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)
323 S_utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
325 I32 count = FILTER_READ(idx+1, sv, maxlen);
329 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
330 tend = utf16_to_utf8((U16*)SvPVX(sv), tmps, SvCUR(sv));
331 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 :
817 * Forces the next token to be a version number.
821 S_force_version(pTHX_ char *s)
823 OP *version = Nullop;
824 bool is_vstr = FALSE;
835 for (; isDIGIT(*d) || *d == '_' || *d == '.'; d++);
836 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
838 /* real VERSION number -- GBARR */
839 version = yylval.opval;
841 SV *ver = cSVOPx(version)->op_sv;
842 SvUPGRADE(ver, SVt_PVIV);
843 SvIOKp_on(ver); /* hint that it is a version */
848 /* NOTE: The parser sees the package name and the VERSION swapped */
849 PL_nextval[PL_nexttoke].opval = version;
857 * Tokenize a quoted string passed in as an SV. It finds the next
858 * chunk, up to end of string or a backslash. It may make a new
859 * SV containing that chunk (if HINT_NEW_STRING is on). It also
864 S_tokeq(pTHX_ SV *sv)
875 s = SvPV_force(sv, len);
879 while (s < send && *s != '\\')
884 if ( PL_hints & HINT_NEW_STRING )
885 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
888 if (s + 1 < send && (s[1] == '\\'))
889 s++; /* all that, just for this */
894 SvCUR_set(sv, d - SvPVX(sv));
896 if ( PL_hints & HINT_NEW_STRING )
897 return new_constant(NULL, 0, "q", sv, pv, "q");
902 * Now come three functions related to double-quote context,
903 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
904 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
905 * interact with PL_lex_state, and create fake ( ... ) argument lists
906 * to handle functions and concatenation.
907 * They assume that whoever calls them will be setting up a fake
908 * join call, because each subthing puts a ',' after it. This lets
911 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
913 * (I'm not sure whether the spurious commas at the end of lcfirst's
914 * arguments and join's arguments are created or not).
919 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
921 * Pattern matching will set PL_lex_op to the pattern-matching op to
922 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
924 * OP_CONST and OP_READLINE are easy--just make the new op and return.
926 * Everything else becomes a FUNC.
928 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
929 * had an OP_CONST or OP_READLINE). This just sets us up for a
930 * call to S_sublex_push().
936 register I32 op_type = yylval.ival;
938 if (op_type == OP_NULL) {
939 yylval.opval = PL_lex_op;
943 if (op_type == OP_CONST || op_type == OP_READLINE) {
944 SV *sv = tokeq(PL_lex_stuff);
946 if (SvTYPE(sv) == SVt_PVIV) {
947 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
953 nsv = newSVpvn(p, len);
957 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
958 PL_lex_stuff = Nullsv;
962 PL_sublex_info.super_state = PL_lex_state;
963 PL_sublex_info.sub_inwhat = op_type;
964 PL_sublex_info.sub_op = PL_lex_op;
965 PL_lex_state = LEX_INTERPPUSH;
969 yylval.opval = PL_lex_op;
979 * Create a new scope to save the lexing state. The scope will be
980 * ended in S_sublex_done. Returns a '(', starting the function arguments
981 * to the uc, lc, etc. found before.
982 * Sets PL_lex_state to LEX_INTERPCONCAT.
991 PL_lex_state = PL_sublex_info.super_state;
992 SAVEI32(PL_lex_dojoin);
993 SAVEI32(PL_lex_brackets);
994 SAVEI32(PL_lex_casemods);
995 SAVEI32(PL_lex_starts);
996 SAVEI32(PL_lex_state);
997 SAVEVPTR(PL_lex_inpat);
998 SAVEI32(PL_lex_inwhat);
999 SAVECOPLINE(PL_curcop);
1000 SAVEPPTR(PL_bufptr);
1001 SAVEPPTR(PL_oldbufptr);
1002 SAVEPPTR(PL_oldoldbufptr);
1003 SAVEPPTR(PL_linestart);
1004 SAVESPTR(PL_linestr);
1005 SAVEPPTR(PL_lex_brackstack);
1006 SAVEPPTR(PL_lex_casestack);
1008 PL_linestr = PL_lex_stuff;
1009 PL_lex_stuff = Nullsv;
1011 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1012 = SvPVX(PL_linestr);
1013 PL_bufend += SvCUR(PL_linestr);
1014 SAVEFREESV(PL_linestr);
1016 PL_lex_dojoin = FALSE;
1017 PL_lex_brackets = 0;
1018 New(899, PL_lex_brackstack, 120, char);
1019 New(899, PL_lex_casestack, 12, char);
1020 SAVEFREEPV(PL_lex_brackstack);
1021 SAVEFREEPV(PL_lex_casestack);
1022 PL_lex_casemods = 0;
1023 *PL_lex_casestack = '\0';
1025 PL_lex_state = LEX_INTERPCONCAT;
1026 CopLINE_set(PL_curcop, PL_multi_start);
1028 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1029 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1030 PL_lex_inpat = PL_sublex_info.sub_op;
1032 PL_lex_inpat = Nullop;
1039 * Restores lexer state after a S_sublex_push.
1045 if (!PL_lex_starts++) {
1046 PL_expect = XOPERATOR;
1047 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn("",0));
1051 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1052 PL_lex_state = LEX_INTERPCASEMOD;
1056 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1057 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1058 PL_linestr = PL_lex_repl;
1060 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1061 PL_bufend += SvCUR(PL_linestr);
1062 SAVEFREESV(PL_linestr);
1063 PL_lex_dojoin = FALSE;
1064 PL_lex_brackets = 0;
1065 PL_lex_casemods = 0;
1066 *PL_lex_casestack = '\0';
1068 if (SvEVALED(PL_lex_repl)) {
1069 PL_lex_state = LEX_INTERPNORMAL;
1071 /* we don't clear PL_lex_repl here, so that we can check later
1072 whether this is an evalled subst; that means we rely on the
1073 logic to ensure sublex_done() is called again only via the
1074 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1077 PL_lex_state = LEX_INTERPCONCAT;
1078 PL_lex_repl = Nullsv;
1084 PL_bufend = SvPVX(PL_linestr);
1085 PL_bufend += SvCUR(PL_linestr);
1086 PL_expect = XOPERATOR;
1087 PL_sublex_info.sub_inwhat = 0;
1095 Extracts a pattern, double-quoted string, or transliteration. This
1098 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1099 processing a pattern (PL_lex_inpat is true), a transliteration
1100 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1102 Returns a pointer to the character scanned up to. Iff this is
1103 advanced from the start pointer supplied (ie if anything was
1104 successfully parsed), will leave an OP for the substring scanned
1105 in yylval. Caller must intuit reason for not parsing further
1106 by looking at the next characters herself.
1110 double-quoted style: \r and \n
1111 regexp special ones: \D \s
1113 backrefs: \1 (deprecated in substitution replacements)
1114 case and quoting: \U \Q \E
1115 stops on @ and $, but not for $ as tail anchor
1117 In transliterations:
1118 characters are VERY literal, except for - not at the start or end
1119 of the string, which indicates a range. scan_const expands the
1120 range to the full set of intermediate characters.
1122 In double-quoted strings:
1124 double-quoted style: \r and \n
1126 backrefs: \1 (deprecated)
1127 case and quoting: \U \Q \E
1130 scan_const does *not* construct ops to handle interpolated strings.
1131 It stops processing as soon as it finds an embedded $ or @ variable
1132 and leaves it to the caller to work out what's going on.
1134 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1136 $ in pattern could be $foo or could be tail anchor. Assumption:
1137 it's a tail anchor if $ is the last thing in the string, or if it's
1138 followed by one of ")| \n\t"
1140 \1 (backreferences) are turned into $1
1142 The structure of the code is
1143 while (there's a character to process) {
1144 handle transliteration ranges
1145 skip regexp comments
1146 skip # initiated comments in //x patterns
1147 check for embedded @foo
1148 check for embedded scalars
1150 leave intact backslashes from leave (below)
1151 deprecate \1 in strings and sub replacements
1152 handle string-changing backslashes \l \U \Q \E, etc.
1153 switch (what was escaped) {
1154 handle - in a transliteration (becomes a literal -)
1155 handle \132 octal characters
1156 handle 0x15 hex characters
1157 handle \cV (control V)
1158 handle printf backslashes (\f, \r, \n, etc)
1160 } (end if backslash)
1161 } (end while character to read)
1166 S_scan_const(pTHX_ char *start)
1168 register char *send = PL_bufend; /* end of the constant */
1169 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1170 register char *s = start; /* start of the constant */
1171 register char *d = SvPVX(sv); /* destination for copies */
1172 bool dorange = FALSE; /* are we in a translit range? */
1173 bool has_utf = FALSE; /* embedded \x{} */
1175 I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1176 ? (PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1178 I32 thisutf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1179 ? (PL_sublex_info.sub_op->op_private & (PL_lex_repl ?
1180 OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF))
1182 const char *leaveit = /* set of acceptably-backslashed characters */
1184 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1187 while (s < send || dorange) {
1188 /* get transliterations out of the way (they're most literal) */
1189 if (PL_lex_inwhat == OP_TRANS) {
1190 /* expand a range A-Z to the full set of characters. AIE! */
1192 I32 i; /* current expanded character */
1193 I32 min; /* first character in range */
1194 I32 max; /* last character in range */
1196 i = d - SvPVX(sv); /* remember current offset */
1197 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1198 d = SvPVX(sv) + i; /* refresh d after realloc */
1199 d -= 2; /* eat the first char and the - */
1201 min = (U8)*d; /* first char in range */
1202 max = (U8)d[1]; /* last char in range */
1205 if ((isLOWER(min) && isLOWER(max)) ||
1206 (isUPPER(min) && isUPPER(max))) {
1208 for (i = min; i <= max; i++)
1212 for (i = min; i <= max; i++)
1219 for (i = min; i <= max; i++)
1222 /* mark the range as done, and continue */
1227 /* range begins (ignore - as first or last char) */
1228 else if (*s == '-' && s+1 < send && s != start) {
1230 *d++ = (char)0xff; /* use illegal utf8 byte--see pmtrans */
1239 /* if we get here, we're not doing a transliteration */
1241 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1242 except for the last char, which will be done separately. */
1243 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1245 while (s < send && *s != ')')
1247 } else if (s[2] == '{'
1248 || s[2] == 'p' && s[3] == '{') { /* This should march regcomp.c */
1250 char *regparse = s + (s[2] == '{' ? 3 : 4);
1253 while (count && (c = *regparse)) {
1254 if (c == '\\' && regparse[1])
1262 if (*regparse != ')') {
1263 regparse--; /* Leave one char for continuation. */
1264 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1266 while (s < regparse)
1271 /* likewise skip #-initiated comments in //x patterns */
1272 else if (*s == '#' && PL_lex_inpat &&
1273 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1274 while (s+1 < send && *s != '\n')
1278 /* check for embedded arrays (@foo, @:foo, @'foo, @{foo}, @$foo) */
1279 else if (*s == '@' && s[1]
1280 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$", s[1])))
1283 /* check for embedded scalars. only stop if we're sure it's a
1286 else if (*s == '$') {
1287 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1289 if (s + 1 < send && !strchr("()| \n\t", s[1]))
1290 break; /* in regexp, $ might be tail anchor */
1293 /* (now in tr/// code again) */
1295 if (*s & 0x80 && thisutf) {
1296 dTHR; /* only for ckWARN */
1297 if (ckWARN(WARN_UTF8)) {
1298 (void)utf8_to_uv((U8*)s, &len); /* could cvt latin-1 to utf8 here... */
1307 has_utf = TRUE; /* assume valid utf8 */
1311 if (*s == '\\' && s+1 < send) {
1314 /* some backslashes we leave behind */
1315 if (*leaveit && *s && strchr(leaveit, *s)) {
1321 /* deprecate \1 in strings and substitution replacements */
1322 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1323 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1325 dTHR; /* only for ckWARN */
1326 if (ckWARN(WARN_SYNTAX))
1327 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1332 /* string-change backslash escapes */
1333 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1338 /* if we get here, it's either a quoted -, or a digit */
1341 /* quoted - in transliterations */
1343 if (PL_lex_inwhat == OP_TRANS) {
1351 if (ckWARN(WARN_MISC) && isALPHA(*s))
1352 Perl_warner(aTHX_ WARN_MISC,
1353 "Unrecognized escape \\%c passed through",
1355 /* default action is to copy the quoted character */
1360 /* \132 indicates an octal constant */
1361 case '0': case '1': case '2': case '3':
1362 case '4': case '5': case '6': case '7':
1363 *d++ = (char)scan_oct(s, 3, &len);
1367 /* \x24 indicates a hex constant */
1371 char* e = strchr(s, '}');
1375 yyerror("Missing right brace on \\x{}");
1378 /* note: utf always shorter than hex */
1379 uv = (UV)scan_hex(s + 1, e - s - 1, &len);
1381 d = (char*)uv_to_utf8((U8*)d, uv);
1389 /* XXX collapse this branch into the one above */
1390 UV uv = (UV)scan_hex(s, 2, &len);
1391 if (utf && PL_lex_inwhat == OP_TRANS &&
1392 utf != (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1394 d = (char*)uv_to_utf8((U8*)d, uv); /* doing a CU or UC */
1398 if (uv >= 127 && UTF) {
1400 if (ckWARN(WARN_UTF8))
1401 Perl_warner(aTHX_ WARN_UTF8,
1402 "\\x%.*s will produce malformed UTF-8 character; use \\x{%.*s} for that",
1403 (int)len,s,(int)len,s);
1411 /* \N{latin small letter a} is a named character */
1415 char* e = strchr(s, '}');
1424 yyerror("Missing right brace on \\N{}");
1428 res = newSVpvn(s + 1, e - s - 1);
1429 res = new_constant( Nullch, 0, "charnames",
1430 res, Nullsv, "\\N{...}" );
1431 str = SvPV(res,len);
1432 if (len > e - s + 4) {
1433 char *odest = SvPVX(sv);
1435 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1436 d = SvPVX(sv) + (d - odest);
1438 Copy(str, d, len, char);
1445 yyerror("Missing braces on \\N{}");
1448 /* \c is a control character */
1462 /* printf-style backslashes, formfeeds, newlines, etc */
1480 *d++ = '\047'; /* CP 1047 */
1483 *d++ = '\057'; /* CP 1047 */
1497 } /* end if (backslash) */
1500 } /* while loop to process each character */
1502 /* terminate the string and set up the sv */
1504 SvCUR_set(sv, d - SvPVX(sv));
1509 /* shrink the sv if we allocated more than we used */
1510 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1511 SvLEN_set(sv, SvCUR(sv) + 1);
1512 Renew(SvPVX(sv), SvLEN(sv), char);
1515 /* return the substring (via yylval) only if we parsed anything */
1516 if (s > PL_bufptr) {
1517 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1518 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1520 ( PL_lex_inwhat == OP_TRANS
1522 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1525 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1532 * Returns TRUE if there's more to the expression (e.g., a subscript),
1535 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1537 * ->[ and ->{ return TRUE
1538 * { and [ outside a pattern are always subscripts, so return TRUE
1539 * if we're outside a pattern and it's not { or [, then return FALSE
1540 * if we're in a pattern and the first char is a {
1541 * {4,5} (any digits around the comma) returns FALSE
1542 * if we're in a pattern and the first char is a [
1544 * [SOMETHING] has a funky algorithm to decide whether it's a
1545 * character class or not. It has to deal with things like
1546 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1547 * anything else returns TRUE
1550 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1553 S_intuit_more(pTHX_ register char *s)
1555 if (PL_lex_brackets)
1557 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1559 if (*s != '{' && *s != '[')
1564 /* In a pattern, so maybe we have {n,m}. */
1581 /* On the other hand, maybe we have a character class */
1584 if (*s == ']' || *s == '^')
1587 /* this is terrifying, and it works */
1588 int weight = 2; /* let's weigh the evidence */
1590 unsigned char un_char = 255, last_un_char;
1591 char *send = strchr(s,']');
1592 char tmpbuf[sizeof PL_tokenbuf * 4];
1594 if (!send) /* has to be an expression */
1597 Zero(seen,256,char);
1600 else if (isDIGIT(*s)) {
1602 if (isDIGIT(s[1]) && s[2] == ']')
1608 for (; s < send; s++) {
1609 last_un_char = un_char;
1610 un_char = (unsigned char)*s;
1615 weight -= seen[un_char] * 10;
1616 if (isALNUM_lazy_if(s+1,UTF)) {
1617 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1618 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1623 else if (*s == '$' && s[1] &&
1624 strchr("[#!%*<>()-=",s[1])) {
1625 if (/*{*/ strchr("])} =",s[2]))
1634 if (strchr("wds]",s[1]))
1636 else if (seen['\''] || seen['"'])
1638 else if (strchr("rnftbxcav",s[1]))
1640 else if (isDIGIT(s[1])) {
1642 while (s[1] && isDIGIT(s[1]))
1652 if (strchr("aA01! ",last_un_char))
1654 if (strchr("zZ79~",s[1]))
1656 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1657 weight -= 5; /* cope with negative subscript */
1660 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1661 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1666 if (keyword(tmpbuf, d - tmpbuf))
1669 if (un_char == last_un_char + 1)
1671 weight -= seen[un_char];
1676 if (weight >= 0) /* probably a character class */
1686 * Does all the checking to disambiguate
1688 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1689 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1691 * First argument is the stuff after the first token, e.g. "bar".
1693 * Not a method if bar is a filehandle.
1694 * Not a method if foo is a subroutine prototyped to take a filehandle.
1695 * Not a method if it's really "Foo $bar"
1696 * Method if it's "foo $bar"
1697 * Not a method if it's really "print foo $bar"
1698 * Method if it's really "foo package::" (interpreted as package->foo)
1699 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1700 * Not a method if bar is a filehandle or package, but is quoted with
1705 S_intuit_method(pTHX_ char *start, GV *gv)
1707 char *s = start + (*start == '$');
1708 char tmpbuf[sizeof PL_tokenbuf];
1716 if ((cv = GvCVu(gv))) {
1717 char *proto = SvPVX(cv);
1727 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1728 /* start is the beginning of the possible filehandle/object,
1729 * and s is the end of it
1730 * tmpbuf is a copy of it
1733 if (*start == '$') {
1734 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1739 return *s == '(' ? FUNCMETH : METHOD;
1741 if (!keyword(tmpbuf, len)) {
1742 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1747 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1748 if (indirgv && GvCVu(indirgv))
1750 /* filehandle or package name makes it a method */
1751 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1753 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1754 return 0; /* no assumptions -- "=>" quotes bearword */
1756 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1757 newSVpvn(tmpbuf,len));
1758 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1762 return *s == '(' ? FUNCMETH : METHOD;
1770 * Return a string of Perl code to load the debugger. If PERL5DB
1771 * is set, it will return the contents of that, otherwise a
1772 * compile-time require of perl5db.pl.
1779 char *pdb = PerlEnv_getenv("PERL5DB");
1783 SETERRNO(0,SS$_NORMAL);
1784 return "BEGIN { require 'perl5db.pl' }";
1790 /* Encoded script support. filter_add() effectively inserts a
1791 * 'pre-processing' function into the current source input stream.
1792 * Note that the filter function only applies to the current source file
1793 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1795 * The datasv parameter (which may be NULL) can be used to pass
1796 * private data to this instance of the filter. The filter function
1797 * can recover the SV using the FILTER_DATA macro and use it to
1798 * store private buffers and state information.
1800 * The supplied datasv parameter is upgraded to a PVIO type
1801 * and the IoDIRP field is used to store the function pointer,
1802 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1803 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1804 * private use must be set using malloc'd pointers.
1808 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1813 if (!PL_rsfp_filters)
1814 PL_rsfp_filters = newAV();
1816 datasv = NEWSV(255,0);
1817 if (!SvUPGRADE(datasv, SVt_PVIO))
1818 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1819 IoDIRP(datasv) = (DIR*)funcp; /* stash funcp into spare field */
1820 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1821 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1822 funcp, SvPV_nolen(datasv)));
1823 av_unshift(PL_rsfp_filters, 1);
1824 av_store(PL_rsfp_filters, 0, datasv) ;
1829 /* Delete most recently added instance of this filter function. */
1831 Perl_filter_del(pTHX_ filter_t funcp)
1834 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1835 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1837 /* if filter is on top of stack (usual case) just pop it off */
1838 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1839 if (IoDIRP(datasv) == (DIR*)funcp) {
1840 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1841 IoDIRP(datasv) = (DIR*)NULL;
1842 sv_free(av_pop(PL_rsfp_filters));
1846 /* we need to search for the correct entry and clear it */
1847 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1851 /* Invoke the n'th filter function for the current rsfp. */
1853 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1856 /* 0 = read one text line */
1861 if (!PL_rsfp_filters)
1863 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1864 /* Provide a default input filter to make life easy. */
1865 /* Note that we append to the line. This is handy. */
1866 DEBUG_P(PerlIO_printf(Perl_debug_log,
1867 "filter_read %d: from rsfp\n", idx));
1871 int old_len = SvCUR(buf_sv) ;
1873 /* ensure buf_sv is large enough */
1874 SvGROW(buf_sv, old_len + maxlen) ;
1875 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1876 if (PerlIO_error(PL_rsfp))
1877 return -1; /* error */
1879 return 0 ; /* end of file */
1881 SvCUR_set(buf_sv, old_len + len) ;
1884 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1885 if (PerlIO_error(PL_rsfp))
1886 return -1; /* error */
1888 return 0 ; /* end of file */
1891 return SvCUR(buf_sv);
1893 /* Skip this filter slot if filter has been deleted */
1894 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1895 DEBUG_P(PerlIO_printf(Perl_debug_log,
1896 "filter_read %d: skipped (filter deleted)\n",
1898 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1900 /* Get function pointer hidden within datasv */
1901 funcp = (filter_t)IoDIRP(datasv);
1902 DEBUG_P(PerlIO_printf(Perl_debug_log,
1903 "filter_read %d: via function %p (%s)\n",
1904 idx, funcp, SvPV_nolen(datasv)));
1905 /* Call function. The function is expected to */
1906 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1907 /* Return: <0:error, =0:eof, >0:not eof */
1908 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1912 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1914 #ifdef PERL_CR_FILTER
1915 if (!PL_rsfp_filters) {
1916 filter_add(S_cr_textfilter,NULL);
1919 if (PL_rsfp_filters) {
1922 SvCUR_set(sv, 0); /* start with empty line */
1923 if (FILTER_READ(0, sv, 0) > 0)
1924 return ( SvPVX(sv) ) ;
1929 return (sv_gets(sv, fp, append));
1934 static char* exp_name[] =
1935 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
1936 "ATTRTERM", "TERMBLOCK"
1943 Works out what to call the token just pulled out of the input
1944 stream. The yacc parser takes care of taking the ops we return and
1945 stitching them into a tree.
1951 if read an identifier
1952 if we're in a my declaration
1953 croak if they tried to say my($foo::bar)
1954 build the ops for a my() declaration
1955 if it's an access to a my() variable
1956 are we in a sort block?
1957 croak if my($a); $a <=> $b
1958 build ops for access to a my() variable
1959 if in a dq string, and they've said @foo and we can't find @foo
1961 build ops for a bareword
1962 if we already built the token before, use it.
1966 #ifdef USE_PURE_BISON
1967 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
1980 #ifdef USE_PURE_BISON
1981 yylval_pointer = lvalp;
1982 yychar_pointer = lcharp;
1985 /* check if there's an identifier for us to look at */
1986 if (PL_pending_ident) {
1987 /* pit holds the identifier we read and pending_ident is reset */
1988 char pit = PL_pending_ident;
1989 PL_pending_ident = 0;
1991 /* if we're in a my(), we can't allow dynamics here.
1992 $foo'bar has already been turned into $foo::bar, so
1993 just check for colons.
1995 if it's a legal name, the OP is a PADANY.
1998 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
1999 if (strchr(PL_tokenbuf,':'))
2000 yyerror(Perl_form(aTHX_ "No package name allowed for "
2001 "variable %s in \"our\"",
2003 tmp = pad_allocmy(PL_tokenbuf);
2006 if (strchr(PL_tokenbuf,':'))
2007 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
2009 yylval.opval = newOP(OP_PADANY, 0);
2010 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
2016 build the ops for accesses to a my() variable.
2018 Deny my($a) or my($b) in a sort block, *if* $a or $b is
2019 then used in a comparison. This catches most, but not
2020 all cases. For instance, it catches
2021 sort { my($a); $a <=> $b }
2023 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2024 (although why you'd do that is anyone's guess).
2027 if (!strchr(PL_tokenbuf,':')) {
2029 /* Check for single character per-thread SVs */
2030 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2031 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2032 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2034 yylval.opval = newOP(OP_THREADSV, 0);
2035 yylval.opval->op_targ = tmp;
2038 #endif /* USE_THREADS */
2039 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2040 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2041 /* might be an "our" variable" */
2042 if (SvFLAGS(namesv) & SVpad_OUR) {
2043 /* build ops for a bareword */
2044 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2045 sv_catpvn(sym, "::", 2);
2046 sv_catpv(sym, PL_tokenbuf+1);
2047 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2048 yylval.opval->op_private = OPpCONST_ENTERED;
2049 gv_fetchpv(SvPVX(sym),
2051 ? (GV_ADDMULTI | GV_ADDINEVAL)
2054 ((PL_tokenbuf[0] == '$') ? SVt_PV
2055 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2060 /* if it's a sort block and they're naming $a or $b */
2061 if (PL_last_lop_op == OP_SORT &&
2062 PL_tokenbuf[0] == '$' &&
2063 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2066 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2067 d < PL_bufend && *d != '\n';
2070 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2071 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2077 yylval.opval = newOP(OP_PADANY, 0);
2078 yylval.opval->op_targ = tmp;
2084 Whine if they've said @foo in a doublequoted string,
2085 and @foo isn't a variable we can find in the symbol
2088 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2089 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2090 if (!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2091 yyerror(Perl_form(aTHX_ "In string, %s now must be written as \\%s",
2092 PL_tokenbuf, PL_tokenbuf));
2095 /* build ops for a bareword */
2096 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2097 yylval.opval->op_private = OPpCONST_ENTERED;
2098 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2099 ((PL_tokenbuf[0] == '$') ? SVt_PV
2100 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2105 /* no identifier pending identification */
2107 switch (PL_lex_state) {
2109 case LEX_NORMAL: /* Some compilers will produce faster */
2110 case LEX_INTERPNORMAL: /* code if we comment these out. */
2114 /* when we've already built the next token, just pull it out of the queue */
2117 yylval = PL_nextval[PL_nexttoke];
2119 PL_lex_state = PL_lex_defer;
2120 PL_expect = PL_lex_expect;
2121 PL_lex_defer = LEX_NORMAL;
2123 return(PL_nexttype[PL_nexttoke]);
2125 /* interpolated case modifiers like \L \U, including \Q and \E.
2126 when we get here, PL_bufptr is at the \
2128 case LEX_INTERPCASEMOD:
2130 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2131 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2133 /* handle \E or end of string */
2134 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2138 if (PL_lex_casemods) {
2139 oldmod = PL_lex_casestack[--PL_lex_casemods];
2140 PL_lex_casestack[PL_lex_casemods] = '\0';
2142 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2144 PL_lex_state = LEX_INTERPCONCAT;
2148 if (PL_bufptr != PL_bufend)
2150 PL_lex_state = LEX_INTERPCONCAT;
2155 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2156 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2157 if (strchr("LU", *s) &&
2158 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2160 PL_lex_casestack[--PL_lex_casemods] = '\0';
2163 if (PL_lex_casemods > 10) {
2164 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2165 if (newlb != PL_lex_casestack) {
2167 PL_lex_casestack = newlb;
2170 PL_lex_casestack[PL_lex_casemods++] = *s;
2171 PL_lex_casestack[PL_lex_casemods] = '\0';
2172 PL_lex_state = LEX_INTERPCONCAT;
2173 PL_nextval[PL_nexttoke].ival = 0;
2176 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2178 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2180 PL_nextval[PL_nexttoke].ival = OP_LC;
2182 PL_nextval[PL_nexttoke].ival = OP_UC;
2184 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2186 Perl_croak(aTHX_ "panic: yylex");
2189 if (PL_lex_starts) {
2198 case LEX_INTERPPUSH:
2199 return sublex_push();
2201 case LEX_INTERPSTART:
2202 if (PL_bufptr == PL_bufend)
2203 return sublex_done();
2205 PL_lex_dojoin = (*PL_bufptr == '@');
2206 PL_lex_state = LEX_INTERPNORMAL;
2207 if (PL_lex_dojoin) {
2208 PL_nextval[PL_nexttoke].ival = 0;
2211 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2212 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2213 force_next(PRIVATEREF);
2215 force_ident("\"", '$');
2216 #endif /* USE_THREADS */
2217 PL_nextval[PL_nexttoke].ival = 0;
2219 PL_nextval[PL_nexttoke].ival = 0;
2221 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2224 if (PL_lex_starts++) {
2230 case LEX_INTERPENDMAYBE:
2231 if (intuit_more(PL_bufptr)) {
2232 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2238 if (PL_lex_dojoin) {
2239 PL_lex_dojoin = FALSE;
2240 PL_lex_state = LEX_INTERPCONCAT;
2243 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2244 && SvEVALED(PL_lex_repl))
2246 if (PL_bufptr != PL_bufend)
2247 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2248 PL_lex_repl = Nullsv;
2251 case LEX_INTERPCONCAT:
2253 if (PL_lex_brackets)
2254 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2256 if (PL_bufptr == PL_bufend)
2257 return sublex_done();
2259 if (SvIVX(PL_linestr) == '\'') {
2260 SV *sv = newSVsv(PL_linestr);
2263 else if ( PL_hints & HINT_NEW_RE )
2264 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2265 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2269 s = scan_const(PL_bufptr);
2271 PL_lex_state = LEX_INTERPCASEMOD;
2273 PL_lex_state = LEX_INTERPSTART;
2276 if (s != PL_bufptr) {
2277 PL_nextval[PL_nexttoke] = yylval;
2280 if (PL_lex_starts++)
2290 PL_lex_state = LEX_NORMAL;
2291 s = scan_formline(PL_bufptr);
2292 if (!PL_lex_formbrack)
2298 PL_oldoldbufptr = PL_oldbufptr;
2301 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2302 exp_name[PL_expect], s);
2308 if (isIDFIRST_lazy_if(s,UTF))
2310 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2313 goto fake_eof; /* emulate EOF on ^D or ^Z */
2318 if (PL_lex_brackets)
2319 yyerror("Missing right curly or square bracket");
2322 if (s++ < PL_bufend)
2323 goto retry; /* ignore stray nulls */
2326 if (!PL_in_eval && !PL_preambled) {
2327 PL_preambled = TRUE;
2328 sv_setpv(PL_linestr,incl_perldb());
2329 if (SvCUR(PL_linestr))
2330 sv_catpv(PL_linestr,";");
2332 while(AvFILLp(PL_preambleav) >= 0) {
2333 SV *tmpsv = av_shift(PL_preambleav);
2334 sv_catsv(PL_linestr, tmpsv);
2335 sv_catpv(PL_linestr, ";");
2338 sv_free((SV*)PL_preambleav);
2339 PL_preambleav = NULL;
2341 if (PL_minus_n || PL_minus_p) {
2342 sv_catpv(PL_linestr, "LINE: while (<>) {");
2344 sv_catpv(PL_linestr,"chomp;");
2346 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2348 GvIMPORTED_AV_on(gv);
2350 if (strchr("/'\"", *PL_splitstr)
2351 && strchr(PL_splitstr + 1, *PL_splitstr))
2352 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2355 s = "'~#\200\1'"; /* surely one char is unused...*/
2356 while (s[1] && strchr(PL_splitstr, *s)) s++;
2358 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2359 "q" + (delim == '\''), delim);
2360 for (s = PL_splitstr; *s; s++) {
2362 sv_catpvn(PL_linestr, "\\", 1);
2363 sv_catpvn(PL_linestr, s, 1);
2365 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2369 sv_catpv(PL_linestr,"@F=split(' ');");
2372 sv_catpv(PL_linestr, "\n");
2373 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2374 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2375 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2376 SV *sv = NEWSV(85,0);
2378 sv_upgrade(sv, SVt_PVMG);
2379 sv_setsv(sv,PL_linestr);
2380 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2385 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2388 if (PL_preprocess && !PL_in_eval)
2389 (void)PerlProc_pclose(PL_rsfp);
2390 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2391 PerlIO_clearerr(PL_rsfp);
2393 (void)PerlIO_close(PL_rsfp);
2395 PL_doextract = FALSE;
2397 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2398 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2399 sv_catpv(PL_linestr,";}");
2400 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2401 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2402 PL_minus_n = PL_minus_p = 0;
2405 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2406 sv_setpv(PL_linestr,"");
2407 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2410 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2411 PL_doextract = FALSE;
2413 /* Incest with pod. */
2414 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2415 sv_setpv(PL_linestr, "");
2416 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2417 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2418 PL_doextract = FALSE;
2422 } while (PL_doextract);
2423 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2424 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2425 SV *sv = NEWSV(85,0);
2427 sv_upgrade(sv, SVt_PVMG);
2428 sv_setsv(sv,PL_linestr);
2429 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2431 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2432 if (CopLINE(PL_curcop) == 1) {
2433 while (s < PL_bufend && isSPACE(*s))
2435 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2439 if (*s == '#' && *(s+1) == '!')
2441 #ifdef ALTERNATE_SHEBANG
2443 static char as[] = ALTERNATE_SHEBANG;
2444 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2445 d = s + (sizeof(as) - 1);
2447 #endif /* ALTERNATE_SHEBANG */
2456 while (*d && !isSPACE(*d))
2460 #ifdef ARG_ZERO_IS_SCRIPT
2461 if (ipathend > ipath) {
2463 * HP-UX (at least) sets argv[0] to the script name,
2464 * which makes $^X incorrect. And Digital UNIX and Linux,
2465 * at least, set argv[0] to the basename of the Perl
2466 * interpreter. So, having found "#!", we'll set it right.
2468 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2469 assert(SvPOK(x) || SvGMAGICAL(x));
2470 if (sv_eq(x, CopFILESV(PL_curcop))) {
2471 sv_setpvn(x, ipath, ipathend - ipath);
2474 TAINT_NOT; /* $^X is always tainted, but that's OK */
2476 #endif /* ARG_ZERO_IS_SCRIPT */
2481 d = instr(s,"perl -");
2483 d = instr(s,"perl");
2485 /* avoid getting into infinite loops when shebang
2486 * line contains "Perl" rather than "perl" */
2488 for (d = ipathend-4; d >= ipath; --d) {
2489 if ((*d == 'p' || *d == 'P')
2490 && !ibcmp(d, "perl", 4))
2500 #ifdef ALTERNATE_SHEBANG
2502 * If the ALTERNATE_SHEBANG on this system starts with a
2503 * character that can be part of a Perl expression, then if
2504 * we see it but not "perl", we're probably looking at the
2505 * start of Perl code, not a request to hand off to some
2506 * other interpreter. Similarly, if "perl" is there, but
2507 * not in the first 'word' of the line, we assume the line
2508 * contains the start of the Perl program.
2510 if (d && *s != '#') {
2512 while (*c && !strchr("; \t\r\n\f\v#", *c))
2515 d = Nullch; /* "perl" not in first word; ignore */
2517 *s = '#'; /* Don't try to parse shebang line */
2519 #endif /* ALTERNATE_SHEBANG */
2524 !instr(s,"indir") &&
2525 instr(PL_origargv[0],"perl"))
2531 while (s < PL_bufend && isSPACE(*s))
2533 if (s < PL_bufend) {
2534 Newz(899,newargv,PL_origargc+3,char*);
2536 while (s < PL_bufend && !isSPACE(*s))
2539 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2542 newargv = PL_origargv;
2544 PerlProc_execv(ipath, newargv);
2545 Perl_croak(aTHX_ "Can't exec %s", ipath);
2548 U32 oldpdb = PL_perldb;
2549 bool oldn = PL_minus_n;
2550 bool oldp = PL_minus_p;
2552 while (*d && !isSPACE(*d)) d++;
2553 while (*d == ' ' || *d == '\t') d++;
2557 if (*d == 'M' || *d == 'm') {
2559 while (*d && !isSPACE(*d)) d++;
2560 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2563 d = moreswitches(d);
2565 if (PERLDB_LINE && !oldpdb ||
2566 ( PL_minus_n || PL_minus_p ) && !(oldn || oldp) )
2567 /* if we have already added "LINE: while (<>) {",
2568 we must not do it again */
2570 sv_setpv(PL_linestr, "");
2571 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2572 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2573 PL_preambled = FALSE;
2575 (void)gv_fetchfile(PL_origfilename);
2582 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2584 PL_lex_state = LEX_FORMLINE;
2589 #ifdef PERL_STRICT_CR
2590 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2592 "(Maybe you didn't strip carriage returns after a network transfer?)\n");
2594 case ' ': case '\t': case '\f': case 013:
2599 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2601 while (s < d && *s != '\n')
2606 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2608 PL_lex_state = LEX_FORMLINE;
2618 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2623 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2626 if (strnEQ(s,"=>",2)) {
2627 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2628 OPERATOR('-'); /* unary minus */
2630 PL_last_uni = PL_oldbufptr;
2631 PL_last_lop_op = OP_FTEREAD; /* good enough */
2633 case 'r': FTST(OP_FTEREAD);
2634 case 'w': FTST(OP_FTEWRITE);
2635 case 'x': FTST(OP_FTEEXEC);
2636 case 'o': FTST(OP_FTEOWNED);
2637 case 'R': FTST(OP_FTRREAD);
2638 case 'W': FTST(OP_FTRWRITE);
2639 case 'X': FTST(OP_FTREXEC);
2640 case 'O': FTST(OP_FTROWNED);
2641 case 'e': FTST(OP_FTIS);
2642 case 'z': FTST(OP_FTZERO);
2643 case 's': FTST(OP_FTSIZE);
2644 case 'f': FTST(OP_FTFILE);
2645 case 'd': FTST(OP_FTDIR);
2646 case 'l': FTST(OP_FTLINK);
2647 case 'p': FTST(OP_FTPIPE);
2648 case 'S': FTST(OP_FTSOCK);
2649 case 'u': FTST(OP_FTSUID);
2650 case 'g': FTST(OP_FTSGID);
2651 case 'k': FTST(OP_FTSVTX);
2652 case 'b': FTST(OP_FTBLK);
2653 case 'c': FTST(OP_FTCHR);
2654 case 't': FTST(OP_FTTTY);
2655 case 'T': FTST(OP_FTTEXT);
2656 case 'B': FTST(OP_FTBINARY);
2657 case 'M': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTMTIME);
2658 case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME);
2659 case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME);
2661 Perl_croak(aTHX_ "Unrecognized file test: -%c", (int)tmp);
2668 if (PL_expect == XOPERATOR)
2673 else if (*s == '>') {
2676 if (isIDFIRST_lazy_if(s,UTF)) {
2677 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2685 if (PL_expect == XOPERATOR)
2688 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2690 OPERATOR('-'); /* unary minus */
2697 if (PL_expect == XOPERATOR)
2702 if (PL_expect == XOPERATOR)
2705 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2711 if (PL_expect != XOPERATOR) {
2712 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2713 PL_expect = XOPERATOR;
2714 force_ident(PL_tokenbuf, '*');
2727 if (PL_expect == XOPERATOR) {
2731 PL_tokenbuf[0] = '%';
2732 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2733 if (!PL_tokenbuf[1]) {
2735 yyerror("Final % should be \\% or %name");
2738 PL_pending_ident = '%';
2757 switch (PL_expect) {
2760 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2762 PL_bufptr = s; /* update in case we back off */
2768 PL_expect = XTERMBLOCK;
2772 while (isIDFIRST_lazy_if(s,UTF)) {
2773 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2774 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2775 if (tmp < 0) tmp = -tmp;
2790 d = scan_str(d,TRUE,TRUE);
2793 SvREFCNT_dec(PL_lex_stuff);
2794 PL_lex_stuff = Nullsv;
2796 /* MUST advance bufptr here to avoid bogus
2797 "at end of line" context messages from yyerror().
2799 PL_bufptr = s + len;
2800 yyerror("Unterminated attribute parameter in attribute list");
2803 return 0; /* EOF indicator */
2807 SV *sv = newSVpvn(s, len);
2808 sv_catsv(sv, PL_lex_stuff);
2809 attrs = append_elem(OP_LIST, attrs,
2810 newSVOP(OP_CONST, 0, sv));
2811 SvREFCNT_dec(PL_lex_stuff);
2812 PL_lex_stuff = Nullsv;
2815 attrs = append_elem(OP_LIST, attrs,
2816 newSVOP(OP_CONST, 0,
2820 if (*s == ':' && s[1] != ':')
2823 break; /* require real whitespace or :'s */
2825 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
2826 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
2827 char q = ((*s == '\'') ? '"' : '\'');
2828 /* If here for an expression, and parsed no attrs, back off. */
2829 if (tmp == '=' && !attrs) {
2833 /* MUST advance bufptr here to avoid bogus "at end of line"
2834 context messages from yyerror().
2838 yyerror("Unterminated attribute list");
2840 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
2848 PL_nextval[PL_nexttoke].opval = attrs;
2856 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
2857 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
2862 if (CopLINE(PL_curcop) < PL_copline)
2863 PL_copline = CopLINE(PL_curcop);
2874 if (PL_lex_brackets <= 0)
2875 yyerror("Unmatched right square bracket");
2878 if (PL_lex_state == LEX_INTERPNORMAL) {
2879 if (PL_lex_brackets == 0) {
2880 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
2881 PL_lex_state = LEX_INTERPEND;
2888 if (PL_lex_brackets > 100) {
2889 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
2890 if (newlb != PL_lex_brackstack) {
2892 PL_lex_brackstack = newlb;
2895 switch (PL_expect) {
2897 if (PL_lex_formbrack) {
2901 if (PL_oldoldbufptr == PL_last_lop)
2902 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2904 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2905 OPERATOR(HASHBRACK);
2907 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2910 PL_tokenbuf[0] = '\0';
2911 if (d < PL_bufend && *d == '-') {
2912 PL_tokenbuf[0] = '-';
2914 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2917 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
2918 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
2920 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2923 char minus = (PL_tokenbuf[0] == '-');
2924 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
2932 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
2937 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2942 if (PL_oldoldbufptr == PL_last_lop)
2943 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2945 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2948 OPERATOR(HASHBRACK);
2949 /* This hack serves to disambiguate a pair of curlies
2950 * as being a block or an anon hash. Normally, expectation
2951 * determines that, but in cases where we're not in a
2952 * position to expect anything in particular (like inside
2953 * eval"") we have to resolve the ambiguity. This code
2954 * covers the case where the first term in the curlies is a
2955 * quoted string. Most other cases need to be explicitly
2956 * disambiguated by prepending a `+' before the opening
2957 * curly in order to force resolution as an anon hash.
2959 * XXX should probably propagate the outer expectation
2960 * into eval"" to rely less on this hack, but that could
2961 * potentially break current behavior of eval"".
2965 if (*s == '\'' || *s == '"' || *s == '`') {
2966 /* common case: get past first string, handling escapes */
2967 for (t++; t < PL_bufend && *t != *s;)
2968 if (*t++ == '\\' && (*t == '\\' || *t == *s))
2972 else if (*s == 'q') {
2975 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
2979 char open, close, term;
2982 while (t < PL_bufend && isSPACE(*t))
2986 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
2990 for (t++; t < PL_bufend; t++) {
2991 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
2993 else if (*t == open)
2997 for (t++; t < PL_bufend; t++) {
2998 if (*t == '\\' && t+1 < PL_bufend)
3000 else if (*t == close && --brackets <= 0)
3002 else if (*t == open)
3008 else if (isALNUM_lazy_if(t,UTF)) {
3010 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3013 while (t < PL_bufend && isSPACE(*t))
3015 /* if comma follows first term, call it an anon hash */
3016 /* XXX it could be a comma expression with loop modifiers */
3017 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3018 || (*t == '=' && t[1] == '>')))
3019 OPERATOR(HASHBRACK);
3020 if (PL_expect == XREF)
3023 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3029 yylval.ival = CopLINE(PL_curcop);
3030 if (isSPACE(*s) || *s == '#')
3031 PL_copline = NOLINE; /* invalidate current command line number */
3036 if (PL_lex_brackets <= 0)
3037 yyerror("Unmatched right curly bracket");
3039 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3040 if (PL_lex_brackets < PL_lex_formbrack)
3041 PL_lex_formbrack = 0;
3042 if (PL_lex_state == LEX_INTERPNORMAL) {
3043 if (PL_lex_brackets == 0) {
3044 if (PL_expect & XFAKEBRACK) {
3045 PL_expect &= XENUMMASK;
3046 PL_lex_state = LEX_INTERPEND;
3048 return yylex(); /* ignore fake brackets */
3050 if (*s == '-' && s[1] == '>')
3051 PL_lex_state = LEX_INTERPENDMAYBE;
3052 else if (*s != '[' && *s != '{')
3053 PL_lex_state = LEX_INTERPEND;
3056 if (PL_expect & XFAKEBRACK) {
3057 PL_expect &= XENUMMASK;
3059 return yylex(); /* ignore fake brackets */
3069 if (PL_expect == XOPERATOR) {
3070 if (ckWARN(WARN_SEMICOLON)
3071 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3073 CopLINE_dec(PL_curcop);
3074 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3075 CopLINE_inc(PL_curcop);
3080 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3082 PL_expect = XOPERATOR;
3083 force_ident(PL_tokenbuf, '&');
3087 yylval.ival = (OPpENTERSUB_AMPER<<8);
3106 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3107 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3109 if (PL_expect == XSTATE && isALPHA(tmp) &&
3110 (s == PL_linestart+1 || s[-2] == '\n') )
3112 if (PL_in_eval && !PL_rsfp) {
3117 if (strnEQ(s,"=cut",4)) {
3131 PL_doextract = TRUE;
3134 if (PL_lex_brackets < PL_lex_formbrack) {
3136 #ifdef PERL_STRICT_CR
3137 for (t = s; *t == ' ' || *t == '\t'; t++) ;
3139 for (t = s; *t == ' ' || *t == '\t' || *t == '\r'; t++) ;
3141 if (*t == '\n' || *t == '#') {
3159 if (PL_expect != XOPERATOR) {
3160 if (s[1] != '<' && !strchr(s,'>'))
3163 s = scan_heredoc(s);
3165 s = scan_inputsymbol(s);
3166 TERM(sublex_start());
3171 SHop(OP_LEFT_SHIFT);
3185 SHop(OP_RIGHT_SHIFT);
3194 if (PL_expect == XOPERATOR) {
3195 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3198 return ','; /* grandfather non-comma-format format */
3202 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3203 PL_tokenbuf[0] = '@';
3204 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3205 sizeof PL_tokenbuf - 1, FALSE);
3206 if (PL_expect == XOPERATOR)
3207 no_op("Array length", s);
3208 if (!PL_tokenbuf[1])
3210 PL_expect = XOPERATOR;
3211 PL_pending_ident = '#';
3215 PL_tokenbuf[0] = '$';
3216 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3217 sizeof PL_tokenbuf - 1, FALSE);
3218 if (PL_expect == XOPERATOR)
3220 if (!PL_tokenbuf[1]) {
3222 yyerror("Final $ should be \\$ or $name");
3226 /* This kludge not intended to be bulletproof. */
3227 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3228 yylval.opval = newSVOP(OP_CONST, 0,
3229 newSViv((IV)PL_compiling.cop_arybase));
3230 yylval.opval->op_private = OPpCONST_ARYBASE;
3236 if (PL_lex_state == LEX_NORMAL)
3239 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3242 PL_tokenbuf[0] = '@';
3243 if (ckWARN(WARN_SYNTAX)) {
3245 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3248 PL_bufptr = skipspace(PL_bufptr);
3249 while (t < PL_bufend && *t != ']')
3251 Perl_warner(aTHX_ WARN_SYNTAX,
3252 "Multidimensional syntax %.*s not supported",
3253 (t - PL_bufptr) + 1, PL_bufptr);
3257 else if (*s == '{') {
3258 PL_tokenbuf[0] = '%';
3259 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3260 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3262 char tmpbuf[sizeof PL_tokenbuf];
3264 for (t++; isSPACE(*t); t++) ;
3265 if (isIDFIRST_lazy_if(t,UTF)) {
3266 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3267 for (; isSPACE(*t); t++) ;
3268 if (*t == ';' && get_cv(tmpbuf, FALSE))
3269 Perl_warner(aTHX_ WARN_SYNTAX,
3270 "You need to quote \"%s\"", tmpbuf);
3276 PL_expect = XOPERATOR;
3277 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3278 bool islop = (PL_last_lop == PL_oldoldbufptr);
3279 if (!islop || PL_last_lop_op == OP_GREPSTART)
3280 PL_expect = XOPERATOR;
3281 else if (strchr("$@\"'`q", *s))
3282 PL_expect = XTERM; /* e.g. print $fh "foo" */
3283 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3284 PL_expect = XTERM; /* e.g. print $fh &sub */
3285 else if (isIDFIRST_lazy_if(s,UTF)) {
3286 char tmpbuf[sizeof PL_tokenbuf];
3287 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3288 if (tmp = keyword(tmpbuf, len)) {
3289 /* binary operators exclude handle interpretations */
3301 PL_expect = XTERM; /* e.g. print $fh length() */
3306 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3307 if (gv && GvCVu(gv))
3308 PL_expect = XTERM; /* e.g. print $fh subr() */
3311 else if (isDIGIT(*s))
3312 PL_expect = XTERM; /* e.g. print $fh 3 */
3313 else if (*s == '.' && isDIGIT(s[1]))
3314 PL_expect = XTERM; /* e.g. print $fh .3 */
3315 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3316 PL_expect = XTERM; /* e.g. print $fh -1 */
3317 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3318 PL_expect = XTERM; /* print $fh <<"EOF" */
3320 PL_pending_ident = '$';
3324 if (PL_expect == XOPERATOR)
3326 PL_tokenbuf[0] = '@';
3327 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3328 if (!PL_tokenbuf[1]) {
3330 yyerror("Final @ should be \\@ or @name");
3333 if (PL_lex_state == LEX_NORMAL)
3335 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3337 PL_tokenbuf[0] = '%';
3339 /* Warn about @ where they meant $. */
3340 if (ckWARN(WARN_SYNTAX)) {
3341 if (*s == '[' || *s == '{') {
3343 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3345 if (*t == '}' || *t == ']') {
3347 PL_bufptr = skipspace(PL_bufptr);
3348 Perl_warner(aTHX_ WARN_SYNTAX,
3349 "Scalar value %.*s better written as $%.*s",
3350 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3355 PL_pending_ident = '@';
3358 case '/': /* may either be division or pattern */
3359 case '?': /* may either be conditional or pattern */
3360 if (PL_expect != XOPERATOR) {
3361 /* Disable warning on "study /blah/" */
3362 if (PL_oldoldbufptr == PL_last_uni
3363 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3364 || memNE(PL_last_uni, "study", 5)
3365 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3367 s = scan_pat(s,OP_MATCH);
3368 TERM(sublex_start());
3376 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3377 #ifdef PERL_STRICT_CR
3380 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3382 && (s == PL_linestart || s[-1] == '\n') )
3384 PL_lex_formbrack = 0;
3388 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3394 yylval.ival = OPf_SPECIAL;
3400 if (PL_expect != XOPERATOR)
3405 case '0': case '1': case '2': case '3': case '4':
3406 case '5': case '6': case '7': case '8': case '9':
3408 if (PL_expect == XOPERATOR)
3413 s = scan_str(s,FALSE,FALSE);
3414 if (PL_expect == XOPERATOR) {
3415 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3418 return ','; /* grandfather non-comma-format format */
3424 missingterm((char*)0);
3425 yylval.ival = OP_CONST;
3426 TERM(sublex_start());
3429 s = scan_str(s,FALSE,FALSE);
3430 if (PL_expect == XOPERATOR) {
3431 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3434 return ','; /* grandfather non-comma-format format */
3440 missingterm((char*)0);
3441 yylval.ival = OP_CONST;
3442 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3443 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3444 yylval.ival = OP_STRINGIFY;
3448 TERM(sublex_start());
3451 s = scan_str(s,FALSE,FALSE);
3452 if (PL_expect == XOPERATOR)
3453 no_op("Backticks",s);
3455 missingterm((char*)0);
3456 yylval.ival = OP_BACKTICK;
3458 TERM(sublex_start());
3462 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3463 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3465 if (PL_expect == XOPERATOR)
3466 no_op("Backslash",s);
3470 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3474 while (isDIGIT(*start))
3476 if (*start == '.' && isDIGIT(start[1])) {
3480 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3481 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF)) {
3485 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3495 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3535 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3537 /* Some keywords can be followed by any delimiter, including ':' */
3538 tmp = (len == 1 && strchr("msyq", PL_tokenbuf[0]) ||
3539 len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3540 (PL_tokenbuf[0] == 'q' &&
3541 strchr("qwxr", PL_tokenbuf[1]))));
3543 /* x::* is just a word, unless x is "CORE" */
3544 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3548 while (d < PL_bufend && isSPACE(*d))
3549 d++; /* no comments skipped here, or s### is misparsed */
3551 /* Is this a label? */
3552 if (!tmp && PL_expect == XSTATE
3553 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3555 yylval.pval = savepv(PL_tokenbuf);
3560 /* Check for keywords */
3561 tmp = keyword(PL_tokenbuf, len);
3563 /* Is this a word before a => operator? */
3564 if (strnEQ(d,"=>",2)) {
3566 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3567 yylval.opval->op_private = OPpCONST_BARE;
3571 if (tmp < 0) { /* second-class keyword? */
3572 GV *ogv = Nullgv; /* override (winner) */
3573 GV *hgv = Nullgv; /* hidden (loser) */
3574 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3576 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3579 if (GvIMPORTED_CV(gv))
3581 else if (! CvMETHOD(cv))
3585 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3586 (gv = *gvp) != (GV*)&PL_sv_undef &&
3587 GvCVu(gv) && GvIMPORTED_CV(gv))
3593 tmp = 0; /* overridden by import or by GLOBAL */
3596 && -tmp==KEY_lock /* XXX generalizable kludge */
3598 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3600 tmp = 0; /* any sub overrides "weak" keyword */
3602 else { /* no override */
3606 if (ckWARN(WARN_AMBIGUOUS) && hgv
3607 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3608 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3609 "Ambiguous call resolved as CORE::%s(), %s",
3610 GvENAME(hgv), "qualify as such or use &");
3617 default: /* not a keyword */
3620 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3622 /* Get the rest if it looks like a package qualifier */
3624 if (*s == '\'' || *s == ':' && s[1] == ':') {
3626 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3629 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3630 *s == '\'' ? "'" : "::");
3634 if (PL_expect == XOPERATOR) {
3635 if (PL_bufptr == PL_linestart) {
3636 CopLINE_dec(PL_curcop);
3637 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3638 CopLINE_inc(PL_curcop);
3641 no_op("Bareword",s);
3644 /* Look for a subroutine with this name in current package,
3645 unless name is "Foo::", in which case Foo is a bearword
3646 (and a package name). */
3649 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3651 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3652 Perl_warner(aTHX_ WARN_BAREWORD,
3653 "Bareword \"%s\" refers to nonexistent package",
3656 PL_tokenbuf[len] = '\0';
3663 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3666 /* if we saw a global override before, get the right name */
3669 sv = newSVpvn("CORE::GLOBAL::",14);
3670 sv_catpv(sv,PL_tokenbuf);
3673 sv = newSVpv(PL_tokenbuf,0);
3675 /* Presume this is going to be a bareword of some sort. */
3678 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3679 yylval.opval->op_private = OPpCONST_BARE;
3681 /* And if "Foo::", then that's what it certainly is. */
3686 /* See if it's the indirect object for a list operator. */
3688 if (PL_oldoldbufptr &&
3689 PL_oldoldbufptr < PL_bufptr &&
3690 (PL_oldoldbufptr == PL_last_lop
3691 || PL_oldoldbufptr == PL_last_uni) &&
3692 /* NO SKIPSPACE BEFORE HERE! */
3693 (PL_expect == XREF ||
3694 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3696 bool immediate_paren = *s == '(';
3698 /* (Now we can afford to cross potential line boundary.) */
3701 /* Two barewords in a row may indicate method call. */
3703 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3706 /* If not a declared subroutine, it's an indirect object. */
3707 /* (But it's an indir obj regardless for sort.) */
3709 if ((PL_last_lop_op == OP_SORT ||
3710 (!immediate_paren && (!gv || !GvCVu(gv)))) &&
3711 (PL_last_lop_op != OP_MAPSTART &&
3712 PL_last_lop_op != OP_GREPSTART))
3714 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3719 /* If followed by a paren, it's certainly a subroutine. */
3721 PL_expect = XOPERATOR;
3725 if (gv && GvCVu(gv)) {
3726 for (d = s + 1; *d == ' ' || *d == '\t'; d++) ;
3727 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3732 PL_nextval[PL_nexttoke].opval = yylval.opval;
3733 PL_expect = XOPERATOR;
3739 /* If followed by var or block, call it a method (unless sub) */
3741 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3742 PL_last_lop = PL_oldbufptr;
3743 PL_last_lop_op = OP_METHOD;
3747 /* If followed by a bareword, see if it looks like indir obj. */
3749 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3752 /* Not a method, so call it a subroutine (if defined) */
3754 if (gv && GvCVu(gv)) {
3756 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3757 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3758 "Ambiguous use of -%s resolved as -&%s()",
3759 PL_tokenbuf, PL_tokenbuf);
3760 /* Check for a constant sub */
3762 if ((sv = cv_const_sv(cv))) {
3764 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3765 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3766 yylval.opval->op_private = 0;
3770 /* Resolve to GV now. */
3771 op_free(yylval.opval);
3772 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3773 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3774 PL_last_lop = PL_oldbufptr;
3775 PL_last_lop_op = OP_ENTERSUB;
3776 /* Is there a prototype? */
3779 char *proto = SvPV((SV*)cv, len);
3782 if (strEQ(proto, "$"))
3784 if (*proto == '&' && *s == '{') {
3785 sv_setpv(PL_subname,"__ANON__");
3789 PL_nextval[PL_nexttoke].opval = yylval.opval;
3795 /* Call it a bare word */
3797 if (PL_hints & HINT_STRICT_SUBS)
3798 yylval.opval->op_private |= OPpCONST_STRICT;
3801 if (ckWARN(WARN_RESERVED)) {
3802 if (lastchar != '-') {
3803 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
3805 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
3812 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
3813 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3814 "Operator or semicolon missing before %c%s",
3815 lastchar, PL_tokenbuf);
3816 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3817 "Ambiguous use of %c resolved as operator %c",
3818 lastchar, lastchar);
3824 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3825 newSVpv(CopFILE(PL_curcop),0));
3829 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3830 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
3833 case KEY___PACKAGE__:
3834 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3836 ? newSVsv(PL_curstname)
3845 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
3846 char *pname = "main";
3847 if (PL_tokenbuf[2] == 'D')
3848 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
3849 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
3852 GvIOp(gv) = newIO();
3853 IoIFP(GvIOp(gv)) = PL_rsfp;
3854 #if defined(HAS_FCNTL) && defined(F_SETFD)
3856 int fd = PerlIO_fileno(PL_rsfp);
3857 fcntl(fd,F_SETFD,fd >= 3);
3860 /* Mark this internal pseudo-handle as clean */
3861 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
3863 IoTYPE(GvIOp(gv)) = '|';
3864 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
3865 IoTYPE(GvIOp(gv)) = '-';
3867 IoTYPE(GvIOp(gv)) = '<';
3868 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
3869 /* if the script was opened in binmode, we need to revert
3870 * it to text mode for compatibility; but only iff it has CRs
3871 * XXX this is a questionable hack at best. */
3872 if (PL_bufend-PL_bufptr > 2
3873 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
3876 if (IoTYPE(GvIOp(gv)) == '<') {
3877 loc = PerlIO_tell(PL_rsfp);
3878 (void)PerlIO_seek(PL_rsfp, 0L, 0);
3880 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
3881 #if defined(__BORLANDC__)
3882 /* XXX see note in do_binmode() */
3883 ((FILE*)PL_rsfp)->flags |= _F_BIN;
3886 PerlIO_seek(PL_rsfp, loc, 0);
3901 if (PL_expect == XSTATE) {
3908 if (*s == ':' && s[1] == ':') {
3911 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3912 tmp = keyword(PL_tokenbuf, len);
3926 LOP(OP_ACCEPT,XTERM);
3932 LOP(OP_ATAN2,XTERM);
3941 LOP(OP_BLESS,XTERM);
3950 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
3967 if (!PL_cryptseen) {
3968 PL_cryptseen = TRUE;
3972 LOP(OP_CRYPT,XTERM);
3975 if (ckWARN(WARN_CHMOD)) {
3976 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
3977 if (*d != '0' && isDIGIT(*d))
3978 Perl_warner(aTHX_ WARN_CHMOD,
3979 "chmod() mode argument is missing initial 0");
3981 LOP(OP_CHMOD,XTERM);
3984 LOP(OP_CHOWN,XTERM);
3987 LOP(OP_CONNECT,XTERM);
4003 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4007 PL_hints |= HINT_BLOCK_SCOPE;
4017 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4018 LOP(OP_DBMOPEN,XTERM);
4024 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4031 yylval.ival = CopLINE(PL_curcop);
4045 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4046 UNIBRACK(OP_ENTEREVAL);
4061 case KEY_endhostent:
4067 case KEY_endservent:
4070 case KEY_endprotoent:
4081 yylval.ival = CopLINE(PL_curcop);
4083 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4085 if ((PL_bufend - p) >= 3 &&
4086 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4088 else if ((PL_bufend - p) >= 4 &&
4089 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4092 if (isIDFIRST_lazy_if(p,UTF)) {
4093 p = scan_ident(p, PL_bufend,
4094 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4098 Perl_croak(aTHX_ "Missing $ on loop variable");
4103 LOP(OP_FORMLINE,XTERM);
4109 LOP(OP_FCNTL,XTERM);
4115 LOP(OP_FLOCK,XTERM);
4124 LOP(OP_GREPSTART, XREF);
4127 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4142 case KEY_getpriority:
4143 LOP(OP_GETPRIORITY,XTERM);
4145 case KEY_getprotobyname:
4148 case KEY_getprotobynumber:
4149 LOP(OP_GPBYNUMBER,XTERM);
4151 case KEY_getprotoent:
4163 case KEY_getpeername:
4164 UNI(OP_GETPEERNAME);
4166 case KEY_gethostbyname:
4169 case KEY_gethostbyaddr:
4170 LOP(OP_GHBYADDR,XTERM);
4172 case KEY_gethostent:
4175 case KEY_getnetbyname:
4178 case KEY_getnetbyaddr:
4179 LOP(OP_GNBYADDR,XTERM);
4184 case KEY_getservbyname:
4185 LOP(OP_GSBYNAME,XTERM);
4187 case KEY_getservbyport:
4188 LOP(OP_GSBYPORT,XTERM);
4190 case KEY_getservent:
4193 case KEY_getsockname:
4194 UNI(OP_GETSOCKNAME);
4196 case KEY_getsockopt:
4197 LOP(OP_GSOCKOPT,XTERM);
4219 yylval.ival = CopLINE(PL_curcop);
4223 LOP(OP_INDEX,XTERM);
4229 LOP(OP_IOCTL,XTERM);
4241 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4273 LOP(OP_LISTEN,XTERM);
4282 s = scan_pat(s,OP_MATCH);
4283 TERM(sublex_start());
4286 LOP(OP_MAPSTART, XREF);
4289 LOP(OP_MKDIR,XTERM);
4292 LOP(OP_MSGCTL,XTERM);
4295 LOP(OP_MSGGET,XTERM);
4298 LOP(OP_MSGRCV,XTERM);
4301 LOP(OP_MSGSND,XTERM);
4307 if (isIDFIRST_lazy_if(s,UTF)) {
4308 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4309 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4311 PL_in_my_stash = gv_stashpv(PL_tokenbuf, FALSE);
4312 if (!PL_in_my_stash) {
4315 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4323 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4330 if (PL_expect != XSTATE)
4331 yyerror("\"no\" not allowed in expression");
4332 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4333 s = force_version(s);
4338 if (*s == '(' || (s = skipspace(s), *s == '('))
4345 if (isIDFIRST_lazy_if(s,UTF)) {
4347 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4349 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4350 Perl_warner(aTHX_ WARN_PRECEDENCE,
4351 "Precedence problem: open %.*s should be open(%.*s)",
4357 yylval.ival = OP_OR;
4367 LOP(OP_OPEN_DIR,XTERM);
4370 checkcomma(s,PL_tokenbuf,"filehandle");
4374 checkcomma(s,PL_tokenbuf,"filehandle");
4393 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4397 LOP(OP_PIPE_OP,XTERM);
4400 s = scan_str(s,FALSE,FALSE);
4402 missingterm((char*)0);
4403 yylval.ival = OP_CONST;
4404 TERM(sublex_start());
4410 s = scan_str(s,FALSE,FALSE);
4412 missingterm((char*)0);
4414 if (SvCUR(PL_lex_stuff)) {
4417 d = SvPV_force(PL_lex_stuff, len);
4419 for (; isSPACE(*d) && len; --len, ++d) ;
4422 if (!warned && ckWARN(WARN_QW)) {
4423 for (; !isSPACE(*d) && len; --len, ++d) {
4425 Perl_warner(aTHX_ WARN_QW,
4426 "Possible attempt to separate words with commas");
4429 else if (*d == '#') {
4430 Perl_warner(aTHX_ WARN_QW,
4431 "Possible attempt to put comments in qw() list");
4437 for (; !isSPACE(*d) && len; --len, ++d) ;
4439 words = append_elem(OP_LIST, words,
4440 newSVOP(OP_CONST, 0, newSVpvn(b, d-b)));
4444 PL_nextval[PL_nexttoke].opval = words;
4449 SvREFCNT_dec(PL_lex_stuff);
4450 PL_lex_stuff = Nullsv;
4455 s = scan_str(s,FALSE,FALSE);
4457 missingterm((char*)0);
4458 yylval.ival = OP_STRINGIFY;
4459 if (SvIVX(PL_lex_stuff) == '\'')
4460 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4461 TERM(sublex_start());
4464 s = scan_pat(s,OP_QR);
4465 TERM(sublex_start());
4468 s = scan_str(s,FALSE,FALSE);
4470 missingterm((char*)0);
4471 yylval.ival = OP_BACKTICK;
4473 TERM(sublex_start());
4480 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4481 s = force_version(s);
4484 *PL_tokenbuf = '\0';
4485 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4486 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4487 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4489 yyerror("<> should be quotes");
4497 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4501 LOP(OP_RENAME,XTERM);
4510 LOP(OP_RINDEX,XTERM);
4533 LOP(OP_REVERSE,XTERM);
4544 TERM(sublex_start());
4546 TOKEN(1); /* force error */
4555 LOP(OP_SELECT,XTERM);
4561 LOP(OP_SEMCTL,XTERM);
4564 LOP(OP_SEMGET,XTERM);
4567 LOP(OP_SEMOP,XTERM);
4573 LOP(OP_SETPGRP,XTERM);
4575 case KEY_setpriority:
4576 LOP(OP_SETPRIORITY,XTERM);
4578 case KEY_sethostent:
4584 case KEY_setservent:
4587 case KEY_setprotoent:
4597 LOP(OP_SEEKDIR,XTERM);
4599 case KEY_setsockopt:
4600 LOP(OP_SSOCKOPT,XTERM);
4606 LOP(OP_SHMCTL,XTERM);
4609 LOP(OP_SHMGET,XTERM);
4612 LOP(OP_SHMREAD,XTERM);
4615 LOP(OP_SHMWRITE,XTERM);
4618 LOP(OP_SHUTDOWN,XTERM);
4627 LOP(OP_SOCKET,XTERM);
4629 case KEY_socketpair:
4630 LOP(OP_SOCKPAIR,XTERM);
4633 checkcomma(s,PL_tokenbuf,"subroutine name");
4635 if (*s == ';' || *s == ')') /* probably a close */
4636 Perl_croak(aTHX_ "sort is now a reserved word");
4638 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4642 LOP(OP_SPLIT,XTERM);
4645 LOP(OP_SPRINTF,XTERM);
4648 LOP(OP_SPLICE,XTERM);
4663 LOP(OP_SUBSTR,XTERM);
4669 char tmpbuf[sizeof PL_tokenbuf];
4671 expectation attrful;
4672 bool have_name, have_proto;
4677 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4678 (*s == ':' && s[1] == ':'))
4681 attrful = XATTRBLOCK;
4682 /* remember buffer pos'n for later force_word */
4683 tboffset = s - PL_oldbufptr;
4684 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4685 if (strchr(tmpbuf, ':'))
4686 sv_setpv(PL_subname, tmpbuf);
4688 sv_setsv(PL_subname,PL_curstname);
4689 sv_catpvn(PL_subname,"::",2);
4690 sv_catpvn(PL_subname,tmpbuf,len);
4697 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4698 PL_expect = XTERMBLOCK;
4699 attrful = XATTRTERM;
4700 sv_setpv(PL_subname,"?");
4704 if (key == KEY_format) {
4706 PL_lex_formbrack = PL_lex_brackets + 1;
4708 (void) force_word(PL_oldbufptr + tboffset, WORD,
4713 /* Look for a prototype */
4717 s = scan_str(s,FALSE,FALSE);
4720 SvREFCNT_dec(PL_lex_stuff);
4721 PL_lex_stuff = Nullsv;
4722 Perl_croak(aTHX_ "Prototype not terminated");
4725 d = SvPVX(PL_lex_stuff);
4727 for (p = d; *p; ++p) {
4732 SvCUR(PL_lex_stuff) = tmp;
4740 if (*s == ':' && s[1] != ':')
4741 PL_expect = attrful;
4744 PL_nextval[PL_nexttoke].opval =
4745 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4746 PL_lex_stuff = Nullsv;
4750 sv_setpv(PL_subname,"__ANON__");
4753 (void) force_word(PL_oldbufptr + tboffset, WORD,
4762 LOP(OP_SYSTEM,XREF);
4765 LOP(OP_SYMLINK,XTERM);
4768 LOP(OP_SYSCALL,XTERM);
4771 LOP(OP_SYSOPEN,XTERM);
4774 LOP(OP_SYSSEEK,XTERM);
4777 LOP(OP_SYSREAD,XTERM);
4780 LOP(OP_SYSWRITE,XTERM);
4784 TERM(sublex_start());
4805 LOP(OP_TRUNCATE,XTERM);
4817 yylval.ival = CopLINE(PL_curcop);
4821 yylval.ival = CopLINE(PL_curcop);
4825 LOP(OP_UNLINK,XTERM);
4831 LOP(OP_UNPACK,XTERM);
4834 LOP(OP_UTIME,XTERM);
4837 if (ckWARN(WARN_UMASK)) {
4838 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4839 if (*d != '0' && isDIGIT(*d))
4840 Perl_warner(aTHX_ WARN_UMASK,
4841 "umask: argument is missing initial 0");
4846 LOP(OP_UNSHIFT,XTERM);
4849 if (PL_expect != XSTATE)
4850 yyerror("\"use\" not allowed in expression");
4852 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4853 s = force_version(s);
4854 if (*s == ';' || (s = skipspace(s), *s == ';')) {
4855 PL_nextval[PL_nexttoke].opval = Nullop;
4860 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4861 s = force_version(s);
4873 yylval.ival = CopLINE(PL_curcop);
4877 PL_hints |= HINT_BLOCK_SCOPE;
4884 LOP(OP_WAITPID,XTERM);
4892 static char ctl_l[2];
4894 if (ctl_l[0] == '\0')
4895 ctl_l[0] = toCTRL('L');
4896 gv_fetchpv(ctl_l,TRUE, SVt_PV);
4899 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
4904 if (PL_expect == XOPERATOR)
4910 yylval.ival = OP_XOR;
4915 TERM(sublex_start());
4921 Perl_keyword(pTHX_ register char *d, I32 len)
4926 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
4927 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
4928 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
4929 if (strEQ(d,"__DATA__")) return KEY___DATA__;
4930 if (strEQ(d,"__END__")) return KEY___END__;
4934 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
4939 if (strEQ(d,"and")) return -KEY_and;
4940 if (strEQ(d,"abs")) return -KEY_abs;
4943 if (strEQ(d,"alarm")) return -KEY_alarm;
4944 if (strEQ(d,"atan2")) return -KEY_atan2;
4947 if (strEQ(d,"accept")) return -KEY_accept;
4952 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
4955 if (strEQ(d,"bless")) return -KEY_bless;
4956 if (strEQ(d,"bind")) return -KEY_bind;
4957 if (strEQ(d,"binmode")) return -KEY_binmode;
4960 if (strEQ(d,"CORE")) return -KEY_CORE;
4961 if (strEQ(d,"CHECK")) return KEY_CHECK;
4966 if (strEQ(d,"cmp")) return -KEY_cmp;
4967 if (strEQ(d,"chr")) return -KEY_chr;
4968 if (strEQ(d,"cos")) return -KEY_cos;
4971 if (strEQ(d,"chop")) return KEY_chop;
4974 if (strEQ(d,"close")) return -KEY_close;
4975 if (strEQ(d,"chdir")) return -KEY_chdir;
4976 if (strEQ(d,"chomp")) return KEY_chomp;
4977 if (strEQ(d,"chmod")) return -KEY_chmod;
4978 if (strEQ(d,"chown")) return -KEY_chown;
4979 if (strEQ(d,"crypt")) return -KEY_crypt;
4982 if (strEQ(d,"chroot")) return -KEY_chroot;
4983 if (strEQ(d,"caller")) return -KEY_caller;
4986 if (strEQ(d,"connect")) return -KEY_connect;
4989 if (strEQ(d,"closedir")) return -KEY_closedir;
4990 if (strEQ(d,"continue")) return -KEY_continue;
4995 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5000 if (strEQ(d,"do")) return KEY_do;
5003 if (strEQ(d,"die")) return -KEY_die;
5006 if (strEQ(d,"dump")) return -KEY_dump;
5009 if (strEQ(d,"delete")) return KEY_delete;
5012 if (strEQ(d,"defined")) return KEY_defined;
5013 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5016 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5021 if (strEQ(d,"EQ")) { deprecate(d); return -KEY_eq;}
5022 if (strEQ(d,"END")) return KEY_END;
5027 if (strEQ(d,"eq")) return -KEY_eq;
5030 if (strEQ(d,"eof")) return -KEY_eof;
5031 if (strEQ(d,"exp")) return -KEY_exp;
5034 if (strEQ(d,"else")) return KEY_else;
5035 if (strEQ(d,"exit")) return -KEY_exit;
5036 if (strEQ(d,"eval")) return KEY_eval;
5037 if (strEQ(d,"exec")) return -KEY_exec;
5038 if (strEQ(d,"each")) return KEY_each;
5041 if (strEQ(d,"elsif")) return KEY_elsif;
5044 if (strEQ(d,"exists")) return KEY_exists;
5045 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5048 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5049 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5052 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5055 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5056 if (strEQ(d,"endservent")) return -KEY_endservent;
5059 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5066 if (strEQ(d,"for")) return KEY_for;
5069 if (strEQ(d,"fork")) return -KEY_fork;
5072 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5073 if (strEQ(d,"flock")) return -KEY_flock;
5076 if (strEQ(d,"format")) return KEY_format;
5077 if (strEQ(d,"fileno")) return -KEY_fileno;
5080 if (strEQ(d,"foreach")) return KEY_foreach;
5083 if (strEQ(d,"formline")) return -KEY_formline;
5089 if (strEQ(d,"GT")) { deprecate(d); return -KEY_gt;}
5090 if (strEQ(d,"GE")) { deprecate(d); return -KEY_ge;}
5094 if (strnEQ(d,"get",3)) {
5099 if (strEQ(d,"ppid")) return -KEY_getppid;
5100 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5103 if (strEQ(d,"pwent")) return -KEY_getpwent;
5104 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5105 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5108 if (strEQ(d,"peername")) return -KEY_getpeername;
5109 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5110 if (strEQ(d,"priority")) return -KEY_getpriority;
5113 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5116 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5120 else if (*d == 'h') {
5121 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5122 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5123 if (strEQ(d,"hostent")) return -KEY_gethostent;
5125 else if (*d == 'n') {
5126 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5127 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5128 if (strEQ(d,"netent")) return -KEY_getnetent;
5130 else if (*d == 's') {
5131 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5132 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5133 if (strEQ(d,"servent")) return -KEY_getservent;
5134 if (strEQ(d,"sockname")) return -KEY_getsockname;
5135 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5137 else if (*d == 'g') {
5138 if (strEQ(d,"grent")) return -KEY_getgrent;
5139 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5140 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5142 else if (*d == 'l') {
5143 if (strEQ(d,"login")) return -KEY_getlogin;
5145 else if (strEQ(d,"c")) return -KEY_getc;
5150 if (strEQ(d,"gt")) return -KEY_gt;
5151 if (strEQ(d,"ge")) return -KEY_ge;
5154 if (strEQ(d,"grep")) return KEY_grep;
5155 if (strEQ(d,"goto")) return KEY_goto;
5156 if (strEQ(d,"glob")) return KEY_glob;
5159 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5164 if (strEQ(d,"hex")) return -KEY_hex;
5167 if (strEQ(d,"INIT")) return KEY_INIT;
5172 if (strEQ(d,"if")) return KEY_if;
5175 if (strEQ(d,"int")) return -KEY_int;
5178 if (strEQ(d,"index")) return -KEY_index;
5179 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5184 if (strEQ(d,"join")) return -KEY_join;
5188 if (strEQ(d,"keys")) return KEY_keys;
5189 if (strEQ(d,"kill")) return -KEY_kill;
5194 if (strEQ(d,"LT")) { deprecate(d); return -KEY_lt;}
5195 if (strEQ(d,"LE")) { deprecate(d); return -KEY_le;}
5201 if (strEQ(d,"lt")) return -KEY_lt;
5202 if (strEQ(d,"le")) return -KEY_le;
5203 if (strEQ(d,"lc")) return -KEY_lc;
5206 if (strEQ(d,"log")) return -KEY_log;
5209 if (strEQ(d,"last")) return KEY_last;
5210 if (strEQ(d,"link")) return -KEY_link;
5211 if (strEQ(d,"lock")) return -KEY_lock;
5214 if (strEQ(d,"local")) return KEY_local;
5215 if (strEQ(d,"lstat")) return -KEY_lstat;
5218 if (strEQ(d,"length")) return -KEY_length;
5219 if (strEQ(d,"listen")) return -KEY_listen;
5222 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5225 if (strEQ(d,"localtime")) return -KEY_localtime;
5231 case 1: return KEY_m;
5233 if (strEQ(d,"my")) return KEY_my;
5236 if (strEQ(d,"map")) return KEY_map;
5239 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5242 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5243 if (strEQ(d,"msgget")) return -KEY_msgget;
5244 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5245 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5250 if (strEQ(d,"NE")) { deprecate(d); return -KEY_ne;}
5253 if (strEQ(d,"next")) return KEY_next;
5254 if (strEQ(d,"ne")) return -KEY_ne;
5255 if (strEQ(d,"not")) return -KEY_not;
5256 if (strEQ(d,"no")) return KEY_no;
5261 if (strEQ(d,"or")) return -KEY_or;
5264 if (strEQ(d,"ord")) return -KEY_ord;
5265 if (strEQ(d,"oct")) return -KEY_oct;
5266 if (strEQ(d,"our")) return KEY_our;
5269 if (strEQ(d,"open")) return -KEY_open;
5272 if (strEQ(d,"opendir")) return -KEY_opendir;
5279 if (strEQ(d,"pop")) return KEY_pop;
5280 if (strEQ(d,"pos")) return KEY_pos;
5283 if (strEQ(d,"push")) return KEY_push;
5284 if (strEQ(d,"pack")) return -KEY_pack;
5285 if (strEQ(d,"pipe")) return -KEY_pipe;
5288 if (strEQ(d,"print")) return KEY_print;
5291 if (strEQ(d,"printf")) return KEY_printf;
5294 if (strEQ(d,"package")) return KEY_package;
5297 if (strEQ(d,"prototype")) return KEY_prototype;
5302 if (strEQ(d,"q")) return KEY_q;
5303 if (strEQ(d,"qr")) return KEY_qr;
5304 if (strEQ(d,"qq")) return KEY_qq;
5305 if (strEQ(d,"qw")) return KEY_qw;
5306 if (strEQ(d,"qx")) return KEY_qx;
5308 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5313 if (strEQ(d,"ref")) return -KEY_ref;
5316 if (strEQ(d,"read")) return -KEY_read;
5317 if (strEQ(d,"rand")) return -KEY_rand;
5318 if (strEQ(d,"recv")) return -KEY_recv;
5319 if (strEQ(d,"redo")) return KEY_redo;
5322 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5323 if (strEQ(d,"reset")) return -KEY_reset;
5326 if (strEQ(d,"return")) return KEY_return;
5327 if (strEQ(d,"rename")) return -KEY_rename;
5328 if (strEQ(d,"rindex")) return -KEY_rindex;
5331 if (strEQ(d,"require")) return -KEY_require;
5332 if (strEQ(d,"reverse")) return -KEY_reverse;
5333 if (strEQ(d,"readdir")) return -KEY_readdir;
5336 if (strEQ(d,"readlink")) return -KEY_readlink;
5337 if (strEQ(d,"readline")) return -KEY_readline;
5338 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5341 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5347 case 0: return KEY_s;
5349 if (strEQ(d,"scalar")) return KEY_scalar;
5354 if (strEQ(d,"seek")) return -KEY_seek;
5355 if (strEQ(d,"send")) return -KEY_send;
5358 if (strEQ(d,"semop")) return -KEY_semop;
5361 if (strEQ(d,"select")) return -KEY_select;
5362 if (strEQ(d,"semctl")) return -KEY_semctl;
5363 if (strEQ(d,"semget")) return -KEY_semget;
5366 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5367 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5370 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5371 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5374 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5377 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5378 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5379 if (strEQ(d,"setservent")) return -KEY_setservent;
5382 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5383 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5390 if (strEQ(d,"shift")) return KEY_shift;
5393 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5394 if (strEQ(d,"shmget")) return -KEY_shmget;
5397 if (strEQ(d,"shmread")) return -KEY_shmread;
5400 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5401 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5406 if (strEQ(d,"sin")) return -KEY_sin;
5409 if (strEQ(d,"sleep")) return -KEY_sleep;
5412 if (strEQ(d,"sort")) return KEY_sort;
5413 if (strEQ(d,"socket")) return -KEY_socket;
5414 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5417 if (strEQ(d,"split")) return KEY_split;
5418 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5419 if (strEQ(d,"splice")) return KEY_splice;
5422 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5425 if (strEQ(d,"srand")) return -KEY_srand;
5428 if (strEQ(d,"stat")) return -KEY_stat;
5429 if (strEQ(d,"study")) return KEY_study;
5432 if (strEQ(d,"substr")) return -KEY_substr;
5433 if (strEQ(d,"sub")) return KEY_sub;
5438 if (strEQ(d,"system")) return -KEY_system;
5441 if (strEQ(d,"symlink")) return -KEY_symlink;
5442 if (strEQ(d,"syscall")) return -KEY_syscall;
5443 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5444 if (strEQ(d,"sysread")) return -KEY_sysread;
5445 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5448 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5457 if (strEQ(d,"tr")) return KEY_tr;
5460 if (strEQ(d,"tie")) return KEY_tie;
5463 if (strEQ(d,"tell")) return -KEY_tell;
5464 if (strEQ(d,"tied")) return KEY_tied;
5465 if (strEQ(d,"time")) return -KEY_time;
5468 if (strEQ(d,"times")) return -KEY_times;
5471 if (strEQ(d,"telldir")) return -KEY_telldir;
5474 if (strEQ(d,"truncate")) return -KEY_truncate;
5481 if (strEQ(d,"uc")) return -KEY_uc;
5484 if (strEQ(d,"use")) return KEY_use;
5487 if (strEQ(d,"undef")) return KEY_undef;
5488 if (strEQ(d,"until")) return KEY_until;
5489 if (strEQ(d,"untie")) return KEY_untie;
5490 if (strEQ(d,"utime")) return -KEY_utime;
5491 if (strEQ(d,"umask")) return -KEY_umask;
5494 if (strEQ(d,"unless")) return KEY_unless;
5495 if (strEQ(d,"unpack")) return -KEY_unpack;
5496 if (strEQ(d,"unlink")) return -KEY_unlink;
5499 if (strEQ(d,"unshift")) return KEY_unshift;
5500 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5505 if (strEQ(d,"values")) return -KEY_values;
5506 if (strEQ(d,"vec")) return -KEY_vec;
5511 if (strEQ(d,"warn")) return -KEY_warn;
5512 if (strEQ(d,"wait")) return -KEY_wait;
5515 if (strEQ(d,"while")) return KEY_while;
5516 if (strEQ(d,"write")) return -KEY_write;
5519 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5522 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5527 if (len == 1) return -KEY_x;
5528 if (strEQ(d,"xor")) return -KEY_xor;
5531 if (len == 1) return KEY_y;
5540 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5544 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5545 dTHR; /* only for ckWARN */
5546 if (ckWARN(WARN_SYNTAX)) {
5548 for (w = s+2; *w && level; w++) {
5555 for (; *w && isSPACE(*w); w++) ;
5556 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5557 Perl_warner(aTHX_ WARN_SYNTAX,
5558 "%s (...) interpreted as function",name);
5561 while (s < PL_bufend && isSPACE(*s))
5565 while (s < PL_bufend && isSPACE(*s))
5567 if (isIDFIRST_lazy_if(s,UTF)) {
5569 while (isALNUM_lazy_if(s,UTF))
5571 while (s < PL_bufend && isSPACE(*s))
5576 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5580 Perl_croak(aTHX_ "No comma allowed after %s", what);
5585 /* Either returns sv, or mortalizes sv and returns a new SV*.
5586 Best used as sv=new_constant(..., sv, ...).
5587 If s, pv are NULL, calls subroutine with one argument,
5588 and type is used with error messages only. */
5591 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5595 HV *table = GvHV(PL_hintgv); /* ^H */
5599 const char *why, *why1, *why2;
5601 if (!(PL_hints & HINT_LOCALIZE_HH)) {
5604 why = "%^H is not localized";
5608 msg = Perl_newSVpvf(aTHX_ "constant(%s): %s%s%s",
5609 (type ? type: "undef"), why1, why2, why);
5610 yyerror(SvPVX(msg));
5615 why = "%^H is not defined";
5618 cvp = hv_fetch(table, key, strlen(key), FALSE);
5619 if (!cvp || !SvOK(*cvp)) {
5620 why = "} is not defined";
5625 sv_2mortal(sv); /* Parent created it permanently */
5628 pv = sv_2mortal(newSVpvn(s, len));
5630 typesv = sv_2mortal(newSVpv(type, 0));
5632 typesv = &PL_sv_undef;
5634 PUSHSTACKi(PERLSI_OVERLOAD);
5647 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5651 /* Check the eval first */
5652 if (!PL_in_eval && SvTRUE(ERRSV))
5655 sv_catpv(ERRSV, "Propagated");
5656 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5658 res = SvREFCNT_inc(sv);
5662 (void)SvREFCNT_inc(res);
5671 why = "}} did not return a defined value";
5672 why1 = "Call to &{$^H{";
5682 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5684 register char *d = dest;
5685 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5688 Perl_croak(aTHX_ ident_too_long);
5689 if (isALNUM(*s)) /* UTF handled below */
5691 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
5696 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5700 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5701 char *t = s + UTF8SKIP(s);
5702 while (*t & 0x80 && is_utf8_mark((U8*)t))
5704 if (d + (t - s) > e)
5705 Perl_croak(aTHX_ ident_too_long);
5706 Copy(s, d, t - s, char);
5719 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5729 e = d + destlen - 3; /* two-character token, ending NUL */
5731 while (isDIGIT(*s)) {
5733 Perl_croak(aTHX_ ident_too_long);
5740 Perl_croak(aTHX_ ident_too_long);
5741 if (isALNUM(*s)) /* UTF handled below */
5743 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
5748 else if (*s == ':' && s[1] == ':') {
5752 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5753 char *t = s + UTF8SKIP(s);
5754 while (*t & 0x80 && is_utf8_mark((U8*)t))
5756 if (d + (t - s) > e)
5757 Perl_croak(aTHX_ ident_too_long);
5758 Copy(s, d, t - s, char);
5769 if (PL_lex_state != LEX_NORMAL)
5770 PL_lex_state = LEX_INTERPENDMAYBE;
5773 if (*s == '$' && s[1] &&
5774 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
5787 if (*d == '^' && *s && isCONTROLVAR(*s)) {
5792 if (isSPACE(s[-1])) {
5795 if (ch != ' ' && ch != '\t') {
5801 if (isIDFIRST_lazy_if(d,UTF)) {
5805 while (e < send && isALNUM_lazy_if(e,UTF) || *e == ':') {
5807 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
5810 Copy(s, d, e - s, char);
5815 while ((isALNUM(*s) || *s == ':') && d < e)
5818 Perl_croak(aTHX_ ident_too_long);
5821 while (s < send && (*s == ' ' || *s == '\t')) s++;
5822 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
5823 dTHR; /* only for ckWARN */
5824 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
5825 const char *brack = *s == '[' ? "[...]" : "{...}";
5826 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5827 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
5828 funny, dest, brack, funny, dest, brack);
5831 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
5835 /* Handle extended ${^Foo} variables
5836 * 1999-02-27 mjd-perl-patch@plover.com */
5837 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
5841 while (isALNUM(*s) && d < e) {
5845 Perl_croak(aTHX_ ident_too_long);
5850 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
5851 PL_lex_state = LEX_INTERPEND;
5854 if (PL_lex_state == LEX_NORMAL) {
5855 dTHR; /* only for ckWARN */
5856 if (ckWARN(WARN_AMBIGUOUS) &&
5857 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
5859 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5860 "Ambiguous use of %c{%s} resolved to %c%s",
5861 funny, dest, funny, dest);
5866 s = bracket; /* let the parser handle it */
5870 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
5871 PL_lex_state = LEX_INTERPEND;
5876 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
5881 *pmfl |= PMf_GLOBAL;
5883 *pmfl |= PMf_CONTINUE;
5887 *pmfl |= PMf_MULTILINE;
5889 *pmfl |= PMf_SINGLELINE;
5891 *pmfl |= PMf_EXTENDED;
5895 S_scan_pat(pTHX_ char *start, I32 type)
5900 s = scan_str(start,FALSE,FALSE);
5903 SvREFCNT_dec(PL_lex_stuff);
5904 PL_lex_stuff = Nullsv;
5905 Perl_croak(aTHX_ "Search pattern not terminated");
5908 pm = (PMOP*)newPMOP(type, 0);
5909 if (PL_multi_open == '?')
5910 pm->op_pmflags |= PMf_ONCE;
5912 while (*s && strchr("iomsx", *s))
5913 pmflag(&pm->op_pmflags,*s++);
5916 while (*s && strchr("iogcmsx", *s))
5917 pmflag(&pm->op_pmflags,*s++);
5919 pm->op_pmpermflags = pm->op_pmflags;
5921 PL_lex_op = (OP*)pm;
5922 yylval.ival = OP_MATCH;
5927 S_scan_subst(pTHX_ char *start)
5934 yylval.ival = OP_NULL;
5936 s = scan_str(start,FALSE,FALSE);
5940 SvREFCNT_dec(PL_lex_stuff);
5941 PL_lex_stuff = Nullsv;
5942 Perl_croak(aTHX_ "Substitution pattern not terminated");
5945 if (s[-1] == PL_multi_open)
5948 first_start = PL_multi_start;
5949 s = scan_str(s,FALSE,FALSE);
5952 SvREFCNT_dec(PL_lex_stuff);
5953 PL_lex_stuff = Nullsv;
5955 SvREFCNT_dec(PL_lex_repl);
5956 PL_lex_repl = Nullsv;
5957 Perl_croak(aTHX_ "Substitution replacement not terminated");
5959 PL_multi_start = first_start; /* so whole substitution is taken together */
5961 pm = (PMOP*)newPMOP(OP_SUBST, 0);
5967 else if (strchr("iogcmsx", *s))
5968 pmflag(&pm->op_pmflags,*s++);
5975 PL_sublex_info.super_bufptr = s;
5976 PL_sublex_info.super_bufend = PL_bufend;
5978 pm->op_pmflags |= PMf_EVAL;
5979 repl = newSVpvn("",0);
5981 sv_catpv(repl, es ? "eval " : "do ");
5982 sv_catpvn(repl, "{ ", 2);
5983 sv_catsv(repl, PL_lex_repl);
5984 sv_catpvn(repl, " };", 2);
5986 SvREFCNT_dec(PL_lex_repl);
5990 pm->op_pmpermflags = pm->op_pmflags;
5991 PL_lex_op = (OP*)pm;
5992 yylval.ival = OP_SUBST;
5997 S_scan_trans(pTHX_ char *start)
6008 yylval.ival = OP_NULL;
6010 s = scan_str(start,FALSE,FALSE);
6013 SvREFCNT_dec(PL_lex_stuff);
6014 PL_lex_stuff = Nullsv;
6015 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6017 if (s[-1] == PL_multi_open)
6020 s = scan_str(s,FALSE,FALSE);
6023 SvREFCNT_dec(PL_lex_stuff);
6024 PL_lex_stuff = Nullsv;
6026 SvREFCNT_dec(PL_lex_repl);
6027 PL_lex_repl = Nullsv;
6028 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6032 o = newSVOP(OP_TRANS, 0, 0);
6033 utf8 = OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF;
6036 New(803,tbl,256,short);
6037 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6041 complement = del = squash = 0;
6042 while (strchr("cdsCU", *s)) {
6044 complement = OPpTRANS_COMPLEMENT;
6046 del = OPpTRANS_DELETE;
6048 squash = OPpTRANS_SQUASH;
6053 utf8 &= ~OPpTRANS_FROM_UTF;
6055 utf8 |= OPpTRANS_FROM_UTF;
6059 utf8 &= ~OPpTRANS_TO_UTF;
6061 utf8 |= OPpTRANS_TO_UTF;
6064 Perl_croak(aTHX_ "Too many /C and /U options");
6069 o->op_private = del|squash|complement|utf8;
6072 yylval.ival = OP_TRANS;
6077 S_scan_heredoc(pTHX_ register char *s)
6081 I32 op_type = OP_SCALAR;
6088 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6092 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6095 for (peek = s; *peek == ' ' || *peek == '\t'; peek++) ;
6096 if (*peek && strchr("`'\"",*peek)) {
6099 s = delimcpy(d, e, s, PL_bufend, term, &len);
6109 if (!isALNUM_lazy_if(s,UTF))
6110 deprecate("bare << to mean <<\"\"");
6111 for (; isALNUM_lazy_if(s,UTF); s++) {
6116 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6117 Perl_croak(aTHX_ "Delimiter for here document is too long");
6120 len = d - PL_tokenbuf;
6121 #ifndef PERL_STRICT_CR
6122 d = strchr(s, '\r');
6126 while (s < PL_bufend) {
6132 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6141 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6146 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6147 herewas = newSVpvn(s,PL_bufend-s);
6149 s--, herewas = newSVpvn(s,d-s);
6150 s += SvCUR(herewas);
6152 tmpstr = NEWSV(87,79);
6153 sv_upgrade(tmpstr, SVt_PVIV);
6158 else if (term == '`') {
6159 op_type = OP_BACKTICK;
6160 SvIVX(tmpstr) = '\\';
6164 PL_multi_start = CopLINE(PL_curcop);
6165 PL_multi_open = PL_multi_close = '<';
6166 term = *PL_tokenbuf;
6167 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6168 char *bufptr = PL_sublex_info.super_bufptr;
6169 char *bufend = PL_sublex_info.super_bufend;
6170 char *olds = s - SvCUR(herewas);
6171 s = strchr(bufptr, '\n');
6175 while (s < bufend &&
6176 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6178 CopLINE_inc(PL_curcop);
6181 CopLINE_set(PL_curcop, PL_multi_start);
6182 missingterm(PL_tokenbuf);
6184 sv_setpvn(herewas,bufptr,d-bufptr+1);
6185 sv_setpvn(tmpstr,d+1,s-d);
6187 sv_catpvn(herewas,s,bufend-s);
6188 (void)strcpy(bufptr,SvPVX(herewas));
6195 while (s < PL_bufend &&
6196 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6198 CopLINE_inc(PL_curcop);
6200 if (s >= PL_bufend) {
6201 CopLINE_set(PL_curcop, PL_multi_start);
6202 missingterm(PL_tokenbuf);
6204 sv_setpvn(tmpstr,d+1,s-d);
6206 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6208 sv_catpvn(herewas,s,PL_bufend-s);
6209 sv_setsv(PL_linestr,herewas);
6210 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6211 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6214 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6215 while (s >= PL_bufend) { /* multiple line string? */
6217 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6218 CopLINE_set(PL_curcop, PL_multi_start);
6219 missingterm(PL_tokenbuf);
6221 CopLINE_inc(PL_curcop);
6222 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6223 #ifndef PERL_STRICT_CR
6224 if (PL_bufend - PL_linestart >= 2) {
6225 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6226 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6228 PL_bufend[-2] = '\n';
6230 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6232 else if (PL_bufend[-1] == '\r')
6233 PL_bufend[-1] = '\n';
6235 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6236 PL_bufend[-1] = '\n';
6238 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6239 SV *sv = NEWSV(88,0);
6241 sv_upgrade(sv, SVt_PVMG);
6242 sv_setsv(sv,PL_linestr);
6243 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6245 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6248 sv_catsv(PL_linestr,herewas);
6249 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6253 sv_catsv(tmpstr,PL_linestr);
6258 PL_multi_end = CopLINE(PL_curcop);
6259 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6260 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6261 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6263 SvREFCNT_dec(herewas);
6264 PL_lex_stuff = tmpstr;
6265 yylval.ival = op_type;
6270 takes: current position in input buffer
6271 returns: new position in input buffer
6272 side-effects: yylval and lex_op are set.
6277 <FH> read from filehandle
6278 <pkg::FH> read from package qualified filehandle
6279 <pkg'FH> read from package qualified filehandle
6280 <$fh> read from filehandle in $fh
6286 S_scan_inputsymbol(pTHX_ char *start)
6288 register char *s = start; /* current position in buffer */
6294 d = PL_tokenbuf; /* start of temp holding space */
6295 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6296 end = strchr(s, '\n');
6299 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6301 /* die if we didn't have space for the contents of the <>,
6302 or if it didn't end, or if we see a newline
6305 if (len >= sizeof PL_tokenbuf)
6306 Perl_croak(aTHX_ "Excessively long <> operator");
6308 Perl_croak(aTHX_ "Unterminated <> operator");
6313 Remember, only scalar variables are interpreted as filehandles by
6314 this code. Anything more complex (e.g., <$fh{$num}>) will be
6315 treated as a glob() call.
6316 This code makes use of the fact that except for the $ at the front,
6317 a scalar variable and a filehandle look the same.
6319 if (*d == '$' && d[1]) d++;
6321 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6322 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6325 /* If we've tried to read what we allow filehandles to look like, and
6326 there's still text left, then it must be a glob() and not a getline.
6327 Use scan_str to pull out the stuff between the <> and treat it
6328 as nothing more than a string.
6331 if (d - PL_tokenbuf != len) {
6332 yylval.ival = OP_GLOB;
6334 s = scan_str(start,FALSE,FALSE);
6336 Perl_croak(aTHX_ "Glob not terminated");
6340 /* we're in a filehandle read situation */
6343 /* turn <> into <ARGV> */
6345 (void)strcpy(d,"ARGV");
6347 /* if <$fh>, create the ops to turn the variable into a
6353 /* try to find it in the pad for this block, otherwise find
6354 add symbol table ops
6356 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6357 OP *o = newOP(OP_PADSV, 0);
6359 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6362 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6363 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6364 newUNOP(OP_RV2SV, 0,
6365 newGVOP(OP_GV, 0, gv)));
6367 PL_lex_op->op_flags |= OPf_SPECIAL;
6368 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6369 yylval.ival = OP_NULL;
6372 /* If it's none of the above, it must be a literal filehandle
6373 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6375 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6376 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6377 yylval.ival = OP_NULL;
6386 takes: start position in buffer
6387 keep_quoted preserve \ on the embedded delimiter(s)
6388 keep_delims preserve the delimiters around the string
6389 returns: position to continue reading from buffer
6390 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6391 updates the read buffer.
6393 This subroutine pulls a string out of the input. It is called for:
6394 q single quotes q(literal text)
6395 ' single quotes 'literal text'
6396 qq double quotes qq(interpolate $here please)
6397 " double quotes "interpolate $here please"
6398 qx backticks qx(/bin/ls -l)
6399 ` backticks `/bin/ls -l`
6400 qw quote words @EXPORT_OK = qw( func() $spam )
6401 m// regexp match m/this/
6402 s/// regexp substitute s/this/that/
6403 tr/// string transliterate tr/this/that/
6404 y/// string transliterate y/this/that/
6405 ($*@) sub prototypes sub foo ($)
6406 (stuff) sub attr parameters sub foo : attr(stuff)
6407 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6409 In most of these cases (all but <>, patterns and transliterate)
6410 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6411 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6412 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6415 It skips whitespace before the string starts, and treats the first
6416 character as the delimiter. If the delimiter is one of ([{< then
6417 the corresponding "close" character )]}> is used as the closing
6418 delimiter. It allows quoting of delimiters, and if the string has
6419 balanced delimiters ([{<>}]) it allows nesting.
6421 The lexer always reads these strings into lex_stuff, except in the
6422 case of the operators which take *two* arguments (s/// and tr///)
6423 when it checks to see if lex_stuff is full (presumably with the 1st
6424 arg to s or tr) and if so puts the string into lex_repl.
6429 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6432 SV *sv; /* scalar value: string */
6433 char *tmps; /* temp string, used for delimiter matching */
6434 register char *s = start; /* current position in the buffer */
6435 register char term; /* terminating character */
6436 register char *to; /* current position in the sv's data */
6437 I32 brackets = 1; /* bracket nesting level */
6438 bool has_utf = FALSE; /* is there any utf8 content? */
6440 /* skip space before the delimiter */
6444 /* mark where we are, in case we need to report errors */
6447 /* after skipping whitespace, the next character is the terminator */
6449 if ((term & 0x80) && UTF)
6452 /* mark where we are */
6453 PL_multi_start = CopLINE(PL_curcop);
6454 PL_multi_open = term;
6456 /* find corresponding closing delimiter */
6457 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6459 PL_multi_close = term;
6461 /* create a new SV to hold the contents. 87 is leak category, I'm
6462 assuming. 79 is the SV's initial length. What a random number. */
6464 sv_upgrade(sv, SVt_PVIV);
6466 (void)SvPOK_only(sv); /* validate pointer */
6468 /* move past delimiter and try to read a complete string */
6470 sv_catpvn(sv, s, 1);
6473 /* extend sv if need be */
6474 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6475 /* set 'to' to the next character in the sv's string */
6476 to = SvPVX(sv)+SvCUR(sv);
6478 /* if open delimiter is the close delimiter read unbridle */
6479 if (PL_multi_open == PL_multi_close) {
6480 for (; s < PL_bufend; s++,to++) {
6481 /* embedded newlines increment the current line number */
6482 if (*s == '\n' && !PL_rsfp)
6483 CopLINE_inc(PL_curcop);
6484 /* handle quoted delimiters */
6485 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6486 if (!keep_quoted && s[1] == term)
6488 /* any other quotes are simply copied straight through */
6492 /* terminate when run out of buffer (the for() condition), or
6493 have found the terminator */
6494 else if (*s == term)
6496 else if (!has_utf && (*s & 0x80) && UTF)
6502 /* if the terminator isn't the same as the start character (e.g.,
6503 matched brackets), we have to allow more in the quoting, and
6504 be prepared for nested brackets.
6507 /* read until we run out of string, or we find the terminator */
6508 for (; s < PL_bufend; s++,to++) {
6509 /* embedded newlines increment the line count */
6510 if (*s == '\n' && !PL_rsfp)
6511 CopLINE_inc(PL_curcop);
6512 /* backslashes can escape the open or closing characters */
6513 if (*s == '\\' && s+1 < PL_bufend) {
6515 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6520 /* allow nested opens and closes */
6521 else if (*s == PL_multi_close && --brackets <= 0)
6523 else if (*s == PL_multi_open)
6525 else if (!has_utf && (*s & 0x80) && UTF)
6530 /* terminate the copied string and update the sv's end-of-string */
6532 SvCUR_set(sv, to - SvPVX(sv));
6535 * this next chunk reads more into the buffer if we're not done yet
6539 break; /* handle case where we are done yet :-) */
6541 #ifndef PERL_STRICT_CR
6542 if (to - SvPVX(sv) >= 2) {
6543 if ((to[-2] == '\r' && to[-1] == '\n') ||
6544 (to[-2] == '\n' && to[-1] == '\r'))
6548 SvCUR_set(sv, to - SvPVX(sv));
6550 else if (to[-1] == '\r')
6553 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6557 /* if we're out of file, or a read fails, bail and reset the current
6558 line marker so we can report where the unterminated string began
6561 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6563 CopLINE_set(PL_curcop, PL_multi_start);
6566 /* we read a line, so increment our line counter */
6567 CopLINE_inc(PL_curcop);
6569 /* update debugger info */
6570 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6571 SV *sv = NEWSV(88,0);
6573 sv_upgrade(sv, SVt_PVMG);
6574 sv_setsv(sv,PL_linestr);
6575 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6578 /* having changed the buffer, we must update PL_bufend */
6579 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6582 /* at this point, we have successfully read the delimited string */
6585 sv_catpvn(sv, s, 1);
6588 PL_multi_end = CopLINE(PL_curcop);
6591 /* if we allocated too much space, give some back */
6592 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6593 SvLEN_set(sv, SvCUR(sv) + 1);
6594 Renew(SvPVX(sv), SvLEN(sv), char);
6597 /* decide whether this is the first or second quoted string we've read
6610 takes: pointer to position in buffer
6611 returns: pointer to new position in buffer
6612 side-effects: builds ops for the constant in yylval.op
6614 Read a number in any of the formats that Perl accepts:
6616 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6617 [\d_]+(\.[\d_]*)?[Ee](\d+)
6619 Underbars (_) are allowed in decimal numbers. If -w is on,
6620 underbars before a decimal point must be at three digit intervals.
6622 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6625 If it reads a number without a decimal point or an exponent, it will
6626 try converting the number to an integer and see if it can do so
6627 without loss of precision.
6631 Perl_scan_num(pTHX_ char *start)
6633 register char *s = start; /* current position in buffer */
6634 register char *d; /* destination in temp buffer */
6635 register char *e; /* end of temp buffer */
6636 IV tryiv; /* used to see if it can be an IV */
6637 NV value; /* number read, as a double */
6638 SV *sv = Nullsv; /* place to put the converted number */
6639 bool floatit; /* boolean: int or float? */
6640 char *lastub = 0; /* position of last underbar */
6641 static char number_too_long[] = "Number too long";
6643 /* We use the first character to decide what type of number this is */
6647 Perl_croak(aTHX_ "panic: scan_num");
6649 /* if it starts with a 0, it could be an octal number, a decimal in
6650 0.13 disguise, or a hexadecimal number, or a binary number. */
6654 u holds the "number so far"
6655 shift the power of 2 of the base
6656 (hex == 4, octal == 3, binary == 1)
6657 overflowed was the number more than we can hold?
6659 Shift is used when we add a digit. It also serves as an "are
6660 we in octal/hex/binary?" indicator to disallow hex characters
6667 bool overflowed = FALSE;
6668 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6669 static char* bases[5] = { "", "binary", "", "octal",
6671 static char* Bases[5] = { "", "Binary", "", "Octal",
6673 static char *maxima[5] = { "",
6674 "0b11111111111111111111111111111111",
6678 char *base, *Base, *max;
6684 } else if (s[1] == 'b') {
6688 /* check for a decimal in disguise */
6689 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6691 /* so it must be octal */
6695 base = bases[shift];
6696 Base = Bases[shift];
6697 max = maxima[shift];
6699 /* read the rest of the number */
6701 /* x is used in the overflow test,
6702 b is the digit we're adding on. */
6707 /* if we don't mention it, we're done */
6716 /* 8 and 9 are not octal */
6719 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6723 case '2': case '3': case '4':
6724 case '5': case '6': case '7':
6726 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6730 b = *s++ & 15; /* ASCII digit -> value of digit */
6734 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6735 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6736 /* make sure they said 0x */
6741 /* Prepare to put the digit we have onto the end
6742 of the number so far. We check for overflows.
6747 x = u << shift; /* make room for the digit */
6749 if ((x >> shift) != u
6750 && !(PL_hints & HINT_NEW_BINARY)) {
6754 if (ckWARN_d(WARN_OVERFLOW))
6755 Perl_warner(aTHX_ WARN_OVERFLOW,
6756 "Integer overflow in %s number",
6759 u = x | b; /* add the digit to the end */
6762 n *= nvshift[shift];
6763 /* If an NV has not enough bits in its
6764 * mantissa to represent an UV this summing of
6765 * small low-order numbers is a waste of time
6766 * (because the NV cannot preserve the
6767 * low-order bits anyway): we could just
6768 * remember when did we overflow and in the
6769 * end just multiply n by the right
6777 /* if we get here, we had success: make a scalar value from
6784 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6785 Perl_warner(aTHX_ WARN_PORTABLE,
6786 "%s number > %s non-portable",
6793 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6794 Perl_warner(aTHX_ WARN_PORTABLE,
6795 "%s number > %s non-portable",
6800 if (PL_hints & HINT_NEW_BINARY)
6801 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6806 handle decimal numbers.
6807 we're also sent here when we read a 0 as the first digit
6809 case '1': case '2': case '3': case '4': case '5':
6810 case '6': case '7': case '8': case '9': case '.':
6813 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
6816 /* read next group of digits and _ and copy into d */
6817 while (isDIGIT(*s) || *s == '_') {
6818 /* skip underscores, checking for misplaced ones
6822 dTHR; /* only for ckWARN */
6823 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
6824 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6828 /* check for end of fixed-length buffer */
6830 Perl_croak(aTHX_ number_too_long);
6831 /* if we're ok, copy the character */
6836 /* final misplaced underbar check */
6837 if (lastub && s - lastub != 3) {
6839 if (ckWARN(WARN_SYNTAX))
6840 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6843 /* read a decimal portion if there is one. avoid
6844 3..5 being interpreted as the number 3. followed
6847 if (*s == '.' && s[1] != '.') {
6851 /* copy, ignoring underbars, until we run out of
6852 digits. Note: no misplaced underbar checks!
6854 for (; isDIGIT(*s) || *s == '_'; s++) {
6855 /* fixed length buffer check */
6857 Perl_croak(aTHX_ number_too_long);
6863 /* read exponent part, if present */
6864 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
6868 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
6869 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
6871 /* allow positive or negative exponent */
6872 if (*s == '+' || *s == '-')
6875 /* read digits of exponent (no underbars :-) */
6876 while (isDIGIT(*s)) {
6878 Perl_croak(aTHX_ number_too_long);
6883 /* terminate the string */
6886 /* make an sv from the string */
6889 value = Atof(PL_tokenbuf);
6892 See if we can make do with an integer value without loss of
6893 precision. We use I_V to cast to an int, because some
6894 compilers have issues. Then we try casting it back and see
6895 if it was the same. We only do this if we know we
6896 specifically read an integer.
6898 Note: if floatit is true, then we don't need to do the
6902 if (!floatit && (NV)tryiv == value)
6903 sv_setiv(sv, tryiv);
6905 sv_setnv(sv, value);
6906 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
6907 (PL_hints & HINT_NEW_INTEGER) )
6908 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
6909 (floatit ? "float" : "integer"),
6912 /* if it starts with a v, it could be a version number */
6917 while (isDIGIT(*pos))
6919 if (!isALPHA(*pos)) {
6921 U8 tmpbuf[UTF8_MAXLEN];
6924 s++; /* get past 'v' */
6927 sv_setpvn(sv, "", 0);
6930 if (*s == '0' && isDIGIT(s[1]))
6931 yyerror("Octal number in vector unsupported");
6933 tmpend = uv_to_utf8(tmpbuf, rev);
6934 utf8 = utf8 || rev > 127;
6935 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
6936 if (*pos == '.' && isDIGIT(pos[1]))
6942 while (isDIGIT(*pos))
6950 sv_utf8_downgrade(sv, TRUE);
6957 /* make the op for the constant and return */
6960 yylval.opval = newSVOP(OP_CONST, 0, sv);
6962 yylval.opval = Nullop;
6968 S_scan_formline(pTHX_ register char *s)
6973 SV *stuff = newSVpvn("",0);
6974 bool needargs = FALSE;
6977 if (*s == '.' || *s == '}') {
6979 #ifdef PERL_STRICT_CR
6980 for (t = s+1;*t == ' ' || *t == '\t'; t++) ;
6982 for (t = s+1;*t == ' ' || *t == '\t' || *t == '\r'; t++) ;
6984 if (*t == '\n' || t == PL_bufend)
6987 if (PL_in_eval && !PL_rsfp) {
6988 eol = strchr(s,'\n');
6993 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6995 for (t = s; t < eol; t++) {
6996 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
6998 goto enough; /* ~~ must be first line in formline */
7000 if (*t == '@' || *t == '^')
7003 sv_catpvn(stuff, s, eol-s);
7004 #ifndef PERL_STRICT_CR
7005 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7006 char *end = SvPVX(stuff) + SvCUR(stuff);
7015 s = filter_gets(PL_linestr, PL_rsfp, 0);
7016 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7017 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7020 yyerror("Format not terminated");
7030 PL_lex_state = LEX_NORMAL;
7031 PL_nextval[PL_nexttoke].ival = 0;
7035 PL_lex_state = LEX_FORMLINE;
7036 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7038 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7042 SvREFCNT_dec(stuff);
7043 PL_lex_formbrack = 0;
7054 PL_cshlen = strlen(PL_cshname);
7059 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7062 I32 oldsavestack_ix = PL_savestack_ix;
7063 CV* outsidecv = PL_compcv;
7067 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7069 SAVEI32(PL_subline);
7070 save_item(PL_subname);
7073 SAVESPTR(PL_comppad_name);
7074 SAVESPTR(PL_compcv);
7075 SAVEI32(PL_comppad_name_fill);
7076 SAVEI32(PL_min_intro_pending);
7077 SAVEI32(PL_max_intro_pending);
7078 SAVEI32(PL_pad_reset_pending);
7080 PL_compcv = (CV*)NEWSV(1104,0);
7081 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7082 CvFLAGS(PL_compcv) |= flags;
7084 PL_comppad = newAV();
7085 av_push(PL_comppad, Nullsv);
7086 PL_curpad = AvARRAY(PL_comppad);
7087 PL_comppad_name = newAV();
7088 PL_comppad_name_fill = 0;
7089 PL_min_intro_pending = 0;
7091 PL_subline = CopLINE(PL_curcop);
7093 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7094 PL_curpad[0] = (SV*)newAV();
7095 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7096 #endif /* USE_THREADS */
7098 comppadlist = newAV();
7099 AvREAL_off(comppadlist);
7100 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7101 av_store(comppadlist, 1, (SV*)PL_comppad);
7103 CvPADLIST(PL_compcv) = comppadlist;
7104 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7106 CvOWNER(PL_compcv) = 0;
7107 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7108 MUTEX_INIT(CvMUTEXP(PL_compcv));
7109 #endif /* USE_THREADS */
7111 return oldsavestack_ix;
7115 Perl_yywarn(pTHX_ char *s)
7118 PL_in_eval |= EVAL_WARNONLY;
7120 PL_in_eval &= ~EVAL_WARNONLY;
7125 Perl_yyerror(pTHX_ char *s)
7129 char *context = NULL;
7133 if (!yychar || (yychar == ';' && !PL_rsfp))
7135 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7136 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7137 while (isSPACE(*PL_oldoldbufptr))
7139 context = PL_oldoldbufptr;
7140 contlen = PL_bufptr - PL_oldoldbufptr;
7142 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7143 PL_oldbufptr != PL_bufptr) {
7144 while (isSPACE(*PL_oldbufptr))
7146 context = PL_oldbufptr;
7147 contlen = PL_bufptr - PL_oldbufptr;
7149 else if (yychar > 255)
7150 where = "next token ???";
7151 #ifdef USE_PURE_BISON
7152 /* GNU Bison sets the value -2 */
7153 else if (yychar == -2) {
7155 else if ((yychar & 127) == 127) {
7157 if (PL_lex_state == LEX_NORMAL ||
7158 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7159 where = "at end of line";
7160 else if (PL_lex_inpat)
7161 where = "within pattern";
7163 where = "within string";
7166 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7168 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7169 else if (isPRINT_LC(yychar))
7170 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7172 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7173 where = SvPVX(where_sv);
7175 msg = sv_2mortal(newSVpv(s, 0));
7176 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7177 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7179 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7181 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7182 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7183 Perl_sv_catpvf(aTHX_ msg,
7184 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7185 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7188 if (PL_in_eval & EVAL_WARNONLY)
7189 Perl_warn(aTHX_ "%"SVf, msg);
7192 if (PL_error_count >= 10)
7193 Perl_croak(aTHX_ "%s has too many errors.\n", CopFILE(PL_curcop));
7195 PL_in_my_stash = Nullhv;
7206 * Restore a source filter.
7210 restore_rsfp(pTHXo_ void *f)
7212 PerlIO *fp = (PerlIO*)f;
7214 if (PL_rsfp == PerlIO_stdin())
7215 PerlIO_clearerr(PL_rsfp);
7216 else if (PL_rsfp && (PL_rsfp != fp))
7217 PerlIO_close(PL_rsfp);