3 * Copyright (c) 1991-2000, 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(pTHXo_ void *f);
32 #define XFAKEBRACK 128
35 /*#define UTF (SvUTF8(PL_linestr) && !(PL_hints & HINT_BYTE))*/
36 #define UTF (PL_hints & HINT_UTF8)
38 /* In variables name $^X, these are the legal values for X.
39 * 1999-02-27 mjd-perl-patch@plover.com */
40 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
42 /* LEX_* are values for PL_lex_state, the state of the lexer.
43 * They are arranged oddly so that the guard on the switch statement
44 * can get by with a single comparison (if the compiler is smart enough).
47 /* #define LEX_NOTPARSING 11 is done in perl.h. */
50 #define LEX_INTERPNORMAL 9
51 #define LEX_INTERPCASEMOD 8
52 #define LEX_INTERPPUSH 7
53 #define LEX_INTERPSTART 6
54 #define LEX_INTERPEND 5
55 #define LEX_INTERPENDMAYBE 4
56 #define LEX_INTERPCONCAT 3
57 #define LEX_INTERPCONST 2
58 #define LEX_FORMLINE 1
59 #define LEX_KNOWNEXT 0
61 /* XXX If this causes problems, set i_unistd=undef in the hint file. */
63 # include <unistd.h> /* Needed for execv() */
72 YYSTYPE* yylval_pointer = NULL;
73 int* yychar_pointer = NULL;
76 # define yylval (*yylval_pointer)
77 # define yychar (*yychar_pointer)
78 # define PERL_YYLEX_PARAM yylval_pointer,yychar_pointer
80 # define yylex() Perl_yylex(aTHX_ yylval_pointer, yychar_pointer)
85 /* CLINE is a macro that ensures PL_copline has a sane value */
90 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
93 * Convenience functions to return different tokens and prime the
94 * lexer for the next token. They all take an argument.
96 * TOKEN : generic token (used for '(', DOLSHARP, etc)
97 * OPERATOR : generic operator
98 * AOPERATOR : assignment operator
99 * PREBLOCK : beginning the block after an if, while, foreach, ...
100 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
101 * PREREF : *EXPR where EXPR is not a simple identifier
102 * TERM : expression term
103 * LOOPX : loop exiting command (goto, last, dump, etc)
104 * FTST : file test operator
105 * FUN0 : zero-argument function
106 * FUN1 : not used, except for not, which isn't a UNIOP
107 * BOop : bitwise or or xor
109 * SHop : shift operator
110 * PWop : power operator
111 * PMop : pattern-matching operator
112 * Aop : addition-level operator
113 * Mop : multiplication-level operator
114 * Eop : equality-testing operator
115 * Rop : relational operator <= != gt
117 * Also see LOP and lop() below.
120 #define TOKEN(retval) return (PL_bufptr = s,(int)retval)
121 #define OPERATOR(retval) return (PL_expect = XTERM,PL_bufptr = s,(int)retval)
122 #define AOPERATOR(retval) return ao((PL_expect = XTERM,PL_bufptr = s,(int)retval))
123 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
124 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
125 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s,(int)retval)
126 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR,PL_bufptr = s,(int)retval)
127 #define LOOPX(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
128 #define FTST(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)UNIOP)
129 #define FUN0(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
130 #define FUN1(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
131 #define BOop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
132 #define BAop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
133 #define SHop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
134 #define PWop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
135 #define PMop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
136 #define Aop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
137 #define Mop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
138 #define Eop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
139 #define Rop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)RELOP)
141 /* This bit of chicanery makes a unary function followed by
142 * a parenthesis into a function with one argument, highest precedence.
144 #define UNI(f) return(yylval.ival = f, \
147 PL_last_uni = PL_oldbufptr, \
148 PL_last_lop_op = f, \
149 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
151 #define UNIBRACK(f) return(yylval.ival = f, \
153 PL_last_uni = PL_oldbufptr, \
154 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
156 /* grandfather return to old style */
157 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
162 * This subroutine detects &&= and ||= and turns an ANDAND or OROR
163 * into an OP_ANDASSIGN or OP_ORASSIGN
167 S_ao(pTHX_ int toketype)
169 if (*PL_bufptr == '=') {
171 if (toketype == ANDAND)
172 yylval.ival = OP_ANDASSIGN;
173 else if (toketype == OROR)
174 yylval.ival = OP_ORASSIGN;
182 * When Perl expects an operator and finds something else, no_op
183 * prints the warning. It always prints "<something> found where
184 * operator expected. It prints "Missing semicolon on previous line?"
185 * if the surprise occurs at the start of the line. "do you need to
186 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
187 * where the compiler doesn't know if foo is a method call or a function.
188 * It prints "Missing operator before end of line" if there's nothing
189 * after the missing operator, or "... before <...>" if there is something
190 * after the missing operator.
194 S_no_op(pTHX_ char *what, char *s)
196 char *oldbp = PL_bufptr;
197 bool is_first = (PL_oldbufptr == PL_linestart);
205 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
207 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
208 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
210 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
211 if (t < PL_bufptr && isSPACE(*t))
212 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
213 t - PL_oldoldbufptr, PL_oldoldbufptr);
216 Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
222 * Complain about missing quote/regexp/heredoc terminator.
223 * If it's called with (char *)NULL then it cauterizes the line buffer.
224 * If we're in a delimited string and the delimiter is a control
225 * character, it's reformatted into a two-char sequence like ^C.
230 S_missingterm(pTHX_ char *s)
235 char *nl = strrchr(s,'\n');
241 iscntrl(PL_multi_close)
243 PL_multi_close < 32 || PL_multi_close == 127
247 tmpbuf[1] = toCTRL(PL_multi_close);
253 *tmpbuf = PL_multi_close;
257 q = strchr(s,'"') ? '\'' : '"';
258 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
266 Perl_deprecate(pTHX_ char *s)
269 if (ckWARN(WARN_DEPRECATED))
270 Perl_warner(aTHX_ WARN_DEPRECATED, "Use of %s is deprecated", s);
275 * Deprecate a comma-less variable list.
281 deprecate("comma-less variable list");
285 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
286 * utf16-to-utf8-reversed.
289 #ifdef PERL_CR_FILTER
293 register char *s = SvPVX(sv);
294 register char *e = s + SvCUR(sv);
295 /* outer loop optimized to do nothing if there are no CR-LFs */
297 if (*s++ == '\r' && *s == '\n') {
298 /* hit a CR-LF, need to copy the rest */
299 register char *d = s - 1;
302 if (*s == '\r' && s[1] == '\n')
313 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
315 I32 count = FILTER_READ(idx+1, sv, maxlen);
316 if (count > 0 && !maxlen)
324 S_utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
326 I32 count = FILTER_READ(idx+1, sv, maxlen);
330 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
331 tend = utf16_to_utf8((U16*)SvPVX(sv), tmps, SvCUR(sv));
332 sv_usepvn(sv, (char*)tmps, tend - tmps);
338 S_utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
340 I32 count = FILTER_READ(idx+1, sv, maxlen);
344 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
345 tend = utf16_to_utf8_reversed((U16*)SvPVX(sv), tmps, SvCUR(sv));
346 sv_usepvn(sv, (char*)tmps, tend - tmps);
354 * Initialize variables. Uses the Perl save_stack to save its state (for
355 * recursive calls to the parser).
359 Perl_lex_start(pTHX_ SV *line)
365 SAVEI32(PL_lex_dojoin);
366 SAVEI32(PL_lex_brackets);
367 SAVEI32(PL_lex_casemods);
368 SAVEI32(PL_lex_starts);
369 SAVEI32(PL_lex_state);
370 SAVEVPTR(PL_lex_inpat);
371 SAVEI32(PL_lex_inwhat);
372 if (PL_lex_state == LEX_KNOWNEXT) {
373 I32 toke = PL_nexttoke;
374 while (--toke >= 0) {
375 SAVEI32(PL_nexttype[toke]);
376 SAVEVPTR(PL_nextval[toke]);
378 SAVEI32(PL_nexttoke);
381 SAVECOPLINE(PL_curcop);
384 SAVEPPTR(PL_oldbufptr);
385 SAVEPPTR(PL_oldoldbufptr);
386 SAVEPPTR(PL_linestart);
387 SAVESPTR(PL_linestr);
388 SAVEPPTR(PL_lex_brackstack);
389 SAVEPPTR(PL_lex_casestack);
390 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
391 SAVESPTR(PL_lex_stuff);
392 SAVEI32(PL_lex_defer);
393 SAVEI32(PL_sublex_info.sub_inwhat);
394 SAVESPTR(PL_lex_repl);
396 SAVEINT(PL_lex_expect);
398 PL_lex_state = LEX_NORMAL;
402 New(899, PL_lex_brackstack, 120, char);
403 New(899, PL_lex_casestack, 12, char);
404 SAVEFREEPV(PL_lex_brackstack);
405 SAVEFREEPV(PL_lex_casestack);
407 *PL_lex_casestack = '\0';
410 PL_lex_stuff = Nullsv;
411 PL_lex_repl = Nullsv;
414 PL_sublex_info.sub_inwhat = 0;
416 if (SvREADONLY(PL_linestr))
417 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
418 s = SvPV(PL_linestr, len);
419 if (len && s[len-1] != ';') {
420 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
421 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
422 sv_catpvn(PL_linestr, "\n;", 2);
424 SvTEMP_off(PL_linestr);
425 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
426 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
428 PL_rs = newSVpvn("\n", 1);
434 * Finalizer for lexing operations. Must be called when the parser is
435 * done with the lexer.
441 PL_doextract = FALSE;
446 * This subroutine has nothing to do with tilting, whether at windmills
447 * or pinball tables. Its name is short for "increment line". It
448 * increments the current line number in CopLINE(PL_curcop) and checks
449 * to see whether the line starts with a comment of the form
450 * # line 500 "foo.pm"
451 * If so, it sets the current line number and file to the values in the comment.
455 S_incline(pTHX_ char *s)
463 CopLINE_inc(PL_curcop);
466 while (*s == ' ' || *s == '\t') s++;
467 if (strnEQ(s, "line", 4))
471 if (*s == ' ' || *s == '\t')
475 while (*s == ' ' || *s == '\t') s++;
481 while (*s == ' ' || *s == '\t')
483 if (*s == '"' && (t = strchr(s+1, '"'))) {
488 for (t = s; !isSPACE(*t); t++) ;
491 while (*e == ' ' || *e == '\t' || *e == '\r' || *e == '\f')
493 if (*e != '\n' && *e != '\0')
494 return; /* false alarm */
499 CopFILE_set(PL_curcop, s);
501 CopLINE_set(PL_curcop, atoi(n)-1);
506 * Called to gobble the appropriate amount and type of whitespace.
507 * Skips comments as well.
511 S_skipspace(pTHX_ register char *s)
514 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
515 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
521 SSize_t oldprevlen, oldoldprevlen;
522 SSize_t oldloplen, oldunilen;
523 while (s < PL_bufend && isSPACE(*s)) {
524 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
529 if (s < PL_bufend && *s == '#') {
530 while (s < PL_bufend && *s != '\n')
534 if (PL_in_eval && !PL_rsfp) {
541 /* only continue to recharge the buffer if we're at the end
542 * of the buffer, we're not reading from a source filter, and
543 * we're in normal lexing mode
545 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
546 PL_lex_state == LEX_FORMLINE)
549 /* try to recharge the buffer */
550 if ((s = filter_gets(PL_linestr, PL_rsfp,
551 (prevlen = SvCUR(PL_linestr)))) == Nullch)
553 /* end of file. Add on the -p or -n magic */
554 if (PL_minus_n || PL_minus_p) {
555 sv_setpv(PL_linestr,PL_minus_p ?
556 ";}continue{print or die qq(-p destination: $!\\n)" :
558 sv_catpv(PL_linestr,";}");
559 PL_minus_n = PL_minus_p = 0;
562 sv_setpv(PL_linestr,";");
564 /* reset variables for next time we lex */
565 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
567 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
569 /* Close the filehandle. Could be from -P preprocessor,
570 * STDIN, or a regular file. If we were reading code from
571 * STDIN (because the commandline held no -e or filename)
572 * then we don't close it, we reset it so the code can
573 * read from STDIN too.
576 if (PL_preprocess && !PL_in_eval)
577 (void)PerlProc_pclose(PL_rsfp);
578 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
579 PerlIO_clearerr(PL_rsfp);
581 (void)PerlIO_close(PL_rsfp);
586 /* not at end of file, so we only read another line */
587 /* make corresponding updates to old pointers, for yyerror() */
588 oldprevlen = PL_oldbufptr - PL_bufend;
589 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
591 oldunilen = PL_last_uni - PL_bufend;
593 oldloplen = PL_last_lop - PL_bufend;
594 PL_linestart = PL_bufptr = s + prevlen;
595 PL_bufend = s + SvCUR(PL_linestr);
597 PL_oldbufptr = s + oldprevlen;
598 PL_oldoldbufptr = s + oldoldprevlen;
600 PL_last_uni = s + oldunilen;
602 PL_last_lop = s + oldloplen;
605 /* debugger active and we're not compiling the debugger code,
606 * so store the line into the debugger's array of lines
608 if (PERLDB_LINE && PL_curstash != PL_debstash) {
609 SV *sv = NEWSV(85,0);
611 sv_upgrade(sv, SVt_PVMG);
612 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
613 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
620 * Check the unary operators to ensure there's no ambiguity in how they're
621 * used. An ambiguous piece of code would be:
623 * This doesn't mean rand() + 5. Because rand() is a unary operator,
624 * the +5 is its argument.
634 if (PL_oldoldbufptr != PL_last_uni)
636 while (isSPACE(*PL_last_uni))
638 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
639 if ((t = strchr(s, '(')) && t < PL_bufptr)
641 if (ckWARN_d(WARN_AMBIGUOUS)){
644 Perl_warner(aTHX_ WARN_AMBIGUOUS,
645 "Warning: Use of \"%s\" without parens is ambiguous",
651 /* workaround to replace the UNI() macro with a function. Only the
652 * hints/uts.sh file mentions this. Other comments elsewhere in the
653 * source indicate Microport Unix might need it too.
659 #define UNI(f) return uni(f,s)
662 S_uni(pTHX_ I32 f, char *s)
667 PL_last_uni = PL_oldbufptr;
678 #endif /* CRIPPLED_CC */
681 * LOP : macro to build a list operator. Its behaviour has been replaced
682 * with a subroutine, S_lop() for which LOP is just another name.
685 #define LOP(f,x) return lop(f,x,s)
689 * Build a list operator (or something that might be one). The rules:
690 * - if we have a next token, then it's a list operator [why?]
691 * - if the next thing is an opening paren, then it's a function
692 * - else it's a list operator
696 S_lop(pTHX_ I32 f, int x, char *s)
703 PL_last_lop = PL_oldbufptr;
718 * When the lexer realizes it knows the next token (for instance,
719 * it is reordering tokens for the parser) then it can call S_force_next
720 * to know what token to return the next time the lexer is called. Caller
721 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
722 * handles the token correctly.
726 S_force_next(pTHX_ I32 type)
728 PL_nexttype[PL_nexttoke] = type;
730 if (PL_lex_state != LEX_KNOWNEXT) {
731 PL_lex_defer = PL_lex_state;
732 PL_lex_expect = PL_expect;
733 PL_lex_state = LEX_KNOWNEXT;
739 * When the lexer knows the next thing is a word (for instance, it has
740 * just seen -> and it knows that the next char is a word char, then
741 * it calls S_force_word to stick the next word into the PL_next lookahead.
744 * char *start : buffer position (must be within PL_linestr)
745 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
746 * int check_keyword : if true, Perl checks to make sure the word isn't
747 * a keyword (do this if the word is a label, e.g. goto FOO)
748 * int allow_pack : if true, : characters will also be allowed (require,
750 * int allow_initial_tick : used by the "sub" lexer only.
754 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
759 start = skipspace(start);
761 if (isIDFIRST_lazy_if(s,UTF) ||
762 (allow_pack && *s == ':') ||
763 (allow_initial_tick && *s == '\'') )
765 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
766 if (check_keyword && keyword(PL_tokenbuf, len))
768 if (token == METHOD) {
773 PL_expect = XOPERATOR;
776 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
777 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
785 * Called when the lexer wants $foo *foo &foo etc, but the program
786 * text only contains the "foo" portion. The first argument is a pointer
787 * to the "foo", and the second argument is the type symbol to prefix.
788 * Forces the next token to be a "WORD".
789 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
793 S_force_ident(pTHX_ register char *s, int kind)
796 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
797 PL_nextval[PL_nexttoke].opval = o;
800 dTHR; /* just for in_eval */
801 o->op_private = OPpCONST_ENTERED;
802 /* XXX see note in pp_entereval() for why we forgo typo
803 warnings if the symbol must be introduced in an eval.
805 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
806 kind == '$' ? SVt_PV :
807 kind == '@' ? SVt_PVAV :
808 kind == '%' ? SVt_PVHV :
816 Perl_str_to_version(pTHX_ SV *sv)
821 char *start = SvPVx(sv,len);
822 bool utf = SvUTF8(sv);
823 char *end = start + len;
824 while (start < end) {
828 n = utf8_to_uv((U8*)start, &skip);
833 retval += ((NV)n)/nshift;
842 * Forces the next token to be a version number.
846 S_force_version(pTHX_ char *s)
848 OP *version = Nullop;
857 for (; isDIGIT(*d) || *d == '_' || *d == '.'; d++);
858 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
861 version = yylval.opval;
862 ver = cSVOPx(version)->op_sv;
863 if (SvPOK(ver) && !SvNIOK(ver)) {
864 (void)SvUPGRADE(ver, SVt_PVNV);
865 SvNVX(ver) = str_to_version(ver);
866 SvNOK_on(ver); /* hint that it is a version */
871 /* NOTE: The parser sees the package name and the VERSION swapped */
872 PL_nextval[PL_nexttoke].opval = version;
880 * Tokenize a quoted string passed in as an SV. It finds the next
881 * chunk, up to end of string or a backslash. It may make a new
882 * SV containing that chunk (if HINT_NEW_STRING is on). It also
887 S_tokeq(pTHX_ SV *sv)
898 s = SvPV_force(sv, len);
902 while (s < send && *s != '\\')
907 if ( PL_hints & HINT_NEW_STRING )
908 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
911 if (s + 1 < send && (s[1] == '\\'))
912 s++; /* all that, just for this */
917 SvCUR_set(sv, d - SvPVX(sv));
919 if ( PL_hints & HINT_NEW_STRING )
920 return new_constant(NULL, 0, "q", sv, pv, "q");
925 * Now come three functions related to double-quote context,
926 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
927 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
928 * interact with PL_lex_state, and create fake ( ... ) argument lists
929 * to handle functions and concatenation.
930 * They assume that whoever calls them will be setting up a fake
931 * join call, because each subthing puts a ',' after it. This lets
934 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
936 * (I'm not sure whether the spurious commas at the end of lcfirst's
937 * arguments and join's arguments are created or not).
942 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
944 * Pattern matching will set PL_lex_op to the pattern-matching op to
945 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
947 * OP_CONST and OP_READLINE are easy--just make the new op and return.
949 * Everything else becomes a FUNC.
951 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
952 * had an OP_CONST or OP_READLINE). This just sets us up for a
953 * call to S_sublex_push().
959 register I32 op_type = yylval.ival;
961 if (op_type == OP_NULL) {
962 yylval.opval = PL_lex_op;
966 if (op_type == OP_CONST || op_type == OP_READLINE) {
967 SV *sv = tokeq(PL_lex_stuff);
969 if (SvTYPE(sv) == SVt_PVIV) {
970 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
976 nsv = newSVpvn(p, len);
980 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
981 PL_lex_stuff = Nullsv;
985 PL_sublex_info.super_state = PL_lex_state;
986 PL_sublex_info.sub_inwhat = op_type;
987 PL_sublex_info.sub_op = PL_lex_op;
988 PL_lex_state = LEX_INTERPPUSH;
992 yylval.opval = PL_lex_op;
1002 * Create a new scope to save the lexing state. The scope will be
1003 * ended in S_sublex_done. Returns a '(', starting the function arguments
1004 * to the uc, lc, etc. found before.
1005 * Sets PL_lex_state to LEX_INTERPCONCAT.
1014 PL_lex_state = PL_sublex_info.super_state;
1015 SAVEI32(PL_lex_dojoin);
1016 SAVEI32(PL_lex_brackets);
1017 SAVEI32(PL_lex_casemods);
1018 SAVEI32(PL_lex_starts);
1019 SAVEI32(PL_lex_state);
1020 SAVEVPTR(PL_lex_inpat);
1021 SAVEI32(PL_lex_inwhat);
1022 SAVECOPLINE(PL_curcop);
1023 SAVEPPTR(PL_bufptr);
1024 SAVEPPTR(PL_oldbufptr);
1025 SAVEPPTR(PL_oldoldbufptr);
1026 SAVEPPTR(PL_linestart);
1027 SAVESPTR(PL_linestr);
1028 SAVEPPTR(PL_lex_brackstack);
1029 SAVEPPTR(PL_lex_casestack);
1031 PL_linestr = PL_lex_stuff;
1032 PL_lex_stuff = Nullsv;
1034 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1035 = SvPVX(PL_linestr);
1036 PL_bufend += SvCUR(PL_linestr);
1037 SAVEFREESV(PL_linestr);
1039 PL_lex_dojoin = FALSE;
1040 PL_lex_brackets = 0;
1041 New(899, PL_lex_brackstack, 120, char);
1042 New(899, PL_lex_casestack, 12, char);
1043 SAVEFREEPV(PL_lex_brackstack);
1044 SAVEFREEPV(PL_lex_casestack);
1045 PL_lex_casemods = 0;
1046 *PL_lex_casestack = '\0';
1048 PL_lex_state = LEX_INTERPCONCAT;
1049 CopLINE_set(PL_curcop, PL_multi_start);
1051 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1052 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1053 PL_lex_inpat = PL_sublex_info.sub_op;
1055 PL_lex_inpat = Nullop;
1062 * Restores lexer state after a S_sublex_push.
1068 if (!PL_lex_starts++) {
1069 PL_expect = XOPERATOR;
1070 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn("",0));
1074 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1075 PL_lex_state = LEX_INTERPCASEMOD;
1079 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1080 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1081 PL_linestr = PL_lex_repl;
1083 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1084 PL_bufend += SvCUR(PL_linestr);
1085 SAVEFREESV(PL_linestr);
1086 PL_lex_dojoin = FALSE;
1087 PL_lex_brackets = 0;
1088 PL_lex_casemods = 0;
1089 *PL_lex_casestack = '\0';
1091 if (SvEVALED(PL_lex_repl)) {
1092 PL_lex_state = LEX_INTERPNORMAL;
1094 /* we don't clear PL_lex_repl here, so that we can check later
1095 whether this is an evalled subst; that means we rely on the
1096 logic to ensure sublex_done() is called again only via the
1097 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1100 PL_lex_state = LEX_INTERPCONCAT;
1101 PL_lex_repl = Nullsv;
1107 PL_bufend = SvPVX(PL_linestr);
1108 PL_bufend += SvCUR(PL_linestr);
1109 PL_expect = XOPERATOR;
1110 PL_sublex_info.sub_inwhat = 0;
1118 Extracts a pattern, double-quoted string, or transliteration. This
1121 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1122 processing a pattern (PL_lex_inpat is true), a transliteration
1123 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1125 Returns a pointer to the character scanned up to. Iff this is
1126 advanced from the start pointer supplied (ie if anything was
1127 successfully parsed), will leave an OP for the substring scanned
1128 in yylval. Caller must intuit reason for not parsing further
1129 by looking at the next characters herself.
1133 double-quoted style: \r and \n
1134 regexp special ones: \D \s
1136 backrefs: \1 (deprecated in substitution replacements)
1137 case and quoting: \U \Q \E
1138 stops on @ and $, but not for $ as tail anchor
1140 In transliterations:
1141 characters are VERY literal, except for - not at the start or end
1142 of the string, which indicates a range. scan_const expands the
1143 range to the full set of intermediate characters.
1145 In double-quoted strings:
1147 double-quoted style: \r and \n
1149 backrefs: \1 (deprecated)
1150 case and quoting: \U \Q \E
1153 scan_const does *not* construct ops to handle interpolated strings.
1154 It stops processing as soon as it finds an embedded $ or @ variable
1155 and leaves it to the caller to work out what's going on.
1157 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1159 $ in pattern could be $foo or could be tail anchor. Assumption:
1160 it's a tail anchor if $ is the last thing in the string, or if it's
1161 followed by one of ")| \n\t"
1163 \1 (backreferences) are turned into $1
1165 The structure of the code is
1166 while (there's a character to process) {
1167 handle transliteration ranges
1168 skip regexp comments
1169 skip # initiated comments in //x patterns
1170 check for embedded @foo
1171 check for embedded scalars
1173 leave intact backslashes from leave (below)
1174 deprecate \1 in strings and sub replacements
1175 handle string-changing backslashes \l \U \Q \E, etc.
1176 switch (what was escaped) {
1177 handle - in a transliteration (becomes a literal -)
1178 handle \132 octal characters
1179 handle 0x15 hex characters
1180 handle \cV (control V)
1181 handle printf backslashes (\f, \r, \n, etc)
1183 } (end if backslash)
1184 } (end while character to read)
1189 S_scan_const(pTHX_ char *start)
1191 register char *send = PL_bufend; /* end of the constant */
1192 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1193 register char *s = start; /* start of the constant */
1194 register char *d = SvPVX(sv); /* destination for copies */
1195 bool dorange = FALSE; /* are we in a translit range? */
1196 bool has_utf = FALSE; /* embedded \x{} */
1200 I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1201 ? (PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1203 I32 thisutf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1204 ? (PL_sublex_info.sub_op->op_private & (PL_lex_repl ?
1205 OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF))
1207 const char *leaveit = /* set of acceptably-backslashed characters */
1209 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1212 while (s < send || dorange) {
1213 /* get transliterations out of the way (they're most literal) */
1214 if (PL_lex_inwhat == OP_TRANS) {
1215 /* expand a range A-Z to the full set of characters. AIE! */
1217 I32 i; /* current expanded character */
1218 I32 min; /* first character in range */
1219 I32 max; /* last character in range */
1221 i = d - SvPVX(sv); /* remember current offset */
1222 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1223 d = SvPVX(sv) + i; /* refresh d after realloc */
1224 d -= 2; /* eat the first char and the - */
1226 min = (U8)*d; /* first char in range */
1227 max = (U8)d[1]; /* last char in range */
1230 if ((isLOWER(min) && isLOWER(max)) ||
1231 (isUPPER(min) && isUPPER(max))) {
1233 for (i = min; i <= max; i++)
1237 for (i = min; i <= max; i++)
1244 for (i = min; i <= max; i++)
1247 /* mark the range as done, and continue */
1252 /* range begins (ignore - as first or last char) */
1253 else if (*s == '-' && s+1 < send && s != start) {
1255 *d++ = (char)0xff; /* use illegal utf8 byte--see pmtrans */
1264 /* if we get here, we're not doing a transliteration */
1266 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1267 except for the last char, which will be done separately. */
1268 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1270 while (s < send && *s != ')')
1273 else if (s[2] == '{' /* This should match regcomp.c */
1274 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1277 char *regparse = s + (s[2] == '{' ? 3 : 4);
1280 while (count && (c = *regparse)) {
1281 if (c == '\\' && regparse[1])
1289 if (*regparse != ')') {
1290 regparse--; /* Leave one char for continuation. */
1291 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1293 while (s < regparse)
1298 /* likewise skip #-initiated comments in //x patterns */
1299 else if (*s == '#' && PL_lex_inpat &&
1300 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1301 while (s+1 < send && *s != '\n')
1305 /* check for embedded arrays (@foo, @:foo, @'foo, @{foo}, @$foo) */
1306 else if (*s == '@' && s[1]
1307 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$", s[1])))
1310 /* check for embedded scalars. only stop if we're sure it's a
1313 else if (*s == '$') {
1314 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1316 if (s + 1 < send && !strchr("()| \n\t", s[1]))
1317 break; /* in regexp, $ might be tail anchor */
1320 /* (now in tr/// code again) */
1322 if (*s & 0x80 && thisutf) {
1323 (void)utf8_to_uv((U8*)s, &len);
1325 /* illegal UTF8, make it valid */
1326 char *old_pvx = SvPVX(sv);
1327 /* need space for one extra char (NOTE: SvCUR() not set here) */
1328 d = SvGROW(sv, SvLEN(sv) + 1) + (d - old_pvx);
1329 d = (char*)uv_to_utf8((U8*)d, (U8)*s++);
1340 if (*s == '\\' && s+1 < send) {
1343 /* some backslashes we leave behind */
1344 if (*leaveit && *s && strchr(leaveit, *s)) {
1350 /* deprecate \1 in strings and substitution replacements */
1351 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1352 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1354 dTHR; /* only for ckWARN */
1355 if (ckWARN(WARN_SYNTAX))
1356 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1361 /* string-change backslash escapes */
1362 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1367 /* if we get here, it's either a quoted -, or a digit */
1370 /* quoted - in transliterations */
1372 if (PL_lex_inwhat == OP_TRANS) {
1380 if (ckWARN(WARN_MISC) && isALPHA(*s))
1381 Perl_warner(aTHX_ WARN_MISC,
1382 "Unrecognized escape \\%c passed through",
1384 /* default action is to copy the quoted character */
1389 /* \132 indicates an octal constant */
1390 case '0': case '1': case '2': case '3':
1391 case '4': case '5': case '6': case '7':
1392 uv = (UV)scan_oct(s, 3, &len);
1394 goto NUM_ESCAPE_INSERT;
1396 /* \x24 indicates a hex constant */
1400 char* e = strchr(s, '}');
1402 yyerror("Missing right brace on \\x{}");
1405 uv = (UV)scan_hex(s + 1, e - s - 1, &len);
1409 uv = (UV)scan_hex(s, 2, &len);
1414 /* Insert oct or hex escaped character.
1415 * There will always enough room in sv since such escapes will
1416 * be longer than any utf8 sequence they can end up as
1419 if (!thisutf && !has_utf && uv > 255) {
1420 /* might need to recode whatever we have accumulated so far
1421 * if it contains any hibit chars
1425 for (c = SvPVX(sv); c < d; c++) {
1430 char *old_pvx = SvPVX(sv);
1432 d = SvGROW(sv, SvCUR(sv) + hicount + 1) + (d - old_pvx);
1441 uv_to_utf8((U8*)dst, (U8)*src--);
1451 if (thisutf || uv > 255) {
1452 d = (char*)uv_to_utf8((U8*)d, uv);
1464 /* \N{latin small letter a} is a named character */
1468 char* e = strchr(s, '}');
1474 yyerror("Missing right brace on \\N{}");
1478 res = newSVpvn(s + 1, e - s - 1);
1479 res = new_constant( Nullch, 0, "charnames",
1480 res, Nullsv, "\\N{...}" );
1481 str = SvPV(res,len);
1482 if (!has_utf && SvUTF8(res)) {
1483 char *ostart = SvPVX(sv);
1484 SvCUR_set(sv, d - ostart);
1486 sv_utf8_upgrade(sv);
1487 d = SvPVX(sv) + SvCUR(sv);
1490 if (len > e - s + 4) {
1491 char *odest = SvPVX(sv);
1493 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1494 d = SvPVX(sv) + (d - odest);
1496 Copy(str, d, len, char);
1503 yyerror("Missing braces on \\N{}");
1506 /* \c is a control character */
1520 /* printf-style backslashes, formfeeds, newlines, etc */
1538 *d++ = '\047'; /* CP 1047 */
1541 *d++ = '\057'; /* CP 1047 */
1555 } /* end if (backslash) */
1558 } /* while loop to process each character */
1560 /* terminate the string and set up the sv */
1562 SvCUR_set(sv, d - SvPVX(sv));
1567 /* shrink the sv if we allocated more than we used */
1568 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1569 SvLEN_set(sv, SvCUR(sv) + 1);
1570 Renew(SvPVX(sv), SvLEN(sv), char);
1573 /* return the substring (via yylval) only if we parsed anything */
1574 if (s > PL_bufptr) {
1575 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1576 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1578 ( PL_lex_inwhat == OP_TRANS
1580 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1583 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1590 * Returns TRUE if there's more to the expression (e.g., a subscript),
1593 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1595 * ->[ and ->{ return TRUE
1596 * { and [ outside a pattern are always subscripts, so return TRUE
1597 * if we're outside a pattern and it's not { or [, then return FALSE
1598 * if we're in a pattern and the first char is a {
1599 * {4,5} (any digits around the comma) returns FALSE
1600 * if we're in a pattern and the first char is a [
1602 * [SOMETHING] has a funky algorithm to decide whether it's a
1603 * character class or not. It has to deal with things like
1604 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1605 * anything else returns TRUE
1608 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1611 S_intuit_more(pTHX_ register char *s)
1613 if (PL_lex_brackets)
1615 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1617 if (*s != '{' && *s != '[')
1622 /* In a pattern, so maybe we have {n,m}. */
1639 /* On the other hand, maybe we have a character class */
1642 if (*s == ']' || *s == '^')
1645 /* this is terrifying, and it works */
1646 int weight = 2; /* let's weigh the evidence */
1648 unsigned char un_char = 255, last_un_char;
1649 char *send = strchr(s,']');
1650 char tmpbuf[sizeof PL_tokenbuf * 4];
1652 if (!send) /* has to be an expression */
1655 Zero(seen,256,char);
1658 else if (isDIGIT(*s)) {
1660 if (isDIGIT(s[1]) && s[2] == ']')
1666 for (; s < send; s++) {
1667 last_un_char = un_char;
1668 un_char = (unsigned char)*s;
1673 weight -= seen[un_char] * 10;
1674 if (isALNUM_lazy_if(s+1,UTF)) {
1675 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1676 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1681 else if (*s == '$' && s[1] &&
1682 strchr("[#!%*<>()-=",s[1])) {
1683 if (/*{*/ strchr("])} =",s[2]))
1692 if (strchr("wds]",s[1]))
1694 else if (seen['\''] || seen['"'])
1696 else if (strchr("rnftbxcav",s[1]))
1698 else if (isDIGIT(s[1])) {
1700 while (s[1] && isDIGIT(s[1]))
1710 if (strchr("aA01! ",last_un_char))
1712 if (strchr("zZ79~",s[1]))
1714 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1715 weight -= 5; /* cope with negative subscript */
1718 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1719 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1724 if (keyword(tmpbuf, d - tmpbuf))
1727 if (un_char == last_un_char + 1)
1729 weight -= seen[un_char];
1734 if (weight >= 0) /* probably a character class */
1744 * Does all the checking to disambiguate
1746 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1747 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1749 * First argument is the stuff after the first token, e.g. "bar".
1751 * Not a method if bar is a filehandle.
1752 * Not a method if foo is a subroutine prototyped to take a filehandle.
1753 * Not a method if it's really "Foo $bar"
1754 * Method if it's "foo $bar"
1755 * Not a method if it's really "print foo $bar"
1756 * Method if it's really "foo package::" (interpreted as package->foo)
1757 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1758 * Not a method if bar is a filehandle or package, but is quoted with
1763 S_intuit_method(pTHX_ char *start, GV *gv)
1765 char *s = start + (*start == '$');
1766 char tmpbuf[sizeof PL_tokenbuf];
1774 if ((cv = GvCVu(gv))) {
1775 char *proto = SvPVX(cv);
1785 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1786 /* start is the beginning of the possible filehandle/object,
1787 * and s is the end of it
1788 * tmpbuf is a copy of it
1791 if (*start == '$') {
1792 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1797 return *s == '(' ? FUNCMETH : METHOD;
1799 if (!keyword(tmpbuf, len)) {
1800 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1805 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1806 if (indirgv && GvCVu(indirgv))
1808 /* filehandle or package name makes it a method */
1809 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1811 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1812 return 0; /* no assumptions -- "=>" quotes bearword */
1814 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1815 newSVpvn(tmpbuf,len));
1816 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1820 return *s == '(' ? FUNCMETH : METHOD;
1828 * Return a string of Perl code to load the debugger. If PERL5DB
1829 * is set, it will return the contents of that, otherwise a
1830 * compile-time require of perl5db.pl.
1837 char *pdb = PerlEnv_getenv("PERL5DB");
1841 SETERRNO(0,SS$_NORMAL);
1842 return "BEGIN { require 'perl5db.pl' }";
1848 /* Encoded script support. filter_add() effectively inserts a
1849 * 'pre-processing' function into the current source input stream.
1850 * Note that the filter function only applies to the current source file
1851 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1853 * The datasv parameter (which may be NULL) can be used to pass
1854 * private data to this instance of the filter. The filter function
1855 * can recover the SV using the FILTER_DATA macro and use it to
1856 * store private buffers and state information.
1858 * The supplied datasv parameter is upgraded to a PVIO type
1859 * and the IoDIRP field is used to store the function pointer,
1860 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1861 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1862 * private use must be set using malloc'd pointers.
1866 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1871 if (!PL_rsfp_filters)
1872 PL_rsfp_filters = newAV();
1874 datasv = NEWSV(255,0);
1875 if (!SvUPGRADE(datasv, SVt_PVIO))
1876 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1877 IoDIRP(datasv) = (DIR*)funcp; /* stash funcp into spare field */
1878 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1879 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1880 funcp, SvPV_nolen(datasv)));
1881 av_unshift(PL_rsfp_filters, 1);
1882 av_store(PL_rsfp_filters, 0, datasv) ;
1887 /* Delete most recently added instance of this filter function. */
1889 Perl_filter_del(pTHX_ filter_t funcp)
1892 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1893 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1895 /* if filter is on top of stack (usual case) just pop it off */
1896 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1897 if (IoDIRP(datasv) == (DIR*)funcp) {
1898 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1899 IoDIRP(datasv) = (DIR*)NULL;
1900 sv_free(av_pop(PL_rsfp_filters));
1904 /* we need to search for the correct entry and clear it */
1905 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1909 /* Invoke the n'th filter function for the current rsfp. */
1911 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1914 /* 0 = read one text line */
1919 if (!PL_rsfp_filters)
1921 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1922 /* Provide a default input filter to make life easy. */
1923 /* Note that we append to the line. This is handy. */
1924 DEBUG_P(PerlIO_printf(Perl_debug_log,
1925 "filter_read %d: from rsfp\n", idx));
1929 int old_len = SvCUR(buf_sv) ;
1931 /* ensure buf_sv is large enough */
1932 SvGROW(buf_sv, old_len + maxlen) ;
1933 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1934 if (PerlIO_error(PL_rsfp))
1935 return -1; /* error */
1937 return 0 ; /* end of file */
1939 SvCUR_set(buf_sv, old_len + len) ;
1942 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1943 if (PerlIO_error(PL_rsfp))
1944 return -1; /* error */
1946 return 0 ; /* end of file */
1949 return SvCUR(buf_sv);
1951 /* Skip this filter slot if filter has been deleted */
1952 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1953 DEBUG_P(PerlIO_printf(Perl_debug_log,
1954 "filter_read %d: skipped (filter deleted)\n",
1956 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1958 /* Get function pointer hidden within datasv */
1959 funcp = (filter_t)IoDIRP(datasv);
1960 DEBUG_P(PerlIO_printf(Perl_debug_log,
1961 "filter_read %d: via function %p (%s)\n",
1962 idx, funcp, SvPV_nolen(datasv)));
1963 /* Call function. The function is expected to */
1964 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1965 /* Return: <0:error, =0:eof, >0:not eof */
1966 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1970 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1972 #ifdef PERL_CR_FILTER
1973 if (!PL_rsfp_filters) {
1974 filter_add(S_cr_textfilter,NULL);
1977 if (PL_rsfp_filters) {
1980 SvCUR_set(sv, 0); /* start with empty line */
1981 if (FILTER_READ(0, sv, 0) > 0)
1982 return ( SvPVX(sv) ) ;
1987 return (sv_gets(sv, fp, append));
1992 static char* exp_name[] =
1993 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
1994 "ATTRTERM", "TERMBLOCK"
2001 Works out what to call the token just pulled out of the input
2002 stream. The yacc parser takes care of taking the ops we return and
2003 stitching them into a tree.
2009 if read an identifier
2010 if we're in a my declaration
2011 croak if they tried to say my($foo::bar)
2012 build the ops for a my() declaration
2013 if it's an access to a my() variable
2014 are we in a sort block?
2015 croak if my($a); $a <=> $b
2016 build ops for access to a my() variable
2017 if in a dq string, and they've said @foo and we can't find @foo
2019 build ops for a bareword
2020 if we already built the token before, use it.
2024 #ifdef USE_PURE_BISON
2025 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
2038 #ifdef USE_PURE_BISON
2039 yylval_pointer = lvalp;
2040 yychar_pointer = lcharp;
2043 /* check if there's an identifier for us to look at */
2044 if (PL_pending_ident) {
2045 /* pit holds the identifier we read and pending_ident is reset */
2046 char pit = PL_pending_ident;
2047 PL_pending_ident = 0;
2049 /* if we're in a my(), we can't allow dynamics here.
2050 $foo'bar has already been turned into $foo::bar, so
2051 just check for colons.
2053 if it's a legal name, the OP is a PADANY.
2056 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
2057 if (strchr(PL_tokenbuf,':'))
2058 yyerror(Perl_form(aTHX_ "No package name allowed for "
2059 "variable %s in \"our\"",
2061 tmp = pad_allocmy(PL_tokenbuf);
2064 if (strchr(PL_tokenbuf,':'))
2065 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
2067 yylval.opval = newOP(OP_PADANY, 0);
2068 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
2074 build the ops for accesses to a my() variable.
2076 Deny my($a) or my($b) in a sort block, *if* $a or $b is
2077 then used in a comparison. This catches most, but not
2078 all cases. For instance, it catches
2079 sort { my($a); $a <=> $b }
2081 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2082 (although why you'd do that is anyone's guess).
2085 if (!strchr(PL_tokenbuf,':')) {
2087 /* Check for single character per-thread SVs */
2088 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2089 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2090 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2092 yylval.opval = newOP(OP_THREADSV, 0);
2093 yylval.opval->op_targ = tmp;
2096 #endif /* USE_THREADS */
2097 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2098 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2099 /* might be an "our" variable" */
2100 if (SvFLAGS(namesv) & SVpad_OUR) {
2101 /* build ops for a bareword */
2102 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2103 sv_catpvn(sym, "::", 2);
2104 sv_catpv(sym, PL_tokenbuf+1);
2105 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2106 yylval.opval->op_private = OPpCONST_ENTERED;
2107 gv_fetchpv(SvPVX(sym),
2109 ? (GV_ADDMULTI | GV_ADDINEVAL)
2112 ((PL_tokenbuf[0] == '$') ? SVt_PV
2113 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2118 /* if it's a sort block and they're naming $a or $b */
2119 if (PL_last_lop_op == OP_SORT &&
2120 PL_tokenbuf[0] == '$' &&
2121 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2124 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2125 d < PL_bufend && *d != '\n';
2128 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2129 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2135 yylval.opval = newOP(OP_PADANY, 0);
2136 yylval.opval->op_targ = tmp;
2142 Whine if they've said @foo in a doublequoted string,
2143 and @foo isn't a variable we can find in the symbol
2146 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2147 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2148 if (!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2149 yyerror(Perl_form(aTHX_ "In string, %s now must be written as \\%s",
2150 PL_tokenbuf, PL_tokenbuf));
2153 /* build ops for a bareword */
2154 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2155 yylval.opval->op_private = OPpCONST_ENTERED;
2156 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2157 ((PL_tokenbuf[0] == '$') ? SVt_PV
2158 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2163 /* no identifier pending identification */
2165 switch (PL_lex_state) {
2167 case LEX_NORMAL: /* Some compilers will produce faster */
2168 case LEX_INTERPNORMAL: /* code if we comment these out. */
2172 /* when we've already built the next token, just pull it out of the queue */
2175 yylval = PL_nextval[PL_nexttoke];
2177 PL_lex_state = PL_lex_defer;
2178 PL_expect = PL_lex_expect;
2179 PL_lex_defer = LEX_NORMAL;
2181 return(PL_nexttype[PL_nexttoke]);
2183 /* interpolated case modifiers like \L \U, including \Q and \E.
2184 when we get here, PL_bufptr is at the \
2186 case LEX_INTERPCASEMOD:
2188 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2189 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2191 /* handle \E or end of string */
2192 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2196 if (PL_lex_casemods) {
2197 oldmod = PL_lex_casestack[--PL_lex_casemods];
2198 PL_lex_casestack[PL_lex_casemods] = '\0';
2200 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2202 PL_lex_state = LEX_INTERPCONCAT;
2206 if (PL_bufptr != PL_bufend)
2208 PL_lex_state = LEX_INTERPCONCAT;
2213 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2214 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2215 if (strchr("LU", *s) &&
2216 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2218 PL_lex_casestack[--PL_lex_casemods] = '\0';
2221 if (PL_lex_casemods > 10) {
2222 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2223 if (newlb != PL_lex_casestack) {
2225 PL_lex_casestack = newlb;
2228 PL_lex_casestack[PL_lex_casemods++] = *s;
2229 PL_lex_casestack[PL_lex_casemods] = '\0';
2230 PL_lex_state = LEX_INTERPCONCAT;
2231 PL_nextval[PL_nexttoke].ival = 0;
2234 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2236 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2238 PL_nextval[PL_nexttoke].ival = OP_LC;
2240 PL_nextval[PL_nexttoke].ival = OP_UC;
2242 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2244 Perl_croak(aTHX_ "panic: yylex");
2247 if (PL_lex_starts) {
2256 case LEX_INTERPPUSH:
2257 return sublex_push();
2259 case LEX_INTERPSTART:
2260 if (PL_bufptr == PL_bufend)
2261 return sublex_done();
2263 PL_lex_dojoin = (*PL_bufptr == '@');
2264 PL_lex_state = LEX_INTERPNORMAL;
2265 if (PL_lex_dojoin) {
2266 PL_nextval[PL_nexttoke].ival = 0;
2269 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2270 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2271 force_next(PRIVATEREF);
2273 force_ident("\"", '$');
2274 #endif /* USE_THREADS */
2275 PL_nextval[PL_nexttoke].ival = 0;
2277 PL_nextval[PL_nexttoke].ival = 0;
2279 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2282 if (PL_lex_starts++) {
2288 case LEX_INTERPENDMAYBE:
2289 if (intuit_more(PL_bufptr)) {
2290 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2296 if (PL_lex_dojoin) {
2297 PL_lex_dojoin = FALSE;
2298 PL_lex_state = LEX_INTERPCONCAT;
2301 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2302 && SvEVALED(PL_lex_repl))
2304 if (PL_bufptr != PL_bufend)
2305 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2306 PL_lex_repl = Nullsv;
2309 case LEX_INTERPCONCAT:
2311 if (PL_lex_brackets)
2312 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2314 if (PL_bufptr == PL_bufend)
2315 return sublex_done();
2317 if (SvIVX(PL_linestr) == '\'') {
2318 SV *sv = newSVsv(PL_linestr);
2321 else if ( PL_hints & HINT_NEW_RE )
2322 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2323 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2327 s = scan_const(PL_bufptr);
2329 PL_lex_state = LEX_INTERPCASEMOD;
2331 PL_lex_state = LEX_INTERPSTART;
2334 if (s != PL_bufptr) {
2335 PL_nextval[PL_nexttoke] = yylval;
2338 if (PL_lex_starts++)
2348 PL_lex_state = LEX_NORMAL;
2349 s = scan_formline(PL_bufptr);
2350 if (!PL_lex_formbrack)
2356 PL_oldoldbufptr = PL_oldbufptr;
2359 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2360 exp_name[PL_expect], s);
2366 if (isIDFIRST_lazy_if(s,UTF))
2368 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2371 goto fake_eof; /* emulate EOF on ^D or ^Z */
2376 if (PL_lex_brackets)
2377 yyerror("Missing right curly or square bracket");
2380 if (s++ < PL_bufend)
2381 goto retry; /* ignore stray nulls */
2384 if (!PL_in_eval && !PL_preambled) {
2385 PL_preambled = TRUE;
2386 sv_setpv(PL_linestr,incl_perldb());
2387 if (SvCUR(PL_linestr))
2388 sv_catpv(PL_linestr,";");
2390 while(AvFILLp(PL_preambleav) >= 0) {
2391 SV *tmpsv = av_shift(PL_preambleav);
2392 sv_catsv(PL_linestr, tmpsv);
2393 sv_catpv(PL_linestr, ";");
2396 sv_free((SV*)PL_preambleav);
2397 PL_preambleav = NULL;
2399 if (PL_minus_n || PL_minus_p) {
2400 sv_catpv(PL_linestr, "LINE: while (<>) {");
2402 sv_catpv(PL_linestr,"chomp;");
2404 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2406 GvIMPORTED_AV_on(gv);
2408 if (strchr("/'\"", *PL_splitstr)
2409 && strchr(PL_splitstr + 1, *PL_splitstr))
2410 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2413 s = "'~#\200\1'"; /* surely one char is unused...*/
2414 while (s[1] && strchr(PL_splitstr, *s)) s++;
2416 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2417 "q" + (delim == '\''), delim);
2418 for (s = PL_splitstr; *s; s++) {
2420 sv_catpvn(PL_linestr, "\\", 1);
2421 sv_catpvn(PL_linestr, s, 1);
2423 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2427 sv_catpv(PL_linestr,"@F=split(' ');");
2430 sv_catpv(PL_linestr, "\n");
2431 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2432 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2433 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2434 SV *sv = NEWSV(85,0);
2436 sv_upgrade(sv, SVt_PVMG);
2437 sv_setsv(sv,PL_linestr);
2438 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2443 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2446 if (PL_preprocess && !PL_in_eval)
2447 (void)PerlProc_pclose(PL_rsfp);
2448 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2449 PerlIO_clearerr(PL_rsfp);
2451 (void)PerlIO_close(PL_rsfp);
2453 PL_doextract = FALSE;
2455 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2456 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2457 sv_catpv(PL_linestr,";}");
2458 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2459 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2460 PL_minus_n = PL_minus_p = 0;
2463 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2464 sv_setpv(PL_linestr,"");
2465 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2468 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2469 PL_doextract = FALSE;
2471 /* Incest with pod. */
2472 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2473 sv_setpv(PL_linestr, "");
2474 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2475 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2476 PL_doextract = FALSE;
2480 } while (PL_doextract);
2481 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2482 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2483 SV *sv = NEWSV(85,0);
2485 sv_upgrade(sv, SVt_PVMG);
2486 sv_setsv(sv,PL_linestr);
2487 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2489 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2490 if (CopLINE(PL_curcop) == 1) {
2491 while (s < PL_bufend && isSPACE(*s))
2493 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2497 if (*s == '#' && *(s+1) == '!')
2499 #ifdef ALTERNATE_SHEBANG
2501 static char as[] = ALTERNATE_SHEBANG;
2502 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2503 d = s + (sizeof(as) - 1);
2505 #endif /* ALTERNATE_SHEBANG */
2514 while (*d && !isSPACE(*d))
2518 #ifdef ARG_ZERO_IS_SCRIPT
2519 if (ipathend > ipath) {
2521 * HP-UX (at least) sets argv[0] to the script name,
2522 * which makes $^X incorrect. And Digital UNIX and Linux,
2523 * at least, set argv[0] to the basename of the Perl
2524 * interpreter. So, having found "#!", we'll set it right.
2526 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2527 assert(SvPOK(x) || SvGMAGICAL(x));
2528 if (sv_eq(x, CopFILESV(PL_curcop))) {
2529 sv_setpvn(x, ipath, ipathend - ipath);
2532 TAINT_NOT; /* $^X is always tainted, but that's OK */
2534 #endif /* ARG_ZERO_IS_SCRIPT */
2539 d = instr(s,"perl -");
2541 d = instr(s,"perl");
2543 /* avoid getting into infinite loops when shebang
2544 * line contains "Perl" rather than "perl" */
2546 for (d = ipathend-4; d >= ipath; --d) {
2547 if ((*d == 'p' || *d == 'P')
2548 && !ibcmp(d, "perl", 4))
2558 #ifdef ALTERNATE_SHEBANG
2560 * If the ALTERNATE_SHEBANG on this system starts with a
2561 * character that can be part of a Perl expression, then if
2562 * we see it but not "perl", we're probably looking at the
2563 * start of Perl code, not a request to hand off to some
2564 * other interpreter. Similarly, if "perl" is there, but
2565 * not in the first 'word' of the line, we assume the line
2566 * contains the start of the Perl program.
2568 if (d && *s != '#') {
2570 while (*c && !strchr("; \t\r\n\f\v#", *c))
2573 d = Nullch; /* "perl" not in first word; ignore */
2575 *s = '#'; /* Don't try to parse shebang line */
2577 #endif /* ALTERNATE_SHEBANG */
2582 !instr(s,"indir") &&
2583 instr(PL_origargv[0],"perl"))
2589 while (s < PL_bufend && isSPACE(*s))
2591 if (s < PL_bufend) {
2592 Newz(899,newargv,PL_origargc+3,char*);
2594 while (s < PL_bufend && !isSPACE(*s))
2597 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2600 newargv = PL_origargv;
2602 PerlProc_execv(ipath, newargv);
2603 Perl_croak(aTHX_ "Can't exec %s", ipath);
2606 U32 oldpdb = PL_perldb;
2607 bool oldn = PL_minus_n;
2608 bool oldp = PL_minus_p;
2610 while (*d && !isSPACE(*d)) d++;
2611 while (*d == ' ' || *d == '\t') d++;
2615 if (*d == 'M' || *d == 'm') {
2617 while (*d && !isSPACE(*d)) d++;
2618 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2621 d = moreswitches(d);
2623 if ((PERLDB_LINE && !oldpdb) ||
2624 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2625 /* if we have already added "LINE: while (<>) {",
2626 we must not do it again */
2628 sv_setpv(PL_linestr, "");
2629 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2630 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2631 PL_preambled = FALSE;
2633 (void)gv_fetchfile(PL_origfilename);
2640 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2642 PL_lex_state = LEX_FORMLINE;
2647 #ifdef PERL_STRICT_CR
2648 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2650 "(Maybe you didn't strip carriage returns after a network transfer?)\n");
2652 case ' ': case '\t': case '\f': case 013:
2657 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2659 while (s < d && *s != '\n')
2664 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2666 PL_lex_state = LEX_FORMLINE;
2676 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2681 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2684 if (strnEQ(s,"=>",2)) {
2685 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2686 OPERATOR('-'); /* unary minus */
2688 PL_last_uni = PL_oldbufptr;
2689 PL_last_lop_op = OP_FTEREAD; /* good enough */
2691 case 'r': FTST(OP_FTEREAD);
2692 case 'w': FTST(OP_FTEWRITE);
2693 case 'x': FTST(OP_FTEEXEC);
2694 case 'o': FTST(OP_FTEOWNED);
2695 case 'R': FTST(OP_FTRREAD);
2696 case 'W': FTST(OP_FTRWRITE);
2697 case 'X': FTST(OP_FTREXEC);
2698 case 'O': FTST(OP_FTROWNED);
2699 case 'e': FTST(OP_FTIS);
2700 case 'z': FTST(OP_FTZERO);
2701 case 's': FTST(OP_FTSIZE);
2702 case 'f': FTST(OP_FTFILE);
2703 case 'd': FTST(OP_FTDIR);
2704 case 'l': FTST(OP_FTLINK);
2705 case 'p': FTST(OP_FTPIPE);
2706 case 'S': FTST(OP_FTSOCK);
2707 case 'u': FTST(OP_FTSUID);
2708 case 'g': FTST(OP_FTSGID);
2709 case 'k': FTST(OP_FTSVTX);
2710 case 'b': FTST(OP_FTBLK);
2711 case 'c': FTST(OP_FTCHR);
2712 case 't': FTST(OP_FTTTY);
2713 case 'T': FTST(OP_FTTEXT);
2714 case 'B': FTST(OP_FTBINARY);
2715 case 'M': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTMTIME);
2716 case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME);
2717 case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME);
2719 Perl_croak(aTHX_ "Unrecognized file test: -%c", (int)tmp);
2726 if (PL_expect == XOPERATOR)
2731 else if (*s == '>') {
2734 if (isIDFIRST_lazy_if(s,UTF)) {
2735 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2743 if (PL_expect == XOPERATOR)
2746 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2748 OPERATOR('-'); /* unary minus */
2755 if (PL_expect == XOPERATOR)
2760 if (PL_expect == XOPERATOR)
2763 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2769 if (PL_expect != XOPERATOR) {
2770 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2771 PL_expect = XOPERATOR;
2772 force_ident(PL_tokenbuf, '*');
2785 if (PL_expect == XOPERATOR) {
2789 PL_tokenbuf[0] = '%';
2790 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2791 if (!PL_tokenbuf[1]) {
2793 yyerror("Final % should be \\% or %name");
2796 PL_pending_ident = '%';
2815 switch (PL_expect) {
2818 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2820 PL_bufptr = s; /* update in case we back off */
2826 PL_expect = XTERMBLOCK;
2830 while (isIDFIRST_lazy_if(s,UTF)) {
2831 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2832 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2833 if (tmp < 0) tmp = -tmp;
2848 d = scan_str(d,TRUE,TRUE);
2851 SvREFCNT_dec(PL_lex_stuff);
2852 PL_lex_stuff = Nullsv;
2854 /* MUST advance bufptr here to avoid bogus
2855 "at end of line" context messages from yyerror().
2857 PL_bufptr = s + len;
2858 yyerror("Unterminated attribute parameter in attribute list");
2861 return 0; /* EOF indicator */
2865 SV *sv = newSVpvn(s, len);
2866 sv_catsv(sv, PL_lex_stuff);
2867 attrs = append_elem(OP_LIST, attrs,
2868 newSVOP(OP_CONST, 0, sv));
2869 SvREFCNT_dec(PL_lex_stuff);
2870 PL_lex_stuff = Nullsv;
2873 attrs = append_elem(OP_LIST, attrs,
2874 newSVOP(OP_CONST, 0,
2878 if (*s == ':' && s[1] != ':')
2881 break; /* require real whitespace or :'s */
2883 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
2884 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
2885 char q = ((*s == '\'') ? '"' : '\'');
2886 /* If here for an expression, and parsed no attrs, back off. */
2887 if (tmp == '=' && !attrs) {
2891 /* MUST advance bufptr here to avoid bogus "at end of line"
2892 context messages from yyerror().
2896 yyerror("Unterminated attribute list");
2898 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
2906 PL_nextval[PL_nexttoke].opval = attrs;
2914 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
2915 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
2920 if (CopLINE(PL_curcop) < PL_copline)
2921 PL_copline = CopLINE(PL_curcop);
2932 if (PL_lex_brackets <= 0)
2933 yyerror("Unmatched right square bracket");
2936 if (PL_lex_state == LEX_INTERPNORMAL) {
2937 if (PL_lex_brackets == 0) {
2938 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
2939 PL_lex_state = LEX_INTERPEND;
2946 if (PL_lex_brackets > 100) {
2947 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
2948 if (newlb != PL_lex_brackstack) {
2950 PL_lex_brackstack = newlb;
2953 switch (PL_expect) {
2955 if (PL_lex_formbrack) {
2959 if (PL_oldoldbufptr == PL_last_lop)
2960 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2962 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2963 OPERATOR(HASHBRACK);
2965 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2968 PL_tokenbuf[0] = '\0';
2969 if (d < PL_bufend && *d == '-') {
2970 PL_tokenbuf[0] = '-';
2972 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2975 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
2976 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
2978 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2981 char minus = (PL_tokenbuf[0] == '-');
2982 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
2990 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
2995 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3000 if (PL_oldoldbufptr == PL_last_lop)
3001 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3003 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3006 OPERATOR(HASHBRACK);
3007 /* This hack serves to disambiguate a pair of curlies
3008 * as being a block or an anon hash. Normally, expectation
3009 * determines that, but in cases where we're not in a
3010 * position to expect anything in particular (like inside
3011 * eval"") we have to resolve the ambiguity. This code
3012 * covers the case where the first term in the curlies is a
3013 * quoted string. Most other cases need to be explicitly
3014 * disambiguated by prepending a `+' before the opening
3015 * curly in order to force resolution as an anon hash.
3017 * XXX should probably propagate the outer expectation
3018 * into eval"" to rely less on this hack, but that could
3019 * potentially break current behavior of eval"".
3023 if (*s == '\'' || *s == '"' || *s == '`') {
3024 /* common case: get past first string, handling escapes */
3025 for (t++; t < PL_bufend && *t != *s;)
3026 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3030 else if (*s == 'q') {
3033 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3037 char open, close, term;
3040 while (t < PL_bufend && isSPACE(*t))
3044 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3048 for (t++; t < PL_bufend; t++) {
3049 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3051 else if (*t == open)
3055 for (t++; t < PL_bufend; t++) {
3056 if (*t == '\\' && t+1 < PL_bufend)
3058 else if (*t == close && --brackets <= 0)
3060 else if (*t == open)
3066 else if (isALNUM_lazy_if(t,UTF)) {
3068 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3071 while (t < PL_bufend && isSPACE(*t))
3073 /* if comma follows first term, call it an anon hash */
3074 /* XXX it could be a comma expression with loop modifiers */
3075 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3076 || (*t == '=' && t[1] == '>')))
3077 OPERATOR(HASHBRACK);
3078 if (PL_expect == XREF)
3081 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3087 yylval.ival = CopLINE(PL_curcop);
3088 if (isSPACE(*s) || *s == '#')
3089 PL_copline = NOLINE; /* invalidate current command line number */
3094 if (PL_lex_brackets <= 0)
3095 yyerror("Unmatched right curly bracket");
3097 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3098 if (PL_lex_brackets < PL_lex_formbrack)
3099 PL_lex_formbrack = 0;
3100 if (PL_lex_state == LEX_INTERPNORMAL) {
3101 if (PL_lex_brackets == 0) {
3102 if (PL_expect & XFAKEBRACK) {
3103 PL_expect &= XENUMMASK;
3104 PL_lex_state = LEX_INTERPEND;
3106 return yylex(); /* ignore fake brackets */
3108 if (*s == '-' && s[1] == '>')
3109 PL_lex_state = LEX_INTERPENDMAYBE;
3110 else if (*s != '[' && *s != '{')
3111 PL_lex_state = LEX_INTERPEND;
3114 if (PL_expect & XFAKEBRACK) {
3115 PL_expect &= XENUMMASK;
3117 return yylex(); /* ignore fake brackets */
3127 if (PL_expect == XOPERATOR) {
3128 if (ckWARN(WARN_SEMICOLON)
3129 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3131 CopLINE_dec(PL_curcop);
3132 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3133 CopLINE_inc(PL_curcop);
3138 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3140 PL_expect = XOPERATOR;
3141 force_ident(PL_tokenbuf, '&');
3145 yylval.ival = (OPpENTERSUB_AMPER<<8);
3164 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3165 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3167 if (PL_expect == XSTATE && isALPHA(tmp) &&
3168 (s == PL_linestart+1 || s[-2] == '\n') )
3170 if (PL_in_eval && !PL_rsfp) {
3175 if (strnEQ(s,"=cut",4)) {
3189 PL_doextract = TRUE;
3192 if (PL_lex_brackets < PL_lex_formbrack) {
3194 #ifdef PERL_STRICT_CR
3195 for (t = s; *t == ' ' || *t == '\t'; t++) ;
3197 for (t = s; *t == ' ' || *t == '\t' || *t == '\r'; t++) ;
3199 if (*t == '\n' || *t == '#') {
3217 if (PL_expect != XOPERATOR) {
3218 if (s[1] != '<' && !strchr(s,'>'))
3221 s = scan_heredoc(s);
3223 s = scan_inputsymbol(s);
3224 TERM(sublex_start());
3229 SHop(OP_LEFT_SHIFT);
3243 SHop(OP_RIGHT_SHIFT);
3252 if (PL_expect == XOPERATOR) {
3253 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3256 return ','; /* grandfather non-comma-format format */
3260 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3261 PL_tokenbuf[0] = '@';
3262 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3263 sizeof PL_tokenbuf - 1, FALSE);
3264 if (PL_expect == XOPERATOR)
3265 no_op("Array length", s);
3266 if (!PL_tokenbuf[1])
3268 PL_expect = XOPERATOR;
3269 PL_pending_ident = '#';
3273 PL_tokenbuf[0] = '$';
3274 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3275 sizeof PL_tokenbuf - 1, FALSE);
3276 if (PL_expect == XOPERATOR)
3278 if (!PL_tokenbuf[1]) {
3280 yyerror("Final $ should be \\$ or $name");
3284 /* This kludge not intended to be bulletproof. */
3285 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3286 yylval.opval = newSVOP(OP_CONST, 0,
3287 newSViv(PL_compiling.cop_arybase));
3288 yylval.opval->op_private = OPpCONST_ARYBASE;
3294 if (PL_lex_state == LEX_NORMAL)
3297 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3300 PL_tokenbuf[0] = '@';
3301 if (ckWARN(WARN_SYNTAX)) {
3303 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3306 PL_bufptr = skipspace(PL_bufptr);
3307 while (t < PL_bufend && *t != ']')
3309 Perl_warner(aTHX_ WARN_SYNTAX,
3310 "Multidimensional syntax %.*s not supported",
3311 (t - PL_bufptr) + 1, PL_bufptr);
3315 else if (*s == '{') {
3316 PL_tokenbuf[0] = '%';
3317 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3318 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3320 char tmpbuf[sizeof PL_tokenbuf];
3322 for (t++; isSPACE(*t); t++) ;
3323 if (isIDFIRST_lazy_if(t,UTF)) {
3324 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3325 for (; isSPACE(*t); t++) ;
3326 if (*t == ';' && get_cv(tmpbuf, FALSE))
3327 Perl_warner(aTHX_ WARN_SYNTAX,
3328 "You need to quote \"%s\"", tmpbuf);
3334 PL_expect = XOPERATOR;
3335 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3336 bool islop = (PL_last_lop == PL_oldoldbufptr);
3337 if (!islop || PL_last_lop_op == OP_GREPSTART)
3338 PL_expect = XOPERATOR;
3339 else if (strchr("$@\"'`q", *s))
3340 PL_expect = XTERM; /* e.g. print $fh "foo" */
3341 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3342 PL_expect = XTERM; /* e.g. print $fh &sub */
3343 else if (isIDFIRST_lazy_if(s,UTF)) {
3344 char tmpbuf[sizeof PL_tokenbuf];
3345 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3346 if ((tmp = keyword(tmpbuf, len))) {
3347 /* binary operators exclude handle interpretations */
3359 PL_expect = XTERM; /* e.g. print $fh length() */
3364 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3365 if (gv && GvCVu(gv))
3366 PL_expect = XTERM; /* e.g. print $fh subr() */
3369 else if (isDIGIT(*s))
3370 PL_expect = XTERM; /* e.g. print $fh 3 */
3371 else if (*s == '.' && isDIGIT(s[1]))
3372 PL_expect = XTERM; /* e.g. print $fh .3 */
3373 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3374 PL_expect = XTERM; /* e.g. print $fh -1 */
3375 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3376 PL_expect = XTERM; /* print $fh <<"EOF" */
3378 PL_pending_ident = '$';
3382 if (PL_expect == XOPERATOR)
3384 PL_tokenbuf[0] = '@';
3385 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3386 if (!PL_tokenbuf[1]) {
3388 yyerror("Final @ should be \\@ or @name");
3391 if (PL_lex_state == LEX_NORMAL)
3393 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3395 PL_tokenbuf[0] = '%';
3397 /* Warn about @ where they meant $. */
3398 if (ckWARN(WARN_SYNTAX)) {
3399 if (*s == '[' || *s == '{') {
3401 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3403 if (*t == '}' || *t == ']') {
3405 PL_bufptr = skipspace(PL_bufptr);
3406 Perl_warner(aTHX_ WARN_SYNTAX,
3407 "Scalar value %.*s better written as $%.*s",
3408 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3413 PL_pending_ident = '@';
3416 case '/': /* may either be division or pattern */
3417 case '?': /* may either be conditional or pattern */
3418 if (PL_expect != XOPERATOR) {
3419 /* Disable warning on "study /blah/" */
3420 if (PL_oldoldbufptr == PL_last_uni
3421 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3422 || memNE(PL_last_uni, "study", 5)
3423 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3425 s = scan_pat(s,OP_MATCH);
3426 TERM(sublex_start());
3434 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3435 #ifdef PERL_STRICT_CR
3438 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3440 && (s == PL_linestart || s[-1] == '\n') )
3442 PL_lex_formbrack = 0;
3446 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3452 yylval.ival = OPf_SPECIAL;
3458 if (PL_expect != XOPERATOR)
3463 case '0': case '1': case '2': case '3': case '4':
3464 case '5': case '6': case '7': case '8': case '9':
3466 if (PL_expect == XOPERATOR)
3471 s = scan_str(s,FALSE,FALSE);
3472 if (PL_expect == XOPERATOR) {
3473 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3476 return ','; /* grandfather non-comma-format format */
3482 missingterm((char*)0);
3483 yylval.ival = OP_CONST;
3484 TERM(sublex_start());
3487 s = scan_str(s,FALSE,FALSE);
3488 if (PL_expect == XOPERATOR) {
3489 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3492 return ','; /* grandfather non-comma-format format */
3498 missingterm((char*)0);
3499 yylval.ival = OP_CONST;
3500 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3501 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3502 yylval.ival = OP_STRINGIFY;
3506 TERM(sublex_start());
3509 s = scan_str(s,FALSE,FALSE);
3510 if (PL_expect == XOPERATOR)
3511 no_op("Backticks",s);
3513 missingterm((char*)0);
3514 yylval.ival = OP_BACKTICK;
3516 TERM(sublex_start());
3520 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3521 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3523 if (PL_expect == XOPERATOR)
3524 no_op("Backslash",s);
3528 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3532 while (isDIGIT(*start) || *start == '_')
3534 if (*start == '.' && isDIGIT(start[1])) {
3538 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3539 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF)) {
3543 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3553 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3592 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3594 /* Some keywords can be followed by any delimiter, including ':' */
3595 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3596 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3597 (PL_tokenbuf[0] == 'q' &&
3598 strchr("qwxr", PL_tokenbuf[1])))));
3600 /* x::* is just a word, unless x is "CORE" */
3601 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3605 while (d < PL_bufend && isSPACE(*d))
3606 d++; /* no comments skipped here, or s### is misparsed */
3608 /* Is this a label? */
3609 if (!tmp && PL_expect == XSTATE
3610 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3612 yylval.pval = savepv(PL_tokenbuf);
3617 /* Check for keywords */
3618 tmp = keyword(PL_tokenbuf, len);
3620 /* Is this a word before a => operator? */
3621 if (strnEQ(d,"=>",2)) {
3623 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3624 yylval.opval->op_private = OPpCONST_BARE;
3628 if (tmp < 0) { /* second-class keyword? */
3629 GV *ogv = Nullgv; /* override (winner) */
3630 GV *hgv = Nullgv; /* hidden (loser) */
3631 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3633 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3636 if (GvIMPORTED_CV(gv))
3638 else if (! CvMETHOD(cv))
3642 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3643 (gv = *gvp) != (GV*)&PL_sv_undef &&
3644 GvCVu(gv) && GvIMPORTED_CV(gv))
3650 tmp = 0; /* overridden by import or by GLOBAL */
3653 && -tmp==KEY_lock /* XXX generalizable kludge */
3655 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3657 tmp = 0; /* any sub overrides "weak" keyword */
3659 else { /* no override */
3663 if (ckWARN(WARN_AMBIGUOUS) && hgv
3664 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3665 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3666 "Ambiguous call resolved as CORE::%s(), %s",
3667 GvENAME(hgv), "qualify as such or use &");
3674 default: /* not a keyword */
3677 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3679 /* Get the rest if it looks like a package qualifier */
3681 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3683 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3686 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3687 *s == '\'' ? "'" : "::");
3691 if (PL_expect == XOPERATOR) {
3692 if (PL_bufptr == PL_linestart) {
3693 CopLINE_dec(PL_curcop);
3694 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3695 CopLINE_inc(PL_curcop);
3698 no_op("Bareword",s);
3701 /* Look for a subroutine with this name in current package,
3702 unless name is "Foo::", in which case Foo is a bearword
3703 (and a package name). */
3706 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3708 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3709 Perl_warner(aTHX_ WARN_BAREWORD,
3710 "Bareword \"%s\" refers to nonexistent package",
3713 PL_tokenbuf[len] = '\0';
3720 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3723 /* if we saw a global override before, get the right name */
3726 sv = newSVpvn("CORE::GLOBAL::",14);
3727 sv_catpv(sv,PL_tokenbuf);
3730 sv = newSVpv(PL_tokenbuf,0);
3732 /* Presume this is going to be a bareword of some sort. */
3735 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3736 yylval.opval->op_private = OPpCONST_BARE;
3738 /* And if "Foo::", then that's what it certainly is. */
3743 /* See if it's the indirect object for a list operator. */
3745 if (PL_oldoldbufptr &&
3746 PL_oldoldbufptr < PL_bufptr &&
3747 (PL_oldoldbufptr == PL_last_lop
3748 || PL_oldoldbufptr == PL_last_uni) &&
3749 /* NO SKIPSPACE BEFORE HERE! */
3750 (PL_expect == XREF ||
3751 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3753 bool immediate_paren = *s == '(';
3755 /* (Now we can afford to cross potential line boundary.) */
3758 /* Two barewords in a row may indicate method call. */
3760 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3763 /* If not a declared subroutine, it's an indirect object. */
3764 /* (But it's an indir obj regardless for sort.) */
3766 if ((PL_last_lop_op == OP_SORT ||
3767 (!immediate_paren && (!gv || !GvCVu(gv)))) &&
3768 (PL_last_lop_op != OP_MAPSTART &&
3769 PL_last_lop_op != OP_GREPSTART))
3771 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3776 /* If followed by a paren, it's certainly a subroutine. */
3778 PL_expect = XOPERATOR;
3782 if (gv && GvCVu(gv)) {
3783 for (d = s + 1; *d == ' ' || *d == '\t'; d++) ;
3784 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3789 PL_nextval[PL_nexttoke].opval = yylval.opval;
3790 PL_expect = XOPERATOR;
3796 /* If followed by var or block, call it a method (unless sub) */
3798 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3799 PL_last_lop = PL_oldbufptr;
3800 PL_last_lop_op = OP_METHOD;
3804 /* If followed by a bareword, see if it looks like indir obj. */
3806 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3809 /* Not a method, so call it a subroutine (if defined) */
3811 if (gv && GvCVu(gv)) {
3813 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3814 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3815 "Ambiguous use of -%s resolved as -&%s()",
3816 PL_tokenbuf, PL_tokenbuf);
3817 /* Check for a constant sub */
3819 if ((sv = cv_const_sv(cv))) {
3821 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3822 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3823 yylval.opval->op_private = 0;
3827 /* Resolve to GV now. */
3828 op_free(yylval.opval);
3829 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3830 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3831 PL_last_lop = PL_oldbufptr;
3832 PL_last_lop_op = OP_ENTERSUB;
3833 /* Is there a prototype? */
3836 char *proto = SvPV((SV*)cv, len);
3839 if (strEQ(proto, "$"))
3841 if (*proto == '&' && *s == '{') {
3842 sv_setpv(PL_subname,"__ANON__");
3846 PL_nextval[PL_nexttoke].opval = yylval.opval;
3852 /* Call it a bare word */
3854 if (PL_hints & HINT_STRICT_SUBS)
3855 yylval.opval->op_private |= OPpCONST_STRICT;
3858 if (ckWARN(WARN_RESERVED)) {
3859 if (lastchar != '-') {
3860 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
3862 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
3869 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
3870 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3871 "Operator or semicolon missing before %c%s",
3872 lastchar, PL_tokenbuf);
3873 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3874 "Ambiguous use of %c resolved as operator %c",
3875 lastchar, lastchar);
3881 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3882 newSVpv(CopFILE(PL_curcop),0));
3886 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3887 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
3890 case KEY___PACKAGE__:
3891 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3893 ? newSVsv(PL_curstname)
3902 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
3903 char *pname = "main";
3904 if (PL_tokenbuf[2] == 'D')
3905 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
3906 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
3909 GvIOp(gv) = newIO();
3910 IoIFP(GvIOp(gv)) = PL_rsfp;
3911 #if defined(HAS_FCNTL) && defined(F_SETFD)
3913 int fd = PerlIO_fileno(PL_rsfp);
3914 fcntl(fd,F_SETFD,fd >= 3);
3917 /* Mark this internal pseudo-handle as clean */
3918 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
3920 IoTYPE(GvIOp(gv)) = '|';
3921 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
3922 IoTYPE(GvIOp(gv)) = '-';
3924 IoTYPE(GvIOp(gv)) = '<';
3925 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
3926 /* if the script was opened in binmode, we need to revert
3927 * it to text mode for compatibility; but only iff it has CRs
3928 * XXX this is a questionable hack at best. */
3929 if (PL_bufend-PL_bufptr > 2
3930 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
3933 if (IoTYPE(GvIOp(gv)) == '<') {
3934 loc = PerlIO_tell(PL_rsfp);
3935 (void)PerlIO_seek(PL_rsfp, 0L, 0);
3937 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
3938 #if defined(__BORLANDC__)
3939 /* XXX see note in do_binmode() */
3940 ((FILE*)PL_rsfp)->flags |= _F_BIN;
3943 PerlIO_seek(PL_rsfp, loc, 0);
3958 if (PL_expect == XSTATE) {
3965 if (*s == ':' && s[1] == ':') {
3968 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3969 if (!(tmp = keyword(PL_tokenbuf, len)))
3970 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
3984 LOP(OP_ACCEPT,XTERM);
3990 LOP(OP_ATAN2,XTERM);
3996 LOP(OP_BINMODE,XTERM);
3999 LOP(OP_BLESS,XTERM);
4008 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4025 if (!PL_cryptseen) {
4026 PL_cryptseen = TRUE;
4030 LOP(OP_CRYPT,XTERM);
4033 if (ckWARN(WARN_CHMOD)) {
4034 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4035 if (*d != '0' && isDIGIT(*d))
4036 Perl_warner(aTHX_ WARN_CHMOD,
4037 "chmod() mode argument is missing initial 0");
4039 LOP(OP_CHMOD,XTERM);
4042 LOP(OP_CHOWN,XTERM);
4045 LOP(OP_CONNECT,XTERM);
4061 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4065 PL_hints |= HINT_BLOCK_SCOPE;
4075 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4076 LOP(OP_DBMOPEN,XTERM);
4082 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4089 yylval.ival = CopLINE(PL_curcop);
4103 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4104 UNIBRACK(OP_ENTEREVAL);
4119 case KEY_endhostent:
4125 case KEY_endservent:
4128 case KEY_endprotoent:
4139 yylval.ival = CopLINE(PL_curcop);
4141 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4143 if ((PL_bufend - p) >= 3 &&
4144 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4146 else if ((PL_bufend - p) >= 4 &&
4147 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4150 if (isIDFIRST_lazy_if(p,UTF)) {
4151 p = scan_ident(p, PL_bufend,
4152 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4156 Perl_croak(aTHX_ "Missing $ on loop variable");
4161 LOP(OP_FORMLINE,XTERM);
4167 LOP(OP_FCNTL,XTERM);
4173 LOP(OP_FLOCK,XTERM);
4182 LOP(OP_GREPSTART, XREF);
4185 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4200 case KEY_getpriority:
4201 LOP(OP_GETPRIORITY,XTERM);
4203 case KEY_getprotobyname:
4206 case KEY_getprotobynumber:
4207 LOP(OP_GPBYNUMBER,XTERM);
4209 case KEY_getprotoent:
4221 case KEY_getpeername:
4222 UNI(OP_GETPEERNAME);
4224 case KEY_gethostbyname:
4227 case KEY_gethostbyaddr:
4228 LOP(OP_GHBYADDR,XTERM);
4230 case KEY_gethostent:
4233 case KEY_getnetbyname:
4236 case KEY_getnetbyaddr:
4237 LOP(OP_GNBYADDR,XTERM);
4242 case KEY_getservbyname:
4243 LOP(OP_GSBYNAME,XTERM);
4245 case KEY_getservbyport:
4246 LOP(OP_GSBYPORT,XTERM);
4248 case KEY_getservent:
4251 case KEY_getsockname:
4252 UNI(OP_GETSOCKNAME);
4254 case KEY_getsockopt:
4255 LOP(OP_GSOCKOPT,XTERM);
4277 yylval.ival = CopLINE(PL_curcop);
4281 LOP(OP_INDEX,XTERM);
4287 LOP(OP_IOCTL,XTERM);
4299 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4331 LOP(OP_LISTEN,XTERM);
4340 s = scan_pat(s,OP_MATCH);
4341 TERM(sublex_start());
4344 LOP(OP_MAPSTART, XREF);
4347 LOP(OP_MKDIR,XTERM);
4350 LOP(OP_MSGCTL,XTERM);
4353 LOP(OP_MSGGET,XTERM);
4356 LOP(OP_MSGRCV,XTERM);
4359 LOP(OP_MSGSND,XTERM);
4365 if (isIDFIRST_lazy_if(s,UTF)) {
4366 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4367 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4369 PL_in_my_stash = gv_stashpv(PL_tokenbuf, FALSE);
4370 if (!PL_in_my_stash) {
4373 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4381 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4388 if (PL_expect != XSTATE)
4389 yyerror("\"no\" not allowed in expression");
4390 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4391 s = force_version(s);
4396 if (*s == '(' || (s = skipspace(s), *s == '('))
4403 if (isIDFIRST_lazy_if(s,UTF)) {
4405 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4407 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4408 Perl_warner(aTHX_ WARN_PRECEDENCE,
4409 "Precedence problem: open %.*s should be open(%.*s)",
4415 yylval.ival = OP_OR;
4425 LOP(OP_OPEN_DIR,XTERM);
4428 checkcomma(s,PL_tokenbuf,"filehandle");
4432 checkcomma(s,PL_tokenbuf,"filehandle");
4451 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4455 LOP(OP_PIPE_OP,XTERM);
4458 s = scan_str(s,FALSE,FALSE);
4460 missingterm((char*)0);
4461 yylval.ival = OP_CONST;
4462 TERM(sublex_start());
4468 s = scan_str(s,FALSE,FALSE);
4470 missingterm((char*)0);
4472 if (SvCUR(PL_lex_stuff)) {
4475 d = SvPV_force(PL_lex_stuff, len);
4477 for (; isSPACE(*d) && len; --len, ++d) ;
4480 if (!warned && ckWARN(WARN_QW)) {
4481 for (; !isSPACE(*d) && len; --len, ++d) {
4483 Perl_warner(aTHX_ WARN_QW,
4484 "Possible attempt to separate words with commas");
4487 else if (*d == '#') {
4488 Perl_warner(aTHX_ WARN_QW,
4489 "Possible attempt to put comments in qw() list");
4495 for (; !isSPACE(*d) && len; --len, ++d) ;
4497 words = append_elem(OP_LIST, words,
4498 newSVOP(OP_CONST, 0, newSVpvn(b, d-b)));
4502 PL_nextval[PL_nexttoke].opval = words;
4507 SvREFCNT_dec(PL_lex_stuff);
4508 PL_lex_stuff = Nullsv;
4513 s = scan_str(s,FALSE,FALSE);
4515 missingterm((char*)0);
4516 yylval.ival = OP_STRINGIFY;
4517 if (SvIVX(PL_lex_stuff) == '\'')
4518 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4519 TERM(sublex_start());
4522 s = scan_pat(s,OP_QR);
4523 TERM(sublex_start());
4526 s = scan_str(s,FALSE,FALSE);
4528 missingterm((char*)0);
4529 yylval.ival = OP_BACKTICK;
4531 TERM(sublex_start());
4538 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4539 s = force_version(s);
4542 *PL_tokenbuf = '\0';
4543 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4544 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4545 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4547 yyerror("<> should be quotes");
4555 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4559 LOP(OP_RENAME,XTERM);
4568 LOP(OP_RINDEX,XTERM);
4591 LOP(OP_REVERSE,XTERM);
4602 TERM(sublex_start());
4604 TOKEN(1); /* force error */
4613 LOP(OP_SELECT,XTERM);
4619 LOP(OP_SEMCTL,XTERM);
4622 LOP(OP_SEMGET,XTERM);
4625 LOP(OP_SEMOP,XTERM);
4631 LOP(OP_SETPGRP,XTERM);
4633 case KEY_setpriority:
4634 LOP(OP_SETPRIORITY,XTERM);
4636 case KEY_sethostent:
4642 case KEY_setservent:
4645 case KEY_setprotoent:
4655 LOP(OP_SEEKDIR,XTERM);
4657 case KEY_setsockopt:
4658 LOP(OP_SSOCKOPT,XTERM);
4664 LOP(OP_SHMCTL,XTERM);
4667 LOP(OP_SHMGET,XTERM);
4670 LOP(OP_SHMREAD,XTERM);
4673 LOP(OP_SHMWRITE,XTERM);
4676 LOP(OP_SHUTDOWN,XTERM);
4685 LOP(OP_SOCKET,XTERM);
4687 case KEY_socketpair:
4688 LOP(OP_SOCKPAIR,XTERM);
4691 checkcomma(s,PL_tokenbuf,"subroutine name");
4693 if (*s == ';' || *s == ')') /* probably a close */
4694 Perl_croak(aTHX_ "sort is now a reserved word");
4696 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4700 LOP(OP_SPLIT,XTERM);
4703 LOP(OP_SPRINTF,XTERM);
4706 LOP(OP_SPLICE,XTERM);
4721 LOP(OP_SUBSTR,XTERM);
4727 char tmpbuf[sizeof PL_tokenbuf];
4729 expectation attrful;
4730 bool have_name, have_proto;
4735 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4736 (*s == ':' && s[1] == ':'))
4739 attrful = XATTRBLOCK;
4740 /* remember buffer pos'n for later force_word */
4741 tboffset = s - PL_oldbufptr;
4742 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4743 if (strchr(tmpbuf, ':'))
4744 sv_setpv(PL_subname, tmpbuf);
4746 sv_setsv(PL_subname,PL_curstname);
4747 sv_catpvn(PL_subname,"::",2);
4748 sv_catpvn(PL_subname,tmpbuf,len);
4755 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4756 PL_expect = XTERMBLOCK;
4757 attrful = XATTRTERM;
4758 sv_setpv(PL_subname,"?");
4762 if (key == KEY_format) {
4764 PL_lex_formbrack = PL_lex_brackets + 1;
4766 (void) force_word(PL_oldbufptr + tboffset, WORD,
4771 /* Look for a prototype */
4775 s = scan_str(s,FALSE,FALSE);
4778 SvREFCNT_dec(PL_lex_stuff);
4779 PL_lex_stuff = Nullsv;
4780 Perl_croak(aTHX_ "Prototype not terminated");
4783 d = SvPVX(PL_lex_stuff);
4785 for (p = d; *p; ++p) {
4790 SvCUR(PL_lex_stuff) = tmp;
4798 if (*s == ':' && s[1] != ':')
4799 PL_expect = attrful;
4802 PL_nextval[PL_nexttoke].opval =
4803 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4804 PL_lex_stuff = Nullsv;
4808 sv_setpv(PL_subname,"__ANON__");
4811 (void) force_word(PL_oldbufptr + tboffset, WORD,
4820 LOP(OP_SYSTEM,XREF);
4823 LOP(OP_SYMLINK,XTERM);
4826 LOP(OP_SYSCALL,XTERM);
4829 LOP(OP_SYSOPEN,XTERM);
4832 LOP(OP_SYSSEEK,XTERM);
4835 LOP(OP_SYSREAD,XTERM);
4838 LOP(OP_SYSWRITE,XTERM);
4842 TERM(sublex_start());
4863 LOP(OP_TRUNCATE,XTERM);
4875 yylval.ival = CopLINE(PL_curcop);
4879 yylval.ival = CopLINE(PL_curcop);
4883 LOP(OP_UNLINK,XTERM);
4889 LOP(OP_UNPACK,XTERM);
4892 LOP(OP_UTIME,XTERM);
4895 if (ckWARN(WARN_UMASK)) {
4896 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4897 if (*d != '0' && isDIGIT(*d))
4898 Perl_warner(aTHX_ WARN_UMASK,
4899 "umask: argument is missing initial 0");
4904 LOP(OP_UNSHIFT,XTERM);
4907 if (PL_expect != XSTATE)
4908 yyerror("\"use\" not allowed in expression");
4910 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4911 s = force_version(s);
4912 if (*s == ';' || (s = skipspace(s), *s == ';')) {
4913 PL_nextval[PL_nexttoke].opval = Nullop;
4918 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4919 s = force_version(s);
4931 yylval.ival = CopLINE(PL_curcop);
4935 PL_hints |= HINT_BLOCK_SCOPE;
4942 LOP(OP_WAITPID,XTERM);
4950 static char ctl_l[2];
4952 if (ctl_l[0] == '\0')
4953 ctl_l[0] = toCTRL('L');
4954 gv_fetchpv(ctl_l,TRUE, SVt_PV);
4957 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
4962 if (PL_expect == XOPERATOR)
4968 yylval.ival = OP_XOR;
4973 TERM(sublex_start());
4979 Perl_keyword(pTHX_ register char *d, I32 len)
4984 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
4985 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
4986 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
4987 if (strEQ(d,"__DATA__")) return KEY___DATA__;
4988 if (strEQ(d,"__END__")) return KEY___END__;
4992 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
4997 if (strEQ(d,"and")) return -KEY_and;
4998 if (strEQ(d,"abs")) return -KEY_abs;
5001 if (strEQ(d,"alarm")) return -KEY_alarm;
5002 if (strEQ(d,"atan2")) return -KEY_atan2;
5005 if (strEQ(d,"accept")) return -KEY_accept;
5010 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5013 if (strEQ(d,"bless")) return -KEY_bless;
5014 if (strEQ(d,"bind")) return -KEY_bind;
5015 if (strEQ(d,"binmode")) return -KEY_binmode;
5018 if (strEQ(d,"CORE")) return -KEY_CORE;
5019 if (strEQ(d,"CHECK")) return KEY_CHECK;
5024 if (strEQ(d,"cmp")) return -KEY_cmp;
5025 if (strEQ(d,"chr")) return -KEY_chr;
5026 if (strEQ(d,"cos")) return -KEY_cos;
5029 if (strEQ(d,"chop")) return KEY_chop;
5032 if (strEQ(d,"close")) return -KEY_close;
5033 if (strEQ(d,"chdir")) return -KEY_chdir;
5034 if (strEQ(d,"chomp")) return KEY_chomp;
5035 if (strEQ(d,"chmod")) return -KEY_chmod;
5036 if (strEQ(d,"chown")) return -KEY_chown;
5037 if (strEQ(d,"crypt")) return -KEY_crypt;
5040 if (strEQ(d,"chroot")) return -KEY_chroot;
5041 if (strEQ(d,"caller")) return -KEY_caller;
5044 if (strEQ(d,"connect")) return -KEY_connect;
5047 if (strEQ(d,"closedir")) return -KEY_closedir;
5048 if (strEQ(d,"continue")) return -KEY_continue;
5053 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5058 if (strEQ(d,"do")) return KEY_do;
5061 if (strEQ(d,"die")) return -KEY_die;
5064 if (strEQ(d,"dump")) return -KEY_dump;
5067 if (strEQ(d,"delete")) return KEY_delete;
5070 if (strEQ(d,"defined")) return KEY_defined;
5071 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5074 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5079 if (strEQ(d,"EQ")) { deprecate(d); return -KEY_eq;}
5080 if (strEQ(d,"END")) return KEY_END;
5085 if (strEQ(d,"eq")) return -KEY_eq;
5088 if (strEQ(d,"eof")) return -KEY_eof;
5089 if (strEQ(d,"exp")) return -KEY_exp;
5092 if (strEQ(d,"else")) return KEY_else;
5093 if (strEQ(d,"exit")) return -KEY_exit;
5094 if (strEQ(d,"eval")) return KEY_eval;
5095 if (strEQ(d,"exec")) return -KEY_exec;
5096 if (strEQ(d,"each")) return KEY_each;
5099 if (strEQ(d,"elsif")) return KEY_elsif;
5102 if (strEQ(d,"exists")) return KEY_exists;
5103 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5106 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5107 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5110 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5113 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5114 if (strEQ(d,"endservent")) return -KEY_endservent;
5117 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5124 if (strEQ(d,"for")) return KEY_for;
5127 if (strEQ(d,"fork")) return -KEY_fork;
5130 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5131 if (strEQ(d,"flock")) return -KEY_flock;
5134 if (strEQ(d,"format")) return KEY_format;
5135 if (strEQ(d,"fileno")) return -KEY_fileno;
5138 if (strEQ(d,"foreach")) return KEY_foreach;
5141 if (strEQ(d,"formline")) return -KEY_formline;
5147 if (strEQ(d,"GT")) { deprecate(d); return -KEY_gt;}
5148 if (strEQ(d,"GE")) { deprecate(d); return -KEY_ge;}
5152 if (strnEQ(d,"get",3)) {
5157 if (strEQ(d,"ppid")) return -KEY_getppid;
5158 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5161 if (strEQ(d,"pwent")) return -KEY_getpwent;
5162 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5163 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5166 if (strEQ(d,"peername")) return -KEY_getpeername;
5167 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5168 if (strEQ(d,"priority")) return -KEY_getpriority;
5171 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5174 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5178 else if (*d == 'h') {
5179 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5180 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5181 if (strEQ(d,"hostent")) return -KEY_gethostent;
5183 else if (*d == 'n') {
5184 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5185 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5186 if (strEQ(d,"netent")) return -KEY_getnetent;
5188 else if (*d == 's') {
5189 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5190 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5191 if (strEQ(d,"servent")) return -KEY_getservent;
5192 if (strEQ(d,"sockname")) return -KEY_getsockname;
5193 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5195 else if (*d == 'g') {
5196 if (strEQ(d,"grent")) return -KEY_getgrent;
5197 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5198 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5200 else if (*d == 'l') {
5201 if (strEQ(d,"login")) return -KEY_getlogin;
5203 else if (strEQ(d,"c")) return -KEY_getc;
5208 if (strEQ(d,"gt")) return -KEY_gt;
5209 if (strEQ(d,"ge")) return -KEY_ge;
5212 if (strEQ(d,"grep")) return KEY_grep;
5213 if (strEQ(d,"goto")) return KEY_goto;
5214 if (strEQ(d,"glob")) return KEY_glob;
5217 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5222 if (strEQ(d,"hex")) return -KEY_hex;
5225 if (strEQ(d,"INIT")) return KEY_INIT;
5230 if (strEQ(d,"if")) return KEY_if;
5233 if (strEQ(d,"int")) return -KEY_int;
5236 if (strEQ(d,"index")) return -KEY_index;
5237 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5242 if (strEQ(d,"join")) return -KEY_join;
5246 if (strEQ(d,"keys")) return KEY_keys;
5247 if (strEQ(d,"kill")) return -KEY_kill;
5252 if (strEQ(d,"LT")) { deprecate(d); return -KEY_lt;}
5253 if (strEQ(d,"LE")) { deprecate(d); return -KEY_le;}
5259 if (strEQ(d,"lt")) return -KEY_lt;
5260 if (strEQ(d,"le")) return -KEY_le;
5261 if (strEQ(d,"lc")) return -KEY_lc;
5264 if (strEQ(d,"log")) return -KEY_log;
5267 if (strEQ(d,"last")) return KEY_last;
5268 if (strEQ(d,"link")) return -KEY_link;
5269 if (strEQ(d,"lock")) return -KEY_lock;
5272 if (strEQ(d,"local")) return KEY_local;
5273 if (strEQ(d,"lstat")) return -KEY_lstat;
5276 if (strEQ(d,"length")) return -KEY_length;
5277 if (strEQ(d,"listen")) return -KEY_listen;
5280 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5283 if (strEQ(d,"localtime")) return -KEY_localtime;
5289 case 1: return KEY_m;
5291 if (strEQ(d,"my")) return KEY_my;
5294 if (strEQ(d,"map")) return KEY_map;
5297 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5300 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5301 if (strEQ(d,"msgget")) return -KEY_msgget;
5302 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5303 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5308 if (strEQ(d,"NE")) { deprecate(d); return -KEY_ne;}
5311 if (strEQ(d,"next")) return KEY_next;
5312 if (strEQ(d,"ne")) return -KEY_ne;
5313 if (strEQ(d,"not")) return -KEY_not;
5314 if (strEQ(d,"no")) return KEY_no;
5319 if (strEQ(d,"or")) return -KEY_or;
5322 if (strEQ(d,"ord")) return -KEY_ord;
5323 if (strEQ(d,"oct")) return -KEY_oct;
5324 if (strEQ(d,"our")) return KEY_our;
5327 if (strEQ(d,"open")) return -KEY_open;
5330 if (strEQ(d,"opendir")) return -KEY_opendir;
5337 if (strEQ(d,"pop")) return KEY_pop;
5338 if (strEQ(d,"pos")) return KEY_pos;
5341 if (strEQ(d,"push")) return KEY_push;
5342 if (strEQ(d,"pack")) return -KEY_pack;
5343 if (strEQ(d,"pipe")) return -KEY_pipe;
5346 if (strEQ(d,"print")) return KEY_print;
5349 if (strEQ(d,"printf")) return KEY_printf;
5352 if (strEQ(d,"package")) return KEY_package;
5355 if (strEQ(d,"prototype")) return KEY_prototype;
5360 if (strEQ(d,"q")) return KEY_q;
5361 if (strEQ(d,"qr")) return KEY_qr;
5362 if (strEQ(d,"qq")) return KEY_qq;
5363 if (strEQ(d,"qw")) return KEY_qw;
5364 if (strEQ(d,"qx")) return KEY_qx;
5366 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5371 if (strEQ(d,"ref")) return -KEY_ref;
5374 if (strEQ(d,"read")) return -KEY_read;
5375 if (strEQ(d,"rand")) return -KEY_rand;
5376 if (strEQ(d,"recv")) return -KEY_recv;
5377 if (strEQ(d,"redo")) return KEY_redo;
5380 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5381 if (strEQ(d,"reset")) return -KEY_reset;
5384 if (strEQ(d,"return")) return KEY_return;
5385 if (strEQ(d,"rename")) return -KEY_rename;
5386 if (strEQ(d,"rindex")) return -KEY_rindex;
5389 if (strEQ(d,"require")) return -KEY_require;
5390 if (strEQ(d,"reverse")) return -KEY_reverse;
5391 if (strEQ(d,"readdir")) return -KEY_readdir;
5394 if (strEQ(d,"readlink")) return -KEY_readlink;
5395 if (strEQ(d,"readline")) return -KEY_readline;
5396 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5399 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5405 case 0: return KEY_s;
5407 if (strEQ(d,"scalar")) return KEY_scalar;
5412 if (strEQ(d,"seek")) return -KEY_seek;
5413 if (strEQ(d,"send")) return -KEY_send;
5416 if (strEQ(d,"semop")) return -KEY_semop;
5419 if (strEQ(d,"select")) return -KEY_select;
5420 if (strEQ(d,"semctl")) return -KEY_semctl;
5421 if (strEQ(d,"semget")) return -KEY_semget;
5424 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5425 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5428 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5429 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5432 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5435 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5436 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5437 if (strEQ(d,"setservent")) return -KEY_setservent;
5440 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5441 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5448 if (strEQ(d,"shift")) return KEY_shift;
5451 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5452 if (strEQ(d,"shmget")) return -KEY_shmget;
5455 if (strEQ(d,"shmread")) return -KEY_shmread;
5458 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5459 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5464 if (strEQ(d,"sin")) return -KEY_sin;
5467 if (strEQ(d,"sleep")) return -KEY_sleep;
5470 if (strEQ(d,"sort")) return KEY_sort;
5471 if (strEQ(d,"socket")) return -KEY_socket;
5472 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5475 if (strEQ(d,"split")) return KEY_split;
5476 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5477 if (strEQ(d,"splice")) return KEY_splice;
5480 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5483 if (strEQ(d,"srand")) return -KEY_srand;
5486 if (strEQ(d,"stat")) return -KEY_stat;
5487 if (strEQ(d,"study")) return KEY_study;
5490 if (strEQ(d,"substr")) return -KEY_substr;
5491 if (strEQ(d,"sub")) return KEY_sub;
5496 if (strEQ(d,"system")) return -KEY_system;
5499 if (strEQ(d,"symlink")) return -KEY_symlink;
5500 if (strEQ(d,"syscall")) return -KEY_syscall;
5501 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5502 if (strEQ(d,"sysread")) return -KEY_sysread;
5503 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5506 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5515 if (strEQ(d,"tr")) return KEY_tr;
5518 if (strEQ(d,"tie")) return KEY_tie;
5521 if (strEQ(d,"tell")) return -KEY_tell;
5522 if (strEQ(d,"tied")) return KEY_tied;
5523 if (strEQ(d,"time")) return -KEY_time;
5526 if (strEQ(d,"times")) return -KEY_times;
5529 if (strEQ(d,"telldir")) return -KEY_telldir;
5532 if (strEQ(d,"truncate")) return -KEY_truncate;
5539 if (strEQ(d,"uc")) return -KEY_uc;
5542 if (strEQ(d,"use")) return KEY_use;
5545 if (strEQ(d,"undef")) return KEY_undef;
5546 if (strEQ(d,"until")) return KEY_until;
5547 if (strEQ(d,"untie")) return KEY_untie;
5548 if (strEQ(d,"utime")) return -KEY_utime;
5549 if (strEQ(d,"umask")) return -KEY_umask;
5552 if (strEQ(d,"unless")) return KEY_unless;
5553 if (strEQ(d,"unpack")) return -KEY_unpack;
5554 if (strEQ(d,"unlink")) return -KEY_unlink;
5557 if (strEQ(d,"unshift")) return KEY_unshift;
5558 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5563 if (strEQ(d,"values")) return -KEY_values;
5564 if (strEQ(d,"vec")) return -KEY_vec;
5569 if (strEQ(d,"warn")) return -KEY_warn;
5570 if (strEQ(d,"wait")) return -KEY_wait;
5573 if (strEQ(d,"while")) return KEY_while;
5574 if (strEQ(d,"write")) return -KEY_write;
5577 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5580 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5585 if (len == 1) return -KEY_x;
5586 if (strEQ(d,"xor")) return -KEY_xor;
5589 if (len == 1) return KEY_y;
5598 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5602 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5603 dTHR; /* only for ckWARN */
5604 if (ckWARN(WARN_SYNTAX)) {
5606 for (w = s+2; *w && level; w++) {
5613 for (; *w && isSPACE(*w); w++) ;
5614 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5615 Perl_warner(aTHX_ WARN_SYNTAX,
5616 "%s (...) interpreted as function",name);
5619 while (s < PL_bufend && isSPACE(*s))
5623 while (s < PL_bufend && isSPACE(*s))
5625 if (isIDFIRST_lazy_if(s,UTF)) {
5627 while (isALNUM_lazy_if(s,UTF))
5629 while (s < PL_bufend && isSPACE(*s))
5634 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5638 Perl_croak(aTHX_ "No comma allowed after %s", what);
5643 /* Either returns sv, or mortalizes sv and returns a new SV*.
5644 Best used as sv=new_constant(..., sv, ...).
5645 If s, pv are NULL, calls subroutine with one argument,
5646 and type is used with error messages only. */
5649 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5653 HV *table = GvHV(PL_hintgv); /* ^H */
5657 const char *why1, *why2, *why3;
5659 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
5662 why1 = "%^H is not consistent";
5663 why2 = strEQ(key,"charnames")
5664 ? " (missing \"use charnames ...\"?)"
5668 msg = Perl_newSVpvf(aTHX_ "constant(%s): %s%s%s",
5669 (type ? type: "undef"), why1, why2, why3);
5670 yyerror(SvPVX(msg));
5674 cvp = hv_fetch(table, key, strlen(key), FALSE);
5675 if (!cvp || !SvOK(*cvp)) {
5678 why3 = "} is not defined";
5681 sv_2mortal(sv); /* Parent created it permanently */
5684 pv = sv_2mortal(newSVpvn(s, len));
5686 typesv = sv_2mortal(newSVpv(type, 0));
5688 typesv = &PL_sv_undef;
5690 PUSHSTACKi(PERLSI_OVERLOAD);
5703 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5707 /* Check the eval first */
5708 if (!PL_in_eval && SvTRUE(ERRSV)) {
5710 sv_catpv(ERRSV, "Propagated");
5711 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5713 res = SvREFCNT_inc(sv);
5717 (void)SvREFCNT_inc(res);
5726 why1 = "Call to &{$^H{";
5728 why3 = "}} did not return a defined value";
5737 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5739 register char *d = dest;
5740 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5743 Perl_croak(aTHX_ ident_too_long);
5744 if (isALNUM(*s)) /* UTF handled below */
5746 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
5751 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5755 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5756 char *t = s + UTF8SKIP(s);
5757 while (*t & 0x80 && is_utf8_mark((U8*)t))
5759 if (d + (t - s) > e)
5760 Perl_croak(aTHX_ ident_too_long);
5761 Copy(s, d, t - s, char);
5774 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5784 e = d + destlen - 3; /* two-character token, ending NUL */
5786 while (isDIGIT(*s)) {
5788 Perl_croak(aTHX_ ident_too_long);
5795 Perl_croak(aTHX_ ident_too_long);
5796 if (isALNUM(*s)) /* UTF handled below */
5798 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
5803 else if (*s == ':' && s[1] == ':') {
5807 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5808 char *t = s + UTF8SKIP(s);
5809 while (*t & 0x80 && is_utf8_mark((U8*)t))
5811 if (d + (t - s) > e)
5812 Perl_croak(aTHX_ ident_too_long);
5813 Copy(s, d, t - s, char);
5824 if (PL_lex_state != LEX_NORMAL)
5825 PL_lex_state = LEX_INTERPENDMAYBE;
5828 if (*s == '$' && s[1] &&
5829 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
5842 if (*d == '^' && *s && isCONTROLVAR(*s)) {
5847 if (isSPACE(s[-1])) {
5850 if (ch != ' ' && ch != '\t') {
5856 if (isIDFIRST_lazy_if(d,UTF)) {
5860 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
5862 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
5865 Copy(s, d, e - s, char);
5870 while ((isALNUM(*s) || *s == ':') && d < e)
5873 Perl_croak(aTHX_ ident_too_long);
5876 while (s < send && (*s == ' ' || *s == '\t')) s++;
5877 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
5878 dTHR; /* only for ckWARN */
5879 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
5880 const char *brack = *s == '[' ? "[...]" : "{...}";
5881 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5882 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
5883 funny, dest, brack, funny, dest, brack);
5886 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
5890 /* Handle extended ${^Foo} variables
5891 * 1999-02-27 mjd-perl-patch@plover.com */
5892 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
5896 while (isALNUM(*s) && d < e) {
5900 Perl_croak(aTHX_ ident_too_long);
5905 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
5906 PL_lex_state = LEX_INTERPEND;
5909 if (PL_lex_state == LEX_NORMAL) {
5910 dTHR; /* only for ckWARN */
5911 if (ckWARN(WARN_AMBIGUOUS) &&
5912 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
5914 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5915 "Ambiguous use of %c{%s} resolved to %c%s",
5916 funny, dest, funny, dest);
5921 s = bracket; /* let the parser handle it */
5925 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
5926 PL_lex_state = LEX_INTERPEND;
5931 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
5936 *pmfl |= PMf_GLOBAL;
5938 *pmfl |= PMf_CONTINUE;
5942 *pmfl |= PMf_MULTILINE;
5944 *pmfl |= PMf_SINGLELINE;
5946 *pmfl |= PMf_EXTENDED;
5950 S_scan_pat(pTHX_ char *start, I32 type)
5955 s = scan_str(start,FALSE,FALSE);
5958 SvREFCNT_dec(PL_lex_stuff);
5959 PL_lex_stuff = Nullsv;
5960 Perl_croak(aTHX_ "Search pattern not terminated");
5963 pm = (PMOP*)newPMOP(type, 0);
5964 if (PL_multi_open == '?')
5965 pm->op_pmflags |= PMf_ONCE;
5967 while (*s && strchr("iomsx", *s))
5968 pmflag(&pm->op_pmflags,*s++);
5971 while (*s && strchr("iogcmsx", *s))
5972 pmflag(&pm->op_pmflags,*s++);
5974 pm->op_pmpermflags = pm->op_pmflags;
5976 PL_lex_op = (OP*)pm;
5977 yylval.ival = OP_MATCH;
5982 S_scan_subst(pTHX_ char *start)
5989 yylval.ival = OP_NULL;
5991 s = scan_str(start,FALSE,FALSE);
5995 SvREFCNT_dec(PL_lex_stuff);
5996 PL_lex_stuff = Nullsv;
5997 Perl_croak(aTHX_ "Substitution pattern not terminated");
6000 if (s[-1] == PL_multi_open)
6003 first_start = PL_multi_start;
6004 s = scan_str(s,FALSE,FALSE);
6007 SvREFCNT_dec(PL_lex_stuff);
6008 PL_lex_stuff = Nullsv;
6010 SvREFCNT_dec(PL_lex_repl);
6011 PL_lex_repl = Nullsv;
6012 Perl_croak(aTHX_ "Substitution replacement not terminated");
6014 PL_multi_start = first_start; /* so whole substitution is taken together */
6016 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6022 else if (strchr("iogcmsx", *s))
6023 pmflag(&pm->op_pmflags,*s++);
6030 PL_sublex_info.super_bufptr = s;
6031 PL_sublex_info.super_bufend = PL_bufend;
6033 pm->op_pmflags |= PMf_EVAL;
6034 repl = newSVpvn("",0);
6036 sv_catpv(repl, es ? "eval " : "do ");
6037 sv_catpvn(repl, "{ ", 2);
6038 sv_catsv(repl, PL_lex_repl);
6039 sv_catpvn(repl, " };", 2);
6041 SvREFCNT_dec(PL_lex_repl);
6045 pm->op_pmpermflags = pm->op_pmflags;
6046 PL_lex_op = (OP*)pm;
6047 yylval.ival = OP_SUBST;
6052 S_scan_trans(pTHX_ char *start)
6063 yylval.ival = OP_NULL;
6065 s = scan_str(start,FALSE,FALSE);
6068 SvREFCNT_dec(PL_lex_stuff);
6069 PL_lex_stuff = Nullsv;
6070 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6072 if (s[-1] == PL_multi_open)
6075 s = scan_str(s,FALSE,FALSE);
6078 SvREFCNT_dec(PL_lex_stuff);
6079 PL_lex_stuff = Nullsv;
6081 SvREFCNT_dec(PL_lex_repl);
6082 PL_lex_repl = Nullsv;
6083 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6087 o = newSVOP(OP_TRANS, 0, 0);
6088 utf8 = OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF;
6091 New(803,tbl,256,short);
6092 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6096 complement = del = squash = 0;
6097 while (strchr("cdsCU", *s)) {
6099 complement = OPpTRANS_COMPLEMENT;
6101 del = OPpTRANS_DELETE;
6103 squash = OPpTRANS_SQUASH;
6108 utf8 &= ~OPpTRANS_FROM_UTF;
6110 utf8 |= OPpTRANS_FROM_UTF;
6114 utf8 &= ~OPpTRANS_TO_UTF;
6116 utf8 |= OPpTRANS_TO_UTF;
6119 Perl_croak(aTHX_ "Too many /C and /U options");
6124 o->op_private = del|squash|complement|utf8;
6127 yylval.ival = OP_TRANS;
6132 S_scan_heredoc(pTHX_ register char *s)
6136 I32 op_type = OP_SCALAR;
6143 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6147 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6150 for (peek = s; *peek == ' ' || *peek == '\t'; peek++) ;
6151 if (*peek && strchr("`'\"",*peek)) {
6154 s = delimcpy(d, e, s, PL_bufend, term, &len);
6164 if (!isALNUM_lazy_if(s,UTF))
6165 deprecate("bare << to mean <<\"\"");
6166 for (; isALNUM_lazy_if(s,UTF); s++) {
6171 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6172 Perl_croak(aTHX_ "Delimiter for here document is too long");
6175 len = d - PL_tokenbuf;
6176 #ifndef PERL_STRICT_CR
6177 d = strchr(s, '\r');
6181 while (s < PL_bufend) {
6187 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6196 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6201 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6202 herewas = newSVpvn(s,PL_bufend-s);
6204 s--, herewas = newSVpvn(s,d-s);
6205 s += SvCUR(herewas);
6207 tmpstr = NEWSV(87,79);
6208 sv_upgrade(tmpstr, SVt_PVIV);
6213 else if (term == '`') {
6214 op_type = OP_BACKTICK;
6215 SvIVX(tmpstr) = '\\';
6219 PL_multi_start = CopLINE(PL_curcop);
6220 PL_multi_open = PL_multi_close = '<';
6221 term = *PL_tokenbuf;
6222 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6223 char *bufptr = PL_sublex_info.super_bufptr;
6224 char *bufend = PL_sublex_info.super_bufend;
6225 char *olds = s - SvCUR(herewas);
6226 s = strchr(bufptr, '\n');
6230 while (s < bufend &&
6231 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6233 CopLINE_inc(PL_curcop);
6236 CopLINE_set(PL_curcop, PL_multi_start);
6237 missingterm(PL_tokenbuf);
6239 sv_setpvn(herewas,bufptr,d-bufptr+1);
6240 sv_setpvn(tmpstr,d+1,s-d);
6242 sv_catpvn(herewas,s,bufend-s);
6243 (void)strcpy(bufptr,SvPVX(herewas));
6250 while (s < PL_bufend &&
6251 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6253 CopLINE_inc(PL_curcop);
6255 if (s >= PL_bufend) {
6256 CopLINE_set(PL_curcop, PL_multi_start);
6257 missingterm(PL_tokenbuf);
6259 sv_setpvn(tmpstr,d+1,s-d);
6261 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6263 sv_catpvn(herewas,s,PL_bufend-s);
6264 sv_setsv(PL_linestr,herewas);
6265 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6266 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6269 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6270 while (s >= PL_bufend) { /* multiple line string? */
6272 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6273 CopLINE_set(PL_curcop, PL_multi_start);
6274 missingterm(PL_tokenbuf);
6276 CopLINE_inc(PL_curcop);
6277 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6278 #ifndef PERL_STRICT_CR
6279 if (PL_bufend - PL_linestart >= 2) {
6280 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6281 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6283 PL_bufend[-2] = '\n';
6285 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6287 else if (PL_bufend[-1] == '\r')
6288 PL_bufend[-1] = '\n';
6290 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6291 PL_bufend[-1] = '\n';
6293 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6294 SV *sv = NEWSV(88,0);
6296 sv_upgrade(sv, SVt_PVMG);
6297 sv_setsv(sv,PL_linestr);
6298 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6300 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6303 sv_catsv(PL_linestr,herewas);
6304 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6308 sv_catsv(tmpstr,PL_linestr);
6313 PL_multi_end = CopLINE(PL_curcop);
6314 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6315 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6316 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6318 SvREFCNT_dec(herewas);
6319 PL_lex_stuff = tmpstr;
6320 yylval.ival = op_type;
6325 takes: current position in input buffer
6326 returns: new position in input buffer
6327 side-effects: yylval and lex_op are set.
6332 <FH> read from filehandle
6333 <pkg::FH> read from package qualified filehandle
6334 <pkg'FH> read from package qualified filehandle
6335 <$fh> read from filehandle in $fh
6341 S_scan_inputsymbol(pTHX_ char *start)
6343 register char *s = start; /* current position in buffer */
6349 d = PL_tokenbuf; /* start of temp holding space */
6350 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6351 end = strchr(s, '\n');
6354 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6356 /* die if we didn't have space for the contents of the <>,
6357 or if it didn't end, or if we see a newline
6360 if (len >= sizeof PL_tokenbuf)
6361 Perl_croak(aTHX_ "Excessively long <> operator");
6363 Perl_croak(aTHX_ "Unterminated <> operator");
6368 Remember, only scalar variables are interpreted as filehandles by
6369 this code. Anything more complex (e.g., <$fh{$num}>) will be
6370 treated as a glob() call.
6371 This code makes use of the fact that except for the $ at the front,
6372 a scalar variable and a filehandle look the same.
6374 if (*d == '$' && d[1]) d++;
6376 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6377 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6380 /* If we've tried to read what we allow filehandles to look like, and
6381 there's still text left, then it must be a glob() and not a getline.
6382 Use scan_str to pull out the stuff between the <> and treat it
6383 as nothing more than a string.
6386 if (d - PL_tokenbuf != len) {
6387 yylval.ival = OP_GLOB;
6389 s = scan_str(start,FALSE,FALSE);
6391 Perl_croak(aTHX_ "Glob not terminated");
6395 /* we're in a filehandle read situation */
6398 /* turn <> into <ARGV> */
6400 (void)strcpy(d,"ARGV");
6402 /* if <$fh>, create the ops to turn the variable into a
6408 /* try to find it in the pad for this block, otherwise find
6409 add symbol table ops
6411 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6412 OP *o = newOP(OP_PADSV, 0);
6414 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6417 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6418 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6419 newUNOP(OP_RV2SV, 0,
6420 newGVOP(OP_GV, 0, gv)));
6422 PL_lex_op->op_flags |= OPf_SPECIAL;
6423 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6424 yylval.ival = OP_NULL;
6427 /* If it's none of the above, it must be a literal filehandle
6428 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6430 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6431 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6432 yylval.ival = OP_NULL;
6441 takes: start position in buffer
6442 keep_quoted preserve \ on the embedded delimiter(s)
6443 keep_delims preserve the delimiters around the string
6444 returns: position to continue reading from buffer
6445 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6446 updates the read buffer.
6448 This subroutine pulls a string out of the input. It is called for:
6449 q single quotes q(literal text)
6450 ' single quotes 'literal text'
6451 qq double quotes qq(interpolate $here please)
6452 " double quotes "interpolate $here please"
6453 qx backticks qx(/bin/ls -l)
6454 ` backticks `/bin/ls -l`
6455 qw quote words @EXPORT_OK = qw( func() $spam )
6456 m// regexp match m/this/
6457 s/// regexp substitute s/this/that/
6458 tr/// string transliterate tr/this/that/
6459 y/// string transliterate y/this/that/
6460 ($*@) sub prototypes sub foo ($)
6461 (stuff) sub attr parameters sub foo : attr(stuff)
6462 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6464 In most of these cases (all but <>, patterns and transliterate)
6465 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6466 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6467 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6470 It skips whitespace before the string starts, and treats the first
6471 character as the delimiter. If the delimiter is one of ([{< then
6472 the corresponding "close" character )]}> is used as the closing
6473 delimiter. It allows quoting of delimiters, and if the string has
6474 balanced delimiters ([{<>}]) it allows nesting.
6476 The lexer always reads these strings into lex_stuff, except in the
6477 case of the operators which take *two* arguments (s/// and tr///)
6478 when it checks to see if lex_stuff is full (presumably with the 1st
6479 arg to s or tr) and if so puts the string into lex_repl.
6484 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6487 SV *sv; /* scalar value: string */
6488 char *tmps; /* temp string, used for delimiter matching */
6489 register char *s = start; /* current position in the buffer */
6490 register char term; /* terminating character */
6491 register char *to; /* current position in the sv's data */
6492 I32 brackets = 1; /* bracket nesting level */
6493 bool has_utf = FALSE; /* is there any utf8 content? */
6495 /* skip space before the delimiter */
6499 /* mark where we are, in case we need to report errors */
6502 /* after skipping whitespace, the next character is the terminator */
6504 if ((term & 0x80) && UTF)
6507 /* mark where we are */
6508 PL_multi_start = CopLINE(PL_curcop);
6509 PL_multi_open = term;
6511 /* find corresponding closing delimiter */
6512 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6514 PL_multi_close = term;
6516 /* create a new SV to hold the contents. 87 is leak category, I'm
6517 assuming. 79 is the SV's initial length. What a random number. */
6519 sv_upgrade(sv, SVt_PVIV);
6521 (void)SvPOK_only(sv); /* validate pointer */
6523 /* move past delimiter and try to read a complete string */
6525 sv_catpvn(sv, s, 1);
6528 /* extend sv if need be */
6529 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6530 /* set 'to' to the next character in the sv's string */
6531 to = SvPVX(sv)+SvCUR(sv);
6533 /* if open delimiter is the close delimiter read unbridle */
6534 if (PL_multi_open == PL_multi_close) {
6535 for (; s < PL_bufend; s++,to++) {
6536 /* embedded newlines increment the current line number */
6537 if (*s == '\n' && !PL_rsfp)
6538 CopLINE_inc(PL_curcop);
6539 /* handle quoted delimiters */
6540 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6541 if (!keep_quoted && s[1] == term)
6543 /* any other quotes are simply copied straight through */
6547 /* terminate when run out of buffer (the for() condition), or
6548 have found the terminator */
6549 else if (*s == term)
6551 else if (!has_utf && (*s & 0x80) && UTF)
6557 /* if the terminator isn't the same as the start character (e.g.,
6558 matched brackets), we have to allow more in the quoting, and
6559 be prepared for nested brackets.
6562 /* read until we run out of string, or we find the terminator */
6563 for (; s < PL_bufend; s++,to++) {
6564 /* embedded newlines increment the line count */
6565 if (*s == '\n' && !PL_rsfp)
6566 CopLINE_inc(PL_curcop);
6567 /* backslashes can escape the open or closing characters */
6568 if (*s == '\\' && s+1 < PL_bufend) {
6570 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6575 /* allow nested opens and closes */
6576 else if (*s == PL_multi_close && --brackets <= 0)
6578 else if (*s == PL_multi_open)
6580 else if (!has_utf && (*s & 0x80) && UTF)
6585 /* terminate the copied string and update the sv's end-of-string */
6587 SvCUR_set(sv, to - SvPVX(sv));
6590 * this next chunk reads more into the buffer if we're not done yet
6594 break; /* handle case where we are done yet :-) */
6596 #ifndef PERL_STRICT_CR
6597 if (to - SvPVX(sv) >= 2) {
6598 if ((to[-2] == '\r' && to[-1] == '\n') ||
6599 (to[-2] == '\n' && to[-1] == '\r'))
6603 SvCUR_set(sv, to - SvPVX(sv));
6605 else if (to[-1] == '\r')
6608 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6612 /* if we're out of file, or a read fails, bail and reset the current
6613 line marker so we can report where the unterminated string began
6616 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6618 CopLINE_set(PL_curcop, PL_multi_start);
6621 /* we read a line, so increment our line counter */
6622 CopLINE_inc(PL_curcop);
6624 /* update debugger info */
6625 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6626 SV *sv = NEWSV(88,0);
6628 sv_upgrade(sv, SVt_PVMG);
6629 sv_setsv(sv,PL_linestr);
6630 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6633 /* having changed the buffer, we must update PL_bufend */
6634 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6637 /* at this point, we have successfully read the delimited string */
6640 sv_catpvn(sv, s, 1);
6643 PL_multi_end = CopLINE(PL_curcop);
6646 /* if we allocated too much space, give some back */
6647 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6648 SvLEN_set(sv, SvCUR(sv) + 1);
6649 Renew(SvPVX(sv), SvLEN(sv), char);
6652 /* decide whether this is the first or second quoted string we've read
6665 takes: pointer to position in buffer
6666 returns: pointer to new position in buffer
6667 side-effects: builds ops for the constant in yylval.op
6669 Read a number in any of the formats that Perl accepts:
6671 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6672 [\d_]+(\.[\d_]*)?[Ee](\d+)
6674 Underbars (_) are allowed in decimal numbers. If -w is on,
6675 underbars before a decimal point must be at three digit intervals.
6677 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6680 If it reads a number without a decimal point or an exponent, it will
6681 try converting the number to an integer and see if it can do so
6682 without loss of precision.
6686 Perl_scan_num(pTHX_ char *start)
6688 register char *s = start; /* current position in buffer */
6689 register char *d; /* destination in temp buffer */
6690 register char *e; /* end of temp buffer */
6691 NV value; /* number read, as a double */
6692 SV *sv = Nullsv; /* place to put the converted number */
6693 bool floatit; /* boolean: int or float? */
6694 char *lastub = 0; /* position of last underbar */
6695 static char number_too_long[] = "Number too long";
6697 /* We use the first character to decide what type of number this is */
6701 Perl_croak(aTHX_ "panic: scan_num");
6703 /* if it starts with a 0, it could be an octal number, a decimal in
6704 0.13 disguise, or a hexadecimal number, or a binary number. */
6708 u holds the "number so far"
6709 shift the power of 2 of the base
6710 (hex == 4, octal == 3, binary == 1)
6711 overflowed was the number more than we can hold?
6713 Shift is used when we add a digit. It also serves as an "are
6714 we in octal/hex/binary?" indicator to disallow hex characters
6721 bool overflowed = FALSE;
6722 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6723 static char* bases[5] = { "", "binary", "", "octal",
6725 static char* Bases[5] = { "", "Binary", "", "Octal",
6727 static char *maxima[5] = { "",
6728 "0b11111111111111111111111111111111",
6732 char *base, *Base, *max;
6738 } else if (s[1] == 'b') {
6742 /* check for a decimal in disguise */
6743 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6745 /* so it must be octal */
6749 base = bases[shift];
6750 Base = Bases[shift];
6751 max = maxima[shift];
6753 /* read the rest of the number */
6755 /* x is used in the overflow test,
6756 b is the digit we're adding on. */
6761 /* if we don't mention it, we're done */
6770 /* 8 and 9 are not octal */
6773 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6777 case '2': case '3': case '4':
6778 case '5': case '6': case '7':
6780 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6784 b = *s++ & 15; /* ASCII digit -> value of digit */
6788 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6789 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6790 /* make sure they said 0x */
6795 /* Prepare to put the digit we have onto the end
6796 of the number so far. We check for overflows.
6801 x = u << shift; /* make room for the digit */
6803 if ((x >> shift) != u
6804 && !(PL_hints & HINT_NEW_BINARY)) {
6808 if (ckWARN_d(WARN_OVERFLOW))
6809 Perl_warner(aTHX_ WARN_OVERFLOW,
6810 "Integer overflow in %s number",
6813 u = x | b; /* add the digit to the end */
6816 n *= nvshift[shift];
6817 /* If an NV has not enough bits in its
6818 * mantissa to represent an UV this summing of
6819 * small low-order numbers is a waste of time
6820 * (because the NV cannot preserve the
6821 * low-order bits anyway): we could just
6822 * remember when did we overflow and in the
6823 * end just multiply n by the right
6831 /* if we get here, we had success: make a scalar value from
6838 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6839 Perl_warner(aTHX_ WARN_PORTABLE,
6840 "%s number > %s non-portable",
6847 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6848 Perl_warner(aTHX_ WARN_PORTABLE,
6849 "%s number > %s non-portable",
6854 if (PL_hints & HINT_NEW_BINARY)
6855 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6860 handle decimal numbers.
6861 we're also sent here when we read a 0 as the first digit
6863 case '1': case '2': case '3': case '4': case '5':
6864 case '6': case '7': case '8': case '9': case '.':
6867 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
6870 /* read next group of digits and _ and copy into d */
6871 while (isDIGIT(*s) || *s == '_') {
6872 /* skip underscores, checking for misplaced ones
6876 dTHR; /* only for ckWARN */
6877 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
6878 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6882 /* check for end of fixed-length buffer */
6884 Perl_croak(aTHX_ number_too_long);
6885 /* if we're ok, copy the character */
6890 /* final misplaced underbar check */
6891 if (lastub && s - lastub != 3) {
6893 if (ckWARN(WARN_SYNTAX))
6894 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6897 /* read a decimal portion if there is one. avoid
6898 3..5 being interpreted as the number 3. followed
6901 if (*s == '.' && s[1] != '.') {
6905 /* copy, ignoring underbars, until we run out of
6906 digits. Note: no misplaced underbar checks!
6908 for (; isDIGIT(*s) || *s == '_'; s++) {
6909 /* fixed length buffer check */
6911 Perl_croak(aTHX_ number_too_long);
6915 if (*s == '.' && isDIGIT(s[1])) {
6916 /* oops, it's really a v-string, but without the "v" */
6922 /* read exponent part, if present */
6923 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
6927 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
6928 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
6930 /* allow positive or negative exponent */
6931 if (*s == '+' || *s == '-')
6934 /* read digits of exponent (no underbars :-) */
6935 while (isDIGIT(*s)) {
6937 Perl_croak(aTHX_ number_too_long);
6942 /* terminate the string */
6945 /* make an sv from the string */
6948 /* unfortunately this monster needs to be on one line or
6949 makedepend will be confused. */
6950 #if (defined(USE_64_BIT_INT) && (!defined(HAS_STRTOLL)|| !defined(HAS_STRTOULL))) || (!defined(USE_64_BIT_INT) && (!defined(HAS_STRTOL) || !defined(HAS_STRTOUL)))
6953 No working strto[u]l[l]. Since atoi() doesn't do range checks,
6954 we need to do this the hard way.
6957 value = Atof(PL_tokenbuf);
6960 See if we can make do with an integer value without loss of
6961 precision. We use I_V to cast to an int, because some
6962 compilers have issues. Then we try casting it back and see
6963 if it was the same. We only do this if we know we
6964 specifically read an integer.
6966 Note: if floatit is true, then we don't need to do the
6970 UV tryuv = U_V(value);
6971 if (!floatit && (NV)tryuv == value) {
6972 if (tryuv <= IV_MAX)
6973 sv_setiv(sv, (IV)tryuv);
6975 sv_setuv(sv, tryuv);
6978 sv_setnv(sv, value);
6982 strtol/strtoll sets errno to ERANGE if the number is too big
6983 for an integer. We try to do an integer conversion first
6984 if no characters indicating "float" have been found.
6991 if (*PL_tokenbuf == '-')
6992 iv = Strtol(PL_tokenbuf, (char**)NULL, 10);
6994 uv = Strtoul(PL_tokenbuf, (char**)NULL, 10);
6996 floatit = TRUE; /* probably just too large */
6997 else if (*PL_tokenbuf == '-')
7003 value = Atof(PL_tokenbuf);
7004 sv_setnv(sv, value);
7007 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7008 (PL_hints & HINT_NEW_INTEGER) )
7009 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7010 (floatit ? "float" : "integer"),
7014 /* if it starts with a v, it could be a v-string */
7020 while (isDIGIT(*pos) || *pos == '_')
7022 if (!isALPHA(*pos)) {
7024 U8 tmpbuf[UTF8_MAXLEN];
7027 s++; /* get past 'v' */
7030 sv_setpvn(sv, "", 0);
7033 if (*s == '0' && isDIGIT(s[1]))
7034 yyerror("Octal number in vector unsupported");
7037 /* this is atoi() that tolerates underscores */
7040 while (--end >= s) {
7045 rev += (*end - '0') * mult;
7047 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
7048 Perl_warner(aTHX_ WARN_OVERFLOW,
7049 "Integer overflow in decimal number");
7052 tmpend = uv_to_utf8(tmpbuf, rev);
7053 utf8 = utf8 || rev > 127;
7054 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
7055 if (*pos == '.' && isDIGIT(pos[1]))
7061 while (isDIGIT(*pos) || *pos == '_')
7069 sv_utf8_downgrade(sv, TRUE);
7076 /* make the op for the constant and return */
7079 yylval.opval = newSVOP(OP_CONST, 0, sv);
7081 yylval.opval = Nullop;
7087 S_scan_formline(pTHX_ register char *s)
7092 SV *stuff = newSVpvn("",0);
7093 bool needargs = FALSE;
7096 if (*s == '.' || *s == '}') {
7098 #ifdef PERL_STRICT_CR
7099 for (t = s+1;*t == ' ' || *t == '\t'; t++) ;
7101 for (t = s+1;*t == ' ' || *t == '\t' || *t == '\r'; t++) ;
7103 if (*t == '\n' || t == PL_bufend)
7106 if (PL_in_eval && !PL_rsfp) {
7107 eol = strchr(s,'\n');
7112 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7114 for (t = s; t < eol; t++) {
7115 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7117 goto enough; /* ~~ must be first line in formline */
7119 if (*t == '@' || *t == '^')
7122 sv_catpvn(stuff, s, eol-s);
7123 #ifndef PERL_STRICT_CR
7124 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7125 char *end = SvPVX(stuff) + SvCUR(stuff);
7134 s = filter_gets(PL_linestr, PL_rsfp, 0);
7135 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7136 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7139 yyerror("Format not terminated");
7149 PL_lex_state = LEX_NORMAL;
7150 PL_nextval[PL_nexttoke].ival = 0;
7154 PL_lex_state = LEX_FORMLINE;
7155 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7157 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7161 SvREFCNT_dec(stuff);
7162 PL_lex_formbrack = 0;
7173 PL_cshlen = strlen(PL_cshname);
7178 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7181 I32 oldsavestack_ix = PL_savestack_ix;
7182 CV* outsidecv = PL_compcv;
7186 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7188 SAVEI32(PL_subline);
7189 save_item(PL_subname);
7192 SAVESPTR(PL_comppad_name);
7193 SAVESPTR(PL_compcv);
7194 SAVEI32(PL_comppad_name_fill);
7195 SAVEI32(PL_min_intro_pending);
7196 SAVEI32(PL_max_intro_pending);
7197 SAVEI32(PL_pad_reset_pending);
7199 PL_compcv = (CV*)NEWSV(1104,0);
7200 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7201 CvFLAGS(PL_compcv) |= flags;
7203 PL_comppad = newAV();
7204 av_push(PL_comppad, Nullsv);
7205 PL_curpad = AvARRAY(PL_comppad);
7206 PL_comppad_name = newAV();
7207 PL_comppad_name_fill = 0;
7208 PL_min_intro_pending = 0;
7210 PL_subline = CopLINE(PL_curcop);
7212 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7213 PL_curpad[0] = (SV*)newAV();
7214 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7215 #endif /* USE_THREADS */
7217 comppadlist = newAV();
7218 AvREAL_off(comppadlist);
7219 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7220 av_store(comppadlist, 1, (SV*)PL_comppad);
7222 CvPADLIST(PL_compcv) = comppadlist;
7223 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7225 CvOWNER(PL_compcv) = 0;
7226 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7227 MUTEX_INIT(CvMUTEXP(PL_compcv));
7228 #endif /* USE_THREADS */
7230 return oldsavestack_ix;
7234 Perl_yywarn(pTHX_ char *s)
7237 PL_in_eval |= EVAL_WARNONLY;
7239 PL_in_eval &= ~EVAL_WARNONLY;
7244 Perl_yyerror(pTHX_ char *s)
7248 char *context = NULL;
7252 if (!yychar || (yychar == ';' && !PL_rsfp))
7254 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7255 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7256 while (isSPACE(*PL_oldoldbufptr))
7258 context = PL_oldoldbufptr;
7259 contlen = PL_bufptr - PL_oldoldbufptr;
7261 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7262 PL_oldbufptr != PL_bufptr) {
7263 while (isSPACE(*PL_oldbufptr))
7265 context = PL_oldbufptr;
7266 contlen = PL_bufptr - PL_oldbufptr;
7268 else if (yychar > 255)
7269 where = "next token ???";
7270 #ifdef USE_PURE_BISON
7271 /* GNU Bison sets the value -2 */
7272 else if (yychar == -2) {
7274 else if ((yychar & 127) == 127) {
7276 if (PL_lex_state == LEX_NORMAL ||
7277 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7278 where = "at end of line";
7279 else if (PL_lex_inpat)
7280 where = "within pattern";
7282 where = "within string";
7285 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7287 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7288 else if (isPRINT_LC(yychar))
7289 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7291 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7292 where = SvPVX(where_sv);
7294 msg = sv_2mortal(newSVpv(s, 0));
7295 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7296 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7298 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7300 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7301 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7302 Perl_sv_catpvf(aTHX_ msg,
7303 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7304 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7307 if (PL_in_eval & EVAL_WARNONLY)
7308 Perl_warn(aTHX_ "%"SVf, msg);
7311 if (PL_error_count >= 10) {
7312 if (PL_in_eval && SvCUR(ERRSV))
7313 Perl_croak(aTHX_ "%_%s has too many errors.\n",
7314 ERRSV, CopFILE(PL_curcop));
7316 Perl_croak(aTHX_ "%s has too many errors.\n",
7317 CopFILE(PL_curcop));
7320 PL_in_my_stash = Nullhv;
7331 * Restore a source filter.
7335 restore_rsfp(pTHXo_ void *f)
7337 PerlIO *fp = (PerlIO*)f;
7339 if (PL_rsfp == PerlIO_stdin())
7340 PerlIO_clearerr(PL_rsfp);
7341 else if (PL_rsfp && (PL_rsfp != fp))
7342 PerlIO_close(PL_rsfp);