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(pTHX_ void *f);
31 #ifndef PERL_NO_UTF16_FILTER
32 static I32 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen);
33 static I32 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen);
36 #define XFAKEBRACK 128
39 #ifdef USE_UTF8_SCRIPTS
40 # define UTF (!IN_BYTES)
42 # ifdef EBCDIC /* For now 'use utf8' does not affect tokenizer on EBCDIC */
43 # define UTF (PL_linestr && DO_UTF8(PL_linestr))
45 # define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
49 /* In variables named $^X, these are the legal values for X.
50 * 1999-02-27 mjd-perl-patch@plover.com */
51 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
53 /* On MacOS, respect nonbreaking spaces */
54 #ifdef MACOS_TRADITIONAL
55 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
57 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
60 /* LEX_* are values for PL_lex_state, the state of the lexer.
61 * They are arranged oddly so that the guard on the switch statement
62 * can get by with a single comparison (if the compiler is smart enough).
65 /* #define LEX_NOTPARSING 11 is done in perl.h. */
68 #define LEX_INTERPNORMAL 9
69 #define LEX_INTERPCASEMOD 8
70 #define LEX_INTERPPUSH 7
71 #define LEX_INTERPSTART 6
72 #define LEX_INTERPEND 5
73 #define LEX_INTERPENDMAYBE 4
74 #define LEX_INTERPCONCAT 3
75 #define LEX_INTERPCONST 2
76 #define LEX_FORMLINE 1
77 #define LEX_KNOWNEXT 0
85 # define YYMAXLEVEL 100
87 YYSTYPE* yylval_pointer[YYMAXLEVEL];
88 int* yychar_pointer[YYMAXLEVEL];
92 # define yylval (*yylval_pointer[yyactlevel])
93 # define yychar (*yychar_pointer[yyactlevel])
94 # define PERL_YYLEX_PARAM yylval_pointer[yyactlevel],yychar_pointer[yyactlevel]
96 # define yylex() Perl_yylex_r(aTHX_ yylval_pointer[yyactlevel],yychar_pointer[yyactlevel])
101 /* CLINE is a macro that ensures PL_copline has a sane value */
106 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
109 * Convenience functions to return different tokens and prime the
110 * lexer for the next token. They all take an argument.
112 * TOKEN : generic token (used for '(', DOLSHARP, etc)
113 * OPERATOR : generic operator
114 * AOPERATOR : assignment operator
115 * PREBLOCK : beginning the block after an if, while, foreach, ...
116 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
117 * PREREF : *EXPR where EXPR is not a simple identifier
118 * TERM : expression term
119 * LOOPX : loop exiting command (goto, last, dump, etc)
120 * FTST : file test operator
121 * FUN0 : zero-argument function
122 * FUN1 : not used, except for not, which isn't a UNIOP
123 * BOop : bitwise or or xor
125 * SHop : shift operator
126 * PWop : power operator
127 * PMop : pattern-matching operator
128 * Aop : addition-level operator
129 * Mop : multiplication-level operator
130 * Eop : equality-testing operator
131 * Rop : relational operator <= != gt
133 * Also see LOP and lop() below.
136 /* Note that REPORT() and REPORT2() will be expressions that supply
137 * their own trailing comma, not suitable for statements as such. */
138 #ifdef DEBUGGING /* Serve -DT. */
139 # define REPORT(x,retval) tokereport(x,s,(int)retval),
140 # define REPORT2(x,retval) tokereport(x,s, yylval.ival),
142 # define REPORT(x,retval)
143 # define REPORT2(x,retval)
146 #define TOKEN(retval) return (REPORT2("token",retval) PL_bufptr = s,(int)retval)
147 #define OPERATOR(retval) return (REPORT2("operator",retval) PL_expect = XTERM, PL_bufptr = s,(int)retval)
148 #define AOPERATOR(retval) return ao((REPORT2("aop",retval) PL_expect = XTERM, PL_bufptr = s,(int)retval))
149 #define PREBLOCK(retval) return (REPORT2("preblock",retval) PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
150 #define PRETERMBLOCK(retval) return (REPORT2("pretermblock",retval) PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
151 #define PREREF(retval) return (REPORT2("preref",retval) PL_expect = XREF,PL_bufptr = s,(int)retval)
152 #define TERM(retval) return (CLINE, REPORT2("term",retval) PL_expect = XOPERATOR, PL_bufptr = s,(int)retval)
153 #define LOOPX(f) return(yylval.ival=f, REPORT("loopx",f) PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
154 #define FTST(f) return(yylval.ival=f, REPORT("ftst",f) PL_expect = XTERM,PL_bufptr = s,(int)UNIOP)
155 #define FUN0(f) return(yylval.ival = f, REPORT("fun0",f) PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
156 #define FUN1(f) return(yylval.ival = f, REPORT("fun1",f) PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
157 #define BOop(f) return ao((yylval.ival=f, REPORT("bitorop",f) PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
158 #define BAop(f) return ao((yylval.ival=f, REPORT("bitandop",f) PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
159 #define SHop(f) return ao((yylval.ival=f, REPORT("shiftop",f) PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
160 #define PWop(f) return ao((yylval.ival=f, REPORT("powop",f) PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
161 #define PMop(f) return(yylval.ival=f, REPORT("matchop",f) PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
162 #define Aop(f) return ao((yylval.ival=f, REPORT("add",f) PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
163 #define Mop(f) return ao((yylval.ival=f, REPORT("mul",f) PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
164 #define Eop(f) return(yylval.ival=f, REPORT("eq",f) PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
165 #define Rop(f) return(yylval.ival=f, REPORT("rel",f) PL_expect = XTERM,PL_bufptr = s,(int)RELOP)
167 /* This bit of chicanery makes a unary function followed by
168 * a parenthesis into a function with one argument, highest precedence.
170 #define UNI(f) return(yylval.ival = f, \
174 PL_last_uni = PL_oldbufptr, \
175 PL_last_lop_op = f, \
176 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
178 #define UNIBRACK(f) return(yylval.ival = f, \
181 PL_last_uni = PL_oldbufptr, \
182 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
184 /* grandfather return to old style */
185 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
190 S_tokereport(pTHX_ char *thing, char* s, I32 rv)
193 SV* report = newSVpv(thing, 0);
194 Perl_sv_catpvf(aTHX_ report, ":line %d:%"IVdf":", CopLINE(PL_curcop),
197 if (s - PL_bufptr > 0)
198 sv_catpvn(report, PL_bufptr, s - PL_bufptr);
200 if (PL_oldbufptr && *PL_oldbufptr)
201 sv_catpv(report, PL_tokenbuf);
203 PerlIO_printf(Perl_debug_log, "### %s\n", SvPV_nolen(report));
212 * This subroutine detects &&= and ||= and turns an ANDAND or OROR
213 * into an OP_ANDASSIGN or OP_ORASSIGN
217 S_ao(pTHX_ int toketype)
219 if (*PL_bufptr == '=') {
221 if (toketype == ANDAND)
222 yylval.ival = OP_ANDASSIGN;
223 else if (toketype == OROR)
224 yylval.ival = OP_ORASSIGN;
232 * When Perl expects an operator and finds something else, no_op
233 * prints the warning. It always prints "<something> found where
234 * operator expected. It prints "Missing semicolon on previous line?"
235 * if the surprise occurs at the start of the line. "do you need to
236 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
237 * where the compiler doesn't know if foo is a method call or a function.
238 * It prints "Missing operator before end of line" if there's nothing
239 * after the missing operator, or "... before <...>" if there is something
240 * after the missing operator.
244 S_no_op(pTHX_ char *what, char *s)
246 char *oldbp = PL_bufptr;
247 bool is_first = (PL_oldbufptr == PL_linestart);
253 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
255 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
256 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
258 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
259 if (t < PL_bufptr && isSPACE(*t))
260 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
261 t - PL_oldoldbufptr, PL_oldoldbufptr);
265 Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
272 * Complain about missing quote/regexp/heredoc terminator.
273 * If it's called with (char *)NULL then it cauterizes the line buffer.
274 * If we're in a delimited string and the delimiter is a control
275 * character, it's reformatted into a two-char sequence like ^C.
280 S_missingterm(pTHX_ char *s)
285 char *nl = strrchr(s,'\n');
291 iscntrl(PL_multi_close)
293 PL_multi_close < 32 || PL_multi_close == 127
297 tmpbuf[1] = toCTRL(PL_multi_close);
303 *tmpbuf = PL_multi_close;
307 q = strchr(s,'"') ? '\'' : '"';
308 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
316 Perl_deprecate(pTHX_ char *s)
318 if (ckWARN(WARN_DEPRECATED))
319 Perl_warner(aTHX_ WARN_DEPRECATED, "Use of %s is deprecated", s);
324 * Deprecate a comma-less variable list.
330 deprecate("comma-less variable list");
334 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
335 * utf16-to-utf8-reversed.
338 #ifdef PERL_CR_FILTER
342 register char *s = SvPVX(sv);
343 register char *e = s + SvCUR(sv);
344 /* outer loop optimized to do nothing if there are no CR-LFs */
346 if (*s++ == '\r' && *s == '\n') {
347 /* hit a CR-LF, need to copy the rest */
348 register char *d = s - 1;
351 if (*s == '\r' && s[1] == '\n')
362 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
364 I32 count = FILTER_READ(idx+1, sv, maxlen);
365 if (count > 0 && !maxlen)
373 * Initialize variables. Uses the Perl save_stack to save its state (for
374 * recursive calls to the parser).
378 Perl_lex_start(pTHX_ SV *line)
383 SAVEI32(PL_lex_dojoin);
384 SAVEI32(PL_lex_brackets);
385 SAVEI32(PL_lex_casemods);
386 SAVEI32(PL_lex_starts);
387 SAVEI32(PL_lex_state);
388 SAVEVPTR(PL_lex_inpat);
389 SAVEI32(PL_lex_inwhat);
390 if (PL_lex_state == LEX_KNOWNEXT) {
391 I32 toke = PL_nexttoke;
392 while (--toke >= 0) {
393 SAVEI32(PL_nexttype[toke]);
394 SAVEVPTR(PL_nextval[toke]);
396 SAVEI32(PL_nexttoke);
398 SAVECOPLINE(PL_curcop);
401 SAVEPPTR(PL_oldbufptr);
402 SAVEPPTR(PL_oldoldbufptr);
403 SAVEPPTR(PL_last_lop);
404 SAVEPPTR(PL_last_uni);
405 SAVEPPTR(PL_linestart);
406 SAVESPTR(PL_linestr);
407 SAVEPPTR(PL_lex_brackstack);
408 SAVEPPTR(PL_lex_casestack);
409 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
410 SAVESPTR(PL_lex_stuff);
411 SAVEI32(PL_lex_defer);
412 SAVEI32(PL_sublex_info.sub_inwhat);
413 SAVESPTR(PL_lex_repl);
415 SAVEINT(PL_lex_expect);
417 PL_lex_state = LEX_NORMAL;
421 New(899, PL_lex_brackstack, 120, char);
422 New(899, PL_lex_casestack, 12, char);
423 SAVEFREEPV(PL_lex_brackstack);
424 SAVEFREEPV(PL_lex_casestack);
426 *PL_lex_casestack = '\0';
429 PL_lex_stuff = Nullsv;
430 PL_lex_repl = Nullsv;
434 PL_sublex_info.sub_inwhat = 0;
436 if (SvREADONLY(PL_linestr))
437 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
438 s = SvPV(PL_linestr, len);
439 if (len && s[len-1] != ';') {
440 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
441 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
442 sv_catpvn(PL_linestr, "\n;", 2);
444 SvTEMP_off(PL_linestr);
445 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
446 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
447 PL_last_lop = PL_last_uni = Nullch;
449 PL_rs = newSVpvn("\n", 1);
455 * Finalizer for lexing operations. Must be called when the parser is
456 * done with the lexer.
462 PL_doextract = FALSE;
467 * This subroutine has nothing to do with tilting, whether at windmills
468 * or pinball tables. Its name is short for "increment line". It
469 * increments the current line number in CopLINE(PL_curcop) and checks
470 * to see whether the line starts with a comment of the form
471 * # line 500 "foo.pm"
472 * If so, it sets the current line number and file to the values in the comment.
476 S_incline(pTHX_ char *s)
483 CopLINE_inc(PL_curcop);
486 while (SPACE_OR_TAB(*s)) s++;
487 if (strnEQ(s, "line", 4))
491 if (SPACE_OR_TAB(*s))
495 while (SPACE_OR_TAB(*s)) s++;
501 while (SPACE_OR_TAB(*s))
503 if (*s == '"' && (t = strchr(s+1, '"'))) {
508 for (t = s; !isSPACE(*t); t++) ;
511 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
513 if (*e != '\n' && *e != '\0')
514 return; /* false alarm */
520 Safefree(CopFILE(PL_curcop));
522 SvREFCNT_dec(CopFILEGV(PL_curcop));
524 CopFILE_set(PL_curcop, s);
527 CopLINE_set(PL_curcop, atoi(n)-1);
532 * Called to gobble the appropriate amount and type of whitespace.
533 * Skips comments as well.
537 S_skipspace(pTHX_ register char *s)
539 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
540 while (s < PL_bufend && SPACE_OR_TAB(*s))
546 SSize_t oldprevlen, oldoldprevlen;
547 SSize_t oldloplen = 0, oldunilen = 0;
548 while (s < PL_bufend && isSPACE(*s)) {
549 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
554 if (s < PL_bufend && *s == '#') {
555 while (s < PL_bufend && *s != '\n')
559 if (PL_in_eval && !PL_rsfp) {
566 /* only continue to recharge the buffer if we're at the end
567 * of the buffer, we're not reading from a source filter, and
568 * we're in normal lexing mode
570 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
571 PL_lex_state == LEX_FORMLINE)
574 /* try to recharge the buffer */
575 if ((s = filter_gets(PL_linestr, PL_rsfp,
576 (prevlen = SvCUR(PL_linestr)))) == Nullch)
578 /* end of file. Add on the -p or -n magic */
579 if (PL_minus_n || PL_minus_p) {
580 sv_setpv(PL_linestr,PL_minus_p ?
581 ";}continue{print or die qq(-p destination: $!\\n)" :
583 sv_catpv(PL_linestr,";}");
584 PL_minus_n = PL_minus_p = 0;
587 sv_setpv(PL_linestr,";");
589 /* reset variables for next time we lex */
590 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
592 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
593 PL_last_lop = PL_last_uni = Nullch;
595 /* Close the filehandle. Could be from -P preprocessor,
596 * STDIN, or a regular file. If we were reading code from
597 * STDIN (because the commandline held no -e or filename)
598 * then we don't close it, we reset it so the code can
599 * read from STDIN too.
602 if (PL_preprocess && !PL_in_eval)
603 (void)PerlProc_pclose(PL_rsfp);
604 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
605 PerlIO_clearerr(PL_rsfp);
607 (void)PerlIO_close(PL_rsfp);
612 /* not at end of file, so we only read another line */
613 /* make corresponding updates to old pointers, for yyerror() */
614 oldprevlen = PL_oldbufptr - PL_bufend;
615 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
617 oldunilen = PL_last_uni - PL_bufend;
619 oldloplen = PL_last_lop - PL_bufend;
620 PL_linestart = PL_bufptr = s + prevlen;
621 PL_bufend = s + SvCUR(PL_linestr);
623 PL_oldbufptr = s + oldprevlen;
624 PL_oldoldbufptr = s + oldoldprevlen;
626 PL_last_uni = s + oldunilen;
628 PL_last_lop = s + oldloplen;
631 /* debugger active and we're not compiling the debugger code,
632 * so store the line into the debugger's array of lines
634 if (PERLDB_LINE && PL_curstash != PL_debstash) {
635 SV *sv = NEWSV(85,0);
637 sv_upgrade(sv, SVt_PVMG);
638 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
639 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
646 * Check the unary operators to ensure there's no ambiguity in how they're
647 * used. An ambiguous piece of code would be:
649 * This doesn't mean rand() + 5. Because rand() is a unary operator,
650 * the +5 is its argument.
659 if (PL_oldoldbufptr != PL_last_uni)
661 while (isSPACE(*PL_last_uni))
663 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
664 if ((t = strchr(s, '(')) && t < PL_bufptr)
666 if (ckWARN_d(WARN_AMBIGUOUS)){
669 Perl_warner(aTHX_ WARN_AMBIGUOUS,
670 "Warning: Use of \"%s\" without parens is ambiguous",
676 /* workaround to replace the UNI() macro with a function. Only the
677 * hints/uts.sh file mentions this. Other comments elsewhere in the
678 * source indicate Microport Unix might need it too.
684 #define UNI(f) return uni(f,s)
687 S_uni(pTHX_ I32 f, char *s)
692 PL_last_uni = PL_oldbufptr;
703 #endif /* CRIPPLED_CC */
706 * LOP : macro to build a list operator. Its behaviour has been replaced
707 * with a subroutine, S_lop() for which LOP is just another name.
710 #define LOP(f,x) return lop(f,x,s)
714 * Build a list operator (or something that might be one). The rules:
715 * - if we have a next token, then it's a list operator [why?]
716 * - if the next thing is an opening paren, then it's a function
717 * - else it's a list operator
721 S_lop(pTHX_ I32 f, int x, char *s)
728 PL_last_lop = PL_oldbufptr;
743 * When the lexer realizes it knows the next token (for instance,
744 * it is reordering tokens for the parser) then it can call S_force_next
745 * to know what token to return the next time the lexer is called. Caller
746 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
747 * handles the token correctly.
751 S_force_next(pTHX_ I32 type)
753 PL_nexttype[PL_nexttoke] = type;
755 if (PL_lex_state != LEX_KNOWNEXT) {
756 PL_lex_defer = PL_lex_state;
757 PL_lex_expect = PL_expect;
758 PL_lex_state = LEX_KNOWNEXT;
764 * When the lexer knows the next thing is a word (for instance, it has
765 * just seen -> and it knows that the next char is a word char, then
766 * it calls S_force_word to stick the next word into the PL_next lookahead.
769 * char *start : buffer position (must be within PL_linestr)
770 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
771 * int check_keyword : if true, Perl checks to make sure the word isn't
772 * a keyword (do this if the word is a label, e.g. goto FOO)
773 * int allow_pack : if true, : characters will also be allowed (require,
775 * int allow_initial_tick : used by the "sub" lexer only.
779 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
784 start = skipspace(start);
786 if (isIDFIRST_lazy_if(s,UTF) ||
787 (allow_pack && *s == ':') ||
788 (allow_initial_tick && *s == '\'') )
790 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
791 if (check_keyword && keyword(PL_tokenbuf, len))
793 if (token == METHOD) {
798 PL_expect = XOPERATOR;
801 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
802 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
810 * Called when the lexer wants $foo *foo &foo etc, but the program
811 * text only contains the "foo" portion. The first argument is a pointer
812 * to the "foo", and the second argument is the type symbol to prefix.
813 * Forces the next token to be a "WORD".
814 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
818 S_force_ident(pTHX_ register char *s, int kind)
821 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
822 PL_nextval[PL_nexttoke].opval = o;
825 o->op_private = OPpCONST_ENTERED;
826 /* XXX see note in pp_entereval() for why we forgo typo
827 warnings if the symbol must be introduced in an eval.
829 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
830 kind == '$' ? SVt_PV :
831 kind == '@' ? SVt_PVAV :
832 kind == '%' ? SVt_PVHV :
840 Perl_str_to_version(pTHX_ SV *sv)
845 char *start = SvPVx(sv,len);
846 bool utf = SvUTF8(sv) ? TRUE : FALSE;
847 char *end = start + len;
848 while (start < end) {
852 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
857 retval += ((NV)n)/nshift;
866 * Forces the next token to be a version number.
867 * If the next token appears to be an invalid version number, (e.g. "v2b"),
868 * and if "guessing" is TRUE, then no new token is created (and the caller
869 * must use an alternative parsing method).
873 S_force_version(pTHX_ char *s, int guessing)
875 OP *version = Nullop;
884 while (isDIGIT(*d) || *d == '_' || *d == '.')
886 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
888 s = scan_num(s, &yylval);
889 version = yylval.opval;
890 ver = cSVOPx(version)->op_sv;
891 if (SvPOK(ver) && !SvNIOK(ver)) {
892 (void)SvUPGRADE(ver, SVt_PVNV);
893 SvNVX(ver) = str_to_version(ver);
894 SvNOK_on(ver); /* hint that it is a version */
901 /* NOTE: The parser sees the package name and the VERSION swapped */
902 PL_nextval[PL_nexttoke].opval = version;
910 * Tokenize a quoted string passed in as an SV. It finds the next
911 * chunk, up to end of string or a backslash. It may make a new
912 * SV containing that chunk (if HINT_NEW_STRING is on). It also
917 S_tokeq(pTHX_ SV *sv)
928 s = SvPV_force(sv, len);
929 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
932 while (s < send && *s != '\\')
937 if ( PL_hints & HINT_NEW_STRING ) {
938 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
944 if (s + 1 < send && (s[1] == '\\'))
945 s++; /* all that, just for this */
950 SvCUR_set(sv, d - SvPVX(sv));
952 if ( PL_hints & HINT_NEW_STRING )
953 return new_constant(NULL, 0, "q", sv, pv, "q");
958 * Now come three functions related to double-quote context,
959 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
960 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
961 * interact with PL_lex_state, and create fake ( ... ) argument lists
962 * to handle functions and concatenation.
963 * They assume that whoever calls them will be setting up a fake
964 * join call, because each subthing puts a ',' after it. This lets
967 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
969 * (I'm not sure whether the spurious commas at the end of lcfirst's
970 * arguments and join's arguments are created or not).
975 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
977 * Pattern matching will set PL_lex_op to the pattern-matching op to
978 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
980 * OP_CONST and OP_READLINE are easy--just make the new op and return.
982 * Everything else becomes a FUNC.
984 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
985 * had an OP_CONST or OP_READLINE). This just sets us up for a
986 * call to S_sublex_push().
992 register I32 op_type = yylval.ival;
994 if (op_type == OP_NULL) {
995 yylval.opval = PL_lex_op;
999 if (op_type == OP_CONST || op_type == OP_READLINE) {
1000 SV *sv = tokeq(PL_lex_stuff);
1002 if (SvTYPE(sv) == SVt_PVIV) {
1003 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
1009 nsv = newSVpvn(p, len);
1015 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
1016 PL_lex_stuff = Nullsv;
1020 PL_sublex_info.super_state = PL_lex_state;
1021 PL_sublex_info.sub_inwhat = op_type;
1022 PL_sublex_info.sub_op = PL_lex_op;
1023 PL_lex_state = LEX_INTERPPUSH;
1027 yylval.opval = PL_lex_op;
1037 * Create a new scope to save the lexing state. The scope will be
1038 * ended in S_sublex_done. Returns a '(', starting the function arguments
1039 * to the uc, lc, etc. found before.
1040 * Sets PL_lex_state to LEX_INTERPCONCAT.
1048 PL_lex_state = PL_sublex_info.super_state;
1049 SAVEI32(PL_lex_dojoin);
1050 SAVEI32(PL_lex_brackets);
1051 SAVEI32(PL_lex_casemods);
1052 SAVEI32(PL_lex_starts);
1053 SAVEI32(PL_lex_state);
1054 SAVEVPTR(PL_lex_inpat);
1055 SAVEI32(PL_lex_inwhat);
1056 SAVECOPLINE(PL_curcop);
1057 SAVEPPTR(PL_bufptr);
1058 SAVEPPTR(PL_bufend);
1059 SAVEPPTR(PL_oldbufptr);
1060 SAVEPPTR(PL_oldoldbufptr);
1061 SAVEPPTR(PL_last_lop);
1062 SAVEPPTR(PL_last_uni);
1063 SAVEPPTR(PL_linestart);
1064 SAVESPTR(PL_linestr);
1065 SAVEPPTR(PL_lex_brackstack);
1066 SAVEPPTR(PL_lex_casestack);
1068 PL_linestr = PL_lex_stuff;
1069 PL_lex_stuff = Nullsv;
1071 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1072 = SvPVX(PL_linestr);
1073 PL_bufend += SvCUR(PL_linestr);
1074 PL_last_lop = PL_last_uni = Nullch;
1075 SAVEFREESV(PL_linestr);
1077 PL_lex_dojoin = FALSE;
1078 PL_lex_brackets = 0;
1079 New(899, PL_lex_brackstack, 120, char);
1080 New(899, PL_lex_casestack, 12, char);
1081 SAVEFREEPV(PL_lex_brackstack);
1082 SAVEFREEPV(PL_lex_casestack);
1083 PL_lex_casemods = 0;
1084 *PL_lex_casestack = '\0';
1086 PL_lex_state = LEX_INTERPCONCAT;
1087 CopLINE_set(PL_curcop, PL_multi_start);
1089 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1090 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1091 PL_lex_inpat = PL_sublex_info.sub_op;
1093 PL_lex_inpat = Nullop;
1100 * Restores lexer state after a S_sublex_push.
1106 if (!PL_lex_starts++) {
1107 SV *sv = newSVpvn("",0);
1108 if (SvUTF8(PL_linestr))
1110 PL_expect = XOPERATOR;
1111 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1115 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1116 PL_lex_state = LEX_INTERPCASEMOD;
1120 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1121 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1122 PL_linestr = PL_lex_repl;
1124 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1125 PL_bufend += SvCUR(PL_linestr);
1126 PL_last_lop = PL_last_uni = Nullch;
1127 SAVEFREESV(PL_linestr);
1128 PL_lex_dojoin = FALSE;
1129 PL_lex_brackets = 0;
1130 PL_lex_casemods = 0;
1131 *PL_lex_casestack = '\0';
1133 if (SvEVALED(PL_lex_repl)) {
1134 PL_lex_state = LEX_INTERPNORMAL;
1136 /* we don't clear PL_lex_repl here, so that we can check later
1137 whether this is an evalled subst; that means we rely on the
1138 logic to ensure sublex_done() is called again only via the
1139 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1142 PL_lex_state = LEX_INTERPCONCAT;
1143 PL_lex_repl = Nullsv;
1149 PL_bufend = SvPVX(PL_linestr);
1150 PL_bufend += SvCUR(PL_linestr);
1151 PL_expect = XOPERATOR;
1152 PL_sublex_info.sub_inwhat = 0;
1160 Extracts a pattern, double-quoted string, or transliteration. This
1163 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1164 processing a pattern (PL_lex_inpat is true), a transliteration
1165 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1167 Returns a pointer to the character scanned up to. Iff this is
1168 advanced from the start pointer supplied (ie if anything was
1169 successfully parsed), will leave an OP for the substring scanned
1170 in yylval. Caller must intuit reason for not parsing further
1171 by looking at the next characters herself.
1175 double-quoted style: \r and \n
1176 regexp special ones: \D \s
1178 backrefs: \1 (deprecated in substitution replacements)
1179 case and quoting: \U \Q \E
1180 stops on @ and $, but not for $ as tail anchor
1182 In transliterations:
1183 characters are VERY literal, except for - not at the start or end
1184 of the string, which indicates a range. scan_const expands the
1185 range to the full set of intermediate characters.
1187 In double-quoted strings:
1189 double-quoted style: \r and \n
1191 backrefs: \1 (deprecated)
1192 case and quoting: \U \Q \E
1195 scan_const does *not* construct ops to handle interpolated strings.
1196 It stops processing as soon as it finds an embedded $ or @ variable
1197 and leaves it to the caller to work out what's going on.
1199 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1201 $ in pattern could be $foo or could be tail anchor. Assumption:
1202 it's a tail anchor if $ is the last thing in the string, or if it's
1203 followed by one of ")| \n\t"
1205 \1 (backreferences) are turned into $1
1207 The structure of the code is
1208 while (there's a character to process) {
1209 handle transliteration ranges
1210 skip regexp comments
1211 skip # initiated comments in //x patterns
1212 check for embedded @foo
1213 check for embedded scalars
1215 leave intact backslashes from leave (below)
1216 deprecate \1 in strings and sub replacements
1217 handle string-changing backslashes \l \U \Q \E, etc.
1218 switch (what was escaped) {
1219 handle - in a transliteration (becomes a literal -)
1220 handle \132 octal characters
1221 handle 0x15 hex characters
1222 handle \cV (control V)
1223 handle printf backslashes (\f, \r, \n, etc)
1225 } (end if backslash)
1226 } (end while character to read)
1231 S_scan_const(pTHX_ char *start)
1233 register char *send = PL_bufend; /* end of the constant */
1234 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1235 register char *s = start; /* start of the constant */
1236 register char *d = SvPVX(sv); /* destination for copies */
1237 bool dorange = FALSE; /* are we in a translit range? */
1238 bool didrange = FALSE; /* did we just finish a range? */
1239 I32 has_utf8 = FALSE; /* Output constant is UTF8 */
1240 I32 this_utf8 = UTF; /* The source string is assumed to be UTF8 */
1243 const char *leaveit = /* set of acceptably-backslashed characters */
1245 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1248 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1249 /* If we are doing a trans and we know we want UTF8 set expectation */
1250 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
1251 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1255 while (s < send || dorange) {
1256 /* get transliterations out of the way (they're most literal) */
1257 if (PL_lex_inwhat == OP_TRANS) {
1258 /* expand a range A-Z to the full set of characters. AIE! */
1260 I32 i; /* current expanded character */
1261 I32 min; /* first character in range */
1262 I32 max; /* last character in range */
1265 char *c = (char*)utf8_hop((U8*)d, -1);
1269 *c = (char)UTF_TO_NATIVE(0xff);
1270 /* mark the range as done, and continue */
1276 i = d - SvPVX(sv); /* remember current offset */
1277 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1278 d = SvPVX(sv) + i; /* refresh d after realloc */
1279 d -= 2; /* eat the first char and the - */
1281 min = (U8)*d; /* first char in range */
1282 max = (U8)d[1]; /* last char in range */
1286 "Invalid [] range \"%c-%c\" in transliteration operator",
1287 (char)min, (char)max);
1291 if ((isLOWER(min) && isLOWER(max)) ||
1292 (isUPPER(min) && isUPPER(max))) {
1294 for (i = min; i <= max; i++)
1296 *d++ = NATIVE_TO_NEED(has_utf8,i);
1298 for (i = min; i <= max; i++)
1300 *d++ = NATIVE_TO_NEED(has_utf8,i);
1305 for (i = min; i <= max; i++)
1308 /* mark the range as done, and continue */
1314 /* range begins (ignore - as first or last char) */
1315 else if (*s == '-' && s+1 < send && s != start) {
1317 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
1320 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
1332 /* if we get here, we're not doing a transliteration */
1334 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1335 except for the last char, which will be done separately. */
1336 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1338 while (s < send && *s != ')')
1339 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1341 else if (s[2] == '{' /* This should match regcomp.c */
1342 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1345 char *regparse = s + (s[2] == '{' ? 3 : 4);
1348 while (count && (c = *regparse)) {
1349 if (c == '\\' && regparse[1])
1357 if (*regparse != ')') {
1358 regparse--; /* Leave one char for continuation. */
1359 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1361 while (s < regparse)
1362 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1366 /* likewise skip #-initiated comments in //x patterns */
1367 else if (*s == '#' && PL_lex_inpat &&
1368 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1369 while (s+1 < send && *s != '\n')
1370 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1373 /* check for embedded arrays
1374 (@foo, @:foo, @'foo, @{foo}, @$foo, @+, @-)
1376 else if (*s == '@' && s[1]
1377 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1])))
1380 /* check for embedded scalars. only stop if we're sure it's a
1383 else if (*s == '$') {
1384 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1386 if (s + 1 < send && !strchr("()| \r\n\t", s[1]))
1387 break; /* in regexp, $ might be tail anchor */
1390 /* End of else if chain - OP_TRANS rejoin rest */
1393 if (*s == '\\' && s+1 < send) {
1396 /* some backslashes we leave behind */
1397 if (*leaveit && *s && strchr(leaveit, *s)) {
1398 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
1399 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1403 /* deprecate \1 in strings and substitution replacements */
1404 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1405 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1407 if (ckWARN(WARN_SYNTAX))
1408 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1413 /* string-change backslash escapes */
1414 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1419 /* if we get here, it's either a quoted -, or a digit */
1422 /* quoted - in transliterations */
1424 if (PL_lex_inwhat == OP_TRANS) {
1431 if (ckWARN(WARN_MISC) && isALNUM(*s))
1432 Perl_warner(aTHX_ WARN_MISC,
1433 "Unrecognized escape \\%c passed through",
1435 /* default action is to copy the quoted character */
1436 goto default_action;
1439 /* \132 indicates an octal constant */
1440 case '0': case '1': case '2': case '3':
1441 case '4': case '5': case '6': case '7':
1445 uv = grok_oct(s, &len, &flags, NULL);
1448 goto NUM_ESCAPE_INSERT;
1450 /* \x24 indicates a hex constant */
1454 char* e = strchr(s, '}');
1455 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1456 PERL_SCAN_DISALLOW_PREFIX;
1461 yyerror("Missing right brace on \\x{}");
1465 uv = grok_hex(s, &len, &flags, NULL);
1471 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
1472 uv = grok_hex(s, &len, &flags, NULL);
1478 /* Insert oct or hex escaped character.
1479 * There will always enough room in sv since such
1480 * escapes will be longer than any UTF-8 sequence
1481 * they can end up as. */
1483 /* We need to map to chars to ASCII before doing the tests
1486 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(uv))) {
1487 if (!has_utf8 && uv > 255) {
1488 /* Might need to recode whatever we have
1489 * accumulated so far if it contains any
1492 * (Can't we keep track of that and avoid
1493 * this rescan? --jhi)
1497 for (c = (U8 *) SvPVX(sv); c < (U8 *)d; c++) {
1498 if (!NATIVE_IS_INVARIANT(*c)) {
1503 STRLEN offset = d - SvPVX(sv);
1505 d = SvGROW(sv, SvLEN(sv) + hicount + 1) + offset;
1509 while (src >= (U8 *)SvPVX(sv)) {
1510 if (!NATIVE_IS_INVARIANT(*src)) {
1511 U8 ch = NATIVE_TO_ASCII(*src);
1512 *dst-- = UTF8_EIGHT_BIT_LO(ch);
1513 *dst-- = UTF8_EIGHT_BIT_HI(ch);
1523 if (has_utf8 || uv > 255) {
1524 d = (char*)uvchr_to_utf8((U8*)d, uv);
1526 if (PL_lex_inwhat == OP_TRANS &&
1527 PL_sublex_info.sub_op) {
1528 PL_sublex_info.sub_op->op_private |=
1529 (PL_lex_repl ? OPpTRANS_FROM_UTF
1542 /* \N{latin small letter a} is a named character */
1546 char* e = strchr(s, '}');
1552 yyerror("Missing right brace on \\N{}");
1556 res = newSVpvn(s + 1, e - s - 1);
1557 res = new_constant( Nullch, 0, "charnames",
1558 res, Nullsv, "\\N{...}" );
1560 sv_utf8_upgrade(res);
1561 str = SvPV(res,len);
1562 if (!has_utf8 && SvUTF8(res)) {
1563 char *ostart = SvPVX(sv);
1564 SvCUR_set(sv, d - ostart);
1567 sv_utf8_upgrade(sv);
1568 /* this just broke our allocation above... */
1569 SvGROW(sv, send - start);
1570 d = SvPVX(sv) + SvCUR(sv);
1573 if (len > e - s + 4) {
1574 char *odest = SvPVX(sv);
1576 SvGROW(sv, (SvLEN(sv) + len - (e - s + 4)));
1577 d = SvPVX(sv) + (d - odest);
1579 Copy(str, d, len, char);
1586 yyerror("Missing braces on \\N{}");
1589 /* \c is a control character */
1598 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
1602 /* printf-style backslashes, formfeeds, newlines, etc */
1604 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
1607 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
1610 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
1613 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
1616 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
1619 *d++ = ASCII_TO_NEED(has_utf8,'\033');
1622 *d++ = ASCII_TO_NEED(has_utf8,'\007');
1628 } /* end if (backslash) */
1631 /* If we started with encoded form, or already know we want it
1632 and then encode the next character */
1633 if ((has_utf8 || this_utf8) && !NATIVE_IS_INVARIANT((U8)(*s))) {
1635 UV uv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
1636 STRLEN need = UNISKIP(NATIVE_TO_UNI(uv));
1639 /* encoded value larger than old, need extra space (NOTE: SvCUR() not set here) */
1640 STRLEN off = d - SvPVX(sv);
1641 d = SvGROW(sv, SvLEN(sv) + (need-len)) + off;
1643 d = (char*)uvchr_to_utf8((U8*)d, uv);
1647 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1649 } /* while loop to process each character */
1651 /* terminate the string and set up the sv */
1653 SvCUR_set(sv, d - SvPVX(sv));
1654 if (SvCUR(sv) >= SvLEN(sv))
1655 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
1660 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1661 PL_sublex_info.sub_op->op_private |=
1662 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1666 /* shrink the sv if we allocated more than we used */
1667 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1668 SvLEN_set(sv, SvCUR(sv) + 1);
1669 Renew(SvPVX(sv), SvLEN(sv), char);
1672 /* return the substring (via yylval) only if we parsed anything */
1673 if (s > PL_bufptr) {
1674 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1675 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1677 ( PL_lex_inwhat == OP_TRANS
1679 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1682 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1689 * Returns TRUE if there's more to the expression (e.g., a subscript),
1692 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1694 * ->[ and ->{ return TRUE
1695 * { and [ outside a pattern are always subscripts, so return TRUE
1696 * if we're outside a pattern and it's not { or [, then return FALSE
1697 * if we're in a pattern and the first char is a {
1698 * {4,5} (any digits around the comma) returns FALSE
1699 * if we're in a pattern and the first char is a [
1701 * [SOMETHING] has a funky algorithm to decide whether it's a
1702 * character class or not. It has to deal with things like
1703 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1704 * anything else returns TRUE
1707 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1710 S_intuit_more(pTHX_ register char *s)
1712 if (PL_lex_brackets)
1714 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1716 if (*s != '{' && *s != '[')
1721 /* In a pattern, so maybe we have {n,m}. */
1738 /* On the other hand, maybe we have a character class */
1741 if (*s == ']' || *s == '^')
1744 /* this is terrifying, and it works */
1745 int weight = 2; /* let's weigh the evidence */
1747 unsigned char un_char = 255, last_un_char;
1748 char *send = strchr(s,']');
1749 char tmpbuf[sizeof PL_tokenbuf * 4];
1751 if (!send) /* has to be an expression */
1754 Zero(seen,256,char);
1757 else if (isDIGIT(*s)) {
1759 if (isDIGIT(s[1]) && s[2] == ']')
1765 for (; s < send; s++) {
1766 last_un_char = un_char;
1767 un_char = (unsigned char)*s;
1772 weight -= seen[un_char] * 10;
1773 if (isALNUM_lazy_if(s+1,UTF)) {
1774 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1775 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1780 else if (*s == '$' && s[1] &&
1781 strchr("[#!%*<>()-=",s[1])) {
1782 if (/*{*/ strchr("])} =",s[2]))
1791 if (strchr("wds]",s[1]))
1793 else if (seen['\''] || seen['"'])
1795 else if (strchr("rnftbxcav",s[1]))
1797 else if (isDIGIT(s[1])) {
1799 while (s[1] && isDIGIT(s[1]))
1809 if (strchr("aA01! ",last_un_char))
1811 if (strchr("zZ79~",s[1]))
1813 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1814 weight -= 5; /* cope with negative subscript */
1817 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1818 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1823 if (keyword(tmpbuf, d - tmpbuf))
1826 if (un_char == last_un_char + 1)
1828 weight -= seen[un_char];
1833 if (weight >= 0) /* probably a character class */
1843 * Does all the checking to disambiguate
1845 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1846 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1848 * First argument is the stuff after the first token, e.g. "bar".
1850 * Not a method if bar is a filehandle.
1851 * Not a method if foo is a subroutine prototyped to take a filehandle.
1852 * Not a method if it's really "Foo $bar"
1853 * Method if it's "foo $bar"
1854 * Not a method if it's really "print foo $bar"
1855 * Method if it's really "foo package::" (interpreted as package->foo)
1856 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1857 * Not a method if bar is a filehandle or package, but is quoted with
1862 S_intuit_method(pTHX_ char *start, GV *gv)
1864 char *s = start + (*start == '$');
1865 char tmpbuf[sizeof PL_tokenbuf];
1873 if ((cv = GvCVu(gv))) {
1874 char *proto = SvPVX(cv);
1884 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1885 /* start is the beginning of the possible filehandle/object,
1886 * and s is the end of it
1887 * tmpbuf is a copy of it
1890 if (*start == '$') {
1891 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1896 return *s == '(' ? FUNCMETH : METHOD;
1898 if (!keyword(tmpbuf, len)) {
1899 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1904 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1905 if (indirgv && GvCVu(indirgv))
1907 /* filehandle or package name makes it a method */
1908 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1910 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1911 return 0; /* no assumptions -- "=>" quotes bearword */
1913 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1914 newSVpvn(tmpbuf,len));
1915 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1919 return *s == '(' ? FUNCMETH : METHOD;
1927 * Return a string of Perl code to load the debugger. If PERL5DB
1928 * is set, it will return the contents of that, otherwise a
1929 * compile-time require of perl5db.pl.
1936 char *pdb = PerlEnv_getenv("PERL5DB");
1940 SETERRNO(0,SS$_NORMAL);
1941 return "BEGIN { require 'perl5db.pl' }";
1947 /* Encoded script support. filter_add() effectively inserts a
1948 * 'pre-processing' function into the current source input stream.
1949 * Note that the filter function only applies to the current source file
1950 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1952 * The datasv parameter (which may be NULL) can be used to pass
1953 * private data to this instance of the filter. The filter function
1954 * can recover the SV using the FILTER_DATA macro and use it to
1955 * store private buffers and state information.
1957 * The supplied datasv parameter is upgraded to a PVIO type
1958 * and the IoDIRP/IoANY field is used to store the function pointer,
1959 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1960 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1961 * private use must be set using malloc'd pointers.
1965 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1970 if (!PL_rsfp_filters)
1971 PL_rsfp_filters = newAV();
1973 datasv = NEWSV(255,0);
1974 if (!SvUPGRADE(datasv, SVt_PVIO))
1975 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1976 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
1977 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1978 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1979 funcp, SvPV_nolen(datasv)));
1980 av_unshift(PL_rsfp_filters, 1);
1981 av_store(PL_rsfp_filters, 0, datasv) ;
1986 /* Delete most recently added instance of this filter function. */
1988 Perl_filter_del(pTHX_ filter_t funcp)
1991 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1992 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1994 /* if filter is on top of stack (usual case) just pop it off */
1995 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1996 if (IoANY(datasv) == (void *)funcp) {
1997 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1998 IoANY(datasv) = (void *)NULL;
1999 sv_free(av_pop(PL_rsfp_filters));
2003 /* we need to search for the correct entry and clear it */
2004 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
2008 /* Invoke the n'th filter function for the current rsfp. */
2010 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
2013 /* 0 = read one text line */
2018 if (!PL_rsfp_filters)
2020 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
2021 /* Provide a default input filter to make life easy. */
2022 /* Note that we append to the line. This is handy. */
2023 DEBUG_P(PerlIO_printf(Perl_debug_log,
2024 "filter_read %d: from rsfp\n", idx));
2028 int old_len = SvCUR(buf_sv) ;
2030 /* ensure buf_sv is large enough */
2031 SvGROW(buf_sv, old_len + maxlen) ;
2032 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
2033 if (PerlIO_error(PL_rsfp))
2034 return -1; /* error */
2036 return 0 ; /* end of file */
2038 SvCUR_set(buf_sv, old_len + len) ;
2041 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2042 if (PerlIO_error(PL_rsfp))
2043 return -1; /* error */
2045 return 0 ; /* end of file */
2048 return SvCUR(buf_sv);
2050 /* Skip this filter slot if filter has been deleted */
2051 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
2052 DEBUG_P(PerlIO_printf(Perl_debug_log,
2053 "filter_read %d: skipped (filter deleted)\n",
2055 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
2057 /* Get function pointer hidden within datasv */
2058 funcp = (filter_t)IoANY(datasv);
2059 DEBUG_P(PerlIO_printf(Perl_debug_log,
2060 "filter_read %d: via function %p (%s)\n",
2061 idx, funcp, SvPV_nolen(datasv)));
2062 /* Call function. The function is expected to */
2063 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2064 /* Return: <0:error, =0:eof, >0:not eof */
2065 return (*funcp)(aTHX_ idx, buf_sv, maxlen);
2069 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2071 #ifdef PERL_CR_FILTER
2072 if (!PL_rsfp_filters) {
2073 filter_add(S_cr_textfilter,NULL);
2076 if (PL_rsfp_filters) {
2079 SvCUR_set(sv, 0); /* start with empty line */
2080 if (FILTER_READ(0, sv, 0) > 0)
2081 return ( SvPVX(sv) ) ;
2086 return (sv_gets(sv, fp, append));
2090 S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2094 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2098 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2099 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2101 return GvHV(gv); /* Foo:: */
2104 /* use constant CLASS => 'MyClass' */
2105 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2107 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2108 pkgname = SvPV_nolen(sv);
2112 return gv_stashpv(pkgname, FALSE);
2116 static char* exp_name[] =
2117 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2118 "ATTRTERM", "TERMBLOCK"
2125 Works out what to call the token just pulled out of the input
2126 stream. The yacc parser takes care of taking the ops we return and
2127 stitching them into a tree.
2133 if read an identifier
2134 if we're in a my declaration
2135 croak if they tried to say my($foo::bar)
2136 build the ops for a my() declaration
2137 if it's an access to a my() variable
2138 are we in a sort block?
2139 croak if my($a); $a <=> $b
2140 build ops for access to a my() variable
2141 if in a dq string, and they've said @foo and we can't find @foo
2143 build ops for a bareword
2144 if we already built the token before, use it.
2147 #ifdef USE_PURE_BISON
2149 Perl_yylex_r(pTHX_ YYSTYPE *lvalp, int *lcharp)
2154 yylval_pointer[yyactlevel] = lvalp;
2155 yychar_pointer[yyactlevel] = lcharp;
2156 if (yyactlevel >= YYMAXLEVEL)
2157 Perl_croak(aTHX_ "panic: YYMAXLEVEL");
2159 r = Perl_yylex(aTHX);
2169 #pragma segment Perl_yylex
2182 /* check if there's an identifier for us to look at */
2183 if (PL_pending_ident)
2184 return S_pending_ident(aTHX);
2186 /* no identifier pending identification */
2188 switch (PL_lex_state) {
2190 case LEX_NORMAL: /* Some compilers will produce faster */
2191 case LEX_INTERPNORMAL: /* code if we comment these out. */
2195 /* when we've already built the next token, just pull it out of the queue */
2198 yylval = PL_nextval[PL_nexttoke];
2200 PL_lex_state = PL_lex_defer;
2201 PL_expect = PL_lex_expect;
2202 PL_lex_defer = LEX_NORMAL;
2204 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2205 "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr,
2206 (IV)PL_nexttype[PL_nexttoke]); });
2208 return(PL_nexttype[PL_nexttoke]);
2210 /* interpolated case modifiers like \L \U, including \Q and \E.
2211 when we get here, PL_bufptr is at the \
2213 case LEX_INTERPCASEMOD:
2215 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2216 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2218 /* handle \E or end of string */
2219 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2223 if (PL_lex_casemods) {
2224 oldmod = PL_lex_casestack[--PL_lex_casemods];
2225 PL_lex_casestack[PL_lex_casemods] = '\0';
2227 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2229 PL_lex_state = LEX_INTERPCONCAT;
2233 if (PL_bufptr != PL_bufend)
2235 PL_lex_state = LEX_INTERPCONCAT;
2239 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2240 "### Saw case modifier at '%s'\n", PL_bufptr); });
2242 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2243 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2244 if (strchr("LU", *s) &&
2245 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2247 PL_lex_casestack[--PL_lex_casemods] = '\0';
2250 if (PL_lex_casemods > 10) {
2251 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2252 if (newlb != PL_lex_casestack) {
2254 PL_lex_casestack = newlb;
2257 PL_lex_casestack[PL_lex_casemods++] = *s;
2258 PL_lex_casestack[PL_lex_casemods] = '\0';
2259 PL_lex_state = LEX_INTERPCONCAT;
2260 PL_nextval[PL_nexttoke].ival = 0;
2263 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2265 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2267 PL_nextval[PL_nexttoke].ival = OP_LC;
2269 PL_nextval[PL_nexttoke].ival = OP_UC;
2271 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2273 Perl_croak(aTHX_ "panic: yylex");
2276 if (PL_lex_starts) {
2285 case LEX_INTERPPUSH:
2286 return sublex_push();
2288 case LEX_INTERPSTART:
2289 if (PL_bufptr == PL_bufend)
2290 return sublex_done();
2291 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2292 "### Interpolated variable at '%s'\n", PL_bufptr); });
2294 PL_lex_dojoin = (*PL_bufptr == '@');
2295 PL_lex_state = LEX_INTERPNORMAL;
2296 if (PL_lex_dojoin) {
2297 PL_nextval[PL_nexttoke].ival = 0;
2299 #ifdef USE_5005THREADS
2300 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2301 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2302 force_next(PRIVATEREF);
2304 force_ident("\"", '$');
2305 #endif /* USE_5005THREADS */
2306 PL_nextval[PL_nexttoke].ival = 0;
2308 PL_nextval[PL_nexttoke].ival = 0;
2310 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2313 if (PL_lex_starts++) {
2319 case LEX_INTERPENDMAYBE:
2320 if (intuit_more(PL_bufptr)) {
2321 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2327 if (PL_lex_dojoin) {
2328 PL_lex_dojoin = FALSE;
2329 PL_lex_state = LEX_INTERPCONCAT;
2332 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2333 && SvEVALED(PL_lex_repl))
2335 if (PL_bufptr != PL_bufend)
2336 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2337 PL_lex_repl = Nullsv;
2340 case LEX_INTERPCONCAT:
2342 if (PL_lex_brackets)
2343 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2345 if (PL_bufptr == PL_bufend)
2346 return sublex_done();
2348 if (SvIVX(PL_linestr) == '\'') {
2349 SV *sv = newSVsv(PL_linestr);
2352 else if ( PL_hints & HINT_NEW_RE )
2353 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2354 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2358 s = scan_const(PL_bufptr);
2360 PL_lex_state = LEX_INTERPCASEMOD;
2362 PL_lex_state = LEX_INTERPSTART;
2365 if (s != PL_bufptr) {
2366 PL_nextval[PL_nexttoke] = yylval;
2369 if (PL_lex_starts++)
2379 PL_lex_state = LEX_NORMAL;
2380 s = scan_formline(PL_bufptr);
2381 if (!PL_lex_formbrack)
2387 PL_oldoldbufptr = PL_oldbufptr;
2390 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2391 exp_name[PL_expect], s);
2397 if (isIDFIRST_lazy_if(s,UTF))
2399 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2402 goto fake_eof; /* emulate EOF on ^D or ^Z */
2407 if (PL_lex_brackets)
2408 yyerror("Missing right curly or square bracket");
2409 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2410 "### Tokener got EOF\n");
2414 if (s++ < PL_bufend)
2415 goto retry; /* ignore stray nulls */
2418 if (!PL_in_eval && !PL_preambled) {
2419 PL_preambled = TRUE;
2420 sv_setpv(PL_linestr,incl_perldb());
2421 if (SvCUR(PL_linestr))
2422 sv_catpv(PL_linestr,";");
2424 while(AvFILLp(PL_preambleav) >= 0) {
2425 SV *tmpsv = av_shift(PL_preambleav);
2426 sv_catsv(PL_linestr, tmpsv);
2427 sv_catpv(PL_linestr, ";");
2430 sv_free((SV*)PL_preambleav);
2431 PL_preambleav = NULL;
2433 if (PL_minus_n || PL_minus_p) {
2434 sv_catpv(PL_linestr, "LINE: while (<>) {");
2436 sv_catpv(PL_linestr,"chomp;");
2439 if (strchr("/'\"", *PL_splitstr)
2440 && strchr(PL_splitstr + 1, *PL_splitstr))
2441 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2444 s = "'~#\200\1'"; /* surely one char is unused...*/
2445 while (s[1] && strchr(PL_splitstr, *s)) s++;
2447 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s%c",
2448 "q" + (delim == '\''), delim);
2449 for (s = PL_splitstr; *s; s++) {
2451 sv_catpvn(PL_linestr, "\\", 1);
2452 sv_catpvn(PL_linestr, s, 1);
2454 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2458 sv_catpv(PL_linestr,"our @F=split(' ');");
2461 sv_catpv(PL_linestr, "\n");
2462 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2463 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2464 PL_last_lop = PL_last_uni = Nullch;
2465 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2466 SV *sv = NEWSV(85,0);
2468 sv_upgrade(sv, SVt_PVMG);
2469 sv_setsv(sv,PL_linestr);
2470 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2475 bof = PL_rsfp ? TRUE : FALSE;
2476 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2479 if (PL_preprocess && !PL_in_eval)
2480 (void)PerlProc_pclose(PL_rsfp);
2481 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2482 PerlIO_clearerr(PL_rsfp);
2484 (void)PerlIO_close(PL_rsfp);
2486 PL_doextract = FALSE;
2488 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2489 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2490 sv_catpv(PL_linestr,";}");
2491 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2492 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2493 PL_last_lop = PL_last_uni = Nullch;
2494 PL_minus_n = PL_minus_p = 0;
2497 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2498 PL_last_lop = PL_last_uni = Nullch;
2499 sv_setpv(PL_linestr,"");
2500 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2502 /* if it looks like the start of a BOM, check if it in fact is */
2503 else if (bof && (!*s || *(U8*)s == 0xEF || *(U8*)s >= 0xFE)) {
2504 #ifdef PERLIO_IS_STDIO
2505 # ifdef __GNU_LIBRARY__
2506 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
2507 # define FTELL_FOR_PIPE_IS_BROKEN
2511 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2512 # define FTELL_FOR_PIPE_IS_BROKEN
2517 #ifdef FTELL_FOR_PIPE_IS_BROKEN
2518 /* This loses the possibility to detect the bof
2519 * situation on perl -P when the libc5 is being used.
2520 * Workaround? Maybe attach some extra state to PL_rsfp?
2523 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2525 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2528 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2529 s = swallow_bom((U8*)s);
2533 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2534 PL_doextract = FALSE;
2536 /* Incest with pod. */
2537 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2538 sv_setpv(PL_linestr, "");
2539 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2540 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2541 PL_last_lop = PL_last_uni = Nullch;
2542 PL_doextract = FALSE;
2546 } while (PL_doextract);
2547 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2548 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2549 SV *sv = NEWSV(85,0);
2551 sv_upgrade(sv, SVt_PVMG);
2552 sv_setsv(sv,PL_linestr);
2553 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2555 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2556 PL_last_lop = PL_last_uni = Nullch;
2557 if (CopLINE(PL_curcop) == 1) {
2558 while (s < PL_bufend && isSPACE(*s))
2560 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2564 if (*s == '#' && *(s+1) == '!')
2566 #ifdef ALTERNATE_SHEBANG
2568 static char as[] = ALTERNATE_SHEBANG;
2569 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2570 d = s + (sizeof(as) - 1);
2572 #endif /* ALTERNATE_SHEBANG */
2581 while (*d && !isSPACE(*d))
2585 #ifdef ARG_ZERO_IS_SCRIPT
2586 if (ipathend > ipath) {
2588 * HP-UX (at least) sets argv[0] to the script name,
2589 * which makes $^X incorrect. And Digital UNIX and Linux,
2590 * at least, set argv[0] to the basename of the Perl
2591 * interpreter. So, having found "#!", we'll set it right.
2593 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2594 assert(SvPOK(x) || SvGMAGICAL(x));
2595 if (sv_eq(x, CopFILESV(PL_curcop))) {
2596 sv_setpvn(x, ipath, ipathend - ipath);
2599 TAINT_NOT; /* $^X is always tainted, but that's OK */
2601 #endif /* ARG_ZERO_IS_SCRIPT */
2606 d = instr(s,"perl -");
2608 d = instr(s,"perl");
2610 /* avoid getting into infinite loops when shebang
2611 * line contains "Perl" rather than "perl" */
2613 for (d = ipathend-4; d >= ipath; --d) {
2614 if ((*d == 'p' || *d == 'P')
2615 && !ibcmp(d, "perl", 4))
2625 #ifdef ALTERNATE_SHEBANG
2627 * If the ALTERNATE_SHEBANG on this system starts with a
2628 * character that can be part of a Perl expression, then if
2629 * we see it but not "perl", we're probably looking at the
2630 * start of Perl code, not a request to hand off to some
2631 * other interpreter. Similarly, if "perl" is there, but
2632 * not in the first 'word' of the line, we assume the line
2633 * contains the start of the Perl program.
2635 if (d && *s != '#') {
2637 while (*c && !strchr("; \t\r\n\f\v#", *c))
2640 d = Nullch; /* "perl" not in first word; ignore */
2642 *s = '#'; /* Don't try to parse shebang line */
2644 #endif /* ALTERNATE_SHEBANG */
2645 #ifndef MACOS_TRADITIONAL
2650 !instr(s,"indir") &&
2651 instr(PL_origargv[0],"perl"))
2657 while (s < PL_bufend && isSPACE(*s))
2659 if (s < PL_bufend) {
2660 Newz(899,newargv,PL_origargc+3,char*);
2662 while (s < PL_bufend && !isSPACE(*s))
2665 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2668 newargv = PL_origargv;
2670 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
2671 Perl_croak(aTHX_ "Can't exec %s", ipath);
2675 U32 oldpdb = PL_perldb;
2676 bool oldn = PL_minus_n;
2677 bool oldp = PL_minus_p;
2679 while (*d && !isSPACE(*d)) d++;
2680 while (SPACE_OR_TAB(*d)) d++;
2684 if (*d == 'M' || *d == 'm') {
2686 while (*d && !isSPACE(*d)) d++;
2687 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2690 d = moreswitches(d);
2692 if ((PERLDB_LINE && !oldpdb) ||
2693 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2694 /* if we have already added "LINE: while (<>) {",
2695 we must not do it again */
2697 sv_setpv(PL_linestr, "");
2698 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2699 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2700 PL_last_lop = PL_last_uni = Nullch;
2701 PL_preambled = FALSE;
2703 (void)gv_fetchfile(PL_origfilename);
2710 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2712 PL_lex_state = LEX_FORMLINE;
2717 #ifdef PERL_STRICT_CR
2718 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2720 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2722 case ' ': case '\t': case '\f': case 013:
2723 #ifdef MACOS_TRADITIONAL
2730 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2731 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2732 /* handle eval qq[#line 1 "foo"\n ...] */
2733 CopLINE_dec(PL_curcop);
2737 while (s < d && *s != '\n')
2741 else if (s > d) /* Found by Ilya: feed random input to Perl. */
2742 Perl_croak(aTHX_ "panic: input overflow");
2744 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2746 PL_lex_state = LEX_FORMLINE;
2756 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2763 while (s < PL_bufend && SPACE_OR_TAB(*s))
2766 if (strnEQ(s,"=>",2)) {
2767 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2768 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2769 "### Saw unary minus before =>, forcing word '%s'\n", s);
2771 OPERATOR('-'); /* unary minus */
2773 PL_last_uni = PL_oldbufptr;
2775 case 'r': ftst = OP_FTEREAD; break;
2776 case 'w': ftst = OP_FTEWRITE; break;
2777 case 'x': ftst = OP_FTEEXEC; break;
2778 case 'o': ftst = OP_FTEOWNED; break;
2779 case 'R': ftst = OP_FTRREAD; break;
2780 case 'W': ftst = OP_FTRWRITE; break;
2781 case 'X': ftst = OP_FTREXEC; break;
2782 case 'O': ftst = OP_FTROWNED; break;
2783 case 'e': ftst = OP_FTIS; break;
2784 case 'z': ftst = OP_FTZERO; break;
2785 case 's': ftst = OP_FTSIZE; break;
2786 case 'f': ftst = OP_FTFILE; break;
2787 case 'd': ftst = OP_FTDIR; break;
2788 case 'l': ftst = OP_FTLINK; break;
2789 case 'p': ftst = OP_FTPIPE; break;
2790 case 'S': ftst = OP_FTSOCK; break;
2791 case 'u': ftst = OP_FTSUID; break;
2792 case 'g': ftst = OP_FTSGID; break;
2793 case 'k': ftst = OP_FTSVTX; break;
2794 case 'b': ftst = OP_FTBLK; break;
2795 case 'c': ftst = OP_FTCHR; break;
2796 case 't': ftst = OP_FTTTY; break;
2797 case 'T': ftst = OP_FTTEXT; break;
2798 case 'B': ftst = OP_FTBINARY; break;
2799 case 'M': case 'A': case 'C':
2800 gv_fetchpv("\024",TRUE, SVt_PV);
2802 case 'M': ftst = OP_FTMTIME; break;
2803 case 'A': ftst = OP_FTATIME; break;
2804 case 'C': ftst = OP_FTCTIME; break;
2812 PL_last_lop_op = ftst;
2813 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2814 "### Saw file test %c\n", (int)ftst);
2819 /* Assume it was a minus followed by a one-letter named
2820 * subroutine call (or a -bareword), then. */
2821 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2822 "### %c looked like a file test but was not\n",
2831 if (PL_expect == XOPERATOR)
2836 else if (*s == '>') {
2839 if (isIDFIRST_lazy_if(s,UTF)) {
2840 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2848 if (PL_expect == XOPERATOR)
2851 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2853 OPERATOR('-'); /* unary minus */
2860 if (PL_expect == XOPERATOR)
2865 if (PL_expect == XOPERATOR)
2868 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2874 if (PL_expect != XOPERATOR) {
2875 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2876 PL_expect = XOPERATOR;
2877 force_ident(PL_tokenbuf, '*');
2890 if (PL_expect == XOPERATOR) {
2894 PL_tokenbuf[0] = '%';
2895 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2896 if (!PL_tokenbuf[1]) {
2898 yyerror("Final % should be \\% or %name");
2901 PL_pending_ident = '%';
2920 switch (PL_expect) {
2923 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2925 PL_bufptr = s; /* update in case we back off */
2931 PL_expect = XTERMBLOCK;
2935 while (isIDFIRST_lazy_if(s,UTF)) {
2936 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2937 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2938 if (tmp < 0) tmp = -tmp;
2953 d = scan_str(d,TRUE,TRUE);
2955 /* MUST advance bufptr here to avoid bogus
2956 "at end of line" context messages from yyerror().
2958 PL_bufptr = s + len;
2959 yyerror("Unterminated attribute parameter in attribute list");
2962 return 0; /* EOF indicator */
2966 SV *sv = newSVpvn(s, len);
2967 sv_catsv(sv, PL_lex_stuff);
2968 attrs = append_elem(OP_LIST, attrs,
2969 newSVOP(OP_CONST, 0, sv));
2970 SvREFCNT_dec(PL_lex_stuff);
2971 PL_lex_stuff = Nullsv;
2974 if (!PL_in_my && len == 6 && strnEQ(s, "lvalue", len))
2975 CvLVALUE_on(PL_compcv);
2976 else if (!PL_in_my && len == 6 && strnEQ(s, "locked", len))
2977 CvLOCKED_on(PL_compcv);
2978 else if (!PL_in_my && len == 6 && strnEQ(s, "method", len))
2979 CvMETHOD_on(PL_compcv);
2981 else if (PL_in_my == KEY_our && len == 6 && strnEQ(s, "unique", len))
2982 GvUNIQUE_on(cGVOPx_gv(yylval.opval));
2984 /* After we've set the flags, it could be argued that
2985 we don't need to do the attributes.pm-based setting
2986 process, and shouldn't bother appending recognized
2987 flags. To experiment with that, uncomment the
2988 following "else": */
2990 attrs = append_elem(OP_LIST, attrs,
2991 newSVOP(OP_CONST, 0,
2995 if (*s == ':' && s[1] != ':')
2998 break; /* require real whitespace or :'s */
3000 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3001 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
3002 char q = ((*s == '\'') ? '"' : '\'');
3003 /* If here for an expression, and parsed no attrs, back off. */
3004 if (tmp == '=' && !attrs) {
3008 /* MUST advance bufptr here to avoid bogus "at end of line"
3009 context messages from yyerror().
3013 yyerror("Unterminated attribute list");
3015 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
3023 PL_nextval[PL_nexttoke].opval = attrs;
3031 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3032 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3048 if (PL_lex_brackets <= 0)
3049 yyerror("Unmatched right square bracket");
3052 if (PL_lex_state == LEX_INTERPNORMAL) {
3053 if (PL_lex_brackets == 0) {
3054 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3055 PL_lex_state = LEX_INTERPEND;
3062 if (PL_lex_brackets > 100) {
3063 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
3064 if (newlb != PL_lex_brackstack) {
3066 PL_lex_brackstack = newlb;
3069 switch (PL_expect) {
3071 if (PL_lex_formbrack) {
3075 if (PL_oldoldbufptr == PL_last_lop)
3076 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3078 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3079 OPERATOR(HASHBRACK);
3081 while (s < PL_bufend && SPACE_OR_TAB(*s))
3084 PL_tokenbuf[0] = '\0';
3085 if (d < PL_bufend && *d == '-') {
3086 PL_tokenbuf[0] = '-';
3088 while (d < PL_bufend && SPACE_OR_TAB(*d))
3091 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3092 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3094 while (d < PL_bufend && SPACE_OR_TAB(*d))
3097 char minus = (PL_tokenbuf[0] == '-');
3098 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3106 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3111 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3116 if (PL_oldoldbufptr == PL_last_lop)
3117 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3119 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3122 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
3124 /* This hack is to get the ${} in the message. */
3126 yyerror("syntax error");
3129 OPERATOR(HASHBRACK);
3131 /* This hack serves to disambiguate a pair of curlies
3132 * as being a block or an anon hash. Normally, expectation
3133 * determines that, but in cases where we're not in a
3134 * position to expect anything in particular (like inside
3135 * eval"") we have to resolve the ambiguity. This code
3136 * covers the case where the first term in the curlies is a
3137 * quoted string. Most other cases need to be explicitly
3138 * disambiguated by prepending a `+' before the opening
3139 * curly in order to force resolution as an anon hash.
3141 * XXX should probably propagate the outer expectation
3142 * into eval"" to rely less on this hack, but that could
3143 * potentially break current behavior of eval"".
3147 if (*s == '\'' || *s == '"' || *s == '`') {
3148 /* common case: get past first string, handling escapes */
3149 for (t++; t < PL_bufend && *t != *s;)
3150 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3154 else if (*s == 'q') {
3157 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3161 char open, close, term;
3164 while (t < PL_bufend && isSPACE(*t))
3168 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3172 for (t++; t < PL_bufend; t++) {
3173 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3175 else if (*t == open)
3179 for (t++; t < PL_bufend; t++) {
3180 if (*t == '\\' && t+1 < PL_bufend)
3182 else if (*t == close && --brackets <= 0)
3184 else if (*t == open)
3190 else if (isALNUM_lazy_if(t,UTF)) {
3192 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3195 while (t < PL_bufend && isSPACE(*t))
3197 /* if comma follows first term, call it an anon hash */
3198 /* XXX it could be a comma expression with loop modifiers */
3199 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3200 || (*t == '=' && t[1] == '>')))
3201 OPERATOR(HASHBRACK);
3202 if (PL_expect == XREF)
3205 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3211 yylval.ival = CopLINE(PL_curcop);
3212 if (isSPACE(*s) || *s == '#')
3213 PL_copline = NOLINE; /* invalidate current command line number */
3218 if (PL_lex_brackets <= 0)
3219 yyerror("Unmatched right curly bracket");
3221 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3222 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3223 PL_lex_formbrack = 0;
3224 if (PL_lex_state == LEX_INTERPNORMAL) {
3225 if (PL_lex_brackets == 0) {
3226 if (PL_expect & XFAKEBRACK) {
3227 PL_expect &= XENUMMASK;
3228 PL_lex_state = LEX_INTERPEND;
3230 return yylex(); /* ignore fake brackets */
3232 if (*s == '-' && s[1] == '>')
3233 PL_lex_state = LEX_INTERPENDMAYBE;
3234 else if (*s != '[' && *s != '{')
3235 PL_lex_state = LEX_INTERPEND;
3238 if (PL_expect & XFAKEBRACK) {
3239 PL_expect &= XENUMMASK;
3241 return yylex(); /* ignore fake brackets */
3251 if (PL_expect == XOPERATOR) {
3252 if (ckWARN(WARN_SEMICOLON)
3253 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3255 CopLINE_dec(PL_curcop);
3256 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3257 CopLINE_inc(PL_curcop);
3262 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3264 PL_expect = XOPERATOR;
3265 force_ident(PL_tokenbuf, '&');
3269 yylval.ival = (OPpENTERSUB_AMPER<<8);
3288 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3289 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3291 if (PL_expect == XSTATE && isALPHA(tmp) &&
3292 (s == PL_linestart+1 || s[-2] == '\n') )
3294 if (PL_in_eval && !PL_rsfp) {
3299 if (strnEQ(s,"=cut",4)) {
3313 PL_doextract = TRUE;
3316 if (PL_lex_brackets < PL_lex_formbrack) {
3318 #ifdef PERL_STRICT_CR
3319 for (t = s; SPACE_OR_TAB(*t); t++) ;
3321 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3323 if (*t == '\n' || *t == '#') {
3341 if (PL_expect != XOPERATOR) {
3342 if (s[1] != '<' && !strchr(s,'>'))
3345 s = scan_heredoc(s);
3347 s = scan_inputsymbol(s);
3348 TERM(sublex_start());
3353 SHop(OP_LEFT_SHIFT);
3367 SHop(OP_RIGHT_SHIFT);
3376 if (PL_expect == XOPERATOR) {
3377 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3380 return ','; /* grandfather non-comma-format format */
3384 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3385 PL_tokenbuf[0] = '@';
3386 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3387 sizeof PL_tokenbuf - 1, FALSE);
3388 if (PL_expect == XOPERATOR)
3389 no_op("Array length", s);
3390 if (!PL_tokenbuf[1])
3392 PL_expect = XOPERATOR;
3393 PL_pending_ident = '#';
3397 PL_tokenbuf[0] = '$';
3398 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3399 sizeof PL_tokenbuf - 1, FALSE);
3400 if (PL_expect == XOPERATOR)
3402 if (!PL_tokenbuf[1]) {
3404 yyerror("Final $ should be \\$ or $name");
3408 /* This kludge not intended to be bulletproof. */
3409 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3410 yylval.opval = newSVOP(OP_CONST, 0,
3411 newSViv(PL_compiling.cop_arybase));
3412 yylval.opval->op_private = OPpCONST_ARYBASE;
3418 if (PL_lex_state == LEX_NORMAL)
3421 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3424 PL_tokenbuf[0] = '@';
3425 if (ckWARN(WARN_SYNTAX)) {
3427 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3430 PL_bufptr = skipspace(PL_bufptr);
3431 while (t < PL_bufend && *t != ']')
3433 Perl_warner(aTHX_ WARN_SYNTAX,
3434 "Multidimensional syntax %.*s not supported",
3435 (t - PL_bufptr) + 1, PL_bufptr);
3439 else if (*s == '{') {
3440 PL_tokenbuf[0] = '%';
3441 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3442 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3444 char tmpbuf[sizeof PL_tokenbuf];
3446 for (t++; isSPACE(*t); t++) ;
3447 if (isIDFIRST_lazy_if(t,UTF)) {
3448 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3449 for (; isSPACE(*t); t++) ;
3450 if (*t == ';' && get_cv(tmpbuf, FALSE))
3451 Perl_warner(aTHX_ WARN_SYNTAX,
3452 "You need to quote \"%s\"", tmpbuf);
3458 PL_expect = XOPERATOR;
3459 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3460 bool islop = (PL_last_lop == PL_oldoldbufptr);
3461 if (!islop || PL_last_lop_op == OP_GREPSTART)
3462 PL_expect = XOPERATOR;
3463 else if (strchr("$@\"'`q", *s))
3464 PL_expect = XTERM; /* e.g. print $fh "foo" */
3465 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3466 PL_expect = XTERM; /* e.g. print $fh &sub */
3467 else if (isIDFIRST_lazy_if(s,UTF)) {
3468 char tmpbuf[sizeof PL_tokenbuf];
3469 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3470 if ((tmp = keyword(tmpbuf, len))) {
3471 /* binary operators exclude handle interpretations */
3483 PL_expect = XTERM; /* e.g. print $fh length() */
3488 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3489 if (gv && GvCVu(gv))
3490 PL_expect = XTERM; /* e.g. print $fh subr() */
3493 else if (isDIGIT(*s))
3494 PL_expect = XTERM; /* e.g. print $fh 3 */
3495 else if (*s == '.' && isDIGIT(s[1]))
3496 PL_expect = XTERM; /* e.g. print $fh .3 */
3497 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3498 PL_expect = XTERM; /* e.g. print $fh -1 */
3499 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3500 PL_expect = XTERM; /* print $fh <<"EOF" */
3502 PL_pending_ident = '$';
3506 if (PL_expect == XOPERATOR)
3508 PL_tokenbuf[0] = '@';
3509 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3510 if (!PL_tokenbuf[1]) {
3512 yyerror("Final @ should be \\@ or @name");
3515 if (PL_lex_state == LEX_NORMAL)
3517 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3519 PL_tokenbuf[0] = '%';
3521 /* Warn about @ where they meant $. */
3522 if (ckWARN(WARN_SYNTAX)) {
3523 if (*s == '[' || *s == '{') {
3525 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3527 if (*t == '}' || *t == ']') {
3529 PL_bufptr = skipspace(PL_bufptr);
3530 Perl_warner(aTHX_ WARN_SYNTAX,
3531 "Scalar value %.*s better written as $%.*s",
3532 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3537 PL_pending_ident = '@';
3540 case '/': /* may either be division or pattern */
3541 case '?': /* may either be conditional or pattern */
3542 if (PL_expect != XOPERATOR) {
3543 /* Disable warning on "study /blah/" */
3544 if (PL_oldoldbufptr == PL_last_uni
3545 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3546 || memNE(PL_last_uni, "study", 5)
3547 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3549 s = scan_pat(s,OP_MATCH);
3550 TERM(sublex_start());
3558 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3559 #ifdef PERL_STRICT_CR
3562 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3564 && (s == PL_linestart || s[-1] == '\n') )
3566 PL_lex_formbrack = 0;
3570 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3576 yylval.ival = OPf_SPECIAL;
3582 if (PL_expect != XOPERATOR)
3587 case '0': case '1': case '2': case '3': case '4':
3588 case '5': case '6': case '7': case '8': case '9':
3589 s = scan_num(s, &yylval);
3590 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3591 "### Saw number in '%s'\n", s);
3593 if (PL_expect == XOPERATOR)
3598 s = scan_str(s,FALSE,FALSE);
3599 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3600 "### Saw string before '%s'\n", s);
3602 if (PL_expect == XOPERATOR) {
3603 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3606 return ','; /* grandfather non-comma-format format */
3612 missingterm((char*)0);
3613 yylval.ival = OP_CONST;
3614 TERM(sublex_start());
3617 s = scan_str(s,FALSE,FALSE);
3618 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3619 "### Saw string before '%s'\n", s);
3621 if (PL_expect == XOPERATOR) {
3622 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3625 return ','; /* grandfather non-comma-format format */
3631 missingterm((char*)0);
3632 yylval.ival = OP_CONST;
3633 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3634 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
3635 yylval.ival = OP_STRINGIFY;
3639 TERM(sublex_start());
3642 s = scan_str(s,FALSE,FALSE);
3643 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3644 "### Saw backtick string before '%s'\n", s);
3646 if (PL_expect == XOPERATOR)
3647 no_op("Backticks",s);
3649 missingterm((char*)0);
3650 yylval.ival = OP_BACKTICK;
3652 TERM(sublex_start());
3656 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3657 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3659 if (PL_expect == XOPERATOR)
3660 no_op("Backslash",s);
3664 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3668 while (isDIGIT(*start) || *start == '_')
3670 if (*start == '.' && isDIGIT(start[1])) {
3671 s = scan_num(s, &yylval);
3674 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3675 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF || PL_expect == XSTATE)) {
3679 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3682 s = scan_num(s, &yylval);
3689 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3728 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3730 /* Some keywords can be followed by any delimiter, including ':' */
3731 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3732 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3733 (PL_tokenbuf[0] == 'q' &&
3734 strchr("qwxr", PL_tokenbuf[1])))));
3736 /* x::* is just a word, unless x is "CORE" */
3737 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3741 while (d < PL_bufend && isSPACE(*d))
3742 d++; /* no comments skipped here, or s### is misparsed */
3744 /* Is this a label? */
3745 if (!tmp && PL_expect == XSTATE
3746 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3748 yylval.pval = savepv(PL_tokenbuf);
3753 /* Check for keywords */
3754 tmp = keyword(PL_tokenbuf, len);
3756 /* Is this a word before a => operator? */
3757 if (*d == '=' && d[1] == '>') {
3759 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3760 yylval.opval->op_private = OPpCONST_BARE;
3761 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
3762 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3766 if (tmp < 0) { /* second-class keyword? */
3767 GV *ogv = Nullgv; /* override (winner) */
3768 GV *hgv = Nullgv; /* hidden (loser) */
3769 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3771 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3774 if (GvIMPORTED_CV(gv))
3776 else if (! CvMETHOD(cv))
3780 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3781 (gv = *gvp) != (GV*)&PL_sv_undef &&
3782 GvCVu(gv) && GvIMPORTED_CV(gv))
3788 tmp = 0; /* overridden by import or by GLOBAL */
3791 && -tmp==KEY_lock /* XXX generalizable kludge */
3793 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3795 tmp = 0; /* any sub overrides "weak" keyword */
3797 else { /* no override */
3801 if (ckWARN(WARN_AMBIGUOUS) && hgv
3802 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3803 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3804 "Ambiguous call resolved as CORE::%s(), %s",
3805 GvENAME(hgv), "qualify as such or use &");
3812 default: /* not a keyword */
3816 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3818 /* Get the rest if it looks like a package qualifier */
3820 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3822 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3825 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3826 *s == '\'' ? "'" : "::");
3831 if (PL_expect == XOPERATOR) {
3832 if (PL_bufptr == PL_linestart) {
3833 CopLINE_dec(PL_curcop);
3834 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3835 CopLINE_inc(PL_curcop);
3838 no_op("Bareword",s);
3841 /* Look for a subroutine with this name in current package,
3842 unless name is "Foo::", in which case Foo is a bearword
3843 (and a package name). */
3846 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3848 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3849 Perl_warner(aTHX_ WARN_BAREWORD,
3850 "Bareword \"%s\" refers to nonexistent package",
3853 PL_tokenbuf[len] = '\0';
3860 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3863 /* if we saw a global override before, get the right name */
3866 sv = newSVpvn("CORE::GLOBAL::",14);
3867 sv_catpv(sv,PL_tokenbuf);
3870 sv = newSVpv(PL_tokenbuf,0);
3872 /* Presume this is going to be a bareword of some sort. */
3875 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3876 yylval.opval->op_private = OPpCONST_BARE;
3878 /* And if "Foo::", then that's what it certainly is. */
3883 /* See if it's the indirect object for a list operator. */
3885 if (PL_oldoldbufptr &&
3886 PL_oldoldbufptr < PL_bufptr &&
3887 (PL_oldoldbufptr == PL_last_lop
3888 || PL_oldoldbufptr == PL_last_uni) &&
3889 /* NO SKIPSPACE BEFORE HERE! */
3890 (PL_expect == XREF ||
3891 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3893 bool immediate_paren = *s == '(';
3895 /* (Now we can afford to cross potential line boundary.) */
3898 /* Two barewords in a row may indicate method call. */
3900 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3903 /* If not a declared subroutine, it's an indirect object. */
3904 /* (But it's an indir obj regardless for sort.) */
3906 if ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
3907 ((!gv || !GvCVu(gv)) &&
3908 (PL_last_lop_op != OP_MAPSTART &&
3909 PL_last_lop_op != OP_GREPSTART))))
3911 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3916 PL_expect = XOPERATOR;
3919 /* Is this a word before a => operator? */
3920 if (*s == '=' && s[1] == '>' && !pkgname) {
3922 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
3923 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
3924 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3928 /* If followed by a paren, it's certainly a subroutine. */
3931 if (gv && GvCVu(gv)) {
3932 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
3933 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3938 PL_nextval[PL_nexttoke].opval = yylval.opval;
3939 PL_expect = XOPERATOR;
3945 /* If followed by var or block, call it a method (unless sub) */
3947 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3948 PL_last_lop = PL_oldbufptr;
3949 PL_last_lop_op = OP_METHOD;
3953 /* If followed by a bareword, see if it looks like indir obj. */
3955 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3958 /* Not a method, so call it a subroutine (if defined) */
3960 if (gv && GvCVu(gv)) {
3962 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3963 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3964 "Ambiguous use of -%s resolved as -&%s()",
3965 PL_tokenbuf, PL_tokenbuf);
3966 /* Check for a constant sub */
3968 if ((sv = cv_const_sv(cv))) {
3970 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3971 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3972 yylval.opval->op_private = 0;
3976 /* Resolve to GV now. */
3977 op_free(yylval.opval);
3978 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3979 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3980 PL_last_lop = PL_oldbufptr;
3981 PL_last_lop_op = OP_ENTERSUB;
3982 /* Is there a prototype? */
3985 char *proto = SvPV((SV*)cv, len);
3988 if (strEQ(proto, "$"))
3990 if (*proto == '&' && *s == '{') {
3991 sv_setpv(PL_subname,"__ANON__");
3995 PL_nextval[PL_nexttoke].opval = yylval.opval;
4001 /* Call it a bare word */
4003 if (PL_hints & HINT_STRICT_SUBS)
4004 yylval.opval->op_private |= OPpCONST_STRICT;
4007 if (ckWARN(WARN_RESERVED)) {
4008 if (lastchar != '-') {
4009 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4010 if (!*d && strNE(PL_tokenbuf,"main"))
4011 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
4018 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
4019 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4020 "Operator or semicolon missing before %c%s",
4021 lastchar, PL_tokenbuf);
4022 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4023 "Ambiguous use of %c resolved as operator %c",
4024 lastchar, lastchar);
4030 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4031 newSVpv(CopFILE(PL_curcop),0));
4035 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4036 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4039 case KEY___PACKAGE__:
4040 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4042 ? newSVsv(PL_curstname)
4051 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4052 char *pname = "main";
4053 if (PL_tokenbuf[2] == 'D')
4054 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
4055 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4058 GvIOp(gv) = newIO();
4059 IoIFP(GvIOp(gv)) = PL_rsfp;
4060 #if defined(HAS_FCNTL) && defined(F_SETFD)
4062 int fd = PerlIO_fileno(PL_rsfp);
4063 fcntl(fd,F_SETFD,fd >= 3);
4066 /* Mark this internal pseudo-handle as clean */
4067 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4069 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4070 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4071 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4073 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4074 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4075 /* if the script was opened in binmode, we need to revert
4076 * it to text mode for compatibility; but only iff it has CRs
4077 * XXX this is a questionable hack at best. */
4078 if (PL_bufend-PL_bufptr > 2
4079 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4082 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4083 loc = PerlIO_tell(PL_rsfp);
4084 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4087 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
4089 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4090 #endif /* NETWARE */
4091 #ifdef PERLIO_IS_STDIO /* really? */
4092 # if defined(__BORLANDC__)
4093 /* XXX see note in do_binmode() */
4094 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
4098 PerlIO_seek(PL_rsfp, loc, 0);
4102 #ifdef PERLIO_LAYERS
4103 if (UTF && !IN_BYTES)
4104 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
4117 if (PL_expect == XSTATE) {
4124 if (*s == ':' && s[1] == ':') {
4127 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4128 if (!(tmp = keyword(PL_tokenbuf, len)))
4129 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4143 LOP(OP_ACCEPT,XTERM);
4149 LOP(OP_ATAN2,XTERM);
4155 LOP(OP_BINMODE,XTERM);
4158 LOP(OP_BLESS,XTERM);
4167 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4184 if (!PL_cryptseen) {
4185 PL_cryptseen = TRUE;
4189 LOP(OP_CRYPT,XTERM);
4192 LOP(OP_CHMOD,XTERM);
4195 LOP(OP_CHOWN,XTERM);
4198 LOP(OP_CONNECT,XTERM);
4214 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4218 PL_hints |= HINT_BLOCK_SCOPE;
4228 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4229 LOP(OP_DBMOPEN,XTERM);
4235 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4242 yylval.ival = CopLINE(PL_curcop);
4256 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4257 UNIBRACK(OP_ENTEREVAL);
4272 case KEY_endhostent:
4278 case KEY_endservent:
4281 case KEY_endprotoent:
4292 yylval.ival = CopLINE(PL_curcop);
4294 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4296 if ((PL_bufend - p) >= 3 &&
4297 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4299 else if ((PL_bufend - p) >= 4 &&
4300 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4303 if (isIDFIRST_lazy_if(p,UTF)) {
4304 p = scan_ident(p, PL_bufend,
4305 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4309 Perl_croak(aTHX_ "Missing $ on loop variable");
4314 LOP(OP_FORMLINE,XTERM);
4320 LOP(OP_FCNTL,XTERM);
4326 LOP(OP_FLOCK,XTERM);
4335 LOP(OP_GREPSTART, XREF);
4338 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4353 case KEY_getpriority:
4354 LOP(OP_GETPRIORITY,XTERM);
4356 case KEY_getprotobyname:
4359 case KEY_getprotobynumber:
4360 LOP(OP_GPBYNUMBER,XTERM);
4362 case KEY_getprotoent:
4374 case KEY_getpeername:
4375 UNI(OP_GETPEERNAME);
4377 case KEY_gethostbyname:
4380 case KEY_gethostbyaddr:
4381 LOP(OP_GHBYADDR,XTERM);
4383 case KEY_gethostent:
4386 case KEY_getnetbyname:
4389 case KEY_getnetbyaddr:
4390 LOP(OP_GNBYADDR,XTERM);
4395 case KEY_getservbyname:
4396 LOP(OP_GSBYNAME,XTERM);
4398 case KEY_getservbyport:
4399 LOP(OP_GSBYPORT,XTERM);
4401 case KEY_getservent:
4404 case KEY_getsockname:
4405 UNI(OP_GETSOCKNAME);
4407 case KEY_getsockopt:
4408 LOP(OP_GSOCKOPT,XTERM);
4430 yylval.ival = CopLINE(PL_curcop);
4434 LOP(OP_INDEX,XTERM);
4440 LOP(OP_IOCTL,XTERM);
4452 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4484 LOP(OP_LISTEN,XTERM);
4493 s = scan_pat(s,OP_MATCH);
4494 TERM(sublex_start());
4497 LOP(OP_MAPSTART, XREF);
4500 LOP(OP_MKDIR,XTERM);
4503 LOP(OP_MSGCTL,XTERM);
4506 LOP(OP_MSGGET,XTERM);
4509 LOP(OP_MSGRCV,XTERM);
4512 LOP(OP_MSGSND,XTERM);
4518 if (isIDFIRST_lazy_if(s,UTF)) {
4519 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4520 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4522 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4523 if (!PL_in_my_stash) {
4526 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4534 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4541 if (PL_expect != XSTATE)
4542 yyerror("\"no\" not allowed in expression");
4543 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4544 s = force_version(s, FALSE);
4549 if (*s == '(' || (s = skipspace(s), *s == '('))
4556 if (isIDFIRST_lazy_if(s,UTF)) {
4558 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4560 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4561 Perl_warner(aTHX_ WARN_PRECEDENCE,
4562 "Precedence problem: open %.*s should be open(%.*s)",
4568 yylval.ival = OP_OR;
4578 LOP(OP_OPEN_DIR,XTERM);
4581 checkcomma(s,PL_tokenbuf,"filehandle");
4585 checkcomma(s,PL_tokenbuf,"filehandle");
4604 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4608 LOP(OP_PIPE_OP,XTERM);
4611 s = scan_str(s,FALSE,FALSE);
4613 missingterm((char*)0);
4614 yylval.ival = OP_CONST;
4615 TERM(sublex_start());
4621 s = scan_str(s,FALSE,FALSE);
4623 missingterm((char*)0);
4625 if (SvCUR(PL_lex_stuff)) {
4628 d = SvPV_force(PL_lex_stuff, len);
4631 for (; isSPACE(*d) && len; --len, ++d) ;
4634 if (!warned && ckWARN(WARN_QW)) {
4635 for (; !isSPACE(*d) && len; --len, ++d) {
4637 Perl_warner(aTHX_ WARN_QW,
4638 "Possible attempt to separate words with commas");
4641 else if (*d == '#') {
4642 Perl_warner(aTHX_ WARN_QW,
4643 "Possible attempt to put comments in qw() list");
4649 for (; !isSPACE(*d) && len; --len, ++d) ;
4651 sv = newSVpvn(b, d-b);
4652 if (DO_UTF8(PL_lex_stuff))
4654 words = append_elem(OP_LIST, words,
4655 newSVOP(OP_CONST, 0, tokeq(sv)));
4659 PL_nextval[PL_nexttoke].opval = words;
4664 SvREFCNT_dec(PL_lex_stuff);
4665 PL_lex_stuff = Nullsv;
4671 s = scan_str(s,FALSE,FALSE);
4673 missingterm((char*)0);
4674 yylval.ival = OP_STRINGIFY;
4675 if (SvIVX(PL_lex_stuff) == '\'')
4676 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4677 TERM(sublex_start());
4680 s = scan_pat(s,OP_QR);
4681 TERM(sublex_start());
4684 s = scan_str(s,FALSE,FALSE);
4686 missingterm((char*)0);
4687 yylval.ival = OP_BACKTICK;
4689 TERM(sublex_start());
4697 s = force_version(s, FALSE);
4699 else if (*s != 'v' || !isDIGIT(s[1])
4700 || (s = force_version(s, TRUE), *s == 'v'))
4702 *PL_tokenbuf = '\0';
4703 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4704 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4705 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4707 yyerror("<> should be quotes");
4715 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4719 LOP(OP_RENAME,XTERM);
4728 LOP(OP_RINDEX,XTERM);
4751 LOP(OP_REVERSE,XTERM);
4762 TERM(sublex_start());
4764 TOKEN(1); /* force error */
4773 LOP(OP_SELECT,XTERM);
4779 LOP(OP_SEMCTL,XTERM);
4782 LOP(OP_SEMGET,XTERM);
4785 LOP(OP_SEMOP,XTERM);
4791 LOP(OP_SETPGRP,XTERM);
4793 case KEY_setpriority:
4794 LOP(OP_SETPRIORITY,XTERM);
4796 case KEY_sethostent:
4802 case KEY_setservent:
4805 case KEY_setprotoent:
4815 LOP(OP_SEEKDIR,XTERM);
4817 case KEY_setsockopt:
4818 LOP(OP_SSOCKOPT,XTERM);
4824 LOP(OP_SHMCTL,XTERM);
4827 LOP(OP_SHMGET,XTERM);
4830 LOP(OP_SHMREAD,XTERM);
4833 LOP(OP_SHMWRITE,XTERM);
4836 LOP(OP_SHUTDOWN,XTERM);
4845 LOP(OP_SOCKET,XTERM);
4847 case KEY_socketpair:
4848 LOP(OP_SOCKPAIR,XTERM);
4851 checkcomma(s,PL_tokenbuf,"subroutine name");
4853 if (*s == ';' || *s == ')') /* probably a close */
4854 Perl_croak(aTHX_ "sort is now a reserved word");
4856 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4860 LOP(OP_SPLIT,XTERM);
4863 LOP(OP_SPRINTF,XTERM);
4866 LOP(OP_SPLICE,XTERM);
4881 LOP(OP_SUBSTR,XTERM);
4887 char tmpbuf[sizeof PL_tokenbuf];
4888 SSize_t tboffset = 0;
4889 expectation attrful;
4890 bool have_name, have_proto;
4895 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4896 (*s == ':' && s[1] == ':'))
4899 attrful = XATTRBLOCK;
4900 /* remember buffer pos'n for later force_word */
4901 tboffset = s - PL_oldbufptr;
4902 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4903 if (strchr(tmpbuf, ':'))
4904 sv_setpv(PL_subname, tmpbuf);
4906 sv_setsv(PL_subname,PL_curstname);
4907 sv_catpvn(PL_subname,"::",2);
4908 sv_catpvn(PL_subname,tmpbuf,len);
4915 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4916 PL_expect = XTERMBLOCK;
4917 attrful = XATTRTERM;
4918 sv_setpv(PL_subname,"?");
4922 if (key == KEY_format) {
4924 PL_lex_formbrack = PL_lex_brackets + 1;
4926 (void) force_word(PL_oldbufptr + tboffset, WORD,
4931 /* Look for a prototype */
4935 s = scan_str(s,FALSE,FALSE);
4937 Perl_croak(aTHX_ "Prototype not terminated");
4939 d = SvPVX(PL_lex_stuff);
4941 for (p = d; *p; ++p) {
4946 SvCUR(PL_lex_stuff) = tmp;
4954 if (*s == ':' && s[1] != ':')
4955 PL_expect = attrful;
4958 PL_nextval[PL_nexttoke].opval =
4959 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4960 PL_lex_stuff = Nullsv;
4964 sv_setpv(PL_subname,"__ANON__");
4967 (void) force_word(PL_oldbufptr + tboffset, WORD,
4976 LOP(OP_SYSTEM,XREF);
4979 LOP(OP_SYMLINK,XTERM);
4982 LOP(OP_SYSCALL,XTERM);
4985 LOP(OP_SYSOPEN,XTERM);
4988 LOP(OP_SYSSEEK,XTERM);
4991 LOP(OP_SYSREAD,XTERM);
4994 LOP(OP_SYSWRITE,XTERM);
4998 TERM(sublex_start());
5019 LOP(OP_TRUNCATE,XTERM);
5031 yylval.ival = CopLINE(PL_curcop);
5035 yylval.ival = CopLINE(PL_curcop);
5039 LOP(OP_UNLINK,XTERM);
5045 LOP(OP_UNPACK,XTERM);
5048 LOP(OP_UTIME,XTERM);
5054 LOP(OP_UNSHIFT,XTERM);
5057 if (PL_expect != XSTATE)
5058 yyerror("\"use\" not allowed in expression");
5060 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
5061 s = force_version(s, TRUE);
5062 if (*s == ';' || (s = skipspace(s), *s == ';')) {
5063 PL_nextval[PL_nexttoke].opval = Nullop;
5066 else if (*s == 'v') {
5067 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5068 s = force_version(s, FALSE);
5072 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5073 s = force_version(s, FALSE);
5085 yylval.ival = CopLINE(PL_curcop);
5089 PL_hints |= HINT_BLOCK_SCOPE;
5096 LOP(OP_WAITPID,XTERM);
5104 static char ctl_l[2];
5106 if (ctl_l[0] == '\0')
5107 ctl_l[0] = toCTRL('L');
5108 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5111 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5116 if (PL_expect == XOPERATOR)
5122 yylval.ival = OP_XOR;
5127 TERM(sublex_start());
5132 #pragma segment Main
5136 S_pending_ident(pTHX)
5140 /* pit holds the identifier we read and pending_ident is reset */
5141 char pit = PL_pending_ident;
5142 PL_pending_ident = 0;
5144 DEBUG_T({ PerlIO_printf(Perl_debug_log,
5145 "### Tokener saw identifier '%s'\n", PL_tokenbuf); });
5147 /* if we're in a my(), we can't allow dynamics here.
5148 $foo'bar has already been turned into $foo::bar, so
5149 just check for colons.
5151 if it's a legal name, the OP is a PADANY.
5154 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
5155 if (strchr(PL_tokenbuf,':'))
5156 yyerror(Perl_form(aTHX_ "No package name allowed for "
5157 "variable %s in \"our\"",
5159 tmp = pad_allocmy(PL_tokenbuf);
5162 if (strchr(PL_tokenbuf,':'))
5163 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
5165 yylval.opval = newOP(OP_PADANY, 0);
5166 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
5172 build the ops for accesses to a my() variable.
5174 Deny my($a) or my($b) in a sort block, *if* $a or $b is
5175 then used in a comparison. This catches most, but not
5176 all cases. For instance, it catches
5177 sort { my($a); $a <=> $b }
5179 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
5180 (although why you'd do that is anyone's guess).
5183 if (!strchr(PL_tokenbuf,':')) {
5184 #ifdef USE_5005THREADS
5185 /* Check for single character per-thread SVs */
5186 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
5187 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
5188 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
5190 yylval.opval = newOP(OP_THREADSV, 0);
5191 yylval.opval->op_targ = tmp;
5194 #endif /* USE_5005THREADS */
5195 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
5196 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
5197 /* might be an "our" variable" */
5198 if (SvFLAGS(namesv) & SVpad_OUR) {
5199 /* build ops for a bareword */
5200 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
5201 sv_catpvn(sym, "::", 2);
5202 sv_catpv(sym, PL_tokenbuf+1);
5203 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
5204 yylval.opval->op_private = OPpCONST_ENTERED;
5205 gv_fetchpv(SvPVX(sym),
5207 ? (GV_ADDMULTI | GV_ADDINEVAL)
5210 ((PL_tokenbuf[0] == '$') ? SVt_PV
5211 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5216 /* if it's a sort block and they're naming $a or $b */
5217 if (PL_last_lop_op == OP_SORT &&
5218 PL_tokenbuf[0] == '$' &&
5219 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
5222 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
5223 d < PL_bufend && *d != '\n';
5226 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
5227 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
5233 yylval.opval = newOP(OP_PADANY, 0);
5234 yylval.opval->op_targ = tmp;
5240 Whine if they've said @foo in a doublequoted string,
5241 and @foo isn't a variable we can find in the symbol
5244 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
5245 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
5246 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
5247 && ckWARN(WARN_AMBIGUOUS))
5249 /* Downgraded from fatal to warning 20000522 mjd */
5250 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5251 "Possible unintended interpolation of %s in string",
5256 /* build ops for a bareword */
5257 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
5258 yylval.opval->op_private = OPpCONST_ENTERED;
5259 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
5260 ((PL_tokenbuf[0] == '$') ? SVt_PV
5261 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5267 Perl_keyword(pTHX_ register char *d, I32 len)
5272 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5273 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5274 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5275 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5276 if (strEQ(d,"__END__")) return KEY___END__;
5280 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5285 if (strEQ(d,"and")) return -KEY_and;
5286 if (strEQ(d,"abs")) return -KEY_abs;
5289 if (strEQ(d,"alarm")) return -KEY_alarm;
5290 if (strEQ(d,"atan2")) return -KEY_atan2;
5293 if (strEQ(d,"accept")) return -KEY_accept;
5298 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5301 if (strEQ(d,"bless")) return -KEY_bless;
5302 if (strEQ(d,"bind")) return -KEY_bind;
5303 if (strEQ(d,"binmode")) return -KEY_binmode;
5306 if (strEQ(d,"CORE")) return -KEY_CORE;
5307 if (strEQ(d,"CHECK")) return KEY_CHECK;
5312 if (strEQ(d,"cmp")) return -KEY_cmp;
5313 if (strEQ(d,"chr")) return -KEY_chr;
5314 if (strEQ(d,"cos")) return -KEY_cos;
5317 if (strEQ(d,"chop")) return -KEY_chop;
5320 if (strEQ(d,"close")) return -KEY_close;
5321 if (strEQ(d,"chdir")) return -KEY_chdir;
5322 if (strEQ(d,"chomp")) return -KEY_chomp;
5323 if (strEQ(d,"chmod")) return -KEY_chmod;
5324 if (strEQ(d,"chown")) return -KEY_chown;
5325 if (strEQ(d,"crypt")) return -KEY_crypt;
5328 if (strEQ(d,"chroot")) return -KEY_chroot;
5329 if (strEQ(d,"caller")) return -KEY_caller;
5332 if (strEQ(d,"connect")) return -KEY_connect;
5335 if (strEQ(d,"closedir")) return -KEY_closedir;
5336 if (strEQ(d,"continue")) return -KEY_continue;
5341 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5346 if (strEQ(d,"do")) return KEY_do;
5349 if (strEQ(d,"die")) return -KEY_die;
5352 if (strEQ(d,"dump")) return -KEY_dump;
5355 if (strEQ(d,"delete")) return KEY_delete;
5358 if (strEQ(d,"defined")) return KEY_defined;
5359 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5362 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5367 if (strEQ(d,"END")) return KEY_END;
5372 if (strEQ(d,"eq")) return -KEY_eq;
5375 if (strEQ(d,"eof")) return -KEY_eof;
5376 if (strEQ(d,"exp")) return -KEY_exp;
5379 if (strEQ(d,"else")) return KEY_else;
5380 if (strEQ(d,"exit")) return -KEY_exit;
5381 if (strEQ(d,"eval")) return KEY_eval;
5382 if (strEQ(d,"exec")) return -KEY_exec;
5383 if (strEQ(d,"each")) return -KEY_each;
5386 if (strEQ(d,"elsif")) return KEY_elsif;
5389 if (strEQ(d,"exists")) return KEY_exists;
5390 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5393 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5394 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5397 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5400 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5401 if (strEQ(d,"endservent")) return -KEY_endservent;
5404 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5411 if (strEQ(d,"for")) return KEY_for;
5414 if (strEQ(d,"fork")) return -KEY_fork;
5417 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5418 if (strEQ(d,"flock")) return -KEY_flock;
5421 if (strEQ(d,"format")) return KEY_format;
5422 if (strEQ(d,"fileno")) return -KEY_fileno;
5425 if (strEQ(d,"foreach")) return KEY_foreach;
5428 if (strEQ(d,"formline")) return -KEY_formline;
5433 if (strnEQ(d,"get",3)) {
5438 if (strEQ(d,"ppid")) return -KEY_getppid;
5439 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5442 if (strEQ(d,"pwent")) return -KEY_getpwent;
5443 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5444 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5447 if (strEQ(d,"peername")) return -KEY_getpeername;
5448 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5449 if (strEQ(d,"priority")) return -KEY_getpriority;
5452 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5455 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5459 else if (*d == 'h') {
5460 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5461 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5462 if (strEQ(d,"hostent")) return -KEY_gethostent;
5464 else if (*d == 'n') {
5465 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5466 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5467 if (strEQ(d,"netent")) return -KEY_getnetent;
5469 else if (*d == 's') {
5470 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5471 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5472 if (strEQ(d,"servent")) return -KEY_getservent;
5473 if (strEQ(d,"sockname")) return -KEY_getsockname;
5474 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5476 else if (*d == 'g') {
5477 if (strEQ(d,"grent")) return -KEY_getgrent;
5478 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5479 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5481 else if (*d == 'l') {
5482 if (strEQ(d,"login")) return -KEY_getlogin;
5484 else if (strEQ(d,"c")) return -KEY_getc;
5489 if (strEQ(d,"gt")) return -KEY_gt;
5490 if (strEQ(d,"ge")) return -KEY_ge;
5493 if (strEQ(d,"grep")) return KEY_grep;
5494 if (strEQ(d,"goto")) return KEY_goto;
5495 if (strEQ(d,"glob")) return KEY_glob;
5498 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5503 if (strEQ(d,"hex")) return -KEY_hex;
5506 if (strEQ(d,"INIT")) return KEY_INIT;
5511 if (strEQ(d,"if")) return KEY_if;
5514 if (strEQ(d,"int")) return -KEY_int;
5517 if (strEQ(d,"index")) return -KEY_index;
5518 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5523 if (strEQ(d,"join")) return -KEY_join;
5527 if (strEQ(d,"keys")) return -KEY_keys;
5528 if (strEQ(d,"kill")) return -KEY_kill;
5534 if (strEQ(d,"lt")) return -KEY_lt;
5535 if (strEQ(d,"le")) return -KEY_le;
5536 if (strEQ(d,"lc")) return -KEY_lc;
5539 if (strEQ(d,"log")) return -KEY_log;
5542 if (strEQ(d,"last")) return KEY_last;
5543 if (strEQ(d,"link")) return -KEY_link;
5544 if (strEQ(d,"lock")) return -KEY_lock;
5547 if (strEQ(d,"local")) return KEY_local;
5548 if (strEQ(d,"lstat")) return -KEY_lstat;
5551 if (strEQ(d,"length")) return -KEY_length;
5552 if (strEQ(d,"listen")) return -KEY_listen;
5555 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5558 if (strEQ(d,"localtime")) return -KEY_localtime;
5564 case 1: return KEY_m;
5566 if (strEQ(d,"my")) return KEY_my;
5569 if (strEQ(d,"map")) return KEY_map;
5572 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5575 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5576 if (strEQ(d,"msgget")) return -KEY_msgget;
5577 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5578 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5583 if (strEQ(d,"next")) return KEY_next;
5584 if (strEQ(d,"ne")) return -KEY_ne;
5585 if (strEQ(d,"not")) return -KEY_not;
5586 if (strEQ(d,"no")) return KEY_no;
5591 if (strEQ(d,"or")) return -KEY_or;
5594 if (strEQ(d,"ord")) return -KEY_ord;
5595 if (strEQ(d,"oct")) return -KEY_oct;
5596 if (strEQ(d,"our")) return KEY_our;
5599 if (strEQ(d,"open")) return -KEY_open;
5602 if (strEQ(d,"opendir")) return -KEY_opendir;
5609 if (strEQ(d,"pop")) return -KEY_pop;
5610 if (strEQ(d,"pos")) return KEY_pos;
5613 if (strEQ(d,"push")) return -KEY_push;
5614 if (strEQ(d,"pack")) return -KEY_pack;
5615 if (strEQ(d,"pipe")) return -KEY_pipe;
5618 if (strEQ(d,"print")) return KEY_print;
5621 if (strEQ(d,"printf")) return KEY_printf;
5624 if (strEQ(d,"package")) return KEY_package;
5627 if (strEQ(d,"prototype")) return KEY_prototype;
5632 if (strEQ(d,"q")) return KEY_q;
5633 if (strEQ(d,"qr")) return KEY_qr;
5634 if (strEQ(d,"qq")) return KEY_qq;
5635 if (strEQ(d,"qw")) return KEY_qw;
5636 if (strEQ(d,"qx")) return KEY_qx;
5638 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5643 if (strEQ(d,"ref")) return -KEY_ref;
5646 if (strEQ(d,"read")) return -KEY_read;
5647 if (strEQ(d,"rand")) return -KEY_rand;
5648 if (strEQ(d,"recv")) return -KEY_recv;
5649 if (strEQ(d,"redo")) return KEY_redo;
5652 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5653 if (strEQ(d,"reset")) return -KEY_reset;
5656 if (strEQ(d,"return")) return KEY_return;
5657 if (strEQ(d,"rename")) return -KEY_rename;
5658 if (strEQ(d,"rindex")) return -KEY_rindex;
5661 if (strEQ(d,"require")) return KEY_require;
5662 if (strEQ(d,"reverse")) return -KEY_reverse;
5663 if (strEQ(d,"readdir")) return -KEY_readdir;
5666 if (strEQ(d,"readlink")) return -KEY_readlink;
5667 if (strEQ(d,"readline")) return -KEY_readline;
5668 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5671 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5677 case 0: return KEY_s;
5679 if (strEQ(d,"scalar")) return KEY_scalar;
5684 if (strEQ(d,"seek")) return -KEY_seek;
5685 if (strEQ(d,"send")) return -KEY_send;
5688 if (strEQ(d,"semop")) return -KEY_semop;
5691 if (strEQ(d,"select")) return -KEY_select;
5692 if (strEQ(d,"semctl")) return -KEY_semctl;
5693 if (strEQ(d,"semget")) return -KEY_semget;
5696 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5697 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5700 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5701 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5704 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5707 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5708 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5709 if (strEQ(d,"setservent")) return -KEY_setservent;
5712 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5713 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5720 if (strEQ(d,"shift")) return -KEY_shift;
5723 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5724 if (strEQ(d,"shmget")) return -KEY_shmget;
5727 if (strEQ(d,"shmread")) return -KEY_shmread;
5730 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5731 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5736 if (strEQ(d,"sin")) return -KEY_sin;
5739 if (strEQ(d,"sleep")) return -KEY_sleep;
5742 if (strEQ(d,"sort")) return KEY_sort;
5743 if (strEQ(d,"socket")) return -KEY_socket;
5744 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5747 if (strEQ(d,"split")) return KEY_split;
5748 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5749 if (strEQ(d,"splice")) return -KEY_splice;
5752 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5755 if (strEQ(d,"srand")) return -KEY_srand;
5758 if (strEQ(d,"stat")) return -KEY_stat;
5759 if (strEQ(d,"study")) return KEY_study;
5762 if (strEQ(d,"substr")) return -KEY_substr;
5763 if (strEQ(d,"sub")) return KEY_sub;
5768 if (strEQ(d,"system")) return -KEY_system;
5771 if (strEQ(d,"symlink")) return -KEY_symlink;
5772 if (strEQ(d,"syscall")) return -KEY_syscall;
5773 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5774 if (strEQ(d,"sysread")) return -KEY_sysread;
5775 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5778 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5787 if (strEQ(d,"tr")) return KEY_tr;
5790 if (strEQ(d,"tie")) return KEY_tie;
5793 if (strEQ(d,"tell")) return -KEY_tell;
5794 if (strEQ(d,"tied")) return KEY_tied;
5795 if (strEQ(d,"time")) return -KEY_time;
5798 if (strEQ(d,"times")) return -KEY_times;
5801 if (strEQ(d,"telldir")) return -KEY_telldir;
5804 if (strEQ(d,"truncate")) return -KEY_truncate;
5811 if (strEQ(d,"uc")) return -KEY_uc;
5814 if (strEQ(d,"use")) return KEY_use;
5817 if (strEQ(d,"undef")) return KEY_undef;
5818 if (strEQ(d,"until")) return KEY_until;
5819 if (strEQ(d,"untie")) return KEY_untie;
5820 if (strEQ(d,"utime")) return -KEY_utime;
5821 if (strEQ(d,"umask")) return -KEY_umask;
5824 if (strEQ(d,"unless")) return KEY_unless;
5825 if (strEQ(d,"unpack")) return -KEY_unpack;
5826 if (strEQ(d,"unlink")) return -KEY_unlink;
5829 if (strEQ(d,"unshift")) return -KEY_unshift;
5830 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5835 if (strEQ(d,"values")) return -KEY_values;
5836 if (strEQ(d,"vec")) return -KEY_vec;
5841 if (strEQ(d,"warn")) return -KEY_warn;
5842 if (strEQ(d,"wait")) return -KEY_wait;
5845 if (strEQ(d,"while")) return KEY_while;
5846 if (strEQ(d,"write")) return -KEY_write;
5849 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5852 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5857 if (len == 1) return -KEY_x;
5858 if (strEQ(d,"xor")) return -KEY_xor;
5861 if (len == 1) return KEY_y;
5870 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5874 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5875 if (ckWARN(WARN_SYNTAX)) {
5877 for (w = s+2; *w && level; w++) {
5884 for (; *w && isSPACE(*w); w++) ;
5885 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5886 Perl_warner(aTHX_ WARN_SYNTAX,
5887 "%s (...) interpreted as function",name);
5890 while (s < PL_bufend && isSPACE(*s))
5894 while (s < PL_bufend && isSPACE(*s))
5896 if (isIDFIRST_lazy_if(s,UTF)) {
5898 while (isALNUM_lazy_if(s,UTF))
5900 while (s < PL_bufend && isSPACE(*s))
5905 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5909 Perl_croak(aTHX_ "No comma allowed after %s", what);
5914 /* Either returns sv, or mortalizes sv and returns a new SV*.
5915 Best used as sv=new_constant(..., sv, ...).
5916 If s, pv are NULL, calls subroutine with one argument,
5917 and type is used with error messages only. */
5920 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5924 HV *table = GvHV(PL_hintgv); /* ^H */
5928 const char *why1, *why2, *why3;
5930 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
5933 why2 = strEQ(key,"charnames")
5934 ? "(possibly a missing \"use charnames ...\")"
5936 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
5937 (type ? type: "undef"), why2);
5939 /* This is convoluted and evil ("goto considered harmful")
5940 * but I do not understand the intricacies of all the different
5941 * failure modes of %^H in here. The goal here is to make
5942 * the most probable error message user-friendly. --jhi */
5947 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
5948 (type ? type: "undef"), why1, why2, why3);
5950 yyerror(SvPVX(msg));
5954 cvp = hv_fetch(table, key, strlen(key), FALSE);
5955 if (!cvp || !SvOK(*cvp)) {
5958 why3 = "} is not defined";
5961 sv_2mortal(sv); /* Parent created it permanently */
5964 pv = sv_2mortal(newSVpvn(s, len));
5966 typesv = sv_2mortal(newSVpv(type, 0));
5968 typesv = &PL_sv_undef;
5970 PUSHSTACKi(PERLSI_OVERLOAD);
5982 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5986 /* Check the eval first */
5987 if (!PL_in_eval && SvTRUE(ERRSV)) {
5989 sv_catpv(ERRSV, "Propagated");
5990 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5992 res = SvREFCNT_inc(sv);
5996 (void)SvREFCNT_inc(res);
6005 why1 = "Call to &{$^H{";
6007 why3 = "}} did not return a defined value";
6016 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
6018 register char *d = dest;
6019 register char *e = d + destlen - 3; /* two-character token, ending NUL */
6022 Perl_croak(aTHX_ ident_too_long);
6023 if (isALNUM(*s)) /* UTF handled below */
6025 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
6030 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
6034 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6035 char *t = s + UTF8SKIP(s);
6036 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6038 if (d + (t - s) > e)
6039 Perl_croak(aTHX_ ident_too_long);
6040 Copy(s, d, t - s, char);
6053 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
6063 e = d + destlen - 3; /* two-character token, ending NUL */
6065 while (isDIGIT(*s)) {
6067 Perl_croak(aTHX_ ident_too_long);
6074 Perl_croak(aTHX_ ident_too_long);
6075 if (isALNUM(*s)) /* UTF handled below */
6077 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
6082 else if (*s == ':' && s[1] == ':') {
6086 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6087 char *t = s + UTF8SKIP(s);
6088 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6090 if (d + (t - s) > e)
6091 Perl_croak(aTHX_ ident_too_long);
6092 Copy(s, d, t - s, char);
6103 if (PL_lex_state != LEX_NORMAL)
6104 PL_lex_state = LEX_INTERPENDMAYBE;
6107 if (*s == '$' && s[1] &&
6108 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
6121 if (*d == '^' && *s && isCONTROLVAR(*s)) {
6126 if (isSPACE(s[-1])) {
6129 if (!SPACE_OR_TAB(ch)) {
6135 if (isIDFIRST_lazy_if(d,UTF)) {
6139 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
6141 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
6144 Copy(s, d, e - s, char);
6149 while ((isALNUM(*s) || *s == ':') && d < e)
6152 Perl_croak(aTHX_ ident_too_long);
6155 while (s < send && SPACE_OR_TAB(*s)) s++;
6156 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
6157 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
6158 const char *brack = *s == '[' ? "[...]" : "{...}";
6159 Perl_warner(aTHX_ WARN_AMBIGUOUS,
6160 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
6161 funny, dest, brack, funny, dest, brack);
6164 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
6168 /* Handle extended ${^Foo} variables
6169 * 1999-02-27 mjd-perl-patch@plover.com */
6170 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
6174 while (isALNUM(*s) && d < e) {
6178 Perl_croak(aTHX_ ident_too_long);
6183 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
6184 PL_lex_state = LEX_INTERPEND;
6187 if (PL_lex_state == LEX_NORMAL) {
6188 if (ckWARN(WARN_AMBIGUOUS) &&
6189 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
6191 Perl_warner(aTHX_ WARN_AMBIGUOUS,
6192 "Ambiguous use of %c{%s} resolved to %c%s",
6193 funny, dest, funny, dest);
6198 s = bracket; /* let the parser handle it */
6202 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
6203 PL_lex_state = LEX_INTERPEND;
6208 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
6213 *pmfl |= PMf_GLOBAL;
6215 *pmfl |= PMf_CONTINUE;
6219 *pmfl |= PMf_MULTILINE;
6221 *pmfl |= PMf_SINGLELINE;
6223 *pmfl |= PMf_EXTENDED;
6227 S_scan_pat(pTHX_ char *start, I32 type)
6232 s = scan_str(start,FALSE,FALSE);
6234 Perl_croak(aTHX_ "Search pattern not terminated");
6236 pm = (PMOP*)newPMOP(type, 0);
6237 if (PL_multi_open == '?')
6238 pm->op_pmflags |= PMf_ONCE;
6240 while (*s && strchr("iomsx", *s))
6241 pmflag(&pm->op_pmflags,*s++);
6244 while (*s && strchr("iogcmsx", *s))
6245 pmflag(&pm->op_pmflags,*s++);
6247 pm->op_pmpermflags = pm->op_pmflags;
6249 PL_lex_op = (OP*)pm;
6250 yylval.ival = OP_MATCH;
6255 S_scan_subst(pTHX_ char *start)
6262 yylval.ival = OP_NULL;
6264 s = scan_str(start,FALSE,FALSE);
6267 Perl_croak(aTHX_ "Substitution pattern not terminated");
6269 if (s[-1] == PL_multi_open)
6272 first_start = PL_multi_start;
6273 s = scan_str(s,FALSE,FALSE);
6276 SvREFCNT_dec(PL_lex_stuff);
6277 PL_lex_stuff = Nullsv;
6279 Perl_croak(aTHX_ "Substitution replacement not terminated");
6281 PL_multi_start = first_start; /* so whole substitution is taken together */
6283 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6289 else if (strchr("iogcmsx", *s))
6290 pmflag(&pm->op_pmflags,*s++);
6297 PL_sublex_info.super_bufptr = s;
6298 PL_sublex_info.super_bufend = PL_bufend;
6300 pm->op_pmflags |= PMf_EVAL;
6301 repl = newSVpvn("",0);
6303 sv_catpv(repl, es ? "eval " : "do ");
6304 sv_catpvn(repl, "{ ", 2);
6305 sv_catsv(repl, PL_lex_repl);
6306 sv_catpvn(repl, " };", 2);
6308 SvREFCNT_dec(PL_lex_repl);
6312 pm->op_pmpermflags = pm->op_pmflags;
6313 PL_lex_op = (OP*)pm;
6314 yylval.ival = OP_SUBST;
6319 S_scan_trans(pTHX_ char *start)
6328 yylval.ival = OP_NULL;
6330 s = scan_str(start,FALSE,FALSE);
6332 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6333 if (s[-1] == PL_multi_open)
6336 s = scan_str(s,FALSE,FALSE);
6339 SvREFCNT_dec(PL_lex_stuff);
6340 PL_lex_stuff = Nullsv;
6342 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6345 complement = del = squash = 0;
6346 while (strchr("cds", *s)) {
6348 complement = OPpTRANS_COMPLEMENT;
6350 del = OPpTRANS_DELETE;
6352 squash = OPpTRANS_SQUASH;
6356 New(803, tbl, complement&&!del?258:256, short);
6357 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6358 o->op_private = del|squash|complement|
6359 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
6360 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
6363 yylval.ival = OP_TRANS;
6368 S_scan_heredoc(pTHX_ register char *s)
6371 I32 op_type = OP_SCALAR;
6378 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6382 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6385 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6386 if (*peek && strchr("`'\"",*peek)) {
6389 s = delimcpy(d, e, s, PL_bufend, term, &len);
6399 if (!isALNUM_lazy_if(s,UTF))
6400 deprecate("bare << to mean <<\"\"");
6401 for (; isALNUM_lazy_if(s,UTF); s++) {
6406 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6407 Perl_croak(aTHX_ "Delimiter for here document is too long");
6410 len = d - PL_tokenbuf;
6411 #ifndef PERL_STRICT_CR
6412 d = strchr(s, '\r');
6416 while (s < PL_bufend) {
6422 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6431 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6436 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6437 herewas = newSVpvn(s,PL_bufend-s);
6439 s--, herewas = newSVpvn(s,d-s);
6440 s += SvCUR(herewas);
6442 tmpstr = NEWSV(87,79);
6443 sv_upgrade(tmpstr, SVt_PVIV);
6448 else if (term == '`') {
6449 op_type = OP_BACKTICK;
6450 SvIVX(tmpstr) = '\\';
6454 PL_multi_start = CopLINE(PL_curcop);
6455 PL_multi_open = PL_multi_close = '<';
6456 term = *PL_tokenbuf;
6457 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6458 char *bufptr = PL_sublex_info.super_bufptr;
6459 char *bufend = PL_sublex_info.super_bufend;
6460 char *olds = s - SvCUR(herewas);
6461 s = strchr(bufptr, '\n');
6465 while (s < bufend &&
6466 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6468 CopLINE_inc(PL_curcop);
6471 CopLINE_set(PL_curcop, PL_multi_start);
6472 missingterm(PL_tokenbuf);
6474 sv_setpvn(herewas,bufptr,d-bufptr+1);
6475 sv_setpvn(tmpstr,d+1,s-d);
6477 sv_catpvn(herewas,s,bufend-s);
6478 (void)strcpy(bufptr,SvPVX(herewas));
6485 while (s < PL_bufend &&
6486 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6488 CopLINE_inc(PL_curcop);
6490 if (s >= PL_bufend) {
6491 CopLINE_set(PL_curcop, PL_multi_start);
6492 missingterm(PL_tokenbuf);
6494 sv_setpvn(tmpstr,d+1,s-d);
6496 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6498 sv_catpvn(herewas,s,PL_bufend-s);
6499 sv_setsv(PL_linestr,herewas);
6500 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6501 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6502 PL_last_lop = PL_last_uni = Nullch;
6505 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6506 while (s >= PL_bufend) { /* multiple line string? */
6508 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6509 CopLINE_set(PL_curcop, PL_multi_start);
6510 missingterm(PL_tokenbuf);
6512 CopLINE_inc(PL_curcop);
6513 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6514 PL_last_lop = PL_last_uni = Nullch;
6515 #ifndef PERL_STRICT_CR
6516 if (PL_bufend - PL_linestart >= 2) {
6517 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6518 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6520 PL_bufend[-2] = '\n';
6522 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6524 else if (PL_bufend[-1] == '\r')
6525 PL_bufend[-1] = '\n';
6527 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6528 PL_bufend[-1] = '\n';
6530 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6531 SV *sv = NEWSV(88,0);
6533 sv_upgrade(sv, SVt_PVMG);
6534 sv_setsv(sv,PL_linestr);
6535 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6537 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6540 sv_catsv(PL_linestr,herewas);
6541 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6545 sv_catsv(tmpstr,PL_linestr);
6550 PL_multi_end = CopLINE(PL_curcop);
6551 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6552 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6553 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6555 SvREFCNT_dec(herewas);
6556 if (UTF && !IN_BYTES && is_utf8_string((U8*)SvPVX(tmpstr), SvCUR(tmpstr)))
6558 PL_lex_stuff = tmpstr;
6559 yylval.ival = op_type;
6564 takes: current position in input buffer
6565 returns: new position in input buffer
6566 side-effects: yylval and lex_op are set.
6571 <FH> read from filehandle
6572 <pkg::FH> read from package qualified filehandle
6573 <pkg'FH> read from package qualified filehandle
6574 <$fh> read from filehandle in $fh
6580 S_scan_inputsymbol(pTHX_ char *start)
6582 register char *s = start; /* current position in buffer */
6588 d = PL_tokenbuf; /* start of temp holding space */
6589 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6590 end = strchr(s, '\n');
6593 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6595 /* die if we didn't have space for the contents of the <>,
6596 or if it didn't end, or if we see a newline
6599 if (len >= sizeof PL_tokenbuf)
6600 Perl_croak(aTHX_ "Excessively long <> operator");
6602 Perl_croak(aTHX_ "Unterminated <> operator");
6607 Remember, only scalar variables are interpreted as filehandles by
6608 this code. Anything more complex (e.g., <$fh{$num}>) will be
6609 treated as a glob() call.
6610 This code makes use of the fact that except for the $ at the front,
6611 a scalar variable and a filehandle look the same.
6613 if (*d == '$' && d[1]) d++;
6615 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6616 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6619 /* If we've tried to read what we allow filehandles to look like, and
6620 there's still text left, then it must be a glob() and not a getline.
6621 Use scan_str to pull out the stuff between the <> and treat it
6622 as nothing more than a string.
6625 if (d - PL_tokenbuf != len) {
6626 yylval.ival = OP_GLOB;
6628 s = scan_str(start,FALSE,FALSE);
6630 Perl_croak(aTHX_ "Glob not terminated");
6634 /* we're in a filehandle read situation */
6637 /* turn <> into <ARGV> */
6639 (void)strcpy(d,"ARGV");
6641 /* if <$fh>, create the ops to turn the variable into a
6647 /* try to find it in the pad for this block, otherwise find
6648 add symbol table ops
6650 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6651 OP *o = newOP(OP_PADSV, 0);
6653 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6656 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6657 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6658 newUNOP(OP_RV2SV, 0,
6659 newGVOP(OP_GV, 0, gv)));
6661 PL_lex_op->op_flags |= OPf_SPECIAL;
6662 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6663 yylval.ival = OP_NULL;
6666 /* If it's none of the above, it must be a literal filehandle
6667 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6669 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6670 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6671 yylval.ival = OP_NULL;
6680 takes: start position in buffer
6681 keep_quoted preserve \ on the embedded delimiter(s)
6682 keep_delims preserve the delimiters around the string
6683 returns: position to continue reading from buffer
6684 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6685 updates the read buffer.
6687 This subroutine pulls a string out of the input. It is called for:
6688 q single quotes q(literal text)
6689 ' single quotes 'literal text'
6690 qq double quotes qq(interpolate $here please)
6691 " double quotes "interpolate $here please"
6692 qx backticks qx(/bin/ls -l)
6693 ` backticks `/bin/ls -l`
6694 qw quote words @EXPORT_OK = qw( func() $spam )
6695 m// regexp match m/this/
6696 s/// regexp substitute s/this/that/
6697 tr/// string transliterate tr/this/that/
6698 y/// string transliterate y/this/that/
6699 ($*@) sub prototypes sub foo ($)
6700 (stuff) sub attr parameters sub foo : attr(stuff)
6701 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6703 In most of these cases (all but <>, patterns and transliterate)
6704 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6705 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6706 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6709 It skips whitespace before the string starts, and treats the first
6710 character as the delimiter. If the delimiter is one of ([{< then
6711 the corresponding "close" character )]}> is used as the closing
6712 delimiter. It allows quoting of delimiters, and if the string has
6713 balanced delimiters ([{<>}]) it allows nesting.
6715 On success, the SV with the resulting string is put into lex_stuff or,
6716 if that is already non-NULL, into lex_repl. The second case occurs only
6717 when parsing the RHS of the special constructs s/// and tr/// (y///).
6718 For convenience, the terminating delimiter character is stuffed into
6723 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6725 SV *sv; /* scalar value: string */
6726 char *tmps; /* temp string, used for delimiter matching */
6727 register char *s = start; /* current position in the buffer */
6728 register char term; /* terminating character */
6729 register char *to; /* current position in the sv's data */
6730 I32 brackets = 1; /* bracket nesting level */
6731 bool has_utf8 = FALSE; /* is there any utf8 content? */
6733 /* skip space before the delimiter */
6737 /* mark where we are, in case we need to report errors */
6740 /* after skipping whitespace, the next character is the terminator */
6742 if (!UTF8_IS_INVARIANT((U8)term) && UTF)
6745 /* mark where we are */
6746 PL_multi_start = CopLINE(PL_curcop);
6747 PL_multi_open = term;
6749 /* find corresponding closing delimiter */
6750 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6752 PL_multi_close = term;
6754 /* create a new SV to hold the contents. 87 is leak category, I'm
6755 assuming. 79 is the SV's initial length. What a random number. */
6757 sv_upgrade(sv, SVt_PVIV);
6759 (void)SvPOK_only(sv); /* validate pointer */
6761 /* move past delimiter and try to read a complete string */
6763 sv_catpvn(sv, s, 1);
6766 /* extend sv if need be */
6767 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6768 /* set 'to' to the next character in the sv's string */
6769 to = SvPVX(sv)+SvCUR(sv);
6771 /* if open delimiter is the close delimiter read unbridle */
6772 if (PL_multi_open == PL_multi_close) {
6773 for (; s < PL_bufend; s++,to++) {
6774 /* embedded newlines increment the current line number */
6775 if (*s == '\n' && !PL_rsfp)
6776 CopLINE_inc(PL_curcop);
6777 /* handle quoted delimiters */
6778 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6779 if (!keep_quoted && s[1] == term)
6781 /* any other quotes are simply copied straight through */
6785 /* terminate when run out of buffer (the for() condition), or
6786 have found the terminator */
6787 else if (*s == term)
6789 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
6795 /* if the terminator isn't the same as the start character (e.g.,
6796 matched brackets), we have to allow more in the quoting, and
6797 be prepared for nested brackets.
6800 /* read until we run out of string, or we find the terminator */
6801 for (; s < PL_bufend; s++,to++) {
6802 /* embedded newlines increment the line count */
6803 if (*s == '\n' && !PL_rsfp)
6804 CopLINE_inc(PL_curcop);
6805 /* backslashes can escape the open or closing characters */
6806 if (*s == '\\' && s+1 < PL_bufend) {
6808 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6813 /* allow nested opens and closes */
6814 else if (*s == PL_multi_close && --brackets <= 0)
6816 else if (*s == PL_multi_open)
6818 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
6823 /* terminate the copied string and update the sv's end-of-string */
6825 SvCUR_set(sv, to - SvPVX(sv));
6828 * this next chunk reads more into the buffer if we're not done yet
6832 break; /* handle case where we are done yet :-) */
6834 #ifndef PERL_STRICT_CR
6835 if (to - SvPVX(sv) >= 2) {
6836 if ((to[-2] == '\r' && to[-1] == '\n') ||
6837 (to[-2] == '\n' && to[-1] == '\r'))
6841 SvCUR_set(sv, to - SvPVX(sv));
6843 else if (to[-1] == '\r')
6846 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6850 /* if we're out of file, or a read fails, bail and reset the current
6851 line marker so we can report where the unterminated string began
6854 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6856 CopLINE_set(PL_curcop, PL_multi_start);
6859 /* we read a line, so increment our line counter */
6860 CopLINE_inc(PL_curcop);
6862 /* update debugger info */
6863 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6864 SV *sv = NEWSV(88,0);
6866 sv_upgrade(sv, SVt_PVMG);
6867 sv_setsv(sv,PL_linestr);
6868 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6871 /* having changed the buffer, we must update PL_bufend */
6872 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6873 PL_last_lop = PL_last_uni = Nullch;
6876 /* at this point, we have successfully read the delimited string */
6879 sv_catpvn(sv, s, 1);
6882 PL_multi_end = CopLINE(PL_curcop);
6885 /* if we allocated too much space, give some back */
6886 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6887 SvLEN_set(sv, SvCUR(sv) + 1);
6888 Renew(SvPVX(sv), SvLEN(sv), char);
6891 /* decide whether this is the first or second quoted string we've read
6904 takes: pointer to position in buffer
6905 returns: pointer to new position in buffer
6906 side-effects: builds ops for the constant in yylval.op
6908 Read a number in any of the formats that Perl accepts:
6910 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
6911 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
6914 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
6916 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6919 If it reads a number without a decimal point or an exponent, it will
6920 try converting the number to an integer and see if it can do so
6921 without loss of precision.
6925 Perl_scan_num(pTHX_ char *start, YYSTYPE* lvalp)
6927 register char *s = start; /* current position in buffer */
6928 register char *d; /* destination in temp buffer */
6929 register char *e; /* end of temp buffer */
6930 NV nv; /* number read, as a double */
6931 SV *sv = Nullsv; /* place to put the converted number */
6932 bool floatit; /* boolean: int or float? */
6933 char *lastub = 0; /* position of last underbar */
6934 static char number_too_long[] = "Number too long";
6936 /* We use the first character to decide what type of number this is */
6940 Perl_croak(aTHX_ "panic: scan_num");
6942 /* if it starts with a 0, it could be an octal number, a decimal in
6943 0.13 disguise, or a hexadecimal number, or a binary number. */
6947 u holds the "number so far"
6948 shift the power of 2 of the base
6949 (hex == 4, octal == 3, binary == 1)
6950 overflowed was the number more than we can hold?
6952 Shift is used when we add a digit. It also serves as an "are
6953 we in octal/hex/binary?" indicator to disallow hex characters
6959 bool overflowed = FALSE;
6960 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6961 static char* bases[5] = { "", "binary", "", "octal",
6963 static char* Bases[5] = { "", "Binary", "", "Octal",
6965 static char *maxima[5] = { "",
6966 "0b11111111111111111111111111111111",
6970 char *base, *Base, *max;
6976 } else if (s[1] == 'b') {
6980 /* check for a decimal in disguise */
6981 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6983 /* so it must be octal */
6990 if (ckWARN(WARN_SYNTAX))
6991 Perl_warner(aTHX_ WARN_SYNTAX,
6992 "Misplaced _ in number");
6996 base = bases[shift];
6997 Base = Bases[shift];
6998 max = maxima[shift];
7000 /* read the rest of the number */
7002 /* x is used in the overflow test,
7003 b is the digit we're adding on. */
7008 /* if we don't mention it, we're done */
7012 /* _ are ignored -- but warned about if consecutive */
7014 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7015 Perl_warner(aTHX_ WARN_SYNTAX,
7016 "Misplaced _ in number");
7020 /* 8 and 9 are not octal */
7023 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
7027 case '2': case '3': case '4':
7028 case '5': case '6': case '7':
7030 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
7034 b = *s++ & 15; /* ASCII digit -> value of digit */
7038 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
7039 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
7040 /* make sure they said 0x */
7045 /* Prepare to put the digit we have onto the end
7046 of the number so far. We check for overflows.
7051 x = u << shift; /* make room for the digit */
7053 if ((x >> shift) != u
7054 && !(PL_hints & HINT_NEW_BINARY)) {
7057 if (ckWARN_d(WARN_OVERFLOW))
7058 Perl_warner(aTHX_ WARN_OVERFLOW,
7059 "Integer overflow in %s number",
7062 u = x | b; /* add the digit to the end */
7065 n *= nvshift[shift];
7066 /* If an NV has not enough bits in its
7067 * mantissa to represent an UV this summing of
7068 * small low-order numbers is a waste of time
7069 * (because the NV cannot preserve the
7070 * low-order bits anyway): we could just
7071 * remember when did we overflow and in the
7072 * end just multiply n by the right
7080 /* if we get here, we had success: make a scalar value from
7085 /* final misplaced underbar check */
7087 if (ckWARN(WARN_SYNTAX))
7088 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
7093 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
7094 Perl_warner(aTHX_ WARN_PORTABLE,
7095 "%s number > %s non-portable",
7101 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
7102 Perl_warner(aTHX_ WARN_PORTABLE,
7103 "%s number > %s non-portable",
7108 if (PL_hints & HINT_NEW_BINARY)
7109 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
7114 handle decimal numbers.
7115 we're also sent here when we read a 0 as the first digit
7117 case '1': case '2': case '3': case '4': case '5':
7118 case '6': case '7': case '8': case '9': case '.':
7121 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
7124 /* read next group of digits and _ and copy into d */
7125 while (isDIGIT(*s) || *s == '_') {
7126 /* skip underscores, checking for misplaced ones
7130 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7131 Perl_warner(aTHX_ WARN_SYNTAX,
7132 "Misplaced _ in number");
7136 /* check for end of fixed-length buffer */
7138 Perl_croak(aTHX_ number_too_long);
7139 /* if we're ok, copy the character */
7144 /* final misplaced underbar check */
7145 if (lastub && s == lastub + 1) {
7146 if (ckWARN(WARN_SYNTAX))
7147 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
7150 /* read a decimal portion if there is one. avoid
7151 3..5 being interpreted as the number 3. followed
7154 if (*s == '.' && s[1] != '.') {
7159 if (ckWARN(WARN_SYNTAX))
7160 Perl_warner(aTHX_ WARN_SYNTAX,
7161 "Misplaced _ in number");
7165 /* copy, ignoring underbars, until we run out of digits.
7167 for (; isDIGIT(*s) || *s == '_'; s++) {
7168 /* fixed length buffer check */
7170 Perl_croak(aTHX_ number_too_long);
7172 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7173 Perl_warner(aTHX_ WARN_SYNTAX,
7174 "Misplaced _ in number");
7180 /* fractional part ending in underbar? */
7182 if (ckWARN(WARN_SYNTAX))
7183 Perl_warner(aTHX_ WARN_SYNTAX,
7184 "Misplaced _ in number");
7186 if (*s == '.' && isDIGIT(s[1])) {
7187 /* oops, it's really a v-string, but without the "v" */
7193 /* read exponent part, if present */
7194 if (*s && strchr("eE",*s) && strchr("+-0123456789_", s[1])) {
7198 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
7199 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
7201 /* stray preinitial _ */
7203 if (ckWARN(WARN_SYNTAX))
7204 Perl_warner(aTHX_ WARN_SYNTAX,
7205 "Misplaced _ in number");
7209 /* allow positive or negative exponent */
7210 if (*s == '+' || *s == '-')
7213 /* stray initial _ */
7215 if (ckWARN(WARN_SYNTAX))
7216 Perl_warner(aTHX_ WARN_SYNTAX,
7217 "Misplaced _ in number");
7221 /* read digits of exponent */
7222 while (isDIGIT(*s) || *s == '_') {
7225 Perl_croak(aTHX_ number_too_long);
7229 if (ckWARN(WARN_SYNTAX) &&
7230 ((lastub && s == lastub + 1) ||
7231 (!isDIGIT(s[1]) && s[1] != '_')))
7232 Perl_warner(aTHX_ WARN_SYNTAX,
7233 "Misplaced _ in number");
7240 /* make an sv from the string */
7244 We try to do an integer conversion first if no characters
7245 indicating "float" have been found.
7250 int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
7252 if (flags == IS_NUMBER_IN_UV) {
7254 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7257 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
7258 if (uv <= (UV) IV_MIN)
7259 sv_setiv(sv, -(IV)uv);
7266 /* terminate the string */
7268 nv = Atof(PL_tokenbuf);
7272 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7273 (PL_hints & HINT_NEW_INTEGER) )
7274 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7275 (floatit ? "float" : "integer"),
7279 /* if it starts with a v, it could be a v-string */
7285 while (isDIGIT(*pos) || *pos == '_')
7287 if (!isALPHA(*pos)) {
7289 U8 tmpbuf[UTF8_MAXLEN+1];
7291 s++; /* get past 'v' */
7294 sv_setpvn(sv, "", 0);
7297 if (*s == '0' && isDIGIT(s[1]))
7298 yyerror("Octal number in vector unsupported");
7301 /* this is atoi() that tolerates underscores */
7304 while (--end >= s) {
7309 rev += (*end - '0') * mult;
7311 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
7312 Perl_warner(aTHX_ WARN_OVERFLOW,
7313 "Integer overflow in decimal number");
7316 /* Append native character for the rev point */
7317 tmpend = uvchr_to_utf8(tmpbuf, rev);
7318 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
7319 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev)))
7321 if (*pos == '.' && isDIGIT(pos[1]))
7327 while (isDIGIT(*pos) || *pos == '_')
7337 /* make the op for the constant and return */
7340 lvalp->opval = newSVOP(OP_CONST, 0, sv);
7342 lvalp->opval = Nullop;
7348 S_scan_formline(pTHX_ register char *s)
7352 SV *stuff = newSVpvn("",0);
7353 bool needargs = FALSE;
7356 if (*s == '.' || *s == /*{*/'}') {
7358 #ifdef PERL_STRICT_CR
7359 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7361 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7363 if (*t == '\n' || t == PL_bufend)
7366 if (PL_in_eval && !PL_rsfp) {
7367 eol = strchr(s,'\n');
7372 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7374 for (t = s; t < eol; t++) {
7375 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7377 goto enough; /* ~~ must be first line in formline */
7379 if (*t == '@' || *t == '^')
7383 sv_catpvn(stuff, s, eol-s);
7384 #ifndef PERL_STRICT_CR
7385 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7386 char *end = SvPVX(stuff) + SvCUR(stuff);
7398 s = filter_gets(PL_linestr, PL_rsfp, 0);
7399 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7400 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7401 PL_last_lop = PL_last_uni = Nullch;
7404 yyerror("Format not terminated");
7414 PL_lex_state = LEX_NORMAL;
7415 PL_nextval[PL_nexttoke].ival = 0;
7419 PL_lex_state = LEX_FORMLINE;
7420 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7422 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7426 SvREFCNT_dec(stuff);
7427 PL_lex_formbrack = 0;
7438 PL_cshlen = strlen(PL_cshname);
7443 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7445 I32 oldsavestack_ix = PL_savestack_ix;
7446 CV* outsidecv = PL_compcv;
7450 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7452 SAVEI32(PL_subline);
7453 save_item(PL_subname);
7456 SAVESPTR(PL_comppad_name);
7457 SAVESPTR(PL_compcv);
7458 SAVEI32(PL_comppad_name_fill);
7459 SAVEI32(PL_min_intro_pending);
7460 SAVEI32(PL_max_intro_pending);
7461 SAVEI32(PL_pad_reset_pending);
7463 PL_compcv = (CV*)NEWSV(1104,0);
7464 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7465 CvFLAGS(PL_compcv) |= flags;
7467 PL_comppad = newAV();
7468 av_push(PL_comppad, Nullsv);
7469 PL_curpad = AvARRAY(PL_comppad);
7470 PL_comppad_name = newAV();
7471 PL_comppad_name_fill = 0;
7472 PL_min_intro_pending = 0;
7474 PL_subline = CopLINE(PL_curcop);
7475 #ifdef USE_5005THREADS
7476 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7477 PL_curpad[0] = (SV*)newAV();
7478 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7479 #endif /* USE_5005THREADS */
7481 comppadlist = newAV();
7482 AvREAL_off(comppadlist);
7483 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7484 av_store(comppadlist, 1, (SV*)PL_comppad);
7486 CvPADLIST(PL_compcv) = comppadlist;
7487 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7488 #ifdef USE_5005THREADS
7489 CvOWNER(PL_compcv) = 0;
7490 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7491 MUTEX_INIT(CvMUTEXP(PL_compcv));
7492 #endif /* USE_5005THREADS */
7494 return oldsavestack_ix;
7498 #pragma segment Perl_yylex
7501 Perl_yywarn(pTHX_ char *s)
7503 PL_in_eval |= EVAL_WARNONLY;
7505 PL_in_eval &= ~EVAL_WARNONLY;
7510 Perl_yyerror(pTHX_ char *s)
7513 char *context = NULL;
7517 if (!yychar || (yychar == ';' && !PL_rsfp))
7519 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7520 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7521 while (isSPACE(*PL_oldoldbufptr))
7523 context = PL_oldoldbufptr;
7524 contlen = PL_bufptr - PL_oldoldbufptr;
7526 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7527 PL_oldbufptr != PL_bufptr) {
7528 while (isSPACE(*PL_oldbufptr))
7530 context = PL_oldbufptr;
7531 contlen = PL_bufptr - PL_oldbufptr;
7533 else if (yychar > 255)
7534 where = "next token ???";
7535 #ifdef USE_PURE_BISON
7536 /* GNU Bison sets the value -2 */
7537 else if (yychar == -2) {
7539 else if ((yychar & 127) == 127) {
7541 if (PL_lex_state == LEX_NORMAL ||
7542 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7543 where = "at end of line";
7544 else if (PL_lex_inpat)
7545 where = "within pattern";
7547 where = "within string";
7550 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7552 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7553 else if (isPRINT_LC(yychar))
7554 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7556 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7557 where = SvPVX(where_sv);
7559 msg = sv_2mortal(newSVpv(s, 0));
7560 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7561 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7563 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7565 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7566 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7567 Perl_sv_catpvf(aTHX_ msg,
7568 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7569 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7572 if (PL_in_eval & EVAL_WARNONLY)
7573 Perl_warn(aTHX_ "%"SVf, msg);
7576 if (PL_error_count >= 10) {
7577 if (PL_in_eval && SvCUR(ERRSV))
7578 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
7579 ERRSV, CopFILE(PL_curcop));
7581 Perl_croak(aTHX_ "%s has too many errors.\n",
7582 CopFILE(PL_curcop));
7585 PL_in_my_stash = Nullhv;
7589 #pragma segment Main
7593 S_swallow_bom(pTHX_ U8 *s)
7596 slen = SvCUR(PL_linestr);
7600 /* UTF-16 little-endian */
7601 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
7602 Perl_croak(aTHX_ "Unsupported script encoding");
7603 #ifndef PERL_NO_UTF16_FILTER
7604 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-LE script encoding\n"));
7606 if (PL_bufend > (char*)s) {
7610 filter_add(utf16rev_textfilter, NULL);
7611 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7612 PL_bufend = (char*)utf16_to_utf8_reversed(s, news,
7613 PL_bufend - (char*)s - 1,
7615 Copy(news, s, newlen, U8);
7616 SvCUR_set(PL_linestr, newlen);
7617 PL_bufend = SvPVX(PL_linestr) + newlen;
7618 news[newlen++] = '\0';
7622 Perl_croak(aTHX_ "Unsupported script encoding");
7627 if (s[1] == 0xFF) { /* UTF-16 big-endian */
7628 #ifndef PERL_NO_UTF16_FILTER
7629 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding\n"));
7631 if (PL_bufend > (char *)s) {
7635 filter_add(utf16_textfilter, NULL);
7636 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7637 PL_bufend = (char*)utf16_to_utf8(s, news,
7638 PL_bufend - (char*)s,
7640 Copy(news, s, newlen, U8);
7641 SvCUR_set(PL_linestr, newlen);
7642 PL_bufend = SvPVX(PL_linestr) + newlen;
7643 news[newlen++] = '\0';
7647 Perl_croak(aTHX_ "Unsupported script encoding");
7652 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
7653 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-8 script encoding\n"));
7658 if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */
7659 s[2] == 0xFE && s[3] == 0xFF)
7661 Perl_croak(aTHX_ "Unsupported script encoding");
7669 * Restore a source filter.
7673 restore_rsfp(pTHX_ void *f)
7675 PerlIO *fp = (PerlIO*)f;
7677 if (PL_rsfp == PerlIO_stdin())
7678 PerlIO_clearerr(PL_rsfp);
7679 else if (PL_rsfp && (PL_rsfp != fp))
7680 PerlIO_close(PL_rsfp);
7684 #ifndef PERL_NO_UTF16_FILTER
7686 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
7688 I32 count = FILTER_READ(idx+1, sv, maxlen);
7693 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7694 if (!*SvPV_nolen(sv))
7695 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7698 tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7699 sv_usepvn(sv, (char*)tmps, tend - tmps);
7705 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
7707 I32 count = FILTER_READ(idx+1, sv, maxlen);
7712 if (!*SvPV_nolen(sv))
7713 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7716 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7717 tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7718 sv_usepvn(sv, (char*)tmps, tend - tmps);