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 == XTERM) {
3474 while (isDIGIT(*start))
3476 if (*start == '.' && isDIGIT(start[1])) {
3483 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3523 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3525 /* Some keywords can be followed by any delimiter, including ':' */
3526 tmp = (len == 1 && strchr("msyq", PL_tokenbuf[0]) ||
3527 len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3528 (PL_tokenbuf[0] == 'q' &&
3529 strchr("qwxr", PL_tokenbuf[1]))));
3531 /* x::* is just a word, unless x is "CORE" */
3532 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3536 while (d < PL_bufend && isSPACE(*d))
3537 d++; /* no comments skipped here, or s### is misparsed */
3539 /* Is this a label? */
3540 if (!tmp && PL_expect == XSTATE
3541 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3543 yylval.pval = savepv(PL_tokenbuf);
3548 /* Check for keywords */
3549 tmp = keyword(PL_tokenbuf, len);
3551 /* Is this a word before a => operator? */
3552 if (strnEQ(d,"=>",2)) {
3554 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3555 yylval.opval->op_private = OPpCONST_BARE;
3559 if (tmp < 0) { /* second-class keyword? */
3560 GV *ogv = Nullgv; /* override (winner) */
3561 GV *hgv = Nullgv; /* hidden (loser) */
3562 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3564 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3567 if (GvIMPORTED_CV(gv))
3569 else if (! CvMETHOD(cv))
3573 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3574 (gv = *gvp) != (GV*)&PL_sv_undef &&
3575 GvCVu(gv) && GvIMPORTED_CV(gv))
3581 tmp = 0; /* overridden by import or by GLOBAL */
3584 && -tmp==KEY_lock /* XXX generalizable kludge */
3586 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3588 tmp = 0; /* any sub overrides "weak" keyword */
3590 else { /* no override */
3594 if (ckWARN(WARN_AMBIGUOUS) && hgv
3595 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3596 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3597 "Ambiguous call resolved as CORE::%s(), %s",
3598 GvENAME(hgv), "qualify as such or use &");
3605 default: /* not a keyword */
3608 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3610 /* Get the rest if it looks like a package qualifier */
3612 if (*s == '\'' || *s == ':' && s[1] == ':') {
3614 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3617 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3618 *s == '\'' ? "'" : "::");
3622 if (PL_expect == XOPERATOR) {
3623 if (PL_bufptr == PL_linestart) {
3624 CopLINE_dec(PL_curcop);
3625 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3626 CopLINE_inc(PL_curcop);
3629 no_op("Bareword",s);
3632 /* Look for a subroutine with this name in current package,
3633 unless name is "Foo::", in which case Foo is a bearword
3634 (and a package name). */
3637 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3639 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3640 Perl_warner(aTHX_ WARN_BAREWORD,
3641 "Bareword \"%s\" refers to nonexistent package",
3644 PL_tokenbuf[len] = '\0';
3651 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3654 /* if we saw a global override before, get the right name */
3657 sv = newSVpvn("CORE::GLOBAL::",14);
3658 sv_catpv(sv,PL_tokenbuf);
3661 sv = newSVpv(PL_tokenbuf,0);
3663 /* Presume this is going to be a bareword of some sort. */
3666 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3667 yylval.opval->op_private = OPpCONST_BARE;
3669 /* And if "Foo::", then that's what it certainly is. */
3674 /* See if it's the indirect object for a list operator. */
3676 if (PL_oldoldbufptr &&
3677 PL_oldoldbufptr < PL_bufptr &&
3678 (PL_oldoldbufptr == PL_last_lop
3679 || PL_oldoldbufptr == PL_last_uni) &&
3680 /* NO SKIPSPACE BEFORE HERE! */
3681 (PL_expect == XREF ||
3682 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3684 bool immediate_paren = *s == '(';
3686 /* (Now we can afford to cross potential line boundary.) */
3689 /* Two barewords in a row may indicate method call. */
3691 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3694 /* If not a declared subroutine, it's an indirect object. */
3695 /* (But it's an indir obj regardless for sort.) */
3697 if ((PL_last_lop_op == OP_SORT ||
3698 (!immediate_paren && (!gv || !GvCVu(gv)))) &&
3699 (PL_last_lop_op != OP_MAPSTART &&
3700 PL_last_lop_op != OP_GREPSTART))
3702 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3707 /* If followed by a paren, it's certainly a subroutine. */
3709 PL_expect = XOPERATOR;
3713 if (gv && GvCVu(gv)) {
3714 for (d = s + 1; *d == ' ' || *d == '\t'; d++) ;
3715 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3720 PL_nextval[PL_nexttoke].opval = yylval.opval;
3721 PL_expect = XOPERATOR;
3727 /* If followed by var or block, call it a method (unless sub) */
3729 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3730 PL_last_lop = PL_oldbufptr;
3731 PL_last_lop_op = OP_METHOD;
3735 /* If followed by a bareword, see if it looks like indir obj. */
3737 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3740 /* Not a method, so call it a subroutine (if defined) */
3742 if (gv && GvCVu(gv)) {
3744 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3745 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3746 "Ambiguous use of -%s resolved as -&%s()",
3747 PL_tokenbuf, PL_tokenbuf);
3748 /* Check for a constant sub */
3750 if ((sv = cv_const_sv(cv))) {
3752 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3753 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3754 yylval.opval->op_private = 0;
3758 /* Resolve to GV now. */
3759 op_free(yylval.opval);
3760 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3761 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3762 PL_last_lop = PL_oldbufptr;
3763 PL_last_lop_op = OP_ENTERSUB;
3764 /* Is there a prototype? */
3767 char *proto = SvPV((SV*)cv, len);
3770 if (strEQ(proto, "$"))
3772 if (*proto == '&' && *s == '{') {
3773 sv_setpv(PL_subname,"__ANON__");
3777 PL_nextval[PL_nexttoke].opval = yylval.opval;
3783 /* Call it a bare word */
3785 if (PL_hints & HINT_STRICT_SUBS)
3786 yylval.opval->op_private |= OPpCONST_STRICT;
3789 if (ckWARN(WARN_RESERVED)) {
3790 if (lastchar != '-') {
3791 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
3793 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
3800 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
3801 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3802 "Operator or semicolon missing before %c%s",
3803 lastchar, PL_tokenbuf);
3804 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3805 "Ambiguous use of %c resolved as operator %c",
3806 lastchar, lastchar);
3812 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3813 newSVpv(CopFILE(PL_curcop),0));
3817 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3818 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
3821 case KEY___PACKAGE__:
3822 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3824 ? newSVsv(PL_curstname)
3833 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
3834 char *pname = "main";
3835 if (PL_tokenbuf[2] == 'D')
3836 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
3837 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
3840 GvIOp(gv) = newIO();
3841 IoIFP(GvIOp(gv)) = PL_rsfp;
3842 #if defined(HAS_FCNTL) && defined(F_SETFD)
3844 int fd = PerlIO_fileno(PL_rsfp);
3845 fcntl(fd,F_SETFD,fd >= 3);
3848 /* Mark this internal pseudo-handle as clean */
3849 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
3851 IoTYPE(GvIOp(gv)) = '|';
3852 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
3853 IoTYPE(GvIOp(gv)) = '-';
3855 IoTYPE(GvIOp(gv)) = '<';
3856 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
3857 /* if the script was opened in binmode, we need to revert
3858 * it to text mode for compatibility; but only iff it has CRs
3859 * XXX this is a questionable hack at best. */
3860 if (PL_bufend-PL_bufptr > 2
3861 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
3864 if (IoTYPE(GvIOp(gv)) == '<') {
3865 loc = PerlIO_tell(PL_rsfp);
3866 (void)PerlIO_seek(PL_rsfp, 0L, 0);
3868 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
3869 #if defined(__BORLANDC__)
3870 /* XXX see note in do_binmode() */
3871 ((FILE*)PL_rsfp)->flags |= _F_BIN;
3874 PerlIO_seek(PL_rsfp, loc, 0);
3889 if (PL_expect == XSTATE) {
3896 if (*s == ':' && s[1] == ':') {
3899 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3900 tmp = keyword(PL_tokenbuf, len);
3914 LOP(OP_ACCEPT,XTERM);
3920 LOP(OP_ATAN2,XTERM);
3929 LOP(OP_BLESS,XTERM);
3938 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
3955 if (!PL_cryptseen) {
3956 PL_cryptseen = TRUE;
3960 LOP(OP_CRYPT,XTERM);
3963 if (ckWARN(WARN_CHMOD)) {
3964 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
3965 if (*d != '0' && isDIGIT(*d))
3966 Perl_warner(aTHX_ WARN_CHMOD,
3967 "chmod() mode argument is missing initial 0");
3969 LOP(OP_CHMOD,XTERM);
3972 LOP(OP_CHOWN,XTERM);
3975 LOP(OP_CONNECT,XTERM);
3991 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3995 PL_hints |= HINT_BLOCK_SCOPE;
4005 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4006 LOP(OP_DBMOPEN,XTERM);
4012 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4019 yylval.ival = CopLINE(PL_curcop);
4033 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4034 UNIBRACK(OP_ENTEREVAL);
4049 case KEY_endhostent:
4055 case KEY_endservent:
4058 case KEY_endprotoent:
4069 yylval.ival = CopLINE(PL_curcop);
4071 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4073 if ((PL_bufend - p) >= 3 &&
4074 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4076 else if ((PL_bufend - p) >= 4 &&
4077 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4080 if (isIDFIRST_lazy_if(p,UTF)) {
4081 p = scan_ident(p, PL_bufend,
4082 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4086 Perl_croak(aTHX_ "Missing $ on loop variable");
4091 LOP(OP_FORMLINE,XTERM);
4097 LOP(OP_FCNTL,XTERM);
4103 LOP(OP_FLOCK,XTERM);
4112 LOP(OP_GREPSTART, XREF);
4115 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4130 case KEY_getpriority:
4131 LOP(OP_GETPRIORITY,XTERM);
4133 case KEY_getprotobyname:
4136 case KEY_getprotobynumber:
4137 LOP(OP_GPBYNUMBER,XTERM);
4139 case KEY_getprotoent:
4151 case KEY_getpeername:
4152 UNI(OP_GETPEERNAME);
4154 case KEY_gethostbyname:
4157 case KEY_gethostbyaddr:
4158 LOP(OP_GHBYADDR,XTERM);
4160 case KEY_gethostent:
4163 case KEY_getnetbyname:
4166 case KEY_getnetbyaddr:
4167 LOP(OP_GNBYADDR,XTERM);
4172 case KEY_getservbyname:
4173 LOP(OP_GSBYNAME,XTERM);
4175 case KEY_getservbyport:
4176 LOP(OP_GSBYPORT,XTERM);
4178 case KEY_getservent:
4181 case KEY_getsockname:
4182 UNI(OP_GETSOCKNAME);
4184 case KEY_getsockopt:
4185 LOP(OP_GSOCKOPT,XTERM);
4207 yylval.ival = CopLINE(PL_curcop);
4211 LOP(OP_INDEX,XTERM);
4217 LOP(OP_IOCTL,XTERM);
4229 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4261 LOP(OP_LISTEN,XTERM);
4270 s = scan_pat(s,OP_MATCH);
4271 TERM(sublex_start());
4274 LOP(OP_MAPSTART, XREF);
4277 LOP(OP_MKDIR,XTERM);
4280 LOP(OP_MSGCTL,XTERM);
4283 LOP(OP_MSGGET,XTERM);
4286 LOP(OP_MSGRCV,XTERM);
4289 LOP(OP_MSGSND,XTERM);
4295 if (isIDFIRST_lazy_if(s,UTF)) {
4296 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4297 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4299 PL_in_my_stash = gv_stashpv(PL_tokenbuf, FALSE);
4300 if (!PL_in_my_stash) {
4303 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4311 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4318 if (PL_expect != XSTATE)
4319 yyerror("\"no\" not allowed in expression");
4320 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4321 s = force_version(s);
4326 if (*s == '(' || (s = skipspace(s), *s == '('))
4333 if (isIDFIRST_lazy_if(s,UTF)) {
4335 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4337 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4338 Perl_warner(aTHX_ WARN_PRECEDENCE,
4339 "Precedence problem: open %.*s should be open(%.*s)",
4345 yylval.ival = OP_OR;
4355 LOP(OP_OPEN_DIR,XTERM);
4358 checkcomma(s,PL_tokenbuf,"filehandle");
4362 checkcomma(s,PL_tokenbuf,"filehandle");
4381 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4385 LOP(OP_PIPE_OP,XTERM);
4388 s = scan_str(s,FALSE,FALSE);
4390 missingterm((char*)0);
4391 yylval.ival = OP_CONST;
4392 TERM(sublex_start());
4398 s = scan_str(s,FALSE,FALSE);
4400 missingterm((char*)0);
4402 if (SvCUR(PL_lex_stuff)) {
4405 d = SvPV_force(PL_lex_stuff, len);
4407 for (; isSPACE(*d) && len; --len, ++d) ;
4410 if (!warned && ckWARN(WARN_QW)) {
4411 for (; !isSPACE(*d) && len; --len, ++d) {
4413 Perl_warner(aTHX_ WARN_QW,
4414 "Possible attempt to separate words with commas");
4417 else if (*d == '#') {
4418 Perl_warner(aTHX_ WARN_QW,
4419 "Possible attempt to put comments in qw() list");
4425 for (; !isSPACE(*d) && len; --len, ++d) ;
4427 words = append_elem(OP_LIST, words,
4428 newSVOP(OP_CONST, 0, newSVpvn(b, d-b)));
4432 PL_nextval[PL_nexttoke].opval = words;
4437 SvREFCNT_dec(PL_lex_stuff);
4438 PL_lex_stuff = Nullsv;
4443 s = scan_str(s,FALSE,FALSE);
4445 missingterm((char*)0);
4446 yylval.ival = OP_STRINGIFY;
4447 if (SvIVX(PL_lex_stuff) == '\'')
4448 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4449 TERM(sublex_start());
4452 s = scan_pat(s,OP_QR);
4453 TERM(sublex_start());
4456 s = scan_str(s,FALSE,FALSE);
4458 missingterm((char*)0);
4459 yylval.ival = OP_BACKTICK;
4461 TERM(sublex_start());
4468 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4469 s = force_version(s);
4472 *PL_tokenbuf = '\0';
4473 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4474 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4475 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4477 yyerror("<> should be quotes");
4485 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4489 LOP(OP_RENAME,XTERM);
4498 LOP(OP_RINDEX,XTERM);
4521 LOP(OP_REVERSE,XTERM);
4532 TERM(sublex_start());
4534 TOKEN(1); /* force error */
4543 LOP(OP_SELECT,XTERM);
4549 LOP(OP_SEMCTL,XTERM);
4552 LOP(OP_SEMGET,XTERM);
4555 LOP(OP_SEMOP,XTERM);
4561 LOP(OP_SETPGRP,XTERM);
4563 case KEY_setpriority:
4564 LOP(OP_SETPRIORITY,XTERM);
4566 case KEY_sethostent:
4572 case KEY_setservent:
4575 case KEY_setprotoent:
4585 LOP(OP_SEEKDIR,XTERM);
4587 case KEY_setsockopt:
4588 LOP(OP_SSOCKOPT,XTERM);
4594 LOP(OP_SHMCTL,XTERM);
4597 LOP(OP_SHMGET,XTERM);
4600 LOP(OP_SHMREAD,XTERM);
4603 LOP(OP_SHMWRITE,XTERM);
4606 LOP(OP_SHUTDOWN,XTERM);
4615 LOP(OP_SOCKET,XTERM);
4617 case KEY_socketpair:
4618 LOP(OP_SOCKPAIR,XTERM);
4621 checkcomma(s,PL_tokenbuf,"subroutine name");
4623 if (*s == ';' || *s == ')') /* probably a close */
4624 Perl_croak(aTHX_ "sort is now a reserved word");
4626 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4630 LOP(OP_SPLIT,XTERM);
4633 LOP(OP_SPRINTF,XTERM);
4636 LOP(OP_SPLICE,XTERM);
4651 LOP(OP_SUBSTR,XTERM);
4657 char tmpbuf[sizeof PL_tokenbuf];
4659 expectation attrful;
4660 bool have_name, have_proto;
4665 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4666 (*s == ':' && s[1] == ':'))
4669 attrful = XATTRBLOCK;
4670 /* remember buffer pos'n for later force_word */
4671 tboffset = s - PL_oldbufptr;
4672 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4673 if (strchr(tmpbuf, ':'))
4674 sv_setpv(PL_subname, tmpbuf);
4676 sv_setsv(PL_subname,PL_curstname);
4677 sv_catpvn(PL_subname,"::",2);
4678 sv_catpvn(PL_subname,tmpbuf,len);
4685 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4686 PL_expect = XTERMBLOCK;
4687 attrful = XATTRTERM;
4688 sv_setpv(PL_subname,"?");
4692 if (key == KEY_format) {
4694 PL_lex_formbrack = PL_lex_brackets + 1;
4696 (void) force_word(PL_oldbufptr + tboffset, WORD,
4701 /* Look for a prototype */
4705 s = scan_str(s,FALSE,FALSE);
4708 SvREFCNT_dec(PL_lex_stuff);
4709 PL_lex_stuff = Nullsv;
4710 Perl_croak(aTHX_ "Prototype not terminated");
4713 d = SvPVX(PL_lex_stuff);
4715 for (p = d; *p; ++p) {
4720 SvCUR(PL_lex_stuff) = tmp;
4728 if (*s == ':' && s[1] != ':')
4729 PL_expect = attrful;
4732 PL_nextval[PL_nexttoke].opval =
4733 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4734 PL_lex_stuff = Nullsv;
4738 sv_setpv(PL_subname,"__ANON__");
4741 (void) force_word(PL_oldbufptr + tboffset, WORD,
4750 LOP(OP_SYSTEM,XREF);
4753 LOP(OP_SYMLINK,XTERM);
4756 LOP(OP_SYSCALL,XTERM);
4759 LOP(OP_SYSOPEN,XTERM);
4762 LOP(OP_SYSSEEK,XTERM);
4765 LOP(OP_SYSREAD,XTERM);
4768 LOP(OP_SYSWRITE,XTERM);
4772 TERM(sublex_start());
4793 LOP(OP_TRUNCATE,XTERM);
4805 yylval.ival = CopLINE(PL_curcop);
4809 yylval.ival = CopLINE(PL_curcop);
4813 LOP(OP_UNLINK,XTERM);
4819 LOP(OP_UNPACK,XTERM);
4822 LOP(OP_UTIME,XTERM);
4825 if (ckWARN(WARN_UMASK)) {
4826 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4827 if (*d != '0' && isDIGIT(*d))
4828 Perl_warner(aTHX_ WARN_UMASK,
4829 "umask: argument is missing initial 0");
4834 LOP(OP_UNSHIFT,XTERM);
4837 if (PL_expect != XSTATE)
4838 yyerror("\"use\" not allowed in expression");
4840 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4841 s = force_version(s);
4842 if (*s == ';' || (s = skipspace(s), *s == ';')) {
4843 PL_nextval[PL_nexttoke].opval = Nullop;
4848 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4849 s = force_version(s);
4861 yylval.ival = CopLINE(PL_curcop);
4865 PL_hints |= HINT_BLOCK_SCOPE;
4872 LOP(OP_WAITPID,XTERM);
4880 static char ctl_l[2];
4882 if (ctl_l[0] == '\0')
4883 ctl_l[0] = toCTRL('L');
4884 gv_fetchpv(ctl_l,TRUE, SVt_PV);
4887 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
4892 if (PL_expect == XOPERATOR)
4898 yylval.ival = OP_XOR;
4903 TERM(sublex_start());
4909 Perl_keyword(pTHX_ register char *d, I32 len)
4914 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
4915 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
4916 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
4917 if (strEQ(d,"__DATA__")) return KEY___DATA__;
4918 if (strEQ(d,"__END__")) return KEY___END__;
4922 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
4927 if (strEQ(d,"and")) return -KEY_and;
4928 if (strEQ(d,"abs")) return -KEY_abs;
4931 if (strEQ(d,"alarm")) return -KEY_alarm;
4932 if (strEQ(d,"atan2")) return -KEY_atan2;
4935 if (strEQ(d,"accept")) return -KEY_accept;
4940 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
4943 if (strEQ(d,"bless")) return -KEY_bless;
4944 if (strEQ(d,"bind")) return -KEY_bind;
4945 if (strEQ(d,"binmode")) return -KEY_binmode;
4948 if (strEQ(d,"CORE")) return -KEY_CORE;
4949 if (strEQ(d,"CHECK")) return KEY_CHECK;
4954 if (strEQ(d,"cmp")) return -KEY_cmp;
4955 if (strEQ(d,"chr")) return -KEY_chr;
4956 if (strEQ(d,"cos")) return -KEY_cos;
4959 if (strEQ(d,"chop")) return KEY_chop;
4962 if (strEQ(d,"close")) return -KEY_close;
4963 if (strEQ(d,"chdir")) return -KEY_chdir;
4964 if (strEQ(d,"chomp")) return KEY_chomp;
4965 if (strEQ(d,"chmod")) return -KEY_chmod;
4966 if (strEQ(d,"chown")) return -KEY_chown;
4967 if (strEQ(d,"crypt")) return -KEY_crypt;
4970 if (strEQ(d,"chroot")) return -KEY_chroot;
4971 if (strEQ(d,"caller")) return -KEY_caller;
4974 if (strEQ(d,"connect")) return -KEY_connect;
4977 if (strEQ(d,"closedir")) return -KEY_closedir;
4978 if (strEQ(d,"continue")) return -KEY_continue;
4983 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
4988 if (strEQ(d,"do")) return KEY_do;
4991 if (strEQ(d,"die")) return -KEY_die;
4994 if (strEQ(d,"dump")) return -KEY_dump;
4997 if (strEQ(d,"delete")) return KEY_delete;
5000 if (strEQ(d,"defined")) return KEY_defined;
5001 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5004 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5009 if (strEQ(d,"EQ")) { deprecate(d); return -KEY_eq;}
5010 if (strEQ(d,"END")) return KEY_END;
5015 if (strEQ(d,"eq")) return -KEY_eq;
5018 if (strEQ(d,"eof")) return -KEY_eof;
5019 if (strEQ(d,"exp")) return -KEY_exp;
5022 if (strEQ(d,"else")) return KEY_else;
5023 if (strEQ(d,"exit")) return -KEY_exit;
5024 if (strEQ(d,"eval")) return KEY_eval;
5025 if (strEQ(d,"exec")) return -KEY_exec;
5026 if (strEQ(d,"each")) return KEY_each;
5029 if (strEQ(d,"elsif")) return KEY_elsif;
5032 if (strEQ(d,"exists")) return KEY_exists;
5033 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5036 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5037 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5040 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5043 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5044 if (strEQ(d,"endservent")) return -KEY_endservent;
5047 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5054 if (strEQ(d,"for")) return KEY_for;
5057 if (strEQ(d,"fork")) return -KEY_fork;
5060 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5061 if (strEQ(d,"flock")) return -KEY_flock;
5064 if (strEQ(d,"format")) return KEY_format;
5065 if (strEQ(d,"fileno")) return -KEY_fileno;
5068 if (strEQ(d,"foreach")) return KEY_foreach;
5071 if (strEQ(d,"formline")) return -KEY_formline;
5077 if (strEQ(d,"GT")) { deprecate(d); return -KEY_gt;}
5078 if (strEQ(d,"GE")) { deprecate(d); return -KEY_ge;}
5082 if (strnEQ(d,"get",3)) {
5087 if (strEQ(d,"ppid")) return -KEY_getppid;
5088 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5091 if (strEQ(d,"pwent")) return -KEY_getpwent;
5092 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5093 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5096 if (strEQ(d,"peername")) return -KEY_getpeername;
5097 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5098 if (strEQ(d,"priority")) return -KEY_getpriority;
5101 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5104 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5108 else if (*d == 'h') {
5109 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5110 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5111 if (strEQ(d,"hostent")) return -KEY_gethostent;
5113 else if (*d == 'n') {
5114 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5115 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5116 if (strEQ(d,"netent")) return -KEY_getnetent;
5118 else if (*d == 's') {
5119 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5120 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5121 if (strEQ(d,"servent")) return -KEY_getservent;
5122 if (strEQ(d,"sockname")) return -KEY_getsockname;
5123 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5125 else if (*d == 'g') {
5126 if (strEQ(d,"grent")) return -KEY_getgrent;
5127 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5128 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5130 else if (*d == 'l') {
5131 if (strEQ(d,"login")) return -KEY_getlogin;
5133 else if (strEQ(d,"c")) return -KEY_getc;
5138 if (strEQ(d,"gt")) return -KEY_gt;
5139 if (strEQ(d,"ge")) return -KEY_ge;
5142 if (strEQ(d,"grep")) return KEY_grep;
5143 if (strEQ(d,"goto")) return KEY_goto;
5144 if (strEQ(d,"glob")) return KEY_glob;
5147 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5152 if (strEQ(d,"hex")) return -KEY_hex;
5155 if (strEQ(d,"INIT")) return KEY_INIT;
5160 if (strEQ(d,"if")) return KEY_if;
5163 if (strEQ(d,"int")) return -KEY_int;
5166 if (strEQ(d,"index")) return -KEY_index;
5167 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5172 if (strEQ(d,"join")) return -KEY_join;
5176 if (strEQ(d,"keys")) return KEY_keys;
5177 if (strEQ(d,"kill")) return -KEY_kill;
5182 if (strEQ(d,"LT")) { deprecate(d); return -KEY_lt;}
5183 if (strEQ(d,"LE")) { deprecate(d); return -KEY_le;}
5189 if (strEQ(d,"lt")) return -KEY_lt;
5190 if (strEQ(d,"le")) return -KEY_le;
5191 if (strEQ(d,"lc")) return -KEY_lc;
5194 if (strEQ(d,"log")) return -KEY_log;
5197 if (strEQ(d,"last")) return KEY_last;
5198 if (strEQ(d,"link")) return -KEY_link;
5199 if (strEQ(d,"lock")) return -KEY_lock;
5202 if (strEQ(d,"local")) return KEY_local;
5203 if (strEQ(d,"lstat")) return -KEY_lstat;
5206 if (strEQ(d,"length")) return -KEY_length;
5207 if (strEQ(d,"listen")) return -KEY_listen;
5210 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5213 if (strEQ(d,"localtime")) return -KEY_localtime;
5219 case 1: return KEY_m;
5221 if (strEQ(d,"my")) return KEY_my;
5224 if (strEQ(d,"map")) return KEY_map;
5227 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5230 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5231 if (strEQ(d,"msgget")) return -KEY_msgget;
5232 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5233 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5238 if (strEQ(d,"NE")) { deprecate(d); return -KEY_ne;}
5241 if (strEQ(d,"next")) return KEY_next;
5242 if (strEQ(d,"ne")) return -KEY_ne;
5243 if (strEQ(d,"not")) return -KEY_not;
5244 if (strEQ(d,"no")) return KEY_no;
5249 if (strEQ(d,"or")) return -KEY_or;
5252 if (strEQ(d,"ord")) return -KEY_ord;
5253 if (strEQ(d,"oct")) return -KEY_oct;
5254 if (strEQ(d,"our")) return KEY_our;
5257 if (strEQ(d,"open")) return -KEY_open;
5260 if (strEQ(d,"opendir")) return -KEY_opendir;
5267 if (strEQ(d,"pop")) return KEY_pop;
5268 if (strEQ(d,"pos")) return KEY_pos;
5271 if (strEQ(d,"push")) return KEY_push;
5272 if (strEQ(d,"pack")) return -KEY_pack;
5273 if (strEQ(d,"pipe")) return -KEY_pipe;
5276 if (strEQ(d,"print")) return KEY_print;
5279 if (strEQ(d,"printf")) return KEY_printf;
5282 if (strEQ(d,"package")) return KEY_package;
5285 if (strEQ(d,"prototype")) return KEY_prototype;
5290 if (strEQ(d,"q")) return KEY_q;
5291 if (strEQ(d,"qr")) return KEY_qr;
5292 if (strEQ(d,"qq")) return KEY_qq;
5293 if (strEQ(d,"qw")) return KEY_qw;
5294 if (strEQ(d,"qx")) return KEY_qx;
5296 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5301 if (strEQ(d,"ref")) return -KEY_ref;
5304 if (strEQ(d,"read")) return -KEY_read;
5305 if (strEQ(d,"rand")) return -KEY_rand;
5306 if (strEQ(d,"recv")) return -KEY_recv;
5307 if (strEQ(d,"redo")) return KEY_redo;
5310 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5311 if (strEQ(d,"reset")) return -KEY_reset;
5314 if (strEQ(d,"return")) return KEY_return;
5315 if (strEQ(d,"rename")) return -KEY_rename;
5316 if (strEQ(d,"rindex")) return -KEY_rindex;
5319 if (strEQ(d,"require")) return -KEY_require;
5320 if (strEQ(d,"reverse")) return -KEY_reverse;
5321 if (strEQ(d,"readdir")) return -KEY_readdir;
5324 if (strEQ(d,"readlink")) return -KEY_readlink;
5325 if (strEQ(d,"readline")) return -KEY_readline;
5326 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5329 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5335 case 0: return KEY_s;
5337 if (strEQ(d,"scalar")) return KEY_scalar;
5342 if (strEQ(d,"seek")) return -KEY_seek;
5343 if (strEQ(d,"send")) return -KEY_send;
5346 if (strEQ(d,"semop")) return -KEY_semop;
5349 if (strEQ(d,"select")) return -KEY_select;
5350 if (strEQ(d,"semctl")) return -KEY_semctl;
5351 if (strEQ(d,"semget")) return -KEY_semget;
5354 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5355 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5358 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5359 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5362 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5365 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5366 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5367 if (strEQ(d,"setservent")) return -KEY_setservent;
5370 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5371 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5378 if (strEQ(d,"shift")) return KEY_shift;
5381 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5382 if (strEQ(d,"shmget")) return -KEY_shmget;
5385 if (strEQ(d,"shmread")) return -KEY_shmread;
5388 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5389 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5394 if (strEQ(d,"sin")) return -KEY_sin;
5397 if (strEQ(d,"sleep")) return -KEY_sleep;
5400 if (strEQ(d,"sort")) return KEY_sort;
5401 if (strEQ(d,"socket")) return -KEY_socket;
5402 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5405 if (strEQ(d,"split")) return KEY_split;
5406 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5407 if (strEQ(d,"splice")) return KEY_splice;
5410 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5413 if (strEQ(d,"srand")) return -KEY_srand;
5416 if (strEQ(d,"stat")) return -KEY_stat;
5417 if (strEQ(d,"study")) return KEY_study;
5420 if (strEQ(d,"substr")) return -KEY_substr;
5421 if (strEQ(d,"sub")) return KEY_sub;
5426 if (strEQ(d,"system")) return -KEY_system;
5429 if (strEQ(d,"symlink")) return -KEY_symlink;
5430 if (strEQ(d,"syscall")) return -KEY_syscall;
5431 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5432 if (strEQ(d,"sysread")) return -KEY_sysread;
5433 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5436 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5445 if (strEQ(d,"tr")) return KEY_tr;
5448 if (strEQ(d,"tie")) return KEY_tie;
5451 if (strEQ(d,"tell")) return -KEY_tell;
5452 if (strEQ(d,"tied")) return KEY_tied;
5453 if (strEQ(d,"time")) return -KEY_time;
5456 if (strEQ(d,"times")) return -KEY_times;
5459 if (strEQ(d,"telldir")) return -KEY_telldir;
5462 if (strEQ(d,"truncate")) return -KEY_truncate;
5469 if (strEQ(d,"uc")) return -KEY_uc;
5472 if (strEQ(d,"use")) return KEY_use;
5475 if (strEQ(d,"undef")) return KEY_undef;
5476 if (strEQ(d,"until")) return KEY_until;
5477 if (strEQ(d,"untie")) return KEY_untie;
5478 if (strEQ(d,"utime")) return -KEY_utime;
5479 if (strEQ(d,"umask")) return -KEY_umask;
5482 if (strEQ(d,"unless")) return KEY_unless;
5483 if (strEQ(d,"unpack")) return -KEY_unpack;
5484 if (strEQ(d,"unlink")) return -KEY_unlink;
5487 if (strEQ(d,"unshift")) return KEY_unshift;
5488 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5493 if (strEQ(d,"values")) return -KEY_values;
5494 if (strEQ(d,"vec")) return -KEY_vec;
5499 if (strEQ(d,"warn")) return -KEY_warn;
5500 if (strEQ(d,"wait")) return -KEY_wait;
5503 if (strEQ(d,"while")) return KEY_while;
5504 if (strEQ(d,"write")) return -KEY_write;
5507 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5510 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5515 if (len == 1) return -KEY_x;
5516 if (strEQ(d,"xor")) return -KEY_xor;
5519 if (len == 1) return KEY_y;
5528 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5532 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5533 dTHR; /* only for ckWARN */
5534 if (ckWARN(WARN_SYNTAX)) {
5536 for (w = s+2; *w && level; w++) {
5543 for (; *w && isSPACE(*w); w++) ;
5544 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5545 Perl_warner(aTHX_ WARN_SYNTAX,
5546 "%s (...) interpreted as function",name);
5549 while (s < PL_bufend && isSPACE(*s))
5553 while (s < PL_bufend && isSPACE(*s))
5555 if (isIDFIRST_lazy_if(s,UTF)) {
5557 while (isALNUM_lazy_if(s,UTF))
5559 while (s < PL_bufend && isSPACE(*s))
5564 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5568 Perl_croak(aTHX_ "No comma allowed after %s", what);
5573 /* Either returns sv, or mortalizes sv and returns a new SV*.
5574 Best used as sv=new_constant(..., sv, ...).
5575 If s, pv are NULL, calls subroutine with one argument,
5576 and type is used with error messages only. */
5579 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5583 HV *table = GvHV(PL_hintgv); /* ^H */
5587 const char *why, *why1, *why2;
5589 if (!(PL_hints & HINT_LOCALIZE_HH)) {
5592 why = "%^H is not localized";
5596 msg = Perl_newSVpvf(aTHX_ "constant(%s): %s%s%s",
5597 (type ? type: "undef"), why1, why2, why);
5598 yyerror(SvPVX(msg));
5603 why = "%^H is not defined";
5606 cvp = hv_fetch(table, key, strlen(key), FALSE);
5607 if (!cvp || !SvOK(*cvp)) {
5608 why = "} is not defined";
5613 sv_2mortal(sv); /* Parent created it permanently */
5616 pv = sv_2mortal(newSVpvn(s, len));
5618 typesv = sv_2mortal(newSVpv(type, 0));
5620 typesv = &PL_sv_undef;
5622 PUSHSTACKi(PERLSI_OVERLOAD);
5635 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5639 /* Check the eval first */
5640 if (!PL_in_eval && SvTRUE(ERRSV))
5643 sv_catpv(ERRSV, "Propagated");
5644 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5646 res = SvREFCNT_inc(sv);
5650 (void)SvREFCNT_inc(res);
5659 why = "}} did not return a defined value";
5660 why1 = "Call to &{$^H{";
5670 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5672 register char *d = dest;
5673 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5676 Perl_croak(aTHX_ ident_too_long);
5677 if (isALNUM(*s)) /* UTF handled below */
5679 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
5684 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5688 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5689 char *t = s + UTF8SKIP(s);
5690 while (*t & 0x80 && is_utf8_mark((U8*)t))
5692 if (d + (t - s) > e)
5693 Perl_croak(aTHX_ ident_too_long);
5694 Copy(s, d, t - s, char);
5707 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5717 e = d + destlen - 3; /* two-character token, ending NUL */
5719 while (isDIGIT(*s)) {
5721 Perl_croak(aTHX_ ident_too_long);
5728 Perl_croak(aTHX_ ident_too_long);
5729 if (isALNUM(*s)) /* UTF handled below */
5731 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
5736 else if (*s == ':' && s[1] == ':') {
5740 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5741 char *t = s + UTF8SKIP(s);
5742 while (*t & 0x80 && is_utf8_mark((U8*)t))
5744 if (d + (t - s) > e)
5745 Perl_croak(aTHX_ ident_too_long);
5746 Copy(s, d, t - s, char);
5757 if (PL_lex_state != LEX_NORMAL)
5758 PL_lex_state = LEX_INTERPENDMAYBE;
5761 if (*s == '$' && s[1] &&
5762 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
5775 if (*d == '^' && *s && isCONTROLVAR(*s)) {
5780 if (isSPACE(s[-1])) {
5783 if (ch != ' ' && ch != '\t') {
5789 if (isIDFIRST_lazy_if(d,UTF)) {
5793 while (e < send && isALNUM_lazy_if(e,UTF) || *e == ':') {
5795 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
5798 Copy(s, d, e - s, char);
5803 while ((isALNUM(*s) || *s == ':') && d < e)
5806 Perl_croak(aTHX_ ident_too_long);
5809 while (s < send && (*s == ' ' || *s == '\t')) s++;
5810 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
5811 dTHR; /* only for ckWARN */
5812 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
5813 const char *brack = *s == '[' ? "[...]" : "{...}";
5814 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5815 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
5816 funny, dest, brack, funny, dest, brack);
5819 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
5823 /* Handle extended ${^Foo} variables
5824 * 1999-02-27 mjd-perl-patch@plover.com */
5825 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
5829 while (isALNUM(*s) && d < e) {
5833 Perl_croak(aTHX_ ident_too_long);
5838 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
5839 PL_lex_state = LEX_INTERPEND;
5842 if (PL_lex_state == LEX_NORMAL) {
5843 dTHR; /* only for ckWARN */
5844 if (ckWARN(WARN_AMBIGUOUS) &&
5845 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
5847 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5848 "Ambiguous use of %c{%s} resolved to %c%s",
5849 funny, dest, funny, dest);
5854 s = bracket; /* let the parser handle it */
5858 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
5859 PL_lex_state = LEX_INTERPEND;
5864 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
5869 *pmfl |= PMf_GLOBAL;
5871 *pmfl |= PMf_CONTINUE;
5875 *pmfl |= PMf_MULTILINE;
5877 *pmfl |= PMf_SINGLELINE;
5879 *pmfl |= PMf_EXTENDED;
5883 S_scan_pat(pTHX_ char *start, I32 type)
5888 s = scan_str(start,FALSE,FALSE);
5891 SvREFCNT_dec(PL_lex_stuff);
5892 PL_lex_stuff = Nullsv;
5893 Perl_croak(aTHX_ "Search pattern not terminated");
5896 pm = (PMOP*)newPMOP(type, 0);
5897 if (PL_multi_open == '?')
5898 pm->op_pmflags |= PMf_ONCE;
5900 while (*s && strchr("iomsx", *s))
5901 pmflag(&pm->op_pmflags,*s++);
5904 while (*s && strchr("iogcmsx", *s))
5905 pmflag(&pm->op_pmflags,*s++);
5907 pm->op_pmpermflags = pm->op_pmflags;
5909 PL_lex_op = (OP*)pm;
5910 yylval.ival = OP_MATCH;
5915 S_scan_subst(pTHX_ char *start)
5922 yylval.ival = OP_NULL;
5924 s = scan_str(start,FALSE,FALSE);
5928 SvREFCNT_dec(PL_lex_stuff);
5929 PL_lex_stuff = Nullsv;
5930 Perl_croak(aTHX_ "Substitution pattern not terminated");
5933 if (s[-1] == PL_multi_open)
5936 first_start = PL_multi_start;
5937 s = scan_str(s,FALSE,FALSE);
5940 SvREFCNT_dec(PL_lex_stuff);
5941 PL_lex_stuff = Nullsv;
5943 SvREFCNT_dec(PL_lex_repl);
5944 PL_lex_repl = Nullsv;
5945 Perl_croak(aTHX_ "Substitution replacement not terminated");
5947 PL_multi_start = first_start; /* so whole substitution is taken together */
5949 pm = (PMOP*)newPMOP(OP_SUBST, 0);
5955 else if (strchr("iogcmsx", *s))
5956 pmflag(&pm->op_pmflags,*s++);
5963 PL_sublex_info.super_bufptr = s;
5964 PL_sublex_info.super_bufend = PL_bufend;
5966 pm->op_pmflags |= PMf_EVAL;
5967 repl = newSVpvn("",0);
5969 sv_catpv(repl, es ? "eval " : "do ");
5970 sv_catpvn(repl, "{ ", 2);
5971 sv_catsv(repl, PL_lex_repl);
5972 sv_catpvn(repl, " };", 2);
5974 SvREFCNT_dec(PL_lex_repl);
5978 pm->op_pmpermflags = pm->op_pmflags;
5979 PL_lex_op = (OP*)pm;
5980 yylval.ival = OP_SUBST;
5985 S_scan_trans(pTHX_ char *start)
5996 yylval.ival = OP_NULL;
5998 s = scan_str(start,FALSE,FALSE);
6001 SvREFCNT_dec(PL_lex_stuff);
6002 PL_lex_stuff = Nullsv;
6003 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6005 if (s[-1] == PL_multi_open)
6008 s = scan_str(s,FALSE,FALSE);
6011 SvREFCNT_dec(PL_lex_stuff);
6012 PL_lex_stuff = Nullsv;
6014 SvREFCNT_dec(PL_lex_repl);
6015 PL_lex_repl = Nullsv;
6016 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6020 o = newSVOP(OP_TRANS, 0, 0);
6021 utf8 = OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF;
6024 New(803,tbl,256,short);
6025 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6029 complement = del = squash = 0;
6030 while (strchr("cdsCU", *s)) {
6032 complement = OPpTRANS_COMPLEMENT;
6034 del = OPpTRANS_DELETE;
6036 squash = OPpTRANS_SQUASH;
6041 utf8 &= ~OPpTRANS_FROM_UTF;
6043 utf8 |= OPpTRANS_FROM_UTF;
6047 utf8 &= ~OPpTRANS_TO_UTF;
6049 utf8 |= OPpTRANS_TO_UTF;
6052 Perl_croak(aTHX_ "Too many /C and /U options");
6057 o->op_private = del|squash|complement|utf8;
6060 yylval.ival = OP_TRANS;
6065 S_scan_heredoc(pTHX_ register char *s)
6069 I32 op_type = OP_SCALAR;
6076 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6080 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6083 for (peek = s; *peek == ' ' || *peek == '\t'; peek++) ;
6084 if (*peek && strchr("`'\"",*peek)) {
6087 s = delimcpy(d, e, s, PL_bufend, term, &len);
6097 if (!isALNUM_lazy_if(s,UTF))
6098 deprecate("bare << to mean <<\"\"");
6099 for (; isALNUM_lazy_if(s,UTF); s++) {
6104 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6105 Perl_croak(aTHX_ "Delimiter for here document is too long");
6108 len = d - PL_tokenbuf;
6109 #ifndef PERL_STRICT_CR
6110 d = strchr(s, '\r');
6114 while (s < PL_bufend) {
6120 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6129 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6134 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6135 herewas = newSVpvn(s,PL_bufend-s);
6137 s--, herewas = newSVpvn(s,d-s);
6138 s += SvCUR(herewas);
6140 tmpstr = NEWSV(87,79);
6141 sv_upgrade(tmpstr, SVt_PVIV);
6146 else if (term == '`') {
6147 op_type = OP_BACKTICK;
6148 SvIVX(tmpstr) = '\\';
6152 PL_multi_start = CopLINE(PL_curcop);
6153 PL_multi_open = PL_multi_close = '<';
6154 term = *PL_tokenbuf;
6155 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6156 char *bufptr = PL_sublex_info.super_bufptr;
6157 char *bufend = PL_sublex_info.super_bufend;
6158 char *olds = s - SvCUR(herewas);
6159 s = strchr(bufptr, '\n');
6163 while (s < bufend &&
6164 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6166 CopLINE_inc(PL_curcop);
6169 CopLINE_set(PL_curcop, PL_multi_start);
6170 missingterm(PL_tokenbuf);
6172 sv_setpvn(herewas,bufptr,d-bufptr+1);
6173 sv_setpvn(tmpstr,d+1,s-d);
6175 sv_catpvn(herewas,s,bufend-s);
6176 (void)strcpy(bufptr,SvPVX(herewas));
6183 while (s < PL_bufend &&
6184 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6186 CopLINE_inc(PL_curcop);
6188 if (s >= PL_bufend) {
6189 CopLINE_set(PL_curcop, PL_multi_start);
6190 missingterm(PL_tokenbuf);
6192 sv_setpvn(tmpstr,d+1,s-d);
6194 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6196 sv_catpvn(herewas,s,PL_bufend-s);
6197 sv_setsv(PL_linestr,herewas);
6198 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6199 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6202 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6203 while (s >= PL_bufend) { /* multiple line string? */
6205 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6206 CopLINE_set(PL_curcop, PL_multi_start);
6207 missingterm(PL_tokenbuf);
6209 CopLINE_inc(PL_curcop);
6210 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6211 #ifndef PERL_STRICT_CR
6212 if (PL_bufend - PL_linestart >= 2) {
6213 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6214 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6216 PL_bufend[-2] = '\n';
6218 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6220 else if (PL_bufend[-1] == '\r')
6221 PL_bufend[-1] = '\n';
6223 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6224 PL_bufend[-1] = '\n';
6226 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6227 SV *sv = NEWSV(88,0);
6229 sv_upgrade(sv, SVt_PVMG);
6230 sv_setsv(sv,PL_linestr);
6231 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6233 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6236 sv_catsv(PL_linestr,herewas);
6237 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6241 sv_catsv(tmpstr,PL_linestr);
6246 PL_multi_end = CopLINE(PL_curcop);
6247 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6248 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6249 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6251 SvREFCNT_dec(herewas);
6252 PL_lex_stuff = tmpstr;
6253 yylval.ival = op_type;
6258 takes: current position in input buffer
6259 returns: new position in input buffer
6260 side-effects: yylval and lex_op are set.
6265 <FH> read from filehandle
6266 <pkg::FH> read from package qualified filehandle
6267 <pkg'FH> read from package qualified filehandle
6268 <$fh> read from filehandle in $fh
6274 S_scan_inputsymbol(pTHX_ char *start)
6276 register char *s = start; /* current position in buffer */
6282 d = PL_tokenbuf; /* start of temp holding space */
6283 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6284 end = strchr(s, '\n');
6287 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6289 /* die if we didn't have space for the contents of the <>,
6290 or if it didn't end, or if we see a newline
6293 if (len >= sizeof PL_tokenbuf)
6294 Perl_croak(aTHX_ "Excessively long <> operator");
6296 Perl_croak(aTHX_ "Unterminated <> operator");
6301 Remember, only scalar variables are interpreted as filehandles by
6302 this code. Anything more complex (e.g., <$fh{$num}>) will be
6303 treated as a glob() call.
6304 This code makes use of the fact that except for the $ at the front,
6305 a scalar variable and a filehandle look the same.
6307 if (*d == '$' && d[1]) d++;
6309 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6310 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6313 /* If we've tried to read what we allow filehandles to look like, and
6314 there's still text left, then it must be a glob() and not a getline.
6315 Use scan_str to pull out the stuff between the <> and treat it
6316 as nothing more than a string.
6319 if (d - PL_tokenbuf != len) {
6320 yylval.ival = OP_GLOB;
6322 s = scan_str(start,FALSE,FALSE);
6324 Perl_croak(aTHX_ "Glob not terminated");
6328 /* we're in a filehandle read situation */
6331 /* turn <> into <ARGV> */
6333 (void)strcpy(d,"ARGV");
6335 /* if <$fh>, create the ops to turn the variable into a
6341 /* try to find it in the pad for this block, otherwise find
6342 add symbol table ops
6344 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6345 OP *o = newOP(OP_PADSV, 0);
6347 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6350 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6351 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6352 newUNOP(OP_RV2SV, 0,
6353 newGVOP(OP_GV, 0, gv)));
6355 PL_lex_op->op_flags |= OPf_SPECIAL;
6356 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6357 yylval.ival = OP_NULL;
6360 /* If it's none of the above, it must be a literal filehandle
6361 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6363 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6364 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6365 yylval.ival = OP_NULL;
6374 takes: start position in buffer
6375 keep_quoted preserve \ on the embedded delimiter(s)
6376 keep_delims preserve the delimiters around the string
6377 returns: position to continue reading from buffer
6378 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6379 updates the read buffer.
6381 This subroutine pulls a string out of the input. It is called for:
6382 q single quotes q(literal text)
6383 ' single quotes 'literal text'
6384 qq double quotes qq(interpolate $here please)
6385 " double quotes "interpolate $here please"
6386 qx backticks qx(/bin/ls -l)
6387 ` backticks `/bin/ls -l`
6388 qw quote words @EXPORT_OK = qw( func() $spam )
6389 m// regexp match m/this/
6390 s/// regexp substitute s/this/that/
6391 tr/// string transliterate tr/this/that/
6392 y/// string transliterate y/this/that/
6393 ($*@) sub prototypes sub foo ($)
6394 (stuff) sub attr parameters sub foo : attr(stuff)
6395 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6397 In most of these cases (all but <>, patterns and transliterate)
6398 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6399 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6400 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6403 It skips whitespace before the string starts, and treats the first
6404 character as the delimiter. If the delimiter is one of ([{< then
6405 the corresponding "close" character )]}> is used as the closing
6406 delimiter. It allows quoting of delimiters, and if the string has
6407 balanced delimiters ([{<>}]) it allows nesting.
6409 The lexer always reads these strings into lex_stuff, except in the
6410 case of the operators which take *two* arguments (s/// and tr///)
6411 when it checks to see if lex_stuff is full (presumably with the 1st
6412 arg to s or tr) and if so puts the string into lex_repl.
6417 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6420 SV *sv; /* scalar value: string */
6421 char *tmps; /* temp string, used for delimiter matching */
6422 register char *s = start; /* current position in the buffer */
6423 register char term; /* terminating character */
6424 register char *to; /* current position in the sv's data */
6425 I32 brackets = 1; /* bracket nesting level */
6426 bool has_utf = FALSE; /* is there any utf8 content? */
6428 /* skip space before the delimiter */
6432 /* mark where we are, in case we need to report errors */
6435 /* after skipping whitespace, the next character is the terminator */
6437 if ((term & 0x80) && UTF)
6440 /* mark where we are */
6441 PL_multi_start = CopLINE(PL_curcop);
6442 PL_multi_open = term;
6444 /* find corresponding closing delimiter */
6445 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6447 PL_multi_close = term;
6449 /* create a new SV to hold the contents. 87 is leak category, I'm
6450 assuming. 79 is the SV's initial length. What a random number. */
6452 sv_upgrade(sv, SVt_PVIV);
6454 (void)SvPOK_only(sv); /* validate pointer */
6456 /* move past delimiter and try to read a complete string */
6458 sv_catpvn(sv, s, 1);
6461 /* extend sv if need be */
6462 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6463 /* set 'to' to the next character in the sv's string */
6464 to = SvPVX(sv)+SvCUR(sv);
6466 /* if open delimiter is the close delimiter read unbridle */
6467 if (PL_multi_open == PL_multi_close) {
6468 for (; s < PL_bufend; s++,to++) {
6469 /* embedded newlines increment the current line number */
6470 if (*s == '\n' && !PL_rsfp)
6471 CopLINE_inc(PL_curcop);
6472 /* handle quoted delimiters */
6473 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6474 if (!keep_quoted && s[1] == term)
6476 /* any other quotes are simply copied straight through */
6480 /* terminate when run out of buffer (the for() condition), or
6481 have found the terminator */
6482 else if (*s == term)
6484 else if (!has_utf && (*s & 0x80) && UTF)
6490 /* if the terminator isn't the same as the start character (e.g.,
6491 matched brackets), we have to allow more in the quoting, and
6492 be prepared for nested brackets.
6495 /* read until we run out of string, or we find the terminator */
6496 for (; s < PL_bufend; s++,to++) {
6497 /* embedded newlines increment the line count */
6498 if (*s == '\n' && !PL_rsfp)
6499 CopLINE_inc(PL_curcop);
6500 /* backslashes can escape the open or closing characters */
6501 if (*s == '\\' && s+1 < PL_bufend) {
6503 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6508 /* allow nested opens and closes */
6509 else if (*s == PL_multi_close && --brackets <= 0)
6511 else if (*s == PL_multi_open)
6513 else if (!has_utf && (*s & 0x80) && UTF)
6518 /* terminate the copied string and update the sv's end-of-string */
6520 SvCUR_set(sv, to - SvPVX(sv));
6523 * this next chunk reads more into the buffer if we're not done yet
6527 break; /* handle case where we are done yet :-) */
6529 #ifndef PERL_STRICT_CR
6530 if (to - SvPVX(sv) >= 2) {
6531 if ((to[-2] == '\r' && to[-1] == '\n') ||
6532 (to[-2] == '\n' && to[-1] == '\r'))
6536 SvCUR_set(sv, to - SvPVX(sv));
6538 else if (to[-1] == '\r')
6541 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6545 /* if we're out of file, or a read fails, bail and reset the current
6546 line marker so we can report where the unterminated string began
6549 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6551 CopLINE_set(PL_curcop, PL_multi_start);
6554 /* we read a line, so increment our line counter */
6555 CopLINE_inc(PL_curcop);
6557 /* update debugger info */
6558 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6559 SV *sv = NEWSV(88,0);
6561 sv_upgrade(sv, SVt_PVMG);
6562 sv_setsv(sv,PL_linestr);
6563 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6566 /* having changed the buffer, we must update PL_bufend */
6567 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6570 /* at this point, we have successfully read the delimited string */
6573 sv_catpvn(sv, s, 1);
6576 PL_multi_end = CopLINE(PL_curcop);
6579 /* if we allocated too much space, give some back */
6580 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6581 SvLEN_set(sv, SvCUR(sv) + 1);
6582 Renew(SvPVX(sv), SvLEN(sv), char);
6585 /* decide whether this is the first or second quoted string we've read
6598 takes: pointer to position in buffer
6599 returns: pointer to new position in buffer
6600 side-effects: builds ops for the constant in yylval.op
6602 Read a number in any of the formats that Perl accepts:
6604 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6605 [\d_]+(\.[\d_]*)?[Ee](\d+)
6607 Underbars (_) are allowed in decimal numbers. If -w is on,
6608 underbars before a decimal point must be at three digit intervals.
6610 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6613 If it reads a number without a decimal point or an exponent, it will
6614 try converting the number to an integer and see if it can do so
6615 without loss of precision.
6619 Perl_scan_num(pTHX_ char *start)
6621 register char *s = start; /* current position in buffer */
6622 register char *d; /* destination in temp buffer */
6623 register char *e; /* end of temp buffer */
6624 IV tryiv; /* used to see if it can be an IV */
6625 NV value; /* number read, as a double */
6626 SV *sv = Nullsv; /* place to put the converted number */
6627 bool floatit; /* boolean: int or float? */
6628 char *lastub = 0; /* position of last underbar */
6629 static char number_too_long[] = "Number too long";
6631 /* We use the first character to decide what type of number this is */
6635 Perl_croak(aTHX_ "panic: scan_num");
6637 /* if it starts with a 0, it could be an octal number, a decimal in
6638 0.13 disguise, or a hexadecimal number, or a binary number. */
6642 u holds the "number so far"
6643 shift the power of 2 of the base
6644 (hex == 4, octal == 3, binary == 1)
6645 overflowed was the number more than we can hold?
6647 Shift is used when we add a digit. It also serves as an "are
6648 we in octal/hex/binary?" indicator to disallow hex characters
6655 bool overflowed = FALSE;
6656 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6657 static char* bases[5] = { "", "binary", "", "octal",
6659 static char* Bases[5] = { "", "Binary", "", "Octal",
6661 static char *maxima[5] = { "",
6662 "0b11111111111111111111111111111111",
6666 char *base, *Base, *max;
6672 } else if (s[1] == 'b') {
6676 /* check for a decimal in disguise */
6677 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6679 /* so it must be octal */
6683 base = bases[shift];
6684 Base = Bases[shift];
6685 max = maxima[shift];
6687 /* read the rest of the number */
6689 /* x is used in the overflow test,
6690 b is the digit we're adding on. */
6695 /* if we don't mention it, we're done */
6704 /* 8 and 9 are not octal */
6707 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6711 case '2': case '3': case '4':
6712 case '5': case '6': case '7':
6714 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6718 b = *s++ & 15; /* ASCII digit -> value of digit */
6722 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6723 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6724 /* make sure they said 0x */
6729 /* Prepare to put the digit we have onto the end
6730 of the number so far. We check for overflows.
6735 x = u << shift; /* make room for the digit */
6737 if ((x >> shift) != u
6738 && !(PL_hints & HINT_NEW_BINARY)) {
6742 if (ckWARN_d(WARN_OVERFLOW))
6743 Perl_warner(aTHX_ WARN_OVERFLOW,
6744 "Integer overflow in %s number",
6747 u = x | b; /* add the digit to the end */
6750 n *= nvshift[shift];
6751 /* If an NV has not enough bits in its
6752 * mantissa to represent an UV this summing of
6753 * small low-order numbers is a waste of time
6754 * (because the NV cannot preserve the
6755 * low-order bits anyway): we could just
6756 * remember when did we overflow and in the
6757 * end just multiply n by the right
6765 /* if we get here, we had success: make a scalar value from
6772 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6773 Perl_warner(aTHX_ WARN_PORTABLE,
6774 "%s number > %s non-portable",
6781 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6782 Perl_warner(aTHX_ WARN_PORTABLE,
6783 "%s number > %s non-portable",
6788 if (PL_hints & HINT_NEW_BINARY)
6789 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6794 handle decimal numbers.
6795 we're also sent here when we read a 0 as the first digit
6797 case '1': case '2': case '3': case '4': case '5':
6798 case '6': case '7': case '8': case '9': case '.':
6801 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
6804 /* read next group of digits and _ and copy into d */
6805 while (isDIGIT(*s) || *s == '_') {
6806 /* skip underscores, checking for misplaced ones
6810 dTHR; /* only for ckWARN */
6811 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
6812 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6816 /* check for end of fixed-length buffer */
6818 Perl_croak(aTHX_ number_too_long);
6819 /* if we're ok, copy the character */
6824 /* final misplaced underbar check */
6825 if (lastub && s - lastub != 3) {
6827 if (ckWARN(WARN_SYNTAX))
6828 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6831 /* read a decimal portion if there is one. avoid
6832 3..5 being interpreted as the number 3. followed
6835 if (*s == '.' && s[1] != '.') {
6839 /* copy, ignoring underbars, until we run out of
6840 digits. Note: no misplaced underbar checks!
6842 for (; isDIGIT(*s) || *s == '_'; s++) {
6843 /* fixed length buffer check */
6845 Perl_croak(aTHX_ number_too_long);
6851 /* read exponent part, if present */
6852 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
6856 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
6857 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
6859 /* allow positive or negative exponent */
6860 if (*s == '+' || *s == '-')
6863 /* read digits of exponent (no underbars :-) */
6864 while (isDIGIT(*s)) {
6866 Perl_croak(aTHX_ number_too_long);
6871 /* terminate the string */
6874 /* make an sv from the string */
6877 value = Atof(PL_tokenbuf);
6880 See if we can make do with an integer value without loss of
6881 precision. We use I_V to cast to an int, because some
6882 compilers have issues. Then we try casting it back and see
6883 if it was the same. We only do this if we know we
6884 specifically read an integer.
6886 Note: if floatit is true, then we don't need to do the
6890 if (!floatit && (NV)tryiv == value)
6891 sv_setiv(sv, tryiv);
6893 sv_setnv(sv, value);
6894 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
6895 (PL_hints & HINT_NEW_INTEGER) )
6896 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
6897 (floatit ? "float" : "integer"),
6900 /* if it starts with a v, it could be a version number */
6905 while (isDIGIT(*pos))
6907 if (*pos == '.' && isDIGIT(pos[1])) {
6909 U8 tmpbuf[UTF8_MAXLEN];
6912 s++; /* get past 'v' */
6915 sv_setpvn(sv, "", 0);
6918 if (*s == '0' && isDIGIT(s[1]))
6919 yyerror("Octal number in vector unsupported");
6922 while (isDIGIT(*pos))
6926 tmpend = uv_to_utf8(tmpbuf, rev);
6930 tmpbuf[0] = (U8)rev;
6931 tmpend = &tmpbuf[1];
6933 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
6934 } while (*pos == '.' && isDIGIT(pos[1]));
6936 if (*s == '0' && isDIGIT(s[1]))
6937 yyerror("Octal number in vector unsupported");
6940 tmpend = uv_to_utf8(tmpbuf, rev);
6941 utf8 = utf8 || rev > 127;
6942 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
6948 sv_utf8_downgrade(sv, TRUE);
6955 /* make the op for the constant and return */
6958 yylval.opval = newSVOP(OP_CONST, 0, sv);
6960 yylval.opval = Nullop;
6966 S_scan_formline(pTHX_ register char *s)
6971 SV *stuff = newSVpvn("",0);
6972 bool needargs = FALSE;
6975 if (*s == '.' || *s == '}') {
6977 #ifdef PERL_STRICT_CR
6978 for (t = s+1;*t == ' ' || *t == '\t'; t++) ;
6980 for (t = s+1;*t == ' ' || *t == '\t' || *t == '\r'; t++) ;
6982 if (*t == '\n' || t == PL_bufend)
6985 if (PL_in_eval && !PL_rsfp) {
6986 eol = strchr(s,'\n');
6991 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6993 for (t = s; t < eol; t++) {
6994 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
6996 goto enough; /* ~~ must be first line in formline */
6998 if (*t == '@' || *t == '^')
7001 sv_catpvn(stuff, s, eol-s);
7002 #ifndef PERL_STRICT_CR
7003 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7004 char *end = SvPVX(stuff) + SvCUR(stuff);
7013 s = filter_gets(PL_linestr, PL_rsfp, 0);
7014 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7015 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7018 yyerror("Format not terminated");
7028 PL_lex_state = LEX_NORMAL;
7029 PL_nextval[PL_nexttoke].ival = 0;
7033 PL_lex_state = LEX_FORMLINE;
7034 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7036 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7040 SvREFCNT_dec(stuff);
7041 PL_lex_formbrack = 0;
7052 PL_cshlen = strlen(PL_cshname);
7057 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7060 I32 oldsavestack_ix = PL_savestack_ix;
7061 CV* outsidecv = PL_compcv;
7065 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7067 SAVEI32(PL_subline);
7068 save_item(PL_subname);
7071 SAVESPTR(PL_comppad_name);
7072 SAVESPTR(PL_compcv);
7073 SAVEI32(PL_comppad_name_fill);
7074 SAVEI32(PL_min_intro_pending);
7075 SAVEI32(PL_max_intro_pending);
7076 SAVEI32(PL_pad_reset_pending);
7078 PL_compcv = (CV*)NEWSV(1104,0);
7079 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7080 CvFLAGS(PL_compcv) |= flags;
7082 PL_comppad = newAV();
7083 av_push(PL_comppad, Nullsv);
7084 PL_curpad = AvARRAY(PL_comppad);
7085 PL_comppad_name = newAV();
7086 PL_comppad_name_fill = 0;
7087 PL_min_intro_pending = 0;
7089 PL_subline = CopLINE(PL_curcop);
7091 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7092 PL_curpad[0] = (SV*)newAV();
7093 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7094 #endif /* USE_THREADS */
7096 comppadlist = newAV();
7097 AvREAL_off(comppadlist);
7098 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7099 av_store(comppadlist, 1, (SV*)PL_comppad);
7101 CvPADLIST(PL_compcv) = comppadlist;
7102 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7104 CvOWNER(PL_compcv) = 0;
7105 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7106 MUTEX_INIT(CvMUTEXP(PL_compcv));
7107 #endif /* USE_THREADS */
7109 return oldsavestack_ix;
7113 Perl_yywarn(pTHX_ char *s)
7116 PL_in_eval |= EVAL_WARNONLY;
7118 PL_in_eval &= ~EVAL_WARNONLY;
7123 Perl_yyerror(pTHX_ char *s)
7127 char *context = NULL;
7131 if (!yychar || (yychar == ';' && !PL_rsfp))
7133 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7134 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7135 while (isSPACE(*PL_oldoldbufptr))
7137 context = PL_oldoldbufptr;
7138 contlen = PL_bufptr - PL_oldoldbufptr;
7140 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7141 PL_oldbufptr != PL_bufptr) {
7142 while (isSPACE(*PL_oldbufptr))
7144 context = PL_oldbufptr;
7145 contlen = PL_bufptr - PL_oldbufptr;
7147 else if (yychar > 255)
7148 where = "next token ???";
7149 #ifdef USE_PURE_BISON
7150 /* GNU Bison sets the value -2 */
7151 else if (yychar == -2) {
7153 else if ((yychar & 127) == 127) {
7155 if (PL_lex_state == LEX_NORMAL ||
7156 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7157 where = "at end of line";
7158 else if (PL_lex_inpat)
7159 where = "within pattern";
7161 where = "within string";
7164 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7166 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7167 else if (isPRINT_LC(yychar))
7168 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7170 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7171 where = SvPVX(where_sv);
7173 msg = sv_2mortal(newSVpv(s, 0));
7174 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7175 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7177 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7179 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7180 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7181 Perl_sv_catpvf(aTHX_ msg,
7182 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7183 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7186 if (PL_in_eval & EVAL_WARNONLY)
7187 Perl_warn(aTHX_ "%"SVf, msg);
7190 if (PL_error_count >= 10)
7191 Perl_croak(aTHX_ "%s has too many errors.\n", CopFILE(PL_curcop));
7193 PL_in_my_stash = Nullhv;
7204 * Restore a source filter.
7208 restore_rsfp(pTHXo_ void *f)
7210 PerlIO *fp = (PerlIO*)f;
7212 if (PL_rsfp == PerlIO_stdin())
7213 PerlIO_clearerr(PL_rsfp);
7214 else if (PL_rsfp && (PL_rsfp != fp))
7215 PerlIO_close(PL_rsfp);