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
68 /* XXX If this causes problems, set i_unistd=undef in the hint file. */
70 # include <unistd.h> /* Needed for execv() */
79 YYSTYPE* yylval_pointer = NULL;
80 int* yychar_pointer = NULL;
83 # define yylval (*yylval_pointer)
84 # define yychar (*yychar_pointer)
85 # define PERL_YYLEX_PARAM yylval_pointer,yychar_pointer
87 # define yylex() Perl_yylex(aTHX_ yylval_pointer, yychar_pointer)
92 /* CLINE is a macro that ensures PL_copline has a sane value */
97 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
100 * Convenience functions to return different tokens and prime the
101 * lexer for the next token. They all take an argument.
103 * TOKEN : generic token (used for '(', DOLSHARP, etc)
104 * OPERATOR : generic operator
105 * AOPERATOR : assignment operator
106 * PREBLOCK : beginning the block after an if, while, foreach, ...
107 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
108 * PREREF : *EXPR where EXPR is not a simple identifier
109 * TERM : expression term
110 * LOOPX : loop exiting command (goto, last, dump, etc)
111 * FTST : file test operator
112 * FUN0 : zero-argument function
113 * FUN1 : not used, except for not, which isn't a UNIOP
114 * BOop : bitwise or or xor
116 * SHop : shift operator
117 * PWop : power operator
118 * PMop : pattern-matching operator
119 * Aop : addition-level operator
120 * Mop : multiplication-level operator
121 * Eop : equality-testing operator
122 * Rop : relational operator <= != gt
124 * Also see LOP and lop() below.
127 #define TOKEN(retval) return (PL_bufptr = s,(int)retval)
128 #define OPERATOR(retval) return (PL_expect = XTERM,PL_bufptr = s,(int)retval)
129 #define AOPERATOR(retval) return ao((PL_expect = XTERM,PL_bufptr = s,(int)retval))
130 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
131 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
132 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s,(int)retval)
133 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR,PL_bufptr = s,(int)retval)
134 #define LOOPX(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
135 #define FTST(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)UNIOP)
136 #define FUN0(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
137 #define FUN1(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
138 #define BOop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
139 #define BAop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
140 #define SHop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
141 #define PWop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
142 #define PMop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
143 #define Aop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
144 #define Mop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
145 #define Eop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
146 #define Rop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)RELOP)
148 /* This bit of chicanery makes a unary function followed by
149 * a parenthesis into a function with one argument, highest precedence.
151 #define UNI(f) return(yylval.ival = f, \
154 PL_last_uni = PL_oldbufptr, \
155 PL_last_lop_op = f, \
156 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
158 #define UNIBRACK(f) return(yylval.ival = f, \
160 PL_last_uni = PL_oldbufptr, \
161 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
163 /* grandfather return to old style */
164 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
169 * This subroutine detects &&= and ||= and turns an ANDAND or OROR
170 * into an OP_ANDASSIGN or OP_ORASSIGN
174 S_ao(pTHX_ int toketype)
176 if (*PL_bufptr == '=') {
178 if (toketype == ANDAND)
179 yylval.ival = OP_ANDASSIGN;
180 else if (toketype == OROR)
181 yylval.ival = OP_ORASSIGN;
189 * When Perl expects an operator and finds something else, no_op
190 * prints the warning. It always prints "<something> found where
191 * operator expected. It prints "Missing semicolon on previous line?"
192 * if the surprise occurs at the start of the line. "do you need to
193 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
194 * where the compiler doesn't know if foo is a method call or a function.
195 * It prints "Missing operator before end of line" if there's nothing
196 * after the missing operator, or "... before <...>" if there is something
197 * after the missing operator.
201 S_no_op(pTHX_ char *what, char *s)
203 char *oldbp = PL_bufptr;
204 bool is_first = (PL_oldbufptr == PL_linestart);
212 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
214 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
215 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
217 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
218 if (t < PL_bufptr && isSPACE(*t))
219 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
220 t - PL_oldoldbufptr, PL_oldoldbufptr);
223 Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
229 * Complain about missing quote/regexp/heredoc terminator.
230 * If it's called with (char *)NULL then it cauterizes the line buffer.
231 * If we're in a delimited string and the delimiter is a control
232 * character, it's reformatted into a two-char sequence like ^C.
237 S_missingterm(pTHX_ char *s)
242 char *nl = strrchr(s,'\n');
248 iscntrl(PL_multi_close)
250 PL_multi_close < 32 || PL_multi_close == 127
254 tmpbuf[1] = toCTRL(PL_multi_close);
260 *tmpbuf = PL_multi_close;
264 q = strchr(s,'"') ? '\'' : '"';
265 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
273 Perl_deprecate(pTHX_ char *s)
276 if (ckWARN(WARN_DEPRECATED))
277 Perl_warner(aTHX_ WARN_DEPRECATED, "Use of %s is deprecated", s);
282 * Deprecate a comma-less variable list.
288 deprecate("comma-less variable list");
292 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
293 * utf16-to-utf8-reversed.
296 #ifdef PERL_CR_FILTER
300 register char *s = SvPVX(sv);
301 register char *e = s + SvCUR(sv);
302 /* outer loop optimized to do nothing if there are no CR-LFs */
304 if (*s++ == '\r' && *s == '\n') {
305 /* hit a CR-LF, need to copy the rest */
306 register char *d = s - 1;
309 if (*s == '\r' && s[1] == '\n')
320 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
322 I32 count = FILTER_READ(idx+1, sv, maxlen);
323 if (count > 0 && !maxlen)
330 S_utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
332 I32 count = FILTER_READ(idx+1, sv, maxlen);
336 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
337 tend = utf16_to_utf8((U16*)SvPVX(sv), tmps, SvCUR(sv));
338 sv_usepvn(sv, (char*)tmps, tend - tmps);
345 S_utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
347 I32 count = FILTER_READ(idx+1, sv, maxlen);
351 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
352 tend = utf16_to_utf8_reversed((U16*)SvPVX(sv), tmps, SvCUR(sv));
353 sv_usepvn(sv, (char*)tmps, tend - tmps);
361 * Initialize variables. Uses the Perl save_stack to save its state (for
362 * recursive calls to the parser).
366 Perl_lex_start(pTHX_ SV *line)
372 SAVEI32(PL_lex_dojoin);
373 SAVEI32(PL_lex_brackets);
374 SAVEI32(PL_lex_casemods);
375 SAVEI32(PL_lex_starts);
376 SAVEI32(PL_lex_state);
377 SAVEVPTR(PL_lex_inpat);
378 SAVEI32(PL_lex_inwhat);
379 if (PL_lex_state == LEX_KNOWNEXT) {
380 I32 toke = PL_nexttoke;
381 while (--toke >= 0) {
382 SAVEI32(PL_nexttype[toke]);
383 SAVEVPTR(PL_nextval[toke]);
385 SAVEI32(PL_nexttoke);
388 SAVECOPLINE(PL_curcop);
391 SAVEPPTR(PL_oldbufptr);
392 SAVEPPTR(PL_oldoldbufptr);
393 SAVEPPTR(PL_linestart);
394 SAVESPTR(PL_linestr);
395 SAVEPPTR(PL_lex_brackstack);
396 SAVEPPTR(PL_lex_casestack);
397 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
398 SAVESPTR(PL_lex_stuff);
399 SAVEI32(PL_lex_defer);
400 SAVEI32(PL_sublex_info.sub_inwhat);
401 SAVESPTR(PL_lex_repl);
403 SAVEINT(PL_lex_expect);
405 PL_lex_state = LEX_NORMAL;
409 New(899, PL_lex_brackstack, 120, char);
410 New(899, PL_lex_casestack, 12, char);
411 SAVEFREEPV(PL_lex_brackstack);
412 SAVEFREEPV(PL_lex_casestack);
414 *PL_lex_casestack = '\0';
417 PL_lex_stuff = Nullsv;
418 PL_lex_repl = Nullsv;
421 PL_sublex_info.sub_inwhat = 0;
423 if (SvREADONLY(PL_linestr))
424 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
425 s = SvPV(PL_linestr, len);
426 if (len && s[len-1] != ';') {
427 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
428 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
429 sv_catpvn(PL_linestr, "\n;", 2);
431 SvTEMP_off(PL_linestr);
432 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
433 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
435 PL_rs = newSVpvn("\n", 1);
441 * Finalizer for lexing operations. Must be called when the parser is
442 * done with the lexer.
448 PL_doextract = FALSE;
453 * This subroutine has nothing to do with tilting, whether at windmills
454 * or pinball tables. Its name is short for "increment line". It
455 * increments the current line number in CopLINE(PL_curcop) and checks
456 * to see whether the line starts with a comment of the form
457 * # line 500 "foo.pm"
458 * If so, it sets the current line number and file to the values in the comment.
462 S_incline(pTHX_ char *s)
470 CopLINE_inc(PL_curcop);
473 while (*s == ' ' || *s == '\t') s++;
474 if (strnEQ(s, "line ", 5)) {
483 while (*s == ' ' || *s == '\t')
485 if (*s == '"' && (t = strchr(s+1, '"')))
489 return; /* false alarm */
490 for (t = s; !isSPACE(*t); t++) ;
495 CopFILE_set(PL_curcop, s);
497 CopLINE_set(PL_curcop, atoi(n)-1);
502 * Called to gobble the appropriate amount and type of whitespace.
503 * Skips comments as well.
507 S_skipspace(pTHX_ register char *s)
510 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
511 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
517 SSize_t oldprevlen, oldoldprevlen;
518 SSize_t oldloplen, oldunilen;
519 while (s < PL_bufend && isSPACE(*s)) {
520 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
525 if (s < PL_bufend && *s == '#') {
526 while (s < PL_bufend && *s != '\n')
530 if (PL_in_eval && !PL_rsfp) {
537 /* only continue to recharge the buffer if we're at the end
538 * of the buffer, we're not reading from a source filter, and
539 * we're in normal lexing mode
541 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
542 PL_lex_state == LEX_FORMLINE)
545 /* try to recharge the buffer */
546 if ((s = filter_gets(PL_linestr, PL_rsfp,
547 (prevlen = SvCUR(PL_linestr)))) == Nullch)
549 /* end of file. Add on the -p or -n magic */
550 if (PL_minus_n || PL_minus_p) {
551 sv_setpv(PL_linestr,PL_minus_p ?
552 ";}continue{print or die qq(-p destination: $!\\n)" :
554 sv_catpv(PL_linestr,";}");
555 PL_minus_n = PL_minus_p = 0;
558 sv_setpv(PL_linestr,";");
560 /* reset variables for next time we lex */
561 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
563 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
565 /* Close the filehandle. Could be from -P preprocessor,
566 * STDIN, or a regular file. If we were reading code from
567 * STDIN (because the commandline held no -e or filename)
568 * then we don't close it, we reset it so the code can
569 * read from STDIN too.
572 if (PL_preprocess && !PL_in_eval)
573 (void)PerlProc_pclose(PL_rsfp);
574 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
575 PerlIO_clearerr(PL_rsfp);
577 (void)PerlIO_close(PL_rsfp);
582 /* not at end of file, so we only read another line */
583 /* make corresponding updates to old pointers, for yyerror() */
584 oldprevlen = PL_oldbufptr - PL_bufend;
585 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
587 oldunilen = PL_last_uni - PL_bufend;
589 oldloplen = PL_last_lop - PL_bufend;
590 PL_linestart = PL_bufptr = s + prevlen;
591 PL_bufend = s + SvCUR(PL_linestr);
593 PL_oldbufptr = s + oldprevlen;
594 PL_oldoldbufptr = s + oldoldprevlen;
596 PL_last_uni = s + oldunilen;
598 PL_last_lop = s + oldloplen;
601 /* debugger active and we're not compiling the debugger code,
602 * so store the line into the debugger's array of lines
604 if (PERLDB_LINE && PL_curstash != PL_debstash) {
605 SV *sv = NEWSV(85,0);
607 sv_upgrade(sv, SVt_PVMG);
608 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
609 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
616 * Check the unary operators to ensure there's no ambiguity in how they're
617 * used. An ambiguous piece of code would be:
619 * This doesn't mean rand() + 5. Because rand() is a unary operator,
620 * the +5 is its argument.
630 if (PL_oldoldbufptr != PL_last_uni)
632 while (isSPACE(*PL_last_uni))
634 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
635 if ((t = strchr(s, '(')) && t < PL_bufptr)
637 if (ckWARN_d(WARN_AMBIGUOUS)){
640 Perl_warner(aTHX_ WARN_AMBIGUOUS,
641 "Warning: Use of \"%s\" without parens is ambiguous",
647 /* workaround to replace the UNI() macro with a function. Only the
648 * hints/uts.sh file mentions this. Other comments elsewhere in the
649 * source indicate Microport Unix might need it too.
655 #define UNI(f) return uni(f,s)
658 S_uni(pTHX_ I32 f, char *s)
663 PL_last_uni = PL_oldbufptr;
674 #endif /* CRIPPLED_CC */
677 * LOP : macro to build a list operator. Its behaviour has been replaced
678 * with a subroutine, S_lop() for which LOP is just another name.
681 #define LOP(f,x) return lop(f,x,s)
685 * Build a list operator (or something that might be one). The rules:
686 * - if we have a next token, then it's a list operator [why?]
687 * - if the next thing is an opening paren, then it's a function
688 * - else it's a list operator
692 S_lop(pTHX_ I32 f, int x, char *s)
699 PL_last_lop = PL_oldbufptr;
714 * When the lexer realizes it knows the next token (for instance,
715 * it is reordering tokens for the parser) then it can call S_force_next
716 * to know what token to return the next time the lexer is called. Caller
717 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
718 * handles the token correctly.
722 S_force_next(pTHX_ I32 type)
724 PL_nexttype[PL_nexttoke] = type;
726 if (PL_lex_state != LEX_KNOWNEXT) {
727 PL_lex_defer = PL_lex_state;
728 PL_lex_expect = PL_expect;
729 PL_lex_state = LEX_KNOWNEXT;
735 * When the lexer knows the next thing is a word (for instance, it has
736 * just seen -> and it knows that the next char is a word char, then
737 * it calls S_force_word to stick the next word into the PL_next lookahead.
740 * char *start : buffer position (must be within PL_linestr)
741 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
742 * int check_keyword : if true, Perl checks to make sure the word isn't
743 * a keyword (do this if the word is a label, e.g. goto FOO)
744 * int allow_pack : if true, : characters will also be allowed (require,
746 * int allow_initial_tick : used by the "sub" lexer only.
750 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
755 start = skipspace(start);
757 if (isIDFIRST_lazy_if(s,UTF) ||
758 (allow_pack && *s == ':') ||
759 (allow_initial_tick && *s == '\'') )
761 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
762 if (check_keyword && keyword(PL_tokenbuf, len))
764 if (token == METHOD) {
769 PL_expect = XOPERATOR;
772 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
773 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
781 * Called when the lexer wants $foo *foo &foo etc, but the program
782 * text only contains the "foo" portion. The first argument is a pointer
783 * to the "foo", and the second argument is the type symbol to prefix.
784 * Forces the next token to be a "WORD".
785 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
789 S_force_ident(pTHX_ register char *s, int kind)
792 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
793 PL_nextval[PL_nexttoke].opval = o;
796 dTHR; /* just for in_eval */
797 o->op_private = OPpCONST_ENTERED;
798 /* XXX see note in pp_entereval() for why we forgo typo
799 warnings if the symbol must be introduced in an eval.
801 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
802 kind == '$' ? SVt_PV :
803 kind == '@' ? SVt_PVAV :
804 kind == '%' ? SVt_PVHV :
813 * Forces the next token to be a version number.
817 S_force_version(pTHX_ char *s)
819 OP *version = Nullop;
823 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
827 for (; isDIGIT(*d) || *d == '_' || *d == '.'; d++);
828 if ((*d == ';' || isSPACE(*d)) && *(skipspace(d)) != ',') {
830 /* real VERSION number -- GBARR */
831 version = yylval.opval;
835 /* NOTE: The parser sees the package name and the VERSION swapped */
836 PL_nextval[PL_nexttoke].opval = version;
844 * Tokenize a quoted string passed in as an SV. It finds the next
845 * chunk, up to end of string or a backslash. It may make a new
846 * SV containing that chunk (if HINT_NEW_STRING is on). It also
851 S_tokeq(pTHX_ SV *sv)
862 s = SvPV_force(sv, len);
866 while (s < send && *s != '\\')
871 if ( PL_hints & HINT_NEW_STRING )
872 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
875 if (s + 1 < send && (s[1] == '\\'))
876 s++; /* all that, just for this */
881 SvCUR_set(sv, d - SvPVX(sv));
883 if ( PL_hints & HINT_NEW_STRING )
884 return new_constant(NULL, 0, "q", sv, pv, "q");
889 * Now come three functions related to double-quote context,
890 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
891 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
892 * interact with PL_lex_state, and create fake ( ... ) argument lists
893 * to handle functions and concatenation.
894 * They assume that whoever calls them will be setting up a fake
895 * join call, because each subthing puts a ',' after it. This lets
898 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
900 * (I'm not sure whether the spurious commas at the end of lcfirst's
901 * arguments and join's arguments are created or not).
906 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
908 * Pattern matching will set PL_lex_op to the pattern-matching op to
909 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
911 * OP_CONST and OP_READLINE are easy--just make the new op and return.
913 * Everything else becomes a FUNC.
915 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
916 * had an OP_CONST or OP_READLINE). This just sets us up for a
917 * call to S_sublex_push().
923 register I32 op_type = yylval.ival;
925 if (op_type == OP_NULL) {
926 yylval.opval = PL_lex_op;
930 if (op_type == OP_CONST || op_type == OP_READLINE) {
931 SV *sv = tokeq(PL_lex_stuff);
933 if (SvTYPE(sv) == SVt_PVIV) {
934 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
940 nsv = newSVpvn(p, len);
944 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
945 PL_lex_stuff = Nullsv;
949 PL_sublex_info.super_state = PL_lex_state;
950 PL_sublex_info.sub_inwhat = op_type;
951 PL_sublex_info.sub_op = PL_lex_op;
952 PL_lex_state = LEX_INTERPPUSH;
956 yylval.opval = PL_lex_op;
966 * Create a new scope to save the lexing state. The scope will be
967 * ended in S_sublex_done. Returns a '(', starting the function arguments
968 * to the uc, lc, etc. found before.
969 * Sets PL_lex_state to LEX_INTERPCONCAT.
978 PL_lex_state = PL_sublex_info.super_state;
979 SAVEI32(PL_lex_dojoin);
980 SAVEI32(PL_lex_brackets);
981 SAVEI32(PL_lex_casemods);
982 SAVEI32(PL_lex_starts);
983 SAVEI32(PL_lex_state);
984 SAVEVPTR(PL_lex_inpat);
985 SAVEI32(PL_lex_inwhat);
986 SAVECOPLINE(PL_curcop);
988 SAVEPPTR(PL_oldbufptr);
989 SAVEPPTR(PL_oldoldbufptr);
990 SAVEPPTR(PL_linestart);
991 SAVESPTR(PL_linestr);
992 SAVEPPTR(PL_lex_brackstack);
993 SAVEPPTR(PL_lex_casestack);
995 PL_linestr = PL_lex_stuff;
996 PL_lex_stuff = Nullsv;
998 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1000 PL_bufend += SvCUR(PL_linestr);
1001 SAVEFREESV(PL_linestr);
1003 PL_lex_dojoin = FALSE;
1004 PL_lex_brackets = 0;
1005 New(899, PL_lex_brackstack, 120, char);
1006 New(899, PL_lex_casestack, 12, char);
1007 SAVEFREEPV(PL_lex_brackstack);
1008 SAVEFREEPV(PL_lex_casestack);
1009 PL_lex_casemods = 0;
1010 *PL_lex_casestack = '\0';
1012 PL_lex_state = LEX_INTERPCONCAT;
1013 CopLINE_set(PL_curcop, PL_multi_start);
1015 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1016 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1017 PL_lex_inpat = PL_sublex_info.sub_op;
1019 PL_lex_inpat = Nullop;
1026 * Restores lexer state after a S_sublex_push.
1032 if (!PL_lex_starts++) {
1033 PL_expect = XOPERATOR;
1034 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn("",0));
1038 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1039 PL_lex_state = LEX_INTERPCASEMOD;
1043 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1044 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1045 PL_linestr = PL_lex_repl;
1047 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1048 PL_bufend += SvCUR(PL_linestr);
1049 SAVEFREESV(PL_linestr);
1050 PL_lex_dojoin = FALSE;
1051 PL_lex_brackets = 0;
1052 PL_lex_casemods = 0;
1053 *PL_lex_casestack = '\0';
1055 if (SvEVALED(PL_lex_repl)) {
1056 PL_lex_state = LEX_INTERPNORMAL;
1058 /* we don't clear PL_lex_repl here, so that we can check later
1059 whether this is an evalled subst; that means we rely on the
1060 logic to ensure sublex_done() is called again only via the
1061 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1064 PL_lex_state = LEX_INTERPCONCAT;
1065 PL_lex_repl = Nullsv;
1071 PL_bufend = SvPVX(PL_linestr);
1072 PL_bufend += SvCUR(PL_linestr);
1073 PL_expect = XOPERATOR;
1074 PL_sublex_info.sub_inwhat = 0;
1082 Extracts a pattern, double-quoted string, or transliteration. This
1085 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1086 processing a pattern (PL_lex_inpat is true), a transliteration
1087 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1089 Returns a pointer to the character scanned up to. Iff this is
1090 advanced from the start pointer supplied (ie if anything was
1091 successfully parsed), will leave an OP for the substring scanned
1092 in yylval. Caller must intuit reason for not parsing further
1093 by looking at the next characters herself.
1097 double-quoted style: \r and \n
1098 regexp special ones: \D \s
1100 backrefs: \1 (deprecated in substitution replacements)
1101 case and quoting: \U \Q \E
1102 stops on @ and $, but not for $ as tail anchor
1104 In transliterations:
1105 characters are VERY literal, except for - not at the start or end
1106 of the string, which indicates a range. scan_const expands the
1107 range to the full set of intermediate characters.
1109 In double-quoted strings:
1111 double-quoted style: \r and \n
1113 backrefs: \1 (deprecated)
1114 case and quoting: \U \Q \E
1117 scan_const does *not* construct ops to handle interpolated strings.
1118 It stops processing as soon as it finds an embedded $ or @ variable
1119 and leaves it to the caller to work out what's going on.
1121 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1123 $ in pattern could be $foo or could be tail anchor. Assumption:
1124 it's a tail anchor if $ is the last thing in the string, or if it's
1125 followed by one of ")| \n\t"
1127 \1 (backreferences) are turned into $1
1129 The structure of the code is
1130 while (there's a character to process) {
1131 handle transliteration ranges
1132 skip regexp comments
1133 skip # initiated comments in //x patterns
1134 check for embedded @foo
1135 check for embedded scalars
1137 leave intact backslashes from leave (below)
1138 deprecate \1 in strings and sub replacements
1139 handle string-changing backslashes \l \U \Q \E, etc.
1140 switch (what was escaped) {
1141 handle - in a transliteration (becomes a literal -)
1142 handle \132 octal characters
1143 handle 0x15 hex characters
1144 handle \cV (control V)
1145 handle printf backslashes (\f, \r, \n, etc)
1147 } (end if backslash)
1148 } (end while character to read)
1153 S_scan_const(pTHX_ char *start)
1155 register char *send = PL_bufend; /* end of the constant */
1156 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1157 register char *s = start; /* start of the constant */
1158 register char *d = SvPVX(sv); /* destination for copies */
1159 bool dorange = FALSE; /* are we in a translit range? */
1160 bool has_utf = FALSE; /* embedded \x{} */
1162 I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1163 ? (PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1165 I32 thisutf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1166 ? (PL_sublex_info.sub_op->op_private & (PL_lex_repl ?
1167 OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF))
1169 const char *leaveit = /* set of acceptably-backslashed characters */
1171 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1174 while (s < send || dorange) {
1175 /* get transliterations out of the way (they're most literal) */
1176 if (PL_lex_inwhat == OP_TRANS) {
1177 /* expand a range A-Z to the full set of characters. AIE! */
1179 I32 i; /* current expanded character */
1180 I32 min; /* first character in range */
1181 I32 max; /* last character in range */
1183 i = d - SvPVX(sv); /* remember current offset */
1184 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1185 d = SvPVX(sv) + i; /* refresh d after realloc */
1186 d -= 2; /* eat the first char and the - */
1188 min = (U8)*d; /* first char in range */
1189 max = (U8)d[1]; /* last char in range */
1192 if ((isLOWER(min) && isLOWER(max)) ||
1193 (isUPPER(min) && isUPPER(max))) {
1195 for (i = min; i <= max; i++)
1199 for (i = min; i <= max; i++)
1206 for (i = min; i <= max; i++)
1209 /* mark the range as done, and continue */
1214 /* range begins (ignore - as first or last char) */
1215 else if (*s == '-' && s+1 < send && s != start) {
1217 *d++ = (char)0xff; /* use illegal utf8 byte--see pmtrans */
1226 /* if we get here, we're not doing a transliteration */
1228 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1229 except for the last char, which will be done separately. */
1230 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1232 while (s < send && *s != ')')
1234 } else if (s[2] == '{'
1235 || s[2] == 'p' && s[3] == '{') { /* This should march regcomp.c */
1237 char *regparse = s + (s[2] == '{' ? 3 : 4);
1240 while (count && (c = *regparse)) {
1241 if (c == '\\' && regparse[1])
1249 if (*regparse != ')') {
1250 regparse--; /* Leave one char for continuation. */
1251 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1253 while (s < regparse)
1258 /* likewise skip #-initiated comments in //x patterns */
1259 else if (*s == '#' && PL_lex_inpat &&
1260 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1261 while (s+1 < send && *s != '\n')
1265 /* check for embedded arrays (@foo, @:foo, @'foo, @{foo}, @$foo) */
1266 else if (*s == '@' && s[1]
1267 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$", s[1])))
1270 /* check for embedded scalars. only stop if we're sure it's a
1273 else if (*s == '$') {
1274 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1276 if (s + 1 < send && !strchr("()| \n\t", s[1]))
1277 break; /* in regexp, $ might be tail anchor */
1280 /* (now in tr/// code again) */
1282 if (*s & 0x80 && thisutf) {
1283 dTHR; /* only for ckWARN */
1284 if (ckWARN(WARN_UTF8)) {
1285 (void)utf8_to_uv((U8*)s, &len); /* could cvt latin-1 to utf8 here... */
1294 has_utf = TRUE; /* assume valid utf8 */
1298 if (*s == '\\' && s+1 < send) {
1301 /* some backslashes we leave behind */
1302 if (*leaveit && *s && strchr(leaveit, *s)) {
1308 /* deprecate \1 in strings and substitution replacements */
1309 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1310 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1312 dTHR; /* only for ckWARN */
1313 if (ckWARN(WARN_SYNTAX))
1314 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1319 /* string-change backslash escapes */
1320 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1325 /* if we get here, it's either a quoted -, or a digit */
1328 /* quoted - in transliterations */
1330 if (PL_lex_inwhat == OP_TRANS) {
1338 if (ckWARN(WARN_UNSAFE) && isALPHA(*s))
1339 Perl_warner(aTHX_ WARN_UNSAFE,
1340 "Unrecognized escape \\%c passed through",
1342 /* default action is to copy the quoted character */
1347 /* \132 indicates an octal constant */
1348 case '0': case '1': case '2': case '3':
1349 case '4': case '5': case '6': case '7':
1350 *d++ = (char)scan_oct(s, 3, &len);
1354 /* \x24 indicates a hex constant */
1358 char* e = strchr(s, '}');
1362 yyerror("Missing right brace on \\x{}");
1365 /* note: utf always shorter than hex */
1366 uv = (UV)scan_hex(s + 1, e - s - 1, &len);
1368 d = (char*)uv_to_utf8((U8*)d, uv);
1376 /* XXX collapse this branch into the one above */
1377 UV uv = (UV)scan_hex(s, 2, &len);
1378 if (utf && PL_lex_inwhat == OP_TRANS &&
1379 utf != (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1381 d = (char*)uv_to_utf8((U8*)d, uv); /* doing a CU or UC */
1385 if (uv >= 127 && UTF) {
1387 if (ckWARN(WARN_UTF8))
1388 Perl_warner(aTHX_ WARN_UTF8,
1389 "\\x%.*s will produce malformed UTF-8 character; use \\x{%.*s} for that",
1390 (int)len,s,(int)len,s);
1398 /* \N{latin small letter a} is a named character */
1402 char* e = strchr(s, '}');
1411 yyerror("Missing right brace on \\N{}");
1415 res = newSVpvn(s + 1, e - s - 1);
1416 res = new_constant( Nullch, 0, "charnames",
1417 res, Nullsv, "\\N{...}" );
1418 str = SvPV(res,len);
1419 if (len > e - s + 4) {
1420 char *odest = SvPVX(sv);
1422 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1423 d = SvPVX(sv) + (d - odest);
1425 Copy(str, d, len, char);
1432 yyerror("Missing braces on \\N{}");
1435 /* \c is a control character */
1449 /* printf-style backslashes, formfeeds, newlines, etc */
1467 *d++ = '\047'; /* CP 1047 */
1470 *d++ = '\057'; /* CP 1047 */
1484 } /* end if (backslash) */
1487 } /* while loop to process each character */
1489 /* terminate the string and set up the sv */
1491 SvCUR_set(sv, d - SvPVX(sv));
1496 /* shrink the sv if we allocated more than we used */
1497 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1498 SvLEN_set(sv, SvCUR(sv) + 1);
1499 Renew(SvPVX(sv), SvLEN(sv), char);
1502 /* return the substring (via yylval) only if we parsed anything */
1503 if (s > PL_bufptr) {
1504 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1505 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1507 ( PL_lex_inwhat == OP_TRANS
1509 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1512 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1519 * Returns TRUE if there's more to the expression (e.g., a subscript),
1522 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1524 * ->[ and ->{ return TRUE
1525 * { and [ outside a pattern are always subscripts, so return TRUE
1526 * if we're outside a pattern and it's not { or [, then return FALSE
1527 * if we're in a pattern and the first char is a {
1528 * {4,5} (any digits around the comma) returns FALSE
1529 * if we're in a pattern and the first char is a [
1531 * [SOMETHING] has a funky algorithm to decide whether it's a
1532 * character class or not. It has to deal with things like
1533 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1534 * anything else returns TRUE
1537 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1540 S_intuit_more(pTHX_ register char *s)
1542 if (PL_lex_brackets)
1544 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1546 if (*s != '{' && *s != '[')
1551 /* In a pattern, so maybe we have {n,m}. */
1568 /* On the other hand, maybe we have a character class */
1571 if (*s == ']' || *s == '^')
1574 /* this is terrifying, and it works */
1575 int weight = 2; /* let's weigh the evidence */
1577 unsigned char un_char = 255, last_un_char;
1578 char *send = strchr(s,']');
1579 char tmpbuf[sizeof PL_tokenbuf * 4];
1581 if (!send) /* has to be an expression */
1584 Zero(seen,256,char);
1587 else if (isDIGIT(*s)) {
1589 if (isDIGIT(s[1]) && s[2] == ']')
1595 for (; s < send; s++) {
1596 last_un_char = un_char;
1597 un_char = (unsigned char)*s;
1602 weight -= seen[un_char] * 10;
1603 if (isALNUM_lazy_if(s+1,UTF)) {
1604 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1605 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1610 else if (*s == '$' && s[1] &&
1611 strchr("[#!%*<>()-=",s[1])) {
1612 if (/*{*/ strchr("])} =",s[2]))
1621 if (strchr("wds]",s[1]))
1623 else if (seen['\''] || seen['"'])
1625 else if (strchr("rnftbxcav",s[1]))
1627 else if (isDIGIT(s[1])) {
1629 while (s[1] && isDIGIT(s[1]))
1639 if (strchr("aA01! ",last_un_char))
1641 if (strchr("zZ79~",s[1]))
1643 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1644 weight -= 5; /* cope with negative subscript */
1647 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1648 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1653 if (keyword(tmpbuf, d - tmpbuf))
1656 if (un_char == last_un_char + 1)
1658 weight -= seen[un_char];
1663 if (weight >= 0) /* probably a character class */
1673 * Does all the checking to disambiguate
1675 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1676 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1678 * First argument is the stuff after the first token, e.g. "bar".
1680 * Not a method if bar is a filehandle.
1681 * Not a method if foo is a subroutine prototyped to take a filehandle.
1682 * Not a method if it's really "Foo $bar"
1683 * Method if it's "foo $bar"
1684 * Not a method if it's really "print foo $bar"
1685 * Method if it's really "foo package::" (interpreted as package->foo)
1686 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1687 * Not a method if bar is a filehandle or package, but is quoted with
1692 S_intuit_method(pTHX_ char *start, GV *gv)
1694 char *s = start + (*start == '$');
1695 char tmpbuf[sizeof PL_tokenbuf];
1703 if ((cv = GvCVu(gv))) {
1704 char *proto = SvPVX(cv);
1714 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1715 /* start is the beginning of the possible filehandle/object,
1716 * and s is the end of it
1717 * tmpbuf is a copy of it
1720 if (*start == '$') {
1721 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1726 return *s == '(' ? FUNCMETH : METHOD;
1728 if (!keyword(tmpbuf, len)) {
1729 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1734 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1735 if (indirgv && GvCVu(indirgv))
1737 /* filehandle or package name makes it a method */
1738 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1740 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1741 return 0; /* no assumptions -- "=>" quotes bearword */
1743 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1744 newSVpvn(tmpbuf,len));
1745 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1749 return *s == '(' ? FUNCMETH : METHOD;
1757 * Return a string of Perl code to load the debugger. If PERL5DB
1758 * is set, it will return the contents of that, otherwise a
1759 * compile-time require of perl5db.pl.
1766 char *pdb = PerlEnv_getenv("PERL5DB");
1770 SETERRNO(0,SS$_NORMAL);
1771 return "BEGIN { require 'perl5db.pl' }";
1777 /* Encoded script support. filter_add() effectively inserts a
1778 * 'pre-processing' function into the current source input stream.
1779 * Note that the filter function only applies to the current source file
1780 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1782 * The datasv parameter (which may be NULL) can be used to pass
1783 * private data to this instance of the filter. The filter function
1784 * can recover the SV using the FILTER_DATA macro and use it to
1785 * store private buffers and state information.
1787 * The supplied datasv parameter is upgraded to a PVIO type
1788 * and the IoDIRP field is used to store the function pointer,
1789 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1790 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1791 * private use must be set using malloc'd pointers.
1795 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1800 if (!PL_rsfp_filters)
1801 PL_rsfp_filters = newAV();
1803 datasv = NEWSV(255,0);
1804 if (!SvUPGRADE(datasv, SVt_PVIO))
1805 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1806 IoDIRP(datasv) = (DIR*)funcp; /* stash funcp into spare field */
1807 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1808 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1809 funcp, SvPV_nolen(datasv)));
1810 av_unshift(PL_rsfp_filters, 1);
1811 av_store(PL_rsfp_filters, 0, datasv) ;
1816 /* Delete most recently added instance of this filter function. */
1818 Perl_filter_del(pTHX_ filter_t funcp)
1821 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1822 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1824 /* if filter is on top of stack (usual case) just pop it off */
1825 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1826 if (IoDIRP(datasv) == (DIR*)funcp) {
1827 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1828 IoDIRP(datasv) = (DIR*)NULL;
1829 sv_free(av_pop(PL_rsfp_filters));
1833 /* we need to search for the correct entry and clear it */
1834 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1838 /* Invoke the n'th filter function for the current rsfp. */
1840 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1843 /* 0 = read one text line */
1848 if (!PL_rsfp_filters)
1850 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1851 /* Provide a default input filter to make life easy. */
1852 /* Note that we append to the line. This is handy. */
1853 DEBUG_P(PerlIO_printf(Perl_debug_log,
1854 "filter_read %d: from rsfp\n", idx));
1858 int old_len = SvCUR(buf_sv) ;
1860 /* ensure buf_sv is large enough */
1861 SvGROW(buf_sv, old_len + maxlen) ;
1862 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1863 if (PerlIO_error(PL_rsfp))
1864 return -1; /* error */
1866 return 0 ; /* end of file */
1868 SvCUR_set(buf_sv, old_len + len) ;
1871 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1872 if (PerlIO_error(PL_rsfp))
1873 return -1; /* error */
1875 return 0 ; /* end of file */
1878 return SvCUR(buf_sv);
1880 /* Skip this filter slot if filter has been deleted */
1881 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1882 DEBUG_P(PerlIO_printf(Perl_debug_log,
1883 "filter_read %d: skipped (filter deleted)\n",
1885 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1887 /* Get function pointer hidden within datasv */
1888 funcp = (filter_t)IoDIRP(datasv);
1889 DEBUG_P(PerlIO_printf(Perl_debug_log,
1890 "filter_read %d: via function %p (%s)\n",
1891 idx, funcp, SvPV_nolen(datasv)));
1892 /* Call function. The function is expected to */
1893 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1894 /* Return: <0:error, =0:eof, >0:not eof */
1895 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1899 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1901 #ifdef PERL_CR_FILTER
1902 if (!PL_rsfp_filters) {
1903 filter_add(S_cr_textfilter,NULL);
1906 if (PL_rsfp_filters) {
1909 SvCUR_set(sv, 0); /* start with empty line */
1910 if (FILTER_READ(0, sv, 0) > 0)
1911 return ( SvPVX(sv) ) ;
1916 return (sv_gets(sv, fp, append));
1921 static char* exp_name[] =
1922 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
1923 "ATTRTERM", "TERMBLOCK"
1930 Works out what to call the token just pulled out of the input
1931 stream. The yacc parser takes care of taking the ops we return and
1932 stitching them into a tree.
1938 if read an identifier
1939 if we're in a my declaration
1940 croak if they tried to say my($foo::bar)
1941 build the ops for a my() declaration
1942 if it's an access to a my() variable
1943 are we in a sort block?
1944 croak if my($a); $a <=> $b
1945 build ops for access to a my() variable
1946 if in a dq string, and they've said @foo and we can't find @foo
1948 build ops for a bareword
1949 if we already built the token before, use it.
1953 #ifdef USE_PURE_BISON
1954 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
1967 #ifdef USE_PURE_BISON
1968 yylval_pointer = lvalp;
1969 yychar_pointer = lcharp;
1972 /* check if there's an identifier for us to look at */
1973 if (PL_pending_ident) {
1974 /* pit holds the identifier we read and pending_ident is reset */
1975 char pit = PL_pending_ident;
1976 PL_pending_ident = 0;
1978 /* if we're in a my(), we can't allow dynamics here.
1979 $foo'bar has already been turned into $foo::bar, so
1980 just check for colons.
1982 if it's a legal name, the OP is a PADANY.
1985 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
1986 if (strchr(PL_tokenbuf,':'))
1987 yyerror(Perl_form(aTHX_ "No package name allowed for "
1988 "variable %s in \"our\"",
1990 tmp = pad_allocmy(PL_tokenbuf);
1993 if (strchr(PL_tokenbuf,':'))
1994 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
1996 yylval.opval = newOP(OP_PADANY, 0);
1997 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
2003 build the ops for accesses to a my() variable.
2005 Deny my($a) or my($b) in a sort block, *if* $a or $b is
2006 then used in a comparison. This catches most, but not
2007 all cases. For instance, it catches
2008 sort { my($a); $a <=> $b }
2010 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2011 (although why you'd do that is anyone's guess).
2014 if (!strchr(PL_tokenbuf,':')) {
2016 /* Check for single character per-thread SVs */
2017 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2018 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2019 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2021 yylval.opval = newOP(OP_THREADSV, 0);
2022 yylval.opval->op_targ = tmp;
2025 #endif /* USE_THREADS */
2026 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2027 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2028 /* might be an "our" variable" */
2029 if (SvFLAGS(namesv) & SVpad_OUR) {
2030 /* build ops for a bareword */
2031 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2032 sv_catpvn(sym, "::", 2);
2033 sv_catpv(sym, PL_tokenbuf+1);
2034 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2035 yylval.opval->op_private = OPpCONST_ENTERED;
2036 gv_fetchpv(SvPVX(sym),
2038 ? (GV_ADDMULTI | GV_ADDINEVAL)
2041 ((PL_tokenbuf[0] == '$') ? SVt_PV
2042 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2047 /* if it's a sort block and they're naming $a or $b */
2048 if (PL_last_lop_op == OP_SORT &&
2049 PL_tokenbuf[0] == '$' &&
2050 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2053 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2054 d < PL_bufend && *d != '\n';
2057 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2058 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2064 yylval.opval = newOP(OP_PADANY, 0);
2065 yylval.opval->op_targ = tmp;
2071 Whine if they've said @foo in a doublequoted string,
2072 and @foo isn't a variable we can find in the symbol
2075 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2076 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2077 if (!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2078 yyerror(Perl_form(aTHX_ "In string, %s now must be written as \\%s",
2079 PL_tokenbuf, PL_tokenbuf));
2082 /* build ops for a bareword */
2083 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2084 yylval.opval->op_private = OPpCONST_ENTERED;
2085 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2086 ((PL_tokenbuf[0] == '$') ? SVt_PV
2087 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2092 /* no identifier pending identification */
2094 switch (PL_lex_state) {
2096 case LEX_NORMAL: /* Some compilers will produce faster */
2097 case LEX_INTERPNORMAL: /* code if we comment these out. */
2101 /* when we've already built the next token, just pull it out of the queue */
2104 yylval = PL_nextval[PL_nexttoke];
2106 PL_lex_state = PL_lex_defer;
2107 PL_expect = PL_lex_expect;
2108 PL_lex_defer = LEX_NORMAL;
2110 return(PL_nexttype[PL_nexttoke]);
2112 /* interpolated case modifiers like \L \U, including \Q and \E.
2113 when we get here, PL_bufptr is at the \
2115 case LEX_INTERPCASEMOD:
2117 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2118 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2120 /* handle \E or end of string */
2121 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2125 if (PL_lex_casemods) {
2126 oldmod = PL_lex_casestack[--PL_lex_casemods];
2127 PL_lex_casestack[PL_lex_casemods] = '\0';
2129 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2131 PL_lex_state = LEX_INTERPCONCAT;
2135 if (PL_bufptr != PL_bufend)
2137 PL_lex_state = LEX_INTERPCONCAT;
2142 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2143 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2144 if (strchr("LU", *s) &&
2145 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2147 PL_lex_casestack[--PL_lex_casemods] = '\0';
2150 if (PL_lex_casemods > 10) {
2151 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2152 if (newlb != PL_lex_casestack) {
2154 PL_lex_casestack = newlb;
2157 PL_lex_casestack[PL_lex_casemods++] = *s;
2158 PL_lex_casestack[PL_lex_casemods] = '\0';
2159 PL_lex_state = LEX_INTERPCONCAT;
2160 PL_nextval[PL_nexttoke].ival = 0;
2163 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2165 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2167 PL_nextval[PL_nexttoke].ival = OP_LC;
2169 PL_nextval[PL_nexttoke].ival = OP_UC;
2171 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2173 Perl_croak(aTHX_ "panic: yylex");
2176 if (PL_lex_starts) {
2185 case LEX_INTERPPUSH:
2186 return sublex_push();
2188 case LEX_INTERPSTART:
2189 if (PL_bufptr == PL_bufend)
2190 return sublex_done();
2192 PL_lex_dojoin = (*PL_bufptr == '@');
2193 PL_lex_state = LEX_INTERPNORMAL;
2194 if (PL_lex_dojoin) {
2195 PL_nextval[PL_nexttoke].ival = 0;
2198 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2199 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2200 force_next(PRIVATEREF);
2202 force_ident("\"", '$');
2203 #endif /* USE_THREADS */
2204 PL_nextval[PL_nexttoke].ival = 0;
2206 PL_nextval[PL_nexttoke].ival = 0;
2208 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2211 if (PL_lex_starts++) {
2217 case LEX_INTERPENDMAYBE:
2218 if (intuit_more(PL_bufptr)) {
2219 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2225 if (PL_lex_dojoin) {
2226 PL_lex_dojoin = FALSE;
2227 PL_lex_state = LEX_INTERPCONCAT;
2230 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2231 && SvEVALED(PL_lex_repl))
2233 if (PL_bufptr != PL_bufend)
2234 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2235 PL_lex_repl = Nullsv;
2238 case LEX_INTERPCONCAT:
2240 if (PL_lex_brackets)
2241 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2243 if (PL_bufptr == PL_bufend)
2244 return sublex_done();
2246 if (SvIVX(PL_linestr) == '\'') {
2247 SV *sv = newSVsv(PL_linestr);
2250 else if ( PL_hints & HINT_NEW_RE )
2251 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2252 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2256 s = scan_const(PL_bufptr);
2258 PL_lex_state = LEX_INTERPCASEMOD;
2260 PL_lex_state = LEX_INTERPSTART;
2263 if (s != PL_bufptr) {
2264 PL_nextval[PL_nexttoke] = yylval;
2267 if (PL_lex_starts++)
2277 PL_lex_state = LEX_NORMAL;
2278 s = scan_formline(PL_bufptr);
2279 if (!PL_lex_formbrack)
2285 PL_oldoldbufptr = PL_oldbufptr;
2288 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2289 exp_name[PL_expect], s);
2295 if (isIDFIRST_lazy_if(s,UTF))
2297 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2300 goto fake_eof; /* emulate EOF on ^D or ^Z */
2305 if (PL_lex_brackets)
2306 yyerror("Missing right curly or square bracket");
2309 if (s++ < PL_bufend)
2310 goto retry; /* ignore stray nulls */
2313 if (!PL_in_eval && !PL_preambled) {
2314 PL_preambled = TRUE;
2315 sv_setpv(PL_linestr,incl_perldb());
2316 if (SvCUR(PL_linestr))
2317 sv_catpv(PL_linestr,";");
2319 while(AvFILLp(PL_preambleav) >= 0) {
2320 SV *tmpsv = av_shift(PL_preambleav);
2321 sv_catsv(PL_linestr, tmpsv);
2322 sv_catpv(PL_linestr, ";");
2325 sv_free((SV*)PL_preambleav);
2326 PL_preambleav = NULL;
2328 if (PL_minus_n || PL_minus_p) {
2329 sv_catpv(PL_linestr, "LINE: while (<>) {");
2331 sv_catpv(PL_linestr,"chomp;");
2333 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2335 GvIMPORTED_AV_on(gv);
2337 if (strchr("/'\"", *PL_splitstr)
2338 && strchr(PL_splitstr + 1, *PL_splitstr))
2339 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2342 s = "'~#\200\1'"; /* surely one char is unused...*/
2343 while (s[1] && strchr(PL_splitstr, *s)) s++;
2345 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2346 "q" + (delim == '\''), delim);
2347 for (s = PL_splitstr; *s; s++) {
2349 sv_catpvn(PL_linestr, "\\", 1);
2350 sv_catpvn(PL_linestr, s, 1);
2352 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2356 sv_catpv(PL_linestr,"@F=split(' ');");
2359 sv_catpv(PL_linestr, "\n");
2360 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2361 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2362 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2363 SV *sv = NEWSV(85,0);
2365 sv_upgrade(sv, SVt_PVMG);
2366 sv_setsv(sv,PL_linestr);
2367 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2372 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2375 if (PL_preprocess && !PL_in_eval)
2376 (void)PerlProc_pclose(PL_rsfp);
2377 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2378 PerlIO_clearerr(PL_rsfp);
2380 (void)PerlIO_close(PL_rsfp);
2382 PL_doextract = FALSE;
2384 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2385 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2386 sv_catpv(PL_linestr,";}");
2387 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2388 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2389 PL_minus_n = PL_minus_p = 0;
2392 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2393 sv_setpv(PL_linestr,"");
2394 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2397 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2398 PL_doextract = FALSE;
2400 /* Incest with pod. */
2401 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2402 sv_setpv(PL_linestr, "");
2403 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2404 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2405 PL_doextract = FALSE;
2409 } while (PL_doextract);
2410 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2411 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2412 SV *sv = NEWSV(85,0);
2414 sv_upgrade(sv, SVt_PVMG);
2415 sv_setsv(sv,PL_linestr);
2416 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2418 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2419 if (CopLINE(PL_curcop) == 1) {
2420 while (s < PL_bufend && isSPACE(*s))
2422 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2426 if (*s == '#' && *(s+1) == '!')
2428 #ifdef ALTERNATE_SHEBANG
2430 static char as[] = ALTERNATE_SHEBANG;
2431 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2432 d = s + (sizeof(as) - 1);
2434 #endif /* ALTERNATE_SHEBANG */
2443 while (*d && !isSPACE(*d))
2447 #ifdef ARG_ZERO_IS_SCRIPT
2448 if (ipathend > ipath) {
2450 * HP-UX (at least) sets argv[0] to the script name,
2451 * which makes $^X incorrect. And Digital UNIX and Linux,
2452 * at least, set argv[0] to the basename of the Perl
2453 * interpreter. So, having found "#!", we'll set it right.
2455 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2456 assert(SvPOK(x) || SvGMAGICAL(x));
2457 if (sv_eq(x, CopFILESV(PL_curcop))) {
2458 sv_setpvn(x, ipath, ipathend - ipath);
2461 TAINT_NOT; /* $^X is always tainted, but that's OK */
2463 #endif /* ARG_ZERO_IS_SCRIPT */
2468 d = instr(s,"perl -");
2470 d = instr(s,"perl");
2472 /* avoid getting into infinite loops when shebang
2473 * line contains "Perl" rather than "perl" */
2475 for (d = ipathend-4; d >= ipath; --d) {
2476 if ((*d == 'p' || *d == 'P')
2477 && !ibcmp(d, "perl", 4))
2487 #ifdef ALTERNATE_SHEBANG
2489 * If the ALTERNATE_SHEBANG on this system starts with a
2490 * character that can be part of a Perl expression, then if
2491 * we see it but not "perl", we're probably looking at the
2492 * start of Perl code, not a request to hand off to some
2493 * other interpreter. Similarly, if "perl" is there, but
2494 * not in the first 'word' of the line, we assume the line
2495 * contains the start of the Perl program.
2497 if (d && *s != '#') {
2499 while (*c && !strchr("; \t\r\n\f\v#", *c))
2502 d = Nullch; /* "perl" not in first word; ignore */
2504 *s = '#'; /* Don't try to parse shebang line */
2506 #endif /* ALTERNATE_SHEBANG */
2511 !instr(s,"indir") &&
2512 instr(PL_origargv[0],"perl"))
2518 while (s < PL_bufend && isSPACE(*s))
2520 if (s < PL_bufend) {
2521 Newz(899,newargv,PL_origargc+3,char*);
2523 while (s < PL_bufend && !isSPACE(*s))
2526 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2529 newargv = PL_origargv;
2531 PerlProc_execv(ipath, newargv);
2532 Perl_croak(aTHX_ "Can't exec %s", ipath);
2535 U32 oldpdb = PL_perldb;
2536 bool oldn = PL_minus_n;
2537 bool oldp = PL_minus_p;
2539 while (*d && !isSPACE(*d)) d++;
2540 while (*d == ' ' || *d == '\t') d++;
2544 if (*d == 'M' || *d == 'm') {
2546 while (*d && !isSPACE(*d)) d++;
2547 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2550 d = moreswitches(d);
2552 if (PERLDB_LINE && !oldpdb ||
2553 ( PL_minus_n || PL_minus_p ) && !(oldn || oldp) )
2554 /* if we have already added "LINE: while (<>) {",
2555 we must not do it again */
2557 sv_setpv(PL_linestr, "");
2558 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2559 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2560 PL_preambled = FALSE;
2562 (void)gv_fetchfile(PL_origfilename);
2569 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2571 PL_lex_state = LEX_FORMLINE;
2576 #ifdef PERL_STRICT_CR
2577 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2579 "(Maybe you didn't strip carriage returns after a network transfer?)\n");
2581 case ' ': case '\t': case '\f': case 013:
2586 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2588 while (s < d && *s != '\n')
2593 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2595 PL_lex_state = LEX_FORMLINE;
2605 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2610 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2613 if (strnEQ(s,"=>",2)) {
2614 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2615 OPERATOR('-'); /* unary minus */
2617 PL_last_uni = PL_oldbufptr;
2618 PL_last_lop_op = OP_FTEREAD; /* good enough */
2620 case 'r': FTST(OP_FTEREAD);
2621 case 'w': FTST(OP_FTEWRITE);
2622 case 'x': FTST(OP_FTEEXEC);
2623 case 'o': FTST(OP_FTEOWNED);
2624 case 'R': FTST(OP_FTRREAD);
2625 case 'W': FTST(OP_FTRWRITE);
2626 case 'X': FTST(OP_FTREXEC);
2627 case 'O': FTST(OP_FTROWNED);
2628 case 'e': FTST(OP_FTIS);
2629 case 'z': FTST(OP_FTZERO);
2630 case 's': FTST(OP_FTSIZE);
2631 case 'f': FTST(OP_FTFILE);
2632 case 'd': FTST(OP_FTDIR);
2633 case 'l': FTST(OP_FTLINK);
2634 case 'p': FTST(OP_FTPIPE);
2635 case 'S': FTST(OP_FTSOCK);
2636 case 'u': FTST(OP_FTSUID);
2637 case 'g': FTST(OP_FTSGID);
2638 case 'k': FTST(OP_FTSVTX);
2639 case 'b': FTST(OP_FTBLK);
2640 case 'c': FTST(OP_FTCHR);
2641 case 't': FTST(OP_FTTTY);
2642 case 'T': FTST(OP_FTTEXT);
2643 case 'B': FTST(OP_FTBINARY);
2644 case 'M': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTMTIME);
2645 case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME);
2646 case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME);
2648 Perl_croak(aTHX_ "Unrecognized file test: -%c", (int)tmp);
2655 if (PL_expect == XOPERATOR)
2660 else if (*s == '>') {
2663 if (isIDFIRST_lazy_if(s,UTF)) {
2664 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2672 if (PL_expect == XOPERATOR)
2675 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2677 OPERATOR('-'); /* unary minus */
2684 if (PL_expect == XOPERATOR)
2689 if (PL_expect == XOPERATOR)
2692 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2698 if (PL_expect != XOPERATOR) {
2699 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2700 PL_expect = XOPERATOR;
2701 force_ident(PL_tokenbuf, '*');
2714 if (PL_expect == XOPERATOR) {
2718 PL_tokenbuf[0] = '%';
2719 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2720 if (!PL_tokenbuf[1]) {
2722 yyerror("Final % should be \\% or %name");
2725 PL_pending_ident = '%';
2744 switch (PL_expect) {
2747 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2749 PL_bufptr = s; /* update in case we back off */
2755 PL_expect = XTERMBLOCK;
2759 while (isIDFIRST_lazy_if(s,UTF)) {
2760 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2761 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2762 if (tmp < 0) tmp = -tmp;
2777 d = scan_str(d,TRUE,TRUE);
2780 SvREFCNT_dec(PL_lex_stuff);
2781 PL_lex_stuff = Nullsv;
2783 /* MUST advance bufptr here to avoid bogus
2784 "at end of line" context messages from yyerror().
2786 PL_bufptr = s + len;
2787 yyerror("Unterminated attribute parameter in attribute list");
2790 return 0; /* EOF indicator */
2794 SV *sv = newSVpvn(s, len);
2795 sv_catsv(sv, PL_lex_stuff);
2796 attrs = append_elem(OP_LIST, attrs,
2797 newSVOP(OP_CONST, 0, sv));
2798 SvREFCNT_dec(PL_lex_stuff);
2799 PL_lex_stuff = Nullsv;
2802 attrs = append_elem(OP_LIST, attrs,
2803 newSVOP(OP_CONST, 0,
2807 if (*s == ':' && s[1] != ':')
2810 break; /* require real whitespace or :'s */
2812 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
2813 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
2814 char q = ((*s == '\'') ? '"' : '\'');
2815 /* If here for an expression, and parsed no attrs, back off. */
2816 if (tmp == '=' && !attrs) {
2820 /* MUST advance bufptr here to avoid bogus "at end of line"
2821 context messages from yyerror().
2825 yyerror("Unterminated attribute list");
2827 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
2835 PL_nextval[PL_nexttoke].opval = attrs;
2843 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
2844 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
2849 if (CopLINE(PL_curcop) < PL_copline)
2850 PL_copline = CopLINE(PL_curcop);
2861 if (PL_lex_brackets <= 0)
2862 yyerror("Unmatched right square bracket");
2865 if (PL_lex_state == LEX_INTERPNORMAL) {
2866 if (PL_lex_brackets == 0) {
2867 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
2868 PL_lex_state = LEX_INTERPEND;
2875 if (PL_lex_brackets > 100) {
2876 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
2877 if (newlb != PL_lex_brackstack) {
2879 PL_lex_brackstack = newlb;
2882 switch (PL_expect) {
2884 if (PL_lex_formbrack) {
2888 if (PL_oldoldbufptr == PL_last_lop)
2889 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2891 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2892 OPERATOR(HASHBRACK);
2894 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2897 PL_tokenbuf[0] = '\0';
2898 if (d < PL_bufend && *d == '-') {
2899 PL_tokenbuf[0] = '-';
2901 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2904 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
2905 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
2907 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2910 char minus = (PL_tokenbuf[0] == '-');
2911 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
2919 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
2924 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2929 if (PL_oldoldbufptr == PL_last_lop)
2930 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2932 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2935 OPERATOR(HASHBRACK);
2936 /* This hack serves to disambiguate a pair of curlies
2937 * as being a block or an anon hash. Normally, expectation
2938 * determines that, but in cases where we're not in a
2939 * position to expect anything in particular (like inside
2940 * eval"") we have to resolve the ambiguity. This code
2941 * covers the case where the first term in the curlies is a
2942 * quoted string. Most other cases need to be explicitly
2943 * disambiguated by prepending a `+' before the opening
2944 * curly in order to force resolution as an anon hash.
2946 * XXX should probably propagate the outer expectation
2947 * into eval"" to rely less on this hack, but that could
2948 * potentially break current behavior of eval"".
2952 if (*s == '\'' || *s == '"' || *s == '`') {
2953 /* common case: get past first string, handling escapes */
2954 for (t++; t < PL_bufend && *t != *s;)
2955 if (*t++ == '\\' && (*t == '\\' || *t == *s))
2959 else if (*s == 'q') {
2962 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
2966 char open, close, term;
2969 while (t < PL_bufend && isSPACE(*t))
2973 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
2977 for (t++; t < PL_bufend; t++) {
2978 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
2980 else if (*t == open)
2984 for (t++; t < PL_bufend; t++) {
2985 if (*t == '\\' && t+1 < PL_bufend)
2987 else if (*t == close && --brackets <= 0)
2989 else if (*t == open)
2995 else if (isALNUM_lazy_if(t,UTF)) {
2997 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3000 while (t < PL_bufend && isSPACE(*t))
3002 /* if comma follows first term, call it an anon hash */
3003 /* XXX it could be a comma expression with loop modifiers */
3004 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3005 || (*t == '=' && t[1] == '>')))
3006 OPERATOR(HASHBRACK);
3007 if (PL_expect == XREF)
3010 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3016 yylval.ival = CopLINE(PL_curcop);
3017 if (isSPACE(*s) || *s == '#')
3018 PL_copline = NOLINE; /* invalidate current command line number */
3023 if (PL_lex_brackets <= 0)
3024 yyerror("Unmatched right curly bracket");
3026 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3027 if (PL_lex_brackets < PL_lex_formbrack)
3028 PL_lex_formbrack = 0;
3029 if (PL_lex_state == LEX_INTERPNORMAL) {
3030 if (PL_lex_brackets == 0) {
3031 if (PL_expect & XFAKEBRACK) {
3032 PL_expect &= XENUMMASK;
3033 PL_lex_state = LEX_INTERPEND;
3035 return yylex(); /* ignore fake brackets */
3037 if (*s == '-' && s[1] == '>')
3038 PL_lex_state = LEX_INTERPENDMAYBE;
3039 else if (*s != '[' && *s != '{')
3040 PL_lex_state = LEX_INTERPEND;
3043 if (PL_expect & XFAKEBRACK) {
3044 PL_expect &= XENUMMASK;
3046 return yylex(); /* ignore fake brackets */
3056 if (PL_expect == XOPERATOR) {
3057 if (ckWARN(WARN_SEMICOLON)
3058 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3060 CopLINE_dec(PL_curcop);
3061 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3062 CopLINE_inc(PL_curcop);
3067 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3069 PL_expect = XOPERATOR;
3070 force_ident(PL_tokenbuf, '&');
3074 yylval.ival = (OPpENTERSUB_AMPER<<8);
3093 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3094 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3096 if (PL_expect == XSTATE && isALPHA(tmp) &&
3097 (s == PL_linestart+1 || s[-2] == '\n') )
3099 if (PL_in_eval && !PL_rsfp) {
3104 if (strnEQ(s,"=cut",4)) {
3118 PL_doextract = TRUE;
3121 if (PL_lex_brackets < PL_lex_formbrack) {
3123 #ifdef PERL_STRICT_CR
3124 for (t = s; *t == ' ' || *t == '\t'; t++) ;
3126 for (t = s; *t == ' ' || *t == '\t' || *t == '\r'; t++) ;
3128 if (*t == '\n' || *t == '#') {
3146 if (PL_expect != XOPERATOR) {
3147 if (s[1] != '<' && !strchr(s,'>'))
3150 s = scan_heredoc(s);
3152 s = scan_inputsymbol(s);
3153 TERM(sublex_start());
3158 SHop(OP_LEFT_SHIFT);
3172 SHop(OP_RIGHT_SHIFT);
3181 if (PL_expect == XOPERATOR) {
3182 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3185 return ','; /* grandfather non-comma-format format */
3189 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3190 PL_tokenbuf[0] = '@';
3191 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3192 sizeof PL_tokenbuf - 1, FALSE);
3193 if (PL_expect == XOPERATOR)
3194 no_op("Array length", s);
3195 if (!PL_tokenbuf[1])
3197 PL_expect = XOPERATOR;
3198 PL_pending_ident = '#';
3202 PL_tokenbuf[0] = '$';
3203 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3204 sizeof PL_tokenbuf - 1, FALSE);
3205 if (PL_expect == XOPERATOR)
3207 if (!PL_tokenbuf[1]) {
3209 yyerror("Final $ should be \\$ or $name");
3213 /* This kludge not intended to be bulletproof. */
3214 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3215 yylval.opval = newSVOP(OP_CONST, 0,
3216 newSViv((IV)PL_compiling.cop_arybase));
3217 yylval.opval->op_private = OPpCONST_ARYBASE;
3223 if (PL_lex_state == LEX_NORMAL)
3226 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3229 PL_tokenbuf[0] = '@';
3230 if (ckWARN(WARN_SYNTAX)) {
3232 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3235 PL_bufptr = skipspace(PL_bufptr);
3236 while (t < PL_bufend && *t != ']')
3238 Perl_warner(aTHX_ WARN_SYNTAX,
3239 "Multidimensional syntax %.*s not supported",
3240 (t - PL_bufptr) + 1, PL_bufptr);
3244 else if (*s == '{') {
3245 PL_tokenbuf[0] = '%';
3246 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3247 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3249 char tmpbuf[sizeof PL_tokenbuf];
3251 for (t++; isSPACE(*t); t++) ;
3252 if (isIDFIRST_lazy_if(t,UTF)) {
3253 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3254 for (; isSPACE(*t); t++) ;
3255 if (*t == ';' && get_cv(tmpbuf, FALSE))
3256 Perl_warner(aTHX_ WARN_SYNTAX,
3257 "You need to quote \"%s\"", tmpbuf);
3263 PL_expect = XOPERATOR;
3264 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3265 bool islop = (PL_last_lop == PL_oldoldbufptr);
3266 if (!islop || PL_last_lop_op == OP_GREPSTART)
3267 PL_expect = XOPERATOR;
3268 else if (strchr("$@\"'`q", *s))
3269 PL_expect = XTERM; /* e.g. print $fh "foo" */
3270 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3271 PL_expect = XTERM; /* e.g. print $fh &sub */
3272 else if (isIDFIRST_lazy_if(s,UTF)) {
3273 char tmpbuf[sizeof PL_tokenbuf];
3274 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3275 if (tmp = keyword(tmpbuf, len)) {
3276 /* binary operators exclude handle interpretations */
3288 PL_expect = XTERM; /* e.g. print $fh length() */
3293 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3294 if (gv && GvCVu(gv))
3295 PL_expect = XTERM; /* e.g. print $fh subr() */
3298 else if (isDIGIT(*s))
3299 PL_expect = XTERM; /* e.g. print $fh 3 */
3300 else if (*s == '.' && isDIGIT(s[1]))
3301 PL_expect = XTERM; /* e.g. print $fh .3 */
3302 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3303 PL_expect = XTERM; /* e.g. print $fh -1 */
3304 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3305 PL_expect = XTERM; /* print $fh <<"EOF" */
3307 PL_pending_ident = '$';
3311 if (PL_expect == XOPERATOR)
3313 PL_tokenbuf[0] = '@';
3314 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3315 if (!PL_tokenbuf[1]) {
3317 yyerror("Final @ should be \\@ or @name");
3320 if (PL_lex_state == LEX_NORMAL)
3322 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3324 PL_tokenbuf[0] = '%';
3326 /* Warn about @ where they meant $. */
3327 if (ckWARN(WARN_SYNTAX)) {
3328 if (*s == '[' || *s == '{') {
3330 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3332 if (*t == '}' || *t == ']') {
3334 PL_bufptr = skipspace(PL_bufptr);
3335 Perl_warner(aTHX_ WARN_SYNTAX,
3336 "Scalar value %.*s better written as $%.*s",
3337 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3342 PL_pending_ident = '@';
3345 case '/': /* may either be division or pattern */
3346 case '?': /* may either be conditional or pattern */
3347 if (PL_expect != XOPERATOR) {
3348 /* Disable warning on "study /blah/" */
3349 if (PL_oldoldbufptr == PL_last_uni
3350 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3351 || memNE(PL_last_uni, "study", 5)
3352 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3354 s = scan_pat(s,OP_MATCH);
3355 TERM(sublex_start());
3363 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3364 #ifdef PERL_STRICT_CR
3367 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3369 && (s == PL_linestart || s[-1] == '\n') )
3371 PL_lex_formbrack = 0;
3375 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3381 yylval.ival = OPf_SPECIAL;
3387 if (PL_expect != XOPERATOR)
3392 case '0': case '1': case '2': case '3': case '4':
3393 case '5': case '6': case '7': case '8': case '9':
3395 if (PL_expect == XOPERATOR)
3400 s = scan_str(s,FALSE,FALSE);
3401 if (PL_expect == XOPERATOR) {
3402 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3405 return ','; /* grandfather non-comma-format format */
3411 missingterm((char*)0);
3412 yylval.ival = OP_CONST;
3413 TERM(sublex_start());
3416 s = scan_str(s,FALSE,FALSE);
3417 if (PL_expect == XOPERATOR) {
3418 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3421 return ','; /* grandfather non-comma-format format */
3427 missingterm((char*)0);
3428 yylval.ival = OP_CONST;
3429 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3430 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3431 yylval.ival = OP_STRINGIFY;
3435 TERM(sublex_start());
3438 s = scan_str(s,FALSE,FALSE);
3439 if (PL_expect == XOPERATOR)
3440 no_op("Backticks",s);
3442 missingterm((char*)0);
3443 yylval.ival = OP_BACKTICK;
3445 TERM(sublex_start());
3449 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3450 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3452 if (PL_expect == XOPERATOR)
3453 no_op("Backslash",s);
3457 if (isDIGIT(s[1]) && PL_expect == XTERM) {
3461 while (isDIGIT(*start))
3463 if (*start == '.' && isDIGIT(start[1])) {
3470 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3510 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3512 /* Some keywords can be followed by any delimiter, including ':' */
3513 tmp = (len == 1 && strchr("msyq", PL_tokenbuf[0]) ||
3514 len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3515 (PL_tokenbuf[0] == 'q' &&
3516 strchr("qwxr", PL_tokenbuf[1]))));
3518 /* x::* is just a word, unless x is "CORE" */
3519 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3523 while (d < PL_bufend && isSPACE(*d))
3524 d++; /* no comments skipped here, or s### is misparsed */
3526 /* Is this a label? */
3527 if (!tmp && PL_expect == XSTATE
3528 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3530 yylval.pval = savepv(PL_tokenbuf);
3535 /* Check for keywords */
3536 tmp = keyword(PL_tokenbuf, len);
3538 /* Is this a word before a => operator? */
3539 if (strnEQ(d,"=>",2)) {
3541 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3542 yylval.opval->op_private = OPpCONST_BARE;
3546 if (tmp < 0) { /* second-class keyword? */
3547 GV *ogv = Nullgv; /* override (winner) */
3548 GV *hgv = Nullgv; /* hidden (loser) */
3549 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3551 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3554 if (GvIMPORTED_CV(gv))
3556 else if (! CvMETHOD(cv))
3560 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3561 (gv = *gvp) != (GV*)&PL_sv_undef &&
3562 GvCVu(gv) && GvIMPORTED_CV(gv))
3568 tmp = 0; /* overridden by import or by GLOBAL */
3571 && -tmp==KEY_lock /* XXX generalizable kludge */
3573 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3575 tmp = 0; /* any sub overrides "weak" keyword */
3577 else { /* no override */
3581 if (ckWARN(WARN_AMBIGUOUS) && hgv
3582 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3583 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3584 "Ambiguous call resolved as CORE::%s(), %s",
3585 GvENAME(hgv), "qualify as such or use &");
3592 default: /* not a keyword */
3595 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3597 /* Get the rest if it looks like a package qualifier */
3599 if (*s == '\'' || *s == ':' && s[1] == ':') {
3601 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3604 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3605 *s == '\'' ? "'" : "::");
3609 if (PL_expect == XOPERATOR) {
3610 if (PL_bufptr == PL_linestart) {
3611 CopLINE_dec(PL_curcop);
3612 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3613 CopLINE_inc(PL_curcop);
3616 no_op("Bareword",s);
3619 /* Look for a subroutine with this name in current package,
3620 unless name is "Foo::", in which case Foo is a bearword
3621 (and a package name). */
3624 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3626 if (ckWARN(WARN_UNSAFE) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3627 Perl_warner(aTHX_ WARN_UNSAFE,
3628 "Bareword \"%s\" refers to nonexistent package",
3631 PL_tokenbuf[len] = '\0';
3638 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3641 /* if we saw a global override before, get the right name */
3644 sv = newSVpvn("CORE::GLOBAL::",14);
3645 sv_catpv(sv,PL_tokenbuf);
3648 sv = newSVpv(PL_tokenbuf,0);
3650 /* Presume this is going to be a bareword of some sort. */
3653 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3654 yylval.opval->op_private = OPpCONST_BARE;
3656 /* And if "Foo::", then that's what it certainly is. */
3661 /* See if it's the indirect object for a list operator. */
3663 if (PL_oldoldbufptr &&
3664 PL_oldoldbufptr < PL_bufptr &&
3665 (PL_oldoldbufptr == PL_last_lop
3666 || PL_oldoldbufptr == PL_last_uni) &&
3667 /* NO SKIPSPACE BEFORE HERE! */
3668 (PL_expect == XREF ||
3669 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3671 bool immediate_paren = *s == '(';
3673 /* (Now we can afford to cross potential line boundary.) */
3676 /* Two barewords in a row may indicate method call. */
3678 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3681 /* If not a declared subroutine, it's an indirect object. */
3682 /* (But it's an indir obj regardless for sort.) */
3684 if ((PL_last_lop_op == OP_SORT ||
3685 (!immediate_paren && (!gv || !GvCVu(gv)))) &&
3686 (PL_last_lop_op != OP_MAPSTART &&
3687 PL_last_lop_op != OP_GREPSTART))
3689 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3694 /* If followed by a paren, it's certainly a subroutine. */
3696 PL_expect = XOPERATOR;
3700 if (gv && GvCVu(gv)) {
3701 for (d = s + 1; *d == ' ' || *d == '\t'; d++) ;
3702 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3707 PL_nextval[PL_nexttoke].opval = yylval.opval;
3708 PL_expect = XOPERATOR;
3714 /* If followed by var or block, call it a method (unless sub) */
3716 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3717 PL_last_lop = PL_oldbufptr;
3718 PL_last_lop_op = OP_METHOD;
3722 /* If followed by a bareword, see if it looks like indir obj. */
3724 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3727 /* Not a method, so call it a subroutine (if defined) */
3729 if (gv && GvCVu(gv)) {
3731 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3732 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3733 "Ambiguous use of -%s resolved as -&%s()",
3734 PL_tokenbuf, PL_tokenbuf);
3735 /* Check for a constant sub */
3737 if ((sv = cv_const_sv(cv))) {
3739 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3740 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3741 yylval.opval->op_private = 0;
3745 /* Resolve to GV now. */
3746 op_free(yylval.opval);
3747 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3748 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3749 PL_last_lop = PL_oldbufptr;
3750 PL_last_lop_op = OP_ENTERSUB;
3751 /* Is there a prototype? */
3754 char *proto = SvPV((SV*)cv, len);
3757 if (strEQ(proto, "$"))
3759 if (*proto == '&' && *s == '{') {
3760 sv_setpv(PL_subname,"__ANON__");
3764 PL_nextval[PL_nexttoke].opval = yylval.opval;
3770 /* Call it a bare word */
3772 if (PL_hints & HINT_STRICT_SUBS)
3773 yylval.opval->op_private |= OPpCONST_STRICT;
3776 if (ckWARN(WARN_RESERVED)) {
3777 if (lastchar != '-') {
3778 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
3780 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
3787 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
3788 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3789 "Operator or semicolon missing before %c%s",
3790 lastchar, PL_tokenbuf);
3791 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3792 "Ambiguous use of %c resolved as operator %c",
3793 lastchar, lastchar);
3799 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3800 newSVpv(CopFILE(PL_curcop),0));
3804 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3805 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
3808 case KEY___PACKAGE__:
3809 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3811 ? newSVsv(PL_curstname)
3820 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
3821 char *pname = "main";
3822 if (PL_tokenbuf[2] == 'D')
3823 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
3824 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
3827 GvIOp(gv) = newIO();
3828 IoIFP(GvIOp(gv)) = PL_rsfp;
3829 #if defined(HAS_FCNTL) && defined(F_SETFD)
3831 int fd = PerlIO_fileno(PL_rsfp);
3832 fcntl(fd,F_SETFD,fd >= 3);
3835 /* Mark this internal pseudo-handle as clean */
3836 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
3838 IoTYPE(GvIOp(gv)) = '|';
3839 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
3840 IoTYPE(GvIOp(gv)) = '-';
3842 IoTYPE(GvIOp(gv)) = '<';
3843 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
3844 /* if the script was opened in binmode, we need to revert
3845 * it to text mode for compatibility; but only iff it has CRs
3846 * XXX this is a questionable hack at best. */
3847 if (PL_bufend-PL_bufptr > 2
3848 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
3851 if (IoTYPE(GvIOp(gv)) == '<') {
3852 loc = PerlIO_tell(PL_rsfp);
3853 (void)PerlIO_seek(PL_rsfp, 0L, 0);
3855 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
3856 #if defined(__BORLANDC__)
3857 /* XXX see note in do_binmode() */
3858 ((FILE*)PL_rsfp)->flags |= _F_BIN;
3861 PerlIO_seek(PL_rsfp, loc, 0);
3876 if (PL_expect == XSTATE) {
3883 if (*s == ':' && s[1] == ':') {
3886 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3887 tmp = keyword(PL_tokenbuf, len);
3901 LOP(OP_ACCEPT,XTERM);
3907 LOP(OP_ATAN2,XTERM);
3916 LOP(OP_BLESS,XTERM);
3925 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
3942 if (!PL_cryptseen) {
3943 PL_cryptseen = TRUE;
3947 LOP(OP_CRYPT,XTERM);
3950 if (ckWARN(WARN_OCTAL)) {
3951 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
3952 if (*d != '0' && isDIGIT(*d))
3953 Perl_warner(aTHX_ WARN_OCTAL,
3954 "chmod: mode argument is missing initial 0");
3956 LOP(OP_CHMOD,XTERM);
3959 LOP(OP_CHOWN,XTERM);
3962 LOP(OP_CONNECT,XTERM);
3978 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3982 PL_hints |= HINT_BLOCK_SCOPE;
3992 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
3993 LOP(OP_DBMOPEN,XTERM);
3999 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4006 yylval.ival = CopLINE(PL_curcop);
4020 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4021 UNIBRACK(OP_ENTEREVAL);
4036 case KEY_endhostent:
4042 case KEY_endservent:
4045 case KEY_endprotoent:
4056 yylval.ival = CopLINE(PL_curcop);
4058 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4060 if ((PL_bufend - p) >= 3 &&
4061 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4063 else if ((PL_bufend - p) >= 4 &&
4064 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4067 if (isIDFIRST_lazy_if(p,UTF)) {
4068 p = scan_ident(p, PL_bufend,
4069 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4073 Perl_croak(aTHX_ "Missing $ on loop variable");
4078 LOP(OP_FORMLINE,XTERM);
4084 LOP(OP_FCNTL,XTERM);
4090 LOP(OP_FLOCK,XTERM);
4099 LOP(OP_GREPSTART, XREF);
4102 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4117 case KEY_getpriority:
4118 LOP(OP_GETPRIORITY,XTERM);
4120 case KEY_getprotobyname:
4123 case KEY_getprotobynumber:
4124 LOP(OP_GPBYNUMBER,XTERM);
4126 case KEY_getprotoent:
4138 case KEY_getpeername:
4139 UNI(OP_GETPEERNAME);
4141 case KEY_gethostbyname:
4144 case KEY_gethostbyaddr:
4145 LOP(OP_GHBYADDR,XTERM);
4147 case KEY_gethostent:
4150 case KEY_getnetbyname:
4153 case KEY_getnetbyaddr:
4154 LOP(OP_GNBYADDR,XTERM);
4159 case KEY_getservbyname:
4160 LOP(OP_GSBYNAME,XTERM);
4162 case KEY_getservbyport:
4163 LOP(OP_GSBYPORT,XTERM);
4165 case KEY_getservent:
4168 case KEY_getsockname:
4169 UNI(OP_GETSOCKNAME);
4171 case KEY_getsockopt:
4172 LOP(OP_GSOCKOPT,XTERM);
4194 yylval.ival = CopLINE(PL_curcop);
4198 LOP(OP_INDEX,XTERM);
4204 LOP(OP_IOCTL,XTERM);
4216 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4248 LOP(OP_LISTEN,XTERM);
4257 s = scan_pat(s,OP_MATCH);
4258 TERM(sublex_start());
4261 LOP(OP_MAPSTART, XREF);
4264 LOP(OP_MKDIR,XTERM);
4267 LOP(OP_MSGCTL,XTERM);
4270 LOP(OP_MSGGET,XTERM);
4273 LOP(OP_MSGRCV,XTERM);
4276 LOP(OP_MSGSND,XTERM);
4282 if (isIDFIRST_lazy_if(s,UTF)) {
4283 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4284 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4286 PL_in_my_stash = gv_stashpv(PL_tokenbuf, FALSE);
4287 if (!PL_in_my_stash) {
4290 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4298 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4305 if (PL_expect != XSTATE)
4306 yyerror("\"no\" not allowed in expression");
4307 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4308 s = force_version(s);
4313 if (*s == '(' || (s = skipspace(s), *s == '('))
4320 if (isIDFIRST_lazy_if(s,UTF)) {
4322 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4324 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_AMBIGUOUS))
4325 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4326 "Precedence problem: open %.*s should be open(%.*s)",
4332 yylval.ival = OP_OR;
4342 LOP(OP_OPEN_DIR,XTERM);
4345 checkcomma(s,PL_tokenbuf,"filehandle");
4349 checkcomma(s,PL_tokenbuf,"filehandle");
4368 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4372 LOP(OP_PIPE_OP,XTERM);
4375 s = scan_str(s,FALSE,FALSE);
4377 missingterm((char*)0);
4378 yylval.ival = OP_CONST;
4379 TERM(sublex_start());
4385 s = scan_str(s,FALSE,FALSE);
4387 missingterm((char*)0);
4389 if (SvCUR(PL_lex_stuff)) {
4392 d = SvPV_force(PL_lex_stuff, len);
4394 for (; isSPACE(*d) && len; --len, ++d) ;
4397 if (!warned && ckWARN(WARN_SYNTAX)) {
4398 for (; !isSPACE(*d) && len; --len, ++d) {
4400 Perl_warner(aTHX_ WARN_SYNTAX,
4401 "Possible attempt to separate words with commas");
4404 else if (*d == '#') {
4405 Perl_warner(aTHX_ WARN_SYNTAX,
4406 "Possible attempt to put comments in qw() list");
4412 for (; !isSPACE(*d) && len; --len, ++d) ;
4414 words = append_elem(OP_LIST, words,
4415 newSVOP(OP_CONST, 0, newSVpvn(b, d-b)));
4419 PL_nextval[PL_nexttoke].opval = words;
4424 SvREFCNT_dec(PL_lex_stuff);
4425 PL_lex_stuff = Nullsv;
4430 s = scan_str(s,FALSE,FALSE);
4432 missingterm((char*)0);
4433 yylval.ival = OP_STRINGIFY;
4434 if (SvIVX(PL_lex_stuff) == '\'')
4435 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4436 TERM(sublex_start());
4439 s = scan_pat(s,OP_QR);
4440 TERM(sublex_start());
4443 s = scan_str(s,FALSE,FALSE);
4445 missingterm((char*)0);
4446 yylval.ival = OP_BACKTICK;
4448 TERM(sublex_start());
4455 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4456 s = force_version(s);
4459 *PL_tokenbuf = '\0';
4460 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4461 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4462 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4464 yyerror("<> should be quotes");
4472 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4476 LOP(OP_RENAME,XTERM);
4485 LOP(OP_RINDEX,XTERM);
4508 LOP(OP_REVERSE,XTERM);
4519 TERM(sublex_start());
4521 TOKEN(1); /* force error */
4530 LOP(OP_SELECT,XTERM);
4536 LOP(OP_SEMCTL,XTERM);
4539 LOP(OP_SEMGET,XTERM);
4542 LOP(OP_SEMOP,XTERM);
4548 LOP(OP_SETPGRP,XTERM);
4550 case KEY_setpriority:
4551 LOP(OP_SETPRIORITY,XTERM);
4553 case KEY_sethostent:
4559 case KEY_setservent:
4562 case KEY_setprotoent:
4572 LOP(OP_SEEKDIR,XTERM);
4574 case KEY_setsockopt:
4575 LOP(OP_SSOCKOPT,XTERM);
4581 LOP(OP_SHMCTL,XTERM);
4584 LOP(OP_SHMGET,XTERM);
4587 LOP(OP_SHMREAD,XTERM);
4590 LOP(OP_SHMWRITE,XTERM);
4593 LOP(OP_SHUTDOWN,XTERM);
4602 LOP(OP_SOCKET,XTERM);
4604 case KEY_socketpair:
4605 LOP(OP_SOCKPAIR,XTERM);
4608 checkcomma(s,PL_tokenbuf,"subroutine name");
4610 if (*s == ';' || *s == ')') /* probably a close */
4611 Perl_croak(aTHX_ "sort is now a reserved word");
4613 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4617 LOP(OP_SPLIT,XTERM);
4620 LOP(OP_SPRINTF,XTERM);
4623 LOP(OP_SPLICE,XTERM);
4638 LOP(OP_SUBSTR,XTERM);
4644 char tmpbuf[sizeof PL_tokenbuf];
4646 expectation attrful;
4647 bool have_name, have_proto;
4652 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4653 (*s == ':' && s[1] == ':'))
4656 attrful = XATTRBLOCK;
4657 /* remember buffer pos'n for later force_word */
4658 tboffset = s - PL_oldbufptr;
4659 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4660 if (strchr(tmpbuf, ':'))
4661 sv_setpv(PL_subname, tmpbuf);
4663 sv_setsv(PL_subname,PL_curstname);
4664 sv_catpvn(PL_subname,"::",2);
4665 sv_catpvn(PL_subname,tmpbuf,len);
4672 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4673 PL_expect = XTERMBLOCK;
4674 attrful = XATTRTERM;
4675 sv_setpv(PL_subname,"?");
4679 if (key == KEY_format) {
4681 PL_lex_formbrack = PL_lex_brackets + 1;
4683 (void) force_word(PL_oldbufptr + tboffset, WORD,
4688 /* Look for a prototype */
4692 s = scan_str(s,FALSE,FALSE);
4695 SvREFCNT_dec(PL_lex_stuff);
4696 PL_lex_stuff = Nullsv;
4697 Perl_croak(aTHX_ "Prototype not terminated");
4700 d = SvPVX(PL_lex_stuff);
4702 for (p = d; *p; ++p) {
4707 SvCUR(PL_lex_stuff) = tmp;
4715 if (*s == ':' && s[1] != ':')
4716 PL_expect = attrful;
4719 PL_nextval[PL_nexttoke].opval =
4720 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4721 PL_lex_stuff = Nullsv;
4725 sv_setpv(PL_subname,"__ANON__");
4728 (void) force_word(PL_oldbufptr + tboffset, WORD,
4737 LOP(OP_SYSTEM,XREF);
4740 LOP(OP_SYMLINK,XTERM);
4743 LOP(OP_SYSCALL,XTERM);
4746 LOP(OP_SYSOPEN,XTERM);
4749 LOP(OP_SYSSEEK,XTERM);
4752 LOP(OP_SYSREAD,XTERM);
4755 LOP(OP_SYSWRITE,XTERM);
4759 TERM(sublex_start());
4780 LOP(OP_TRUNCATE,XTERM);
4792 yylval.ival = CopLINE(PL_curcop);
4796 yylval.ival = CopLINE(PL_curcop);
4800 LOP(OP_UNLINK,XTERM);
4806 LOP(OP_UNPACK,XTERM);
4809 LOP(OP_UTIME,XTERM);
4812 if (ckWARN(WARN_OCTAL)) {
4813 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4814 if (*d != '0' && isDIGIT(*d))
4815 Perl_warner(aTHX_ WARN_OCTAL,
4816 "umask: argument is missing initial 0");
4821 LOP(OP_UNSHIFT,XTERM);
4824 if (PL_expect != XSTATE)
4825 yyerror("\"use\" not allowed in expression");
4827 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4828 s = force_version(s);
4829 if (*s == ';' || (s = skipspace(s), *s == ';')) {
4830 PL_nextval[PL_nexttoke].opval = Nullop;
4835 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4836 s = force_version(s);
4848 yylval.ival = CopLINE(PL_curcop);
4852 PL_hints |= HINT_BLOCK_SCOPE;
4859 LOP(OP_WAITPID,XTERM);
4867 static char ctl_l[2];
4869 if (ctl_l[0] == '\0')
4870 ctl_l[0] = toCTRL('L');
4871 gv_fetchpv(ctl_l,TRUE, SVt_PV);
4874 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
4879 if (PL_expect == XOPERATOR)
4885 yylval.ival = OP_XOR;
4890 TERM(sublex_start());
4896 Perl_keyword(pTHX_ register char *d, I32 len)
4901 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
4902 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
4903 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
4904 if (strEQ(d,"__DATA__")) return KEY___DATA__;
4905 if (strEQ(d,"__END__")) return KEY___END__;
4909 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
4914 if (strEQ(d,"and")) return -KEY_and;
4915 if (strEQ(d,"abs")) return -KEY_abs;
4918 if (strEQ(d,"alarm")) return -KEY_alarm;
4919 if (strEQ(d,"atan2")) return -KEY_atan2;
4922 if (strEQ(d,"accept")) return -KEY_accept;
4927 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
4930 if (strEQ(d,"bless")) return -KEY_bless;
4931 if (strEQ(d,"bind")) return -KEY_bind;
4932 if (strEQ(d,"binmode")) return -KEY_binmode;
4935 if (strEQ(d,"CORE")) return -KEY_CORE;
4936 if (strEQ(d,"CHECK")) return KEY_CHECK;
4941 if (strEQ(d,"cmp")) return -KEY_cmp;
4942 if (strEQ(d,"chr")) return -KEY_chr;
4943 if (strEQ(d,"cos")) return -KEY_cos;
4946 if (strEQ(d,"chop")) return KEY_chop;
4949 if (strEQ(d,"close")) return -KEY_close;
4950 if (strEQ(d,"chdir")) return -KEY_chdir;
4951 if (strEQ(d,"chomp")) return KEY_chomp;
4952 if (strEQ(d,"chmod")) return -KEY_chmod;
4953 if (strEQ(d,"chown")) return -KEY_chown;
4954 if (strEQ(d,"crypt")) return -KEY_crypt;
4957 if (strEQ(d,"chroot")) return -KEY_chroot;
4958 if (strEQ(d,"caller")) return -KEY_caller;
4961 if (strEQ(d,"connect")) return -KEY_connect;
4964 if (strEQ(d,"closedir")) return -KEY_closedir;
4965 if (strEQ(d,"continue")) return -KEY_continue;
4970 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
4975 if (strEQ(d,"do")) return KEY_do;
4978 if (strEQ(d,"die")) return -KEY_die;
4981 if (strEQ(d,"dump")) return -KEY_dump;
4984 if (strEQ(d,"delete")) return KEY_delete;
4987 if (strEQ(d,"defined")) return KEY_defined;
4988 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
4991 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
4996 if (strEQ(d,"EQ")) { deprecate(d); return -KEY_eq;}
4997 if (strEQ(d,"END")) return KEY_END;
5002 if (strEQ(d,"eq")) return -KEY_eq;
5005 if (strEQ(d,"eof")) return -KEY_eof;
5006 if (strEQ(d,"exp")) return -KEY_exp;
5009 if (strEQ(d,"else")) return KEY_else;
5010 if (strEQ(d,"exit")) return -KEY_exit;
5011 if (strEQ(d,"eval")) return KEY_eval;
5012 if (strEQ(d,"exec")) return -KEY_exec;
5013 if (strEQ(d,"each")) return KEY_each;
5016 if (strEQ(d,"elsif")) return KEY_elsif;
5019 if (strEQ(d,"exists")) return KEY_exists;
5020 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5023 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5024 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5027 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5030 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5031 if (strEQ(d,"endservent")) return -KEY_endservent;
5034 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5041 if (strEQ(d,"for")) return KEY_for;
5044 if (strEQ(d,"fork")) return -KEY_fork;
5047 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5048 if (strEQ(d,"flock")) return -KEY_flock;
5051 if (strEQ(d,"format")) return KEY_format;
5052 if (strEQ(d,"fileno")) return -KEY_fileno;
5055 if (strEQ(d,"foreach")) return KEY_foreach;
5058 if (strEQ(d,"formline")) return -KEY_formline;
5064 if (strEQ(d,"GT")) { deprecate(d); return -KEY_gt;}
5065 if (strEQ(d,"GE")) { deprecate(d); return -KEY_ge;}
5069 if (strnEQ(d,"get",3)) {
5074 if (strEQ(d,"ppid")) return -KEY_getppid;
5075 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5078 if (strEQ(d,"pwent")) return -KEY_getpwent;
5079 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5080 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5083 if (strEQ(d,"peername")) return -KEY_getpeername;
5084 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5085 if (strEQ(d,"priority")) return -KEY_getpriority;
5088 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5091 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5095 else if (*d == 'h') {
5096 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5097 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5098 if (strEQ(d,"hostent")) return -KEY_gethostent;
5100 else if (*d == 'n') {
5101 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5102 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5103 if (strEQ(d,"netent")) return -KEY_getnetent;
5105 else if (*d == 's') {
5106 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5107 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5108 if (strEQ(d,"servent")) return -KEY_getservent;
5109 if (strEQ(d,"sockname")) return -KEY_getsockname;
5110 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5112 else if (*d == 'g') {
5113 if (strEQ(d,"grent")) return -KEY_getgrent;
5114 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5115 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5117 else if (*d == 'l') {
5118 if (strEQ(d,"login")) return -KEY_getlogin;
5120 else if (strEQ(d,"c")) return -KEY_getc;
5125 if (strEQ(d,"gt")) return -KEY_gt;
5126 if (strEQ(d,"ge")) return -KEY_ge;
5129 if (strEQ(d,"grep")) return KEY_grep;
5130 if (strEQ(d,"goto")) return KEY_goto;
5131 if (strEQ(d,"glob")) return KEY_glob;
5134 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5139 if (strEQ(d,"hex")) return -KEY_hex;
5142 if (strEQ(d,"INIT")) return KEY_INIT;
5147 if (strEQ(d,"if")) return KEY_if;
5150 if (strEQ(d,"int")) return -KEY_int;
5153 if (strEQ(d,"index")) return -KEY_index;
5154 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5159 if (strEQ(d,"join")) return -KEY_join;
5163 if (strEQ(d,"keys")) return KEY_keys;
5164 if (strEQ(d,"kill")) return -KEY_kill;
5169 if (strEQ(d,"LT")) { deprecate(d); return -KEY_lt;}
5170 if (strEQ(d,"LE")) { deprecate(d); return -KEY_le;}
5176 if (strEQ(d,"lt")) return -KEY_lt;
5177 if (strEQ(d,"le")) return -KEY_le;
5178 if (strEQ(d,"lc")) return -KEY_lc;
5181 if (strEQ(d,"log")) return -KEY_log;
5184 if (strEQ(d,"last")) return KEY_last;
5185 if (strEQ(d,"link")) return -KEY_link;
5186 if (strEQ(d,"lock")) return -KEY_lock;
5189 if (strEQ(d,"local")) return KEY_local;
5190 if (strEQ(d,"lstat")) return -KEY_lstat;
5193 if (strEQ(d,"length")) return -KEY_length;
5194 if (strEQ(d,"listen")) return -KEY_listen;
5197 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5200 if (strEQ(d,"localtime")) return -KEY_localtime;
5206 case 1: return KEY_m;
5208 if (strEQ(d,"my")) return KEY_my;
5211 if (strEQ(d,"map")) return KEY_map;
5214 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5217 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5218 if (strEQ(d,"msgget")) return -KEY_msgget;
5219 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5220 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5225 if (strEQ(d,"NE")) { deprecate(d); return -KEY_ne;}
5228 if (strEQ(d,"next")) return KEY_next;
5229 if (strEQ(d,"ne")) return -KEY_ne;
5230 if (strEQ(d,"not")) return -KEY_not;
5231 if (strEQ(d,"no")) return KEY_no;
5236 if (strEQ(d,"or")) return -KEY_or;
5239 if (strEQ(d,"ord")) return -KEY_ord;
5240 if (strEQ(d,"oct")) return -KEY_oct;
5241 if (strEQ(d,"our")) return KEY_our;
5244 if (strEQ(d,"open")) return -KEY_open;
5247 if (strEQ(d,"opendir")) return -KEY_opendir;
5254 if (strEQ(d,"pop")) return KEY_pop;
5255 if (strEQ(d,"pos")) return KEY_pos;
5258 if (strEQ(d,"push")) return KEY_push;
5259 if (strEQ(d,"pack")) return -KEY_pack;
5260 if (strEQ(d,"pipe")) return -KEY_pipe;
5263 if (strEQ(d,"print")) return KEY_print;
5266 if (strEQ(d,"printf")) return KEY_printf;
5269 if (strEQ(d,"package")) return KEY_package;
5272 if (strEQ(d,"prototype")) return KEY_prototype;
5277 if (strEQ(d,"q")) return KEY_q;
5278 if (strEQ(d,"qr")) return KEY_qr;
5279 if (strEQ(d,"qq")) return KEY_qq;
5280 if (strEQ(d,"qw")) return KEY_qw;
5281 if (strEQ(d,"qx")) return KEY_qx;
5283 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5288 if (strEQ(d,"ref")) return -KEY_ref;
5291 if (strEQ(d,"read")) return -KEY_read;
5292 if (strEQ(d,"rand")) return -KEY_rand;
5293 if (strEQ(d,"recv")) return -KEY_recv;
5294 if (strEQ(d,"redo")) return KEY_redo;
5297 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5298 if (strEQ(d,"reset")) return -KEY_reset;
5301 if (strEQ(d,"return")) return KEY_return;
5302 if (strEQ(d,"rename")) return -KEY_rename;
5303 if (strEQ(d,"rindex")) return -KEY_rindex;
5306 if (strEQ(d,"require")) return -KEY_require;
5307 if (strEQ(d,"reverse")) return -KEY_reverse;
5308 if (strEQ(d,"readdir")) return -KEY_readdir;
5311 if (strEQ(d,"readlink")) return -KEY_readlink;
5312 if (strEQ(d,"readline")) return -KEY_readline;
5313 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5316 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5322 case 0: return KEY_s;
5324 if (strEQ(d,"scalar")) return KEY_scalar;
5329 if (strEQ(d,"seek")) return -KEY_seek;
5330 if (strEQ(d,"send")) return -KEY_send;
5333 if (strEQ(d,"semop")) return -KEY_semop;
5336 if (strEQ(d,"select")) return -KEY_select;
5337 if (strEQ(d,"semctl")) return -KEY_semctl;
5338 if (strEQ(d,"semget")) return -KEY_semget;
5341 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5342 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5345 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5346 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5349 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5352 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5353 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5354 if (strEQ(d,"setservent")) return -KEY_setservent;
5357 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5358 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5365 if (strEQ(d,"shift")) return KEY_shift;
5368 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5369 if (strEQ(d,"shmget")) return -KEY_shmget;
5372 if (strEQ(d,"shmread")) return -KEY_shmread;
5375 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5376 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5381 if (strEQ(d,"sin")) return -KEY_sin;
5384 if (strEQ(d,"sleep")) return -KEY_sleep;
5387 if (strEQ(d,"sort")) return KEY_sort;
5388 if (strEQ(d,"socket")) return -KEY_socket;
5389 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5392 if (strEQ(d,"split")) return KEY_split;
5393 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5394 if (strEQ(d,"splice")) return KEY_splice;
5397 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5400 if (strEQ(d,"srand")) return -KEY_srand;
5403 if (strEQ(d,"stat")) return -KEY_stat;
5404 if (strEQ(d,"study")) return KEY_study;
5407 if (strEQ(d,"substr")) return -KEY_substr;
5408 if (strEQ(d,"sub")) return KEY_sub;
5413 if (strEQ(d,"system")) return -KEY_system;
5416 if (strEQ(d,"symlink")) return -KEY_symlink;
5417 if (strEQ(d,"syscall")) return -KEY_syscall;
5418 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5419 if (strEQ(d,"sysread")) return -KEY_sysread;
5420 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5423 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5432 if (strEQ(d,"tr")) return KEY_tr;
5435 if (strEQ(d,"tie")) return KEY_tie;
5438 if (strEQ(d,"tell")) return -KEY_tell;
5439 if (strEQ(d,"tied")) return KEY_tied;
5440 if (strEQ(d,"time")) return -KEY_time;
5443 if (strEQ(d,"times")) return -KEY_times;
5446 if (strEQ(d,"telldir")) return -KEY_telldir;
5449 if (strEQ(d,"truncate")) return -KEY_truncate;
5456 if (strEQ(d,"uc")) return -KEY_uc;
5459 if (strEQ(d,"use")) return KEY_use;
5462 if (strEQ(d,"undef")) return KEY_undef;
5463 if (strEQ(d,"until")) return KEY_until;
5464 if (strEQ(d,"untie")) return KEY_untie;
5465 if (strEQ(d,"utime")) return -KEY_utime;
5466 if (strEQ(d,"umask")) return -KEY_umask;
5469 if (strEQ(d,"unless")) return KEY_unless;
5470 if (strEQ(d,"unpack")) return -KEY_unpack;
5471 if (strEQ(d,"unlink")) return -KEY_unlink;
5474 if (strEQ(d,"unshift")) return KEY_unshift;
5475 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5480 if (strEQ(d,"values")) return -KEY_values;
5481 if (strEQ(d,"vec")) return -KEY_vec;
5486 if (strEQ(d,"warn")) return -KEY_warn;
5487 if (strEQ(d,"wait")) return -KEY_wait;
5490 if (strEQ(d,"while")) return KEY_while;
5491 if (strEQ(d,"write")) return -KEY_write;
5494 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5497 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5502 if (len == 1) return -KEY_x;
5503 if (strEQ(d,"xor")) return -KEY_xor;
5506 if (len == 1) return KEY_y;
5515 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5519 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5520 dTHR; /* only for ckWARN */
5521 if (ckWARN(WARN_SYNTAX)) {
5523 for (w = s+2; *w && level; w++) {
5530 for (; *w && isSPACE(*w); w++) ;
5531 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5532 Perl_warner(aTHX_ WARN_SYNTAX,
5533 "%s (...) interpreted as function",name);
5536 while (s < PL_bufend && isSPACE(*s))
5540 while (s < PL_bufend && isSPACE(*s))
5542 if (isIDFIRST_lazy_if(s,UTF)) {
5544 while (isALNUM_lazy_if(s,UTF))
5546 while (s < PL_bufend && isSPACE(*s))
5551 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5555 Perl_croak(aTHX_ "No comma allowed after %s", what);
5560 /* Either returns sv, or mortalizes sv and returns a new SV*.
5561 Best used as sv=new_constant(..., sv, ...).
5562 If s, pv are NULL, calls subroutine with one argument,
5563 and type is used with error messages only. */
5566 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5570 HV *table = GvHV(PL_hintgv); /* ^H */
5574 const char *why, *why1, *why2;
5576 if (!(PL_hints & HINT_LOCALIZE_HH)) {
5579 why = "%^H is not localized";
5583 msg = Perl_newSVpvf(aTHX_ "constant(%s): %s%s%s",
5584 (type ? type: "undef"), why1, why2, why);
5585 yyerror(SvPVX(msg));
5590 why = "%^H is not defined";
5593 cvp = hv_fetch(table, key, strlen(key), FALSE);
5594 if (!cvp || !SvOK(*cvp)) {
5595 why = "} is not defined";
5600 sv_2mortal(sv); /* Parent created it permanently */
5603 pv = sv_2mortal(newSVpvn(s, len));
5605 typesv = sv_2mortal(newSVpv(type, 0));
5607 typesv = &PL_sv_undef;
5609 PUSHSTACKi(PERLSI_OVERLOAD);
5622 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5626 /* Check the eval first */
5627 if (!PL_in_eval && SvTRUE(ERRSV))
5630 sv_catpv(ERRSV, "Propagated");
5631 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5633 res = SvREFCNT_inc(sv);
5637 (void)SvREFCNT_inc(res);
5646 why = "}} did not return a defined value";
5647 why1 = "Call to &{$^H{";
5657 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5659 register char *d = dest;
5660 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5663 Perl_croak(aTHX_ ident_too_long);
5664 if (isALNUM(*s)) /* UTF handled below */
5666 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
5671 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5675 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5676 char *t = s + UTF8SKIP(s);
5677 while (*t & 0x80 && is_utf8_mark((U8*)t))
5679 if (d + (t - s) > e)
5680 Perl_croak(aTHX_ ident_too_long);
5681 Copy(s, d, t - s, char);
5694 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5704 e = d + destlen - 3; /* two-character token, ending NUL */
5706 while (isDIGIT(*s)) {
5708 Perl_croak(aTHX_ ident_too_long);
5715 Perl_croak(aTHX_ ident_too_long);
5716 if (isALNUM(*s)) /* UTF handled below */
5718 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
5723 else if (*s == ':' && s[1] == ':') {
5727 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5728 char *t = s + UTF8SKIP(s);
5729 while (*t & 0x80 && is_utf8_mark((U8*)t))
5731 if (d + (t - s) > e)
5732 Perl_croak(aTHX_ ident_too_long);
5733 Copy(s, d, t - s, char);
5744 if (PL_lex_state != LEX_NORMAL)
5745 PL_lex_state = LEX_INTERPENDMAYBE;
5748 if (*s == '$' && s[1] &&
5749 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
5762 if (*d == '^' && *s && isCONTROLVAR(*s)) {
5767 if (isSPACE(s[-1])) {
5770 if (ch != ' ' && ch != '\t') {
5776 if (isIDFIRST_lazy_if(d,UTF)) {
5780 while (e < send && isALNUM_lazy_if(e,UTF) || *e == ':') {
5782 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
5785 Copy(s, d, e - s, char);
5790 while ((isALNUM(*s) || *s == ':') && d < e)
5793 Perl_croak(aTHX_ ident_too_long);
5796 while (s < send && (*s == ' ' || *s == '\t')) s++;
5797 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
5798 dTHR; /* only for ckWARN */
5799 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
5800 const char *brack = *s == '[' ? "[...]" : "{...}";
5801 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5802 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
5803 funny, dest, brack, funny, dest, brack);
5806 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
5810 /* Handle extended ${^Foo} variables
5811 * 1999-02-27 mjd-perl-patch@plover.com */
5812 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
5816 while (isALNUM(*s) && d < e) {
5820 Perl_croak(aTHX_ ident_too_long);
5825 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
5826 PL_lex_state = LEX_INTERPEND;
5829 if (PL_lex_state == LEX_NORMAL) {
5830 dTHR; /* only for ckWARN */
5831 if (ckWARN(WARN_AMBIGUOUS) &&
5832 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
5834 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5835 "Ambiguous use of %c{%s} resolved to %c%s",
5836 funny, dest, funny, dest);
5841 s = bracket; /* let the parser handle it */
5845 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
5846 PL_lex_state = LEX_INTERPEND;
5851 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
5856 *pmfl |= PMf_GLOBAL;
5858 *pmfl |= PMf_CONTINUE;
5862 *pmfl |= PMf_MULTILINE;
5864 *pmfl |= PMf_SINGLELINE;
5866 *pmfl |= PMf_EXTENDED;
5870 S_scan_pat(pTHX_ char *start, I32 type)
5875 s = scan_str(start,FALSE,FALSE);
5878 SvREFCNT_dec(PL_lex_stuff);
5879 PL_lex_stuff = Nullsv;
5880 Perl_croak(aTHX_ "Search pattern not terminated");
5883 pm = (PMOP*)newPMOP(type, 0);
5884 if (PL_multi_open == '?')
5885 pm->op_pmflags |= PMf_ONCE;
5887 while (*s && strchr("iomsx", *s))
5888 pmflag(&pm->op_pmflags,*s++);
5891 while (*s && strchr("iogcmsx", *s))
5892 pmflag(&pm->op_pmflags,*s++);
5894 pm->op_pmpermflags = pm->op_pmflags;
5896 PL_lex_op = (OP*)pm;
5897 yylval.ival = OP_MATCH;
5902 S_scan_subst(pTHX_ char *start)
5909 yylval.ival = OP_NULL;
5911 s = scan_str(start,FALSE,FALSE);
5915 SvREFCNT_dec(PL_lex_stuff);
5916 PL_lex_stuff = Nullsv;
5917 Perl_croak(aTHX_ "Substitution pattern not terminated");
5920 if (s[-1] == PL_multi_open)
5923 first_start = PL_multi_start;
5924 s = scan_str(s,FALSE,FALSE);
5927 SvREFCNT_dec(PL_lex_stuff);
5928 PL_lex_stuff = Nullsv;
5930 SvREFCNT_dec(PL_lex_repl);
5931 PL_lex_repl = Nullsv;
5932 Perl_croak(aTHX_ "Substitution replacement not terminated");
5934 PL_multi_start = first_start; /* so whole substitution is taken together */
5936 pm = (PMOP*)newPMOP(OP_SUBST, 0);
5942 else if (strchr("iogcmsx", *s))
5943 pmflag(&pm->op_pmflags,*s++);
5950 PL_sublex_info.super_bufptr = s;
5951 PL_sublex_info.super_bufend = PL_bufend;
5953 pm->op_pmflags |= PMf_EVAL;
5954 repl = newSVpvn("",0);
5956 sv_catpv(repl, es ? "eval " : "do ");
5957 sv_catpvn(repl, "{ ", 2);
5958 sv_catsv(repl, PL_lex_repl);
5959 sv_catpvn(repl, " };", 2);
5961 SvREFCNT_dec(PL_lex_repl);
5965 pm->op_pmpermflags = pm->op_pmflags;
5966 PL_lex_op = (OP*)pm;
5967 yylval.ival = OP_SUBST;
5972 S_scan_trans(pTHX_ char *start)
5983 yylval.ival = OP_NULL;
5985 s = scan_str(start,FALSE,FALSE);
5988 SvREFCNT_dec(PL_lex_stuff);
5989 PL_lex_stuff = Nullsv;
5990 Perl_croak(aTHX_ "Transliteration pattern not terminated");
5992 if (s[-1] == PL_multi_open)
5995 s = scan_str(s,FALSE,FALSE);
5998 SvREFCNT_dec(PL_lex_stuff);
5999 PL_lex_stuff = Nullsv;
6001 SvREFCNT_dec(PL_lex_repl);
6002 PL_lex_repl = Nullsv;
6003 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6007 o = newSVOP(OP_TRANS, 0, 0);
6008 utf8 = OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF;
6011 New(803,tbl,256,short);
6012 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6016 complement = del = squash = 0;
6017 while (strchr("cdsCU", *s)) {
6019 complement = OPpTRANS_COMPLEMENT;
6021 del = OPpTRANS_DELETE;
6023 squash = OPpTRANS_SQUASH;
6028 utf8 &= ~OPpTRANS_FROM_UTF;
6030 utf8 |= OPpTRANS_FROM_UTF;
6034 utf8 &= ~OPpTRANS_TO_UTF;
6036 utf8 |= OPpTRANS_TO_UTF;
6039 Perl_croak(aTHX_ "Too many /C and /U options");
6044 o->op_private = del|squash|complement|utf8;
6047 yylval.ival = OP_TRANS;
6052 S_scan_heredoc(pTHX_ register char *s)
6056 I32 op_type = OP_SCALAR;
6063 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6067 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6070 for (peek = s; *peek == ' ' || *peek == '\t'; peek++) ;
6071 if (*peek && strchr("`'\"",*peek)) {
6074 s = delimcpy(d, e, s, PL_bufend, term, &len);
6084 if (!isALNUM_lazy_if(s,UTF))
6085 deprecate("bare << to mean <<\"\"");
6086 for (; isALNUM_lazy_if(s,UTF); s++) {
6091 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6092 Perl_croak(aTHX_ "Delimiter for here document is too long");
6095 len = d - PL_tokenbuf;
6096 #ifndef PERL_STRICT_CR
6097 d = strchr(s, '\r');
6101 while (s < PL_bufend) {
6107 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6116 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6121 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6122 herewas = newSVpvn(s,PL_bufend-s);
6124 s--, herewas = newSVpvn(s,d-s);
6125 s += SvCUR(herewas);
6127 tmpstr = NEWSV(87,79);
6128 sv_upgrade(tmpstr, SVt_PVIV);
6133 else if (term == '`') {
6134 op_type = OP_BACKTICK;
6135 SvIVX(tmpstr) = '\\';
6139 PL_multi_start = CopLINE(PL_curcop);
6140 PL_multi_open = PL_multi_close = '<';
6141 term = *PL_tokenbuf;
6142 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6143 char *bufptr = PL_sublex_info.super_bufptr;
6144 char *bufend = PL_sublex_info.super_bufend;
6145 char *olds = s - SvCUR(herewas);
6146 s = strchr(bufptr, '\n');
6150 while (s < bufend &&
6151 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6153 CopLINE_inc(PL_curcop);
6156 CopLINE_set(PL_curcop, PL_multi_start);
6157 missingterm(PL_tokenbuf);
6159 sv_setpvn(herewas,bufptr,d-bufptr+1);
6160 sv_setpvn(tmpstr,d+1,s-d);
6162 sv_catpvn(herewas,s,bufend-s);
6163 (void)strcpy(bufptr,SvPVX(herewas));
6170 while (s < PL_bufend &&
6171 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6173 CopLINE_inc(PL_curcop);
6175 if (s >= PL_bufend) {
6176 CopLINE_set(PL_curcop, PL_multi_start);
6177 missingterm(PL_tokenbuf);
6179 sv_setpvn(tmpstr,d+1,s-d);
6181 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6183 sv_catpvn(herewas,s,PL_bufend-s);
6184 sv_setsv(PL_linestr,herewas);
6185 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6186 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6189 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6190 while (s >= PL_bufend) { /* multiple line string? */
6192 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6193 CopLINE_set(PL_curcop, PL_multi_start);
6194 missingterm(PL_tokenbuf);
6196 CopLINE_inc(PL_curcop);
6197 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6198 #ifndef PERL_STRICT_CR
6199 if (PL_bufend - PL_linestart >= 2) {
6200 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6201 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6203 PL_bufend[-2] = '\n';
6205 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6207 else if (PL_bufend[-1] == '\r')
6208 PL_bufend[-1] = '\n';
6210 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6211 PL_bufend[-1] = '\n';
6213 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6214 SV *sv = NEWSV(88,0);
6216 sv_upgrade(sv, SVt_PVMG);
6217 sv_setsv(sv,PL_linestr);
6218 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6220 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6223 sv_catsv(PL_linestr,herewas);
6224 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6228 sv_catsv(tmpstr,PL_linestr);
6233 PL_multi_end = CopLINE(PL_curcop);
6234 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6235 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6236 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6238 SvREFCNT_dec(herewas);
6239 PL_lex_stuff = tmpstr;
6240 yylval.ival = op_type;
6245 takes: current position in input buffer
6246 returns: new position in input buffer
6247 side-effects: yylval and lex_op are set.
6252 <FH> read from filehandle
6253 <pkg::FH> read from package qualified filehandle
6254 <pkg'FH> read from package qualified filehandle
6255 <$fh> read from filehandle in $fh
6261 S_scan_inputsymbol(pTHX_ char *start)
6263 register char *s = start; /* current position in buffer */
6269 d = PL_tokenbuf; /* start of temp holding space */
6270 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6271 end = strchr(s, '\n');
6274 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6276 /* die if we didn't have space for the contents of the <>,
6277 or if it didn't end, or if we see a newline
6280 if (len >= sizeof PL_tokenbuf)
6281 Perl_croak(aTHX_ "Excessively long <> operator");
6283 Perl_croak(aTHX_ "Unterminated <> operator");
6288 Remember, only scalar variables are interpreted as filehandles by
6289 this code. Anything more complex (e.g., <$fh{$num}>) will be
6290 treated as a glob() call.
6291 This code makes use of the fact that except for the $ at the front,
6292 a scalar variable and a filehandle look the same.
6294 if (*d == '$' && d[1]) d++;
6296 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6297 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6300 /* If we've tried to read what we allow filehandles to look like, and
6301 there's still text left, then it must be a glob() and not a getline.
6302 Use scan_str to pull out the stuff between the <> and treat it
6303 as nothing more than a string.
6306 if (d - PL_tokenbuf != len) {
6307 yylval.ival = OP_GLOB;
6309 s = scan_str(start,FALSE,FALSE);
6311 Perl_croak(aTHX_ "Glob not terminated");
6315 /* we're in a filehandle read situation */
6318 /* turn <> into <ARGV> */
6320 (void)strcpy(d,"ARGV");
6322 /* if <$fh>, create the ops to turn the variable into a
6328 /* try to find it in the pad for this block, otherwise find
6329 add symbol table ops
6331 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6332 OP *o = newOP(OP_PADSV, 0);
6334 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6337 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6338 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6339 newUNOP(OP_RV2SV, 0,
6340 newGVOP(OP_GV, 0, gv)));
6342 PL_lex_op->op_flags |= OPf_SPECIAL;
6343 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6344 yylval.ival = OP_NULL;
6347 /* If it's none of the above, it must be a literal filehandle
6348 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6350 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6351 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6352 yylval.ival = OP_NULL;
6361 takes: start position in buffer
6362 keep_quoted preserve \ on the embedded delimiter(s)
6363 keep_delims preserve the delimiters around the string
6364 returns: position to continue reading from buffer
6365 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6366 updates the read buffer.
6368 This subroutine pulls a string out of the input. It is called for:
6369 q single quotes q(literal text)
6370 ' single quotes 'literal text'
6371 qq double quotes qq(interpolate $here please)
6372 " double quotes "interpolate $here please"
6373 qx backticks qx(/bin/ls -l)
6374 ` backticks `/bin/ls -l`
6375 qw quote words @EXPORT_OK = qw( func() $spam )
6376 m// regexp match m/this/
6377 s/// regexp substitute s/this/that/
6378 tr/// string transliterate tr/this/that/
6379 y/// string transliterate y/this/that/
6380 ($*@) sub prototypes sub foo ($)
6381 (stuff) sub attr parameters sub foo : attr(stuff)
6382 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6384 In most of these cases (all but <>, patterns and transliterate)
6385 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6386 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6387 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6390 It skips whitespace before the string starts, and treats the first
6391 character as the delimiter. If the delimiter is one of ([{< then
6392 the corresponding "close" character )]}> is used as the closing
6393 delimiter. It allows quoting of delimiters, and if the string has
6394 balanced delimiters ([{<>}]) it allows nesting.
6396 The lexer always reads these strings into lex_stuff, except in the
6397 case of the operators which take *two* arguments (s/// and tr///)
6398 when it checks to see if lex_stuff is full (presumably with the 1st
6399 arg to s or tr) and if so puts the string into lex_repl.
6404 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6407 SV *sv; /* scalar value: string */
6408 char *tmps; /* temp string, used for delimiter matching */
6409 register char *s = start; /* current position in the buffer */
6410 register char term; /* terminating character */
6411 register char *to; /* current position in the sv's data */
6412 I32 brackets = 1; /* bracket nesting level */
6413 bool has_utf = FALSE; /* is there any utf8 content? */
6415 /* skip space before the delimiter */
6419 /* mark where we are, in case we need to report errors */
6422 /* after skipping whitespace, the next character is the terminator */
6424 if ((term & 0x80) && UTF)
6427 /* mark where we are */
6428 PL_multi_start = CopLINE(PL_curcop);
6429 PL_multi_open = term;
6431 /* find corresponding closing delimiter */
6432 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6434 PL_multi_close = term;
6436 /* create a new SV to hold the contents. 87 is leak category, I'm
6437 assuming. 79 is the SV's initial length. What a random number. */
6439 sv_upgrade(sv, SVt_PVIV);
6441 (void)SvPOK_only(sv); /* validate pointer */
6443 /* move past delimiter and try to read a complete string */
6445 sv_catpvn(sv, s, 1);
6448 /* extend sv if need be */
6449 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6450 /* set 'to' to the next character in the sv's string */
6451 to = SvPVX(sv)+SvCUR(sv);
6453 /* if open delimiter is the close delimiter read unbridle */
6454 if (PL_multi_open == PL_multi_close) {
6455 for (; s < PL_bufend; s++,to++) {
6456 /* embedded newlines increment the current line number */
6457 if (*s == '\n' && !PL_rsfp)
6458 CopLINE_inc(PL_curcop);
6459 /* handle quoted delimiters */
6460 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6461 if (!keep_quoted && s[1] == term)
6463 /* any other quotes are simply copied straight through */
6467 /* terminate when run out of buffer (the for() condition), or
6468 have found the terminator */
6469 else if (*s == term)
6471 else if (!has_utf && (*s & 0x80) && UTF)
6477 /* if the terminator isn't the same as the start character (e.g.,
6478 matched brackets), we have to allow more in the quoting, and
6479 be prepared for nested brackets.
6482 /* read until we run out of string, or we find the terminator */
6483 for (; s < PL_bufend; s++,to++) {
6484 /* embedded newlines increment the line count */
6485 if (*s == '\n' && !PL_rsfp)
6486 CopLINE_inc(PL_curcop);
6487 /* backslashes can escape the open or closing characters */
6488 if (*s == '\\' && s+1 < PL_bufend) {
6490 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6495 /* allow nested opens and closes */
6496 else if (*s == PL_multi_close && --brackets <= 0)
6498 else if (*s == PL_multi_open)
6500 else if (!has_utf && (*s & 0x80) && UTF)
6505 /* terminate the copied string and update the sv's end-of-string */
6507 SvCUR_set(sv, to - SvPVX(sv));
6510 * this next chunk reads more into the buffer if we're not done yet
6514 break; /* handle case where we are done yet :-) */
6516 #ifndef PERL_STRICT_CR
6517 if (to - SvPVX(sv) >= 2) {
6518 if ((to[-2] == '\r' && to[-1] == '\n') ||
6519 (to[-2] == '\n' && to[-1] == '\r'))
6523 SvCUR_set(sv, to - SvPVX(sv));
6525 else if (to[-1] == '\r')
6528 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6532 /* if we're out of file, or a read fails, bail and reset the current
6533 line marker so we can report where the unterminated string began
6536 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6538 CopLINE_set(PL_curcop, PL_multi_start);
6541 /* we read a line, so increment our line counter */
6542 CopLINE_inc(PL_curcop);
6544 /* update debugger info */
6545 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6546 SV *sv = NEWSV(88,0);
6548 sv_upgrade(sv, SVt_PVMG);
6549 sv_setsv(sv,PL_linestr);
6550 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6553 /* having changed the buffer, we must update PL_bufend */
6554 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6557 /* at this point, we have successfully read the delimited string */
6560 sv_catpvn(sv, s, 1);
6563 PL_multi_end = CopLINE(PL_curcop);
6566 /* if we allocated too much space, give some back */
6567 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6568 SvLEN_set(sv, SvCUR(sv) + 1);
6569 Renew(SvPVX(sv), SvLEN(sv), char);
6572 /* decide whether this is the first or second quoted string we've read
6585 takes: pointer to position in buffer
6586 returns: pointer to new position in buffer
6587 side-effects: builds ops for the constant in yylval.op
6589 Read a number in any of the formats that Perl accepts:
6591 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6592 [\d_]+(\.[\d_]*)?[Ee](\d+)
6594 Underbars (_) are allowed in decimal numbers. If -w is on,
6595 underbars before a decimal point must be at three digit intervals.
6597 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6600 If it reads a number without a decimal point or an exponent, it will
6601 try converting the number to an integer and see if it can do so
6602 without loss of precision.
6606 Perl_scan_num(pTHX_ char *start)
6608 register char *s = start; /* current position in buffer */
6609 register char *d; /* destination in temp buffer */
6610 register char *e; /* end of temp buffer */
6611 IV tryiv; /* used to see if it can be an IV */
6612 NV value; /* number read, as a double */
6613 SV *sv = Nullsv; /* place to put the converted number */
6614 bool floatit; /* boolean: int or float? */
6615 char *lastub = 0; /* position of last underbar */
6616 static char number_too_long[] = "Number too long";
6618 /* We use the first character to decide what type of number this is */
6622 Perl_croak(aTHX_ "panic: scan_num");
6624 /* if it starts with a 0, it could be an octal number, a decimal in
6625 0.13 disguise, or a hexadecimal number, or a binary number. */
6629 u holds the "number so far"
6630 shift the power of 2 of the base
6631 (hex == 4, octal == 3, binary == 1)
6632 overflowed was the number more than we can hold?
6634 Shift is used when we add a digit. It also serves as an "are
6635 we in octal/hex/binary?" indicator to disallow hex characters
6642 bool overflowed = FALSE;
6643 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6644 static char* bases[5] = { "", "binary", "", "octal",
6646 static char* Bases[5] = { "", "Binary", "", "Octal",
6648 static char *maxima[5] = { "",
6649 "0b11111111111111111111111111111111",
6653 char *base, *Base, *max;
6659 } else if (s[1] == 'b') {
6663 /* check for a decimal in disguise */
6664 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6666 /* so it must be octal */
6670 base = bases[shift];
6671 Base = Bases[shift];
6672 max = maxima[shift];
6674 /* read the rest of the number */
6676 /* x is used in the overflow test,
6677 b is the digit we're adding on. */
6682 /* if we don't mention it, we're done */
6691 /* 8 and 9 are not octal */
6694 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6698 case '2': case '3': case '4':
6699 case '5': case '6': case '7':
6701 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6705 b = *s++ & 15; /* ASCII digit -> value of digit */
6709 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6710 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6711 /* make sure they said 0x */
6716 /* Prepare to put the digit we have onto the end
6717 of the number so far. We check for overflows.
6722 x = u << shift; /* make room for the digit */
6724 if ((x >> shift) != u
6725 && !(PL_hints & HINT_NEW_BINARY)) {
6729 if (ckWARN_d(WARN_OVERFLOW))
6730 Perl_warner(aTHX_ WARN_OVERFLOW,
6731 "Integer overflow in %s number",
6734 u = x | b; /* add the digit to the end */
6737 n *= nvshift[shift];
6738 /* If an NV has not enough bits in its
6739 * mantissa to represent an UV this summing of
6740 * small low-order numbers is a waste of time
6741 * (because the NV cannot preserve the
6742 * low-order bits anyway): we could just
6743 * remember when did we overflow and in the
6744 * end just multiply n by the right
6752 /* if we get here, we had success: make a scalar value from
6759 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6760 Perl_warner(aTHX_ WARN_PORTABLE,
6761 "%s number > %s non-portable",
6768 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6769 Perl_warner(aTHX_ WARN_PORTABLE,
6770 "%s number > %s non-portable",
6775 if (PL_hints & HINT_NEW_BINARY)
6776 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6781 handle decimal numbers.
6782 we're also sent here when we read a 0 as the first digit
6784 case '1': case '2': case '3': case '4': case '5':
6785 case '6': case '7': case '8': case '9': case '.':
6788 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
6791 /* read next group of digits and _ and copy into d */
6792 while (isDIGIT(*s) || *s == '_') {
6793 /* skip underscores, checking for misplaced ones
6797 dTHR; /* only for ckWARN */
6798 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
6799 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6803 /* check for end of fixed-length buffer */
6805 Perl_croak(aTHX_ number_too_long);
6806 /* if we're ok, copy the character */
6811 /* final misplaced underbar check */
6812 if (lastub && s - lastub != 3) {
6814 if (ckWARN(WARN_SYNTAX))
6815 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6818 /* read a decimal portion if there is one. avoid
6819 3..5 being interpreted as the number 3. followed
6822 if (*s == '.' && s[1] != '.') {
6826 /* copy, ignoring underbars, until we run out of
6827 digits. Note: no misplaced underbar checks!
6829 for (; isDIGIT(*s) || *s == '_'; s++) {
6830 /* fixed length buffer check */
6832 Perl_croak(aTHX_ number_too_long);
6838 /* read exponent part, if present */
6839 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
6843 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
6844 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
6846 /* allow positive or negative exponent */
6847 if (*s == '+' || *s == '-')
6850 /* read digits of exponent (no underbars :-) */
6851 while (isDIGIT(*s)) {
6853 Perl_croak(aTHX_ number_too_long);
6858 /* terminate the string */
6861 /* make an sv from the string */
6864 value = Atof(PL_tokenbuf);
6867 See if we can make do with an integer value without loss of
6868 precision. We use I_V to cast to an int, because some
6869 compilers have issues. Then we try casting it back and see
6870 if it was the same. We only do this if we know we
6871 specifically read an integer.
6873 Note: if floatit is true, then we don't need to do the
6877 if (!floatit && (NV)tryiv == value)
6878 sv_setiv(sv, tryiv);
6880 sv_setnv(sv, value);
6881 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
6882 (PL_hints & HINT_NEW_INTEGER) )
6883 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
6884 (floatit ? "float" : "integer"),
6887 /* if it starts with a v, it could be a version number */
6892 while (isDIGIT(*pos))
6894 if (*pos == '.' && isDIGIT(pos[1])) {
6900 s++; /* get past 'v' */
6903 SvUPGRADE(sv, SVt_PVNV);
6904 sv_setpvn(sv, "", 0);
6907 if (*s == '0' && isDIGIT(s[1]))
6908 yyerror("Octal number in vector unsupported");
6911 while (isDIGIT(*pos))
6915 tmpend = uv_to_utf8(tmpbuf, rev);
6919 tmpbuf[0] = (U8)rev;
6920 tmpend = &tmpbuf[1];
6923 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
6925 SvNVX(sv) += (NV)rev/nshift;
6927 } while (*pos == '.' && isDIGIT(pos[1]));
6929 if (*s == '0' && isDIGIT(s[1]))
6930 yyerror("Octal number in vector unsupported");
6933 tmpend = uv_to_utf8(tmpbuf, rev);
6934 utf8 = utf8 || rev > 127;
6936 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
6938 SvNVX(sv) += (NV)rev/nshift;
6950 /* make the op for the constant and return */
6953 yylval.opval = newSVOP(OP_CONST, 0, sv);
6955 yylval.opval = Nullop;
6961 S_scan_formline(pTHX_ register char *s)
6966 SV *stuff = newSVpvn("",0);
6967 bool needargs = FALSE;
6970 if (*s == '.' || *s == '}') {
6972 #ifdef PERL_STRICT_CR
6973 for (t = s+1;*t == ' ' || *t == '\t'; t++) ;
6975 for (t = s+1;*t == ' ' || *t == '\t' || *t == '\r'; t++) ;
6977 if (*t == '\n' || t == PL_bufend)
6980 if (PL_in_eval && !PL_rsfp) {
6981 eol = strchr(s,'\n');
6986 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6988 for (t = s; t < eol; t++) {
6989 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
6991 goto enough; /* ~~ must be first line in formline */
6993 if (*t == '@' || *t == '^')
6996 sv_catpvn(stuff, s, eol-s);
6997 #ifndef PERL_STRICT_CR
6998 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
6999 char *end = SvPVX(stuff) + SvCUR(stuff);
7008 s = filter_gets(PL_linestr, PL_rsfp, 0);
7009 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7010 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7013 yyerror("Format not terminated");
7023 PL_lex_state = LEX_NORMAL;
7024 PL_nextval[PL_nexttoke].ival = 0;
7028 PL_lex_state = LEX_FORMLINE;
7029 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7031 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7035 SvREFCNT_dec(stuff);
7036 PL_lex_formbrack = 0;
7047 PL_cshlen = strlen(PL_cshname);
7052 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7055 I32 oldsavestack_ix = PL_savestack_ix;
7056 CV* outsidecv = PL_compcv;
7060 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7062 SAVEI32(PL_subline);
7063 save_item(PL_subname);
7066 SAVESPTR(PL_comppad_name);
7067 SAVESPTR(PL_compcv);
7068 SAVEI32(PL_comppad_name_fill);
7069 SAVEI32(PL_min_intro_pending);
7070 SAVEI32(PL_max_intro_pending);
7071 SAVEI32(PL_pad_reset_pending);
7073 PL_compcv = (CV*)NEWSV(1104,0);
7074 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7075 CvFLAGS(PL_compcv) |= flags;
7077 PL_comppad = newAV();
7078 av_push(PL_comppad, Nullsv);
7079 PL_curpad = AvARRAY(PL_comppad);
7080 PL_comppad_name = newAV();
7081 PL_comppad_name_fill = 0;
7082 PL_min_intro_pending = 0;
7084 PL_subline = CopLINE(PL_curcop);
7086 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7087 PL_curpad[0] = (SV*)newAV();
7088 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7089 #endif /* USE_THREADS */
7091 comppadlist = newAV();
7092 AvREAL_off(comppadlist);
7093 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7094 av_store(comppadlist, 1, (SV*)PL_comppad);
7096 CvPADLIST(PL_compcv) = comppadlist;
7097 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7099 CvOWNER(PL_compcv) = 0;
7100 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7101 MUTEX_INIT(CvMUTEXP(PL_compcv));
7102 #endif /* USE_THREADS */
7104 return oldsavestack_ix;
7108 Perl_yywarn(pTHX_ char *s)
7111 PL_in_eval |= EVAL_WARNONLY;
7113 PL_in_eval &= ~EVAL_WARNONLY;
7118 Perl_yyerror(pTHX_ char *s)
7122 char *context = NULL;
7126 if (!yychar || (yychar == ';' && !PL_rsfp))
7128 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7129 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7130 while (isSPACE(*PL_oldoldbufptr))
7132 context = PL_oldoldbufptr;
7133 contlen = PL_bufptr - PL_oldoldbufptr;
7135 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7136 PL_oldbufptr != PL_bufptr) {
7137 while (isSPACE(*PL_oldbufptr))
7139 context = PL_oldbufptr;
7140 contlen = PL_bufptr - PL_oldbufptr;
7142 else if (yychar > 255)
7143 where = "next token ???";
7144 else if ((yychar & 127) == 127) {
7145 if (PL_lex_state == LEX_NORMAL ||
7146 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7147 where = "at end of line";
7148 else if (PL_lex_inpat)
7149 where = "within pattern";
7151 where = "within string";
7154 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7156 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7157 else if (isPRINT_LC(yychar))
7158 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7160 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7161 where = SvPVX(where_sv);
7163 msg = sv_2mortal(newSVpv(s, 0));
7164 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7165 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7167 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7169 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7170 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7171 Perl_sv_catpvf(aTHX_ msg,
7172 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7173 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7176 if (PL_in_eval & EVAL_WARNONLY)
7177 Perl_warn(aTHX_ "%"SVf, msg);
7180 if (PL_error_count >= 10)
7181 Perl_croak(aTHX_ "%s has too many errors.\n", CopFILE(PL_curcop));
7183 PL_in_my_stash = Nullhv;
7194 * Restore a source filter.
7198 restore_rsfp(pTHXo_ void *f)
7200 PerlIO *fp = (PerlIO*)f;
7202 if (PL_rsfp == PerlIO_stdin())
7203 PerlIO_clearerr(PL_rsfp);
7204 else if (PL_rsfp && (PL_rsfp != fp))
7205 PerlIO_close(PL_rsfp);