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;
453 * Finalizer for lexing operations. Must be called when the parser is
454 * done with the lexer.
460 PL_doextract = FALSE;
465 * This subroutine has nothing to do with tilting, whether at windmills
466 * or pinball tables. Its name is short for "increment line". It
467 * increments the current line number in CopLINE(PL_curcop) and checks
468 * to see whether the line starts with a comment of the form
469 * # line 500 "foo.pm"
470 * If so, it sets the current line number and file to the values in the comment.
474 S_incline(pTHX_ char *s)
481 CopLINE_inc(PL_curcop);
484 while (SPACE_OR_TAB(*s)) s++;
485 if (strnEQ(s, "line", 4))
489 if (SPACE_OR_TAB(*s))
493 while (SPACE_OR_TAB(*s)) s++;
499 while (SPACE_OR_TAB(*s))
501 if (*s == '"' && (t = strchr(s+1, '"'))) {
506 for (t = s; !isSPACE(*t); t++) ;
509 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
511 if (*e != '\n' && *e != '\0')
512 return; /* false alarm */
518 Safefree(CopFILE(PL_curcop));
520 SvREFCNT_dec(CopFILEGV(PL_curcop));
522 CopFILE_set(PL_curcop, s);
525 CopLINE_set(PL_curcop, atoi(n)-1);
530 * Called to gobble the appropriate amount and type of whitespace.
531 * Skips comments as well.
535 S_skipspace(pTHX_ register char *s)
537 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
538 while (s < PL_bufend && SPACE_OR_TAB(*s))
544 SSize_t oldprevlen, oldoldprevlen;
545 SSize_t oldloplen = 0, oldunilen = 0;
546 while (s < PL_bufend && isSPACE(*s)) {
547 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
552 if (s < PL_bufend && *s == '#') {
553 while (s < PL_bufend && *s != '\n')
557 if (PL_in_eval && !PL_rsfp) {
564 /* only continue to recharge the buffer if we're at the end
565 * of the buffer, we're not reading from a source filter, and
566 * we're in normal lexing mode
568 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
569 PL_lex_state == LEX_FORMLINE)
572 /* try to recharge the buffer */
573 if ((s = filter_gets(PL_linestr, PL_rsfp,
574 (prevlen = SvCUR(PL_linestr)))) == Nullch)
576 /* end of file. Add on the -p or -n magic */
577 if (PL_minus_n || PL_minus_p) {
578 sv_setpv(PL_linestr,PL_minus_p ?
579 ";}continue{print or die qq(-p destination: $!\\n)" :
581 sv_catpv(PL_linestr,";}");
582 PL_minus_n = PL_minus_p = 0;
585 sv_setpv(PL_linestr,";");
587 /* reset variables for next time we lex */
588 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
590 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
591 PL_last_lop = PL_last_uni = Nullch;
593 /* Close the filehandle. Could be from -P preprocessor,
594 * STDIN, or a regular file. If we were reading code from
595 * STDIN (because the commandline held no -e or filename)
596 * then we don't close it, we reset it so the code can
597 * read from STDIN too.
600 if (PL_preprocess && !PL_in_eval)
601 (void)PerlProc_pclose(PL_rsfp);
602 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
603 PerlIO_clearerr(PL_rsfp);
605 (void)PerlIO_close(PL_rsfp);
610 /* not at end of file, so we only read another line */
611 /* make corresponding updates to old pointers, for yyerror() */
612 oldprevlen = PL_oldbufptr - PL_bufend;
613 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
615 oldunilen = PL_last_uni - PL_bufend;
617 oldloplen = PL_last_lop - PL_bufend;
618 PL_linestart = PL_bufptr = s + prevlen;
619 PL_bufend = s + SvCUR(PL_linestr);
621 PL_oldbufptr = s + oldprevlen;
622 PL_oldoldbufptr = s + oldoldprevlen;
624 PL_last_uni = s + oldunilen;
626 PL_last_lop = s + oldloplen;
629 /* debugger active and we're not compiling the debugger code,
630 * so store the line into the debugger's array of lines
632 if (PERLDB_LINE && PL_curstash != PL_debstash) {
633 SV *sv = NEWSV(85,0);
635 sv_upgrade(sv, SVt_PVMG);
636 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) { /* I _guess_ 4 is \N{} --jhi */
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");
1658 if (PL_encoding && !has_utf8) {
1659 Perl_sv_recode_to_utf8(aTHX_ sv, PL_encoding);
1664 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1665 PL_sublex_info.sub_op->op_private |=
1666 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1670 /* shrink the sv if we allocated more than we used */
1671 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1672 SvLEN_set(sv, SvCUR(sv) + 1);
1673 Renew(SvPVX(sv), SvLEN(sv), char);
1676 /* return the substring (via yylval) only if we parsed anything */
1677 if (s > PL_bufptr) {
1678 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1679 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1681 ( PL_lex_inwhat == OP_TRANS
1683 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1686 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1693 * Returns TRUE if there's more to the expression (e.g., a subscript),
1696 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1698 * ->[ and ->{ return TRUE
1699 * { and [ outside a pattern are always subscripts, so return TRUE
1700 * if we're outside a pattern and it's not { or [, then return FALSE
1701 * if we're in a pattern and the first char is a {
1702 * {4,5} (any digits around the comma) returns FALSE
1703 * if we're in a pattern and the first char is a [
1705 * [SOMETHING] has a funky algorithm to decide whether it's a
1706 * character class or not. It has to deal with things like
1707 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1708 * anything else returns TRUE
1711 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1714 S_intuit_more(pTHX_ register char *s)
1716 if (PL_lex_brackets)
1718 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1720 if (*s != '{' && *s != '[')
1725 /* In a pattern, so maybe we have {n,m}. */
1742 /* On the other hand, maybe we have a character class */
1745 if (*s == ']' || *s == '^')
1748 /* this is terrifying, and it works */
1749 int weight = 2; /* let's weigh the evidence */
1751 unsigned char un_char = 255, last_un_char;
1752 char *send = strchr(s,']');
1753 char tmpbuf[sizeof PL_tokenbuf * 4];
1755 if (!send) /* has to be an expression */
1758 Zero(seen,256,char);
1761 else if (isDIGIT(*s)) {
1763 if (isDIGIT(s[1]) && s[2] == ']')
1769 for (; s < send; s++) {
1770 last_un_char = un_char;
1771 un_char = (unsigned char)*s;
1776 weight -= seen[un_char] * 10;
1777 if (isALNUM_lazy_if(s+1,UTF)) {
1778 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1779 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1784 else if (*s == '$' && s[1] &&
1785 strchr("[#!%*<>()-=",s[1])) {
1786 if (/*{*/ strchr("])} =",s[2]))
1795 if (strchr("wds]",s[1]))
1797 else if (seen['\''] || seen['"'])
1799 else if (strchr("rnftbxcav",s[1]))
1801 else if (isDIGIT(s[1])) {
1803 while (s[1] && isDIGIT(s[1]))
1813 if (strchr("aA01! ",last_un_char))
1815 if (strchr("zZ79~",s[1]))
1817 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1818 weight -= 5; /* cope with negative subscript */
1821 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1822 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1827 if (keyword(tmpbuf, d - tmpbuf))
1830 if (un_char == last_un_char + 1)
1832 weight -= seen[un_char];
1837 if (weight >= 0) /* probably a character class */
1847 * Does all the checking to disambiguate
1849 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1850 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1852 * First argument is the stuff after the first token, e.g. "bar".
1854 * Not a method if bar is a filehandle.
1855 * Not a method if foo is a subroutine prototyped to take a filehandle.
1856 * Not a method if it's really "Foo $bar"
1857 * Method if it's "foo $bar"
1858 * Not a method if it's really "print foo $bar"
1859 * Method if it's really "foo package::" (interpreted as package->foo)
1860 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1861 * Not a method if bar is a filehandle or package, but is quoted with
1866 S_intuit_method(pTHX_ char *start, GV *gv)
1868 char *s = start + (*start == '$');
1869 char tmpbuf[sizeof PL_tokenbuf];
1877 if ((cv = GvCVu(gv))) {
1878 char *proto = SvPVX(cv);
1888 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1889 /* start is the beginning of the possible filehandle/object,
1890 * and s is the end of it
1891 * tmpbuf is a copy of it
1894 if (*start == '$') {
1895 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1900 return *s == '(' ? FUNCMETH : METHOD;
1902 if (!keyword(tmpbuf, len)) {
1903 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1908 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1909 if (indirgv && GvCVu(indirgv))
1911 /* filehandle or package name makes it a method */
1912 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1914 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1915 return 0; /* no assumptions -- "=>" quotes bearword */
1917 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1918 newSVpvn(tmpbuf,len));
1919 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1923 return *s == '(' ? FUNCMETH : METHOD;
1931 * Return a string of Perl code to load the debugger. If PERL5DB
1932 * is set, it will return the contents of that, otherwise a
1933 * compile-time require of perl5db.pl.
1940 char *pdb = PerlEnv_getenv("PERL5DB");
1944 SETERRNO(0,SS$_NORMAL);
1945 return "BEGIN { require 'perl5db.pl' }";
1951 /* Encoded script support. filter_add() effectively inserts a
1952 * 'pre-processing' function into the current source input stream.
1953 * Note that the filter function only applies to the current source file
1954 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1956 * The datasv parameter (which may be NULL) can be used to pass
1957 * private data to this instance of the filter. The filter function
1958 * can recover the SV using the FILTER_DATA macro and use it to
1959 * store private buffers and state information.
1961 * The supplied datasv parameter is upgraded to a PVIO type
1962 * and the IoDIRP/IoANY field is used to store the function pointer,
1963 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1964 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1965 * private use must be set using malloc'd pointers.
1969 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1974 if (!PL_rsfp_filters)
1975 PL_rsfp_filters = newAV();
1977 datasv = NEWSV(255,0);
1978 if (!SvUPGRADE(datasv, SVt_PVIO))
1979 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1980 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
1981 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1982 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1983 funcp, SvPV_nolen(datasv)));
1984 av_unshift(PL_rsfp_filters, 1);
1985 av_store(PL_rsfp_filters, 0, datasv) ;
1990 /* Delete most recently added instance of this filter function. */
1992 Perl_filter_del(pTHX_ filter_t funcp)
1995 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1996 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1998 /* if filter is on top of stack (usual case) just pop it off */
1999 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
2000 if (IoANY(datasv) == (void *)funcp) {
2001 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
2002 IoANY(datasv) = (void *)NULL;
2003 sv_free(av_pop(PL_rsfp_filters));
2007 /* we need to search for the correct entry and clear it */
2008 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
2012 /* Invoke the n'th filter function for the current rsfp. */
2014 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
2017 /* 0 = read one text line */
2022 if (!PL_rsfp_filters)
2024 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
2025 /* Provide a default input filter to make life easy. */
2026 /* Note that we append to the line. This is handy. */
2027 DEBUG_P(PerlIO_printf(Perl_debug_log,
2028 "filter_read %d: from rsfp\n", idx));
2032 int old_len = SvCUR(buf_sv) ;
2034 /* ensure buf_sv is large enough */
2035 SvGROW(buf_sv, old_len + maxlen) ;
2036 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
2037 if (PerlIO_error(PL_rsfp))
2038 return -1; /* error */
2040 return 0 ; /* end of file */
2042 SvCUR_set(buf_sv, old_len + len) ;
2045 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2046 if (PerlIO_error(PL_rsfp))
2047 return -1; /* error */
2049 return 0 ; /* end of file */
2052 return SvCUR(buf_sv);
2054 /* Skip this filter slot if filter has been deleted */
2055 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
2056 DEBUG_P(PerlIO_printf(Perl_debug_log,
2057 "filter_read %d: skipped (filter deleted)\n",
2059 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
2061 /* Get function pointer hidden within datasv */
2062 funcp = (filter_t)IoANY(datasv);
2063 DEBUG_P(PerlIO_printf(Perl_debug_log,
2064 "filter_read %d: via function %p (%s)\n",
2065 idx, funcp, SvPV_nolen(datasv)));
2066 /* Call function. The function is expected to */
2067 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2068 /* Return: <0:error, =0:eof, >0:not eof */
2069 return (*funcp)(aTHX_ idx, buf_sv, maxlen);
2073 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2075 #ifdef PERL_CR_FILTER
2076 if (!PL_rsfp_filters) {
2077 filter_add(S_cr_textfilter,NULL);
2080 if (PL_rsfp_filters) {
2083 SvCUR_set(sv, 0); /* start with empty line */
2084 if (FILTER_READ(0, sv, 0) > 0)
2085 return ( SvPVX(sv) ) ;
2090 return (sv_gets(sv, fp, append));
2094 S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2098 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2102 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2103 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2105 return GvHV(gv); /* Foo:: */
2108 /* use constant CLASS => 'MyClass' */
2109 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2111 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2112 pkgname = SvPV_nolen(sv);
2116 return gv_stashpv(pkgname, FALSE);
2120 static char* exp_name[] =
2121 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2122 "ATTRTERM", "TERMBLOCK"
2129 Works out what to call the token just pulled out of the input
2130 stream. The yacc parser takes care of taking the ops we return and
2131 stitching them into a tree.
2137 if read an identifier
2138 if we're in a my declaration
2139 croak if they tried to say my($foo::bar)
2140 build the ops for a my() declaration
2141 if it's an access to a my() variable
2142 are we in a sort block?
2143 croak if my($a); $a <=> $b
2144 build ops for access to a my() variable
2145 if in a dq string, and they've said @foo and we can't find @foo
2147 build ops for a bareword
2148 if we already built the token before, use it.
2151 #ifdef USE_PURE_BISON
2153 Perl_yylex_r(pTHX_ YYSTYPE *lvalp, int *lcharp)
2158 yylval_pointer[yyactlevel] = lvalp;
2159 yychar_pointer[yyactlevel] = lcharp;
2160 if (yyactlevel >= YYMAXLEVEL)
2161 Perl_croak(aTHX_ "panic: YYMAXLEVEL");
2163 r = Perl_yylex(aTHX);
2173 #pragma segment Perl_yylex
2186 /* check if there's an identifier for us to look at */
2187 if (PL_pending_ident)
2188 return S_pending_ident(aTHX);
2190 /* no identifier pending identification */
2192 switch (PL_lex_state) {
2194 case LEX_NORMAL: /* Some compilers will produce faster */
2195 case LEX_INTERPNORMAL: /* code if we comment these out. */
2199 /* when we've already built the next token, just pull it out of the queue */
2202 yylval = PL_nextval[PL_nexttoke];
2204 PL_lex_state = PL_lex_defer;
2205 PL_expect = PL_lex_expect;
2206 PL_lex_defer = LEX_NORMAL;
2208 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2209 "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr,
2210 (IV)PL_nexttype[PL_nexttoke]); });
2212 return(PL_nexttype[PL_nexttoke]);
2214 /* interpolated case modifiers like \L \U, including \Q and \E.
2215 when we get here, PL_bufptr is at the \
2217 case LEX_INTERPCASEMOD:
2219 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2220 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2222 /* handle \E or end of string */
2223 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2227 if (PL_lex_casemods) {
2228 oldmod = PL_lex_casestack[--PL_lex_casemods];
2229 PL_lex_casestack[PL_lex_casemods] = '\0';
2231 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2233 PL_lex_state = LEX_INTERPCONCAT;
2237 if (PL_bufptr != PL_bufend)
2239 PL_lex_state = LEX_INTERPCONCAT;
2243 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2244 "### Saw case modifier at '%s'\n", PL_bufptr); });
2246 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2247 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2248 if (strchr("LU", *s) &&
2249 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2251 PL_lex_casestack[--PL_lex_casemods] = '\0';
2254 if (PL_lex_casemods > 10) {
2255 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2256 if (newlb != PL_lex_casestack) {
2258 PL_lex_casestack = newlb;
2261 PL_lex_casestack[PL_lex_casemods++] = *s;
2262 PL_lex_casestack[PL_lex_casemods] = '\0';
2263 PL_lex_state = LEX_INTERPCONCAT;
2264 PL_nextval[PL_nexttoke].ival = 0;
2267 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2269 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2271 PL_nextval[PL_nexttoke].ival = OP_LC;
2273 PL_nextval[PL_nexttoke].ival = OP_UC;
2275 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2277 Perl_croak(aTHX_ "panic: yylex");
2280 if (PL_lex_starts) {
2289 case LEX_INTERPPUSH:
2290 return sublex_push();
2292 case LEX_INTERPSTART:
2293 if (PL_bufptr == PL_bufend)
2294 return sublex_done();
2295 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2296 "### Interpolated variable at '%s'\n", PL_bufptr); });
2298 PL_lex_dojoin = (*PL_bufptr == '@');
2299 PL_lex_state = LEX_INTERPNORMAL;
2300 if (PL_lex_dojoin) {
2301 PL_nextval[PL_nexttoke].ival = 0;
2303 #ifdef USE_5005THREADS
2304 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2305 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2306 force_next(PRIVATEREF);
2308 force_ident("\"", '$');
2309 #endif /* USE_5005THREADS */
2310 PL_nextval[PL_nexttoke].ival = 0;
2312 PL_nextval[PL_nexttoke].ival = 0;
2314 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2317 if (PL_lex_starts++) {
2323 case LEX_INTERPENDMAYBE:
2324 if (intuit_more(PL_bufptr)) {
2325 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2331 if (PL_lex_dojoin) {
2332 PL_lex_dojoin = FALSE;
2333 PL_lex_state = LEX_INTERPCONCAT;
2336 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2337 && SvEVALED(PL_lex_repl))
2339 if (PL_bufptr != PL_bufend)
2340 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2341 PL_lex_repl = Nullsv;
2344 case LEX_INTERPCONCAT:
2346 if (PL_lex_brackets)
2347 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2349 if (PL_bufptr == PL_bufend)
2350 return sublex_done();
2352 if (SvIVX(PL_linestr) == '\'') {
2353 SV *sv = newSVsv(PL_linestr);
2356 else if ( PL_hints & HINT_NEW_RE )
2357 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2358 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2362 s = scan_const(PL_bufptr);
2364 PL_lex_state = LEX_INTERPCASEMOD;
2366 PL_lex_state = LEX_INTERPSTART;
2369 if (s != PL_bufptr) {
2370 PL_nextval[PL_nexttoke] = yylval;
2373 if (PL_lex_starts++)
2383 PL_lex_state = LEX_NORMAL;
2384 s = scan_formline(PL_bufptr);
2385 if (!PL_lex_formbrack)
2391 PL_oldoldbufptr = PL_oldbufptr;
2394 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2395 exp_name[PL_expect], s);
2401 if (isIDFIRST_lazy_if(s,UTF))
2403 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2406 goto fake_eof; /* emulate EOF on ^D or ^Z */
2411 if (PL_lex_brackets)
2412 yyerror("Missing right curly or square bracket");
2413 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2414 "### Tokener got EOF\n");
2418 if (s++ < PL_bufend)
2419 goto retry; /* ignore stray nulls */
2422 if (!PL_in_eval && !PL_preambled) {
2423 PL_preambled = TRUE;
2424 sv_setpv(PL_linestr,incl_perldb());
2425 if (SvCUR(PL_linestr))
2426 sv_catpv(PL_linestr,";");
2428 while(AvFILLp(PL_preambleav) >= 0) {
2429 SV *tmpsv = av_shift(PL_preambleav);
2430 sv_catsv(PL_linestr, tmpsv);
2431 sv_catpv(PL_linestr, ";");
2434 sv_free((SV*)PL_preambleav);
2435 PL_preambleav = NULL;
2437 if (PL_minus_n || PL_minus_p) {
2438 sv_catpv(PL_linestr, "LINE: while (<>) {");
2440 sv_catpv(PL_linestr,"chomp;");
2443 if (strchr("/'\"", *PL_splitstr)
2444 && strchr(PL_splitstr + 1, *PL_splitstr))
2445 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2448 s = "'~#\200\1'"; /* surely one char is unused...*/
2449 while (s[1] && strchr(PL_splitstr, *s)) s++;
2451 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s%c",
2452 "q" + (delim == '\''), delim);
2453 for (s = PL_splitstr; *s; s++) {
2455 sv_catpvn(PL_linestr, "\\", 1);
2456 sv_catpvn(PL_linestr, s, 1);
2458 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2462 sv_catpv(PL_linestr,"our @F=split(' ');");
2465 sv_catpv(PL_linestr, "\n");
2466 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2467 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2468 PL_last_lop = PL_last_uni = Nullch;
2469 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2470 SV *sv = NEWSV(85,0);
2472 sv_upgrade(sv, SVt_PVMG);
2473 sv_setsv(sv,PL_linestr);
2476 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2481 bof = PL_rsfp ? TRUE : FALSE;
2482 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2485 if (PL_preprocess && !PL_in_eval)
2486 (void)PerlProc_pclose(PL_rsfp);
2487 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2488 PerlIO_clearerr(PL_rsfp);
2490 (void)PerlIO_close(PL_rsfp);
2492 PL_doextract = FALSE;
2494 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2495 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2496 sv_catpv(PL_linestr,";}");
2497 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2498 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2499 PL_last_lop = PL_last_uni = Nullch;
2500 PL_minus_n = PL_minus_p = 0;
2503 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2504 PL_last_lop = PL_last_uni = Nullch;
2505 sv_setpv(PL_linestr,"");
2506 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2508 /* if it looks like the start of a BOM, check if it in fact is */
2509 else if (bof && (!*s || *(U8*)s == 0xEF || *(U8*)s >= 0xFE)) {
2510 #ifdef PERLIO_IS_STDIO
2511 # ifdef __GNU_LIBRARY__
2512 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
2513 # define FTELL_FOR_PIPE_IS_BROKEN
2517 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2518 # define FTELL_FOR_PIPE_IS_BROKEN
2523 #ifdef FTELL_FOR_PIPE_IS_BROKEN
2524 /* This loses the possibility to detect the bof
2525 * situation on perl -P when the libc5 is being used.
2526 * Workaround? Maybe attach some extra state to PL_rsfp?
2529 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2531 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2534 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2535 s = swallow_bom((U8*)s);
2539 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2540 PL_doextract = FALSE;
2542 /* Incest with pod. */
2543 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2544 sv_setpv(PL_linestr, "");
2545 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2546 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2547 PL_last_lop = PL_last_uni = Nullch;
2548 PL_doextract = FALSE;
2552 } while (PL_doextract);
2553 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2554 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2555 SV *sv = NEWSV(85,0);
2557 sv_upgrade(sv, SVt_PVMG);
2558 sv_setsv(sv,PL_linestr);
2561 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2563 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2564 PL_last_lop = PL_last_uni = Nullch;
2565 if (CopLINE(PL_curcop) == 1) {
2566 while (s < PL_bufend && isSPACE(*s))
2568 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2572 if (*s == '#' && *(s+1) == '!')
2574 #ifdef ALTERNATE_SHEBANG
2576 static char as[] = ALTERNATE_SHEBANG;
2577 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2578 d = s + (sizeof(as) - 1);
2580 #endif /* ALTERNATE_SHEBANG */
2589 while (*d && !isSPACE(*d))
2593 #ifdef ARG_ZERO_IS_SCRIPT
2594 if (ipathend > ipath) {
2596 * HP-UX (at least) sets argv[0] to the script name,
2597 * which makes $^X incorrect. And Digital UNIX and Linux,
2598 * at least, set argv[0] to the basename of the Perl
2599 * interpreter. So, having found "#!", we'll set it right.
2601 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2602 assert(SvPOK(x) || SvGMAGICAL(x));
2603 if (sv_eq(x, CopFILESV(PL_curcop))) {
2604 sv_setpvn(x, ipath, ipathend - ipath);
2607 TAINT_NOT; /* $^X is always tainted, but that's OK */
2609 #endif /* ARG_ZERO_IS_SCRIPT */
2614 d = instr(s,"perl -");
2616 d = instr(s,"perl");
2618 /* avoid getting into infinite loops when shebang
2619 * line contains "Perl" rather than "perl" */
2621 for (d = ipathend-4; d >= ipath; --d) {
2622 if ((*d == 'p' || *d == 'P')
2623 && !ibcmp(d, "perl", 4))
2633 #ifdef ALTERNATE_SHEBANG
2635 * If the ALTERNATE_SHEBANG on this system starts with a
2636 * character that can be part of a Perl expression, then if
2637 * we see it but not "perl", we're probably looking at the
2638 * start of Perl code, not a request to hand off to some
2639 * other interpreter. Similarly, if "perl" is there, but
2640 * not in the first 'word' of the line, we assume the line
2641 * contains the start of the Perl program.
2643 if (d && *s != '#') {
2645 while (*c && !strchr("; \t\r\n\f\v#", *c))
2648 d = Nullch; /* "perl" not in first word; ignore */
2650 *s = '#'; /* Don't try to parse shebang line */
2652 #endif /* ALTERNATE_SHEBANG */
2653 #ifndef MACOS_TRADITIONAL
2658 !instr(s,"indir") &&
2659 instr(PL_origargv[0],"perl"))
2665 while (s < PL_bufend && isSPACE(*s))
2667 if (s < PL_bufend) {
2668 Newz(899,newargv,PL_origargc+3,char*);
2670 while (s < PL_bufend && !isSPACE(*s))
2673 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2676 newargv = PL_origargv;
2678 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
2679 Perl_croak(aTHX_ "Can't exec %s", ipath);
2683 U32 oldpdb = PL_perldb;
2684 bool oldn = PL_minus_n;
2685 bool oldp = PL_minus_p;
2687 while (*d && !isSPACE(*d)) d++;
2688 while (SPACE_OR_TAB(*d)) d++;
2692 if (*d == 'M' || *d == 'm') {
2694 while (*d && !isSPACE(*d)) d++;
2695 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2698 d = moreswitches(d);
2700 if ((PERLDB_LINE && !oldpdb) ||
2701 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2702 /* if we have already added "LINE: while (<>) {",
2703 we must not do it again */
2705 sv_setpv(PL_linestr, "");
2706 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2707 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2708 PL_last_lop = PL_last_uni = Nullch;
2709 PL_preambled = FALSE;
2711 (void)gv_fetchfile(PL_origfilename);
2718 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2720 PL_lex_state = LEX_FORMLINE;
2725 #ifdef PERL_STRICT_CR
2726 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2728 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2730 case ' ': case '\t': case '\f': case 013:
2731 #ifdef MACOS_TRADITIONAL
2738 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2739 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2740 /* handle eval qq[#line 1 "foo"\n ...] */
2741 CopLINE_dec(PL_curcop);
2745 while (s < d && *s != '\n')
2749 else if (s > d) /* Found by Ilya: feed random input to Perl. */
2750 Perl_croak(aTHX_ "panic: input overflow");
2752 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2754 PL_lex_state = LEX_FORMLINE;
2764 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2771 while (s < PL_bufend && SPACE_OR_TAB(*s))
2774 if (strnEQ(s,"=>",2)) {
2775 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2776 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2777 "### Saw unary minus before =>, forcing word '%s'\n", s);
2779 OPERATOR('-'); /* unary minus */
2781 PL_last_uni = PL_oldbufptr;
2783 case 'r': ftst = OP_FTEREAD; break;
2784 case 'w': ftst = OP_FTEWRITE; break;
2785 case 'x': ftst = OP_FTEEXEC; break;
2786 case 'o': ftst = OP_FTEOWNED; break;
2787 case 'R': ftst = OP_FTRREAD; break;
2788 case 'W': ftst = OP_FTRWRITE; break;
2789 case 'X': ftst = OP_FTREXEC; break;
2790 case 'O': ftst = OP_FTROWNED; break;
2791 case 'e': ftst = OP_FTIS; break;
2792 case 'z': ftst = OP_FTZERO; break;
2793 case 's': ftst = OP_FTSIZE; break;
2794 case 'f': ftst = OP_FTFILE; break;
2795 case 'd': ftst = OP_FTDIR; break;
2796 case 'l': ftst = OP_FTLINK; break;
2797 case 'p': ftst = OP_FTPIPE; break;
2798 case 'S': ftst = OP_FTSOCK; break;
2799 case 'u': ftst = OP_FTSUID; break;
2800 case 'g': ftst = OP_FTSGID; break;
2801 case 'k': ftst = OP_FTSVTX; break;
2802 case 'b': ftst = OP_FTBLK; break;
2803 case 'c': ftst = OP_FTCHR; break;
2804 case 't': ftst = OP_FTTTY; break;
2805 case 'T': ftst = OP_FTTEXT; break;
2806 case 'B': ftst = OP_FTBINARY; break;
2807 case 'M': case 'A': case 'C':
2808 gv_fetchpv("\024",TRUE, SVt_PV);
2810 case 'M': ftst = OP_FTMTIME; break;
2811 case 'A': ftst = OP_FTATIME; break;
2812 case 'C': ftst = OP_FTCTIME; break;
2820 PL_last_lop_op = ftst;
2821 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2822 "### Saw file test %c\n", (int)ftst);
2827 /* Assume it was a minus followed by a one-letter named
2828 * subroutine call (or a -bareword), then. */
2829 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2830 "### %c looked like a file test but was not\n",
2839 if (PL_expect == XOPERATOR)
2844 else if (*s == '>') {
2847 if (isIDFIRST_lazy_if(s,UTF)) {
2848 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2856 if (PL_expect == XOPERATOR)
2859 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2861 OPERATOR('-'); /* unary minus */
2868 if (PL_expect == XOPERATOR)
2873 if (PL_expect == XOPERATOR)
2876 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2882 if (PL_expect != XOPERATOR) {
2883 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2884 PL_expect = XOPERATOR;
2885 force_ident(PL_tokenbuf, '*');
2898 if (PL_expect == XOPERATOR) {
2902 PL_tokenbuf[0] = '%';
2903 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2904 if (!PL_tokenbuf[1]) {
2906 yyerror("Final % should be \\% or %name");
2909 PL_pending_ident = '%';
2928 switch (PL_expect) {
2931 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2933 PL_bufptr = s; /* update in case we back off */
2939 PL_expect = XTERMBLOCK;
2943 while (isIDFIRST_lazy_if(s,UTF)) {
2944 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2945 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2946 if (tmp < 0) tmp = -tmp;
2961 d = scan_str(d,TRUE,TRUE);
2963 /* MUST advance bufptr here to avoid bogus
2964 "at end of line" context messages from yyerror().
2966 PL_bufptr = s + len;
2967 yyerror("Unterminated attribute parameter in attribute list");
2970 return 0; /* EOF indicator */
2974 SV *sv = newSVpvn(s, len);
2975 sv_catsv(sv, PL_lex_stuff);
2976 attrs = append_elem(OP_LIST, attrs,
2977 newSVOP(OP_CONST, 0, sv));
2978 SvREFCNT_dec(PL_lex_stuff);
2979 PL_lex_stuff = Nullsv;
2982 if (!PL_in_my && len == 6 && strnEQ(s, "lvalue", len))
2983 CvLVALUE_on(PL_compcv);
2984 else if (!PL_in_my && len == 6 && strnEQ(s, "locked", len))
2985 CvLOCKED_on(PL_compcv);
2986 else if (!PL_in_my && len == 6 && strnEQ(s, "method", len))
2987 CvMETHOD_on(PL_compcv);
2989 else if (PL_in_my == KEY_our && len == 6 && strnEQ(s, "unique", len))
2990 GvUNIQUE_on(cGVOPx_gv(yylval.opval));
2992 /* After we've set the flags, it could be argued that
2993 we don't need to do the attributes.pm-based setting
2994 process, and shouldn't bother appending recognized
2995 flags. To experiment with that, uncomment the
2996 following "else": */
2998 attrs = append_elem(OP_LIST, attrs,
2999 newSVOP(OP_CONST, 0,
3003 if (*s == ':' && s[1] != ':')
3006 break; /* require real whitespace or :'s */
3008 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3009 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
3010 char q = ((*s == '\'') ? '"' : '\'');
3011 /* If here for an expression, and parsed no attrs, back off. */
3012 if (tmp == '=' && !attrs) {
3016 /* MUST advance bufptr here to avoid bogus "at end of line"
3017 context messages from yyerror().
3021 yyerror("Unterminated attribute list");
3023 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
3031 PL_nextval[PL_nexttoke].opval = attrs;
3039 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3040 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3056 if (PL_lex_brackets <= 0)
3057 yyerror("Unmatched right square bracket");
3060 if (PL_lex_state == LEX_INTERPNORMAL) {
3061 if (PL_lex_brackets == 0) {
3062 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3063 PL_lex_state = LEX_INTERPEND;
3070 if (PL_lex_brackets > 100) {
3071 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
3072 if (newlb != PL_lex_brackstack) {
3074 PL_lex_brackstack = newlb;
3077 switch (PL_expect) {
3079 if (PL_lex_formbrack) {
3083 if (PL_oldoldbufptr == PL_last_lop)
3084 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3086 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3087 OPERATOR(HASHBRACK);
3089 while (s < PL_bufend && SPACE_OR_TAB(*s))
3092 PL_tokenbuf[0] = '\0';
3093 if (d < PL_bufend && *d == '-') {
3094 PL_tokenbuf[0] = '-';
3096 while (d < PL_bufend && SPACE_OR_TAB(*d))
3099 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3100 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3102 while (d < PL_bufend && SPACE_OR_TAB(*d))
3105 char minus = (PL_tokenbuf[0] == '-');
3106 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3114 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3119 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3124 if (PL_oldoldbufptr == PL_last_lop)
3125 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3127 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3130 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
3132 /* This hack is to get the ${} in the message. */
3134 yyerror("syntax error");
3137 OPERATOR(HASHBRACK);
3139 /* This hack serves to disambiguate a pair of curlies
3140 * as being a block or an anon hash. Normally, expectation
3141 * determines that, but in cases where we're not in a
3142 * position to expect anything in particular (like inside
3143 * eval"") we have to resolve the ambiguity. This code
3144 * covers the case where the first term in the curlies is a
3145 * quoted string. Most other cases need to be explicitly
3146 * disambiguated by prepending a `+' before the opening
3147 * curly in order to force resolution as an anon hash.
3149 * XXX should probably propagate the outer expectation
3150 * into eval"" to rely less on this hack, but that could
3151 * potentially break current behavior of eval"".
3155 if (*s == '\'' || *s == '"' || *s == '`') {
3156 /* common case: get past first string, handling escapes */
3157 for (t++; t < PL_bufend && *t != *s;)
3158 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3162 else if (*s == 'q') {
3165 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3169 char open, close, term;
3172 while (t < PL_bufend && isSPACE(*t))
3176 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3180 for (t++; t < PL_bufend; t++) {
3181 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3183 else if (*t == open)
3187 for (t++; t < PL_bufend; t++) {
3188 if (*t == '\\' && t+1 < PL_bufend)
3190 else if (*t == close && --brackets <= 0)
3192 else if (*t == open)
3198 else if (isALNUM_lazy_if(t,UTF)) {
3200 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3203 while (t < PL_bufend && isSPACE(*t))
3205 /* if comma follows first term, call it an anon hash */
3206 /* XXX it could be a comma expression with loop modifiers */
3207 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3208 || (*t == '=' && t[1] == '>')))
3209 OPERATOR(HASHBRACK);
3210 if (PL_expect == XREF)
3213 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3219 yylval.ival = CopLINE(PL_curcop);
3220 if (isSPACE(*s) || *s == '#')
3221 PL_copline = NOLINE; /* invalidate current command line number */
3226 if (PL_lex_brackets <= 0)
3227 yyerror("Unmatched right curly bracket");
3229 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3230 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3231 PL_lex_formbrack = 0;
3232 if (PL_lex_state == LEX_INTERPNORMAL) {
3233 if (PL_lex_brackets == 0) {
3234 if (PL_expect & XFAKEBRACK) {
3235 PL_expect &= XENUMMASK;
3236 PL_lex_state = LEX_INTERPEND;
3238 return yylex(); /* ignore fake brackets */
3240 if (*s == '-' && s[1] == '>')
3241 PL_lex_state = LEX_INTERPENDMAYBE;
3242 else if (*s != '[' && *s != '{')
3243 PL_lex_state = LEX_INTERPEND;
3246 if (PL_expect & XFAKEBRACK) {
3247 PL_expect &= XENUMMASK;
3249 return yylex(); /* ignore fake brackets */
3259 if (PL_expect == XOPERATOR) {
3260 if (ckWARN(WARN_SEMICOLON)
3261 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3263 CopLINE_dec(PL_curcop);
3264 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3265 CopLINE_inc(PL_curcop);
3270 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3272 PL_expect = XOPERATOR;
3273 force_ident(PL_tokenbuf, '&');
3277 yylval.ival = (OPpENTERSUB_AMPER<<8);
3296 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3297 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3299 if (PL_expect == XSTATE && isALPHA(tmp) &&
3300 (s == PL_linestart+1 || s[-2] == '\n') )
3302 if (PL_in_eval && !PL_rsfp) {
3307 if (strnEQ(s,"=cut",4)) {
3321 PL_doextract = TRUE;
3324 if (PL_lex_brackets < PL_lex_formbrack) {
3326 #ifdef PERL_STRICT_CR
3327 for (t = s; SPACE_OR_TAB(*t); t++) ;
3329 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3331 if (*t == '\n' || *t == '#') {
3349 if (PL_expect != XOPERATOR) {
3350 if (s[1] != '<' && !strchr(s,'>'))
3353 s = scan_heredoc(s);
3355 s = scan_inputsymbol(s);
3356 TERM(sublex_start());
3361 SHop(OP_LEFT_SHIFT);
3375 SHop(OP_RIGHT_SHIFT);
3384 if (PL_expect == XOPERATOR) {
3385 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3388 return ','; /* grandfather non-comma-format format */
3392 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3393 PL_tokenbuf[0] = '@';
3394 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3395 sizeof PL_tokenbuf - 1, FALSE);
3396 if (PL_expect == XOPERATOR)
3397 no_op("Array length", s);
3398 if (!PL_tokenbuf[1])
3400 PL_expect = XOPERATOR;
3401 PL_pending_ident = '#';
3405 PL_tokenbuf[0] = '$';
3406 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3407 sizeof PL_tokenbuf - 1, FALSE);
3408 if (PL_expect == XOPERATOR)
3410 if (!PL_tokenbuf[1]) {
3412 yyerror("Final $ should be \\$ or $name");
3416 /* This kludge not intended to be bulletproof. */
3417 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3418 yylval.opval = newSVOP(OP_CONST, 0,
3419 newSViv(PL_compiling.cop_arybase));
3420 yylval.opval->op_private = OPpCONST_ARYBASE;
3426 if (PL_lex_state == LEX_NORMAL)
3429 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3432 PL_tokenbuf[0] = '@';
3433 if (ckWARN(WARN_SYNTAX)) {
3435 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3438 PL_bufptr = skipspace(PL_bufptr);
3439 while (t < PL_bufend && *t != ']')
3441 Perl_warner(aTHX_ WARN_SYNTAX,
3442 "Multidimensional syntax %.*s not supported",
3443 (t - PL_bufptr) + 1, PL_bufptr);
3447 else if (*s == '{') {
3448 PL_tokenbuf[0] = '%';
3449 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3450 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3452 char tmpbuf[sizeof PL_tokenbuf];
3454 for (t++; isSPACE(*t); t++) ;
3455 if (isIDFIRST_lazy_if(t,UTF)) {
3456 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3457 for (; isSPACE(*t); t++) ;
3458 if (*t == ';' && get_cv(tmpbuf, FALSE))
3459 Perl_warner(aTHX_ WARN_SYNTAX,
3460 "You need to quote \"%s\"", tmpbuf);
3466 PL_expect = XOPERATOR;
3467 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3468 bool islop = (PL_last_lop == PL_oldoldbufptr);
3469 if (!islop || PL_last_lop_op == OP_GREPSTART)
3470 PL_expect = XOPERATOR;
3471 else if (strchr("$@\"'`q", *s))
3472 PL_expect = XTERM; /* e.g. print $fh "foo" */
3473 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3474 PL_expect = XTERM; /* e.g. print $fh &sub */
3475 else if (isIDFIRST_lazy_if(s,UTF)) {
3476 char tmpbuf[sizeof PL_tokenbuf];
3477 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3478 if ((tmp = keyword(tmpbuf, len))) {
3479 /* binary operators exclude handle interpretations */
3491 PL_expect = XTERM; /* e.g. print $fh length() */
3496 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3497 if (gv && GvCVu(gv))
3498 PL_expect = XTERM; /* e.g. print $fh subr() */
3501 else if (isDIGIT(*s))
3502 PL_expect = XTERM; /* e.g. print $fh 3 */
3503 else if (*s == '.' && isDIGIT(s[1]))
3504 PL_expect = XTERM; /* e.g. print $fh .3 */
3505 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3506 PL_expect = XTERM; /* e.g. print $fh -1 */
3507 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3508 PL_expect = XTERM; /* print $fh <<"EOF" */
3510 PL_pending_ident = '$';
3514 if (PL_expect == XOPERATOR)
3516 PL_tokenbuf[0] = '@';
3517 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3518 if (!PL_tokenbuf[1]) {
3520 yyerror("Final @ should be \\@ or @name");
3523 if (PL_lex_state == LEX_NORMAL)
3525 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3527 PL_tokenbuf[0] = '%';
3529 /* Warn about @ where they meant $. */
3530 if (ckWARN(WARN_SYNTAX)) {
3531 if (*s == '[' || *s == '{') {
3533 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3535 if (*t == '}' || *t == ']') {
3537 PL_bufptr = skipspace(PL_bufptr);
3538 Perl_warner(aTHX_ WARN_SYNTAX,
3539 "Scalar value %.*s better written as $%.*s",
3540 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3545 PL_pending_ident = '@';
3548 case '/': /* may either be division or pattern */
3549 case '?': /* may either be conditional or pattern */
3550 if (PL_expect != XOPERATOR) {
3551 /* Disable warning on "study /blah/" */
3552 if (PL_oldoldbufptr == PL_last_uni
3553 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3554 || memNE(PL_last_uni, "study", 5)
3555 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3557 s = scan_pat(s,OP_MATCH);
3558 TERM(sublex_start());
3566 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3567 #ifdef PERL_STRICT_CR
3570 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3572 && (s == PL_linestart || s[-1] == '\n') )
3574 PL_lex_formbrack = 0;
3578 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3584 yylval.ival = OPf_SPECIAL;
3590 if (PL_expect != XOPERATOR)
3595 case '0': case '1': case '2': case '3': case '4':
3596 case '5': case '6': case '7': case '8': case '9':
3597 s = scan_num(s, &yylval);
3598 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3599 "### Saw number in '%s'\n", s);
3601 if (PL_expect == XOPERATOR)
3606 s = scan_str(s,FALSE,FALSE);
3607 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3608 "### Saw string before '%s'\n", s);
3610 if (PL_expect == XOPERATOR) {
3611 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3614 return ','; /* grandfather non-comma-format format */
3620 missingterm((char*)0);
3621 yylval.ival = OP_CONST;
3622 TERM(sublex_start());
3625 s = scan_str(s,FALSE,FALSE);
3626 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3627 "### Saw string before '%s'\n", s);
3629 if (PL_expect == XOPERATOR) {
3630 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3633 return ','; /* grandfather non-comma-format format */
3639 missingterm((char*)0);
3640 yylval.ival = OP_CONST;
3641 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3642 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
3643 yylval.ival = OP_STRINGIFY;
3647 TERM(sublex_start());
3650 s = scan_str(s,FALSE,FALSE);
3651 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3652 "### Saw backtick string before '%s'\n", s);
3654 if (PL_expect == XOPERATOR)
3655 no_op("Backticks",s);
3657 missingterm((char*)0);
3658 yylval.ival = OP_BACKTICK;
3660 TERM(sublex_start());
3664 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3665 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3667 if (PL_expect == XOPERATOR)
3668 no_op("Backslash",s);
3672 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3676 while (isDIGIT(*start) || *start == '_')
3678 if (*start == '.' && isDIGIT(start[1])) {
3679 s = scan_num(s, &yylval);
3682 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3683 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF || PL_expect == XSTATE)) {
3687 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3690 s = scan_num(s, &yylval);
3697 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3736 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3738 /* Some keywords can be followed by any delimiter, including ':' */
3739 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3740 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3741 (PL_tokenbuf[0] == 'q' &&
3742 strchr("qwxr", PL_tokenbuf[1])))));
3744 /* x::* is just a word, unless x is "CORE" */
3745 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3749 while (d < PL_bufend && isSPACE(*d))
3750 d++; /* no comments skipped here, or s### is misparsed */
3752 /* Is this a label? */
3753 if (!tmp && PL_expect == XSTATE
3754 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3756 yylval.pval = savepv(PL_tokenbuf);
3761 /* Check for keywords */
3762 tmp = keyword(PL_tokenbuf, len);
3764 /* Is this a word before a => operator? */
3765 if (*d == '=' && d[1] == '>') {
3767 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3768 yylval.opval->op_private = OPpCONST_BARE;
3769 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
3770 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3774 if (tmp < 0) { /* second-class keyword? */
3775 GV *ogv = Nullgv; /* override (winner) */
3776 GV *hgv = Nullgv; /* hidden (loser) */
3777 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3779 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3782 if (GvIMPORTED_CV(gv))
3784 else if (! CvMETHOD(cv))
3788 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3789 (gv = *gvp) != (GV*)&PL_sv_undef &&
3790 GvCVu(gv) && GvIMPORTED_CV(gv))
3796 tmp = 0; /* overridden by import or by GLOBAL */
3799 && -tmp==KEY_lock /* XXX generalizable kludge */
3801 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3803 tmp = 0; /* any sub overrides "weak" keyword */
3805 else { /* no override */
3809 if (ckWARN(WARN_AMBIGUOUS) && hgv
3810 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3811 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3812 "Ambiguous call resolved as CORE::%s(), %s",
3813 GvENAME(hgv), "qualify as such or use &");
3820 default: /* not a keyword */
3824 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3826 /* Get the rest if it looks like a package qualifier */
3828 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3830 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3833 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3834 *s == '\'' ? "'" : "::");
3839 if (PL_expect == XOPERATOR) {
3840 if (PL_bufptr == PL_linestart) {
3841 CopLINE_dec(PL_curcop);
3842 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3843 CopLINE_inc(PL_curcop);
3846 no_op("Bareword",s);
3849 /* Look for a subroutine with this name in current package,
3850 unless name is "Foo::", in which case Foo is a bearword
3851 (and a package name). */
3854 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3856 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3857 Perl_warner(aTHX_ WARN_BAREWORD,
3858 "Bareword \"%s\" refers to nonexistent package",
3861 PL_tokenbuf[len] = '\0';
3868 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3871 /* if we saw a global override before, get the right name */
3874 sv = newSVpvn("CORE::GLOBAL::",14);
3875 sv_catpv(sv,PL_tokenbuf);
3878 sv = newSVpv(PL_tokenbuf,0);
3880 /* Presume this is going to be a bareword of some sort. */
3883 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3884 yylval.opval->op_private = OPpCONST_BARE;
3886 /* And if "Foo::", then that's what it certainly is. */
3891 /* See if it's the indirect object for a list operator. */
3893 if (PL_oldoldbufptr &&
3894 PL_oldoldbufptr < PL_bufptr &&
3895 (PL_oldoldbufptr == PL_last_lop
3896 || PL_oldoldbufptr == PL_last_uni) &&
3897 /* NO SKIPSPACE BEFORE HERE! */
3898 (PL_expect == XREF ||
3899 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3901 bool immediate_paren = *s == '(';
3903 /* (Now we can afford to cross potential line boundary.) */
3906 /* Two barewords in a row may indicate method call. */
3908 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3911 /* If not a declared subroutine, it's an indirect object. */
3912 /* (But it's an indir obj regardless for sort.) */
3914 if ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
3915 ((!gv || !GvCVu(gv)) &&
3916 (PL_last_lop_op != OP_MAPSTART &&
3917 PL_last_lop_op != OP_GREPSTART))))
3919 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3924 PL_expect = XOPERATOR;
3927 /* Is this a word before a => operator? */
3928 if (*s == '=' && s[1] == '>' && !pkgname) {
3930 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
3931 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
3932 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3936 /* If followed by a paren, it's certainly a subroutine. */
3939 if (gv && GvCVu(gv)) {
3940 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
3941 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3946 PL_nextval[PL_nexttoke].opval = yylval.opval;
3947 PL_expect = XOPERATOR;
3953 /* If followed by var or block, call it a method (unless sub) */
3955 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3956 PL_last_lop = PL_oldbufptr;
3957 PL_last_lop_op = OP_METHOD;
3961 /* If followed by a bareword, see if it looks like indir obj. */
3963 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3966 /* Not a method, so call it a subroutine (if defined) */
3968 if (gv && GvCVu(gv)) {
3970 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3971 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3972 "Ambiguous use of -%s resolved as -&%s()",
3973 PL_tokenbuf, PL_tokenbuf);
3974 /* Check for a constant sub */
3976 if ((sv = cv_const_sv(cv))) {
3978 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3979 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3980 yylval.opval->op_private = 0;
3984 /* Resolve to GV now. */
3985 op_free(yylval.opval);
3986 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3987 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3988 PL_last_lop = PL_oldbufptr;
3989 PL_last_lop_op = OP_ENTERSUB;
3990 /* Is there a prototype? */
3993 char *proto = SvPV((SV*)cv, len);
3996 if (strEQ(proto, "$"))
3998 if (*proto == '&' && *s == '{') {
3999 sv_setpv(PL_subname,"__ANON__");
4003 PL_nextval[PL_nexttoke].opval = yylval.opval;
4009 /* Call it a bare word */
4011 if (PL_hints & HINT_STRICT_SUBS)
4012 yylval.opval->op_private |= OPpCONST_STRICT;
4015 if (ckWARN(WARN_RESERVED)) {
4016 if (lastchar != '-') {
4017 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4018 if (!*d && strNE(PL_tokenbuf,"main"))
4019 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
4026 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
4027 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4028 "Operator or semicolon missing before %c%s",
4029 lastchar, PL_tokenbuf);
4030 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4031 "Ambiguous use of %c resolved as operator %c",
4032 lastchar, lastchar);
4038 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4039 newSVpv(CopFILE(PL_curcop),0));
4043 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4044 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4047 case KEY___PACKAGE__:
4048 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4050 ? newSVsv(PL_curstname)
4059 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4060 char *pname = "main";
4061 if (PL_tokenbuf[2] == 'D')
4062 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
4063 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4066 GvIOp(gv) = newIO();
4067 IoIFP(GvIOp(gv)) = PL_rsfp;
4068 #if defined(HAS_FCNTL) && defined(F_SETFD)
4070 int fd = PerlIO_fileno(PL_rsfp);
4071 fcntl(fd,F_SETFD,fd >= 3);
4074 /* Mark this internal pseudo-handle as clean */
4075 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4077 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4078 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4079 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4081 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4082 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4083 /* if the script was opened in binmode, we need to revert
4084 * it to text mode for compatibility; but only iff it has CRs
4085 * XXX this is a questionable hack at best. */
4086 if (PL_bufend-PL_bufptr > 2
4087 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4090 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4091 loc = PerlIO_tell(PL_rsfp);
4092 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4095 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
4097 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4098 #endif /* NETWARE */
4099 #ifdef PERLIO_IS_STDIO /* really? */
4100 # if defined(__BORLANDC__)
4101 /* XXX see note in do_binmode() */
4102 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
4106 PerlIO_seek(PL_rsfp, loc, 0);
4110 #ifdef PERLIO_LAYERS
4111 if (UTF && !IN_BYTES)
4112 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
4125 if (PL_expect == XSTATE) {
4132 if (*s == ':' && s[1] == ':') {
4135 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4136 if (!(tmp = keyword(PL_tokenbuf, len)))
4137 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4151 LOP(OP_ACCEPT,XTERM);
4157 LOP(OP_ATAN2,XTERM);
4163 LOP(OP_BINMODE,XTERM);
4166 LOP(OP_BLESS,XTERM);
4175 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4192 if (!PL_cryptseen) {
4193 PL_cryptseen = TRUE;
4197 LOP(OP_CRYPT,XTERM);
4200 LOP(OP_CHMOD,XTERM);
4203 LOP(OP_CHOWN,XTERM);
4206 LOP(OP_CONNECT,XTERM);
4222 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4226 PL_hints |= HINT_BLOCK_SCOPE;
4236 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4237 LOP(OP_DBMOPEN,XTERM);
4243 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4250 yylval.ival = CopLINE(PL_curcop);
4264 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4265 UNIBRACK(OP_ENTEREVAL);
4280 case KEY_endhostent:
4286 case KEY_endservent:
4289 case KEY_endprotoent:
4300 yylval.ival = CopLINE(PL_curcop);
4302 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4304 if ((PL_bufend - p) >= 3 &&
4305 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4307 else if ((PL_bufend - p) >= 4 &&
4308 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4311 if (isIDFIRST_lazy_if(p,UTF)) {
4312 p = scan_ident(p, PL_bufend,
4313 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4317 Perl_croak(aTHX_ "Missing $ on loop variable");
4322 LOP(OP_FORMLINE,XTERM);
4328 LOP(OP_FCNTL,XTERM);
4334 LOP(OP_FLOCK,XTERM);
4343 LOP(OP_GREPSTART, XREF);
4346 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4361 case KEY_getpriority:
4362 LOP(OP_GETPRIORITY,XTERM);
4364 case KEY_getprotobyname:
4367 case KEY_getprotobynumber:
4368 LOP(OP_GPBYNUMBER,XTERM);
4370 case KEY_getprotoent:
4382 case KEY_getpeername:
4383 UNI(OP_GETPEERNAME);
4385 case KEY_gethostbyname:
4388 case KEY_gethostbyaddr:
4389 LOP(OP_GHBYADDR,XTERM);
4391 case KEY_gethostent:
4394 case KEY_getnetbyname:
4397 case KEY_getnetbyaddr:
4398 LOP(OP_GNBYADDR,XTERM);
4403 case KEY_getservbyname:
4404 LOP(OP_GSBYNAME,XTERM);
4406 case KEY_getservbyport:
4407 LOP(OP_GSBYPORT,XTERM);
4409 case KEY_getservent:
4412 case KEY_getsockname:
4413 UNI(OP_GETSOCKNAME);
4415 case KEY_getsockopt:
4416 LOP(OP_GSOCKOPT,XTERM);
4438 yylval.ival = CopLINE(PL_curcop);
4442 LOP(OP_INDEX,XTERM);
4448 LOP(OP_IOCTL,XTERM);
4460 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4492 LOP(OP_LISTEN,XTERM);
4501 s = scan_pat(s,OP_MATCH);
4502 TERM(sublex_start());
4505 LOP(OP_MAPSTART, XREF);
4508 LOP(OP_MKDIR,XTERM);
4511 LOP(OP_MSGCTL,XTERM);
4514 LOP(OP_MSGGET,XTERM);
4517 LOP(OP_MSGRCV,XTERM);
4520 LOP(OP_MSGSND,XTERM);
4526 if (isIDFIRST_lazy_if(s,UTF)) {
4527 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4528 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4530 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4531 if (!PL_in_my_stash) {
4534 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4542 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4549 if (PL_expect != XSTATE)
4550 yyerror("\"no\" not allowed in expression");
4551 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4552 s = force_version(s, FALSE);
4557 if (*s == '(' || (s = skipspace(s), *s == '('))
4564 if (isIDFIRST_lazy_if(s,UTF)) {
4566 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4568 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4569 Perl_warner(aTHX_ WARN_PRECEDENCE,
4570 "Precedence problem: open %.*s should be open(%.*s)",
4576 yylval.ival = OP_OR;
4586 LOP(OP_OPEN_DIR,XTERM);
4589 checkcomma(s,PL_tokenbuf,"filehandle");
4593 checkcomma(s,PL_tokenbuf,"filehandle");
4612 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4616 LOP(OP_PIPE_OP,XTERM);
4619 s = scan_str(s,FALSE,FALSE);
4621 missingterm((char*)0);
4622 yylval.ival = OP_CONST;
4623 TERM(sublex_start());
4629 s = scan_str(s,FALSE,FALSE);
4631 missingterm((char*)0);
4633 if (SvCUR(PL_lex_stuff)) {
4636 d = SvPV_force(PL_lex_stuff, len);
4639 for (; isSPACE(*d) && len; --len, ++d) ;
4642 if (!warned && ckWARN(WARN_QW)) {
4643 for (; !isSPACE(*d) && len; --len, ++d) {
4645 Perl_warner(aTHX_ WARN_QW,
4646 "Possible attempt to separate words with commas");
4649 else if (*d == '#') {
4650 Perl_warner(aTHX_ WARN_QW,
4651 "Possible attempt to put comments in qw() list");
4657 for (; !isSPACE(*d) && len; --len, ++d) ;
4659 sv = newSVpvn(b, d-b);
4660 if (DO_UTF8(PL_lex_stuff))
4662 words = append_elem(OP_LIST, words,
4663 newSVOP(OP_CONST, 0, tokeq(sv)));
4667 PL_nextval[PL_nexttoke].opval = words;
4672 SvREFCNT_dec(PL_lex_stuff);
4673 PL_lex_stuff = Nullsv;
4679 s = scan_str(s,FALSE,FALSE);
4681 missingterm((char*)0);
4682 yylval.ival = OP_STRINGIFY;
4683 if (SvIVX(PL_lex_stuff) == '\'')
4684 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4685 TERM(sublex_start());
4688 s = scan_pat(s,OP_QR);
4689 TERM(sublex_start());
4692 s = scan_str(s,FALSE,FALSE);
4694 missingterm((char*)0);
4695 yylval.ival = OP_BACKTICK;
4697 TERM(sublex_start());
4705 s = force_version(s, FALSE);
4707 else if (*s != 'v' || !isDIGIT(s[1])
4708 || (s = force_version(s, TRUE), *s == 'v'))
4710 *PL_tokenbuf = '\0';
4711 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4712 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4713 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4715 yyerror("<> should be quotes");
4723 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4727 LOP(OP_RENAME,XTERM);
4736 LOP(OP_RINDEX,XTERM);
4759 LOP(OP_REVERSE,XTERM);
4770 TERM(sublex_start());
4772 TOKEN(1); /* force error */
4781 LOP(OP_SELECT,XTERM);
4787 LOP(OP_SEMCTL,XTERM);
4790 LOP(OP_SEMGET,XTERM);
4793 LOP(OP_SEMOP,XTERM);
4799 LOP(OP_SETPGRP,XTERM);
4801 case KEY_setpriority:
4802 LOP(OP_SETPRIORITY,XTERM);
4804 case KEY_sethostent:
4810 case KEY_setservent:
4813 case KEY_setprotoent:
4823 LOP(OP_SEEKDIR,XTERM);
4825 case KEY_setsockopt:
4826 LOP(OP_SSOCKOPT,XTERM);
4832 LOP(OP_SHMCTL,XTERM);
4835 LOP(OP_SHMGET,XTERM);
4838 LOP(OP_SHMREAD,XTERM);
4841 LOP(OP_SHMWRITE,XTERM);
4844 LOP(OP_SHUTDOWN,XTERM);
4853 LOP(OP_SOCKET,XTERM);
4855 case KEY_socketpair:
4856 LOP(OP_SOCKPAIR,XTERM);
4859 checkcomma(s,PL_tokenbuf,"subroutine name");
4861 if (*s == ';' || *s == ')') /* probably a close */
4862 Perl_croak(aTHX_ "sort is now a reserved word");
4864 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4868 LOP(OP_SPLIT,XTERM);
4871 LOP(OP_SPRINTF,XTERM);
4874 LOP(OP_SPLICE,XTERM);
4889 LOP(OP_SUBSTR,XTERM);
4895 char tmpbuf[sizeof PL_tokenbuf];
4896 SSize_t tboffset = 0;
4897 expectation attrful;
4898 bool have_name, have_proto;
4903 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4904 (*s == ':' && s[1] == ':'))
4907 attrful = XATTRBLOCK;
4908 /* remember buffer pos'n for later force_word */
4909 tboffset = s - PL_oldbufptr;
4910 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4911 if (strchr(tmpbuf, ':'))
4912 sv_setpv(PL_subname, tmpbuf);
4914 sv_setsv(PL_subname,PL_curstname);
4915 sv_catpvn(PL_subname,"::",2);
4916 sv_catpvn(PL_subname,tmpbuf,len);
4923 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4924 PL_expect = XTERMBLOCK;
4925 attrful = XATTRTERM;
4926 sv_setpv(PL_subname,"?");
4930 if (key == KEY_format) {
4932 PL_lex_formbrack = PL_lex_brackets + 1;
4934 (void) force_word(PL_oldbufptr + tboffset, WORD,
4939 /* Look for a prototype */
4943 s = scan_str(s,FALSE,FALSE);
4945 Perl_croak(aTHX_ "Prototype not terminated");
4947 d = SvPVX(PL_lex_stuff);
4949 for (p = d; *p; ++p) {
4954 SvCUR(PL_lex_stuff) = tmp;
4962 if (*s == ':' && s[1] != ':')
4963 PL_expect = attrful;
4966 PL_nextval[PL_nexttoke].opval =
4967 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4968 PL_lex_stuff = Nullsv;
4972 sv_setpv(PL_subname,"__ANON__");
4975 (void) force_word(PL_oldbufptr + tboffset, WORD,
4984 LOP(OP_SYSTEM,XREF);
4987 LOP(OP_SYMLINK,XTERM);
4990 LOP(OP_SYSCALL,XTERM);
4993 LOP(OP_SYSOPEN,XTERM);
4996 LOP(OP_SYSSEEK,XTERM);
4999 LOP(OP_SYSREAD,XTERM);
5002 LOP(OP_SYSWRITE,XTERM);
5006 TERM(sublex_start());
5027 LOP(OP_TRUNCATE,XTERM);
5039 yylval.ival = CopLINE(PL_curcop);
5043 yylval.ival = CopLINE(PL_curcop);
5047 LOP(OP_UNLINK,XTERM);
5053 LOP(OP_UNPACK,XTERM);
5056 LOP(OP_UTIME,XTERM);
5062 LOP(OP_UNSHIFT,XTERM);
5065 if (PL_expect != XSTATE)
5066 yyerror("\"use\" not allowed in expression");
5068 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
5069 s = force_version(s, TRUE);
5070 if (*s == ';' || (s = skipspace(s), *s == ';')) {
5071 PL_nextval[PL_nexttoke].opval = Nullop;
5074 else if (*s == 'v') {
5075 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5076 s = force_version(s, FALSE);
5080 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5081 s = force_version(s, FALSE);
5093 yylval.ival = CopLINE(PL_curcop);
5097 PL_hints |= HINT_BLOCK_SCOPE;
5104 LOP(OP_WAITPID,XTERM);
5113 ctl_l[0] = toCTRL('L');
5115 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5118 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5123 if (PL_expect == XOPERATOR)
5129 yylval.ival = OP_XOR;
5134 TERM(sublex_start());
5139 #pragma segment Main
5143 S_pending_ident(pTHX)
5147 /* pit holds the identifier we read and pending_ident is reset */
5148 char pit = PL_pending_ident;
5149 PL_pending_ident = 0;
5151 DEBUG_T({ PerlIO_printf(Perl_debug_log,
5152 "### Tokener saw identifier '%s'\n", PL_tokenbuf); });
5154 /* if we're in a my(), we can't allow dynamics here.
5155 $foo'bar has already been turned into $foo::bar, so
5156 just check for colons.
5158 if it's a legal name, the OP is a PADANY.
5161 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
5162 if (strchr(PL_tokenbuf,':'))
5163 yyerror(Perl_form(aTHX_ "No package name allowed for "
5164 "variable %s in \"our\"",
5166 tmp = pad_allocmy(PL_tokenbuf);
5169 if (strchr(PL_tokenbuf,':'))
5170 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
5172 yylval.opval = newOP(OP_PADANY, 0);
5173 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
5179 build the ops for accesses to a my() variable.
5181 Deny my($a) or my($b) in a sort block, *if* $a or $b is
5182 then used in a comparison. This catches most, but not
5183 all cases. For instance, it catches
5184 sort { my($a); $a <=> $b }
5186 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
5187 (although why you'd do that is anyone's guess).
5190 if (!strchr(PL_tokenbuf,':')) {
5191 #ifdef USE_5005THREADS
5192 /* Check for single character per-thread SVs */
5193 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
5194 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
5195 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
5197 yylval.opval = newOP(OP_THREADSV, 0);
5198 yylval.opval->op_targ = tmp;
5201 #endif /* USE_5005THREADS */
5202 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
5203 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
5204 /* might be an "our" variable" */
5205 if (SvFLAGS(namesv) & SVpad_OUR) {
5206 /* build ops for a bareword */
5207 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
5208 sv_catpvn(sym, "::", 2);
5209 sv_catpv(sym, PL_tokenbuf+1);
5210 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
5211 yylval.opval->op_private = OPpCONST_ENTERED;
5212 gv_fetchpv(SvPVX(sym),
5214 ? (GV_ADDMULTI | GV_ADDINEVAL)
5217 ((PL_tokenbuf[0] == '$') ? SVt_PV
5218 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5223 /* if it's a sort block and they're naming $a or $b */
5224 if (PL_last_lop_op == OP_SORT &&
5225 PL_tokenbuf[0] == '$' &&
5226 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
5229 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
5230 d < PL_bufend && *d != '\n';
5233 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
5234 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
5240 yylval.opval = newOP(OP_PADANY, 0);
5241 yylval.opval->op_targ = tmp;
5247 Whine if they've said @foo in a doublequoted string,
5248 and @foo isn't a variable we can find in the symbol
5251 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
5252 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
5253 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
5254 && ckWARN(WARN_AMBIGUOUS))
5256 /* Downgraded from fatal to warning 20000522 mjd */
5257 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5258 "Possible unintended interpolation of %s in string",
5263 /* build ops for a bareword */
5264 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
5265 yylval.opval->op_private = OPpCONST_ENTERED;
5266 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
5267 ((PL_tokenbuf[0] == '$') ? SVt_PV
5268 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5274 Perl_keyword(pTHX_ register char *d, I32 len)
5279 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5280 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5281 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5282 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5283 if (strEQ(d,"__END__")) return KEY___END__;
5287 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5292 if (strEQ(d,"and")) return -KEY_and;
5293 if (strEQ(d,"abs")) return -KEY_abs;
5296 if (strEQ(d,"alarm")) return -KEY_alarm;
5297 if (strEQ(d,"atan2")) return -KEY_atan2;
5300 if (strEQ(d,"accept")) return -KEY_accept;
5305 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5308 if (strEQ(d,"bless")) return -KEY_bless;
5309 if (strEQ(d,"bind")) return -KEY_bind;
5310 if (strEQ(d,"binmode")) return -KEY_binmode;
5313 if (strEQ(d,"CORE")) return -KEY_CORE;
5314 if (strEQ(d,"CHECK")) return KEY_CHECK;
5319 if (strEQ(d,"cmp")) return -KEY_cmp;
5320 if (strEQ(d,"chr")) return -KEY_chr;
5321 if (strEQ(d,"cos")) return -KEY_cos;
5324 if (strEQ(d,"chop")) return -KEY_chop;
5327 if (strEQ(d,"close")) return -KEY_close;
5328 if (strEQ(d,"chdir")) return -KEY_chdir;
5329 if (strEQ(d,"chomp")) return -KEY_chomp;
5330 if (strEQ(d,"chmod")) return -KEY_chmod;
5331 if (strEQ(d,"chown")) return -KEY_chown;
5332 if (strEQ(d,"crypt")) return -KEY_crypt;
5335 if (strEQ(d,"chroot")) return -KEY_chroot;
5336 if (strEQ(d,"caller")) return -KEY_caller;
5339 if (strEQ(d,"connect")) return -KEY_connect;
5342 if (strEQ(d,"closedir")) return -KEY_closedir;
5343 if (strEQ(d,"continue")) return -KEY_continue;
5348 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5353 if (strEQ(d,"do")) return KEY_do;
5356 if (strEQ(d,"die")) return -KEY_die;
5359 if (strEQ(d,"dump")) return -KEY_dump;
5362 if (strEQ(d,"delete")) return KEY_delete;
5365 if (strEQ(d,"defined")) return KEY_defined;
5366 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5369 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5374 if (strEQ(d,"END")) return KEY_END;
5379 if (strEQ(d,"eq")) return -KEY_eq;
5382 if (strEQ(d,"eof")) return -KEY_eof;
5383 if (strEQ(d,"exp")) return -KEY_exp;
5386 if (strEQ(d,"else")) return KEY_else;
5387 if (strEQ(d,"exit")) return -KEY_exit;
5388 if (strEQ(d,"eval")) return KEY_eval;
5389 if (strEQ(d,"exec")) return -KEY_exec;
5390 if (strEQ(d,"each")) return -KEY_each;
5393 if (strEQ(d,"elsif")) return KEY_elsif;
5396 if (strEQ(d,"exists")) return KEY_exists;
5397 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5400 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5401 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5404 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5407 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5408 if (strEQ(d,"endservent")) return -KEY_endservent;
5411 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5418 if (strEQ(d,"for")) return KEY_for;
5421 if (strEQ(d,"fork")) return -KEY_fork;
5424 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5425 if (strEQ(d,"flock")) return -KEY_flock;
5428 if (strEQ(d,"format")) return KEY_format;
5429 if (strEQ(d,"fileno")) return -KEY_fileno;
5432 if (strEQ(d,"foreach")) return KEY_foreach;
5435 if (strEQ(d,"formline")) return -KEY_formline;
5440 if (strnEQ(d,"get",3)) {
5445 if (strEQ(d,"ppid")) return -KEY_getppid;
5446 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5449 if (strEQ(d,"pwent")) return -KEY_getpwent;
5450 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5451 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5454 if (strEQ(d,"peername")) return -KEY_getpeername;
5455 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5456 if (strEQ(d,"priority")) return -KEY_getpriority;
5459 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5462 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5466 else if (*d == 'h') {
5467 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5468 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5469 if (strEQ(d,"hostent")) return -KEY_gethostent;
5471 else if (*d == 'n') {
5472 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5473 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5474 if (strEQ(d,"netent")) return -KEY_getnetent;
5476 else if (*d == 's') {
5477 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5478 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5479 if (strEQ(d,"servent")) return -KEY_getservent;
5480 if (strEQ(d,"sockname")) return -KEY_getsockname;
5481 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5483 else if (*d == 'g') {
5484 if (strEQ(d,"grent")) return -KEY_getgrent;
5485 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5486 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5488 else if (*d == 'l') {
5489 if (strEQ(d,"login")) return -KEY_getlogin;
5491 else if (strEQ(d,"c")) return -KEY_getc;
5496 if (strEQ(d,"gt")) return -KEY_gt;
5497 if (strEQ(d,"ge")) return -KEY_ge;
5500 if (strEQ(d,"grep")) return KEY_grep;
5501 if (strEQ(d,"goto")) return KEY_goto;
5502 if (strEQ(d,"glob")) return KEY_glob;
5505 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5510 if (strEQ(d,"hex")) return -KEY_hex;
5513 if (strEQ(d,"INIT")) return KEY_INIT;
5518 if (strEQ(d,"if")) return KEY_if;
5521 if (strEQ(d,"int")) return -KEY_int;
5524 if (strEQ(d,"index")) return -KEY_index;
5525 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5530 if (strEQ(d,"join")) return -KEY_join;
5534 if (strEQ(d,"keys")) return -KEY_keys;
5535 if (strEQ(d,"kill")) return -KEY_kill;
5541 if (strEQ(d,"lt")) return -KEY_lt;
5542 if (strEQ(d,"le")) return -KEY_le;
5543 if (strEQ(d,"lc")) return -KEY_lc;
5546 if (strEQ(d,"log")) return -KEY_log;
5549 if (strEQ(d,"last")) return KEY_last;
5550 if (strEQ(d,"link")) return -KEY_link;
5551 if (strEQ(d,"lock")) return -KEY_lock;
5554 if (strEQ(d,"local")) return KEY_local;
5555 if (strEQ(d,"lstat")) return -KEY_lstat;
5558 if (strEQ(d,"length")) return -KEY_length;
5559 if (strEQ(d,"listen")) return -KEY_listen;
5562 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5565 if (strEQ(d,"localtime")) return -KEY_localtime;
5571 case 1: return KEY_m;
5573 if (strEQ(d,"my")) return KEY_my;
5576 if (strEQ(d,"map")) return KEY_map;
5579 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5582 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5583 if (strEQ(d,"msgget")) return -KEY_msgget;
5584 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5585 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5590 if (strEQ(d,"next")) return KEY_next;
5591 if (strEQ(d,"ne")) return -KEY_ne;
5592 if (strEQ(d,"not")) return -KEY_not;
5593 if (strEQ(d,"no")) return KEY_no;
5598 if (strEQ(d,"or")) return -KEY_or;
5601 if (strEQ(d,"ord")) return -KEY_ord;
5602 if (strEQ(d,"oct")) return -KEY_oct;
5603 if (strEQ(d,"our")) return KEY_our;
5606 if (strEQ(d,"open")) return -KEY_open;
5609 if (strEQ(d,"opendir")) return -KEY_opendir;
5616 if (strEQ(d,"pop")) return -KEY_pop;
5617 if (strEQ(d,"pos")) return KEY_pos;
5620 if (strEQ(d,"push")) return -KEY_push;
5621 if (strEQ(d,"pack")) return -KEY_pack;
5622 if (strEQ(d,"pipe")) return -KEY_pipe;
5625 if (strEQ(d,"print")) return KEY_print;
5628 if (strEQ(d,"printf")) return KEY_printf;
5631 if (strEQ(d,"package")) return KEY_package;
5634 if (strEQ(d,"prototype")) return KEY_prototype;
5639 if (strEQ(d,"q")) return KEY_q;
5640 if (strEQ(d,"qr")) return KEY_qr;
5641 if (strEQ(d,"qq")) return KEY_qq;
5642 if (strEQ(d,"qw")) return KEY_qw;
5643 if (strEQ(d,"qx")) return KEY_qx;
5645 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5650 if (strEQ(d,"ref")) return -KEY_ref;
5653 if (strEQ(d,"read")) return -KEY_read;
5654 if (strEQ(d,"rand")) return -KEY_rand;
5655 if (strEQ(d,"recv")) return -KEY_recv;
5656 if (strEQ(d,"redo")) return KEY_redo;
5659 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5660 if (strEQ(d,"reset")) return -KEY_reset;
5663 if (strEQ(d,"return")) return KEY_return;
5664 if (strEQ(d,"rename")) return -KEY_rename;
5665 if (strEQ(d,"rindex")) return -KEY_rindex;
5668 if (strEQ(d,"require")) return KEY_require;
5669 if (strEQ(d,"reverse")) return -KEY_reverse;
5670 if (strEQ(d,"readdir")) return -KEY_readdir;
5673 if (strEQ(d,"readlink")) return -KEY_readlink;
5674 if (strEQ(d,"readline")) return -KEY_readline;
5675 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5678 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5684 case 0: return KEY_s;
5686 if (strEQ(d,"scalar")) return KEY_scalar;
5691 if (strEQ(d,"seek")) return -KEY_seek;
5692 if (strEQ(d,"send")) return -KEY_send;
5695 if (strEQ(d,"semop")) return -KEY_semop;
5698 if (strEQ(d,"select")) return -KEY_select;
5699 if (strEQ(d,"semctl")) return -KEY_semctl;
5700 if (strEQ(d,"semget")) return -KEY_semget;
5703 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5704 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5707 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5708 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5711 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5714 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5715 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5716 if (strEQ(d,"setservent")) return -KEY_setservent;
5719 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5720 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5727 if (strEQ(d,"shift")) return -KEY_shift;
5730 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5731 if (strEQ(d,"shmget")) return -KEY_shmget;
5734 if (strEQ(d,"shmread")) return -KEY_shmread;
5737 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5738 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5743 if (strEQ(d,"sin")) return -KEY_sin;
5746 if (strEQ(d,"sleep")) return -KEY_sleep;
5749 if (strEQ(d,"sort")) return KEY_sort;
5750 if (strEQ(d,"socket")) return -KEY_socket;
5751 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5754 if (strEQ(d,"split")) return KEY_split;
5755 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5756 if (strEQ(d,"splice")) return -KEY_splice;
5759 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5762 if (strEQ(d,"srand")) return -KEY_srand;
5765 if (strEQ(d,"stat")) return -KEY_stat;
5766 if (strEQ(d,"study")) return KEY_study;
5769 if (strEQ(d,"substr")) return -KEY_substr;
5770 if (strEQ(d,"sub")) return KEY_sub;
5775 if (strEQ(d,"system")) return -KEY_system;
5778 if (strEQ(d,"symlink")) return -KEY_symlink;
5779 if (strEQ(d,"syscall")) return -KEY_syscall;
5780 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5781 if (strEQ(d,"sysread")) return -KEY_sysread;
5782 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5785 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5794 if (strEQ(d,"tr")) return KEY_tr;
5797 if (strEQ(d,"tie")) return KEY_tie;
5800 if (strEQ(d,"tell")) return -KEY_tell;
5801 if (strEQ(d,"tied")) return KEY_tied;
5802 if (strEQ(d,"time")) return -KEY_time;
5805 if (strEQ(d,"times")) return -KEY_times;
5808 if (strEQ(d,"telldir")) return -KEY_telldir;
5811 if (strEQ(d,"truncate")) return -KEY_truncate;
5818 if (strEQ(d,"uc")) return -KEY_uc;
5821 if (strEQ(d,"use")) return KEY_use;
5824 if (strEQ(d,"undef")) return KEY_undef;
5825 if (strEQ(d,"until")) return KEY_until;
5826 if (strEQ(d,"untie")) return KEY_untie;
5827 if (strEQ(d,"utime")) return -KEY_utime;
5828 if (strEQ(d,"umask")) return -KEY_umask;
5831 if (strEQ(d,"unless")) return KEY_unless;
5832 if (strEQ(d,"unpack")) return -KEY_unpack;
5833 if (strEQ(d,"unlink")) return -KEY_unlink;
5836 if (strEQ(d,"unshift")) return -KEY_unshift;
5837 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5842 if (strEQ(d,"values")) return -KEY_values;
5843 if (strEQ(d,"vec")) return -KEY_vec;
5848 if (strEQ(d,"warn")) return -KEY_warn;
5849 if (strEQ(d,"wait")) return -KEY_wait;
5852 if (strEQ(d,"while")) return KEY_while;
5853 if (strEQ(d,"write")) return -KEY_write;
5856 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5859 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5864 if (len == 1) return -KEY_x;
5865 if (strEQ(d,"xor")) return -KEY_xor;
5868 if (len == 1) return KEY_y;
5877 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5881 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5882 if (ckWARN(WARN_SYNTAX)) {
5884 for (w = s+2; *w && level; w++) {
5891 for (; *w && isSPACE(*w); w++) ;
5892 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5893 Perl_warner(aTHX_ WARN_SYNTAX,
5894 "%s (...) interpreted as function",name);
5897 while (s < PL_bufend && isSPACE(*s))
5901 while (s < PL_bufend && isSPACE(*s))
5903 if (isIDFIRST_lazy_if(s,UTF)) {
5905 while (isALNUM_lazy_if(s,UTF))
5907 while (s < PL_bufend && isSPACE(*s))
5912 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5916 Perl_croak(aTHX_ "No comma allowed after %s", what);
5921 /* Either returns sv, or mortalizes sv and returns a new SV*.
5922 Best used as sv=new_constant(..., sv, ...).
5923 If s, pv are NULL, calls subroutine with one argument,
5924 and type is used with error messages only. */
5927 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5931 HV *table = GvHV(PL_hintgv); /* ^H */
5935 const char *why1, *why2, *why3;
5937 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
5940 why2 = strEQ(key,"charnames")
5941 ? "(possibly a missing \"use charnames ...\")"
5943 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
5944 (type ? type: "undef"), why2);
5946 /* This is convoluted and evil ("goto considered harmful")
5947 * but I do not understand the intricacies of all the different
5948 * failure modes of %^H in here. The goal here is to make
5949 * the most probable error message user-friendly. --jhi */
5954 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
5955 (type ? type: "undef"), why1, why2, why3);
5957 yyerror(SvPVX(msg));
5961 cvp = hv_fetch(table, key, strlen(key), FALSE);
5962 if (!cvp || !SvOK(*cvp)) {
5965 why3 = "} is not defined";
5968 sv_2mortal(sv); /* Parent created it permanently */
5971 pv = sv_2mortal(newSVpvn(s, len));
5973 typesv = sv_2mortal(newSVpv(type, 0));
5975 typesv = &PL_sv_undef;
5977 PUSHSTACKi(PERLSI_OVERLOAD);
5989 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5993 /* Check the eval first */
5994 if (!PL_in_eval && SvTRUE(ERRSV)) {
5996 sv_catpv(ERRSV, "Propagated");
5997 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5999 res = SvREFCNT_inc(sv);
6003 (void)SvREFCNT_inc(res);
6012 why1 = "Call to &{$^H{";
6014 why3 = "}} did not return a defined value";
6023 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
6025 register char *d = dest;
6026 register char *e = d + destlen - 3; /* two-character token, ending NUL */
6029 Perl_croak(aTHX_ ident_too_long);
6030 if (isALNUM(*s)) /* UTF handled below */
6032 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
6037 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
6041 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6042 char *t = s + UTF8SKIP(s);
6043 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6045 if (d + (t - s) > e)
6046 Perl_croak(aTHX_ ident_too_long);
6047 Copy(s, d, t - s, char);
6060 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
6070 e = d + destlen - 3; /* two-character token, ending NUL */
6072 while (isDIGIT(*s)) {
6074 Perl_croak(aTHX_ ident_too_long);
6081 Perl_croak(aTHX_ ident_too_long);
6082 if (isALNUM(*s)) /* UTF handled below */
6084 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
6089 else if (*s == ':' && s[1] == ':') {
6093 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6094 char *t = s + UTF8SKIP(s);
6095 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6097 if (d + (t - s) > e)
6098 Perl_croak(aTHX_ ident_too_long);
6099 Copy(s, d, t - s, char);
6110 if (PL_lex_state != LEX_NORMAL)
6111 PL_lex_state = LEX_INTERPENDMAYBE;
6114 if (*s == '$' && s[1] &&
6115 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
6128 if (*d == '^' && *s && isCONTROLVAR(*s)) {
6133 if (isSPACE(s[-1])) {
6136 if (!SPACE_OR_TAB(ch)) {
6142 if (isIDFIRST_lazy_if(d,UTF)) {
6146 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
6148 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
6151 Copy(s, d, e - s, char);
6156 while ((isALNUM(*s) || *s == ':') && d < e)
6159 Perl_croak(aTHX_ ident_too_long);
6162 while (s < send && SPACE_OR_TAB(*s)) s++;
6163 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
6164 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
6165 const char *brack = *s == '[' ? "[...]" : "{...}";
6166 Perl_warner(aTHX_ WARN_AMBIGUOUS,
6167 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
6168 funny, dest, brack, funny, dest, brack);
6171 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
6175 /* Handle extended ${^Foo} variables
6176 * 1999-02-27 mjd-perl-patch@plover.com */
6177 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
6181 while (isALNUM(*s) && d < e) {
6185 Perl_croak(aTHX_ ident_too_long);
6190 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
6191 PL_lex_state = LEX_INTERPEND;
6194 if (PL_lex_state == LEX_NORMAL) {
6195 if (ckWARN(WARN_AMBIGUOUS) &&
6196 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
6198 Perl_warner(aTHX_ WARN_AMBIGUOUS,
6199 "Ambiguous use of %c{%s} resolved to %c%s",
6200 funny, dest, funny, dest);
6205 s = bracket; /* let the parser handle it */
6209 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
6210 PL_lex_state = LEX_INTERPEND;
6215 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
6220 *pmfl |= PMf_GLOBAL;
6222 *pmfl |= PMf_CONTINUE;
6226 *pmfl |= PMf_MULTILINE;
6228 *pmfl |= PMf_SINGLELINE;
6230 *pmfl |= PMf_EXTENDED;
6234 S_scan_pat(pTHX_ char *start, I32 type)
6239 s = scan_str(start,FALSE,FALSE);
6241 Perl_croak(aTHX_ "Search pattern not terminated");
6243 pm = (PMOP*)newPMOP(type, 0);
6244 if (PL_multi_open == '?')
6245 pm->op_pmflags |= PMf_ONCE;
6247 while (*s && strchr("iomsx", *s))
6248 pmflag(&pm->op_pmflags,*s++);
6251 while (*s && strchr("iogcmsx", *s))
6252 pmflag(&pm->op_pmflags,*s++);
6254 pm->op_pmpermflags = pm->op_pmflags;
6256 PL_lex_op = (OP*)pm;
6257 yylval.ival = OP_MATCH;
6262 S_scan_subst(pTHX_ char *start)
6269 yylval.ival = OP_NULL;
6271 s = scan_str(start,FALSE,FALSE);
6274 Perl_croak(aTHX_ "Substitution pattern not terminated");
6276 if (s[-1] == PL_multi_open)
6279 first_start = PL_multi_start;
6280 s = scan_str(s,FALSE,FALSE);
6283 SvREFCNT_dec(PL_lex_stuff);
6284 PL_lex_stuff = Nullsv;
6286 Perl_croak(aTHX_ "Substitution replacement not terminated");
6288 PL_multi_start = first_start; /* so whole substitution is taken together */
6290 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6296 else if (strchr("iogcmsx", *s))
6297 pmflag(&pm->op_pmflags,*s++);
6304 PL_sublex_info.super_bufptr = s;
6305 PL_sublex_info.super_bufend = PL_bufend;
6307 pm->op_pmflags |= PMf_EVAL;
6308 repl = newSVpvn("",0);
6310 sv_catpv(repl, es ? "eval " : "do ");
6311 sv_catpvn(repl, "{ ", 2);
6312 sv_catsv(repl, PL_lex_repl);
6313 sv_catpvn(repl, " };", 2);
6315 SvREFCNT_dec(PL_lex_repl);
6319 pm->op_pmpermflags = pm->op_pmflags;
6320 PL_lex_op = (OP*)pm;
6321 yylval.ival = OP_SUBST;
6326 S_scan_trans(pTHX_ char *start)
6335 yylval.ival = OP_NULL;
6337 s = scan_str(start,FALSE,FALSE);
6339 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6340 if (s[-1] == PL_multi_open)
6343 s = scan_str(s,FALSE,FALSE);
6346 SvREFCNT_dec(PL_lex_stuff);
6347 PL_lex_stuff = Nullsv;
6349 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6352 complement = del = squash = 0;
6353 while (strchr("cds", *s)) {
6355 complement = OPpTRANS_COMPLEMENT;
6357 del = OPpTRANS_DELETE;
6359 squash = OPpTRANS_SQUASH;
6363 New(803, tbl, complement&&!del?258:256, short);
6364 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6365 o->op_private = del|squash|complement|
6366 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
6367 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
6370 yylval.ival = OP_TRANS;
6375 S_scan_heredoc(pTHX_ register char *s)
6378 I32 op_type = OP_SCALAR;
6385 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6389 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6392 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6393 if (*peek && strchr("`'\"",*peek)) {
6396 s = delimcpy(d, e, s, PL_bufend, term, &len);
6406 if (!isALNUM_lazy_if(s,UTF))
6407 deprecate("bare << to mean <<\"\"");
6408 for (; isALNUM_lazy_if(s,UTF); s++) {
6413 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6414 Perl_croak(aTHX_ "Delimiter for here document is too long");
6417 len = d - PL_tokenbuf;
6418 #ifndef PERL_STRICT_CR
6419 d = strchr(s, '\r');
6423 while (s < PL_bufend) {
6429 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6438 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6443 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6444 herewas = newSVpvn(s,PL_bufend-s);
6446 s--, herewas = newSVpvn(s,d-s);
6447 s += SvCUR(herewas);
6449 tmpstr = NEWSV(87,79);
6450 sv_upgrade(tmpstr, SVt_PVIV);
6455 else if (term == '`') {
6456 op_type = OP_BACKTICK;
6457 SvIVX(tmpstr) = '\\';
6461 PL_multi_start = CopLINE(PL_curcop);
6462 PL_multi_open = PL_multi_close = '<';
6463 term = *PL_tokenbuf;
6464 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6465 char *bufptr = PL_sublex_info.super_bufptr;
6466 char *bufend = PL_sublex_info.super_bufend;
6467 char *olds = s - SvCUR(herewas);
6468 s = strchr(bufptr, '\n');
6472 while (s < bufend &&
6473 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6475 CopLINE_inc(PL_curcop);
6478 CopLINE_set(PL_curcop, PL_multi_start);
6479 missingterm(PL_tokenbuf);
6481 sv_setpvn(herewas,bufptr,d-bufptr+1);
6482 sv_setpvn(tmpstr,d+1,s-d);
6484 sv_catpvn(herewas,s,bufend-s);
6485 (void)strcpy(bufptr,SvPVX(herewas));
6492 while (s < PL_bufend &&
6493 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6495 CopLINE_inc(PL_curcop);
6497 if (s >= PL_bufend) {
6498 CopLINE_set(PL_curcop, PL_multi_start);
6499 missingterm(PL_tokenbuf);
6501 sv_setpvn(tmpstr,d+1,s-d);
6503 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6505 sv_catpvn(herewas,s,PL_bufend-s);
6506 sv_setsv(PL_linestr,herewas);
6507 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6508 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6509 PL_last_lop = PL_last_uni = Nullch;
6512 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6513 while (s >= PL_bufend) { /* multiple line string? */
6515 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6516 CopLINE_set(PL_curcop, PL_multi_start);
6517 missingterm(PL_tokenbuf);
6519 CopLINE_inc(PL_curcop);
6520 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6521 PL_last_lop = PL_last_uni = Nullch;
6522 #ifndef PERL_STRICT_CR
6523 if (PL_bufend - PL_linestart >= 2) {
6524 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6525 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6527 PL_bufend[-2] = '\n';
6529 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6531 else if (PL_bufend[-1] == '\r')
6532 PL_bufend[-1] = '\n';
6534 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6535 PL_bufend[-1] = '\n';
6537 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6538 SV *sv = NEWSV(88,0);
6540 sv_upgrade(sv, SVt_PVMG);
6541 sv_setsv(sv,PL_linestr);
6544 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6546 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6549 sv_catsv(PL_linestr,herewas);
6550 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6554 sv_catsv(tmpstr,PL_linestr);
6559 PL_multi_end = CopLINE(PL_curcop);
6560 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6561 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6562 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6564 SvREFCNT_dec(herewas);
6565 if (UTF && !IN_BYTES && is_utf8_string((U8*)SvPVX(tmpstr), SvCUR(tmpstr)))
6567 PL_lex_stuff = tmpstr;
6568 yylval.ival = op_type;
6573 takes: current position in input buffer
6574 returns: new position in input buffer
6575 side-effects: yylval and lex_op are set.
6580 <FH> read from filehandle
6581 <pkg::FH> read from package qualified filehandle
6582 <pkg'FH> read from package qualified filehandle
6583 <$fh> read from filehandle in $fh
6589 S_scan_inputsymbol(pTHX_ char *start)
6591 register char *s = start; /* current position in buffer */
6597 d = PL_tokenbuf; /* start of temp holding space */
6598 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6599 end = strchr(s, '\n');
6602 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6604 /* die if we didn't have space for the contents of the <>,
6605 or if it didn't end, or if we see a newline
6608 if (len >= sizeof PL_tokenbuf)
6609 Perl_croak(aTHX_ "Excessively long <> operator");
6611 Perl_croak(aTHX_ "Unterminated <> operator");
6616 Remember, only scalar variables are interpreted as filehandles by
6617 this code. Anything more complex (e.g., <$fh{$num}>) will be
6618 treated as a glob() call.
6619 This code makes use of the fact that except for the $ at the front,
6620 a scalar variable and a filehandle look the same.
6622 if (*d == '$' && d[1]) d++;
6624 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6625 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6628 /* If we've tried to read what we allow filehandles to look like, and
6629 there's still text left, then it must be a glob() and not a getline.
6630 Use scan_str to pull out the stuff between the <> and treat it
6631 as nothing more than a string.
6634 if (d - PL_tokenbuf != len) {
6635 yylval.ival = OP_GLOB;
6637 s = scan_str(start,FALSE,FALSE);
6639 Perl_croak(aTHX_ "Glob not terminated");
6643 /* we're in a filehandle read situation */
6646 /* turn <> into <ARGV> */
6648 (void)strcpy(d,"ARGV");
6650 /* if <$fh>, create the ops to turn the variable into a
6656 /* try to find it in the pad for this block, otherwise find
6657 add symbol table ops
6659 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6660 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
6661 if (SvFLAGS(namesv) & SVpad_OUR) {
6662 SV *sym = sv_2mortal(newSVpv(HvNAME(GvSTASH(namesv)),0));
6663 sv_catpvn(sym, "::", 2);
6669 OP *o = newOP(OP_PADSV, 0);
6671 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6680 ? (GV_ADDMULTI | GV_ADDINEVAL)
6683 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6684 newUNOP(OP_RV2SV, 0,
6685 newGVOP(OP_GV, 0, gv)));
6687 PL_lex_op->op_flags |= OPf_SPECIAL;
6688 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6689 yylval.ival = OP_NULL;
6692 /* If it's none of the above, it must be a literal filehandle
6693 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6695 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6696 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6697 yylval.ival = OP_NULL;
6706 takes: start position in buffer
6707 keep_quoted preserve \ on the embedded delimiter(s)
6708 keep_delims preserve the delimiters around the string
6709 returns: position to continue reading from buffer
6710 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6711 updates the read buffer.
6713 This subroutine pulls a string out of the input. It is called for:
6714 q single quotes q(literal text)
6715 ' single quotes 'literal text'
6716 qq double quotes qq(interpolate $here please)
6717 " double quotes "interpolate $here please"
6718 qx backticks qx(/bin/ls -l)
6719 ` backticks `/bin/ls -l`
6720 qw quote words @EXPORT_OK = qw( func() $spam )
6721 m// regexp match m/this/
6722 s/// regexp substitute s/this/that/
6723 tr/// string transliterate tr/this/that/
6724 y/// string transliterate y/this/that/
6725 ($*@) sub prototypes sub foo ($)
6726 (stuff) sub attr parameters sub foo : attr(stuff)
6727 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6729 In most of these cases (all but <>, patterns and transliterate)
6730 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6731 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6732 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6735 It skips whitespace before the string starts, and treats the first
6736 character as the delimiter. If the delimiter is one of ([{< then
6737 the corresponding "close" character )]}> is used as the closing
6738 delimiter. It allows quoting of delimiters, and if the string has
6739 balanced delimiters ([{<>}]) it allows nesting.
6741 On success, the SV with the resulting string is put into lex_stuff or,
6742 if that is already non-NULL, into lex_repl. The second case occurs only
6743 when parsing the RHS of the special constructs s/// and tr/// (y///).
6744 For convenience, the terminating delimiter character is stuffed into
6749 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6751 SV *sv; /* scalar value: string */
6752 char *tmps; /* temp string, used for delimiter matching */
6753 register char *s = start; /* current position in the buffer */
6754 register char term; /* terminating character */
6755 register char *to; /* current position in the sv's data */
6756 I32 brackets = 1; /* bracket nesting level */
6757 bool has_utf8 = FALSE; /* is there any utf8 content? */
6759 /* skip space before the delimiter */
6763 /* mark where we are, in case we need to report errors */
6766 /* after skipping whitespace, the next character is the terminator */
6768 if (!UTF8_IS_INVARIANT((U8)term) && UTF)
6771 /* mark where we are */
6772 PL_multi_start = CopLINE(PL_curcop);
6773 PL_multi_open = term;
6775 /* find corresponding closing delimiter */
6776 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6778 PL_multi_close = term;
6780 /* create a new SV to hold the contents. 87 is leak category, I'm
6781 assuming. 79 is the SV's initial length. What a random number. */
6783 sv_upgrade(sv, SVt_PVIV);
6785 (void)SvPOK_only(sv); /* validate pointer */
6787 /* move past delimiter and try to read a complete string */
6789 sv_catpvn(sv, s, 1);
6792 /* extend sv if need be */
6793 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6794 /* set 'to' to the next character in the sv's string */
6795 to = SvPVX(sv)+SvCUR(sv);
6797 /* if open delimiter is the close delimiter read unbridle */
6798 if (PL_multi_open == PL_multi_close) {
6799 for (; s < PL_bufend; s++,to++) {
6800 /* embedded newlines increment the current line number */
6801 if (*s == '\n' && !PL_rsfp)
6802 CopLINE_inc(PL_curcop);
6803 /* handle quoted delimiters */
6804 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6805 if (!keep_quoted && s[1] == term)
6807 /* any other quotes are simply copied straight through */
6811 /* terminate when run out of buffer (the for() condition), or
6812 have found the terminator */
6813 else if (*s == term)
6815 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
6821 /* if the terminator isn't the same as the start character (e.g.,
6822 matched brackets), we have to allow more in the quoting, and
6823 be prepared for nested brackets.
6826 /* read until we run out of string, or we find the terminator */
6827 for (; s < PL_bufend; s++,to++) {
6828 /* embedded newlines increment the line count */
6829 if (*s == '\n' && !PL_rsfp)
6830 CopLINE_inc(PL_curcop);
6831 /* backslashes can escape the open or closing characters */
6832 if (*s == '\\' && s+1 < PL_bufend) {
6834 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6839 /* allow nested opens and closes */
6840 else if (*s == PL_multi_close && --brackets <= 0)
6842 else if (*s == PL_multi_open)
6844 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
6849 /* terminate the copied string and update the sv's end-of-string */
6851 SvCUR_set(sv, to - SvPVX(sv));
6854 * this next chunk reads more into the buffer if we're not done yet
6858 break; /* handle case where we are done yet :-) */
6860 #ifndef PERL_STRICT_CR
6861 if (to - SvPVX(sv) >= 2) {
6862 if ((to[-2] == '\r' && to[-1] == '\n') ||
6863 (to[-2] == '\n' && to[-1] == '\r'))
6867 SvCUR_set(sv, to - SvPVX(sv));
6869 else if (to[-1] == '\r')
6872 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6876 /* if we're out of file, or a read fails, bail and reset the current
6877 line marker so we can report where the unterminated string began
6880 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6882 CopLINE_set(PL_curcop, PL_multi_start);
6885 /* we read a line, so increment our line counter */
6886 CopLINE_inc(PL_curcop);
6888 /* update debugger info */
6889 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6890 SV *sv = NEWSV(88,0);
6892 sv_upgrade(sv, SVt_PVMG);
6893 sv_setsv(sv,PL_linestr);
6896 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6899 /* having changed the buffer, we must update PL_bufend */
6900 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6901 PL_last_lop = PL_last_uni = Nullch;
6904 /* at this point, we have successfully read the delimited string */
6907 sv_catpvn(sv, s, 1);
6910 PL_multi_end = CopLINE(PL_curcop);
6913 /* if we allocated too much space, give some back */
6914 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6915 SvLEN_set(sv, SvCUR(sv) + 1);
6916 Renew(SvPVX(sv), SvLEN(sv), char);
6919 /* decide whether this is the first or second quoted string we've read
6932 takes: pointer to position in buffer
6933 returns: pointer to new position in buffer
6934 side-effects: builds ops for the constant in yylval.op
6936 Read a number in any of the formats that Perl accepts:
6938 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
6939 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
6942 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
6944 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6947 If it reads a number without a decimal point or an exponent, it will
6948 try converting the number to an integer and see if it can do so
6949 without loss of precision.
6953 Perl_scan_num(pTHX_ char *start, YYSTYPE* lvalp)
6955 register char *s = start; /* current position in buffer */
6956 register char *d; /* destination in temp buffer */
6957 register char *e; /* end of temp buffer */
6958 NV nv; /* number read, as a double */
6959 SV *sv = Nullsv; /* place to put the converted number */
6960 bool floatit; /* boolean: int or float? */
6961 char *lastub = 0; /* position of last underbar */
6962 static char number_too_long[] = "Number too long";
6964 /* We use the first character to decide what type of number this is */
6968 Perl_croak(aTHX_ "panic: scan_num");
6970 /* if it starts with a 0, it could be an octal number, a decimal in
6971 0.13 disguise, or a hexadecimal number, or a binary number. */
6975 u holds the "number so far"
6976 shift the power of 2 of the base
6977 (hex == 4, octal == 3, binary == 1)
6978 overflowed was the number more than we can hold?
6980 Shift is used when we add a digit. It also serves as an "are
6981 we in octal/hex/binary?" indicator to disallow hex characters
6987 bool overflowed = FALSE;
6988 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6989 static char* bases[5] = { "", "binary", "", "octal",
6991 static char* Bases[5] = { "", "Binary", "", "Octal",
6993 static char *maxima[5] = { "",
6994 "0b11111111111111111111111111111111",
6998 char *base, *Base, *max;
7004 } else if (s[1] == 'b') {
7008 /* check for a decimal in disguise */
7009 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
7011 /* so it must be octal */
7018 if (ckWARN(WARN_SYNTAX))
7019 Perl_warner(aTHX_ WARN_SYNTAX,
7020 "Misplaced _ in number");
7024 base = bases[shift];
7025 Base = Bases[shift];
7026 max = maxima[shift];
7028 /* read the rest of the number */
7030 /* x is used in the overflow test,
7031 b is the digit we're adding on. */
7036 /* if we don't mention it, we're done */
7040 /* _ are ignored -- but warned about if consecutive */
7042 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7043 Perl_warner(aTHX_ WARN_SYNTAX,
7044 "Misplaced _ in number");
7048 /* 8 and 9 are not octal */
7051 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
7055 case '2': case '3': case '4':
7056 case '5': case '6': case '7':
7058 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
7062 b = *s++ & 15; /* ASCII digit -> value of digit */
7066 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
7067 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
7068 /* make sure they said 0x */
7073 /* Prepare to put the digit we have onto the end
7074 of the number so far. We check for overflows.
7079 x = u << shift; /* make room for the digit */
7081 if ((x >> shift) != u
7082 && !(PL_hints & HINT_NEW_BINARY)) {
7085 if (ckWARN_d(WARN_OVERFLOW))
7086 Perl_warner(aTHX_ WARN_OVERFLOW,
7087 "Integer overflow in %s number",
7090 u = x | b; /* add the digit to the end */
7093 n *= nvshift[shift];
7094 /* If an NV has not enough bits in its
7095 * mantissa to represent an UV this summing of
7096 * small low-order numbers is a waste of time
7097 * (because the NV cannot preserve the
7098 * low-order bits anyway): we could just
7099 * remember when did we overflow and in the
7100 * end just multiply n by the right
7108 /* if we get here, we had success: make a scalar value from
7113 /* final misplaced underbar check */
7115 if (ckWARN(WARN_SYNTAX))
7116 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
7121 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
7122 Perl_warner(aTHX_ WARN_PORTABLE,
7123 "%s number > %s non-portable",
7129 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
7130 Perl_warner(aTHX_ WARN_PORTABLE,
7131 "%s number > %s non-portable",
7136 if (PL_hints & HINT_NEW_BINARY)
7137 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
7142 handle decimal numbers.
7143 we're also sent here when we read a 0 as the first digit
7145 case '1': case '2': case '3': case '4': case '5':
7146 case '6': case '7': case '8': case '9': case '.':
7149 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
7152 /* read next group of digits and _ and copy into d */
7153 while (isDIGIT(*s) || *s == '_') {
7154 /* skip underscores, checking for misplaced ones
7158 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7159 Perl_warner(aTHX_ WARN_SYNTAX,
7160 "Misplaced _ in number");
7164 /* check for end of fixed-length buffer */
7166 Perl_croak(aTHX_ number_too_long);
7167 /* if we're ok, copy the character */
7172 /* final misplaced underbar check */
7173 if (lastub && s == lastub + 1) {
7174 if (ckWARN(WARN_SYNTAX))
7175 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
7178 /* read a decimal portion if there is one. avoid
7179 3..5 being interpreted as the number 3. followed
7182 if (*s == '.' && s[1] != '.') {
7187 if (ckWARN(WARN_SYNTAX))
7188 Perl_warner(aTHX_ WARN_SYNTAX,
7189 "Misplaced _ in number");
7193 /* copy, ignoring underbars, until we run out of digits.
7195 for (; isDIGIT(*s) || *s == '_'; s++) {
7196 /* fixed length buffer check */
7198 Perl_croak(aTHX_ number_too_long);
7200 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7201 Perl_warner(aTHX_ WARN_SYNTAX,
7202 "Misplaced _ in number");
7208 /* fractional part ending in underbar? */
7210 if (ckWARN(WARN_SYNTAX))
7211 Perl_warner(aTHX_ WARN_SYNTAX,
7212 "Misplaced _ in number");
7214 if (*s == '.' && isDIGIT(s[1])) {
7215 /* oops, it's really a v-string, but without the "v" */
7221 /* read exponent part, if present */
7222 if (*s && strchr("eE",*s) && strchr("+-0123456789_", s[1])) {
7226 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
7227 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
7229 /* stray preinitial _ */
7231 if (ckWARN(WARN_SYNTAX))
7232 Perl_warner(aTHX_ WARN_SYNTAX,
7233 "Misplaced _ in number");
7237 /* allow positive or negative exponent */
7238 if (*s == '+' || *s == '-')
7241 /* stray initial _ */
7243 if (ckWARN(WARN_SYNTAX))
7244 Perl_warner(aTHX_ WARN_SYNTAX,
7245 "Misplaced _ in number");
7249 /* read digits of exponent */
7250 while (isDIGIT(*s) || *s == '_') {
7253 Perl_croak(aTHX_ number_too_long);
7257 if (ckWARN(WARN_SYNTAX) &&
7258 ((lastub && s == lastub + 1) ||
7259 (!isDIGIT(s[1]) && s[1] != '_')))
7260 Perl_warner(aTHX_ WARN_SYNTAX,
7261 "Misplaced _ in number");
7268 /* make an sv from the string */
7272 We try to do an integer conversion first if no characters
7273 indicating "float" have been found.
7278 int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
7280 if (flags == IS_NUMBER_IN_UV) {
7282 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7285 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
7286 if (uv <= (UV) IV_MIN)
7287 sv_setiv(sv, -(IV)uv);
7294 /* terminate the string */
7296 nv = Atof(PL_tokenbuf);
7300 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7301 (PL_hints & HINT_NEW_INTEGER) )
7302 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7303 (floatit ? "float" : "integer"),
7307 /* if it starts with a v, it could be a v-string */
7313 while (isDIGIT(*pos) || *pos == '_')
7315 if (!isALPHA(*pos)) {
7317 U8 tmpbuf[UTF8_MAXLEN+1];
7319 s++; /* get past 'v' */
7322 sv_setpvn(sv, "", 0);
7325 if (*s == '0' && isDIGIT(s[1]))
7326 yyerror("Octal number in vector unsupported");
7329 /* this is atoi() that tolerates underscores */
7332 while (--end >= s) {
7337 rev += (*end - '0') * mult;
7339 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
7340 Perl_warner(aTHX_ WARN_OVERFLOW,
7341 "Integer overflow in decimal number");
7344 /* Append native character for the rev point */
7345 tmpend = uvchr_to_utf8(tmpbuf, rev);
7346 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
7347 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev)))
7349 if (*pos == '.' && isDIGIT(pos[1]))
7355 while (isDIGIT(*pos) || *pos == '_')
7365 /* make the op for the constant and return */
7368 lvalp->opval = newSVOP(OP_CONST, 0, sv);
7370 lvalp->opval = Nullop;
7376 S_scan_formline(pTHX_ register char *s)
7380 SV *stuff = newSVpvn("",0);
7381 bool needargs = FALSE;
7384 if (*s == '.' || *s == /*{*/'}') {
7386 #ifdef PERL_STRICT_CR
7387 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7389 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7391 if (*t == '\n' || t == PL_bufend)
7394 if (PL_in_eval && !PL_rsfp) {
7395 eol = strchr(s,'\n');
7400 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7402 for (t = s; t < eol; t++) {
7403 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7405 goto enough; /* ~~ must be first line in formline */
7407 if (*t == '@' || *t == '^')
7411 sv_catpvn(stuff, s, eol-s);
7412 #ifndef PERL_STRICT_CR
7413 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7414 char *end = SvPVX(stuff) + SvCUR(stuff);
7426 s = filter_gets(PL_linestr, PL_rsfp, 0);
7427 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7428 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7429 PL_last_lop = PL_last_uni = Nullch;
7432 yyerror("Format not terminated");
7442 PL_lex_state = LEX_NORMAL;
7443 PL_nextval[PL_nexttoke].ival = 0;
7447 PL_lex_state = LEX_FORMLINE;
7448 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7450 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7454 SvREFCNT_dec(stuff);
7455 PL_lex_formbrack = 0;
7466 PL_cshlen = strlen(PL_cshname);
7471 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7473 I32 oldsavestack_ix = PL_savestack_ix;
7474 CV* outsidecv = PL_compcv;
7478 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7480 SAVEI32(PL_subline);
7481 save_item(PL_subname);
7484 SAVESPTR(PL_comppad_name);
7485 SAVESPTR(PL_compcv);
7486 SAVEI32(PL_comppad_name_fill);
7487 SAVEI32(PL_min_intro_pending);
7488 SAVEI32(PL_max_intro_pending);
7489 SAVEI32(PL_pad_reset_pending);
7491 PL_compcv = (CV*)NEWSV(1104,0);
7492 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7493 CvFLAGS(PL_compcv) |= flags;
7495 PL_comppad = newAV();
7496 av_push(PL_comppad, Nullsv);
7497 PL_curpad = AvARRAY(PL_comppad);
7498 PL_comppad_name = newAV();
7499 PL_comppad_name_fill = 0;
7500 PL_min_intro_pending = 0;
7502 PL_subline = CopLINE(PL_curcop);
7503 #ifdef USE_5005THREADS
7504 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7505 PL_curpad[0] = (SV*)newAV();
7506 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7507 #endif /* USE_5005THREADS */
7509 comppadlist = newAV();
7510 AvREAL_off(comppadlist);
7511 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7512 av_store(comppadlist, 1, (SV*)PL_comppad);
7514 CvPADLIST(PL_compcv) = comppadlist;
7515 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7516 #ifdef USE_5005THREADS
7517 CvOWNER(PL_compcv) = 0;
7518 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7519 MUTEX_INIT(CvMUTEXP(PL_compcv));
7520 #endif /* USE_5005THREADS */
7522 return oldsavestack_ix;
7526 #pragma segment Perl_yylex
7529 Perl_yywarn(pTHX_ char *s)
7531 PL_in_eval |= EVAL_WARNONLY;
7533 PL_in_eval &= ~EVAL_WARNONLY;
7538 Perl_yyerror(pTHX_ char *s)
7541 char *context = NULL;
7545 if (!yychar || (yychar == ';' && !PL_rsfp))
7547 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7548 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7549 while (isSPACE(*PL_oldoldbufptr))
7551 context = PL_oldoldbufptr;
7552 contlen = PL_bufptr - PL_oldoldbufptr;
7554 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7555 PL_oldbufptr != PL_bufptr) {
7556 while (isSPACE(*PL_oldbufptr))
7558 context = PL_oldbufptr;
7559 contlen = PL_bufptr - PL_oldbufptr;
7561 else if (yychar > 255)
7562 where = "next token ???";
7563 #ifdef USE_PURE_BISON
7564 /* GNU Bison sets the value -2 */
7565 else if (yychar == -2) {
7567 else if ((yychar & 127) == 127) {
7569 if (PL_lex_state == LEX_NORMAL ||
7570 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7571 where = "at end of line";
7572 else if (PL_lex_inpat)
7573 where = "within pattern";
7575 where = "within string";
7578 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7580 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7581 else if (isPRINT_LC(yychar))
7582 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7584 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7585 where = SvPVX(where_sv);
7587 msg = sv_2mortal(newSVpv(s, 0));
7588 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7589 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7591 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7593 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7594 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7595 Perl_sv_catpvf(aTHX_ msg,
7596 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7597 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7600 if (PL_in_eval & EVAL_WARNONLY)
7601 Perl_warn(aTHX_ "%"SVf, msg);
7604 if (PL_error_count >= 10) {
7605 if (PL_in_eval && SvCUR(ERRSV))
7606 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
7607 ERRSV, CopFILE(PL_curcop));
7609 Perl_croak(aTHX_ "%s has too many errors.\n",
7610 CopFILE(PL_curcop));
7613 PL_in_my_stash = Nullhv;
7617 #pragma segment Main
7621 S_swallow_bom(pTHX_ U8 *s)
7624 slen = SvCUR(PL_linestr);
7628 /* UTF-16 little-endian */
7629 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
7630 Perl_croak(aTHX_ "Unsupported script encoding");
7631 #ifndef PERL_NO_UTF16_FILTER
7632 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-LE script encoding\n"));
7634 if (PL_bufend > (char*)s) {
7638 filter_add(utf16rev_textfilter, NULL);
7639 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7640 PL_bufend = (char*)utf16_to_utf8_reversed(s, news,
7641 PL_bufend - (char*)s - 1,
7643 Copy(news, s, newlen, U8);
7644 SvCUR_set(PL_linestr, newlen);
7645 PL_bufend = SvPVX(PL_linestr) + newlen;
7646 news[newlen++] = '\0';
7650 Perl_croak(aTHX_ "Unsupported script encoding");
7655 if (s[1] == 0xFF) { /* UTF-16 big-endian */
7656 #ifndef PERL_NO_UTF16_FILTER
7657 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding\n"));
7659 if (PL_bufend > (char *)s) {
7663 filter_add(utf16_textfilter, NULL);
7664 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7665 PL_bufend = (char*)utf16_to_utf8(s, news,
7666 PL_bufend - (char*)s,
7668 Copy(news, s, newlen, U8);
7669 SvCUR_set(PL_linestr, newlen);
7670 PL_bufend = SvPVX(PL_linestr) + newlen;
7671 news[newlen++] = '\0';
7675 Perl_croak(aTHX_ "Unsupported script encoding");
7680 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
7681 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-8 script encoding\n"));
7686 if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */
7687 s[2] == 0xFE && s[3] == 0xFF)
7689 Perl_croak(aTHX_ "Unsupported script encoding");
7697 * Restore a source filter.
7701 restore_rsfp(pTHX_ void *f)
7703 PerlIO *fp = (PerlIO*)f;
7705 if (PL_rsfp == PerlIO_stdin())
7706 PerlIO_clearerr(PL_rsfp);
7707 else if (PL_rsfp && (PL_rsfp != fp))
7708 PerlIO_close(PL_rsfp);
7712 #ifndef PERL_NO_UTF16_FILTER
7714 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
7716 I32 count = FILTER_READ(idx+1, sv, maxlen);
7721 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7722 if (!*SvPV_nolen(sv))
7723 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7726 tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7727 sv_usepvn(sv, (char*)tmps, tend - tmps);
7733 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
7735 I32 count = FILTER_READ(idx+1, sv, maxlen);
7740 if (!*SvPV_nolen(sv))
7741 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7744 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7745 tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7746 sv_usepvn(sv, (char*)tmps, tend - tmps);