3 * Copyright (c) 1991-1999, 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, '}');
1361 yyerror("Missing right brace on \\x{}");
1364 /* note: utf always shorter than hex */
1365 d = (char*)uv_to_utf8((U8*)d,
1366 (UV)scan_hex(s + 1, e - s - 1, &len));
1371 UV uv = (UV)scan_hex(s, 2, &len);
1372 if (utf && PL_lex_inwhat == OP_TRANS &&
1373 utf != (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1375 d = (char*)uv_to_utf8((U8*)d, uv); /* doing a CU or UC */
1379 if (uv >= 127 && UTF) {
1381 if (ckWARN(WARN_UTF8))
1382 Perl_warner(aTHX_ WARN_UTF8,
1383 "\\x%.*s will produce malformed UTF-8 character; use \\x{%.*s} for that",
1384 (int)len,s,(int)len,s);
1392 /* \N{latin small letter a} is a named character */
1396 char* e = strchr(s, '}');
1405 yyerror("Missing right brace on \\N{}");
1409 res = newSVpvn(s + 1, e - s - 1);
1410 res = new_constant( Nullch, 0, "charnames",
1411 res, Nullsv, "\\N{...}" );
1412 str = SvPV(res,len);
1413 if (len > e - s + 4) {
1414 char *odest = SvPVX(sv);
1416 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1417 d = SvPVX(sv) + (d - odest);
1419 Copy(str, d, len, char);
1426 yyerror("Missing braces on \\N{}");
1429 /* \c is a control character */
1443 /* printf-style backslashes, formfeeds, newlines, etc */
1461 *d++ = '\047'; /* CP 1047 */
1464 *d++ = '\057'; /* CP 1047 */
1478 } /* end if (backslash) */
1481 } /* while loop to process each character */
1483 /* terminate the string and set up the sv */
1485 SvCUR_set(sv, d - SvPVX(sv));
1490 /* shrink the sv if we allocated more than we used */
1491 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1492 SvLEN_set(sv, SvCUR(sv) + 1);
1493 Renew(SvPVX(sv), SvLEN(sv), char);
1496 /* return the substring (via yylval) only if we parsed anything */
1497 if (s > PL_bufptr) {
1498 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1499 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1501 ( PL_lex_inwhat == OP_TRANS
1503 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1506 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1513 * Returns TRUE if there's more to the expression (e.g., a subscript),
1516 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1518 * ->[ and ->{ return TRUE
1519 * { and [ outside a pattern are always subscripts, so return TRUE
1520 * if we're outside a pattern and it's not { or [, then return FALSE
1521 * if we're in a pattern and the first char is a {
1522 * {4,5} (any digits around the comma) returns FALSE
1523 * if we're in a pattern and the first char is a [
1525 * [SOMETHING] has a funky algorithm to decide whether it's a
1526 * character class or not. It has to deal with things like
1527 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1528 * anything else returns TRUE
1531 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1534 S_intuit_more(pTHX_ register char *s)
1536 if (PL_lex_brackets)
1538 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1540 if (*s != '{' && *s != '[')
1545 /* In a pattern, so maybe we have {n,m}. */
1562 /* On the other hand, maybe we have a character class */
1565 if (*s == ']' || *s == '^')
1568 /* this is terrifying, and it works */
1569 int weight = 2; /* let's weigh the evidence */
1571 unsigned char un_char = 255, last_un_char;
1572 char *send = strchr(s,']');
1573 char tmpbuf[sizeof PL_tokenbuf * 4];
1575 if (!send) /* has to be an expression */
1578 Zero(seen,256,char);
1581 else if (isDIGIT(*s)) {
1583 if (isDIGIT(s[1]) && s[2] == ']')
1589 for (; s < send; s++) {
1590 last_un_char = un_char;
1591 un_char = (unsigned char)*s;
1596 weight -= seen[un_char] * 10;
1597 if (isALNUM_lazy_if(s+1,UTF)) {
1598 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1599 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1604 else if (*s == '$' && s[1] &&
1605 strchr("[#!%*<>()-=",s[1])) {
1606 if (/*{*/ strchr("])} =",s[2]))
1615 if (strchr("wds]",s[1]))
1617 else if (seen['\''] || seen['"'])
1619 else if (strchr("rnftbxcav",s[1]))
1621 else if (isDIGIT(s[1])) {
1623 while (s[1] && isDIGIT(s[1]))
1633 if (strchr("aA01! ",last_un_char))
1635 if (strchr("zZ79~",s[1]))
1637 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1638 weight -= 5; /* cope with negative subscript */
1641 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1642 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1647 if (keyword(tmpbuf, d - tmpbuf))
1650 if (un_char == last_un_char + 1)
1652 weight -= seen[un_char];
1657 if (weight >= 0) /* probably a character class */
1667 * Does all the checking to disambiguate
1669 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1670 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1672 * First argument is the stuff after the first token, e.g. "bar".
1674 * Not a method if bar is a filehandle.
1675 * Not a method if foo is a subroutine prototyped to take a filehandle.
1676 * Not a method if it's really "Foo $bar"
1677 * Method if it's "foo $bar"
1678 * Not a method if it's really "print foo $bar"
1679 * Method if it's really "foo package::" (interpreted as package->foo)
1680 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1681 * Not a method if bar is a filehandle or package, but is quoted with
1686 S_intuit_method(pTHX_ char *start, GV *gv)
1688 char *s = start + (*start == '$');
1689 char tmpbuf[sizeof PL_tokenbuf];
1697 if ((cv = GvCVu(gv))) {
1698 char *proto = SvPVX(cv);
1708 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1709 /* start is the beginning of the possible filehandle/object,
1710 * and s is the end of it
1711 * tmpbuf is a copy of it
1714 if (*start == '$') {
1715 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1720 return *s == '(' ? FUNCMETH : METHOD;
1722 if (!keyword(tmpbuf, len)) {
1723 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1728 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1729 if (indirgv && GvCVu(indirgv))
1731 /* filehandle or package name makes it a method */
1732 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1734 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1735 return 0; /* no assumptions -- "=>" quotes bearword */
1737 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1738 newSVpvn(tmpbuf,len));
1739 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1743 return *s == '(' ? FUNCMETH : METHOD;
1751 * Return a string of Perl code to load the debugger. If PERL5DB
1752 * is set, it will return the contents of that, otherwise a
1753 * compile-time require of perl5db.pl.
1760 char *pdb = PerlEnv_getenv("PERL5DB");
1764 SETERRNO(0,SS$_NORMAL);
1765 return "BEGIN { require 'perl5db.pl' }";
1771 /* Encoded script support. filter_add() effectively inserts a
1772 * 'pre-processing' function into the current source input stream.
1773 * Note that the filter function only applies to the current source file
1774 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1776 * The datasv parameter (which may be NULL) can be used to pass
1777 * private data to this instance of the filter. The filter function
1778 * can recover the SV using the FILTER_DATA macro and use it to
1779 * store private buffers and state information.
1781 * The supplied datasv parameter is upgraded to a PVIO type
1782 * and the IoDIRP field is used to store the function pointer,
1783 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1784 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1785 * private use must be set using malloc'd pointers.
1789 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1794 if (!PL_rsfp_filters)
1795 PL_rsfp_filters = newAV();
1797 datasv = NEWSV(255,0);
1798 if (!SvUPGRADE(datasv, SVt_PVIO))
1799 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1800 IoDIRP(datasv) = (DIR*)funcp; /* stash funcp into spare field */
1801 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1802 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1803 funcp, SvPV_nolen(datasv)));
1804 av_unshift(PL_rsfp_filters, 1);
1805 av_store(PL_rsfp_filters, 0, datasv) ;
1810 /* Delete most recently added instance of this filter function. */
1812 Perl_filter_del(pTHX_ filter_t funcp)
1815 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1816 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1818 /* if filter is on top of stack (usual case) just pop it off */
1819 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1820 if (IoDIRP(datasv) == (DIR*)funcp) {
1821 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1822 IoDIRP(datasv) = (DIR*)NULL;
1823 sv_free(av_pop(PL_rsfp_filters));
1827 /* we need to search for the correct entry and clear it */
1828 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1832 /* Invoke the n'th filter function for the current rsfp. */
1834 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1837 /* 0 = read one text line */
1842 if (!PL_rsfp_filters)
1844 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1845 /* Provide a default input filter to make life easy. */
1846 /* Note that we append to the line. This is handy. */
1847 DEBUG_P(PerlIO_printf(Perl_debug_log,
1848 "filter_read %d: from rsfp\n", idx));
1852 int old_len = SvCUR(buf_sv) ;
1854 /* ensure buf_sv is large enough */
1855 SvGROW(buf_sv, old_len + maxlen) ;
1856 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1857 if (PerlIO_error(PL_rsfp))
1858 return -1; /* error */
1860 return 0 ; /* end of file */
1862 SvCUR_set(buf_sv, old_len + len) ;
1865 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1866 if (PerlIO_error(PL_rsfp))
1867 return -1; /* error */
1869 return 0 ; /* end of file */
1872 return SvCUR(buf_sv);
1874 /* Skip this filter slot if filter has been deleted */
1875 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1876 DEBUG_P(PerlIO_printf(Perl_debug_log,
1877 "filter_read %d: skipped (filter deleted)\n",
1879 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1881 /* Get function pointer hidden within datasv */
1882 funcp = (filter_t)IoDIRP(datasv);
1883 DEBUG_P(PerlIO_printf(Perl_debug_log,
1884 "filter_read %d: via function %p (%s)\n",
1885 idx, funcp, SvPV_nolen(datasv)));
1886 /* Call function. The function is expected to */
1887 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1888 /* Return: <0:error, =0:eof, >0:not eof */
1889 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1893 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1895 #ifdef PERL_CR_FILTER
1896 if (!PL_rsfp_filters) {
1897 filter_add(S_cr_textfilter,NULL);
1900 if (PL_rsfp_filters) {
1903 SvCUR_set(sv, 0); /* start with empty line */
1904 if (FILTER_READ(0, sv, 0) > 0)
1905 return ( SvPVX(sv) ) ;
1910 return (sv_gets(sv, fp, append));
1915 static char* exp_name[] =
1916 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
1917 "ATTRTERM", "TERMBLOCK"
1924 Works out what to call the token just pulled out of the input
1925 stream. The yacc parser takes care of taking the ops we return and
1926 stitching them into a tree.
1932 if read an identifier
1933 if we're in a my declaration
1934 croak if they tried to say my($foo::bar)
1935 build the ops for a my() declaration
1936 if it's an access to a my() variable
1937 are we in a sort block?
1938 croak if my($a); $a <=> $b
1939 build ops for access to a my() variable
1940 if in a dq string, and they've said @foo and we can't find @foo
1942 build ops for a bareword
1943 if we already built the token before, use it.
1947 #ifdef USE_PURE_BISON
1948 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
1961 #ifdef USE_PURE_BISON
1962 yylval_pointer = lvalp;
1963 yychar_pointer = lcharp;
1966 /* check if there's an identifier for us to look at */
1967 if (PL_pending_ident) {
1968 /* pit holds the identifier we read and pending_ident is reset */
1969 char pit = PL_pending_ident;
1970 PL_pending_ident = 0;
1972 /* if we're in a my(), we can't allow dynamics here.
1973 $foo'bar has already been turned into $foo::bar, so
1974 just check for colons.
1976 if it's a legal name, the OP is a PADANY.
1979 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
1980 if (strchr(PL_tokenbuf,':'))
1981 yyerror(Perl_form(aTHX_ "No package name allowed for "
1982 "variable %s in \"our\"",
1984 tmp = pad_allocmy(PL_tokenbuf);
1987 if (strchr(PL_tokenbuf,':'))
1988 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
1990 yylval.opval = newOP(OP_PADANY, 0);
1991 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
1997 build the ops for accesses to a my() variable.
1999 Deny my($a) or my($b) in a sort block, *if* $a or $b is
2000 then used in a comparison. This catches most, but not
2001 all cases. For instance, it catches
2002 sort { my($a); $a <=> $b }
2004 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2005 (although why you'd do that is anyone's guess).
2008 if (!strchr(PL_tokenbuf,':')) {
2010 /* Check for single character per-thread SVs */
2011 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2012 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2013 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2015 yylval.opval = newOP(OP_THREADSV, 0);
2016 yylval.opval->op_targ = tmp;
2019 #endif /* USE_THREADS */
2020 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2021 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2022 /* might be an "our" variable" */
2023 if (SvFLAGS(namesv) & SVpad_OUR) {
2024 /* build ops for a bareword */
2025 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2026 sv_catpvn(sym, "::", 2);
2027 sv_catpv(sym, PL_tokenbuf+1);
2028 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2029 yylval.opval->op_private = OPpCONST_ENTERED;
2030 gv_fetchpv(SvPVX(sym),
2032 ? (GV_ADDMULTI | GV_ADDINEVAL)
2035 ((PL_tokenbuf[0] == '$') ? SVt_PV
2036 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2041 /* if it's a sort block and they're naming $a or $b */
2042 if (PL_last_lop_op == OP_SORT &&
2043 PL_tokenbuf[0] == '$' &&
2044 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2047 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2048 d < PL_bufend && *d != '\n';
2051 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2052 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2058 yylval.opval = newOP(OP_PADANY, 0);
2059 yylval.opval->op_targ = tmp;
2065 Whine if they've said @foo in a doublequoted string,
2066 and @foo isn't a variable we can find in the symbol
2069 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2070 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2071 if (!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2072 yyerror(Perl_form(aTHX_ "In string, %s now must be written as \\%s",
2073 PL_tokenbuf, PL_tokenbuf));
2076 /* build ops for a bareword */
2077 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2078 yylval.opval->op_private = OPpCONST_ENTERED;
2079 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2080 ((PL_tokenbuf[0] == '$') ? SVt_PV
2081 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2086 /* no identifier pending identification */
2088 switch (PL_lex_state) {
2090 case LEX_NORMAL: /* Some compilers will produce faster */
2091 case LEX_INTERPNORMAL: /* code if we comment these out. */
2095 /* when we've already built the next token, just pull it out of the queue */
2098 yylval = PL_nextval[PL_nexttoke];
2100 PL_lex_state = PL_lex_defer;
2101 PL_expect = PL_lex_expect;
2102 PL_lex_defer = LEX_NORMAL;
2104 return(PL_nexttype[PL_nexttoke]);
2106 /* interpolated case modifiers like \L \U, including \Q and \E.
2107 when we get here, PL_bufptr is at the \
2109 case LEX_INTERPCASEMOD:
2111 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2112 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2114 /* handle \E or end of string */
2115 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2119 if (PL_lex_casemods) {
2120 oldmod = PL_lex_casestack[--PL_lex_casemods];
2121 PL_lex_casestack[PL_lex_casemods] = '\0';
2123 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2125 PL_lex_state = LEX_INTERPCONCAT;
2129 if (PL_bufptr != PL_bufend)
2131 PL_lex_state = LEX_INTERPCONCAT;
2136 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2137 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2138 if (strchr("LU", *s) &&
2139 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2141 PL_lex_casestack[--PL_lex_casemods] = '\0';
2144 if (PL_lex_casemods > 10) {
2145 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2146 if (newlb != PL_lex_casestack) {
2148 PL_lex_casestack = newlb;
2151 PL_lex_casestack[PL_lex_casemods++] = *s;
2152 PL_lex_casestack[PL_lex_casemods] = '\0';
2153 PL_lex_state = LEX_INTERPCONCAT;
2154 PL_nextval[PL_nexttoke].ival = 0;
2157 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2159 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2161 PL_nextval[PL_nexttoke].ival = OP_LC;
2163 PL_nextval[PL_nexttoke].ival = OP_UC;
2165 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2167 Perl_croak(aTHX_ "panic: yylex");
2170 if (PL_lex_starts) {
2179 case LEX_INTERPPUSH:
2180 return sublex_push();
2182 case LEX_INTERPSTART:
2183 if (PL_bufptr == PL_bufend)
2184 return sublex_done();
2186 PL_lex_dojoin = (*PL_bufptr == '@');
2187 PL_lex_state = LEX_INTERPNORMAL;
2188 if (PL_lex_dojoin) {
2189 PL_nextval[PL_nexttoke].ival = 0;
2192 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2193 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2194 force_next(PRIVATEREF);
2196 force_ident("\"", '$');
2197 #endif /* USE_THREADS */
2198 PL_nextval[PL_nexttoke].ival = 0;
2200 PL_nextval[PL_nexttoke].ival = 0;
2202 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2205 if (PL_lex_starts++) {
2211 case LEX_INTERPENDMAYBE:
2212 if (intuit_more(PL_bufptr)) {
2213 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2219 if (PL_lex_dojoin) {
2220 PL_lex_dojoin = FALSE;
2221 PL_lex_state = LEX_INTERPCONCAT;
2224 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2225 && SvEVALED(PL_lex_repl))
2227 if (PL_bufptr != PL_bufend)
2228 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2229 PL_lex_repl = Nullsv;
2232 case LEX_INTERPCONCAT:
2234 if (PL_lex_brackets)
2235 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2237 if (PL_bufptr == PL_bufend)
2238 return sublex_done();
2240 if (SvIVX(PL_linestr) == '\'') {
2241 SV *sv = newSVsv(PL_linestr);
2244 else if ( PL_hints & HINT_NEW_RE )
2245 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2246 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2250 s = scan_const(PL_bufptr);
2252 PL_lex_state = LEX_INTERPCASEMOD;
2254 PL_lex_state = LEX_INTERPSTART;
2257 if (s != PL_bufptr) {
2258 PL_nextval[PL_nexttoke] = yylval;
2261 if (PL_lex_starts++)
2271 PL_lex_state = LEX_NORMAL;
2272 s = scan_formline(PL_bufptr);
2273 if (!PL_lex_formbrack)
2279 PL_oldoldbufptr = PL_oldbufptr;
2282 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2283 exp_name[PL_expect], s);
2289 if (isIDFIRST_lazy_if(s,UTF))
2291 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2294 goto fake_eof; /* emulate EOF on ^D or ^Z */
2299 if (PL_lex_brackets)
2300 yyerror("Missing right curly or square bracket");
2303 if (s++ < PL_bufend)
2304 goto retry; /* ignore stray nulls */
2307 if (!PL_in_eval && !PL_preambled) {
2308 PL_preambled = TRUE;
2309 sv_setpv(PL_linestr,incl_perldb());
2310 if (SvCUR(PL_linestr))
2311 sv_catpv(PL_linestr,";");
2313 while(AvFILLp(PL_preambleav) >= 0) {
2314 SV *tmpsv = av_shift(PL_preambleav);
2315 sv_catsv(PL_linestr, tmpsv);
2316 sv_catpv(PL_linestr, ";");
2319 sv_free((SV*)PL_preambleav);
2320 PL_preambleav = NULL;
2322 if (PL_minus_n || PL_minus_p) {
2323 sv_catpv(PL_linestr, "LINE: while (<>) {");
2325 sv_catpv(PL_linestr,"chomp;");
2327 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2329 GvIMPORTED_AV_on(gv);
2331 if (strchr("/'\"", *PL_splitstr)
2332 && strchr(PL_splitstr + 1, *PL_splitstr))
2333 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2336 s = "'~#\200\1'"; /* surely one char is unused...*/
2337 while (s[1] && strchr(PL_splitstr, *s)) s++;
2339 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2340 "q" + (delim == '\''), delim);
2341 for (s = PL_splitstr; *s; s++) {
2343 sv_catpvn(PL_linestr, "\\", 1);
2344 sv_catpvn(PL_linestr, s, 1);
2346 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2350 sv_catpv(PL_linestr,"@F=split(' ');");
2353 sv_catpv(PL_linestr, "\n");
2354 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2355 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2356 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2357 SV *sv = NEWSV(85,0);
2359 sv_upgrade(sv, SVt_PVMG);
2360 sv_setsv(sv,PL_linestr);
2361 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2366 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2369 if (PL_preprocess && !PL_in_eval)
2370 (void)PerlProc_pclose(PL_rsfp);
2371 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2372 PerlIO_clearerr(PL_rsfp);
2374 (void)PerlIO_close(PL_rsfp);
2376 PL_doextract = FALSE;
2378 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2379 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2380 sv_catpv(PL_linestr,";}");
2381 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2382 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2383 PL_minus_n = PL_minus_p = 0;
2386 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2387 sv_setpv(PL_linestr,"");
2388 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2391 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2392 PL_doextract = FALSE;
2394 /* Incest with pod. */
2395 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2396 sv_setpv(PL_linestr, "");
2397 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2398 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2399 PL_doextract = FALSE;
2403 } while (PL_doextract);
2404 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2405 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2406 SV *sv = NEWSV(85,0);
2408 sv_upgrade(sv, SVt_PVMG);
2409 sv_setsv(sv,PL_linestr);
2410 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2412 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2413 if (CopLINE(PL_curcop) == 1) {
2414 while (s < PL_bufend && isSPACE(*s))
2416 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2420 if (*s == '#' && *(s+1) == '!')
2422 #ifdef ALTERNATE_SHEBANG
2424 static char as[] = ALTERNATE_SHEBANG;
2425 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2426 d = s + (sizeof(as) - 1);
2428 #endif /* ALTERNATE_SHEBANG */
2437 while (*d && !isSPACE(*d))
2441 #ifdef ARG_ZERO_IS_SCRIPT
2442 if (ipathend > ipath) {
2444 * HP-UX (at least) sets argv[0] to the script name,
2445 * which makes $^X incorrect. And Digital UNIX and Linux,
2446 * at least, set argv[0] to the basename of the Perl
2447 * interpreter. So, having found "#!", we'll set it right.
2449 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2450 assert(SvPOK(x) || SvGMAGICAL(x));
2451 if (sv_eq(x, CopFILESV(PL_curcop))) {
2452 sv_setpvn(x, ipath, ipathend - ipath);
2455 TAINT_NOT; /* $^X is always tainted, but that's OK */
2457 #endif /* ARG_ZERO_IS_SCRIPT */
2462 d = instr(s,"perl -");
2464 d = instr(s,"perl");
2466 /* avoid getting into infinite loops when shebang
2467 * line contains "Perl" rather than "perl" */
2469 for (d = ipathend-4; d >= ipath; --d) {
2470 if ((*d == 'p' || *d == 'P')
2471 && !ibcmp(d, "perl", 4))
2481 #ifdef ALTERNATE_SHEBANG
2483 * If the ALTERNATE_SHEBANG on this system starts with a
2484 * character that can be part of a Perl expression, then if
2485 * we see it but not "perl", we're probably looking at the
2486 * start of Perl code, not a request to hand off to some
2487 * other interpreter. Similarly, if "perl" is there, but
2488 * not in the first 'word' of the line, we assume the line
2489 * contains the start of the Perl program.
2491 if (d && *s != '#') {
2493 while (*c && !strchr("; \t\r\n\f\v#", *c))
2496 d = Nullch; /* "perl" not in first word; ignore */
2498 *s = '#'; /* Don't try to parse shebang line */
2500 #endif /* ALTERNATE_SHEBANG */
2505 !instr(s,"indir") &&
2506 instr(PL_origargv[0],"perl"))
2512 while (s < PL_bufend && isSPACE(*s))
2514 if (s < PL_bufend) {
2515 Newz(899,newargv,PL_origargc+3,char*);
2517 while (s < PL_bufend && !isSPACE(*s))
2520 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2523 newargv = PL_origargv;
2525 PerlProc_execv(ipath, newargv);
2526 Perl_croak(aTHX_ "Can't exec %s", ipath);
2529 U32 oldpdb = PL_perldb;
2530 bool oldn = PL_minus_n;
2531 bool oldp = PL_minus_p;
2533 while (*d && !isSPACE(*d)) d++;
2534 while (*d == ' ' || *d == '\t') d++;
2538 if (*d == 'M' || *d == 'm') {
2540 while (*d && !isSPACE(*d)) d++;
2541 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2544 d = moreswitches(d);
2546 if (PERLDB_LINE && !oldpdb ||
2547 ( PL_minus_n || PL_minus_p ) && !(oldn || oldp) )
2548 /* if we have already added "LINE: while (<>) {",
2549 we must not do it again */
2551 sv_setpv(PL_linestr, "");
2552 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2553 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2554 PL_preambled = FALSE;
2556 (void)gv_fetchfile(PL_origfilename);
2563 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2565 PL_lex_state = LEX_FORMLINE;
2570 #ifdef PERL_STRICT_CR
2571 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2573 "(Maybe you didn't strip carriage returns after a network transfer?)\n");
2575 case ' ': case '\t': case '\f': case 013:
2580 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2582 while (s < d && *s != '\n')
2587 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2589 PL_lex_state = LEX_FORMLINE;
2599 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2604 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2607 if (strnEQ(s,"=>",2)) {
2608 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2609 OPERATOR('-'); /* unary minus */
2611 PL_last_uni = PL_oldbufptr;
2612 PL_last_lop_op = OP_FTEREAD; /* good enough */
2614 case 'r': FTST(OP_FTEREAD);
2615 case 'w': FTST(OP_FTEWRITE);
2616 case 'x': FTST(OP_FTEEXEC);
2617 case 'o': FTST(OP_FTEOWNED);
2618 case 'R': FTST(OP_FTRREAD);
2619 case 'W': FTST(OP_FTRWRITE);
2620 case 'X': FTST(OP_FTREXEC);
2621 case 'O': FTST(OP_FTROWNED);
2622 case 'e': FTST(OP_FTIS);
2623 case 'z': FTST(OP_FTZERO);
2624 case 's': FTST(OP_FTSIZE);
2625 case 'f': FTST(OP_FTFILE);
2626 case 'd': FTST(OP_FTDIR);
2627 case 'l': FTST(OP_FTLINK);
2628 case 'p': FTST(OP_FTPIPE);
2629 case 'S': FTST(OP_FTSOCK);
2630 case 'u': FTST(OP_FTSUID);
2631 case 'g': FTST(OP_FTSGID);
2632 case 'k': FTST(OP_FTSVTX);
2633 case 'b': FTST(OP_FTBLK);
2634 case 'c': FTST(OP_FTCHR);
2635 case 't': FTST(OP_FTTTY);
2636 case 'T': FTST(OP_FTTEXT);
2637 case 'B': FTST(OP_FTBINARY);
2638 case 'M': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTMTIME);
2639 case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME);
2640 case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME);
2642 Perl_croak(aTHX_ "Unrecognized file test: -%c", (int)tmp);
2649 if (PL_expect == XOPERATOR)
2654 else if (*s == '>') {
2657 if (isIDFIRST_lazy_if(s,UTF)) {
2658 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2666 if (PL_expect == XOPERATOR)
2669 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2671 OPERATOR('-'); /* unary minus */
2678 if (PL_expect == XOPERATOR)
2683 if (PL_expect == XOPERATOR)
2686 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2692 if (PL_expect != XOPERATOR) {
2693 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2694 PL_expect = XOPERATOR;
2695 force_ident(PL_tokenbuf, '*');
2708 if (PL_expect == XOPERATOR) {
2712 PL_tokenbuf[0] = '%';
2713 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2714 if (!PL_tokenbuf[1]) {
2716 yyerror("Final % should be \\% or %name");
2719 PL_pending_ident = '%';
2738 switch (PL_expect) {
2741 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2743 PL_bufptr = s; /* update in case we back off */
2749 PL_expect = XTERMBLOCK;
2753 while (isIDFIRST_lazy_if(s,UTF)) {
2754 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2755 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2756 if (tmp < 0) tmp = -tmp;
2771 d = scan_str(d,TRUE,TRUE);
2774 SvREFCNT_dec(PL_lex_stuff);
2775 PL_lex_stuff = Nullsv;
2777 /* MUST advance bufptr here to avoid bogus
2778 "at end of line" context messages from yyerror().
2780 PL_bufptr = s + len;
2781 yyerror("Unterminated attribute parameter in attribute list");
2784 return 0; /* EOF indicator */
2788 SV *sv = newSVpvn(s, len);
2789 sv_catsv(sv, PL_lex_stuff);
2790 attrs = append_elem(OP_LIST, attrs,
2791 newSVOP(OP_CONST, 0, sv));
2792 SvREFCNT_dec(PL_lex_stuff);
2793 PL_lex_stuff = Nullsv;
2796 attrs = append_elem(OP_LIST, attrs,
2797 newSVOP(OP_CONST, 0,
2801 if (*s == ':' && s[1] != ':')
2804 break; /* require real whitespace or :'s */
2806 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
2807 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
2808 char q = ((*s == '\'') ? '"' : '\'');
2809 /* If here for an expression, and parsed no attrs, back off. */
2810 if (tmp == '=' && !attrs) {
2814 /* MUST advance bufptr here to avoid bogus "at end of line"
2815 context messages from yyerror().
2819 yyerror("Unterminated attribute list");
2821 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
2829 PL_nextval[PL_nexttoke].opval = attrs;
2837 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
2838 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
2843 if (CopLINE(PL_curcop) < PL_copline)
2844 PL_copline = CopLINE(PL_curcop);
2855 if (PL_lex_brackets <= 0)
2856 yyerror("Unmatched right square bracket");
2859 if (PL_lex_state == LEX_INTERPNORMAL) {
2860 if (PL_lex_brackets == 0) {
2861 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
2862 PL_lex_state = LEX_INTERPEND;
2869 if (PL_lex_brackets > 100) {
2870 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
2871 if (newlb != PL_lex_brackstack) {
2873 PL_lex_brackstack = newlb;
2876 switch (PL_expect) {
2878 if (PL_lex_formbrack) {
2882 if (PL_oldoldbufptr == PL_last_lop)
2883 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2885 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2886 OPERATOR(HASHBRACK);
2888 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2891 PL_tokenbuf[0] = '\0';
2892 if (d < PL_bufend && *d == '-') {
2893 PL_tokenbuf[0] = '-';
2895 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2898 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
2899 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
2901 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2904 char minus = (PL_tokenbuf[0] == '-');
2905 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
2913 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
2918 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2923 if (PL_oldoldbufptr == PL_last_lop)
2924 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2926 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2929 OPERATOR(HASHBRACK);
2930 /* This hack serves to disambiguate a pair of curlies
2931 * as being a block or an anon hash. Normally, expectation
2932 * determines that, but in cases where we're not in a
2933 * position to expect anything in particular (like inside
2934 * eval"") we have to resolve the ambiguity. This code
2935 * covers the case where the first term in the curlies is a
2936 * quoted string. Most other cases need to be explicitly
2937 * disambiguated by prepending a `+' before the opening
2938 * curly in order to force resolution as an anon hash.
2940 * XXX should probably propagate the outer expectation
2941 * into eval"" to rely less on this hack, but that could
2942 * potentially break current behavior of eval"".
2946 if (*s == '\'' || *s == '"' || *s == '`') {
2947 /* common case: get past first string, handling escapes */
2948 for (t++; t < PL_bufend && *t != *s;)
2949 if (*t++ == '\\' && (*t == '\\' || *t == *s))
2953 else if (*s == 'q') {
2956 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
2960 char open, close, term;
2963 while (t < PL_bufend && isSPACE(*t))
2967 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
2971 for (t++; t < PL_bufend; t++) {
2972 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
2974 else if (*t == open)
2978 for (t++; t < PL_bufend; t++) {
2979 if (*t == '\\' && t+1 < PL_bufend)
2981 else if (*t == close && --brackets <= 0)
2983 else if (*t == open)
2989 else if (isALNUM_lazy_if(t,UTF)) {
2991 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
2994 while (t < PL_bufend && isSPACE(*t))
2996 /* if comma follows first term, call it an anon hash */
2997 /* XXX it could be a comma expression with loop modifiers */
2998 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
2999 || (*t == '=' && t[1] == '>')))
3000 OPERATOR(HASHBRACK);
3001 if (PL_expect == XREF)
3004 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3010 yylval.ival = CopLINE(PL_curcop);
3011 if (isSPACE(*s) || *s == '#')
3012 PL_copline = NOLINE; /* invalidate current command line number */
3017 if (PL_lex_brackets <= 0)
3018 yyerror("Unmatched right curly bracket");
3020 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3021 if (PL_lex_brackets < PL_lex_formbrack)
3022 PL_lex_formbrack = 0;
3023 if (PL_lex_state == LEX_INTERPNORMAL) {
3024 if (PL_lex_brackets == 0) {
3025 if (PL_expect & XFAKEBRACK) {
3026 PL_expect &= XENUMMASK;
3027 PL_lex_state = LEX_INTERPEND;
3029 return yylex(); /* ignore fake brackets */
3031 if (*s == '-' && s[1] == '>')
3032 PL_lex_state = LEX_INTERPENDMAYBE;
3033 else if (*s != '[' && *s != '{')
3034 PL_lex_state = LEX_INTERPEND;
3037 if (PL_expect & XFAKEBRACK) {
3038 PL_expect &= XENUMMASK;
3040 return yylex(); /* ignore fake brackets */
3050 if (PL_expect == XOPERATOR) {
3051 if (ckWARN(WARN_SEMICOLON)
3052 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3054 CopLINE_dec(PL_curcop);
3055 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3056 CopLINE_inc(PL_curcop);
3061 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3063 PL_expect = XOPERATOR;
3064 force_ident(PL_tokenbuf, '&');
3068 yylval.ival = (OPpENTERSUB_AMPER<<8);
3087 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3088 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3090 if (PL_expect == XSTATE && isALPHA(tmp) &&
3091 (s == PL_linestart+1 || s[-2] == '\n') )
3093 if (PL_in_eval && !PL_rsfp) {
3098 if (strnEQ(s,"=cut",4)) {
3112 PL_doextract = TRUE;
3115 if (PL_lex_brackets < PL_lex_formbrack) {
3117 #ifdef PERL_STRICT_CR
3118 for (t = s; *t == ' ' || *t == '\t'; t++) ;
3120 for (t = s; *t == ' ' || *t == '\t' || *t == '\r'; t++) ;
3122 if (*t == '\n' || *t == '#') {
3140 if (PL_expect != XOPERATOR) {
3141 if (s[1] != '<' && !strchr(s,'>'))
3144 s = scan_heredoc(s);
3146 s = scan_inputsymbol(s);
3147 TERM(sublex_start());
3152 SHop(OP_LEFT_SHIFT);
3166 SHop(OP_RIGHT_SHIFT);
3175 if (PL_expect == XOPERATOR) {
3176 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3179 return ','; /* grandfather non-comma-format format */
3183 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3184 PL_tokenbuf[0] = '@';
3185 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3186 sizeof PL_tokenbuf - 1, FALSE);
3187 if (PL_expect == XOPERATOR)
3188 no_op("Array length", s);
3189 if (!PL_tokenbuf[1])
3191 PL_expect = XOPERATOR;
3192 PL_pending_ident = '#';
3196 PL_tokenbuf[0] = '$';
3197 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3198 sizeof PL_tokenbuf - 1, FALSE);
3199 if (PL_expect == XOPERATOR)
3201 if (!PL_tokenbuf[1]) {
3203 yyerror("Final $ should be \\$ or $name");
3207 /* This kludge not intended to be bulletproof. */
3208 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3209 yylval.opval = newSVOP(OP_CONST, 0,
3210 newSViv((IV)PL_compiling.cop_arybase));
3211 yylval.opval->op_private = OPpCONST_ARYBASE;
3217 if (PL_lex_state == LEX_NORMAL)
3220 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3223 PL_tokenbuf[0] = '@';
3224 if (ckWARN(WARN_SYNTAX)) {
3226 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3229 PL_bufptr = skipspace(PL_bufptr);
3230 while (t < PL_bufend && *t != ']')
3232 Perl_warner(aTHX_ WARN_SYNTAX,
3233 "Multidimensional syntax %.*s not supported",
3234 (t - PL_bufptr) + 1, PL_bufptr);
3238 else if (*s == '{') {
3239 PL_tokenbuf[0] = '%';
3240 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3241 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3243 char tmpbuf[sizeof PL_tokenbuf];
3245 for (t++; isSPACE(*t); t++) ;
3246 if (isIDFIRST_lazy_if(t,UTF)) {
3247 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3248 for (; isSPACE(*t); t++) ;
3249 if (*t == ';' && get_cv(tmpbuf, FALSE))
3250 Perl_warner(aTHX_ WARN_SYNTAX,
3251 "You need to quote \"%s\"", tmpbuf);
3257 PL_expect = XOPERATOR;
3258 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3259 bool islop = (PL_last_lop == PL_oldoldbufptr);
3260 if (!islop || PL_last_lop_op == OP_GREPSTART)
3261 PL_expect = XOPERATOR;
3262 else if (strchr("$@\"'`q", *s))
3263 PL_expect = XTERM; /* e.g. print $fh "foo" */
3264 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3265 PL_expect = XTERM; /* e.g. print $fh &sub */
3266 else if (isIDFIRST_lazy_if(s,UTF)) {
3267 char tmpbuf[sizeof PL_tokenbuf];
3268 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3269 if (tmp = keyword(tmpbuf, len)) {
3270 /* binary operators exclude handle interpretations */
3282 PL_expect = XTERM; /* e.g. print $fh length() */
3287 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3288 if (gv && GvCVu(gv))
3289 PL_expect = XTERM; /* e.g. print $fh subr() */
3292 else if (isDIGIT(*s))
3293 PL_expect = XTERM; /* e.g. print $fh 3 */
3294 else if (*s == '.' && isDIGIT(s[1]))
3295 PL_expect = XTERM; /* e.g. print $fh .3 */
3296 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3297 PL_expect = XTERM; /* e.g. print $fh -1 */
3298 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3299 PL_expect = XTERM; /* print $fh <<"EOF" */
3301 PL_pending_ident = '$';
3305 if (PL_expect == XOPERATOR)
3307 PL_tokenbuf[0] = '@';
3308 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3309 if (!PL_tokenbuf[1]) {
3311 yyerror("Final @ should be \\@ or @name");
3314 if (PL_lex_state == LEX_NORMAL)
3316 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3318 PL_tokenbuf[0] = '%';
3320 /* Warn about @ where they meant $. */
3321 if (ckWARN(WARN_SYNTAX)) {
3322 if (*s == '[' || *s == '{') {
3324 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3326 if (*t == '}' || *t == ']') {
3328 PL_bufptr = skipspace(PL_bufptr);
3329 Perl_warner(aTHX_ WARN_SYNTAX,
3330 "Scalar value %.*s better written as $%.*s",
3331 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3336 PL_pending_ident = '@';
3339 case '/': /* may either be division or pattern */
3340 case '?': /* may either be conditional or pattern */
3341 if (PL_expect != XOPERATOR) {
3342 /* Disable warning on "study /blah/" */
3343 if (PL_oldoldbufptr == PL_last_uni
3344 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3345 || memNE(PL_last_uni, "study", 5)
3346 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3348 s = scan_pat(s,OP_MATCH);
3349 TERM(sublex_start());
3357 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3358 #ifdef PERL_STRICT_CR
3361 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3363 && (s == PL_linestart || s[-1] == '\n') )
3365 PL_lex_formbrack = 0;
3369 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3375 yylval.ival = OPf_SPECIAL;
3381 if (PL_expect != XOPERATOR)
3386 case '0': case '1': case '2': case '3': case '4':
3387 case '5': case '6': case '7': case '8': case '9':
3389 if (PL_expect == XOPERATOR)
3394 s = scan_str(s,FALSE,FALSE);
3395 if (PL_expect == XOPERATOR) {
3396 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3399 return ','; /* grandfather non-comma-format format */
3405 missingterm((char*)0);
3406 yylval.ival = OP_CONST;
3407 TERM(sublex_start());
3410 s = scan_str(s,FALSE,FALSE);
3411 if (PL_expect == XOPERATOR) {
3412 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3415 return ','; /* grandfather non-comma-format format */
3421 missingterm((char*)0);
3422 yylval.ival = OP_CONST;
3423 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3424 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3425 yylval.ival = OP_STRINGIFY;
3429 TERM(sublex_start());
3432 s = scan_str(s,FALSE,FALSE);
3433 if (PL_expect == XOPERATOR)
3434 no_op("Backticks",s);
3436 missingterm((char*)0);
3437 yylval.ival = OP_BACKTICK;
3439 TERM(sublex_start());
3443 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3444 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3446 if (PL_expect == XOPERATOR)
3447 no_op("Backslash",s);
3451 if (isDIGIT(s[1]) && PL_expect == XTERM) {
3455 while (isDIGIT(*start))
3457 if (*start == '.' && isDIGIT(start[1])) {
3464 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3504 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3506 /* Some keywords can be followed by any delimiter, including ':' */
3507 tmp = (len == 1 && strchr("msyq", PL_tokenbuf[0]) ||
3508 len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3509 (PL_tokenbuf[0] == 'q' &&
3510 strchr("qwxr", PL_tokenbuf[1]))));
3512 /* x::* is just a word, unless x is "CORE" */
3513 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3517 while (d < PL_bufend && isSPACE(*d))
3518 d++; /* no comments skipped here, or s### is misparsed */
3520 /* Is this a label? */
3521 if (!tmp && PL_expect == XSTATE
3522 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3524 yylval.pval = savepv(PL_tokenbuf);
3529 /* Check for keywords */
3530 tmp = keyword(PL_tokenbuf, len);
3532 /* Is this a word before a => operator? */
3533 if (strnEQ(d,"=>",2)) {
3535 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3536 yylval.opval->op_private = OPpCONST_BARE;
3540 if (tmp < 0) { /* second-class keyword? */
3541 GV *ogv = Nullgv; /* override (winner) */
3542 GV *hgv = Nullgv; /* hidden (loser) */
3543 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3545 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3548 if (GvIMPORTED_CV(gv))
3550 else if (! CvMETHOD(cv))
3554 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3555 (gv = *gvp) != (GV*)&PL_sv_undef &&
3556 GvCVu(gv) && GvIMPORTED_CV(gv))
3562 tmp = 0; /* overridden by import or by GLOBAL */
3565 && -tmp==KEY_lock /* XXX generalizable kludge */
3567 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3569 tmp = 0; /* any sub overrides "weak" keyword */
3571 else { /* no override */
3575 if (ckWARN(WARN_AMBIGUOUS) && hgv
3576 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3577 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3578 "Ambiguous call resolved as CORE::%s(), %s",
3579 GvENAME(hgv), "qualify as such or use &");
3586 default: /* not a keyword */
3589 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3591 /* Get the rest if it looks like a package qualifier */
3593 if (*s == '\'' || *s == ':' && s[1] == ':') {
3595 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3598 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3599 *s == '\'' ? "'" : "::");
3603 if (PL_expect == XOPERATOR) {
3604 if (PL_bufptr == PL_linestart) {
3605 CopLINE_dec(PL_curcop);
3606 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3607 CopLINE_inc(PL_curcop);
3610 no_op("Bareword",s);
3613 /* Look for a subroutine with this name in current package,
3614 unless name is "Foo::", in which case Foo is a bearword
3615 (and a package name). */
3618 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3620 if (ckWARN(WARN_UNSAFE) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3621 Perl_warner(aTHX_ WARN_UNSAFE,
3622 "Bareword \"%s\" refers to nonexistent package",
3625 PL_tokenbuf[len] = '\0';
3632 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3635 /* if we saw a global override before, get the right name */
3638 sv = newSVpvn("CORE::GLOBAL::",14);
3639 sv_catpv(sv,PL_tokenbuf);
3642 sv = newSVpv(PL_tokenbuf,0);
3644 /* Presume this is going to be a bareword of some sort. */
3647 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3648 yylval.opval->op_private = OPpCONST_BARE;
3650 /* And if "Foo::", then that's what it certainly is. */
3655 /* See if it's the indirect object for a list operator. */
3657 if (PL_oldoldbufptr &&
3658 PL_oldoldbufptr < PL_bufptr &&
3659 (PL_oldoldbufptr == PL_last_lop
3660 || PL_oldoldbufptr == PL_last_uni) &&
3661 /* NO SKIPSPACE BEFORE HERE! */
3662 (PL_expect == XREF ||
3663 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3665 bool immediate_paren = *s == '(';
3667 /* (Now we can afford to cross potential line boundary.) */
3670 /* Two barewords in a row may indicate method call. */
3672 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3675 /* If not a declared subroutine, it's an indirect object. */
3676 /* (But it's an indir obj regardless for sort.) */
3678 if ((PL_last_lop_op == OP_SORT ||
3679 (!immediate_paren && (!gv || !GvCVu(gv)))) &&
3680 (PL_last_lop_op != OP_MAPSTART &&
3681 PL_last_lop_op != OP_GREPSTART))
3683 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3688 /* If followed by a paren, it's certainly a subroutine. */
3690 PL_expect = XOPERATOR;
3694 if (gv && GvCVu(gv)) {
3695 for (d = s + 1; *d == ' ' || *d == '\t'; d++) ;
3696 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3701 PL_nextval[PL_nexttoke].opval = yylval.opval;
3702 PL_expect = XOPERATOR;
3708 /* If followed by var or block, call it a method (unless sub) */
3710 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3711 PL_last_lop = PL_oldbufptr;
3712 PL_last_lop_op = OP_METHOD;
3716 /* If followed by a bareword, see if it looks like indir obj. */
3718 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3721 /* Not a method, so call it a subroutine (if defined) */
3723 if (gv && GvCVu(gv)) {
3725 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3726 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3727 "Ambiguous use of -%s resolved as -&%s()",
3728 PL_tokenbuf, PL_tokenbuf);
3729 /* Check for a constant sub */
3731 if ((sv = cv_const_sv(cv))) {
3733 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3734 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3735 yylval.opval->op_private = 0;
3739 /* Resolve to GV now. */
3740 op_free(yylval.opval);
3741 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3742 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3743 PL_last_lop = PL_oldbufptr;
3744 PL_last_lop_op = OP_ENTERSUB;
3745 /* Is there a prototype? */
3748 char *proto = SvPV((SV*)cv, len);
3751 if (strEQ(proto, "$"))
3753 if (*proto == '&' && *s == '{') {
3754 sv_setpv(PL_subname,"__ANON__");
3758 PL_nextval[PL_nexttoke].opval = yylval.opval;
3764 /* Call it a bare word */
3766 if (PL_hints & HINT_STRICT_SUBS)
3767 yylval.opval->op_private |= OPpCONST_STRICT;
3770 if (ckWARN(WARN_RESERVED)) {
3771 if (lastchar != '-') {
3772 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
3774 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
3781 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
3782 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3783 "Operator or semicolon missing before %c%s",
3784 lastchar, PL_tokenbuf);
3785 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3786 "Ambiguous use of %c resolved as operator %c",
3787 lastchar, lastchar);
3793 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3794 newSVpv(CopFILE(PL_curcop),0));
3798 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3799 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
3802 case KEY___PACKAGE__:
3803 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3805 ? newSVsv(PL_curstname)
3814 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
3815 char *pname = "main";
3816 if (PL_tokenbuf[2] == 'D')
3817 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
3818 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
3821 GvIOp(gv) = newIO();
3822 IoIFP(GvIOp(gv)) = PL_rsfp;
3823 #if defined(HAS_FCNTL) && defined(F_SETFD)
3825 int fd = PerlIO_fileno(PL_rsfp);
3826 fcntl(fd,F_SETFD,fd >= 3);
3829 /* Mark this internal pseudo-handle as clean */
3830 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
3832 IoTYPE(GvIOp(gv)) = '|';
3833 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
3834 IoTYPE(GvIOp(gv)) = '-';
3836 IoTYPE(GvIOp(gv)) = '<';
3837 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
3838 /* if the script was opened in binmode, we need to revert
3839 * it to text mode for compatibility; but only iff it has CRs
3840 * XXX this is a questionable hack at best. */
3841 if (PL_bufend-PL_bufptr > 2
3842 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
3845 if (IoTYPE(GvIOp(gv)) == '<') {
3846 loc = PerlIO_tell(PL_rsfp);
3847 (void)PerlIO_seek(PL_rsfp, 0L, 0);
3849 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
3850 #if defined(__BORLANDC__)
3851 /* XXX see note in do_binmode() */
3852 ((FILE*)PL_rsfp)->flags |= _F_BIN;
3855 PerlIO_seek(PL_rsfp, loc, 0);
3870 if (PL_expect == XSTATE) {
3877 if (*s == ':' && s[1] == ':') {
3880 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3881 tmp = keyword(PL_tokenbuf, len);
3895 LOP(OP_ACCEPT,XTERM);
3901 LOP(OP_ATAN2,XTERM);
3910 LOP(OP_BLESS,XTERM);
3919 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
3936 if (!PL_cryptseen) {
3937 PL_cryptseen = TRUE;
3941 LOP(OP_CRYPT,XTERM);
3944 if (ckWARN(WARN_OCTAL)) {
3945 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
3946 if (*d != '0' && isDIGIT(*d))
3947 Perl_warner(aTHX_ WARN_OCTAL,
3948 "chmod: mode argument is missing initial 0");
3950 LOP(OP_CHMOD,XTERM);
3953 LOP(OP_CHOWN,XTERM);
3956 LOP(OP_CONNECT,XTERM);
3972 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3976 PL_hints |= HINT_BLOCK_SCOPE;
3986 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
3987 LOP(OP_DBMOPEN,XTERM);
3993 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4000 yylval.ival = CopLINE(PL_curcop);
4014 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4015 UNIBRACK(OP_ENTEREVAL);
4030 case KEY_endhostent:
4036 case KEY_endservent:
4039 case KEY_endprotoent:
4050 yylval.ival = CopLINE(PL_curcop);
4052 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4054 if ((PL_bufend - p) >= 3 &&
4055 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4057 else if ((PL_bufend - p) >= 4 &&
4058 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4061 if (isIDFIRST_lazy_if(p,UTF)) {
4062 p = scan_ident(p, PL_bufend,
4063 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4067 Perl_croak(aTHX_ "Missing $ on loop variable");
4072 LOP(OP_FORMLINE,XTERM);
4078 LOP(OP_FCNTL,XTERM);
4084 LOP(OP_FLOCK,XTERM);
4093 LOP(OP_GREPSTART, XREF);
4096 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4111 case KEY_getpriority:
4112 LOP(OP_GETPRIORITY,XTERM);
4114 case KEY_getprotobyname:
4117 case KEY_getprotobynumber:
4118 LOP(OP_GPBYNUMBER,XTERM);
4120 case KEY_getprotoent:
4132 case KEY_getpeername:
4133 UNI(OP_GETPEERNAME);
4135 case KEY_gethostbyname:
4138 case KEY_gethostbyaddr:
4139 LOP(OP_GHBYADDR,XTERM);
4141 case KEY_gethostent:
4144 case KEY_getnetbyname:
4147 case KEY_getnetbyaddr:
4148 LOP(OP_GNBYADDR,XTERM);
4153 case KEY_getservbyname:
4154 LOP(OP_GSBYNAME,XTERM);
4156 case KEY_getservbyport:
4157 LOP(OP_GSBYPORT,XTERM);
4159 case KEY_getservent:
4162 case KEY_getsockname:
4163 UNI(OP_GETSOCKNAME);
4165 case KEY_getsockopt:
4166 LOP(OP_GSOCKOPT,XTERM);
4188 yylval.ival = CopLINE(PL_curcop);
4192 LOP(OP_INDEX,XTERM);
4198 LOP(OP_IOCTL,XTERM);
4210 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4242 LOP(OP_LISTEN,XTERM);
4251 s = scan_pat(s,OP_MATCH);
4252 TERM(sublex_start());
4255 LOP(OP_MAPSTART, XREF);
4258 LOP(OP_MKDIR,XTERM);
4261 LOP(OP_MSGCTL,XTERM);
4264 LOP(OP_MSGGET,XTERM);
4267 LOP(OP_MSGRCV,XTERM);
4270 LOP(OP_MSGSND,XTERM);
4276 if (isIDFIRST_lazy_if(s,UTF)) {
4277 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4278 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4280 PL_in_my_stash = gv_stashpv(PL_tokenbuf, FALSE);
4281 if (!PL_in_my_stash) {
4284 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4292 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4299 if (PL_expect != XSTATE)
4300 yyerror("\"no\" not allowed in expression");
4301 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4302 s = force_version(s);
4307 if (*s == '(' || (s = skipspace(s), *s == '('))
4314 if (isIDFIRST_lazy_if(s,UTF)) {
4316 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4318 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_AMBIGUOUS))
4319 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4320 "Precedence problem: open %.*s should be open(%.*s)",
4326 yylval.ival = OP_OR;
4336 LOP(OP_OPEN_DIR,XTERM);
4339 checkcomma(s,PL_tokenbuf,"filehandle");
4343 checkcomma(s,PL_tokenbuf,"filehandle");
4362 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4366 LOP(OP_PIPE_OP,XTERM);
4369 s = scan_str(s,FALSE,FALSE);
4371 missingterm((char*)0);
4372 yylval.ival = OP_CONST;
4373 TERM(sublex_start());
4379 s = scan_str(s,FALSE,FALSE);
4381 missingterm((char*)0);
4383 if (SvCUR(PL_lex_stuff)) {
4386 d = SvPV_force(PL_lex_stuff, len);
4388 for (; isSPACE(*d) && len; --len, ++d) ;
4391 if (!warned && ckWARN(WARN_SYNTAX)) {
4392 for (; !isSPACE(*d) && len; --len, ++d) {
4394 Perl_warner(aTHX_ WARN_SYNTAX,
4395 "Possible attempt to separate words with commas");
4398 else if (*d == '#') {
4399 Perl_warner(aTHX_ WARN_SYNTAX,
4400 "Possible attempt to put comments in qw() list");
4406 for (; !isSPACE(*d) && len; --len, ++d) ;
4408 words = append_elem(OP_LIST, words,
4409 newSVOP(OP_CONST, 0, newSVpvn(b, d-b)));
4413 PL_nextval[PL_nexttoke].opval = words;
4418 SvREFCNT_dec(PL_lex_stuff);
4419 PL_lex_stuff = Nullsv;
4424 s = scan_str(s,FALSE,FALSE);
4426 missingterm((char*)0);
4427 yylval.ival = OP_STRINGIFY;
4428 if (SvIVX(PL_lex_stuff) == '\'')
4429 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4430 TERM(sublex_start());
4433 s = scan_pat(s,OP_QR);
4434 TERM(sublex_start());
4437 s = scan_str(s,FALSE,FALSE);
4439 missingterm((char*)0);
4440 yylval.ival = OP_BACKTICK;
4442 TERM(sublex_start());
4449 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4450 s = force_version(s);
4453 *PL_tokenbuf = '\0';
4454 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4455 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4456 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4458 yyerror("<> should be quotes");
4466 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4470 LOP(OP_RENAME,XTERM);
4479 LOP(OP_RINDEX,XTERM);
4502 LOP(OP_REVERSE,XTERM);
4513 TERM(sublex_start());
4515 TOKEN(1); /* force error */
4524 LOP(OP_SELECT,XTERM);
4530 LOP(OP_SEMCTL,XTERM);
4533 LOP(OP_SEMGET,XTERM);
4536 LOP(OP_SEMOP,XTERM);
4542 LOP(OP_SETPGRP,XTERM);
4544 case KEY_setpriority:
4545 LOP(OP_SETPRIORITY,XTERM);
4547 case KEY_sethostent:
4553 case KEY_setservent:
4556 case KEY_setprotoent:
4566 LOP(OP_SEEKDIR,XTERM);
4568 case KEY_setsockopt:
4569 LOP(OP_SSOCKOPT,XTERM);
4575 LOP(OP_SHMCTL,XTERM);
4578 LOP(OP_SHMGET,XTERM);
4581 LOP(OP_SHMREAD,XTERM);
4584 LOP(OP_SHMWRITE,XTERM);
4587 LOP(OP_SHUTDOWN,XTERM);
4596 LOP(OP_SOCKET,XTERM);
4598 case KEY_socketpair:
4599 LOP(OP_SOCKPAIR,XTERM);
4602 checkcomma(s,PL_tokenbuf,"subroutine name");
4604 if (*s == ';' || *s == ')') /* probably a close */
4605 Perl_croak(aTHX_ "sort is now a reserved word");
4607 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4611 LOP(OP_SPLIT,XTERM);
4614 LOP(OP_SPRINTF,XTERM);
4617 LOP(OP_SPLICE,XTERM);
4632 LOP(OP_SUBSTR,XTERM);
4638 char tmpbuf[sizeof PL_tokenbuf];
4640 expectation attrful;
4641 bool have_name, have_proto;
4646 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4647 (*s == ':' && s[1] == ':'))
4650 attrful = XATTRBLOCK;
4651 /* remember buffer pos'n for later force_word */
4652 tboffset = s - PL_oldbufptr;
4653 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4654 if (strchr(tmpbuf, ':'))
4655 sv_setpv(PL_subname, tmpbuf);
4657 sv_setsv(PL_subname,PL_curstname);
4658 sv_catpvn(PL_subname,"::",2);
4659 sv_catpvn(PL_subname,tmpbuf,len);
4666 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4667 PL_expect = XTERMBLOCK;
4668 attrful = XATTRTERM;
4669 sv_setpv(PL_subname,"?");
4673 if (key == KEY_format) {
4675 PL_lex_formbrack = PL_lex_brackets + 1;
4677 (void) force_word(PL_oldbufptr + tboffset, WORD,
4682 /* Look for a prototype */
4686 s = scan_str(s,FALSE,FALSE);
4689 SvREFCNT_dec(PL_lex_stuff);
4690 PL_lex_stuff = Nullsv;
4691 Perl_croak(aTHX_ "Prototype not terminated");
4694 d = SvPVX(PL_lex_stuff);
4696 for (p = d; *p; ++p) {
4701 SvCUR(PL_lex_stuff) = tmp;
4709 if (*s == ':' && s[1] != ':')
4710 PL_expect = attrful;
4713 PL_nextval[PL_nexttoke].opval =
4714 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4715 PL_lex_stuff = Nullsv;
4719 sv_setpv(PL_subname,"__ANON__");
4722 (void) force_word(PL_oldbufptr + tboffset, WORD,
4731 LOP(OP_SYSTEM,XREF);
4734 LOP(OP_SYMLINK,XTERM);
4737 LOP(OP_SYSCALL,XTERM);
4740 LOP(OP_SYSOPEN,XTERM);
4743 LOP(OP_SYSSEEK,XTERM);
4746 LOP(OP_SYSREAD,XTERM);
4749 LOP(OP_SYSWRITE,XTERM);
4753 TERM(sublex_start());
4774 LOP(OP_TRUNCATE,XTERM);
4786 yylval.ival = CopLINE(PL_curcop);
4790 yylval.ival = CopLINE(PL_curcop);
4794 LOP(OP_UNLINK,XTERM);
4800 LOP(OP_UNPACK,XTERM);
4803 LOP(OP_UTIME,XTERM);
4806 if (ckWARN(WARN_OCTAL)) {
4807 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4808 if (*d != '0' && isDIGIT(*d))
4809 Perl_warner(aTHX_ WARN_OCTAL,
4810 "umask: argument is missing initial 0");
4815 LOP(OP_UNSHIFT,XTERM);
4818 if (PL_expect != XSTATE)
4819 yyerror("\"use\" not allowed in expression");
4821 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4822 s = force_version(s);
4823 if (*s == ';' || (s = skipspace(s), *s == ';')) {
4824 PL_nextval[PL_nexttoke].opval = Nullop;
4829 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4830 s = force_version(s);
4842 yylval.ival = CopLINE(PL_curcop);
4846 PL_hints |= HINT_BLOCK_SCOPE;
4853 LOP(OP_WAITPID,XTERM);
4861 static char ctl_l[2];
4863 if (ctl_l[0] == '\0')
4864 ctl_l[0] = toCTRL('L');
4865 gv_fetchpv(ctl_l,TRUE, SVt_PV);
4868 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
4873 if (PL_expect == XOPERATOR)
4879 yylval.ival = OP_XOR;
4884 TERM(sublex_start());
4890 Perl_keyword(pTHX_ register char *d, I32 len)
4895 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
4896 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
4897 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
4898 if (strEQ(d,"__DATA__")) return KEY___DATA__;
4899 if (strEQ(d,"__END__")) return KEY___END__;
4903 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
4908 if (strEQ(d,"and")) return -KEY_and;
4909 if (strEQ(d,"abs")) return -KEY_abs;
4912 if (strEQ(d,"alarm")) return -KEY_alarm;
4913 if (strEQ(d,"atan2")) return -KEY_atan2;
4916 if (strEQ(d,"accept")) return -KEY_accept;
4921 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
4924 if (strEQ(d,"bless")) return -KEY_bless;
4925 if (strEQ(d,"bind")) return -KEY_bind;
4926 if (strEQ(d,"binmode")) return -KEY_binmode;
4929 if (strEQ(d,"CORE")) return -KEY_CORE;
4930 if (strEQ(d,"CHECK")) return KEY_CHECK;
4935 if (strEQ(d,"cmp")) return -KEY_cmp;
4936 if (strEQ(d,"chr")) return -KEY_chr;
4937 if (strEQ(d,"cos")) return -KEY_cos;
4940 if (strEQ(d,"chop")) return KEY_chop;
4943 if (strEQ(d,"close")) return -KEY_close;
4944 if (strEQ(d,"chdir")) return -KEY_chdir;
4945 if (strEQ(d,"chomp")) return KEY_chomp;
4946 if (strEQ(d,"chmod")) return -KEY_chmod;
4947 if (strEQ(d,"chown")) return -KEY_chown;
4948 if (strEQ(d,"crypt")) return -KEY_crypt;
4951 if (strEQ(d,"chroot")) return -KEY_chroot;
4952 if (strEQ(d,"caller")) return -KEY_caller;
4955 if (strEQ(d,"connect")) return -KEY_connect;
4958 if (strEQ(d,"closedir")) return -KEY_closedir;
4959 if (strEQ(d,"continue")) return -KEY_continue;
4964 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
4969 if (strEQ(d,"do")) return KEY_do;
4972 if (strEQ(d,"die")) return -KEY_die;
4975 if (strEQ(d,"dump")) return -KEY_dump;
4978 if (strEQ(d,"delete")) return KEY_delete;
4981 if (strEQ(d,"defined")) return KEY_defined;
4982 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
4985 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
4990 if (strEQ(d,"EQ")) { deprecate(d); return -KEY_eq;}
4991 if (strEQ(d,"END")) return KEY_END;
4996 if (strEQ(d,"eq")) return -KEY_eq;
4999 if (strEQ(d,"eof")) return -KEY_eof;
5000 if (strEQ(d,"exp")) return -KEY_exp;
5003 if (strEQ(d,"else")) return KEY_else;
5004 if (strEQ(d,"exit")) return -KEY_exit;
5005 if (strEQ(d,"eval")) return KEY_eval;
5006 if (strEQ(d,"exec")) return -KEY_exec;
5007 if (strEQ(d,"each")) return KEY_each;
5010 if (strEQ(d,"elsif")) return KEY_elsif;
5013 if (strEQ(d,"exists")) return KEY_exists;
5014 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5017 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5018 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5021 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5024 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5025 if (strEQ(d,"endservent")) return -KEY_endservent;
5028 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5035 if (strEQ(d,"for")) return KEY_for;
5038 if (strEQ(d,"fork")) return -KEY_fork;
5041 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5042 if (strEQ(d,"flock")) return -KEY_flock;
5045 if (strEQ(d,"format")) return KEY_format;
5046 if (strEQ(d,"fileno")) return -KEY_fileno;
5049 if (strEQ(d,"foreach")) return KEY_foreach;
5052 if (strEQ(d,"formline")) return -KEY_formline;
5058 if (strEQ(d,"GT")) { deprecate(d); return -KEY_gt;}
5059 if (strEQ(d,"GE")) { deprecate(d); return -KEY_ge;}
5063 if (strnEQ(d,"get",3)) {
5068 if (strEQ(d,"ppid")) return -KEY_getppid;
5069 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5072 if (strEQ(d,"pwent")) return -KEY_getpwent;
5073 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5074 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5077 if (strEQ(d,"peername")) return -KEY_getpeername;
5078 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5079 if (strEQ(d,"priority")) return -KEY_getpriority;
5082 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5085 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5089 else if (*d == 'h') {
5090 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5091 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5092 if (strEQ(d,"hostent")) return -KEY_gethostent;
5094 else if (*d == 'n') {
5095 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5096 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5097 if (strEQ(d,"netent")) return -KEY_getnetent;
5099 else if (*d == 's') {
5100 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5101 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5102 if (strEQ(d,"servent")) return -KEY_getservent;
5103 if (strEQ(d,"sockname")) return -KEY_getsockname;
5104 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5106 else if (*d == 'g') {
5107 if (strEQ(d,"grent")) return -KEY_getgrent;
5108 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5109 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5111 else if (*d == 'l') {
5112 if (strEQ(d,"login")) return -KEY_getlogin;
5114 else if (strEQ(d,"c")) return -KEY_getc;
5119 if (strEQ(d,"gt")) return -KEY_gt;
5120 if (strEQ(d,"ge")) return -KEY_ge;
5123 if (strEQ(d,"grep")) return KEY_grep;
5124 if (strEQ(d,"goto")) return KEY_goto;
5125 if (strEQ(d,"glob")) return KEY_glob;
5128 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5133 if (strEQ(d,"hex")) return -KEY_hex;
5136 if (strEQ(d,"INIT")) return KEY_INIT;
5141 if (strEQ(d,"if")) return KEY_if;
5144 if (strEQ(d,"int")) return -KEY_int;
5147 if (strEQ(d,"index")) return -KEY_index;
5148 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5153 if (strEQ(d,"join")) return -KEY_join;
5157 if (strEQ(d,"keys")) return KEY_keys;
5158 if (strEQ(d,"kill")) return -KEY_kill;
5163 if (strEQ(d,"LT")) { deprecate(d); return -KEY_lt;}
5164 if (strEQ(d,"LE")) { deprecate(d); return -KEY_le;}
5170 if (strEQ(d,"lt")) return -KEY_lt;
5171 if (strEQ(d,"le")) return -KEY_le;
5172 if (strEQ(d,"lc")) return -KEY_lc;
5175 if (strEQ(d,"log")) return -KEY_log;
5178 if (strEQ(d,"last")) return KEY_last;
5179 if (strEQ(d,"link")) return -KEY_link;
5180 if (strEQ(d,"lock")) return -KEY_lock;
5183 if (strEQ(d,"local")) return KEY_local;
5184 if (strEQ(d,"lstat")) return -KEY_lstat;
5187 if (strEQ(d,"length")) return -KEY_length;
5188 if (strEQ(d,"listen")) return -KEY_listen;
5191 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5194 if (strEQ(d,"localtime")) return -KEY_localtime;
5200 case 1: return KEY_m;
5202 if (strEQ(d,"my")) return KEY_my;
5205 if (strEQ(d,"map")) return KEY_map;
5208 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5211 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5212 if (strEQ(d,"msgget")) return -KEY_msgget;
5213 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5214 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5219 if (strEQ(d,"NE")) { deprecate(d); return -KEY_ne;}
5222 if (strEQ(d,"next")) return KEY_next;
5223 if (strEQ(d,"ne")) return -KEY_ne;
5224 if (strEQ(d,"not")) return -KEY_not;
5225 if (strEQ(d,"no")) return KEY_no;
5230 if (strEQ(d,"or")) return -KEY_or;
5233 if (strEQ(d,"ord")) return -KEY_ord;
5234 if (strEQ(d,"oct")) return -KEY_oct;
5235 if (strEQ(d,"our")) return KEY_our;
5238 if (strEQ(d,"open")) return -KEY_open;
5241 if (strEQ(d,"opendir")) return -KEY_opendir;
5248 if (strEQ(d,"pop")) return KEY_pop;
5249 if (strEQ(d,"pos")) return KEY_pos;
5252 if (strEQ(d,"push")) return KEY_push;
5253 if (strEQ(d,"pack")) return -KEY_pack;
5254 if (strEQ(d,"pipe")) return -KEY_pipe;
5257 if (strEQ(d,"print")) return KEY_print;
5260 if (strEQ(d,"printf")) return KEY_printf;
5263 if (strEQ(d,"package")) return KEY_package;
5266 if (strEQ(d,"prototype")) return KEY_prototype;
5271 if (strEQ(d,"q")) return KEY_q;
5272 if (strEQ(d,"qr")) return KEY_qr;
5273 if (strEQ(d,"qq")) return KEY_qq;
5274 if (strEQ(d,"qw")) return KEY_qw;
5275 if (strEQ(d,"qx")) return KEY_qx;
5277 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5282 if (strEQ(d,"ref")) return -KEY_ref;
5285 if (strEQ(d,"read")) return -KEY_read;
5286 if (strEQ(d,"rand")) return -KEY_rand;
5287 if (strEQ(d,"recv")) return -KEY_recv;
5288 if (strEQ(d,"redo")) return KEY_redo;
5291 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5292 if (strEQ(d,"reset")) return -KEY_reset;
5295 if (strEQ(d,"return")) return KEY_return;
5296 if (strEQ(d,"rename")) return -KEY_rename;
5297 if (strEQ(d,"rindex")) return -KEY_rindex;
5300 if (strEQ(d,"require")) return -KEY_require;
5301 if (strEQ(d,"reverse")) return -KEY_reverse;
5302 if (strEQ(d,"readdir")) return -KEY_readdir;
5305 if (strEQ(d,"readlink")) return -KEY_readlink;
5306 if (strEQ(d,"readline")) return -KEY_readline;
5307 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5310 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5316 case 0: return KEY_s;
5318 if (strEQ(d,"scalar")) return KEY_scalar;
5323 if (strEQ(d,"seek")) return -KEY_seek;
5324 if (strEQ(d,"send")) return -KEY_send;
5327 if (strEQ(d,"semop")) return -KEY_semop;
5330 if (strEQ(d,"select")) return -KEY_select;
5331 if (strEQ(d,"semctl")) return -KEY_semctl;
5332 if (strEQ(d,"semget")) return -KEY_semget;
5335 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5336 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5339 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5340 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5343 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5346 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5347 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5348 if (strEQ(d,"setservent")) return -KEY_setservent;
5351 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5352 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5359 if (strEQ(d,"shift")) return KEY_shift;
5362 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5363 if (strEQ(d,"shmget")) return -KEY_shmget;
5366 if (strEQ(d,"shmread")) return -KEY_shmread;
5369 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5370 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5375 if (strEQ(d,"sin")) return -KEY_sin;
5378 if (strEQ(d,"sleep")) return -KEY_sleep;
5381 if (strEQ(d,"sort")) return KEY_sort;
5382 if (strEQ(d,"socket")) return -KEY_socket;
5383 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5386 if (strEQ(d,"split")) return KEY_split;
5387 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5388 if (strEQ(d,"splice")) return KEY_splice;
5391 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5394 if (strEQ(d,"srand")) return -KEY_srand;
5397 if (strEQ(d,"stat")) return -KEY_stat;
5398 if (strEQ(d,"study")) return KEY_study;
5401 if (strEQ(d,"substr")) return -KEY_substr;
5402 if (strEQ(d,"sub")) return KEY_sub;
5407 if (strEQ(d,"system")) return -KEY_system;
5410 if (strEQ(d,"symlink")) return -KEY_symlink;
5411 if (strEQ(d,"syscall")) return -KEY_syscall;
5412 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5413 if (strEQ(d,"sysread")) return -KEY_sysread;
5414 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5417 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5426 if (strEQ(d,"tr")) return KEY_tr;
5429 if (strEQ(d,"tie")) return KEY_tie;
5432 if (strEQ(d,"tell")) return -KEY_tell;
5433 if (strEQ(d,"tied")) return KEY_tied;
5434 if (strEQ(d,"time")) return -KEY_time;
5437 if (strEQ(d,"times")) return -KEY_times;
5440 if (strEQ(d,"telldir")) return -KEY_telldir;
5443 if (strEQ(d,"truncate")) return -KEY_truncate;
5450 if (strEQ(d,"uc")) return -KEY_uc;
5453 if (strEQ(d,"use")) return KEY_use;
5456 if (strEQ(d,"undef")) return KEY_undef;
5457 if (strEQ(d,"until")) return KEY_until;
5458 if (strEQ(d,"untie")) return KEY_untie;
5459 if (strEQ(d,"utime")) return -KEY_utime;
5460 if (strEQ(d,"umask")) return -KEY_umask;
5463 if (strEQ(d,"unless")) return KEY_unless;
5464 if (strEQ(d,"unpack")) return -KEY_unpack;
5465 if (strEQ(d,"unlink")) return -KEY_unlink;
5468 if (strEQ(d,"unshift")) return KEY_unshift;
5469 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5474 if (strEQ(d,"values")) return -KEY_values;
5475 if (strEQ(d,"vec")) return -KEY_vec;
5480 if (strEQ(d,"warn")) return -KEY_warn;
5481 if (strEQ(d,"wait")) return -KEY_wait;
5484 if (strEQ(d,"while")) return KEY_while;
5485 if (strEQ(d,"write")) return -KEY_write;
5488 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5491 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5496 if (len == 1) return -KEY_x;
5497 if (strEQ(d,"xor")) return -KEY_xor;
5500 if (len == 1) return KEY_y;
5509 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5513 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5514 dTHR; /* only for ckWARN */
5515 if (ckWARN(WARN_SYNTAX)) {
5517 for (w = s+2; *w && level; w++) {
5524 for (; *w && isSPACE(*w); w++) ;
5525 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5526 Perl_warner(aTHX_ WARN_SYNTAX,
5527 "%s (...) interpreted as function",name);
5530 while (s < PL_bufend && isSPACE(*s))
5534 while (s < PL_bufend && isSPACE(*s))
5536 if (isIDFIRST_lazy_if(s,UTF)) {
5538 while (isALNUM_lazy_if(s,UTF))
5540 while (s < PL_bufend && isSPACE(*s))
5545 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5549 Perl_croak(aTHX_ "No comma allowed after %s", what);
5554 /* Either returns sv, or mortalizes sv and returns a new SV*.
5555 Best used as sv=new_constant(..., sv, ...).
5556 If s, pv are NULL, calls subroutine with one argument,
5557 and type is used with error messages only. */
5560 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5564 HV *table = GvHV(PL_hintgv); /* ^H */
5568 const char *why, *why1, *why2;
5570 if (!(PL_hints & HINT_LOCALIZE_HH)) {
5573 why = "%^H is not localized";
5577 msg = Perl_newSVpvf(aTHX_ "constant(%s): %s%s%s",
5578 (type ? type: "undef"), why1, why2, why);
5579 yyerror(SvPVX(msg));
5584 why = "%^H is not defined";
5587 cvp = hv_fetch(table, key, strlen(key), FALSE);
5588 if (!cvp || !SvOK(*cvp)) {
5589 why = "} is not defined";
5594 sv_2mortal(sv); /* Parent created it permanently */
5597 pv = sv_2mortal(newSVpvn(s, len));
5599 typesv = sv_2mortal(newSVpv(type, 0));
5601 typesv = &PL_sv_undef;
5603 PUSHSTACKi(PERLSI_OVERLOAD);
5616 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5620 /* Check the eval first */
5621 if (!PL_in_eval && SvTRUE(ERRSV))
5624 sv_catpv(ERRSV, "Propagated");
5625 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5627 res = SvREFCNT_inc(sv);
5631 (void)SvREFCNT_inc(res);
5640 why = "}} did not return a defined value";
5641 why1 = "Call to &{$^H{";
5651 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5653 register char *d = dest;
5654 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5657 Perl_croak(aTHX_ ident_too_long);
5658 if (isALNUM(*s)) /* UTF handled below */
5660 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
5665 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5669 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5670 char *t = s + UTF8SKIP(s);
5671 while (*t & 0x80 && is_utf8_mark((U8*)t))
5673 if (d + (t - s) > e)
5674 Perl_croak(aTHX_ ident_too_long);
5675 Copy(s, d, t - s, char);
5688 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5698 e = d + destlen - 3; /* two-character token, ending NUL */
5700 while (isDIGIT(*s)) {
5702 Perl_croak(aTHX_ ident_too_long);
5709 Perl_croak(aTHX_ ident_too_long);
5710 if (isALNUM(*s)) /* UTF handled below */
5712 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
5717 else if (*s == ':' && s[1] == ':') {
5721 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5722 char *t = s + UTF8SKIP(s);
5723 while (*t & 0x80 && is_utf8_mark((U8*)t))
5725 if (d + (t - s) > e)
5726 Perl_croak(aTHX_ ident_too_long);
5727 Copy(s, d, t - s, char);
5738 if (PL_lex_state != LEX_NORMAL)
5739 PL_lex_state = LEX_INTERPENDMAYBE;
5742 if (*s == '$' && s[1] &&
5743 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
5756 if (*d == '^' && *s && isCONTROLVAR(*s)) {
5761 if (isSPACE(s[-1])) {
5764 if (ch != ' ' && ch != '\t') {
5770 if (isIDFIRST_lazy_if(d,UTF)) {
5774 while (e < send && isALNUM_lazy_if(e,UTF) || *e == ':') {
5776 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
5779 Copy(s, d, e - s, char);
5784 while ((isALNUM(*s) || *s == ':') && d < e)
5787 Perl_croak(aTHX_ ident_too_long);
5790 while (s < send && (*s == ' ' || *s == '\t')) s++;
5791 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
5792 dTHR; /* only for ckWARN */
5793 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
5794 const char *brack = *s == '[' ? "[...]" : "{...}";
5795 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5796 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
5797 funny, dest, brack, funny, dest, brack);
5800 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
5804 /* Handle extended ${^Foo} variables
5805 * 1999-02-27 mjd-perl-patch@plover.com */
5806 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
5810 while (isALNUM(*s) && d < e) {
5814 Perl_croak(aTHX_ ident_too_long);
5819 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
5820 PL_lex_state = LEX_INTERPEND;
5823 if (PL_lex_state == LEX_NORMAL) {
5824 dTHR; /* only for ckWARN */
5825 if (ckWARN(WARN_AMBIGUOUS) &&
5826 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
5828 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5829 "Ambiguous use of %c{%s} resolved to %c%s",
5830 funny, dest, funny, dest);
5835 s = bracket; /* let the parser handle it */
5839 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
5840 PL_lex_state = LEX_INTERPEND;
5845 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
5850 *pmfl |= PMf_GLOBAL;
5852 *pmfl |= PMf_CONTINUE;
5856 *pmfl |= PMf_MULTILINE;
5858 *pmfl |= PMf_SINGLELINE;
5860 *pmfl |= PMf_EXTENDED;
5864 S_scan_pat(pTHX_ char *start, I32 type)
5869 s = scan_str(start,FALSE,FALSE);
5872 SvREFCNT_dec(PL_lex_stuff);
5873 PL_lex_stuff = Nullsv;
5874 Perl_croak(aTHX_ "Search pattern not terminated");
5877 pm = (PMOP*)newPMOP(type, 0);
5878 if (PL_multi_open == '?')
5879 pm->op_pmflags |= PMf_ONCE;
5881 while (*s && strchr("iomsx", *s))
5882 pmflag(&pm->op_pmflags,*s++);
5885 while (*s && strchr("iogcmsx", *s))
5886 pmflag(&pm->op_pmflags,*s++);
5888 pm->op_pmpermflags = pm->op_pmflags;
5890 PL_lex_op = (OP*)pm;
5891 yylval.ival = OP_MATCH;
5896 S_scan_subst(pTHX_ char *start)
5903 yylval.ival = OP_NULL;
5905 s = scan_str(start,FALSE,FALSE);
5909 SvREFCNT_dec(PL_lex_stuff);
5910 PL_lex_stuff = Nullsv;
5911 Perl_croak(aTHX_ "Substitution pattern not terminated");
5914 if (s[-1] == PL_multi_open)
5917 first_start = PL_multi_start;
5918 s = scan_str(s,FALSE,FALSE);
5921 SvREFCNT_dec(PL_lex_stuff);
5922 PL_lex_stuff = Nullsv;
5924 SvREFCNT_dec(PL_lex_repl);
5925 PL_lex_repl = Nullsv;
5926 Perl_croak(aTHX_ "Substitution replacement not terminated");
5928 PL_multi_start = first_start; /* so whole substitution is taken together */
5930 pm = (PMOP*)newPMOP(OP_SUBST, 0);
5936 else if (strchr("iogcmsx", *s))
5937 pmflag(&pm->op_pmflags,*s++);
5944 PL_sublex_info.super_bufptr = s;
5945 PL_sublex_info.super_bufend = PL_bufend;
5947 pm->op_pmflags |= PMf_EVAL;
5948 repl = newSVpvn("",0);
5950 sv_catpv(repl, es ? "eval " : "do ");
5951 sv_catpvn(repl, "{ ", 2);
5952 sv_catsv(repl, PL_lex_repl);
5953 sv_catpvn(repl, " };", 2);
5955 SvREFCNT_dec(PL_lex_repl);
5959 pm->op_pmpermflags = pm->op_pmflags;
5960 PL_lex_op = (OP*)pm;
5961 yylval.ival = OP_SUBST;
5966 S_scan_trans(pTHX_ char *start)
5977 yylval.ival = OP_NULL;
5979 s = scan_str(start,FALSE,FALSE);
5982 SvREFCNT_dec(PL_lex_stuff);
5983 PL_lex_stuff = Nullsv;
5984 Perl_croak(aTHX_ "Transliteration pattern not terminated");
5986 if (s[-1] == PL_multi_open)
5989 s = scan_str(s,FALSE,FALSE);
5992 SvREFCNT_dec(PL_lex_stuff);
5993 PL_lex_stuff = Nullsv;
5995 SvREFCNT_dec(PL_lex_repl);
5996 PL_lex_repl = Nullsv;
5997 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6001 o = newSVOP(OP_TRANS, 0, 0);
6002 utf8 = OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF;
6005 New(803,tbl,256,short);
6006 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6010 complement = del = squash = 0;
6011 while (strchr("cdsCU", *s)) {
6013 complement = OPpTRANS_COMPLEMENT;
6015 del = OPpTRANS_DELETE;
6017 squash = OPpTRANS_SQUASH;
6022 utf8 &= ~OPpTRANS_FROM_UTF;
6024 utf8 |= OPpTRANS_FROM_UTF;
6028 utf8 &= ~OPpTRANS_TO_UTF;
6030 utf8 |= OPpTRANS_TO_UTF;
6033 Perl_croak(aTHX_ "Too many /C and /U options");
6038 o->op_private = del|squash|complement|utf8;
6041 yylval.ival = OP_TRANS;
6046 S_scan_heredoc(pTHX_ register char *s)
6050 I32 op_type = OP_SCALAR;
6057 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6061 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6064 for (peek = s; *peek == ' ' || *peek == '\t'; peek++) ;
6065 if (*peek && strchr("`'\"",*peek)) {
6068 s = delimcpy(d, e, s, PL_bufend, term, &len);
6078 if (!isALNUM_lazy_if(s,UTF))
6079 deprecate("bare << to mean <<\"\"");
6080 for (; isALNUM_lazy_if(s,UTF); s++) {
6085 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6086 Perl_croak(aTHX_ "Delimiter for here document is too long");
6089 len = d - PL_tokenbuf;
6090 #ifndef PERL_STRICT_CR
6091 d = strchr(s, '\r');
6095 while (s < PL_bufend) {
6101 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6110 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6115 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6116 herewas = newSVpvn(s,PL_bufend-s);
6118 s--, herewas = newSVpvn(s,d-s);
6119 s += SvCUR(herewas);
6121 tmpstr = NEWSV(87,79);
6122 sv_upgrade(tmpstr, SVt_PVIV);
6127 else if (term == '`') {
6128 op_type = OP_BACKTICK;
6129 SvIVX(tmpstr) = '\\';
6133 PL_multi_start = CopLINE(PL_curcop);
6134 PL_multi_open = PL_multi_close = '<';
6135 term = *PL_tokenbuf;
6136 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6137 char *bufptr = PL_sublex_info.super_bufptr;
6138 char *bufend = PL_sublex_info.super_bufend;
6139 char *olds = s - SvCUR(herewas);
6140 s = strchr(bufptr, '\n');
6144 while (s < bufend &&
6145 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6147 CopLINE_inc(PL_curcop);
6150 CopLINE_set(PL_curcop, PL_multi_start);
6151 missingterm(PL_tokenbuf);
6153 sv_setpvn(herewas,bufptr,d-bufptr+1);
6154 sv_setpvn(tmpstr,d+1,s-d);
6156 sv_catpvn(herewas,s,bufend-s);
6157 (void)strcpy(bufptr,SvPVX(herewas));
6164 while (s < PL_bufend &&
6165 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6167 CopLINE_inc(PL_curcop);
6169 if (s >= PL_bufend) {
6170 CopLINE_set(PL_curcop, PL_multi_start);
6171 missingterm(PL_tokenbuf);
6173 sv_setpvn(tmpstr,d+1,s-d);
6175 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6177 sv_catpvn(herewas,s,PL_bufend-s);
6178 sv_setsv(PL_linestr,herewas);
6179 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6180 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6183 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6184 while (s >= PL_bufend) { /* multiple line string? */
6186 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6187 CopLINE_set(PL_curcop, PL_multi_start);
6188 missingterm(PL_tokenbuf);
6190 CopLINE_inc(PL_curcop);
6191 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6192 #ifndef PERL_STRICT_CR
6193 if (PL_bufend - PL_linestart >= 2) {
6194 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6195 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6197 PL_bufend[-2] = '\n';
6199 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6201 else if (PL_bufend[-1] == '\r')
6202 PL_bufend[-1] = '\n';
6204 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6205 PL_bufend[-1] = '\n';
6207 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6208 SV *sv = NEWSV(88,0);
6210 sv_upgrade(sv, SVt_PVMG);
6211 sv_setsv(sv,PL_linestr);
6212 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6214 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6217 sv_catsv(PL_linestr,herewas);
6218 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6222 sv_catsv(tmpstr,PL_linestr);
6227 PL_multi_end = CopLINE(PL_curcop);
6228 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6229 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6230 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6232 SvREFCNT_dec(herewas);
6233 PL_lex_stuff = tmpstr;
6234 yylval.ival = op_type;
6239 takes: current position in input buffer
6240 returns: new position in input buffer
6241 side-effects: yylval and lex_op are set.
6246 <FH> read from filehandle
6247 <pkg::FH> read from package qualified filehandle
6248 <pkg'FH> read from package qualified filehandle
6249 <$fh> read from filehandle in $fh
6255 S_scan_inputsymbol(pTHX_ char *start)
6257 register char *s = start; /* current position in buffer */
6263 d = PL_tokenbuf; /* start of temp holding space */
6264 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6265 end = strchr(s, '\n');
6268 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6270 /* die if we didn't have space for the contents of the <>,
6271 or if it didn't end, or if we see a newline
6274 if (len >= sizeof PL_tokenbuf)
6275 Perl_croak(aTHX_ "Excessively long <> operator");
6277 Perl_croak(aTHX_ "Unterminated <> operator");
6282 Remember, only scalar variables are interpreted as filehandles by
6283 this code. Anything more complex (e.g., <$fh{$num}>) will be
6284 treated as a glob() call.
6285 This code makes use of the fact that except for the $ at the front,
6286 a scalar variable and a filehandle look the same.
6288 if (*d == '$' && d[1]) d++;
6290 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6291 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6294 /* If we've tried to read what we allow filehandles to look like, and
6295 there's still text left, then it must be a glob() and not a getline.
6296 Use scan_str to pull out the stuff between the <> and treat it
6297 as nothing more than a string.
6300 if (d - PL_tokenbuf != len) {
6301 yylval.ival = OP_GLOB;
6303 s = scan_str(start,FALSE,FALSE);
6305 Perl_croak(aTHX_ "Glob not terminated");
6309 /* we're in a filehandle read situation */
6312 /* turn <> into <ARGV> */
6314 (void)strcpy(d,"ARGV");
6316 /* if <$fh>, create the ops to turn the variable into a
6322 /* try to find it in the pad for this block, otherwise find
6323 add symbol table ops
6325 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6326 OP *o = newOP(OP_PADSV, 0);
6328 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6331 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6332 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6333 newUNOP(OP_RV2SV, 0,
6334 newGVOP(OP_GV, 0, gv)));
6336 PL_lex_op->op_flags |= OPf_SPECIAL;
6337 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6338 yylval.ival = OP_NULL;
6341 /* If it's none of the above, it must be a literal filehandle
6342 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6344 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6345 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6346 yylval.ival = OP_NULL;
6355 takes: start position in buffer
6356 keep_quoted preserve \ on the embedded delimiter(s)
6357 keep_delims preserve the delimiters around the string
6358 returns: position to continue reading from buffer
6359 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6360 updates the read buffer.
6362 This subroutine pulls a string out of the input. It is called for:
6363 q single quotes q(literal text)
6364 ' single quotes 'literal text'
6365 qq double quotes qq(interpolate $here please)
6366 " double quotes "interpolate $here please"
6367 qx backticks qx(/bin/ls -l)
6368 ` backticks `/bin/ls -l`
6369 qw quote words @EXPORT_OK = qw( func() $spam )
6370 m// regexp match m/this/
6371 s/// regexp substitute s/this/that/
6372 tr/// string transliterate tr/this/that/
6373 y/// string transliterate y/this/that/
6374 ($*@) sub prototypes sub foo ($)
6375 (stuff) sub attr parameters sub foo : attr(stuff)
6376 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6378 In most of these cases (all but <>, patterns and transliterate)
6379 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6380 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6381 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6384 It skips whitespace before the string starts, and treats the first
6385 character as the delimiter. If the delimiter is one of ([{< then
6386 the corresponding "close" character )]}> is used as the closing
6387 delimiter. It allows quoting of delimiters, and if the string has
6388 balanced delimiters ([{<>}]) it allows nesting.
6390 The lexer always reads these strings into lex_stuff, except in the
6391 case of the operators which take *two* arguments (s/// and tr///)
6392 when it checks to see if lex_stuff is full (presumably with the 1st
6393 arg to s or tr) and if so puts the string into lex_repl.
6398 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6401 SV *sv; /* scalar value: string */
6402 char *tmps; /* temp string, used for delimiter matching */
6403 register char *s = start; /* current position in the buffer */
6404 register char term; /* terminating character */
6405 register char *to; /* current position in the sv's data */
6406 I32 brackets = 1; /* bracket nesting level */
6407 bool has_utf = FALSE; /* is there any utf8 content? */
6409 /* skip space before the delimiter */
6413 /* mark where we are, in case we need to report errors */
6416 /* after skipping whitespace, the next character is the terminator */
6418 if ((term & 0x80) && UTF)
6421 /* mark where we are */
6422 PL_multi_start = CopLINE(PL_curcop);
6423 PL_multi_open = term;
6425 /* find corresponding closing delimiter */
6426 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6428 PL_multi_close = term;
6430 /* create a new SV to hold the contents. 87 is leak category, I'm
6431 assuming. 79 is the SV's initial length. What a random number. */
6433 sv_upgrade(sv, SVt_PVIV);
6435 (void)SvPOK_only(sv); /* validate pointer */
6437 /* move past delimiter and try to read a complete string */
6439 sv_catpvn(sv, s, 1);
6442 /* extend sv if need be */
6443 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6444 /* set 'to' to the next character in the sv's string */
6445 to = SvPVX(sv)+SvCUR(sv);
6447 /* if open delimiter is the close delimiter read unbridle */
6448 if (PL_multi_open == PL_multi_close) {
6449 for (; s < PL_bufend; s++,to++) {
6450 /* embedded newlines increment the current line number */
6451 if (*s == '\n' && !PL_rsfp)
6452 CopLINE_inc(PL_curcop);
6453 /* handle quoted delimiters */
6454 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6455 if (!keep_quoted && s[1] == term)
6457 /* any other quotes are simply copied straight through */
6461 /* terminate when run out of buffer (the for() condition), or
6462 have found the terminator */
6463 else if (*s == term)
6465 else if (!has_utf && (*s & 0x80) && UTF)
6471 /* if the terminator isn't the same as the start character (e.g.,
6472 matched brackets), we have to allow more in the quoting, and
6473 be prepared for nested brackets.
6476 /* read until we run out of string, or we find the terminator */
6477 for (; s < PL_bufend; s++,to++) {
6478 /* embedded newlines increment the line count */
6479 if (*s == '\n' && !PL_rsfp)
6480 CopLINE_inc(PL_curcop);
6481 /* backslashes can escape the open or closing characters */
6482 if (*s == '\\' && s+1 < PL_bufend) {
6484 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6489 /* allow nested opens and closes */
6490 else if (*s == PL_multi_close && --brackets <= 0)
6492 else if (*s == PL_multi_open)
6494 else if (!has_utf && (*s & 0x80) && UTF)
6499 /* terminate the copied string and update the sv's end-of-string */
6501 SvCUR_set(sv, to - SvPVX(sv));
6504 * this next chunk reads more into the buffer if we're not done yet
6508 break; /* handle case where we are done yet :-) */
6510 #ifndef PERL_STRICT_CR
6511 if (to - SvPVX(sv) >= 2) {
6512 if ((to[-2] == '\r' && to[-1] == '\n') ||
6513 (to[-2] == '\n' && to[-1] == '\r'))
6517 SvCUR_set(sv, to - SvPVX(sv));
6519 else if (to[-1] == '\r')
6522 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6526 /* if we're out of file, or a read fails, bail and reset the current
6527 line marker so we can report where the unterminated string began
6530 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6532 CopLINE_set(PL_curcop, PL_multi_start);
6535 /* we read a line, so increment our line counter */
6536 CopLINE_inc(PL_curcop);
6538 /* update debugger info */
6539 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6540 SV *sv = NEWSV(88,0);
6542 sv_upgrade(sv, SVt_PVMG);
6543 sv_setsv(sv,PL_linestr);
6544 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6547 /* having changed the buffer, we must update PL_bufend */
6548 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6551 /* at this point, we have successfully read the delimited string */
6554 sv_catpvn(sv, s, 1);
6557 PL_multi_end = CopLINE(PL_curcop);
6560 /* if we allocated too much space, give some back */
6561 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6562 SvLEN_set(sv, SvCUR(sv) + 1);
6563 Renew(SvPVX(sv), SvLEN(sv), char);
6566 /* decide whether this is the first or second quoted string we've read
6579 takes: pointer to position in buffer
6580 returns: pointer to new position in buffer
6581 side-effects: builds ops for the constant in yylval.op
6583 Read a number in any of the formats that Perl accepts:
6585 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6586 [\d_]+(\.[\d_]*)?[Ee](\d+)
6588 Underbars (_) are allowed in decimal numbers. If -w is on,
6589 underbars before a decimal point must be at three digit intervals.
6591 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6594 If it reads a number without a decimal point or an exponent, it will
6595 try converting the number to an integer and see if it can do so
6596 without loss of precision.
6600 Perl_scan_num(pTHX_ char *start)
6602 register char *s = start; /* current position in buffer */
6603 register char *d; /* destination in temp buffer */
6604 register char *e; /* end of temp buffer */
6605 IV tryiv; /* used to see if it can be an IV */
6606 NV value; /* number read, as a double */
6607 SV *sv = Nullsv; /* place to put the converted number */
6608 bool floatit; /* boolean: int or float? */
6609 char *lastub = 0; /* position of last underbar */
6610 static char number_too_long[] = "Number too long";
6612 /* We use the first character to decide what type of number this is */
6616 Perl_croak(aTHX_ "panic: scan_num");
6618 /* if it starts with a 0, it could be an octal number, a decimal in
6619 0.13 disguise, or a hexadecimal number, or a binary number. */
6623 u holds the "number so far"
6624 shift the power of 2 of the base
6625 (hex == 4, octal == 3, binary == 1)
6626 overflowed was the number more than we can hold?
6628 Shift is used when we add a digit. It also serves as an "are
6629 we in octal/hex/binary?" indicator to disallow hex characters
6636 bool overflowed = FALSE;
6637 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6638 static char* bases[5] = { "", "binary", "", "octal",
6640 static char* Bases[5] = { "", "Binary", "", "Octal",
6642 static char *maxima[5] = { "",
6643 "0b11111111111111111111111111111111",
6647 char *base, *Base, *max;
6653 } else if (s[1] == 'b') {
6657 /* check for a decimal in disguise */
6658 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6660 /* so it must be octal */
6664 base = bases[shift];
6665 Base = Bases[shift];
6666 max = maxima[shift];
6668 /* read the rest of the number */
6670 /* x is used in the overflow test,
6671 b is the digit we're adding on. */
6676 /* if we don't mention it, we're done */
6685 /* 8 and 9 are not octal */
6688 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6692 case '2': case '3': case '4':
6693 case '5': case '6': case '7':
6695 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6699 b = *s++ & 15; /* ASCII digit -> value of digit */
6703 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6704 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6705 /* make sure they said 0x */
6710 /* Prepare to put the digit we have onto the end
6711 of the number so far. We check for overflows.
6716 x = u << shift; /* make room for the digit */
6718 if ((x >> shift) != u
6719 && !(PL_hints & HINT_NEW_BINARY)) {
6723 if (ckWARN_d(WARN_OVERFLOW))
6724 Perl_warner(aTHX_ WARN_OVERFLOW,
6725 "Integer overflow in %s number",
6728 u = x | b; /* add the digit to the end */
6731 n *= nvshift[shift];
6732 /* If an NV has not enough bits in its
6733 * mantissa to represent an UV this summing of
6734 * small low-order numbers is a waste of time
6735 * (because the NV cannot preserve the
6736 * low-order bits anyway): we could just
6737 * remember when did we overflow and in the
6738 * end just multiply n by the right
6746 /* if we get here, we had success: make a scalar value from
6753 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6754 Perl_warner(aTHX_ WARN_PORTABLE,
6755 "%s number > %s non-portable",
6762 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6763 Perl_warner(aTHX_ WARN_PORTABLE,
6764 "%s number > %s non-portable",
6769 if (PL_hints & HINT_NEW_BINARY)
6770 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6775 handle decimal numbers.
6776 we're also sent here when we read a 0 as the first digit
6778 case '1': case '2': case '3': case '4': case '5':
6779 case '6': case '7': case '8': case '9': case '.':
6782 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
6785 /* read next group of digits and _ and copy into d */
6786 while (isDIGIT(*s) || *s == '_') {
6787 /* skip underscores, checking for misplaced ones
6791 dTHR; /* only for ckWARN */
6792 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
6793 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6797 /* check for end of fixed-length buffer */
6799 Perl_croak(aTHX_ number_too_long);
6800 /* if we're ok, copy the character */
6805 /* final misplaced underbar check */
6806 if (lastub && s - lastub != 3) {
6808 if (ckWARN(WARN_SYNTAX))
6809 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6812 /* read a decimal portion if there is one. avoid
6813 3..5 being interpreted as the number 3. followed
6816 if (*s == '.' && s[1] != '.') {
6820 /* copy, ignoring underbars, until we run out of
6821 digits. Note: no misplaced underbar checks!
6823 for (; isDIGIT(*s) || *s == '_'; s++) {
6824 /* fixed length buffer check */
6826 Perl_croak(aTHX_ number_too_long);
6832 /* read exponent part, if present */
6833 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
6837 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
6838 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
6840 /* allow positive or negative exponent */
6841 if (*s == '+' || *s == '-')
6844 /* read digits of exponent (no underbars :-) */
6845 while (isDIGIT(*s)) {
6847 Perl_croak(aTHX_ number_too_long);
6852 /* terminate the string */
6855 /* make an sv from the string */
6858 value = Atof(PL_tokenbuf);
6861 See if we can make do with an integer value without loss of
6862 precision. We use I_V to cast to an int, because some
6863 compilers have issues. Then we try casting it back and see
6864 if it was the same. We only do this if we know we
6865 specifically read an integer.
6867 Note: if floatit is true, then we don't need to do the
6871 if (!floatit && (NV)tryiv == value)
6872 sv_setiv(sv, tryiv);
6874 sv_setnv(sv, value);
6875 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
6876 (PL_hints & HINT_NEW_INTEGER) )
6877 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
6878 (floatit ? "float" : "integer"),
6881 /* if it starts with a v, it could be a version number */
6886 while (isDIGIT(*pos))
6888 if (*pos == '.' && isDIGIT(pos[1])) {
6893 s++; /* get past 'v' */
6896 SvUPGRADE(sv, SVt_PVNV);
6897 sv_setpvn(sv, "", 0);
6900 if (*s == '0' && isDIGIT(s[1]))
6901 yyerror("Octal number in vector unsupported");
6904 while (isDIGIT(*pos))
6907 tmpend = uv_to_utf8(tmpbuf, rev);
6909 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
6911 SvNVX(sv) += (NV)rev/nshift;
6913 } while (*pos == '.' && isDIGIT(pos[1]));
6915 if (*s == '0' && isDIGIT(s[1]))
6916 yyerror("Octal number in vector unsupported");
6919 tmpend = uv_to_utf8(tmpbuf, rev);
6921 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
6923 SvNVX(sv) += (NV)rev/nshift;
6934 /* make the op for the constant and return */
6937 yylval.opval = newSVOP(OP_CONST, 0, sv);
6939 yylval.opval = Nullop;
6945 S_scan_formline(pTHX_ register char *s)
6950 SV *stuff = newSVpvn("",0);
6951 bool needargs = FALSE;
6954 if (*s == '.' || *s == '}') {
6956 #ifdef PERL_STRICT_CR
6957 for (t = s+1;*t == ' ' || *t == '\t'; t++) ;
6959 for (t = s+1;*t == ' ' || *t == '\t' || *t == '\r'; t++) ;
6961 if (*t == '\n' || t == PL_bufend)
6964 if (PL_in_eval && !PL_rsfp) {
6965 eol = strchr(s,'\n');
6970 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6972 for (t = s; t < eol; t++) {
6973 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
6975 goto enough; /* ~~ must be first line in formline */
6977 if (*t == '@' || *t == '^')
6980 sv_catpvn(stuff, s, eol-s);
6981 #ifndef PERL_STRICT_CR
6982 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
6983 char *end = SvPVX(stuff) + SvCUR(stuff);
6992 s = filter_gets(PL_linestr, PL_rsfp, 0);
6993 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
6994 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
6997 yyerror("Format not terminated");
7007 PL_lex_state = LEX_NORMAL;
7008 PL_nextval[PL_nexttoke].ival = 0;
7012 PL_lex_state = LEX_FORMLINE;
7013 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7015 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7019 SvREFCNT_dec(stuff);
7020 PL_lex_formbrack = 0;
7031 PL_cshlen = strlen(PL_cshname);
7036 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7039 I32 oldsavestack_ix = PL_savestack_ix;
7040 CV* outsidecv = PL_compcv;
7044 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7046 SAVEI32(PL_subline);
7047 save_item(PL_subname);
7050 SAVESPTR(PL_comppad_name);
7051 SAVESPTR(PL_compcv);
7052 SAVEI32(PL_comppad_name_fill);
7053 SAVEI32(PL_min_intro_pending);
7054 SAVEI32(PL_max_intro_pending);
7055 SAVEI32(PL_pad_reset_pending);
7057 PL_compcv = (CV*)NEWSV(1104,0);
7058 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7059 CvFLAGS(PL_compcv) |= flags;
7061 PL_comppad = newAV();
7062 av_push(PL_comppad, Nullsv);
7063 PL_curpad = AvARRAY(PL_comppad);
7064 PL_comppad_name = newAV();
7065 PL_comppad_name_fill = 0;
7066 PL_min_intro_pending = 0;
7068 PL_subline = CopLINE(PL_curcop);
7070 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7071 PL_curpad[0] = (SV*)newAV();
7072 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7073 #endif /* USE_THREADS */
7075 comppadlist = newAV();
7076 AvREAL_off(comppadlist);
7077 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7078 av_store(comppadlist, 1, (SV*)PL_comppad);
7080 CvPADLIST(PL_compcv) = comppadlist;
7081 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7083 CvOWNER(PL_compcv) = 0;
7084 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7085 MUTEX_INIT(CvMUTEXP(PL_compcv));
7086 #endif /* USE_THREADS */
7088 return oldsavestack_ix;
7092 Perl_yywarn(pTHX_ char *s)
7095 PL_in_eval |= EVAL_WARNONLY;
7097 PL_in_eval &= ~EVAL_WARNONLY;
7102 Perl_yyerror(pTHX_ char *s)
7106 char *context = NULL;
7110 if (!yychar || (yychar == ';' && !PL_rsfp))
7112 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7113 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7114 while (isSPACE(*PL_oldoldbufptr))
7116 context = PL_oldoldbufptr;
7117 contlen = PL_bufptr - PL_oldoldbufptr;
7119 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7120 PL_oldbufptr != PL_bufptr) {
7121 while (isSPACE(*PL_oldbufptr))
7123 context = PL_oldbufptr;
7124 contlen = PL_bufptr - PL_oldbufptr;
7126 else if (yychar > 255)
7127 where = "next token ???";
7128 else if ((yychar & 127) == 127) {
7129 if (PL_lex_state == LEX_NORMAL ||
7130 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7131 where = "at end of line";
7132 else if (PL_lex_inpat)
7133 where = "within pattern";
7135 where = "within string";
7138 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7140 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7141 else if (isPRINT_LC(yychar))
7142 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7144 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7145 where = SvPVX(where_sv);
7147 msg = sv_2mortal(newSVpv(s, 0));
7148 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7149 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7151 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7153 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7154 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7155 Perl_sv_catpvf(aTHX_ msg,
7156 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7157 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7160 if (PL_in_eval & EVAL_WARNONLY)
7161 Perl_warn(aTHX_ "%"SVf, msg);
7164 if (PL_error_count >= 10)
7165 Perl_croak(aTHX_ "%s has too many errors.\n", CopFILE(PL_curcop));
7167 PL_in_my_stash = Nullhv;
7178 * Restore a source filter.
7182 restore_rsfp(pTHXo_ void *f)
7184 PerlIO *fp = (PerlIO*)f;
7186 if (PL_rsfp == PerlIO_stdin())
7187 PerlIO_clearerr(PL_rsfp);
7188 else if (PL_rsfp && (PL_rsfp != fp))
7189 PerlIO_close(PL_rsfp);