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 if (!has_utf8 && SvUTF8(res)) {
1561 char *ostart = SvPVX(sv);
1562 SvCUR_set(sv, d - ostart);
1565 sv_utf8_upgrade(sv);
1566 /* this just broke our allocation above... */
1567 SvGROW(sv, send - start);
1568 d = SvPVX(sv) + SvCUR(sv);
1571 if (len > e - s + 4) { /* I _guess_ 4 is \N{} --jhi */
1572 char *odest = SvPVX(sv);
1574 SvGROW(sv, (SvLEN(sv) + len - (e - s + 4)));
1575 d = SvPVX(sv) + (d - odest);
1577 Copy(str, d, len, char);
1584 yyerror("Missing braces on \\N{}");
1587 /* \c is a control character */
1596 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
1600 /* printf-style backslashes, formfeeds, newlines, etc */
1602 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
1605 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
1608 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
1611 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
1614 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
1617 *d++ = ASCII_TO_NEED(has_utf8,'\033');
1620 *d++ = ASCII_TO_NEED(has_utf8,'\007');
1626 } /* end if (backslash) */
1629 /* If we started with encoded form, or already know we want it
1630 and then encode the next character */
1631 if ((has_utf8 || this_utf8) && !NATIVE_IS_INVARIANT((U8)(*s))) {
1633 UV uv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
1634 STRLEN need = UNISKIP(NATIVE_TO_UNI(uv));
1637 /* encoded value larger than old, need extra space (NOTE: SvCUR() not set here) */
1638 STRLEN off = d - SvPVX(sv);
1639 d = SvGROW(sv, SvLEN(sv) + (need-len)) + off;
1641 d = (char*)uvchr_to_utf8((U8*)d, uv);
1645 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1647 } /* while loop to process each character */
1649 /* terminate the string and set up the sv */
1651 SvCUR_set(sv, d - SvPVX(sv));
1652 if (SvCUR(sv) >= SvLEN(sv))
1653 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
1656 if (PL_encoding && !has_utf8) {
1657 Perl_sv_recode_to_utf8(aTHX_ sv, PL_encoding);
1662 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1663 PL_sublex_info.sub_op->op_private |=
1664 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1668 /* shrink the sv if we allocated more than we used */
1669 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1670 SvLEN_set(sv, SvCUR(sv) + 1);
1671 Renew(SvPVX(sv), SvLEN(sv), char);
1674 /* return the substring (via yylval) only if we parsed anything */
1675 if (s > PL_bufptr) {
1676 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1677 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1679 ( PL_lex_inwhat == OP_TRANS
1681 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1684 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1691 * Returns TRUE if there's more to the expression (e.g., a subscript),
1694 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1696 * ->[ and ->{ return TRUE
1697 * { and [ outside a pattern are always subscripts, so return TRUE
1698 * if we're outside a pattern and it's not { or [, then return FALSE
1699 * if we're in a pattern and the first char is a {
1700 * {4,5} (any digits around the comma) returns FALSE
1701 * if we're in a pattern and the first char is a [
1703 * [SOMETHING] has a funky algorithm to decide whether it's a
1704 * character class or not. It has to deal with things like
1705 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1706 * anything else returns TRUE
1709 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1712 S_intuit_more(pTHX_ register char *s)
1714 if (PL_lex_brackets)
1716 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1718 if (*s != '{' && *s != '[')
1723 /* In a pattern, so maybe we have {n,m}. */
1740 /* On the other hand, maybe we have a character class */
1743 if (*s == ']' || *s == '^')
1746 /* this is terrifying, and it works */
1747 int weight = 2; /* let's weigh the evidence */
1749 unsigned char un_char = 255, last_un_char;
1750 char *send = strchr(s,']');
1751 char tmpbuf[sizeof PL_tokenbuf * 4];
1753 if (!send) /* has to be an expression */
1756 Zero(seen,256,char);
1759 else if (isDIGIT(*s)) {
1761 if (isDIGIT(s[1]) && s[2] == ']')
1767 for (; s < send; s++) {
1768 last_un_char = un_char;
1769 un_char = (unsigned char)*s;
1774 weight -= seen[un_char] * 10;
1775 if (isALNUM_lazy_if(s+1,UTF)) {
1776 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1777 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1782 else if (*s == '$' && s[1] &&
1783 strchr("[#!%*<>()-=",s[1])) {
1784 if (/*{*/ strchr("])} =",s[2]))
1793 if (strchr("wds]",s[1]))
1795 else if (seen['\''] || seen['"'])
1797 else if (strchr("rnftbxcav",s[1]))
1799 else if (isDIGIT(s[1])) {
1801 while (s[1] && isDIGIT(s[1]))
1811 if (strchr("aA01! ",last_un_char))
1813 if (strchr("zZ79~",s[1]))
1815 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1816 weight -= 5; /* cope with negative subscript */
1819 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1820 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1825 if (keyword(tmpbuf, d - tmpbuf))
1828 if (un_char == last_un_char + 1)
1830 weight -= seen[un_char];
1835 if (weight >= 0) /* probably a character class */
1845 * Does all the checking to disambiguate
1847 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1848 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1850 * First argument is the stuff after the first token, e.g. "bar".
1852 * Not a method if bar is a filehandle.
1853 * Not a method if foo is a subroutine prototyped to take a filehandle.
1854 * Not a method if it's really "Foo $bar"
1855 * Method if it's "foo $bar"
1856 * Not a method if it's really "print foo $bar"
1857 * Method if it's really "foo package::" (interpreted as package->foo)
1858 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1859 * Not a method if bar is a filehandle or package, but is quoted with
1864 S_intuit_method(pTHX_ char *start, GV *gv)
1866 char *s = start + (*start == '$');
1867 char tmpbuf[sizeof PL_tokenbuf];
1875 if ((cv = GvCVu(gv))) {
1876 char *proto = SvPVX(cv);
1886 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1887 /* start is the beginning of the possible filehandle/object,
1888 * and s is the end of it
1889 * tmpbuf is a copy of it
1892 if (*start == '$') {
1893 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1898 return *s == '(' ? FUNCMETH : METHOD;
1900 if (!keyword(tmpbuf, len)) {
1901 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1906 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1907 if (indirgv && GvCVu(indirgv))
1909 /* filehandle or package name makes it a method */
1910 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1912 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1913 return 0; /* no assumptions -- "=>" quotes bearword */
1915 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1916 newSVpvn(tmpbuf,len));
1917 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1921 return *s == '(' ? FUNCMETH : METHOD;
1929 * Return a string of Perl code to load the debugger. If PERL5DB
1930 * is set, it will return the contents of that, otherwise a
1931 * compile-time require of perl5db.pl.
1938 char *pdb = PerlEnv_getenv("PERL5DB");
1942 SETERRNO(0,SS$_NORMAL);
1943 return "BEGIN { require 'perl5db.pl' }";
1949 /* Encoded script support. filter_add() effectively inserts a
1950 * 'pre-processing' function into the current source input stream.
1951 * Note that the filter function only applies to the current source file
1952 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1954 * The datasv parameter (which may be NULL) can be used to pass
1955 * private data to this instance of the filter. The filter function
1956 * can recover the SV using the FILTER_DATA macro and use it to
1957 * store private buffers and state information.
1959 * The supplied datasv parameter is upgraded to a PVIO type
1960 * and the IoDIRP/IoANY field is used to store the function pointer,
1961 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1962 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1963 * private use must be set using malloc'd pointers.
1967 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1972 if (!PL_rsfp_filters)
1973 PL_rsfp_filters = newAV();
1975 datasv = NEWSV(255,0);
1976 if (!SvUPGRADE(datasv, SVt_PVIO))
1977 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1978 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
1979 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1980 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1981 (void*)funcp, SvPV_nolen(datasv)));
1982 av_unshift(PL_rsfp_filters, 1);
1983 av_store(PL_rsfp_filters, 0, datasv) ;
1988 /* Delete most recently added instance of this filter function. */
1990 Perl_filter_del(pTHX_ filter_t funcp)
1993 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", (void*)funcp));
1994 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1996 /* if filter is on top of stack (usual case) just pop it off */
1997 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1998 if (IoANY(datasv) == (void *)funcp) {
1999 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
2000 IoANY(datasv) = (void *)NULL;
2001 sv_free(av_pop(PL_rsfp_filters));
2005 /* we need to search for the correct entry and clear it */
2006 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
2010 /* Invoke the n'th filter function for the current rsfp. */
2012 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
2015 /* 0 = read one text line */
2020 if (!PL_rsfp_filters)
2022 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
2023 /* Provide a default input filter to make life easy. */
2024 /* Note that we append to the line. This is handy. */
2025 DEBUG_P(PerlIO_printf(Perl_debug_log,
2026 "filter_read %d: from rsfp\n", idx));
2030 int old_len = SvCUR(buf_sv) ;
2032 /* ensure buf_sv is large enough */
2033 SvGROW(buf_sv, old_len + maxlen) ;
2034 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
2035 if (PerlIO_error(PL_rsfp))
2036 return -1; /* error */
2038 return 0 ; /* end of file */
2040 SvCUR_set(buf_sv, old_len + len) ;
2043 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2044 if (PerlIO_error(PL_rsfp))
2045 return -1; /* error */
2047 return 0 ; /* end of file */
2050 return SvCUR(buf_sv);
2052 /* Skip this filter slot if filter has been deleted */
2053 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
2054 DEBUG_P(PerlIO_printf(Perl_debug_log,
2055 "filter_read %d: skipped (filter deleted)\n",
2057 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
2059 /* Get function pointer hidden within datasv */
2060 funcp = (filter_t)IoANY(datasv);
2061 DEBUG_P(PerlIO_printf(Perl_debug_log,
2062 "filter_read %d: via function %p (%s)\n",
2063 idx, (void*)funcp, SvPV_nolen(datasv)));
2064 /* Call function. The function is expected to */
2065 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2066 /* Return: <0:error, =0:eof, >0:not eof */
2067 return (*funcp)(aTHX_ idx, buf_sv, maxlen);
2071 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2073 #ifdef PERL_CR_FILTER
2074 if (!PL_rsfp_filters) {
2075 filter_add(S_cr_textfilter,NULL);
2078 if (PL_rsfp_filters) {
2081 SvCUR_set(sv, 0); /* start with empty line */
2082 if (FILTER_READ(0, sv, 0) > 0)
2083 return ( SvPVX(sv) ) ;
2088 return (sv_gets(sv, fp, append));
2092 S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2096 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2100 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2101 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2103 return GvHV(gv); /* Foo:: */
2106 /* use constant CLASS => 'MyClass' */
2107 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2109 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2110 pkgname = SvPV_nolen(sv);
2114 return gv_stashpv(pkgname, FALSE);
2118 static char* exp_name[] =
2119 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2120 "ATTRTERM", "TERMBLOCK"
2127 Works out what to call the token just pulled out of the input
2128 stream. The yacc parser takes care of taking the ops we return and
2129 stitching them into a tree.
2135 if read an identifier
2136 if we're in a my declaration
2137 croak if they tried to say my($foo::bar)
2138 build the ops for a my() declaration
2139 if it's an access to a my() variable
2140 are we in a sort block?
2141 croak if my($a); $a <=> $b
2142 build ops for access to a my() variable
2143 if in a dq string, and they've said @foo and we can't find @foo
2145 build ops for a bareword
2146 if we already built the token before, use it.
2149 #ifdef USE_PURE_BISON
2151 Perl_yylex_r(pTHX_ YYSTYPE *lvalp, int *lcharp)
2156 yylval_pointer[yyactlevel] = lvalp;
2157 yychar_pointer[yyactlevel] = lcharp;
2158 if (yyactlevel >= YYMAXLEVEL)
2159 Perl_croak(aTHX_ "panic: YYMAXLEVEL");
2161 r = Perl_yylex(aTHX);
2171 #pragma segment Perl_yylex
2184 /* check if there's an identifier for us to look at */
2185 if (PL_pending_ident)
2186 return S_pending_ident(aTHX);
2188 /* no identifier pending identification */
2190 switch (PL_lex_state) {
2192 case LEX_NORMAL: /* Some compilers will produce faster */
2193 case LEX_INTERPNORMAL: /* code if we comment these out. */
2197 /* when we've already built the next token, just pull it out of the queue */
2200 yylval = PL_nextval[PL_nexttoke];
2202 PL_lex_state = PL_lex_defer;
2203 PL_expect = PL_lex_expect;
2204 PL_lex_defer = LEX_NORMAL;
2206 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2207 "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr,
2208 (IV)PL_nexttype[PL_nexttoke]); });
2210 return(PL_nexttype[PL_nexttoke]);
2212 /* interpolated case modifiers like \L \U, including \Q and \E.
2213 when we get here, PL_bufptr is at the \
2215 case LEX_INTERPCASEMOD:
2217 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2218 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2220 /* handle \E or end of string */
2221 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2225 if (PL_lex_casemods) {
2226 oldmod = PL_lex_casestack[--PL_lex_casemods];
2227 PL_lex_casestack[PL_lex_casemods] = '\0';
2229 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2231 PL_lex_state = LEX_INTERPCONCAT;
2235 if (PL_bufptr != PL_bufend)
2237 PL_lex_state = LEX_INTERPCONCAT;
2241 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2242 "### Saw case modifier at '%s'\n", PL_bufptr); });
2244 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2245 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2246 if (strchr("LU", *s) &&
2247 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2249 PL_lex_casestack[--PL_lex_casemods] = '\0';
2252 if (PL_lex_casemods > 10) {
2253 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2254 if (newlb != PL_lex_casestack) {
2256 PL_lex_casestack = newlb;
2259 PL_lex_casestack[PL_lex_casemods++] = *s;
2260 PL_lex_casestack[PL_lex_casemods] = '\0';
2261 PL_lex_state = LEX_INTERPCONCAT;
2262 PL_nextval[PL_nexttoke].ival = 0;
2265 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2267 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2269 PL_nextval[PL_nexttoke].ival = OP_LC;
2271 PL_nextval[PL_nexttoke].ival = OP_UC;
2273 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2275 Perl_croak(aTHX_ "panic: yylex");
2278 if (PL_lex_starts) {
2287 case LEX_INTERPPUSH:
2288 return sublex_push();
2290 case LEX_INTERPSTART:
2291 if (PL_bufptr == PL_bufend)
2292 return sublex_done();
2293 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2294 "### Interpolated variable at '%s'\n", PL_bufptr); });
2296 PL_lex_dojoin = (*PL_bufptr == '@');
2297 PL_lex_state = LEX_INTERPNORMAL;
2298 if (PL_lex_dojoin) {
2299 PL_nextval[PL_nexttoke].ival = 0;
2301 #ifdef USE_5005THREADS
2302 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2303 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2304 force_next(PRIVATEREF);
2306 force_ident("\"", '$');
2307 #endif /* USE_5005THREADS */
2308 PL_nextval[PL_nexttoke].ival = 0;
2310 PL_nextval[PL_nexttoke].ival = 0;
2312 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2315 if (PL_lex_starts++) {
2321 case LEX_INTERPENDMAYBE:
2322 if (intuit_more(PL_bufptr)) {
2323 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2329 if (PL_lex_dojoin) {
2330 PL_lex_dojoin = FALSE;
2331 PL_lex_state = LEX_INTERPCONCAT;
2334 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2335 && SvEVALED(PL_lex_repl))
2337 if (PL_bufptr != PL_bufend)
2338 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2339 PL_lex_repl = Nullsv;
2342 case LEX_INTERPCONCAT:
2344 if (PL_lex_brackets)
2345 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2347 if (PL_bufptr == PL_bufend)
2348 return sublex_done();
2350 if (SvIVX(PL_linestr) == '\'') {
2351 SV *sv = newSVsv(PL_linestr);
2354 else if ( PL_hints & HINT_NEW_RE )
2355 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2356 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2360 s = scan_const(PL_bufptr);
2362 PL_lex_state = LEX_INTERPCASEMOD;
2364 PL_lex_state = LEX_INTERPSTART;
2367 if (s != PL_bufptr) {
2368 PL_nextval[PL_nexttoke] = yylval;
2371 if (PL_lex_starts++)
2381 PL_lex_state = LEX_NORMAL;
2382 s = scan_formline(PL_bufptr);
2383 if (!PL_lex_formbrack)
2389 PL_oldoldbufptr = PL_oldbufptr;
2392 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2393 exp_name[PL_expect], s);
2399 if (isIDFIRST_lazy_if(s,UTF))
2401 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2404 goto fake_eof; /* emulate EOF on ^D or ^Z */
2409 if (PL_lex_brackets)
2410 yyerror("Missing right curly or square bracket");
2411 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2412 "### Tokener got EOF\n");
2416 if (s++ < PL_bufend)
2417 goto retry; /* ignore stray nulls */
2420 if (!PL_in_eval && !PL_preambled) {
2421 PL_preambled = TRUE;
2422 sv_setpv(PL_linestr,incl_perldb());
2423 if (SvCUR(PL_linestr))
2424 sv_catpv(PL_linestr,";");
2426 while(AvFILLp(PL_preambleav) >= 0) {
2427 SV *tmpsv = av_shift(PL_preambleav);
2428 sv_catsv(PL_linestr, tmpsv);
2429 sv_catpv(PL_linestr, ";");
2432 sv_free((SV*)PL_preambleav);
2433 PL_preambleav = NULL;
2435 if (PL_minus_n || PL_minus_p) {
2436 sv_catpv(PL_linestr, "LINE: while (<>) {");
2438 sv_catpv(PL_linestr,"chomp;");
2441 if (strchr("/'\"", *PL_splitstr)
2442 && strchr(PL_splitstr + 1, *PL_splitstr))
2443 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
2446 s = "'~#\200\1'"; /* surely one char is unused...*/
2447 while (s[1] && strchr(PL_splitstr, *s)) s++;
2449 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s%c",
2450 "q" + (delim == '\''), delim);
2451 for (s = PL_splitstr; *s; s++) {
2453 sv_catpvn(PL_linestr, "\\", 1);
2454 sv_catpvn(PL_linestr, s, 1);
2456 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2460 sv_catpv(PL_linestr,"our @F=split(' ');");
2463 sv_catpv(PL_linestr, "\n");
2464 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2465 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2466 PL_last_lop = PL_last_uni = Nullch;
2467 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2468 SV *sv = NEWSV(85,0);
2470 sv_upgrade(sv, SVt_PVMG);
2471 sv_setsv(sv,PL_linestr);
2474 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2479 bof = PL_rsfp ? TRUE : FALSE;
2480 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2483 if (PL_preprocess && !PL_in_eval)
2484 (void)PerlProc_pclose(PL_rsfp);
2485 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2486 PerlIO_clearerr(PL_rsfp);
2488 (void)PerlIO_close(PL_rsfp);
2490 PL_doextract = FALSE;
2492 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2493 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2494 sv_catpv(PL_linestr,";}");
2495 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2496 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2497 PL_last_lop = PL_last_uni = Nullch;
2498 PL_minus_n = PL_minus_p = 0;
2501 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2502 PL_last_lop = PL_last_uni = Nullch;
2503 sv_setpv(PL_linestr,"");
2504 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2506 /* if it looks like the start of a BOM, check if it in fact is */
2507 else if (bof && (!*s || *(U8*)s == 0xEF || *(U8*)s >= 0xFE)) {
2508 #ifdef PERLIO_IS_STDIO
2509 # ifdef __GNU_LIBRARY__
2510 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
2511 # define FTELL_FOR_PIPE_IS_BROKEN
2515 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2516 # define FTELL_FOR_PIPE_IS_BROKEN
2521 #ifdef FTELL_FOR_PIPE_IS_BROKEN
2522 /* This loses the possibility to detect the bof
2523 * situation on perl -P when the libc5 is being used.
2524 * Workaround? Maybe attach some extra state to PL_rsfp?
2527 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2529 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2532 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2533 s = swallow_bom((U8*)s);
2537 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2538 PL_doextract = FALSE;
2540 /* Incest with pod. */
2541 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2542 sv_setpv(PL_linestr, "");
2543 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2544 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2545 PL_last_lop = PL_last_uni = Nullch;
2546 PL_doextract = FALSE;
2550 } while (PL_doextract);
2551 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2552 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2553 SV *sv = NEWSV(85,0);
2555 sv_upgrade(sv, SVt_PVMG);
2556 sv_setsv(sv,PL_linestr);
2559 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2561 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2562 PL_last_lop = PL_last_uni = Nullch;
2563 if (CopLINE(PL_curcop) == 1) {
2564 while (s < PL_bufend && isSPACE(*s))
2566 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2570 if (*s == '#' && *(s+1) == '!')
2572 #ifdef ALTERNATE_SHEBANG
2574 static char as[] = ALTERNATE_SHEBANG;
2575 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2576 d = s + (sizeof(as) - 1);
2578 #endif /* ALTERNATE_SHEBANG */
2587 while (*d && !isSPACE(*d))
2591 #ifdef ARG_ZERO_IS_SCRIPT
2592 if (ipathend > ipath) {
2594 * HP-UX (at least) sets argv[0] to the script name,
2595 * which makes $^X incorrect. And Digital UNIX and Linux,
2596 * at least, set argv[0] to the basename of the Perl
2597 * interpreter. So, having found "#!", we'll set it right.
2599 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV)); /* $^X */
2600 assert(SvPOK(x) || SvGMAGICAL(x));
2601 if (sv_eq(x, CopFILESV(PL_curcop))) {
2602 sv_setpvn(x, ipath, ipathend - ipath);
2605 TAINT_NOT; /* $^X is always tainted, but that's OK */
2607 #endif /* ARG_ZERO_IS_SCRIPT */
2612 d = instr(s,"perl -");
2614 d = instr(s,"perl");
2616 /* avoid getting into infinite loops when shebang
2617 * line contains "Perl" rather than "perl" */
2619 for (d = ipathend-4; d >= ipath; --d) {
2620 if ((*d == 'p' || *d == 'P')
2621 && !ibcmp(d, "perl", 4))
2631 #ifdef ALTERNATE_SHEBANG
2633 * If the ALTERNATE_SHEBANG on this system starts with a
2634 * character that can be part of a Perl expression, then if
2635 * we see it but not "perl", we're probably looking at the
2636 * start of Perl code, not a request to hand off to some
2637 * other interpreter. Similarly, if "perl" is there, but
2638 * not in the first 'word' of the line, we assume the line
2639 * contains the start of the Perl program.
2641 if (d && *s != '#') {
2643 while (*c && !strchr("; \t\r\n\f\v#", *c))
2646 d = Nullch; /* "perl" not in first word; ignore */
2648 *s = '#'; /* Don't try to parse shebang line */
2650 #endif /* ALTERNATE_SHEBANG */
2651 #ifndef MACOS_TRADITIONAL
2656 !instr(s,"indir") &&
2657 instr(PL_origargv[0],"perl"))
2663 while (s < PL_bufend && isSPACE(*s))
2665 if (s < PL_bufend) {
2666 Newz(899,newargv,PL_origargc+3,char*);
2668 while (s < PL_bufend && !isSPACE(*s))
2671 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2674 newargv = PL_origargv;
2676 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
2677 Perl_croak(aTHX_ "Can't exec %s", ipath);
2681 U32 oldpdb = PL_perldb;
2682 bool oldn = PL_minus_n;
2683 bool oldp = PL_minus_p;
2685 while (*d && !isSPACE(*d)) d++;
2686 while (SPACE_OR_TAB(*d)) d++;
2689 bool switches_done = PL_doswitches;
2691 if (*d == 'M' || *d == 'm') {
2693 while (*d && !isSPACE(*d)) d++;
2694 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2697 d = moreswitches(d);
2699 if ((PERLDB_LINE && !oldpdb) ||
2700 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2701 /* if we have already added "LINE: while (<>) {",
2702 we must not do it again */
2704 sv_setpv(PL_linestr, "");
2705 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2706 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2707 PL_last_lop = PL_last_uni = Nullch;
2708 PL_preambled = FALSE;
2710 (void)gv_fetchfile(PL_origfilename);
2713 if (PL_doswitches && !switches_done) {
2714 int argc = PL_origargc;
2715 char **argv = PL_origargv;
2718 } while (argc && argv[0][0] == '-' && argv[0][1]);
2719 init_argv_symbols(argc,argv);
2725 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2727 PL_lex_state = LEX_FORMLINE;
2732 #ifdef PERL_STRICT_CR
2733 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2735 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2737 case ' ': case '\t': case '\f': case 013:
2738 #ifdef MACOS_TRADITIONAL
2745 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2746 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2747 /* handle eval qq[#line 1 "foo"\n ...] */
2748 CopLINE_dec(PL_curcop);
2752 while (s < d && *s != '\n')
2756 else if (s > d) /* Found by Ilya: feed random input to Perl. */
2757 Perl_croak(aTHX_ "panic: input overflow");
2759 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2761 PL_lex_state = LEX_FORMLINE;
2771 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2778 while (s < PL_bufend && SPACE_OR_TAB(*s))
2781 if (strnEQ(s,"=>",2)) {
2782 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2783 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2784 "### Saw unary minus before =>, forcing word '%s'\n", s);
2786 OPERATOR('-'); /* unary minus */
2788 PL_last_uni = PL_oldbufptr;
2790 case 'r': ftst = OP_FTEREAD; break;
2791 case 'w': ftst = OP_FTEWRITE; break;
2792 case 'x': ftst = OP_FTEEXEC; break;
2793 case 'o': ftst = OP_FTEOWNED; break;
2794 case 'R': ftst = OP_FTRREAD; break;
2795 case 'W': ftst = OP_FTRWRITE; break;
2796 case 'X': ftst = OP_FTREXEC; break;
2797 case 'O': ftst = OP_FTROWNED; break;
2798 case 'e': ftst = OP_FTIS; break;
2799 case 'z': ftst = OP_FTZERO; break;
2800 case 's': ftst = OP_FTSIZE; break;
2801 case 'f': ftst = OP_FTFILE; break;
2802 case 'd': ftst = OP_FTDIR; break;
2803 case 'l': ftst = OP_FTLINK; break;
2804 case 'p': ftst = OP_FTPIPE; break;
2805 case 'S': ftst = OP_FTSOCK; break;
2806 case 'u': ftst = OP_FTSUID; break;
2807 case 'g': ftst = OP_FTSGID; break;
2808 case 'k': ftst = OP_FTSVTX; break;
2809 case 'b': ftst = OP_FTBLK; break;
2810 case 'c': ftst = OP_FTCHR; break;
2811 case 't': ftst = OP_FTTTY; break;
2812 case 'T': ftst = OP_FTTEXT; break;
2813 case 'B': ftst = OP_FTBINARY; break;
2814 case 'M': case 'A': case 'C':
2815 gv_fetchpv("\024",TRUE, SVt_PV);
2817 case 'M': ftst = OP_FTMTIME; break;
2818 case 'A': ftst = OP_FTATIME; break;
2819 case 'C': ftst = OP_FTCTIME; break;
2827 PL_last_lop_op = ftst;
2828 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2829 "### Saw file test %c\n", (int)ftst);
2834 /* Assume it was a minus followed by a one-letter named
2835 * subroutine call (or a -bareword), then. */
2836 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2837 "### %c looked like a file test but was not\n",
2846 if (PL_expect == XOPERATOR)
2851 else if (*s == '>') {
2854 if (isIDFIRST_lazy_if(s,UTF)) {
2855 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2863 if (PL_expect == XOPERATOR)
2866 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2868 OPERATOR('-'); /* unary minus */
2875 if (PL_expect == XOPERATOR)
2880 if (PL_expect == XOPERATOR)
2883 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2889 if (PL_expect != XOPERATOR) {
2890 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2891 PL_expect = XOPERATOR;
2892 force_ident(PL_tokenbuf, '*');
2905 if (PL_expect == XOPERATOR) {
2909 PL_tokenbuf[0] = '%';
2910 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2911 if (!PL_tokenbuf[1]) {
2913 yyerror("Final % should be \\% or %name");
2916 PL_pending_ident = '%';
2935 switch (PL_expect) {
2938 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2940 PL_bufptr = s; /* update in case we back off */
2946 PL_expect = XTERMBLOCK;
2950 while (isIDFIRST_lazy_if(s,UTF)) {
2951 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2952 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2953 if (tmp < 0) tmp = -tmp;
2968 d = scan_str(d,TRUE,TRUE);
2970 /* MUST advance bufptr here to avoid bogus
2971 "at end of line" context messages from yyerror().
2973 PL_bufptr = s + len;
2974 yyerror("Unterminated attribute parameter in attribute list");
2977 return 0; /* EOF indicator */
2981 SV *sv = newSVpvn(s, len);
2982 sv_catsv(sv, PL_lex_stuff);
2983 attrs = append_elem(OP_LIST, attrs,
2984 newSVOP(OP_CONST, 0, sv));
2985 SvREFCNT_dec(PL_lex_stuff);
2986 PL_lex_stuff = Nullsv;
2989 if (!PL_in_my && len == 6 && strnEQ(s, "lvalue", len))
2990 CvLVALUE_on(PL_compcv);
2991 else if (!PL_in_my && len == 6 && strnEQ(s, "locked", len))
2992 CvLOCKED_on(PL_compcv);
2993 else if (!PL_in_my && len == 6 && strnEQ(s, "method", len))
2994 CvMETHOD_on(PL_compcv);
2996 else if (PL_in_my == KEY_our && len == 6 && strnEQ(s, "unique", len))
2997 GvUNIQUE_on(cGVOPx_gv(yylval.opval));
2999 /* After we've set the flags, it could be argued that
3000 we don't need to do the attributes.pm-based setting
3001 process, and shouldn't bother appending recognized
3002 flags. To experiment with that, uncomment the
3003 following "else": */
3005 attrs = append_elem(OP_LIST, attrs,
3006 newSVOP(OP_CONST, 0,
3010 if (*s == ':' && s[1] != ':')
3013 break; /* require real whitespace or :'s */
3015 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3016 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
3017 char q = ((*s == '\'') ? '"' : '\'');
3018 /* If here for an expression, and parsed no attrs, back off. */
3019 if (tmp == '=' && !attrs) {
3023 /* MUST advance bufptr here to avoid bogus "at end of line"
3024 context messages from yyerror().
3028 yyerror("Unterminated attribute list");
3030 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
3038 PL_nextval[PL_nexttoke].opval = attrs;
3046 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3047 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3063 if (PL_lex_brackets <= 0)
3064 yyerror("Unmatched right square bracket");
3067 if (PL_lex_state == LEX_INTERPNORMAL) {
3068 if (PL_lex_brackets == 0) {
3069 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3070 PL_lex_state = LEX_INTERPEND;
3077 if (PL_lex_brackets > 100) {
3078 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
3079 if (newlb != PL_lex_brackstack) {
3081 PL_lex_brackstack = newlb;
3084 switch (PL_expect) {
3086 if (PL_lex_formbrack) {
3090 if (PL_oldoldbufptr == PL_last_lop)
3091 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3093 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3094 OPERATOR(HASHBRACK);
3096 while (s < PL_bufend && SPACE_OR_TAB(*s))
3099 PL_tokenbuf[0] = '\0';
3100 if (d < PL_bufend && *d == '-') {
3101 PL_tokenbuf[0] = '-';
3103 while (d < PL_bufend && SPACE_OR_TAB(*d))
3106 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3107 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3109 while (d < PL_bufend && SPACE_OR_TAB(*d))
3112 char minus = (PL_tokenbuf[0] == '-');
3113 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3121 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3126 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3131 if (PL_oldoldbufptr == PL_last_lop)
3132 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3134 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3137 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
3139 /* This hack is to get the ${} in the message. */
3141 yyerror("syntax error");
3144 OPERATOR(HASHBRACK);
3146 /* This hack serves to disambiguate a pair of curlies
3147 * as being a block or an anon hash. Normally, expectation
3148 * determines that, but in cases where we're not in a
3149 * position to expect anything in particular (like inside
3150 * eval"") we have to resolve the ambiguity. This code
3151 * covers the case where the first term in the curlies is a
3152 * quoted string. Most other cases need to be explicitly
3153 * disambiguated by prepending a `+' before the opening
3154 * curly in order to force resolution as an anon hash.
3156 * XXX should probably propagate the outer expectation
3157 * into eval"" to rely less on this hack, but that could
3158 * potentially break current behavior of eval"".
3162 if (*s == '\'' || *s == '"' || *s == '`') {
3163 /* common case: get past first string, handling escapes */
3164 for (t++; t < PL_bufend && *t != *s;)
3165 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3169 else if (*s == 'q') {
3172 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3176 char open, close, term;
3179 while (t < PL_bufend && isSPACE(*t))
3183 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3187 for (t++; t < PL_bufend; t++) {
3188 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3190 else if (*t == open)
3194 for (t++; t < PL_bufend; t++) {
3195 if (*t == '\\' && t+1 < PL_bufend)
3197 else if (*t == close && --brackets <= 0)
3199 else if (*t == open)
3205 else if (isALNUM_lazy_if(t,UTF)) {
3207 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3210 while (t < PL_bufend && isSPACE(*t))
3212 /* if comma follows first term, call it an anon hash */
3213 /* XXX it could be a comma expression with loop modifiers */
3214 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3215 || (*t == '=' && t[1] == '>')))
3216 OPERATOR(HASHBRACK);
3217 if (PL_expect == XREF)
3220 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3226 yylval.ival = CopLINE(PL_curcop);
3227 if (isSPACE(*s) || *s == '#')
3228 PL_copline = NOLINE; /* invalidate current command line number */
3233 if (PL_lex_brackets <= 0)
3234 yyerror("Unmatched right curly bracket");
3236 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3237 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3238 PL_lex_formbrack = 0;
3239 if (PL_lex_state == LEX_INTERPNORMAL) {
3240 if (PL_lex_brackets == 0) {
3241 if (PL_expect & XFAKEBRACK) {
3242 PL_expect &= XENUMMASK;
3243 PL_lex_state = LEX_INTERPEND;
3245 return yylex(); /* ignore fake brackets */
3247 if (*s == '-' && s[1] == '>')
3248 PL_lex_state = LEX_INTERPENDMAYBE;
3249 else if (*s != '[' && *s != '{')
3250 PL_lex_state = LEX_INTERPEND;
3253 if (PL_expect & XFAKEBRACK) {
3254 PL_expect &= XENUMMASK;
3256 return yylex(); /* ignore fake brackets */
3266 if (PL_expect == XOPERATOR) {
3267 if (ckWARN(WARN_SEMICOLON)
3268 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3270 CopLINE_dec(PL_curcop);
3271 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3272 CopLINE_inc(PL_curcop);
3277 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3279 PL_expect = XOPERATOR;
3280 force_ident(PL_tokenbuf, '&');
3284 yylval.ival = (OPpENTERSUB_AMPER<<8);
3303 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3304 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3306 if (PL_expect == XSTATE && isALPHA(tmp) &&
3307 (s == PL_linestart+1 || s[-2] == '\n') )
3309 if (PL_in_eval && !PL_rsfp) {
3314 if (strnEQ(s,"=cut",4)) {
3328 PL_doextract = TRUE;
3331 if (PL_lex_brackets < PL_lex_formbrack) {
3333 #ifdef PERL_STRICT_CR
3334 for (t = s; SPACE_OR_TAB(*t); t++) ;
3336 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3338 if (*t == '\n' || *t == '#') {
3356 if (PL_expect != XOPERATOR) {
3357 if (s[1] != '<' && !strchr(s,'>'))
3360 s = scan_heredoc(s);
3362 s = scan_inputsymbol(s);
3363 TERM(sublex_start());
3368 SHop(OP_LEFT_SHIFT);
3382 SHop(OP_RIGHT_SHIFT);
3391 if (PL_expect == XOPERATOR) {
3392 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3395 return ','; /* grandfather non-comma-format format */
3399 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3400 PL_tokenbuf[0] = '@';
3401 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3402 sizeof PL_tokenbuf - 1, FALSE);
3403 if (PL_expect == XOPERATOR)
3404 no_op("Array length", s);
3405 if (!PL_tokenbuf[1])
3407 PL_expect = XOPERATOR;
3408 PL_pending_ident = '#';
3412 PL_tokenbuf[0] = '$';
3413 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3414 sizeof PL_tokenbuf - 1, FALSE);
3415 if (PL_expect == XOPERATOR)
3417 if (!PL_tokenbuf[1]) {
3419 yyerror("Final $ should be \\$ or $name");
3423 /* This kludge not intended to be bulletproof. */
3424 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3425 yylval.opval = newSVOP(OP_CONST, 0,
3426 newSViv(PL_compiling.cop_arybase));
3427 yylval.opval->op_private = OPpCONST_ARYBASE;
3433 if (PL_lex_state == LEX_NORMAL)
3436 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3439 PL_tokenbuf[0] = '@';
3440 if (ckWARN(WARN_SYNTAX)) {
3442 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3445 PL_bufptr = skipspace(PL_bufptr);
3446 while (t < PL_bufend && *t != ']')
3448 Perl_warner(aTHX_ WARN_SYNTAX,
3449 "Multidimensional syntax %.*s not supported",
3450 (t - PL_bufptr) + 1, PL_bufptr);
3454 else if (*s == '{') {
3455 PL_tokenbuf[0] = '%';
3456 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3457 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3459 char tmpbuf[sizeof PL_tokenbuf];
3461 for (t++; isSPACE(*t); t++) ;
3462 if (isIDFIRST_lazy_if(t,UTF)) {
3463 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3464 for (; isSPACE(*t); t++) ;
3465 if (*t == ';' && get_cv(tmpbuf, FALSE))
3466 Perl_warner(aTHX_ WARN_SYNTAX,
3467 "You need to quote \"%s\"", tmpbuf);
3473 PL_expect = XOPERATOR;
3474 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3475 bool islop = (PL_last_lop == PL_oldoldbufptr);
3476 if (!islop || PL_last_lop_op == OP_GREPSTART)
3477 PL_expect = XOPERATOR;
3478 else if (strchr("$@\"'`q", *s))
3479 PL_expect = XTERM; /* e.g. print $fh "foo" */
3480 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3481 PL_expect = XTERM; /* e.g. print $fh &sub */
3482 else if (isIDFIRST_lazy_if(s,UTF)) {
3483 char tmpbuf[sizeof PL_tokenbuf];
3484 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3485 if ((tmp = keyword(tmpbuf, len))) {
3486 /* binary operators exclude handle interpretations */
3498 PL_expect = XTERM; /* e.g. print $fh length() */
3503 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3504 if (gv && GvCVu(gv))
3505 PL_expect = XTERM; /* e.g. print $fh subr() */
3508 else if (isDIGIT(*s))
3509 PL_expect = XTERM; /* e.g. print $fh 3 */
3510 else if (*s == '.' && isDIGIT(s[1]))
3511 PL_expect = XTERM; /* e.g. print $fh .3 */
3512 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3513 PL_expect = XTERM; /* e.g. print $fh -1 */
3514 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3515 PL_expect = XTERM; /* print $fh <<"EOF" */
3517 PL_pending_ident = '$';
3521 if (PL_expect == XOPERATOR)
3523 PL_tokenbuf[0] = '@';
3524 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3525 if (!PL_tokenbuf[1]) {
3527 yyerror("Final @ should be \\@ or @name");
3530 if (PL_lex_state == LEX_NORMAL)
3532 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3534 PL_tokenbuf[0] = '%';
3536 /* Warn about @ where they meant $. */
3537 if (ckWARN(WARN_SYNTAX)) {
3538 if (*s == '[' || *s == '{') {
3540 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3542 if (*t == '}' || *t == ']') {
3544 PL_bufptr = skipspace(PL_bufptr);
3545 Perl_warner(aTHX_ WARN_SYNTAX,
3546 "Scalar value %.*s better written as $%.*s",
3547 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3552 PL_pending_ident = '@';
3555 case '/': /* may either be division or pattern */
3556 case '?': /* may either be conditional or pattern */
3557 if (PL_expect != XOPERATOR) {
3558 /* Disable warning on "study /blah/" */
3559 if (PL_oldoldbufptr == PL_last_uni
3560 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3561 || memNE(PL_last_uni, "study", 5)
3562 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3564 s = scan_pat(s,OP_MATCH);
3565 TERM(sublex_start());
3573 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3574 #ifdef PERL_STRICT_CR
3577 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3579 && (s == PL_linestart || s[-1] == '\n') )
3581 PL_lex_formbrack = 0;
3585 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3591 yylval.ival = OPf_SPECIAL;
3597 if (PL_expect != XOPERATOR)
3602 case '0': case '1': case '2': case '3': case '4':
3603 case '5': case '6': case '7': case '8': case '9':
3604 s = scan_num(s, &yylval);
3605 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3606 "### Saw number in '%s'\n", s);
3608 if (PL_expect == XOPERATOR)
3613 s = scan_str(s,FALSE,FALSE);
3614 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3615 "### Saw string before '%s'\n", s);
3617 if (PL_expect == XOPERATOR) {
3618 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3621 return ','; /* grandfather non-comma-format format */
3627 missingterm((char*)0);
3628 yylval.ival = OP_CONST;
3629 TERM(sublex_start());
3632 s = scan_str(s,FALSE,FALSE);
3633 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3634 "### Saw string before '%s'\n", s);
3636 if (PL_expect == XOPERATOR) {
3637 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3640 return ','; /* grandfather non-comma-format format */
3646 missingterm((char*)0);
3647 yylval.ival = OP_CONST;
3648 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3649 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
3650 yylval.ival = OP_STRINGIFY;
3654 TERM(sublex_start());
3657 s = scan_str(s,FALSE,FALSE);
3658 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3659 "### Saw backtick string before '%s'\n", s);
3661 if (PL_expect == XOPERATOR)
3662 no_op("Backticks",s);
3664 missingterm((char*)0);
3665 yylval.ival = OP_BACKTICK;
3667 TERM(sublex_start());
3671 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3672 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3674 if (PL_expect == XOPERATOR)
3675 no_op("Backslash",s);
3679 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3683 while (isDIGIT(*start) || *start == '_')
3685 if (*start == '.' && isDIGIT(start[1])) {
3686 s = scan_num(s, &yylval);
3689 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3690 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF || PL_expect == XSTATE)) {
3694 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3697 s = scan_num(s, &yylval);
3704 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3743 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3745 /* Some keywords can be followed by any delimiter, including ':' */
3746 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3747 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3748 (PL_tokenbuf[0] == 'q' &&
3749 strchr("qwxr", PL_tokenbuf[1])))));
3751 /* x::* is just a word, unless x is "CORE" */
3752 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3756 while (d < PL_bufend && isSPACE(*d))
3757 d++; /* no comments skipped here, or s### is misparsed */
3759 /* Is this a label? */
3760 if (!tmp && PL_expect == XSTATE
3761 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3763 yylval.pval = savepv(PL_tokenbuf);
3768 /* Check for keywords */
3769 tmp = keyword(PL_tokenbuf, len);
3771 /* Is this a word before a => operator? */
3772 if (*d == '=' && d[1] == '>') {
3774 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3775 yylval.opval->op_private = OPpCONST_BARE;
3776 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
3777 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3781 if (tmp < 0) { /* second-class keyword? */
3782 GV *ogv = Nullgv; /* override (winner) */
3783 GV *hgv = Nullgv; /* hidden (loser) */
3784 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3786 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3789 if (GvIMPORTED_CV(gv))
3791 else if (! CvMETHOD(cv))
3795 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3796 (gv = *gvp) != (GV*)&PL_sv_undef &&
3797 GvCVu(gv) && GvIMPORTED_CV(gv))
3803 tmp = 0; /* overridden by import or by GLOBAL */
3806 && -tmp==KEY_lock /* XXX generalizable kludge */
3808 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3810 tmp = 0; /* any sub overrides "weak" keyword */
3812 else { /* no override */
3814 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
3815 Perl_warner(aTHX_ WARN_MISC,
3816 "dump() better written as CORE::dump()");
3820 if (ckWARN(WARN_AMBIGUOUS) && hgv
3821 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3822 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3823 "Ambiguous call resolved as CORE::%s(), %s",
3824 GvENAME(hgv), "qualify as such or use &");
3831 default: /* not a keyword */
3835 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3837 /* Get the rest if it looks like a package qualifier */
3839 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3841 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3844 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3845 *s == '\'' ? "'" : "::");
3850 if (PL_expect == XOPERATOR) {
3851 if (PL_bufptr == PL_linestart) {
3852 CopLINE_dec(PL_curcop);
3853 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3854 CopLINE_inc(PL_curcop);
3857 no_op("Bareword",s);
3860 /* Look for a subroutine with this name in current package,
3861 unless name is "Foo::", in which case Foo is a bearword
3862 (and a package name). */
3865 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3867 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3868 Perl_warner(aTHX_ WARN_BAREWORD,
3869 "Bareword \"%s\" refers to nonexistent package",
3872 PL_tokenbuf[len] = '\0';
3879 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3882 /* if we saw a global override before, get the right name */
3885 sv = newSVpvn("CORE::GLOBAL::",14);
3886 sv_catpv(sv,PL_tokenbuf);
3889 sv = newSVpv(PL_tokenbuf,0);
3891 /* Presume this is going to be a bareword of some sort. */
3894 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3895 yylval.opval->op_private = OPpCONST_BARE;
3897 /* And if "Foo::", then that's what it certainly is. */
3902 /* See if it's the indirect object for a list operator. */
3904 if (PL_oldoldbufptr &&
3905 PL_oldoldbufptr < PL_bufptr &&
3906 (PL_oldoldbufptr == PL_last_lop
3907 || PL_oldoldbufptr == PL_last_uni) &&
3908 /* NO SKIPSPACE BEFORE HERE! */
3909 (PL_expect == XREF ||
3910 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3912 bool immediate_paren = *s == '(';
3914 /* (Now we can afford to cross potential line boundary.) */
3917 /* Two barewords in a row may indicate method call. */
3919 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3922 /* If not a declared subroutine, it's an indirect object. */
3923 /* (But it's an indir obj regardless for sort.) */
3925 if ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
3926 ((!gv || !GvCVu(gv)) &&
3927 (PL_last_lop_op != OP_MAPSTART &&
3928 PL_last_lop_op != OP_GREPSTART))))
3930 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3935 PL_expect = XOPERATOR;
3938 /* Is this a word before a => operator? */
3939 if (*s == '=' && s[1] == '>' && !pkgname) {
3941 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
3942 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
3943 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3947 /* If followed by a paren, it's certainly a subroutine. */
3950 if (gv && GvCVu(gv)) {
3951 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
3952 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3957 PL_nextval[PL_nexttoke].opval = yylval.opval;
3958 PL_expect = XOPERATOR;
3964 /* If followed by var or block, call it a method (unless sub) */
3966 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3967 PL_last_lop = PL_oldbufptr;
3968 PL_last_lop_op = OP_METHOD;
3972 /* If followed by a bareword, see if it looks like indir obj. */
3974 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3977 /* Not a method, so call it a subroutine (if defined) */
3979 if (gv && GvCVu(gv)) {
3981 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3982 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3983 "Ambiguous use of -%s resolved as -&%s()",
3984 PL_tokenbuf, PL_tokenbuf);
3985 /* Check for a constant sub */
3987 if ((sv = cv_const_sv(cv))) {
3989 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3990 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3991 yylval.opval->op_private = 0;
3995 /* Resolve to GV now. */
3996 op_free(yylval.opval);
3997 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3998 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3999 PL_last_lop = PL_oldbufptr;
4000 PL_last_lop_op = OP_ENTERSUB;
4001 /* Is there a prototype? */
4004 char *proto = SvPV((SV*)cv, len);
4007 if (strEQ(proto, "$"))
4009 if (*proto == '&' && *s == '{') {
4010 sv_setpv(PL_subname,"__ANON__");
4014 PL_nextval[PL_nexttoke].opval = yylval.opval;
4020 /* Call it a bare word */
4022 if (PL_hints & HINT_STRICT_SUBS)
4023 yylval.opval->op_private |= OPpCONST_STRICT;
4026 if (ckWARN(WARN_RESERVED)) {
4027 if (lastchar != '-') {
4028 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4029 if (!*d && strNE(PL_tokenbuf,"main"))
4030 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
4037 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
4038 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4039 "Operator or semicolon missing before %c%s",
4040 lastchar, PL_tokenbuf);
4041 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4042 "Ambiguous use of %c resolved as operator %c",
4043 lastchar, lastchar);
4049 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4050 newSVpv(CopFILE(PL_curcop),0));
4054 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4055 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4058 case KEY___PACKAGE__:
4059 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4061 ? newSVsv(PL_curstname)
4070 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4071 char *pname = "main";
4072 if (PL_tokenbuf[2] == 'D')
4073 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
4074 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4077 GvIOp(gv) = newIO();
4078 IoIFP(GvIOp(gv)) = PL_rsfp;
4079 #if defined(HAS_FCNTL) && defined(F_SETFD)
4081 int fd = PerlIO_fileno(PL_rsfp);
4082 fcntl(fd,F_SETFD,fd >= 3);
4085 /* Mark this internal pseudo-handle as clean */
4086 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4088 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4089 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4090 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4092 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4093 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4094 /* if the script was opened in binmode, we need to revert
4095 * it to text mode for compatibility; but only iff it has CRs
4096 * XXX this is a questionable hack at best. */
4097 if (PL_bufend-PL_bufptr > 2
4098 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4101 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4102 loc = PerlIO_tell(PL_rsfp);
4103 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4106 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
4108 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4109 #endif /* NETWARE */
4110 #ifdef PERLIO_IS_STDIO /* really? */
4111 # if defined(__BORLANDC__)
4112 /* XXX see note in do_binmode() */
4113 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
4117 PerlIO_seek(PL_rsfp, loc, 0);
4121 #ifdef PERLIO_LAYERS
4122 if (UTF && !IN_BYTES)
4123 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
4136 if (PL_expect == XSTATE) {
4143 if (*s == ':' && s[1] == ':') {
4146 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4147 if (!(tmp = keyword(PL_tokenbuf, len)))
4148 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4162 LOP(OP_ACCEPT,XTERM);
4168 LOP(OP_ATAN2,XTERM);
4174 LOP(OP_BINMODE,XTERM);
4177 LOP(OP_BLESS,XTERM);
4186 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4203 if (!PL_cryptseen) {
4204 PL_cryptseen = TRUE;
4208 LOP(OP_CRYPT,XTERM);
4211 LOP(OP_CHMOD,XTERM);
4214 LOP(OP_CHOWN,XTERM);
4217 LOP(OP_CONNECT,XTERM);
4233 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4237 PL_hints |= HINT_BLOCK_SCOPE;
4247 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4248 LOP(OP_DBMOPEN,XTERM);
4254 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4261 yylval.ival = CopLINE(PL_curcop);
4275 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4276 UNIBRACK(OP_ENTEREVAL);
4291 case KEY_endhostent:
4297 case KEY_endservent:
4300 case KEY_endprotoent:
4311 yylval.ival = CopLINE(PL_curcop);
4313 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4315 if ((PL_bufend - p) >= 3 &&
4316 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4318 else if ((PL_bufend - p) >= 4 &&
4319 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4322 if (isIDFIRST_lazy_if(p,UTF)) {
4323 p = scan_ident(p, PL_bufend,
4324 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4328 Perl_croak(aTHX_ "Missing $ on loop variable");
4333 LOP(OP_FORMLINE,XTERM);
4339 LOP(OP_FCNTL,XTERM);
4345 LOP(OP_FLOCK,XTERM);
4354 LOP(OP_GREPSTART, XREF);
4357 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4372 case KEY_getpriority:
4373 LOP(OP_GETPRIORITY,XTERM);
4375 case KEY_getprotobyname:
4378 case KEY_getprotobynumber:
4379 LOP(OP_GPBYNUMBER,XTERM);
4381 case KEY_getprotoent:
4393 case KEY_getpeername:
4394 UNI(OP_GETPEERNAME);
4396 case KEY_gethostbyname:
4399 case KEY_gethostbyaddr:
4400 LOP(OP_GHBYADDR,XTERM);
4402 case KEY_gethostent:
4405 case KEY_getnetbyname:
4408 case KEY_getnetbyaddr:
4409 LOP(OP_GNBYADDR,XTERM);
4414 case KEY_getservbyname:
4415 LOP(OP_GSBYNAME,XTERM);
4417 case KEY_getservbyport:
4418 LOP(OP_GSBYPORT,XTERM);
4420 case KEY_getservent:
4423 case KEY_getsockname:
4424 UNI(OP_GETSOCKNAME);
4426 case KEY_getsockopt:
4427 LOP(OP_GSOCKOPT,XTERM);
4449 yylval.ival = CopLINE(PL_curcop);
4453 LOP(OP_INDEX,XTERM);
4459 LOP(OP_IOCTL,XTERM);
4471 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4503 LOP(OP_LISTEN,XTERM);
4512 s = scan_pat(s,OP_MATCH);
4513 TERM(sublex_start());
4516 LOP(OP_MAPSTART, XREF);
4519 LOP(OP_MKDIR,XTERM);
4522 LOP(OP_MSGCTL,XTERM);
4525 LOP(OP_MSGGET,XTERM);
4528 LOP(OP_MSGRCV,XTERM);
4531 LOP(OP_MSGSND,XTERM);
4537 if (isIDFIRST_lazy_if(s,UTF)) {
4538 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4539 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4541 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4542 if (!PL_in_my_stash) {
4545 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4553 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4560 if (PL_expect != XSTATE)
4561 yyerror("\"no\" not allowed in expression");
4562 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4563 s = force_version(s, FALSE);
4568 if (*s == '(' || (s = skipspace(s), *s == '('))
4575 if (isIDFIRST_lazy_if(s,UTF)) {
4577 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4579 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4580 Perl_warner(aTHX_ WARN_PRECEDENCE,
4581 "Precedence problem: open %.*s should be open(%.*s)",
4587 yylval.ival = OP_OR;
4597 LOP(OP_OPEN_DIR,XTERM);
4600 checkcomma(s,PL_tokenbuf,"filehandle");
4604 checkcomma(s,PL_tokenbuf,"filehandle");
4623 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4627 LOP(OP_PIPE_OP,XTERM);
4630 s = scan_str(s,FALSE,FALSE);
4632 missingterm((char*)0);
4633 yylval.ival = OP_CONST;
4634 TERM(sublex_start());
4640 s = scan_str(s,FALSE,FALSE);
4642 missingterm((char*)0);
4644 if (SvCUR(PL_lex_stuff)) {
4647 d = SvPV_force(PL_lex_stuff, len);
4650 for (; isSPACE(*d) && len; --len, ++d) ;
4653 if (!warned && ckWARN(WARN_QW)) {
4654 for (; !isSPACE(*d) && len; --len, ++d) {
4656 Perl_warner(aTHX_ WARN_QW,
4657 "Possible attempt to separate words with commas");
4660 else if (*d == '#') {
4661 Perl_warner(aTHX_ WARN_QW,
4662 "Possible attempt to put comments in qw() list");
4668 for (; !isSPACE(*d) && len; --len, ++d) ;
4670 sv = newSVpvn(b, d-b);
4671 if (DO_UTF8(PL_lex_stuff))
4673 words = append_elem(OP_LIST, words,
4674 newSVOP(OP_CONST, 0, tokeq(sv)));
4678 PL_nextval[PL_nexttoke].opval = words;
4683 SvREFCNT_dec(PL_lex_stuff);
4684 PL_lex_stuff = Nullsv;
4690 s = scan_str(s,FALSE,FALSE);
4692 missingterm((char*)0);
4693 yylval.ival = OP_STRINGIFY;
4694 if (SvIVX(PL_lex_stuff) == '\'')
4695 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4696 TERM(sublex_start());
4699 s = scan_pat(s,OP_QR);
4700 TERM(sublex_start());
4703 s = scan_str(s,FALSE,FALSE);
4705 missingterm((char*)0);
4706 yylval.ival = OP_BACKTICK;
4708 TERM(sublex_start());
4716 s = force_version(s, FALSE);
4718 else if (*s != 'v' || !isDIGIT(s[1])
4719 || (s = force_version(s, TRUE), *s == 'v'))
4721 *PL_tokenbuf = '\0';
4722 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4723 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4724 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4726 yyerror("<> should be quotes");
4734 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4738 LOP(OP_RENAME,XTERM);
4747 LOP(OP_RINDEX,XTERM);
4770 LOP(OP_REVERSE,XTERM);
4781 TERM(sublex_start());
4783 TOKEN(1); /* force error */
4792 LOP(OP_SELECT,XTERM);
4798 LOP(OP_SEMCTL,XTERM);
4801 LOP(OP_SEMGET,XTERM);
4804 LOP(OP_SEMOP,XTERM);
4810 LOP(OP_SETPGRP,XTERM);
4812 case KEY_setpriority:
4813 LOP(OP_SETPRIORITY,XTERM);
4815 case KEY_sethostent:
4821 case KEY_setservent:
4824 case KEY_setprotoent:
4834 LOP(OP_SEEKDIR,XTERM);
4836 case KEY_setsockopt:
4837 LOP(OP_SSOCKOPT,XTERM);
4843 LOP(OP_SHMCTL,XTERM);
4846 LOP(OP_SHMGET,XTERM);
4849 LOP(OP_SHMREAD,XTERM);
4852 LOP(OP_SHMWRITE,XTERM);
4855 LOP(OP_SHUTDOWN,XTERM);
4864 LOP(OP_SOCKET,XTERM);
4866 case KEY_socketpair:
4867 LOP(OP_SOCKPAIR,XTERM);
4870 checkcomma(s,PL_tokenbuf,"subroutine name");
4872 if (*s == ';' || *s == ')') /* probably a close */
4873 Perl_croak(aTHX_ "sort is now a reserved word");
4875 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4879 LOP(OP_SPLIT,XTERM);
4882 LOP(OP_SPRINTF,XTERM);
4885 LOP(OP_SPLICE,XTERM);
4900 LOP(OP_SUBSTR,XTERM);
4906 char tmpbuf[sizeof PL_tokenbuf];
4907 SSize_t tboffset = 0;
4908 expectation attrful;
4909 bool have_name, have_proto, bad_proto;
4914 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4915 (*s == ':' && s[1] == ':'))
4918 attrful = XATTRBLOCK;
4919 /* remember buffer pos'n for later force_word */
4920 tboffset = s - PL_oldbufptr;
4921 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4922 if (strchr(tmpbuf, ':'))
4923 sv_setpv(PL_subname, tmpbuf);
4925 sv_setsv(PL_subname,PL_curstname);
4926 sv_catpvn(PL_subname,"::",2);
4927 sv_catpvn(PL_subname,tmpbuf,len);
4934 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4935 PL_expect = XTERMBLOCK;
4936 attrful = XATTRTERM;
4937 sv_setpv(PL_subname,"?");
4941 if (key == KEY_format) {
4943 PL_lex_formbrack = PL_lex_brackets + 1;
4945 (void) force_word(PL_oldbufptr + tboffset, WORD,
4950 /* Look for a prototype */
4954 s = scan_str(s,FALSE,FALSE);
4956 Perl_croak(aTHX_ "Prototype not terminated");
4957 /* strip spaces and check for bad characters */
4958 d = SvPVX(PL_lex_stuff);
4961 for (p = d; *p; ++p) {
4964 if (!strchr("$@%*;[]&\\", *p))
4969 if (bad_proto && ckWARN(WARN_SYNTAX))
4970 Perl_warner(aTHX_ WARN_SYNTAX,
4971 "Illegal character in prototype for %s : %s",
4972 SvPVX(PL_subname), d);
4973 SvCUR(PL_lex_stuff) = tmp;
4981 if (*s == ':' && s[1] != ':')
4982 PL_expect = attrful;
4985 PL_nextval[PL_nexttoke].opval =
4986 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4987 PL_lex_stuff = Nullsv;
4991 sv_setpv(PL_subname,"__ANON__");
4994 (void) force_word(PL_oldbufptr + tboffset, WORD,
5003 LOP(OP_SYSTEM,XREF);
5006 LOP(OP_SYMLINK,XTERM);
5009 LOP(OP_SYSCALL,XTERM);
5012 LOP(OP_SYSOPEN,XTERM);
5015 LOP(OP_SYSSEEK,XTERM);
5018 LOP(OP_SYSREAD,XTERM);
5021 LOP(OP_SYSWRITE,XTERM);
5025 TERM(sublex_start());
5046 LOP(OP_TRUNCATE,XTERM);
5058 yylval.ival = CopLINE(PL_curcop);
5062 yylval.ival = CopLINE(PL_curcop);
5066 LOP(OP_UNLINK,XTERM);
5072 LOP(OP_UNPACK,XTERM);
5075 LOP(OP_UTIME,XTERM);
5081 LOP(OP_UNSHIFT,XTERM);
5084 if (PL_expect != XSTATE)
5085 yyerror("\"use\" not allowed in expression");
5087 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
5088 s = force_version(s, TRUE);
5089 if (*s == ';' || (s = skipspace(s), *s == ';')) {
5090 PL_nextval[PL_nexttoke].opval = Nullop;
5093 else if (*s == 'v') {
5094 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5095 s = force_version(s, FALSE);
5099 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5100 s = force_version(s, FALSE);
5112 yylval.ival = CopLINE(PL_curcop);
5116 PL_hints |= HINT_BLOCK_SCOPE;
5123 LOP(OP_WAITPID,XTERM);
5132 ctl_l[0] = toCTRL('L');
5134 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5137 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5142 if (PL_expect == XOPERATOR)
5148 yylval.ival = OP_XOR;
5153 TERM(sublex_start());
5158 #pragma segment Main
5162 S_pending_ident(pTHX)
5166 /* pit holds the identifier we read and pending_ident is reset */
5167 char pit = PL_pending_ident;
5168 PL_pending_ident = 0;
5170 DEBUG_T({ PerlIO_printf(Perl_debug_log,
5171 "### Tokener saw identifier '%s'\n", PL_tokenbuf); });
5173 /* if we're in a my(), we can't allow dynamics here.
5174 $foo'bar has already been turned into $foo::bar, so
5175 just check for colons.
5177 if it's a legal name, the OP is a PADANY.
5180 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
5181 if (strchr(PL_tokenbuf,':'))
5182 yyerror(Perl_form(aTHX_ "No package name allowed for "
5183 "variable %s in \"our\"",
5185 tmp = pad_allocmy(PL_tokenbuf);
5188 if (strchr(PL_tokenbuf,':'))
5189 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
5191 yylval.opval = newOP(OP_PADANY, 0);
5192 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
5198 build the ops for accesses to a my() variable.
5200 Deny my($a) or my($b) in a sort block, *if* $a or $b is
5201 then used in a comparison. This catches most, but not
5202 all cases. For instance, it catches
5203 sort { my($a); $a <=> $b }
5205 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
5206 (although why you'd do that is anyone's guess).
5209 if (!strchr(PL_tokenbuf,':')) {
5210 #ifdef USE_5005THREADS
5211 /* Check for single character per-thread SVs */
5212 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
5213 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
5214 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
5216 yylval.opval = newOP(OP_THREADSV, 0);
5217 yylval.opval->op_targ = tmp;
5220 #endif /* USE_5005THREADS */
5221 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
5222 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
5223 /* might be an "our" variable" */
5224 if (SvFLAGS(namesv) & SVpad_OUR) {
5225 /* build ops for a bareword */
5226 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
5227 sv_catpvn(sym, "::", 2);
5228 sv_catpv(sym, PL_tokenbuf+1);
5229 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
5230 yylval.opval->op_private = OPpCONST_ENTERED;
5231 gv_fetchpv(SvPVX(sym),
5233 ? (GV_ADDMULTI | GV_ADDINEVAL)
5236 ((PL_tokenbuf[0] == '$') ? SVt_PV
5237 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5242 /* if it's a sort block and they're naming $a or $b */
5243 if (PL_last_lop_op == OP_SORT &&
5244 PL_tokenbuf[0] == '$' &&
5245 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
5248 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
5249 d < PL_bufend && *d != '\n';
5252 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
5253 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
5259 yylval.opval = newOP(OP_PADANY, 0);
5260 yylval.opval->op_targ = tmp;
5266 Whine if they've said @foo in a doublequoted string,
5267 and @foo isn't a variable we can find in the symbol
5270 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
5271 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
5272 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
5273 && ckWARN(WARN_AMBIGUOUS))
5275 /* Downgraded from fatal to warning 20000522 mjd */
5276 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5277 "Possible unintended interpolation of %s in string",
5282 /* build ops for a bareword */
5283 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
5284 yylval.opval->op_private = OPpCONST_ENTERED;
5285 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
5286 ((PL_tokenbuf[0] == '$') ? SVt_PV
5287 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5293 Perl_keyword(pTHX_ register char *d, I32 len)
5298 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5299 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5300 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5301 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5302 if (strEQ(d,"__END__")) return KEY___END__;
5306 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5311 if (strEQ(d,"and")) return -KEY_and;
5312 if (strEQ(d,"abs")) return -KEY_abs;
5315 if (strEQ(d,"alarm")) return -KEY_alarm;
5316 if (strEQ(d,"atan2")) return -KEY_atan2;
5319 if (strEQ(d,"accept")) return -KEY_accept;
5324 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5327 if (strEQ(d,"bless")) return -KEY_bless;
5328 if (strEQ(d,"bind")) return -KEY_bind;
5329 if (strEQ(d,"binmode")) return -KEY_binmode;
5332 if (strEQ(d,"CORE")) return -KEY_CORE;
5333 if (strEQ(d,"CHECK")) return KEY_CHECK;
5338 if (strEQ(d,"cmp")) return -KEY_cmp;
5339 if (strEQ(d,"chr")) return -KEY_chr;
5340 if (strEQ(d,"cos")) return -KEY_cos;
5343 if (strEQ(d,"chop")) return -KEY_chop;
5346 if (strEQ(d,"close")) return -KEY_close;
5347 if (strEQ(d,"chdir")) return -KEY_chdir;
5348 if (strEQ(d,"chomp")) return -KEY_chomp;
5349 if (strEQ(d,"chmod")) return -KEY_chmod;
5350 if (strEQ(d,"chown")) return -KEY_chown;
5351 if (strEQ(d,"crypt")) return -KEY_crypt;
5354 if (strEQ(d,"chroot")) return -KEY_chroot;
5355 if (strEQ(d,"caller")) return -KEY_caller;
5358 if (strEQ(d,"connect")) return -KEY_connect;
5361 if (strEQ(d,"closedir")) return -KEY_closedir;
5362 if (strEQ(d,"continue")) return -KEY_continue;
5367 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5372 if (strEQ(d,"do")) return KEY_do;
5375 if (strEQ(d,"die")) return -KEY_die;
5378 if (strEQ(d,"dump")) return -KEY_dump;
5381 if (strEQ(d,"delete")) return KEY_delete;
5384 if (strEQ(d,"defined")) return KEY_defined;
5385 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5388 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5393 if (strEQ(d,"END")) return KEY_END;
5398 if (strEQ(d,"eq")) return -KEY_eq;
5401 if (strEQ(d,"eof")) return -KEY_eof;
5402 if (strEQ(d,"exp")) return -KEY_exp;
5405 if (strEQ(d,"else")) return KEY_else;
5406 if (strEQ(d,"exit")) return -KEY_exit;
5407 if (strEQ(d,"eval")) return KEY_eval;
5408 if (strEQ(d,"exec")) return -KEY_exec;
5409 if (strEQ(d,"each")) return -KEY_each;
5412 if (strEQ(d,"elsif")) return KEY_elsif;
5415 if (strEQ(d,"exists")) return KEY_exists;
5416 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5419 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5420 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5423 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5426 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5427 if (strEQ(d,"endservent")) return -KEY_endservent;
5430 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5437 if (strEQ(d,"for")) return KEY_for;
5440 if (strEQ(d,"fork")) return -KEY_fork;
5443 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5444 if (strEQ(d,"flock")) return -KEY_flock;
5447 if (strEQ(d,"format")) return KEY_format;
5448 if (strEQ(d,"fileno")) return -KEY_fileno;
5451 if (strEQ(d,"foreach")) return KEY_foreach;
5454 if (strEQ(d,"formline")) return -KEY_formline;
5459 if (strnEQ(d,"get",3)) {
5464 if (strEQ(d,"ppid")) return -KEY_getppid;
5465 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5468 if (strEQ(d,"pwent")) return -KEY_getpwent;
5469 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5470 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5473 if (strEQ(d,"peername")) return -KEY_getpeername;
5474 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5475 if (strEQ(d,"priority")) return -KEY_getpriority;
5478 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5481 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5485 else if (*d == 'h') {
5486 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5487 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5488 if (strEQ(d,"hostent")) return -KEY_gethostent;
5490 else if (*d == 'n') {
5491 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5492 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5493 if (strEQ(d,"netent")) return -KEY_getnetent;
5495 else if (*d == 's') {
5496 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5497 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5498 if (strEQ(d,"servent")) return -KEY_getservent;
5499 if (strEQ(d,"sockname")) return -KEY_getsockname;
5500 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5502 else if (*d == 'g') {
5503 if (strEQ(d,"grent")) return -KEY_getgrent;
5504 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5505 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5507 else if (*d == 'l') {
5508 if (strEQ(d,"login")) return -KEY_getlogin;
5510 else if (strEQ(d,"c")) return -KEY_getc;
5515 if (strEQ(d,"gt")) return -KEY_gt;
5516 if (strEQ(d,"ge")) return -KEY_ge;
5519 if (strEQ(d,"grep")) return KEY_grep;
5520 if (strEQ(d,"goto")) return KEY_goto;
5521 if (strEQ(d,"glob")) return KEY_glob;
5524 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5529 if (strEQ(d,"hex")) return -KEY_hex;
5532 if (strEQ(d,"INIT")) return KEY_INIT;
5537 if (strEQ(d,"if")) return KEY_if;
5540 if (strEQ(d,"int")) return -KEY_int;
5543 if (strEQ(d,"index")) return -KEY_index;
5544 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5549 if (strEQ(d,"join")) return -KEY_join;
5553 if (strEQ(d,"keys")) return -KEY_keys;
5554 if (strEQ(d,"kill")) return -KEY_kill;
5560 if (strEQ(d,"lt")) return -KEY_lt;
5561 if (strEQ(d,"le")) return -KEY_le;
5562 if (strEQ(d,"lc")) return -KEY_lc;
5565 if (strEQ(d,"log")) return -KEY_log;
5568 if (strEQ(d,"last")) return KEY_last;
5569 if (strEQ(d,"link")) return -KEY_link;
5570 if (strEQ(d,"lock")) return -KEY_lock;
5573 if (strEQ(d,"local")) return KEY_local;
5574 if (strEQ(d,"lstat")) return -KEY_lstat;
5577 if (strEQ(d,"length")) return -KEY_length;
5578 if (strEQ(d,"listen")) return -KEY_listen;
5581 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5584 if (strEQ(d,"localtime")) return -KEY_localtime;
5590 case 1: return KEY_m;
5592 if (strEQ(d,"my")) return KEY_my;
5595 if (strEQ(d,"map")) return KEY_map;
5598 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5601 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5602 if (strEQ(d,"msgget")) return -KEY_msgget;
5603 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5604 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5609 if (strEQ(d,"next")) return KEY_next;
5610 if (strEQ(d,"ne")) return -KEY_ne;
5611 if (strEQ(d,"not")) return -KEY_not;
5612 if (strEQ(d,"no")) return KEY_no;
5617 if (strEQ(d,"or")) return -KEY_or;
5620 if (strEQ(d,"ord")) return -KEY_ord;
5621 if (strEQ(d,"oct")) return -KEY_oct;
5622 if (strEQ(d,"our")) return KEY_our;
5625 if (strEQ(d,"open")) return -KEY_open;
5628 if (strEQ(d,"opendir")) return -KEY_opendir;
5635 if (strEQ(d,"pop")) return -KEY_pop;
5636 if (strEQ(d,"pos")) return KEY_pos;
5639 if (strEQ(d,"push")) return -KEY_push;
5640 if (strEQ(d,"pack")) return -KEY_pack;
5641 if (strEQ(d,"pipe")) return -KEY_pipe;
5644 if (strEQ(d,"print")) return KEY_print;
5647 if (strEQ(d,"printf")) return KEY_printf;
5650 if (strEQ(d,"package")) return KEY_package;
5653 if (strEQ(d,"prototype")) return KEY_prototype;
5658 if (strEQ(d,"q")) return KEY_q;
5659 if (strEQ(d,"qr")) return KEY_qr;
5660 if (strEQ(d,"qq")) return KEY_qq;
5661 if (strEQ(d,"qw")) return KEY_qw;
5662 if (strEQ(d,"qx")) return KEY_qx;
5664 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5669 if (strEQ(d,"ref")) return -KEY_ref;
5672 if (strEQ(d,"read")) return -KEY_read;
5673 if (strEQ(d,"rand")) return -KEY_rand;
5674 if (strEQ(d,"recv")) return -KEY_recv;
5675 if (strEQ(d,"redo")) return KEY_redo;
5678 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5679 if (strEQ(d,"reset")) return -KEY_reset;
5682 if (strEQ(d,"return")) return KEY_return;
5683 if (strEQ(d,"rename")) return -KEY_rename;
5684 if (strEQ(d,"rindex")) return -KEY_rindex;
5687 if (strEQ(d,"require")) return KEY_require;
5688 if (strEQ(d,"reverse")) return -KEY_reverse;
5689 if (strEQ(d,"readdir")) return -KEY_readdir;
5692 if (strEQ(d,"readlink")) return -KEY_readlink;
5693 if (strEQ(d,"readline")) return -KEY_readline;
5694 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5697 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5703 case 0: return KEY_s;
5705 if (strEQ(d,"scalar")) return KEY_scalar;
5710 if (strEQ(d,"seek")) return -KEY_seek;
5711 if (strEQ(d,"send")) return -KEY_send;
5714 if (strEQ(d,"semop")) return -KEY_semop;
5717 if (strEQ(d,"select")) return -KEY_select;
5718 if (strEQ(d,"semctl")) return -KEY_semctl;
5719 if (strEQ(d,"semget")) return -KEY_semget;
5722 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5723 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5726 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5727 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5730 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5733 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5734 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5735 if (strEQ(d,"setservent")) return -KEY_setservent;
5738 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5739 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5746 if (strEQ(d,"shift")) return -KEY_shift;
5749 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5750 if (strEQ(d,"shmget")) return -KEY_shmget;
5753 if (strEQ(d,"shmread")) return -KEY_shmread;
5756 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5757 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5762 if (strEQ(d,"sin")) return -KEY_sin;
5765 if (strEQ(d,"sleep")) return -KEY_sleep;
5768 if (strEQ(d,"sort")) return KEY_sort;
5769 if (strEQ(d,"socket")) return -KEY_socket;
5770 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5773 if (strEQ(d,"split")) return KEY_split;
5774 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5775 if (strEQ(d,"splice")) return -KEY_splice;
5778 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5781 if (strEQ(d,"srand")) return -KEY_srand;
5784 if (strEQ(d,"stat")) return -KEY_stat;
5785 if (strEQ(d,"study")) return KEY_study;
5788 if (strEQ(d,"substr")) return -KEY_substr;
5789 if (strEQ(d,"sub")) return KEY_sub;
5794 if (strEQ(d,"system")) return -KEY_system;
5797 if (strEQ(d,"symlink")) return -KEY_symlink;
5798 if (strEQ(d,"syscall")) return -KEY_syscall;
5799 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5800 if (strEQ(d,"sysread")) return -KEY_sysread;
5801 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5804 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5813 if (strEQ(d,"tr")) return KEY_tr;
5816 if (strEQ(d,"tie")) return KEY_tie;
5819 if (strEQ(d,"tell")) return -KEY_tell;
5820 if (strEQ(d,"tied")) return KEY_tied;
5821 if (strEQ(d,"time")) return -KEY_time;
5824 if (strEQ(d,"times")) return -KEY_times;
5827 if (strEQ(d,"telldir")) return -KEY_telldir;
5830 if (strEQ(d,"truncate")) return -KEY_truncate;
5837 if (strEQ(d,"uc")) return -KEY_uc;
5840 if (strEQ(d,"use")) return KEY_use;
5843 if (strEQ(d,"undef")) return KEY_undef;
5844 if (strEQ(d,"until")) return KEY_until;
5845 if (strEQ(d,"untie")) return KEY_untie;
5846 if (strEQ(d,"utime")) return -KEY_utime;
5847 if (strEQ(d,"umask")) return -KEY_umask;
5850 if (strEQ(d,"unless")) return KEY_unless;
5851 if (strEQ(d,"unpack")) return -KEY_unpack;
5852 if (strEQ(d,"unlink")) return -KEY_unlink;
5855 if (strEQ(d,"unshift")) return -KEY_unshift;
5856 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5861 if (strEQ(d,"values")) return -KEY_values;
5862 if (strEQ(d,"vec")) return -KEY_vec;
5867 if (strEQ(d,"warn")) return -KEY_warn;
5868 if (strEQ(d,"wait")) return -KEY_wait;
5871 if (strEQ(d,"while")) return KEY_while;
5872 if (strEQ(d,"write")) return -KEY_write;
5875 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5878 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5883 if (len == 1) return -KEY_x;
5884 if (strEQ(d,"xor")) return -KEY_xor;
5887 if (len == 1) return KEY_y;
5896 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5900 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5901 if (ckWARN(WARN_SYNTAX)) {
5903 for (w = s+2; *w && level; w++) {
5910 for (; *w && isSPACE(*w); w++) ;
5911 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5912 Perl_warner(aTHX_ WARN_SYNTAX,
5913 "%s (...) interpreted as function",name);
5916 while (s < PL_bufend && isSPACE(*s))
5920 while (s < PL_bufend && isSPACE(*s))
5922 if (isIDFIRST_lazy_if(s,UTF)) {
5924 while (isALNUM_lazy_if(s,UTF))
5926 while (s < PL_bufend && isSPACE(*s))
5931 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5935 Perl_croak(aTHX_ "No comma allowed after %s", what);
5940 /* Either returns sv, or mortalizes sv and returns a new SV*.
5941 Best used as sv=new_constant(..., sv, ...).
5942 If s, pv are NULL, calls subroutine with one argument,
5943 and type is used with error messages only. */
5946 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5950 HV *table = GvHV(PL_hintgv); /* ^H */
5954 const char *why1, *why2, *why3;
5956 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
5959 why2 = strEQ(key,"charnames")
5960 ? "(possibly a missing \"use charnames ...\")"
5962 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
5963 (type ? type: "undef"), why2);
5965 /* This is convoluted and evil ("goto considered harmful")
5966 * but I do not understand the intricacies of all the different
5967 * failure modes of %^H in here. The goal here is to make
5968 * the most probable error message user-friendly. --jhi */
5973 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
5974 (type ? type: "undef"), why1, why2, why3);
5976 yyerror(SvPVX(msg));
5980 cvp = hv_fetch(table, key, strlen(key), FALSE);
5981 if (!cvp || !SvOK(*cvp)) {
5984 why3 = "} is not defined";
5987 sv_2mortal(sv); /* Parent created it permanently */
5990 pv = sv_2mortal(newSVpvn(s, len));
5992 typesv = sv_2mortal(newSVpv(type, 0));
5994 typesv = &PL_sv_undef;
5996 PUSHSTACKi(PERLSI_OVERLOAD);
6008 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
6012 /* Check the eval first */
6013 if (!PL_in_eval && SvTRUE(ERRSV)) {
6015 sv_catpv(ERRSV, "Propagated");
6016 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
6018 res = SvREFCNT_inc(sv);
6022 (void)SvREFCNT_inc(res);
6031 why1 = "Call to &{$^H{";
6033 why3 = "}} did not return a defined value";
6042 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
6044 register char *d = dest;
6045 register char *e = d + destlen - 3; /* two-character token, ending NUL */
6048 Perl_croak(aTHX_ ident_too_long);
6049 if (isALNUM(*s)) /* UTF handled below */
6051 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
6056 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
6060 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6061 char *t = s + UTF8SKIP(s);
6062 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6064 if (d + (t - s) > e)
6065 Perl_croak(aTHX_ ident_too_long);
6066 Copy(s, d, t - s, char);
6079 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
6089 e = d + destlen - 3; /* two-character token, ending NUL */
6091 while (isDIGIT(*s)) {
6093 Perl_croak(aTHX_ ident_too_long);
6100 Perl_croak(aTHX_ ident_too_long);
6101 if (isALNUM(*s)) /* UTF handled below */
6103 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
6108 else if (*s == ':' && s[1] == ':') {
6112 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6113 char *t = s + UTF8SKIP(s);
6114 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6116 if (d + (t - s) > e)
6117 Perl_croak(aTHX_ ident_too_long);
6118 Copy(s, d, t - s, char);
6129 if (PL_lex_state != LEX_NORMAL)
6130 PL_lex_state = LEX_INTERPENDMAYBE;
6133 if (*s == '$' && s[1] &&
6134 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
6147 if (*d == '^' && *s && isCONTROLVAR(*s)) {
6152 if (isSPACE(s[-1])) {
6155 if (!SPACE_OR_TAB(ch)) {
6161 if (isIDFIRST_lazy_if(d,UTF)) {
6165 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
6167 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
6170 Copy(s, d, e - s, char);
6175 while ((isALNUM(*s) || *s == ':') && d < e)
6178 Perl_croak(aTHX_ ident_too_long);
6181 while (s < send && SPACE_OR_TAB(*s)) s++;
6182 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
6183 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
6184 const char *brack = *s == '[' ? "[...]" : "{...}";
6185 Perl_warner(aTHX_ WARN_AMBIGUOUS,
6186 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
6187 funny, dest, brack, funny, dest, brack);
6190 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
6194 /* Handle extended ${^Foo} variables
6195 * 1999-02-27 mjd-perl-patch@plover.com */
6196 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
6200 while (isALNUM(*s) && d < e) {
6204 Perl_croak(aTHX_ ident_too_long);
6209 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
6210 PL_lex_state = LEX_INTERPEND;
6213 if (PL_lex_state == LEX_NORMAL) {
6214 if (ckWARN(WARN_AMBIGUOUS) &&
6215 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
6217 Perl_warner(aTHX_ WARN_AMBIGUOUS,
6218 "Ambiguous use of %c{%s} resolved to %c%s",
6219 funny, dest, funny, dest);
6224 s = bracket; /* let the parser handle it */
6228 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
6229 PL_lex_state = LEX_INTERPEND;
6234 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
6239 *pmfl |= PMf_GLOBAL;
6241 *pmfl |= PMf_CONTINUE;
6245 *pmfl |= PMf_MULTILINE;
6247 *pmfl |= PMf_SINGLELINE;
6249 *pmfl |= PMf_EXTENDED;
6253 S_scan_pat(pTHX_ char *start, I32 type)
6258 s = scan_str(start,FALSE,FALSE);
6260 Perl_croak(aTHX_ "Search pattern not terminated");
6262 pm = (PMOP*)newPMOP(type, 0);
6263 if (PL_multi_open == '?')
6264 pm->op_pmflags |= PMf_ONCE;
6266 while (*s && strchr("iomsx", *s))
6267 pmflag(&pm->op_pmflags,*s++);
6270 while (*s && strchr("iogcmsx", *s))
6271 pmflag(&pm->op_pmflags,*s++);
6273 pm->op_pmpermflags = pm->op_pmflags;
6275 PL_lex_op = (OP*)pm;
6276 yylval.ival = OP_MATCH;
6281 S_scan_subst(pTHX_ char *start)
6288 yylval.ival = OP_NULL;
6290 s = scan_str(start,FALSE,FALSE);
6293 Perl_croak(aTHX_ "Substitution pattern not terminated");
6295 if (s[-1] == PL_multi_open)
6298 first_start = PL_multi_start;
6299 s = scan_str(s,FALSE,FALSE);
6302 SvREFCNT_dec(PL_lex_stuff);
6303 PL_lex_stuff = Nullsv;
6305 Perl_croak(aTHX_ "Substitution replacement not terminated");
6307 PL_multi_start = first_start; /* so whole substitution is taken together */
6309 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6315 else if (strchr("iogcmsx", *s))
6316 pmflag(&pm->op_pmflags,*s++);
6323 PL_sublex_info.super_bufptr = s;
6324 PL_sublex_info.super_bufend = PL_bufend;
6326 pm->op_pmflags |= PMf_EVAL;
6327 repl = newSVpvn("",0);
6329 sv_catpv(repl, es ? "eval " : "do ");
6330 sv_catpvn(repl, "{ ", 2);
6331 sv_catsv(repl, PL_lex_repl);
6332 sv_catpvn(repl, " };", 2);
6334 SvREFCNT_dec(PL_lex_repl);
6338 pm->op_pmpermflags = pm->op_pmflags;
6339 PL_lex_op = (OP*)pm;
6340 yylval.ival = OP_SUBST;
6345 S_scan_trans(pTHX_ char *start)
6354 yylval.ival = OP_NULL;
6356 s = scan_str(start,FALSE,FALSE);
6358 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6359 if (s[-1] == PL_multi_open)
6362 s = scan_str(s,FALSE,FALSE);
6365 SvREFCNT_dec(PL_lex_stuff);
6366 PL_lex_stuff = Nullsv;
6368 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6371 complement = del = squash = 0;
6372 while (strchr("cds", *s)) {
6374 complement = OPpTRANS_COMPLEMENT;
6376 del = OPpTRANS_DELETE;
6378 squash = OPpTRANS_SQUASH;
6382 New(803, tbl, complement&&!del?258:256, short);
6383 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6384 o->op_private = del|squash|complement|
6385 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
6386 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
6389 yylval.ival = OP_TRANS;
6394 S_scan_heredoc(pTHX_ register char *s)
6397 I32 op_type = OP_SCALAR;
6404 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6408 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6411 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6412 if (*peek && strchr("`'\"",*peek)) {
6415 s = delimcpy(d, e, s, PL_bufend, term, &len);
6425 if (!isALNUM_lazy_if(s,UTF))
6426 deprecate("bare << to mean <<\"\"");
6427 for (; isALNUM_lazy_if(s,UTF); s++) {
6432 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6433 Perl_croak(aTHX_ "Delimiter for here document is too long");
6436 len = d - PL_tokenbuf;
6437 #ifndef PERL_STRICT_CR
6438 d = strchr(s, '\r');
6442 while (s < PL_bufend) {
6448 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6457 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6462 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6463 herewas = newSVpvn(s,PL_bufend-s);
6465 s--, herewas = newSVpvn(s,d-s);
6466 s += SvCUR(herewas);
6468 tmpstr = NEWSV(87,79);
6469 sv_upgrade(tmpstr, SVt_PVIV);
6474 else if (term == '`') {
6475 op_type = OP_BACKTICK;
6476 SvIVX(tmpstr) = '\\';
6480 PL_multi_start = CopLINE(PL_curcop);
6481 PL_multi_open = PL_multi_close = '<';
6482 term = *PL_tokenbuf;
6483 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6484 char *bufptr = PL_sublex_info.super_bufptr;
6485 char *bufend = PL_sublex_info.super_bufend;
6486 char *olds = s - SvCUR(herewas);
6487 s = strchr(bufptr, '\n');
6491 while (s < bufend &&
6492 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6494 CopLINE_inc(PL_curcop);
6497 CopLINE_set(PL_curcop, PL_multi_start);
6498 missingterm(PL_tokenbuf);
6500 sv_setpvn(herewas,bufptr,d-bufptr+1);
6501 sv_setpvn(tmpstr,d+1,s-d);
6503 sv_catpvn(herewas,s,bufend-s);
6504 (void)strcpy(bufptr,SvPVX(herewas));
6511 while (s < PL_bufend &&
6512 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6514 CopLINE_inc(PL_curcop);
6516 if (s >= PL_bufend) {
6517 CopLINE_set(PL_curcop, PL_multi_start);
6518 missingterm(PL_tokenbuf);
6520 sv_setpvn(tmpstr,d+1,s-d);
6522 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6524 sv_catpvn(herewas,s,PL_bufend-s);
6525 sv_setsv(PL_linestr,herewas);
6526 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6527 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6528 PL_last_lop = PL_last_uni = Nullch;
6531 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6532 while (s >= PL_bufend) { /* multiple line string? */
6534 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6535 CopLINE_set(PL_curcop, PL_multi_start);
6536 missingterm(PL_tokenbuf);
6538 CopLINE_inc(PL_curcop);
6539 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6540 PL_last_lop = PL_last_uni = Nullch;
6541 #ifndef PERL_STRICT_CR
6542 if (PL_bufend - PL_linestart >= 2) {
6543 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6544 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6546 PL_bufend[-2] = '\n';
6548 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6550 else if (PL_bufend[-1] == '\r')
6551 PL_bufend[-1] = '\n';
6553 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6554 PL_bufend[-1] = '\n';
6556 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6557 SV *sv = NEWSV(88,0);
6559 sv_upgrade(sv, SVt_PVMG);
6560 sv_setsv(sv,PL_linestr);
6563 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6565 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6568 sv_catsv(PL_linestr,herewas);
6569 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6573 sv_catsv(tmpstr,PL_linestr);
6578 PL_multi_end = CopLINE(PL_curcop);
6579 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6580 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6581 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6583 SvREFCNT_dec(herewas);
6584 if (UTF && !IN_BYTES && is_utf8_string((U8*)SvPVX(tmpstr), SvCUR(tmpstr)))
6586 PL_lex_stuff = tmpstr;
6587 yylval.ival = op_type;
6592 takes: current position in input buffer
6593 returns: new position in input buffer
6594 side-effects: yylval and lex_op are set.
6599 <FH> read from filehandle
6600 <pkg::FH> read from package qualified filehandle
6601 <pkg'FH> read from package qualified filehandle
6602 <$fh> read from filehandle in $fh
6608 S_scan_inputsymbol(pTHX_ char *start)
6610 register char *s = start; /* current position in buffer */
6616 d = PL_tokenbuf; /* start of temp holding space */
6617 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6618 end = strchr(s, '\n');
6621 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6623 /* die if we didn't have space for the contents of the <>,
6624 or if it didn't end, or if we see a newline
6627 if (len >= sizeof PL_tokenbuf)
6628 Perl_croak(aTHX_ "Excessively long <> operator");
6630 Perl_croak(aTHX_ "Unterminated <> operator");
6635 Remember, only scalar variables are interpreted as filehandles by
6636 this code. Anything more complex (e.g., <$fh{$num}>) will be
6637 treated as a glob() call.
6638 This code makes use of the fact that except for the $ at the front,
6639 a scalar variable and a filehandle look the same.
6641 if (*d == '$' && d[1]) d++;
6643 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6644 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6647 /* If we've tried to read what we allow filehandles to look like, and
6648 there's still text left, then it must be a glob() and not a getline.
6649 Use scan_str to pull out the stuff between the <> and treat it
6650 as nothing more than a string.
6653 if (d - PL_tokenbuf != len) {
6654 yylval.ival = OP_GLOB;
6656 s = scan_str(start,FALSE,FALSE);
6658 Perl_croak(aTHX_ "Glob not terminated");
6662 bool readline_overriden = FALSE;
6663 GV *gv_readline = Nullgv;
6665 /* we're in a filehandle read situation */
6668 /* turn <> into <ARGV> */
6670 (void)strcpy(d,"ARGV");
6672 /* Check whether readline() is overriden */
6673 if (((gv_readline = gv_fetchpv("readline", FALSE, SVt_PVCV))
6674 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
6676 ((gvp = (GV**)hv_fetch(PL_globalstash, "readline", 8, FALSE))
6677 && (gv_readline = *gvp) != (GV*)&PL_sv_undef
6678 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
6679 readline_overriden = TRUE;
6681 /* if <$fh>, create the ops to turn the variable into a
6687 /* try to find it in the pad for this block, otherwise find
6688 add symbol table ops
6690 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6691 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
6692 if (SvFLAGS(namesv) & SVpad_OUR) {
6693 SV *sym = sv_2mortal(newSVpv(HvNAME(GvSTASH(namesv)),0));
6694 sv_catpvn(sym, "::", 2);
6700 OP *o = newOP(OP_PADSV, 0);
6702 PL_lex_op = readline_overriden
6703 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6704 append_elem(OP_LIST, o,
6705 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
6706 : (OP*)newUNOP(OP_READLINE, 0, o);
6715 ? (GV_ADDMULTI | GV_ADDINEVAL)
6718 PL_lex_op = readline_overriden
6719 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6720 append_elem(OP_LIST,
6721 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
6722 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
6723 : (OP*)newUNOP(OP_READLINE, 0,
6724 newUNOP(OP_RV2SV, 0,
6725 newGVOP(OP_GV, 0, gv)));
6727 PL_lex_op->op_flags |= OPf_SPECIAL;
6728 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6729 yylval.ival = OP_NULL;
6732 /* If it's none of the above, it must be a literal filehandle
6733 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6735 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6736 PL_lex_op = readline_overriden
6737 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6738 append_elem(OP_LIST,
6739 newGVOP(OP_GV, 0, gv),
6740 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
6741 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6742 yylval.ival = OP_NULL;
6751 takes: start position in buffer
6752 keep_quoted preserve \ on the embedded delimiter(s)
6753 keep_delims preserve the delimiters around the string
6754 returns: position to continue reading from buffer
6755 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6756 updates the read buffer.
6758 This subroutine pulls a string out of the input. It is called for:
6759 q single quotes q(literal text)
6760 ' single quotes 'literal text'
6761 qq double quotes qq(interpolate $here please)
6762 " double quotes "interpolate $here please"
6763 qx backticks qx(/bin/ls -l)
6764 ` backticks `/bin/ls -l`
6765 qw quote words @EXPORT_OK = qw( func() $spam )
6766 m// regexp match m/this/
6767 s/// regexp substitute s/this/that/
6768 tr/// string transliterate tr/this/that/
6769 y/// string transliterate y/this/that/
6770 ($*@) sub prototypes sub foo ($)
6771 (stuff) sub attr parameters sub foo : attr(stuff)
6772 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6774 In most of these cases (all but <>, patterns and transliterate)
6775 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6776 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6777 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6780 It skips whitespace before the string starts, and treats the first
6781 character as the delimiter. If the delimiter is one of ([{< then
6782 the corresponding "close" character )]}> is used as the closing
6783 delimiter. It allows quoting of delimiters, and if the string has
6784 balanced delimiters ([{<>}]) it allows nesting.
6786 On success, the SV with the resulting string is put into lex_stuff or,
6787 if that is already non-NULL, into lex_repl. The second case occurs only
6788 when parsing the RHS of the special constructs s/// and tr/// (y///).
6789 For convenience, the terminating delimiter character is stuffed into
6794 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6796 SV *sv; /* scalar value: string */
6797 char *tmps; /* temp string, used for delimiter matching */
6798 register char *s = start; /* current position in the buffer */
6799 register char term; /* terminating character */
6800 register char *to; /* current position in the sv's data */
6801 I32 brackets = 1; /* bracket nesting level */
6802 bool has_utf8 = FALSE; /* is there any utf8 content? */
6804 /* skip space before the delimiter */
6808 /* mark where we are, in case we need to report errors */
6811 /* after skipping whitespace, the next character is the terminator */
6813 if (!UTF8_IS_INVARIANT((U8)term) && UTF)
6816 /* mark where we are */
6817 PL_multi_start = CopLINE(PL_curcop);
6818 PL_multi_open = term;
6820 /* find corresponding closing delimiter */
6821 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6823 PL_multi_close = term;
6825 /* create a new SV to hold the contents. 87 is leak category, I'm
6826 assuming. 79 is the SV's initial length. What a random number. */
6828 sv_upgrade(sv, SVt_PVIV);
6830 (void)SvPOK_only(sv); /* validate pointer */
6832 /* move past delimiter and try to read a complete string */
6834 sv_catpvn(sv, s, 1);
6837 /* extend sv if need be */
6838 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6839 /* set 'to' to the next character in the sv's string */
6840 to = SvPVX(sv)+SvCUR(sv);
6842 /* if open delimiter is the close delimiter read unbridle */
6843 if (PL_multi_open == PL_multi_close) {
6844 for (; s < PL_bufend; s++,to++) {
6845 /* embedded newlines increment the current line number */
6846 if (*s == '\n' && !PL_rsfp)
6847 CopLINE_inc(PL_curcop);
6848 /* handle quoted delimiters */
6849 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6850 if (!keep_quoted && s[1] == term)
6852 /* any other quotes are simply copied straight through */
6856 /* terminate when run out of buffer (the for() condition), or
6857 have found the terminator */
6858 else if (*s == term)
6860 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
6866 /* if the terminator isn't the same as the start character (e.g.,
6867 matched brackets), we have to allow more in the quoting, and
6868 be prepared for nested brackets.
6871 /* read until we run out of string, or we find the terminator */
6872 for (; s < PL_bufend; s++,to++) {
6873 /* embedded newlines increment the line count */
6874 if (*s == '\n' && !PL_rsfp)
6875 CopLINE_inc(PL_curcop);
6876 /* backslashes can escape the open or closing characters */
6877 if (*s == '\\' && s+1 < PL_bufend) {
6879 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6884 /* allow nested opens and closes */
6885 else if (*s == PL_multi_close && --brackets <= 0)
6887 else if (*s == PL_multi_open)
6889 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
6894 /* terminate the copied string and update the sv's end-of-string */
6896 SvCUR_set(sv, to - SvPVX(sv));
6899 * this next chunk reads more into the buffer if we're not done yet
6903 break; /* handle case where we are done yet :-) */
6905 #ifndef PERL_STRICT_CR
6906 if (to - SvPVX(sv) >= 2) {
6907 if ((to[-2] == '\r' && to[-1] == '\n') ||
6908 (to[-2] == '\n' && to[-1] == '\r'))
6912 SvCUR_set(sv, to - SvPVX(sv));
6914 else if (to[-1] == '\r')
6917 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6921 /* if we're out of file, or a read fails, bail and reset the current
6922 line marker so we can report where the unterminated string began
6925 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6927 CopLINE_set(PL_curcop, PL_multi_start);
6930 /* we read a line, so increment our line counter */
6931 CopLINE_inc(PL_curcop);
6933 /* update debugger info */
6934 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6935 SV *sv = NEWSV(88,0);
6937 sv_upgrade(sv, SVt_PVMG);
6938 sv_setsv(sv,PL_linestr);
6941 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6944 /* having changed the buffer, we must update PL_bufend */
6945 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6946 PL_last_lop = PL_last_uni = Nullch;
6949 /* at this point, we have successfully read the delimited string */
6952 sv_catpvn(sv, s, 1);
6955 PL_multi_end = CopLINE(PL_curcop);
6958 /* if we allocated too much space, give some back */
6959 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6960 SvLEN_set(sv, SvCUR(sv) + 1);
6961 Renew(SvPVX(sv), SvLEN(sv), char);
6964 /* decide whether this is the first or second quoted string we've read
6977 takes: pointer to position in buffer
6978 returns: pointer to new position in buffer
6979 side-effects: builds ops for the constant in yylval.op
6981 Read a number in any of the formats that Perl accepts:
6983 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
6984 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
6987 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
6989 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6992 If it reads a number without a decimal point or an exponent, it will
6993 try converting the number to an integer and see if it can do so
6994 without loss of precision.
6998 Perl_scan_num(pTHX_ char *start, YYSTYPE* lvalp)
7000 register char *s = start; /* current position in buffer */
7001 register char *d; /* destination in temp buffer */
7002 register char *e; /* end of temp buffer */
7003 NV nv; /* number read, as a double */
7004 SV *sv = Nullsv; /* place to put the converted number */
7005 bool floatit; /* boolean: int or float? */
7006 char *lastub = 0; /* position of last underbar */
7007 static char number_too_long[] = "Number too long";
7009 /* We use the first character to decide what type of number this is */
7013 Perl_croak(aTHX_ "panic: scan_num");
7015 /* if it starts with a 0, it could be an octal number, a decimal in
7016 0.13 disguise, or a hexadecimal number, or a binary number. */
7020 u holds the "number so far"
7021 shift the power of 2 of the base
7022 (hex == 4, octal == 3, binary == 1)
7023 overflowed was the number more than we can hold?
7025 Shift is used when we add a digit. It also serves as an "are
7026 we in octal/hex/binary?" indicator to disallow hex characters
7032 bool overflowed = FALSE;
7033 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
7034 static char* bases[5] = { "", "binary", "", "octal",
7036 static char* Bases[5] = { "", "Binary", "", "Octal",
7038 static char *maxima[5] = { "",
7039 "0b11111111111111111111111111111111",
7043 char *base, *Base, *max;
7049 } else if (s[1] == 'b') {
7053 /* check for a decimal in disguise */
7054 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
7056 /* so it must be octal */
7063 if (ckWARN(WARN_SYNTAX))
7064 Perl_warner(aTHX_ WARN_SYNTAX,
7065 "Misplaced _ in number");
7069 base = bases[shift];
7070 Base = Bases[shift];
7071 max = maxima[shift];
7073 /* read the rest of the number */
7075 /* x is used in the overflow test,
7076 b is the digit we're adding on. */
7081 /* if we don't mention it, we're done */
7085 /* _ are ignored -- but warned about if consecutive */
7087 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7088 Perl_warner(aTHX_ WARN_SYNTAX,
7089 "Misplaced _ in number");
7093 /* 8 and 9 are not octal */
7096 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
7100 case '2': case '3': case '4':
7101 case '5': case '6': case '7':
7103 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
7107 b = *s++ & 15; /* ASCII digit -> value of digit */
7111 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
7112 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
7113 /* make sure they said 0x */
7118 /* Prepare to put the digit we have onto the end
7119 of the number so far. We check for overflows.
7124 x = u << shift; /* make room for the digit */
7126 if ((x >> shift) != u
7127 && !(PL_hints & HINT_NEW_BINARY)) {
7130 if (ckWARN_d(WARN_OVERFLOW))
7131 Perl_warner(aTHX_ WARN_OVERFLOW,
7132 "Integer overflow in %s number",
7135 u = x | b; /* add the digit to the end */
7138 n *= nvshift[shift];
7139 /* If an NV has not enough bits in its
7140 * mantissa to represent an UV this summing of
7141 * small low-order numbers is a waste of time
7142 * (because the NV cannot preserve the
7143 * low-order bits anyway): we could just
7144 * remember when did we overflow and in the
7145 * end just multiply n by the right
7153 /* if we get here, we had success: make a scalar value from
7158 /* final misplaced underbar check */
7160 if (ckWARN(WARN_SYNTAX))
7161 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
7166 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
7167 Perl_warner(aTHX_ WARN_PORTABLE,
7168 "%s number > %s non-portable",
7174 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
7175 Perl_warner(aTHX_ WARN_PORTABLE,
7176 "%s number > %s non-portable",
7181 if (PL_hints & HINT_NEW_BINARY)
7182 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
7187 handle decimal numbers.
7188 we're also sent here when we read a 0 as the first digit
7190 case '1': case '2': case '3': case '4': case '5':
7191 case '6': case '7': case '8': case '9': case '.':
7194 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
7197 /* read next group of digits and _ and copy into d */
7198 while (isDIGIT(*s) || *s == '_') {
7199 /* skip underscores, checking for misplaced ones
7203 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7204 Perl_warner(aTHX_ WARN_SYNTAX,
7205 "Misplaced _ in number");
7209 /* check for end of fixed-length buffer */
7211 Perl_croak(aTHX_ number_too_long);
7212 /* if we're ok, copy the character */
7217 /* final misplaced underbar check */
7218 if (lastub && s == lastub + 1) {
7219 if (ckWARN(WARN_SYNTAX))
7220 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
7223 /* read a decimal portion if there is one. avoid
7224 3..5 being interpreted as the number 3. followed
7227 if (*s == '.' && s[1] != '.') {
7232 if (ckWARN(WARN_SYNTAX))
7233 Perl_warner(aTHX_ WARN_SYNTAX,
7234 "Misplaced _ in number");
7238 /* copy, ignoring underbars, until we run out of digits.
7240 for (; isDIGIT(*s) || *s == '_'; s++) {
7241 /* fixed length buffer check */
7243 Perl_croak(aTHX_ number_too_long);
7245 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7246 Perl_warner(aTHX_ WARN_SYNTAX,
7247 "Misplaced _ in number");
7253 /* fractional part ending in underbar? */
7255 if (ckWARN(WARN_SYNTAX))
7256 Perl_warner(aTHX_ WARN_SYNTAX,
7257 "Misplaced _ in number");
7259 if (*s == '.' && isDIGIT(s[1])) {
7260 /* oops, it's really a v-string, but without the "v" */
7266 /* read exponent part, if present */
7267 if (*s && strchr("eE",*s) && strchr("+-0123456789_", s[1])) {
7271 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
7272 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
7274 /* stray preinitial _ */
7276 if (ckWARN(WARN_SYNTAX))
7277 Perl_warner(aTHX_ WARN_SYNTAX,
7278 "Misplaced _ in number");
7282 /* allow positive or negative exponent */
7283 if (*s == '+' || *s == '-')
7286 /* stray initial _ */
7288 if (ckWARN(WARN_SYNTAX))
7289 Perl_warner(aTHX_ WARN_SYNTAX,
7290 "Misplaced _ in number");
7294 /* read digits of exponent */
7295 while (isDIGIT(*s) || *s == '_') {
7298 Perl_croak(aTHX_ number_too_long);
7302 if (ckWARN(WARN_SYNTAX) &&
7303 ((lastub && s == lastub + 1) ||
7304 (!isDIGIT(s[1]) && s[1] != '_')))
7305 Perl_warner(aTHX_ WARN_SYNTAX,
7306 "Misplaced _ in number");
7313 /* make an sv from the string */
7317 We try to do an integer conversion first if no characters
7318 indicating "float" have been found.
7323 int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
7325 if (flags == IS_NUMBER_IN_UV) {
7327 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7330 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
7331 if (uv <= (UV) IV_MIN)
7332 sv_setiv(sv, -(IV)uv);
7339 /* terminate the string */
7341 nv = Atof(PL_tokenbuf);
7345 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7346 (PL_hints & HINT_NEW_INTEGER) )
7347 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7348 (floatit ? "float" : "integer"),
7352 /* if it starts with a v, it could be a v-string */
7355 sv = NEWSV(92,5); /* preallocate storage space */
7356 s = new_vstring(s,sv);
7360 /* make the op for the constant and return */
7363 lvalp->opval = newSVOP(OP_CONST, 0, sv);
7365 lvalp->opval = Nullop;
7371 S_scan_formline(pTHX_ register char *s)
7375 SV *stuff = newSVpvn("",0);
7376 bool needargs = FALSE;
7379 if (*s == '.' || *s == /*{*/'}') {
7381 #ifdef PERL_STRICT_CR
7382 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7384 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7386 if (*t == '\n' || t == PL_bufend)
7389 if (PL_in_eval && !PL_rsfp) {
7390 eol = strchr(s,'\n');
7395 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7397 for (t = s; t < eol; t++) {
7398 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7400 goto enough; /* ~~ must be first line in formline */
7402 if (*t == '@' || *t == '^')
7406 sv_catpvn(stuff, s, eol-s);
7407 #ifndef PERL_STRICT_CR
7408 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7409 char *end = SvPVX(stuff) + SvCUR(stuff);
7421 s = filter_gets(PL_linestr, PL_rsfp, 0);
7422 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7423 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7424 PL_last_lop = PL_last_uni = Nullch;
7427 yyerror("Format not terminated");
7437 PL_lex_state = LEX_NORMAL;
7438 PL_nextval[PL_nexttoke].ival = 0;
7442 PL_lex_state = LEX_FORMLINE;
7443 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7445 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7449 SvREFCNT_dec(stuff);
7450 PL_lex_formbrack = 0;
7461 PL_cshlen = strlen(PL_cshname);
7466 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7468 I32 oldsavestack_ix = PL_savestack_ix;
7469 CV* outsidecv = PL_compcv;
7473 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7475 SAVEI32(PL_subline);
7476 save_item(PL_subname);
7479 SAVESPTR(PL_comppad_name);
7480 SAVESPTR(PL_compcv);
7481 SAVEI32(PL_comppad_name_fill);
7482 SAVEI32(PL_min_intro_pending);
7483 SAVEI32(PL_max_intro_pending);
7484 SAVEI32(PL_pad_reset_pending);
7486 PL_compcv = (CV*)NEWSV(1104,0);
7487 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7488 CvFLAGS(PL_compcv) |= flags;
7490 PL_comppad = newAV();
7491 av_push(PL_comppad, Nullsv);
7492 PL_curpad = AvARRAY(PL_comppad);
7493 PL_comppad_name = newAV();
7494 PL_comppad_name_fill = 0;
7495 PL_min_intro_pending = 0;
7497 PL_subline = CopLINE(PL_curcop);
7498 #ifdef USE_5005THREADS
7499 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7500 PL_curpad[0] = (SV*)newAV();
7501 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7502 #endif /* USE_5005THREADS */
7504 comppadlist = newAV();
7505 AvREAL_off(comppadlist);
7506 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7507 av_store(comppadlist, 1, (SV*)PL_comppad);
7509 CvPADLIST(PL_compcv) = comppadlist;
7510 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7511 #ifdef USE_5005THREADS
7512 CvOWNER(PL_compcv) = 0;
7513 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7514 MUTEX_INIT(CvMUTEXP(PL_compcv));
7515 #endif /* USE_5005THREADS */
7517 return oldsavestack_ix;
7521 #pragma segment Perl_yylex
7524 Perl_yywarn(pTHX_ char *s)
7526 PL_in_eval |= EVAL_WARNONLY;
7528 PL_in_eval &= ~EVAL_WARNONLY;
7533 Perl_yyerror(pTHX_ char *s)
7536 char *context = NULL;
7540 if (!yychar || (yychar == ';' && !PL_rsfp))
7542 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7543 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7544 while (isSPACE(*PL_oldoldbufptr))
7546 context = PL_oldoldbufptr;
7547 contlen = PL_bufptr - PL_oldoldbufptr;
7549 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7550 PL_oldbufptr != PL_bufptr) {
7551 while (isSPACE(*PL_oldbufptr))
7553 context = PL_oldbufptr;
7554 contlen = PL_bufptr - PL_oldbufptr;
7556 else if (yychar > 255)
7557 where = "next token ???";
7558 #ifdef USE_PURE_BISON
7559 /* GNU Bison sets the value -2 */
7560 else if (yychar == -2) {
7562 else if ((yychar & 127) == 127) {
7564 if (PL_lex_state == LEX_NORMAL ||
7565 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7566 where = "at end of line";
7567 else if (PL_lex_inpat)
7568 where = "within pattern";
7570 where = "within string";
7573 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7575 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7576 else if (isPRINT_LC(yychar))
7577 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7579 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7580 where = SvPVX(where_sv);
7582 msg = sv_2mortal(newSVpv(s, 0));
7583 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7584 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7586 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7588 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7589 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7590 Perl_sv_catpvf(aTHX_ msg,
7591 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7592 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7595 if (PL_in_eval & EVAL_WARNONLY)
7596 Perl_warn(aTHX_ "%"SVf, msg);
7599 if (PL_error_count >= 10) {
7600 if (PL_in_eval && SvCUR(ERRSV))
7601 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
7602 ERRSV, CopFILE(PL_curcop));
7604 Perl_croak(aTHX_ "%s has too many errors.\n",
7605 CopFILE(PL_curcop));
7608 PL_in_my_stash = Nullhv;
7612 #pragma segment Main
7616 S_swallow_bom(pTHX_ U8 *s)
7619 slen = SvCUR(PL_linestr);
7623 /* UTF-16 little-endian */
7624 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
7625 Perl_croak(aTHX_ "Unsupported script encoding");
7626 #ifndef PERL_NO_UTF16_FILTER
7627 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-LE script encoding\n"));
7629 if (PL_bufend > (char*)s) {
7633 filter_add(utf16rev_textfilter, NULL);
7634 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7635 PL_bufend = (char*)utf16_to_utf8_reversed(s, news,
7636 PL_bufend - (char*)s - 1,
7638 Copy(news, s, newlen, U8);
7639 SvCUR_set(PL_linestr, newlen);
7640 PL_bufend = SvPVX(PL_linestr) + newlen;
7641 news[newlen++] = '\0';
7645 Perl_croak(aTHX_ "Unsupported script encoding");
7650 if (s[1] == 0xFF) { /* UTF-16 big-endian */
7651 #ifndef PERL_NO_UTF16_FILTER
7652 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding\n"));
7654 if (PL_bufend > (char *)s) {
7658 filter_add(utf16_textfilter, NULL);
7659 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7660 PL_bufend = (char*)utf16_to_utf8(s, news,
7661 PL_bufend - (char*)s,
7663 Copy(news, s, newlen, U8);
7664 SvCUR_set(PL_linestr, newlen);
7665 PL_bufend = SvPVX(PL_linestr) + newlen;
7666 news[newlen++] = '\0';
7670 Perl_croak(aTHX_ "Unsupported script encoding");
7675 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
7676 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-8 script encoding\n"));
7681 if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */
7682 s[2] == 0xFE && s[3] == 0xFF)
7684 Perl_croak(aTHX_ "Unsupported script encoding");
7692 * Restore a source filter.
7696 restore_rsfp(pTHX_ void *f)
7698 PerlIO *fp = (PerlIO*)f;
7700 if (PL_rsfp == PerlIO_stdin())
7701 PerlIO_clearerr(PL_rsfp);
7702 else if (PL_rsfp && (PL_rsfp != fp))
7703 PerlIO_close(PL_rsfp);
7707 #ifndef PERL_NO_UTF16_FILTER
7709 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
7711 I32 count = FILTER_READ(idx+1, sv, maxlen);
7716 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7717 if (!*SvPV_nolen(sv))
7718 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7721 tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7722 sv_usepvn(sv, (char*)tmps, tend - tmps);
7728 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
7730 I32 count = FILTER_READ(idx+1, sv, maxlen);
7735 if (!*SvPV_nolen(sv))
7736 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7739 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7740 tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7741 sv_usepvn(sv, (char*)tmps, tend - tmps);