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 (len > e - s + 4) {
1483 char *odest = SvPVX(sv);
1485 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1486 d = SvPVX(sv) + (d - odest);
1488 Copy(str, d, len, char);
1495 yyerror("Missing braces on \\N{}");
1498 /* \c is a control character */
1512 /* printf-style backslashes, formfeeds, newlines, etc */
1530 *d++ = '\047'; /* CP 1047 */
1533 *d++ = '\057'; /* CP 1047 */
1547 } /* end if (backslash) */
1550 } /* while loop to process each character */
1552 /* terminate the string and set up the sv */
1554 SvCUR_set(sv, d - SvPVX(sv));
1559 /* shrink the sv if we allocated more than we used */
1560 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1561 SvLEN_set(sv, SvCUR(sv) + 1);
1562 Renew(SvPVX(sv), SvLEN(sv), char);
1565 /* return the substring (via yylval) only if we parsed anything */
1566 if (s > PL_bufptr) {
1567 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1568 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1570 ( PL_lex_inwhat == OP_TRANS
1572 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1575 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1582 * Returns TRUE if there's more to the expression (e.g., a subscript),
1585 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1587 * ->[ and ->{ return TRUE
1588 * { and [ outside a pattern are always subscripts, so return TRUE
1589 * if we're outside a pattern and it's not { or [, then return FALSE
1590 * if we're in a pattern and the first char is a {
1591 * {4,5} (any digits around the comma) returns FALSE
1592 * if we're in a pattern and the first char is a [
1594 * [SOMETHING] has a funky algorithm to decide whether it's a
1595 * character class or not. It has to deal with things like
1596 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1597 * anything else returns TRUE
1600 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1603 S_intuit_more(pTHX_ register char *s)
1605 if (PL_lex_brackets)
1607 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1609 if (*s != '{' && *s != '[')
1614 /* In a pattern, so maybe we have {n,m}. */
1631 /* On the other hand, maybe we have a character class */
1634 if (*s == ']' || *s == '^')
1637 /* this is terrifying, and it works */
1638 int weight = 2; /* let's weigh the evidence */
1640 unsigned char un_char = 255, last_un_char;
1641 char *send = strchr(s,']');
1642 char tmpbuf[sizeof PL_tokenbuf * 4];
1644 if (!send) /* has to be an expression */
1647 Zero(seen,256,char);
1650 else if (isDIGIT(*s)) {
1652 if (isDIGIT(s[1]) && s[2] == ']')
1658 for (; s < send; s++) {
1659 last_un_char = un_char;
1660 un_char = (unsigned char)*s;
1665 weight -= seen[un_char] * 10;
1666 if (isALNUM_lazy_if(s+1,UTF)) {
1667 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1668 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1673 else if (*s == '$' && s[1] &&
1674 strchr("[#!%*<>()-=",s[1])) {
1675 if (/*{*/ strchr("])} =",s[2]))
1684 if (strchr("wds]",s[1]))
1686 else if (seen['\''] || seen['"'])
1688 else if (strchr("rnftbxcav",s[1]))
1690 else if (isDIGIT(s[1])) {
1692 while (s[1] && isDIGIT(s[1]))
1702 if (strchr("aA01! ",last_un_char))
1704 if (strchr("zZ79~",s[1]))
1706 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1707 weight -= 5; /* cope with negative subscript */
1710 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1711 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1716 if (keyword(tmpbuf, d - tmpbuf))
1719 if (un_char == last_un_char + 1)
1721 weight -= seen[un_char];
1726 if (weight >= 0) /* probably a character class */
1736 * Does all the checking to disambiguate
1738 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1739 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1741 * First argument is the stuff after the first token, e.g. "bar".
1743 * Not a method if bar is a filehandle.
1744 * Not a method if foo is a subroutine prototyped to take a filehandle.
1745 * Not a method if it's really "Foo $bar"
1746 * Method if it's "foo $bar"
1747 * Not a method if it's really "print foo $bar"
1748 * Method if it's really "foo package::" (interpreted as package->foo)
1749 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1750 * Not a method if bar is a filehandle or package, but is quoted with
1755 S_intuit_method(pTHX_ char *start, GV *gv)
1757 char *s = start + (*start == '$');
1758 char tmpbuf[sizeof PL_tokenbuf];
1766 if ((cv = GvCVu(gv))) {
1767 char *proto = SvPVX(cv);
1777 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1778 /* start is the beginning of the possible filehandle/object,
1779 * and s is the end of it
1780 * tmpbuf is a copy of it
1783 if (*start == '$') {
1784 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1789 return *s == '(' ? FUNCMETH : METHOD;
1791 if (!keyword(tmpbuf, len)) {
1792 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1797 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1798 if (indirgv && GvCVu(indirgv))
1800 /* filehandle or package name makes it a method */
1801 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1803 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1804 return 0; /* no assumptions -- "=>" quotes bearword */
1806 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1807 newSVpvn(tmpbuf,len));
1808 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1812 return *s == '(' ? FUNCMETH : METHOD;
1820 * Return a string of Perl code to load the debugger. If PERL5DB
1821 * is set, it will return the contents of that, otherwise a
1822 * compile-time require of perl5db.pl.
1829 char *pdb = PerlEnv_getenv("PERL5DB");
1833 SETERRNO(0,SS$_NORMAL);
1834 return "BEGIN { require 'perl5db.pl' }";
1840 /* Encoded script support. filter_add() effectively inserts a
1841 * 'pre-processing' function into the current source input stream.
1842 * Note that the filter function only applies to the current source file
1843 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1845 * The datasv parameter (which may be NULL) can be used to pass
1846 * private data to this instance of the filter. The filter function
1847 * can recover the SV using the FILTER_DATA macro and use it to
1848 * store private buffers and state information.
1850 * The supplied datasv parameter is upgraded to a PVIO type
1851 * and the IoDIRP field is used to store the function pointer,
1852 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1853 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1854 * private use must be set using malloc'd pointers.
1858 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1863 if (!PL_rsfp_filters)
1864 PL_rsfp_filters = newAV();
1866 datasv = NEWSV(255,0);
1867 if (!SvUPGRADE(datasv, SVt_PVIO))
1868 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1869 IoDIRP(datasv) = (DIR*)funcp; /* stash funcp into spare field */
1870 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1871 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1872 funcp, SvPV_nolen(datasv)));
1873 av_unshift(PL_rsfp_filters, 1);
1874 av_store(PL_rsfp_filters, 0, datasv) ;
1879 /* Delete most recently added instance of this filter function. */
1881 Perl_filter_del(pTHX_ filter_t funcp)
1884 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1885 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1887 /* if filter is on top of stack (usual case) just pop it off */
1888 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1889 if (IoDIRP(datasv) == (DIR*)funcp) {
1890 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1891 IoDIRP(datasv) = (DIR*)NULL;
1892 sv_free(av_pop(PL_rsfp_filters));
1896 /* we need to search for the correct entry and clear it */
1897 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1901 /* Invoke the n'th filter function for the current rsfp. */
1903 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1906 /* 0 = read one text line */
1911 if (!PL_rsfp_filters)
1913 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1914 /* Provide a default input filter to make life easy. */
1915 /* Note that we append to the line. This is handy. */
1916 DEBUG_P(PerlIO_printf(Perl_debug_log,
1917 "filter_read %d: from rsfp\n", idx));
1921 int old_len = SvCUR(buf_sv) ;
1923 /* ensure buf_sv is large enough */
1924 SvGROW(buf_sv, old_len + maxlen) ;
1925 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1926 if (PerlIO_error(PL_rsfp))
1927 return -1; /* error */
1929 return 0 ; /* end of file */
1931 SvCUR_set(buf_sv, old_len + len) ;
1934 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1935 if (PerlIO_error(PL_rsfp))
1936 return -1; /* error */
1938 return 0 ; /* end of file */
1941 return SvCUR(buf_sv);
1943 /* Skip this filter slot if filter has been deleted */
1944 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1945 DEBUG_P(PerlIO_printf(Perl_debug_log,
1946 "filter_read %d: skipped (filter deleted)\n",
1948 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1950 /* Get function pointer hidden within datasv */
1951 funcp = (filter_t)IoDIRP(datasv);
1952 DEBUG_P(PerlIO_printf(Perl_debug_log,
1953 "filter_read %d: via function %p (%s)\n",
1954 idx, funcp, SvPV_nolen(datasv)));
1955 /* Call function. The function is expected to */
1956 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1957 /* Return: <0:error, =0:eof, >0:not eof */
1958 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1962 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1964 #ifdef PERL_CR_FILTER
1965 if (!PL_rsfp_filters) {
1966 filter_add(S_cr_textfilter,NULL);
1969 if (PL_rsfp_filters) {
1972 SvCUR_set(sv, 0); /* start with empty line */
1973 if (FILTER_READ(0, sv, 0) > 0)
1974 return ( SvPVX(sv) ) ;
1979 return (sv_gets(sv, fp, append));
1984 static char* exp_name[] =
1985 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
1986 "ATTRTERM", "TERMBLOCK"
1993 Works out what to call the token just pulled out of the input
1994 stream. The yacc parser takes care of taking the ops we return and
1995 stitching them into a tree.
2001 if read an identifier
2002 if we're in a my declaration
2003 croak if they tried to say my($foo::bar)
2004 build the ops for a my() declaration
2005 if it's an access to a my() variable
2006 are we in a sort block?
2007 croak if my($a); $a <=> $b
2008 build ops for access to a my() variable
2009 if in a dq string, and they've said @foo and we can't find @foo
2011 build ops for a bareword
2012 if we already built the token before, use it.
2016 #ifdef USE_PURE_BISON
2017 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
2030 #ifdef USE_PURE_BISON
2031 yylval_pointer = lvalp;
2032 yychar_pointer = lcharp;
2035 /* check if there's an identifier for us to look at */
2036 if (PL_pending_ident) {
2037 /* pit holds the identifier we read and pending_ident is reset */
2038 char pit = PL_pending_ident;
2039 PL_pending_ident = 0;
2041 /* if we're in a my(), we can't allow dynamics here.
2042 $foo'bar has already been turned into $foo::bar, so
2043 just check for colons.
2045 if it's a legal name, the OP is a PADANY.
2048 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
2049 if (strchr(PL_tokenbuf,':'))
2050 yyerror(Perl_form(aTHX_ "No package name allowed for "
2051 "variable %s in \"our\"",
2053 tmp = pad_allocmy(PL_tokenbuf);
2056 if (strchr(PL_tokenbuf,':'))
2057 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
2059 yylval.opval = newOP(OP_PADANY, 0);
2060 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
2066 build the ops for accesses to a my() variable.
2068 Deny my($a) or my($b) in a sort block, *if* $a or $b is
2069 then used in a comparison. This catches most, but not
2070 all cases. For instance, it catches
2071 sort { my($a); $a <=> $b }
2073 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2074 (although why you'd do that is anyone's guess).
2077 if (!strchr(PL_tokenbuf,':')) {
2079 /* Check for single character per-thread SVs */
2080 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2081 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2082 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2084 yylval.opval = newOP(OP_THREADSV, 0);
2085 yylval.opval->op_targ = tmp;
2088 #endif /* USE_THREADS */
2089 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2090 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2091 /* might be an "our" variable" */
2092 if (SvFLAGS(namesv) & SVpad_OUR) {
2093 /* build ops for a bareword */
2094 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2095 sv_catpvn(sym, "::", 2);
2096 sv_catpv(sym, PL_tokenbuf+1);
2097 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2098 yylval.opval->op_private = OPpCONST_ENTERED;
2099 gv_fetchpv(SvPVX(sym),
2101 ? (GV_ADDMULTI | GV_ADDINEVAL)
2104 ((PL_tokenbuf[0] == '$') ? SVt_PV
2105 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2110 /* if it's a sort block and they're naming $a or $b */
2111 if (PL_last_lop_op == OP_SORT &&
2112 PL_tokenbuf[0] == '$' &&
2113 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2116 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2117 d < PL_bufend && *d != '\n';
2120 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2121 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2127 yylval.opval = newOP(OP_PADANY, 0);
2128 yylval.opval->op_targ = tmp;
2134 Whine if they've said @foo in a doublequoted string,
2135 and @foo isn't a variable we can find in the symbol
2138 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2139 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2140 if (!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2141 yyerror(Perl_form(aTHX_ "In string, %s now must be written as \\%s",
2142 PL_tokenbuf, PL_tokenbuf));
2145 /* build ops for a bareword */
2146 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2147 yylval.opval->op_private = OPpCONST_ENTERED;
2148 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2149 ((PL_tokenbuf[0] == '$') ? SVt_PV
2150 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2155 /* no identifier pending identification */
2157 switch (PL_lex_state) {
2159 case LEX_NORMAL: /* Some compilers will produce faster */
2160 case LEX_INTERPNORMAL: /* code if we comment these out. */
2164 /* when we've already built the next token, just pull it out of the queue */
2167 yylval = PL_nextval[PL_nexttoke];
2169 PL_lex_state = PL_lex_defer;
2170 PL_expect = PL_lex_expect;
2171 PL_lex_defer = LEX_NORMAL;
2173 return(PL_nexttype[PL_nexttoke]);
2175 /* interpolated case modifiers like \L \U, including \Q and \E.
2176 when we get here, PL_bufptr is at the \
2178 case LEX_INTERPCASEMOD:
2180 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2181 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2183 /* handle \E or end of string */
2184 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2188 if (PL_lex_casemods) {
2189 oldmod = PL_lex_casestack[--PL_lex_casemods];
2190 PL_lex_casestack[PL_lex_casemods] = '\0';
2192 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2194 PL_lex_state = LEX_INTERPCONCAT;
2198 if (PL_bufptr != PL_bufend)
2200 PL_lex_state = LEX_INTERPCONCAT;
2205 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2206 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2207 if (strchr("LU", *s) &&
2208 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2210 PL_lex_casestack[--PL_lex_casemods] = '\0';
2213 if (PL_lex_casemods > 10) {
2214 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2215 if (newlb != PL_lex_casestack) {
2217 PL_lex_casestack = newlb;
2220 PL_lex_casestack[PL_lex_casemods++] = *s;
2221 PL_lex_casestack[PL_lex_casemods] = '\0';
2222 PL_lex_state = LEX_INTERPCONCAT;
2223 PL_nextval[PL_nexttoke].ival = 0;
2226 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2228 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2230 PL_nextval[PL_nexttoke].ival = OP_LC;
2232 PL_nextval[PL_nexttoke].ival = OP_UC;
2234 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2236 Perl_croak(aTHX_ "panic: yylex");
2239 if (PL_lex_starts) {
2248 case LEX_INTERPPUSH:
2249 return sublex_push();
2251 case LEX_INTERPSTART:
2252 if (PL_bufptr == PL_bufend)
2253 return sublex_done();
2255 PL_lex_dojoin = (*PL_bufptr == '@');
2256 PL_lex_state = LEX_INTERPNORMAL;
2257 if (PL_lex_dojoin) {
2258 PL_nextval[PL_nexttoke].ival = 0;
2261 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2262 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2263 force_next(PRIVATEREF);
2265 force_ident("\"", '$');
2266 #endif /* USE_THREADS */
2267 PL_nextval[PL_nexttoke].ival = 0;
2269 PL_nextval[PL_nexttoke].ival = 0;
2271 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2274 if (PL_lex_starts++) {
2280 case LEX_INTERPENDMAYBE:
2281 if (intuit_more(PL_bufptr)) {
2282 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2288 if (PL_lex_dojoin) {
2289 PL_lex_dojoin = FALSE;
2290 PL_lex_state = LEX_INTERPCONCAT;
2293 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2294 && SvEVALED(PL_lex_repl))
2296 if (PL_bufptr != PL_bufend)
2297 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2298 PL_lex_repl = Nullsv;
2301 case LEX_INTERPCONCAT:
2303 if (PL_lex_brackets)
2304 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2306 if (PL_bufptr == PL_bufend)
2307 return sublex_done();
2309 if (SvIVX(PL_linestr) == '\'') {
2310 SV *sv = newSVsv(PL_linestr);
2313 else if ( PL_hints & HINT_NEW_RE )
2314 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2315 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2319 s = scan_const(PL_bufptr);
2321 PL_lex_state = LEX_INTERPCASEMOD;
2323 PL_lex_state = LEX_INTERPSTART;
2326 if (s != PL_bufptr) {
2327 PL_nextval[PL_nexttoke] = yylval;
2330 if (PL_lex_starts++)
2340 PL_lex_state = LEX_NORMAL;
2341 s = scan_formline(PL_bufptr);
2342 if (!PL_lex_formbrack)
2348 PL_oldoldbufptr = PL_oldbufptr;
2351 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2352 exp_name[PL_expect], s);
2358 if (isIDFIRST_lazy_if(s,UTF))
2360 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2363 goto fake_eof; /* emulate EOF on ^D or ^Z */
2368 if (PL_lex_brackets)
2369 yyerror("Missing right curly or square bracket");
2372 if (s++ < PL_bufend)
2373 goto retry; /* ignore stray nulls */
2376 if (!PL_in_eval && !PL_preambled) {
2377 PL_preambled = TRUE;
2378 sv_setpv(PL_linestr,incl_perldb());
2379 if (SvCUR(PL_linestr))
2380 sv_catpv(PL_linestr,";");
2382 while(AvFILLp(PL_preambleav) >= 0) {
2383 SV *tmpsv = av_shift(PL_preambleav);
2384 sv_catsv(PL_linestr, tmpsv);
2385 sv_catpv(PL_linestr, ";");
2388 sv_free((SV*)PL_preambleav);
2389 PL_preambleav = NULL;
2391 if (PL_minus_n || PL_minus_p) {
2392 sv_catpv(PL_linestr, "LINE: while (<>) {");
2394 sv_catpv(PL_linestr,"chomp;");
2396 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2398 GvIMPORTED_AV_on(gv);
2400 if (strchr("/'\"", *PL_splitstr)
2401 && strchr(PL_splitstr + 1, *PL_splitstr))
2402 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2405 s = "'~#\200\1'"; /* surely one char is unused...*/
2406 while (s[1] && strchr(PL_splitstr, *s)) s++;
2408 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2409 "q" + (delim == '\''), delim);
2410 for (s = PL_splitstr; *s; s++) {
2412 sv_catpvn(PL_linestr, "\\", 1);
2413 sv_catpvn(PL_linestr, s, 1);
2415 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2419 sv_catpv(PL_linestr,"@F=split(' ');");
2422 sv_catpv(PL_linestr, "\n");
2423 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2424 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2425 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2426 SV *sv = NEWSV(85,0);
2428 sv_upgrade(sv, SVt_PVMG);
2429 sv_setsv(sv,PL_linestr);
2430 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2435 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2438 if (PL_preprocess && !PL_in_eval)
2439 (void)PerlProc_pclose(PL_rsfp);
2440 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2441 PerlIO_clearerr(PL_rsfp);
2443 (void)PerlIO_close(PL_rsfp);
2445 PL_doextract = FALSE;
2447 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2448 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2449 sv_catpv(PL_linestr,";}");
2450 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2451 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2452 PL_minus_n = PL_minus_p = 0;
2455 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2456 sv_setpv(PL_linestr,"");
2457 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2460 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2461 PL_doextract = FALSE;
2463 /* Incest with pod. */
2464 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2465 sv_setpv(PL_linestr, "");
2466 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2467 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2468 PL_doextract = FALSE;
2472 } while (PL_doextract);
2473 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2474 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2475 SV *sv = NEWSV(85,0);
2477 sv_upgrade(sv, SVt_PVMG);
2478 sv_setsv(sv,PL_linestr);
2479 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2481 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2482 if (CopLINE(PL_curcop) == 1) {
2483 while (s < PL_bufend && isSPACE(*s))
2485 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2489 if (*s == '#' && *(s+1) == '!')
2491 #ifdef ALTERNATE_SHEBANG
2493 static char as[] = ALTERNATE_SHEBANG;
2494 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2495 d = s + (sizeof(as) - 1);
2497 #endif /* ALTERNATE_SHEBANG */
2506 while (*d && !isSPACE(*d))
2510 #ifdef ARG_ZERO_IS_SCRIPT
2511 if (ipathend > ipath) {
2513 * HP-UX (at least) sets argv[0] to the script name,
2514 * which makes $^X incorrect. And Digital UNIX and Linux,
2515 * at least, set argv[0] to the basename of the Perl
2516 * interpreter. So, having found "#!", we'll set it right.
2518 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2519 assert(SvPOK(x) || SvGMAGICAL(x));
2520 if (sv_eq(x, CopFILESV(PL_curcop))) {
2521 sv_setpvn(x, ipath, ipathend - ipath);
2524 TAINT_NOT; /* $^X is always tainted, but that's OK */
2526 #endif /* ARG_ZERO_IS_SCRIPT */
2531 d = instr(s,"perl -");
2533 d = instr(s,"perl");
2535 /* avoid getting into infinite loops when shebang
2536 * line contains "Perl" rather than "perl" */
2538 for (d = ipathend-4; d >= ipath; --d) {
2539 if ((*d == 'p' || *d == 'P')
2540 && !ibcmp(d, "perl", 4))
2550 #ifdef ALTERNATE_SHEBANG
2552 * If the ALTERNATE_SHEBANG on this system starts with a
2553 * character that can be part of a Perl expression, then if
2554 * we see it but not "perl", we're probably looking at the
2555 * start of Perl code, not a request to hand off to some
2556 * other interpreter. Similarly, if "perl" is there, but
2557 * not in the first 'word' of the line, we assume the line
2558 * contains the start of the Perl program.
2560 if (d && *s != '#') {
2562 while (*c && !strchr("; \t\r\n\f\v#", *c))
2565 d = Nullch; /* "perl" not in first word; ignore */
2567 *s = '#'; /* Don't try to parse shebang line */
2569 #endif /* ALTERNATE_SHEBANG */
2574 !instr(s,"indir") &&
2575 instr(PL_origargv[0],"perl"))
2581 while (s < PL_bufend && isSPACE(*s))
2583 if (s < PL_bufend) {
2584 Newz(899,newargv,PL_origargc+3,char*);
2586 while (s < PL_bufend && !isSPACE(*s))
2589 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2592 newargv = PL_origargv;
2594 PerlProc_execv(ipath, newargv);
2595 Perl_croak(aTHX_ "Can't exec %s", ipath);
2598 U32 oldpdb = PL_perldb;
2599 bool oldn = PL_minus_n;
2600 bool oldp = PL_minus_p;
2602 while (*d && !isSPACE(*d)) d++;
2603 while (*d == ' ' || *d == '\t') d++;
2607 if (*d == 'M' || *d == 'm') {
2609 while (*d && !isSPACE(*d)) d++;
2610 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2613 d = moreswitches(d);
2615 if ((PERLDB_LINE && !oldpdb) ||
2616 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2617 /* if we have already added "LINE: while (<>) {",
2618 we must not do it again */
2620 sv_setpv(PL_linestr, "");
2621 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2622 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2623 PL_preambled = FALSE;
2625 (void)gv_fetchfile(PL_origfilename);
2632 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2634 PL_lex_state = LEX_FORMLINE;
2639 #ifdef PERL_STRICT_CR
2640 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2642 "(Maybe you didn't strip carriage returns after a network transfer?)\n");
2644 case ' ': case '\t': case '\f': case 013:
2649 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2651 while (s < d && *s != '\n')
2656 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2658 PL_lex_state = LEX_FORMLINE;
2668 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2673 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2676 if (strnEQ(s,"=>",2)) {
2677 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2678 OPERATOR('-'); /* unary minus */
2680 PL_last_uni = PL_oldbufptr;
2681 PL_last_lop_op = OP_FTEREAD; /* good enough */
2683 case 'r': FTST(OP_FTEREAD);
2684 case 'w': FTST(OP_FTEWRITE);
2685 case 'x': FTST(OP_FTEEXEC);
2686 case 'o': FTST(OP_FTEOWNED);
2687 case 'R': FTST(OP_FTRREAD);
2688 case 'W': FTST(OP_FTRWRITE);
2689 case 'X': FTST(OP_FTREXEC);
2690 case 'O': FTST(OP_FTROWNED);
2691 case 'e': FTST(OP_FTIS);
2692 case 'z': FTST(OP_FTZERO);
2693 case 's': FTST(OP_FTSIZE);
2694 case 'f': FTST(OP_FTFILE);
2695 case 'd': FTST(OP_FTDIR);
2696 case 'l': FTST(OP_FTLINK);
2697 case 'p': FTST(OP_FTPIPE);
2698 case 'S': FTST(OP_FTSOCK);
2699 case 'u': FTST(OP_FTSUID);
2700 case 'g': FTST(OP_FTSGID);
2701 case 'k': FTST(OP_FTSVTX);
2702 case 'b': FTST(OP_FTBLK);
2703 case 'c': FTST(OP_FTCHR);
2704 case 't': FTST(OP_FTTTY);
2705 case 'T': FTST(OP_FTTEXT);
2706 case 'B': FTST(OP_FTBINARY);
2707 case 'M': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTMTIME);
2708 case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME);
2709 case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME);
2711 Perl_croak(aTHX_ "Unrecognized file test: -%c", (int)tmp);
2718 if (PL_expect == XOPERATOR)
2723 else if (*s == '>') {
2726 if (isIDFIRST_lazy_if(s,UTF)) {
2727 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2735 if (PL_expect == XOPERATOR)
2738 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2740 OPERATOR('-'); /* unary minus */
2747 if (PL_expect == XOPERATOR)
2752 if (PL_expect == XOPERATOR)
2755 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2761 if (PL_expect != XOPERATOR) {
2762 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2763 PL_expect = XOPERATOR;
2764 force_ident(PL_tokenbuf, '*');
2777 if (PL_expect == XOPERATOR) {
2781 PL_tokenbuf[0] = '%';
2782 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2783 if (!PL_tokenbuf[1]) {
2785 yyerror("Final % should be \\% or %name");
2788 PL_pending_ident = '%';
2807 switch (PL_expect) {
2810 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2812 PL_bufptr = s; /* update in case we back off */
2818 PL_expect = XTERMBLOCK;
2822 while (isIDFIRST_lazy_if(s,UTF)) {
2823 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2824 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2825 if (tmp < 0) tmp = -tmp;
2840 d = scan_str(d,TRUE,TRUE);
2843 SvREFCNT_dec(PL_lex_stuff);
2844 PL_lex_stuff = Nullsv;
2846 /* MUST advance bufptr here to avoid bogus
2847 "at end of line" context messages from yyerror().
2849 PL_bufptr = s + len;
2850 yyerror("Unterminated attribute parameter in attribute list");
2853 return 0; /* EOF indicator */
2857 SV *sv = newSVpvn(s, len);
2858 sv_catsv(sv, PL_lex_stuff);
2859 attrs = append_elem(OP_LIST, attrs,
2860 newSVOP(OP_CONST, 0, sv));
2861 SvREFCNT_dec(PL_lex_stuff);
2862 PL_lex_stuff = Nullsv;
2865 attrs = append_elem(OP_LIST, attrs,
2866 newSVOP(OP_CONST, 0,
2870 if (*s == ':' && s[1] != ':')
2873 break; /* require real whitespace or :'s */
2875 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
2876 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
2877 char q = ((*s == '\'') ? '"' : '\'');
2878 /* If here for an expression, and parsed no attrs, back off. */
2879 if (tmp == '=' && !attrs) {
2883 /* MUST advance bufptr here to avoid bogus "at end of line"
2884 context messages from yyerror().
2888 yyerror("Unterminated attribute list");
2890 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
2898 PL_nextval[PL_nexttoke].opval = attrs;
2906 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
2907 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
2912 if (CopLINE(PL_curcop) < PL_copline)
2913 PL_copline = CopLINE(PL_curcop);
2924 if (PL_lex_brackets <= 0)
2925 yyerror("Unmatched right square bracket");
2928 if (PL_lex_state == LEX_INTERPNORMAL) {
2929 if (PL_lex_brackets == 0) {
2930 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
2931 PL_lex_state = LEX_INTERPEND;
2938 if (PL_lex_brackets > 100) {
2939 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
2940 if (newlb != PL_lex_brackstack) {
2942 PL_lex_brackstack = newlb;
2945 switch (PL_expect) {
2947 if (PL_lex_formbrack) {
2951 if (PL_oldoldbufptr == PL_last_lop)
2952 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2954 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2955 OPERATOR(HASHBRACK);
2957 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2960 PL_tokenbuf[0] = '\0';
2961 if (d < PL_bufend && *d == '-') {
2962 PL_tokenbuf[0] = '-';
2964 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2967 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
2968 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
2970 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2973 char minus = (PL_tokenbuf[0] == '-');
2974 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
2982 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
2987 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2992 if (PL_oldoldbufptr == PL_last_lop)
2993 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2995 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2998 OPERATOR(HASHBRACK);
2999 /* This hack serves to disambiguate a pair of curlies
3000 * as being a block or an anon hash. Normally, expectation
3001 * determines that, but in cases where we're not in a
3002 * position to expect anything in particular (like inside
3003 * eval"") we have to resolve the ambiguity. This code
3004 * covers the case where the first term in the curlies is a
3005 * quoted string. Most other cases need to be explicitly
3006 * disambiguated by prepending a `+' before the opening
3007 * curly in order to force resolution as an anon hash.
3009 * XXX should probably propagate the outer expectation
3010 * into eval"" to rely less on this hack, but that could
3011 * potentially break current behavior of eval"".
3015 if (*s == '\'' || *s == '"' || *s == '`') {
3016 /* common case: get past first string, handling escapes */
3017 for (t++; t < PL_bufend && *t != *s;)
3018 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3022 else if (*s == 'q') {
3025 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3029 char open, close, term;
3032 while (t < PL_bufend && isSPACE(*t))
3036 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3040 for (t++; t < PL_bufend; t++) {
3041 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3043 else if (*t == open)
3047 for (t++; t < PL_bufend; t++) {
3048 if (*t == '\\' && t+1 < PL_bufend)
3050 else if (*t == close && --brackets <= 0)
3052 else if (*t == open)
3058 else if (isALNUM_lazy_if(t,UTF)) {
3060 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3063 while (t < PL_bufend && isSPACE(*t))
3065 /* if comma follows first term, call it an anon hash */
3066 /* XXX it could be a comma expression with loop modifiers */
3067 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3068 || (*t == '=' && t[1] == '>')))
3069 OPERATOR(HASHBRACK);
3070 if (PL_expect == XREF)
3073 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3079 yylval.ival = CopLINE(PL_curcop);
3080 if (isSPACE(*s) || *s == '#')
3081 PL_copline = NOLINE; /* invalidate current command line number */
3086 if (PL_lex_brackets <= 0)
3087 yyerror("Unmatched right curly bracket");
3089 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3090 if (PL_lex_brackets < PL_lex_formbrack)
3091 PL_lex_formbrack = 0;
3092 if (PL_lex_state == LEX_INTERPNORMAL) {
3093 if (PL_lex_brackets == 0) {
3094 if (PL_expect & XFAKEBRACK) {
3095 PL_expect &= XENUMMASK;
3096 PL_lex_state = LEX_INTERPEND;
3098 return yylex(); /* ignore fake brackets */
3100 if (*s == '-' && s[1] == '>')
3101 PL_lex_state = LEX_INTERPENDMAYBE;
3102 else if (*s != '[' && *s != '{')
3103 PL_lex_state = LEX_INTERPEND;
3106 if (PL_expect & XFAKEBRACK) {
3107 PL_expect &= XENUMMASK;
3109 return yylex(); /* ignore fake brackets */
3119 if (PL_expect == XOPERATOR) {
3120 if (ckWARN(WARN_SEMICOLON)
3121 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3123 CopLINE_dec(PL_curcop);
3124 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3125 CopLINE_inc(PL_curcop);
3130 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3132 PL_expect = XOPERATOR;
3133 force_ident(PL_tokenbuf, '&');
3137 yylval.ival = (OPpENTERSUB_AMPER<<8);
3156 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3157 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3159 if (PL_expect == XSTATE && isALPHA(tmp) &&
3160 (s == PL_linestart+1 || s[-2] == '\n') )
3162 if (PL_in_eval && !PL_rsfp) {
3167 if (strnEQ(s,"=cut",4)) {
3181 PL_doextract = TRUE;
3184 if (PL_lex_brackets < PL_lex_formbrack) {
3186 #ifdef PERL_STRICT_CR
3187 for (t = s; *t == ' ' || *t == '\t'; t++) ;
3189 for (t = s; *t == ' ' || *t == '\t' || *t == '\r'; t++) ;
3191 if (*t == '\n' || *t == '#') {
3209 if (PL_expect != XOPERATOR) {
3210 if (s[1] != '<' && !strchr(s,'>'))
3213 s = scan_heredoc(s);
3215 s = scan_inputsymbol(s);
3216 TERM(sublex_start());
3221 SHop(OP_LEFT_SHIFT);
3235 SHop(OP_RIGHT_SHIFT);
3244 if (PL_expect == XOPERATOR) {
3245 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3248 return ','; /* grandfather non-comma-format format */
3252 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3253 PL_tokenbuf[0] = '@';
3254 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3255 sizeof PL_tokenbuf - 1, FALSE);
3256 if (PL_expect == XOPERATOR)
3257 no_op("Array length", s);
3258 if (!PL_tokenbuf[1])
3260 PL_expect = XOPERATOR;
3261 PL_pending_ident = '#';
3265 PL_tokenbuf[0] = '$';
3266 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3267 sizeof PL_tokenbuf - 1, FALSE);
3268 if (PL_expect == XOPERATOR)
3270 if (!PL_tokenbuf[1]) {
3272 yyerror("Final $ should be \\$ or $name");
3276 /* This kludge not intended to be bulletproof. */
3277 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3278 yylval.opval = newSVOP(OP_CONST, 0,
3279 newSViv((IV)PL_compiling.cop_arybase));
3280 yylval.opval->op_private = OPpCONST_ARYBASE;
3286 if (PL_lex_state == LEX_NORMAL)
3289 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3292 PL_tokenbuf[0] = '@';
3293 if (ckWARN(WARN_SYNTAX)) {
3295 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3298 PL_bufptr = skipspace(PL_bufptr);
3299 while (t < PL_bufend && *t != ']')
3301 Perl_warner(aTHX_ WARN_SYNTAX,
3302 "Multidimensional syntax %.*s not supported",
3303 (t - PL_bufptr) + 1, PL_bufptr);
3307 else if (*s == '{') {
3308 PL_tokenbuf[0] = '%';
3309 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3310 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3312 char tmpbuf[sizeof PL_tokenbuf];
3314 for (t++; isSPACE(*t); t++) ;
3315 if (isIDFIRST_lazy_if(t,UTF)) {
3316 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3317 for (; isSPACE(*t); t++) ;
3318 if (*t == ';' && get_cv(tmpbuf, FALSE))
3319 Perl_warner(aTHX_ WARN_SYNTAX,
3320 "You need to quote \"%s\"", tmpbuf);
3326 PL_expect = XOPERATOR;
3327 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3328 bool islop = (PL_last_lop == PL_oldoldbufptr);
3329 if (!islop || PL_last_lop_op == OP_GREPSTART)
3330 PL_expect = XOPERATOR;
3331 else if (strchr("$@\"'`q", *s))
3332 PL_expect = XTERM; /* e.g. print $fh "foo" */
3333 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3334 PL_expect = XTERM; /* e.g. print $fh &sub */
3335 else if (isIDFIRST_lazy_if(s,UTF)) {
3336 char tmpbuf[sizeof PL_tokenbuf];
3337 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3338 if ((tmp = keyword(tmpbuf, len))) {
3339 /* binary operators exclude handle interpretations */
3351 PL_expect = XTERM; /* e.g. print $fh length() */
3356 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3357 if (gv && GvCVu(gv))
3358 PL_expect = XTERM; /* e.g. print $fh subr() */
3361 else if (isDIGIT(*s))
3362 PL_expect = XTERM; /* e.g. print $fh 3 */
3363 else if (*s == '.' && isDIGIT(s[1]))
3364 PL_expect = XTERM; /* e.g. print $fh .3 */
3365 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3366 PL_expect = XTERM; /* e.g. print $fh -1 */
3367 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3368 PL_expect = XTERM; /* print $fh <<"EOF" */
3370 PL_pending_ident = '$';
3374 if (PL_expect == XOPERATOR)
3376 PL_tokenbuf[0] = '@';
3377 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3378 if (!PL_tokenbuf[1]) {
3380 yyerror("Final @ should be \\@ or @name");
3383 if (PL_lex_state == LEX_NORMAL)
3385 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3387 PL_tokenbuf[0] = '%';
3389 /* Warn about @ where they meant $. */
3390 if (ckWARN(WARN_SYNTAX)) {
3391 if (*s == '[' || *s == '{') {
3393 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3395 if (*t == '}' || *t == ']') {
3397 PL_bufptr = skipspace(PL_bufptr);
3398 Perl_warner(aTHX_ WARN_SYNTAX,
3399 "Scalar value %.*s better written as $%.*s",
3400 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3405 PL_pending_ident = '@';
3408 case '/': /* may either be division or pattern */
3409 case '?': /* may either be conditional or pattern */
3410 if (PL_expect != XOPERATOR) {
3411 /* Disable warning on "study /blah/" */
3412 if (PL_oldoldbufptr == PL_last_uni
3413 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3414 || memNE(PL_last_uni, "study", 5)
3415 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3417 s = scan_pat(s,OP_MATCH);
3418 TERM(sublex_start());
3426 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3427 #ifdef PERL_STRICT_CR
3430 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3432 && (s == PL_linestart || s[-1] == '\n') )
3434 PL_lex_formbrack = 0;
3438 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3444 yylval.ival = OPf_SPECIAL;
3450 if (PL_expect != XOPERATOR)
3455 case '0': case '1': case '2': case '3': case '4':
3456 case '5': case '6': case '7': case '8': case '9':
3458 if (PL_expect == XOPERATOR)
3463 s = scan_str(s,FALSE,FALSE);
3464 if (PL_expect == XOPERATOR) {
3465 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3468 return ','; /* grandfather non-comma-format format */
3474 missingterm((char*)0);
3475 yylval.ival = OP_CONST;
3476 TERM(sublex_start());
3479 s = scan_str(s,FALSE,FALSE);
3480 if (PL_expect == XOPERATOR) {
3481 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3484 return ','; /* grandfather non-comma-format format */
3490 missingterm((char*)0);
3491 yylval.ival = OP_CONST;
3492 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3493 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3494 yylval.ival = OP_STRINGIFY;
3498 TERM(sublex_start());
3501 s = scan_str(s,FALSE,FALSE);
3502 if (PL_expect == XOPERATOR)
3503 no_op("Backticks",s);
3505 missingterm((char*)0);
3506 yylval.ival = OP_BACKTICK;
3508 TERM(sublex_start());
3512 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3513 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3515 if (PL_expect == XOPERATOR)
3516 no_op("Backslash",s);
3520 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3524 while (isDIGIT(*start) || *start == '_')
3526 if (*start == '.' && isDIGIT(start[1])) {
3530 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3531 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF)) {
3535 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3545 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3584 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3586 /* Some keywords can be followed by any delimiter, including ':' */
3587 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3588 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3589 (PL_tokenbuf[0] == 'q' &&
3590 strchr("qwxr", PL_tokenbuf[1])))));
3592 /* x::* is just a word, unless x is "CORE" */
3593 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3597 while (d < PL_bufend && isSPACE(*d))
3598 d++; /* no comments skipped here, or s### is misparsed */
3600 /* Is this a label? */
3601 if (!tmp && PL_expect == XSTATE
3602 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3604 yylval.pval = savepv(PL_tokenbuf);
3609 /* Check for keywords */
3610 tmp = keyword(PL_tokenbuf, len);
3612 /* Is this a word before a => operator? */
3613 if (strnEQ(d,"=>",2)) {
3615 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3616 yylval.opval->op_private = OPpCONST_BARE;
3620 if (tmp < 0) { /* second-class keyword? */
3621 GV *ogv = Nullgv; /* override (winner) */
3622 GV *hgv = Nullgv; /* hidden (loser) */
3623 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3625 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3628 if (GvIMPORTED_CV(gv))
3630 else if (! CvMETHOD(cv))
3634 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3635 (gv = *gvp) != (GV*)&PL_sv_undef &&
3636 GvCVu(gv) && GvIMPORTED_CV(gv))
3642 tmp = 0; /* overridden by import or by GLOBAL */
3645 && -tmp==KEY_lock /* XXX generalizable kludge */
3647 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3649 tmp = 0; /* any sub overrides "weak" keyword */
3651 else { /* no override */
3655 if (ckWARN(WARN_AMBIGUOUS) && hgv
3656 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3657 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3658 "Ambiguous call resolved as CORE::%s(), %s",
3659 GvENAME(hgv), "qualify as such or use &");
3666 default: /* not a keyword */
3669 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3671 /* Get the rest if it looks like a package qualifier */
3673 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3675 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3678 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3679 *s == '\'' ? "'" : "::");
3683 if (PL_expect == XOPERATOR) {
3684 if (PL_bufptr == PL_linestart) {
3685 CopLINE_dec(PL_curcop);
3686 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3687 CopLINE_inc(PL_curcop);
3690 no_op("Bareword",s);
3693 /* Look for a subroutine with this name in current package,
3694 unless name is "Foo::", in which case Foo is a bearword
3695 (and a package name). */
3698 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3700 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3701 Perl_warner(aTHX_ WARN_BAREWORD,
3702 "Bareword \"%s\" refers to nonexistent package",
3705 PL_tokenbuf[len] = '\0';
3712 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3715 /* if we saw a global override before, get the right name */
3718 sv = newSVpvn("CORE::GLOBAL::",14);
3719 sv_catpv(sv,PL_tokenbuf);
3722 sv = newSVpv(PL_tokenbuf,0);
3724 /* Presume this is going to be a bareword of some sort. */
3727 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3728 yylval.opval->op_private = OPpCONST_BARE;
3730 /* And if "Foo::", then that's what it certainly is. */
3735 /* See if it's the indirect object for a list operator. */
3737 if (PL_oldoldbufptr &&
3738 PL_oldoldbufptr < PL_bufptr &&
3739 (PL_oldoldbufptr == PL_last_lop
3740 || PL_oldoldbufptr == PL_last_uni) &&
3741 /* NO SKIPSPACE BEFORE HERE! */
3742 (PL_expect == XREF ||
3743 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3745 bool immediate_paren = *s == '(';
3747 /* (Now we can afford to cross potential line boundary.) */
3750 /* Two barewords in a row may indicate method call. */
3752 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3755 /* If not a declared subroutine, it's an indirect object. */
3756 /* (But it's an indir obj regardless for sort.) */
3758 if ((PL_last_lop_op == OP_SORT ||
3759 (!immediate_paren && (!gv || !GvCVu(gv)))) &&
3760 (PL_last_lop_op != OP_MAPSTART &&
3761 PL_last_lop_op != OP_GREPSTART))
3763 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3768 /* If followed by a paren, it's certainly a subroutine. */
3770 PL_expect = XOPERATOR;
3774 if (gv && GvCVu(gv)) {
3775 for (d = s + 1; *d == ' ' || *d == '\t'; d++) ;
3776 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3781 PL_nextval[PL_nexttoke].opval = yylval.opval;
3782 PL_expect = XOPERATOR;
3788 /* If followed by var or block, call it a method (unless sub) */
3790 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3791 PL_last_lop = PL_oldbufptr;
3792 PL_last_lop_op = OP_METHOD;
3796 /* If followed by a bareword, see if it looks like indir obj. */
3798 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3801 /* Not a method, so call it a subroutine (if defined) */
3803 if (gv && GvCVu(gv)) {
3805 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3806 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3807 "Ambiguous use of -%s resolved as -&%s()",
3808 PL_tokenbuf, PL_tokenbuf);
3809 /* Check for a constant sub */
3811 if ((sv = cv_const_sv(cv))) {
3813 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3814 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3815 yylval.opval->op_private = 0;
3819 /* Resolve to GV now. */
3820 op_free(yylval.opval);
3821 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3822 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3823 PL_last_lop = PL_oldbufptr;
3824 PL_last_lop_op = OP_ENTERSUB;
3825 /* Is there a prototype? */
3828 char *proto = SvPV((SV*)cv, len);
3831 if (strEQ(proto, "$"))
3833 if (*proto == '&' && *s == '{') {
3834 sv_setpv(PL_subname,"__ANON__");
3838 PL_nextval[PL_nexttoke].opval = yylval.opval;
3844 /* Call it a bare word */
3846 if (PL_hints & HINT_STRICT_SUBS)
3847 yylval.opval->op_private |= OPpCONST_STRICT;
3850 if (ckWARN(WARN_RESERVED)) {
3851 if (lastchar != '-') {
3852 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
3854 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
3861 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
3862 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3863 "Operator or semicolon missing before %c%s",
3864 lastchar, PL_tokenbuf);
3865 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3866 "Ambiguous use of %c resolved as operator %c",
3867 lastchar, lastchar);
3873 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3874 newSVpv(CopFILE(PL_curcop),0));
3878 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3879 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
3882 case KEY___PACKAGE__:
3883 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3885 ? newSVsv(PL_curstname)
3894 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
3895 char *pname = "main";
3896 if (PL_tokenbuf[2] == 'D')
3897 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
3898 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
3901 GvIOp(gv) = newIO();
3902 IoIFP(GvIOp(gv)) = PL_rsfp;
3903 #if defined(HAS_FCNTL) && defined(F_SETFD)
3905 int fd = PerlIO_fileno(PL_rsfp);
3906 fcntl(fd,F_SETFD,fd >= 3);
3909 /* Mark this internal pseudo-handle as clean */
3910 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
3912 IoTYPE(GvIOp(gv)) = '|';
3913 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
3914 IoTYPE(GvIOp(gv)) = '-';
3916 IoTYPE(GvIOp(gv)) = '<';
3917 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
3918 /* if the script was opened in binmode, we need to revert
3919 * it to text mode for compatibility; but only iff it has CRs
3920 * XXX this is a questionable hack at best. */
3921 if (PL_bufend-PL_bufptr > 2
3922 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
3925 if (IoTYPE(GvIOp(gv)) == '<') {
3926 loc = PerlIO_tell(PL_rsfp);
3927 (void)PerlIO_seek(PL_rsfp, 0L, 0);
3929 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
3930 #if defined(__BORLANDC__)
3931 /* XXX see note in do_binmode() */
3932 ((FILE*)PL_rsfp)->flags |= _F_BIN;
3935 PerlIO_seek(PL_rsfp, loc, 0);
3950 if (PL_expect == XSTATE) {
3957 if (*s == ':' && s[1] == ':') {
3960 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3961 tmp = keyword(PL_tokenbuf, len);
3975 LOP(OP_ACCEPT,XTERM);
3981 LOP(OP_ATAN2,XTERM);
3987 LOP(OP_BINMODE,XTERM);
3990 LOP(OP_BLESS,XTERM);
3999 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4016 if (!PL_cryptseen) {
4017 PL_cryptseen = TRUE;
4021 LOP(OP_CRYPT,XTERM);
4024 if (ckWARN(WARN_CHMOD)) {
4025 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4026 if (*d != '0' && isDIGIT(*d))
4027 Perl_warner(aTHX_ WARN_CHMOD,
4028 "chmod() mode argument is missing initial 0");
4030 LOP(OP_CHMOD,XTERM);
4033 LOP(OP_CHOWN,XTERM);
4036 LOP(OP_CONNECT,XTERM);
4052 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4056 PL_hints |= HINT_BLOCK_SCOPE;
4066 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4067 LOP(OP_DBMOPEN,XTERM);
4073 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4080 yylval.ival = CopLINE(PL_curcop);
4094 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4095 UNIBRACK(OP_ENTEREVAL);
4110 case KEY_endhostent:
4116 case KEY_endservent:
4119 case KEY_endprotoent:
4130 yylval.ival = CopLINE(PL_curcop);
4132 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4134 if ((PL_bufend - p) >= 3 &&
4135 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4137 else if ((PL_bufend - p) >= 4 &&
4138 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4141 if (isIDFIRST_lazy_if(p,UTF)) {
4142 p = scan_ident(p, PL_bufend,
4143 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4147 Perl_croak(aTHX_ "Missing $ on loop variable");
4152 LOP(OP_FORMLINE,XTERM);
4158 LOP(OP_FCNTL,XTERM);
4164 LOP(OP_FLOCK,XTERM);
4173 LOP(OP_GREPSTART, XREF);
4176 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4191 case KEY_getpriority:
4192 LOP(OP_GETPRIORITY,XTERM);
4194 case KEY_getprotobyname:
4197 case KEY_getprotobynumber:
4198 LOP(OP_GPBYNUMBER,XTERM);
4200 case KEY_getprotoent:
4212 case KEY_getpeername:
4213 UNI(OP_GETPEERNAME);
4215 case KEY_gethostbyname:
4218 case KEY_gethostbyaddr:
4219 LOP(OP_GHBYADDR,XTERM);
4221 case KEY_gethostent:
4224 case KEY_getnetbyname:
4227 case KEY_getnetbyaddr:
4228 LOP(OP_GNBYADDR,XTERM);
4233 case KEY_getservbyname:
4234 LOP(OP_GSBYNAME,XTERM);
4236 case KEY_getservbyport:
4237 LOP(OP_GSBYPORT,XTERM);
4239 case KEY_getservent:
4242 case KEY_getsockname:
4243 UNI(OP_GETSOCKNAME);
4245 case KEY_getsockopt:
4246 LOP(OP_GSOCKOPT,XTERM);
4268 yylval.ival = CopLINE(PL_curcop);
4272 LOP(OP_INDEX,XTERM);
4278 LOP(OP_IOCTL,XTERM);
4290 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4322 LOP(OP_LISTEN,XTERM);
4331 s = scan_pat(s,OP_MATCH);
4332 TERM(sublex_start());
4335 LOP(OP_MAPSTART, XREF);
4338 LOP(OP_MKDIR,XTERM);
4341 LOP(OP_MSGCTL,XTERM);
4344 LOP(OP_MSGGET,XTERM);
4347 LOP(OP_MSGRCV,XTERM);
4350 LOP(OP_MSGSND,XTERM);
4356 if (isIDFIRST_lazy_if(s,UTF)) {
4357 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4358 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4360 PL_in_my_stash = gv_stashpv(PL_tokenbuf, FALSE);
4361 if (!PL_in_my_stash) {
4364 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4372 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4379 if (PL_expect != XSTATE)
4380 yyerror("\"no\" not allowed in expression");
4381 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4382 s = force_version(s);
4387 if (*s == '(' || (s = skipspace(s), *s == '('))
4394 if (isIDFIRST_lazy_if(s,UTF)) {
4396 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4398 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4399 Perl_warner(aTHX_ WARN_PRECEDENCE,
4400 "Precedence problem: open %.*s should be open(%.*s)",
4406 yylval.ival = OP_OR;
4416 LOP(OP_OPEN_DIR,XTERM);
4419 checkcomma(s,PL_tokenbuf,"filehandle");
4423 checkcomma(s,PL_tokenbuf,"filehandle");
4442 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4446 LOP(OP_PIPE_OP,XTERM);
4449 s = scan_str(s,FALSE,FALSE);
4451 missingterm((char*)0);
4452 yylval.ival = OP_CONST;
4453 TERM(sublex_start());
4459 s = scan_str(s,FALSE,FALSE);
4461 missingterm((char*)0);
4463 if (SvCUR(PL_lex_stuff)) {
4466 d = SvPV_force(PL_lex_stuff, len);
4468 for (; isSPACE(*d) && len; --len, ++d) ;
4471 if (!warned && ckWARN(WARN_QW)) {
4472 for (; !isSPACE(*d) && len; --len, ++d) {
4474 Perl_warner(aTHX_ WARN_QW,
4475 "Possible attempt to separate words with commas");
4478 else if (*d == '#') {
4479 Perl_warner(aTHX_ WARN_QW,
4480 "Possible attempt to put comments in qw() list");
4486 for (; !isSPACE(*d) && len; --len, ++d) ;
4488 words = append_elem(OP_LIST, words,
4489 newSVOP(OP_CONST, 0, newSVpvn(b, d-b)));
4493 PL_nextval[PL_nexttoke].opval = words;
4498 SvREFCNT_dec(PL_lex_stuff);
4499 PL_lex_stuff = Nullsv;
4504 s = scan_str(s,FALSE,FALSE);
4506 missingterm((char*)0);
4507 yylval.ival = OP_STRINGIFY;
4508 if (SvIVX(PL_lex_stuff) == '\'')
4509 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4510 TERM(sublex_start());
4513 s = scan_pat(s,OP_QR);
4514 TERM(sublex_start());
4517 s = scan_str(s,FALSE,FALSE);
4519 missingterm((char*)0);
4520 yylval.ival = OP_BACKTICK;
4522 TERM(sublex_start());
4529 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4530 s = force_version(s);
4533 *PL_tokenbuf = '\0';
4534 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4535 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4536 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4538 yyerror("<> should be quotes");
4546 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4550 LOP(OP_RENAME,XTERM);
4559 LOP(OP_RINDEX,XTERM);
4582 LOP(OP_REVERSE,XTERM);
4593 TERM(sublex_start());
4595 TOKEN(1); /* force error */
4604 LOP(OP_SELECT,XTERM);
4610 LOP(OP_SEMCTL,XTERM);
4613 LOP(OP_SEMGET,XTERM);
4616 LOP(OP_SEMOP,XTERM);
4622 LOP(OP_SETPGRP,XTERM);
4624 case KEY_setpriority:
4625 LOP(OP_SETPRIORITY,XTERM);
4627 case KEY_sethostent:
4633 case KEY_setservent:
4636 case KEY_setprotoent:
4646 LOP(OP_SEEKDIR,XTERM);
4648 case KEY_setsockopt:
4649 LOP(OP_SSOCKOPT,XTERM);
4655 LOP(OP_SHMCTL,XTERM);
4658 LOP(OP_SHMGET,XTERM);
4661 LOP(OP_SHMREAD,XTERM);
4664 LOP(OP_SHMWRITE,XTERM);
4667 LOP(OP_SHUTDOWN,XTERM);
4676 LOP(OP_SOCKET,XTERM);
4678 case KEY_socketpair:
4679 LOP(OP_SOCKPAIR,XTERM);
4682 checkcomma(s,PL_tokenbuf,"subroutine name");
4684 if (*s == ';' || *s == ')') /* probably a close */
4685 Perl_croak(aTHX_ "sort is now a reserved word");
4687 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4691 LOP(OP_SPLIT,XTERM);
4694 LOP(OP_SPRINTF,XTERM);
4697 LOP(OP_SPLICE,XTERM);
4712 LOP(OP_SUBSTR,XTERM);
4718 char tmpbuf[sizeof PL_tokenbuf];
4720 expectation attrful;
4721 bool have_name, have_proto;
4726 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4727 (*s == ':' && s[1] == ':'))
4730 attrful = XATTRBLOCK;
4731 /* remember buffer pos'n for later force_word */
4732 tboffset = s - PL_oldbufptr;
4733 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4734 if (strchr(tmpbuf, ':'))
4735 sv_setpv(PL_subname, tmpbuf);
4737 sv_setsv(PL_subname,PL_curstname);
4738 sv_catpvn(PL_subname,"::",2);
4739 sv_catpvn(PL_subname,tmpbuf,len);
4746 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4747 PL_expect = XTERMBLOCK;
4748 attrful = XATTRTERM;
4749 sv_setpv(PL_subname,"?");
4753 if (key == KEY_format) {
4755 PL_lex_formbrack = PL_lex_brackets + 1;
4757 (void) force_word(PL_oldbufptr + tboffset, WORD,
4762 /* Look for a prototype */
4766 s = scan_str(s,FALSE,FALSE);
4769 SvREFCNT_dec(PL_lex_stuff);
4770 PL_lex_stuff = Nullsv;
4771 Perl_croak(aTHX_ "Prototype not terminated");
4774 d = SvPVX(PL_lex_stuff);
4776 for (p = d; *p; ++p) {
4781 SvCUR(PL_lex_stuff) = tmp;
4789 if (*s == ':' && s[1] != ':')
4790 PL_expect = attrful;
4793 PL_nextval[PL_nexttoke].opval =
4794 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4795 PL_lex_stuff = Nullsv;
4799 sv_setpv(PL_subname,"__ANON__");
4802 (void) force_word(PL_oldbufptr + tboffset, WORD,
4811 LOP(OP_SYSTEM,XREF);
4814 LOP(OP_SYMLINK,XTERM);
4817 LOP(OP_SYSCALL,XTERM);
4820 LOP(OP_SYSOPEN,XTERM);
4823 LOP(OP_SYSSEEK,XTERM);
4826 LOP(OP_SYSREAD,XTERM);
4829 LOP(OP_SYSWRITE,XTERM);
4833 TERM(sublex_start());
4854 LOP(OP_TRUNCATE,XTERM);
4866 yylval.ival = CopLINE(PL_curcop);
4870 yylval.ival = CopLINE(PL_curcop);
4874 LOP(OP_UNLINK,XTERM);
4880 LOP(OP_UNPACK,XTERM);
4883 LOP(OP_UTIME,XTERM);
4886 if (ckWARN(WARN_UMASK)) {
4887 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4888 if (*d != '0' && isDIGIT(*d))
4889 Perl_warner(aTHX_ WARN_UMASK,
4890 "umask: argument is missing initial 0");
4895 LOP(OP_UNSHIFT,XTERM);
4898 if (PL_expect != XSTATE)
4899 yyerror("\"use\" not allowed in expression");
4901 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4902 s = force_version(s);
4903 if (*s == ';' || (s = skipspace(s), *s == ';')) {
4904 PL_nextval[PL_nexttoke].opval = Nullop;
4909 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4910 s = force_version(s);
4922 yylval.ival = CopLINE(PL_curcop);
4926 PL_hints |= HINT_BLOCK_SCOPE;
4933 LOP(OP_WAITPID,XTERM);
4941 static char ctl_l[2];
4943 if (ctl_l[0] == '\0')
4944 ctl_l[0] = toCTRL('L');
4945 gv_fetchpv(ctl_l,TRUE, SVt_PV);
4948 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
4953 if (PL_expect == XOPERATOR)
4959 yylval.ival = OP_XOR;
4964 TERM(sublex_start());
4970 Perl_keyword(pTHX_ register char *d, I32 len)
4975 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
4976 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
4977 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
4978 if (strEQ(d,"__DATA__")) return KEY___DATA__;
4979 if (strEQ(d,"__END__")) return KEY___END__;
4983 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
4988 if (strEQ(d,"and")) return -KEY_and;
4989 if (strEQ(d,"abs")) return -KEY_abs;
4992 if (strEQ(d,"alarm")) return -KEY_alarm;
4993 if (strEQ(d,"atan2")) return -KEY_atan2;
4996 if (strEQ(d,"accept")) return -KEY_accept;
5001 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5004 if (strEQ(d,"bless")) return -KEY_bless;
5005 if (strEQ(d,"bind")) return -KEY_bind;
5006 if (strEQ(d,"binmode")) return -KEY_binmode;
5009 if (strEQ(d,"CORE")) return -KEY_CORE;
5010 if (strEQ(d,"CHECK")) return KEY_CHECK;
5015 if (strEQ(d,"cmp")) return -KEY_cmp;
5016 if (strEQ(d,"chr")) return -KEY_chr;
5017 if (strEQ(d,"cos")) return -KEY_cos;
5020 if (strEQ(d,"chop")) return KEY_chop;
5023 if (strEQ(d,"close")) return -KEY_close;
5024 if (strEQ(d,"chdir")) return -KEY_chdir;
5025 if (strEQ(d,"chomp")) return KEY_chomp;
5026 if (strEQ(d,"chmod")) return -KEY_chmod;
5027 if (strEQ(d,"chown")) return -KEY_chown;
5028 if (strEQ(d,"crypt")) return -KEY_crypt;
5031 if (strEQ(d,"chroot")) return -KEY_chroot;
5032 if (strEQ(d,"caller")) return -KEY_caller;
5035 if (strEQ(d,"connect")) return -KEY_connect;
5038 if (strEQ(d,"closedir")) return -KEY_closedir;
5039 if (strEQ(d,"continue")) return -KEY_continue;
5044 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5049 if (strEQ(d,"do")) return KEY_do;
5052 if (strEQ(d,"die")) return -KEY_die;
5055 if (strEQ(d,"dump")) return -KEY_dump;
5058 if (strEQ(d,"delete")) return KEY_delete;
5061 if (strEQ(d,"defined")) return KEY_defined;
5062 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5065 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5070 if (strEQ(d,"EQ")) { deprecate(d); return -KEY_eq;}
5071 if (strEQ(d,"END")) return KEY_END;
5076 if (strEQ(d,"eq")) return -KEY_eq;
5079 if (strEQ(d,"eof")) return -KEY_eof;
5080 if (strEQ(d,"exp")) return -KEY_exp;
5083 if (strEQ(d,"else")) return KEY_else;
5084 if (strEQ(d,"exit")) return -KEY_exit;
5085 if (strEQ(d,"eval")) return KEY_eval;
5086 if (strEQ(d,"exec")) return -KEY_exec;
5087 if (strEQ(d,"each")) return KEY_each;
5090 if (strEQ(d,"elsif")) return KEY_elsif;
5093 if (strEQ(d,"exists")) return KEY_exists;
5094 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5097 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5098 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5101 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5104 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5105 if (strEQ(d,"endservent")) return -KEY_endservent;
5108 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5115 if (strEQ(d,"for")) return KEY_for;
5118 if (strEQ(d,"fork")) return -KEY_fork;
5121 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5122 if (strEQ(d,"flock")) return -KEY_flock;
5125 if (strEQ(d,"format")) return KEY_format;
5126 if (strEQ(d,"fileno")) return -KEY_fileno;
5129 if (strEQ(d,"foreach")) return KEY_foreach;
5132 if (strEQ(d,"formline")) return -KEY_formline;
5138 if (strEQ(d,"GT")) { deprecate(d); return -KEY_gt;}
5139 if (strEQ(d,"GE")) { deprecate(d); return -KEY_ge;}
5143 if (strnEQ(d,"get",3)) {
5148 if (strEQ(d,"ppid")) return -KEY_getppid;
5149 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5152 if (strEQ(d,"pwent")) return -KEY_getpwent;
5153 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5154 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5157 if (strEQ(d,"peername")) return -KEY_getpeername;
5158 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5159 if (strEQ(d,"priority")) return -KEY_getpriority;
5162 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5165 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5169 else if (*d == 'h') {
5170 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5171 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5172 if (strEQ(d,"hostent")) return -KEY_gethostent;
5174 else if (*d == 'n') {
5175 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5176 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5177 if (strEQ(d,"netent")) return -KEY_getnetent;
5179 else if (*d == 's') {
5180 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5181 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5182 if (strEQ(d,"servent")) return -KEY_getservent;
5183 if (strEQ(d,"sockname")) return -KEY_getsockname;
5184 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5186 else if (*d == 'g') {
5187 if (strEQ(d,"grent")) return -KEY_getgrent;
5188 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5189 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5191 else if (*d == 'l') {
5192 if (strEQ(d,"login")) return -KEY_getlogin;
5194 else if (strEQ(d,"c")) return -KEY_getc;
5199 if (strEQ(d,"gt")) return -KEY_gt;
5200 if (strEQ(d,"ge")) return -KEY_ge;
5203 if (strEQ(d,"grep")) return KEY_grep;
5204 if (strEQ(d,"goto")) return KEY_goto;
5205 if (strEQ(d,"glob")) return KEY_glob;
5208 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5213 if (strEQ(d,"hex")) return -KEY_hex;
5216 if (strEQ(d,"INIT")) return KEY_INIT;
5221 if (strEQ(d,"if")) return KEY_if;
5224 if (strEQ(d,"int")) return -KEY_int;
5227 if (strEQ(d,"index")) return -KEY_index;
5228 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5233 if (strEQ(d,"join")) return -KEY_join;
5237 if (strEQ(d,"keys")) return KEY_keys;
5238 if (strEQ(d,"kill")) return -KEY_kill;
5243 if (strEQ(d,"LT")) { deprecate(d); return -KEY_lt;}
5244 if (strEQ(d,"LE")) { deprecate(d); return -KEY_le;}
5250 if (strEQ(d,"lt")) return -KEY_lt;
5251 if (strEQ(d,"le")) return -KEY_le;
5252 if (strEQ(d,"lc")) return -KEY_lc;
5255 if (strEQ(d,"log")) return -KEY_log;
5258 if (strEQ(d,"last")) return KEY_last;
5259 if (strEQ(d,"link")) return -KEY_link;
5260 if (strEQ(d,"lock")) return -KEY_lock;
5263 if (strEQ(d,"local")) return KEY_local;
5264 if (strEQ(d,"lstat")) return -KEY_lstat;
5267 if (strEQ(d,"length")) return -KEY_length;
5268 if (strEQ(d,"listen")) return -KEY_listen;
5271 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5274 if (strEQ(d,"localtime")) return -KEY_localtime;
5280 case 1: return KEY_m;
5282 if (strEQ(d,"my")) return KEY_my;
5285 if (strEQ(d,"map")) return KEY_map;
5288 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5291 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5292 if (strEQ(d,"msgget")) return -KEY_msgget;
5293 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5294 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5299 if (strEQ(d,"NE")) { deprecate(d); return -KEY_ne;}
5302 if (strEQ(d,"next")) return KEY_next;
5303 if (strEQ(d,"ne")) return -KEY_ne;
5304 if (strEQ(d,"not")) return -KEY_not;
5305 if (strEQ(d,"no")) return KEY_no;
5310 if (strEQ(d,"or")) return -KEY_or;
5313 if (strEQ(d,"ord")) return -KEY_ord;
5314 if (strEQ(d,"oct")) return -KEY_oct;
5315 if (strEQ(d,"our")) return KEY_our;
5318 if (strEQ(d,"open")) return -KEY_open;
5321 if (strEQ(d,"opendir")) return -KEY_opendir;
5328 if (strEQ(d,"pop")) return KEY_pop;
5329 if (strEQ(d,"pos")) return KEY_pos;
5332 if (strEQ(d,"push")) return KEY_push;
5333 if (strEQ(d,"pack")) return -KEY_pack;
5334 if (strEQ(d,"pipe")) return -KEY_pipe;
5337 if (strEQ(d,"print")) return KEY_print;
5340 if (strEQ(d,"printf")) return KEY_printf;
5343 if (strEQ(d,"package")) return KEY_package;
5346 if (strEQ(d,"prototype")) return KEY_prototype;
5351 if (strEQ(d,"q")) return KEY_q;
5352 if (strEQ(d,"qr")) return KEY_qr;
5353 if (strEQ(d,"qq")) return KEY_qq;
5354 if (strEQ(d,"qw")) return KEY_qw;
5355 if (strEQ(d,"qx")) return KEY_qx;
5357 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5362 if (strEQ(d,"ref")) return -KEY_ref;
5365 if (strEQ(d,"read")) return -KEY_read;
5366 if (strEQ(d,"rand")) return -KEY_rand;
5367 if (strEQ(d,"recv")) return -KEY_recv;
5368 if (strEQ(d,"redo")) return KEY_redo;
5371 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5372 if (strEQ(d,"reset")) return -KEY_reset;
5375 if (strEQ(d,"return")) return KEY_return;
5376 if (strEQ(d,"rename")) return -KEY_rename;
5377 if (strEQ(d,"rindex")) return -KEY_rindex;
5380 if (strEQ(d,"require")) return -KEY_require;
5381 if (strEQ(d,"reverse")) return -KEY_reverse;
5382 if (strEQ(d,"readdir")) return -KEY_readdir;
5385 if (strEQ(d,"readlink")) return -KEY_readlink;
5386 if (strEQ(d,"readline")) return -KEY_readline;
5387 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5390 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5396 case 0: return KEY_s;
5398 if (strEQ(d,"scalar")) return KEY_scalar;
5403 if (strEQ(d,"seek")) return -KEY_seek;
5404 if (strEQ(d,"send")) return -KEY_send;
5407 if (strEQ(d,"semop")) return -KEY_semop;
5410 if (strEQ(d,"select")) return -KEY_select;
5411 if (strEQ(d,"semctl")) return -KEY_semctl;
5412 if (strEQ(d,"semget")) return -KEY_semget;
5415 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5416 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5419 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5420 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5423 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5426 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5427 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5428 if (strEQ(d,"setservent")) return -KEY_setservent;
5431 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5432 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5439 if (strEQ(d,"shift")) return KEY_shift;
5442 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5443 if (strEQ(d,"shmget")) return -KEY_shmget;
5446 if (strEQ(d,"shmread")) return -KEY_shmread;
5449 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5450 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5455 if (strEQ(d,"sin")) return -KEY_sin;
5458 if (strEQ(d,"sleep")) return -KEY_sleep;
5461 if (strEQ(d,"sort")) return KEY_sort;
5462 if (strEQ(d,"socket")) return -KEY_socket;
5463 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5466 if (strEQ(d,"split")) return KEY_split;
5467 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5468 if (strEQ(d,"splice")) return KEY_splice;
5471 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5474 if (strEQ(d,"srand")) return -KEY_srand;
5477 if (strEQ(d,"stat")) return -KEY_stat;
5478 if (strEQ(d,"study")) return KEY_study;
5481 if (strEQ(d,"substr")) return -KEY_substr;
5482 if (strEQ(d,"sub")) return KEY_sub;
5487 if (strEQ(d,"system")) return -KEY_system;
5490 if (strEQ(d,"symlink")) return -KEY_symlink;
5491 if (strEQ(d,"syscall")) return -KEY_syscall;
5492 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5493 if (strEQ(d,"sysread")) return -KEY_sysread;
5494 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5497 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5506 if (strEQ(d,"tr")) return KEY_tr;
5509 if (strEQ(d,"tie")) return KEY_tie;
5512 if (strEQ(d,"tell")) return -KEY_tell;
5513 if (strEQ(d,"tied")) return KEY_tied;
5514 if (strEQ(d,"time")) return -KEY_time;
5517 if (strEQ(d,"times")) return -KEY_times;
5520 if (strEQ(d,"telldir")) return -KEY_telldir;
5523 if (strEQ(d,"truncate")) return -KEY_truncate;
5530 if (strEQ(d,"uc")) return -KEY_uc;
5533 if (strEQ(d,"use")) return KEY_use;
5536 if (strEQ(d,"undef")) return KEY_undef;
5537 if (strEQ(d,"until")) return KEY_until;
5538 if (strEQ(d,"untie")) return KEY_untie;
5539 if (strEQ(d,"utime")) return -KEY_utime;
5540 if (strEQ(d,"umask")) return -KEY_umask;
5543 if (strEQ(d,"unless")) return KEY_unless;
5544 if (strEQ(d,"unpack")) return -KEY_unpack;
5545 if (strEQ(d,"unlink")) return -KEY_unlink;
5548 if (strEQ(d,"unshift")) return KEY_unshift;
5549 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5554 if (strEQ(d,"values")) return -KEY_values;
5555 if (strEQ(d,"vec")) return -KEY_vec;
5560 if (strEQ(d,"warn")) return -KEY_warn;
5561 if (strEQ(d,"wait")) return -KEY_wait;
5564 if (strEQ(d,"while")) return KEY_while;
5565 if (strEQ(d,"write")) return -KEY_write;
5568 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5571 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5576 if (len == 1) return -KEY_x;
5577 if (strEQ(d,"xor")) return -KEY_xor;
5580 if (len == 1) return KEY_y;
5589 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5593 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5594 dTHR; /* only for ckWARN */
5595 if (ckWARN(WARN_SYNTAX)) {
5597 for (w = s+2; *w && level; w++) {
5604 for (; *w && isSPACE(*w); w++) ;
5605 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5606 Perl_warner(aTHX_ WARN_SYNTAX,
5607 "%s (...) interpreted as function",name);
5610 while (s < PL_bufend && isSPACE(*s))
5614 while (s < PL_bufend && isSPACE(*s))
5616 if (isIDFIRST_lazy_if(s,UTF)) {
5618 while (isALNUM_lazy_if(s,UTF))
5620 while (s < PL_bufend && isSPACE(*s))
5625 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5629 Perl_croak(aTHX_ "No comma allowed after %s", what);
5634 /* Either returns sv, or mortalizes sv and returns a new SV*.
5635 Best used as sv=new_constant(..., sv, ...).
5636 If s, pv are NULL, calls subroutine with one argument,
5637 and type is used with error messages only. */
5640 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5644 HV *table = GvHV(PL_hintgv); /* ^H */
5648 const char *why, *why1, *why2;
5650 if (!(PL_hints & HINT_LOCALIZE_HH)) {
5653 why = "%^H is not localized";
5657 msg = Perl_newSVpvf(aTHX_ "constant(%s): %s%s%s",
5658 (type ? type: "undef"), why1, why2, why);
5659 yyerror(SvPVX(msg));
5664 why = "%^H is not defined";
5667 cvp = hv_fetch(table, key, strlen(key), FALSE);
5668 if (!cvp || !SvOK(*cvp)) {
5669 why = "} is not defined";
5674 sv_2mortal(sv); /* Parent created it permanently */
5677 pv = sv_2mortal(newSVpvn(s, len));
5679 typesv = sv_2mortal(newSVpv(type, 0));
5681 typesv = &PL_sv_undef;
5683 PUSHSTACKi(PERLSI_OVERLOAD);
5696 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5700 /* Check the eval first */
5701 if (!PL_in_eval && SvTRUE(ERRSV)) {
5703 sv_catpv(ERRSV, "Propagated");
5704 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5706 res = SvREFCNT_inc(sv);
5710 (void)SvREFCNT_inc(res);
5719 why = "}} did not return a defined value";
5720 why1 = "Call to &{$^H{";
5730 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5732 register char *d = dest;
5733 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5736 Perl_croak(aTHX_ ident_too_long);
5737 if (isALNUM(*s)) /* UTF handled below */
5739 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
5744 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5748 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5749 char *t = s + UTF8SKIP(s);
5750 while (*t & 0x80 && is_utf8_mark((U8*)t))
5752 if (d + (t - s) > e)
5753 Perl_croak(aTHX_ ident_too_long);
5754 Copy(s, d, t - s, char);
5767 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5777 e = d + destlen - 3; /* two-character token, ending NUL */
5779 while (isDIGIT(*s)) {
5781 Perl_croak(aTHX_ ident_too_long);
5788 Perl_croak(aTHX_ ident_too_long);
5789 if (isALNUM(*s)) /* UTF handled below */
5791 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
5796 else if (*s == ':' && s[1] == ':') {
5800 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5801 char *t = s + UTF8SKIP(s);
5802 while (*t & 0x80 && is_utf8_mark((U8*)t))
5804 if (d + (t - s) > e)
5805 Perl_croak(aTHX_ ident_too_long);
5806 Copy(s, d, t - s, char);
5817 if (PL_lex_state != LEX_NORMAL)
5818 PL_lex_state = LEX_INTERPENDMAYBE;
5821 if (*s == '$' && s[1] &&
5822 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
5835 if (*d == '^' && *s && isCONTROLVAR(*s)) {
5840 if (isSPACE(s[-1])) {
5843 if (ch != ' ' && ch != '\t') {
5849 if (isIDFIRST_lazy_if(d,UTF)) {
5853 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
5855 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
5858 Copy(s, d, e - s, char);
5863 while ((isALNUM(*s) || *s == ':') && d < e)
5866 Perl_croak(aTHX_ ident_too_long);
5869 while (s < send && (*s == ' ' || *s == '\t')) s++;
5870 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
5871 dTHR; /* only for ckWARN */
5872 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
5873 const char *brack = *s == '[' ? "[...]" : "{...}";
5874 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5875 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
5876 funny, dest, brack, funny, dest, brack);
5879 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
5883 /* Handle extended ${^Foo} variables
5884 * 1999-02-27 mjd-perl-patch@plover.com */
5885 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
5889 while (isALNUM(*s) && d < e) {
5893 Perl_croak(aTHX_ ident_too_long);
5898 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
5899 PL_lex_state = LEX_INTERPEND;
5902 if (PL_lex_state == LEX_NORMAL) {
5903 dTHR; /* only for ckWARN */
5904 if (ckWARN(WARN_AMBIGUOUS) &&
5905 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
5907 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5908 "Ambiguous use of %c{%s} resolved to %c%s",
5909 funny, dest, funny, dest);
5914 s = bracket; /* let the parser handle it */
5918 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
5919 PL_lex_state = LEX_INTERPEND;
5924 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
5929 *pmfl |= PMf_GLOBAL;
5931 *pmfl |= PMf_CONTINUE;
5935 *pmfl |= PMf_MULTILINE;
5937 *pmfl |= PMf_SINGLELINE;
5939 *pmfl |= PMf_EXTENDED;
5943 S_scan_pat(pTHX_ char *start, I32 type)
5948 s = scan_str(start,FALSE,FALSE);
5951 SvREFCNT_dec(PL_lex_stuff);
5952 PL_lex_stuff = Nullsv;
5953 Perl_croak(aTHX_ "Search pattern not terminated");
5956 pm = (PMOP*)newPMOP(type, 0);
5957 if (PL_multi_open == '?')
5958 pm->op_pmflags |= PMf_ONCE;
5960 while (*s && strchr("iomsx", *s))
5961 pmflag(&pm->op_pmflags,*s++);
5964 while (*s && strchr("iogcmsx", *s))
5965 pmflag(&pm->op_pmflags,*s++);
5967 pm->op_pmpermflags = pm->op_pmflags;
5969 PL_lex_op = (OP*)pm;
5970 yylval.ival = OP_MATCH;
5975 S_scan_subst(pTHX_ char *start)
5982 yylval.ival = OP_NULL;
5984 s = scan_str(start,FALSE,FALSE);
5988 SvREFCNT_dec(PL_lex_stuff);
5989 PL_lex_stuff = Nullsv;
5990 Perl_croak(aTHX_ "Substitution pattern not terminated");
5993 if (s[-1] == PL_multi_open)
5996 first_start = PL_multi_start;
5997 s = scan_str(s,FALSE,FALSE);
6000 SvREFCNT_dec(PL_lex_stuff);
6001 PL_lex_stuff = Nullsv;
6003 SvREFCNT_dec(PL_lex_repl);
6004 PL_lex_repl = Nullsv;
6005 Perl_croak(aTHX_ "Substitution replacement not terminated");
6007 PL_multi_start = first_start; /* so whole substitution is taken together */
6009 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6015 else if (strchr("iogcmsx", *s))
6016 pmflag(&pm->op_pmflags,*s++);
6023 PL_sublex_info.super_bufptr = s;
6024 PL_sublex_info.super_bufend = PL_bufend;
6026 pm->op_pmflags |= PMf_EVAL;
6027 repl = newSVpvn("",0);
6029 sv_catpv(repl, es ? "eval " : "do ");
6030 sv_catpvn(repl, "{ ", 2);
6031 sv_catsv(repl, PL_lex_repl);
6032 sv_catpvn(repl, " };", 2);
6034 SvREFCNT_dec(PL_lex_repl);
6038 pm->op_pmpermflags = pm->op_pmflags;
6039 PL_lex_op = (OP*)pm;
6040 yylval.ival = OP_SUBST;
6045 S_scan_trans(pTHX_ char *start)
6056 yylval.ival = OP_NULL;
6058 s = scan_str(start,FALSE,FALSE);
6061 SvREFCNT_dec(PL_lex_stuff);
6062 PL_lex_stuff = Nullsv;
6063 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6065 if (s[-1] == PL_multi_open)
6068 s = scan_str(s,FALSE,FALSE);
6071 SvREFCNT_dec(PL_lex_stuff);
6072 PL_lex_stuff = Nullsv;
6074 SvREFCNT_dec(PL_lex_repl);
6075 PL_lex_repl = Nullsv;
6076 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6080 o = newSVOP(OP_TRANS, 0, 0);
6081 utf8 = OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF;
6084 New(803,tbl,256,short);
6085 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6089 complement = del = squash = 0;
6090 while (strchr("cdsCU", *s)) {
6092 complement = OPpTRANS_COMPLEMENT;
6094 del = OPpTRANS_DELETE;
6096 squash = OPpTRANS_SQUASH;
6101 utf8 &= ~OPpTRANS_FROM_UTF;
6103 utf8 |= OPpTRANS_FROM_UTF;
6107 utf8 &= ~OPpTRANS_TO_UTF;
6109 utf8 |= OPpTRANS_TO_UTF;
6112 Perl_croak(aTHX_ "Too many /C and /U options");
6117 o->op_private = del|squash|complement|utf8;
6120 yylval.ival = OP_TRANS;
6125 S_scan_heredoc(pTHX_ register char *s)
6129 I32 op_type = OP_SCALAR;
6136 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6140 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6143 for (peek = s; *peek == ' ' || *peek == '\t'; peek++) ;
6144 if (*peek && strchr("`'\"",*peek)) {
6147 s = delimcpy(d, e, s, PL_bufend, term, &len);
6157 if (!isALNUM_lazy_if(s,UTF))
6158 deprecate("bare << to mean <<\"\"");
6159 for (; isALNUM_lazy_if(s,UTF); s++) {
6164 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6165 Perl_croak(aTHX_ "Delimiter for here document is too long");
6168 len = d - PL_tokenbuf;
6169 #ifndef PERL_STRICT_CR
6170 d = strchr(s, '\r');
6174 while (s < PL_bufend) {
6180 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6189 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6194 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6195 herewas = newSVpvn(s,PL_bufend-s);
6197 s--, herewas = newSVpvn(s,d-s);
6198 s += SvCUR(herewas);
6200 tmpstr = NEWSV(87,79);
6201 sv_upgrade(tmpstr, SVt_PVIV);
6206 else if (term == '`') {
6207 op_type = OP_BACKTICK;
6208 SvIVX(tmpstr) = '\\';
6212 PL_multi_start = CopLINE(PL_curcop);
6213 PL_multi_open = PL_multi_close = '<';
6214 term = *PL_tokenbuf;
6215 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6216 char *bufptr = PL_sublex_info.super_bufptr;
6217 char *bufend = PL_sublex_info.super_bufend;
6218 char *olds = s - SvCUR(herewas);
6219 s = strchr(bufptr, '\n');
6223 while (s < bufend &&
6224 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6226 CopLINE_inc(PL_curcop);
6229 CopLINE_set(PL_curcop, PL_multi_start);
6230 missingterm(PL_tokenbuf);
6232 sv_setpvn(herewas,bufptr,d-bufptr+1);
6233 sv_setpvn(tmpstr,d+1,s-d);
6235 sv_catpvn(herewas,s,bufend-s);
6236 (void)strcpy(bufptr,SvPVX(herewas));
6243 while (s < PL_bufend &&
6244 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6246 CopLINE_inc(PL_curcop);
6248 if (s >= PL_bufend) {
6249 CopLINE_set(PL_curcop, PL_multi_start);
6250 missingterm(PL_tokenbuf);
6252 sv_setpvn(tmpstr,d+1,s-d);
6254 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6256 sv_catpvn(herewas,s,PL_bufend-s);
6257 sv_setsv(PL_linestr,herewas);
6258 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6259 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6262 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6263 while (s >= PL_bufend) { /* multiple line string? */
6265 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6266 CopLINE_set(PL_curcop, PL_multi_start);
6267 missingterm(PL_tokenbuf);
6269 CopLINE_inc(PL_curcop);
6270 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6271 #ifndef PERL_STRICT_CR
6272 if (PL_bufend - PL_linestart >= 2) {
6273 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6274 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6276 PL_bufend[-2] = '\n';
6278 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6280 else if (PL_bufend[-1] == '\r')
6281 PL_bufend[-1] = '\n';
6283 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6284 PL_bufend[-1] = '\n';
6286 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6287 SV *sv = NEWSV(88,0);
6289 sv_upgrade(sv, SVt_PVMG);
6290 sv_setsv(sv,PL_linestr);
6291 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6293 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6296 sv_catsv(PL_linestr,herewas);
6297 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6301 sv_catsv(tmpstr,PL_linestr);
6306 PL_multi_end = CopLINE(PL_curcop);
6307 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6308 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6309 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6311 SvREFCNT_dec(herewas);
6312 PL_lex_stuff = tmpstr;
6313 yylval.ival = op_type;
6318 takes: current position in input buffer
6319 returns: new position in input buffer
6320 side-effects: yylval and lex_op are set.
6325 <FH> read from filehandle
6326 <pkg::FH> read from package qualified filehandle
6327 <pkg'FH> read from package qualified filehandle
6328 <$fh> read from filehandle in $fh
6334 S_scan_inputsymbol(pTHX_ char *start)
6336 register char *s = start; /* current position in buffer */
6342 d = PL_tokenbuf; /* start of temp holding space */
6343 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6344 end = strchr(s, '\n');
6347 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6349 /* die if we didn't have space for the contents of the <>,
6350 or if it didn't end, or if we see a newline
6353 if (len >= sizeof PL_tokenbuf)
6354 Perl_croak(aTHX_ "Excessively long <> operator");
6356 Perl_croak(aTHX_ "Unterminated <> operator");
6361 Remember, only scalar variables are interpreted as filehandles by
6362 this code. Anything more complex (e.g., <$fh{$num}>) will be
6363 treated as a glob() call.
6364 This code makes use of the fact that except for the $ at the front,
6365 a scalar variable and a filehandle look the same.
6367 if (*d == '$' && d[1]) d++;
6369 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6370 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6373 /* If we've tried to read what we allow filehandles to look like, and
6374 there's still text left, then it must be a glob() and not a getline.
6375 Use scan_str to pull out the stuff between the <> and treat it
6376 as nothing more than a string.
6379 if (d - PL_tokenbuf != len) {
6380 yylval.ival = OP_GLOB;
6382 s = scan_str(start,FALSE,FALSE);
6384 Perl_croak(aTHX_ "Glob not terminated");
6388 /* we're in a filehandle read situation */
6391 /* turn <> into <ARGV> */
6393 (void)strcpy(d,"ARGV");
6395 /* if <$fh>, create the ops to turn the variable into a
6401 /* try to find it in the pad for this block, otherwise find
6402 add symbol table ops
6404 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6405 OP *o = newOP(OP_PADSV, 0);
6407 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6410 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6411 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6412 newUNOP(OP_RV2SV, 0,
6413 newGVOP(OP_GV, 0, gv)));
6415 PL_lex_op->op_flags |= OPf_SPECIAL;
6416 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6417 yylval.ival = OP_NULL;
6420 /* If it's none of the above, it must be a literal filehandle
6421 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6423 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6424 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6425 yylval.ival = OP_NULL;
6434 takes: start position in buffer
6435 keep_quoted preserve \ on the embedded delimiter(s)
6436 keep_delims preserve the delimiters around the string
6437 returns: position to continue reading from buffer
6438 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6439 updates the read buffer.
6441 This subroutine pulls a string out of the input. It is called for:
6442 q single quotes q(literal text)
6443 ' single quotes 'literal text'
6444 qq double quotes qq(interpolate $here please)
6445 " double quotes "interpolate $here please"
6446 qx backticks qx(/bin/ls -l)
6447 ` backticks `/bin/ls -l`
6448 qw quote words @EXPORT_OK = qw( func() $spam )
6449 m// regexp match m/this/
6450 s/// regexp substitute s/this/that/
6451 tr/// string transliterate tr/this/that/
6452 y/// string transliterate y/this/that/
6453 ($*@) sub prototypes sub foo ($)
6454 (stuff) sub attr parameters sub foo : attr(stuff)
6455 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6457 In most of these cases (all but <>, patterns and transliterate)
6458 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6459 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6460 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6463 It skips whitespace before the string starts, and treats the first
6464 character as the delimiter. If the delimiter is one of ([{< then
6465 the corresponding "close" character )]}> is used as the closing
6466 delimiter. It allows quoting of delimiters, and if the string has
6467 balanced delimiters ([{<>}]) it allows nesting.
6469 The lexer always reads these strings into lex_stuff, except in the
6470 case of the operators which take *two* arguments (s/// and tr///)
6471 when it checks to see if lex_stuff is full (presumably with the 1st
6472 arg to s or tr) and if so puts the string into lex_repl.
6477 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6480 SV *sv; /* scalar value: string */
6481 char *tmps; /* temp string, used for delimiter matching */
6482 register char *s = start; /* current position in the buffer */
6483 register char term; /* terminating character */
6484 register char *to; /* current position in the sv's data */
6485 I32 brackets = 1; /* bracket nesting level */
6486 bool has_utf = FALSE; /* is there any utf8 content? */
6488 /* skip space before the delimiter */
6492 /* mark where we are, in case we need to report errors */
6495 /* after skipping whitespace, the next character is the terminator */
6497 if ((term & 0x80) && UTF)
6500 /* mark where we are */
6501 PL_multi_start = CopLINE(PL_curcop);
6502 PL_multi_open = term;
6504 /* find corresponding closing delimiter */
6505 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6507 PL_multi_close = term;
6509 /* create a new SV to hold the contents. 87 is leak category, I'm
6510 assuming. 79 is the SV's initial length. What a random number. */
6512 sv_upgrade(sv, SVt_PVIV);
6514 (void)SvPOK_only(sv); /* validate pointer */
6516 /* move past delimiter and try to read a complete string */
6518 sv_catpvn(sv, s, 1);
6521 /* extend sv if need be */
6522 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6523 /* set 'to' to the next character in the sv's string */
6524 to = SvPVX(sv)+SvCUR(sv);
6526 /* if open delimiter is the close delimiter read unbridle */
6527 if (PL_multi_open == PL_multi_close) {
6528 for (; s < PL_bufend; s++,to++) {
6529 /* embedded newlines increment the current line number */
6530 if (*s == '\n' && !PL_rsfp)
6531 CopLINE_inc(PL_curcop);
6532 /* handle quoted delimiters */
6533 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6534 if (!keep_quoted && s[1] == term)
6536 /* any other quotes are simply copied straight through */
6540 /* terminate when run out of buffer (the for() condition), or
6541 have found the terminator */
6542 else if (*s == term)
6544 else if (!has_utf && (*s & 0x80) && UTF)
6550 /* if the terminator isn't the same as the start character (e.g.,
6551 matched brackets), we have to allow more in the quoting, and
6552 be prepared for nested brackets.
6555 /* read until we run out of string, or we find the terminator */
6556 for (; s < PL_bufend; s++,to++) {
6557 /* embedded newlines increment the line count */
6558 if (*s == '\n' && !PL_rsfp)
6559 CopLINE_inc(PL_curcop);
6560 /* backslashes can escape the open or closing characters */
6561 if (*s == '\\' && s+1 < PL_bufend) {
6563 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6568 /* allow nested opens and closes */
6569 else if (*s == PL_multi_close && --brackets <= 0)
6571 else if (*s == PL_multi_open)
6573 else if (!has_utf && (*s & 0x80) && UTF)
6578 /* terminate the copied string and update the sv's end-of-string */
6580 SvCUR_set(sv, to - SvPVX(sv));
6583 * this next chunk reads more into the buffer if we're not done yet
6587 break; /* handle case where we are done yet :-) */
6589 #ifndef PERL_STRICT_CR
6590 if (to - SvPVX(sv) >= 2) {
6591 if ((to[-2] == '\r' && to[-1] == '\n') ||
6592 (to[-2] == '\n' && to[-1] == '\r'))
6596 SvCUR_set(sv, to - SvPVX(sv));
6598 else if (to[-1] == '\r')
6601 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6605 /* if we're out of file, or a read fails, bail and reset the current
6606 line marker so we can report where the unterminated string began
6609 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6611 CopLINE_set(PL_curcop, PL_multi_start);
6614 /* we read a line, so increment our line counter */
6615 CopLINE_inc(PL_curcop);
6617 /* update debugger info */
6618 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6619 SV *sv = NEWSV(88,0);
6621 sv_upgrade(sv, SVt_PVMG);
6622 sv_setsv(sv,PL_linestr);
6623 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6626 /* having changed the buffer, we must update PL_bufend */
6627 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6630 /* at this point, we have successfully read the delimited string */
6633 sv_catpvn(sv, s, 1);
6636 PL_multi_end = CopLINE(PL_curcop);
6639 /* if we allocated too much space, give some back */
6640 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6641 SvLEN_set(sv, SvCUR(sv) + 1);
6642 Renew(SvPVX(sv), SvLEN(sv), char);
6645 /* decide whether this is the first or second quoted string we've read
6658 takes: pointer to position in buffer
6659 returns: pointer to new position in buffer
6660 side-effects: builds ops for the constant in yylval.op
6662 Read a number in any of the formats that Perl accepts:
6664 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6665 [\d_]+(\.[\d_]*)?[Ee](\d+)
6667 Underbars (_) are allowed in decimal numbers. If -w is on,
6668 underbars before a decimal point must be at three digit intervals.
6670 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6673 If it reads a number without a decimal point or an exponent, it will
6674 try converting the number to an integer and see if it can do so
6675 without loss of precision.
6679 Perl_scan_num(pTHX_ char *start)
6681 register char *s = start; /* current position in buffer */
6682 register char *d; /* destination in temp buffer */
6683 register char *e; /* end of temp buffer */
6684 NV value; /* number read, as a double */
6685 SV *sv = Nullsv; /* place to put the converted number */
6686 bool floatit; /* boolean: int or float? */
6687 char *lastub = 0; /* position of last underbar */
6688 static char number_too_long[] = "Number too long";
6690 /* We use the first character to decide what type of number this is */
6694 Perl_croak(aTHX_ "panic: scan_num");
6696 /* if it starts with a 0, it could be an octal number, a decimal in
6697 0.13 disguise, or a hexadecimal number, or a binary number. */
6701 u holds the "number so far"
6702 shift the power of 2 of the base
6703 (hex == 4, octal == 3, binary == 1)
6704 overflowed was the number more than we can hold?
6706 Shift is used when we add a digit. It also serves as an "are
6707 we in octal/hex/binary?" indicator to disallow hex characters
6714 bool overflowed = FALSE;
6715 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6716 static char* bases[5] = { "", "binary", "", "octal",
6718 static char* Bases[5] = { "", "Binary", "", "Octal",
6720 static char *maxima[5] = { "",
6721 "0b11111111111111111111111111111111",
6725 char *base, *Base, *max;
6731 } else if (s[1] == 'b') {
6735 /* check for a decimal in disguise */
6736 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6738 /* so it must be octal */
6742 base = bases[shift];
6743 Base = Bases[shift];
6744 max = maxima[shift];
6746 /* read the rest of the number */
6748 /* x is used in the overflow test,
6749 b is the digit we're adding on. */
6754 /* if we don't mention it, we're done */
6763 /* 8 and 9 are not octal */
6766 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6770 case '2': case '3': case '4':
6771 case '5': case '6': case '7':
6773 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6777 b = *s++ & 15; /* ASCII digit -> value of digit */
6781 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6782 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6783 /* make sure they said 0x */
6788 /* Prepare to put the digit we have onto the end
6789 of the number so far. We check for overflows.
6794 x = u << shift; /* make room for the digit */
6796 if ((x >> shift) != u
6797 && !(PL_hints & HINT_NEW_BINARY)) {
6801 if (ckWARN_d(WARN_OVERFLOW))
6802 Perl_warner(aTHX_ WARN_OVERFLOW,
6803 "Integer overflow in %s number",
6806 u = x | b; /* add the digit to the end */
6809 n *= nvshift[shift];
6810 /* If an NV has not enough bits in its
6811 * mantissa to represent an UV this summing of
6812 * small low-order numbers is a waste of time
6813 * (because the NV cannot preserve the
6814 * low-order bits anyway): we could just
6815 * remember when did we overflow and in the
6816 * end just multiply n by the right
6824 /* if we get here, we had success: make a scalar value from
6831 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6832 Perl_warner(aTHX_ WARN_PORTABLE,
6833 "%s number > %s non-portable",
6840 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6841 Perl_warner(aTHX_ WARN_PORTABLE,
6842 "%s number > %s non-portable",
6847 if (PL_hints & HINT_NEW_BINARY)
6848 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6853 handle decimal numbers.
6854 we're also sent here when we read a 0 as the first digit
6856 case '1': case '2': case '3': case '4': case '5':
6857 case '6': case '7': case '8': case '9': case '.':
6860 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
6863 /* read next group of digits and _ and copy into d */
6864 while (isDIGIT(*s) || *s == '_') {
6865 /* skip underscores, checking for misplaced ones
6869 dTHR; /* only for ckWARN */
6870 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
6871 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6875 /* check for end of fixed-length buffer */
6877 Perl_croak(aTHX_ number_too_long);
6878 /* if we're ok, copy the character */
6883 /* final misplaced underbar check */
6884 if (lastub && s - lastub != 3) {
6886 if (ckWARN(WARN_SYNTAX))
6887 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6890 /* read a decimal portion if there is one. avoid
6891 3..5 being interpreted as the number 3. followed
6894 if (*s == '.' && s[1] != '.') {
6898 /* copy, ignoring underbars, until we run out of
6899 digits. Note: no misplaced underbar checks!
6901 for (; isDIGIT(*s) || *s == '_'; s++) {
6902 /* fixed length buffer check */
6904 Perl_croak(aTHX_ number_too_long);
6908 if (*s == '.' && isDIGIT(s[1])) {
6909 /* oops, it's really a v-string, but without the "v" */
6915 /* read exponent part, if present */
6916 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
6920 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
6921 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
6923 /* allow positive or negative exponent */
6924 if (*s == '+' || *s == '-')
6927 /* read digits of exponent (no underbars :-) */
6928 while (isDIGIT(*s)) {
6930 Perl_croak(aTHX_ number_too_long);
6935 /* terminate the string */
6938 /* make an sv from the string */
6941 /* unfortunately this monster needs to be on one line or
6942 makedepend will be confused. */
6943 #if (defined(USE_64_BIT_INT) && (!defined(HAS_STRTOLL)|| !defined(HAS_STRTOULL))) || (!defined(USE_64_BIT_INT) && (!defined(HAS_STRTOL) || !defined(HAS_STRTOUL)))
6946 No working strto[u]l[l]. Since atoi() doesn't do range checks,
6947 we need to do this the hard way.
6950 value = Atof(PL_tokenbuf);
6953 See if we can make do with an integer value without loss of
6954 precision. We use I_V to cast to an int, because some
6955 compilers have issues. Then we try casting it back and see
6956 if it was the same. We only do this if we know we
6957 specifically read an integer.
6959 Note: if floatit is true, then we don't need to do the
6963 UV tryuv = U_V(value);
6964 if (!floatit && (NV)tryuv == value) {
6965 if (tryuv <= IV_MAX)
6966 sv_setiv(sv, (IV)tryuv);
6968 sv_setuv(sv, tryuv);
6971 sv_setnv(sv, value);
6975 strtol/strtoll sets errno to ERANGE if the number is too big
6976 for an integer. We try to do an integer conversion first
6977 if no characters indicating "float" have been found.
6984 if (*PL_tokenbuf == '-')
6985 iv = Atol(PL_tokenbuf);
6987 uv = Atoul(PL_tokenbuf);
6989 floatit = TRUE; /* probably just too large */
6990 else if (*PL_tokenbuf == '-')
6998 value = Atof(PL_tokenbuf);
7000 Perl_die(aTHX_ "unparseable float");
7002 sv_setnv(sv, value);
7005 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7006 (PL_hints & HINT_NEW_INTEGER) )
7007 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7008 (floatit ? "float" : "integer"),
7012 /* if it starts with a v, it could be a v-string */
7018 while (isDIGIT(*pos) || *pos == '_')
7020 if (!isALPHA(*pos)) {
7022 U8 tmpbuf[UTF8_MAXLEN];
7025 s++; /* get past 'v' */
7028 sv_setpvn(sv, "", 0);
7031 if (*s == '0' && isDIGIT(s[1]))
7032 yyerror("Octal number in vector unsupported");
7035 /* this is atoi() that tolerates underscores */
7038 while (--end >= s) {
7043 rev += (*end - '0') * mult;
7045 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
7046 Perl_warner(aTHX_ WARN_OVERFLOW,
7047 "Integer overflow in decimal number");
7050 tmpend = uv_to_utf8(tmpbuf, rev);
7051 utf8 = utf8 || rev > 127;
7052 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
7053 if (*pos == '.' && isDIGIT(pos[1]))
7059 while (isDIGIT(*pos) || *pos == '_')
7067 sv_utf8_downgrade(sv, TRUE);
7074 /* make the op for the constant and return */
7077 yylval.opval = newSVOP(OP_CONST, 0, sv);
7079 yylval.opval = Nullop;
7085 S_scan_formline(pTHX_ register char *s)
7090 SV *stuff = newSVpvn("",0);
7091 bool needargs = FALSE;
7094 if (*s == '.' || *s == '}') {
7096 #ifdef PERL_STRICT_CR
7097 for (t = s+1;*t == ' ' || *t == '\t'; t++) ;
7099 for (t = s+1;*t == ' ' || *t == '\t' || *t == '\r'; t++) ;
7101 if (*t == '\n' || t == PL_bufend)
7104 if (PL_in_eval && !PL_rsfp) {
7105 eol = strchr(s,'\n');
7110 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7112 for (t = s; t < eol; t++) {
7113 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7115 goto enough; /* ~~ must be first line in formline */
7117 if (*t == '@' || *t == '^')
7120 sv_catpvn(stuff, s, eol-s);
7121 #ifndef PERL_STRICT_CR
7122 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7123 char *end = SvPVX(stuff) + SvCUR(stuff);
7132 s = filter_gets(PL_linestr, PL_rsfp, 0);
7133 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7134 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7137 yyerror("Format not terminated");
7147 PL_lex_state = LEX_NORMAL;
7148 PL_nextval[PL_nexttoke].ival = 0;
7152 PL_lex_state = LEX_FORMLINE;
7153 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7155 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7159 SvREFCNT_dec(stuff);
7160 PL_lex_formbrack = 0;
7171 PL_cshlen = strlen(PL_cshname);
7176 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7179 I32 oldsavestack_ix = PL_savestack_ix;
7180 CV* outsidecv = PL_compcv;
7184 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7186 SAVEI32(PL_subline);
7187 save_item(PL_subname);
7190 SAVESPTR(PL_comppad_name);
7191 SAVESPTR(PL_compcv);
7192 SAVEI32(PL_comppad_name_fill);
7193 SAVEI32(PL_min_intro_pending);
7194 SAVEI32(PL_max_intro_pending);
7195 SAVEI32(PL_pad_reset_pending);
7197 PL_compcv = (CV*)NEWSV(1104,0);
7198 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7199 CvFLAGS(PL_compcv) |= flags;
7201 PL_comppad = newAV();
7202 av_push(PL_comppad, Nullsv);
7203 PL_curpad = AvARRAY(PL_comppad);
7204 PL_comppad_name = newAV();
7205 PL_comppad_name_fill = 0;
7206 PL_min_intro_pending = 0;
7208 PL_subline = CopLINE(PL_curcop);
7210 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7211 PL_curpad[0] = (SV*)newAV();
7212 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7213 #endif /* USE_THREADS */
7215 comppadlist = newAV();
7216 AvREAL_off(comppadlist);
7217 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7218 av_store(comppadlist, 1, (SV*)PL_comppad);
7220 CvPADLIST(PL_compcv) = comppadlist;
7221 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7223 CvOWNER(PL_compcv) = 0;
7224 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7225 MUTEX_INIT(CvMUTEXP(PL_compcv));
7226 #endif /* USE_THREADS */
7228 return oldsavestack_ix;
7232 Perl_yywarn(pTHX_ char *s)
7235 PL_in_eval |= EVAL_WARNONLY;
7237 PL_in_eval &= ~EVAL_WARNONLY;
7242 Perl_yyerror(pTHX_ char *s)
7246 char *context = NULL;
7250 if (!yychar || (yychar == ';' && !PL_rsfp))
7252 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7253 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7254 while (isSPACE(*PL_oldoldbufptr))
7256 context = PL_oldoldbufptr;
7257 contlen = PL_bufptr - PL_oldoldbufptr;
7259 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7260 PL_oldbufptr != PL_bufptr) {
7261 while (isSPACE(*PL_oldbufptr))
7263 context = PL_oldbufptr;
7264 contlen = PL_bufptr - PL_oldbufptr;
7266 else if (yychar > 255)
7267 where = "next token ???";
7268 #ifdef USE_PURE_BISON
7269 /* GNU Bison sets the value -2 */
7270 else if (yychar == -2) {
7272 else if ((yychar & 127) == 127) {
7274 if (PL_lex_state == LEX_NORMAL ||
7275 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7276 where = "at end of line";
7277 else if (PL_lex_inpat)
7278 where = "within pattern";
7280 where = "within string";
7283 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7285 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7286 else if (isPRINT_LC(yychar))
7287 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7289 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7290 where = SvPVX(where_sv);
7292 msg = sv_2mortal(newSVpv(s, 0));
7293 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7294 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7296 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7298 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7299 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7300 Perl_sv_catpvf(aTHX_ msg,
7301 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7302 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7305 if (PL_in_eval & EVAL_WARNONLY)
7306 Perl_warn(aTHX_ "%"SVf, msg);
7309 if (PL_error_count >= 10)
7310 Perl_croak(aTHX_ "%s has too many errors.\n", CopFILE(PL_curcop));
7312 PL_in_my_stash = Nullhv;
7323 * Restore a source filter.
7327 restore_rsfp(pTHXo_ void *f)
7329 PerlIO *fp = (PerlIO*)f;
7331 if (PL_rsfp == PerlIO_stdin())
7332 PerlIO_clearerr(PL_rsfp);
7333 else if (PL_rsfp && (PL_rsfp != fp))
7334 PerlIO_close(PL_rsfp);