3 * Copyright (c) 1991-2002, 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);
323 Perl_deprecate_old(pTHX_ char *s)
325 /* This function should NOT be called for any new deprecated warnings */
326 /* Use Perl_deprecate instead */
328 /* It is here to maintain backward compatibility with the pre-5.8 */
329 /* warnings category hierarchy. The "deprecated" category used to */
330 /* live under the "syntax" category. It is now a top-level category */
331 /* in its own right. */
333 if (ckWARN2(WARN_DEPRECATED, WARN_SYNTAX))
334 Perl_warner(aTHX_ packWARN2(WARN_DEPRECATED, WARN_SYNTAX),
335 "Use of %s is deprecated", s);
340 * Deprecate a comma-less variable list.
346 deprecate_old("comma-less variable list");
350 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
351 * utf16-to-utf8-reversed.
354 #ifdef PERL_CR_FILTER
358 register char *s = SvPVX(sv);
359 register char *e = s + SvCUR(sv);
360 /* outer loop optimized to do nothing if there are no CR-LFs */
362 if (*s++ == '\r' && *s == '\n') {
363 /* hit a CR-LF, need to copy the rest */
364 register char *d = s - 1;
367 if (*s == '\r' && s[1] == '\n')
378 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
380 I32 count = FILTER_READ(idx+1, sv, maxlen);
381 if (count > 0 && !maxlen)
389 * Initialize variables. Uses the Perl save_stack to save its state (for
390 * recursive calls to the parser).
394 Perl_lex_start(pTHX_ SV *line)
399 SAVEI32(PL_lex_dojoin);
400 SAVEI32(PL_lex_brackets);
401 SAVEI32(PL_lex_casemods);
402 SAVEI32(PL_lex_starts);
403 SAVEI32(PL_lex_state);
404 SAVEVPTR(PL_lex_inpat);
405 SAVEI32(PL_lex_inwhat);
406 if (PL_lex_state == LEX_KNOWNEXT) {
407 I32 toke = PL_nexttoke;
408 while (--toke >= 0) {
409 SAVEI32(PL_nexttype[toke]);
410 SAVEVPTR(PL_nextval[toke]);
412 SAVEI32(PL_nexttoke);
414 SAVECOPLINE(PL_curcop);
417 SAVEPPTR(PL_oldbufptr);
418 SAVEPPTR(PL_oldoldbufptr);
419 SAVEPPTR(PL_last_lop);
420 SAVEPPTR(PL_last_uni);
421 SAVEPPTR(PL_linestart);
422 SAVESPTR(PL_linestr);
423 SAVEPPTR(PL_lex_brackstack);
424 SAVEPPTR(PL_lex_casestack);
425 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
426 SAVESPTR(PL_lex_stuff);
427 SAVEI32(PL_lex_defer);
428 SAVEI32(PL_sublex_info.sub_inwhat);
429 SAVESPTR(PL_lex_repl);
431 SAVEINT(PL_lex_expect);
433 PL_lex_state = LEX_NORMAL;
437 New(899, PL_lex_brackstack, 120, char);
438 New(899, PL_lex_casestack, 12, char);
439 SAVEFREEPV(PL_lex_brackstack);
440 SAVEFREEPV(PL_lex_casestack);
442 *PL_lex_casestack = '\0';
445 PL_lex_stuff = Nullsv;
446 PL_lex_repl = Nullsv;
450 PL_sublex_info.sub_inwhat = 0;
452 if (SvREADONLY(PL_linestr))
453 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
454 s = SvPV(PL_linestr, len);
455 if (len && s[len-1] != ';') {
456 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
457 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
458 sv_catpvn(PL_linestr, "\n;", 2);
460 SvTEMP_off(PL_linestr);
461 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
462 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
463 PL_last_lop = PL_last_uni = Nullch;
469 * Finalizer for lexing operations. Must be called when the parser is
470 * done with the lexer.
476 PL_doextract = FALSE;
481 * This subroutine has nothing to do with tilting, whether at windmills
482 * or pinball tables. Its name is short for "increment line". It
483 * increments the current line number in CopLINE(PL_curcop) and checks
484 * to see whether the line starts with a comment of the form
485 * # line 500 "foo.pm"
486 * If so, it sets the current line number and file to the values in the comment.
490 S_incline(pTHX_ char *s)
497 CopLINE_inc(PL_curcop);
500 while (SPACE_OR_TAB(*s)) s++;
501 if (strnEQ(s, "line", 4))
505 if (SPACE_OR_TAB(*s))
509 while (SPACE_OR_TAB(*s)) s++;
515 while (SPACE_OR_TAB(*s))
517 if (*s == '"' && (t = strchr(s+1, '"'))) {
522 for (t = s; !isSPACE(*t); t++) ;
525 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
527 if (*e != '\n' && *e != '\0')
528 return; /* false alarm */
533 CopFILE_free(PL_curcop);
534 CopFILE_set(PL_curcop, s);
537 CopLINE_set(PL_curcop, atoi(n)-1);
542 * Called to gobble the appropriate amount and type of whitespace.
543 * Skips comments as well.
547 S_skipspace(pTHX_ register char *s)
549 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
550 while (s < PL_bufend && SPACE_OR_TAB(*s))
556 SSize_t oldprevlen, oldoldprevlen;
557 SSize_t oldloplen = 0, oldunilen = 0;
558 while (s < PL_bufend && isSPACE(*s)) {
559 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
564 if (s < PL_bufend && *s == '#') {
565 while (s < PL_bufend && *s != '\n')
569 if (PL_in_eval && !PL_rsfp) {
576 /* only continue to recharge the buffer if we're at the end
577 * of the buffer, we're not reading from a source filter, and
578 * we're in normal lexing mode
580 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
581 PL_lex_state == LEX_FORMLINE)
584 /* try to recharge the buffer */
585 if ((s = filter_gets(PL_linestr, PL_rsfp,
586 (prevlen = SvCUR(PL_linestr)))) == Nullch)
588 /* end of file. Add on the -p or -n magic */
589 if (PL_minus_n || PL_minus_p) {
590 sv_setpv(PL_linestr,PL_minus_p ?
591 ";}continue{print or die qq(-p destination: $!\\n)" :
593 sv_catpv(PL_linestr,";}");
594 PL_minus_n = PL_minus_p = 0;
597 sv_setpv(PL_linestr,";");
599 /* reset variables for next time we lex */
600 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
602 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
603 PL_last_lop = PL_last_uni = Nullch;
605 /* Close the filehandle. Could be from -P preprocessor,
606 * STDIN, or a regular file. If we were reading code from
607 * STDIN (because the commandline held no -e or filename)
608 * then we don't close it, we reset it so the code can
609 * read from STDIN too.
612 if (PL_preprocess && !PL_in_eval)
613 (void)PerlProc_pclose(PL_rsfp);
614 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
615 PerlIO_clearerr(PL_rsfp);
617 (void)PerlIO_close(PL_rsfp);
622 /* not at end of file, so we only read another line */
623 /* make corresponding updates to old pointers, for yyerror() */
624 oldprevlen = PL_oldbufptr - PL_bufend;
625 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
627 oldunilen = PL_last_uni - PL_bufend;
629 oldloplen = PL_last_lop - PL_bufend;
630 PL_linestart = PL_bufptr = s + prevlen;
631 PL_bufend = s + SvCUR(PL_linestr);
633 PL_oldbufptr = s + oldprevlen;
634 PL_oldoldbufptr = s + oldoldprevlen;
636 PL_last_uni = s + oldunilen;
638 PL_last_lop = s + oldloplen;
641 /* debugger active and we're not compiling the debugger code,
642 * so store the line into the debugger's array of lines
644 if (PERLDB_LINE && PL_curstash != PL_debstash) {
645 SV *sv = NEWSV(85,0);
647 sv_upgrade(sv, SVt_PVMG);
648 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
651 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
658 * Check the unary operators to ensure there's no ambiguity in how they're
659 * used. An ambiguous piece of code would be:
661 * This doesn't mean rand() + 5. Because rand() is a unary operator,
662 * the +5 is its argument.
671 if (PL_oldoldbufptr != PL_last_uni)
673 while (isSPACE(*PL_last_uni))
675 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
676 if ((t = strchr(s, '(')) && t < PL_bufptr)
678 if (ckWARN_d(WARN_AMBIGUOUS)){
681 Perl_warner(aTHX_ WARN_AMBIGUOUS,
682 "Warning: Use of \"%s\" without parens is ambiguous",
688 /* workaround to replace the UNI() macro with a function. Only the
689 * hints/uts.sh file mentions this. Other comments elsewhere in the
690 * source indicate Microport Unix might need it too.
696 #define UNI(f) return uni(f,s)
699 S_uni(pTHX_ I32 f, char *s)
704 PL_last_uni = PL_oldbufptr;
715 #endif /* CRIPPLED_CC */
718 * LOP : macro to build a list operator. Its behaviour has been replaced
719 * with a subroutine, S_lop() for which LOP is just another name.
722 #define LOP(f,x) return lop(f,x,s)
726 * Build a list operator (or something that might be one). The rules:
727 * - if we have a next token, then it's a list operator [why?]
728 * - if the next thing is an opening paren, then it's a function
729 * - else it's a list operator
733 S_lop(pTHX_ I32 f, int x, char *s)
740 PL_last_lop = PL_oldbufptr;
755 * When the lexer realizes it knows the next token (for instance,
756 * it is reordering tokens for the parser) then it can call S_force_next
757 * to know what token to return the next time the lexer is called. Caller
758 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
759 * handles the token correctly.
763 S_force_next(pTHX_ I32 type)
765 PL_nexttype[PL_nexttoke] = type;
767 if (PL_lex_state != LEX_KNOWNEXT) {
768 PL_lex_defer = PL_lex_state;
769 PL_lex_expect = PL_expect;
770 PL_lex_state = LEX_KNOWNEXT;
776 * When the lexer knows the next thing is a word (for instance, it has
777 * just seen -> and it knows that the next char is a word char, then
778 * it calls S_force_word to stick the next word into the PL_next lookahead.
781 * char *start : buffer position (must be within PL_linestr)
782 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
783 * int check_keyword : if true, Perl checks to make sure the word isn't
784 * a keyword (do this if the word is a label, e.g. goto FOO)
785 * int allow_pack : if true, : characters will also be allowed (require,
787 * int allow_initial_tick : used by the "sub" lexer only.
791 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
796 start = skipspace(start);
798 if (isIDFIRST_lazy_if(s,UTF) ||
799 (allow_pack && *s == ':') ||
800 (allow_initial_tick && *s == '\'') )
802 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
803 if (check_keyword && keyword(PL_tokenbuf, len))
805 if (token == METHOD) {
810 PL_expect = XOPERATOR;
813 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
814 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
822 * Called when the lexer wants $foo *foo &foo etc, but the program
823 * text only contains the "foo" portion. The first argument is a pointer
824 * to the "foo", and the second argument is the type symbol to prefix.
825 * Forces the next token to be a "WORD".
826 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
830 S_force_ident(pTHX_ register char *s, int kind)
833 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
834 PL_nextval[PL_nexttoke].opval = o;
837 o->op_private = OPpCONST_ENTERED;
838 /* XXX see note in pp_entereval() for why we forgo typo
839 warnings if the symbol must be introduced in an eval.
841 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
842 kind == '$' ? SVt_PV :
843 kind == '@' ? SVt_PVAV :
844 kind == '%' ? SVt_PVHV :
852 Perl_str_to_version(pTHX_ SV *sv)
857 char *start = SvPVx(sv,len);
858 bool utf = SvUTF8(sv) ? TRUE : FALSE;
859 char *end = start + len;
860 while (start < end) {
864 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
869 retval += ((NV)n)/nshift;
878 * Forces the next token to be a version number.
879 * If the next token appears to be an invalid version number, (e.g. "v2b"),
880 * and if "guessing" is TRUE, then no new token is created (and the caller
881 * must use an alternative parsing method).
885 S_force_version(pTHX_ char *s, int guessing)
887 OP *version = Nullop;
896 while (isDIGIT(*d) || *d == '_' || *d == '.')
898 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
900 s = scan_num(s, &yylval);
901 version = yylval.opval;
902 ver = cSVOPx(version)->op_sv;
903 if (SvPOK(ver) && !SvNIOK(ver)) {
904 (void)SvUPGRADE(ver, SVt_PVNV);
905 SvNVX(ver) = str_to_version(ver);
906 SvNOK_on(ver); /* hint that it is a version */
913 /* NOTE: The parser sees the package name and the VERSION swapped */
914 PL_nextval[PL_nexttoke].opval = version;
922 * Tokenize a quoted string passed in as an SV. It finds the next
923 * chunk, up to end of string or a backslash. It may make a new
924 * SV containing that chunk (if HINT_NEW_STRING is on). It also
929 S_tokeq(pTHX_ SV *sv)
940 s = SvPV_force(sv, len);
941 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
944 while (s < send && *s != '\\')
949 if ( PL_hints & HINT_NEW_STRING ) {
950 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
956 if (s + 1 < send && (s[1] == '\\'))
957 s++; /* all that, just for this */
962 SvCUR_set(sv, d - SvPVX(sv));
964 if ( PL_hints & HINT_NEW_STRING )
965 return new_constant(NULL, 0, "q", sv, pv, "q");
970 * Now come three functions related to double-quote context,
971 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
972 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
973 * interact with PL_lex_state, and create fake ( ... ) argument lists
974 * to handle functions and concatenation.
975 * They assume that whoever calls them will be setting up a fake
976 * join call, because each subthing puts a ',' after it. This lets
979 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
981 * (I'm not sure whether the spurious commas at the end of lcfirst's
982 * arguments and join's arguments are created or not).
987 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
989 * Pattern matching will set PL_lex_op to the pattern-matching op to
990 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
992 * OP_CONST and OP_READLINE are easy--just make the new op and return.
994 * Everything else becomes a FUNC.
996 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
997 * had an OP_CONST or OP_READLINE). This just sets us up for a
998 * call to S_sublex_push().
1002 S_sublex_start(pTHX)
1004 register I32 op_type = yylval.ival;
1006 if (op_type == OP_NULL) {
1007 yylval.opval = PL_lex_op;
1011 if (op_type == OP_CONST || op_type == OP_READLINE) {
1012 SV *sv = tokeq(PL_lex_stuff);
1014 if (SvTYPE(sv) == SVt_PVIV) {
1015 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
1021 nsv = newSVpvn(p, len);
1027 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
1028 PL_lex_stuff = Nullsv;
1032 PL_sublex_info.super_state = PL_lex_state;
1033 PL_sublex_info.sub_inwhat = op_type;
1034 PL_sublex_info.sub_op = PL_lex_op;
1035 PL_lex_state = LEX_INTERPPUSH;
1039 yylval.opval = PL_lex_op;
1049 * Create a new scope to save the lexing state. The scope will be
1050 * ended in S_sublex_done. Returns a '(', starting the function arguments
1051 * to the uc, lc, etc. found before.
1052 * Sets PL_lex_state to LEX_INTERPCONCAT.
1060 PL_lex_state = PL_sublex_info.super_state;
1061 SAVEI32(PL_lex_dojoin);
1062 SAVEI32(PL_lex_brackets);
1063 SAVEI32(PL_lex_casemods);
1064 SAVEI32(PL_lex_starts);
1065 SAVEI32(PL_lex_state);
1066 SAVEVPTR(PL_lex_inpat);
1067 SAVEI32(PL_lex_inwhat);
1068 SAVECOPLINE(PL_curcop);
1069 SAVEPPTR(PL_bufptr);
1070 SAVEPPTR(PL_bufend);
1071 SAVEPPTR(PL_oldbufptr);
1072 SAVEPPTR(PL_oldoldbufptr);
1073 SAVEPPTR(PL_last_lop);
1074 SAVEPPTR(PL_last_uni);
1075 SAVEPPTR(PL_linestart);
1076 SAVESPTR(PL_linestr);
1077 SAVEPPTR(PL_lex_brackstack);
1078 SAVEPPTR(PL_lex_casestack);
1080 PL_linestr = PL_lex_stuff;
1081 PL_lex_stuff = Nullsv;
1083 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1084 = SvPVX(PL_linestr);
1085 PL_bufend += SvCUR(PL_linestr);
1086 PL_last_lop = PL_last_uni = Nullch;
1087 SAVEFREESV(PL_linestr);
1089 PL_lex_dojoin = FALSE;
1090 PL_lex_brackets = 0;
1091 New(899, PL_lex_brackstack, 120, char);
1092 New(899, PL_lex_casestack, 12, char);
1093 SAVEFREEPV(PL_lex_brackstack);
1094 SAVEFREEPV(PL_lex_casestack);
1095 PL_lex_casemods = 0;
1096 *PL_lex_casestack = '\0';
1098 PL_lex_state = LEX_INTERPCONCAT;
1099 CopLINE_set(PL_curcop, PL_multi_start);
1101 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1102 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1103 PL_lex_inpat = PL_sublex_info.sub_op;
1105 PL_lex_inpat = Nullop;
1112 * Restores lexer state after a S_sublex_push.
1118 if (!PL_lex_starts++) {
1119 SV *sv = newSVpvn("",0);
1120 if (SvUTF8(PL_linestr))
1122 PL_expect = XOPERATOR;
1123 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1127 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1128 PL_lex_state = LEX_INTERPCASEMOD;
1132 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1133 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1134 PL_linestr = PL_lex_repl;
1136 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1137 PL_bufend += SvCUR(PL_linestr);
1138 PL_last_lop = PL_last_uni = Nullch;
1139 SAVEFREESV(PL_linestr);
1140 PL_lex_dojoin = FALSE;
1141 PL_lex_brackets = 0;
1142 PL_lex_casemods = 0;
1143 *PL_lex_casestack = '\0';
1145 if (SvEVALED(PL_lex_repl)) {
1146 PL_lex_state = LEX_INTERPNORMAL;
1148 /* we don't clear PL_lex_repl here, so that we can check later
1149 whether this is an evalled subst; that means we rely on the
1150 logic to ensure sublex_done() is called again only via the
1151 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1154 PL_lex_state = LEX_INTERPCONCAT;
1155 PL_lex_repl = Nullsv;
1161 PL_bufend = SvPVX(PL_linestr);
1162 PL_bufend += SvCUR(PL_linestr);
1163 PL_expect = XOPERATOR;
1164 PL_sublex_info.sub_inwhat = 0;
1172 Extracts a pattern, double-quoted string, or transliteration. This
1175 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1176 processing a pattern (PL_lex_inpat is true), a transliteration
1177 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1179 Returns a pointer to the character scanned up to. Iff this is
1180 advanced from the start pointer supplied (ie if anything was
1181 successfully parsed), will leave an OP for the substring scanned
1182 in yylval. Caller must intuit reason for not parsing further
1183 by looking at the next characters herself.
1187 double-quoted style: \r and \n
1188 regexp special ones: \D \s
1190 backrefs: \1 (deprecated in substitution replacements)
1191 case and quoting: \U \Q \E
1192 stops on @ and $, but not for $ as tail anchor
1194 In transliterations:
1195 characters are VERY literal, except for - not at the start or end
1196 of the string, which indicates a range. scan_const expands the
1197 range to the full set of intermediate characters.
1199 In double-quoted strings:
1201 double-quoted style: \r and \n
1203 backrefs: \1 (deprecated)
1204 case and quoting: \U \Q \E
1207 scan_const does *not* construct ops to handle interpolated strings.
1208 It stops processing as soon as it finds an embedded $ or @ variable
1209 and leaves it to the caller to work out what's going on.
1211 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1213 $ in pattern could be $foo or could be tail anchor. Assumption:
1214 it's a tail anchor if $ is the last thing in the string, or if it's
1215 followed by one of ")| \n\t"
1217 \1 (backreferences) are turned into $1
1219 The structure of the code is
1220 while (there's a character to process) {
1221 handle transliteration ranges
1222 skip regexp comments
1223 skip # initiated comments in //x patterns
1224 check for embedded @foo
1225 check for embedded scalars
1227 leave intact backslashes from leave (below)
1228 deprecate \1 in strings and sub replacements
1229 handle string-changing backslashes \l \U \Q \E, etc.
1230 switch (what was escaped) {
1231 handle - in a transliteration (becomes a literal -)
1232 handle \132 octal characters
1233 handle 0x15 hex characters
1234 handle \cV (control V)
1235 handle printf backslashes (\f, \r, \n, etc)
1237 } (end if backslash)
1238 } (end while character to read)
1243 S_scan_const(pTHX_ char *start)
1245 register char *send = PL_bufend; /* end of the constant */
1246 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1247 register char *s = start; /* start of the constant */
1248 register char *d = SvPVX(sv); /* destination for copies */
1249 bool dorange = FALSE; /* are we in a translit range? */
1250 bool didrange = FALSE; /* did we just finish a range? */
1251 I32 has_utf8 = FALSE; /* Output constant is UTF8 */
1252 I32 this_utf8 = UTF; /* The source string is assumed to be UTF8 */
1255 const char *leaveit = /* set of acceptably-backslashed characters */
1257 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1260 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1261 /* If we are doing a trans and we know we want UTF8 set expectation */
1262 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
1263 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1267 while (s < send || dorange) {
1268 /* get transliterations out of the way (they're most literal) */
1269 if (PL_lex_inwhat == OP_TRANS) {
1270 /* expand a range A-Z to the full set of characters. AIE! */
1272 I32 i; /* current expanded character */
1273 I32 min; /* first character in range */
1274 I32 max; /* last character in range */
1277 char *c = (char*)utf8_hop((U8*)d, -1);
1281 *c = (char)UTF_TO_NATIVE(0xff);
1282 /* mark the range as done, and continue */
1288 i = d - SvPVX(sv); /* remember current offset */
1289 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1290 d = SvPVX(sv) + i; /* refresh d after realloc */
1291 d -= 2; /* eat the first char and the - */
1293 min = (U8)*d; /* first char in range */
1294 max = (U8)d[1]; /* last char in range */
1298 "Invalid [] range \"%c-%c\" in transliteration operator",
1299 (char)min, (char)max);
1303 if ((isLOWER(min) && isLOWER(max)) ||
1304 (isUPPER(min) && isUPPER(max))) {
1306 for (i = min; i <= max; i++)
1308 *d++ = NATIVE_TO_NEED(has_utf8,i);
1310 for (i = min; i <= max; i++)
1312 *d++ = NATIVE_TO_NEED(has_utf8,i);
1317 for (i = min; i <= max; i++)
1320 /* mark the range as done, and continue */
1326 /* range begins (ignore - as first or last char) */
1327 else if (*s == '-' && s+1 < send && s != start) {
1329 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
1332 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
1344 /* if we get here, we're not doing a transliteration */
1346 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1347 except for the last char, which will be done separately. */
1348 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1350 while (s < send && *s != ')')
1351 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1353 else if (s[2] == '{' /* This should match regcomp.c */
1354 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1357 char *regparse = s + (s[2] == '{' ? 3 : 4);
1360 while (count && (c = *regparse)) {
1361 if (c == '\\' && regparse[1])
1369 if (*regparse != ')') {
1370 regparse--; /* Leave one char for continuation. */
1371 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1373 while (s < regparse)
1374 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1378 /* likewise skip #-initiated comments in //x patterns */
1379 else if (*s == '#' && PL_lex_inpat &&
1380 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1381 while (s+1 < send && *s != '\n')
1382 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1385 /* check for embedded arrays
1386 (@foo, @:foo, @'foo, @{foo}, @$foo, @+, @-)
1388 else if (*s == '@' && s[1]
1389 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1])))
1392 /* check for embedded scalars. only stop if we're sure it's a
1395 else if (*s == '$') {
1396 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1398 if (s + 1 < send && !strchr("()| \r\n\t", s[1]))
1399 break; /* in regexp, $ might be tail anchor */
1402 /* End of else if chain - OP_TRANS rejoin rest */
1405 if (*s == '\\' && s+1 < send) {
1408 /* some backslashes we leave behind */
1409 if (*leaveit && *s && strchr(leaveit, *s)) {
1410 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
1411 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1415 /* deprecate \1 in strings and substitution replacements */
1416 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1417 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1419 if (ckWARN(WARN_SYNTAX))
1420 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1425 /* string-change backslash escapes */
1426 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1431 /* if we get here, it's either a quoted -, or a digit */
1434 /* quoted - in transliterations */
1436 if (PL_lex_inwhat == OP_TRANS) {
1443 if (ckWARN(WARN_MISC) &&
1446 Perl_warner(aTHX_ WARN_MISC,
1447 "Unrecognized escape \\%c passed through",
1449 /* default action is to copy the quoted character */
1450 goto default_action;
1453 /* \132 indicates an octal constant */
1454 case '0': case '1': case '2': case '3':
1455 case '4': case '5': case '6': case '7':
1459 uv = grok_oct(s, &len, &flags, NULL);
1462 goto NUM_ESCAPE_INSERT;
1464 /* \x24 indicates a hex constant */
1468 char* e = strchr(s, '}');
1469 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1470 PERL_SCAN_DISALLOW_PREFIX;
1475 yyerror("Missing right brace on \\x{}");
1479 uv = grok_hex(s, &len, &flags, NULL);
1485 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
1486 uv = grok_hex(s, &len, &flags, NULL);
1492 /* Insert oct or hex escaped character.
1493 * There will always enough room in sv since such
1494 * escapes will be longer than any UTF-8 sequence
1495 * they can end up as. */
1497 /* We need to map to chars to ASCII before doing the tests
1500 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(uv))) {
1501 if (!has_utf8 && uv > 255) {
1502 /* Might need to recode whatever we have
1503 * accumulated so far if it contains any
1506 * (Can't we keep track of that and avoid
1507 * this rescan? --jhi)
1511 for (c = (U8 *) SvPVX(sv); c < (U8 *)d; c++) {
1512 if (!NATIVE_IS_INVARIANT(*c)) {
1517 STRLEN offset = d - SvPVX(sv);
1519 d = SvGROW(sv, SvLEN(sv) + hicount + 1) + offset;
1523 while (src >= (U8 *)SvPVX(sv)) {
1524 if (!NATIVE_IS_INVARIANT(*src)) {
1525 U8 ch = NATIVE_TO_ASCII(*src);
1526 *dst-- = UTF8_EIGHT_BIT_LO(ch);
1527 *dst-- = UTF8_EIGHT_BIT_HI(ch);
1537 if (has_utf8 || uv > 255) {
1538 d = (char*)uvchr_to_utf8((U8*)d, uv);
1540 if (PL_lex_inwhat == OP_TRANS &&
1541 PL_sublex_info.sub_op) {
1542 PL_sublex_info.sub_op->op_private |=
1543 (PL_lex_repl ? OPpTRANS_FROM_UTF
1556 /* \N{LATIN SMALL LETTER A} is a named character */
1560 char* e = strchr(s, '}');
1566 yyerror("Missing right brace on \\N{}");
1570 res = newSVpvn(s + 1, e - s - 1);
1571 res = new_constant( Nullch, 0, "charnames",
1572 res, Nullsv, "\\N{...}" );
1574 sv_utf8_upgrade(res);
1575 str = SvPV(res,len);
1576 #ifdef EBCDIC_NEVER_MIND
1577 /* charnames uses pack U and that has been
1578 * recently changed to do the below uni->native
1579 * mapping, so this would be redundant (and wrong,
1580 * the code point would be doubly converted).
1581 * But leave this in just in case the pack U change
1582 * gets revoked, but the semantics is still
1583 * desireable for charnames. --jhi */
1585 UV uv = utf8_to_uvchr((U8*)str, 0);
1588 U8 tmpbuf[UTF8_MAXLEN+1], *d;
1590 d = uvchr_to_utf8(tmpbuf, UNI_TO_NATIVE(uv));
1591 sv_setpvn(res, (char *)tmpbuf, d - tmpbuf);
1592 str = SvPV(res, len);
1596 if (!has_utf8 && SvUTF8(res)) {
1597 char *ostart = SvPVX(sv);
1598 SvCUR_set(sv, d - ostart);
1601 sv_utf8_upgrade(sv);
1602 /* this just broke our allocation above... */
1603 SvGROW(sv, send - start);
1604 d = SvPVX(sv) + SvCUR(sv);
1607 if (len > e - s + 4) { /* I _guess_ 4 is \N{} --jhi */
1608 char *odest = SvPVX(sv);
1610 SvGROW(sv, (SvLEN(sv) + len - (e - s + 4)));
1611 d = SvPVX(sv) + (d - odest);
1613 Copy(str, d, len, char);
1620 yyerror("Missing braces on \\N{}");
1623 /* \c is a control character */
1632 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
1636 /* printf-style backslashes, formfeeds, newlines, etc */
1638 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
1641 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
1644 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
1647 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
1650 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
1653 *d++ = ASCII_TO_NEED(has_utf8,'\033');
1656 *d++ = ASCII_TO_NEED(has_utf8,'\007');
1662 } /* end if (backslash) */
1665 /* If we started with encoded form, or already know we want it
1666 and then encode the next character */
1667 if ((has_utf8 || this_utf8) && !NATIVE_IS_INVARIANT((U8)(*s))) {
1669 UV uv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
1670 STRLEN need = UNISKIP(NATIVE_TO_UNI(uv));
1673 /* encoded value larger than old, need extra space (NOTE: SvCUR() not set here) */
1674 STRLEN off = d - SvPVX(sv);
1675 d = SvGROW(sv, SvLEN(sv) + (need-len)) + off;
1677 d = (char*)uvchr_to_utf8((U8*)d, uv);
1681 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1683 } /* while loop to process each character */
1685 /* terminate the string and set up the sv */
1687 SvCUR_set(sv, d - SvPVX(sv));
1688 if (SvCUR(sv) >= SvLEN(sv))
1689 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
1692 if (PL_encoding && !has_utf8) {
1693 Perl_sv_recode_to_utf8(aTHX_ sv, PL_encoding);
1698 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1699 PL_sublex_info.sub_op->op_private |=
1700 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1704 /* shrink the sv if we allocated more than we used */
1705 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1706 SvLEN_set(sv, SvCUR(sv) + 1);
1707 Renew(SvPVX(sv), SvLEN(sv), char);
1710 /* return the substring (via yylval) only if we parsed anything */
1711 if (s > PL_bufptr) {
1712 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1713 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1715 ( PL_lex_inwhat == OP_TRANS
1717 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1720 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1727 * Returns TRUE if there's more to the expression (e.g., a subscript),
1730 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1732 * ->[ and ->{ return TRUE
1733 * { and [ outside a pattern are always subscripts, so return TRUE
1734 * if we're outside a pattern and it's not { or [, then return FALSE
1735 * if we're in a pattern and the first char is a {
1736 * {4,5} (any digits around the comma) returns FALSE
1737 * if we're in a pattern and the first char is a [
1739 * [SOMETHING] has a funky algorithm to decide whether it's a
1740 * character class or not. It has to deal with things like
1741 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1742 * anything else returns TRUE
1745 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1748 S_intuit_more(pTHX_ register char *s)
1750 if (PL_lex_brackets)
1752 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1754 if (*s != '{' && *s != '[')
1759 /* In a pattern, so maybe we have {n,m}. */
1776 /* On the other hand, maybe we have a character class */
1779 if (*s == ']' || *s == '^')
1782 /* this is terrifying, and it works */
1783 int weight = 2; /* let's weigh the evidence */
1785 unsigned char un_char = 255, last_un_char;
1786 char *send = strchr(s,']');
1787 char tmpbuf[sizeof PL_tokenbuf * 4];
1789 if (!send) /* has to be an expression */
1792 Zero(seen,256,char);
1795 else if (isDIGIT(*s)) {
1797 if (isDIGIT(s[1]) && s[2] == ']')
1803 for (; s < send; s++) {
1804 last_un_char = un_char;
1805 un_char = (unsigned char)*s;
1810 weight -= seen[un_char] * 10;
1811 if (isALNUM_lazy_if(s+1,UTF)) {
1812 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1813 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1818 else if (*s == '$' && s[1] &&
1819 strchr("[#!%*<>()-=",s[1])) {
1820 if (/*{*/ strchr("])} =",s[2]))
1829 if (strchr("wds]",s[1]))
1831 else if (seen['\''] || seen['"'])
1833 else if (strchr("rnftbxcav",s[1]))
1835 else if (isDIGIT(s[1])) {
1837 while (s[1] && isDIGIT(s[1]))
1847 if (strchr("aA01! ",last_un_char))
1849 if (strchr("zZ79~",s[1]))
1851 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1852 weight -= 5; /* cope with negative subscript */
1855 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1856 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1861 if (keyword(tmpbuf, d - tmpbuf))
1864 if (un_char == last_un_char + 1)
1866 weight -= seen[un_char];
1871 if (weight >= 0) /* probably a character class */
1881 * Does all the checking to disambiguate
1883 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1884 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1886 * First argument is the stuff after the first token, e.g. "bar".
1888 * Not a method if bar is a filehandle.
1889 * Not a method if foo is a subroutine prototyped to take a filehandle.
1890 * Not a method if it's really "Foo $bar"
1891 * Method if it's "foo $bar"
1892 * Not a method if it's really "print foo $bar"
1893 * Method if it's really "foo package::" (interpreted as package->foo)
1894 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1895 * Not a method if bar is a filehandle or package, but is quoted with
1900 S_intuit_method(pTHX_ char *start, GV *gv)
1902 char *s = start + (*start == '$');
1903 char tmpbuf[sizeof PL_tokenbuf];
1911 if ((cv = GvCVu(gv))) {
1912 char *proto = SvPVX(cv);
1922 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1923 /* start is the beginning of the possible filehandle/object,
1924 * and s is the end of it
1925 * tmpbuf is a copy of it
1928 if (*start == '$') {
1929 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1934 return *s == '(' ? FUNCMETH : METHOD;
1936 if (!keyword(tmpbuf, len)) {
1937 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1942 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1943 if (indirgv && GvCVu(indirgv))
1945 /* filehandle or package name makes it a method */
1946 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1948 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1949 return 0; /* no assumptions -- "=>" quotes bearword */
1951 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1952 newSVpvn(tmpbuf,len));
1953 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1957 return *s == '(' ? FUNCMETH : METHOD;
1965 * Return a string of Perl code to load the debugger. If PERL5DB
1966 * is set, it will return the contents of that, otherwise a
1967 * compile-time require of perl5db.pl.
1974 char *pdb = PerlEnv_getenv("PERL5DB");
1978 SETERRNO(0,SS$_NORMAL);
1979 return "BEGIN { require 'perl5db.pl' }";
1985 /* Encoded script support. filter_add() effectively inserts a
1986 * 'pre-processing' function into the current source input stream.
1987 * Note that the filter function only applies to the current source file
1988 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1990 * The datasv parameter (which may be NULL) can be used to pass
1991 * private data to this instance of the filter. The filter function
1992 * can recover the SV using the FILTER_DATA macro and use it to
1993 * store private buffers and state information.
1995 * The supplied datasv parameter is upgraded to a PVIO type
1996 * and the IoDIRP/IoANY field is used to store the function pointer,
1997 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1998 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1999 * private use must be set using malloc'd pointers.
2003 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
2008 if (!PL_rsfp_filters)
2009 PL_rsfp_filters = newAV();
2011 datasv = NEWSV(255,0);
2012 if (!SvUPGRADE(datasv, SVt_PVIO))
2013 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
2014 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
2015 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
2016 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
2017 (void*)funcp, SvPV_nolen(datasv)));
2018 av_unshift(PL_rsfp_filters, 1);
2019 av_store(PL_rsfp_filters, 0, datasv) ;
2024 /* Delete most recently added instance of this filter function. */
2026 Perl_filter_del(pTHX_ filter_t funcp)
2029 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", (void*)funcp));
2030 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
2032 /* if filter is on top of stack (usual case) just pop it off */
2033 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
2034 if (IoANY(datasv) == (void *)funcp) {
2035 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
2036 IoANY(datasv) = (void *)NULL;
2037 sv_free(av_pop(PL_rsfp_filters));
2041 /* we need to search for the correct entry and clear it */
2042 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
2046 /* Invoke the n'th filter function for the current rsfp. */
2048 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
2051 /* 0 = read one text line */
2056 if (!PL_rsfp_filters)
2058 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
2059 /* Provide a default input filter to make life easy. */
2060 /* Note that we append to the line. This is handy. */
2061 DEBUG_P(PerlIO_printf(Perl_debug_log,
2062 "filter_read %d: from rsfp\n", idx));
2066 int old_len = SvCUR(buf_sv) ;
2068 /* ensure buf_sv is large enough */
2069 SvGROW(buf_sv, old_len + maxlen) ;
2070 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
2071 if (PerlIO_error(PL_rsfp))
2072 return -1; /* error */
2074 return 0 ; /* end of file */
2076 SvCUR_set(buf_sv, old_len + len) ;
2079 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2080 if (PerlIO_error(PL_rsfp))
2081 return -1; /* error */
2083 return 0 ; /* end of file */
2086 return SvCUR(buf_sv);
2088 /* Skip this filter slot if filter has been deleted */
2089 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
2090 DEBUG_P(PerlIO_printf(Perl_debug_log,
2091 "filter_read %d: skipped (filter deleted)\n",
2093 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
2095 /* Get function pointer hidden within datasv */
2096 funcp = (filter_t)IoANY(datasv);
2097 DEBUG_P(PerlIO_printf(Perl_debug_log,
2098 "filter_read %d: via function %p (%s)\n",
2099 idx, (void*)funcp, SvPV_nolen(datasv)));
2100 /* Call function. The function is expected to */
2101 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2102 /* Return: <0:error, =0:eof, >0:not eof */
2103 return (*funcp)(aTHX_ idx, buf_sv, maxlen);
2107 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2109 #ifdef PERL_CR_FILTER
2110 if (!PL_rsfp_filters) {
2111 filter_add(S_cr_textfilter,NULL);
2114 if (PL_rsfp_filters) {
2117 SvCUR_set(sv, 0); /* start with empty line */
2118 if (FILTER_READ(0, sv, 0) > 0)
2119 return ( SvPVX(sv) ) ;
2124 return (sv_gets(sv, fp, append));
2128 S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2132 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2136 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2137 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2139 return GvHV(gv); /* Foo:: */
2142 /* use constant CLASS => 'MyClass' */
2143 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2145 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2146 pkgname = SvPV_nolen(sv);
2150 return gv_stashpv(pkgname, FALSE);
2154 static char* exp_name[] =
2155 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2156 "ATTRTERM", "TERMBLOCK"
2163 Works out what to call the token just pulled out of the input
2164 stream. The yacc parser takes care of taking the ops we return and
2165 stitching them into a tree.
2171 if read an identifier
2172 if we're in a my declaration
2173 croak if they tried to say my($foo::bar)
2174 build the ops for a my() declaration
2175 if it's an access to a my() variable
2176 are we in a sort block?
2177 croak if my($a); $a <=> $b
2178 build ops for access to a my() variable
2179 if in a dq string, and they've said @foo and we can't find @foo
2181 build ops for a bareword
2182 if we already built the token before, use it.
2185 #ifdef USE_PURE_BISON
2187 Perl_yylex_r(pTHX_ YYSTYPE *lvalp, int *lcharp)
2192 yylval_pointer[yyactlevel] = lvalp;
2193 yychar_pointer[yyactlevel] = lcharp;
2194 if (yyactlevel >= YYMAXLEVEL)
2195 Perl_croak(aTHX_ "panic: YYMAXLEVEL");
2197 r = Perl_yylex(aTHX);
2207 #pragma segment Perl_yylex
2220 /* check if there's an identifier for us to look at */
2221 if (PL_pending_ident)
2222 return S_pending_ident(aTHX);
2224 /* no identifier pending identification */
2226 switch (PL_lex_state) {
2228 case LEX_NORMAL: /* Some compilers will produce faster */
2229 case LEX_INTERPNORMAL: /* code if we comment these out. */
2233 /* when we've already built the next token, just pull it out of the queue */
2236 yylval = PL_nextval[PL_nexttoke];
2238 PL_lex_state = PL_lex_defer;
2239 PL_expect = PL_lex_expect;
2240 PL_lex_defer = LEX_NORMAL;
2242 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2243 "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr,
2244 (IV)PL_nexttype[PL_nexttoke]); });
2246 return(PL_nexttype[PL_nexttoke]);
2248 /* interpolated case modifiers like \L \U, including \Q and \E.
2249 when we get here, PL_bufptr is at the \
2251 case LEX_INTERPCASEMOD:
2253 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2254 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2256 /* handle \E or end of string */
2257 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2261 if (PL_lex_casemods) {
2262 oldmod = PL_lex_casestack[--PL_lex_casemods];
2263 PL_lex_casestack[PL_lex_casemods] = '\0';
2265 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2267 PL_lex_state = LEX_INTERPCONCAT;
2271 if (PL_bufptr != PL_bufend)
2273 PL_lex_state = LEX_INTERPCONCAT;
2277 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2278 "### Saw case modifier at '%s'\n", PL_bufptr); });
2280 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2281 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2282 if (strchr("LU", *s) &&
2283 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2285 PL_lex_casestack[--PL_lex_casemods] = '\0';
2288 if (PL_lex_casemods > 10) {
2289 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2290 if (newlb != PL_lex_casestack) {
2292 PL_lex_casestack = newlb;
2295 PL_lex_casestack[PL_lex_casemods++] = *s;
2296 PL_lex_casestack[PL_lex_casemods] = '\0';
2297 PL_lex_state = LEX_INTERPCONCAT;
2298 PL_nextval[PL_nexttoke].ival = 0;
2301 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2303 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2305 PL_nextval[PL_nexttoke].ival = OP_LC;
2307 PL_nextval[PL_nexttoke].ival = OP_UC;
2309 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2311 Perl_croak(aTHX_ "panic: yylex");
2314 if (PL_lex_starts) {
2323 case LEX_INTERPPUSH:
2324 return sublex_push();
2326 case LEX_INTERPSTART:
2327 if (PL_bufptr == PL_bufend)
2328 return sublex_done();
2329 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2330 "### Interpolated variable at '%s'\n", PL_bufptr); });
2332 PL_lex_dojoin = (*PL_bufptr == '@');
2333 PL_lex_state = LEX_INTERPNORMAL;
2334 if (PL_lex_dojoin) {
2335 PL_nextval[PL_nexttoke].ival = 0;
2337 #ifdef USE_5005THREADS
2338 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2339 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2340 force_next(PRIVATEREF);
2342 force_ident("\"", '$');
2343 #endif /* USE_5005THREADS */
2344 PL_nextval[PL_nexttoke].ival = 0;
2346 PL_nextval[PL_nexttoke].ival = 0;
2348 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2351 if (PL_lex_starts++) {
2357 case LEX_INTERPENDMAYBE:
2358 if (intuit_more(PL_bufptr)) {
2359 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2365 if (PL_lex_dojoin) {
2366 PL_lex_dojoin = FALSE;
2367 PL_lex_state = LEX_INTERPCONCAT;
2370 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2371 && SvEVALED(PL_lex_repl))
2373 if (PL_bufptr != PL_bufend)
2374 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2375 PL_lex_repl = Nullsv;
2378 case LEX_INTERPCONCAT:
2380 if (PL_lex_brackets)
2381 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2383 if (PL_bufptr == PL_bufend)
2384 return sublex_done();
2386 if (SvIVX(PL_linestr) == '\'') {
2387 SV *sv = newSVsv(PL_linestr);
2390 else if ( PL_hints & HINT_NEW_RE )
2391 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2392 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2396 s = scan_const(PL_bufptr);
2398 PL_lex_state = LEX_INTERPCASEMOD;
2400 PL_lex_state = LEX_INTERPSTART;
2403 if (s != PL_bufptr) {
2404 PL_nextval[PL_nexttoke] = yylval;
2407 if (PL_lex_starts++)
2417 PL_lex_state = LEX_NORMAL;
2418 s = scan_formline(PL_bufptr);
2419 if (!PL_lex_formbrack)
2425 PL_oldoldbufptr = PL_oldbufptr;
2428 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2429 exp_name[PL_expect], s);
2435 if (isIDFIRST_lazy_if(s,UTF))
2437 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2440 goto fake_eof; /* emulate EOF on ^D or ^Z */
2445 if (PL_lex_brackets)
2446 yyerror("Missing right curly or square bracket");
2447 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2448 "### Tokener got EOF\n");
2452 if (s++ < PL_bufend)
2453 goto retry; /* ignore stray nulls */
2456 if (!PL_in_eval && !PL_preambled) {
2457 PL_preambled = TRUE;
2458 sv_setpv(PL_linestr,incl_perldb());
2459 if (SvCUR(PL_linestr))
2460 sv_catpv(PL_linestr,";");
2462 while(AvFILLp(PL_preambleav) >= 0) {
2463 SV *tmpsv = av_shift(PL_preambleav);
2464 sv_catsv(PL_linestr, tmpsv);
2465 sv_catpv(PL_linestr, ";");
2468 sv_free((SV*)PL_preambleav);
2469 PL_preambleav = NULL;
2471 if (PL_minus_n || PL_minus_p) {
2472 sv_catpv(PL_linestr, "LINE: while (<>) {");
2474 sv_catpv(PL_linestr,"chomp;");
2477 if (strchr("/'\"", *PL_splitstr)
2478 && strchr(PL_splitstr + 1, *PL_splitstr))
2479 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
2482 s = "'~#\200\1'"; /* surely one char is unused...*/
2483 while (s[1] && strchr(PL_splitstr, *s)) s++;
2485 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s%c",
2486 "q" + (delim == '\''), delim);
2487 for (s = PL_splitstr; *s; s++) {
2489 sv_catpvn(PL_linestr, "\\", 1);
2490 sv_catpvn(PL_linestr, s, 1);
2492 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2496 sv_catpv(PL_linestr,"our @F=split(' ');");
2499 sv_catpv(PL_linestr, "\n");
2500 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2501 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2502 PL_last_lop = PL_last_uni = Nullch;
2503 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2504 SV *sv = NEWSV(85,0);
2506 sv_upgrade(sv, SVt_PVMG);
2507 sv_setsv(sv,PL_linestr);
2510 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2515 bof = PL_rsfp ? TRUE : FALSE;
2516 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2519 if (PL_preprocess && !PL_in_eval)
2520 (void)PerlProc_pclose(PL_rsfp);
2521 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2522 PerlIO_clearerr(PL_rsfp);
2524 (void)PerlIO_close(PL_rsfp);
2526 PL_doextract = FALSE;
2528 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2529 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2530 sv_catpv(PL_linestr,";}");
2531 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2532 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2533 PL_last_lop = PL_last_uni = Nullch;
2534 PL_minus_n = PL_minus_p = 0;
2537 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2538 PL_last_lop = PL_last_uni = Nullch;
2539 sv_setpv(PL_linestr,"");
2540 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2542 /* if it looks like the start of a BOM, check if it in fact is */
2543 else if (bof && (!*s || *(U8*)s == 0xEF || *(U8*)s >= 0xFE)) {
2544 #ifdef PERLIO_IS_STDIO
2545 # ifdef __GNU_LIBRARY__
2546 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
2547 # define FTELL_FOR_PIPE_IS_BROKEN
2551 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2552 # define FTELL_FOR_PIPE_IS_BROKEN
2557 #ifdef FTELL_FOR_PIPE_IS_BROKEN
2558 /* This loses the possibility to detect the bof
2559 * situation on perl -P when the libc5 is being used.
2560 * Workaround? Maybe attach some extra state to PL_rsfp?
2563 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2565 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2568 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2569 s = swallow_bom((U8*)s);
2573 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2574 PL_doextract = FALSE;
2576 /* Incest with pod. */
2577 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2578 sv_setpv(PL_linestr, "");
2579 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2580 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2581 PL_last_lop = PL_last_uni = Nullch;
2582 PL_doextract = FALSE;
2586 } while (PL_doextract);
2587 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2588 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2589 SV *sv = NEWSV(85,0);
2591 sv_upgrade(sv, SVt_PVMG);
2592 sv_setsv(sv,PL_linestr);
2595 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2597 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2598 PL_last_lop = PL_last_uni = Nullch;
2599 if (CopLINE(PL_curcop) == 1) {
2600 while (s < PL_bufend && isSPACE(*s))
2602 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2606 if (*s == '#' && *(s+1) == '!')
2608 #ifdef ALTERNATE_SHEBANG
2610 static char as[] = ALTERNATE_SHEBANG;
2611 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2612 d = s + (sizeof(as) - 1);
2614 #endif /* ALTERNATE_SHEBANG */
2623 while (*d && !isSPACE(*d))
2627 #ifdef ARG_ZERO_IS_SCRIPT
2628 if (ipathend > ipath) {
2630 * HP-UX (at least) sets argv[0] to the script name,
2631 * which makes $^X incorrect. And Digital UNIX and Linux,
2632 * at least, set argv[0] to the basename of the Perl
2633 * interpreter. So, having found "#!", we'll set it right.
2635 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV)); /* $^X */
2636 assert(SvPOK(x) || SvGMAGICAL(x));
2637 if (sv_eq(x, CopFILESV(PL_curcop))) {
2638 sv_setpvn(x, ipath, ipathend - ipath);
2641 TAINT_NOT; /* $^X is always tainted, but that's OK */
2643 #endif /* ARG_ZERO_IS_SCRIPT */
2648 d = instr(s,"perl -");
2650 d = instr(s,"perl");
2652 /* avoid getting into infinite loops when shebang
2653 * line contains "Perl" rather than "perl" */
2655 for (d = ipathend-4; d >= ipath; --d) {
2656 if ((*d == 'p' || *d == 'P')
2657 && !ibcmp(d, "perl", 4))
2667 #ifdef ALTERNATE_SHEBANG
2669 * If the ALTERNATE_SHEBANG on this system starts with a
2670 * character that can be part of a Perl expression, then if
2671 * we see it but not "perl", we're probably looking at the
2672 * start of Perl code, not a request to hand off to some
2673 * other interpreter. Similarly, if "perl" is there, but
2674 * not in the first 'word' of the line, we assume the line
2675 * contains the start of the Perl program.
2677 if (d && *s != '#') {
2679 while (*c && !strchr("; \t\r\n\f\v#", *c))
2682 d = Nullch; /* "perl" not in first word; ignore */
2684 *s = '#'; /* Don't try to parse shebang line */
2686 #endif /* ALTERNATE_SHEBANG */
2687 #ifndef MACOS_TRADITIONAL
2692 !instr(s,"indir") &&
2693 instr(PL_origargv[0],"perl"))
2699 while (s < PL_bufend && isSPACE(*s))
2701 if (s < PL_bufend) {
2702 Newz(899,newargv,PL_origargc+3,char*);
2704 while (s < PL_bufend && !isSPACE(*s))
2707 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2710 newargv = PL_origargv;
2712 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
2713 Perl_croak(aTHX_ "Can't exec %s", ipath);
2717 U32 oldpdb = PL_perldb;
2718 bool oldn = PL_minus_n;
2719 bool oldp = PL_minus_p;
2721 while (*d && !isSPACE(*d)) d++;
2722 while (SPACE_OR_TAB(*d)) d++;
2725 bool switches_done = PL_doswitches;
2727 if (*d == 'M' || *d == 'm') {
2729 while (*d && !isSPACE(*d)) d++;
2730 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2733 d = moreswitches(d);
2735 if ((PERLDB_LINE && !oldpdb) ||
2736 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2737 /* if we have already added "LINE: while (<>) {",
2738 we must not do it again */
2740 sv_setpv(PL_linestr, "");
2741 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2742 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2743 PL_last_lop = PL_last_uni = Nullch;
2744 PL_preambled = FALSE;
2746 (void)gv_fetchfile(PL_origfilename);
2749 if (PL_doswitches && !switches_done) {
2750 int argc = PL_origargc;
2751 char **argv = PL_origargv;
2754 } while (argc && argv[0][0] == '-' && argv[0][1]);
2755 init_argv_symbols(argc,argv);
2761 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2763 PL_lex_state = LEX_FORMLINE;
2768 #ifdef PERL_STRICT_CR
2769 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2771 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2773 case ' ': case '\t': case '\f': case 013:
2774 #ifdef MACOS_TRADITIONAL
2781 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2782 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2783 /* handle eval qq[#line 1 "foo"\n ...] */
2784 CopLINE_dec(PL_curcop);
2788 while (s < d && *s != '\n')
2792 else if (s > d) /* Found by Ilya: feed random input to Perl. */
2793 Perl_croak(aTHX_ "panic: input overflow");
2795 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2797 PL_lex_state = LEX_FORMLINE;
2807 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2814 while (s < PL_bufend && SPACE_OR_TAB(*s))
2817 if (strnEQ(s,"=>",2)) {
2818 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2819 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2820 "### Saw unary minus before =>, forcing word '%s'\n", s);
2822 OPERATOR('-'); /* unary minus */
2824 PL_last_uni = PL_oldbufptr;
2826 case 'r': ftst = OP_FTEREAD; break;
2827 case 'w': ftst = OP_FTEWRITE; break;
2828 case 'x': ftst = OP_FTEEXEC; break;
2829 case 'o': ftst = OP_FTEOWNED; break;
2830 case 'R': ftst = OP_FTRREAD; break;
2831 case 'W': ftst = OP_FTRWRITE; break;
2832 case 'X': ftst = OP_FTREXEC; break;
2833 case 'O': ftst = OP_FTROWNED; break;
2834 case 'e': ftst = OP_FTIS; break;
2835 case 'z': ftst = OP_FTZERO; break;
2836 case 's': ftst = OP_FTSIZE; break;
2837 case 'f': ftst = OP_FTFILE; break;
2838 case 'd': ftst = OP_FTDIR; break;
2839 case 'l': ftst = OP_FTLINK; break;
2840 case 'p': ftst = OP_FTPIPE; break;
2841 case 'S': ftst = OP_FTSOCK; break;
2842 case 'u': ftst = OP_FTSUID; break;
2843 case 'g': ftst = OP_FTSGID; break;
2844 case 'k': ftst = OP_FTSVTX; break;
2845 case 'b': ftst = OP_FTBLK; break;
2846 case 'c': ftst = OP_FTCHR; break;
2847 case 't': ftst = OP_FTTTY; break;
2848 case 'T': ftst = OP_FTTEXT; break;
2849 case 'B': ftst = OP_FTBINARY; break;
2850 case 'M': case 'A': case 'C':
2851 gv_fetchpv("\024",TRUE, SVt_PV);
2853 case 'M': ftst = OP_FTMTIME; break;
2854 case 'A': ftst = OP_FTATIME; break;
2855 case 'C': ftst = OP_FTCTIME; break;
2863 PL_last_lop_op = ftst;
2864 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2865 "### Saw file test %c\n", (int)ftst);
2870 /* Assume it was a minus followed by a one-letter named
2871 * subroutine call (or a -bareword), then. */
2872 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2873 "### %c looked like a file test but was not\n",
2882 if (PL_expect == XOPERATOR)
2887 else if (*s == '>') {
2890 if (isIDFIRST_lazy_if(s,UTF)) {
2891 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2899 if (PL_expect == XOPERATOR)
2902 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2904 OPERATOR('-'); /* unary minus */
2911 if (PL_expect == XOPERATOR)
2916 if (PL_expect == XOPERATOR)
2919 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2925 if (PL_expect != XOPERATOR) {
2926 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2927 PL_expect = XOPERATOR;
2928 force_ident(PL_tokenbuf, '*');
2941 if (PL_expect == XOPERATOR) {
2945 PL_tokenbuf[0] = '%';
2946 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2947 if (!PL_tokenbuf[1]) {
2949 yyerror("Final % should be \\% or %name");
2952 PL_pending_ident = '%';
2971 switch (PL_expect) {
2974 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2976 PL_bufptr = s; /* update in case we back off */
2982 PL_expect = XTERMBLOCK;
2986 while (isIDFIRST_lazy_if(s,UTF)) {
2987 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2988 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2989 if (tmp < 0) tmp = -tmp;
3004 d = scan_str(d,TRUE,TRUE);
3006 /* MUST advance bufptr here to avoid bogus
3007 "at end of line" context messages from yyerror().
3009 PL_bufptr = s + len;
3010 yyerror("Unterminated attribute parameter in attribute list");
3013 return 0; /* EOF indicator */
3017 SV *sv = newSVpvn(s, len);
3018 sv_catsv(sv, PL_lex_stuff);
3019 attrs = append_elem(OP_LIST, attrs,
3020 newSVOP(OP_CONST, 0, sv));
3021 SvREFCNT_dec(PL_lex_stuff);
3022 PL_lex_stuff = Nullsv;
3025 if (!PL_in_my && len == 6 && strnEQ(s, "lvalue", len))
3026 CvLVALUE_on(PL_compcv);
3027 else if (!PL_in_my && len == 6 && strnEQ(s, "locked", len))
3028 CvLOCKED_on(PL_compcv);
3029 else if (!PL_in_my && len == 6 && strnEQ(s, "method", len))
3030 CvMETHOD_on(PL_compcv);
3032 else if (PL_in_my == KEY_our && len == 6 && strnEQ(s, "unique", len))
3033 GvUNIQUE_on(cGVOPx_gv(yylval.opval));
3035 /* After we've set the flags, it could be argued that
3036 we don't need to do the attributes.pm-based setting
3037 process, and shouldn't bother appending recognized
3038 flags. To experiment with that, uncomment the
3039 following "else": */
3041 attrs = append_elem(OP_LIST, attrs,
3042 newSVOP(OP_CONST, 0,
3046 if (*s == ':' && s[1] != ':')
3049 break; /* require real whitespace or :'s */
3051 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3052 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
3053 char q = ((*s == '\'') ? '"' : '\'');
3054 /* If here for an expression, and parsed no attrs, back off. */
3055 if (tmp == '=' && !attrs) {
3059 /* MUST advance bufptr here to avoid bogus "at end of line"
3060 context messages from yyerror().
3064 yyerror("Unterminated attribute list");
3066 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
3074 PL_nextval[PL_nexttoke].opval = attrs;
3082 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3083 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3099 if (PL_lex_brackets <= 0)
3100 yyerror("Unmatched right square bracket");
3103 if (PL_lex_state == LEX_INTERPNORMAL) {
3104 if (PL_lex_brackets == 0) {
3105 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3106 PL_lex_state = LEX_INTERPEND;
3113 if (PL_lex_brackets > 100) {
3114 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
3115 if (newlb != PL_lex_brackstack) {
3117 PL_lex_brackstack = newlb;
3120 switch (PL_expect) {
3122 if (PL_lex_formbrack) {
3126 if (PL_oldoldbufptr == PL_last_lop)
3127 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3129 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3130 OPERATOR(HASHBRACK);
3132 while (s < PL_bufend && SPACE_OR_TAB(*s))
3135 PL_tokenbuf[0] = '\0';
3136 if (d < PL_bufend && *d == '-') {
3137 PL_tokenbuf[0] = '-';
3139 while (d < PL_bufend && SPACE_OR_TAB(*d))
3142 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3143 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3145 while (d < PL_bufend && SPACE_OR_TAB(*d))
3148 char minus = (PL_tokenbuf[0] == '-');
3149 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3157 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3162 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3167 if (PL_oldoldbufptr == PL_last_lop)
3168 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3170 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3173 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
3175 /* This hack is to get the ${} in the message. */
3177 yyerror("syntax error");
3180 OPERATOR(HASHBRACK);
3182 /* This hack serves to disambiguate a pair of curlies
3183 * as being a block or an anon hash. Normally, expectation
3184 * determines that, but in cases where we're not in a
3185 * position to expect anything in particular (like inside
3186 * eval"") we have to resolve the ambiguity. This code
3187 * covers the case where the first term in the curlies is a
3188 * quoted string. Most other cases need to be explicitly
3189 * disambiguated by prepending a `+' before the opening
3190 * curly in order to force resolution as an anon hash.
3192 * XXX should probably propagate the outer expectation
3193 * into eval"" to rely less on this hack, but that could
3194 * potentially break current behavior of eval"".
3198 if (*s == '\'' || *s == '"' || *s == '`') {
3199 /* common case: get past first string, handling escapes */
3200 for (t++; t < PL_bufend && *t != *s;)
3201 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3205 else if (*s == 'q') {
3208 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3212 char open, close, term;
3215 while (t < PL_bufend && isSPACE(*t))
3219 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3223 for (t++; t < PL_bufend; t++) {
3224 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3226 else if (*t == open)
3230 for (t++; t < PL_bufend; t++) {
3231 if (*t == '\\' && t+1 < PL_bufend)
3233 else if (*t == close && --brackets <= 0)
3235 else if (*t == open)
3241 else if (isALNUM_lazy_if(t,UTF)) {
3243 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3246 while (t < PL_bufend && isSPACE(*t))
3248 /* if comma follows first term, call it an anon hash */
3249 /* XXX it could be a comma expression with loop modifiers */
3250 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3251 || (*t == '=' && t[1] == '>')))
3252 OPERATOR(HASHBRACK);
3253 if (PL_expect == XREF)
3256 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3262 yylval.ival = CopLINE(PL_curcop);
3263 if (isSPACE(*s) || *s == '#')
3264 PL_copline = NOLINE; /* invalidate current command line number */
3269 if (PL_lex_brackets <= 0)
3270 yyerror("Unmatched right curly bracket");
3272 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3273 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3274 PL_lex_formbrack = 0;
3275 if (PL_lex_state == LEX_INTERPNORMAL) {
3276 if (PL_lex_brackets == 0) {
3277 if (PL_expect & XFAKEBRACK) {
3278 PL_expect &= XENUMMASK;
3279 PL_lex_state = LEX_INTERPEND;
3281 return yylex(); /* ignore fake brackets */
3283 if (*s == '-' && s[1] == '>')
3284 PL_lex_state = LEX_INTERPENDMAYBE;
3285 else if (*s != '[' && *s != '{')
3286 PL_lex_state = LEX_INTERPEND;
3289 if (PL_expect & XFAKEBRACK) {
3290 PL_expect &= XENUMMASK;
3292 return yylex(); /* ignore fake brackets */
3302 if (PL_expect == XOPERATOR) {
3303 if (ckWARN(WARN_SEMICOLON)
3304 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3306 CopLINE_dec(PL_curcop);
3307 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3308 CopLINE_inc(PL_curcop);
3313 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3315 PL_expect = XOPERATOR;
3316 force_ident(PL_tokenbuf, '&');
3320 yylval.ival = (OPpENTERSUB_AMPER<<8);
3339 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3340 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3342 if (PL_expect == XSTATE && isALPHA(tmp) &&
3343 (s == PL_linestart+1 || s[-2] == '\n') )
3345 if (PL_in_eval && !PL_rsfp) {
3350 if (strnEQ(s,"=cut",4)) {
3364 PL_doextract = TRUE;
3367 if (PL_lex_brackets < PL_lex_formbrack) {
3369 #ifdef PERL_STRICT_CR
3370 for (t = s; SPACE_OR_TAB(*t); t++) ;
3372 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3374 if (*t == '\n' || *t == '#') {
3392 if (PL_expect != XOPERATOR) {
3393 if (s[1] != '<' && !strchr(s,'>'))
3396 s = scan_heredoc(s);
3398 s = scan_inputsymbol(s);
3399 TERM(sublex_start());
3404 SHop(OP_LEFT_SHIFT);
3418 SHop(OP_RIGHT_SHIFT);
3427 if (PL_expect == XOPERATOR) {
3428 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3431 return ','; /* grandfather non-comma-format format */
3435 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3436 PL_tokenbuf[0] = '@';
3437 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3438 sizeof PL_tokenbuf - 1, FALSE);
3439 if (PL_expect == XOPERATOR)
3440 no_op("Array length", s);
3441 if (!PL_tokenbuf[1])
3443 PL_expect = XOPERATOR;
3444 PL_pending_ident = '#';
3448 PL_tokenbuf[0] = '$';
3449 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3450 sizeof PL_tokenbuf - 1, FALSE);
3451 if (PL_expect == XOPERATOR)
3453 if (!PL_tokenbuf[1]) {
3455 yyerror("Final $ should be \\$ or $name");
3459 /* This kludge not intended to be bulletproof. */
3460 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3461 yylval.opval = newSVOP(OP_CONST, 0,
3462 newSViv(PL_compiling.cop_arybase));
3463 yylval.opval->op_private = OPpCONST_ARYBASE;
3469 if (PL_lex_state == LEX_NORMAL)
3472 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3475 PL_tokenbuf[0] = '@';
3476 if (ckWARN(WARN_SYNTAX)) {
3478 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3481 PL_bufptr = skipspace(PL_bufptr);
3482 while (t < PL_bufend && *t != ']')
3484 Perl_warner(aTHX_ WARN_SYNTAX,
3485 "Multidimensional syntax %.*s not supported",
3486 (t - PL_bufptr) + 1, PL_bufptr);
3490 else if (*s == '{') {
3491 PL_tokenbuf[0] = '%';
3492 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3493 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3495 char tmpbuf[sizeof PL_tokenbuf];
3497 for (t++; isSPACE(*t); t++) ;
3498 if (isIDFIRST_lazy_if(t,UTF)) {
3499 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3500 for (; isSPACE(*t); t++) ;
3501 if (*t == ';' && get_cv(tmpbuf, FALSE))
3502 Perl_warner(aTHX_ WARN_SYNTAX,
3503 "You need to quote \"%s\"", tmpbuf);
3509 PL_expect = XOPERATOR;
3510 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3511 bool islop = (PL_last_lop == PL_oldoldbufptr);
3512 if (!islop || PL_last_lop_op == OP_GREPSTART)
3513 PL_expect = XOPERATOR;
3514 else if (strchr("$@\"'`q", *s))
3515 PL_expect = XTERM; /* e.g. print $fh "foo" */
3516 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3517 PL_expect = XTERM; /* e.g. print $fh &sub */
3518 else if (isIDFIRST_lazy_if(s,UTF)) {
3519 char tmpbuf[sizeof PL_tokenbuf];
3520 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3521 if ((tmp = keyword(tmpbuf, len))) {
3522 /* binary operators exclude handle interpretations */
3534 PL_expect = XTERM; /* e.g. print $fh length() */
3539 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3540 if (gv && GvCVu(gv))
3541 PL_expect = XTERM; /* e.g. print $fh subr() */
3544 else if (isDIGIT(*s))
3545 PL_expect = XTERM; /* e.g. print $fh 3 */
3546 else if (*s == '.' && isDIGIT(s[1]))
3547 PL_expect = XTERM; /* e.g. print $fh .3 */
3548 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3549 PL_expect = XTERM; /* e.g. print $fh -1 */
3550 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3551 PL_expect = XTERM; /* print $fh <<"EOF" */
3553 PL_pending_ident = '$';
3557 if (PL_expect == XOPERATOR)
3559 PL_tokenbuf[0] = '@';
3560 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3561 if (!PL_tokenbuf[1]) {
3563 yyerror("Final @ should be \\@ or @name");
3566 if (PL_lex_state == LEX_NORMAL)
3568 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3570 PL_tokenbuf[0] = '%';
3572 /* Warn about @ where they meant $. */
3573 if (ckWARN(WARN_SYNTAX)) {
3574 if (*s == '[' || *s == '{') {
3576 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3578 if (*t == '}' || *t == ']') {
3580 PL_bufptr = skipspace(PL_bufptr);
3581 Perl_warner(aTHX_ WARN_SYNTAX,
3582 "Scalar value %.*s better written as $%.*s",
3583 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3588 PL_pending_ident = '@';
3591 case '/': /* may either be division or pattern */
3592 case '?': /* may either be conditional or pattern */
3593 if (PL_expect != XOPERATOR) {
3594 /* Disable warning on "study /blah/" */
3595 if (PL_oldoldbufptr == PL_last_uni
3596 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3597 || memNE(PL_last_uni, "study", 5)
3598 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3600 s = scan_pat(s,OP_MATCH);
3601 TERM(sublex_start());
3609 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3610 #ifdef PERL_STRICT_CR
3613 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3615 && (s == PL_linestart || s[-1] == '\n') )
3617 PL_lex_formbrack = 0;
3621 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3627 yylval.ival = OPf_SPECIAL;
3633 if (PL_expect != XOPERATOR)
3638 case '0': case '1': case '2': case '3': case '4':
3639 case '5': case '6': case '7': case '8': case '9':
3640 s = scan_num(s, &yylval);
3641 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3642 "### Saw number in '%s'\n", s);
3644 if (PL_expect == XOPERATOR)
3649 s = scan_str(s,FALSE,FALSE);
3650 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3651 "### Saw string before '%s'\n", s);
3653 if (PL_expect == XOPERATOR) {
3654 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3657 return ','; /* grandfather non-comma-format format */
3663 missingterm((char*)0);
3664 yylval.ival = OP_CONST;
3665 TERM(sublex_start());
3668 s = scan_str(s,FALSE,FALSE);
3669 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3670 "### Saw string before '%s'\n", s);
3672 if (PL_expect == XOPERATOR) {
3673 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3676 return ','; /* grandfather non-comma-format format */
3682 missingterm((char*)0);
3683 yylval.ival = OP_CONST;
3684 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3685 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
3686 yylval.ival = OP_STRINGIFY;
3690 TERM(sublex_start());
3693 s = scan_str(s,FALSE,FALSE);
3694 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3695 "### Saw backtick string before '%s'\n", s);
3697 if (PL_expect == XOPERATOR)
3698 no_op("Backticks",s);
3700 missingterm((char*)0);
3701 yylval.ival = OP_BACKTICK;
3703 TERM(sublex_start());
3707 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3708 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3710 if (PL_expect == XOPERATOR)
3711 no_op("Backslash",s);
3715 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3719 while (isDIGIT(*start) || *start == '_')
3721 if (*start == '.' && isDIGIT(start[1])) {
3722 s = scan_num(s, &yylval);
3725 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3726 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF || PL_expect == XSTATE)) {
3730 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3733 s = scan_num(s, &yylval);
3740 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3779 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3781 /* Some keywords can be followed by any delimiter, including ':' */
3782 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3783 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3784 (PL_tokenbuf[0] == 'q' &&
3785 strchr("qwxr", PL_tokenbuf[1])))));
3787 /* x::* is just a word, unless x is "CORE" */
3788 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3792 while (d < PL_bufend && isSPACE(*d))
3793 d++; /* no comments skipped here, or s### is misparsed */
3795 /* Is this a label? */
3796 if (!tmp && PL_expect == XSTATE
3797 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3799 yylval.pval = savepv(PL_tokenbuf);
3804 /* Check for keywords */
3805 tmp = keyword(PL_tokenbuf, len);
3807 /* Is this a word before a => operator? */
3808 if (*d == '=' && d[1] == '>') {
3810 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3811 yylval.opval->op_private = OPpCONST_BARE;
3812 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
3813 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3817 if (tmp < 0) { /* second-class keyword? */
3818 GV *ogv = Nullgv; /* override (winner) */
3819 GV *hgv = Nullgv; /* hidden (loser) */
3820 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3822 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3825 if (GvIMPORTED_CV(gv))
3827 else if (! CvMETHOD(cv))
3831 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3832 (gv = *gvp) != (GV*)&PL_sv_undef &&
3833 GvCVu(gv) && GvIMPORTED_CV(gv))
3839 tmp = 0; /* overridden by import or by GLOBAL */
3842 && -tmp==KEY_lock /* XXX generalizable kludge */
3844 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3846 tmp = 0; /* any sub overrides "weak" keyword */
3848 else { /* no override */
3850 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
3851 Perl_warner(aTHX_ WARN_MISC,
3852 "dump() better written as CORE::dump()");
3856 if (ckWARN(WARN_AMBIGUOUS) && hgv
3857 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3858 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3859 "Ambiguous call resolved as CORE::%s(), %s",
3860 GvENAME(hgv), "qualify as such or use &");
3867 default: /* not a keyword */
3871 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3873 /* Get the rest if it looks like a package qualifier */
3875 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3877 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3880 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3881 *s == '\'' ? "'" : "::");
3886 if (PL_expect == XOPERATOR) {
3887 if (PL_bufptr == PL_linestart) {
3888 CopLINE_dec(PL_curcop);
3889 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3890 CopLINE_inc(PL_curcop);
3893 no_op("Bareword",s);
3896 /* Look for a subroutine with this name in current package,
3897 unless name is "Foo::", in which case Foo is a bearword
3898 (and a package name). */
3901 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3903 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3904 Perl_warner(aTHX_ WARN_BAREWORD,
3905 "Bareword \"%s\" refers to nonexistent package",
3908 PL_tokenbuf[len] = '\0';
3915 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3918 /* if we saw a global override before, get the right name */
3921 sv = newSVpvn("CORE::GLOBAL::",14);
3922 sv_catpv(sv,PL_tokenbuf);
3925 sv = newSVpv(PL_tokenbuf,0);
3927 /* Presume this is going to be a bareword of some sort. */
3930 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3931 yylval.opval->op_private = OPpCONST_BARE;
3933 /* And if "Foo::", then that's what it certainly is. */
3938 /* See if it's the indirect object for a list operator. */
3940 if (PL_oldoldbufptr &&
3941 PL_oldoldbufptr < PL_bufptr &&
3942 (PL_oldoldbufptr == PL_last_lop
3943 || PL_oldoldbufptr == PL_last_uni) &&
3944 /* NO SKIPSPACE BEFORE HERE! */
3945 (PL_expect == XREF ||
3946 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3948 bool immediate_paren = *s == '(';
3950 /* (Now we can afford to cross potential line boundary.) */
3953 /* Two barewords in a row may indicate method call. */
3955 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3958 /* If not a declared subroutine, it's an indirect object. */
3959 /* (But it's an indir obj regardless for sort.) */
3961 if ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
3962 ((!gv || !GvCVu(gv)) &&
3963 (PL_last_lop_op != OP_MAPSTART &&
3964 PL_last_lop_op != OP_GREPSTART))))
3966 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3971 PL_expect = XOPERATOR;
3974 /* Is this a word before a => operator? */
3975 if (*s == '=' && s[1] == '>' && !pkgname) {
3977 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
3978 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
3979 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3983 /* If followed by a paren, it's certainly a subroutine. */
3986 if (gv && GvCVu(gv)) {
3987 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
3988 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3993 PL_nextval[PL_nexttoke].opval = yylval.opval;
3994 PL_expect = XOPERATOR;
4000 /* If followed by var or block, call it a method (unless sub) */
4002 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
4003 PL_last_lop = PL_oldbufptr;
4004 PL_last_lop_op = OP_METHOD;
4008 /* If followed by a bareword, see if it looks like indir obj. */
4010 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
4013 /* Not a method, so call it a subroutine (if defined) */
4015 if (gv && GvCVu(gv)) {
4017 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
4018 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4019 "Ambiguous use of -%s resolved as -&%s()",
4020 PL_tokenbuf, PL_tokenbuf);
4021 /* Check for a constant sub */
4023 if ((sv = cv_const_sv(cv))) {
4025 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
4026 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
4027 yylval.opval->op_private = 0;
4031 /* Resolve to GV now. */
4032 op_free(yylval.opval);
4033 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
4034 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
4035 PL_last_lop = PL_oldbufptr;
4036 PL_last_lop_op = OP_ENTERSUB;
4037 /* Is there a prototype? */
4040 char *proto = SvPV((SV*)cv, len);
4043 if (strEQ(proto, "$"))
4045 if (*proto == '&' && *s == '{') {
4046 sv_setpv(PL_subname, PL_curstash ?
4047 "__ANON__" : "__ANON__::__ANON__");
4051 PL_nextval[PL_nexttoke].opval = yylval.opval;
4057 /* Call it a bare word */
4059 if (PL_hints & HINT_STRICT_SUBS)
4060 yylval.opval->op_private |= OPpCONST_STRICT;
4063 if (ckWARN(WARN_RESERVED)) {
4064 if (lastchar != '-') {
4065 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4066 if (!*d && strNE(PL_tokenbuf,"main"))
4067 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
4074 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
4075 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4076 "Operator or semicolon missing before %c%s",
4077 lastchar, PL_tokenbuf);
4078 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4079 "Ambiguous use of %c resolved as operator %c",
4080 lastchar, lastchar);
4086 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4087 newSVpv(CopFILE(PL_curcop),0));
4091 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4092 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4095 case KEY___PACKAGE__:
4096 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4098 ? newSVsv(PL_curstname)
4107 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4108 char *pname = "main";
4109 if (PL_tokenbuf[2] == 'D')
4110 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
4111 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4114 GvIOp(gv) = newIO();
4115 IoIFP(GvIOp(gv)) = PL_rsfp;
4116 #if defined(HAS_FCNTL) && defined(F_SETFD)
4118 int fd = PerlIO_fileno(PL_rsfp);
4119 fcntl(fd,F_SETFD,fd >= 3);
4122 /* Mark this internal pseudo-handle as clean */
4123 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4125 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4126 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4127 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4129 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4130 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4131 /* if the script was opened in binmode, we need to revert
4132 * it to text mode for compatibility; but only iff it has CRs
4133 * XXX this is a questionable hack at best. */
4134 if (PL_bufend-PL_bufptr > 2
4135 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4138 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4139 loc = PerlIO_tell(PL_rsfp);
4140 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4143 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
4145 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4146 #endif /* NETWARE */
4147 #ifdef PERLIO_IS_STDIO /* really? */
4148 # if defined(__BORLANDC__)
4149 /* XXX see note in do_binmode() */
4150 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
4154 PerlIO_seek(PL_rsfp, loc, 0);
4158 #ifdef PERLIO_LAYERS
4159 if (UTF && !IN_BYTES)
4160 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
4173 if (PL_expect == XSTATE) {
4180 if (*s == ':' && s[1] == ':') {
4183 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4184 if (!(tmp = keyword(PL_tokenbuf, len)))
4185 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4199 LOP(OP_ACCEPT,XTERM);
4205 LOP(OP_ATAN2,XTERM);
4211 LOP(OP_BINMODE,XTERM);
4214 LOP(OP_BLESS,XTERM);
4223 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4240 if (!PL_cryptseen) {
4241 PL_cryptseen = TRUE;
4245 LOP(OP_CRYPT,XTERM);
4248 LOP(OP_CHMOD,XTERM);
4251 LOP(OP_CHOWN,XTERM);
4254 LOP(OP_CONNECT,XTERM);
4270 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4274 PL_hints |= HINT_BLOCK_SCOPE;
4284 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4285 LOP(OP_DBMOPEN,XTERM);
4291 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4298 yylval.ival = CopLINE(PL_curcop);
4312 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4313 UNIBRACK(OP_ENTEREVAL);
4328 case KEY_endhostent:
4334 case KEY_endservent:
4337 case KEY_endprotoent:
4348 yylval.ival = CopLINE(PL_curcop);
4350 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4352 if ((PL_bufend - p) >= 3 &&
4353 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4355 else if ((PL_bufend - p) >= 4 &&
4356 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4359 if (isIDFIRST_lazy_if(p,UTF)) {
4360 p = scan_ident(p, PL_bufend,
4361 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4365 Perl_croak(aTHX_ "Missing $ on loop variable");
4370 LOP(OP_FORMLINE,XTERM);
4376 LOP(OP_FCNTL,XTERM);
4382 LOP(OP_FLOCK,XTERM);
4391 LOP(OP_GREPSTART, XREF);
4394 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4409 case KEY_getpriority:
4410 LOP(OP_GETPRIORITY,XTERM);
4412 case KEY_getprotobyname:
4415 case KEY_getprotobynumber:
4416 LOP(OP_GPBYNUMBER,XTERM);
4418 case KEY_getprotoent:
4430 case KEY_getpeername:
4431 UNI(OP_GETPEERNAME);
4433 case KEY_gethostbyname:
4436 case KEY_gethostbyaddr:
4437 LOP(OP_GHBYADDR,XTERM);
4439 case KEY_gethostent:
4442 case KEY_getnetbyname:
4445 case KEY_getnetbyaddr:
4446 LOP(OP_GNBYADDR,XTERM);
4451 case KEY_getservbyname:
4452 LOP(OP_GSBYNAME,XTERM);
4454 case KEY_getservbyport:
4455 LOP(OP_GSBYPORT,XTERM);
4457 case KEY_getservent:
4460 case KEY_getsockname:
4461 UNI(OP_GETSOCKNAME);
4463 case KEY_getsockopt:
4464 LOP(OP_GSOCKOPT,XTERM);
4486 yylval.ival = CopLINE(PL_curcop);
4490 LOP(OP_INDEX,XTERM);
4496 LOP(OP_IOCTL,XTERM);
4508 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4540 LOP(OP_LISTEN,XTERM);
4549 s = scan_pat(s,OP_MATCH);
4550 TERM(sublex_start());
4553 LOP(OP_MAPSTART, XREF);
4556 LOP(OP_MKDIR,XTERM);
4559 LOP(OP_MSGCTL,XTERM);
4562 LOP(OP_MSGGET,XTERM);
4565 LOP(OP_MSGRCV,XTERM);
4568 LOP(OP_MSGSND,XTERM);
4574 if (isIDFIRST_lazy_if(s,UTF)) {
4575 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4576 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4578 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4579 if (!PL_in_my_stash) {
4582 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4590 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4597 if (PL_expect != XSTATE)
4598 yyerror("\"no\" not allowed in expression");
4599 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4600 s = force_version(s, FALSE);
4605 if (*s == '(' || (s = skipspace(s), *s == '('))
4612 if (isIDFIRST_lazy_if(s,UTF)) {
4614 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4616 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4617 Perl_warner(aTHX_ WARN_PRECEDENCE,
4618 "Precedence problem: open %.*s should be open(%.*s)",
4624 yylval.ival = OP_OR;
4634 LOP(OP_OPEN_DIR,XTERM);
4637 checkcomma(s,PL_tokenbuf,"filehandle");
4641 checkcomma(s,PL_tokenbuf,"filehandle");
4660 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4664 LOP(OP_PIPE_OP,XTERM);
4667 s = scan_str(s,FALSE,FALSE);
4669 missingterm((char*)0);
4670 yylval.ival = OP_CONST;
4671 TERM(sublex_start());
4677 s = scan_str(s,FALSE,FALSE);
4679 missingterm((char*)0);
4681 if (SvCUR(PL_lex_stuff)) {
4684 d = SvPV_force(PL_lex_stuff, len);
4687 for (; isSPACE(*d) && len; --len, ++d) ;
4690 if (!warned && ckWARN(WARN_QW)) {
4691 for (; !isSPACE(*d) && len; --len, ++d) {
4693 Perl_warner(aTHX_ WARN_QW,
4694 "Possible attempt to separate words with commas");
4697 else if (*d == '#') {
4698 Perl_warner(aTHX_ WARN_QW,
4699 "Possible attempt to put comments in qw() list");
4705 for (; !isSPACE(*d) && len; --len, ++d) ;
4707 sv = newSVpvn(b, d-b);
4708 if (DO_UTF8(PL_lex_stuff))
4710 words = append_elem(OP_LIST, words,
4711 newSVOP(OP_CONST, 0, tokeq(sv)));
4715 PL_nextval[PL_nexttoke].opval = words;
4720 SvREFCNT_dec(PL_lex_stuff);
4721 PL_lex_stuff = Nullsv;
4727 s = scan_str(s,FALSE,FALSE);
4729 missingterm((char*)0);
4730 yylval.ival = OP_STRINGIFY;
4731 if (SvIVX(PL_lex_stuff) == '\'')
4732 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4733 TERM(sublex_start());
4736 s = scan_pat(s,OP_QR);
4737 TERM(sublex_start());
4740 s = scan_str(s,FALSE,FALSE);
4742 missingterm((char*)0);
4743 yylval.ival = OP_BACKTICK;
4745 TERM(sublex_start());
4753 s = force_version(s, FALSE);
4755 else if (*s != 'v' || !isDIGIT(s[1])
4756 || (s = force_version(s, TRUE), *s == 'v'))
4758 *PL_tokenbuf = '\0';
4759 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4760 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4761 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4763 yyerror("<> should be quotes");
4771 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4775 LOP(OP_RENAME,XTERM);
4784 LOP(OP_RINDEX,XTERM);
4807 LOP(OP_REVERSE,XTERM);
4818 TERM(sublex_start());
4820 TOKEN(1); /* force error */
4829 LOP(OP_SELECT,XTERM);
4835 LOP(OP_SEMCTL,XTERM);
4838 LOP(OP_SEMGET,XTERM);
4841 LOP(OP_SEMOP,XTERM);
4847 LOP(OP_SETPGRP,XTERM);
4849 case KEY_setpriority:
4850 LOP(OP_SETPRIORITY,XTERM);
4852 case KEY_sethostent:
4858 case KEY_setservent:
4861 case KEY_setprotoent:
4871 LOP(OP_SEEKDIR,XTERM);
4873 case KEY_setsockopt:
4874 LOP(OP_SSOCKOPT,XTERM);
4880 LOP(OP_SHMCTL,XTERM);
4883 LOP(OP_SHMGET,XTERM);
4886 LOP(OP_SHMREAD,XTERM);
4889 LOP(OP_SHMWRITE,XTERM);
4892 LOP(OP_SHUTDOWN,XTERM);
4901 LOP(OP_SOCKET,XTERM);
4903 case KEY_socketpair:
4904 LOP(OP_SOCKPAIR,XTERM);
4907 checkcomma(s,PL_tokenbuf,"subroutine name");
4909 if (*s == ';' || *s == ')') /* probably a close */
4910 Perl_croak(aTHX_ "sort is now a reserved word");
4912 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4916 LOP(OP_SPLIT,XTERM);
4919 LOP(OP_SPRINTF,XTERM);
4922 LOP(OP_SPLICE,XTERM);
4937 LOP(OP_SUBSTR,XTERM);
4943 char tmpbuf[sizeof PL_tokenbuf];
4944 SSize_t tboffset = 0;
4945 expectation attrful;
4946 bool have_name, have_proto, bad_proto;
4951 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4952 (*s == ':' && s[1] == ':'))
4955 attrful = XATTRBLOCK;
4956 /* remember buffer pos'n for later force_word */
4957 tboffset = s - PL_oldbufptr;
4958 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4959 if (strchr(tmpbuf, ':'))
4960 sv_setpv(PL_subname, tmpbuf);
4962 sv_setsv(PL_subname,PL_curstname);
4963 sv_catpvn(PL_subname,"::",2);
4964 sv_catpvn(PL_subname,tmpbuf,len);
4971 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4972 PL_expect = XTERMBLOCK;
4973 attrful = XATTRTERM;
4974 sv_setpv(PL_subname,"?");
4978 if (key == KEY_format) {
4980 PL_lex_formbrack = PL_lex_brackets + 1;
4982 (void) force_word(PL_oldbufptr + tboffset, WORD,
4987 /* Look for a prototype */
4991 s = scan_str(s,FALSE,FALSE);
4993 Perl_croak(aTHX_ "Prototype not terminated");
4994 /* strip spaces and check for bad characters */
4995 d = SvPVX(PL_lex_stuff);
4998 for (p = d; *p; ++p) {
5001 if (!strchr("$@%*;[]&\\", *p))
5006 if (bad_proto && ckWARN(WARN_SYNTAX))
5007 Perl_warner(aTHX_ WARN_SYNTAX,
5008 "Illegal character in prototype for %s : %s",
5009 SvPVX(PL_subname), d);
5010 SvCUR(PL_lex_stuff) = tmp;
5018 if (*s == ':' && s[1] != ':')
5019 PL_expect = attrful;
5022 PL_nextval[PL_nexttoke].opval =
5023 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
5024 PL_lex_stuff = Nullsv;
5028 sv_setpv(PL_subname,
5029 PL_curstash ? "__ANON__" : "__ANON__::__ANON__");
5032 (void) force_word(PL_oldbufptr + tboffset, WORD,
5041 LOP(OP_SYSTEM,XREF);
5044 LOP(OP_SYMLINK,XTERM);
5047 LOP(OP_SYSCALL,XTERM);
5050 LOP(OP_SYSOPEN,XTERM);
5053 LOP(OP_SYSSEEK,XTERM);
5056 LOP(OP_SYSREAD,XTERM);
5059 LOP(OP_SYSWRITE,XTERM);
5063 TERM(sublex_start());
5084 LOP(OP_TRUNCATE,XTERM);
5096 yylval.ival = CopLINE(PL_curcop);
5100 yylval.ival = CopLINE(PL_curcop);
5104 LOP(OP_UNLINK,XTERM);
5110 LOP(OP_UNPACK,XTERM);
5113 LOP(OP_UTIME,XTERM);
5119 LOP(OP_UNSHIFT,XTERM);
5122 if (PL_expect != XSTATE)
5123 yyerror("\"use\" not allowed in expression");
5125 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
5126 s = force_version(s, TRUE);
5127 if (*s == ';' || (s = skipspace(s), *s == ';')) {
5128 PL_nextval[PL_nexttoke].opval = Nullop;
5131 else if (*s == 'v') {
5132 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5133 s = force_version(s, FALSE);
5137 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5138 s = force_version(s, FALSE);
5150 yylval.ival = CopLINE(PL_curcop);
5154 PL_hints |= HINT_BLOCK_SCOPE;
5161 LOP(OP_WAITPID,XTERM);
5170 ctl_l[0] = toCTRL('L');
5172 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5175 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5180 if (PL_expect == XOPERATOR)
5186 yylval.ival = OP_XOR;
5191 TERM(sublex_start());
5196 #pragma segment Main
5200 S_pending_ident(pTHX)
5204 /* pit holds the identifier we read and pending_ident is reset */
5205 char pit = PL_pending_ident;
5206 PL_pending_ident = 0;
5208 DEBUG_T({ PerlIO_printf(Perl_debug_log,
5209 "### Tokener saw identifier '%s'\n", PL_tokenbuf); });
5211 /* if we're in a my(), we can't allow dynamics here.
5212 $foo'bar has already been turned into $foo::bar, so
5213 just check for colons.
5215 if it's a legal name, the OP is a PADANY.
5218 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
5219 if (strchr(PL_tokenbuf,':'))
5220 yyerror(Perl_form(aTHX_ "No package name allowed for "
5221 "variable %s in \"our\"",
5223 tmp = pad_allocmy(PL_tokenbuf);
5226 if (strchr(PL_tokenbuf,':'))
5227 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
5229 yylval.opval = newOP(OP_PADANY, 0);
5230 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
5236 build the ops for accesses to a my() variable.
5238 Deny my($a) or my($b) in a sort block, *if* $a or $b is
5239 then used in a comparison. This catches most, but not
5240 all cases. For instance, it catches
5241 sort { my($a); $a <=> $b }
5243 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
5244 (although why you'd do that is anyone's guess).
5247 if (!strchr(PL_tokenbuf,':')) {
5248 #ifdef USE_5005THREADS
5249 /* Check for single character per-thread SVs */
5250 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
5251 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
5252 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
5254 yylval.opval = newOP(OP_THREADSV, 0);
5255 yylval.opval->op_targ = tmp;
5258 #endif /* USE_5005THREADS */
5259 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
5260 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
5261 /* might be an "our" variable" */
5262 if (SvFLAGS(namesv) & SVpad_OUR) {
5263 /* build ops for a bareword */
5264 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
5265 sv_catpvn(sym, "::", 2);
5266 sv_catpv(sym, PL_tokenbuf+1);
5267 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
5268 yylval.opval->op_private = OPpCONST_ENTERED;
5269 gv_fetchpv(SvPVX(sym),
5271 ? (GV_ADDMULTI | GV_ADDINEVAL)
5274 ((PL_tokenbuf[0] == '$') ? SVt_PV
5275 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5280 /* if it's a sort block and they're naming $a or $b */
5281 if (PL_last_lop_op == OP_SORT &&
5282 PL_tokenbuf[0] == '$' &&
5283 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
5286 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
5287 d < PL_bufend && *d != '\n';
5290 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
5291 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
5297 yylval.opval = newOP(OP_PADANY, 0);
5298 yylval.opval->op_targ = tmp;
5304 Whine if they've said @foo in a doublequoted string,
5305 and @foo isn't a variable we can find in the symbol
5308 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
5309 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
5310 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
5311 && ckWARN(WARN_AMBIGUOUS))
5313 /* Downgraded from fatal to warning 20000522 mjd */
5314 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5315 "Possible unintended interpolation of %s in string",
5320 /* build ops for a bareword */
5321 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
5322 yylval.opval->op_private = OPpCONST_ENTERED;
5323 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
5324 ((PL_tokenbuf[0] == '$') ? SVt_PV
5325 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5331 Perl_keyword(pTHX_ register char *d, I32 len)
5336 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5337 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5338 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5339 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5340 if (strEQ(d,"__END__")) return KEY___END__;
5344 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5349 if (strEQ(d,"and")) return -KEY_and;
5350 if (strEQ(d,"abs")) return -KEY_abs;
5353 if (strEQ(d,"alarm")) return -KEY_alarm;
5354 if (strEQ(d,"atan2")) return -KEY_atan2;
5357 if (strEQ(d,"accept")) return -KEY_accept;
5362 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5365 if (strEQ(d,"bless")) return -KEY_bless;
5366 if (strEQ(d,"bind")) return -KEY_bind;
5367 if (strEQ(d,"binmode")) return -KEY_binmode;
5370 if (strEQ(d,"CORE")) return -KEY_CORE;
5371 if (strEQ(d,"CHECK")) return KEY_CHECK;
5376 if (strEQ(d,"cmp")) return -KEY_cmp;
5377 if (strEQ(d,"chr")) return -KEY_chr;
5378 if (strEQ(d,"cos")) return -KEY_cos;
5381 if (strEQ(d,"chop")) return -KEY_chop;
5384 if (strEQ(d,"close")) return -KEY_close;
5385 if (strEQ(d,"chdir")) return -KEY_chdir;
5386 if (strEQ(d,"chomp")) return -KEY_chomp;
5387 if (strEQ(d,"chmod")) return -KEY_chmod;
5388 if (strEQ(d,"chown")) return -KEY_chown;
5389 if (strEQ(d,"crypt")) return -KEY_crypt;
5392 if (strEQ(d,"chroot")) return -KEY_chroot;
5393 if (strEQ(d,"caller")) return -KEY_caller;
5396 if (strEQ(d,"connect")) return -KEY_connect;
5399 if (strEQ(d,"closedir")) return -KEY_closedir;
5400 if (strEQ(d,"continue")) return -KEY_continue;
5405 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5410 if (strEQ(d,"do")) return KEY_do;
5413 if (strEQ(d,"die")) return -KEY_die;
5416 if (strEQ(d,"dump")) return -KEY_dump;
5419 if (strEQ(d,"delete")) return KEY_delete;
5422 if (strEQ(d,"defined")) return KEY_defined;
5423 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5426 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5431 if (strEQ(d,"END")) return KEY_END;
5436 if (strEQ(d,"eq")) return -KEY_eq;
5439 if (strEQ(d,"eof")) return -KEY_eof;
5440 if (strEQ(d,"exp")) return -KEY_exp;
5443 if (strEQ(d,"else")) return KEY_else;
5444 if (strEQ(d,"exit")) return -KEY_exit;
5445 if (strEQ(d,"eval")) return KEY_eval;
5446 if (strEQ(d,"exec")) return -KEY_exec;
5447 if (strEQ(d,"each")) return -KEY_each;
5450 if (strEQ(d,"elsif")) return KEY_elsif;
5453 if (strEQ(d,"exists")) return KEY_exists;
5454 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5457 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5458 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5461 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5464 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5465 if (strEQ(d,"endservent")) return -KEY_endservent;
5468 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5475 if (strEQ(d,"for")) return KEY_for;
5478 if (strEQ(d,"fork")) return -KEY_fork;
5481 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5482 if (strEQ(d,"flock")) return -KEY_flock;
5485 if (strEQ(d,"format")) return KEY_format;
5486 if (strEQ(d,"fileno")) return -KEY_fileno;
5489 if (strEQ(d,"foreach")) return KEY_foreach;
5492 if (strEQ(d,"formline")) return -KEY_formline;
5497 if (strnEQ(d,"get",3)) {
5502 if (strEQ(d,"ppid")) return -KEY_getppid;
5503 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5506 if (strEQ(d,"pwent")) return -KEY_getpwent;
5507 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5508 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5511 if (strEQ(d,"peername")) return -KEY_getpeername;
5512 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5513 if (strEQ(d,"priority")) return -KEY_getpriority;
5516 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5519 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5523 else if (*d == 'h') {
5524 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5525 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5526 if (strEQ(d,"hostent")) return -KEY_gethostent;
5528 else if (*d == 'n') {
5529 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5530 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5531 if (strEQ(d,"netent")) return -KEY_getnetent;
5533 else if (*d == 's') {
5534 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5535 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5536 if (strEQ(d,"servent")) return -KEY_getservent;
5537 if (strEQ(d,"sockname")) return -KEY_getsockname;
5538 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5540 else if (*d == 'g') {
5541 if (strEQ(d,"grent")) return -KEY_getgrent;
5542 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5543 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5545 else if (*d == 'l') {
5546 if (strEQ(d,"login")) return -KEY_getlogin;
5548 else if (strEQ(d,"c")) return -KEY_getc;
5553 if (strEQ(d,"gt")) return -KEY_gt;
5554 if (strEQ(d,"ge")) return -KEY_ge;
5557 if (strEQ(d,"grep")) return KEY_grep;
5558 if (strEQ(d,"goto")) return KEY_goto;
5559 if (strEQ(d,"glob")) return KEY_glob;
5562 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5567 if (strEQ(d,"hex")) return -KEY_hex;
5570 if (strEQ(d,"INIT")) return KEY_INIT;
5575 if (strEQ(d,"if")) return KEY_if;
5578 if (strEQ(d,"int")) return -KEY_int;
5581 if (strEQ(d,"index")) return -KEY_index;
5582 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5587 if (strEQ(d,"join")) return -KEY_join;
5591 if (strEQ(d,"keys")) return -KEY_keys;
5592 if (strEQ(d,"kill")) return -KEY_kill;
5598 if (strEQ(d,"lt")) return -KEY_lt;
5599 if (strEQ(d,"le")) return -KEY_le;
5600 if (strEQ(d,"lc")) return -KEY_lc;
5603 if (strEQ(d,"log")) return -KEY_log;
5606 if (strEQ(d,"last")) return KEY_last;
5607 if (strEQ(d,"link")) return -KEY_link;
5608 if (strEQ(d,"lock")) return -KEY_lock;
5611 if (strEQ(d,"local")) return KEY_local;
5612 if (strEQ(d,"lstat")) return -KEY_lstat;
5615 if (strEQ(d,"length")) return -KEY_length;
5616 if (strEQ(d,"listen")) return -KEY_listen;
5619 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5622 if (strEQ(d,"localtime")) return -KEY_localtime;
5628 case 1: return KEY_m;
5630 if (strEQ(d,"my")) return KEY_my;
5633 if (strEQ(d,"map")) return KEY_map;
5636 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5639 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5640 if (strEQ(d,"msgget")) return -KEY_msgget;
5641 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5642 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5647 if (strEQ(d,"next")) return KEY_next;
5648 if (strEQ(d,"ne")) return -KEY_ne;
5649 if (strEQ(d,"not")) return -KEY_not;
5650 if (strEQ(d,"no")) return KEY_no;
5655 if (strEQ(d,"or")) return -KEY_or;
5658 if (strEQ(d,"ord")) return -KEY_ord;
5659 if (strEQ(d,"oct")) return -KEY_oct;
5660 if (strEQ(d,"our")) return KEY_our;
5663 if (strEQ(d,"open")) return -KEY_open;
5666 if (strEQ(d,"opendir")) return -KEY_opendir;
5673 if (strEQ(d,"pop")) return -KEY_pop;
5674 if (strEQ(d,"pos")) return KEY_pos;
5677 if (strEQ(d,"push")) return -KEY_push;
5678 if (strEQ(d,"pack")) return -KEY_pack;
5679 if (strEQ(d,"pipe")) return -KEY_pipe;
5682 if (strEQ(d,"print")) return KEY_print;
5685 if (strEQ(d,"printf")) return KEY_printf;
5688 if (strEQ(d,"package")) return KEY_package;
5691 if (strEQ(d,"prototype")) return KEY_prototype;
5696 if (strEQ(d,"q")) return KEY_q;
5697 if (strEQ(d,"qr")) return KEY_qr;
5698 if (strEQ(d,"qq")) return KEY_qq;
5699 if (strEQ(d,"qw")) return KEY_qw;
5700 if (strEQ(d,"qx")) return KEY_qx;
5702 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5707 if (strEQ(d,"ref")) return -KEY_ref;
5710 if (strEQ(d,"read")) return -KEY_read;
5711 if (strEQ(d,"rand")) return -KEY_rand;
5712 if (strEQ(d,"recv")) return -KEY_recv;
5713 if (strEQ(d,"redo")) return KEY_redo;
5716 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5717 if (strEQ(d,"reset")) return -KEY_reset;
5720 if (strEQ(d,"return")) return KEY_return;
5721 if (strEQ(d,"rename")) return -KEY_rename;
5722 if (strEQ(d,"rindex")) return -KEY_rindex;
5725 if (strEQ(d,"require")) return KEY_require;
5726 if (strEQ(d,"reverse")) return -KEY_reverse;
5727 if (strEQ(d,"readdir")) return -KEY_readdir;
5730 if (strEQ(d,"readlink")) return -KEY_readlink;
5731 if (strEQ(d,"readline")) return -KEY_readline;
5732 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5735 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5741 case 0: return KEY_s;
5743 if (strEQ(d,"scalar")) return KEY_scalar;
5748 if (strEQ(d,"seek")) return -KEY_seek;
5749 if (strEQ(d,"send")) return -KEY_send;
5752 if (strEQ(d,"semop")) return -KEY_semop;
5755 if (strEQ(d,"select")) return -KEY_select;
5756 if (strEQ(d,"semctl")) return -KEY_semctl;
5757 if (strEQ(d,"semget")) return -KEY_semget;
5760 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5761 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5764 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5765 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5768 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5771 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5772 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5773 if (strEQ(d,"setservent")) return -KEY_setservent;
5776 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5777 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5784 if (strEQ(d,"shift")) return -KEY_shift;
5787 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5788 if (strEQ(d,"shmget")) return -KEY_shmget;
5791 if (strEQ(d,"shmread")) return -KEY_shmread;
5794 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5795 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5800 if (strEQ(d,"sin")) return -KEY_sin;
5803 if (strEQ(d,"sleep")) return -KEY_sleep;
5806 if (strEQ(d,"sort")) return KEY_sort;
5807 if (strEQ(d,"socket")) return -KEY_socket;
5808 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5811 if (strEQ(d,"split")) return KEY_split;
5812 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5813 if (strEQ(d,"splice")) return -KEY_splice;
5816 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5819 if (strEQ(d,"srand")) return -KEY_srand;
5822 if (strEQ(d,"stat")) return -KEY_stat;
5823 if (strEQ(d,"study")) return KEY_study;
5826 if (strEQ(d,"substr")) return -KEY_substr;
5827 if (strEQ(d,"sub")) return KEY_sub;
5832 if (strEQ(d,"system")) return -KEY_system;
5835 if (strEQ(d,"symlink")) return -KEY_symlink;
5836 if (strEQ(d,"syscall")) return -KEY_syscall;
5837 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5838 if (strEQ(d,"sysread")) return -KEY_sysread;
5839 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5842 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5851 if (strEQ(d,"tr")) return KEY_tr;
5854 if (strEQ(d,"tie")) return KEY_tie;
5857 if (strEQ(d,"tell")) return -KEY_tell;
5858 if (strEQ(d,"tied")) return KEY_tied;
5859 if (strEQ(d,"time")) return -KEY_time;
5862 if (strEQ(d,"times")) return -KEY_times;
5865 if (strEQ(d,"telldir")) return -KEY_telldir;
5868 if (strEQ(d,"truncate")) return -KEY_truncate;
5875 if (strEQ(d,"uc")) return -KEY_uc;
5878 if (strEQ(d,"use")) return KEY_use;
5881 if (strEQ(d,"undef")) return KEY_undef;
5882 if (strEQ(d,"until")) return KEY_until;
5883 if (strEQ(d,"untie")) return KEY_untie;
5884 if (strEQ(d,"utime")) return -KEY_utime;
5885 if (strEQ(d,"umask")) return -KEY_umask;
5888 if (strEQ(d,"unless")) return KEY_unless;
5889 if (strEQ(d,"unpack")) return -KEY_unpack;
5890 if (strEQ(d,"unlink")) return -KEY_unlink;
5893 if (strEQ(d,"unshift")) return -KEY_unshift;
5894 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5899 if (strEQ(d,"values")) return -KEY_values;
5900 if (strEQ(d,"vec")) return -KEY_vec;
5905 if (strEQ(d,"warn")) return -KEY_warn;
5906 if (strEQ(d,"wait")) return -KEY_wait;
5909 if (strEQ(d,"while")) return KEY_while;
5910 if (strEQ(d,"write")) return -KEY_write;
5913 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5916 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5921 if (len == 1) return -KEY_x;
5922 if (strEQ(d,"xor")) return -KEY_xor;
5925 if (len == 1) return KEY_y;
5934 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5938 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5939 if (ckWARN(WARN_SYNTAX)) {
5941 for (w = s+2; *w && level; w++) {
5948 for (; *w && isSPACE(*w); w++) ;
5949 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5950 Perl_warner(aTHX_ WARN_SYNTAX,
5951 "%s (...) interpreted as function",name);
5954 while (s < PL_bufend && isSPACE(*s))
5958 while (s < PL_bufend && isSPACE(*s))
5960 if (isIDFIRST_lazy_if(s,UTF)) {
5962 while (isALNUM_lazy_if(s,UTF))
5964 while (s < PL_bufend && isSPACE(*s))
5969 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5973 Perl_croak(aTHX_ "No comma allowed after %s", what);
5978 /* Either returns sv, or mortalizes sv and returns a new SV*.
5979 Best used as sv=new_constant(..., sv, ...).
5980 If s, pv are NULL, calls subroutine with one argument,
5981 and type is used with error messages only. */
5984 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5988 HV *table = GvHV(PL_hintgv); /* ^H */
5992 const char *why1, *why2, *why3;
5994 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
5997 why2 = strEQ(key,"charnames")
5998 ? "(possibly a missing \"use charnames ...\")"
6000 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
6001 (type ? type: "undef"), why2);
6003 /* This is convoluted and evil ("goto considered harmful")
6004 * but I do not understand the intricacies of all the different
6005 * failure modes of %^H in here. The goal here is to make
6006 * the most probable error message user-friendly. --jhi */
6011 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
6012 (type ? type: "undef"), why1, why2, why3);
6014 yyerror(SvPVX(msg));
6018 cvp = hv_fetch(table, key, strlen(key), FALSE);
6019 if (!cvp || !SvOK(*cvp)) {
6022 why3 = "} is not defined";
6025 sv_2mortal(sv); /* Parent created it permanently */
6028 pv = sv_2mortal(newSVpvn(s, len));
6030 typesv = sv_2mortal(newSVpv(type, 0));
6032 typesv = &PL_sv_undef;
6034 PUSHSTACKi(PERLSI_OVERLOAD);
6046 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
6050 /* Check the eval first */
6051 if (!PL_in_eval && SvTRUE(ERRSV)) {
6053 sv_catpv(ERRSV, "Propagated");
6054 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
6056 res = SvREFCNT_inc(sv);
6060 (void)SvREFCNT_inc(res);
6069 why1 = "Call to &{$^H{";
6071 why3 = "}} did not return a defined value";
6080 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
6082 register char *d = dest;
6083 register char *e = d + destlen - 3; /* two-character token, ending NUL */
6086 Perl_croak(aTHX_ ident_too_long);
6087 if (isALNUM(*s)) /* UTF handled below */
6089 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
6094 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
6098 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6099 char *t = s + UTF8SKIP(s);
6100 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6102 if (d + (t - s) > e)
6103 Perl_croak(aTHX_ ident_too_long);
6104 Copy(s, d, t - s, char);
6117 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
6127 e = d + destlen - 3; /* two-character token, ending NUL */
6129 while (isDIGIT(*s)) {
6131 Perl_croak(aTHX_ ident_too_long);
6138 Perl_croak(aTHX_ ident_too_long);
6139 if (isALNUM(*s)) /* UTF handled below */
6141 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
6146 else if (*s == ':' && s[1] == ':') {
6150 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6151 char *t = s + UTF8SKIP(s);
6152 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6154 if (d + (t - s) > e)
6155 Perl_croak(aTHX_ ident_too_long);
6156 Copy(s, d, t - s, char);
6167 if (PL_lex_state != LEX_NORMAL)
6168 PL_lex_state = LEX_INTERPENDMAYBE;
6171 if (*s == '$' && s[1] &&
6172 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
6185 if (*d == '^' && *s && isCONTROLVAR(*s)) {
6190 if (isSPACE(s[-1])) {
6193 if (!SPACE_OR_TAB(ch)) {
6199 if (isIDFIRST_lazy_if(d,UTF)) {
6203 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
6205 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
6208 Copy(s, d, e - s, char);
6213 while ((isALNUM(*s) || *s == ':') && d < e)
6216 Perl_croak(aTHX_ ident_too_long);
6219 while (s < send && SPACE_OR_TAB(*s)) s++;
6220 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
6221 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
6222 const char *brack = *s == '[' ? "[...]" : "{...}";
6223 Perl_warner(aTHX_ WARN_AMBIGUOUS,
6224 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
6225 funny, dest, brack, funny, dest, brack);
6228 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
6232 /* Handle extended ${^Foo} variables
6233 * 1999-02-27 mjd-perl-patch@plover.com */
6234 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
6238 while (isALNUM(*s) && d < e) {
6242 Perl_croak(aTHX_ ident_too_long);
6247 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
6248 PL_lex_state = LEX_INTERPEND;
6251 if (PL_lex_state == LEX_NORMAL) {
6252 if (ckWARN(WARN_AMBIGUOUS) &&
6253 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
6255 Perl_warner(aTHX_ WARN_AMBIGUOUS,
6256 "Ambiguous use of %c{%s} resolved to %c%s",
6257 funny, dest, funny, dest);
6262 s = bracket; /* let the parser handle it */
6266 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
6267 PL_lex_state = LEX_INTERPEND;
6272 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
6277 *pmfl |= PMf_GLOBAL;
6279 *pmfl |= PMf_CONTINUE;
6283 *pmfl |= PMf_MULTILINE;
6285 *pmfl |= PMf_SINGLELINE;
6287 *pmfl |= PMf_EXTENDED;
6291 S_scan_pat(pTHX_ char *start, I32 type)
6296 s = scan_str(start,FALSE,FALSE);
6298 Perl_croak(aTHX_ "Search pattern not terminated");
6300 pm = (PMOP*)newPMOP(type, 0);
6301 if (PL_multi_open == '?')
6302 pm->op_pmflags |= PMf_ONCE;
6304 while (*s && strchr("iomsx", *s))
6305 pmflag(&pm->op_pmflags,*s++);
6308 while (*s && strchr("iogcmsx", *s))
6309 pmflag(&pm->op_pmflags,*s++);
6311 pm->op_pmpermflags = pm->op_pmflags;
6313 PL_lex_op = (OP*)pm;
6314 yylval.ival = OP_MATCH;
6319 S_scan_subst(pTHX_ char *start)
6326 yylval.ival = OP_NULL;
6328 s = scan_str(start,FALSE,FALSE);
6331 Perl_croak(aTHX_ "Substitution pattern not terminated");
6333 if (s[-1] == PL_multi_open)
6336 first_start = PL_multi_start;
6337 s = scan_str(s,FALSE,FALSE);
6340 SvREFCNT_dec(PL_lex_stuff);
6341 PL_lex_stuff = Nullsv;
6343 Perl_croak(aTHX_ "Substitution replacement not terminated");
6345 PL_multi_start = first_start; /* so whole substitution is taken together */
6347 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6353 else if (strchr("iogcmsx", *s))
6354 pmflag(&pm->op_pmflags,*s++);
6361 PL_sublex_info.super_bufptr = s;
6362 PL_sublex_info.super_bufend = PL_bufend;
6364 pm->op_pmflags |= PMf_EVAL;
6365 repl = newSVpvn("",0);
6367 sv_catpv(repl, es ? "eval " : "do ");
6368 sv_catpvn(repl, "{ ", 2);
6369 sv_catsv(repl, PL_lex_repl);
6370 sv_catpvn(repl, " };", 2);
6372 SvREFCNT_dec(PL_lex_repl);
6376 pm->op_pmpermflags = pm->op_pmflags;
6377 PL_lex_op = (OP*)pm;
6378 yylval.ival = OP_SUBST;
6383 S_scan_trans(pTHX_ char *start)
6392 yylval.ival = OP_NULL;
6394 s = scan_str(start,FALSE,FALSE);
6396 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6397 if (s[-1] == PL_multi_open)
6400 s = scan_str(s,FALSE,FALSE);
6403 SvREFCNT_dec(PL_lex_stuff);
6404 PL_lex_stuff = Nullsv;
6406 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6409 complement = del = squash = 0;
6410 while (strchr("cds", *s)) {
6412 complement = OPpTRANS_COMPLEMENT;
6414 del = OPpTRANS_DELETE;
6416 squash = OPpTRANS_SQUASH;
6420 New(803, tbl, complement&&!del?258:256, short);
6421 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6422 o->op_private = del|squash|complement|
6423 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
6424 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
6427 yylval.ival = OP_TRANS;
6432 S_scan_heredoc(pTHX_ register char *s)
6435 I32 op_type = OP_SCALAR;
6442 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6446 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6449 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6450 if (*peek && strchr("`'\"",*peek)) {
6453 s = delimcpy(d, e, s, PL_bufend, term, &len);
6463 if (!isALNUM_lazy_if(s,UTF))
6464 deprecate_old("bare << to mean <<\"\"");
6465 for (; isALNUM_lazy_if(s,UTF); s++) {
6470 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6471 Perl_croak(aTHX_ "Delimiter for here document is too long");
6474 len = d - PL_tokenbuf;
6475 #ifndef PERL_STRICT_CR
6476 d = strchr(s, '\r');
6480 while (s < PL_bufend) {
6486 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6495 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6500 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6501 herewas = newSVpvn(s,PL_bufend-s);
6503 s--, herewas = newSVpvn(s,d-s);
6504 s += SvCUR(herewas);
6506 tmpstr = NEWSV(87,79);
6507 sv_upgrade(tmpstr, SVt_PVIV);
6512 else if (term == '`') {
6513 op_type = OP_BACKTICK;
6514 SvIVX(tmpstr) = '\\';
6518 PL_multi_start = CopLINE(PL_curcop);
6519 PL_multi_open = PL_multi_close = '<';
6520 term = *PL_tokenbuf;
6521 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6522 char *bufptr = PL_sublex_info.super_bufptr;
6523 char *bufend = PL_sublex_info.super_bufend;
6524 char *olds = s - SvCUR(herewas);
6525 s = strchr(bufptr, '\n');
6529 while (s < bufend &&
6530 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6532 CopLINE_inc(PL_curcop);
6535 CopLINE_set(PL_curcop, PL_multi_start);
6536 missingterm(PL_tokenbuf);
6538 sv_setpvn(herewas,bufptr,d-bufptr+1);
6539 sv_setpvn(tmpstr,d+1,s-d);
6541 sv_catpvn(herewas,s,bufend-s);
6542 (void)strcpy(bufptr,SvPVX(herewas));
6549 while (s < PL_bufend &&
6550 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6552 CopLINE_inc(PL_curcop);
6554 if (s >= PL_bufend) {
6555 CopLINE_set(PL_curcop, PL_multi_start);
6556 missingterm(PL_tokenbuf);
6558 sv_setpvn(tmpstr,d+1,s-d);
6560 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6562 sv_catpvn(herewas,s,PL_bufend-s);
6563 sv_setsv(PL_linestr,herewas);
6564 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6565 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6566 PL_last_lop = PL_last_uni = Nullch;
6569 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6570 while (s >= PL_bufend) { /* multiple line string? */
6572 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6573 CopLINE_set(PL_curcop, PL_multi_start);
6574 missingterm(PL_tokenbuf);
6576 CopLINE_inc(PL_curcop);
6577 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6578 PL_last_lop = PL_last_uni = Nullch;
6579 #ifndef PERL_STRICT_CR
6580 if (PL_bufend - PL_linestart >= 2) {
6581 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6582 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6584 PL_bufend[-2] = '\n';
6586 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6588 else if (PL_bufend[-1] == '\r')
6589 PL_bufend[-1] = '\n';
6591 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6592 PL_bufend[-1] = '\n';
6594 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6595 SV *sv = NEWSV(88,0);
6597 sv_upgrade(sv, SVt_PVMG);
6598 sv_setsv(sv,PL_linestr);
6601 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6603 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6606 sv_catsv(PL_linestr,herewas);
6607 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6611 sv_catsv(tmpstr,PL_linestr);
6616 PL_multi_end = CopLINE(PL_curcop);
6617 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6618 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6619 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6621 SvREFCNT_dec(herewas);
6622 if (UTF && !IN_BYTES && is_utf8_string((U8*)SvPVX(tmpstr), SvCUR(tmpstr)))
6624 PL_lex_stuff = tmpstr;
6625 yylval.ival = op_type;
6630 takes: current position in input buffer
6631 returns: new position in input buffer
6632 side-effects: yylval and lex_op are set.
6637 <FH> read from filehandle
6638 <pkg::FH> read from package qualified filehandle
6639 <pkg'FH> read from package qualified filehandle
6640 <$fh> read from filehandle in $fh
6646 S_scan_inputsymbol(pTHX_ char *start)
6648 register char *s = start; /* current position in buffer */
6654 d = PL_tokenbuf; /* start of temp holding space */
6655 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6656 end = strchr(s, '\n');
6659 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6661 /* die if we didn't have space for the contents of the <>,
6662 or if it didn't end, or if we see a newline
6665 if (len >= sizeof PL_tokenbuf)
6666 Perl_croak(aTHX_ "Excessively long <> operator");
6668 Perl_croak(aTHX_ "Unterminated <> operator");
6673 Remember, only scalar variables are interpreted as filehandles by
6674 this code. Anything more complex (e.g., <$fh{$num}>) will be
6675 treated as a glob() call.
6676 This code makes use of the fact that except for the $ at the front,
6677 a scalar variable and a filehandle look the same.
6679 if (*d == '$' && d[1]) d++;
6681 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6682 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6685 /* If we've tried to read what we allow filehandles to look like, and
6686 there's still text left, then it must be a glob() and not a getline.
6687 Use scan_str to pull out the stuff between the <> and treat it
6688 as nothing more than a string.
6691 if (d - PL_tokenbuf != len) {
6692 yylval.ival = OP_GLOB;
6694 s = scan_str(start,FALSE,FALSE);
6696 Perl_croak(aTHX_ "Glob not terminated");
6700 bool readline_overriden = FALSE;
6701 GV *gv_readline = Nullgv;
6703 /* we're in a filehandle read situation */
6706 /* turn <> into <ARGV> */
6708 (void)strcpy(d,"ARGV");
6710 /* Check whether readline() is overriden */
6711 if (((gv_readline = gv_fetchpv("readline", FALSE, SVt_PVCV))
6712 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
6714 ((gvp = (GV**)hv_fetch(PL_globalstash, "readline", 8, FALSE))
6715 && (gv_readline = *gvp) != (GV*)&PL_sv_undef
6716 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
6717 readline_overriden = TRUE;
6719 /* if <$fh>, create the ops to turn the variable into a
6725 /* try to find it in the pad for this block, otherwise find
6726 add symbol table ops
6728 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6729 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
6730 if (SvFLAGS(namesv) & SVpad_OUR) {
6731 SV *sym = sv_2mortal(newSVpv(HvNAME(GvSTASH(namesv)),0));
6732 sv_catpvn(sym, "::", 2);
6738 OP *o = newOP(OP_PADSV, 0);
6740 PL_lex_op = readline_overriden
6741 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6742 append_elem(OP_LIST, o,
6743 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
6744 : (OP*)newUNOP(OP_READLINE, 0, o);
6753 ? (GV_ADDMULTI | GV_ADDINEVAL)
6756 PL_lex_op = readline_overriden
6757 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6758 append_elem(OP_LIST,
6759 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
6760 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
6761 : (OP*)newUNOP(OP_READLINE, 0,
6762 newUNOP(OP_RV2SV, 0,
6763 newGVOP(OP_GV, 0, gv)));
6765 if (!readline_overriden)
6766 PL_lex_op->op_flags |= OPf_SPECIAL;
6767 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6768 yylval.ival = OP_NULL;
6771 /* If it's none of the above, it must be a literal filehandle
6772 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6774 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6775 PL_lex_op = readline_overriden
6776 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6777 append_elem(OP_LIST,
6778 newGVOP(OP_GV, 0, gv),
6779 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
6780 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6781 yylval.ival = OP_NULL;
6790 takes: start position in buffer
6791 keep_quoted preserve \ on the embedded delimiter(s)
6792 keep_delims preserve the delimiters around the string
6793 returns: position to continue reading from buffer
6794 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6795 updates the read buffer.
6797 This subroutine pulls a string out of the input. It is called for:
6798 q single quotes q(literal text)
6799 ' single quotes 'literal text'
6800 qq double quotes qq(interpolate $here please)
6801 " double quotes "interpolate $here please"
6802 qx backticks qx(/bin/ls -l)
6803 ` backticks `/bin/ls -l`
6804 qw quote words @EXPORT_OK = qw( func() $spam )
6805 m// regexp match m/this/
6806 s/// regexp substitute s/this/that/
6807 tr/// string transliterate tr/this/that/
6808 y/// string transliterate y/this/that/
6809 ($*@) sub prototypes sub foo ($)
6810 (stuff) sub attr parameters sub foo : attr(stuff)
6811 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6813 In most of these cases (all but <>, patterns and transliterate)
6814 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6815 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6816 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6819 It skips whitespace before the string starts, and treats the first
6820 character as the delimiter. If the delimiter is one of ([{< then
6821 the corresponding "close" character )]}> is used as the closing
6822 delimiter. It allows quoting of delimiters, and if the string has
6823 balanced delimiters ([{<>}]) it allows nesting.
6825 On success, the SV with the resulting string is put into lex_stuff or,
6826 if that is already non-NULL, into lex_repl. The second case occurs only
6827 when parsing the RHS of the special constructs s/// and tr/// (y///).
6828 For convenience, the terminating delimiter character is stuffed into
6833 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6835 SV *sv; /* scalar value: string */
6836 char *tmps; /* temp string, used for delimiter matching */
6837 register char *s = start; /* current position in the buffer */
6838 register char term; /* terminating character */
6839 register char *to; /* current position in the sv's data */
6840 I32 brackets = 1; /* bracket nesting level */
6841 bool has_utf8 = FALSE; /* is there any utf8 content? */
6843 /* skip space before the delimiter */
6847 /* mark where we are, in case we need to report errors */
6850 /* after skipping whitespace, the next character is the terminator */
6852 if (!UTF8_IS_INVARIANT((U8)term) && UTF)
6855 /* mark where we are */
6856 PL_multi_start = CopLINE(PL_curcop);
6857 PL_multi_open = term;
6859 /* find corresponding closing delimiter */
6860 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6862 PL_multi_close = term;
6864 /* create a new SV to hold the contents. 87 is leak category, I'm
6865 assuming. 79 is the SV's initial length. What a random number. */
6867 sv_upgrade(sv, SVt_PVIV);
6869 (void)SvPOK_only(sv); /* validate pointer */
6871 /* move past delimiter and try to read a complete string */
6873 sv_catpvn(sv, s, 1);
6876 /* extend sv if need be */
6877 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6878 /* set 'to' to the next character in the sv's string */
6879 to = SvPVX(sv)+SvCUR(sv);
6881 /* if open delimiter is the close delimiter read unbridle */
6882 if (PL_multi_open == PL_multi_close) {
6883 for (; s < PL_bufend; s++,to++) {
6884 /* embedded newlines increment the current line number */
6885 if (*s == '\n' && !PL_rsfp)
6886 CopLINE_inc(PL_curcop);
6887 /* handle quoted delimiters */
6888 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6889 if (!keep_quoted && s[1] == term)
6891 /* any other quotes are simply copied straight through */
6895 /* terminate when run out of buffer (the for() condition), or
6896 have found the terminator */
6897 else if (*s == term)
6899 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
6905 /* if the terminator isn't the same as the start character (e.g.,
6906 matched brackets), we have to allow more in the quoting, and
6907 be prepared for nested brackets.
6910 /* read until we run out of string, or we find the terminator */
6911 for (; s < PL_bufend; s++,to++) {
6912 /* embedded newlines increment the line count */
6913 if (*s == '\n' && !PL_rsfp)
6914 CopLINE_inc(PL_curcop);
6915 /* backslashes can escape the open or closing characters */
6916 if (*s == '\\' && s+1 < PL_bufend) {
6918 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6923 /* allow nested opens and closes */
6924 else if (*s == PL_multi_close && --brackets <= 0)
6926 else if (*s == PL_multi_open)
6928 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
6933 /* terminate the copied string and update the sv's end-of-string */
6935 SvCUR_set(sv, to - SvPVX(sv));
6938 * this next chunk reads more into the buffer if we're not done yet
6942 break; /* handle case where we are done yet :-) */
6944 #ifndef PERL_STRICT_CR
6945 if (to - SvPVX(sv) >= 2) {
6946 if ((to[-2] == '\r' && to[-1] == '\n') ||
6947 (to[-2] == '\n' && to[-1] == '\r'))
6951 SvCUR_set(sv, to - SvPVX(sv));
6953 else if (to[-1] == '\r')
6956 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6960 /* if we're out of file, or a read fails, bail and reset the current
6961 line marker so we can report where the unterminated string began
6964 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6966 CopLINE_set(PL_curcop, PL_multi_start);
6969 /* we read a line, so increment our line counter */
6970 CopLINE_inc(PL_curcop);
6972 /* update debugger info */
6973 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6974 SV *sv = NEWSV(88,0);
6976 sv_upgrade(sv, SVt_PVMG);
6977 sv_setsv(sv,PL_linestr);
6980 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6983 /* having changed the buffer, we must update PL_bufend */
6984 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6985 PL_last_lop = PL_last_uni = Nullch;
6988 /* at this point, we have successfully read the delimited string */
6991 sv_catpvn(sv, s, 1);
6994 PL_multi_end = CopLINE(PL_curcop);
6997 /* if we allocated too much space, give some back */
6998 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6999 SvLEN_set(sv, SvCUR(sv) + 1);
7000 Renew(SvPVX(sv), SvLEN(sv), char);
7003 /* decide whether this is the first or second quoted string we've read
7016 takes: pointer to position in buffer
7017 returns: pointer to new position in buffer
7018 side-effects: builds ops for the constant in yylval.op
7020 Read a number in any of the formats that Perl accepts:
7022 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
7023 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
7026 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
7028 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
7031 If it reads a number without a decimal point or an exponent, it will
7032 try converting the number to an integer and see if it can do so
7033 without loss of precision.
7037 Perl_scan_num(pTHX_ char *start, YYSTYPE* lvalp)
7039 register char *s = start; /* current position in buffer */
7040 register char *d; /* destination in temp buffer */
7041 register char *e; /* end of temp buffer */
7042 NV nv; /* number read, as a double */
7043 SV *sv = Nullsv; /* place to put the converted number */
7044 bool floatit; /* boolean: int or float? */
7045 char *lastub = 0; /* position of last underbar */
7046 static char number_too_long[] = "Number too long";
7048 /* We use the first character to decide what type of number this is */
7052 Perl_croak(aTHX_ "panic: scan_num");
7054 /* if it starts with a 0, it could be an octal number, a decimal in
7055 0.13 disguise, or a hexadecimal number, or a binary number. */
7059 u holds the "number so far"
7060 shift the power of 2 of the base
7061 (hex == 4, octal == 3, binary == 1)
7062 overflowed was the number more than we can hold?
7064 Shift is used when we add a digit. It also serves as an "are
7065 we in octal/hex/binary?" indicator to disallow hex characters
7071 bool overflowed = FALSE;
7072 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
7073 static char* bases[5] = { "", "binary", "", "octal",
7075 static char* Bases[5] = { "", "Binary", "", "Octal",
7077 static char *maxima[5] = { "",
7078 "0b11111111111111111111111111111111",
7082 char *base, *Base, *max;
7088 } else if (s[1] == 'b') {
7092 /* check for a decimal in disguise */
7093 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
7095 /* so it must be octal */
7102 if (ckWARN(WARN_SYNTAX))
7103 Perl_warner(aTHX_ WARN_SYNTAX,
7104 "Misplaced _ in number");
7108 base = bases[shift];
7109 Base = Bases[shift];
7110 max = maxima[shift];
7112 /* read the rest of the number */
7114 /* x is used in the overflow test,
7115 b is the digit we're adding on. */
7120 /* if we don't mention it, we're done */
7124 /* _ are ignored -- but warned about if consecutive */
7126 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7127 Perl_warner(aTHX_ WARN_SYNTAX,
7128 "Misplaced _ in number");
7132 /* 8 and 9 are not octal */
7135 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
7139 case '2': case '3': case '4':
7140 case '5': case '6': case '7':
7142 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
7146 b = *s++ & 15; /* ASCII digit -> value of digit */
7150 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
7151 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
7152 /* make sure they said 0x */
7157 /* Prepare to put the digit we have onto the end
7158 of the number so far. We check for overflows.
7163 x = u << shift; /* make room for the digit */
7165 if ((x >> shift) != u
7166 && !(PL_hints & HINT_NEW_BINARY)) {
7169 if (ckWARN_d(WARN_OVERFLOW))
7170 Perl_warner(aTHX_ WARN_OVERFLOW,
7171 "Integer overflow in %s number",
7174 u = x | b; /* add the digit to the end */
7177 n *= nvshift[shift];
7178 /* If an NV has not enough bits in its
7179 * mantissa to represent an UV this summing of
7180 * small low-order numbers is a waste of time
7181 * (because the NV cannot preserve the
7182 * low-order bits anyway): we could just
7183 * remember when did we overflow and in the
7184 * end just multiply n by the right
7192 /* if we get here, we had success: make a scalar value from
7197 /* final misplaced underbar check */
7199 if (ckWARN(WARN_SYNTAX))
7200 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
7205 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
7206 Perl_warner(aTHX_ WARN_PORTABLE,
7207 "%s number > %s non-portable",
7213 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
7214 Perl_warner(aTHX_ WARN_PORTABLE,
7215 "%s number > %s non-portable",
7220 if (PL_hints & HINT_NEW_BINARY)
7221 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
7226 handle decimal numbers.
7227 we're also sent here when we read a 0 as the first digit
7229 case '1': case '2': case '3': case '4': case '5':
7230 case '6': case '7': case '8': case '9': case '.':
7233 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
7236 /* read next group of digits and _ and copy into d */
7237 while (isDIGIT(*s) || *s == '_') {
7238 /* skip underscores, checking for misplaced ones
7242 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7243 Perl_warner(aTHX_ WARN_SYNTAX,
7244 "Misplaced _ in number");
7248 /* check for end of fixed-length buffer */
7250 Perl_croak(aTHX_ number_too_long);
7251 /* if we're ok, copy the character */
7256 /* final misplaced underbar check */
7257 if (lastub && s == lastub + 1) {
7258 if (ckWARN(WARN_SYNTAX))
7259 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
7262 /* read a decimal portion if there is one. avoid
7263 3..5 being interpreted as the number 3. followed
7266 if (*s == '.' && s[1] != '.') {
7271 if (ckWARN(WARN_SYNTAX))
7272 Perl_warner(aTHX_ WARN_SYNTAX,
7273 "Misplaced _ in number");
7277 /* copy, ignoring underbars, until we run out of digits.
7279 for (; isDIGIT(*s) || *s == '_'; s++) {
7280 /* fixed length buffer check */
7282 Perl_croak(aTHX_ number_too_long);
7284 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7285 Perl_warner(aTHX_ WARN_SYNTAX,
7286 "Misplaced _ in number");
7292 /* fractional part ending in underbar? */
7294 if (ckWARN(WARN_SYNTAX))
7295 Perl_warner(aTHX_ WARN_SYNTAX,
7296 "Misplaced _ in number");
7298 if (*s == '.' && isDIGIT(s[1])) {
7299 /* oops, it's really a v-string, but without the "v" */
7305 /* read exponent part, if present */
7306 if (*s && strchr("eE",*s) && strchr("+-0123456789_", s[1])) {
7310 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
7311 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
7313 /* stray preinitial _ */
7315 if (ckWARN(WARN_SYNTAX))
7316 Perl_warner(aTHX_ WARN_SYNTAX,
7317 "Misplaced _ in number");
7321 /* allow positive or negative exponent */
7322 if (*s == '+' || *s == '-')
7325 /* stray initial _ */
7327 if (ckWARN(WARN_SYNTAX))
7328 Perl_warner(aTHX_ WARN_SYNTAX,
7329 "Misplaced _ in number");
7333 /* read digits of exponent */
7334 while (isDIGIT(*s) || *s == '_') {
7337 Perl_croak(aTHX_ number_too_long);
7341 if (ckWARN(WARN_SYNTAX) &&
7342 ((lastub && s == lastub + 1) ||
7343 (!isDIGIT(s[1]) && s[1] != '_')))
7344 Perl_warner(aTHX_ WARN_SYNTAX,
7345 "Misplaced _ in number");
7352 /* make an sv from the string */
7356 We try to do an integer conversion first if no characters
7357 indicating "float" have been found.
7362 int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
7364 if (flags == IS_NUMBER_IN_UV) {
7366 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7369 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
7370 if (uv <= (UV) IV_MIN)
7371 sv_setiv(sv, -(IV)uv);
7378 /* terminate the string */
7380 nv = Atof(PL_tokenbuf);
7384 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7385 (PL_hints & HINT_NEW_INTEGER) )
7386 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7387 (floatit ? "float" : "integer"),
7391 /* if it starts with a v, it could be a v-string */
7394 sv = NEWSV(92,5); /* preallocate storage space */
7395 s = new_vstring(s,sv);
7399 /* make the op for the constant and return */
7402 lvalp->opval = newSVOP(OP_CONST, 0, sv);
7404 lvalp->opval = Nullop;
7410 S_scan_formline(pTHX_ register char *s)
7414 SV *stuff = newSVpvn("",0);
7415 bool needargs = FALSE;
7418 if (*s == '.' || *s == /*{*/'}') {
7420 #ifdef PERL_STRICT_CR
7421 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7423 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7425 if (*t == '\n' || t == PL_bufend)
7428 if (PL_in_eval && !PL_rsfp) {
7429 eol = strchr(s,'\n');
7434 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7436 for (t = s; t < eol; t++) {
7437 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7439 goto enough; /* ~~ must be first line in formline */
7441 if (*t == '@' || *t == '^')
7445 sv_catpvn(stuff, s, eol-s);
7446 #ifndef PERL_STRICT_CR
7447 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7448 char *end = SvPVX(stuff) + SvCUR(stuff);
7460 s = filter_gets(PL_linestr, PL_rsfp, 0);
7461 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7462 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7463 PL_last_lop = PL_last_uni = Nullch;
7466 yyerror("Format not terminated");
7476 PL_lex_state = LEX_NORMAL;
7477 PL_nextval[PL_nexttoke].ival = 0;
7481 PL_lex_state = LEX_FORMLINE;
7482 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7484 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7488 SvREFCNT_dec(stuff);
7489 PL_lex_formbrack = 0;
7500 PL_cshlen = strlen(PL_cshname);
7505 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7507 I32 oldsavestack_ix = PL_savestack_ix;
7508 CV* outsidecv = PL_compcv;
7512 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7514 SAVEI32(PL_subline);
7515 save_item(PL_subname);
7518 SAVESPTR(PL_comppad_name);
7519 SAVESPTR(PL_compcv);
7520 SAVEI32(PL_comppad_name_fill);
7521 SAVEI32(PL_min_intro_pending);
7522 SAVEI32(PL_max_intro_pending);
7523 SAVEI32(PL_pad_reset_pending);
7525 PL_compcv = (CV*)NEWSV(1104,0);
7526 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7527 CvFLAGS(PL_compcv) |= flags;
7529 PL_comppad = newAV();
7530 av_push(PL_comppad, Nullsv);
7531 PL_curpad = AvARRAY(PL_comppad);
7532 PL_comppad_name = newAV();
7533 PL_comppad_name_fill = 0;
7534 PL_min_intro_pending = 0;
7536 PL_subline = CopLINE(PL_curcop);
7537 #ifdef USE_5005THREADS
7538 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7539 PL_curpad[0] = (SV*)newAV();
7540 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7541 #endif /* USE_5005THREADS */
7543 comppadlist = newAV();
7544 AvREAL_off(comppadlist);
7545 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7546 av_store(comppadlist, 1, (SV*)PL_comppad);
7548 CvPADLIST(PL_compcv) = comppadlist;
7549 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7550 #ifdef USE_5005THREADS
7551 CvOWNER(PL_compcv) = 0;
7552 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7553 MUTEX_INIT(CvMUTEXP(PL_compcv));
7554 #endif /* USE_5005THREADS */
7556 return oldsavestack_ix;
7560 #pragma segment Perl_yylex
7563 Perl_yywarn(pTHX_ char *s)
7565 PL_in_eval |= EVAL_WARNONLY;
7567 PL_in_eval &= ~EVAL_WARNONLY;
7572 Perl_yyerror(pTHX_ char *s)
7575 char *context = NULL;
7579 if (!yychar || (yychar == ';' && !PL_rsfp))
7581 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7582 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7583 while (isSPACE(*PL_oldoldbufptr))
7585 context = PL_oldoldbufptr;
7586 contlen = PL_bufptr - PL_oldoldbufptr;
7588 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7589 PL_oldbufptr != PL_bufptr) {
7590 while (isSPACE(*PL_oldbufptr))
7592 context = PL_oldbufptr;
7593 contlen = PL_bufptr - PL_oldbufptr;
7595 else if (yychar > 255)
7596 where = "next token ???";
7597 #ifdef USE_PURE_BISON
7598 /* GNU Bison sets the value -2 */
7599 else if (yychar == -2) {
7601 else if ((yychar & 127) == 127) {
7603 if (PL_lex_state == LEX_NORMAL ||
7604 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7605 where = "at end of line";
7606 else if (PL_lex_inpat)
7607 where = "within pattern";
7609 where = "within string";
7612 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7614 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7615 else if (isPRINT_LC(yychar))
7616 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7618 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7619 where = SvPVX(where_sv);
7621 msg = sv_2mortal(newSVpv(s, 0));
7622 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7623 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7625 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7627 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7628 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7629 Perl_sv_catpvf(aTHX_ msg,
7630 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7631 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7634 if (PL_in_eval & EVAL_WARNONLY)
7635 Perl_warn(aTHX_ "%"SVf, msg);
7638 if (PL_error_count >= 10) {
7639 if (PL_in_eval && SvCUR(ERRSV))
7640 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
7641 ERRSV, OutCopFILE(PL_curcop));
7643 Perl_croak(aTHX_ "%s has too many errors.\n",
7644 OutCopFILE(PL_curcop));
7647 PL_in_my_stash = Nullhv;
7651 #pragma segment Main
7655 S_swallow_bom(pTHX_ U8 *s)
7658 slen = SvCUR(PL_linestr);
7662 /* UTF-16 little-endian */
7663 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
7664 Perl_croak(aTHX_ "Unsupported script encoding");
7665 #ifndef PERL_NO_UTF16_FILTER
7666 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-LE script encoding\n"));
7668 if (PL_bufend > (char*)s) {
7672 filter_add(utf16rev_textfilter, NULL);
7673 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7674 PL_bufend = (char*)utf16_to_utf8_reversed(s, news,
7675 PL_bufend - (char*)s - 1,
7677 Copy(news, s, newlen, U8);
7678 SvCUR_set(PL_linestr, newlen);
7679 PL_bufend = SvPVX(PL_linestr) + newlen;
7680 news[newlen++] = '\0';
7684 Perl_croak(aTHX_ "Unsupported script encoding");
7689 if (s[1] == 0xFF) { /* UTF-16 big-endian */
7690 #ifndef PERL_NO_UTF16_FILTER
7691 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding\n"));
7693 if (PL_bufend > (char *)s) {
7697 filter_add(utf16_textfilter, NULL);
7698 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7699 PL_bufend = (char*)utf16_to_utf8(s, news,
7700 PL_bufend - (char*)s,
7702 Copy(news, s, newlen, U8);
7703 SvCUR_set(PL_linestr, newlen);
7704 PL_bufend = SvPVX(PL_linestr) + newlen;
7705 news[newlen++] = '\0';
7709 Perl_croak(aTHX_ "Unsupported script encoding");
7714 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
7715 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-8 script encoding\n"));
7720 if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */
7721 s[2] == 0xFE && s[3] == 0xFF)
7723 Perl_croak(aTHX_ "Unsupported script encoding");
7731 * Restore a source filter.
7735 restore_rsfp(pTHX_ void *f)
7737 PerlIO *fp = (PerlIO*)f;
7739 if (PL_rsfp == PerlIO_stdin())
7740 PerlIO_clearerr(PL_rsfp);
7741 else if (PL_rsfp && (PL_rsfp != fp))
7742 PerlIO_close(PL_rsfp);
7746 #ifndef PERL_NO_UTF16_FILTER
7748 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
7750 I32 count = FILTER_READ(idx+1, sv, maxlen);
7755 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7756 if (!*SvPV_nolen(sv))
7757 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7760 tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7761 sv_usepvn(sv, (char*)tmps, tend - tmps);
7767 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
7769 I32 count = FILTER_READ(idx+1, sv, maxlen);
7774 if (!*SvPV_nolen(sv))
7775 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7778 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7779 tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7780 sv_usepvn(sv, (char*)tmps, tend - tmps);