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);
324 * Deprecate a comma-less variable list.
330 deprecate("comma-less variable list");
334 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
335 * utf16-to-utf8-reversed.
338 #ifdef PERL_CR_FILTER
342 register char *s = SvPVX(sv);
343 register char *e = s + SvCUR(sv);
344 /* outer loop optimized to do nothing if there are no CR-LFs */
346 if (*s++ == '\r' && *s == '\n') {
347 /* hit a CR-LF, need to copy the rest */
348 register char *d = s - 1;
351 if (*s == '\r' && s[1] == '\n')
362 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
364 I32 count = FILTER_READ(idx+1, sv, maxlen);
365 if (count > 0 && !maxlen)
373 * Initialize variables. Uses the Perl save_stack to save its state (for
374 * recursive calls to the parser).
378 Perl_lex_start(pTHX_ SV *line)
383 SAVEI32(PL_lex_dojoin);
384 SAVEI32(PL_lex_brackets);
385 SAVEI32(PL_lex_casemods);
386 SAVEI32(PL_lex_starts);
387 SAVEI32(PL_lex_state);
388 SAVEVPTR(PL_lex_inpat);
389 SAVEI32(PL_lex_inwhat);
390 if (PL_lex_state == LEX_KNOWNEXT) {
391 I32 toke = PL_nexttoke;
392 while (--toke >= 0) {
393 SAVEI32(PL_nexttype[toke]);
394 SAVEVPTR(PL_nextval[toke]);
396 SAVEI32(PL_nexttoke);
398 SAVECOPLINE(PL_curcop);
401 SAVEPPTR(PL_oldbufptr);
402 SAVEPPTR(PL_oldoldbufptr);
403 SAVEPPTR(PL_last_lop);
404 SAVEPPTR(PL_last_uni);
405 SAVEPPTR(PL_linestart);
406 SAVESPTR(PL_linestr);
407 SAVEPPTR(PL_lex_brackstack);
408 SAVEPPTR(PL_lex_casestack);
409 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
410 SAVESPTR(PL_lex_stuff);
411 SAVEI32(PL_lex_defer);
412 SAVEI32(PL_sublex_info.sub_inwhat);
413 SAVESPTR(PL_lex_repl);
415 SAVEINT(PL_lex_expect);
417 PL_lex_state = LEX_NORMAL;
421 New(899, PL_lex_brackstack, 120, char);
422 New(899, PL_lex_casestack, 12, char);
423 SAVEFREEPV(PL_lex_brackstack);
424 SAVEFREEPV(PL_lex_casestack);
426 *PL_lex_casestack = '\0';
429 PL_lex_stuff = Nullsv;
430 PL_lex_repl = Nullsv;
434 PL_sublex_info.sub_inwhat = 0;
436 if (SvREADONLY(PL_linestr))
437 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
438 s = SvPV(PL_linestr, len);
439 if (len && s[len-1] != ';') {
440 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
441 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
442 sv_catpvn(PL_linestr, "\n;", 2);
444 SvTEMP_off(PL_linestr);
445 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
446 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
447 PL_last_lop = PL_last_uni = Nullch;
453 * Finalizer for lexing operations. Must be called when the parser is
454 * done with the lexer.
460 PL_doextract = FALSE;
465 * This subroutine has nothing to do with tilting, whether at windmills
466 * or pinball tables. Its name is short for "increment line". It
467 * increments the current line number in CopLINE(PL_curcop) and checks
468 * to see whether the line starts with a comment of the form
469 * # line 500 "foo.pm"
470 * If so, it sets the current line number and file to the values in the comment.
474 S_incline(pTHX_ char *s)
481 CopLINE_inc(PL_curcop);
484 while (SPACE_OR_TAB(*s)) s++;
485 if (strnEQ(s, "line", 4))
489 if (SPACE_OR_TAB(*s))
493 while (SPACE_OR_TAB(*s)) s++;
499 while (SPACE_OR_TAB(*s))
501 if (*s == '"' && (t = strchr(s+1, '"'))) {
506 for (t = s; !isSPACE(*t); t++) ;
509 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
511 if (*e != '\n' && *e != '\0')
512 return; /* false alarm */
517 CopFILE_free(PL_curcop);
518 CopFILE_set(PL_curcop, s);
521 CopLINE_set(PL_curcop, atoi(n)-1);
526 * Called to gobble the appropriate amount and type of whitespace.
527 * Skips comments as well.
531 S_skipspace(pTHX_ register char *s)
533 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
534 while (s < PL_bufend && SPACE_OR_TAB(*s))
540 SSize_t oldprevlen, oldoldprevlen;
541 SSize_t oldloplen = 0, oldunilen = 0;
542 while (s < PL_bufend && isSPACE(*s)) {
543 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
548 if (s < PL_bufend && *s == '#') {
549 while (s < PL_bufend && *s != '\n')
553 if (PL_in_eval && !PL_rsfp) {
560 /* only continue to recharge the buffer if we're at the end
561 * of the buffer, we're not reading from a source filter, and
562 * we're in normal lexing mode
564 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
565 PL_lex_state == LEX_FORMLINE)
568 /* try to recharge the buffer */
569 if ((s = filter_gets(PL_linestr, PL_rsfp,
570 (prevlen = SvCUR(PL_linestr)))) == Nullch)
572 /* end of file. Add on the -p or -n magic */
573 if (PL_minus_n || PL_minus_p) {
574 sv_setpv(PL_linestr,PL_minus_p ?
575 ";}continue{print or die qq(-p destination: $!\\n)" :
577 sv_catpv(PL_linestr,";}");
578 PL_minus_n = PL_minus_p = 0;
581 sv_setpv(PL_linestr,";");
583 /* reset variables for next time we lex */
584 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
586 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
587 PL_last_lop = PL_last_uni = Nullch;
589 /* Close the filehandle. Could be from -P preprocessor,
590 * STDIN, or a regular file. If we were reading code from
591 * STDIN (because the commandline held no -e or filename)
592 * then we don't close it, we reset it so the code can
593 * read from STDIN too.
596 if (PL_preprocess && !PL_in_eval)
597 (void)PerlProc_pclose(PL_rsfp);
598 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
599 PerlIO_clearerr(PL_rsfp);
601 (void)PerlIO_close(PL_rsfp);
606 /* not at end of file, so we only read another line */
607 /* make corresponding updates to old pointers, for yyerror() */
608 oldprevlen = PL_oldbufptr - PL_bufend;
609 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
611 oldunilen = PL_last_uni - PL_bufend;
613 oldloplen = PL_last_lop - PL_bufend;
614 PL_linestart = PL_bufptr = s + prevlen;
615 PL_bufend = s + SvCUR(PL_linestr);
617 PL_oldbufptr = s + oldprevlen;
618 PL_oldoldbufptr = s + oldoldprevlen;
620 PL_last_uni = s + oldunilen;
622 PL_last_lop = s + oldloplen;
625 /* debugger active and we're not compiling the debugger code,
626 * so store the line into the debugger's array of lines
628 if (PERLDB_LINE && PL_curstash != PL_debstash) {
629 SV *sv = NEWSV(85,0);
631 sv_upgrade(sv, SVt_PVMG);
632 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
635 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
642 * Check the unary operators to ensure there's no ambiguity in how they're
643 * used. An ambiguous piece of code would be:
645 * This doesn't mean rand() + 5. Because rand() is a unary operator,
646 * the +5 is its argument.
655 if (PL_oldoldbufptr != PL_last_uni)
657 while (isSPACE(*PL_last_uni))
659 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
660 if ((t = strchr(s, '(')) && t < PL_bufptr)
662 if (ckWARN_d(WARN_AMBIGUOUS)){
665 Perl_warner(aTHX_ WARN_AMBIGUOUS,
666 "Warning: Use of \"%s\" without parens is ambiguous",
672 /* workaround to replace the UNI() macro with a function. Only the
673 * hints/uts.sh file mentions this. Other comments elsewhere in the
674 * source indicate Microport Unix might need it too.
680 #define UNI(f) return uni(f,s)
683 S_uni(pTHX_ I32 f, char *s)
688 PL_last_uni = PL_oldbufptr;
699 #endif /* CRIPPLED_CC */
702 * LOP : macro to build a list operator. Its behaviour has been replaced
703 * with a subroutine, S_lop() for which LOP is just another name.
706 #define LOP(f,x) return lop(f,x,s)
710 * Build a list operator (or something that might be one). The rules:
711 * - if we have a next token, then it's a list operator [why?]
712 * - if the next thing is an opening paren, then it's a function
713 * - else it's a list operator
717 S_lop(pTHX_ I32 f, int x, char *s)
724 PL_last_lop = PL_oldbufptr;
739 * When the lexer realizes it knows the next token (for instance,
740 * it is reordering tokens for the parser) then it can call S_force_next
741 * to know what token to return the next time the lexer is called. Caller
742 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
743 * handles the token correctly.
747 S_force_next(pTHX_ I32 type)
749 PL_nexttype[PL_nexttoke] = type;
751 if (PL_lex_state != LEX_KNOWNEXT) {
752 PL_lex_defer = PL_lex_state;
753 PL_lex_expect = PL_expect;
754 PL_lex_state = LEX_KNOWNEXT;
760 * When the lexer knows the next thing is a word (for instance, it has
761 * just seen -> and it knows that the next char is a word char, then
762 * it calls S_force_word to stick the next word into the PL_next lookahead.
765 * char *start : buffer position (must be within PL_linestr)
766 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
767 * int check_keyword : if true, Perl checks to make sure the word isn't
768 * a keyword (do this if the word is a label, e.g. goto FOO)
769 * int allow_pack : if true, : characters will also be allowed (require,
771 * int allow_initial_tick : used by the "sub" lexer only.
775 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
780 start = skipspace(start);
782 if (isIDFIRST_lazy_if(s,UTF) ||
783 (allow_pack && *s == ':') ||
784 (allow_initial_tick && *s == '\'') )
786 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
787 if (check_keyword && keyword(PL_tokenbuf, len))
789 if (token == METHOD) {
794 PL_expect = XOPERATOR;
797 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
798 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
806 * Called when the lexer wants $foo *foo &foo etc, but the program
807 * text only contains the "foo" portion. The first argument is a pointer
808 * to the "foo", and the second argument is the type symbol to prefix.
809 * Forces the next token to be a "WORD".
810 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
814 S_force_ident(pTHX_ register char *s, int kind)
817 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
818 PL_nextval[PL_nexttoke].opval = o;
821 o->op_private = OPpCONST_ENTERED;
822 /* XXX see note in pp_entereval() for why we forgo typo
823 warnings if the symbol must be introduced in an eval.
825 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
826 kind == '$' ? SVt_PV :
827 kind == '@' ? SVt_PVAV :
828 kind == '%' ? SVt_PVHV :
836 Perl_str_to_version(pTHX_ SV *sv)
841 char *start = SvPVx(sv,len);
842 bool utf = SvUTF8(sv) ? TRUE : FALSE;
843 char *end = start + len;
844 while (start < end) {
848 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
853 retval += ((NV)n)/nshift;
862 * Forces the next token to be a version number.
863 * If the next token appears to be an invalid version number, (e.g. "v2b"),
864 * and if "guessing" is TRUE, then no new token is created (and the caller
865 * must use an alternative parsing method).
869 S_force_version(pTHX_ char *s, int guessing)
871 OP *version = Nullop;
880 while (isDIGIT(*d) || *d == '_' || *d == '.')
882 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
884 s = scan_num(s, &yylval);
885 version = yylval.opval;
886 ver = cSVOPx(version)->op_sv;
887 if (SvPOK(ver) && !SvNIOK(ver)) {
888 (void)SvUPGRADE(ver, SVt_PVNV);
889 SvNVX(ver) = str_to_version(ver);
890 SvNOK_on(ver); /* hint that it is a version */
897 /* NOTE: The parser sees the package name and the VERSION swapped */
898 PL_nextval[PL_nexttoke].opval = version;
906 * Tokenize a quoted string passed in as an SV. It finds the next
907 * chunk, up to end of string or a backslash. It may make a new
908 * SV containing that chunk (if HINT_NEW_STRING is on). It also
913 S_tokeq(pTHX_ SV *sv)
924 s = SvPV_force(sv, len);
925 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
928 while (s < send && *s != '\\')
933 if ( PL_hints & HINT_NEW_STRING ) {
934 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
940 if (s + 1 < send && (s[1] == '\\'))
941 s++; /* all that, just for this */
946 SvCUR_set(sv, d - SvPVX(sv));
948 if ( PL_hints & HINT_NEW_STRING )
949 return new_constant(NULL, 0, "q", sv, pv, "q");
954 * Now come three functions related to double-quote context,
955 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
956 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
957 * interact with PL_lex_state, and create fake ( ... ) argument lists
958 * to handle functions and concatenation.
959 * They assume that whoever calls them will be setting up a fake
960 * join call, because each subthing puts a ',' after it. This lets
963 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
965 * (I'm not sure whether the spurious commas at the end of lcfirst's
966 * arguments and join's arguments are created or not).
971 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
973 * Pattern matching will set PL_lex_op to the pattern-matching op to
974 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
976 * OP_CONST and OP_READLINE are easy--just make the new op and return.
978 * Everything else becomes a FUNC.
980 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
981 * had an OP_CONST or OP_READLINE). This just sets us up for a
982 * call to S_sublex_push().
988 register I32 op_type = yylval.ival;
990 if (op_type == OP_NULL) {
991 yylval.opval = PL_lex_op;
995 if (op_type == OP_CONST || op_type == OP_READLINE) {
996 SV *sv = tokeq(PL_lex_stuff);
998 if (SvTYPE(sv) == SVt_PVIV) {
999 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
1005 nsv = newSVpvn(p, len);
1011 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
1012 PL_lex_stuff = Nullsv;
1016 PL_sublex_info.super_state = PL_lex_state;
1017 PL_sublex_info.sub_inwhat = op_type;
1018 PL_sublex_info.sub_op = PL_lex_op;
1019 PL_lex_state = LEX_INTERPPUSH;
1023 yylval.opval = PL_lex_op;
1033 * Create a new scope to save the lexing state. The scope will be
1034 * ended in S_sublex_done. Returns a '(', starting the function arguments
1035 * to the uc, lc, etc. found before.
1036 * Sets PL_lex_state to LEX_INTERPCONCAT.
1044 PL_lex_state = PL_sublex_info.super_state;
1045 SAVEI32(PL_lex_dojoin);
1046 SAVEI32(PL_lex_brackets);
1047 SAVEI32(PL_lex_casemods);
1048 SAVEI32(PL_lex_starts);
1049 SAVEI32(PL_lex_state);
1050 SAVEVPTR(PL_lex_inpat);
1051 SAVEI32(PL_lex_inwhat);
1052 SAVECOPLINE(PL_curcop);
1053 SAVEPPTR(PL_bufptr);
1054 SAVEPPTR(PL_bufend);
1055 SAVEPPTR(PL_oldbufptr);
1056 SAVEPPTR(PL_oldoldbufptr);
1057 SAVEPPTR(PL_last_lop);
1058 SAVEPPTR(PL_last_uni);
1059 SAVEPPTR(PL_linestart);
1060 SAVESPTR(PL_linestr);
1061 SAVEPPTR(PL_lex_brackstack);
1062 SAVEPPTR(PL_lex_casestack);
1064 PL_linestr = PL_lex_stuff;
1065 PL_lex_stuff = Nullsv;
1067 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1068 = SvPVX(PL_linestr);
1069 PL_bufend += SvCUR(PL_linestr);
1070 PL_last_lop = PL_last_uni = Nullch;
1071 SAVEFREESV(PL_linestr);
1073 PL_lex_dojoin = FALSE;
1074 PL_lex_brackets = 0;
1075 New(899, PL_lex_brackstack, 120, char);
1076 New(899, PL_lex_casestack, 12, char);
1077 SAVEFREEPV(PL_lex_brackstack);
1078 SAVEFREEPV(PL_lex_casestack);
1079 PL_lex_casemods = 0;
1080 *PL_lex_casestack = '\0';
1082 PL_lex_state = LEX_INTERPCONCAT;
1083 CopLINE_set(PL_curcop, PL_multi_start);
1085 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1086 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1087 PL_lex_inpat = PL_sublex_info.sub_op;
1089 PL_lex_inpat = Nullop;
1096 * Restores lexer state after a S_sublex_push.
1102 if (!PL_lex_starts++) {
1103 SV *sv = newSVpvn("",0);
1104 if (SvUTF8(PL_linestr))
1106 PL_expect = XOPERATOR;
1107 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1111 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1112 PL_lex_state = LEX_INTERPCASEMOD;
1116 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1117 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1118 PL_linestr = PL_lex_repl;
1120 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1121 PL_bufend += SvCUR(PL_linestr);
1122 PL_last_lop = PL_last_uni = Nullch;
1123 SAVEFREESV(PL_linestr);
1124 PL_lex_dojoin = FALSE;
1125 PL_lex_brackets = 0;
1126 PL_lex_casemods = 0;
1127 *PL_lex_casestack = '\0';
1129 if (SvEVALED(PL_lex_repl)) {
1130 PL_lex_state = LEX_INTERPNORMAL;
1132 /* we don't clear PL_lex_repl here, so that we can check later
1133 whether this is an evalled subst; that means we rely on the
1134 logic to ensure sublex_done() is called again only via the
1135 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1138 PL_lex_state = LEX_INTERPCONCAT;
1139 PL_lex_repl = Nullsv;
1145 PL_bufend = SvPVX(PL_linestr);
1146 PL_bufend += SvCUR(PL_linestr);
1147 PL_expect = XOPERATOR;
1148 PL_sublex_info.sub_inwhat = 0;
1156 Extracts a pattern, double-quoted string, or transliteration. This
1159 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1160 processing a pattern (PL_lex_inpat is true), a transliteration
1161 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1163 Returns a pointer to the character scanned up to. Iff this is
1164 advanced from the start pointer supplied (ie if anything was
1165 successfully parsed), will leave an OP for the substring scanned
1166 in yylval. Caller must intuit reason for not parsing further
1167 by looking at the next characters herself.
1171 double-quoted style: \r and \n
1172 regexp special ones: \D \s
1174 backrefs: \1 (deprecated in substitution replacements)
1175 case and quoting: \U \Q \E
1176 stops on @ and $, but not for $ as tail anchor
1178 In transliterations:
1179 characters are VERY literal, except for - not at the start or end
1180 of the string, which indicates a range. scan_const expands the
1181 range to the full set of intermediate characters.
1183 In double-quoted strings:
1185 double-quoted style: \r and \n
1187 backrefs: \1 (deprecated)
1188 case and quoting: \U \Q \E
1191 scan_const does *not* construct ops to handle interpolated strings.
1192 It stops processing as soon as it finds an embedded $ or @ variable
1193 and leaves it to the caller to work out what's going on.
1195 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1197 $ in pattern could be $foo or could be tail anchor. Assumption:
1198 it's a tail anchor if $ is the last thing in the string, or if it's
1199 followed by one of ")| \n\t"
1201 \1 (backreferences) are turned into $1
1203 The structure of the code is
1204 while (there's a character to process) {
1205 handle transliteration ranges
1206 skip regexp comments
1207 skip # initiated comments in //x patterns
1208 check for embedded @foo
1209 check for embedded scalars
1211 leave intact backslashes from leave (below)
1212 deprecate \1 in strings and sub replacements
1213 handle string-changing backslashes \l \U \Q \E, etc.
1214 switch (what was escaped) {
1215 handle - in a transliteration (becomes a literal -)
1216 handle \132 octal characters
1217 handle 0x15 hex characters
1218 handle \cV (control V)
1219 handle printf backslashes (\f, \r, \n, etc)
1221 } (end if backslash)
1222 } (end while character to read)
1227 S_scan_const(pTHX_ char *start)
1229 register char *send = PL_bufend; /* end of the constant */
1230 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1231 register char *s = start; /* start of the constant */
1232 register char *d = SvPVX(sv); /* destination for copies */
1233 bool dorange = FALSE; /* are we in a translit range? */
1234 bool didrange = FALSE; /* did we just finish a range? */
1235 I32 has_utf8 = FALSE; /* Output constant is UTF8 */
1236 I32 this_utf8 = UTF; /* The source string is assumed to be UTF8 */
1239 const char *leaveit = /* set of acceptably-backslashed characters */
1241 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1244 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1245 /* If we are doing a trans and we know we want UTF8 set expectation */
1246 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
1247 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1251 while (s < send || dorange) {
1252 /* get transliterations out of the way (they're most literal) */
1253 if (PL_lex_inwhat == OP_TRANS) {
1254 /* expand a range A-Z to the full set of characters. AIE! */
1256 I32 i; /* current expanded character */
1257 I32 min; /* first character in range */
1258 I32 max; /* last character in range */
1261 char *c = (char*)utf8_hop((U8*)d, -1);
1265 *c = (char)UTF_TO_NATIVE(0xff);
1266 /* mark the range as done, and continue */
1272 i = d - SvPVX(sv); /* remember current offset */
1273 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1274 d = SvPVX(sv) + i; /* refresh d after realloc */
1275 d -= 2; /* eat the first char and the - */
1277 min = (U8)*d; /* first char in range */
1278 max = (U8)d[1]; /* last char in range */
1282 "Invalid [] range \"%c-%c\" in transliteration operator",
1283 (char)min, (char)max);
1287 if ((isLOWER(min) && isLOWER(max)) ||
1288 (isUPPER(min) && isUPPER(max))) {
1290 for (i = min; i <= max; i++)
1292 *d++ = NATIVE_TO_NEED(has_utf8,i);
1294 for (i = min; i <= max; i++)
1296 *d++ = NATIVE_TO_NEED(has_utf8,i);
1301 for (i = min; i <= max; i++)
1304 /* mark the range as done, and continue */
1310 /* range begins (ignore - as first or last char) */
1311 else if (*s == '-' && s+1 < send && s != start) {
1313 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
1316 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
1328 /* if we get here, we're not doing a transliteration */
1330 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1331 except for the last char, which will be done separately. */
1332 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1334 while (s < send && *s != ')')
1335 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1337 else if (s[2] == '{' /* This should match regcomp.c */
1338 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1341 char *regparse = s + (s[2] == '{' ? 3 : 4);
1344 while (count && (c = *regparse)) {
1345 if (c == '\\' && regparse[1])
1353 if (*regparse != ')') {
1354 regparse--; /* Leave one char for continuation. */
1355 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1357 while (s < regparse)
1358 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1362 /* likewise skip #-initiated comments in //x patterns */
1363 else if (*s == '#' && PL_lex_inpat &&
1364 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1365 while (s+1 < send && *s != '\n')
1366 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1369 /* check for embedded arrays
1370 (@foo, @:foo, @'foo, @{foo}, @$foo, @+, @-)
1372 else if (*s == '@' && s[1]
1373 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1])))
1376 /* check for embedded scalars. only stop if we're sure it's a
1379 else if (*s == '$') {
1380 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1382 if (s + 1 < send && !strchr("()| \r\n\t", s[1]))
1383 break; /* in regexp, $ might be tail anchor */
1386 /* End of else if chain - OP_TRANS rejoin rest */
1389 if (*s == '\\' && s+1 < send) {
1392 /* some backslashes we leave behind */
1393 if (*leaveit && *s && strchr(leaveit, *s)) {
1394 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
1395 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1399 /* deprecate \1 in strings and substitution replacements */
1400 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1401 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1403 if (ckWARN(WARN_SYNTAX))
1404 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1409 /* string-change backslash escapes */
1410 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1415 /* if we get here, it's either a quoted -, or a digit */
1418 /* quoted - in transliterations */
1420 if (PL_lex_inwhat == OP_TRANS) {
1427 if (ckWARN(WARN_MISC) &&
1430 Perl_warner(aTHX_ WARN_MISC,
1431 "Unrecognized escape \\%c passed through",
1433 /* default action is to copy the quoted character */
1434 goto default_action;
1437 /* \132 indicates an octal constant */
1438 case '0': case '1': case '2': case '3':
1439 case '4': case '5': case '6': case '7':
1443 uv = grok_oct(s, &len, &flags, NULL);
1446 goto NUM_ESCAPE_INSERT;
1448 /* \x24 indicates a hex constant */
1452 char* e = strchr(s, '}');
1453 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1454 PERL_SCAN_DISALLOW_PREFIX;
1459 yyerror("Missing right brace on \\x{}");
1463 uv = grok_hex(s, &len, &flags, NULL);
1469 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
1470 uv = grok_hex(s, &len, &flags, NULL);
1476 /* Insert oct or hex escaped character.
1477 * There will always enough room in sv since such
1478 * escapes will be longer than any UTF-8 sequence
1479 * they can end up as. */
1481 /* We need to map to chars to ASCII before doing the tests
1484 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(uv))) {
1485 if (!has_utf8 && uv > 255) {
1486 /* Might need to recode whatever we have
1487 * accumulated so far if it contains any
1490 * (Can't we keep track of that and avoid
1491 * this rescan? --jhi)
1495 for (c = (U8 *) SvPVX(sv); c < (U8 *)d; c++) {
1496 if (!NATIVE_IS_INVARIANT(*c)) {
1501 STRLEN offset = d - SvPVX(sv);
1503 d = SvGROW(sv, SvLEN(sv) + hicount + 1) + offset;
1507 while (src >= (U8 *)SvPVX(sv)) {
1508 if (!NATIVE_IS_INVARIANT(*src)) {
1509 U8 ch = NATIVE_TO_ASCII(*src);
1510 *dst-- = UTF8_EIGHT_BIT_LO(ch);
1511 *dst-- = UTF8_EIGHT_BIT_HI(ch);
1521 if (has_utf8 || uv > 255) {
1522 d = (char*)uvchr_to_utf8((U8*)d, uv);
1524 if (PL_lex_inwhat == OP_TRANS &&
1525 PL_sublex_info.sub_op) {
1526 PL_sublex_info.sub_op->op_private |=
1527 (PL_lex_repl ? OPpTRANS_FROM_UTF
1540 /* \N{LATIN SMALL LETTER A} is a named character */
1544 char* e = strchr(s, '}');
1550 yyerror("Missing right brace on \\N{}");
1554 res = newSVpvn(s + 1, e - s - 1);
1555 res = new_constant( Nullch, 0, "charnames",
1556 res, Nullsv, "\\N{...}" );
1558 sv_utf8_upgrade(res);
1559 str = SvPV(res,len);
1560 #ifdef EBCDIC_NEVER_MIND
1561 /* charnames uses pack U and that has been
1562 * recently changed to do the below uni->native
1563 * mapping, so this would be redundant (and wrong,
1564 * the code point would be doubly converted).
1565 * But leave this in just in case the pack U change
1566 * gets revoked, but the semantics is still
1567 * desireable for charnames. --jhi */
1569 UV uv = utf8_to_uvchr((U8*)str, 0);
1572 U8 tmpbuf[UTF8_MAXLEN+1], *d;
1574 d = uvchr_to_utf8(tmpbuf, UNI_TO_NATIVE(uv));
1575 sv_setpvn(res, (char *)tmpbuf, d - tmpbuf);
1576 str = SvPV(res, len);
1580 if (!has_utf8 && SvUTF8(res)) {
1581 char *ostart = SvPVX(sv);
1582 SvCUR_set(sv, d - ostart);
1585 sv_utf8_upgrade(sv);
1586 /* this just broke our allocation above... */
1587 SvGROW(sv, send - start);
1588 d = SvPVX(sv) + SvCUR(sv);
1591 if (len > e - s + 4) { /* I _guess_ 4 is \N{} --jhi */
1592 char *odest = SvPVX(sv);
1594 SvGROW(sv, (SvLEN(sv) + len - (e - s + 4)));
1595 d = SvPVX(sv) + (d - odest);
1597 Copy(str, d, len, char);
1604 yyerror("Missing braces on \\N{}");
1607 /* \c is a control character */
1616 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
1620 /* printf-style backslashes, formfeeds, newlines, etc */
1622 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
1625 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
1628 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
1631 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
1634 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
1637 *d++ = ASCII_TO_NEED(has_utf8,'\033');
1640 *d++ = ASCII_TO_NEED(has_utf8,'\007');
1646 } /* end if (backslash) */
1649 /* If we started with encoded form, or already know we want it
1650 and then encode the next character */
1651 if ((has_utf8 || this_utf8) && !NATIVE_IS_INVARIANT((U8)(*s))) {
1653 UV uv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
1654 STRLEN need = UNISKIP(NATIVE_TO_UNI(uv));
1657 /* encoded value larger than old, need extra space (NOTE: SvCUR() not set here) */
1658 STRLEN off = d - SvPVX(sv);
1659 d = SvGROW(sv, SvLEN(sv) + (need-len)) + off;
1661 d = (char*)uvchr_to_utf8((U8*)d, uv);
1665 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1667 } /* while loop to process each character */
1669 /* terminate the string and set up the sv */
1671 SvCUR_set(sv, d - SvPVX(sv));
1672 if (SvCUR(sv) >= SvLEN(sv))
1673 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
1676 if (PL_encoding && !has_utf8) {
1677 Perl_sv_recode_to_utf8(aTHX_ sv, PL_encoding);
1682 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1683 PL_sublex_info.sub_op->op_private |=
1684 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1688 /* shrink the sv if we allocated more than we used */
1689 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1690 SvLEN_set(sv, SvCUR(sv) + 1);
1691 Renew(SvPVX(sv), SvLEN(sv), char);
1694 /* return the substring (via yylval) only if we parsed anything */
1695 if (s > PL_bufptr) {
1696 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1697 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1699 ( PL_lex_inwhat == OP_TRANS
1701 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1704 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1711 * Returns TRUE if there's more to the expression (e.g., a subscript),
1714 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1716 * ->[ and ->{ return TRUE
1717 * { and [ outside a pattern are always subscripts, so return TRUE
1718 * if we're outside a pattern and it's not { or [, then return FALSE
1719 * if we're in a pattern and the first char is a {
1720 * {4,5} (any digits around the comma) returns FALSE
1721 * if we're in a pattern and the first char is a [
1723 * [SOMETHING] has a funky algorithm to decide whether it's a
1724 * character class or not. It has to deal with things like
1725 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1726 * anything else returns TRUE
1729 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1732 S_intuit_more(pTHX_ register char *s)
1734 if (PL_lex_brackets)
1736 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1738 if (*s != '{' && *s != '[')
1743 /* In a pattern, so maybe we have {n,m}. */
1760 /* On the other hand, maybe we have a character class */
1763 if (*s == ']' || *s == '^')
1766 /* this is terrifying, and it works */
1767 int weight = 2; /* let's weigh the evidence */
1769 unsigned char un_char = 255, last_un_char;
1770 char *send = strchr(s,']');
1771 char tmpbuf[sizeof PL_tokenbuf * 4];
1773 if (!send) /* has to be an expression */
1776 Zero(seen,256,char);
1779 else if (isDIGIT(*s)) {
1781 if (isDIGIT(s[1]) && s[2] == ']')
1787 for (; s < send; s++) {
1788 last_un_char = un_char;
1789 un_char = (unsigned char)*s;
1794 weight -= seen[un_char] * 10;
1795 if (isALNUM_lazy_if(s+1,UTF)) {
1796 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1797 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1802 else if (*s == '$' && s[1] &&
1803 strchr("[#!%*<>()-=",s[1])) {
1804 if (/*{*/ strchr("])} =",s[2]))
1813 if (strchr("wds]",s[1]))
1815 else if (seen['\''] || seen['"'])
1817 else if (strchr("rnftbxcav",s[1]))
1819 else if (isDIGIT(s[1])) {
1821 while (s[1] && isDIGIT(s[1]))
1831 if (strchr("aA01! ",last_un_char))
1833 if (strchr("zZ79~",s[1]))
1835 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1836 weight -= 5; /* cope with negative subscript */
1839 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1840 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1845 if (keyword(tmpbuf, d - tmpbuf))
1848 if (un_char == last_un_char + 1)
1850 weight -= seen[un_char];
1855 if (weight >= 0) /* probably a character class */
1865 * Does all the checking to disambiguate
1867 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1868 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1870 * First argument is the stuff after the first token, e.g. "bar".
1872 * Not a method if bar is a filehandle.
1873 * Not a method if foo is a subroutine prototyped to take a filehandle.
1874 * Not a method if it's really "Foo $bar"
1875 * Method if it's "foo $bar"
1876 * Not a method if it's really "print foo $bar"
1877 * Method if it's really "foo package::" (interpreted as package->foo)
1878 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1879 * Not a method if bar is a filehandle or package, but is quoted with
1884 S_intuit_method(pTHX_ char *start, GV *gv)
1886 char *s = start + (*start == '$');
1887 char tmpbuf[sizeof PL_tokenbuf];
1895 if ((cv = GvCVu(gv))) {
1896 char *proto = SvPVX(cv);
1906 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1907 /* start is the beginning of the possible filehandle/object,
1908 * and s is the end of it
1909 * tmpbuf is a copy of it
1912 if (*start == '$') {
1913 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1918 return *s == '(' ? FUNCMETH : METHOD;
1920 if (!keyword(tmpbuf, len)) {
1921 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1926 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1927 if (indirgv && GvCVu(indirgv))
1929 /* filehandle or package name makes it a method */
1930 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1932 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1933 return 0; /* no assumptions -- "=>" quotes bearword */
1935 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1936 newSVpvn(tmpbuf,len));
1937 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1941 return *s == '(' ? FUNCMETH : METHOD;
1949 * Return a string of Perl code to load the debugger. If PERL5DB
1950 * is set, it will return the contents of that, otherwise a
1951 * compile-time require of perl5db.pl.
1958 char *pdb = PerlEnv_getenv("PERL5DB");
1962 SETERRNO(0,SS$_NORMAL);
1963 return "BEGIN { require 'perl5db.pl' }";
1969 /* Encoded script support. filter_add() effectively inserts a
1970 * 'pre-processing' function into the current source input stream.
1971 * Note that the filter function only applies to the current source file
1972 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1974 * The datasv parameter (which may be NULL) can be used to pass
1975 * private data to this instance of the filter. The filter function
1976 * can recover the SV using the FILTER_DATA macro and use it to
1977 * store private buffers and state information.
1979 * The supplied datasv parameter is upgraded to a PVIO type
1980 * and the IoDIRP/IoANY field is used to store the function pointer,
1981 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1982 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1983 * private use must be set using malloc'd pointers.
1987 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1992 if (!PL_rsfp_filters)
1993 PL_rsfp_filters = newAV();
1995 datasv = NEWSV(255,0);
1996 if (!SvUPGRADE(datasv, SVt_PVIO))
1997 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1998 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
1999 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
2000 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
2001 (void*)funcp, SvPV_nolen(datasv)));
2002 av_unshift(PL_rsfp_filters, 1);
2003 av_store(PL_rsfp_filters, 0, datasv) ;
2008 /* Delete most recently added instance of this filter function. */
2010 Perl_filter_del(pTHX_ filter_t funcp)
2013 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", (void*)funcp));
2014 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
2016 /* if filter is on top of stack (usual case) just pop it off */
2017 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
2018 if (IoANY(datasv) == (void *)funcp) {
2019 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
2020 IoANY(datasv) = (void *)NULL;
2021 sv_free(av_pop(PL_rsfp_filters));
2025 /* we need to search for the correct entry and clear it */
2026 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
2030 /* Invoke the n'th filter function for the current rsfp. */
2032 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
2035 /* 0 = read one text line */
2040 if (!PL_rsfp_filters)
2042 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
2043 /* Provide a default input filter to make life easy. */
2044 /* Note that we append to the line. This is handy. */
2045 DEBUG_P(PerlIO_printf(Perl_debug_log,
2046 "filter_read %d: from rsfp\n", idx));
2050 int old_len = SvCUR(buf_sv) ;
2052 /* ensure buf_sv is large enough */
2053 SvGROW(buf_sv, old_len + maxlen) ;
2054 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
2055 if (PerlIO_error(PL_rsfp))
2056 return -1; /* error */
2058 return 0 ; /* end of file */
2060 SvCUR_set(buf_sv, old_len + len) ;
2063 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2064 if (PerlIO_error(PL_rsfp))
2065 return -1; /* error */
2067 return 0 ; /* end of file */
2070 return SvCUR(buf_sv);
2072 /* Skip this filter slot if filter has been deleted */
2073 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
2074 DEBUG_P(PerlIO_printf(Perl_debug_log,
2075 "filter_read %d: skipped (filter deleted)\n",
2077 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
2079 /* Get function pointer hidden within datasv */
2080 funcp = (filter_t)IoANY(datasv);
2081 DEBUG_P(PerlIO_printf(Perl_debug_log,
2082 "filter_read %d: via function %p (%s)\n",
2083 idx, (void*)funcp, SvPV_nolen(datasv)));
2084 /* Call function. The function is expected to */
2085 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2086 /* Return: <0:error, =0:eof, >0:not eof */
2087 return (*funcp)(aTHX_ idx, buf_sv, maxlen);
2091 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2093 #ifdef PERL_CR_FILTER
2094 if (!PL_rsfp_filters) {
2095 filter_add(S_cr_textfilter,NULL);
2098 if (PL_rsfp_filters) {
2101 SvCUR_set(sv, 0); /* start with empty line */
2102 if (FILTER_READ(0, sv, 0) > 0)
2103 return ( SvPVX(sv) ) ;
2108 return (sv_gets(sv, fp, append));
2112 S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2116 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2120 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2121 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2123 return GvHV(gv); /* Foo:: */
2126 /* use constant CLASS => 'MyClass' */
2127 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2129 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2130 pkgname = SvPV_nolen(sv);
2134 return gv_stashpv(pkgname, FALSE);
2138 static char* exp_name[] =
2139 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2140 "ATTRTERM", "TERMBLOCK"
2147 Works out what to call the token just pulled out of the input
2148 stream. The yacc parser takes care of taking the ops we return and
2149 stitching them into a tree.
2155 if read an identifier
2156 if we're in a my declaration
2157 croak if they tried to say my($foo::bar)
2158 build the ops for a my() declaration
2159 if it's an access to a my() variable
2160 are we in a sort block?
2161 croak if my($a); $a <=> $b
2162 build ops for access to a my() variable
2163 if in a dq string, and they've said @foo and we can't find @foo
2165 build ops for a bareword
2166 if we already built the token before, use it.
2169 #ifdef USE_PURE_BISON
2171 Perl_yylex_r(pTHX_ YYSTYPE *lvalp, int *lcharp)
2176 yylval_pointer[yyactlevel] = lvalp;
2177 yychar_pointer[yyactlevel] = lcharp;
2178 if (yyactlevel >= YYMAXLEVEL)
2179 Perl_croak(aTHX_ "panic: YYMAXLEVEL");
2181 r = Perl_yylex(aTHX);
2191 #pragma segment Perl_yylex
2204 /* check if there's an identifier for us to look at */
2205 if (PL_pending_ident)
2206 return S_pending_ident(aTHX);
2208 /* no identifier pending identification */
2210 switch (PL_lex_state) {
2212 case LEX_NORMAL: /* Some compilers will produce faster */
2213 case LEX_INTERPNORMAL: /* code if we comment these out. */
2217 /* when we've already built the next token, just pull it out of the queue */
2220 yylval = PL_nextval[PL_nexttoke];
2222 PL_lex_state = PL_lex_defer;
2223 PL_expect = PL_lex_expect;
2224 PL_lex_defer = LEX_NORMAL;
2226 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2227 "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr,
2228 (IV)PL_nexttype[PL_nexttoke]); });
2230 return(PL_nexttype[PL_nexttoke]);
2232 /* interpolated case modifiers like \L \U, including \Q and \E.
2233 when we get here, PL_bufptr is at the \
2235 case LEX_INTERPCASEMOD:
2237 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2238 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2240 /* handle \E or end of string */
2241 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2245 if (PL_lex_casemods) {
2246 oldmod = PL_lex_casestack[--PL_lex_casemods];
2247 PL_lex_casestack[PL_lex_casemods] = '\0';
2249 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2251 PL_lex_state = LEX_INTERPCONCAT;
2255 if (PL_bufptr != PL_bufend)
2257 PL_lex_state = LEX_INTERPCONCAT;
2261 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2262 "### Saw case modifier at '%s'\n", PL_bufptr); });
2264 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2265 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2266 if (strchr("LU", *s) &&
2267 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2269 PL_lex_casestack[--PL_lex_casemods] = '\0';
2272 if (PL_lex_casemods > 10) {
2273 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2274 if (newlb != PL_lex_casestack) {
2276 PL_lex_casestack = newlb;
2279 PL_lex_casestack[PL_lex_casemods++] = *s;
2280 PL_lex_casestack[PL_lex_casemods] = '\0';
2281 PL_lex_state = LEX_INTERPCONCAT;
2282 PL_nextval[PL_nexttoke].ival = 0;
2285 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2287 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2289 PL_nextval[PL_nexttoke].ival = OP_LC;
2291 PL_nextval[PL_nexttoke].ival = OP_UC;
2293 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2295 Perl_croak(aTHX_ "panic: yylex");
2298 if (PL_lex_starts) {
2307 case LEX_INTERPPUSH:
2308 return sublex_push();
2310 case LEX_INTERPSTART:
2311 if (PL_bufptr == PL_bufend)
2312 return sublex_done();
2313 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2314 "### Interpolated variable at '%s'\n", PL_bufptr); });
2316 PL_lex_dojoin = (*PL_bufptr == '@');
2317 PL_lex_state = LEX_INTERPNORMAL;
2318 if (PL_lex_dojoin) {
2319 PL_nextval[PL_nexttoke].ival = 0;
2321 #ifdef USE_5005THREADS
2322 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2323 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2324 force_next(PRIVATEREF);
2326 force_ident("\"", '$');
2327 #endif /* USE_5005THREADS */
2328 PL_nextval[PL_nexttoke].ival = 0;
2330 PL_nextval[PL_nexttoke].ival = 0;
2332 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2335 if (PL_lex_starts++) {
2341 case LEX_INTERPENDMAYBE:
2342 if (intuit_more(PL_bufptr)) {
2343 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2349 if (PL_lex_dojoin) {
2350 PL_lex_dojoin = FALSE;
2351 PL_lex_state = LEX_INTERPCONCAT;
2354 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2355 && SvEVALED(PL_lex_repl))
2357 if (PL_bufptr != PL_bufend)
2358 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2359 PL_lex_repl = Nullsv;
2362 case LEX_INTERPCONCAT:
2364 if (PL_lex_brackets)
2365 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2367 if (PL_bufptr == PL_bufend)
2368 return sublex_done();
2370 if (SvIVX(PL_linestr) == '\'') {
2371 SV *sv = newSVsv(PL_linestr);
2374 else if ( PL_hints & HINT_NEW_RE )
2375 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2376 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2380 s = scan_const(PL_bufptr);
2382 PL_lex_state = LEX_INTERPCASEMOD;
2384 PL_lex_state = LEX_INTERPSTART;
2387 if (s != PL_bufptr) {
2388 PL_nextval[PL_nexttoke] = yylval;
2391 if (PL_lex_starts++)
2401 PL_lex_state = LEX_NORMAL;
2402 s = scan_formline(PL_bufptr);
2403 if (!PL_lex_formbrack)
2409 PL_oldoldbufptr = PL_oldbufptr;
2412 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2413 exp_name[PL_expect], s);
2419 if (isIDFIRST_lazy_if(s,UTF))
2421 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2424 goto fake_eof; /* emulate EOF on ^D or ^Z */
2429 if (PL_lex_brackets)
2430 yyerror("Missing right curly or square bracket");
2431 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2432 "### Tokener got EOF\n");
2436 if (s++ < PL_bufend)
2437 goto retry; /* ignore stray nulls */
2440 if (!PL_in_eval && !PL_preambled) {
2441 PL_preambled = TRUE;
2442 sv_setpv(PL_linestr,incl_perldb());
2443 if (SvCUR(PL_linestr))
2444 sv_catpv(PL_linestr,";");
2446 while(AvFILLp(PL_preambleav) >= 0) {
2447 SV *tmpsv = av_shift(PL_preambleav);
2448 sv_catsv(PL_linestr, tmpsv);
2449 sv_catpv(PL_linestr, ";");
2452 sv_free((SV*)PL_preambleav);
2453 PL_preambleav = NULL;
2455 if (PL_minus_n || PL_minus_p) {
2456 sv_catpv(PL_linestr, "LINE: while (<>) {");
2458 sv_catpv(PL_linestr,"chomp;");
2461 if (strchr("/'\"", *PL_splitstr)
2462 && strchr(PL_splitstr + 1, *PL_splitstr))
2463 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
2466 s = "'~#\200\1'"; /* surely one char is unused...*/
2467 while (s[1] && strchr(PL_splitstr, *s)) s++;
2469 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s%c",
2470 "q" + (delim == '\''), delim);
2471 for (s = PL_splitstr; *s; s++) {
2473 sv_catpvn(PL_linestr, "\\", 1);
2474 sv_catpvn(PL_linestr, s, 1);
2476 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2480 sv_catpv(PL_linestr,"our @F=split(' ');");
2483 sv_catpv(PL_linestr, "\n");
2484 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2485 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2486 PL_last_lop = PL_last_uni = Nullch;
2487 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2488 SV *sv = NEWSV(85,0);
2490 sv_upgrade(sv, SVt_PVMG);
2491 sv_setsv(sv,PL_linestr);
2494 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2499 bof = PL_rsfp ? TRUE : FALSE;
2500 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2503 if (PL_preprocess && !PL_in_eval)
2504 (void)PerlProc_pclose(PL_rsfp);
2505 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2506 PerlIO_clearerr(PL_rsfp);
2508 (void)PerlIO_close(PL_rsfp);
2510 PL_doextract = FALSE;
2512 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2513 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2514 sv_catpv(PL_linestr,";}");
2515 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2516 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2517 PL_last_lop = PL_last_uni = Nullch;
2518 PL_minus_n = PL_minus_p = 0;
2521 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2522 PL_last_lop = PL_last_uni = Nullch;
2523 sv_setpv(PL_linestr,"");
2524 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2526 /* if it looks like the start of a BOM, check if it in fact is */
2527 else if (bof && (!*s || *(U8*)s == 0xEF || *(U8*)s >= 0xFE)) {
2528 #ifdef PERLIO_IS_STDIO
2529 # ifdef __GNU_LIBRARY__
2530 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
2531 # define FTELL_FOR_PIPE_IS_BROKEN
2535 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2536 # define FTELL_FOR_PIPE_IS_BROKEN
2541 #ifdef FTELL_FOR_PIPE_IS_BROKEN
2542 /* This loses the possibility to detect the bof
2543 * situation on perl -P when the libc5 is being used.
2544 * Workaround? Maybe attach some extra state to PL_rsfp?
2547 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2549 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2552 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2553 s = swallow_bom((U8*)s);
2557 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2558 PL_doextract = FALSE;
2560 /* Incest with pod. */
2561 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2562 sv_setpv(PL_linestr, "");
2563 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2564 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2565 PL_last_lop = PL_last_uni = Nullch;
2566 PL_doextract = FALSE;
2570 } while (PL_doextract);
2571 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2572 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2573 SV *sv = NEWSV(85,0);
2575 sv_upgrade(sv, SVt_PVMG);
2576 sv_setsv(sv,PL_linestr);
2579 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2581 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2582 PL_last_lop = PL_last_uni = Nullch;
2583 if (CopLINE(PL_curcop) == 1) {
2584 while (s < PL_bufend && isSPACE(*s))
2586 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2590 if (*s == '#' && *(s+1) == '!')
2592 #ifdef ALTERNATE_SHEBANG
2594 static char as[] = ALTERNATE_SHEBANG;
2595 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2596 d = s + (sizeof(as) - 1);
2598 #endif /* ALTERNATE_SHEBANG */
2607 while (*d && !isSPACE(*d))
2611 #ifdef ARG_ZERO_IS_SCRIPT
2612 if (ipathend > ipath) {
2614 * HP-UX (at least) sets argv[0] to the script name,
2615 * which makes $^X incorrect. And Digital UNIX and Linux,
2616 * at least, set argv[0] to the basename of the Perl
2617 * interpreter. So, having found "#!", we'll set it right.
2619 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV)); /* $^X */
2620 assert(SvPOK(x) || SvGMAGICAL(x));
2621 if (sv_eq(x, CopFILESV(PL_curcop))) {
2622 sv_setpvn(x, ipath, ipathend - ipath);
2625 TAINT_NOT; /* $^X is always tainted, but that's OK */
2627 #endif /* ARG_ZERO_IS_SCRIPT */
2632 d = instr(s,"perl -");
2634 d = instr(s,"perl");
2636 /* avoid getting into infinite loops when shebang
2637 * line contains "Perl" rather than "perl" */
2639 for (d = ipathend-4; d >= ipath; --d) {
2640 if ((*d == 'p' || *d == 'P')
2641 && !ibcmp(d, "perl", 4))
2651 #ifdef ALTERNATE_SHEBANG
2653 * If the ALTERNATE_SHEBANG on this system starts with a
2654 * character that can be part of a Perl expression, then if
2655 * we see it but not "perl", we're probably looking at the
2656 * start of Perl code, not a request to hand off to some
2657 * other interpreter. Similarly, if "perl" is there, but
2658 * not in the first 'word' of the line, we assume the line
2659 * contains the start of the Perl program.
2661 if (d && *s != '#') {
2663 while (*c && !strchr("; \t\r\n\f\v#", *c))
2666 d = Nullch; /* "perl" not in first word; ignore */
2668 *s = '#'; /* Don't try to parse shebang line */
2670 #endif /* ALTERNATE_SHEBANG */
2671 #ifndef MACOS_TRADITIONAL
2676 !instr(s,"indir") &&
2677 instr(PL_origargv[0],"perl"))
2683 while (s < PL_bufend && isSPACE(*s))
2685 if (s < PL_bufend) {
2686 Newz(899,newargv,PL_origargc+3,char*);
2688 while (s < PL_bufend && !isSPACE(*s))
2691 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2694 newargv = PL_origargv;
2696 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
2697 Perl_croak(aTHX_ "Can't exec %s", ipath);
2701 U32 oldpdb = PL_perldb;
2702 bool oldn = PL_minus_n;
2703 bool oldp = PL_minus_p;
2705 while (*d && !isSPACE(*d)) d++;
2706 while (SPACE_OR_TAB(*d)) d++;
2709 bool switches_done = PL_doswitches;
2711 if (*d == 'M' || *d == 'm') {
2713 while (*d && !isSPACE(*d)) d++;
2714 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2717 d = moreswitches(d);
2719 if ((PERLDB_LINE && !oldpdb) ||
2720 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2721 /* if we have already added "LINE: while (<>) {",
2722 we must not do it again */
2724 sv_setpv(PL_linestr, "");
2725 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2726 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2727 PL_last_lop = PL_last_uni = Nullch;
2728 PL_preambled = FALSE;
2730 (void)gv_fetchfile(PL_origfilename);
2733 if (PL_doswitches && !switches_done) {
2734 int argc = PL_origargc;
2735 char **argv = PL_origargv;
2738 } while (argc && argv[0][0] == '-' && argv[0][1]);
2739 init_argv_symbols(argc,argv);
2745 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2747 PL_lex_state = LEX_FORMLINE;
2752 #ifdef PERL_STRICT_CR
2753 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2755 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2757 case ' ': case '\t': case '\f': case 013:
2758 #ifdef MACOS_TRADITIONAL
2765 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2766 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2767 /* handle eval qq[#line 1 "foo"\n ...] */
2768 CopLINE_dec(PL_curcop);
2772 while (s < d && *s != '\n')
2776 else if (s > d) /* Found by Ilya: feed random input to Perl. */
2777 Perl_croak(aTHX_ "panic: input overflow");
2779 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2781 PL_lex_state = LEX_FORMLINE;
2791 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2798 while (s < PL_bufend && SPACE_OR_TAB(*s))
2801 if (strnEQ(s,"=>",2)) {
2802 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2803 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2804 "### Saw unary minus before =>, forcing word '%s'\n", s);
2806 OPERATOR('-'); /* unary minus */
2808 PL_last_uni = PL_oldbufptr;
2810 case 'r': ftst = OP_FTEREAD; break;
2811 case 'w': ftst = OP_FTEWRITE; break;
2812 case 'x': ftst = OP_FTEEXEC; break;
2813 case 'o': ftst = OP_FTEOWNED; break;
2814 case 'R': ftst = OP_FTRREAD; break;
2815 case 'W': ftst = OP_FTRWRITE; break;
2816 case 'X': ftst = OP_FTREXEC; break;
2817 case 'O': ftst = OP_FTROWNED; break;
2818 case 'e': ftst = OP_FTIS; break;
2819 case 'z': ftst = OP_FTZERO; break;
2820 case 's': ftst = OP_FTSIZE; break;
2821 case 'f': ftst = OP_FTFILE; break;
2822 case 'd': ftst = OP_FTDIR; break;
2823 case 'l': ftst = OP_FTLINK; break;
2824 case 'p': ftst = OP_FTPIPE; break;
2825 case 'S': ftst = OP_FTSOCK; break;
2826 case 'u': ftst = OP_FTSUID; break;
2827 case 'g': ftst = OP_FTSGID; break;
2828 case 'k': ftst = OP_FTSVTX; break;
2829 case 'b': ftst = OP_FTBLK; break;
2830 case 'c': ftst = OP_FTCHR; break;
2831 case 't': ftst = OP_FTTTY; break;
2832 case 'T': ftst = OP_FTTEXT; break;
2833 case 'B': ftst = OP_FTBINARY; break;
2834 case 'M': case 'A': case 'C':
2835 gv_fetchpv("\024",TRUE, SVt_PV);
2837 case 'M': ftst = OP_FTMTIME; break;
2838 case 'A': ftst = OP_FTATIME; break;
2839 case 'C': ftst = OP_FTCTIME; break;
2847 PL_last_lop_op = ftst;
2848 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2849 "### Saw file test %c\n", (int)ftst);
2854 /* Assume it was a minus followed by a one-letter named
2855 * subroutine call (or a -bareword), then. */
2856 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2857 "### %c looked like a file test but was not\n",
2866 if (PL_expect == XOPERATOR)
2871 else if (*s == '>') {
2874 if (isIDFIRST_lazy_if(s,UTF)) {
2875 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2883 if (PL_expect == XOPERATOR)
2886 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2888 OPERATOR('-'); /* unary minus */
2895 if (PL_expect == XOPERATOR)
2900 if (PL_expect == XOPERATOR)
2903 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2909 if (PL_expect != XOPERATOR) {
2910 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2911 PL_expect = XOPERATOR;
2912 force_ident(PL_tokenbuf, '*');
2925 if (PL_expect == XOPERATOR) {
2929 PL_tokenbuf[0] = '%';
2930 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2931 if (!PL_tokenbuf[1]) {
2933 yyerror("Final % should be \\% or %name");
2936 PL_pending_ident = '%';
2955 switch (PL_expect) {
2958 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2960 PL_bufptr = s; /* update in case we back off */
2966 PL_expect = XTERMBLOCK;
2970 while (isIDFIRST_lazy_if(s,UTF)) {
2971 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2972 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2973 if (tmp < 0) tmp = -tmp;
2988 d = scan_str(d,TRUE,TRUE);
2990 /* MUST advance bufptr here to avoid bogus
2991 "at end of line" context messages from yyerror().
2993 PL_bufptr = s + len;
2994 yyerror("Unterminated attribute parameter in attribute list");
2997 return 0; /* EOF indicator */
3001 SV *sv = newSVpvn(s, len);
3002 sv_catsv(sv, PL_lex_stuff);
3003 attrs = append_elem(OP_LIST, attrs,
3004 newSVOP(OP_CONST, 0, sv));
3005 SvREFCNT_dec(PL_lex_stuff);
3006 PL_lex_stuff = Nullsv;
3009 if (!PL_in_my && len == 6 && strnEQ(s, "lvalue", len))
3010 CvLVALUE_on(PL_compcv);
3011 else if (!PL_in_my && len == 6 && strnEQ(s, "locked", len))
3012 CvLOCKED_on(PL_compcv);
3013 else if (!PL_in_my && len == 6 && strnEQ(s, "method", len))
3014 CvMETHOD_on(PL_compcv);
3016 else if (PL_in_my == KEY_our && len == 6 && strnEQ(s, "unique", len))
3017 GvUNIQUE_on(cGVOPx_gv(yylval.opval));
3019 /* After we've set the flags, it could be argued that
3020 we don't need to do the attributes.pm-based setting
3021 process, and shouldn't bother appending recognized
3022 flags. To experiment with that, uncomment the
3023 following "else": */
3025 attrs = append_elem(OP_LIST, attrs,
3026 newSVOP(OP_CONST, 0,
3030 if (*s == ':' && s[1] != ':')
3033 break; /* require real whitespace or :'s */
3035 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3036 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
3037 char q = ((*s == '\'') ? '"' : '\'');
3038 /* If here for an expression, and parsed no attrs, back off. */
3039 if (tmp == '=' && !attrs) {
3043 /* MUST advance bufptr here to avoid bogus "at end of line"
3044 context messages from yyerror().
3048 yyerror("Unterminated attribute list");
3050 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
3058 PL_nextval[PL_nexttoke].opval = attrs;
3066 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3067 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3083 if (PL_lex_brackets <= 0)
3084 yyerror("Unmatched right square bracket");
3087 if (PL_lex_state == LEX_INTERPNORMAL) {
3088 if (PL_lex_brackets == 0) {
3089 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3090 PL_lex_state = LEX_INTERPEND;
3097 if (PL_lex_brackets > 100) {
3098 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
3099 if (newlb != PL_lex_brackstack) {
3101 PL_lex_brackstack = newlb;
3104 switch (PL_expect) {
3106 if (PL_lex_formbrack) {
3110 if (PL_oldoldbufptr == PL_last_lop)
3111 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3113 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3114 OPERATOR(HASHBRACK);
3116 while (s < PL_bufend && SPACE_OR_TAB(*s))
3119 PL_tokenbuf[0] = '\0';
3120 if (d < PL_bufend && *d == '-') {
3121 PL_tokenbuf[0] = '-';
3123 while (d < PL_bufend && SPACE_OR_TAB(*d))
3126 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3127 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3129 while (d < PL_bufend && SPACE_OR_TAB(*d))
3132 char minus = (PL_tokenbuf[0] == '-');
3133 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3141 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3146 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3151 if (PL_oldoldbufptr == PL_last_lop)
3152 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3154 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3157 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
3159 /* This hack is to get the ${} in the message. */
3161 yyerror("syntax error");
3164 OPERATOR(HASHBRACK);
3166 /* This hack serves to disambiguate a pair of curlies
3167 * as being a block or an anon hash. Normally, expectation
3168 * determines that, but in cases where we're not in a
3169 * position to expect anything in particular (like inside
3170 * eval"") we have to resolve the ambiguity. This code
3171 * covers the case where the first term in the curlies is a
3172 * quoted string. Most other cases need to be explicitly
3173 * disambiguated by prepending a `+' before the opening
3174 * curly in order to force resolution as an anon hash.
3176 * XXX should probably propagate the outer expectation
3177 * into eval"" to rely less on this hack, but that could
3178 * potentially break current behavior of eval"".
3182 if (*s == '\'' || *s == '"' || *s == '`') {
3183 /* common case: get past first string, handling escapes */
3184 for (t++; t < PL_bufend && *t != *s;)
3185 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3189 else if (*s == 'q') {
3192 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3196 char open, close, term;
3199 while (t < PL_bufend && isSPACE(*t))
3203 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3207 for (t++; t < PL_bufend; t++) {
3208 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3210 else if (*t == open)
3214 for (t++; t < PL_bufend; t++) {
3215 if (*t == '\\' && t+1 < PL_bufend)
3217 else if (*t == close && --brackets <= 0)
3219 else if (*t == open)
3225 else if (isALNUM_lazy_if(t,UTF)) {
3227 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3230 while (t < PL_bufend && isSPACE(*t))
3232 /* if comma follows first term, call it an anon hash */
3233 /* XXX it could be a comma expression with loop modifiers */
3234 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3235 || (*t == '=' && t[1] == '>')))
3236 OPERATOR(HASHBRACK);
3237 if (PL_expect == XREF)
3240 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3246 yylval.ival = CopLINE(PL_curcop);
3247 if (isSPACE(*s) || *s == '#')
3248 PL_copline = NOLINE; /* invalidate current command line number */
3253 if (PL_lex_brackets <= 0)
3254 yyerror("Unmatched right curly bracket");
3256 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3257 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3258 PL_lex_formbrack = 0;
3259 if (PL_lex_state == LEX_INTERPNORMAL) {
3260 if (PL_lex_brackets == 0) {
3261 if (PL_expect & XFAKEBRACK) {
3262 PL_expect &= XENUMMASK;
3263 PL_lex_state = LEX_INTERPEND;
3265 return yylex(); /* ignore fake brackets */
3267 if (*s == '-' && s[1] == '>')
3268 PL_lex_state = LEX_INTERPENDMAYBE;
3269 else if (*s != '[' && *s != '{')
3270 PL_lex_state = LEX_INTERPEND;
3273 if (PL_expect & XFAKEBRACK) {
3274 PL_expect &= XENUMMASK;
3276 return yylex(); /* ignore fake brackets */
3286 if (PL_expect == XOPERATOR) {
3287 if (ckWARN(WARN_SEMICOLON)
3288 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3290 CopLINE_dec(PL_curcop);
3291 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3292 CopLINE_inc(PL_curcop);
3297 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3299 PL_expect = XOPERATOR;
3300 force_ident(PL_tokenbuf, '&');
3304 yylval.ival = (OPpENTERSUB_AMPER<<8);
3323 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3324 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3326 if (PL_expect == XSTATE && isALPHA(tmp) &&
3327 (s == PL_linestart+1 || s[-2] == '\n') )
3329 if (PL_in_eval && !PL_rsfp) {
3334 if (strnEQ(s,"=cut",4)) {
3348 PL_doextract = TRUE;
3351 if (PL_lex_brackets < PL_lex_formbrack) {
3353 #ifdef PERL_STRICT_CR
3354 for (t = s; SPACE_OR_TAB(*t); t++) ;
3356 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3358 if (*t == '\n' || *t == '#') {
3376 if (PL_expect != XOPERATOR) {
3377 if (s[1] != '<' && !strchr(s,'>'))
3380 s = scan_heredoc(s);
3382 s = scan_inputsymbol(s);
3383 TERM(sublex_start());
3388 SHop(OP_LEFT_SHIFT);
3402 SHop(OP_RIGHT_SHIFT);
3411 if (PL_expect == XOPERATOR) {
3412 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3415 return ','; /* grandfather non-comma-format format */
3419 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3420 PL_tokenbuf[0] = '@';
3421 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3422 sizeof PL_tokenbuf - 1, FALSE);
3423 if (PL_expect == XOPERATOR)
3424 no_op("Array length", s);
3425 if (!PL_tokenbuf[1])
3427 PL_expect = XOPERATOR;
3428 PL_pending_ident = '#';
3432 PL_tokenbuf[0] = '$';
3433 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3434 sizeof PL_tokenbuf - 1, FALSE);
3435 if (PL_expect == XOPERATOR)
3437 if (!PL_tokenbuf[1]) {
3439 yyerror("Final $ should be \\$ or $name");
3443 /* This kludge not intended to be bulletproof. */
3444 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3445 yylval.opval = newSVOP(OP_CONST, 0,
3446 newSViv(PL_compiling.cop_arybase));
3447 yylval.opval->op_private = OPpCONST_ARYBASE;
3453 if (PL_lex_state == LEX_NORMAL)
3456 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3459 PL_tokenbuf[0] = '@';
3460 if (ckWARN(WARN_SYNTAX)) {
3462 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3465 PL_bufptr = skipspace(PL_bufptr);
3466 while (t < PL_bufend && *t != ']')
3468 Perl_warner(aTHX_ WARN_SYNTAX,
3469 "Multidimensional syntax %.*s not supported",
3470 (t - PL_bufptr) + 1, PL_bufptr);
3474 else if (*s == '{') {
3475 PL_tokenbuf[0] = '%';
3476 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3477 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3479 char tmpbuf[sizeof PL_tokenbuf];
3481 for (t++; isSPACE(*t); t++) ;
3482 if (isIDFIRST_lazy_if(t,UTF)) {
3483 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3484 for (; isSPACE(*t); t++) ;
3485 if (*t == ';' && get_cv(tmpbuf, FALSE))
3486 Perl_warner(aTHX_ WARN_SYNTAX,
3487 "You need to quote \"%s\"", tmpbuf);
3493 PL_expect = XOPERATOR;
3494 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3495 bool islop = (PL_last_lop == PL_oldoldbufptr);
3496 if (!islop || PL_last_lop_op == OP_GREPSTART)
3497 PL_expect = XOPERATOR;
3498 else if (strchr("$@\"'`q", *s))
3499 PL_expect = XTERM; /* e.g. print $fh "foo" */
3500 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3501 PL_expect = XTERM; /* e.g. print $fh &sub */
3502 else if (isIDFIRST_lazy_if(s,UTF)) {
3503 char tmpbuf[sizeof PL_tokenbuf];
3504 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3505 if ((tmp = keyword(tmpbuf, len))) {
3506 /* binary operators exclude handle interpretations */
3518 PL_expect = XTERM; /* e.g. print $fh length() */
3523 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3524 if (gv && GvCVu(gv))
3525 PL_expect = XTERM; /* e.g. print $fh subr() */
3528 else if (isDIGIT(*s))
3529 PL_expect = XTERM; /* e.g. print $fh 3 */
3530 else if (*s == '.' && isDIGIT(s[1]))
3531 PL_expect = XTERM; /* e.g. print $fh .3 */
3532 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3533 PL_expect = XTERM; /* e.g. print $fh -1 */
3534 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3535 PL_expect = XTERM; /* print $fh <<"EOF" */
3537 PL_pending_ident = '$';
3541 if (PL_expect == XOPERATOR)
3543 PL_tokenbuf[0] = '@';
3544 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3545 if (!PL_tokenbuf[1]) {
3547 yyerror("Final @ should be \\@ or @name");
3550 if (PL_lex_state == LEX_NORMAL)
3552 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3554 PL_tokenbuf[0] = '%';
3556 /* Warn about @ where they meant $. */
3557 if (ckWARN(WARN_SYNTAX)) {
3558 if (*s == '[' || *s == '{') {
3560 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3562 if (*t == '}' || *t == ']') {
3564 PL_bufptr = skipspace(PL_bufptr);
3565 Perl_warner(aTHX_ WARN_SYNTAX,
3566 "Scalar value %.*s better written as $%.*s",
3567 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3572 PL_pending_ident = '@';
3575 case '/': /* may either be division or pattern */
3576 case '?': /* may either be conditional or pattern */
3577 if (PL_expect != XOPERATOR) {
3578 /* Disable warning on "study /blah/" */
3579 if (PL_oldoldbufptr == PL_last_uni
3580 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3581 || memNE(PL_last_uni, "study", 5)
3582 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3584 s = scan_pat(s,OP_MATCH);
3585 TERM(sublex_start());
3593 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3594 #ifdef PERL_STRICT_CR
3597 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3599 && (s == PL_linestart || s[-1] == '\n') )
3601 PL_lex_formbrack = 0;
3605 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3611 yylval.ival = OPf_SPECIAL;
3617 if (PL_expect != XOPERATOR)
3622 case '0': case '1': case '2': case '3': case '4':
3623 case '5': case '6': case '7': case '8': case '9':
3624 s = scan_num(s, &yylval);
3625 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3626 "### Saw number in '%s'\n", s);
3628 if (PL_expect == XOPERATOR)
3633 s = scan_str(s,FALSE,FALSE);
3634 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3635 "### Saw string before '%s'\n", s);
3637 if (PL_expect == XOPERATOR) {
3638 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3641 return ','; /* grandfather non-comma-format format */
3647 missingterm((char*)0);
3648 yylval.ival = OP_CONST;
3649 TERM(sublex_start());
3652 s = scan_str(s,FALSE,FALSE);
3653 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3654 "### Saw string before '%s'\n", s);
3656 if (PL_expect == XOPERATOR) {
3657 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3660 return ','; /* grandfather non-comma-format format */
3666 missingterm((char*)0);
3667 yylval.ival = OP_CONST;
3668 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3669 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
3670 yylval.ival = OP_STRINGIFY;
3674 TERM(sublex_start());
3677 s = scan_str(s,FALSE,FALSE);
3678 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3679 "### Saw backtick string before '%s'\n", s);
3681 if (PL_expect == XOPERATOR)
3682 no_op("Backticks",s);
3684 missingterm((char*)0);
3685 yylval.ival = OP_BACKTICK;
3687 TERM(sublex_start());
3691 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3692 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3694 if (PL_expect == XOPERATOR)
3695 no_op("Backslash",s);
3699 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3703 while (isDIGIT(*start) || *start == '_')
3705 if (*start == '.' && isDIGIT(start[1])) {
3706 s = scan_num(s, &yylval);
3709 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3710 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF || PL_expect == XSTATE)) {
3714 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3717 s = scan_num(s, &yylval);
3724 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3763 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3765 /* Some keywords can be followed by any delimiter, including ':' */
3766 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3767 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3768 (PL_tokenbuf[0] == 'q' &&
3769 strchr("qwxr", PL_tokenbuf[1])))));
3771 /* x::* is just a word, unless x is "CORE" */
3772 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3776 while (d < PL_bufend && isSPACE(*d))
3777 d++; /* no comments skipped here, or s### is misparsed */
3779 /* Is this a label? */
3780 if (!tmp && PL_expect == XSTATE
3781 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3783 yylval.pval = savepv(PL_tokenbuf);
3788 /* Check for keywords */
3789 tmp = keyword(PL_tokenbuf, len);
3791 /* Is this a word before a => operator? */
3792 if (*d == '=' && d[1] == '>') {
3794 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3795 yylval.opval->op_private = OPpCONST_BARE;
3796 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
3797 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3801 if (tmp < 0) { /* second-class keyword? */
3802 GV *ogv = Nullgv; /* override (winner) */
3803 GV *hgv = Nullgv; /* hidden (loser) */
3804 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3806 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3809 if (GvIMPORTED_CV(gv))
3811 else if (! CvMETHOD(cv))
3815 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3816 (gv = *gvp) != (GV*)&PL_sv_undef &&
3817 GvCVu(gv) && GvIMPORTED_CV(gv))
3823 tmp = 0; /* overridden by import or by GLOBAL */
3826 && -tmp==KEY_lock /* XXX generalizable kludge */
3828 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3830 tmp = 0; /* any sub overrides "weak" keyword */
3832 else { /* no override */
3834 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
3835 Perl_warner(aTHX_ WARN_MISC,
3836 "dump() better written as CORE::dump()");
3840 if (ckWARN(WARN_AMBIGUOUS) && hgv
3841 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3842 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3843 "Ambiguous call resolved as CORE::%s(), %s",
3844 GvENAME(hgv), "qualify as such or use &");
3851 default: /* not a keyword */
3855 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3857 /* Get the rest if it looks like a package qualifier */
3859 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3861 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3864 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3865 *s == '\'' ? "'" : "::");
3870 if (PL_expect == XOPERATOR) {
3871 if (PL_bufptr == PL_linestart) {
3872 CopLINE_dec(PL_curcop);
3873 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3874 CopLINE_inc(PL_curcop);
3877 no_op("Bareword",s);
3880 /* Look for a subroutine with this name in current package,
3881 unless name is "Foo::", in which case Foo is a bearword
3882 (and a package name). */
3885 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3887 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3888 Perl_warner(aTHX_ WARN_BAREWORD,
3889 "Bareword \"%s\" refers to nonexistent package",
3892 PL_tokenbuf[len] = '\0';
3899 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3902 /* if we saw a global override before, get the right name */
3905 sv = newSVpvn("CORE::GLOBAL::",14);
3906 sv_catpv(sv,PL_tokenbuf);
3909 sv = newSVpv(PL_tokenbuf,0);
3911 /* Presume this is going to be a bareword of some sort. */
3914 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3915 yylval.opval->op_private = OPpCONST_BARE;
3917 /* And if "Foo::", then that's what it certainly is. */
3922 /* See if it's the indirect object for a list operator. */
3924 if (PL_oldoldbufptr &&
3925 PL_oldoldbufptr < PL_bufptr &&
3926 (PL_oldoldbufptr == PL_last_lop
3927 || PL_oldoldbufptr == PL_last_uni) &&
3928 /* NO SKIPSPACE BEFORE HERE! */
3929 (PL_expect == XREF ||
3930 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3932 bool immediate_paren = *s == '(';
3934 /* (Now we can afford to cross potential line boundary.) */
3937 /* Two barewords in a row may indicate method call. */
3939 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3942 /* If not a declared subroutine, it's an indirect object. */
3943 /* (But it's an indir obj regardless for sort.) */
3945 if ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
3946 ((!gv || !GvCVu(gv)) &&
3947 (PL_last_lop_op != OP_MAPSTART &&
3948 PL_last_lop_op != OP_GREPSTART))))
3950 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3955 PL_expect = XOPERATOR;
3958 /* Is this a word before a => operator? */
3959 if (*s == '=' && s[1] == '>' && !pkgname) {
3961 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
3962 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
3963 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3967 /* If followed by a paren, it's certainly a subroutine. */
3970 if (gv && GvCVu(gv)) {
3971 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
3972 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3977 PL_nextval[PL_nexttoke].opval = yylval.opval;
3978 PL_expect = XOPERATOR;
3984 /* If followed by var or block, call it a method (unless sub) */
3986 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3987 PL_last_lop = PL_oldbufptr;
3988 PL_last_lop_op = OP_METHOD;
3992 /* If followed by a bareword, see if it looks like indir obj. */
3994 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3997 /* Not a method, so call it a subroutine (if defined) */
3999 if (gv && GvCVu(gv)) {
4001 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
4002 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4003 "Ambiguous use of -%s resolved as -&%s()",
4004 PL_tokenbuf, PL_tokenbuf);
4005 /* Check for a constant sub */
4007 if ((sv = cv_const_sv(cv))) {
4009 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
4010 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
4011 yylval.opval->op_private = 0;
4015 /* Resolve to GV now. */
4016 op_free(yylval.opval);
4017 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
4018 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
4019 PL_last_lop = PL_oldbufptr;
4020 PL_last_lop_op = OP_ENTERSUB;
4021 /* Is there a prototype? */
4024 char *proto = SvPV((SV*)cv, len);
4027 if (strEQ(proto, "$"))
4029 if (*proto == '&' && *s == '{') {
4030 sv_setpv(PL_subname,"__ANON__");
4034 PL_nextval[PL_nexttoke].opval = yylval.opval;
4040 /* Call it a bare word */
4042 if (PL_hints & HINT_STRICT_SUBS)
4043 yylval.opval->op_private |= OPpCONST_STRICT;
4046 if (ckWARN(WARN_RESERVED)) {
4047 if (lastchar != '-') {
4048 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4049 if (!*d && strNE(PL_tokenbuf,"main"))
4050 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
4057 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
4058 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4059 "Operator or semicolon missing before %c%s",
4060 lastchar, PL_tokenbuf);
4061 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4062 "Ambiguous use of %c resolved as operator %c",
4063 lastchar, lastchar);
4069 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4070 newSVpv(CopFILE(PL_curcop),0));
4074 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4075 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4078 case KEY___PACKAGE__:
4079 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4081 ? newSVsv(PL_curstname)
4090 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4091 char *pname = "main";
4092 if (PL_tokenbuf[2] == 'D')
4093 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
4094 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4097 GvIOp(gv) = newIO();
4098 IoIFP(GvIOp(gv)) = PL_rsfp;
4099 #if defined(HAS_FCNTL) && defined(F_SETFD)
4101 int fd = PerlIO_fileno(PL_rsfp);
4102 fcntl(fd,F_SETFD,fd >= 3);
4105 /* Mark this internal pseudo-handle as clean */
4106 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4108 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4109 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4110 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4112 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4113 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4114 /* if the script was opened in binmode, we need to revert
4115 * it to text mode for compatibility; but only iff it has CRs
4116 * XXX this is a questionable hack at best. */
4117 if (PL_bufend-PL_bufptr > 2
4118 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4121 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4122 loc = PerlIO_tell(PL_rsfp);
4123 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4126 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
4128 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4129 #endif /* NETWARE */
4130 #ifdef PERLIO_IS_STDIO /* really? */
4131 # if defined(__BORLANDC__)
4132 /* XXX see note in do_binmode() */
4133 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
4137 PerlIO_seek(PL_rsfp, loc, 0);
4141 #ifdef PERLIO_LAYERS
4142 if (UTF && !IN_BYTES)
4143 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
4156 if (PL_expect == XSTATE) {
4163 if (*s == ':' && s[1] == ':') {
4166 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4167 if (!(tmp = keyword(PL_tokenbuf, len)))
4168 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4182 LOP(OP_ACCEPT,XTERM);
4188 LOP(OP_ATAN2,XTERM);
4194 LOP(OP_BINMODE,XTERM);
4197 LOP(OP_BLESS,XTERM);
4206 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4223 if (!PL_cryptseen) {
4224 PL_cryptseen = TRUE;
4228 LOP(OP_CRYPT,XTERM);
4231 LOP(OP_CHMOD,XTERM);
4234 LOP(OP_CHOWN,XTERM);
4237 LOP(OP_CONNECT,XTERM);
4253 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4257 PL_hints |= HINT_BLOCK_SCOPE;
4267 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4268 LOP(OP_DBMOPEN,XTERM);
4274 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4281 yylval.ival = CopLINE(PL_curcop);
4295 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4296 UNIBRACK(OP_ENTEREVAL);
4311 case KEY_endhostent:
4317 case KEY_endservent:
4320 case KEY_endprotoent:
4331 yylval.ival = CopLINE(PL_curcop);
4333 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4335 if ((PL_bufend - p) >= 3 &&
4336 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4338 else if ((PL_bufend - p) >= 4 &&
4339 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4342 if (isIDFIRST_lazy_if(p,UTF)) {
4343 p = scan_ident(p, PL_bufend,
4344 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4348 Perl_croak(aTHX_ "Missing $ on loop variable");
4353 LOP(OP_FORMLINE,XTERM);
4359 LOP(OP_FCNTL,XTERM);
4365 LOP(OP_FLOCK,XTERM);
4374 LOP(OP_GREPSTART, XREF);
4377 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4392 case KEY_getpriority:
4393 LOP(OP_GETPRIORITY,XTERM);
4395 case KEY_getprotobyname:
4398 case KEY_getprotobynumber:
4399 LOP(OP_GPBYNUMBER,XTERM);
4401 case KEY_getprotoent:
4413 case KEY_getpeername:
4414 UNI(OP_GETPEERNAME);
4416 case KEY_gethostbyname:
4419 case KEY_gethostbyaddr:
4420 LOP(OP_GHBYADDR,XTERM);
4422 case KEY_gethostent:
4425 case KEY_getnetbyname:
4428 case KEY_getnetbyaddr:
4429 LOP(OP_GNBYADDR,XTERM);
4434 case KEY_getservbyname:
4435 LOP(OP_GSBYNAME,XTERM);
4437 case KEY_getservbyport:
4438 LOP(OP_GSBYPORT,XTERM);
4440 case KEY_getservent:
4443 case KEY_getsockname:
4444 UNI(OP_GETSOCKNAME);
4446 case KEY_getsockopt:
4447 LOP(OP_GSOCKOPT,XTERM);
4469 yylval.ival = CopLINE(PL_curcop);
4473 LOP(OP_INDEX,XTERM);
4479 LOP(OP_IOCTL,XTERM);
4491 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4523 LOP(OP_LISTEN,XTERM);
4532 s = scan_pat(s,OP_MATCH);
4533 TERM(sublex_start());
4536 LOP(OP_MAPSTART, XREF);
4539 LOP(OP_MKDIR,XTERM);
4542 LOP(OP_MSGCTL,XTERM);
4545 LOP(OP_MSGGET,XTERM);
4548 LOP(OP_MSGRCV,XTERM);
4551 LOP(OP_MSGSND,XTERM);
4557 if (isIDFIRST_lazy_if(s,UTF)) {
4558 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4559 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4561 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4562 if (!PL_in_my_stash) {
4565 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4573 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4580 if (PL_expect != XSTATE)
4581 yyerror("\"no\" not allowed in expression");
4582 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4583 s = force_version(s, FALSE);
4588 if (*s == '(' || (s = skipspace(s), *s == '('))
4595 if (isIDFIRST_lazy_if(s,UTF)) {
4597 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4599 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4600 Perl_warner(aTHX_ WARN_PRECEDENCE,
4601 "Precedence problem: open %.*s should be open(%.*s)",
4607 yylval.ival = OP_OR;
4617 LOP(OP_OPEN_DIR,XTERM);
4620 checkcomma(s,PL_tokenbuf,"filehandle");
4624 checkcomma(s,PL_tokenbuf,"filehandle");
4643 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4647 LOP(OP_PIPE_OP,XTERM);
4650 s = scan_str(s,FALSE,FALSE);
4652 missingterm((char*)0);
4653 yylval.ival = OP_CONST;
4654 TERM(sublex_start());
4660 s = scan_str(s,FALSE,FALSE);
4662 missingterm((char*)0);
4664 if (SvCUR(PL_lex_stuff)) {
4667 d = SvPV_force(PL_lex_stuff, len);
4670 for (; isSPACE(*d) && len; --len, ++d) ;
4673 if (!warned && ckWARN(WARN_QW)) {
4674 for (; !isSPACE(*d) && len; --len, ++d) {
4676 Perl_warner(aTHX_ WARN_QW,
4677 "Possible attempt to separate words with commas");
4680 else if (*d == '#') {
4681 Perl_warner(aTHX_ WARN_QW,
4682 "Possible attempt to put comments in qw() list");
4688 for (; !isSPACE(*d) && len; --len, ++d) ;
4690 sv = newSVpvn(b, d-b);
4691 if (DO_UTF8(PL_lex_stuff))
4693 words = append_elem(OP_LIST, words,
4694 newSVOP(OP_CONST, 0, tokeq(sv)));
4698 PL_nextval[PL_nexttoke].opval = words;
4703 SvREFCNT_dec(PL_lex_stuff);
4704 PL_lex_stuff = Nullsv;
4710 s = scan_str(s,FALSE,FALSE);
4712 missingterm((char*)0);
4713 yylval.ival = OP_STRINGIFY;
4714 if (SvIVX(PL_lex_stuff) == '\'')
4715 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4716 TERM(sublex_start());
4719 s = scan_pat(s,OP_QR);
4720 TERM(sublex_start());
4723 s = scan_str(s,FALSE,FALSE);
4725 missingterm((char*)0);
4726 yylval.ival = OP_BACKTICK;
4728 TERM(sublex_start());
4736 s = force_version(s, FALSE);
4738 else if (*s != 'v' || !isDIGIT(s[1])
4739 || (s = force_version(s, TRUE), *s == 'v'))
4741 *PL_tokenbuf = '\0';
4742 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4743 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4744 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4746 yyerror("<> should be quotes");
4754 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4758 LOP(OP_RENAME,XTERM);
4767 LOP(OP_RINDEX,XTERM);
4790 LOP(OP_REVERSE,XTERM);
4801 TERM(sublex_start());
4803 TOKEN(1); /* force error */
4812 LOP(OP_SELECT,XTERM);
4818 LOP(OP_SEMCTL,XTERM);
4821 LOP(OP_SEMGET,XTERM);
4824 LOP(OP_SEMOP,XTERM);
4830 LOP(OP_SETPGRP,XTERM);
4832 case KEY_setpriority:
4833 LOP(OP_SETPRIORITY,XTERM);
4835 case KEY_sethostent:
4841 case KEY_setservent:
4844 case KEY_setprotoent:
4854 LOP(OP_SEEKDIR,XTERM);
4856 case KEY_setsockopt:
4857 LOP(OP_SSOCKOPT,XTERM);
4863 LOP(OP_SHMCTL,XTERM);
4866 LOP(OP_SHMGET,XTERM);
4869 LOP(OP_SHMREAD,XTERM);
4872 LOP(OP_SHMWRITE,XTERM);
4875 LOP(OP_SHUTDOWN,XTERM);
4884 LOP(OP_SOCKET,XTERM);
4886 case KEY_socketpair:
4887 LOP(OP_SOCKPAIR,XTERM);
4890 checkcomma(s,PL_tokenbuf,"subroutine name");
4892 if (*s == ';' || *s == ')') /* probably a close */
4893 Perl_croak(aTHX_ "sort is now a reserved word");
4895 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4899 LOP(OP_SPLIT,XTERM);
4902 LOP(OP_SPRINTF,XTERM);
4905 LOP(OP_SPLICE,XTERM);
4920 LOP(OP_SUBSTR,XTERM);
4926 char tmpbuf[sizeof PL_tokenbuf];
4927 SSize_t tboffset = 0;
4928 expectation attrful;
4929 bool have_name, have_proto, bad_proto;
4934 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4935 (*s == ':' && s[1] == ':'))
4938 attrful = XATTRBLOCK;
4939 /* remember buffer pos'n for later force_word */
4940 tboffset = s - PL_oldbufptr;
4941 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4942 if (strchr(tmpbuf, ':'))
4943 sv_setpv(PL_subname, tmpbuf);
4945 sv_setsv(PL_subname,PL_curstname);
4946 sv_catpvn(PL_subname,"::",2);
4947 sv_catpvn(PL_subname,tmpbuf,len);
4954 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4955 PL_expect = XTERMBLOCK;
4956 attrful = XATTRTERM;
4957 sv_setpv(PL_subname,"?");
4961 if (key == KEY_format) {
4963 PL_lex_formbrack = PL_lex_brackets + 1;
4965 (void) force_word(PL_oldbufptr + tboffset, WORD,
4970 /* Look for a prototype */
4974 s = scan_str(s,FALSE,FALSE);
4976 Perl_croak(aTHX_ "Prototype not terminated");
4977 /* strip spaces and check for bad characters */
4978 d = SvPVX(PL_lex_stuff);
4981 for (p = d; *p; ++p) {
4984 if (!strchr("$@%*;[]&\\", *p))
4989 if (bad_proto && ckWARN(WARN_SYNTAX))
4990 Perl_warner(aTHX_ WARN_SYNTAX,
4991 "Illegal character in prototype for %s : %s",
4992 SvPVX(PL_subname), d);
4993 SvCUR(PL_lex_stuff) = tmp;
5001 if (*s == ':' && s[1] != ':')
5002 PL_expect = attrful;
5005 PL_nextval[PL_nexttoke].opval =
5006 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
5007 PL_lex_stuff = Nullsv;
5011 sv_setpv(PL_subname,"__ANON__");
5014 (void) force_word(PL_oldbufptr + tboffset, WORD,
5023 LOP(OP_SYSTEM,XREF);
5026 LOP(OP_SYMLINK,XTERM);
5029 LOP(OP_SYSCALL,XTERM);
5032 LOP(OP_SYSOPEN,XTERM);
5035 LOP(OP_SYSSEEK,XTERM);
5038 LOP(OP_SYSREAD,XTERM);
5041 LOP(OP_SYSWRITE,XTERM);
5045 TERM(sublex_start());
5066 LOP(OP_TRUNCATE,XTERM);
5078 yylval.ival = CopLINE(PL_curcop);
5082 yylval.ival = CopLINE(PL_curcop);
5086 LOP(OP_UNLINK,XTERM);
5092 LOP(OP_UNPACK,XTERM);
5095 LOP(OP_UTIME,XTERM);
5101 LOP(OP_UNSHIFT,XTERM);
5104 if (PL_expect != XSTATE)
5105 yyerror("\"use\" not allowed in expression");
5107 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
5108 s = force_version(s, TRUE);
5109 if (*s == ';' || (s = skipspace(s), *s == ';')) {
5110 PL_nextval[PL_nexttoke].opval = Nullop;
5113 else if (*s == 'v') {
5114 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5115 s = force_version(s, FALSE);
5119 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5120 s = force_version(s, FALSE);
5132 yylval.ival = CopLINE(PL_curcop);
5136 PL_hints |= HINT_BLOCK_SCOPE;
5143 LOP(OP_WAITPID,XTERM);
5152 ctl_l[0] = toCTRL('L');
5154 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5157 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5162 if (PL_expect == XOPERATOR)
5168 yylval.ival = OP_XOR;
5173 TERM(sublex_start());
5178 #pragma segment Main
5182 S_pending_ident(pTHX)
5186 /* pit holds the identifier we read and pending_ident is reset */
5187 char pit = PL_pending_ident;
5188 PL_pending_ident = 0;
5190 DEBUG_T({ PerlIO_printf(Perl_debug_log,
5191 "### Tokener saw identifier '%s'\n", PL_tokenbuf); });
5193 /* if we're in a my(), we can't allow dynamics here.
5194 $foo'bar has already been turned into $foo::bar, so
5195 just check for colons.
5197 if it's a legal name, the OP is a PADANY.
5200 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
5201 if (strchr(PL_tokenbuf,':'))
5202 yyerror(Perl_form(aTHX_ "No package name allowed for "
5203 "variable %s in \"our\"",
5205 tmp = pad_allocmy(PL_tokenbuf);
5208 if (strchr(PL_tokenbuf,':'))
5209 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
5211 yylval.opval = newOP(OP_PADANY, 0);
5212 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
5218 build the ops for accesses to a my() variable.
5220 Deny my($a) or my($b) in a sort block, *if* $a or $b is
5221 then used in a comparison. This catches most, but not
5222 all cases. For instance, it catches
5223 sort { my($a); $a <=> $b }
5225 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
5226 (although why you'd do that is anyone's guess).
5229 if (!strchr(PL_tokenbuf,':')) {
5230 #ifdef USE_5005THREADS
5231 /* Check for single character per-thread SVs */
5232 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
5233 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
5234 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
5236 yylval.opval = newOP(OP_THREADSV, 0);
5237 yylval.opval->op_targ = tmp;
5240 #endif /* USE_5005THREADS */
5241 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
5242 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
5243 /* might be an "our" variable" */
5244 if (SvFLAGS(namesv) & SVpad_OUR) {
5245 /* build ops for a bareword */
5246 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
5247 sv_catpvn(sym, "::", 2);
5248 sv_catpv(sym, PL_tokenbuf+1);
5249 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
5250 yylval.opval->op_private = OPpCONST_ENTERED;
5251 gv_fetchpv(SvPVX(sym),
5253 ? (GV_ADDMULTI | GV_ADDINEVAL)
5256 ((PL_tokenbuf[0] == '$') ? SVt_PV
5257 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5262 /* if it's a sort block and they're naming $a or $b */
5263 if (PL_last_lop_op == OP_SORT &&
5264 PL_tokenbuf[0] == '$' &&
5265 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
5268 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
5269 d < PL_bufend && *d != '\n';
5272 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
5273 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
5279 yylval.opval = newOP(OP_PADANY, 0);
5280 yylval.opval->op_targ = tmp;
5286 Whine if they've said @foo in a doublequoted string,
5287 and @foo isn't a variable we can find in the symbol
5290 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
5291 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
5292 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
5293 && ckWARN(WARN_AMBIGUOUS))
5295 /* Downgraded from fatal to warning 20000522 mjd */
5296 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5297 "Possible unintended interpolation of %s in string",
5302 /* build ops for a bareword */
5303 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
5304 yylval.opval->op_private = OPpCONST_ENTERED;
5305 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
5306 ((PL_tokenbuf[0] == '$') ? SVt_PV
5307 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5313 Perl_keyword(pTHX_ register char *d, I32 len)
5318 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5319 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5320 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5321 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5322 if (strEQ(d,"__END__")) return KEY___END__;
5326 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5331 if (strEQ(d,"and")) return -KEY_and;
5332 if (strEQ(d,"abs")) return -KEY_abs;
5335 if (strEQ(d,"alarm")) return -KEY_alarm;
5336 if (strEQ(d,"atan2")) return -KEY_atan2;
5339 if (strEQ(d,"accept")) return -KEY_accept;
5344 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5347 if (strEQ(d,"bless")) return -KEY_bless;
5348 if (strEQ(d,"bind")) return -KEY_bind;
5349 if (strEQ(d,"binmode")) return -KEY_binmode;
5352 if (strEQ(d,"CORE")) return -KEY_CORE;
5353 if (strEQ(d,"CHECK")) return KEY_CHECK;
5358 if (strEQ(d,"cmp")) return -KEY_cmp;
5359 if (strEQ(d,"chr")) return -KEY_chr;
5360 if (strEQ(d,"cos")) return -KEY_cos;
5363 if (strEQ(d,"chop")) return -KEY_chop;
5366 if (strEQ(d,"close")) return -KEY_close;
5367 if (strEQ(d,"chdir")) return -KEY_chdir;
5368 if (strEQ(d,"chomp")) return -KEY_chomp;
5369 if (strEQ(d,"chmod")) return -KEY_chmod;
5370 if (strEQ(d,"chown")) return -KEY_chown;
5371 if (strEQ(d,"crypt")) return -KEY_crypt;
5374 if (strEQ(d,"chroot")) return -KEY_chroot;
5375 if (strEQ(d,"caller")) return -KEY_caller;
5378 if (strEQ(d,"connect")) return -KEY_connect;
5381 if (strEQ(d,"closedir")) return -KEY_closedir;
5382 if (strEQ(d,"continue")) return -KEY_continue;
5387 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5392 if (strEQ(d,"do")) return KEY_do;
5395 if (strEQ(d,"die")) return -KEY_die;
5398 if (strEQ(d,"dump")) return -KEY_dump;
5401 if (strEQ(d,"delete")) return KEY_delete;
5404 if (strEQ(d,"defined")) return KEY_defined;
5405 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5408 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5413 if (strEQ(d,"END")) return KEY_END;
5418 if (strEQ(d,"eq")) return -KEY_eq;
5421 if (strEQ(d,"eof")) return -KEY_eof;
5422 if (strEQ(d,"exp")) return -KEY_exp;
5425 if (strEQ(d,"else")) return KEY_else;
5426 if (strEQ(d,"exit")) return -KEY_exit;
5427 if (strEQ(d,"eval")) return KEY_eval;
5428 if (strEQ(d,"exec")) return -KEY_exec;
5429 if (strEQ(d,"each")) return -KEY_each;
5432 if (strEQ(d,"elsif")) return KEY_elsif;
5435 if (strEQ(d,"exists")) return KEY_exists;
5436 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5439 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5440 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5443 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5446 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5447 if (strEQ(d,"endservent")) return -KEY_endservent;
5450 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5457 if (strEQ(d,"for")) return KEY_for;
5460 if (strEQ(d,"fork")) return -KEY_fork;
5463 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5464 if (strEQ(d,"flock")) return -KEY_flock;
5467 if (strEQ(d,"format")) return KEY_format;
5468 if (strEQ(d,"fileno")) return -KEY_fileno;
5471 if (strEQ(d,"foreach")) return KEY_foreach;
5474 if (strEQ(d,"formline")) return -KEY_formline;
5479 if (strnEQ(d,"get",3)) {
5484 if (strEQ(d,"ppid")) return -KEY_getppid;
5485 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5488 if (strEQ(d,"pwent")) return -KEY_getpwent;
5489 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5490 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5493 if (strEQ(d,"peername")) return -KEY_getpeername;
5494 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5495 if (strEQ(d,"priority")) return -KEY_getpriority;
5498 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5501 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5505 else if (*d == 'h') {
5506 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5507 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5508 if (strEQ(d,"hostent")) return -KEY_gethostent;
5510 else if (*d == 'n') {
5511 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5512 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5513 if (strEQ(d,"netent")) return -KEY_getnetent;
5515 else if (*d == 's') {
5516 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5517 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5518 if (strEQ(d,"servent")) return -KEY_getservent;
5519 if (strEQ(d,"sockname")) return -KEY_getsockname;
5520 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5522 else if (*d == 'g') {
5523 if (strEQ(d,"grent")) return -KEY_getgrent;
5524 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5525 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5527 else if (*d == 'l') {
5528 if (strEQ(d,"login")) return -KEY_getlogin;
5530 else if (strEQ(d,"c")) return -KEY_getc;
5535 if (strEQ(d,"gt")) return -KEY_gt;
5536 if (strEQ(d,"ge")) return -KEY_ge;
5539 if (strEQ(d,"grep")) return KEY_grep;
5540 if (strEQ(d,"goto")) return KEY_goto;
5541 if (strEQ(d,"glob")) return KEY_glob;
5544 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5549 if (strEQ(d,"hex")) return -KEY_hex;
5552 if (strEQ(d,"INIT")) return KEY_INIT;
5557 if (strEQ(d,"if")) return KEY_if;
5560 if (strEQ(d,"int")) return -KEY_int;
5563 if (strEQ(d,"index")) return -KEY_index;
5564 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5569 if (strEQ(d,"join")) return -KEY_join;
5573 if (strEQ(d,"keys")) return -KEY_keys;
5574 if (strEQ(d,"kill")) return -KEY_kill;
5580 if (strEQ(d,"lt")) return -KEY_lt;
5581 if (strEQ(d,"le")) return -KEY_le;
5582 if (strEQ(d,"lc")) return -KEY_lc;
5585 if (strEQ(d,"log")) return -KEY_log;
5588 if (strEQ(d,"last")) return KEY_last;
5589 if (strEQ(d,"link")) return -KEY_link;
5590 if (strEQ(d,"lock")) return -KEY_lock;
5593 if (strEQ(d,"local")) return KEY_local;
5594 if (strEQ(d,"lstat")) return -KEY_lstat;
5597 if (strEQ(d,"length")) return -KEY_length;
5598 if (strEQ(d,"listen")) return -KEY_listen;
5601 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5604 if (strEQ(d,"localtime")) return -KEY_localtime;
5610 case 1: return KEY_m;
5612 if (strEQ(d,"my")) return KEY_my;
5615 if (strEQ(d,"map")) return KEY_map;
5618 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5621 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5622 if (strEQ(d,"msgget")) return -KEY_msgget;
5623 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5624 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5629 if (strEQ(d,"next")) return KEY_next;
5630 if (strEQ(d,"ne")) return -KEY_ne;
5631 if (strEQ(d,"not")) return -KEY_not;
5632 if (strEQ(d,"no")) return KEY_no;
5637 if (strEQ(d,"or")) return -KEY_or;
5640 if (strEQ(d,"ord")) return -KEY_ord;
5641 if (strEQ(d,"oct")) return -KEY_oct;
5642 if (strEQ(d,"our")) return KEY_our;
5645 if (strEQ(d,"open")) return -KEY_open;
5648 if (strEQ(d,"opendir")) return -KEY_opendir;
5655 if (strEQ(d,"pop")) return -KEY_pop;
5656 if (strEQ(d,"pos")) return KEY_pos;
5659 if (strEQ(d,"push")) return -KEY_push;
5660 if (strEQ(d,"pack")) return -KEY_pack;
5661 if (strEQ(d,"pipe")) return -KEY_pipe;
5664 if (strEQ(d,"print")) return KEY_print;
5667 if (strEQ(d,"printf")) return KEY_printf;
5670 if (strEQ(d,"package")) return KEY_package;
5673 if (strEQ(d,"prototype")) return KEY_prototype;
5678 if (strEQ(d,"q")) return KEY_q;
5679 if (strEQ(d,"qr")) return KEY_qr;
5680 if (strEQ(d,"qq")) return KEY_qq;
5681 if (strEQ(d,"qw")) return KEY_qw;
5682 if (strEQ(d,"qx")) return KEY_qx;
5684 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5689 if (strEQ(d,"ref")) return -KEY_ref;
5692 if (strEQ(d,"read")) return -KEY_read;
5693 if (strEQ(d,"rand")) return -KEY_rand;
5694 if (strEQ(d,"recv")) return -KEY_recv;
5695 if (strEQ(d,"redo")) return KEY_redo;
5698 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5699 if (strEQ(d,"reset")) return -KEY_reset;
5702 if (strEQ(d,"return")) return KEY_return;
5703 if (strEQ(d,"rename")) return -KEY_rename;
5704 if (strEQ(d,"rindex")) return -KEY_rindex;
5707 if (strEQ(d,"require")) return KEY_require;
5708 if (strEQ(d,"reverse")) return -KEY_reverse;
5709 if (strEQ(d,"readdir")) return -KEY_readdir;
5712 if (strEQ(d,"readlink")) return -KEY_readlink;
5713 if (strEQ(d,"readline")) return -KEY_readline;
5714 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5717 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5723 case 0: return KEY_s;
5725 if (strEQ(d,"scalar")) return KEY_scalar;
5730 if (strEQ(d,"seek")) return -KEY_seek;
5731 if (strEQ(d,"send")) return -KEY_send;
5734 if (strEQ(d,"semop")) return -KEY_semop;
5737 if (strEQ(d,"select")) return -KEY_select;
5738 if (strEQ(d,"semctl")) return -KEY_semctl;
5739 if (strEQ(d,"semget")) return -KEY_semget;
5742 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5743 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5746 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5747 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5750 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5753 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5754 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5755 if (strEQ(d,"setservent")) return -KEY_setservent;
5758 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5759 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5766 if (strEQ(d,"shift")) return -KEY_shift;
5769 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5770 if (strEQ(d,"shmget")) return -KEY_shmget;
5773 if (strEQ(d,"shmread")) return -KEY_shmread;
5776 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5777 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5782 if (strEQ(d,"sin")) return -KEY_sin;
5785 if (strEQ(d,"sleep")) return -KEY_sleep;
5788 if (strEQ(d,"sort")) return KEY_sort;
5789 if (strEQ(d,"socket")) return -KEY_socket;
5790 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5793 if (strEQ(d,"split")) return KEY_split;
5794 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5795 if (strEQ(d,"splice")) return -KEY_splice;
5798 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5801 if (strEQ(d,"srand")) return -KEY_srand;
5804 if (strEQ(d,"stat")) return -KEY_stat;
5805 if (strEQ(d,"study")) return KEY_study;
5808 if (strEQ(d,"substr")) return -KEY_substr;
5809 if (strEQ(d,"sub")) return KEY_sub;
5814 if (strEQ(d,"system")) return -KEY_system;
5817 if (strEQ(d,"symlink")) return -KEY_symlink;
5818 if (strEQ(d,"syscall")) return -KEY_syscall;
5819 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5820 if (strEQ(d,"sysread")) return -KEY_sysread;
5821 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5824 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5833 if (strEQ(d,"tr")) return KEY_tr;
5836 if (strEQ(d,"tie")) return KEY_tie;
5839 if (strEQ(d,"tell")) return -KEY_tell;
5840 if (strEQ(d,"tied")) return KEY_tied;
5841 if (strEQ(d,"time")) return -KEY_time;
5844 if (strEQ(d,"times")) return -KEY_times;
5847 if (strEQ(d,"telldir")) return -KEY_telldir;
5850 if (strEQ(d,"truncate")) return -KEY_truncate;
5857 if (strEQ(d,"uc")) return -KEY_uc;
5860 if (strEQ(d,"use")) return KEY_use;
5863 if (strEQ(d,"undef")) return KEY_undef;
5864 if (strEQ(d,"until")) return KEY_until;
5865 if (strEQ(d,"untie")) return KEY_untie;
5866 if (strEQ(d,"utime")) return -KEY_utime;
5867 if (strEQ(d,"umask")) return -KEY_umask;
5870 if (strEQ(d,"unless")) return KEY_unless;
5871 if (strEQ(d,"unpack")) return -KEY_unpack;
5872 if (strEQ(d,"unlink")) return -KEY_unlink;
5875 if (strEQ(d,"unshift")) return -KEY_unshift;
5876 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5881 if (strEQ(d,"values")) return -KEY_values;
5882 if (strEQ(d,"vec")) return -KEY_vec;
5887 if (strEQ(d,"warn")) return -KEY_warn;
5888 if (strEQ(d,"wait")) return -KEY_wait;
5891 if (strEQ(d,"while")) return KEY_while;
5892 if (strEQ(d,"write")) return -KEY_write;
5895 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5898 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5903 if (len == 1) return -KEY_x;
5904 if (strEQ(d,"xor")) return -KEY_xor;
5907 if (len == 1) return KEY_y;
5916 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5920 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5921 if (ckWARN(WARN_SYNTAX)) {
5923 for (w = s+2; *w && level; w++) {
5930 for (; *w && isSPACE(*w); w++) ;
5931 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5932 Perl_warner(aTHX_ WARN_SYNTAX,
5933 "%s (...) interpreted as function",name);
5936 while (s < PL_bufend && isSPACE(*s))
5940 while (s < PL_bufend && isSPACE(*s))
5942 if (isIDFIRST_lazy_if(s,UTF)) {
5944 while (isALNUM_lazy_if(s,UTF))
5946 while (s < PL_bufend && isSPACE(*s))
5951 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5955 Perl_croak(aTHX_ "No comma allowed after %s", what);
5960 /* Either returns sv, or mortalizes sv and returns a new SV*.
5961 Best used as sv=new_constant(..., sv, ...).
5962 If s, pv are NULL, calls subroutine with one argument,
5963 and type is used with error messages only. */
5966 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5970 HV *table = GvHV(PL_hintgv); /* ^H */
5974 const char *why1, *why2, *why3;
5976 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
5979 why2 = strEQ(key,"charnames")
5980 ? "(possibly a missing \"use charnames ...\")"
5982 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
5983 (type ? type: "undef"), why2);
5985 /* This is convoluted and evil ("goto considered harmful")
5986 * but I do not understand the intricacies of all the different
5987 * failure modes of %^H in here. The goal here is to make
5988 * the most probable error message user-friendly. --jhi */
5993 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
5994 (type ? type: "undef"), why1, why2, why3);
5996 yyerror(SvPVX(msg));
6000 cvp = hv_fetch(table, key, strlen(key), FALSE);
6001 if (!cvp || !SvOK(*cvp)) {
6004 why3 = "} is not defined";
6007 sv_2mortal(sv); /* Parent created it permanently */
6010 pv = sv_2mortal(newSVpvn(s, len));
6012 typesv = sv_2mortal(newSVpv(type, 0));
6014 typesv = &PL_sv_undef;
6016 PUSHSTACKi(PERLSI_OVERLOAD);
6028 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
6032 /* Check the eval first */
6033 if (!PL_in_eval && SvTRUE(ERRSV)) {
6035 sv_catpv(ERRSV, "Propagated");
6036 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
6038 res = SvREFCNT_inc(sv);
6042 (void)SvREFCNT_inc(res);
6051 why1 = "Call to &{$^H{";
6053 why3 = "}} did not return a defined value";
6062 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
6064 register char *d = dest;
6065 register char *e = d + destlen - 3; /* two-character token, ending NUL */
6068 Perl_croak(aTHX_ ident_too_long);
6069 if (isALNUM(*s)) /* UTF handled below */
6071 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
6076 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
6080 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6081 char *t = s + UTF8SKIP(s);
6082 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6084 if (d + (t - s) > e)
6085 Perl_croak(aTHX_ ident_too_long);
6086 Copy(s, d, t - s, char);
6099 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
6109 e = d + destlen - 3; /* two-character token, ending NUL */
6111 while (isDIGIT(*s)) {
6113 Perl_croak(aTHX_ ident_too_long);
6120 Perl_croak(aTHX_ ident_too_long);
6121 if (isALNUM(*s)) /* UTF handled below */
6123 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
6128 else if (*s == ':' && s[1] == ':') {
6132 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6133 char *t = s + UTF8SKIP(s);
6134 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6136 if (d + (t - s) > e)
6137 Perl_croak(aTHX_ ident_too_long);
6138 Copy(s, d, t - s, char);
6149 if (PL_lex_state != LEX_NORMAL)
6150 PL_lex_state = LEX_INTERPENDMAYBE;
6153 if (*s == '$' && s[1] &&
6154 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
6167 if (*d == '^' && *s && isCONTROLVAR(*s)) {
6172 if (isSPACE(s[-1])) {
6175 if (!SPACE_OR_TAB(ch)) {
6181 if (isIDFIRST_lazy_if(d,UTF)) {
6185 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
6187 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
6190 Copy(s, d, e - s, char);
6195 while ((isALNUM(*s) || *s == ':') && d < e)
6198 Perl_croak(aTHX_ ident_too_long);
6201 while (s < send && SPACE_OR_TAB(*s)) s++;
6202 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
6203 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
6204 const char *brack = *s == '[' ? "[...]" : "{...}";
6205 Perl_warner(aTHX_ WARN_AMBIGUOUS,
6206 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
6207 funny, dest, brack, funny, dest, brack);
6210 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
6214 /* Handle extended ${^Foo} variables
6215 * 1999-02-27 mjd-perl-patch@plover.com */
6216 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
6220 while (isALNUM(*s) && d < e) {
6224 Perl_croak(aTHX_ ident_too_long);
6229 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
6230 PL_lex_state = LEX_INTERPEND;
6233 if (PL_lex_state == LEX_NORMAL) {
6234 if (ckWARN(WARN_AMBIGUOUS) &&
6235 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
6237 Perl_warner(aTHX_ WARN_AMBIGUOUS,
6238 "Ambiguous use of %c{%s} resolved to %c%s",
6239 funny, dest, funny, dest);
6244 s = bracket; /* let the parser handle it */
6248 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
6249 PL_lex_state = LEX_INTERPEND;
6254 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
6259 *pmfl |= PMf_GLOBAL;
6261 *pmfl |= PMf_CONTINUE;
6265 *pmfl |= PMf_MULTILINE;
6267 *pmfl |= PMf_SINGLELINE;
6269 *pmfl |= PMf_EXTENDED;
6273 S_scan_pat(pTHX_ char *start, I32 type)
6278 s = scan_str(start,FALSE,FALSE);
6280 Perl_croak(aTHX_ "Search pattern not terminated");
6282 pm = (PMOP*)newPMOP(type, 0);
6283 if (PL_multi_open == '?')
6284 pm->op_pmflags |= PMf_ONCE;
6286 while (*s && strchr("iomsx", *s))
6287 pmflag(&pm->op_pmflags,*s++);
6290 while (*s && strchr("iogcmsx", *s))
6291 pmflag(&pm->op_pmflags,*s++);
6293 pm->op_pmpermflags = pm->op_pmflags;
6295 PL_lex_op = (OP*)pm;
6296 yylval.ival = OP_MATCH;
6301 S_scan_subst(pTHX_ char *start)
6308 yylval.ival = OP_NULL;
6310 s = scan_str(start,FALSE,FALSE);
6313 Perl_croak(aTHX_ "Substitution pattern not terminated");
6315 if (s[-1] == PL_multi_open)
6318 first_start = PL_multi_start;
6319 s = scan_str(s,FALSE,FALSE);
6322 SvREFCNT_dec(PL_lex_stuff);
6323 PL_lex_stuff = Nullsv;
6325 Perl_croak(aTHX_ "Substitution replacement not terminated");
6327 PL_multi_start = first_start; /* so whole substitution is taken together */
6329 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6335 else if (strchr("iogcmsx", *s))
6336 pmflag(&pm->op_pmflags,*s++);
6343 PL_sublex_info.super_bufptr = s;
6344 PL_sublex_info.super_bufend = PL_bufend;
6346 pm->op_pmflags |= PMf_EVAL;
6347 repl = newSVpvn("",0);
6349 sv_catpv(repl, es ? "eval " : "do ");
6350 sv_catpvn(repl, "{ ", 2);
6351 sv_catsv(repl, PL_lex_repl);
6352 sv_catpvn(repl, " };", 2);
6354 SvREFCNT_dec(PL_lex_repl);
6358 pm->op_pmpermflags = pm->op_pmflags;
6359 PL_lex_op = (OP*)pm;
6360 yylval.ival = OP_SUBST;
6365 S_scan_trans(pTHX_ char *start)
6374 yylval.ival = OP_NULL;
6376 s = scan_str(start,FALSE,FALSE);
6378 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6379 if (s[-1] == PL_multi_open)
6382 s = scan_str(s,FALSE,FALSE);
6385 SvREFCNT_dec(PL_lex_stuff);
6386 PL_lex_stuff = Nullsv;
6388 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6391 complement = del = squash = 0;
6392 while (strchr("cds", *s)) {
6394 complement = OPpTRANS_COMPLEMENT;
6396 del = OPpTRANS_DELETE;
6398 squash = OPpTRANS_SQUASH;
6402 New(803, tbl, complement&&!del?258:256, short);
6403 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6404 o->op_private = del|squash|complement|
6405 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
6406 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
6409 yylval.ival = OP_TRANS;
6414 S_scan_heredoc(pTHX_ register char *s)
6417 I32 op_type = OP_SCALAR;
6424 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6428 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6431 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6432 if (*peek && strchr("`'\"",*peek)) {
6435 s = delimcpy(d, e, s, PL_bufend, term, &len);
6445 if (!isALNUM_lazy_if(s,UTF))
6446 deprecate("bare << to mean <<\"\"");
6447 for (; isALNUM_lazy_if(s,UTF); s++) {
6452 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6453 Perl_croak(aTHX_ "Delimiter for here document is too long");
6456 len = d - PL_tokenbuf;
6457 #ifndef PERL_STRICT_CR
6458 d = strchr(s, '\r');
6462 while (s < PL_bufend) {
6468 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6477 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6482 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6483 herewas = newSVpvn(s,PL_bufend-s);
6485 s--, herewas = newSVpvn(s,d-s);
6486 s += SvCUR(herewas);
6488 tmpstr = NEWSV(87,79);
6489 sv_upgrade(tmpstr, SVt_PVIV);
6494 else if (term == '`') {
6495 op_type = OP_BACKTICK;
6496 SvIVX(tmpstr) = '\\';
6500 PL_multi_start = CopLINE(PL_curcop);
6501 PL_multi_open = PL_multi_close = '<';
6502 term = *PL_tokenbuf;
6503 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6504 char *bufptr = PL_sublex_info.super_bufptr;
6505 char *bufend = PL_sublex_info.super_bufend;
6506 char *olds = s - SvCUR(herewas);
6507 s = strchr(bufptr, '\n');
6511 while (s < bufend &&
6512 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6514 CopLINE_inc(PL_curcop);
6517 CopLINE_set(PL_curcop, PL_multi_start);
6518 missingterm(PL_tokenbuf);
6520 sv_setpvn(herewas,bufptr,d-bufptr+1);
6521 sv_setpvn(tmpstr,d+1,s-d);
6523 sv_catpvn(herewas,s,bufend-s);
6524 (void)strcpy(bufptr,SvPVX(herewas));
6531 while (s < PL_bufend &&
6532 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6534 CopLINE_inc(PL_curcop);
6536 if (s >= PL_bufend) {
6537 CopLINE_set(PL_curcop, PL_multi_start);
6538 missingterm(PL_tokenbuf);
6540 sv_setpvn(tmpstr,d+1,s-d);
6542 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6544 sv_catpvn(herewas,s,PL_bufend-s);
6545 sv_setsv(PL_linestr,herewas);
6546 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6547 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6548 PL_last_lop = PL_last_uni = Nullch;
6551 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6552 while (s >= PL_bufend) { /* multiple line string? */
6554 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6555 CopLINE_set(PL_curcop, PL_multi_start);
6556 missingterm(PL_tokenbuf);
6558 CopLINE_inc(PL_curcop);
6559 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6560 PL_last_lop = PL_last_uni = Nullch;
6561 #ifndef PERL_STRICT_CR
6562 if (PL_bufend - PL_linestart >= 2) {
6563 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6564 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6566 PL_bufend[-2] = '\n';
6568 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6570 else if (PL_bufend[-1] == '\r')
6571 PL_bufend[-1] = '\n';
6573 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6574 PL_bufend[-1] = '\n';
6576 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6577 SV *sv = NEWSV(88,0);
6579 sv_upgrade(sv, SVt_PVMG);
6580 sv_setsv(sv,PL_linestr);
6583 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6585 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6588 sv_catsv(PL_linestr,herewas);
6589 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6593 sv_catsv(tmpstr,PL_linestr);
6598 PL_multi_end = CopLINE(PL_curcop);
6599 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6600 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6601 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6603 SvREFCNT_dec(herewas);
6604 if (UTF && !IN_BYTES && is_utf8_string((U8*)SvPVX(tmpstr), SvCUR(tmpstr)))
6606 PL_lex_stuff = tmpstr;
6607 yylval.ival = op_type;
6612 takes: current position in input buffer
6613 returns: new position in input buffer
6614 side-effects: yylval and lex_op are set.
6619 <FH> read from filehandle
6620 <pkg::FH> read from package qualified filehandle
6621 <pkg'FH> read from package qualified filehandle
6622 <$fh> read from filehandle in $fh
6628 S_scan_inputsymbol(pTHX_ char *start)
6630 register char *s = start; /* current position in buffer */
6636 d = PL_tokenbuf; /* start of temp holding space */
6637 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6638 end = strchr(s, '\n');
6641 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6643 /* die if we didn't have space for the contents of the <>,
6644 or if it didn't end, or if we see a newline
6647 if (len >= sizeof PL_tokenbuf)
6648 Perl_croak(aTHX_ "Excessively long <> operator");
6650 Perl_croak(aTHX_ "Unterminated <> operator");
6655 Remember, only scalar variables are interpreted as filehandles by
6656 this code. Anything more complex (e.g., <$fh{$num}>) will be
6657 treated as a glob() call.
6658 This code makes use of the fact that except for the $ at the front,
6659 a scalar variable and a filehandle look the same.
6661 if (*d == '$' && d[1]) d++;
6663 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6664 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6667 /* If we've tried to read what we allow filehandles to look like, and
6668 there's still text left, then it must be a glob() and not a getline.
6669 Use scan_str to pull out the stuff between the <> and treat it
6670 as nothing more than a string.
6673 if (d - PL_tokenbuf != len) {
6674 yylval.ival = OP_GLOB;
6676 s = scan_str(start,FALSE,FALSE);
6678 Perl_croak(aTHX_ "Glob not terminated");
6682 bool readline_overriden = FALSE;
6683 GV *gv_readline = Nullgv;
6685 /* we're in a filehandle read situation */
6688 /* turn <> into <ARGV> */
6690 (void)strcpy(d,"ARGV");
6692 /* Check whether readline() is overriden */
6693 if (((gv_readline = gv_fetchpv("readline", FALSE, SVt_PVCV))
6694 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
6696 ((gvp = (GV**)hv_fetch(PL_globalstash, "readline", 8, FALSE))
6697 && (gv_readline = *gvp) != (GV*)&PL_sv_undef
6698 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
6699 readline_overriden = TRUE;
6701 /* if <$fh>, create the ops to turn the variable into a
6707 /* try to find it in the pad for this block, otherwise find
6708 add symbol table ops
6710 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6711 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
6712 if (SvFLAGS(namesv) & SVpad_OUR) {
6713 SV *sym = sv_2mortal(newSVpv(HvNAME(GvSTASH(namesv)),0));
6714 sv_catpvn(sym, "::", 2);
6720 OP *o = newOP(OP_PADSV, 0);
6722 PL_lex_op = readline_overriden
6723 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6724 append_elem(OP_LIST, o,
6725 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
6726 : (OP*)newUNOP(OP_READLINE, 0, o);
6735 ? (GV_ADDMULTI | GV_ADDINEVAL)
6738 PL_lex_op = readline_overriden
6739 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6740 append_elem(OP_LIST,
6741 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
6742 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
6743 : (OP*)newUNOP(OP_READLINE, 0,
6744 newUNOP(OP_RV2SV, 0,
6745 newGVOP(OP_GV, 0, gv)));
6747 PL_lex_op->op_flags |= OPf_SPECIAL;
6748 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6749 yylval.ival = OP_NULL;
6752 /* If it's none of the above, it must be a literal filehandle
6753 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6755 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6756 PL_lex_op = readline_overriden
6757 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6758 append_elem(OP_LIST,
6759 newGVOP(OP_GV, 0, gv),
6760 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
6761 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6762 yylval.ival = OP_NULL;
6771 takes: start position in buffer
6772 keep_quoted preserve \ on the embedded delimiter(s)
6773 keep_delims preserve the delimiters around the string
6774 returns: position to continue reading from buffer
6775 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6776 updates the read buffer.
6778 This subroutine pulls a string out of the input. It is called for:
6779 q single quotes q(literal text)
6780 ' single quotes 'literal text'
6781 qq double quotes qq(interpolate $here please)
6782 " double quotes "interpolate $here please"
6783 qx backticks qx(/bin/ls -l)
6784 ` backticks `/bin/ls -l`
6785 qw quote words @EXPORT_OK = qw( func() $spam )
6786 m// regexp match m/this/
6787 s/// regexp substitute s/this/that/
6788 tr/// string transliterate tr/this/that/
6789 y/// string transliterate y/this/that/
6790 ($*@) sub prototypes sub foo ($)
6791 (stuff) sub attr parameters sub foo : attr(stuff)
6792 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6794 In most of these cases (all but <>, patterns and transliterate)
6795 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6796 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6797 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6800 It skips whitespace before the string starts, and treats the first
6801 character as the delimiter. If the delimiter is one of ([{< then
6802 the corresponding "close" character )]}> is used as the closing
6803 delimiter. It allows quoting of delimiters, and if the string has
6804 balanced delimiters ([{<>}]) it allows nesting.
6806 On success, the SV with the resulting string is put into lex_stuff or,
6807 if that is already non-NULL, into lex_repl. The second case occurs only
6808 when parsing the RHS of the special constructs s/// and tr/// (y///).
6809 For convenience, the terminating delimiter character is stuffed into
6814 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6816 SV *sv; /* scalar value: string */
6817 char *tmps; /* temp string, used for delimiter matching */
6818 register char *s = start; /* current position in the buffer */
6819 register char term; /* terminating character */
6820 register char *to; /* current position in the sv's data */
6821 I32 brackets = 1; /* bracket nesting level */
6822 bool has_utf8 = FALSE; /* is there any utf8 content? */
6824 /* skip space before the delimiter */
6828 /* mark where we are, in case we need to report errors */
6831 /* after skipping whitespace, the next character is the terminator */
6833 if (!UTF8_IS_INVARIANT((U8)term) && UTF)
6836 /* mark where we are */
6837 PL_multi_start = CopLINE(PL_curcop);
6838 PL_multi_open = term;
6840 /* find corresponding closing delimiter */
6841 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6843 PL_multi_close = term;
6845 /* create a new SV to hold the contents. 87 is leak category, I'm
6846 assuming. 79 is the SV's initial length. What a random number. */
6848 sv_upgrade(sv, SVt_PVIV);
6850 (void)SvPOK_only(sv); /* validate pointer */
6852 /* move past delimiter and try to read a complete string */
6854 sv_catpvn(sv, s, 1);
6857 /* extend sv if need be */
6858 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6859 /* set 'to' to the next character in the sv's string */
6860 to = SvPVX(sv)+SvCUR(sv);
6862 /* if open delimiter is the close delimiter read unbridle */
6863 if (PL_multi_open == PL_multi_close) {
6864 for (; s < PL_bufend; s++,to++) {
6865 /* embedded newlines increment the current line number */
6866 if (*s == '\n' && !PL_rsfp)
6867 CopLINE_inc(PL_curcop);
6868 /* handle quoted delimiters */
6869 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6870 if (!keep_quoted && s[1] == term)
6872 /* any other quotes are simply copied straight through */
6876 /* terminate when run out of buffer (the for() condition), or
6877 have found the terminator */
6878 else if (*s == term)
6880 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
6886 /* if the terminator isn't the same as the start character (e.g.,
6887 matched brackets), we have to allow more in the quoting, and
6888 be prepared for nested brackets.
6891 /* read until we run out of string, or we find the terminator */
6892 for (; s < PL_bufend; s++,to++) {
6893 /* embedded newlines increment the line count */
6894 if (*s == '\n' && !PL_rsfp)
6895 CopLINE_inc(PL_curcop);
6896 /* backslashes can escape the open or closing characters */
6897 if (*s == '\\' && s+1 < PL_bufend) {
6899 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6904 /* allow nested opens and closes */
6905 else if (*s == PL_multi_close && --brackets <= 0)
6907 else if (*s == PL_multi_open)
6909 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
6914 /* terminate the copied string and update the sv's end-of-string */
6916 SvCUR_set(sv, to - SvPVX(sv));
6919 * this next chunk reads more into the buffer if we're not done yet
6923 break; /* handle case where we are done yet :-) */
6925 #ifndef PERL_STRICT_CR
6926 if (to - SvPVX(sv) >= 2) {
6927 if ((to[-2] == '\r' && to[-1] == '\n') ||
6928 (to[-2] == '\n' && to[-1] == '\r'))
6932 SvCUR_set(sv, to - SvPVX(sv));
6934 else if (to[-1] == '\r')
6937 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6941 /* if we're out of file, or a read fails, bail and reset the current
6942 line marker so we can report where the unterminated string began
6945 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6947 CopLINE_set(PL_curcop, PL_multi_start);
6950 /* we read a line, so increment our line counter */
6951 CopLINE_inc(PL_curcop);
6953 /* update debugger info */
6954 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6955 SV *sv = NEWSV(88,0);
6957 sv_upgrade(sv, SVt_PVMG);
6958 sv_setsv(sv,PL_linestr);
6961 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6964 /* having changed the buffer, we must update PL_bufend */
6965 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6966 PL_last_lop = PL_last_uni = Nullch;
6969 /* at this point, we have successfully read the delimited string */
6972 sv_catpvn(sv, s, 1);
6975 PL_multi_end = CopLINE(PL_curcop);
6978 /* if we allocated too much space, give some back */
6979 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6980 SvLEN_set(sv, SvCUR(sv) + 1);
6981 Renew(SvPVX(sv), SvLEN(sv), char);
6984 /* decide whether this is the first or second quoted string we've read
6997 takes: pointer to position in buffer
6998 returns: pointer to new position in buffer
6999 side-effects: builds ops for the constant in yylval.op
7001 Read a number in any of the formats that Perl accepts:
7003 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
7004 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
7007 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
7009 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
7012 If it reads a number without a decimal point or an exponent, it will
7013 try converting the number to an integer and see if it can do so
7014 without loss of precision.
7018 Perl_scan_num(pTHX_ char *start, YYSTYPE* lvalp)
7020 register char *s = start; /* current position in buffer */
7021 register char *d; /* destination in temp buffer */
7022 register char *e; /* end of temp buffer */
7023 NV nv; /* number read, as a double */
7024 SV *sv = Nullsv; /* place to put the converted number */
7025 bool floatit; /* boolean: int or float? */
7026 char *lastub = 0; /* position of last underbar */
7027 static char number_too_long[] = "Number too long";
7029 /* We use the first character to decide what type of number this is */
7033 Perl_croak(aTHX_ "panic: scan_num");
7035 /* if it starts with a 0, it could be an octal number, a decimal in
7036 0.13 disguise, or a hexadecimal number, or a binary number. */
7040 u holds the "number so far"
7041 shift the power of 2 of the base
7042 (hex == 4, octal == 3, binary == 1)
7043 overflowed was the number more than we can hold?
7045 Shift is used when we add a digit. It also serves as an "are
7046 we in octal/hex/binary?" indicator to disallow hex characters
7052 bool overflowed = FALSE;
7053 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
7054 static char* bases[5] = { "", "binary", "", "octal",
7056 static char* Bases[5] = { "", "Binary", "", "Octal",
7058 static char *maxima[5] = { "",
7059 "0b11111111111111111111111111111111",
7063 char *base, *Base, *max;
7069 } else if (s[1] == 'b') {
7073 /* check for a decimal in disguise */
7074 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
7076 /* so it must be octal */
7083 if (ckWARN(WARN_SYNTAX))
7084 Perl_warner(aTHX_ WARN_SYNTAX,
7085 "Misplaced _ in number");
7089 base = bases[shift];
7090 Base = Bases[shift];
7091 max = maxima[shift];
7093 /* read the rest of the number */
7095 /* x is used in the overflow test,
7096 b is the digit we're adding on. */
7101 /* if we don't mention it, we're done */
7105 /* _ are ignored -- but warned about if consecutive */
7107 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7108 Perl_warner(aTHX_ WARN_SYNTAX,
7109 "Misplaced _ in number");
7113 /* 8 and 9 are not octal */
7116 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
7120 case '2': case '3': case '4':
7121 case '5': case '6': case '7':
7123 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
7127 b = *s++ & 15; /* ASCII digit -> value of digit */
7131 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
7132 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
7133 /* make sure they said 0x */
7138 /* Prepare to put the digit we have onto the end
7139 of the number so far. We check for overflows.
7144 x = u << shift; /* make room for the digit */
7146 if ((x >> shift) != u
7147 && !(PL_hints & HINT_NEW_BINARY)) {
7150 if (ckWARN_d(WARN_OVERFLOW))
7151 Perl_warner(aTHX_ WARN_OVERFLOW,
7152 "Integer overflow in %s number",
7155 u = x | b; /* add the digit to the end */
7158 n *= nvshift[shift];
7159 /* If an NV has not enough bits in its
7160 * mantissa to represent an UV this summing of
7161 * small low-order numbers is a waste of time
7162 * (because the NV cannot preserve the
7163 * low-order bits anyway): we could just
7164 * remember when did we overflow and in the
7165 * end just multiply n by the right
7173 /* if we get here, we had success: make a scalar value from
7178 /* final misplaced underbar check */
7180 if (ckWARN(WARN_SYNTAX))
7181 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
7186 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
7187 Perl_warner(aTHX_ WARN_PORTABLE,
7188 "%s number > %s non-portable",
7194 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
7195 Perl_warner(aTHX_ WARN_PORTABLE,
7196 "%s number > %s non-portable",
7201 if (PL_hints & HINT_NEW_BINARY)
7202 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
7207 handle decimal numbers.
7208 we're also sent here when we read a 0 as the first digit
7210 case '1': case '2': case '3': case '4': case '5':
7211 case '6': case '7': case '8': case '9': case '.':
7214 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
7217 /* read next group of digits and _ and copy into d */
7218 while (isDIGIT(*s) || *s == '_') {
7219 /* skip underscores, checking for misplaced ones
7223 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7224 Perl_warner(aTHX_ WARN_SYNTAX,
7225 "Misplaced _ in number");
7229 /* check for end of fixed-length buffer */
7231 Perl_croak(aTHX_ number_too_long);
7232 /* if we're ok, copy the character */
7237 /* final misplaced underbar check */
7238 if (lastub && s == lastub + 1) {
7239 if (ckWARN(WARN_SYNTAX))
7240 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
7243 /* read a decimal portion if there is one. avoid
7244 3..5 being interpreted as the number 3. followed
7247 if (*s == '.' && s[1] != '.') {
7252 if (ckWARN(WARN_SYNTAX))
7253 Perl_warner(aTHX_ WARN_SYNTAX,
7254 "Misplaced _ in number");
7258 /* copy, ignoring underbars, until we run out of digits.
7260 for (; isDIGIT(*s) || *s == '_'; s++) {
7261 /* fixed length buffer check */
7263 Perl_croak(aTHX_ number_too_long);
7265 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7266 Perl_warner(aTHX_ WARN_SYNTAX,
7267 "Misplaced _ in number");
7273 /* fractional part ending in underbar? */
7275 if (ckWARN(WARN_SYNTAX))
7276 Perl_warner(aTHX_ WARN_SYNTAX,
7277 "Misplaced _ in number");
7279 if (*s == '.' && isDIGIT(s[1])) {
7280 /* oops, it's really a v-string, but without the "v" */
7286 /* read exponent part, if present */
7287 if (*s && strchr("eE",*s) && strchr("+-0123456789_", s[1])) {
7291 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
7292 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
7294 /* stray preinitial _ */
7296 if (ckWARN(WARN_SYNTAX))
7297 Perl_warner(aTHX_ WARN_SYNTAX,
7298 "Misplaced _ in number");
7302 /* allow positive or negative exponent */
7303 if (*s == '+' || *s == '-')
7306 /* stray initial _ */
7308 if (ckWARN(WARN_SYNTAX))
7309 Perl_warner(aTHX_ WARN_SYNTAX,
7310 "Misplaced _ in number");
7314 /* read digits of exponent */
7315 while (isDIGIT(*s) || *s == '_') {
7318 Perl_croak(aTHX_ number_too_long);
7322 if (ckWARN(WARN_SYNTAX) &&
7323 ((lastub && s == lastub + 1) ||
7324 (!isDIGIT(s[1]) && s[1] != '_')))
7325 Perl_warner(aTHX_ WARN_SYNTAX,
7326 "Misplaced _ in number");
7333 /* make an sv from the string */
7337 We try to do an integer conversion first if no characters
7338 indicating "float" have been found.
7343 int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
7345 if (flags == IS_NUMBER_IN_UV) {
7347 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7350 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
7351 if (uv <= (UV) IV_MIN)
7352 sv_setiv(sv, -(IV)uv);
7359 /* terminate the string */
7361 nv = Atof(PL_tokenbuf);
7365 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7366 (PL_hints & HINT_NEW_INTEGER) )
7367 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7368 (floatit ? "float" : "integer"),
7372 /* if it starts with a v, it could be a v-string */
7375 sv = NEWSV(92,5); /* preallocate storage space */
7376 s = new_vstring(s,sv);
7380 /* make the op for the constant and return */
7383 lvalp->opval = newSVOP(OP_CONST, 0, sv);
7385 lvalp->opval = Nullop;
7391 S_scan_formline(pTHX_ register char *s)
7395 SV *stuff = newSVpvn("",0);
7396 bool needargs = FALSE;
7399 if (*s == '.' || *s == /*{*/'}') {
7401 #ifdef PERL_STRICT_CR
7402 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7404 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7406 if (*t == '\n' || t == PL_bufend)
7409 if (PL_in_eval && !PL_rsfp) {
7410 eol = strchr(s,'\n');
7415 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7417 for (t = s; t < eol; t++) {
7418 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7420 goto enough; /* ~~ must be first line in formline */
7422 if (*t == '@' || *t == '^')
7426 sv_catpvn(stuff, s, eol-s);
7427 #ifndef PERL_STRICT_CR
7428 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7429 char *end = SvPVX(stuff) + SvCUR(stuff);
7441 s = filter_gets(PL_linestr, PL_rsfp, 0);
7442 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7443 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7444 PL_last_lop = PL_last_uni = Nullch;
7447 yyerror("Format not terminated");
7457 PL_lex_state = LEX_NORMAL;
7458 PL_nextval[PL_nexttoke].ival = 0;
7462 PL_lex_state = LEX_FORMLINE;
7463 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7465 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7469 SvREFCNT_dec(stuff);
7470 PL_lex_formbrack = 0;
7481 PL_cshlen = strlen(PL_cshname);
7486 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7488 I32 oldsavestack_ix = PL_savestack_ix;
7489 CV* outsidecv = PL_compcv;
7493 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7495 SAVEI32(PL_subline);
7496 save_item(PL_subname);
7499 SAVESPTR(PL_comppad_name);
7500 SAVESPTR(PL_compcv);
7501 SAVEI32(PL_comppad_name_fill);
7502 SAVEI32(PL_min_intro_pending);
7503 SAVEI32(PL_max_intro_pending);
7504 SAVEI32(PL_pad_reset_pending);
7506 PL_compcv = (CV*)NEWSV(1104,0);
7507 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7508 CvFLAGS(PL_compcv) |= flags;
7510 PL_comppad = newAV();
7511 av_push(PL_comppad, Nullsv);
7512 PL_curpad = AvARRAY(PL_comppad);
7513 PL_comppad_name = newAV();
7514 PL_comppad_name_fill = 0;
7515 PL_min_intro_pending = 0;
7517 PL_subline = CopLINE(PL_curcop);
7518 #ifdef USE_5005THREADS
7519 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7520 PL_curpad[0] = (SV*)newAV();
7521 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7522 #endif /* USE_5005THREADS */
7524 comppadlist = newAV();
7525 AvREAL_off(comppadlist);
7526 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7527 av_store(comppadlist, 1, (SV*)PL_comppad);
7529 CvPADLIST(PL_compcv) = comppadlist;
7530 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7531 #ifdef USE_5005THREADS
7532 CvOWNER(PL_compcv) = 0;
7533 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7534 MUTEX_INIT(CvMUTEXP(PL_compcv));
7535 #endif /* USE_5005THREADS */
7537 return oldsavestack_ix;
7541 #pragma segment Perl_yylex
7544 Perl_yywarn(pTHX_ char *s)
7546 PL_in_eval |= EVAL_WARNONLY;
7548 PL_in_eval &= ~EVAL_WARNONLY;
7553 Perl_yyerror(pTHX_ char *s)
7556 char *context = NULL;
7560 if (!yychar || (yychar == ';' && !PL_rsfp))
7562 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7563 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7564 while (isSPACE(*PL_oldoldbufptr))
7566 context = PL_oldoldbufptr;
7567 contlen = PL_bufptr - PL_oldoldbufptr;
7569 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7570 PL_oldbufptr != PL_bufptr) {
7571 while (isSPACE(*PL_oldbufptr))
7573 context = PL_oldbufptr;
7574 contlen = PL_bufptr - PL_oldbufptr;
7576 else if (yychar > 255)
7577 where = "next token ???";
7578 #ifdef USE_PURE_BISON
7579 /* GNU Bison sets the value -2 */
7580 else if (yychar == -2) {
7582 else if ((yychar & 127) == 127) {
7584 if (PL_lex_state == LEX_NORMAL ||
7585 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7586 where = "at end of line";
7587 else if (PL_lex_inpat)
7588 where = "within pattern";
7590 where = "within string";
7593 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7595 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7596 else if (isPRINT_LC(yychar))
7597 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7599 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7600 where = SvPVX(where_sv);
7602 msg = sv_2mortal(newSVpv(s, 0));
7603 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7604 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7606 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7608 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7609 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7610 Perl_sv_catpvf(aTHX_ msg,
7611 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7612 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7615 if (PL_in_eval & EVAL_WARNONLY)
7616 Perl_warn(aTHX_ "%"SVf, msg);
7619 if (PL_error_count >= 10) {
7620 if (PL_in_eval && SvCUR(ERRSV))
7621 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
7622 ERRSV, OutCopFILE(PL_curcop));
7624 Perl_croak(aTHX_ "%s has too many errors.\n",
7625 OutCopFILE(PL_curcop));
7628 PL_in_my_stash = Nullhv;
7632 #pragma segment Main
7636 S_swallow_bom(pTHX_ U8 *s)
7639 slen = SvCUR(PL_linestr);
7643 /* UTF-16 little-endian */
7644 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
7645 Perl_croak(aTHX_ "Unsupported script encoding");
7646 #ifndef PERL_NO_UTF16_FILTER
7647 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-LE script encoding\n"));
7649 if (PL_bufend > (char*)s) {
7653 filter_add(utf16rev_textfilter, NULL);
7654 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7655 PL_bufend = (char*)utf16_to_utf8_reversed(s, news,
7656 PL_bufend - (char*)s - 1,
7658 Copy(news, s, newlen, U8);
7659 SvCUR_set(PL_linestr, newlen);
7660 PL_bufend = SvPVX(PL_linestr) + newlen;
7661 news[newlen++] = '\0';
7665 Perl_croak(aTHX_ "Unsupported script encoding");
7670 if (s[1] == 0xFF) { /* UTF-16 big-endian */
7671 #ifndef PERL_NO_UTF16_FILTER
7672 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding\n"));
7674 if (PL_bufend > (char *)s) {
7678 filter_add(utf16_textfilter, NULL);
7679 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7680 PL_bufend = (char*)utf16_to_utf8(s, news,
7681 PL_bufend - (char*)s,
7683 Copy(news, s, newlen, U8);
7684 SvCUR_set(PL_linestr, newlen);
7685 PL_bufend = SvPVX(PL_linestr) + newlen;
7686 news[newlen++] = '\0';
7690 Perl_croak(aTHX_ "Unsupported script encoding");
7695 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
7696 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-8 script encoding\n"));
7701 if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */
7702 s[2] == 0xFE && s[3] == 0xFF)
7704 Perl_croak(aTHX_ "Unsupported script encoding");
7712 * Restore a source filter.
7716 restore_rsfp(pTHX_ void *f)
7718 PerlIO *fp = (PerlIO*)f;
7720 if (PL_rsfp == PerlIO_stdin())
7721 PerlIO_clearerr(PL_rsfp);
7722 else if (PL_rsfp && (PL_rsfp != fp))
7723 PerlIO_close(PL_rsfp);
7727 #ifndef PERL_NO_UTF16_FILTER
7729 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
7731 I32 count = FILTER_READ(idx+1, sv, maxlen);
7736 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7737 if (!*SvPV_nolen(sv))
7738 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7741 tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7742 sv_usepvn(sv, (char*)tmps, tend - tmps);
7748 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
7750 I32 count = FILTER_READ(idx+1, sv, maxlen);
7755 if (!*SvPV_nolen(sv))
7756 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7759 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7760 tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7761 sv_usepvn(sv, (char*)tmps, tend - tmps);