3 * Copyright (c) 1991-2001, 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
78 # define YYMAXLEVEL 100
80 YYSTYPE* yylval_pointer[YYMAXLEVEL];
81 int* yychar_pointer[YYMAXLEVEL];
85 # define yylval (*yylval_pointer[yyactlevel])
86 # define yychar (*yychar_pointer[yyactlevel])
87 # define PERL_YYLEX_PARAM yylval_pointer[yyactlevel],yychar_pointer[yyactlevel]
89 # define yylex() Perl_yylex_r(aTHX_ yylval_pointer[yyactlevel],yychar_pointer[yyactlevel])
94 /* CLINE is a macro that ensures PL_copline has a sane value */
99 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
102 * Convenience functions to return different tokens and prime the
103 * lexer for the next token. They all take an argument.
105 * TOKEN : generic token (used for '(', DOLSHARP, etc)
106 * OPERATOR : generic operator
107 * AOPERATOR : assignment operator
108 * PREBLOCK : beginning the block after an if, while, foreach, ...
109 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
110 * PREREF : *EXPR where EXPR is not a simple identifier
111 * TERM : expression term
112 * LOOPX : loop exiting command (goto, last, dump, etc)
113 * FTST : file test operator
114 * FUN0 : zero-argument function
115 * FUN1 : not used, except for not, which isn't a UNIOP
116 * BOop : bitwise or or xor
118 * SHop : shift operator
119 * PWop : power operator
120 * PMop : pattern-matching operator
121 * Aop : addition-level operator
122 * Mop : multiplication-level operator
123 * Eop : equality-testing operator
124 * Rop : relational operator <= != gt
126 * Also see LOP and lop() below.
129 #ifdef DEBUGGING /* Serve -DT. */
130 # define REPORT(x,retval) tokereport(x,s,(int)retval)
131 # define REPORT2(x,retval) tokereport(x,s, yylval.ival)
133 # define REPORT(x,retval) 1
134 # define REPORT2(x,retval) 1
137 #define TOKEN(retval) return (REPORT2("token",retval), PL_bufptr = s,(int)retval)
138 #define OPERATOR(retval) return (REPORT2("operator",retval), PL_expect = XTERM, PL_bufptr = s,(int)retval)
139 #define AOPERATOR(retval) return ao((REPORT2("aop",retval), PL_expect = XTERM, PL_bufptr = s,(int)retval))
140 #define PREBLOCK(retval) return (REPORT2("preblock",retval), PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
141 #define PRETERMBLOCK(retval) return (REPORT2("pretermblock",retval), PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
142 #define PREREF(retval) return (REPORT2("preref",retval), PL_expect = XREF,PL_bufptr = s,(int)retval)
143 #define TERM(retval) return (CLINE, REPORT2("term",retval), PL_expect = XOPERATOR, PL_bufptr = s,(int)retval)
144 #define LOOPX(f) return(yylval.ival=f, REPORT("loopx",f), PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
145 #define FTST(f) return(yylval.ival=f, REPORT("ftst",f), PL_expect = XTERM,PL_bufptr = s,(int)UNIOP)
146 #define FUN0(f) return(yylval.ival = f, REPORT("fun0",f), PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
147 #define FUN1(f) return(yylval.ival = f, REPORT("fun1",f), PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
148 #define BOop(f) return ao((yylval.ival=f, REPORT("bitorop",f), PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
149 #define BAop(f) return ao((yylval.ival=f, REPORT("bitandop",f), PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
150 #define SHop(f) return ao((yylval.ival=f, REPORT("shiftop",f), PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
151 #define PWop(f) return ao((yylval.ival=f, REPORT("powop",f), PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
152 #define PMop(f) return(yylval.ival=f, REPORT("matchop",f), PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
153 #define Aop(f) return ao((yylval.ival=f, REPORT("add",f), PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
154 #define Mop(f) return ao((yylval.ival=f, REPORT("mul",f), PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
155 #define Eop(f) return(yylval.ival=f, REPORT("eq",f), PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
156 #define Rop(f) return(yylval.ival=f, REPORT("rel",f), PL_expect = XTERM,PL_bufptr = s,(int)RELOP)
158 /* This bit of chicanery makes a unary function followed by
159 * a parenthesis into a function with one argument, highest precedence.
161 #define UNI(f) return(yylval.ival = f, \
165 PL_last_uni = PL_oldbufptr, \
166 PL_last_lop_op = f, \
167 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
169 #define UNIBRACK(f) return(yylval.ival = f, \
172 PL_last_uni = PL_oldbufptr, \
173 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
175 /* grandfather return to old style */
176 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
179 S_tokereport(pTHX_ char *thing, char* s, I32 rv)
183 report = newSVpv(thing, 0);
184 Perl_sv_catpvf(aTHX_ report, ":line %i:%i:", CopLINE(PL_curcop), rv);
186 if (s - PL_bufptr > 0)
187 sv_catpvn(report, PL_bufptr, s - PL_bufptr);
189 if (PL_oldbufptr && *PL_oldbufptr)
190 sv_catpv(report, PL_tokenbuf);
192 PerlIO_printf(Perl_debug_log, "### %s\n", SvPV_nolen(report));
199 * This subroutine detects &&= and ||= and turns an ANDAND or OROR
200 * into an OP_ANDASSIGN or OP_ORASSIGN
204 S_ao(pTHX_ int toketype)
206 if (*PL_bufptr == '=') {
208 if (toketype == ANDAND)
209 yylval.ival = OP_ANDASSIGN;
210 else if (toketype == OROR)
211 yylval.ival = OP_ORASSIGN;
219 * When Perl expects an operator and finds something else, no_op
220 * prints the warning. It always prints "<something> found where
221 * operator expected. It prints "Missing semicolon on previous line?"
222 * if the surprise occurs at the start of the line. "do you need to
223 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
224 * where the compiler doesn't know if foo is a method call or a function.
225 * It prints "Missing operator before end of line" if there's nothing
226 * after the missing operator, or "... before <...>" if there is something
227 * after the missing operator.
231 S_no_op(pTHX_ char *what, char *s)
233 char *oldbp = PL_bufptr;
234 bool is_first = (PL_oldbufptr == PL_linestart);
240 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
242 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
243 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
245 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
246 if (t < PL_bufptr && isSPACE(*t))
247 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
248 t - PL_oldoldbufptr, PL_oldoldbufptr);
252 Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
259 * Complain about missing quote/regexp/heredoc terminator.
260 * If it's called with (char *)NULL then it cauterizes the line buffer.
261 * If we're in a delimited string and the delimiter is a control
262 * character, it's reformatted into a two-char sequence like ^C.
267 S_missingterm(pTHX_ char *s)
272 char *nl = strrchr(s,'\n');
278 iscntrl(PL_multi_close)
280 PL_multi_close < 32 || PL_multi_close == 127
284 tmpbuf[1] = toCTRL(PL_multi_close);
290 *tmpbuf = PL_multi_close;
294 q = strchr(s,'"') ? '\'' : '"';
295 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
303 Perl_deprecate(pTHX_ char *s)
305 if (ckWARN(WARN_DEPRECATED))
306 Perl_warner(aTHX_ WARN_DEPRECATED, "Use of %s is deprecated", s);
311 * Deprecate a comma-less variable list.
317 deprecate("comma-less variable list");
321 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
322 * utf16-to-utf8-reversed.
325 #ifdef PERL_CR_FILTER
329 register char *s = SvPVX(sv);
330 register char *e = s + SvCUR(sv);
331 /* outer loop optimized to do nothing if there are no CR-LFs */
333 if (*s++ == '\r' && *s == '\n') {
334 /* hit a CR-LF, need to copy the rest */
335 register char *d = s - 1;
338 if (*s == '\r' && s[1] == '\n')
349 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
351 I32 count = FILTER_READ(idx+1, sv, maxlen);
352 if (count > 0 && !maxlen)
360 * Initialize variables. Uses the Perl save_stack to save its state (for
361 * recursive calls to the parser).
365 Perl_lex_start(pTHX_ SV *line)
370 SAVEI32(PL_lex_dojoin);
371 SAVEI32(PL_lex_brackets);
372 SAVEI32(PL_lex_casemods);
373 SAVEI32(PL_lex_starts);
374 SAVEI32(PL_lex_state);
375 SAVEVPTR(PL_lex_inpat);
376 SAVEI32(PL_lex_inwhat);
377 if (PL_lex_state == LEX_KNOWNEXT) {
378 I32 toke = PL_nexttoke;
379 while (--toke >= 0) {
380 SAVEI32(PL_nexttype[toke]);
381 SAVEVPTR(PL_nextval[toke]);
383 SAVEI32(PL_nexttoke);
385 SAVECOPLINE(PL_curcop);
388 SAVEPPTR(PL_oldbufptr);
389 SAVEPPTR(PL_oldoldbufptr);
390 SAVEPPTR(PL_linestart);
391 SAVESPTR(PL_linestr);
392 SAVEPPTR(PL_lex_brackstack);
393 SAVEPPTR(PL_lex_casestack);
394 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
395 SAVESPTR(PL_lex_stuff);
396 SAVEI32(PL_lex_defer);
397 SAVEI32(PL_sublex_info.sub_inwhat);
398 SAVESPTR(PL_lex_repl);
400 SAVEINT(PL_lex_expect);
402 PL_lex_state = LEX_NORMAL;
406 New(899, PL_lex_brackstack, 120, char);
407 New(899, PL_lex_casestack, 12, char);
408 SAVEFREEPV(PL_lex_brackstack);
409 SAVEFREEPV(PL_lex_casestack);
411 *PL_lex_casestack = '\0';
414 PL_lex_stuff = Nullsv;
415 PL_lex_repl = Nullsv;
419 PL_sublex_info.sub_inwhat = 0;
421 if (SvREADONLY(PL_linestr))
422 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
423 s = SvPV(PL_linestr, len);
424 if (len && s[len-1] != ';') {
425 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
426 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
427 sv_catpvn(PL_linestr, "\n;", 2);
429 SvTEMP_off(PL_linestr);
430 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
431 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
433 PL_rs = newSVpvn("\n", 1);
439 * Finalizer for lexing operations. Must be called when the parser is
440 * done with the lexer.
446 PL_doextract = FALSE;
451 * This subroutine has nothing to do with tilting, whether at windmills
452 * or pinball tables. Its name is short for "increment line". It
453 * increments the current line number in CopLINE(PL_curcop) and checks
454 * to see whether the line starts with a comment of the form
455 * # line 500 "foo.pm"
456 * If so, it sets the current line number and file to the values in the comment.
460 S_incline(pTHX_ char *s)
467 CopLINE_inc(PL_curcop);
470 while (SPACE_OR_TAB(*s)) s++;
471 if (strnEQ(s, "line", 4))
475 if (SPACE_OR_TAB(*s))
479 while (SPACE_OR_TAB(*s)) s++;
485 while (SPACE_OR_TAB(*s))
487 if (*s == '"' && (t = strchr(s+1, '"'))) {
492 for (t = s; !isSPACE(*t); t++) ;
495 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
497 if (*e != '\n' && *e != '\0')
498 return; /* false alarm */
504 Safefree(CopFILE(PL_curcop));
506 SvREFCNT_dec(CopFILEGV(PL_curcop));
508 CopFILE_set(PL_curcop, s);
511 CopLINE_set(PL_curcop, atoi(n)-1);
516 * Called to gobble the appropriate amount and type of whitespace.
517 * Skips comments as well.
521 S_skipspace(pTHX_ register char *s)
523 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
524 while (s < PL_bufend && SPACE_OR_TAB(*s))
530 SSize_t oldprevlen, oldoldprevlen;
531 SSize_t oldloplen, oldunilen;
532 while (s < PL_bufend && isSPACE(*s)) {
533 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
538 if (s < PL_bufend && *s == '#') {
539 while (s < PL_bufend && *s != '\n')
543 if (PL_in_eval && !PL_rsfp) {
550 /* only continue to recharge the buffer if we're at the end
551 * of the buffer, we're not reading from a source filter, and
552 * we're in normal lexing mode
554 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
555 PL_lex_state == LEX_FORMLINE)
558 /* try to recharge the buffer */
559 if ((s = filter_gets(PL_linestr, PL_rsfp,
560 (prevlen = SvCUR(PL_linestr)))) == Nullch)
562 /* end of file. Add on the -p or -n magic */
563 if (PL_minus_n || PL_minus_p) {
564 sv_setpv(PL_linestr,PL_minus_p ?
565 ";}continue{print or die qq(-p destination: $!\\n)" :
567 sv_catpv(PL_linestr,";}");
568 PL_minus_n = PL_minus_p = 0;
571 sv_setpv(PL_linestr,";");
573 /* reset variables for next time we lex */
574 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
576 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
578 /* Close the filehandle. Could be from -P preprocessor,
579 * STDIN, or a regular file. If we were reading code from
580 * STDIN (because the commandline held no -e or filename)
581 * then we don't close it, we reset it so the code can
582 * read from STDIN too.
585 if (PL_preprocess && !PL_in_eval)
586 (void)PerlProc_pclose(PL_rsfp);
587 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
588 PerlIO_clearerr(PL_rsfp);
590 (void)PerlIO_close(PL_rsfp);
595 /* not at end of file, so we only read another line */
596 /* make corresponding updates to old pointers, for yyerror() */
597 oldprevlen = PL_oldbufptr - PL_bufend;
598 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
600 oldunilen = PL_last_uni - PL_bufend;
602 oldloplen = PL_last_lop - PL_bufend;
603 PL_linestart = PL_bufptr = s + prevlen;
604 PL_bufend = s + SvCUR(PL_linestr);
606 PL_oldbufptr = s + oldprevlen;
607 PL_oldoldbufptr = s + oldoldprevlen;
609 PL_last_uni = s + oldunilen;
611 PL_last_lop = s + oldloplen;
614 /* debugger active and we're not compiling the debugger code,
615 * so store the line into the debugger's array of lines
617 if (PERLDB_LINE && PL_curstash != PL_debstash) {
618 SV *sv = NEWSV(85,0);
620 sv_upgrade(sv, SVt_PVMG);
621 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
622 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
629 * Check the unary operators to ensure there's no ambiguity in how they're
630 * used. An ambiguous piece of code would be:
632 * This doesn't mean rand() + 5. Because rand() is a unary operator,
633 * the +5 is its argument.
642 if (PL_oldoldbufptr != PL_last_uni)
644 while (isSPACE(*PL_last_uni))
646 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
647 if ((t = strchr(s, '(')) && t < PL_bufptr)
649 if (ckWARN_d(WARN_AMBIGUOUS)){
652 Perl_warner(aTHX_ WARN_AMBIGUOUS,
653 "Warning: Use of \"%s\" without parens is ambiguous",
659 /* workaround to replace the UNI() macro with a function. Only the
660 * hints/uts.sh file mentions this. Other comments elsewhere in the
661 * source indicate Microport Unix might need it too.
667 #define UNI(f) return uni(f,s)
670 S_uni(pTHX_ I32 f, char *s)
675 PL_last_uni = PL_oldbufptr;
686 #endif /* CRIPPLED_CC */
689 * LOP : macro to build a list operator. Its behaviour has been replaced
690 * with a subroutine, S_lop() for which LOP is just another name.
693 #define LOP(f,x) return lop(f,x,s)
697 * Build a list operator (or something that might be one). The rules:
698 * - if we have a next token, then it's a list operator [why?]
699 * - if the next thing is an opening paren, then it's a function
700 * - else it's a list operator
704 S_lop(pTHX_ I32 f, int x, char *s)
711 PL_last_lop = PL_oldbufptr;
726 * When the lexer realizes it knows the next token (for instance,
727 * it is reordering tokens for the parser) then it can call S_force_next
728 * to know what token to return the next time the lexer is called. Caller
729 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
730 * handles the token correctly.
734 S_force_next(pTHX_ I32 type)
736 PL_nexttype[PL_nexttoke] = type;
738 if (PL_lex_state != LEX_KNOWNEXT) {
739 PL_lex_defer = PL_lex_state;
740 PL_lex_expect = PL_expect;
741 PL_lex_state = LEX_KNOWNEXT;
747 * When the lexer knows the next thing is a word (for instance, it has
748 * just seen -> and it knows that the next char is a word char, then
749 * it calls S_force_word to stick the next word into the PL_next lookahead.
752 * char *start : buffer position (must be within PL_linestr)
753 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
754 * int check_keyword : if true, Perl checks to make sure the word isn't
755 * a keyword (do this if the word is a label, e.g. goto FOO)
756 * int allow_pack : if true, : characters will also be allowed (require,
758 * int allow_initial_tick : used by the "sub" lexer only.
762 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
767 start = skipspace(start);
769 if (isIDFIRST_lazy_if(s,UTF) ||
770 (allow_pack && *s == ':') ||
771 (allow_initial_tick && *s == '\'') )
773 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
774 if (check_keyword && keyword(PL_tokenbuf, len))
776 if (token == METHOD) {
781 PL_expect = XOPERATOR;
784 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
785 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
793 * Called when the lexer wants $foo *foo &foo etc, but the program
794 * text only contains the "foo" portion. The first argument is a pointer
795 * to the "foo", and the second argument is the type symbol to prefix.
796 * Forces the next token to be a "WORD".
797 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
801 S_force_ident(pTHX_ register char *s, int kind)
804 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
805 PL_nextval[PL_nexttoke].opval = o;
808 o->op_private = OPpCONST_ENTERED;
809 /* XXX see note in pp_entereval() for why we forgo typo
810 warnings if the symbol must be introduced in an eval.
812 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
813 kind == '$' ? SVt_PV :
814 kind == '@' ? SVt_PVAV :
815 kind == '%' ? SVt_PVHV :
823 Perl_str_to_version(pTHX_ SV *sv)
828 char *start = SvPVx(sv,len);
829 bool utf = SvUTF8(sv) ? TRUE : FALSE;
830 char *end = start + len;
831 while (start < end) {
835 n = utf8_to_uv((U8*)start, len, &skip, 0);
840 retval += ((NV)n)/nshift;
849 * Forces the next token to be a version number.
853 S_force_version(pTHX_ char *s)
855 OP *version = Nullop;
864 for (; isDIGIT(*d) || *d == '_' || *d == '.'; d++);
865 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
867 s = scan_num(s, &yylval);
868 version = yylval.opval;
869 ver = cSVOPx(version)->op_sv;
870 if (SvPOK(ver) && !SvNIOK(ver)) {
871 (void)SvUPGRADE(ver, SVt_PVNV);
872 SvNVX(ver) = str_to_version(ver);
873 SvNOK_on(ver); /* hint that it is a version */
878 /* NOTE: The parser sees the package name and the VERSION swapped */
879 PL_nextval[PL_nexttoke].opval = version;
887 * Tokenize a quoted string passed in as an SV. It finds the next
888 * chunk, up to end of string or a backslash. It may make a new
889 * SV containing that chunk (if HINT_NEW_STRING is on). It also
894 S_tokeq(pTHX_ SV *sv)
905 s = SvPV_force(sv, len);
906 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
909 while (s < send && *s != '\\')
914 if ( PL_hints & HINT_NEW_STRING )
915 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
918 if (s + 1 < send && (s[1] == '\\'))
919 s++; /* all that, just for this */
924 SvCUR_set(sv, d - SvPVX(sv));
926 if ( PL_hints & HINT_NEW_STRING )
927 return new_constant(NULL, 0, "q", sv, pv, "q");
932 * Now come three functions related to double-quote context,
933 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
934 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
935 * interact with PL_lex_state, and create fake ( ... ) argument lists
936 * to handle functions and concatenation.
937 * They assume that whoever calls them will be setting up a fake
938 * join call, because each subthing puts a ',' after it. This lets
941 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
943 * (I'm not sure whether the spurious commas at the end of lcfirst's
944 * arguments and join's arguments are created or not).
949 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
951 * Pattern matching will set PL_lex_op to the pattern-matching op to
952 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
954 * OP_CONST and OP_READLINE are easy--just make the new op and return.
956 * Everything else becomes a FUNC.
958 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
959 * had an OP_CONST or OP_READLINE). This just sets us up for a
960 * call to S_sublex_push().
966 register I32 op_type = yylval.ival;
968 if (op_type == OP_NULL) {
969 yylval.opval = PL_lex_op;
973 if (op_type == OP_CONST || op_type == OP_READLINE) {
974 SV *sv = tokeq(PL_lex_stuff);
976 if (SvTYPE(sv) == SVt_PVIV) {
977 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
983 nsv = newSVpvn(p, len);
989 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
990 PL_lex_stuff = Nullsv;
994 PL_sublex_info.super_state = PL_lex_state;
995 PL_sublex_info.sub_inwhat = op_type;
996 PL_sublex_info.sub_op = PL_lex_op;
997 PL_lex_state = LEX_INTERPPUSH;
1001 yylval.opval = PL_lex_op;
1011 * Create a new scope to save the lexing state. The scope will be
1012 * ended in S_sublex_done. Returns a '(', starting the function arguments
1013 * to the uc, lc, etc. found before.
1014 * Sets PL_lex_state to LEX_INTERPCONCAT.
1022 PL_lex_state = PL_sublex_info.super_state;
1023 SAVEI32(PL_lex_dojoin);
1024 SAVEI32(PL_lex_brackets);
1025 SAVEI32(PL_lex_casemods);
1026 SAVEI32(PL_lex_starts);
1027 SAVEI32(PL_lex_state);
1028 SAVEVPTR(PL_lex_inpat);
1029 SAVEI32(PL_lex_inwhat);
1030 SAVECOPLINE(PL_curcop);
1031 SAVEPPTR(PL_bufptr);
1032 SAVEPPTR(PL_oldbufptr);
1033 SAVEPPTR(PL_oldoldbufptr);
1034 SAVEPPTR(PL_linestart);
1035 SAVESPTR(PL_linestr);
1036 SAVEPPTR(PL_lex_brackstack);
1037 SAVEPPTR(PL_lex_casestack);
1039 PL_linestr = PL_lex_stuff;
1040 PL_lex_stuff = Nullsv;
1042 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1043 = SvPVX(PL_linestr);
1044 PL_bufend += SvCUR(PL_linestr);
1045 SAVEFREESV(PL_linestr);
1047 PL_lex_dojoin = FALSE;
1048 PL_lex_brackets = 0;
1049 New(899, PL_lex_brackstack, 120, char);
1050 New(899, PL_lex_casestack, 12, char);
1051 SAVEFREEPV(PL_lex_brackstack);
1052 SAVEFREEPV(PL_lex_casestack);
1053 PL_lex_casemods = 0;
1054 *PL_lex_casestack = '\0';
1056 PL_lex_state = LEX_INTERPCONCAT;
1057 CopLINE_set(PL_curcop, PL_multi_start);
1059 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1060 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1061 PL_lex_inpat = PL_sublex_info.sub_op;
1063 PL_lex_inpat = Nullop;
1070 * Restores lexer state after a S_sublex_push.
1076 if (!PL_lex_starts++) {
1077 SV *sv = newSVpvn("",0);
1078 if (SvUTF8(PL_linestr))
1080 PL_expect = XOPERATOR;
1081 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1085 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1086 PL_lex_state = LEX_INTERPCASEMOD;
1090 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1091 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1092 PL_linestr = PL_lex_repl;
1094 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1095 PL_bufend += SvCUR(PL_linestr);
1096 SAVEFREESV(PL_linestr);
1097 PL_lex_dojoin = FALSE;
1098 PL_lex_brackets = 0;
1099 PL_lex_casemods = 0;
1100 *PL_lex_casestack = '\0';
1102 if (SvEVALED(PL_lex_repl)) {
1103 PL_lex_state = LEX_INTERPNORMAL;
1105 /* we don't clear PL_lex_repl here, so that we can check later
1106 whether this is an evalled subst; that means we rely on the
1107 logic to ensure sublex_done() is called again only via the
1108 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1111 PL_lex_state = LEX_INTERPCONCAT;
1112 PL_lex_repl = Nullsv;
1118 PL_bufend = SvPVX(PL_linestr);
1119 PL_bufend += SvCUR(PL_linestr);
1120 PL_expect = XOPERATOR;
1121 PL_sublex_info.sub_inwhat = 0;
1129 Extracts a pattern, double-quoted string, or transliteration. This
1132 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1133 processing a pattern (PL_lex_inpat is true), a transliteration
1134 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1136 Returns a pointer to the character scanned up to. Iff this is
1137 advanced from the start pointer supplied (ie if anything was
1138 successfully parsed), will leave an OP for the substring scanned
1139 in yylval. Caller must intuit reason for not parsing further
1140 by looking at the next characters herself.
1144 double-quoted style: \r and \n
1145 regexp special ones: \D \s
1147 backrefs: \1 (deprecated in substitution replacements)
1148 case and quoting: \U \Q \E
1149 stops on @ and $, but not for $ as tail anchor
1151 In transliterations:
1152 characters are VERY literal, except for - not at the start or end
1153 of the string, which indicates a range. scan_const expands the
1154 range to the full set of intermediate characters.
1156 In double-quoted strings:
1158 double-quoted style: \r and \n
1160 backrefs: \1 (deprecated)
1161 case and quoting: \U \Q \E
1164 scan_const does *not* construct ops to handle interpolated strings.
1165 It stops processing as soon as it finds an embedded $ or @ variable
1166 and leaves it to the caller to work out what's going on.
1168 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1170 $ in pattern could be $foo or could be tail anchor. Assumption:
1171 it's a tail anchor if $ is the last thing in the string, or if it's
1172 followed by one of ")| \n\t"
1174 \1 (backreferences) are turned into $1
1176 The structure of the code is
1177 while (there's a character to process) {
1178 handle transliteration ranges
1179 skip regexp comments
1180 skip # initiated comments in //x patterns
1181 check for embedded @foo
1182 check for embedded scalars
1184 leave intact backslashes from leave (below)
1185 deprecate \1 in strings and sub replacements
1186 handle string-changing backslashes \l \U \Q \E, etc.
1187 switch (what was escaped) {
1188 handle - in a transliteration (becomes a literal -)
1189 handle \132 octal characters
1190 handle 0x15 hex characters
1191 handle \cV (control V)
1192 handle printf backslashes (\f, \r, \n, etc)
1194 } (end if backslash)
1195 } (end while character to read)
1200 S_scan_const(pTHX_ char *start)
1202 register char *send = PL_bufend; /* end of the constant */
1203 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1204 register char *s = start; /* start of the constant */
1205 register char *d = SvPVX(sv); /* destination for copies */
1206 bool dorange = FALSE; /* are we in a translit range? */
1207 bool didrange = FALSE; /* did we just finish a range? */
1208 bool has_utf8 = (PL_linestr && SvUTF8(PL_linestr));
1209 /* the constant is UTF8 */
1212 I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1213 ? (PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1215 I32 this_utf8 = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1216 ? (PL_sublex_info.sub_op->op_private & (PL_lex_repl ?
1217 OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF))
1219 const char *leaveit = /* set of acceptably-backslashed characters */
1221 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1224 while (s < send || dorange) {
1225 /* get transliterations out of the way (they're most literal) */
1226 if (PL_lex_inwhat == OP_TRANS) {
1227 /* expand a range A-Z to the full set of characters. AIE! */
1229 I32 i; /* current expanded character */
1230 I32 min; /* first character in range */
1231 I32 max; /* last character in range */
1233 i = d - SvPVX(sv); /* remember current offset */
1234 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1235 d = SvPVX(sv) + i; /* refresh d after realloc */
1236 d -= 2; /* eat the first char and the - */
1238 min = (U8)*d; /* first char in range */
1239 max = (U8)d[1]; /* last char in range */
1243 "Invalid [] range \"%c-%c\" in transliteration operator",
1244 (char)min, (char)max);
1248 if ((isLOWER(min) && isLOWER(max)) ||
1249 (isUPPER(min) && isUPPER(max))) {
1251 for (i = min; i <= max; i++)
1255 for (i = min; i <= max; i++)
1262 for (i = min; i <= max; i++)
1265 /* mark the range as done, and continue */
1271 /* range begins (ignore - as first or last char) */
1272 else if (*s == '-' && s+1 < send && s != start) {
1274 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
1277 *d++ = (char)0xff; /* use illegal utf8 byte--see pmtrans */
1289 /* if we get here, we're not doing a transliteration */
1291 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1292 except for the last char, which will be done separately. */
1293 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1295 while (s < send && *s != ')')
1298 else if (s[2] == '{' /* This should match regcomp.c */
1299 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1302 char *regparse = s + (s[2] == '{' ? 3 : 4);
1305 while (count && (c = *regparse)) {
1306 if (c == '\\' && regparse[1])
1314 if (*regparse != ')') {
1315 regparse--; /* Leave one char for continuation. */
1316 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1318 while (s < regparse)
1323 /* likewise skip #-initiated comments in //x patterns */
1324 else if (*s == '#' && PL_lex_inpat &&
1325 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1326 while (s+1 < send && *s != '\n')
1330 /* check for embedded arrays
1331 (@foo, @:foo, @'foo, @{foo}, @$foo, @+, @-)
1333 else if (*s == '@' && s[1]
1334 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1])))
1337 /* check for embedded scalars. only stop if we're sure it's a
1340 else if (*s == '$') {
1341 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1343 if (s + 1 < send && !strchr("()| \n\t", s[1]))
1344 break; /* in regexp, $ might be tail anchor */
1348 if (*s == '\\' && s+1 < send) {
1351 /* some backslashes we leave behind */
1352 if (*leaveit && *s && strchr(leaveit, *s)) {
1358 /* deprecate \1 in strings and substitution replacements */
1359 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1360 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1362 if (ckWARN(WARN_SYNTAX))
1363 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1368 /* string-change backslash escapes */
1369 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1374 /* if we get here, it's either a quoted -, or a digit */
1377 /* quoted - in transliterations */
1379 if (PL_lex_inwhat == OP_TRANS) {
1386 if (ckWARN(WARN_MISC) && isALNUM(*s))
1387 Perl_warner(aTHX_ WARN_MISC,
1388 "Unrecognized escape \\%c passed through",
1390 /* default action is to copy the quoted character */
1391 goto default_action;
1394 /* \132 indicates an octal constant */
1395 case '0': case '1': case '2': case '3':
1396 case '4': case '5': case '6': case '7':
1398 STRLEN len = 0; /* disallow underscores */
1399 uv = (UV)scan_oct(s, 3, &len);
1402 goto NUM_ESCAPE_INSERT;
1404 /* \x24 indicates a hex constant */
1408 char* e = strchr(s, '}');
1410 yyerror("Missing right brace on \\x{}");
1414 STRLEN len = 1; /* allow underscores */
1415 uv = (UV)scan_hex(s + 1, e - s - 1, &len);
1421 STRLEN len = 0; /* disallow underscores */
1422 uv = (UV)scan_hex(s, 2, &len);
1428 /* Insert oct or hex escaped character.
1429 * There will always enough room in sv since such
1430 * escapes will be longer than any UT-F8 sequence
1431 * they can end up as. */
1433 /* This spot is wrong for EBCDIC. Characters like
1434 * the lowercase letters and digits are >127 in EBCDIC,
1435 * so here they would need to be mapped to the Unicode
1436 * repertoire. --jhi */
1439 if (!has_utf8 && uv > 255) {
1440 /* Might need to recode whatever we have
1441 * accumulated so far if it contains any
1444 * (Can't we keep track of that and avoid
1445 * this rescan? --jhi)
1450 for (c = SvPVX(sv); c < d; c++) {
1451 if (UTF8_IS_CONTINUED(*c))
1455 char *old_pvx = SvPVX(sv);
1459 SvCUR(sv) + hicount + 1) +
1467 if (UTF8_IS_CONTINUED(*src)) {
1468 *dst-- = UTF8_EIGHT_BIT_LO(*src);
1469 *dst-- = UTF8_EIGHT_BIT_HI(*src--);
1478 if (has_utf8 || uv > 255) {
1479 d = (char*)uv_to_utf8((U8*)d, uv);
1481 if (PL_lex_inwhat == OP_TRANS &&
1482 PL_sublex_info.sub_op) {
1483 PL_sublex_info.sub_op->op_private |=
1484 (PL_lex_repl ? OPpTRANS_FROM_UTF
1498 /* \N{latin small letter a} is a named character */
1502 char* e = strchr(s, '}');
1508 yyerror("Missing right brace on \\N{}");
1512 res = newSVpvn(s + 1, e - s - 1);
1513 res = new_constant( Nullch, 0, "charnames",
1514 res, Nullsv, "\\N{...}" );
1516 sv_utf8_upgrade(res);
1517 str = SvPV(res,len);
1518 if (!has_utf8 && SvUTF8(res)) {
1519 char *ostart = SvPVX(sv);
1520 SvCUR_set(sv, d - ostart);
1523 sv_utf8_upgrade(sv);
1524 /* this just broke our allocation above... */
1525 SvGROW(sv, send - start);
1526 d = SvPVX(sv) + SvCUR(sv);
1529 if (len > e - s + 4) {
1530 char *odest = SvPVX(sv);
1532 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1533 d = SvPVX(sv) + (d - odest);
1535 Copy(str, d, len, char);
1542 yyerror("Missing braces on \\N{}");
1545 /* \c is a control character */
1562 /* printf-style backslashes, formfeeds, newlines, etc */
1580 *d++ = '\047'; /* CP 1047 */
1583 *d++ = '\057'; /* CP 1047 */
1597 } /* end if (backslash) */
1600 if (UTF8_IS_CONTINUED(*s) && (this_utf8 || has_utf8)) {
1601 STRLEN len = (STRLEN) -1;
1604 uv = utf8_to_uv((U8*)s, send - s, &len, 0);
1606 if (len == (STRLEN)-1) {
1607 /* Illegal UTF8 (a high-bit byte), make it valid. */
1608 char *old_pvx = SvPVX(sv);
1609 /* need space for one extra char (NOTE: SvCUR() not set here) */
1610 d = SvGROW(sv, SvLEN(sv) + 1) + (d - old_pvx);
1611 d = (char*)uv_to_utf8((U8*)d, (U8)*s++);
1618 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1619 PL_sublex_info.sub_op->op_private |=
1620 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1627 } /* while loop to process each character */
1629 /* terminate the string and set up the sv */
1631 SvCUR_set(sv, d - SvPVX(sv));
1636 /* shrink the sv if we allocated more than we used */
1637 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1638 SvLEN_set(sv, SvCUR(sv) + 1);
1639 Renew(SvPVX(sv), SvLEN(sv), char);
1642 /* return the substring (via yylval) only if we parsed anything */
1643 if (s > PL_bufptr) {
1644 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1645 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1647 ( PL_lex_inwhat == OP_TRANS
1649 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1652 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1659 * Returns TRUE if there's more to the expression (e.g., a subscript),
1662 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1664 * ->[ and ->{ return TRUE
1665 * { and [ outside a pattern are always subscripts, so return TRUE
1666 * if we're outside a pattern and it's not { or [, then return FALSE
1667 * if we're in a pattern and the first char is a {
1668 * {4,5} (any digits around the comma) returns FALSE
1669 * if we're in a pattern and the first char is a [
1671 * [SOMETHING] has a funky algorithm to decide whether it's a
1672 * character class or not. It has to deal with things like
1673 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1674 * anything else returns TRUE
1677 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1680 S_intuit_more(pTHX_ register char *s)
1682 if (PL_lex_brackets)
1684 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1686 if (*s != '{' && *s != '[')
1691 /* In a pattern, so maybe we have {n,m}. */
1708 /* On the other hand, maybe we have a character class */
1711 if (*s == ']' || *s == '^')
1714 /* this is terrifying, and it works */
1715 int weight = 2; /* let's weigh the evidence */
1717 unsigned char un_char = 255, last_un_char;
1718 char *send = strchr(s,']');
1719 char tmpbuf[sizeof PL_tokenbuf * 4];
1721 if (!send) /* has to be an expression */
1724 Zero(seen,256,char);
1727 else if (isDIGIT(*s)) {
1729 if (isDIGIT(s[1]) && s[2] == ']')
1735 for (; s < send; s++) {
1736 last_un_char = un_char;
1737 un_char = (unsigned char)*s;
1742 weight -= seen[un_char] * 10;
1743 if (isALNUM_lazy_if(s+1,UTF)) {
1744 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1745 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1750 else if (*s == '$' && s[1] &&
1751 strchr("[#!%*<>()-=",s[1])) {
1752 if (/*{*/ strchr("])} =",s[2]))
1761 if (strchr("wds]",s[1]))
1763 else if (seen['\''] || seen['"'])
1765 else if (strchr("rnftbxcav",s[1]))
1767 else if (isDIGIT(s[1])) {
1769 while (s[1] && isDIGIT(s[1]))
1779 if (strchr("aA01! ",last_un_char))
1781 if (strchr("zZ79~",s[1]))
1783 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1784 weight -= 5; /* cope with negative subscript */
1787 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1788 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1793 if (keyword(tmpbuf, d - tmpbuf))
1796 if (un_char == last_un_char + 1)
1798 weight -= seen[un_char];
1803 if (weight >= 0) /* probably a character class */
1813 * Does all the checking to disambiguate
1815 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1816 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1818 * First argument is the stuff after the first token, e.g. "bar".
1820 * Not a method if bar is a filehandle.
1821 * Not a method if foo is a subroutine prototyped to take a filehandle.
1822 * Not a method if it's really "Foo $bar"
1823 * Method if it's "foo $bar"
1824 * Not a method if it's really "print foo $bar"
1825 * Method if it's really "foo package::" (interpreted as package->foo)
1826 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1827 * Not a method if bar is a filehandle or package, but is quoted with
1832 S_intuit_method(pTHX_ char *start, GV *gv)
1834 char *s = start + (*start == '$');
1835 char tmpbuf[sizeof PL_tokenbuf];
1843 if ((cv = GvCVu(gv))) {
1844 char *proto = SvPVX(cv);
1854 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1855 /* start is the beginning of the possible filehandle/object,
1856 * and s is the end of it
1857 * tmpbuf is a copy of it
1860 if (*start == '$') {
1861 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1866 return *s == '(' ? FUNCMETH : METHOD;
1868 if (!keyword(tmpbuf, len)) {
1869 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1874 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1875 if (indirgv && GvCVu(indirgv))
1877 /* filehandle or package name makes it a method */
1878 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1880 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1881 return 0; /* no assumptions -- "=>" quotes bearword */
1883 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1884 newSVpvn(tmpbuf,len));
1885 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1889 return *s == '(' ? FUNCMETH : METHOD;
1897 * Return a string of Perl code to load the debugger. If PERL5DB
1898 * is set, it will return the contents of that, otherwise a
1899 * compile-time require of perl5db.pl.
1906 char *pdb = PerlEnv_getenv("PERL5DB");
1910 SETERRNO(0,SS$_NORMAL);
1911 return "BEGIN { require 'perl5db.pl' }";
1917 /* Encoded script support. filter_add() effectively inserts a
1918 * 'pre-processing' function into the current source input stream.
1919 * Note that the filter function only applies to the current source file
1920 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1922 * The datasv parameter (which may be NULL) can be used to pass
1923 * private data to this instance of the filter. The filter function
1924 * can recover the SV using the FILTER_DATA macro and use it to
1925 * store private buffers and state information.
1927 * The supplied datasv parameter is upgraded to a PVIO type
1928 * and the IoDIRP/IoANY field is used to store the function pointer,
1929 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1930 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1931 * private use must be set using malloc'd pointers.
1935 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1940 if (!PL_rsfp_filters)
1941 PL_rsfp_filters = newAV();
1943 datasv = NEWSV(255,0);
1944 if (!SvUPGRADE(datasv, SVt_PVIO))
1945 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1946 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
1947 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1948 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1949 funcp, SvPV_nolen(datasv)));
1950 av_unshift(PL_rsfp_filters, 1);
1951 av_store(PL_rsfp_filters, 0, datasv) ;
1956 /* Delete most recently added instance of this filter function. */
1958 Perl_filter_del(pTHX_ filter_t funcp)
1961 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1962 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1964 /* if filter is on top of stack (usual case) just pop it off */
1965 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1966 if (IoANY(datasv) == (void *)funcp) {
1967 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1968 IoANY(datasv) = (void *)NULL;
1969 sv_free(av_pop(PL_rsfp_filters));
1973 /* we need to search for the correct entry and clear it */
1974 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1978 /* Invoke the n'th filter function for the current rsfp. */
1980 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1983 /* 0 = read one text line */
1988 if (!PL_rsfp_filters)
1990 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1991 /* Provide a default input filter to make life easy. */
1992 /* Note that we append to the line. This is handy. */
1993 DEBUG_P(PerlIO_printf(Perl_debug_log,
1994 "filter_read %d: from rsfp\n", idx));
1998 int old_len = SvCUR(buf_sv) ;
2000 /* ensure buf_sv is large enough */
2001 SvGROW(buf_sv, old_len + maxlen) ;
2002 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
2003 if (PerlIO_error(PL_rsfp))
2004 return -1; /* error */
2006 return 0 ; /* end of file */
2008 SvCUR_set(buf_sv, old_len + len) ;
2011 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2012 if (PerlIO_error(PL_rsfp))
2013 return -1; /* error */
2015 return 0 ; /* end of file */
2018 return SvCUR(buf_sv);
2020 /* Skip this filter slot if filter has been deleted */
2021 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
2022 DEBUG_P(PerlIO_printf(Perl_debug_log,
2023 "filter_read %d: skipped (filter deleted)\n",
2025 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
2027 /* Get function pointer hidden within datasv */
2028 funcp = (filter_t)IoANY(datasv);
2029 DEBUG_P(PerlIO_printf(Perl_debug_log,
2030 "filter_read %d: via function %p (%s)\n",
2031 idx, funcp, SvPV_nolen(datasv)));
2032 /* Call function. The function is expected to */
2033 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2034 /* Return: <0:error, =0:eof, >0:not eof */
2035 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
2039 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2041 #ifdef PERL_CR_FILTER
2042 if (!PL_rsfp_filters) {
2043 filter_add(S_cr_textfilter,NULL);
2046 if (PL_rsfp_filters) {
2049 SvCUR_set(sv, 0); /* start with empty line */
2050 if (FILTER_READ(0, sv, 0) > 0)
2051 return ( SvPVX(sv) ) ;
2056 return (sv_gets(sv, fp, append));
2060 S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2064 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2068 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2069 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2071 return GvHV(gv); /* Foo:: */
2074 /* use constant CLASS => 'MyClass' */
2075 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2077 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2078 pkgname = SvPV_nolen(sv);
2082 return gv_stashpv(pkgname, FALSE);
2086 static char* exp_name[] =
2087 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2088 "ATTRTERM", "TERMBLOCK"
2095 Works out what to call the token just pulled out of the input
2096 stream. The yacc parser takes care of taking the ops we return and
2097 stitching them into a tree.
2103 if read an identifier
2104 if we're in a my declaration
2105 croak if they tried to say my($foo::bar)
2106 build the ops for a my() declaration
2107 if it's an access to a my() variable
2108 are we in a sort block?
2109 croak if my($a); $a <=> $b
2110 build ops for access to a my() variable
2111 if in a dq string, and they've said @foo and we can't find @foo
2113 build ops for a bareword
2114 if we already built the token before, use it.
2117 #ifdef USE_PURE_BISON
2119 Perl_yylex_r(pTHX_ YYSTYPE *lvalp, int *lcharp)
2124 yylval_pointer[yyactlevel] = lvalp;
2125 yychar_pointer[yyactlevel] = lcharp;
2126 if (yyactlevel >= YYMAXLEVEL)
2127 Perl_croak(aTHX_ "panic: YYMAXLEVEL");
2129 r = Perl_yylex(aTHX);
2139 #pragma segment Perl_yylex
2152 /* check if there's an identifier for us to look at */
2153 if (PL_pending_ident) {
2154 /* pit holds the identifier we read and pending_ident is reset */
2155 char pit = PL_pending_ident;
2156 PL_pending_ident = 0;
2158 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2159 "### Tokener saw identifier '%s'\n", PL_tokenbuf); })
2161 /* if we're in a my(), we can't allow dynamics here.
2162 $foo'bar has already been turned into $foo::bar, so
2163 just check for colons.
2165 if it's a legal name, the OP is a PADANY.
2168 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
2169 if (strchr(PL_tokenbuf,':'))
2170 yyerror(Perl_form(aTHX_ "No package name allowed for "
2171 "variable %s in \"our\"",
2173 tmp = pad_allocmy(PL_tokenbuf);
2176 if (strchr(PL_tokenbuf,':'))
2177 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
2179 yylval.opval = newOP(OP_PADANY, 0);
2180 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
2186 build the ops for accesses to a my() variable.
2188 Deny my($a) or my($b) in a sort block, *if* $a or $b is
2189 then used in a comparison. This catches most, but not
2190 all cases. For instance, it catches
2191 sort { my($a); $a <=> $b }
2193 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2194 (although why you'd do that is anyone's guess).
2197 if (!strchr(PL_tokenbuf,':')) {
2199 /* Check for single character per-thread SVs */
2200 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2201 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2202 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2204 yylval.opval = newOP(OP_THREADSV, 0);
2205 yylval.opval->op_targ = tmp;
2208 #endif /* USE_THREADS */
2209 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2210 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2211 /* might be an "our" variable" */
2212 if (SvFLAGS(namesv) & SVpad_OUR) {
2213 /* build ops for a bareword */
2214 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2215 sv_catpvn(sym, "::", 2);
2216 sv_catpv(sym, PL_tokenbuf+1);
2217 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2218 yylval.opval->op_private = OPpCONST_ENTERED;
2219 gv_fetchpv(SvPVX(sym),
2221 ? (GV_ADDMULTI | GV_ADDINEVAL)
2224 ((PL_tokenbuf[0] == '$') ? SVt_PV
2225 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2230 /* if it's a sort block and they're naming $a or $b */
2231 if (PL_last_lop_op == OP_SORT &&
2232 PL_tokenbuf[0] == '$' &&
2233 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2236 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2237 d < PL_bufend && *d != '\n';
2240 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2241 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2247 yylval.opval = newOP(OP_PADANY, 0);
2248 yylval.opval->op_targ = tmp;
2254 Whine if they've said @foo in a doublequoted string,
2255 and @foo isn't a variable we can find in the symbol
2258 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2259 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2260 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2261 && ckWARN(WARN_AMBIGUOUS))
2263 /* Downgraded from fatal to warning 20000522 mjd */
2264 Perl_warner(aTHX_ WARN_AMBIGUOUS,
2265 "Possible unintended interpolation of %s in string",
2270 /* build ops for a bareword */
2271 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2272 yylval.opval->op_private = OPpCONST_ENTERED;
2273 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2274 ((PL_tokenbuf[0] == '$') ? SVt_PV
2275 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2280 /* no identifier pending identification */
2282 switch (PL_lex_state) {
2284 case LEX_NORMAL: /* Some compilers will produce faster */
2285 case LEX_INTERPNORMAL: /* code if we comment these out. */
2289 /* when we've already built the next token, just pull it out of the queue */
2292 yylval = PL_nextval[PL_nexttoke];
2294 PL_lex_state = PL_lex_defer;
2295 PL_expect = PL_lex_expect;
2296 PL_lex_defer = LEX_NORMAL;
2298 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2299 "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr,
2300 (IV)PL_nexttype[PL_nexttoke]); })
2302 return(PL_nexttype[PL_nexttoke]);
2304 /* interpolated case modifiers like \L \U, including \Q and \E.
2305 when we get here, PL_bufptr is at the \
2307 case LEX_INTERPCASEMOD:
2309 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2310 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2312 /* handle \E or end of string */
2313 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2317 if (PL_lex_casemods) {
2318 oldmod = PL_lex_casestack[--PL_lex_casemods];
2319 PL_lex_casestack[PL_lex_casemods] = '\0';
2321 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2323 PL_lex_state = LEX_INTERPCONCAT;
2327 if (PL_bufptr != PL_bufend)
2329 PL_lex_state = LEX_INTERPCONCAT;
2333 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2334 "### Saw case modifier at '%s'\n", PL_bufptr); })
2336 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2337 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2338 if (strchr("LU", *s) &&
2339 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2341 PL_lex_casestack[--PL_lex_casemods] = '\0';
2344 if (PL_lex_casemods > 10) {
2345 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2346 if (newlb != PL_lex_casestack) {
2348 PL_lex_casestack = newlb;
2351 PL_lex_casestack[PL_lex_casemods++] = *s;
2352 PL_lex_casestack[PL_lex_casemods] = '\0';
2353 PL_lex_state = LEX_INTERPCONCAT;
2354 PL_nextval[PL_nexttoke].ival = 0;
2357 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2359 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2361 PL_nextval[PL_nexttoke].ival = OP_LC;
2363 PL_nextval[PL_nexttoke].ival = OP_UC;
2365 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2367 Perl_croak(aTHX_ "panic: yylex");
2370 if (PL_lex_starts) {
2379 case LEX_INTERPPUSH:
2380 return sublex_push();
2382 case LEX_INTERPSTART:
2383 if (PL_bufptr == PL_bufend)
2384 return sublex_done();
2385 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2386 "### Interpolated variable at '%s'\n", PL_bufptr); })
2388 PL_lex_dojoin = (*PL_bufptr == '@');
2389 PL_lex_state = LEX_INTERPNORMAL;
2390 if (PL_lex_dojoin) {
2391 PL_nextval[PL_nexttoke].ival = 0;
2394 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2395 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2396 force_next(PRIVATEREF);
2398 force_ident("\"", '$');
2399 #endif /* USE_THREADS */
2400 PL_nextval[PL_nexttoke].ival = 0;
2402 PL_nextval[PL_nexttoke].ival = 0;
2404 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2407 if (PL_lex_starts++) {
2413 case LEX_INTERPENDMAYBE:
2414 if (intuit_more(PL_bufptr)) {
2415 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2421 if (PL_lex_dojoin) {
2422 PL_lex_dojoin = FALSE;
2423 PL_lex_state = LEX_INTERPCONCAT;
2426 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2427 && SvEVALED(PL_lex_repl))
2429 if (PL_bufptr != PL_bufend)
2430 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2431 PL_lex_repl = Nullsv;
2434 case LEX_INTERPCONCAT:
2436 if (PL_lex_brackets)
2437 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2439 if (PL_bufptr == PL_bufend)
2440 return sublex_done();
2442 if (SvIVX(PL_linestr) == '\'') {
2443 SV *sv = newSVsv(PL_linestr);
2446 else if ( PL_hints & HINT_NEW_RE )
2447 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2448 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2452 s = scan_const(PL_bufptr);
2454 PL_lex_state = LEX_INTERPCASEMOD;
2456 PL_lex_state = LEX_INTERPSTART;
2459 if (s != PL_bufptr) {
2460 PL_nextval[PL_nexttoke] = yylval;
2463 if (PL_lex_starts++)
2473 PL_lex_state = LEX_NORMAL;
2474 s = scan_formline(PL_bufptr);
2475 if (!PL_lex_formbrack)
2481 PL_oldoldbufptr = PL_oldbufptr;
2484 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2485 exp_name[PL_expect], s);
2491 if (isIDFIRST_lazy_if(s,UTF))
2493 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2496 goto fake_eof; /* emulate EOF on ^D or ^Z */
2501 if (PL_lex_brackets)
2502 yyerror("Missing right curly or square bracket");
2503 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2504 "### Tokener got EOF\n");
2508 if (s++ < PL_bufend)
2509 goto retry; /* ignore stray nulls */
2512 if (!PL_in_eval && !PL_preambled) {
2513 PL_preambled = TRUE;
2514 sv_setpv(PL_linestr,incl_perldb());
2515 if (SvCUR(PL_linestr))
2516 sv_catpv(PL_linestr,";");
2518 while(AvFILLp(PL_preambleav) >= 0) {
2519 SV *tmpsv = av_shift(PL_preambleav);
2520 sv_catsv(PL_linestr, tmpsv);
2521 sv_catpv(PL_linestr, ";");
2524 sv_free((SV*)PL_preambleav);
2525 PL_preambleav = NULL;
2527 if (PL_minus_n || PL_minus_p) {
2528 sv_catpv(PL_linestr, "LINE: while (<>) {");
2530 sv_catpv(PL_linestr,"chomp;");
2532 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2534 GvIMPORTED_AV_on(gv);
2536 if (strchr("/'\"", *PL_splitstr)
2537 && strchr(PL_splitstr + 1, *PL_splitstr))
2538 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2541 s = "'~#\200\1'"; /* surely one char is unused...*/
2542 while (s[1] && strchr(PL_splitstr, *s)) s++;
2544 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2545 "q" + (delim == '\''), delim);
2546 for (s = PL_splitstr; *s; s++) {
2548 sv_catpvn(PL_linestr, "\\", 1);
2549 sv_catpvn(PL_linestr, s, 1);
2551 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2555 sv_catpv(PL_linestr,"@F=split(' ');");
2558 sv_catpv(PL_linestr, "\n");
2559 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2560 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2561 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2562 SV *sv = NEWSV(85,0);
2564 sv_upgrade(sv, SVt_PVMG);
2565 sv_setsv(sv,PL_linestr);
2566 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2571 bof = PL_rsfp ? TRUE : FALSE;
2572 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2575 if (PL_preprocess && !PL_in_eval)
2576 (void)PerlProc_pclose(PL_rsfp);
2577 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2578 PerlIO_clearerr(PL_rsfp);
2580 (void)PerlIO_close(PL_rsfp);
2582 PL_doextract = FALSE;
2584 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2585 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2586 sv_catpv(PL_linestr,";}");
2587 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2588 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2589 PL_minus_n = PL_minus_p = 0;
2592 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2593 sv_setpv(PL_linestr,"");
2594 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2596 /* if it looks like the start of a BOM, check if it in fact is */
2597 else if (bof && (!*s || *(U8*)s == 0xEF || *(U8*)s >= 0xFE)) {
2598 #ifdef PERLIO_IS_STDIO
2599 # ifdef __GNU_LIBRARY__
2600 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
2601 # define FTELL_FOR_PIPE_IS_BROKEN
2605 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2606 # define FTELL_FOR_PIPE_IS_BROKEN
2611 #ifdef FTELL_FOR_PIPE_IS_BROKEN
2612 /* This loses the possibility to detect the bof
2613 * situation on perl -P when the libc5 is being used.
2614 * Workaround? Maybe attach some extra state to PL_rsfp?
2617 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2619 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2622 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2623 s = swallow_bom((U8*)s);
2627 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2628 PL_doextract = FALSE;
2630 /* Incest with pod. */
2631 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2632 sv_setpv(PL_linestr, "");
2633 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2634 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2635 PL_doextract = FALSE;
2639 } while (PL_doextract);
2640 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2641 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2642 SV *sv = NEWSV(85,0);
2644 sv_upgrade(sv, SVt_PVMG);
2645 sv_setsv(sv,PL_linestr);
2646 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2648 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2649 if (CopLINE(PL_curcop) == 1) {
2650 while (s < PL_bufend && isSPACE(*s))
2652 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2656 if (*s == '#' && *(s+1) == '!')
2658 #ifdef ALTERNATE_SHEBANG
2660 static char as[] = ALTERNATE_SHEBANG;
2661 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2662 d = s + (sizeof(as) - 1);
2664 #endif /* ALTERNATE_SHEBANG */
2673 while (*d && !isSPACE(*d))
2677 #ifdef ARG_ZERO_IS_SCRIPT
2678 if (ipathend > ipath) {
2680 * HP-UX (at least) sets argv[0] to the script name,
2681 * which makes $^X incorrect. And Digital UNIX and Linux,
2682 * at least, set argv[0] to the basename of the Perl
2683 * interpreter. So, having found "#!", we'll set it right.
2685 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2686 assert(SvPOK(x) || SvGMAGICAL(x));
2687 if (sv_eq(x, CopFILESV(PL_curcop))) {
2688 sv_setpvn(x, ipath, ipathend - ipath);
2691 TAINT_NOT; /* $^X is always tainted, but that's OK */
2693 #endif /* ARG_ZERO_IS_SCRIPT */
2698 d = instr(s,"perl -");
2700 d = instr(s,"perl");
2702 /* avoid getting into infinite loops when shebang
2703 * line contains "Perl" rather than "perl" */
2705 for (d = ipathend-4; d >= ipath; --d) {
2706 if ((*d == 'p' || *d == 'P')
2707 && !ibcmp(d, "perl", 4))
2717 #ifdef ALTERNATE_SHEBANG
2719 * If the ALTERNATE_SHEBANG on this system starts with a
2720 * character that can be part of a Perl expression, then if
2721 * we see it but not "perl", we're probably looking at the
2722 * start of Perl code, not a request to hand off to some
2723 * other interpreter. Similarly, if "perl" is there, but
2724 * not in the first 'word' of the line, we assume the line
2725 * contains the start of the Perl program.
2727 if (d && *s != '#') {
2729 while (*c && !strchr("; \t\r\n\f\v#", *c))
2732 d = Nullch; /* "perl" not in first word; ignore */
2734 *s = '#'; /* Don't try to parse shebang line */
2736 #endif /* ALTERNATE_SHEBANG */
2737 #ifndef MACOS_TRADITIONAL
2742 !instr(s,"indir") &&
2743 instr(PL_origargv[0],"perl"))
2749 while (s < PL_bufend && isSPACE(*s))
2751 if (s < PL_bufend) {
2752 Newz(899,newargv,PL_origargc+3,char*);
2754 while (s < PL_bufend && !isSPACE(*s))
2757 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2760 newargv = PL_origargv;
2762 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
2763 Perl_croak(aTHX_ "Can't exec %s", ipath);
2767 U32 oldpdb = PL_perldb;
2768 bool oldn = PL_minus_n;
2769 bool oldp = PL_minus_p;
2771 while (*d && !isSPACE(*d)) d++;
2772 while (SPACE_OR_TAB(*d)) d++;
2776 if (*d == 'M' || *d == 'm') {
2778 while (*d && !isSPACE(*d)) d++;
2779 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2782 d = moreswitches(d);
2784 if ((PERLDB_LINE && !oldpdb) ||
2785 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2786 /* if we have already added "LINE: while (<>) {",
2787 we must not do it again */
2789 sv_setpv(PL_linestr, "");
2790 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2791 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2792 PL_preambled = FALSE;
2794 (void)gv_fetchfile(PL_origfilename);
2801 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2803 PL_lex_state = LEX_FORMLINE;
2808 #ifdef PERL_STRICT_CR
2809 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2811 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2813 case ' ': case '\t': case '\f': case 013:
2814 #ifdef MACOS_TRADITIONAL
2821 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2822 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2823 /* handle eval qq[#line 1 "foo"\n ...] */
2824 CopLINE_dec(PL_curcop);
2828 while (s < d && *s != '\n')
2833 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2835 PL_lex_state = LEX_FORMLINE;
2845 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2852 while (s < PL_bufend && SPACE_OR_TAB(*s))
2855 if (strnEQ(s,"=>",2)) {
2856 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2857 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2858 "### Saw unary minus before =>, forcing word '%s'\n", s);
2860 OPERATOR('-'); /* unary minus */
2862 PL_last_uni = PL_oldbufptr;
2864 case 'r': ftst = OP_FTEREAD; break;
2865 case 'w': ftst = OP_FTEWRITE; break;
2866 case 'x': ftst = OP_FTEEXEC; break;
2867 case 'o': ftst = OP_FTEOWNED; break;
2868 case 'R': ftst = OP_FTRREAD; break;
2869 case 'W': ftst = OP_FTRWRITE; break;
2870 case 'X': ftst = OP_FTREXEC; break;
2871 case 'O': ftst = OP_FTROWNED; break;
2872 case 'e': ftst = OP_FTIS; break;
2873 case 'z': ftst = OP_FTZERO; break;
2874 case 's': ftst = OP_FTSIZE; break;
2875 case 'f': ftst = OP_FTFILE; break;
2876 case 'd': ftst = OP_FTDIR; break;
2877 case 'l': ftst = OP_FTLINK; break;
2878 case 'p': ftst = OP_FTPIPE; break;
2879 case 'S': ftst = OP_FTSOCK; break;
2880 case 'u': ftst = OP_FTSUID; break;
2881 case 'g': ftst = OP_FTSGID; break;
2882 case 'k': ftst = OP_FTSVTX; break;
2883 case 'b': ftst = OP_FTBLK; break;
2884 case 'c': ftst = OP_FTCHR; break;
2885 case 't': ftst = OP_FTTTY; break;
2886 case 'T': ftst = OP_FTTEXT; break;
2887 case 'B': ftst = OP_FTBINARY; break;
2888 case 'M': case 'A': case 'C':
2889 gv_fetchpv("\024",TRUE, SVt_PV);
2891 case 'M': ftst = OP_FTMTIME; break;
2892 case 'A': ftst = OP_FTATIME; break;
2893 case 'C': ftst = OP_FTCTIME; break;
2901 PL_last_lop_op = ftst;
2902 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2903 "### Saw file test %c\n", (int)ftst);
2908 /* Assume it was a minus followed by a one-letter named
2909 * subroutine call (or a -bareword), then. */
2910 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2911 "### %c looked like a file test but was not\n",
2920 if (PL_expect == XOPERATOR)
2925 else if (*s == '>') {
2928 if (isIDFIRST_lazy_if(s,UTF)) {
2929 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2937 if (PL_expect == XOPERATOR)
2940 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2942 OPERATOR('-'); /* unary minus */
2949 if (PL_expect == XOPERATOR)
2954 if (PL_expect == XOPERATOR)
2957 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2963 if (PL_expect != XOPERATOR) {
2964 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2965 PL_expect = XOPERATOR;
2966 force_ident(PL_tokenbuf, '*');
2979 if (PL_expect == XOPERATOR) {
2983 PL_tokenbuf[0] = '%';
2984 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2985 if (!PL_tokenbuf[1]) {
2987 yyerror("Final % should be \\% or %name");
2990 PL_pending_ident = '%';
3009 switch (PL_expect) {
3012 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
3014 PL_bufptr = s; /* update in case we back off */
3020 PL_expect = XTERMBLOCK;
3024 while (isIDFIRST_lazy_if(s,UTF)) {
3025 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3026 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
3027 if (tmp < 0) tmp = -tmp;
3042 d = scan_str(d,TRUE,TRUE);
3045 SvREFCNT_dec(PL_lex_stuff);
3046 PL_lex_stuff = Nullsv;
3048 /* MUST advance bufptr here to avoid bogus
3049 "at end of line" context messages from yyerror().
3051 PL_bufptr = s + len;
3052 yyerror("Unterminated attribute parameter in attribute list");
3055 return 0; /* EOF indicator */
3059 SV *sv = newSVpvn(s, len);
3060 sv_catsv(sv, PL_lex_stuff);
3061 attrs = append_elem(OP_LIST, attrs,
3062 newSVOP(OP_CONST, 0, sv));
3063 SvREFCNT_dec(PL_lex_stuff);
3064 PL_lex_stuff = Nullsv;
3067 if (!PL_in_my && len == 6 && strnEQ(s, "lvalue", len))
3068 CvLVALUE_on(PL_compcv);
3069 else if (!PL_in_my && len == 6 && strnEQ(s, "locked", len))
3070 CvLOCKED_on(PL_compcv);
3071 else if (!PL_in_my && len == 6 && strnEQ(s, "method", len))
3072 CvMETHOD_on(PL_compcv);
3074 else if (PL_in_my == KEY_our && len == 6 && strnEQ(s, "shared", len))
3075 GvSHARED_on(cGVOPx_gv(yylval.opval));
3077 /* After we've set the flags, it could be argued that
3078 we don't need to do the attributes.pm-based setting
3079 process, and shouldn't bother appending recognized
3080 flags. To experiment with that, uncomment the
3081 following "else": */
3083 attrs = append_elem(OP_LIST, attrs,
3084 newSVOP(OP_CONST, 0,
3088 if (*s == ':' && s[1] != ':')
3091 break; /* require real whitespace or :'s */
3093 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3094 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
3095 char q = ((*s == '\'') ? '"' : '\'');
3096 /* If here for an expression, and parsed no attrs, back off. */
3097 if (tmp == '=' && !attrs) {
3101 /* MUST advance bufptr here to avoid bogus "at end of line"
3102 context messages from yyerror().
3106 yyerror("Unterminated attribute list");
3108 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
3116 PL_nextval[PL_nexttoke].opval = attrs;
3124 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3125 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3141 if (PL_lex_brackets <= 0)
3142 yyerror("Unmatched right square bracket");
3145 if (PL_lex_state == LEX_INTERPNORMAL) {
3146 if (PL_lex_brackets == 0) {
3147 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3148 PL_lex_state = LEX_INTERPEND;
3155 if (PL_lex_brackets > 100) {
3156 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
3157 if (newlb != PL_lex_brackstack) {
3159 PL_lex_brackstack = newlb;
3162 switch (PL_expect) {
3164 if (PL_lex_formbrack) {
3168 if (PL_oldoldbufptr == PL_last_lop)
3169 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3171 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3172 OPERATOR(HASHBRACK);
3174 while (s < PL_bufend && SPACE_OR_TAB(*s))
3177 PL_tokenbuf[0] = '\0';
3178 if (d < PL_bufend && *d == '-') {
3179 PL_tokenbuf[0] = '-';
3181 while (d < PL_bufend && SPACE_OR_TAB(*d))
3184 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3185 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3187 while (d < PL_bufend && SPACE_OR_TAB(*d))
3190 char minus = (PL_tokenbuf[0] == '-');
3191 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3192 if (UTF && !IN_BYTE && is_utf8_string((U8*)PL_tokenbuf, 0) &&
3193 PL_nextval[PL_nexttoke-1].opval)
3194 SvUTF8_on(((SVOP*)PL_nextval[PL_nexttoke-1].opval)->op_sv);
3202 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3207 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3212 if (PL_oldoldbufptr == PL_last_lop)
3213 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3215 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3218 OPERATOR(HASHBRACK);
3219 /* This hack serves to disambiguate a pair of curlies
3220 * as being a block or an anon hash. Normally, expectation
3221 * determines that, but in cases where we're not in a
3222 * position to expect anything in particular (like inside
3223 * eval"") we have to resolve the ambiguity. This code
3224 * covers the case where the first term in the curlies is a
3225 * quoted string. Most other cases need to be explicitly
3226 * disambiguated by prepending a `+' before the opening
3227 * curly in order to force resolution as an anon hash.
3229 * XXX should probably propagate the outer expectation
3230 * into eval"" to rely less on this hack, but that could
3231 * potentially break current behavior of eval"".
3235 if (*s == '\'' || *s == '"' || *s == '`') {
3236 /* common case: get past first string, handling escapes */
3237 for (t++; t < PL_bufend && *t != *s;)
3238 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3242 else if (*s == 'q') {
3245 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3249 char open, close, term;
3252 while (t < PL_bufend && isSPACE(*t))
3256 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3260 for (t++; t < PL_bufend; t++) {
3261 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3263 else if (*t == open)
3267 for (t++; t < PL_bufend; t++) {
3268 if (*t == '\\' && t+1 < PL_bufend)
3270 else if (*t == close && --brackets <= 0)
3272 else if (*t == open)
3278 else if (isALNUM_lazy_if(t,UTF)) {
3280 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3283 while (t < PL_bufend && isSPACE(*t))
3285 /* if comma follows first term, call it an anon hash */
3286 /* XXX it could be a comma expression with loop modifiers */
3287 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3288 || (*t == '=' && t[1] == '>')))
3289 OPERATOR(HASHBRACK);
3290 if (PL_expect == XREF)
3293 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3299 yylval.ival = CopLINE(PL_curcop);
3300 if (isSPACE(*s) || *s == '#')
3301 PL_copline = NOLINE; /* invalidate current command line number */
3306 if (PL_lex_brackets <= 0)
3307 yyerror("Unmatched right curly bracket");
3309 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3310 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3311 PL_lex_formbrack = 0;
3312 if (PL_lex_state == LEX_INTERPNORMAL) {
3313 if (PL_lex_brackets == 0) {
3314 if (PL_expect & XFAKEBRACK) {
3315 PL_expect &= XENUMMASK;
3316 PL_lex_state = LEX_INTERPEND;
3318 return yylex(); /* ignore fake brackets */
3320 if (*s == '-' && s[1] == '>')
3321 PL_lex_state = LEX_INTERPENDMAYBE;
3322 else if (*s != '[' && *s != '{')
3323 PL_lex_state = LEX_INTERPEND;
3326 if (PL_expect & XFAKEBRACK) {
3327 PL_expect &= XENUMMASK;
3329 return yylex(); /* ignore fake brackets */
3339 if (PL_expect == XOPERATOR) {
3340 if (ckWARN(WARN_SEMICOLON)
3341 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3343 CopLINE_dec(PL_curcop);
3344 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3345 CopLINE_inc(PL_curcop);
3350 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3352 PL_expect = XOPERATOR;
3353 force_ident(PL_tokenbuf, '&');
3357 yylval.ival = (OPpENTERSUB_AMPER<<8);
3376 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3377 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3379 if (PL_expect == XSTATE && isALPHA(tmp) &&
3380 (s == PL_linestart+1 || s[-2] == '\n') )
3382 if (PL_in_eval && !PL_rsfp) {
3387 if (strnEQ(s,"=cut",4)) {
3401 PL_doextract = TRUE;
3404 if (PL_lex_brackets < PL_lex_formbrack) {
3406 #ifdef PERL_STRICT_CR
3407 for (t = s; SPACE_OR_TAB(*t); t++) ;
3409 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3411 if (*t == '\n' || *t == '#') {
3429 if (PL_expect != XOPERATOR) {
3430 if (s[1] != '<' && !strchr(s,'>'))
3433 s = scan_heredoc(s);
3435 s = scan_inputsymbol(s);
3436 TERM(sublex_start());
3441 SHop(OP_LEFT_SHIFT);
3455 SHop(OP_RIGHT_SHIFT);
3464 if (PL_expect == XOPERATOR) {
3465 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3468 return ','; /* grandfather non-comma-format format */
3472 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3473 PL_tokenbuf[0] = '@';
3474 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3475 sizeof PL_tokenbuf - 1, FALSE);
3476 if (PL_expect == XOPERATOR)
3477 no_op("Array length", s);
3478 if (!PL_tokenbuf[1])
3480 PL_expect = XOPERATOR;
3481 PL_pending_ident = '#';
3485 PL_tokenbuf[0] = '$';
3486 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3487 sizeof PL_tokenbuf - 1, FALSE);
3488 if (PL_expect == XOPERATOR)
3490 if (!PL_tokenbuf[1]) {
3492 yyerror("Final $ should be \\$ or $name");
3496 /* This kludge not intended to be bulletproof. */
3497 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3498 yylval.opval = newSVOP(OP_CONST, 0,
3499 newSViv(PL_compiling.cop_arybase));
3500 yylval.opval->op_private = OPpCONST_ARYBASE;
3506 if (PL_lex_state == LEX_NORMAL)
3509 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3512 PL_tokenbuf[0] = '@';
3513 if (ckWARN(WARN_SYNTAX)) {
3515 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3518 PL_bufptr = skipspace(PL_bufptr);
3519 while (t < PL_bufend && *t != ']')
3521 Perl_warner(aTHX_ WARN_SYNTAX,
3522 "Multidimensional syntax %.*s not supported",
3523 (t - PL_bufptr) + 1, PL_bufptr);
3527 else if (*s == '{') {
3528 PL_tokenbuf[0] = '%';
3529 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3530 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3532 char tmpbuf[sizeof PL_tokenbuf];
3534 for (t++; isSPACE(*t); t++) ;
3535 if (isIDFIRST_lazy_if(t,UTF)) {
3536 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3537 for (; isSPACE(*t); t++) ;
3538 if (*t == ';' && get_cv(tmpbuf, FALSE))
3539 Perl_warner(aTHX_ WARN_SYNTAX,
3540 "You need to quote \"%s\"", tmpbuf);
3546 PL_expect = XOPERATOR;
3547 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3548 bool islop = (PL_last_lop == PL_oldoldbufptr);
3549 if (!islop || PL_last_lop_op == OP_GREPSTART)
3550 PL_expect = XOPERATOR;
3551 else if (strchr("$@\"'`q", *s))
3552 PL_expect = XTERM; /* e.g. print $fh "foo" */
3553 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3554 PL_expect = XTERM; /* e.g. print $fh &sub */
3555 else if (isIDFIRST_lazy_if(s,UTF)) {
3556 char tmpbuf[sizeof PL_tokenbuf];
3557 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3558 if ((tmp = keyword(tmpbuf, len))) {
3559 /* binary operators exclude handle interpretations */
3571 PL_expect = XTERM; /* e.g. print $fh length() */
3576 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3577 if (gv && GvCVu(gv))
3578 PL_expect = XTERM; /* e.g. print $fh subr() */
3581 else if (isDIGIT(*s))
3582 PL_expect = XTERM; /* e.g. print $fh 3 */
3583 else if (*s == '.' && isDIGIT(s[1]))
3584 PL_expect = XTERM; /* e.g. print $fh .3 */
3585 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3586 PL_expect = XTERM; /* e.g. print $fh -1 */
3587 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3588 PL_expect = XTERM; /* print $fh <<"EOF" */
3590 PL_pending_ident = '$';
3594 if (PL_expect == XOPERATOR)
3596 PL_tokenbuf[0] = '@';
3597 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3598 if (!PL_tokenbuf[1]) {
3600 yyerror("Final @ should be \\@ or @name");
3603 if (PL_lex_state == LEX_NORMAL)
3605 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3607 PL_tokenbuf[0] = '%';
3609 /* Warn about @ where they meant $. */
3610 if (ckWARN(WARN_SYNTAX)) {
3611 if (*s == '[' || *s == '{') {
3613 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3615 if (*t == '}' || *t == ']') {
3617 PL_bufptr = skipspace(PL_bufptr);
3618 Perl_warner(aTHX_ WARN_SYNTAX,
3619 "Scalar value %.*s better written as $%.*s",
3620 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3625 PL_pending_ident = '@';
3628 case '/': /* may either be division or pattern */
3629 case '?': /* may either be conditional or pattern */
3630 if (PL_expect != XOPERATOR) {
3631 /* Disable warning on "study /blah/" */
3632 if (PL_oldoldbufptr == PL_last_uni
3633 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3634 || memNE(PL_last_uni, "study", 5)
3635 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3637 s = scan_pat(s,OP_MATCH);
3638 TERM(sublex_start());
3646 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3647 #ifdef PERL_STRICT_CR
3650 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3652 && (s == PL_linestart || s[-1] == '\n') )
3654 PL_lex_formbrack = 0;
3658 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3664 yylval.ival = OPf_SPECIAL;
3670 if (PL_expect != XOPERATOR)
3675 case '0': case '1': case '2': case '3': case '4':
3676 case '5': case '6': case '7': case '8': case '9':
3677 s = scan_num(s, &yylval);
3678 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3679 "### Saw number in '%s'\n", s);
3681 if (PL_expect == XOPERATOR)
3686 s = scan_str(s,FALSE,FALSE);
3687 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3688 "### Saw string in '%s'\n", s);
3690 if (PL_expect == XOPERATOR) {
3691 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3694 return ','; /* grandfather non-comma-format format */
3700 missingterm((char*)0);
3701 yylval.ival = OP_CONST;
3702 TERM(sublex_start());
3705 s = scan_str(s,FALSE,FALSE);
3706 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3707 "### Saw string in '%s'\n", s);
3709 if (PL_expect == XOPERATOR) {
3710 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3713 return ','; /* grandfather non-comma-format format */
3719 missingterm((char*)0);
3720 yylval.ival = OP_CONST;
3721 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3722 if (*d == '$' || *d == '@' || *d == '\\' || UTF8_IS_CONTINUED(*d)) {
3723 yylval.ival = OP_STRINGIFY;
3727 TERM(sublex_start());
3730 s = scan_str(s,FALSE,FALSE);
3731 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3732 "### Saw backtick string in '%s'\n", s);
3734 if (PL_expect == XOPERATOR)
3735 no_op("Backticks",s);
3737 missingterm((char*)0);
3738 yylval.ival = OP_BACKTICK;
3740 TERM(sublex_start());
3744 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3745 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3747 if (PL_expect == XOPERATOR)
3748 no_op("Backslash",s);
3752 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3756 while (isDIGIT(*start) || *start == '_')
3758 if (*start == '.' && isDIGIT(start[1])) {
3759 s = scan_num(s, &yylval);
3762 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3763 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF)) {
3767 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3770 s = scan_num(s, &yylval);
3777 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3816 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3818 /* Some keywords can be followed by any delimiter, including ':' */
3819 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3820 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3821 (PL_tokenbuf[0] == 'q' &&
3822 strchr("qwxr", PL_tokenbuf[1])))));
3824 /* x::* is just a word, unless x is "CORE" */
3825 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3829 while (d < PL_bufend && isSPACE(*d))
3830 d++; /* no comments skipped here, or s### is misparsed */
3832 /* Is this a label? */
3833 if (!tmp && PL_expect == XSTATE
3834 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3836 yylval.pval = savepv(PL_tokenbuf);
3841 /* Check for keywords */
3842 tmp = keyword(PL_tokenbuf, len);
3844 /* Is this a word before a => operator? */
3845 if (*d == '=' && d[1] == '>') {
3847 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3848 yylval.opval->op_private = OPpCONST_BARE;
3849 if (UTF && !IN_BYTE && is_utf8_string((U8*)PL_tokenbuf, len))
3850 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3854 if (tmp < 0) { /* second-class keyword? */
3855 GV *ogv = Nullgv; /* override (winner) */
3856 GV *hgv = Nullgv; /* hidden (loser) */
3857 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3859 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3862 if (GvIMPORTED_CV(gv))
3864 else if (! CvMETHOD(cv))
3868 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3869 (gv = *gvp) != (GV*)&PL_sv_undef &&
3870 GvCVu(gv) && GvIMPORTED_CV(gv))
3876 tmp = 0; /* overridden by import or by GLOBAL */
3879 && -tmp==KEY_lock /* XXX generalizable kludge */
3881 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3883 tmp = 0; /* any sub overrides "weak" keyword */
3885 else { /* no override */
3889 if (ckWARN(WARN_AMBIGUOUS) && hgv
3890 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3891 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3892 "Ambiguous call resolved as CORE::%s(), %s",
3893 GvENAME(hgv), "qualify as such or use &");
3900 default: /* not a keyword */
3903 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3905 /* Get the rest if it looks like a package qualifier */
3907 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3909 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3912 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3913 *s == '\'' ? "'" : "::");
3917 if (PL_expect == XOPERATOR) {
3918 if (PL_bufptr == PL_linestart) {
3919 CopLINE_dec(PL_curcop);
3920 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3921 CopLINE_inc(PL_curcop);
3924 no_op("Bareword",s);
3927 /* Look for a subroutine with this name in current package,
3928 unless name is "Foo::", in which case Foo is a bearword
3929 (and a package name). */
3932 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3934 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3935 Perl_warner(aTHX_ WARN_BAREWORD,
3936 "Bareword \"%s\" refers to nonexistent package",
3939 PL_tokenbuf[len] = '\0';
3946 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3949 /* if we saw a global override before, get the right name */
3952 sv = newSVpvn("CORE::GLOBAL::",14);
3953 sv_catpv(sv,PL_tokenbuf);
3956 sv = newSVpv(PL_tokenbuf,0);
3958 /* Presume this is going to be a bareword of some sort. */
3961 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3962 yylval.opval->op_private = OPpCONST_BARE;
3964 /* And if "Foo::", then that's what it certainly is. */
3969 /* See if it's the indirect object for a list operator. */
3971 if (PL_oldoldbufptr &&
3972 PL_oldoldbufptr < PL_bufptr &&
3973 (PL_oldoldbufptr == PL_last_lop
3974 || PL_oldoldbufptr == PL_last_uni) &&
3975 /* NO SKIPSPACE BEFORE HERE! */
3976 (PL_expect == XREF ||
3977 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3979 bool immediate_paren = *s == '(';
3981 /* (Now we can afford to cross potential line boundary.) */
3984 /* Two barewords in a row may indicate method call. */
3986 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3989 /* If not a declared subroutine, it's an indirect object. */
3990 /* (But it's an indir obj regardless for sort.) */
3992 if ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
3993 ((!gv || !GvCVu(gv)) &&
3994 (PL_last_lop_op != OP_MAPSTART &&
3995 PL_last_lop_op != OP_GREPSTART))))
3997 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
4003 PL_expect = XOPERATOR;
4006 /* Is this a word before a => operator? */
4007 if (*s == '=' && s[1] == '>') {
4009 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
4010 if (UTF && !IN_BYTE && is_utf8_string((U8*)PL_tokenbuf, len))
4011 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
4015 /* If followed by a paren, it's certainly a subroutine. */
4018 if (gv && GvCVu(gv)) {
4019 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
4020 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
4025 PL_nextval[PL_nexttoke].opval = yylval.opval;
4026 PL_expect = XOPERATOR;
4032 /* If followed by var or block, call it a method (unless sub) */
4034 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
4035 PL_last_lop = PL_oldbufptr;
4036 PL_last_lop_op = OP_METHOD;
4040 /* If followed by a bareword, see if it looks like indir obj. */
4042 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
4045 /* Not a method, so call it a subroutine (if defined) */
4047 if (gv && GvCVu(gv)) {
4049 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
4050 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4051 "Ambiguous use of -%s resolved as -&%s()",
4052 PL_tokenbuf, PL_tokenbuf);
4053 /* Check for a constant sub */
4055 if ((sv = cv_const_sv(cv))) {
4057 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
4058 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
4059 yylval.opval->op_private = 0;
4063 /* Resolve to GV now. */
4064 op_free(yylval.opval);
4065 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
4066 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
4067 PL_last_lop = PL_oldbufptr;
4068 PL_last_lop_op = OP_ENTERSUB;
4069 /* Is there a prototype? */
4072 char *proto = SvPV((SV*)cv, len);
4075 if (strEQ(proto, "$"))
4077 if (*proto == '&' && *s == '{') {
4078 sv_setpv(PL_subname,"__ANON__");
4082 PL_nextval[PL_nexttoke].opval = yylval.opval;
4088 /* Call it a bare word */
4090 if (PL_hints & HINT_STRICT_SUBS)
4091 yylval.opval->op_private |= OPpCONST_STRICT;
4094 if (ckWARN(WARN_RESERVED)) {
4095 if (lastchar != '-') {
4096 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4098 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
4105 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
4106 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4107 "Operator or semicolon missing before %c%s",
4108 lastchar, PL_tokenbuf);
4109 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4110 "Ambiguous use of %c resolved as operator %c",
4111 lastchar, lastchar);
4117 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4118 newSVpv(CopFILE(PL_curcop),0));
4122 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4123 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4126 case KEY___PACKAGE__:
4127 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4129 ? newSVsv(PL_curstname)
4138 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4139 char *pname = "main";
4140 if (PL_tokenbuf[2] == 'D')
4141 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
4142 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4145 GvIOp(gv) = newIO();
4146 IoIFP(GvIOp(gv)) = PL_rsfp;
4147 #if defined(HAS_FCNTL) && defined(F_SETFD)
4149 int fd = PerlIO_fileno(PL_rsfp);
4150 fcntl(fd,F_SETFD,fd >= 3);
4153 /* Mark this internal pseudo-handle as clean */
4154 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4156 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4157 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4158 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4160 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4161 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4162 /* if the script was opened in binmode, we need to revert
4163 * it to text mode for compatibility; but only iff it has CRs
4164 * XXX this is a questionable hack at best. */
4165 if (PL_bufend-PL_bufptr > 2
4166 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4169 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4170 loc = PerlIO_tell(PL_rsfp);
4171 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4173 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4175 PerlIO_seek(PL_rsfp, loc, 0);
4179 #ifdef PERLIO_LAYERS
4180 if (UTF && !IN_BYTE)
4181 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
4194 if (PL_expect == XSTATE) {
4201 if (*s == ':' && s[1] == ':') {
4204 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4205 if (!(tmp = keyword(PL_tokenbuf, len)))
4206 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4220 LOP(OP_ACCEPT,XTERM);
4226 LOP(OP_ATAN2,XTERM);
4232 LOP(OP_BINMODE,XTERM);
4235 LOP(OP_BLESS,XTERM);
4244 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4261 if (!PL_cryptseen) {
4262 PL_cryptseen = TRUE;
4266 LOP(OP_CRYPT,XTERM);
4269 if (ckWARN(WARN_CHMOD)) {
4270 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4271 if (*d != '0' && isDIGIT(*d))
4272 Perl_warner(aTHX_ WARN_CHMOD,
4273 "chmod() mode argument is missing initial 0");
4275 LOP(OP_CHMOD,XTERM);
4278 LOP(OP_CHOWN,XTERM);
4281 LOP(OP_CONNECT,XTERM);
4297 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4301 PL_hints |= HINT_BLOCK_SCOPE;
4311 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4312 LOP(OP_DBMOPEN,XTERM);
4318 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4325 yylval.ival = CopLINE(PL_curcop);
4339 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4340 UNIBRACK(OP_ENTEREVAL);
4355 case KEY_endhostent:
4361 case KEY_endservent:
4364 case KEY_endprotoent:
4375 yylval.ival = CopLINE(PL_curcop);
4377 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4379 if ((PL_bufend - p) >= 3 &&
4380 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4382 else if ((PL_bufend - p) >= 4 &&
4383 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4386 if (isIDFIRST_lazy_if(p,UTF)) {
4387 p = scan_ident(p, PL_bufend,
4388 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4392 Perl_croak(aTHX_ "Missing $ on loop variable");
4397 LOP(OP_FORMLINE,XTERM);
4403 LOP(OP_FCNTL,XTERM);
4409 LOP(OP_FLOCK,XTERM);
4418 LOP(OP_GREPSTART, XREF);
4421 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4436 case KEY_getpriority:
4437 LOP(OP_GETPRIORITY,XTERM);
4439 case KEY_getprotobyname:
4442 case KEY_getprotobynumber:
4443 LOP(OP_GPBYNUMBER,XTERM);
4445 case KEY_getprotoent:
4457 case KEY_getpeername:
4458 UNI(OP_GETPEERNAME);
4460 case KEY_gethostbyname:
4463 case KEY_gethostbyaddr:
4464 LOP(OP_GHBYADDR,XTERM);
4466 case KEY_gethostent:
4469 case KEY_getnetbyname:
4472 case KEY_getnetbyaddr:
4473 LOP(OP_GNBYADDR,XTERM);
4478 case KEY_getservbyname:
4479 LOP(OP_GSBYNAME,XTERM);
4481 case KEY_getservbyport:
4482 LOP(OP_GSBYPORT,XTERM);
4484 case KEY_getservent:
4487 case KEY_getsockname:
4488 UNI(OP_GETSOCKNAME);
4490 case KEY_getsockopt:
4491 LOP(OP_GSOCKOPT,XTERM);
4513 yylval.ival = CopLINE(PL_curcop);
4517 LOP(OP_INDEX,XTERM);
4523 LOP(OP_IOCTL,XTERM);
4535 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4567 LOP(OP_LISTEN,XTERM);
4576 s = scan_pat(s,OP_MATCH);
4577 TERM(sublex_start());
4580 LOP(OP_MAPSTART, XREF);
4583 LOP(OP_MKDIR,XTERM);
4586 LOP(OP_MSGCTL,XTERM);
4589 LOP(OP_MSGGET,XTERM);
4592 LOP(OP_MSGRCV,XTERM);
4595 LOP(OP_MSGSND,XTERM);
4601 if (isIDFIRST_lazy_if(s,UTF)) {
4602 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4603 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4605 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4606 if (!PL_in_my_stash) {
4609 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4617 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4624 if (PL_expect != XSTATE)
4625 yyerror("\"no\" not allowed in expression");
4626 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4627 s = force_version(s);
4632 if (*s == '(' || (s = skipspace(s), *s == '('))
4639 if (isIDFIRST_lazy_if(s,UTF)) {
4641 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4643 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4644 Perl_warner(aTHX_ WARN_PRECEDENCE,
4645 "Precedence problem: open %.*s should be open(%.*s)",
4651 yylval.ival = OP_OR;
4661 LOP(OP_OPEN_DIR,XTERM);
4664 checkcomma(s,PL_tokenbuf,"filehandle");
4668 checkcomma(s,PL_tokenbuf,"filehandle");
4687 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4691 LOP(OP_PIPE_OP,XTERM);
4694 s = scan_str(s,FALSE,FALSE);
4696 missingterm((char*)0);
4697 yylval.ival = OP_CONST;
4698 TERM(sublex_start());
4704 s = scan_str(s,FALSE,FALSE);
4706 missingterm((char*)0);
4708 if (SvCUR(PL_lex_stuff)) {
4711 d = SvPV_force(PL_lex_stuff, len);
4714 for (; isSPACE(*d) && len; --len, ++d) ;
4717 if (!warned && ckWARN(WARN_QW)) {
4718 for (; !isSPACE(*d) && len; --len, ++d) {
4720 Perl_warner(aTHX_ WARN_QW,
4721 "Possible attempt to separate words with commas");
4724 else if (*d == '#') {
4725 Perl_warner(aTHX_ WARN_QW,
4726 "Possible attempt to put comments in qw() list");
4732 for (; !isSPACE(*d) && len; --len, ++d) ;
4734 sv = newSVpvn(b, d-b);
4735 if (DO_UTF8(PL_lex_stuff))
4737 words = append_elem(OP_LIST, words,
4738 newSVOP(OP_CONST, 0, tokeq(sv)));
4742 PL_nextval[PL_nexttoke].opval = words;
4747 SvREFCNT_dec(PL_lex_stuff);
4748 PL_lex_stuff = Nullsv;
4754 s = scan_str(s,FALSE,FALSE);
4755 if (tmp == KEY_qu &&
4756 is_utf8_string((U8*)SvPVX(PL_lex_stuff), SvCUR(PL_lex_stuff)))
4757 SvUTF8_on(PL_lex_stuff);
4759 missingterm((char*)0);
4760 yylval.ival = OP_STRINGIFY;
4761 if (SvIVX(PL_lex_stuff) == '\'')
4762 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4763 TERM(sublex_start());
4766 s = scan_pat(s,OP_QR);
4767 TERM(sublex_start());
4770 s = scan_str(s,FALSE,FALSE);
4772 missingterm((char*)0);
4773 yylval.ival = OP_BACKTICK;
4775 TERM(sublex_start());
4782 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4783 s = force_version(s);
4786 *PL_tokenbuf = '\0';
4787 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4788 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4789 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4791 yyerror("<> should be quotes");
4799 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4803 LOP(OP_RENAME,XTERM);
4812 LOP(OP_RINDEX,XTERM);
4835 LOP(OP_REVERSE,XTERM);
4846 TERM(sublex_start());
4848 TOKEN(1); /* force error */
4857 LOP(OP_SELECT,XTERM);
4863 LOP(OP_SEMCTL,XTERM);
4866 LOP(OP_SEMGET,XTERM);
4869 LOP(OP_SEMOP,XTERM);
4875 LOP(OP_SETPGRP,XTERM);
4877 case KEY_setpriority:
4878 LOP(OP_SETPRIORITY,XTERM);
4880 case KEY_sethostent:
4886 case KEY_setservent:
4889 case KEY_setprotoent:
4899 LOP(OP_SEEKDIR,XTERM);
4901 case KEY_setsockopt:
4902 LOP(OP_SSOCKOPT,XTERM);
4908 LOP(OP_SHMCTL,XTERM);
4911 LOP(OP_SHMGET,XTERM);
4914 LOP(OP_SHMREAD,XTERM);
4917 LOP(OP_SHMWRITE,XTERM);
4920 LOP(OP_SHUTDOWN,XTERM);
4929 LOP(OP_SOCKET,XTERM);
4931 case KEY_socketpair:
4932 LOP(OP_SOCKPAIR,XTERM);
4935 checkcomma(s,PL_tokenbuf,"subroutine name");
4937 if (*s == ';' || *s == ')') /* probably a close */
4938 Perl_croak(aTHX_ "sort is now a reserved word");
4940 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4944 LOP(OP_SPLIT,XTERM);
4947 LOP(OP_SPRINTF,XTERM);
4950 LOP(OP_SPLICE,XTERM);
4965 LOP(OP_SUBSTR,XTERM);
4971 char tmpbuf[sizeof PL_tokenbuf];
4973 expectation attrful;
4974 bool have_name, have_proto;
4979 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4980 (*s == ':' && s[1] == ':'))
4983 attrful = XATTRBLOCK;
4984 /* remember buffer pos'n for later force_word */
4985 tboffset = s - PL_oldbufptr;
4986 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4987 if (strchr(tmpbuf, ':'))
4988 sv_setpv(PL_subname, tmpbuf);
4990 sv_setsv(PL_subname,PL_curstname);
4991 sv_catpvn(PL_subname,"::",2);
4992 sv_catpvn(PL_subname,tmpbuf,len);
4999 Perl_croak(aTHX_ "Missing name in \"my sub\"");
5000 PL_expect = XTERMBLOCK;
5001 attrful = XATTRTERM;
5002 sv_setpv(PL_subname,"?");
5006 if (key == KEY_format) {
5008 PL_lex_formbrack = PL_lex_brackets + 1;
5010 (void) force_word(PL_oldbufptr + tboffset, WORD,
5015 /* Look for a prototype */
5019 s = scan_str(s,FALSE,FALSE);
5022 SvREFCNT_dec(PL_lex_stuff);
5023 PL_lex_stuff = Nullsv;
5024 Perl_croak(aTHX_ "Prototype not terminated");
5027 d = SvPVX(PL_lex_stuff);
5029 for (p = d; *p; ++p) {
5034 SvCUR(PL_lex_stuff) = tmp;
5042 if (*s == ':' && s[1] != ':')
5043 PL_expect = attrful;
5046 PL_nextval[PL_nexttoke].opval =
5047 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
5048 PL_lex_stuff = Nullsv;
5052 sv_setpv(PL_subname,"__ANON__");
5055 (void) force_word(PL_oldbufptr + tboffset, WORD,
5064 LOP(OP_SYSTEM,XREF);
5067 LOP(OP_SYMLINK,XTERM);
5070 LOP(OP_SYSCALL,XTERM);
5073 LOP(OP_SYSOPEN,XTERM);
5076 LOP(OP_SYSSEEK,XTERM);
5079 LOP(OP_SYSREAD,XTERM);
5082 LOP(OP_SYSWRITE,XTERM);
5086 TERM(sublex_start());
5107 LOP(OP_TRUNCATE,XTERM);
5119 yylval.ival = CopLINE(PL_curcop);
5123 yylval.ival = CopLINE(PL_curcop);
5127 LOP(OP_UNLINK,XTERM);
5133 LOP(OP_UNPACK,XTERM);
5136 LOP(OP_UTIME,XTERM);
5139 if (ckWARN(WARN_UMASK)) {
5140 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
5141 if (*d != '0' && isDIGIT(*d))
5142 Perl_warner(aTHX_ WARN_UMASK,
5143 "umask: argument is missing initial 0");
5148 LOP(OP_UNSHIFT,XTERM);
5151 if (PL_expect != XSTATE)
5152 yyerror("\"use\" not allowed in expression");
5154 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
5155 s = force_version(s);
5156 if (*s == ';' || (s = skipspace(s), *s == ';')) {
5157 PL_nextval[PL_nexttoke].opval = Nullop;
5162 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5163 s = force_version(s);
5175 yylval.ival = CopLINE(PL_curcop);
5179 PL_hints |= HINT_BLOCK_SCOPE;
5186 LOP(OP_WAITPID,XTERM);
5194 static char ctl_l[2];
5196 if (ctl_l[0] == '\0')
5197 ctl_l[0] = toCTRL('L');
5198 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5201 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5206 if (PL_expect == XOPERATOR)
5212 yylval.ival = OP_XOR;
5217 TERM(sublex_start());
5222 #pragma segment Main
5226 Perl_keyword(pTHX_ register char *d, I32 len)
5231 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5232 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5233 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5234 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5235 if (strEQ(d,"__END__")) return KEY___END__;
5239 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5244 if (strEQ(d,"and")) return -KEY_and;
5245 if (strEQ(d,"abs")) return -KEY_abs;
5248 if (strEQ(d,"alarm")) return -KEY_alarm;
5249 if (strEQ(d,"atan2")) return -KEY_atan2;
5252 if (strEQ(d,"accept")) return -KEY_accept;
5257 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5260 if (strEQ(d,"bless")) return -KEY_bless;
5261 if (strEQ(d,"bind")) return -KEY_bind;
5262 if (strEQ(d,"binmode")) return -KEY_binmode;
5265 if (strEQ(d,"CORE")) return -KEY_CORE;
5266 if (strEQ(d,"CHECK")) return KEY_CHECK;
5271 if (strEQ(d,"cmp")) return -KEY_cmp;
5272 if (strEQ(d,"chr")) return -KEY_chr;
5273 if (strEQ(d,"cos")) return -KEY_cos;
5276 if (strEQ(d,"chop")) return -KEY_chop;
5279 if (strEQ(d,"close")) return -KEY_close;
5280 if (strEQ(d,"chdir")) return -KEY_chdir;
5281 if (strEQ(d,"chomp")) return -KEY_chomp;
5282 if (strEQ(d,"chmod")) return -KEY_chmod;
5283 if (strEQ(d,"chown")) return -KEY_chown;
5284 if (strEQ(d,"crypt")) return -KEY_crypt;
5287 if (strEQ(d,"chroot")) return -KEY_chroot;
5288 if (strEQ(d,"caller")) return -KEY_caller;
5291 if (strEQ(d,"connect")) return -KEY_connect;
5294 if (strEQ(d,"closedir")) return -KEY_closedir;
5295 if (strEQ(d,"continue")) return -KEY_continue;
5300 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5305 if (strEQ(d,"do")) return KEY_do;
5308 if (strEQ(d,"die")) return -KEY_die;
5311 if (strEQ(d,"dump")) return -KEY_dump;
5314 if (strEQ(d,"delete")) return KEY_delete;
5317 if (strEQ(d,"defined")) return KEY_defined;
5318 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5321 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5326 if (strEQ(d,"END")) return KEY_END;
5331 if (strEQ(d,"eq")) return -KEY_eq;
5334 if (strEQ(d,"eof")) return -KEY_eof;
5335 if (strEQ(d,"exp")) return -KEY_exp;
5338 if (strEQ(d,"else")) return KEY_else;
5339 if (strEQ(d,"exit")) return -KEY_exit;
5340 if (strEQ(d,"eval")) return KEY_eval;
5341 if (strEQ(d,"exec")) return -KEY_exec;
5342 if (strEQ(d,"each")) return -KEY_each;
5345 if (strEQ(d,"elsif")) return KEY_elsif;
5348 if (strEQ(d,"exists")) return KEY_exists;
5349 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5352 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5353 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5356 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5359 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5360 if (strEQ(d,"endservent")) return -KEY_endservent;
5363 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5370 if (strEQ(d,"for")) return KEY_for;
5373 if (strEQ(d,"fork")) return -KEY_fork;
5376 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5377 if (strEQ(d,"flock")) return -KEY_flock;
5380 if (strEQ(d,"format")) return KEY_format;
5381 if (strEQ(d,"fileno")) return -KEY_fileno;
5384 if (strEQ(d,"foreach")) return KEY_foreach;
5387 if (strEQ(d,"formline")) return -KEY_formline;
5392 if (strnEQ(d,"get",3)) {
5397 if (strEQ(d,"ppid")) return -KEY_getppid;
5398 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5401 if (strEQ(d,"pwent")) return -KEY_getpwent;
5402 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5403 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5406 if (strEQ(d,"peername")) return -KEY_getpeername;
5407 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5408 if (strEQ(d,"priority")) return -KEY_getpriority;
5411 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5414 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5418 else if (*d == 'h') {
5419 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5420 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5421 if (strEQ(d,"hostent")) return -KEY_gethostent;
5423 else if (*d == 'n') {
5424 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5425 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5426 if (strEQ(d,"netent")) return -KEY_getnetent;
5428 else if (*d == 's') {
5429 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5430 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5431 if (strEQ(d,"servent")) return -KEY_getservent;
5432 if (strEQ(d,"sockname")) return -KEY_getsockname;
5433 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5435 else if (*d == 'g') {
5436 if (strEQ(d,"grent")) return -KEY_getgrent;
5437 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5438 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5440 else if (*d == 'l') {
5441 if (strEQ(d,"login")) return -KEY_getlogin;
5443 else if (strEQ(d,"c")) return -KEY_getc;
5448 if (strEQ(d,"gt")) return -KEY_gt;
5449 if (strEQ(d,"ge")) return -KEY_ge;
5452 if (strEQ(d,"grep")) return KEY_grep;
5453 if (strEQ(d,"goto")) return KEY_goto;
5454 if (strEQ(d,"glob")) return KEY_glob;
5457 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5462 if (strEQ(d,"hex")) return -KEY_hex;
5465 if (strEQ(d,"INIT")) return KEY_INIT;
5470 if (strEQ(d,"if")) return KEY_if;
5473 if (strEQ(d,"int")) return -KEY_int;
5476 if (strEQ(d,"index")) return -KEY_index;
5477 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5482 if (strEQ(d,"join")) return -KEY_join;
5486 if (strEQ(d,"keys")) return -KEY_keys;
5487 if (strEQ(d,"kill")) return -KEY_kill;
5493 if (strEQ(d,"lt")) return -KEY_lt;
5494 if (strEQ(d,"le")) return -KEY_le;
5495 if (strEQ(d,"lc")) return -KEY_lc;
5498 if (strEQ(d,"log")) return -KEY_log;
5501 if (strEQ(d,"last")) return KEY_last;
5502 if (strEQ(d,"link")) return -KEY_link;
5503 if (strEQ(d,"lock")) return -KEY_lock;
5506 if (strEQ(d,"local")) return KEY_local;
5507 if (strEQ(d,"lstat")) return -KEY_lstat;
5510 if (strEQ(d,"length")) return -KEY_length;
5511 if (strEQ(d,"listen")) return -KEY_listen;
5514 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5517 if (strEQ(d,"localtime")) return -KEY_localtime;
5523 case 1: return KEY_m;
5525 if (strEQ(d,"my")) return KEY_my;
5528 if (strEQ(d,"map")) return KEY_map;
5531 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5534 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5535 if (strEQ(d,"msgget")) return -KEY_msgget;
5536 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5537 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5542 if (strEQ(d,"next")) return KEY_next;
5543 if (strEQ(d,"ne")) return -KEY_ne;
5544 if (strEQ(d,"not")) return -KEY_not;
5545 if (strEQ(d,"no")) return KEY_no;
5550 if (strEQ(d,"or")) return -KEY_or;
5553 if (strEQ(d,"ord")) return -KEY_ord;
5554 if (strEQ(d,"oct")) return -KEY_oct;
5555 if (strEQ(d,"our")) return KEY_our;
5558 if (strEQ(d,"open")) return -KEY_open;
5561 if (strEQ(d,"opendir")) return -KEY_opendir;
5568 if (strEQ(d,"pop")) return -KEY_pop;
5569 if (strEQ(d,"pos")) return KEY_pos;
5572 if (strEQ(d,"push")) return -KEY_push;
5573 if (strEQ(d,"pack")) return -KEY_pack;
5574 if (strEQ(d,"pipe")) return -KEY_pipe;
5577 if (strEQ(d,"print")) return KEY_print;
5580 if (strEQ(d,"printf")) return KEY_printf;
5583 if (strEQ(d,"package")) return KEY_package;
5586 if (strEQ(d,"prototype")) return KEY_prototype;
5591 if (strEQ(d,"q")) return KEY_q;
5592 if (strEQ(d,"qr")) return KEY_qr;
5593 if (strEQ(d,"qq")) return KEY_qq;
5594 if (strEQ(d,"qu")) return KEY_qu;
5595 if (strEQ(d,"qw")) return KEY_qw;
5596 if (strEQ(d,"qx")) return KEY_qx;
5598 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5603 if (strEQ(d,"ref")) return -KEY_ref;
5606 if (strEQ(d,"read")) return -KEY_read;
5607 if (strEQ(d,"rand")) return -KEY_rand;
5608 if (strEQ(d,"recv")) return -KEY_recv;
5609 if (strEQ(d,"redo")) return KEY_redo;
5612 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5613 if (strEQ(d,"reset")) return -KEY_reset;
5616 if (strEQ(d,"return")) return KEY_return;
5617 if (strEQ(d,"rename")) return -KEY_rename;
5618 if (strEQ(d,"rindex")) return -KEY_rindex;
5621 if (strEQ(d,"require")) return -KEY_require;
5622 if (strEQ(d,"reverse")) return -KEY_reverse;
5623 if (strEQ(d,"readdir")) return -KEY_readdir;
5626 if (strEQ(d,"readlink")) return -KEY_readlink;
5627 if (strEQ(d,"readline")) return -KEY_readline;
5628 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5631 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5637 case 0: return KEY_s;
5639 if (strEQ(d,"scalar")) return KEY_scalar;
5644 if (strEQ(d,"seek")) return -KEY_seek;
5645 if (strEQ(d,"send")) return -KEY_send;
5648 if (strEQ(d,"semop")) return -KEY_semop;
5651 if (strEQ(d,"select")) return -KEY_select;
5652 if (strEQ(d,"semctl")) return -KEY_semctl;
5653 if (strEQ(d,"semget")) return -KEY_semget;
5656 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5657 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5660 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5661 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5664 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5667 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5668 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5669 if (strEQ(d,"setservent")) return -KEY_setservent;
5672 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5673 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5680 if (strEQ(d,"shift")) return -KEY_shift;
5683 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5684 if (strEQ(d,"shmget")) return -KEY_shmget;
5687 if (strEQ(d,"shmread")) return -KEY_shmread;
5690 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5691 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5696 if (strEQ(d,"sin")) return -KEY_sin;
5699 if (strEQ(d,"sleep")) return -KEY_sleep;
5702 if (strEQ(d,"sort")) return KEY_sort;
5703 if (strEQ(d,"socket")) return -KEY_socket;
5704 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5707 if (strEQ(d,"split")) return KEY_split;
5708 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5709 if (strEQ(d,"splice")) return -KEY_splice;
5712 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5715 if (strEQ(d,"srand")) return -KEY_srand;
5718 if (strEQ(d,"stat")) return -KEY_stat;
5719 if (strEQ(d,"study")) return KEY_study;
5722 if (strEQ(d,"substr")) return -KEY_substr;
5723 if (strEQ(d,"sub")) return KEY_sub;
5728 if (strEQ(d,"system")) return -KEY_system;
5731 if (strEQ(d,"symlink")) return -KEY_symlink;
5732 if (strEQ(d,"syscall")) return -KEY_syscall;
5733 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5734 if (strEQ(d,"sysread")) return -KEY_sysread;
5735 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5738 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5747 if (strEQ(d,"tr")) return KEY_tr;
5750 if (strEQ(d,"tie")) return KEY_tie;
5753 if (strEQ(d,"tell")) return -KEY_tell;
5754 if (strEQ(d,"tied")) return KEY_tied;
5755 if (strEQ(d,"time")) return -KEY_time;
5758 if (strEQ(d,"times")) return -KEY_times;
5761 if (strEQ(d,"telldir")) return -KEY_telldir;
5764 if (strEQ(d,"truncate")) return -KEY_truncate;
5771 if (strEQ(d,"uc")) return -KEY_uc;
5774 if (strEQ(d,"use")) return KEY_use;
5777 if (strEQ(d,"undef")) return KEY_undef;
5778 if (strEQ(d,"until")) return KEY_until;
5779 if (strEQ(d,"untie")) return KEY_untie;
5780 if (strEQ(d,"utime")) return -KEY_utime;
5781 if (strEQ(d,"umask")) return -KEY_umask;
5784 if (strEQ(d,"unless")) return KEY_unless;
5785 if (strEQ(d,"unpack")) return -KEY_unpack;
5786 if (strEQ(d,"unlink")) return -KEY_unlink;
5789 if (strEQ(d,"unshift")) return -KEY_unshift;
5790 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5795 if (strEQ(d,"values")) return -KEY_values;
5796 if (strEQ(d,"vec")) return -KEY_vec;
5801 if (strEQ(d,"warn")) return -KEY_warn;
5802 if (strEQ(d,"wait")) return -KEY_wait;
5805 if (strEQ(d,"while")) return KEY_while;
5806 if (strEQ(d,"write")) return -KEY_write;
5809 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5812 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5817 if (len == 1) return -KEY_x;
5818 if (strEQ(d,"xor")) return -KEY_xor;
5821 if (len == 1) return KEY_y;
5830 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5834 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5835 if (ckWARN(WARN_SYNTAX)) {
5837 for (w = s+2; *w && level; w++) {
5844 for (; *w && isSPACE(*w); w++) ;
5845 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5846 Perl_warner(aTHX_ WARN_SYNTAX,
5847 "%s (...) interpreted as function",name);
5850 while (s < PL_bufend && isSPACE(*s))
5854 while (s < PL_bufend && isSPACE(*s))
5856 if (isIDFIRST_lazy_if(s,UTF)) {
5858 while (isALNUM_lazy_if(s,UTF))
5860 while (s < PL_bufend && isSPACE(*s))
5865 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5869 Perl_croak(aTHX_ "No comma allowed after %s", what);
5874 /* Either returns sv, or mortalizes sv and returns a new SV*.
5875 Best used as sv=new_constant(..., sv, ...).
5876 If s, pv are NULL, calls subroutine with one argument,
5877 and type is used with error messages only. */
5880 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5884 HV *table = GvHV(PL_hintgv); /* ^H */
5888 const char *why1, *why2, *why3;
5890 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
5893 why2 = strEQ(key,"charnames")
5894 ? "(possibly a missing \"use charnames ...\")"
5896 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
5897 (type ? type: "undef"), why2);
5899 /* This is convoluted and evil ("goto considered harmful")
5900 * but I do not understand the intricacies of all the different
5901 * failure modes of %^H in here. The goal here is to make
5902 * the most probable error message user-friendly. --jhi */
5907 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
5908 (type ? type: "undef"), why1, why2, why3);
5910 yyerror(SvPVX(msg));
5914 cvp = hv_fetch(table, key, strlen(key), FALSE);
5915 if (!cvp || !SvOK(*cvp)) {
5918 why3 = "} is not defined";
5921 sv_2mortal(sv); /* Parent created it permanently */
5924 pv = sv_2mortal(newSVpvn(s, len));
5926 typesv = sv_2mortal(newSVpv(type, 0));
5928 typesv = &PL_sv_undef;
5930 PUSHSTACKi(PERLSI_OVERLOAD);
5942 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5946 /* Check the eval first */
5947 if (!PL_in_eval && SvTRUE(ERRSV)) {
5949 sv_catpv(ERRSV, "Propagated");
5950 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5952 res = SvREFCNT_inc(sv);
5956 (void)SvREFCNT_inc(res);
5965 why1 = "Call to &{$^H{";
5967 why3 = "}} did not return a defined value";
5976 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5978 register char *d = dest;
5979 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5982 Perl_croak(aTHX_ ident_too_long);
5983 if (isALNUM(*s)) /* UTF handled below */
5985 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
5990 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5994 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
5995 char *t = s + UTF8SKIP(s);
5996 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
5998 if (d + (t - s) > e)
5999 Perl_croak(aTHX_ ident_too_long);
6000 Copy(s, d, t - s, char);
6013 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
6023 e = d + destlen - 3; /* two-character token, ending NUL */
6025 while (isDIGIT(*s)) {
6027 Perl_croak(aTHX_ ident_too_long);
6034 Perl_croak(aTHX_ ident_too_long);
6035 if (isALNUM(*s)) /* UTF handled below */
6037 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
6042 else if (*s == ':' && s[1] == ':') {
6046 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6047 char *t = s + UTF8SKIP(s);
6048 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6050 if (d + (t - s) > e)
6051 Perl_croak(aTHX_ ident_too_long);
6052 Copy(s, d, t - s, char);
6063 if (PL_lex_state != LEX_NORMAL)
6064 PL_lex_state = LEX_INTERPENDMAYBE;
6067 if (*s == '$' && s[1] &&
6068 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
6081 if (*d == '^' && *s && isCONTROLVAR(*s)) {
6086 if (isSPACE(s[-1])) {
6089 if (!SPACE_OR_TAB(ch)) {
6095 if (isIDFIRST_lazy_if(d,UTF)) {
6099 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
6101 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
6104 Copy(s, d, e - s, char);
6109 while ((isALNUM(*s) || *s == ':') && d < e)
6112 Perl_croak(aTHX_ ident_too_long);
6115 while (s < send && SPACE_OR_TAB(*s)) s++;
6116 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
6117 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
6118 const char *brack = *s == '[' ? "[...]" : "{...}";
6119 Perl_warner(aTHX_ WARN_AMBIGUOUS,
6120 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
6121 funny, dest, brack, funny, dest, brack);
6124 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
6128 /* Handle extended ${^Foo} variables
6129 * 1999-02-27 mjd-perl-patch@plover.com */
6130 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
6134 while (isALNUM(*s) && d < e) {
6138 Perl_croak(aTHX_ ident_too_long);
6143 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
6144 PL_lex_state = LEX_INTERPEND;
6147 if (PL_lex_state == LEX_NORMAL) {
6148 if (ckWARN(WARN_AMBIGUOUS) &&
6149 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
6151 Perl_warner(aTHX_ WARN_AMBIGUOUS,
6152 "Ambiguous use of %c{%s} resolved to %c%s",
6153 funny, dest, funny, dest);
6158 s = bracket; /* let the parser handle it */
6162 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
6163 PL_lex_state = LEX_INTERPEND;
6168 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
6173 *pmfl |= PMf_GLOBAL;
6175 *pmfl |= PMf_CONTINUE;
6179 *pmfl |= PMf_MULTILINE;
6181 *pmfl |= PMf_SINGLELINE;
6183 *pmfl |= PMf_EXTENDED;
6187 S_scan_pat(pTHX_ char *start, I32 type)
6192 s = scan_str(start,FALSE,FALSE);
6195 SvREFCNT_dec(PL_lex_stuff);
6196 PL_lex_stuff = Nullsv;
6197 Perl_croak(aTHX_ "Search pattern not terminated");
6200 pm = (PMOP*)newPMOP(type, 0);
6201 if (PL_multi_open == '?')
6202 pm->op_pmflags |= PMf_ONCE;
6204 while (*s && strchr("iomsx", *s))
6205 pmflag(&pm->op_pmflags,*s++);
6208 while (*s && strchr("iogcmsx", *s))
6209 pmflag(&pm->op_pmflags,*s++);
6211 pm->op_pmpermflags = pm->op_pmflags;
6213 PL_lex_op = (OP*)pm;
6214 yylval.ival = OP_MATCH;
6219 S_scan_subst(pTHX_ char *start)
6226 yylval.ival = OP_NULL;
6228 s = scan_str(start,FALSE,FALSE);
6232 SvREFCNT_dec(PL_lex_stuff);
6233 PL_lex_stuff = Nullsv;
6234 Perl_croak(aTHX_ "Substitution pattern not terminated");
6237 if (s[-1] == PL_multi_open)
6240 first_start = PL_multi_start;
6241 s = scan_str(s,FALSE,FALSE);
6244 SvREFCNT_dec(PL_lex_stuff);
6245 PL_lex_stuff = Nullsv;
6247 SvREFCNT_dec(PL_lex_repl);
6248 PL_lex_repl = Nullsv;
6249 Perl_croak(aTHX_ "Substitution replacement not terminated");
6251 PL_multi_start = first_start; /* so whole substitution is taken together */
6253 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6259 else if (strchr("iogcmsx", *s))
6260 pmflag(&pm->op_pmflags,*s++);
6267 PL_sublex_info.super_bufptr = s;
6268 PL_sublex_info.super_bufend = PL_bufend;
6270 pm->op_pmflags |= PMf_EVAL;
6271 repl = newSVpvn("",0);
6273 sv_catpv(repl, es ? "eval " : "do ");
6274 sv_catpvn(repl, "{ ", 2);
6275 sv_catsv(repl, PL_lex_repl);
6276 sv_catpvn(repl, " };", 2);
6278 SvREFCNT_dec(PL_lex_repl);
6282 pm->op_pmpermflags = pm->op_pmflags;
6283 PL_lex_op = (OP*)pm;
6284 yylval.ival = OP_SUBST;
6289 S_scan_trans(pTHX_ char *start)
6300 yylval.ival = OP_NULL;
6302 s = scan_str(start,FALSE,FALSE);
6305 SvREFCNT_dec(PL_lex_stuff);
6306 PL_lex_stuff = Nullsv;
6307 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6309 if (s[-1] == PL_multi_open)
6312 s = scan_str(s,FALSE,FALSE);
6315 SvREFCNT_dec(PL_lex_stuff);
6316 PL_lex_stuff = Nullsv;
6318 SvREFCNT_dec(PL_lex_repl);
6319 PL_lex_repl = Nullsv;
6320 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6323 New(803,tbl,256,short);
6324 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6326 complement = del = squash = 0;
6327 while (strchr("cds", *s)) {
6329 complement = OPpTRANS_COMPLEMENT;
6331 del = OPpTRANS_DELETE;
6333 squash = OPpTRANS_SQUASH;
6336 o->op_private = del|squash|complement|
6337 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
6338 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
6341 yylval.ival = OP_TRANS;
6346 S_scan_heredoc(pTHX_ register char *s)
6349 I32 op_type = OP_SCALAR;
6356 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6360 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6363 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6364 if (*peek && strchr("`'\"",*peek)) {
6367 s = delimcpy(d, e, s, PL_bufend, term, &len);
6377 if (!isALNUM_lazy_if(s,UTF))
6378 deprecate("bare << to mean <<\"\"");
6379 for (; isALNUM_lazy_if(s,UTF); s++) {
6384 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6385 Perl_croak(aTHX_ "Delimiter for here document is too long");
6388 len = d - PL_tokenbuf;
6389 #ifndef PERL_STRICT_CR
6390 d = strchr(s, '\r');
6394 while (s < PL_bufend) {
6400 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6409 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6414 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6415 herewas = newSVpvn(s,PL_bufend-s);
6417 s--, herewas = newSVpvn(s,d-s);
6418 s += SvCUR(herewas);
6420 tmpstr = NEWSV(87,79);
6421 sv_upgrade(tmpstr, SVt_PVIV);
6426 else if (term == '`') {
6427 op_type = OP_BACKTICK;
6428 SvIVX(tmpstr) = '\\';
6432 PL_multi_start = CopLINE(PL_curcop);
6433 PL_multi_open = PL_multi_close = '<';
6434 term = *PL_tokenbuf;
6435 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6436 char *bufptr = PL_sublex_info.super_bufptr;
6437 char *bufend = PL_sublex_info.super_bufend;
6438 char *olds = s - SvCUR(herewas);
6439 s = strchr(bufptr, '\n');
6443 while (s < bufend &&
6444 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6446 CopLINE_inc(PL_curcop);
6449 CopLINE_set(PL_curcop, PL_multi_start);
6450 missingterm(PL_tokenbuf);
6452 sv_setpvn(herewas,bufptr,d-bufptr+1);
6453 sv_setpvn(tmpstr,d+1,s-d);
6455 sv_catpvn(herewas,s,bufend-s);
6456 (void)strcpy(bufptr,SvPVX(herewas));
6463 while (s < PL_bufend &&
6464 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6466 CopLINE_inc(PL_curcop);
6468 if (s >= PL_bufend) {
6469 CopLINE_set(PL_curcop, PL_multi_start);
6470 missingterm(PL_tokenbuf);
6472 sv_setpvn(tmpstr,d+1,s-d);
6474 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6476 sv_catpvn(herewas,s,PL_bufend-s);
6477 sv_setsv(PL_linestr,herewas);
6478 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6479 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6482 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6483 while (s >= PL_bufend) { /* multiple line string? */
6485 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6486 CopLINE_set(PL_curcop, PL_multi_start);
6487 missingterm(PL_tokenbuf);
6489 CopLINE_inc(PL_curcop);
6490 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6491 #ifndef PERL_STRICT_CR
6492 if (PL_bufend - PL_linestart >= 2) {
6493 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6494 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6496 PL_bufend[-2] = '\n';
6498 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6500 else if (PL_bufend[-1] == '\r')
6501 PL_bufend[-1] = '\n';
6503 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6504 PL_bufend[-1] = '\n';
6506 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6507 SV *sv = NEWSV(88,0);
6509 sv_upgrade(sv, SVt_PVMG);
6510 sv_setsv(sv,PL_linestr);
6511 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6513 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6516 sv_catsv(PL_linestr,herewas);
6517 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6521 sv_catsv(tmpstr,PL_linestr);
6526 PL_multi_end = CopLINE(PL_curcop);
6527 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6528 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6529 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6531 SvREFCNT_dec(herewas);
6532 if (UTF && !IN_BYTE && is_utf8_string((U8*)SvPVX(tmpstr), SvCUR(tmpstr)))
6534 PL_lex_stuff = tmpstr;
6535 yylval.ival = op_type;
6540 takes: current position in input buffer
6541 returns: new position in input buffer
6542 side-effects: yylval and lex_op are set.
6547 <FH> read from filehandle
6548 <pkg::FH> read from package qualified filehandle
6549 <pkg'FH> read from package qualified filehandle
6550 <$fh> read from filehandle in $fh
6556 S_scan_inputsymbol(pTHX_ char *start)
6558 register char *s = start; /* current position in buffer */
6564 d = PL_tokenbuf; /* start of temp holding space */
6565 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6566 end = strchr(s, '\n');
6569 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6571 /* die if we didn't have space for the contents of the <>,
6572 or if it didn't end, or if we see a newline
6575 if (len >= sizeof PL_tokenbuf)
6576 Perl_croak(aTHX_ "Excessively long <> operator");
6578 Perl_croak(aTHX_ "Unterminated <> operator");
6583 Remember, only scalar variables are interpreted as filehandles by
6584 this code. Anything more complex (e.g., <$fh{$num}>) will be
6585 treated as a glob() call.
6586 This code makes use of the fact that except for the $ at the front,
6587 a scalar variable and a filehandle look the same.
6589 if (*d == '$' && d[1]) d++;
6591 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6592 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6595 /* If we've tried to read what we allow filehandles to look like, and
6596 there's still text left, then it must be a glob() and not a getline.
6597 Use scan_str to pull out the stuff between the <> and treat it
6598 as nothing more than a string.
6601 if (d - PL_tokenbuf != len) {
6602 yylval.ival = OP_GLOB;
6604 s = scan_str(start,FALSE,FALSE);
6606 Perl_croak(aTHX_ "Glob not terminated");
6610 /* we're in a filehandle read situation */
6613 /* turn <> into <ARGV> */
6615 (void)strcpy(d,"ARGV");
6617 /* if <$fh>, create the ops to turn the variable into a
6623 /* try to find it in the pad for this block, otherwise find
6624 add symbol table ops
6626 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6627 OP *o = newOP(OP_PADSV, 0);
6629 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6632 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6633 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6634 newUNOP(OP_RV2SV, 0,
6635 newGVOP(OP_GV, 0, gv)));
6637 PL_lex_op->op_flags |= OPf_SPECIAL;
6638 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6639 yylval.ival = OP_NULL;
6642 /* If it's none of the above, it must be a literal filehandle
6643 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6645 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6646 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6647 yylval.ival = OP_NULL;
6656 takes: start position in buffer
6657 keep_quoted preserve \ on the embedded delimiter(s)
6658 keep_delims preserve the delimiters around the string
6659 returns: position to continue reading from buffer
6660 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6661 updates the read buffer.
6663 This subroutine pulls a string out of the input. It is called for:
6664 q single quotes q(literal text)
6665 ' single quotes 'literal text'
6666 qq double quotes qq(interpolate $here please)
6667 " double quotes "interpolate $here please"
6668 qx backticks qx(/bin/ls -l)
6669 ` backticks `/bin/ls -l`
6670 qw quote words @EXPORT_OK = qw( func() $spam )
6671 m// regexp match m/this/
6672 s/// regexp substitute s/this/that/
6673 tr/// string transliterate tr/this/that/
6674 y/// string transliterate y/this/that/
6675 ($*@) sub prototypes sub foo ($)
6676 (stuff) sub attr parameters sub foo : attr(stuff)
6677 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6679 In most of these cases (all but <>, patterns and transliterate)
6680 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6681 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6682 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6685 It skips whitespace before the string starts, and treats the first
6686 character as the delimiter. If the delimiter is one of ([{< then
6687 the corresponding "close" character )]}> is used as the closing
6688 delimiter. It allows quoting of delimiters, and if the string has
6689 balanced delimiters ([{<>}]) it allows nesting.
6691 The lexer always reads these strings into lex_stuff, except in the
6692 case of the operators which take *two* arguments (s/// and tr///)
6693 when it checks to see if lex_stuff is full (presumably with the 1st
6694 arg to s or tr) and if so puts the string into lex_repl.
6699 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6701 SV *sv; /* scalar value: string */
6702 char *tmps; /* temp string, used for delimiter matching */
6703 register char *s = start; /* current position in the buffer */
6704 register char term; /* terminating character */
6705 register char *to; /* current position in the sv's data */
6706 I32 brackets = 1; /* bracket nesting level */
6707 bool has_utf8 = FALSE; /* is there any utf8 content? */
6709 /* skip space before the delimiter */
6713 /* mark where we are, in case we need to report errors */
6716 /* after skipping whitespace, the next character is the terminator */
6718 if (UTF8_IS_CONTINUED(term) && UTF)
6721 /* mark where we are */
6722 PL_multi_start = CopLINE(PL_curcop);
6723 PL_multi_open = term;
6725 /* find corresponding closing delimiter */
6726 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6728 PL_multi_close = term;
6730 /* create a new SV to hold the contents. 87 is leak category, I'm
6731 assuming. 79 is the SV's initial length. What a random number. */
6733 sv_upgrade(sv, SVt_PVIV);
6735 (void)SvPOK_only(sv); /* validate pointer */
6737 /* move past delimiter and try to read a complete string */
6739 sv_catpvn(sv, s, 1);
6742 /* extend sv if need be */
6743 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6744 /* set 'to' to the next character in the sv's string */
6745 to = SvPVX(sv)+SvCUR(sv);
6747 /* if open delimiter is the close delimiter read unbridle */
6748 if (PL_multi_open == PL_multi_close) {
6749 for (; s < PL_bufend; s++,to++) {
6750 /* embedded newlines increment the current line number */
6751 if (*s == '\n' && !PL_rsfp)
6752 CopLINE_inc(PL_curcop);
6753 /* handle quoted delimiters */
6754 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6755 if (!keep_quoted && s[1] == term)
6757 /* any other quotes are simply copied straight through */
6761 /* terminate when run out of buffer (the for() condition), or
6762 have found the terminator */
6763 else if (*s == term)
6765 else if (!has_utf8 && UTF8_IS_CONTINUED(*s) && UTF)
6771 /* if the terminator isn't the same as the start character (e.g.,
6772 matched brackets), we have to allow more in the quoting, and
6773 be prepared for nested brackets.
6776 /* read until we run out of string, or we find the terminator */
6777 for (; s < PL_bufend; s++,to++) {
6778 /* embedded newlines increment the line count */
6779 if (*s == '\n' && !PL_rsfp)
6780 CopLINE_inc(PL_curcop);
6781 /* backslashes can escape the open or closing characters */
6782 if (*s == '\\' && s+1 < PL_bufend) {
6784 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6789 /* allow nested opens and closes */
6790 else if (*s == PL_multi_close && --brackets <= 0)
6792 else if (*s == PL_multi_open)
6794 else if (!has_utf8 && UTF8_IS_CONTINUED(*s) && UTF)
6799 /* terminate the copied string and update the sv's end-of-string */
6801 SvCUR_set(sv, to - SvPVX(sv));
6804 * this next chunk reads more into the buffer if we're not done yet
6808 break; /* handle case where we are done yet :-) */
6810 #ifndef PERL_STRICT_CR
6811 if (to - SvPVX(sv) >= 2) {
6812 if ((to[-2] == '\r' && to[-1] == '\n') ||
6813 (to[-2] == '\n' && to[-1] == '\r'))
6817 SvCUR_set(sv, to - SvPVX(sv));
6819 else if (to[-1] == '\r')
6822 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6826 /* if we're out of file, or a read fails, bail and reset the current
6827 line marker so we can report where the unterminated string began
6830 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6832 CopLINE_set(PL_curcop, PL_multi_start);
6835 /* we read a line, so increment our line counter */
6836 CopLINE_inc(PL_curcop);
6838 /* update debugger info */
6839 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6840 SV *sv = NEWSV(88,0);
6842 sv_upgrade(sv, SVt_PVMG);
6843 sv_setsv(sv,PL_linestr);
6844 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6847 /* having changed the buffer, we must update PL_bufend */
6848 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6851 /* at this point, we have successfully read the delimited string */
6854 sv_catpvn(sv, s, 1);
6857 PL_multi_end = CopLINE(PL_curcop);
6860 /* if we allocated too much space, give some back */
6861 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6862 SvLEN_set(sv, SvCUR(sv) + 1);
6863 Renew(SvPVX(sv), SvLEN(sv), char);
6866 /* decide whether this is the first or second quoted string we've read
6879 takes: pointer to position in buffer
6880 returns: pointer to new position in buffer
6881 side-effects: builds ops for the constant in yylval.op
6883 Read a number in any of the formats that Perl accepts:
6885 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6886 [\d_]+(\.[\d_]*)?[Ee](\d+)
6888 Underbars (_) are allowed in decimal numbers. If -w is on,
6889 underbars before a decimal point must be at three digit intervals.
6891 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6894 If it reads a number without a decimal point or an exponent, it will
6895 try converting the number to an integer and see if it can do so
6896 without loss of precision.
6900 Perl_scan_num(pTHX_ char *start, YYSTYPE* lvalp)
6902 register char *s = start; /* current position in buffer */
6903 register char *d; /* destination in temp buffer */
6904 register char *e; /* end of temp buffer */
6905 NV nv; /* number read, as a double */
6906 SV *sv = Nullsv; /* place to put the converted number */
6907 bool floatit; /* boolean: int or float? */
6908 char *lastub = 0; /* position of last underbar */
6909 static char number_too_long[] = "Number too long";
6911 /* We use the first character to decide what type of number this is */
6915 Perl_croak(aTHX_ "panic: scan_num");
6917 /* if it starts with a 0, it could be an octal number, a decimal in
6918 0.13 disguise, or a hexadecimal number, or a binary number. */
6922 u holds the "number so far"
6923 shift the power of 2 of the base
6924 (hex == 4, octal == 3, binary == 1)
6925 overflowed was the number more than we can hold?
6927 Shift is used when we add a digit. It also serves as an "are
6928 we in octal/hex/binary?" indicator to disallow hex characters
6934 bool overflowed = FALSE;
6935 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6936 static char* bases[5] = { "", "binary", "", "octal",
6938 static char* Bases[5] = { "", "Binary", "", "Octal",
6940 static char *maxima[5] = { "",
6941 "0b11111111111111111111111111111111",
6945 char *base, *Base, *max;
6951 } else if (s[1] == 'b') {
6955 /* check for a decimal in disguise */
6956 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6958 /* so it must be octal */
6962 base = bases[shift];
6963 Base = Bases[shift];
6964 max = maxima[shift];
6966 /* read the rest of the number */
6968 /* x is used in the overflow test,
6969 b is the digit we're adding on. */
6974 /* if we don't mention it, we're done */
6983 /* 8 and 9 are not octal */
6986 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6990 case '2': case '3': case '4':
6991 case '5': case '6': case '7':
6993 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6997 b = *s++ & 15; /* ASCII digit -> value of digit */
7001 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
7002 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
7003 /* make sure they said 0x */
7008 /* Prepare to put the digit we have onto the end
7009 of the number so far. We check for overflows.
7014 x = u << shift; /* make room for the digit */
7016 if ((x >> shift) != u
7017 && !(PL_hints & HINT_NEW_BINARY)) {
7020 if (ckWARN_d(WARN_OVERFLOW))
7021 Perl_warner(aTHX_ WARN_OVERFLOW,
7022 "Integer overflow in %s number",
7025 u = x | b; /* add the digit to the end */
7028 n *= nvshift[shift];
7029 /* If an NV has not enough bits in its
7030 * mantissa to represent an UV this summing of
7031 * small low-order numbers is a waste of time
7032 * (because the NV cannot preserve the
7033 * low-order bits anyway): we could just
7034 * remember when did we overflow and in the
7035 * end just multiply n by the right
7043 /* if we get here, we had success: make a scalar value from
7049 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
7050 Perl_warner(aTHX_ WARN_PORTABLE,
7051 "%s number > %s non-portable",
7057 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
7058 Perl_warner(aTHX_ WARN_PORTABLE,
7059 "%s number > %s non-portable",
7064 if (PL_hints & HINT_NEW_BINARY)
7065 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
7070 handle decimal numbers.
7071 we're also sent here when we read a 0 as the first digit
7073 case '1': case '2': case '3': case '4': case '5':
7074 case '6': case '7': case '8': case '9': case '.':
7077 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
7080 /* read next group of digits and _ and copy into d */
7081 while (isDIGIT(*s) || *s == '_') {
7082 /* skip underscores, checking for misplaced ones
7086 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
7087 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
7091 /* check for end of fixed-length buffer */
7093 Perl_croak(aTHX_ number_too_long);
7094 /* if we're ok, copy the character */
7099 /* final misplaced underbar check */
7100 if (lastub && s - lastub != 3) {
7101 if (ckWARN(WARN_SYNTAX))
7102 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
7105 /* read a decimal portion if there is one. avoid
7106 3..5 being interpreted as the number 3. followed
7109 if (*s == '.' && s[1] != '.') {
7113 /* copy, ignoring underbars, until we run out of
7114 digits. Note: no misplaced underbar checks!
7116 for (; isDIGIT(*s) || *s == '_'; s++) {
7117 /* fixed length buffer check */
7119 Perl_croak(aTHX_ number_too_long);
7123 if (*s == '.' && isDIGIT(s[1])) {
7124 /* oops, it's really a v-string, but without the "v" */
7130 /* read exponent part, if present */
7131 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
7135 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
7136 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
7138 /* allow positive or negative exponent */
7139 if (*s == '+' || *s == '-')
7142 /* read digits of exponent (no underbars :-) */
7143 while (isDIGIT(*s)) {
7145 Perl_croak(aTHX_ number_too_long);
7150 /* terminate the string */
7153 /* make an sv from the string */
7156 #if defined(Strtol) && defined(Strtoul)
7159 strtol/strtoll sets errno to ERANGE if the number is too big
7160 for an integer. We try to do an integer conversion first
7161 if no characters indicating "float" have been found.
7168 if (*PL_tokenbuf == '-')
7169 iv = Strtol(PL_tokenbuf, (char**)NULL, 10);
7171 uv = Strtoul(PL_tokenbuf, (char**)NULL, 10);
7173 floatit = TRUE; /* Probably just too large. */
7174 else if (*PL_tokenbuf == '-')
7176 else if (uv <= IV_MAX)
7177 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7182 nv = Atof(PL_tokenbuf);
7187 No working strtou?ll?.
7189 Unfortunately atol() doesn't do range checks (returning
7190 LONG_MIN/LONG_MAX, and setting errno to ERANGE on overflows)
7191 everywhere [1], so we cannot use use atol() (or atoll()).
7192 If we could, they would be used, as Atol(), very much like
7193 Strtol() and Strtoul() are used above.
7195 [1] XXX Configure test needed to check for atol()
7196 (and atoll()) overflow behaviour XXX
7200 We need to do this the hard way. */
7202 nv = Atof(PL_tokenbuf);
7204 /* See if we can make do with an integer value without loss of
7205 precision. We use U_V to cast to a UV, because some
7206 compilers have issues. Then we try casting it back and see
7207 if it was the same [1]. We only do this if we know we
7208 specifically read an integer. If floatit is true, then we
7209 don't need to do the conversion at all.
7211 [1] Note that this is lossy if our NVs cannot preserve our
7212 UVs. There are metaconfig defines NV_PRESERVES_UV (a boolean)
7213 and NV_PRESERVES_UV_BITS (a number), but in general we really
7214 do hope all such potentially lossy platforms have strtou?ll?
7215 to do a lossless IV/UV conversion.
7217 Maybe could do some tricks with DBL_DIG, LDBL_DIG and
7218 DBL_MANT_DIG and LDBL_MANT_DIG (these are already available
7219 as NV_DIG and NV_MANT_DIG)?
7225 if (!floatit && (NV)uv == nv) {
7227 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7235 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7236 (PL_hints & HINT_NEW_INTEGER) )
7237 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7238 (floatit ? "float" : "integer"),
7242 /* if it starts with a v, it could be a v-string */
7248 while (isDIGIT(*pos) || *pos == '_')
7250 if (!isALPHA(*pos)) {
7252 U8 tmpbuf[UTF8_MAXLEN+1];
7254 s++; /* get past 'v' */
7257 sv_setpvn(sv, "", 0);
7260 if (*s == '0' && isDIGIT(s[1]))
7261 yyerror("Octal number in vector unsupported");
7264 /* this is atoi() that tolerates underscores */
7267 while (--end >= s) {
7272 rev += (*end - '0') * mult;
7274 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
7275 Perl_warner(aTHX_ WARN_OVERFLOW,
7276 "Integer overflow in decimal number");
7279 tmpend = uv_to_utf8(tmpbuf, rev);
7282 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
7283 if (*pos == '.' && isDIGIT(pos[1]))
7289 while (isDIGIT(*pos) || *pos == '_')
7298 sv_utf8_downgrade(sv, TRUE);
7305 /* make the op for the constant and return */
7308 lvalp->opval = newSVOP(OP_CONST, 0, sv);
7310 lvalp->opval = Nullop;
7316 S_scan_formline(pTHX_ register char *s)
7320 SV *stuff = newSVpvn("",0);
7321 bool needargs = FALSE;
7324 if (*s == '.' || *s == /*{*/'}') {
7326 #ifdef PERL_STRICT_CR
7327 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7329 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7331 if (*t == '\n' || t == PL_bufend)
7334 if (PL_in_eval && !PL_rsfp) {
7335 eol = strchr(s,'\n');
7340 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7342 for (t = s; t < eol; t++) {
7343 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7345 goto enough; /* ~~ must be first line in formline */
7347 if (*t == '@' || *t == '^')
7350 sv_catpvn(stuff, s, eol-s);
7351 #ifndef PERL_STRICT_CR
7352 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7353 char *end = SvPVX(stuff) + SvCUR(stuff);
7362 s = filter_gets(PL_linestr, PL_rsfp, 0);
7363 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7364 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7367 yyerror("Format not terminated");
7377 PL_lex_state = LEX_NORMAL;
7378 PL_nextval[PL_nexttoke].ival = 0;
7382 PL_lex_state = LEX_FORMLINE;
7383 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7385 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7389 SvREFCNT_dec(stuff);
7390 PL_lex_formbrack = 0;
7401 PL_cshlen = strlen(PL_cshname);
7406 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7408 I32 oldsavestack_ix = PL_savestack_ix;
7409 CV* outsidecv = PL_compcv;
7413 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7415 SAVEI32(PL_subline);
7416 save_item(PL_subname);
7419 SAVESPTR(PL_comppad_name);
7420 SAVESPTR(PL_compcv);
7421 SAVEI32(PL_comppad_name_fill);
7422 SAVEI32(PL_min_intro_pending);
7423 SAVEI32(PL_max_intro_pending);
7424 SAVEI32(PL_pad_reset_pending);
7426 PL_compcv = (CV*)NEWSV(1104,0);
7427 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7428 CvFLAGS(PL_compcv) |= flags;
7430 PL_comppad = newAV();
7431 av_push(PL_comppad, Nullsv);
7432 PL_curpad = AvARRAY(PL_comppad);
7433 PL_comppad_name = newAV();
7434 PL_comppad_name_fill = 0;
7435 PL_min_intro_pending = 0;
7437 PL_subline = CopLINE(PL_curcop);
7439 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7440 PL_curpad[0] = (SV*)newAV();
7441 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7442 #endif /* USE_THREADS */
7444 comppadlist = newAV();
7445 AvREAL_off(comppadlist);
7446 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7447 av_store(comppadlist, 1, (SV*)PL_comppad);
7449 CvPADLIST(PL_compcv) = comppadlist;
7450 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7452 CvOWNER(PL_compcv) = 0;
7453 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7454 MUTEX_INIT(CvMUTEXP(PL_compcv));
7455 #endif /* USE_THREADS */
7457 return oldsavestack_ix;
7461 #pragma segment Perl_yylex
7464 Perl_yywarn(pTHX_ char *s)
7466 PL_in_eval |= EVAL_WARNONLY;
7468 PL_in_eval &= ~EVAL_WARNONLY;
7473 Perl_yyerror(pTHX_ char *s)
7476 char *context = NULL;
7480 if (!yychar || (yychar == ';' && !PL_rsfp))
7482 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7483 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7484 while (isSPACE(*PL_oldoldbufptr))
7486 context = PL_oldoldbufptr;
7487 contlen = PL_bufptr - PL_oldoldbufptr;
7489 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7490 PL_oldbufptr != PL_bufptr) {
7491 while (isSPACE(*PL_oldbufptr))
7493 context = PL_oldbufptr;
7494 contlen = PL_bufptr - PL_oldbufptr;
7496 else if (yychar > 255)
7497 where = "next token ???";
7498 #ifdef USE_PURE_BISON
7499 /* GNU Bison sets the value -2 */
7500 else if (yychar == -2) {
7502 else if ((yychar & 127) == 127) {
7504 if (PL_lex_state == LEX_NORMAL ||
7505 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7506 where = "at end of line";
7507 else if (PL_lex_inpat)
7508 where = "within pattern";
7510 where = "within string";
7513 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7515 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7516 else if (isPRINT_LC(yychar))
7517 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7519 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7520 where = SvPVX(where_sv);
7522 msg = sv_2mortal(newSVpv(s, 0));
7523 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7524 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7526 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7528 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7529 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7530 Perl_sv_catpvf(aTHX_ msg,
7531 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7532 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7535 if (PL_in_eval & EVAL_WARNONLY)
7536 Perl_warn(aTHX_ "%"SVf, msg);
7539 if (PL_error_count >= 10) {
7540 if (PL_in_eval && SvCUR(ERRSV))
7541 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
7542 ERRSV, CopFILE(PL_curcop));
7544 Perl_croak(aTHX_ "%s has too many errors.\n",
7545 CopFILE(PL_curcop));
7548 PL_in_my_stash = Nullhv;
7552 #pragma segment Main
7556 S_swallow_bom(pTHX_ U8 *s)
7559 slen = SvCUR(PL_linestr);
7563 /* UTF-16 little-endian */
7564 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
7565 Perl_croak(aTHX_ "Unsupported script encoding");
7566 #ifndef PERL_NO_UTF16_FILTER
7567 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-LE script encoding\n"));
7569 if (PL_bufend > (char*)s) {
7573 filter_add(utf16rev_textfilter, NULL);
7574 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7575 PL_bufend = (char*)utf16_to_utf8_reversed(s, news,
7576 PL_bufend - (char*)s - 1,
7578 Copy(news, s, newlen, U8);
7579 SvCUR_set(PL_linestr, newlen);
7580 PL_bufend = SvPVX(PL_linestr) + newlen;
7581 news[newlen++] = '\0';
7585 Perl_croak(aTHX_ "Unsupported script encoding");
7590 if (s[1] == 0xFF) { /* UTF-16 big-endian */
7591 #ifndef PERL_NO_UTF16_FILTER
7592 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding\n"));
7594 if (PL_bufend > (char *)s) {
7598 filter_add(utf16_textfilter, NULL);
7599 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7600 PL_bufend = (char*)utf16_to_utf8(s, news,
7601 PL_bufend - (char*)s,
7603 Copy(news, s, newlen, U8);
7604 SvCUR_set(PL_linestr, newlen);
7605 PL_bufend = SvPVX(PL_linestr) + newlen;
7606 news[newlen++] = '\0';
7610 Perl_croak(aTHX_ "Unsupported script encoding");
7615 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
7616 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-8 script encoding\n"));
7621 if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */
7622 s[2] == 0xFE && s[3] == 0xFF)
7624 Perl_croak(aTHX_ "Unsupported script encoding");
7636 * Restore a source filter.
7640 restore_rsfp(pTHXo_ void *f)
7642 PerlIO *fp = (PerlIO*)f;
7644 if (PL_rsfp == PerlIO_stdin())
7645 PerlIO_clearerr(PL_rsfp);
7646 else if (PL_rsfp && (PL_rsfp != fp))
7647 PerlIO_close(PL_rsfp);
7651 #ifndef PERL_NO_UTF16_FILTER
7653 utf16_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7655 I32 count = FILTER_READ(idx+1, sv, maxlen);
7660 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7661 if (!*SvPV_nolen(sv))
7662 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7665 tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7666 sv_usepvn(sv, (char*)tmps, tend - tmps);
7672 utf16rev_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7674 I32 count = FILTER_READ(idx+1, sv, maxlen);
7679 if (!*SvPV_nolen(sv))
7680 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7683 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7684 tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7685 sv_usepvn(sv, (char*)tmps, tend - tmps);