERM);
4272 LOP(OP_GREPSTART, XREF);
4275 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4290 case KEY_getpriority:
4291 LOP(OP_GETPRIORITY,XTERM);
4293 case KEY_getprotobyname:
4296 case KEY_getprotobynumber:
4297 LOP(OP_GPBYNUMBER,XTERM);
4299 case KEY_getprotoent:
4311 case KEY_getpeername:
4312 UNI(OP_GETPEERNAME);
4314 case KEY_gethostbyname:
4317 case KEY_gethostbyaddr:
4318 LOP(OP_GHBYADDR,XTERM);
4320 case KEY_gethostent:
4323 case KEY_getnetbyname:
4326 case KEY_getnetbyaddr:
4327 LOP(OP_GNBYADDR,XTERM);
4332 case KEY_getservbyname:
4333 LOP(OP_GSBYNAME,XTERM);
4335 case KEY_getservbyport:
4336 LOP(OP_GSBYPORT,XTERM);
4338 case KEY_getservent:
4341 case KEY_getsockname:
4342 UNI(OP_GETSOCKNAME);
4344 case KEY_getsockopt:
4345 LOP(OP_GSOCKOPT,XTERM);
4367 yylval.ival = CopLINE(PL_curcop);
4371 LOP(OP_INDEX,XTERM);
4377 LOP(OP_IOCTL,XTERM);
4389 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4421 LOP(OP_LISTEN,XTERM);
4430 s = scan_pat(s,OP_MATCH);
4431 TERM(sublex_start());
4434 LOP(OP_MAPSTART, XREF);
4437 LOP(OP_MKDIR,XTERM);
4440 LOP(OP_MSGCTL,XTERM);
4443 LOP(OP_MSGGET,XTERM);
4446 LOP(OP_MSGRCV,XTERM);
4449 LOP(OP_MSGSND,XTERM);
4455 if (isIDFIRST_lazy_if(s,UTF)) {
4456 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4457 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4459 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4460 if (!PL_in_my_stash) {
4463 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4471 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4478 if (PL_expect != XSTATE)
4479 yyerror("\"no\" not allowed in expression");
4480 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4481 s = force_version(s);
4486 if (*s == '(' || (s = skipspace(s), *s == '('))
4493 if (isIDFIRST_lazy_if(s,UTF)) {
4495 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4497 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4498 Perl_warner(aTHX_ WARN_PRECEDENCE,
4499 "Precedence problem: open %.*s should be open(%.*s)",
4505 yylval.ival = OP_OR;
4515 LOP(OP_OPEN_DIR,XTERM);
4518 checkcomma(s,PL_tokenbuf,"filehandle");
4522 checkcomma(s,PL_tokenbuf,"filehandle");
4541 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4545 LOP(OP_PIPE_OP,XTERM);
4548 s = scan_str(s,FALSE,FALSE);
4550 missingterm((char*)0);
4551 yylval.ival = OP_CONST;
4552 TERM(sublex_start());
4558 s = scan_str(s,FALSE,FALSE);
4560 missingterm((char*)0);
4562 if (SvCUR(PL_lex_stuff)) {
4565 d = SvPV_force(PL_lex_stuff, len);
4567 for (; isSPACE(*d) && len; --len, ++d) ;
4570 if (!warned && ckWARN(WARN_QW)) {
4571 for (; !isSPACE(*d) && len; --len, ++d) {
4573 Perl_warner(aTHX_ WARN_QW,
4574 "Possible attempt to separate words with commas");
4577 else if (*d == '#') {
4578 Perl_warner(aTHX_ WARN_QW,
4579 "Possible attempt to put comments in qw() list");
4585 for (; !isSPACE(*d) && len; --len, ++d) ;
4587 words = append_elem(OP_LIST, words,
4588 newSVOP(OP_CONST, 0, tokeq(newSVpvn(b, d-b))));
4592 PL_nextval[PL_nexttoke].opval = words;
4597 SvREFCNT_dec(PL_lex_stuff);
4598 PL_lex_stuff = Nullsv;
4603 s = scan_str(s,FALSE,FALSE);
4605 missingterm((char*)0);
4606 yylval.ival = OP_STRINGIFY;
4607 if (SvIVX(PL_lex_stuff) == '\'')
4608 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4609 TERM(sublex_start());
4612 s = scan_pat(s,OP_QR);
4613 TERM(sublex_start());
4616 s = scan_str(s,FALSE,FALSE);
4618 missingterm((char*)0);
4619 yylval.ival = OP_BACKTICK;
4621 TERM(sublex_start());
4628 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4629 s = force_version(s);
4632 *PL_tokenbuf = '\0';
4633 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4634 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4635 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4637 yyerror("<> should be quotes");
4645 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4649 LOP(OP_RENAME,XTERM);
4658 LOP(OP_RINDEX,XTERM);
4681 LOP(OP_REVERSE,XTERM);
4692 TERM(sublex_start());
4694 TOKEN(1); /* force error */
4703 LOP(OP_SELECT,XTERM);
4709 LOP(OP_SEMCTL,XTERM);
4712 LOP(OP_SEMGET,XTERM);
4715 LOP(OP_SEMOP,XTERM);
4721 LOP(OP_SETPGRP,XTERM);
4723 case KEY_setpriority:
4724 LOP(OP_SETPRIORITY,XTERM);
4726 case KEY_sethostent:
4732 case KEY_setservent:
4735 case KEY_setprotoent:
4745 LOP(OP_SEEKDIR,XTERM);
4747 case KEY_setsockopt:
4748 LOP(OP_SSOCKOPT,XTERM);
4754 LOP(OP_SHMCTL,XTERM);
4757 LOP(OP_SHMGET,XTERM);
4760 LOP(OP_SHMREAD,XTERM);
4763 LOP(OP_SHMWRITE,XTERM);
4766 LOP(OP_SHUTDOWN,XTERM);
4775 LOP(OP_SOCKET,XTERM);
4777 case KEY_socketpair:
4778 LOP(OP_SOCKPAIR,XTERM);
4781 checkcomma(s,PL_tokenbuf,"subroutine name");
4783 if (*s == ';' || *s == ')') /* probably a close */
4784 Perl_croak(aTHX_ "sort is now a reserved word");
4786 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4790 LOP(OP_SPLIT,XTERM);
4793 LOP(OP_SPRINTF,XTERM);
4796 LOP(OP_SPLICE,XTERM);
4811 LOP(OP_SUBSTR,XTERM);
4817 char tmpbuf[sizeof PL_tokenbuf];
4819 expectation attrful;
4820 bool have_name, have_proto;
4825 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4826 (*s == ':' && s[1] == ':'))
4829 attrful = XATTRBLOCK;
4830 /* remember buffer pos'n for later force_word */
4831 tboffset = s - PL_oldbufptr;
4832 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4833 if (strchr(tmpbuf, ':'))
4834 sv_setpv(PL_subname, tmpbuf);
4836 sv_setsv(PL_subname,PL_curstname);
4837 sv_catpvn(PL_subname,"::",2);
4838 sv_catpvn(PL_subname,tmpbuf,len);
4845 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4846 PL_expect = XTERMBLOCK;
4847 attrful = XATTRTERM;
4848 sv_setpv(PL_subname,"?");
4852 if (key == KEY_format) {
4854 PL_lex_formbrack = PL_lex_brackets + 1;
4856 (void) force_word(PL_oldbufptr + tboffset, WORD,
4861 /* Look for a prototype */
4865 s = scan_str(s,FALSE,FALSE);
4868 SvREFCNT_dec(PL_lex_stuff);
4869 PL_lex_stuff = Nullsv;
4870 Perl_croak(aTHX_ "Prototype not terminated");
4873 d = SvPVX(PL_lex_stuff);
4875 for (p = d; *p; ++p) {
4880 SvCUR(PL_lex_stuff) = tmp;
4888 if (*s == ':' && s[1] != ':')
4889 PL_expect = attrful;
4892 PL_nextval[PL_nexttoke].opval =
4893 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4894 PL_lex_stuff = Nullsv;
4898 sv_setpv(PL_subname,"__ANON__");
4901 (void) force_word(PL_oldbufptr + tboffset, WORD,
4910 LOP(OP_SYSTEM,XREF);
4913 LOP(OP_SYMLINK,XTERM);
4916 LOP(OP_SYSCALL,XTERM);
4919 LOP(OP_SYSOPEN,XTERM);
4922 LOP(OP_SYSSEEK,XTERM);
4925 LOP(OP_SYSREAD,XTERM);
4928 LOP(OP_SYSWRITE,XTERM);
4932 TERM(sublex_start());
4953 LOP(OP_TRUNCATE,XTERM);
4965 yylval.ival = CopLINE(PL_curcop);
4969 yylval.ival = CopLINE(PL_curcop);
4973 LOP(OP_UNLINK,XTERM);
4979 LOP(OP_UNPACK,XTERM);
4982 LOP(OP_UTIME,XTERM);
4985 if (ckWARN(WARN_UMASK)) {
4986 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4987 if (*d != '0' && isDIGIT(*d))
4988 Perl_warner(aTHX_ WARN_UMASK,
4989 "umask: argument is missing initial 0");
4994 LOP(OP_UNSHIFT,XTERM);
4997 if (PL_expect != XSTATE)
4998 yyerror("\"use\" not allowed in expression");
5000 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
5001 s = force_version(s);
5002 if (*s == ';' || (s = skipspace(s), *s == ';')) {
5003 PL_nextval[PL_nexttoke].opval = Nullop;
5008 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5009 s = force_version(s);
5021 yylval.ival = CopLINE(PL_curcop);
5025 PL_hints |= HINT_BLOCK_SCOPE;
5032 LOP(OP_WAITPID,XTERM);
5040 static char ctl_l[2];
5042 if (ctl_l[0] == '\0')
5043 ctl_l[0] = toCTRL('L');
5044 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5047 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5052 if (PL_expect == XOPERATOR)
5058 yylval.ival = OP_XOR;
5063 TERM(sublex_start());
5068 #pragma segment Main
5072 Perl_keyword(pTHX_ register char *d, I32 len)
5077 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5078 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5079 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5080 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5081 if (strEQ(d,"__END__")) return KEY___END__;
5085 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5090 if (strEQ(d,"and")) return -KEY_and;
5091 if (strEQ(d,"abs")) return -KEY_abs;
5094 if (strEQ(d,"alarm")) return -KEY_alarm;
5095 if (strEQ(d,"atan2")) return -KEY_atan2;
5098 if (strEQ(d,"accept")) return -KEY_accept;
5103 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5106 if (strEQ(d,"bless")) return -KEY_bless;
5107 if (strEQ(d,"bind")) return -KEY_bind;
5108 if (strEQ(d,"binmode")) return -KEY_binmode;
5111 if (strEQ(d,"CORE")) return -KEY_CORE;
5112 if (strEQ(d,"CHECK")) return KEY_CHECK;
5117 if (strEQ(d,"cmp")) return -KEY_cmp;
5118 if (strEQ(d,"chr")) return -KEY_chr;
5119 if (strEQ(d,"cos")) return -KEY_cos;
5122 if (strEQ(d,"chop")) return -KEY_chop;
5125 if (strEQ(d,"close")) return -KEY_close;
5126 if (strEQ(d,"chdir")) return -KEY_chdir;
5127 if (strEQ(d,"chomp")) return -KEY_chomp;
5128 if (strEQ(d,"chmod")) return -KEY_chmod;
5129 if (strEQ(d,"chown")) return -KEY_chown;
5130 if (strEQ(d,"crypt")) return -KEY_crypt;
5133 if (strEQ(d,"chroot")) return -KEY_chroot;
5134 if (strEQ(d,"caller")) return -KEY_caller;
5137 if (strEQ(d,"connect")) return -KEY_connect;
5140 if (strEQ(d,"closedir")) return -KEY_closedir;
5141 if (strEQ(d,"continue")) return -KEY_continue;
5146 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5151 if (strEQ(d,"do")) return KEY_do;
5154 if (strEQ(d,"die")) return -KEY_die;
5157 if (strEQ(d,"dump")) return -KEY_dump;
5160 if (strEQ(d,"delete")) return KEY_delete;
5163 if (strEQ(d,"defined")) return KEY_defined;
5164 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5167 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5172 if (strEQ(d,"END")) return KEY_END;
5177 if (strEQ(d,"eq")) return -KEY_eq;
5180 if (strEQ(d,"eof")) return -KEY_eof;
5181 if (strEQ(d,"exp")) return -KEY_exp;
5184 if (strEQ(d,"else")) return KEY_else;
5185 if (strEQ(d,"exit")) return -KEY_exit;
5186 if (strEQ(d,"eval")) return KEY_eval;
5187 if (strEQ(d,"exec")) return -KEY_exec;
5188 if (strEQ(d,"each")) return KEY_each;
5191 if (strEQ(d,"elsif")) return KEY_elsif;
5194 if (strEQ(d,"exists")) return KEY_exists;
5195 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5198 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5199 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5202 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5205 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5206 if (strEQ(d,"endservent")) return -KEY_endservent;
5209 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5216 if (strEQ(d,"for")) return KEY_for;
5219 if (strEQ(d,"fork")) return -KEY_fork;
5222 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5223 if (strEQ(d,"flock")) return -KEY_flock;
5226 if (strEQ(d,"format")) return KEY_format;
5227 if (strEQ(d,"fileno")) return -KEY_fileno;
5230 if (strEQ(d,"foreach")) return KEY_foreach;
5233 if (strEQ(d,"formline")) return -KEY_formline;
5238 if (strnEQ(d,"get",3)) {
5243 if (strEQ(d,"ppid")) return -KEY_getppid;
5244 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5247 if (strEQ(d,"pwent")) return -KEY_getpwent;
5248 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5249 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5252 if (strEQ(d,"peername")) return -KEY_getpeername;
5253 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5254 if (strEQ(d,"priority")) return -KEY_getpriority;
5257 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5260 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5264 else if (*d == 'h') {
5265 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5266 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5267 if (strEQ(d,"hostent")) return -KEY_gethostent;
5269 else if (*d == 'n') {
5270 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5271 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5272 if (strEQ(d,"netent")) return -KEY_getnetent;
5274 else if (*d == 's') {
5275 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5276 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5277 if (strEQ(d,"servent")) return -KEY_getservent;
5278 if (strEQ(d,"sockname")) return -KEY_getsockname;
5279 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5281 else if (*d == 'g') {
5282 if (strEQ(d,"grent")) return -KEY_getgrent;
5283 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5284 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5286 else if (*d == 'l') {
5287 if (strEQ(d,"login")) return -KEY_getlogin;
5289 else if (strEQ(d,"c")) return -KEY_getc;
5294 if (strEQ(d,"gt")) return -KEY_gt;
5295 if (strEQ(d,"ge")) return -KEY_ge;
5298 if (strEQ(d,"grep")) return KEY_grep;
5299 if (strEQ(d,"goto")) return KEY_goto;
5300 if (strEQ(d,"glob")) return KEY_glob;
5303 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5308 if (strEQ(d,"hex")) return -KEY_hex;
5311 if (strEQ(d,"INIT")) return KEY_INIT;
5316 if (strEQ(d,"if")) return KEY_if;
5319 if (strEQ(d,"int")) return -KEY_int;
5322 if (strEQ(d,"index")) return -KEY_index;
5323 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5328 if (strEQ(d,"join")) return -KEY_join;
5332 if (strEQ(d,"keys")) return KEY_keys;
5333 if (strEQ(d,"kill")) return -KEY_kill;
5339 if (strEQ(d,"lt")) return -KEY_lt;
5340 if (strEQ(d,"le")) return -KEY_le;
5341 if (strEQ(d,"lc")) return -KEY_lc;
5344 if (strEQ(d,"log")) return -KEY_log;
5347 if (strEQ(d,"last")) return KEY_last;
5348 if (strEQ(d,"link")) return -KEY_link;
5349 if (strEQ(d,"lock")) return -KEY_lock;
5352 if (strEQ(d,"local")) return KEY_local;
5353 if (strEQ(d,"lstat")) return -KEY_lstat;
5356 if (strEQ(d,"length")) return -KEY_length;
5357 if (strEQ(d,"listen")) return -KEY_listen;
5360 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5363 if (strEQ(d,"localtime")) return -KEY_localtime;
5369 case 1: return KEY_m;
5371 if (strEQ(d,"my")) return KEY_my;
5374 if (strEQ(d,"map")) return KEY_map;
5377 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5380 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5381 if (strEQ(d,"msgget")) return -KEY_msgget;
5382 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5383 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5388 if (strEQ(d,"next")) return KEY_next;
5389 if (strEQ(d,"ne")) return -KEY_ne;
5390 if (strEQ(d,"not")) return -KEY_not;
5391 if (strEQ(d,"no")) return KEY_no;
5396 if (strEQ(d,"or")) return -KEY_or;
5399 if (strEQ(d,"ord")) return -KEY_ord;
5400 if (strEQ(d,"oct")) return -KEY_oct;
5401 if (strEQ(d,"our")) return KEY_our;
5404 if (strEQ(d,"open")) return -KEY_open;
5407 if (strEQ(d,"opendir")) return -KEY_opendir;
5414 if (strEQ(d,"pop")) return KEY_pop;
5415 if (strEQ(d,"pos")) return KEY_pos;
5418 if (strEQ(d,"push")) return KEY_push;
5419 if (strEQ(d,"pack")) return -KEY_pack;
5420 if (strEQ(d,"pipe")) return -KEY_pipe;
5423 if (strEQ(d,"print")) return KEY_print;
5426 if (strEQ(d,"printf")) return KEY_printf;
5429 if (strEQ(d,"package")) return KEY_package;
5432 if (strEQ(d,"prototype")) return KEY_prototype;
5437 if (strEQ(d,"q")) return KEY_q;
5438 if (strEQ(d,"qr")) return KEY_qr;
5439 if (strEQ(d,"qq")) return KEY_qq;
5440 if (strEQ(d,"qw")) return KEY_qw;
5441 if (strEQ(d,"qx")) return KEY_qx;
5443 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5448 if (strEQ(d,"ref")) return -KEY_ref;
5451 if (strEQ(d,"read")) return -KEY_read;
5452 if (strEQ(d,"rand")) return -KEY_rand;
5453 if (strEQ(d,"recv")) return -KEY_recv;
5454 if (strEQ(d,"redo")) return KEY_redo;
5457 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5458 if (strEQ(d,"reset")) return -KEY_reset;
5461 if (strEQ(d,"return")) return KEY_return;
5462 if (strEQ(d,"rename")) return -KEY_rename;
5463 if (strEQ(d,"rindex")) return -KEY_rindex;
5466 if (strEQ(d,"require")) return -KEY_require;
5467 if (strEQ(d,"reverse")) return -KEY_reverse;
5468 if (strEQ(d,"readdir")) return -KEY_readdir;
5471 if (strEQ(d,"readlink")) return -KEY_readlink;
5472 if (strEQ(d,"readline")) return -KEY_readline;
5473 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5476 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5482 case 0: return KEY_s;
5484 if (strEQ(d,"scalar")) return KEY_scalar;
5489 if (strEQ(d,"seek")) return -KEY_seek;
5490 if (strEQ(d,"send")) return -KEY_send;
5493 if (strEQ(d,"semop")) return -KEY_semop;
5496 if (strEQ(d,"select")) return -KEY_select;
5497 if (strEQ(d,"semctl")) return -KEY_semctl;
5498 if (strEQ(d,"semget")) return -KEY_semget;
5501 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5502 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5505 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5506 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5509 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5512 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5513 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5514 if (strEQ(d,"setservent")) return -KEY_setservent;
5517 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5518 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5525 if (strEQ(d,"shift")) return KEY_shift;
5528 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5529 if (strEQ(d,"shmget")) return -KEY_shmget;
5532 if (strEQ(d,"shmread")) return -KEY_shmread;
5535 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5536 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5541 if (strEQ(d,"sin")) return -KEY_sin;
5544 if (strEQ(d,"sleep")) return -KEY_sleep;
5547 if (strEQ(d,"sort")) return KEY_sort;
5548 if (strEQ(d,"socket")) return -KEY_socket;
5549 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5552 if (strEQ(d,"split")) return KEY_split;
5553 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5554 if (strEQ(d,"splice")) return KEY_splice;
5557 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5560 if (strEQ(d,"srand")) return -KEY_srand;
5563 if (strEQ(d,"stat")) return -KEY_stat;
5564 if (strEQ(d,"study")) return KEY_study;
5567 if (strEQ(d,"substr")) return -KEY_substr;
5568 if (strEQ(d,"sub")) return KEY_sub;
5573 if (strEQ(d,"system")) return -KEY_system;
5576 if (strEQ(d,"symlink")) return -KEY_symlink;
5577 if (strEQ(d,"syscall")) return -KEY_syscall;
5578 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5579 if (strEQ(d,"sysread")) return -KEY_sysread;
5580 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5583 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5592 if (strEQ(d,"tr")) return KEY_tr;
5595 if (strEQ(d,"tie")) return KEY_tie;
5598 if (strEQ(d,"tell")) return -KEY_tell;
5599 if (strEQ(d,"tied")) return KEY_tied;
5600 if (strEQ(d,"time")) return -KEY_time;
5603 if (strEQ(d,"times")) return -KEY_times;
5606 if (strEQ(d,"telldir")) return -KEY_telldir;
5609 if (strEQ(d,"truncate")) return -KEY_truncate;
5616 if (strEQ(d,"uc")) return -KEY_uc;
5619 if (strEQ(d,"use")) return KEY_use;
5622 if (strEQ(d,"undef")) return KEY_undef;
5623 if (strEQ(d,"until")) return KEY_until;
5624 if (strEQ(d,"untie")) return KEY_untie;
5625 if (strEQ(d,"utime")) return -KEY_utime;
5626 if (strEQ(d,"umask")) return -KEY_umask;
5629 if (strEQ(d,"unless")) return KEY_unless;
5630 if (strEQ(d,"unpack")) return -KEY_unpack;
5631 if (strEQ(d,"unlink")) return -KEY_unlink;
5634 if (strEQ(d,"unshift")) return KEY_unshift;
5635 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5640 if (strEQ(d,"values")) return -KEY_values;
5641 if (strEQ(d,"vec")) return -KEY_vec;
5646 if (strEQ(d,"warn")) return -KEY_warn;
5647 if (strEQ(d,"wait")) return -KEY_wait;
5650 if (strEQ(d,"while")) return KEY_while;
5651 if (strEQ(d,"write")) return -KEY_write;
5654 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5657 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5662 if (len == 1) return -KEY_x;
5663 if (strEQ(d,"xor")) return -KEY_xor;
5666 if (len == 1) return KEY_y;
5675 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5679 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5680 dTHR; /* only for ckWARN */
5681 if (ckWARN(WARN_SYNTAX)) {
5683 for (w = s+2; *w && level; w++) {
5690 for (; *w && isSPACE(*w); w++) ;
5691 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5692 Perl_warner(aTHX_ WARN_SYNTAX,
5693 "%s (...) interpreted as function",name);
5696 while (s < PL_bufend && isSPACE(*s))
5700 while (s < PL_bufend && isSPACE(*s))
5702 if (isIDFIRST_lazy_if(s,UTF)) {
5704 while (isALNUM_lazy_if(s,UTF))
5706 while (s < PL_bufend && isSPACE(*s))
5711 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5715 Perl_croak(aTHX_ "No comma allowed after %s", what);
5720 /* Either returns sv, or mortalizes sv and returns a new SV*.
5721 Best used as sv=new_constant(..., sv, ...).
5722 If s, pv are NULL, calls subroutine with one argument,
5723 and type is used with error messages only. */
5726 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5730 HV *table = GvHV(PL_hintgv); /* ^H */
5734 const char *why1, *why2, *why3;
5736 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
5739 why1 = "%^H is not consistent";
5740 why2 = strEQ(key,"charnames")
5741 ? " (missing \"use charnames ...\"?)"
5745 msg = Perl_newSVpvf(aTHX_ "constant(%s): %s%s%s",
5746 (type ? type: "undef"), why1, why2, why3);
5747 yyerror(SvPVX(msg));
5751 cvp = hv_fetch(table, key, strlen(key), FALSE);
5752 if (!cvp || !SvOK(*cvp)) {
5755 why3 = "} is not defined";
5758 sv_2mortal(sv); /* Parent created it permanently */
5761 pv = sv_2mortal(newSVpvn(s, len));
5763 typesv = sv_2mortal(newSVpv(type, 0));
5765 typesv = &PL_sv_undef;
5767 PUSHSTACKi(PERLSI_OVERLOAD);
5779 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5783 /* Check the eval first */
5784 if (!PL_in_eval && SvTRUE(ERRSV)) {
5786 sv_catpv(ERRSV, "Propagated");
5787 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5789 res = SvREFCNT_inc(sv);
5793 (void)SvREFCNT_inc(res);
5802 why1 = "Call to &{$^H{";
5804 why3 = "}} did not return a defined value";
5813 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5815 register char *d = dest;
5816 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5819 Perl_croak(aTHX_ ident_too_long);
5820 if (isALNUM(*s)) /* UTF handled below */
5822 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
5827 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5831 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5832 char *t = s + UTF8SKIP(s);
5833 while (*t & 0x80 && is_utf8_mark((U8*)t))
5835 if (d + (t - s) > e)
5836 Perl_croak(aTHX_ ident_too_long);
5837 Copy(s, d, t - s, char);
5850 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5860 e = d + destlen - 3; /* two-character token, ending NUL */
5862 while (isDIGIT(*s)) {
5864 Perl_croak(aTHX_ ident_too_long);
5871 Perl_croak(aTHX_ ident_too_long);
5872 if (isALNUM(*s)) /* UTF handled below */
5874 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
5879 else if (*s == ':' && s[1] == ':') {
5883 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5884 char *t = s + UTF8SKIP(s);
5885 while (*t & 0x80 && is_utf8_mark((U8*)t))
5887 if (d + (t - s) > e)
5888 Perl_croak(aTHX_ ident_too_long);
5889 Copy(s, d, t - s, char);
5900 if (PL_lex_state != LEX_NORMAL)
5901 PL_lex_state = LEX_INTERPENDMAYBE;
5904 if (*s == '$' && s[1] &&
5905 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
5918 if (*d == '^' && *s && isCONTROLVAR(*s)) {
5923 if (isSPACE(s[-1])) {
5926 if (!SPACE_OR_TAB(ch)) {
5932 if (isIDFIRST_lazy_if(d,UTF)) {
5936 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
5938 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
5941 Copy(s, d, e - s, char);
5946 while ((isALNUM(*s) || *s == ':') && d < e)
5949 Perl_croak(aTHX_ ident_too_long);
5952 while (s < send && SPACE_OR_TAB(*s)) s++;
5953 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
5954 dTHR; /* only for ckWARN */
5955 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
5956 const char *brack = *s == '[' ? "[...]" : "{...}";
5957 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5958 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
5959 funny, dest, brack, funny, dest, brack);
5962 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
5966 /* Handle extended ${^Foo} variables
5967 * 1999-02-27 mjd-perl-patch@plover.com */
5968 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
5972 while (isALNUM(*s) && d < e) {
5976 Perl_croak(aTHX_ ident_too_long);
5981 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
5982 PL_lex_state = LEX_INTERPEND;
5985 if (PL_lex_state == LEX_NORMAL) {
5986 dTHR; /* only for ckWARN */
5987 if (ckWARN(WARN_AMBIGUOUS) &&
5988 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
5990 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5991 "Ambiguous use of %c{%s} resolved to %c%s",
5992 funny, dest, funny, dest);
5997 s = bracket; /* let the parser handle it */
6001 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
6002 PL_lex_state = LEX_INTERPEND;
6007 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
6012 *pmfl |= PMf_GLOBAL;
6014 *pmfl |= PMf_CONTINUE;
6018 *pmfl |= PMf_MULTILINE;
6020 *pmfl |= PMf_SINGLELINE;
6022 *pmfl |= PMf_EXTENDED;
6026 S_scan_pat(pTHX_ char *start, I32 type)
6031 s = scan_str(start,FALSE,FALSE);
6034 SvREFCNT_dec(PL_lex_stuff);
6035 PL_lex_stuff = Nullsv;
6036 Perl_croak(aTHX_ "Search pattern not terminated");
6039 pm = (PMOP*)newPMOP(type, 0);
6040 if (PL_multi_open == '?')
6041 pm->op_pmflags |= PMf_ONCE;
6043 while (*s && strchr("iomsx", *s))
6044 pmflag(&pm->op_pmflags,*s++);
6047 while (*s && strchr("iogcmsx", *s))
6048 pmflag(&pm->op_pmflags,*s++);
6050 pm->op_pmpermflags = pm->op_pmflags;
6052 PL_lex_op = (OP*)pm;
6053 yylval.ival = OP_MATCH;
6058 S_scan_subst(pTHX_ char *start)
6065 yylval.ival = OP_NULL;
6067 s = scan_str(start,FALSE,FALSE);
6071 SvREFCNT_dec(PL_lex_stuff);
6072 PL_lex_stuff = Nullsv;
6073 Perl_croak(aTHX_ "Substitution pattern not terminated");
6076 if (s[-1] == PL_multi_open)
6079 first_start = PL_multi_start;
6080 s = scan_str(s,FALSE,FALSE);
6083 SvREFCNT_dec(PL_lex_stuff);
6084 PL_lex_stuff = Nullsv;
6086 SvREFCNT_dec(PL_lex_repl);
6087 PL_lex_repl = Nullsv;
6088 Perl_croak(aTHX_ "Substitution replacement not terminated");
6090 PL_multi_start = first_start; /* so whole substitution is taken together */
6092 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6098 else if (strchr("iogcmsx", *s))
6099 pmflag(&pm->op_pmflags,*s++);
6106 PL_sublex_info.super_bufptr = s;
6107 PL_sublex_info.super_bufend = PL_bufend;
6109 pm->op_pmflags |= PMf_EVAL;
6110 repl = newSVpvn("",0);
6112 sv_catpv(repl, es ? "eval " : "do ");
6113 sv_catpvn(repl, "{ ", 2);
6114 sv_catsv(repl, PL_lex_repl);
6115 sv_catpvn(repl, " };", 2);
6117 SvREFCNT_dec(PL_lex_repl);
6121 pm->op_pmpermflags = pm->op_pmflags;
6122 PL_lex_op = (OP*)pm;
6123 yylval.ival = OP_SUBST;
6128 S_scan_trans(pTHX_ char *start)
6139 yylval.ival = OP_NULL;
6141 s = scan_str(start,FALSE,FALSE);
6144 SvREFCNT_dec(PL_lex_stuff);
6145 PL_lex_stuff = Nullsv;
6146 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6148 if (s[-1] == PL_multi_open)
6151 s = scan_str(s,FALSE,FALSE);
6154 SvREFCNT_dec(PL_lex_stuff);
6155 PL_lex_stuff = Nullsv;
6157 SvREFCNT_dec(PL_lex_repl);
6158 PL_lex_repl = Nullsv;
6159 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6162 New(803,tbl,256,short);
6163 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6165 complement = del = squash = 0;
6166 while (strchr("cds", *s)) {
6168 complement = OPpTRANS_COMPLEMENT;
6170 del = OPpTRANS_DELETE;
6172 squash = OPpTRANS_SQUASH;
6175 o->op_private = del|squash|complement;
6178 yylval.ival = OP_TRANS;
6183 S_scan_heredoc(pTHX_ register char *s)
6187 I32 op_type = OP_SCALAR;
6194 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6198 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6201 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6202 if (*peek && strchr("`'\"",*peek)) {
6205 s = delimcpy(d, e, s, PL_bufend, term, &len);
6215 if (!isALNUM_lazy_if(s,UTF))
6216 deprecate("bare << to mean <<\"\"");
6217 for (; isALNUM_lazy_if(s,UTF); s++) {
6222 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6223 Perl_croak(aTHX_ "Delimiter for here document is too long");
6226 len = d - PL_tokenbuf;
6227 #ifndef PERL_STRICT_CR
6228 d = strchr(s, '\r');
6232 while (s < PL_bufend) {
6238 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6247 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6252 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6253 herewas = newSVpvn(s,PL_bufend-s);
6255 s--, herewas = newSVpvn(s,d-s);
6256 s += SvCUR(herewas);
6258 tmpstr = NEWSV(87,79);
6259 sv_upgrade(tmpstr, SVt_PVIV);
6264 else if (term == '`') {
6265 op_type = OP_BACKTICK;
6266 SvIVX(tmpstr) = '\\';
6270 PL_multi_start = CopLINE(PL_curcop);
6271 PL_multi_open = PL_multi_close = '<';
6272 term = *PL_tokenbuf;
6273 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6274 char *bufptr = PL_sublex_info.super_bufptr;
6275 char *bufend = PL_sublex_info.super_bufend;
6276 char *olds = s - SvCUR(herewas);
6277 s = strchr(bufptr, '\n');
6281 while (s < bufend &&
6282 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6284 CopLINE_inc(PL_curcop);
6287 CopLINE_set(PL_curcop, PL_multi_start);
6288 missingterm(PL_tokenbuf);
6290 sv_setpvn(herewas,bufptr,d-bufptr+1);
6291 sv_setpvn(tmpstr,d+1,s-d);
6293 sv_catpvn(herewas,s,bufend-s);
6294 (void)strcpy(bufptr,SvPVX(herewas));
6301 while (s < PL_bufend &&
6302 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6304 CopLINE_inc(PL_curcop);
6306 if (s >= PL_bufend) {
6307 CopLINE_set(PL_curcop, PL_multi_start);
6308 missingterm(PL_tokenbuf);
6310 sv_setpvn(tmpstr,d+1,s-d);
6312 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6314 sv_catpvn(herewas,s,PL_bufend-s);
6315 sv_setsv(PL_linestr,herewas);
6316 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6317 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6320 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6321 while (s >= PL_bufend) { /* multiple line string? */
6323 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6324 CopLINE_set(PL_curcop, PL_multi_start);
6325 missingterm(PL_tokenbuf);
6327 CopLINE_inc(PL_curcop);
6328 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6329 #ifndef PERL_STRICT_CR
6330 if (PL_bufend - PL_linestart >= 2) {
6331 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6332 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6334 PL_bufend[-2] = '\n';
6336 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6338 else if (PL_bufend[-1] == '\r')
6339 PL_bufend[-1] = '\n';
6341 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6342 PL_bufend[-1] = '\n';
6344 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6345 SV *sv = NEWSV(88,0);
6347 sv_upgrade(sv, SVt_PVMG);
6348 sv_setsv(sv,PL_linestr);
6349 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6351 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6354 sv_catsv(PL_linestr,herewas);
6355 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6359 sv_catsv(tmpstr,PL_linestr);
6364 PL_multi_end = CopLINE(PL_curcop);
6365 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6366 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6367 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6369 SvREFCNT_dec(herewas);
6370 PL_lex_stuff = tmpstr;
6371 yylval.ival = op_type;
6376 takes: current position in input buffer
6377 returns: new position in input buffer
6378 side-effects: yylval and lex_op are set.
6383 <FH> read from filehandle
6384 <pkg::FH> read from package qualified filehandle
6385 <pkg'FH> read from package qualified filehandle
6386 <$fh> read from filehandle in $fh
6392 S_scan_inputsymbol(pTHX_ char *start)
6394 register char *s = start; /* current position in buffer */
6400 d = PL_tokenbuf; /* start of temp holding space */
6401 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6402 end = strchr(s, '\n');
6405 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6407 /* die if we didn't have space for the contents of the <>,
6408 or if it didn't end, or if we see a newline
6411 if (len >= sizeof PL_tokenbuf)
6412 Perl_croak(aTHX_ "Excessively long <> operator");
6414 Perl_croak(aTHX_ "Unterminated <> operator");
6419 Remember, only scalar variables are interpreted as filehandles by
6420 this code. Anything more complex (e.g., <$fh{$num}>) will be
6421 treated as a glob() call.
6422 This code makes use of the fact that except for the $ at the front,
6423 a scalar variable and a filehandle look the same.
6425 if (*d == '$' && d[1]) d++;
6427 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6428 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6431 /* If we've tried to read what we allow filehandles to look like, and
6432 there's still text left, then it must be a glob() and not a getline.
6433 Use scan_str to pull out the stuff between the <> and treat it
6434 as nothing more than a string.
6437 if (d - PL_tokenbuf != len) {
6438 yylval.ival = OP_GLOB;
6440 s = scan_str(start,FALSE,FALSE);
6442 Perl_croak(aTHX_ "Glob not terminated");
6446 /* we're in a filehandle read situation */
6449 /* turn <> into <ARGV> */
6451 (void)strcpy(d,"ARGV");
6453 /* if <$fh>, create the ops to turn the variable into a
6459 /* try to find it in the pad for this block, otherwise find
6460 add symbol table ops
6462 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6463 OP *o = newOP(OP_PADSV, 0);
6465 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6468 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6469 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6470 newUNOP(OP_RV2SV, 0,
6471 newGVOP(OP_GV, 0, gv)));
6473 PL_lex_op->op_flags |= OPf_SPECIAL;
6474 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6475 yylval.ival = OP_NULL;
6478 /* If it's none of the above, it must be a literal filehandle
6479 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6481 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6482 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6483 yylval.ival = OP_NULL;
6492 takes: start position in buffer
6493 keep_quoted preserve \ on the embedded delimiter(s)
6494 keep_delims preserve the delimiters around the string
6495 returns: position to continue reading from buffer
6496 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6497 updates the read buffer.
6499 This subroutine pulls a string out of the input. It is called for:
6500 q single quotes q(literal text)
6501 ' single quotes 'literal text'
6502 qq double quotes qq(interpolate $here please)
6503 " double quotes "interpolate $here please"
6504 qx backticks qx(/bin/ls -l)
6505 ` backticks `/bin/ls -l`
6506 qw quote words @EXPORT_OK = qw( func() $spam )
6507 m// regexp match m/this/
6508 s/// regexp substitute s/this/that/
6509 tr/// string transliterate tr/this/that/
6510 y/// string transliterate y/this/that/
6511 ($*@) sub prototypes sub foo ($)
6512 (stuff) sub attr parameters sub foo : attr(stuff)
6513 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6515 In most of these cases (all but <>, patterns and transliterate)
6516 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6517 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6518 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6521 It skips whitespace before the string starts, and treats the first
6522 character as the delimiter. If the delimiter is one of ([{< then
6523 the corresponding "close" character )]}> is used as the closing
6524 delimiter. It allows quoting of delimiters, and if the string has
6525 balanced delimiters ([{<>}]) it allows nesting.
6527 The lexer always reads these strings into lex_stuff, except in the
6528 case of the operators which take *two* arguments (s/// and tr///)
6529 when it checks to see if lex_stuff is full (presumably with the 1st
6530 arg to s or tr) and if so puts the string into lex_repl.
6535 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6538 SV *sv; /* scalar value: string */
6539 char *tmps; /* temp string, used for delimiter matching */
6540 register char *s = start; /* current position in the buffer */
6541 register char term; /* terminating character */
6542 register char *to; /* current position in the sv's data */
6543 I32 brackets = 1; /* bracket nesting level */
6544 bool has_utf = FALSE; /* is there any utf8 content? */
6546 /* skip space before the delimiter */
6550 /* mark where we are, in case we need to report errors */
6553 /* after skipping whitespace, the next character is the terminator */
6555 if ((term & 0x80) && UTF)
6558 /* mark where we are */
6559 PL_multi_start = CopLINE(PL_curcop);
6560 PL_multi_open = term;
6562 /* find corresponding closing delimiter */
6563 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6565 PL_multi_close = term;
6567 /* create a new SV to hold the contents. 87 is leak category, I'm
6568 assuming. 79 is the SV's initial length. What a random number. */
6570 sv_upgrade(sv, SVt_PVIV);
6572 (void)SvPOK_only(sv); /* validate pointer */
6574 /* move past delimiter and try to read a complete string */
6576 sv_catpvn(sv, s, 1);
6579 /* extend sv if need be */
6580 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6581 /* set 'to' to the next character in the sv's string */
6582 to = SvPVX(sv)+SvCUR(sv);
6584 /* if open delimiter is the close delimiter read unbridle */
6585 if (PL_multi_open == PL_multi_close) {
6586 for (; s < PL_bufend; s++,to++) {
6587 /* embedded newlines increment the current line number */
6588 if (*s == '\n' && !PL_rsfp)
6589 CopLINE_inc(PL_curcop);
6590 /* handle quoted delimiters */
6591 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6592 if (!keep_quoted && s[1] == term)
6594 /* any other quotes are simply copied straight through */
6598 /* terminate when run out of buffer (the for() condition), or
6599 have found the terminator */
6600 else if (*s == term)
6602 else if (!has_utf && (*s & 0x80) && UTF)
6608 /* if the terminator isn't the same as the start character (e.g.,
6609 matched brackets), we have to allow more in the quoting, and
6610 be prepared for nested brackets.
6613 /* read until we run out of string, or we find the terminator */
6614 for (; s < PL_bufend; s++,to++) {
6615 /* embedded newlines increment the line count */
6616 if (*s == '\n' && !PL_rsfp)
6617 CopLINE_inc(PL_curcop);
6618 /* backslashes can escape the open or closing characters */
6619 if (*s == '\\' && s+1 < PL_bufend) {
6621 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6626 /* allow nested opens and closes */
6627 else if (*s == PL_multi_close && --brackets <= 0)
6629 else if (*s == PL_multi_open)
6631 else if (!has_utf && (*s & 0x80) && UTF)
6636 /* terminate the copied string and update the sv's end-of-string */
6638 SvCUR_set(sv, to - SvPVX(sv));
6641 * this next chunk reads more into the buffer if we're not done yet
6645 break; /* handle case where we are done yet :-) */
6647 #ifndef PERL_STRICT_CR
6648 if (to - SvPVX(sv) >= 2) {
6649 if ((to[-2] == '\r' && to[-1] == '\n') ||
6650 (to[-2] == '\n' && to[-1] == '\r'))
6654 SvCUR_set(sv, to - SvPVX(sv));
6656 else if (to[-1] == '\r')
6659 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6663 /* if we're out of file, or a read fails, bail and reset the current
6664 line marker so we can report where the unterminated string began
6667 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6669 CopLINE_set(PL_curcop, PL_multi_start);
6672 /* we read a line, so increment our line counter */
6673 CopLINE_inc(PL_curcop);
6675 /* update debugger info */
6676 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6677 SV *sv = NEWSV(88,0);
6679 sv_upgrade(sv, SVt_PVMG);
6680 sv_setsv(sv,PL_linestr);
6681 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6684 /* having changed the buffer, we must update PL_bufend */
6685 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6688 /* at this point, we have successfully read the delimited string */
6691 sv_catpvn(sv, s, 1);
6694 PL_multi_end = CopLINE(PL_curcop);
6697 /* if we allocated too much space, give some back */
6698 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6699 SvLEN_set(sv, SvCUR(sv) + 1);
6700 Renew(SvPVX(sv), SvLEN(sv), char);
6703 /* decide whether this is the first or second quoted string we've read
6716 takes: pointer to position in buffer
6717 returns: pointer to new position in buffer
6718 side-effects: builds ops for the constant in yylval.op
6720 Read a number in any of the formats that Perl accepts:
6722 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6723 [\d_]+(\.[\d_]*)?[Ee](\d+)
6725 Underbars (_) are allowed in decimal numbers. If -w is on,
6726 underbars before a decimal point must be at three digit intervals.
6728 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6731 If it reads a number without a decimal point or an exponent, it will
6732 try converting the number to an integer and see if it can do so
6733 without loss of precision.
6737 Perl_scan_num(pTHX_ char *start, YYSTYPE* lvalp)
6739 register char *s = start; /* current position in buffer */
6740 register char *d; /* destination in temp buffer */
6741 register char *e; /* end of temp buffer */
6742 NV nv; /* number read, as a double */
6743 SV *sv = Nullsv; /* place to put the converted number */
6744 bool floatit; /* boolean: int or float? */
6745 char *lastub = 0; /* position of last underbar */
6746 static char number_too_long[] = "Number too long";
6748 /* We use the first character to decide what type of number this is */
6752 Perl_croak(aTHX_ "panic: scan_num");
6754 /* if it starts with a 0, it could be an octal number, a decimal in
6755 0.13 disguise, or a hexadecimal number, or a binary number. */
6759 u holds the "number so far"
6760 shift the power of 2 of the base
6761 (hex == 4, octal == 3, binary == 1)
6762 overflowed was the number more than we can hold?
6764 Shift is used when we add a digit. It also serves as an "are
6765 we in octal/hex/binary?" indicator to disallow hex characters
6772 bool overflowed = FALSE;
6773 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6774 static char* bases[5] = { "", "binary", "", "octal",
6776 static char* Bases[5] = { "", "Binary", "", "Octal",
6778 static char *maxima[5] = { "",
6779 "0b11111111111111111111111111111111",
6783 char *base, *Base, *max;
6789 } else if (s[1] == 'b') {
6793 /* check for a decimal in disguise */
6794 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6796 /* so it must be octal */
6800 base = bases[shift];
6801 Base = Bases[shift];
6802 max = maxima[shift];
6804 /* read the rest of the number */
6806 /* x is used in the overflow test,
6807 b is the digit we're adding on. */
6812 /* if we don't mention it, we're done */
6821 /* 8 and 9 are not octal */
6824 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6828 case '2': case '3': case '4':
6829 case '5': case '6': case '7':
6831 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6835 b = *s++ & 15; /* ASCII digit -> value of digit */
6839 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6840 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6841 /* make sure they said 0x */
6846 /* Prepare to put the digit we have onto the end
6847 of the number so far. We check for overflows.
6852 x = u << shift; /* make room for the digit */
6854 if ((x >> shift) != u
6855 && !(PL_hints & HINT_NEW_BINARY)) {
6859 if (ckWARN_d(WARN_OVERFLOW))
6860 Perl_warner(aTHX_ WARN_OVERFLOW,
6861 "Integer overflow in %s number",
6864 u = x | b; /* add the digit to the end */
6867 n *= nvshift[shift];
6868 /* If an NV has not enough bits in its
6869 * mantissa to represent an UV this summing of
6870 * small low-order numbers is a waste of time
6871 * (because the NV cannot preserve the
6872 * low-order bits anyway): we could just
6873 * remember when did we overflow and in the
6874 * end just multiply n by the right
6882 /* if we get here, we had success: make a scalar value from
6889 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6890 Perl_warner(aTHX_ WARN_PORTABLE,
6891 "%s number > %s non-portable",
6898 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6899 Perl_warner(aTHX_ WARN_PORTABLE,
6900 "%s number > %s non-portable",
6905 if (PL_hints & HINT_NEW_BINARY)
6906 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6911 handle decimal numbers.
6912 we're also sent here when we read a 0 as the first digit
6914 case '1': case '2': case '3': case '4': case '5':
6915 case '6': case '7': case '8': case '9': case '.':
6918 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
6921 /* read next group of digits and _ and copy into d */
6922 while (isDIGIT(*s) || *s == '_') {
6923 /* skip underscores, checking for misplaced ones
6927 dTHR; /* only for ckWARN */
6928 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
6929 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6933 /* check for end of fixed-length buffer */
6935 Perl_croak(aTHX_ number_too_long);
6936 /* if we're ok, copy the character */
6941 /* final misplaced underbar check */
6942 if (lastub && s - lastub != 3) {
6944 if (ckWARN(WARN_SYNTAX))
6945 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6948 /* read a decimal portion if there is one. avoid
6949 3..5 being interpreted as the number 3. followed
6952 if (*s == '.' && s[1] != '.') {
6956 /* copy, ignoring underbars, until we run out of
6957 digits. Note: no misplaced underbar checks!
6959 for (; isDIGIT(*s) || *s == '_'; s++) {
6960 /* fixed length buffer check */
6962 Perl_croak(aTHX_ number_too_long);
6966 if (*s == '.' && isDIGIT(s[1])) {
6967 /* oops, it's really a v-string, but without the "v" */
6973 /* read exponent part, if present */
6974 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
6978 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
6979 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
6981 /* allow positive or negative exponent */
6982 if (*s == '+' || *s == '-')
6985 /* read digits of exponent (no underbars :-) */
6986 while (isDIGIT(*s)) {
6988 Perl_croak(aTHX_ number_too_long);
6993 /* terminate the string */
6996 /* make an sv from the string */
6999 #if defined(Strtol) && defined(Strtoul)
7002 strtol/strtoll sets errno to ERANGE if the number is too big
7003 for an integer. We try to do an integer conversion first
7004 if no characters indicating "float" have been found.
7011 if (*PL_tokenbuf == '-')
7012 iv = Strtol(PL_tokenbuf, (char**)NULL, 10);
7014 uv = Strtoul(PL_tokenbuf, (char**)NULL, 10);
7016 floatit = TRUE; /* Probably just too large. */
7017 else if (*PL_tokenbuf == '-')
7019 else if (uv <= IV_MAX)
7020 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7025 nv = Atof(PL_tokenbuf);
7030 No working strtou?ll?.
7032 Unfortunately atol() doesn't do range checks (returning
7033 LONG_MIN/LONG_MAX, and setting errno to ERANGE on overflows)
7034 everywhere [1], so we cannot use use atol() (or atoll()).
7035 If we could, they would be used, as Atol(), very much like
7036 Strtol() and Strtoul() are used above.
7038 [1] XXX Configure test needed to check for atol()
7039 (and atoll()) overflow behaviour XXX
7043 We need to do this the hard way. */
7045 nv = Atof(PL_tokenbuf);
7047 /* See if we can make do with an integer value without loss of
7048 precision. We use U_V to cast to a UV, because some
7049 compilers have issues. Then we try casting it back and see
7050 if it was the same [1]. We only do this if we know we
7051 specifically read an integer. If floatit is true, then we
7052 don't need to do the conversion at all.
7054 [1] Note that this is lossy if our NVs cannot preserve our
7055 UVs. There are metaconfig defines NV_PRESERVES_UV (a boolean)
7056 and NV_PRESERVES_UV_BITS (a number), but in general we really
7057 do hope all such potentially lossy platforms have strtou?ll?
7058 to do a lossless IV/UV conversion.
7060 Maybe could do some tricks with DBL_DIG, LDBL_DIG and
7061 DBL_MANT_DIG and LDBL_MANT_DIG (these are already available
7062 as NV_DIG and NV_MANT_DIG)?
7068 if (!floatit && (NV)uv == nv) {
7070 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7078 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7079 (PL_hints & HINT_NEW_INTEGER) )
7080 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7081 (floatit ? "float" : "integer"),
7085 /* if it starts with a v, it could be a v-string */
7091 while (isDIGIT(*pos) || *pos == '_')
7093 if (!isALPHA(*pos)) {
7095 U8 tmpbuf[UTF8_MAXLEN];
7098 s++; /* get past 'v' */
7101 sv_setpvn(sv, "", 0);
7104 if (*s == '0' && isDIGIT(s[1]))
7105 yyerror("Octal number in vector unsupported");
7108 /* this is atoi() that tolerates underscores */
7111 while (--end >= s) {
7116 rev += (*end - '0') * mult;
7118 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
7119 Perl_warner(aTHX_ WARN_OVERFLOW,
7120 "Integer overflow in decimal number");
7123 tmpend = uv_to_utf8(tmpbuf, rev);
7124 utf8 = utf8 || rev > 127;
7125 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
7126 if (*pos == '.' && isDIGIT(pos[1]))
7132 while (isDIGIT(*pos) || *pos == '_')
7140 sv_utf8_downgrade(sv, TRUE);
7147 /* make the op for the constant and return */
7150 lvalp->opval = newSVOP(OP_CONST, 0, sv);
7152 lvalp->opval = Nullop;
7158 S_scan_formline(pTHX_ register char *s)
7163 SV *stuff = newSVpvn("",0);
7164 bool needargs = FALSE;
7167 if (*s == '.' || *s == /*{*/'}') {
7169 #ifdef PERL_STRICT_CR
7170 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7172 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7174 if (*t == '\n' || t == PL_bufend)
7177 if (PL_in_eval && !PL_rsfp) {
7178 eol = strchr(s,'\n');
7183 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7185 for (t = s; t < eol; t++) {
7186 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7188 goto enough; /* ~~ must be first line in formline */
7190 if (*t == '@' || *t == '^')
7193 sv_catpvn(stuff, s, eol-s);
7194 #ifndef PERL_STRICT_CR
7195 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7196 char *end = SvPVX(stuff) + SvCUR(stuff);
7205 s = filter_gets(PL_linestr, PL_rsfp, 0);
7206 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7207 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7210 yyerror("Format not terminated");
7220 PL_lex_state = LEX_NORMAL;
7221 PL_nextval[PL_nexttoke].ival = 0;
7225 PL_lex_state = LEX_FORMLINE;
7226 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7228 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7232 SvREFCNT_dec(stuff);
7233 PL_lex_formbrack = 0;
7244 PL_cshlen = strlen(PL_cshname);
7249 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7252 I32 oldsavestack_ix = PL_savestack_ix;
7253 CV* outsidecv = PL_compcv;
7257 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7259 SAVEI32(PL_subline);
7260 save_item(PL_subname);
7263 SAVESPTR(PL_comppad_name);
7264 SAVESPTR(PL_compcv);
7265 SAVEI32(PL_comppad_name_fill);
7266 SAVEI32(PL_min_intro_pending);
7267 SAVEI32(PL_max_intro_pending);
7268 SAVEI32(PL_pad_reset_pending);
7270 PL_compcv = (CV*)NEWSV(1104,0);
7271 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7272 CvFLAGS(PL_compcv) |= flags;
7274 PL_comppad = newAV();
7275 av_push(PL_comppad, Nullsv);
7276 PL_curpad = AvARRAY(PL_comppad);
7277 PL_comppad_name = newAV();
7278 PL_comppad_name_fill = 0;
7279 PL_min_intro_pending = 0;
7281 PL_subline = CopLINE(PL_curcop);
7283 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7284 PL_curpad[0] = (SV*)newAV();
7285 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7286 #endif /* USE_THREADS */
7288 comppadlist = newAV();
7289 AvREAL_off(comppadlist);
7290 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7291 av_store(comppadlist, 1, (SV*)PL_comppad);
7293 CvPADLIST(PL_compcv) = comppadlist;
7294 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7296 CvOWNER(PL_compcv) = 0;
7297 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7298 MUTEX_INIT(CvMUTEXP(PL_compcv));
7299 #endif /* USE_THREADS */
7301 return oldsavestack_ix;
7305 Perl_yywarn(pTHX_ char *s)
7308 PL_in_eval |= EVAL_WARNONLY;
7310 PL_in_eval &= ~EVAL_WARNONLY;
7315 Perl_yyerror(pTHX_ char *s)
7319 char *context = NULL;
7323 if (!yychar || (yychar == ';' && !PL_rsfp))
7325 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7326 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7327 while (isSPACE(*PL_oldoldbufptr))
7329 context = PL_oldoldbufptr;
7330 contlen = PL_bufptr - PL_oldoldbufptr;
7332 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7333 PL_oldbufptr != PL_bufptr) {
7334 while (isSPACE(*PL_oldbufptr))
7336 context = PL_oldbufptr;
7337 contlen = PL_bufptr - PL_oldbufptr;
7339 else if (yychar > 255)
7340 where = "next token ???";
7341 #ifdef USE_PURE_BISON
7342 /* GNU Bison sets the value -2 */
7343 else if (yychar == -2) {
7345 else if ((yychar & 127) == 127) {
7347 if (PL_lex_state == LEX_NORMAL ||
7348 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7349 where = "at end of line";
7350 else if (PL_lex_inpat)
7351 where = "within pattern";
7353 where = "within string";
7356 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7358 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7359 else if (isPRINT_LC(yychar))
7360 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7362 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7363 where = SvPVX(where_sv);
7365 msg = sv_2mortal(newSVpv(s, 0));
7366 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7367 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7369 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7371 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7372 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7373 Perl_sv_catpvf(aTHX_ msg,
7374 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7375 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7378 if (PL_in_eval & EVAL_WARNONLY)
7379 Perl_warn(aTHX_ "%"SVf, msg);
7382 if (PL_error_count >= 10) {
7383 if (PL_in_eval && SvCUR(ERRSV))
7384 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
7385 ERRSV, CopFILE(PL_curcop));
7387 Perl_croak(aTHX_ "%s has too many errors.\n",
7388 CopFILE(PL_curcop));
7391 PL_in_my_stash = Nullhv;
7396 S_swallow_bom(pTHX_ U8 *s)
7399 slen = SvCUR(PL_linestr);
7403 /* UTF-16 little-endian */
7404 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
7405 Perl_croak(aTHX_ "Unsupported script encoding");
7406 #ifndef PERL_NO_UTF16_FILTER
7407 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-LE script encoding\n"));
7409 if (PL_bufend > (char*)s) {
7413 filter_add(utf16rev_textfilter, NULL);
7414 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7415 PL_bufend = (char*)utf16_to_utf8_reversed(s, news,
7416 PL_bufend - (char*)s - 1,
7418 Copy(news, s, newlen, U8);
7419 SvCUR_set(PL_linestr, newlen);
7420 PL_bufend = SvPVX(PL_linestr) + newlen;
7421 news[newlen++] = '\0';
7425 Perl_croak(aTHX_ "Unsupported script encoding");
7430 if (s[1] == 0xFF) { /* UTF-16 big-endian */
7431 #ifndef PERL_NO_UTF16_FILTER
7432 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding\n"));
7434 if (PL_bufend > (char *)s) {
7438 filter_add(utf16_textfilter, NULL);
7439 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7440 PL_bufend = (char*)utf16_to_utf8(s, news,
7441 PL_bufend - (char*)s,
7443 Copy(news, s, newlen, U8);
7444 SvCUR_set(PL_linestr, newlen);
7445 PL_bufend = SvPVX(PL_linestr) + newlen;
7446 news[newlen++] = '\0';
7450 Perl_croak(aTHX_ "Unsupported script encoding");
7455 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
7456 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-8 script encoding\n"));
7461 if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */
7462 s[2] == 0xFE && s[3] == 0xFF)
7464 Perl_croak(aTHX_ "Unsupported script encoding");
7476 * Restore a source filter.
7480 restore_rsfp(pTHXo_ void *f)
7482 PerlIO *fp = (PerlIO*)f;
7484 if (PL_rsfp == PerlIO_stdin())
7485 PerlIO_clearerr(PL_rsfp);
7486 else if (PL_rsfp && (PL_rsfp != fp))
7487 PerlIO_close(PL_rsfp);
7491 #ifndef PERL_NO_UTF16_FILTER
7493 utf16_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7495 I32 count = FILTER_READ(idx+1, sv, maxlen);
7500 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7501 if (!*SvPV_nolen(sv))
7502 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7505 tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7506 sv_usepvn(sv, (char*)tmps, tend - tmps);
7512 utf16rev_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7514 I32 count = FILTER_READ(idx+1, sv, maxlen);
7519 if (!*SvPV_nolen(sv))
7520 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7523 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7524 tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7525 sv_usepvn(sv, (char*)tmps, tend - tmps);