3 * Copyright (c) 1991-2000, Larry Wall
5 * You may distribute under the terms of either the GNU General Public
6 * License or the Artistic License, as specified in the README file.
11 * "It all comes from here, the stench and the peril." --Frodo
15 * This file is the lexer for Perl. It's closely linked to the
18 * The main routine is yylex(), which returns the next token.
22 #define PERL_IN_TOKE_C
25 #define yychar PL_yychar
26 #define yylval PL_yylval
28 static char ident_too_long[] = "Identifier too long";
30 static void restore_rsfp(pTHXo_ void *f);
32 #define XFAKEBRACK 128
35 /*#define UTF (SvUTF8(PL_linestr) && !(PL_hints & HINT_BYTE))*/
36 #define UTF (PL_hints & HINT_UTF8)
38 /* In variables name $^X, these are the legal values for X.
39 * 1999-02-27 mjd-perl-patch@plover.com */
40 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
42 /* On MacOS, respect nonbreaking spaces */
43 #ifdef MACOS_TRADITIONAL
44 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
46 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
49 /* LEX_* are values for PL_lex_state, the state of the lexer.
50 * They are arranged oddly so that the guard on the switch statement
51 * can get by with a single comparison (if the compiler is smart enough).
54 /* #define LEX_NOTPARSING 11 is done in perl.h. */
57 #define LEX_INTERPNORMAL 9
58 #define LEX_INTERPCASEMOD 8
59 #define LEX_INTERPPUSH 7
60 #define LEX_INTERPSTART 6
61 #define LEX_INTERPEND 5
62 #define LEX_INTERPENDMAYBE 4
63 #define LEX_INTERPCONCAT 3
64 #define LEX_INTERPCONST 2
65 #define LEX_FORMLINE 1
66 #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)
329 #ifdef PERL_UTF16_FILTER
331 S_utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
333 I32 count = FILTER_READ(idx+1, sv, maxlen);
337 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
338 tend = utf16_to_utf8((U16*)SvPVX(sv), tmps, SvCUR(sv));
339 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 (SPACE_OR_TAB(*s)) s++;
474 if (strnEQ(s, "line", 4))
478 if (*s == ' ' || *s == '\t')
482 while (SPACE_OR_TAB(*s)) s++;
488 while (SPACE_OR_TAB(*s))
490 if (*s == '"' && (t = strchr(s+1, '"'))) {
495 for (t = s; !isSPACE(*t); t++) ;
498 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
500 if (*e != '\n' && *e != '\0')
501 return; /* false alarm */
507 Safefree(CopFILE(PL_curcop));
509 SvREFCNT_dec(CopFILEGV(PL_curcop));
511 CopFILE_set(PL_curcop, s);
514 CopLINE_set(PL_curcop, atoi(n)-1);
519 * Called to gobble the appropriate amount and type of whitespace.
520 * Skips comments as well.
524 S_skipspace(pTHX_ register char *s)
527 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
528 while (s < PL_bufend && SPACE_OR_TAB(*s))
534 SSize_t oldprevlen, oldoldprevlen;
535 SSize_t oldloplen, oldunilen;
536 while (s < PL_bufend && isSPACE(*s)) {
537 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
542 if (s < PL_bufend && *s == '#') {
543 while (s < PL_bufend && *s != '\n')
547 if (PL_in_eval && !PL_rsfp) {
554 /* only continue to recharge the buffer if we're at the end
555 * of the buffer, we're not reading from a source filter, and
556 * we're in normal lexing mode
558 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
559 PL_lex_state == LEX_FORMLINE)
562 /* try to recharge the buffer */
563 if ((s = filter_gets(PL_linestr, PL_rsfp,
564 (prevlen = SvCUR(PL_linestr)))) == Nullch)
566 /* end of file. Add on the -p or -n magic */
567 if (PL_minus_n || PL_minus_p) {
568 sv_setpv(PL_linestr,PL_minus_p ?
569 ";}continue{print or die qq(-p destination: $!\\n)" :
571 sv_catpv(PL_linestr,";}");
572 PL_minus_n = PL_minus_p = 0;
575 sv_setpv(PL_linestr,";");
577 /* reset variables for next time we lex */
578 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
580 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
582 /* Close the filehandle. Could be from -P preprocessor,
583 * STDIN, or a regular file. If we were reading code from
584 * STDIN (because the commandline held no -e or filename)
585 * then we don't close it, we reset it so the code can
586 * read from STDIN too.
589 if (PL_preprocess && !PL_in_eval)
590 (void)PerlProc_pclose(PL_rsfp);
591 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
592 PerlIO_clearerr(PL_rsfp);
594 (void)PerlIO_close(PL_rsfp);
599 /* not at end of file, so we only read another line */
600 /* make corresponding updates to old pointers, for yyerror() */
601 oldprevlen = PL_oldbufptr - PL_bufend;
602 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
604 oldunilen = PL_last_uni - PL_bufend;
606 oldloplen = PL_last_lop - PL_bufend;
607 PL_linestart = PL_bufptr = s + prevlen;
608 PL_bufend = s + SvCUR(PL_linestr);
610 PL_oldbufptr = s + oldprevlen;
611 PL_oldoldbufptr = s + oldoldprevlen;
613 PL_last_uni = s + oldunilen;
615 PL_last_lop = s + oldloplen;
618 /* debugger active and we're not compiling the debugger code,
619 * so store the line into the debugger's array of lines
621 if (PERLDB_LINE && PL_curstash != PL_debstash) {
622 SV *sv = NEWSV(85,0);
624 sv_upgrade(sv, SVt_PVMG);
625 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
626 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
633 * Check the unary operators to ensure there's no ambiguity in how they're
634 * used. An ambiguous piece of code would be:
636 * This doesn't mean rand() + 5. Because rand() is a unary operator,
637 * the +5 is its argument.
647 if (PL_oldoldbufptr != PL_last_uni)
649 while (isSPACE(*PL_last_uni))
651 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
652 if ((t = strchr(s, '(')) && t < PL_bufptr)
654 if (ckWARN_d(WARN_AMBIGUOUS)){
657 Perl_warner(aTHX_ WARN_AMBIGUOUS,
658 "Warning: Use of \"%s\" without parens is ambiguous",
664 /* workaround to replace the UNI() macro with a function. Only the
665 * hints/uts.sh file mentions this. Other comments elsewhere in the
666 * source indicate Microport Unix might need it too.
672 #define UNI(f) return uni(f,s)
675 S_uni(pTHX_ I32 f, char *s)
680 PL_last_uni = PL_oldbufptr;
691 #endif /* CRIPPLED_CC */
694 * LOP : macro to build a list operator. Its behaviour has been replaced
695 * with a subroutine, S_lop() for which LOP is just another name.
698 #define LOP(f,x) return lop(f,x,s)
702 * Build a list operator (or something that might be one). The rules:
703 * - if we have a next token, then it's a list operator [why?]
704 * - if the next thing is an opening paren, then it's a function
705 * - else it's a list operator
709 S_lop(pTHX_ I32 f, int x, char *s)
716 PL_last_lop = PL_oldbufptr;
731 * When the lexer realizes it knows the next token (for instance,
732 * it is reordering tokens for the parser) then it can call S_force_next
733 * to know what token to return the next time the lexer is called. Caller
734 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
735 * handles the token correctly.
739 S_force_next(pTHX_ I32 type)
741 PL_nexttype[PL_nexttoke] = type;
743 if (PL_lex_state != LEX_KNOWNEXT) {
744 PL_lex_defer = PL_lex_state;
745 PL_lex_expect = PL_expect;
746 PL_lex_state = LEX_KNOWNEXT;
752 * When the lexer knows the next thing is a word (for instance, it has
753 * just seen -> and it knows that the next char is a word char, then
754 * it calls S_force_word to stick the next word into the PL_next lookahead.
757 * char *start : buffer position (must be within PL_linestr)
758 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
759 * int check_keyword : if true, Perl checks to make sure the word isn't
760 * a keyword (do this if the word is a label, e.g. goto FOO)
761 * int allow_pack : if true, : characters will also be allowed (require,
763 * int allow_initial_tick : used by the "sub" lexer only.
767 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
772 start = skipspace(start);
774 if (isIDFIRST_lazy_if(s,UTF) ||
775 (allow_pack && *s == ':') ||
776 (allow_initial_tick && *s == '\'') )
778 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
779 if (check_keyword && keyword(PL_tokenbuf, len))
781 if (token == METHOD) {
786 PL_expect = XOPERATOR;
789 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
790 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
798 * Called when the lexer wants $foo *foo &foo etc, but the program
799 * text only contains the "foo" portion. The first argument is a pointer
800 * to the "foo", and the second argument is the type symbol to prefix.
801 * Forces the next token to be a "WORD".
802 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
806 S_force_ident(pTHX_ register char *s, int kind)
809 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
810 PL_nextval[PL_nexttoke].opval = o;
813 dTHR; /* just for in_eval */
814 o->op_private = OPpCONST_ENTERED;
815 /* XXX see note in pp_entereval() for why we forgo typo
816 warnings if the symbol must be introduced in an eval.
818 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
819 kind == '$' ? SVt_PV :
820 kind == '@' ? SVt_PVAV :
821 kind == '%' ? SVt_PVHV :
829 Perl_str_to_version(pTHX_ SV *sv)
834 char *start = SvPVx(sv,len);
835 bool utf = SvUTF8(sv) ? TRUE : FALSE;
836 char *end = start + len;
837 while (start < end) {
841 n = utf8_to_uv((U8*)start, &skip);
846 retval += ((NV)n)/nshift;
855 * Forces the next token to be a version number.
859 S_force_version(pTHX_ char *s)
861 OP *version = Nullop;
870 for (; isDIGIT(*d) || *d == '_' || *d == '.'; d++);
871 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
874 version = yylval.opval;
875 ver = cSVOPx(version)->op_sv;
876 if (SvPOK(ver) && !SvNIOK(ver)) {
877 (void)SvUPGRADE(ver, SVt_PVNV);
878 SvNVX(ver) = str_to_version(ver);
879 SvNOK_on(ver); /* hint that it is a version */
884 /* NOTE: The parser sees the package name and the VERSION swapped */
885 PL_nextval[PL_nexttoke].opval = version;
893 * Tokenize a quoted string passed in as an SV. It finds the next
894 * chunk, up to end of string or a backslash. It may make a new
895 * SV containing that chunk (if HINT_NEW_STRING is on). It also
900 S_tokeq(pTHX_ SV *sv)
911 s = SvPV_force(sv, len);
912 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
915 while (s < send && *s != '\\')
920 if ( PL_hints & HINT_NEW_STRING )
921 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
924 if (s + 1 < send && (s[1] == '\\'))
925 s++; /* all that, just for this */
930 SvCUR_set(sv, d - SvPVX(sv));
932 if ( PL_hints & HINT_NEW_STRING )
933 return new_constant(NULL, 0, "q", sv, pv, "q");
938 * Now come three functions related to double-quote context,
939 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
940 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
941 * interact with PL_lex_state, and create fake ( ... ) argument lists
942 * to handle functions and concatenation.
943 * They assume that whoever calls them will be setting up a fake
944 * join call, because each subthing puts a ',' after it. This lets
947 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
949 * (I'm not sure whether the spurious commas at the end of lcfirst's
950 * arguments and join's arguments are created or not).
955 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
957 * Pattern matching will set PL_lex_op to the pattern-matching op to
958 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
960 * OP_CONST and OP_READLINE are easy--just make the new op and return.
962 * Everything else becomes a FUNC.
964 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
965 * had an OP_CONST or OP_READLINE). This just sets us up for a
966 * call to S_sublex_push().
972 register I32 op_type = yylval.ival;
974 if (op_type == OP_NULL) {
975 yylval.opval = PL_lex_op;
979 if (op_type == OP_CONST || op_type == OP_READLINE) {
980 SV *sv = tokeq(PL_lex_stuff);
982 if (SvTYPE(sv) == SVt_PVIV) {
983 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
989 nsv = newSVpvn(p, len);
995 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
996 PL_lex_stuff = Nullsv;
1000 PL_sublex_info.super_state = PL_lex_state;
1001 PL_sublex_info.sub_inwhat = op_type;
1002 PL_sublex_info.sub_op = PL_lex_op;
1003 PL_lex_state = LEX_INTERPPUSH;
1007 yylval.opval = PL_lex_op;
1017 * Create a new scope to save the lexing state. The scope will be
1018 * ended in S_sublex_done. Returns a '(', starting the function arguments
1019 * to the uc, lc, etc. found before.
1020 * Sets PL_lex_state to LEX_INTERPCONCAT.
1029 PL_lex_state = PL_sublex_info.super_state;
1030 SAVEI32(PL_lex_dojoin);
1031 SAVEI32(PL_lex_brackets);
1032 SAVEI32(PL_lex_casemods);
1033 SAVEI32(PL_lex_starts);
1034 SAVEI32(PL_lex_state);
1035 SAVEVPTR(PL_lex_inpat);
1036 SAVEI32(PL_lex_inwhat);
1037 SAVECOPLINE(PL_curcop);
1038 SAVEPPTR(PL_bufptr);
1039 SAVEPPTR(PL_oldbufptr);
1040 SAVEPPTR(PL_oldoldbufptr);
1041 SAVEPPTR(PL_linestart);
1042 SAVESPTR(PL_linestr);
1043 SAVEPPTR(PL_lex_brackstack);
1044 SAVEPPTR(PL_lex_casestack);
1046 PL_linestr = PL_lex_stuff;
1047 PL_lex_stuff = Nullsv;
1049 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1050 = SvPVX(PL_linestr);
1051 PL_bufend += SvCUR(PL_linestr);
1052 SAVEFREESV(PL_linestr);
1054 PL_lex_dojoin = FALSE;
1055 PL_lex_brackets = 0;
1056 New(899, PL_lex_brackstack, 120, char);
1057 New(899, PL_lex_casestack, 12, char);
1058 SAVEFREEPV(PL_lex_brackstack);
1059 SAVEFREEPV(PL_lex_casestack);
1060 PL_lex_casemods = 0;
1061 *PL_lex_casestack = '\0';
1063 PL_lex_state = LEX_INTERPCONCAT;
1064 CopLINE_set(PL_curcop, PL_multi_start);
1066 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1067 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1068 PL_lex_inpat = PL_sublex_info.sub_op;
1070 PL_lex_inpat = Nullop;
1077 * Restores lexer state after a S_sublex_push.
1083 if (!PL_lex_starts++) {
1084 PL_expect = XOPERATOR;
1085 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn("",0));
1089 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1090 PL_lex_state = LEX_INTERPCASEMOD;
1094 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1095 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1096 PL_linestr = PL_lex_repl;
1098 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1099 PL_bufend += SvCUR(PL_linestr);
1100 SAVEFREESV(PL_linestr);
1101 PL_lex_dojoin = FALSE;
1102 PL_lex_brackets = 0;
1103 PL_lex_casemods = 0;
1104 *PL_lex_casestack = '\0';
1106 if (SvEVALED(PL_lex_repl)) {
1107 PL_lex_state = LEX_INTERPNORMAL;
1109 /* we don't clear PL_lex_repl here, so that we can check later
1110 whether this is an evalled subst; that means we rely on the
1111 logic to ensure sublex_done() is called again only via the
1112 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1115 PL_lex_state = LEX_INTERPCONCAT;
1116 PL_lex_repl = Nullsv;
1122 PL_bufend = SvPVX(PL_linestr);
1123 PL_bufend += SvCUR(PL_linestr);
1124 PL_expect = XOPERATOR;
1125 PL_sublex_info.sub_inwhat = 0;
1133 Extracts a pattern, double-quoted string, or transliteration. This
1136 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1137 processing a pattern (PL_lex_inpat is true), a transliteration
1138 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1140 Returns a pointer to the character scanned up to. Iff this is
1141 advanced from the start pointer supplied (ie if anything was
1142 successfully parsed), will leave an OP for the substring scanned
1143 in yylval. Caller must intuit reason for not parsing further
1144 by looking at the next characters herself.
1148 double-quoted style: \r and \n
1149 regexp special ones: \D \s
1151 backrefs: \1 (deprecated in substitution replacements)
1152 case and quoting: \U \Q \E
1153 stops on @ and $, but not for $ as tail anchor
1155 In transliterations:
1156 characters are VERY literal, except for - not at the start or end
1157 of the string, which indicates a range. scan_const expands the
1158 range to the full set of intermediate characters.
1160 In double-quoted strings:
1162 double-quoted style: \r and \n
1164 backrefs: \1 (deprecated)
1165 case and quoting: \U \Q \E
1168 scan_const does *not* construct ops to handle interpolated strings.
1169 It stops processing as soon as it finds an embedded $ or @ variable
1170 and leaves it to the caller to work out what's going on.
1172 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1174 $ in pattern could be $foo or could be tail anchor. Assumption:
1175 it's a tail anchor if $ is the last thing in the string, or if it's
1176 followed by one of ")| \n\t"
1178 \1 (backreferences) are turned into $1
1180 The structure of the code is
1181 while (there's a character to process) {
1182 handle transliteration ranges
1183 skip regexp comments
1184 skip # initiated comments in //x patterns
1185 check for embedded @foo
1186 check for embedded scalars
1188 leave intact backslashes from leave (below)
1189 deprecate \1 in strings and sub replacements
1190 handle string-changing backslashes \l \U \Q \E, etc.
1191 switch (what was escaped) {
1192 handle - in a transliteration (becomes a literal -)
1193 handle \132 octal characters
1194 handle 0x15 hex characters
1195 handle \cV (control V)
1196 handle printf backslashes (\f, \r, \n, etc)
1198 } (end if backslash)
1199 } (end while character to read)
1204 S_scan_const(pTHX_ char *start)
1206 register char *send = PL_bufend; /* end of the constant */
1207 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1208 register char *s = start; /* start of the constant */
1209 register char *d = SvPVX(sv); /* destination for copies */
1210 bool dorange = FALSE; /* are we in a translit range? */
1211 bool has_utf = FALSE; /* embedded \x{} */
1215 I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1216 ? (PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1218 I32 thisutf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1219 ? (PL_sublex_info.sub_op->op_private & (PL_lex_repl ?
1220 OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF))
1222 const char *leaveit = /* set of acceptably-backslashed characters */
1224 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1227 while (s < send || dorange) {
1228 /* get transliterations out of the way (they're most literal) */
1229 if (PL_lex_inwhat == OP_TRANS) {
1230 /* expand a range A-Z to the full set of characters. AIE! */
1232 I32 i; /* current expanded character */
1233 I32 min; /* first character in range */
1234 I32 max; /* last character in range */
1236 i = d - SvPVX(sv); /* remember current offset */
1237 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1238 d = SvPVX(sv) + i; /* refresh d after realloc */
1239 d -= 2; /* eat the first char and the - */
1241 min = (U8)*d; /* first char in range */
1242 max = (U8)d[1]; /* last char in range */
1245 if ((isLOWER(min) && isLOWER(max)) ||
1246 (isUPPER(min) && isUPPER(max))) {
1248 for (i = min; i <= max; i++)
1252 for (i = min; i <= max; i++)
1259 for (i = min; i <= max; i++)
1262 /* mark the range as done, and continue */
1267 /* range begins (ignore - as first or last char) */
1268 else if (*s == '-' && s+1 < send && s != start) {
1270 *d++ = (char)0xff; /* use illegal utf8 byte--see pmtrans */
1279 /* if we get here, we're not doing a transliteration */
1281 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1282 except for the last char, which will be done separately. */
1283 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1285 while (s < send && *s != ')')
1288 else if (s[2] == '{' /* This should match regcomp.c */
1289 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1292 char *regparse = s + (s[2] == '{' ? 3 : 4);
1295 while (count && (c = *regparse)) {
1296 if (c == '\\' && regparse[1])
1304 if (*regparse != ')') {
1305 regparse--; /* Leave one char for continuation. */
1306 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1308 while (s < regparse)
1313 /* likewise skip #-initiated comments in //x patterns */
1314 else if (*s == '#' && PL_lex_inpat &&
1315 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1316 while (s+1 < send && *s != '\n')
1320 /* check for embedded arrays (@foo, @:foo, @'foo, @{foo}, @$foo) */
1321 else if (*s == '@' && s[1]
1322 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$", s[1])))
1325 /* check for embedded scalars. only stop if we're sure it's a
1328 else if (*s == '$') {
1329 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1331 if (s + 1 < send && !strchr("()| \n\t", s[1]))
1332 break; /* in regexp, $ might be tail anchor */
1335 /* (now in tr/// code again) */
1337 if (*s & 0x80 && thisutf) {
1338 (void)utf8_to_uv((U8*)s, &len);
1340 /* illegal UTF8, make it valid */
1341 char *old_pvx = SvPVX(sv);
1342 /* need space for one extra char (NOTE: SvCUR() not set here) */
1343 d = SvGROW(sv, SvLEN(sv) + 1) + (d - old_pvx);
1344 d = (char*)uv_to_utf8((U8*)d, (U8)*s++);
1355 if (*s == '\\' && s+1 < send) {
1358 /* some backslashes we leave behind */
1359 if (*leaveit && *s && strchr(leaveit, *s)) {
1365 /* deprecate \1 in strings and substitution replacements */
1366 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1367 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1369 dTHR; /* only for ckWARN */
1370 if (ckWARN(WARN_SYNTAX))
1371 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1376 /* string-change backslash escapes */
1377 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1382 /* if we get here, it's either a quoted -, or a digit */
1385 /* quoted - in transliterations */
1387 if (PL_lex_inwhat == OP_TRANS) {
1395 if (ckWARN(WARN_MISC) && isALNUM(*s) && *s != '_')
1396 Perl_warner(aTHX_ WARN_MISC,
1397 "Unrecognized escape \\%c passed through",
1399 /* default action is to copy the quoted character */
1404 /* \132 indicates an octal constant */
1405 case '0': case '1': case '2': case '3':
1406 case '4': case '5': case '6': case '7':
1407 len = 0; /* disallow underscores */
1408 uv = (UV)scan_oct(s, 3, &len);
1410 goto NUM_ESCAPE_INSERT;
1412 /* \x24 indicates a hex constant */
1416 char* e = strchr(s, '}');
1418 yyerror("Missing right brace on \\x{}");
1421 len = 1; /* allow underscores */
1422 uv = (UV)scan_hex(s + 1, e - s - 1, &len);
1426 len = 0; /* disallow underscores */
1427 uv = (UV)scan_hex(s, 2, &len);
1432 /* Insert oct or hex escaped character.
1433 * There will always enough room in sv since such escapes will
1434 * be longer than any utf8 sequence they can end up as
1437 if (!thisutf && !has_utf && uv > 255) {
1438 /* might need to recode whatever we have accumulated so far
1439 * if it contains any hibit chars
1443 for (c = SvPVX(sv); c < d; c++) {
1448 char *old_pvx = SvPVX(sv);
1450 d = SvGROW(sv, SvCUR(sv) + hicount + 1) + (d - old_pvx);
1459 uv_to_utf8((U8*)dst, (U8)*src--);
1469 if (thisutf || uv > 255) {
1470 d = (char*)uv_to_utf8((U8*)d, uv);
1482 /* \N{latin small letter a} is a named character */
1486 char* e = strchr(s, '}');
1492 yyerror("Missing right brace on \\N{}");
1496 res = newSVpvn(s + 1, e - s - 1);
1497 res = new_constant( Nullch, 0, "charnames",
1498 res, Nullsv, "\\N{...}" );
1499 str = SvPV(res,len);
1500 if (!has_utf && SvUTF8(res)) {
1501 char *ostart = SvPVX(sv);
1502 SvCUR_set(sv, d - ostart);
1504 sv_utf8_upgrade(sv);
1505 d = SvPVX(sv) + SvCUR(sv);
1508 if (len > e - s + 4) {
1509 char *odest = SvPVX(sv);
1511 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1512 d = SvPVX(sv) + (d - odest);
1514 Copy(str, d, len, char);
1521 yyerror("Missing braces on \\N{}");
1524 /* \c is a control character */
1539 /* printf-style backslashes, formfeeds, newlines, etc */
1557 *d++ = '\047'; /* CP 1047 */
1560 *d++ = '\057'; /* CP 1047 */
1574 } /* end if (backslash) */
1577 } /* while loop to process each character */
1579 /* terminate the string and set up the sv */
1581 SvCUR_set(sv, d - SvPVX(sv));
1586 /* shrink the sv if we allocated more than we used */
1587 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1588 SvLEN_set(sv, SvCUR(sv) + 1);
1589 Renew(SvPVX(sv), SvLEN(sv), char);
1592 /* return the substring (via yylval) only if we parsed anything */
1593 if (s > PL_bufptr) {
1594 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1595 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1597 ( PL_lex_inwhat == OP_TRANS
1599 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1602 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1609 * Returns TRUE if there's more to the expression (e.g., a subscript),
1612 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1614 * ->[ and ->{ return TRUE
1615 * { and [ outside a pattern are always subscripts, so return TRUE
1616 * if we're outside a pattern and it's not { or [, then return FALSE
1617 * if we're in a pattern and the first char is a {
1618 * {4,5} (any digits around the comma) returns FALSE
1619 * if we're in a pattern and the first char is a [
1621 * [SOMETHING] has a funky algorithm to decide whether it's a
1622 * character class or not. It has to deal with things like
1623 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1624 * anything else returns TRUE
1627 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1630 S_intuit_more(pTHX_ register char *s)
1632 if (PL_lex_brackets)
1634 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1636 if (*s != '{' && *s != '[')
1641 /* In a pattern, so maybe we have {n,m}. */
1658 /* On the other hand, maybe we have a character class */
1661 if (*s == ']' || *s == '^')
1664 /* this is terrifying, and it works */
1665 int weight = 2; /* let's weigh the evidence */
1667 unsigned char un_char = 255, last_un_char;
1668 char *send = strchr(s,']');
1669 char tmpbuf[sizeof PL_tokenbuf * 4];
1671 if (!send) /* has to be an expression */
1674 Zero(seen,256,char);
1677 else if (isDIGIT(*s)) {
1679 if (isDIGIT(s[1]) && s[2] == ']')
1685 for (; s < send; s++) {
1686 last_un_char = un_char;
1687 un_char = (unsigned char)*s;
1692 weight -= seen[un_char] * 10;
1693 if (isALNUM_lazy_if(s+1,UTF)) {
1694 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1695 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1700 else if (*s == '$' && s[1] &&
1701 strchr("[#!%*<>()-=",s[1])) {
1702 if (/*{*/ strchr("])} =",s[2]))
1711 if (strchr("wds]",s[1]))
1713 else if (seen['\''] || seen['"'])
1715 else if (strchr("rnftbxcav",s[1]))
1717 else if (isDIGIT(s[1])) {
1719 while (s[1] && isDIGIT(s[1]))
1729 if (strchr("aA01! ",last_un_char))
1731 if (strchr("zZ79~",s[1]))
1733 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1734 weight -= 5; /* cope with negative subscript */
1737 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1738 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1743 if (keyword(tmpbuf, d - tmpbuf))
1746 if (un_char == last_un_char + 1)
1748 weight -= seen[un_char];
1753 if (weight >= 0) /* probably a character class */
1763 * Does all the checking to disambiguate
1765 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1766 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1768 * First argument is the stuff after the first token, e.g. "bar".
1770 * Not a method if bar is a filehandle.
1771 * Not a method if foo is a subroutine prototyped to take a filehandle.
1772 * Not a method if it's really "Foo $bar"
1773 * Method if it's "foo $bar"
1774 * Not a method if it's really "print foo $bar"
1775 * Method if it's really "foo package::" (interpreted as package->foo)
1776 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1777 * Not a method if bar is a filehandle or package, but is quoted with
1782 S_intuit_method(pTHX_ char *start, GV *gv)
1784 char *s = start + (*start == '$');
1785 char tmpbuf[sizeof PL_tokenbuf];
1793 if ((cv = GvCVu(gv))) {
1794 char *proto = SvPVX(cv);
1804 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1805 /* start is the beginning of the possible filehandle/object,
1806 * and s is the end of it
1807 * tmpbuf is a copy of it
1810 if (*start == '$') {
1811 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1816 return *s == '(' ? FUNCMETH : METHOD;
1818 if (!keyword(tmpbuf, len)) {
1819 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1824 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1825 if (indirgv && GvCVu(indirgv))
1827 /* filehandle or package name makes it a method */
1828 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1830 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1831 return 0; /* no assumptions -- "=>" quotes bearword */
1833 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1834 newSVpvn(tmpbuf,len));
1835 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1839 return *s == '(' ? FUNCMETH : METHOD;
1847 * Return a string of Perl code to load the debugger. If PERL5DB
1848 * is set, it will return the contents of that, otherwise a
1849 * compile-time require of perl5db.pl.
1856 char *pdb = PerlEnv_getenv("PERL5DB");
1860 SETERRNO(0,SS$_NORMAL);
1861 return "BEGIN { require 'perl5db.pl' }";
1867 /* Encoded script support. filter_add() effectively inserts a
1868 * 'pre-processing' function into the current source input stream.
1869 * Note that the filter function only applies to the current source file
1870 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1872 * The datasv parameter (which may be NULL) can be used to pass
1873 * private data to this instance of the filter. The filter function
1874 * can recover the SV using the FILTER_DATA macro and use it to
1875 * store private buffers and state information.
1877 * The supplied datasv parameter is upgraded to a PVIO type
1878 * and the IoDIRP field is used to store the function pointer,
1879 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1880 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1881 * private use must be set using malloc'd pointers.
1885 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1890 if (!PL_rsfp_filters)
1891 PL_rsfp_filters = newAV();
1893 datasv = NEWSV(255,0);
1894 if (!SvUPGRADE(datasv, SVt_PVIO))
1895 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1896 IoDIRP(datasv) = (DIR*)funcp; /* stash funcp into spare field */
1897 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1898 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1899 funcp, SvPV_nolen(datasv)));
1900 av_unshift(PL_rsfp_filters, 1);
1901 av_store(PL_rsfp_filters, 0, datasv) ;
1906 /* Delete most recently added instance of this filter function. */
1908 Perl_filter_del(pTHX_ filter_t funcp)
1911 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1912 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1914 /* if filter is on top of stack (usual case) just pop it off */
1915 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1916 if (IoDIRP(datasv) == (DIR*)funcp) {
1917 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1918 IoDIRP(datasv) = (DIR*)NULL;
1919 sv_free(av_pop(PL_rsfp_filters));
1923 /* we need to search for the correct entry and clear it */
1924 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1928 /* Invoke the n'th filter function for the current rsfp. */
1930 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1933 /* 0 = read one text line */
1938 if (!PL_rsfp_filters)
1940 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1941 /* Provide a default input filter to make life easy. */
1942 /* Note that we append to the line. This is handy. */
1943 DEBUG_P(PerlIO_printf(Perl_debug_log,
1944 "filter_read %d: from rsfp\n", idx));
1948 int old_len = SvCUR(buf_sv) ;
1950 /* ensure buf_sv is large enough */
1951 SvGROW(buf_sv, old_len + maxlen) ;
1952 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1953 if (PerlIO_error(PL_rsfp))
1954 return -1; /* error */
1956 return 0 ; /* end of file */
1958 SvCUR_set(buf_sv, old_len + len) ;
1961 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1962 if (PerlIO_error(PL_rsfp))
1963 return -1; /* error */
1965 return 0 ; /* end of file */
1968 return SvCUR(buf_sv);
1970 /* Skip this filter slot if filter has been deleted */
1971 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1972 DEBUG_P(PerlIO_printf(Perl_debug_log,
1973 "filter_read %d: skipped (filter deleted)\n",
1975 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1977 /* Get function pointer hidden within datasv */
1978 funcp = (filter_t)IoDIRP(datasv);
1979 DEBUG_P(PerlIO_printf(Perl_debug_log,
1980 "filter_read %d: via function %p (%s)\n",
1981 idx, funcp, SvPV_nolen(datasv)));
1982 /* Call function. The function is expected to */
1983 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1984 /* Return: <0:error, =0:eof, >0:not eof */
1985 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1989 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1991 #ifdef PERL_CR_FILTER
1992 if (!PL_rsfp_filters) {
1993 filter_add(S_cr_textfilter,NULL);
1996 if (PL_rsfp_filters) {
1999 SvCUR_set(sv, 0); /* start with empty line */
2000 if (FILTER_READ(0, sv, 0) > 0)
2001 return ( SvPVX(sv) ) ;
2006 return (sv_gets(sv, fp, append));
2009 STATIC HV *S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2013 if (*pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2017 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2018 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV))) {
2019 return GvHV(gv); /* Foo:: */
2022 /* use constant CLASS => 'MyClass' */
2023 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2025 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2026 pkgname = SvPV_nolen(sv);
2030 return gv_stashpv(pkgname, FALSE);
2034 static char* exp_name[] =
2035 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2036 "ATTRTERM", "TERMBLOCK"
2043 Works out what to call the token just pulled out of the input
2044 stream. The yacc parser takes care of taking the ops we return and
2045 stitching them into a tree.
2051 if read an identifier
2052 if we're in a my declaration
2053 croak if they tried to say my($foo::bar)
2054 build the ops for a my() declaration
2055 if it's an access to a my() variable
2056 are we in a sort block?
2057 croak if my($a); $a <=> $b
2058 build ops for access to a my() variable
2059 if in a dq string, and they've said @foo and we can't find @foo
2061 build ops for a bareword
2062 if we already built the token before, use it.
2066 #pragma segment Perl_yylex
2069 #ifdef USE_PURE_BISON
2070 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
2083 #ifdef USE_PURE_BISON
2084 yylval_pointer = lvalp;
2085 yychar_pointer = lcharp;
2088 /* check if there's an identifier for us to look at */
2089 if (PL_pending_ident) {
2090 /* pit holds the identifier we read and pending_ident is reset */
2091 char pit = PL_pending_ident;
2092 PL_pending_ident = 0;
2094 /* if we're in a my(), we can't allow dynamics here.
2095 $foo'bar has already been turned into $foo::bar, so
2096 just check for colons.
2098 if it's a legal name, the OP is a PADANY.
2101 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
2102 if (strchr(PL_tokenbuf,':'))
2103 yyerror(Perl_form(aTHX_ "No package name allowed for "
2104 "variable %s in \"our\"",
2106 tmp = pad_allocmy(PL_tokenbuf);
2109 if (strchr(PL_tokenbuf,':'))
2110 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
2112 yylval.opval = newOP(OP_PADANY, 0);
2113 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
2119 build the ops for accesses to a my() variable.
2121 Deny my($a) or my($b) in a sort block, *if* $a or $b is
2122 then used in a comparison. This catches most, but not
2123 all cases. For instance, it catches
2124 sort { my($a); $a <=> $b }
2126 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2127 (although why you'd do that is anyone's guess).
2130 if (!strchr(PL_tokenbuf,':')) {
2132 /* Check for single character per-thread SVs */
2133 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2134 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2135 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2137 yylval.opval = newOP(OP_THREADSV, 0);
2138 yylval.opval->op_targ = tmp;
2141 #endif /* USE_THREADS */
2142 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2143 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2144 /* might be an "our" variable" */
2145 if (SvFLAGS(namesv) & SVpad_OUR) {
2146 /* build ops for a bareword */
2147 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2148 sv_catpvn(sym, "::", 2);
2149 sv_catpv(sym, PL_tokenbuf+1);
2150 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2151 yylval.opval->op_private = OPpCONST_ENTERED;
2152 gv_fetchpv(SvPVX(sym),
2154 ? (GV_ADDMULTI | GV_ADDINEVAL)
2157 ((PL_tokenbuf[0] == '$') ? SVt_PV
2158 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2163 /* if it's a sort block and they're naming $a or $b */
2164 if (PL_last_lop_op == OP_SORT &&
2165 PL_tokenbuf[0] == '$' &&
2166 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2169 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2170 d < PL_bufend && *d != '\n';
2173 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2174 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2180 yylval.opval = newOP(OP_PADANY, 0);
2181 yylval.opval->op_targ = tmp;
2187 Whine if they've said @foo in a doublequoted string,
2188 and @foo isn't a variable we can find in the symbol
2191 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2192 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2193 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2194 && ckWARN(WARN_AMBIGUOUS))
2196 /* Downgraded from fatal to warning 20000522 mjd */
2197 Perl_warner(aTHX_ WARN_AMBIGUOUS,
2198 "Possible unintended interpolation of %s in string",
2203 /* build ops for a bareword */
2204 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2205 yylval.opval->op_private = OPpCONST_ENTERED;
2206 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2207 ((PL_tokenbuf[0] == '$') ? SVt_PV
2208 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2213 /* no identifier pending identification */
2215 switch (PL_lex_state) {
2217 case LEX_NORMAL: /* Some compilers will produce faster */
2218 case LEX_INTERPNORMAL: /* code if we comment these out. */
2222 /* when we've already built the next token, just pull it out of the queue */
2225 yylval = PL_nextval[PL_nexttoke];
2227 PL_lex_state = PL_lex_defer;
2228 PL_expect = PL_lex_expect;
2229 PL_lex_defer = LEX_NORMAL;
2231 return(PL_nexttype[PL_nexttoke]);
2233 /* interpolated case modifiers like \L \U, including \Q and \E.
2234 when we get here, PL_bufptr is at the \
2236 case LEX_INTERPCASEMOD:
2238 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2239 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2241 /* handle \E or end of string */
2242 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2246 if (PL_lex_casemods) {
2247 oldmod = PL_lex_casestack[--PL_lex_casemods];
2248 PL_lex_casestack[PL_lex_casemods] = '\0';
2250 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2252 PL_lex_state = LEX_INTERPCONCAT;
2256 if (PL_bufptr != PL_bufend)
2258 PL_lex_state = LEX_INTERPCONCAT;
2263 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2264 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2265 if (strchr("LU", *s) &&
2266 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2268 PL_lex_casestack[--PL_lex_casemods] = '\0';
2271 if (PL_lex_casemods > 10) {
2272 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2273 if (newlb != PL_lex_casestack) {
2275 PL_lex_casestack = newlb;
2278 PL_lex_casestack[PL_lex_casemods++] = *s;
2279 PL_lex_casestack[PL_lex_casemods] = '\0';
2280 PL_lex_state = LEX_INTERPCONCAT;
2281 PL_nextval[PL_nexttoke].ival = 0;
2284 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2286 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2288 PL_nextval[PL_nexttoke].ival = OP_LC;
2290 PL_nextval[PL_nexttoke].ival = OP_UC;
2292 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2294 Perl_croak(aTHX_ "panic: yylex");
2297 if (PL_lex_starts) {
2306 case LEX_INTERPPUSH:
2307 return sublex_push();
2309 case LEX_INTERPSTART:
2310 if (PL_bufptr == PL_bufend)
2311 return sublex_done();
2313 PL_lex_dojoin = (*PL_bufptr == '@');
2314 PL_lex_state = LEX_INTERPNORMAL;
2315 if (PL_lex_dojoin) {
2316 PL_nextval[PL_nexttoke].ival = 0;
2319 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2320 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2321 force_next(PRIVATEREF);
2323 force_ident("\"", '$');
2324 #endif /* USE_THREADS */
2325 PL_nextval[PL_nexttoke].ival = 0;
2327 PL_nextval[PL_nexttoke].ival = 0;
2329 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2332 if (PL_lex_starts++) {
2338 case LEX_INTERPENDMAYBE:
2339 if (intuit_more(PL_bufptr)) {
2340 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2346 if (PL_lex_dojoin) {
2347 PL_lex_dojoin = FALSE;
2348 PL_lex_state = LEX_INTERPCONCAT;
2351 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2352 && SvEVALED(PL_lex_repl))
2354 if (PL_bufptr != PL_bufend)
2355 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2356 PL_lex_repl = Nullsv;
2359 case LEX_INTERPCONCAT:
2361 if (PL_lex_brackets)
2362 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2364 if (PL_bufptr == PL_bufend)
2365 return sublex_done();
2367 if (SvIVX(PL_linestr) == '\'') {
2368 SV *sv = newSVsv(PL_linestr);
2371 else if ( PL_hints & HINT_NEW_RE )
2372 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2373 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2377 s = scan_const(PL_bufptr);
2379 PL_lex_state = LEX_INTERPCASEMOD;
2381 PL_lex_state = LEX_INTERPSTART;
2384 if (s != PL_bufptr) {
2385 PL_nextval[PL_nexttoke] = yylval;
2388 if (PL_lex_starts++)
2398 PL_lex_state = LEX_NORMAL;
2399 s = scan_formline(PL_bufptr);
2400 if (!PL_lex_formbrack)
2406 PL_oldoldbufptr = PL_oldbufptr;
2409 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2410 exp_name[PL_expect], s);
2416 if (isIDFIRST_lazy_if(s,UTF))
2418 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2421 goto fake_eof; /* emulate EOF on ^D or ^Z */
2426 if (PL_lex_brackets)
2427 yyerror("Missing right curly or square bracket");
2430 if (s++ < PL_bufend)
2431 goto retry; /* ignore stray nulls */
2434 if (!PL_in_eval && !PL_preambled) {
2435 PL_preambled = TRUE;
2436 sv_setpv(PL_linestr,incl_perldb());
2437 if (SvCUR(PL_linestr))
2438 sv_catpv(PL_linestr,";");
2440 while(AvFILLp(PL_preambleav) >= 0) {
2441 SV *tmpsv = av_shift(PL_preambleav);
2442 sv_catsv(PL_linestr, tmpsv);
2443 sv_catpv(PL_linestr, ";");
2446 sv_free((SV*)PL_preambleav);
2447 PL_preambleav = NULL;
2449 if (PL_minus_n || PL_minus_p) {
2450 sv_catpv(PL_linestr, "LINE: while (<>) {");
2452 sv_catpv(PL_linestr,"chomp;");
2454 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2456 GvIMPORTED_AV_on(gv);
2458 if (strchr("/'\"", *PL_splitstr)
2459 && strchr(PL_splitstr + 1, *PL_splitstr))
2460 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2463 s = "'~#\200\1'"; /* surely one char is unused...*/
2464 while (s[1] && strchr(PL_splitstr, *s)) s++;
2466 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2467 "q" + (delim == '\''), delim);
2468 for (s = PL_splitstr; *s; s++) {
2470 sv_catpvn(PL_linestr, "\\", 1);
2471 sv_catpvn(PL_linestr, s, 1);
2473 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2477 sv_catpv(PL_linestr,"@F=split(' ');");
2480 sv_catpv(PL_linestr, "\n");
2481 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2482 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2483 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2484 SV *sv = NEWSV(85,0);
2486 sv_upgrade(sv, SVt_PVMG);
2487 sv_setsv(sv,PL_linestr);
2488 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2494 bof = PL_rsfp && (PerlIO_tell(PL_rsfp)==0); /* *Before* read! */
2495 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2498 if (PL_preprocess && !PL_in_eval)
2499 (void)PerlProc_pclose(PL_rsfp);
2500 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2501 PerlIO_clearerr(PL_rsfp);
2503 (void)PerlIO_close(PL_rsfp);
2505 PL_doextract = FALSE;
2507 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2508 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2509 sv_catpv(PL_linestr,";}");
2510 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2511 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2512 PL_minus_n = PL_minus_p = 0;
2515 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2516 sv_setpv(PL_linestr,"");
2517 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2520 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2521 PL_doextract = FALSE;
2523 /* Incest with pod. */
2524 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2525 sv_setpv(PL_linestr, "");
2526 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2527 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2528 PL_doextract = FALSE;
2534 } while (PL_doextract);
2535 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2536 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2537 SV *sv = NEWSV(85,0);
2539 sv_upgrade(sv, SVt_PVMG);
2540 sv_setsv(sv,PL_linestr);
2541 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2543 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2544 if (CopLINE(PL_curcop) == 1) {
2545 while (s < PL_bufend && isSPACE(*s))
2547 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2551 if (*s == '#' && *(s+1) == '!')
2553 #ifdef ALTERNATE_SHEBANG
2555 static char as[] = ALTERNATE_SHEBANG;
2556 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2557 d = s + (sizeof(as) - 1);
2559 #endif /* ALTERNATE_SHEBANG */
2568 while (*d && !isSPACE(*d))
2572 #ifdef ARG_ZERO_IS_SCRIPT
2573 if (ipathend > ipath) {
2575 * HP-UX (at least) sets argv[0] to the script name,
2576 * which makes $^X incorrect. And Digital UNIX and Linux,
2577 * at least, set argv[0] to the basename of the Perl
2578 * interpreter. So, having found "#!", we'll set it right.
2580 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2581 assert(SvPOK(x) || SvGMAGICAL(x));
2582 if (sv_eq(x, CopFILESV(PL_curcop))) {
2583 sv_setpvn(x, ipath, ipathend - ipath);
2586 TAINT_NOT; /* $^X is always tainted, but that's OK */
2588 #endif /* ARG_ZERO_IS_SCRIPT */
2593 d = instr(s,"perl -");
2595 d = instr(s,"perl");
2597 /* avoid getting into infinite loops when shebang
2598 * line contains "Perl" rather than "perl" */
2600 for (d = ipathend-4; d >= ipath; --d) {
2601 if ((*d == 'p' || *d == 'P')
2602 && !ibcmp(d, "perl", 4))
2612 #ifdef ALTERNATE_SHEBANG
2614 * If the ALTERNATE_SHEBANG on this system starts with a
2615 * character that can be part of a Perl expression, then if
2616 * we see it but not "perl", we're probably looking at the
2617 * start of Perl code, not a request to hand off to some
2618 * other interpreter. Similarly, if "perl" is there, but
2619 * not in the first 'word' of the line, we assume the line
2620 * contains the start of the Perl program.
2622 if (d && *s != '#') {
2624 while (*c && !strchr("; \t\r\n\f\v#", *c))
2627 d = Nullch; /* "perl" not in first word; ignore */
2629 *s = '#'; /* Don't try to parse shebang line */
2631 #endif /* ALTERNATE_SHEBANG */
2632 #ifndef MACOS_TRADITIONAL
2637 !instr(s,"indir") &&
2638 instr(PL_origargv[0],"perl"))
2644 while (s < PL_bufend && isSPACE(*s))
2646 if (s < PL_bufend) {
2647 Newz(899,newargv,PL_origargc+3,char*);
2649 while (s < PL_bufend && !isSPACE(*s))
2652 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2655 newargv = PL_origargv;
2657 PerlProc_execv(ipath, newargv);
2658 Perl_croak(aTHX_ "Can't exec %s", ipath);
2662 U32 oldpdb = PL_perldb;
2663 bool oldn = PL_minus_n;
2664 bool oldp = PL_minus_p;
2666 while (*d && !isSPACE(*d)) d++;
2667 while (SPACE_OR_TAB(*d)) d++;
2671 if (*d == 'M' || *d == 'm') {
2673 while (*d && !isSPACE(*d)) d++;
2674 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2677 d = moreswitches(d);
2679 if ((PERLDB_LINE && !oldpdb) ||
2680 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2681 /* if we have already added "LINE: while (<>) {",
2682 we must not do it again */
2684 sv_setpv(PL_linestr, "");
2685 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2686 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2687 PL_preambled = FALSE;
2689 (void)gv_fetchfile(PL_origfilename);
2696 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2698 PL_lex_state = LEX_FORMLINE;
2703 #ifdef PERL_STRICT_CR
2704 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2706 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2708 case ' ': case '\t': case '\f': case 013:
2709 #ifdef MACOS_TRADITIONAL
2716 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2717 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2718 /* handle eval qq[#line 1 "foo"\n ...] */
2719 CopLINE_dec(PL_curcop);
2723 while (s < d && *s != '\n')
2728 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2730 PL_lex_state = LEX_FORMLINE;
2740 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2745 while (s < PL_bufend && SPACE_OR_TAB(*s))
2748 if (strnEQ(s,"=>",2)) {
2749 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2750 OPERATOR('-'); /* unary minus */
2752 PL_last_uni = PL_oldbufptr;
2753 PL_last_lop_op = OP_FTEREAD; /* good enough */
2755 case 'r': FTST(OP_FTEREAD);
2756 case 'w': FTST(OP_FTEWRITE);
2757 case 'x': FTST(OP_FTEEXEC);
2758 case 'o': FTST(OP_FTEOWNED);
2759 case 'R': FTST(OP_FTRREAD);
2760 case 'W': FTST(OP_FTRWRITE);
2761 case 'X': FTST(OP_FTREXEC);
2762 case 'O': FTST(OP_FTROWNED);
2763 case 'e': FTST(OP_FTIS);
2764 case 'z': FTST(OP_FTZERO);
2765 case 's': FTST(OP_FTSIZE);
2766 case 'f': FTST(OP_FTFILE);
2767 case 'd': FTST(OP_FTDIR);
2768 case 'l': FTST(OP_FTLINK);
2769 case 'p': FTST(OP_FTPIPE);
2770 case 'S': FTST(OP_FTSOCK);
2771 case 'u': FTST(OP_FTSUID);
2772 case 'g': FTST(OP_FTSGID);
2773 case 'k': FTST(OP_FTSVTX);
2774 case 'b': FTST(OP_FTBLK);
2775 case 'c': FTST(OP_FTCHR);
2776 case 't': FTST(OP_FTTTY);
2777 case 'T': FTST(OP_FTTEXT);
2778 case 'B': FTST(OP_FTBINARY);
2779 case 'M': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTMTIME);
2780 case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME);
2781 case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME);
2783 Perl_croak(aTHX_ "Unrecognized file test: -%c", (int)tmp);
2790 if (PL_expect == XOPERATOR)
2795 else if (*s == '>') {
2798 if (isIDFIRST_lazy_if(s,UTF)) {
2799 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2807 if (PL_expect == XOPERATOR)
2810 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2812 OPERATOR('-'); /* unary minus */
2819 if (PL_expect == XOPERATOR)
2824 if (PL_expect == XOPERATOR)
2827 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2833 if (PL_expect != XOPERATOR) {
2834 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2835 PL_expect = XOPERATOR;
2836 force_ident(PL_tokenbuf, '*');
2849 if (PL_expect == XOPERATOR) {
2853 PL_tokenbuf[0] = '%';
2854 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2855 if (!PL_tokenbuf[1]) {
2857 yyerror("Final % should be \\% or %name");
2860 PL_pending_ident = '%';
2879 switch (PL_expect) {
2882 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2884 PL_bufptr = s; /* update in case we back off */
2890 PL_expect = XTERMBLOCK;
2894 while (isIDFIRST_lazy_if(s,UTF)) {
2895 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2896 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2897 if (tmp < 0) tmp = -tmp;
2912 d = scan_str(d,TRUE,TRUE);
2915 SvREFCNT_dec(PL_lex_stuff);
2916 PL_lex_stuff = Nullsv;
2918 /* MUST advance bufptr here to avoid bogus
2919 "at end of line" context messages from yyerror().
2921 PL_bufptr = s + len;
2922 yyerror("Unterminated attribute parameter in attribute list");
2925 return 0; /* EOF indicator */
2929 SV *sv = newSVpvn(s, len);
2930 sv_catsv(sv, PL_lex_stuff);
2931 attrs = append_elem(OP_LIST, attrs,
2932 newSVOP(OP_CONST, 0, sv));
2933 SvREFCNT_dec(PL_lex_stuff);
2934 PL_lex_stuff = Nullsv;
2937 attrs = append_elem(OP_LIST, attrs,
2938 newSVOP(OP_CONST, 0,
2942 if (*s == ':' && s[1] != ':')
2945 break; /* require real whitespace or :'s */
2947 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
2948 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
2949 char q = ((*s == '\'') ? '"' : '\'');
2950 /* If here for an expression, and parsed no attrs, back off. */
2951 if (tmp == '=' && !attrs) {
2955 /* MUST advance bufptr here to avoid bogus "at end of line"
2956 context messages from yyerror().
2960 yyerror("Unterminated attribute list");
2962 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
2970 PL_nextval[PL_nexttoke].opval = attrs;
2978 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
2979 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
2995 if (PL_lex_brackets <= 0)
2996 yyerror("Unmatched right square bracket");
2999 if (PL_lex_state == LEX_INTERPNORMAL) {
3000 if (PL_lex_brackets == 0) {
3001 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3002 PL_lex_state = LEX_INTERPEND;
3009 if (PL_lex_brackets > 100) {
3010 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
3011 if (newlb != PL_lex_brackstack) {
3013 PL_lex_brackstack = newlb;
3016 switch (PL_expect) {
3018 if (PL_lex_formbrack) {
3022 if (PL_oldoldbufptr == PL_last_lop)
3023 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3025 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3026 OPERATOR(HASHBRACK);
3028 while (s < PL_bufend && SPACE_OR_TAB(*s))
3031 PL_tokenbuf[0] = '\0';
3032 if (d < PL_bufend && *d == '-') {
3033 PL_tokenbuf[0] = '-';
3035 while (d < PL_bufend && SPACE_OR_TAB(*d))
3038 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3039 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3041 while (d < PL_bufend && SPACE_OR_TAB(*d))
3044 char minus = (PL_tokenbuf[0] == '-');
3045 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3053 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3058 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3063 if (PL_oldoldbufptr == PL_last_lop)
3064 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3066 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3069 OPERATOR(HASHBRACK);
3070 /* This hack serves to disambiguate a pair of curlies
3071 * as being a block or an anon hash. Normally, expectation
3072 * determines that, but in cases where we're not in a
3073 * position to expect anything in particular (like inside
3074 * eval"") we have to resolve the ambiguity. This code
3075 * covers the case where the first term in the curlies is a
3076 * quoted string. Most other cases need to be explicitly
3077 * disambiguated by prepending a `+' before the opening
3078 * curly in order to force resolution as an anon hash.
3080 * XXX should probably propagate the outer expectation
3081 * into eval"" to rely less on this hack, but that could
3082 * potentially break current behavior of eval"".
3086 if (*s == '\'' || *s == '"' || *s == '`') {
3087 /* common case: get past first string, handling escapes */
3088 for (t++; t < PL_bufend && *t != *s;)
3089 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3093 else if (*s == 'q') {
3096 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3100 char open, close, term;
3103 while (t < PL_bufend && isSPACE(*t))
3107 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3111 for (t++; t < PL_bufend; t++) {
3112 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3114 else if (*t == open)
3118 for (t++; t < PL_bufend; t++) {
3119 if (*t == '\\' && t+1 < PL_bufend)
3121 else if (*t == close && --brackets <= 0)
3123 else if (*t == open)
3129 else if (isALNUM_lazy_if(t,UTF)) {
3131 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3134 while (t < PL_bufend && isSPACE(*t))
3136 /* if comma follows first term, call it an anon hash */
3137 /* XXX it could be a comma expression with loop modifiers */
3138 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3139 || (*t == '=' && t[1] == '>')))
3140 OPERATOR(HASHBRACK);
3141 if (PL_expect == XREF)
3144 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3150 yylval.ival = CopLINE(PL_curcop);
3151 if (isSPACE(*s) || *s == '#')
3152 PL_copline = NOLINE; /* invalidate current command line number */
3157 if (PL_lex_brackets <= 0)
3158 yyerror("Unmatched right curly bracket");
3160 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3161 if (PL_lex_brackets < PL_lex_formbrack)
3162 PL_lex_formbrack = 0;
3163 if (PL_lex_state == LEX_INTERPNORMAL) {
3164 if (PL_lex_brackets == 0) {
3165 if (PL_expect & XFAKEBRACK) {
3166 PL_expect &= XENUMMASK;
3167 PL_lex_state = LEX_INTERPEND;
3169 return yylex(); /* ignore fake brackets */
3171 if (*s == '-' && s[1] == '>')
3172 PL_lex_state = LEX_INTERPENDMAYBE;
3173 else if (*s != '[' && *s != '{')
3174 PL_lex_state = LEX_INTERPEND;
3177 if (PL_expect & XFAKEBRACK) {
3178 PL_expect &= XENUMMASK;
3180 return yylex(); /* ignore fake brackets */
3190 if (PL_expect == XOPERATOR) {
3191 if (ckWARN(WARN_SEMICOLON)
3192 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3194 CopLINE_dec(PL_curcop);
3195 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3196 CopLINE_inc(PL_curcop);
3201 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3203 PL_expect = XOPERATOR;
3204 force_ident(PL_tokenbuf, '&');
3208 yylval.ival = (OPpENTERSUB_AMPER<<8);
3227 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3228 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3230 if (PL_expect == XSTATE && isALPHA(tmp) &&
3231 (s == PL_linestart+1 || s[-2] == '\n') )
3233 if (PL_in_eval && !PL_rsfp) {
3238 if (strnEQ(s,"=cut",4)) {
3252 PL_doextract = TRUE;
3255 if (PL_lex_brackets < PL_lex_formbrack) {
3257 #ifdef PERL_STRICT_CR
3258 for (t = s; SPACE_OR_TAB(*t); t++) ;
3260 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3262 if (*t == '\n' || *t == '#') {
3280 if (PL_expect != XOPERATOR) {
3281 if (s[1] != '<' && !strchr(s,'>'))
3284 s = scan_heredoc(s);
3286 s = scan_inputsymbol(s);
3287 TERM(sublex_start());
3292 SHop(OP_LEFT_SHIFT);
3306 SHop(OP_RIGHT_SHIFT);
3315 if (PL_expect == XOPERATOR) {
3316 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3319 return ','; /* grandfather non-comma-format format */
3323 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3324 PL_tokenbuf[0] = '@';
3325 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3326 sizeof PL_tokenbuf - 1, FALSE);
3327 if (PL_expect == XOPERATOR)
3328 no_op("Array length", s);
3329 if (!PL_tokenbuf[1])
3331 PL_expect = XOPERATOR;
3332 PL_pending_ident = '#';
3336 PL_tokenbuf[0] = '$';
3337 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3338 sizeof PL_tokenbuf - 1, FALSE);
3339 if (PL_expect == XOPERATOR)
3341 if (!PL_tokenbuf[1]) {
3343 yyerror("Final $ should be \\$ or $name");
3347 /* This kludge not intended to be bulletproof. */
3348 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3349 yylval.opval = newSVOP(OP_CONST, 0,
3350 newSViv(PL_compiling.cop_arybase));
3351 yylval.opval->op_private = OPpCONST_ARYBASE;
3357 if (PL_lex_state == LEX_NORMAL)
3360 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3363 PL_tokenbuf[0] = '@';
3364 if (ckWARN(WARN_SYNTAX)) {
3366 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3369 PL_bufptr = skipspace(PL_bufptr);
3370 while (t < PL_bufend && *t != ']')
3372 Perl_warner(aTHX_ WARN_SYNTAX,
3373 "Multidimensional syntax %.*s not supported",
3374 (t - PL_bufptr) + 1, PL_bufptr);
3378 else if (*s == '{') {
3379 PL_tokenbuf[0] = '%';
3380 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3381 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3383 char tmpbuf[sizeof PL_tokenbuf];
3385 for (t++; isSPACE(*t); t++) ;
3386 if (isIDFIRST_lazy_if(t,UTF)) {
3387 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3388 for (; isSPACE(*t); t++) ;
3389 if (*t == ';' && get_cv(tmpbuf, FALSE))
3390 Perl_warner(aTHX_ WARN_SYNTAX,
3391 "You need to quote \"%s\"", tmpbuf);
3397 PL_expect = XOPERATOR;
3398 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3399 bool islop = (PL_last_lop == PL_oldoldbufptr);
3400 if (!islop || PL_last_lop_op == OP_GREPSTART)
3401 PL_expect = XOPERATOR;
3402 else if (strchr("$@\"'`q", *s))
3403 PL_expect = XTERM; /* e.g. print $fh "foo" */
3404 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3405 PL_expect = XTERM; /* e.g. print $fh &sub */
3406 else if (isIDFIRST_lazy_if(s,UTF)) {
3407 char tmpbuf[sizeof PL_tokenbuf];
3408 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3409 if ((tmp = keyword(tmpbuf, len))) {
3410 /* binary operators exclude handle interpretations */
3422 PL_expect = XTERM; /* e.g. print $fh length() */
3427 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3428 if (gv && GvCVu(gv))
3429 PL_expect = XTERM; /* e.g. print $fh subr() */
3432 else if (isDIGIT(*s))
3433 PL_expect = XTERM; /* e.g. print $fh 3 */
3434 else if (*s == '.' && isDIGIT(s[1]))
3435 PL_expect = XTERM; /* e.g. print $fh .3 */
3436 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3437 PL_expect = XTERM; /* e.g. print $fh -1 */
3438 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3439 PL_expect = XTERM; /* print $fh <<"EOF" */
3441 PL_pending_ident = '$';
3445 if (PL_expect == XOPERATOR)
3447 PL_tokenbuf[0] = '@';
3448 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3449 if (!PL_tokenbuf[1]) {
3451 yyerror("Final @ should be \\@ or @name");
3454 if (PL_lex_state == LEX_NORMAL)
3456 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3458 PL_tokenbuf[0] = '%';
3460 /* Warn about @ where they meant $. */
3461 if (ckWARN(WARN_SYNTAX)) {
3462 if (*s == '[' || *s == '{') {
3464 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3466 if (*t == '}' || *t == ']') {
3468 PL_bufptr = skipspace(PL_bufptr);
3469 Perl_warner(aTHX_ WARN_SYNTAX,
3470 "Scalar value %.*s better written as $%.*s",
3471 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3476 PL_pending_ident = '@';
3479 case '/': /* may either be division or pattern */
3480 case '?': /* may either be conditional or pattern */
3481 if (PL_expect != XOPERATOR) {
3482 /* Disable warning on "study /blah/" */
3483 if (PL_oldoldbufptr == PL_last_uni
3484 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3485 || memNE(PL_last_uni, "study", 5)
3486 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3488 s = scan_pat(s,OP_MATCH);
3489 TERM(sublex_start());
3497 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3498 #ifdef PERL_STRICT_CR
3501 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3503 && (s == PL_linestart || s[-1] == '\n') )
3505 PL_lex_formbrack = 0;
3509 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3515 yylval.ival = OPf_SPECIAL;
3521 if (PL_expect != XOPERATOR)
3526 case '0': case '1': case '2': case '3': case '4':
3527 case '5': case '6': case '7': case '8': case '9':
3529 if (PL_expect == XOPERATOR)
3534 s = scan_str(s,FALSE,FALSE);
3535 if (PL_expect == XOPERATOR) {
3536 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3539 return ','; /* grandfather non-comma-format format */
3545 missingterm((char*)0);
3546 yylval.ival = OP_CONST;
3547 TERM(sublex_start());
3550 s = scan_str(s,FALSE,FALSE);
3551 if (PL_expect == XOPERATOR) {
3552 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3555 return ','; /* grandfather non-comma-format format */
3561 missingterm((char*)0);
3562 yylval.ival = OP_CONST;
3563 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3564 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3565 yylval.ival = OP_STRINGIFY;
3569 TERM(sublex_start());
3572 s = scan_str(s,FALSE,FALSE);
3573 if (PL_expect == XOPERATOR)
3574 no_op("Backticks",s);
3576 missingterm((char*)0);
3577 yylval.ival = OP_BACKTICK;
3579 TERM(sublex_start());
3583 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3584 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3586 if (PL_expect == XOPERATOR)
3587 no_op("Backslash",s);
3591 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3595 while (isDIGIT(*start) || *start == '_')
3597 if (*start == '.' && isDIGIT(start[1])) {
3601 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3602 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF)) {
3606 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3616 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3655 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3657 /* Some keywords can be followed by any delimiter, including ':' */
3658 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3659 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3660 (PL_tokenbuf[0] == 'q' &&
3661 strchr("qwxr", PL_tokenbuf[1])))));
3663 /* x::* is just a word, unless x is "CORE" */
3664 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3668 while (d < PL_bufend && isSPACE(*d))
3669 d++; /* no comments skipped here, or s### is misparsed */
3671 /* Is this a label? */
3672 if (!tmp && PL_expect == XSTATE
3673 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3675 yylval.pval = savepv(PL_tokenbuf);
3680 /* Check for keywords */
3681 tmp = keyword(PL_tokenbuf, len);
3683 /* Is this a word before a => operator? */
3684 if (*d == '=' && d[1] == '>') {
3686 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3687 yylval.opval->op_private = OPpCONST_BARE;
3691 if (tmp < 0) { /* second-class keyword? */
3692 GV *ogv = Nullgv; /* override (winner) */
3693 GV *hgv = Nullgv; /* hidden (loser) */
3694 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3696 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3699 if (GvIMPORTED_CV(gv))
3701 else if (! CvMETHOD(cv))
3705 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3706 (gv = *gvp) != (GV*)&PL_sv_undef &&
3707 GvCVu(gv) && GvIMPORTED_CV(gv))
3713 tmp = 0; /* overridden by import or by GLOBAL */
3716 && -tmp==KEY_lock /* XXX generalizable kludge */
3718 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3720 tmp = 0; /* any sub overrides "weak" keyword */
3722 else { /* no override */
3726 if (ckWARN(WARN_AMBIGUOUS) && hgv
3727 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3728 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3729 "Ambiguous call resolved as CORE::%s(), %s",
3730 GvENAME(hgv), "qualify as such or use &");
3737 default: /* not a keyword */
3740 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3742 /* Get the rest if it looks like a package qualifier */
3744 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3746 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3749 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3750 *s == '\'' ? "'" : "::");
3754 if (PL_expect == XOPERATOR) {
3755 if (PL_bufptr == PL_linestart) {
3756 CopLINE_dec(PL_curcop);
3757 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3758 CopLINE_inc(PL_curcop);
3761 no_op("Bareword",s);
3764 /* Look for a subroutine with this name in current package,
3765 unless name is "Foo::", in which case Foo is a bearword
3766 (and a package name). */
3769 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3771 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3772 Perl_warner(aTHX_ WARN_BAREWORD,
3773 "Bareword \"%s\" refers to nonexistent package",
3776 PL_tokenbuf[len] = '\0';
3783 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3786 /* if we saw a global override before, get the right name */
3789 sv = newSVpvn("CORE::GLOBAL::",14);
3790 sv_catpv(sv,PL_tokenbuf);
3793 sv = newSVpv(PL_tokenbuf,0);
3795 /* Presume this is going to be a bareword of some sort. */
3798 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3799 yylval.opval->op_private = OPpCONST_BARE;
3801 /* And if "Foo::", then that's what it certainly is. */
3806 /* See if it's the indirect object for a list operator. */
3808 if (PL_oldoldbufptr &&
3809 PL_oldoldbufptr < PL_bufptr &&
3810 (PL_oldoldbufptr == PL_last_lop
3811 || PL_oldoldbufptr == PL_last_uni) &&
3812 /* NO SKIPSPACE BEFORE HERE! */
3813 (PL_expect == XREF ||
3814 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3816 bool immediate_paren = *s == '(';
3818 /* (Now we can afford to cross potential line boundary.) */
3821 /* Two barewords in a row may indicate method call. */
3823 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3826 /* If not a declared subroutine, it's an indirect object. */
3827 /* (But it's an indir obj regardless for sort.) */
3829 if ((PL_last_lop_op == OP_SORT ||
3830 (!immediate_paren && (!gv || !GvCVu(gv)))) &&
3831 (PL_last_lop_op != OP_MAPSTART &&
3832 PL_last_lop_op != OP_GREPSTART))
3834 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3840 PL_expect = XOPERATOR;
3843 /* Is this a word before a => operator? */
3844 if (*s == '=' && s[1] == '>') {
3846 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
3850 /* If followed by a paren, it's certainly a subroutine. */
3853 if (gv && GvCVu(gv)) {
3854 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
3855 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3860 PL_nextval[PL_nexttoke].opval = yylval.opval;
3861 PL_expect = XOPERATOR;
3867 /* If followed by var or block, call it a method (unless sub) */
3869 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3870 PL_last_lop = PL_oldbufptr;
3871 PL_last_lop_op = OP_METHOD;
3875 /* If followed by a bareword, see if it looks like indir obj. */
3877 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3880 /* Not a method, so call it a subroutine (if defined) */
3882 if (gv && GvCVu(gv)) {
3884 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3885 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3886 "Ambiguous use of -%s resolved as -&%s()",
3887 PL_tokenbuf, PL_tokenbuf);
3888 /* Check for a constant sub */
3890 if ((sv = cv_const_sv(cv))) {
3892 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3893 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3894 yylval.opval->op_private = 0;
3898 /* Resolve to GV now. */
3899 op_free(yylval.opval);
3900 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3901 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3902 PL_last_lop = PL_oldbufptr;
3903 PL_last_lop_op = OP_ENTERSUB;
3904 /* Is there a prototype? */
3907 char *proto = SvPV((SV*)cv, len);
3910 if (strEQ(proto, "$"))
3912 if (*proto == '&' && *s == '{') {
3913 sv_setpv(PL_subname,"__ANON__");
3917 PL_nextval[PL_nexttoke].opval = yylval.opval;
3923 /* Call it a bare word */
3925 if (PL_hints & HINT_STRICT_SUBS)
3926 yylval.opval->op_private |= OPpCONST_STRICT;
3929 if (ckWARN(WARN_RESERVED)) {
3930 if (lastchar != '-') {
3931 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
3933 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
3940 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
3941 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3942 "Operator or semicolon missing before %c%s",
3943 lastchar, PL_tokenbuf);
3944 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3945 "Ambiguous use of %c resolved as operator %c",
3946 lastchar, lastchar);
3952 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3953 newSVpv(CopFILE(PL_curcop),0));
3957 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3958 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
3961 case KEY___PACKAGE__:
3962 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3964 ? newSVsv(PL_curstname)
3973 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
3974 char *pname = "main";
3975 if (PL_tokenbuf[2] == 'D')
3976 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
3977 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
3980 GvIOp(gv) = newIO();
3981 IoIFP(GvIOp(gv)) = PL_rsfp;
3982 #if defined(HAS_FCNTL) && defined(F_SETFD)
3984 int fd = PerlIO_fileno(PL_rsfp);
3985 fcntl(fd,F_SETFD,fd >= 3);
3988 /* Mark this internal pseudo-handle as clean */
3989 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
3991 IoTYPE(GvIOp(gv)) = '|';
3992 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
3993 IoTYPE(GvIOp(gv)) = '-';
3995 IoTYPE(GvIOp(gv)) = '<';
3996 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
3997 /* if the script was opened in binmode, we need to revert
3998 * it to text mode for compatibility; but only iff it has CRs
3999 * XXX this is a questionable hack at best. */
4000 if (PL_bufend-PL_bufptr > 2
4001 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4004 if (IoTYPE(GvIOp(gv)) == '<') {
4005 loc = PerlIO_tell(PL_rsfp);
4006 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4008 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4009 #if defined(__BORLANDC__)
4010 /* XXX see note in do_binmode() */
4011 ((FILE*)PL_rsfp)->flags |= _F_BIN;
4014 PerlIO_seek(PL_rsfp, loc, 0);
4029 if (PL_expect == XSTATE) {
4036 if (*s == ':' && s[1] == ':') {
4039 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4040 if (!(tmp = keyword(PL_tokenbuf, len)))
4041 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4055 LOP(OP_ACCEPT,XTERM);
4061 LOP(OP_ATAN2,XTERM);
4067 LOP(OP_BINMODE,XTERM);
4070 LOP(OP_BLESS,XTERM);
4079 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4096 if (!PL_cryptseen) {
4097 PL_cryptseen = TRUE;
4101 LOP(OP_CRYPT,XTERM);
4104 if (ckWARN(WARN_CHMOD)) {
4105 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4106 if (*d != '0' && isDIGIT(*d))
4107 Perl_warner(aTHX_ WARN_CHMOD,
4108 "chmod() mode argument is missing initial 0");
4110 LOP(OP_CHMOD,XTERM);
4113 LOP(OP_CHOWN,XTERM);
4116 LOP(OP_CONNECT,XTERM);
4132 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4136 PL_hints |= HINT_BLOCK_SCOPE;
4146 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4147 LOP(OP_DBMOPEN,XTERM);
4153 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4160 yylval.ival = CopLINE(PL_curcop);
4174 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4175 UNIBRACK(OP_ENTEREVAL);
4190 case KEY_endhostent:
4196 case KEY_endservent:
4199 case KEY_endprotoent:
4210 yylval.ival = CopLINE(PL_curcop);
4212 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4214 if ((PL_bufend - p) >= 3 &&
4215 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4217 else if ((PL_bufend - p) >= 4 &&
4218 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4221 if (isIDFIRST_lazy_if(p,UTF)) {
4222 p = scan_ident(p, PL_bufend,
4223 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4227 Perl_croak(aTHX_ "Missing $ on loop variable");
4232 LOP(OP_FORMLINE,XTERM);
4238 LOP(OP_FCNTL,XTERM);
4244 LOP(OP_FLOCK,XTERM);
4253 LOP(OP_GREPSTART, XREF);
4256 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4271 case KEY_getpriority:
4272 LOP(OP_GETPRIORITY,XTERM);
4274 case KEY_getprotobyname:
4277 case KEY_getprotobynumber:
4278 LOP(OP_GPBYNUMBER,XTERM);
4280 case KEY_getprotoent:
4292 case KEY_getpeername:
4293 UNI(OP_GETPEERNAME);
4295 case KEY_gethostbyname:
4298 case KEY_gethostbyaddr:
4299 LOP(OP_GHBYADDR,XTERM);
4301 case KEY_gethostent:
4304 case KEY_getnetbyname:
4307 case KEY_getnetbyaddr:
4308 LOP(OP_GNBYADDR,XTERM);
4313 case KEY_getservbyname:
4314 LOP(OP_GSBYNAME,XTERM);
4316 case KEY_getservbyport:
4317 LOP(OP_GSBYPORT,XTERM);
4319 case KEY_getservent:
4322 case KEY_getsockname:
4323 UNI(OP_GETSOCKNAME);
4325 case KEY_getsockopt:
4326 LOP(OP_GSOCKOPT,XTERM);
4348 yylval.ival = CopLINE(PL_curcop);
4352 LOP(OP_INDEX,XTERM);
4358 LOP(OP_IOCTL,XTERM);
4370 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4402 LOP(OP_LISTEN,XTERM);
4411 s = scan_pat(s,OP_MATCH);
4412 TERM(sublex_start());
4415 LOP(OP_MAPSTART, XREF);
4418 LOP(OP_MKDIR,XTERM);
4421 LOP(OP_MSGCTL,XTERM);
4424 LOP(OP_MSGGET,XTERM);
4427 LOP(OP_MSGRCV,XTERM);
4430 LOP(OP_MSGSND,XTERM);
4436 if (isIDFIRST_lazy_if(s,UTF)) {
4437 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4438 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4440 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4441 if (!PL_in_my_stash) {
4444 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4452 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4459 if (PL_expect != XSTATE)
4460 yyerror("\"no\" not allowed in expression");
4461 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4462 s = force_version(s);
4467 if (*s == '(' || (s = skipspace(s), *s == '('))
4474 if (isIDFIRST_lazy_if(s,UTF)) {
4476 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4478 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4479 Perl_warner(aTHX_ WARN_PRECEDENCE,
4480 "Precedence problem: open %.*s should be open(%.*s)",
4486 yylval.ival = OP_OR;
4496 LOP(OP_OPEN_DIR,XTERM);
4499 checkcomma(s,PL_tokenbuf,"filehandle");
4503 checkcomma(s,PL_tokenbuf,"filehandle");
4522 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4526 LOP(OP_PIPE_OP,XTERM);
4529 s = scan_str(s,FALSE,FALSE);
4531 missingterm((char*)0);
4532 yylval.ival = OP_CONST;
4533 TERM(sublex_start());
4539 s = scan_str(s,FALSE,FALSE);
4541 missingterm((char*)0);
4543 if (SvCUR(PL_lex_stuff)) {
4546 d = SvPV_force(PL_lex_stuff, len);
4548 for (; isSPACE(*d) && len; --len, ++d) ;
4551 if (!warned && ckWARN(WARN_QW)) {
4552 for (; !isSPACE(*d) && len; --len, ++d) {
4554 Perl_warner(aTHX_ WARN_QW,
4555 "Possible attempt to separate words with commas");
4558 else if (*d == '#') {
4559 Perl_warner(aTHX_ WARN_QW,
4560 "Possible attempt to put comments in qw() list");
4566 for (; !isSPACE(*d) && len; --len, ++d) ;
4568 words = append_elem(OP_LIST, words,
4569 newSVOP(OP_CONST, 0, tokeq(newSVpvn(b, d-b))));
4573 PL_nextval[PL_nexttoke].opval = words;
4578 SvREFCNT_dec(PL_lex_stuff);
4579 PL_lex_stuff = Nullsv;
4584 s = scan_str(s,FALSE,FALSE);
4586 missingterm((char*)0);
4587 yylval.ival = OP_STRINGIFY;
4588 if (SvIVX(PL_lex_stuff) == '\'')
4589 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4590 TERM(sublex_start());
4593 s = scan_pat(s,OP_QR);
4594 TERM(sublex_start());
4597 s = scan_str(s,FALSE,FALSE);
4599 missingterm((char*)0);
4600 yylval.ival = OP_BACKTICK;
4602 TERM(sublex_start());
4609 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4610 s = force_version(s);
4613 *PL_tokenbuf = '\0';
4614 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4615 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4616 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4618 yyerror("<> should be quotes");
4626 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4630 LOP(OP_RENAME,XTERM);
4639 LOP(OP_RINDEX,XTERM);
4662 LOP(OP_REVERSE,XTERM);
4673 TERM(sublex_start());
4675 TOKEN(1); /* force error */
4684 LOP(OP_SELECT,XTERM);
4690 LOP(OP_SEMCTL,XTERM);
4693 LOP(OP_SEMGET,XTERM);
4696 LOP(OP_SEMOP,XTERM);
4702 LOP(OP_SETPGRP,XTERM);
4704 case KEY_setpriority:
4705 LOP(OP_SETPRIORITY,XTERM);
4707 case KEY_sethostent:
4713 case KEY_setservent:
4716 case KEY_setprotoent:
4726 LOP(OP_SEEKDIR,XTERM);
4728 case KEY_setsockopt:
4729 LOP(OP_SSOCKOPT,XTERM);
4735 LOP(OP_SHMCTL,XTERM);
4738 LOP(OP_SHMGET,XTERM);
4741 LOP(OP_SHMREAD,XTERM);
4744 LOP(OP_SHMWRITE,XTERM);
4747 LOP(OP_SHUTDOWN,XTERM);
4756 LOP(OP_SOCKET,XTERM);
4758 case KEY_socketpair:
4759 LOP(OP_SOCKPAIR,XTERM);
4762 checkcomma(s,PL_tokenbuf,"subroutine name");
4764 if (*s == ';' || *s == ')') /* probably a close */
4765 Perl_croak(aTHX_ "sort is now a reserved word");
4767 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4771 LOP(OP_SPLIT,XTERM);
4774 LOP(OP_SPRINTF,XTERM);
4777 LOP(OP_SPLICE,XTERM);
4792 LOP(OP_SUBSTR,XTERM);
4798 char tmpbuf[sizeof PL_tokenbuf];
4800 expectation attrful;
4801 bool have_name, have_proto;
4806 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4807 (*s == ':' && s[1] == ':'))
4810 attrful = XATTRBLOCK;
4811 /* remember buffer pos'n for later force_word */
4812 tboffset = s - PL_oldbufptr;
4813 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4814 if (strchr(tmpbuf, ':'))
4815 sv_setpv(PL_subname, tmpbuf);
4817 sv_setsv(PL_subname,PL_curstname);
4818 sv_catpvn(PL_subname,"::",2);
4819 sv_catpvn(PL_subname,tmpbuf,len);
4826 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4827 PL_expect = XTERMBLOCK;
4828 attrful = XATTRTERM;
4829 sv_setpv(PL_subname,"?");
4833 if (key == KEY_format) {
4835 PL_lex_formbrack = PL_lex_brackets + 1;
4837 (void) force_word(PL_oldbufptr + tboffset, WORD,
4842 /* Look for a prototype */
4846 s = scan_str(s,FALSE,FALSE);
4849 SvREFCNT_dec(PL_lex_stuff);
4850 PL_lex_stuff = Nullsv;
4851 Perl_croak(aTHX_ "Prototype not terminated");
4854 d = SvPVX(PL_lex_stuff);
4856 for (p = d; *p; ++p) {
4861 SvCUR(PL_lex_stuff) = tmp;
4869 if (*s == ':' && s[1] != ':')
4870 PL_expect = attrful;
4873 PL_nextval[PL_nexttoke].opval =
4874 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4875 PL_lex_stuff = Nullsv;
4879 sv_setpv(PL_subname,"__ANON__");
4882 (void) force_word(PL_oldbufptr + tboffset, WORD,
4891 LOP(OP_SYSTEM,XREF);
4894 LOP(OP_SYMLINK,XTERM);
4897 LOP(OP_SYSCALL,XTERM);
4900 LOP(OP_SYSOPEN,XTERM);
4903 LOP(OP_SYSSEEK,XTERM);
4906 LOP(OP_SYSREAD,XTERM);
4909 LOP(OP_SYSWRITE,XTERM);
4913 TERM(sublex_start());
4934 LOP(OP_TRUNCATE,XTERM);
4946 yylval.ival = CopLINE(PL_curcop);
4950 yylval.ival = CopLINE(PL_curcop);
4954 LOP(OP_UNLINK,XTERM);
4960 LOP(OP_UNPACK,XTERM);
4963 LOP(OP_UTIME,XTERM);
4966 if (ckWARN(WARN_UMASK)) {
4967 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4968 if (*d != '0' && isDIGIT(*d))
4969 Perl_warner(aTHX_ WARN_UMASK,
4970 "umask: argument is missing initial 0");
4975 LOP(OP_UNSHIFT,XTERM);
4978 if (PL_expect != XSTATE)
4979 yyerror("\"use\" not allowed in expression");
4981 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4982 s = force_version(s);
4983 if (*s == ';' || (s = skipspace(s), *s == ';')) {
4984 PL_nextval[PL_nexttoke].opval = Nullop;
4989 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4990 s = force_version(s);
5002 yylval.ival = CopLINE(PL_curcop);
5006 PL_hints |= HINT_BLOCK_SCOPE;
5013 LOP(OP_WAITPID,XTERM);
5021 static char ctl_l[2];
5023 if (ctl_l[0] == '\0')
5024 ctl_l[0] = toCTRL('L');
5025 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5028 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5033 if (PL_expect == XOPERATOR)
5039 yylval.ival = OP_XOR;
5044 TERM(sublex_start());
5049 #pragma segment Main
5053 Perl_keyword(pTHX_ register char *d, I32 len)
5058 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5059 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5060 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5061 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5062 if (strEQ(d,"__END__")) return KEY___END__;
5066 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5071 if (strEQ(d,"and")) return -KEY_and;
5072 if (strEQ(d,"abs")) return -KEY_abs;
5075 if (strEQ(d,"alarm")) return -KEY_alarm;
5076 if (strEQ(d,"atan2")) return -KEY_atan2;
5079 if (strEQ(d,"accept")) return -KEY_accept;
5084 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5087 if (strEQ(d,"bless")) return -KEY_bless;
5088 if (strEQ(d,"bind")) return -KEY_bind;
5089 if (strEQ(d,"binmode")) return -KEY_binmode;
5092 if (strEQ(d,"CORE")) return -KEY_CORE;
5093 if (strEQ(d,"CHECK")) return KEY_CHECK;
5098 if (strEQ(d,"cmp")) return -KEY_cmp;
5099 if (strEQ(d,"chr")) return -KEY_chr;
5100 if (strEQ(d,"cos")) return -KEY_cos;
5103 if (strEQ(d,"chop")) return KEY_chop;
5106 if (strEQ(d,"close")) return -KEY_close;
5107 if (strEQ(d,"chdir")) return -KEY_chdir;
5108 if (strEQ(d,"chomp")) return KEY_chomp;
5109 if (strEQ(d,"chmod")) return -KEY_chmod;
5110 if (strEQ(d,"chown")) return -KEY_chown;
5111 if (strEQ(d,"crypt")) return -KEY_crypt;
5114 if (strEQ(d,"chroot")) return -KEY_chroot;
5115 if (strEQ(d,"caller")) return -KEY_caller;
5118 if (strEQ(d,"connect")) return -KEY_connect;
5121 if (strEQ(d,"closedir")) return -KEY_closedir;
5122 if (strEQ(d,"continue")) return -KEY_continue;
5127 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5132 if (strEQ(d,"do")) return KEY_do;
5135 if (strEQ(d,"die")) return -KEY_die;
5138 if (strEQ(d,"dump")) return -KEY_dump;
5141 if (strEQ(d,"delete")) return KEY_delete;
5144 if (strEQ(d,"defined")) return KEY_defined;
5145 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5148 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5153 if (strEQ(d,"EQ")) { deprecate(d); return -KEY_eq;}
5154 if (strEQ(d,"END")) return KEY_END;
5159 if (strEQ(d,"eq")) return -KEY_eq;
5162 if (strEQ(d,"eof")) return -KEY_eof;
5163 if (strEQ(d,"exp")) return -KEY_exp;
5166 if (strEQ(d,"else")) return KEY_else;
5167 if (strEQ(d,"exit")) return -KEY_exit;
5168 if (strEQ(d,"eval")) return KEY_eval;
5169 if (strEQ(d,"exec")) return -KEY_exec;
5170 if (strEQ(d,"each")) return KEY_each;
5173 if (strEQ(d,"elsif")) return KEY_elsif;
5176 if (strEQ(d,"exists")) return KEY_exists;
5177 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5180 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5181 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5184 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5187 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5188 if (strEQ(d,"endservent")) return -KEY_endservent;
5191 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5198 if (strEQ(d,"for")) return KEY_for;
5201 if (strEQ(d,"fork")) return -KEY_fork;
5204 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5205 if (strEQ(d,"flock")) return -KEY_flock;
5208 if (strEQ(d,"format")) return KEY_format;
5209 if (strEQ(d,"fileno")) return -KEY_fileno;
5212 if (strEQ(d,"foreach")) return KEY_foreach;
5215 if (strEQ(d,"formline")) return -KEY_formline;
5221 if (strEQ(d,"GT")) { deprecate(d); return -KEY_gt;}
5222 if (strEQ(d,"GE")) { deprecate(d); return -KEY_ge;}
5226 if (strnEQ(d,"get",3)) {
5231 if (strEQ(d,"ppid")) return -KEY_getppid;
5232 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5235 if (strEQ(d,"pwent")) return -KEY_getpwent;
5236 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5237 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5240 if (strEQ(d,"peername")) return -KEY_getpeername;
5241 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5242 if (strEQ(d,"priority")) return -KEY_getpriority;
5245 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5248 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5252 else if (*d == 'h') {
5253 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5254 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5255 if (strEQ(d,"hostent")) return -KEY_gethostent;
5257 else if (*d == 'n') {
5258 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5259 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5260 if (strEQ(d,"netent")) return -KEY_getnetent;
5262 else if (*d == 's') {
5263 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5264 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5265 if (strEQ(d,"servent")) return -KEY_getservent;
5266 if (strEQ(d,"sockname")) return -KEY_getsockname;
5267 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5269 else if (*d == 'g') {
5270 if (strEQ(d,"grent")) return -KEY_getgrent;
5271 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5272 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5274 else if (*d == 'l') {
5275 if (strEQ(d,"login")) return -KEY_getlogin;
5277 else if (strEQ(d,"c")) return -KEY_getc;
5282 if (strEQ(d,"gt")) return -KEY_gt;
5283 if (strEQ(d,"ge")) return -KEY_ge;
5286 if (strEQ(d,"grep")) return KEY_grep;
5287 if (strEQ(d,"goto")) return KEY_goto;
5288 if (strEQ(d,"glob")) return KEY_glob;
5291 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5296 if (strEQ(d,"hex")) return -KEY_hex;
5299 if (strEQ(d,"INIT")) return KEY_INIT;
5304 if (strEQ(d,"if")) return KEY_if;
5307 if (strEQ(d,"int")) return -KEY_int;
5310 if (strEQ(d,"index")) return -KEY_index;
5311 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5316 if (strEQ(d,"join")) return -KEY_join;
5320 if (strEQ(d,"keys")) return KEY_keys;
5321 if (strEQ(d,"kill")) return -KEY_kill;
5326 if (strEQ(d,"LT")) { deprecate(d); return -KEY_lt;}
5327 if (strEQ(d,"LE")) { deprecate(d); return -KEY_le;}
5333 if (strEQ(d,"lt")) return -KEY_lt;
5334 if (strEQ(d,"le")) return -KEY_le;
5335 if (strEQ(d,"lc")) return -KEY_lc;
5338 if (strEQ(d,"log")) return -KEY_log;
5341 if (strEQ(d,"last")) return KEY_last;
5342 if (strEQ(d,"link")) return -KEY_link;
5343 if (strEQ(d,"lock")) return -KEY_lock;
5346 if (strEQ(d,"local")) return KEY_local;
5347 if (strEQ(d,"lstat")) return -KEY_lstat;
5350 if (strEQ(d,"length")) return -KEY_length;
5351 if (strEQ(d,"listen")) return -KEY_listen;
5354 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5357 if (strEQ(d,"localtime")) return -KEY_localtime;
5363 case 1: return KEY_m;
5365 if (strEQ(d,"my")) return KEY_my;
5368 if (strEQ(d,"map")) return KEY_map;
5371 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5374 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5375 if (strEQ(d,"msgget")) return -KEY_msgget;
5376 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5377 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5382 if (strEQ(d,"NE")) { deprecate(d); return -KEY_ne;}
5385 if (strEQ(d,"next")) return KEY_next;
5386 if (strEQ(d,"ne")) return -KEY_ne;
5387 if (strEQ(d,"not")) return -KEY_not;
5388 if (strEQ(d,"no")) return KEY_no;
5393 if (strEQ(d,"or")) return -KEY_or;
5396 if (strEQ(d,"ord")) return -KEY_ord;
5397 if (strEQ(d,"oct")) return -KEY_oct;
5398 if (strEQ(d,"our")) return KEY_our;
5401 if (strEQ(d,"open")) return -KEY_open;
5404 if (strEQ(d,"opendir")) return -KEY_opendir;
5411 if (strEQ(d,"pop")) return KEY_pop;
5412 if (strEQ(d,"pos")) return KEY_pos;
5415 if (strEQ(d,"push")) return KEY_push;
5416 if (strEQ(d,"pack")) return -KEY_pack;
5417 if (strEQ(d,"pipe")) return -KEY_pipe;
5420 if (strEQ(d,"print")) return KEY_print;
5423 if (strEQ(d,"printf")) return KEY_printf;
5426 if (strEQ(d,"package")) return KEY_package;
5429 if (strEQ(d,"prototype")) return KEY_prototype;
5434 if (strEQ(d,"q")) return KEY_q;
5435 if (strEQ(d,"qr")) return KEY_qr;
5436 if (strEQ(d,"qq")) return KEY_qq;
5437 if (strEQ(d,"qw")) return KEY_qw;
5438 if (strEQ(d,"qx")) return KEY_qx;
5440 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5445 if (strEQ(d,"ref")) return -KEY_ref;
5448 if (strEQ(d,"read")) return -KEY_read;
5449 if (strEQ(d,"rand")) return -KEY_rand;
5450 if (strEQ(d,"recv")) return -KEY_recv;
5451 if (strEQ(d,"redo")) return KEY_redo;
5454 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5455 if (strEQ(d,"reset")) return -KEY_reset;
5458 if (strEQ(d,"return")) return KEY_return;
5459 if (strEQ(d,"rename")) return -KEY_rename;
5460 if (strEQ(d,"rindex")) return -KEY_rindex;
5463 if (strEQ(d,"require")) return -KEY_require;
5464 if (strEQ(d,"reverse")) return -KEY_reverse;
5465 if (strEQ(d,"readdir")) return -KEY_readdir;
5468 if (strEQ(d,"readlink")) return -KEY_readlink;
5469 if (strEQ(d,"readline")) return -KEY_readline;
5470 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5473 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5479 case 0: return KEY_s;
5481 if (strEQ(d,"scalar")) return KEY_scalar;
5486 if (strEQ(d,"seek")) return -KEY_seek;
5487 if (strEQ(d,"send")) return -KEY_send;
5490 if (strEQ(d,"semop")) return -KEY_semop;
5493 if (strEQ(d,"select")) return -KEY_select;
5494 if (strEQ(d,"semctl")) return -KEY_semctl;
5495 if (strEQ(d,"semget")) return -KEY_semget;
5498 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5499 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5502 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5503 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5506 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5509 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5510 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5511 if (strEQ(d,"setservent")) return -KEY_setservent;
5514 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5515 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5522 if (strEQ(d,"shift")) return KEY_shift;
5525 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5526 if (strEQ(d,"shmget")) return -KEY_shmget;
5529 if (strEQ(d,"shmread")) return -KEY_shmread;
5532 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5533 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5538 if (strEQ(d,"sin")) return -KEY_sin;
5541 if (strEQ(d,"sleep")) return -KEY_sleep;
5544 if (strEQ(d,"sort")) return KEY_sort;
5545 if (strEQ(d,"socket")) return -KEY_socket;
5546 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5549 if (strEQ(d,"split")) return KEY_split;
5550 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5551 if (strEQ(d,"splice")) return KEY_splice;
5554 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5557 if (strEQ(d,"srand")) return -KEY_srand;
5560 if (strEQ(d,"stat")) return -KEY_stat;
5561 if (strEQ(d,"study")) return KEY_study;
5564 if (strEQ(d,"substr")) return -KEY_substr;
5565 if (strEQ(d,"sub")) return KEY_sub;
5570 if (strEQ(d,"system")) return -KEY_system;
5573 if (strEQ(d,"symlink")) return -KEY_symlink;
5574 if (strEQ(d,"syscall")) return -KEY_syscall;
5575 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5576 if (strEQ(d,"sysread")) return -KEY_sysread;
5577 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5580 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5589 if (strEQ(d,"tr")) return KEY_tr;
5592 if (strEQ(d,"tie")) return KEY_tie;
5595 if (strEQ(d,"tell")) return -KEY_tell;
5596 if (strEQ(d,"tied")) return KEY_tied;
5597 if (strEQ(d,"time")) return -KEY_time;
5600 if (strEQ(d,"times")) return -KEY_times;
5603 if (strEQ(d,"telldir")) return -KEY_telldir;
5606 if (strEQ(d,"truncate")) return -KEY_truncate;
5613 if (strEQ(d,"uc")) return -KEY_uc;
5616 if (strEQ(d,"use")) return KEY_use;
5619 if (strEQ(d,"undef")) return KEY_undef;
5620 if (strEQ(d,"until")) return KEY_until;
5621 if (strEQ(d,"untie")) return KEY_untie;
5622 if (strEQ(d,"utime")) return -KEY_utime;
5623 if (strEQ(d,"umask")) return -KEY_umask;
5626 if (strEQ(d,"unless")) return KEY_unless;
5627 if (strEQ(d,"unpack")) return -KEY_unpack;
5628 if (strEQ(d,"unlink")) return -KEY_unlink;
5631 if (strEQ(d,"unshift")) return KEY_unshift;
5632 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5637 if (strEQ(d,"values")) return -KEY_values;
5638 if (strEQ(d,"vec")) return -KEY_vec;
5643 if (strEQ(d,"warn")) return -KEY_warn;
5644 if (strEQ(d,"wait")) return -KEY_wait;
5647 if (strEQ(d,"while")) return KEY_while;
5648 if (strEQ(d,"write")) return -KEY_write;
5651 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5654 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5659 if (len == 1) return -KEY_x;
5660 if (strEQ(d,"xor")) return -KEY_xor;
5663 if (len == 1) return KEY_y;
5672 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5676 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5677 dTHR; /* only for ckWARN */
5678 if (ckWARN(WARN_SYNTAX)) {
5680 for (w = s+2; *w && level; w++) {
5687 for (; *w && isSPACE(*w); w++) ;
5688 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5689 Perl_warner(aTHX_ WARN_SYNTAX,
5690 "%s (...) interpreted as function",name);
5693 while (s < PL_bufend && isSPACE(*s))
5697 while (s < PL_bufend && isSPACE(*s))
5699 if (isIDFIRST_lazy_if(s,UTF)) {
5701 while (isALNUM_lazy_if(s,UTF))
5703 while (s < PL_bufend && isSPACE(*s))
5708 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5712 Perl_croak(aTHX_ "No comma allowed after %s", what);
5717 /* Either returns sv, or mortalizes sv and returns a new SV*.
5718 Best used as sv=new_constant(..., sv, ...).
5719 If s, pv are NULL, calls subroutine with one argument,
5720 and type is used with error messages only. */
5723 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5727 HV *table = GvHV(PL_hintgv); /* ^H */
5731 const char *why1, *why2, *why3;
5733 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
5736 why1 = "%^H is not consistent";
5737 why2 = strEQ(key,"charnames")
5738 ? " (missing \"use charnames ...\"?)"
5742 msg = Perl_newSVpvf(aTHX_ "constant(%s): %s%s%s",
5743 (type ? type: "undef"), why1, why2, why3);
5744 yyerror(SvPVX(msg));
5748 cvp = hv_fetch(table, key, strlen(key), FALSE);
5749 if (!cvp || !SvOK(*cvp)) {
5752 why3 = "} is not defined";
5755 sv_2mortal(sv); /* Parent created it permanently */
5758 pv = sv_2mortal(newSVpvn(s, len));
5760 typesv = sv_2mortal(newSVpv(type, 0));
5762 typesv = &PL_sv_undef;
5764 PUSHSTACKi(PERLSI_OVERLOAD);
5777 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5781 /* Check the eval first */
5782 if (!PL_in_eval && SvTRUE(ERRSV)) {
5784 sv_catpv(ERRSV, "Propagated");
5785 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5787 res = SvREFCNT_inc(sv);
5791 (void)SvREFCNT_inc(res);
5800 why1 = "Call to &{$^H{";
5802 why3 = "}} did not return a defined value";
5811 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5813 register char *d = dest;
5814 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5817 Perl_croak(aTHX_ ident_too_long);
5818 if (isALNUM(*s)) /* UTF handled below */
5820 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
5825 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5829 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5830 char *t = s + UTF8SKIP(s);
5831 while (*t & 0x80 && is_utf8_mark((U8*)t))
5833 if (d + (t - s) > e)
5834 Perl_croak(aTHX_ ident_too_long);
5835 Copy(s, d, t - s, char);
5848 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5858 e = d + destlen - 3; /* two-character token, ending NUL */
5860 while (isDIGIT(*s)) {
5862 Perl_croak(aTHX_ ident_too_long);
5869 Perl_croak(aTHX_ ident_too_long);
5870 if (isALNUM(*s)) /* UTF handled below */
5872 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
5877 else if (*s == ':' && s[1] == ':') {
5881 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5882 char *t = s + UTF8SKIP(s);
5883 while (*t & 0x80 && is_utf8_mark((U8*)t))
5885 if (d + (t - s) > e)
5886 Perl_croak(aTHX_ ident_too_long);
5887 Copy(s, d, t - s, char);
5898 if (PL_lex_state != LEX_NORMAL)
5899 PL_lex_state = LEX_INTERPENDMAYBE;
5902 if (*s == '$' && s[1] &&
5903 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
5916 if (*d == '^' && *s && isCONTROLVAR(*s)) {
5921 if (isSPACE(s[-1])) {
5924 if (!SPACE_OR_TAB(ch)) {
5930 if (isIDFIRST_lazy_if(d,UTF)) {
5934 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
5936 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
5939 Copy(s, d, e - s, char);
5944 while ((isALNUM(*s) || *s == ':') && d < e)
5947 Perl_croak(aTHX_ ident_too_long);
5950 while (s < send && SPACE_OR_TAB(*s)) s++;
5951 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
5952 dTHR; /* only for ckWARN */
5953 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
5954 const char *brack = *s == '[' ? "[...]" : "{...}";
5955 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5956 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
5957 funny, dest, brack, funny, dest, brack);
5960 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
5964 /* Handle extended ${^Foo} variables
5965 * 1999-02-27 mjd-perl-patch@plover.com */
5966 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
5970 while (isALNUM(*s) && d < e) {
5974 Perl_croak(aTHX_ ident_too_long);
5979 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
5980 PL_lex_state = LEX_INTERPEND;
5983 if (PL_lex_state == LEX_NORMAL) {
5984 dTHR; /* only for ckWARN */
5985 if (ckWARN(WARN_AMBIGUOUS) &&
5986 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
5988 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5989 "Ambiguous use of %c{%s} resolved to %c%s",
5990 funny, dest, funny, dest);
5995 s = bracket; /* let the parser handle it */
5999 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
6000 PL_lex_state = LEX_INTERPEND;
6005 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
6010 *pmfl |= PMf_GLOBAL;
6012 *pmfl |= PMf_CONTINUE;
6016 *pmfl |= PMf_MULTILINE;
6018 *pmfl |= PMf_SINGLELINE;
6020 *pmfl |= PMf_EXTENDED;
6024 S_scan_pat(pTHX_ char *start, I32 type)
6029 s = scan_str(start,FALSE,FALSE);
6032 SvREFCNT_dec(PL_lex_stuff);
6033 PL_lex_stuff = Nullsv;
6034 Perl_croak(aTHX_ "Search pattern not terminated");
6037 pm = (PMOP*)newPMOP(type, 0);
6038 if (PL_multi_open == '?')
6039 pm->op_pmflags |= PMf_ONCE;
6041 while (*s && strchr("iomsx", *s))
6042 pmflag(&pm->op_pmflags,*s++);
6045 while (*s && strchr("iogcmsx", *s))
6046 pmflag(&pm->op_pmflags,*s++);
6048 pm->op_pmpermflags = pm->op_pmflags;
6050 PL_lex_op = (OP*)pm;
6051 yylval.ival = OP_MATCH;
6056 S_scan_subst(pTHX_ char *start)
6063 yylval.ival = OP_NULL;
6065 s = scan_str(start,FALSE,FALSE);
6069 SvREFCNT_dec(PL_lex_stuff);
6070 PL_lex_stuff = Nullsv;
6071 Perl_croak(aTHX_ "Substitution pattern not terminated");
6074 if (s[-1] == PL_multi_open)
6077 first_start = PL_multi_start;
6078 s = scan_str(s,FALSE,FALSE);
6081 SvREFCNT_dec(PL_lex_stuff);
6082 PL_lex_stuff = Nullsv;
6084 SvREFCNT_dec(PL_lex_repl);
6085 PL_lex_repl = Nullsv;
6086 Perl_croak(aTHX_ "Substitution replacement not terminated");
6088 PL_multi_start = first_start; /* so whole substitution is taken together */
6090 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6096 else if (strchr("iogcmsx", *s))
6097 pmflag(&pm->op_pmflags,*s++);
6104 PL_sublex_info.super_bufptr = s;
6105 PL_sublex_info.super_bufend = PL_bufend;
6107 pm->op_pmflags |= PMf_EVAL;
6108 repl = newSVpvn("",0);
6110 sv_catpv(repl, es ? "eval " : "do ");
6111 sv_catpvn(repl, "{ ", 2);
6112 sv_catsv(repl, PL_lex_repl);
6113 sv_catpvn(repl, " };", 2);
6115 SvREFCNT_dec(PL_lex_repl);
6119 pm->op_pmpermflags = pm->op_pmflags;
6120 PL_lex_op = (OP*)pm;
6121 yylval.ival = OP_SUBST;
6126 S_scan_trans(pTHX_ char *start)
6137 yylval.ival = OP_NULL;
6139 s = scan_str(start,FALSE,FALSE);
6142 SvREFCNT_dec(PL_lex_stuff);
6143 PL_lex_stuff = Nullsv;
6144 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6146 if (s[-1] == PL_multi_open)
6149 s = scan_str(s,FALSE,FALSE);
6152 SvREFCNT_dec(PL_lex_stuff);
6153 PL_lex_stuff = Nullsv;
6155 SvREFCNT_dec(PL_lex_repl);
6156 PL_lex_repl = Nullsv;
6157 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6160 New(803,tbl,256,short);
6161 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6163 complement = del = squash = 0;
6164 while (strchr("cds", *s)) {
6166 complement = OPpTRANS_COMPLEMENT;
6168 del = OPpTRANS_DELETE;
6170 squash = OPpTRANS_SQUASH;
6173 o->op_private = del|squash|complement;
6176 yylval.ival = OP_TRANS;
6181 S_scan_heredoc(pTHX_ register char *s)
6185 I32 op_type = OP_SCALAR;
6192 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6196 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6199 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6200 if (*peek && strchr("`'\"",*peek)) {
6203 s = delimcpy(d, e, s, PL_bufend, term, &len);
6213 if (!isALNUM_lazy_if(s,UTF))
6214 deprecate("bare << to mean <<\"\"");
6215 for (; isALNUM_lazy_if(s,UTF); s++) {
6220 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6221 Perl_croak(aTHX_ "Delimiter for here document is too long");
6224 len = d - PL_tokenbuf;
6225 #ifndef PERL_STRICT_CR
6226 d = strchr(s, '\r');
6230 while (s < PL_bufend) {
6236 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6245 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6250 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6251 herewas = newSVpvn(s,PL_bufend-s);
6253 s--, herewas = newSVpvn(s,d-s);
6254 s += SvCUR(herewas);
6256 tmpstr = NEWSV(87,79);
6257 sv_upgrade(tmpstr, SVt_PVIV);
6262 else if (term == '`') {
6263 op_type = OP_BACKTICK;
6264 SvIVX(tmpstr) = '\\';
6268 PL_multi_start = CopLINE(PL_curcop);
6269 PL_multi_open = PL_multi_close = '<';
6270 term = *PL_tokenbuf;
6271 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6272 char *bufptr = PL_sublex_info.super_bufptr;
6273 char *bufend = PL_sublex_info.super_bufend;
6274 char *olds = s - SvCUR(herewas);
6275 s = strchr(bufptr, '\n');
6279 while (s < bufend &&
6280 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6282 CopLINE_inc(PL_curcop);
6285 CopLINE_set(PL_curcop, PL_multi_start);
6286 missingterm(PL_tokenbuf);
6288 sv_setpvn(herewas,bufptr,d-bufptr+1);
6289 sv_setpvn(tmpstr,d+1,s-d);
6291 sv_catpvn(herewas,s,bufend-s);
6292 (void)strcpy(bufptr,SvPVX(herewas));
6299 while (s < PL_bufend &&
6300 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6302 CopLINE_inc(PL_curcop);
6304 if (s >= PL_bufend) {
6305 CopLINE_set(PL_curcop, PL_multi_start);
6306 missingterm(PL_tokenbuf);
6308 sv_setpvn(tmpstr,d+1,s-d);
6310 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6312 sv_catpvn(herewas,s,PL_bufend-s);
6313 sv_setsv(PL_linestr,herewas);
6314 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6315 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6318 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6319 while (s >= PL_bufend) { /* multiple line string? */
6321 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6322 CopLINE_set(PL_curcop, PL_multi_start);
6323 missingterm(PL_tokenbuf);
6325 CopLINE_inc(PL_curcop);
6326 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6327 #ifndef PERL_STRICT_CR
6328 if (PL_bufend - PL_linestart >= 2) {
6329 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6330 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6332 PL_bufend[-2] = '\n';
6334 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6336 else if (PL_bufend[-1] == '\r')
6337 PL_bufend[-1] = '\n';
6339 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6340 PL_bufend[-1] = '\n';
6342 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6343 SV *sv = NEWSV(88,0);
6345 sv_upgrade(sv, SVt_PVMG);
6346 sv_setsv(sv,PL_linestr);
6347 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6349 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6352 sv_catsv(PL_linestr,herewas);
6353 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6357 sv_catsv(tmpstr,PL_linestr);
6362 PL_multi_end = CopLINE(PL_curcop);
6363 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6364 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6365 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6367 SvREFCNT_dec(herewas);
6368 PL_lex_stuff = tmpstr;
6369 yylval.ival = op_type;
6374 takes: current position in input buffer
6375 returns: new position in input buffer
6376 side-effects: yylval and lex_op are set.
6381 <FH> read from filehandle
6382 <pkg::FH> read from package qualified filehandle
6383 <pkg'FH> read from package qualified filehandle
6384 <$fh> read from filehandle in $fh
6390 S_scan_inputsymbol(pTHX_ char *start)
6392 register char *s = start; /* current position in buffer */
6398 d = PL_tokenbuf; /* start of temp holding space */
6399 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6400 end = strchr(s, '\n');
6403 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6405 /* die if we didn't have space for the contents of the <>,
6406 or if it didn't end, or if we see a newline
6409 if (len >= sizeof PL_tokenbuf)
6410 Perl_croak(aTHX_ "Excessively long <> operator");
6412 Perl_croak(aTHX_ "Unterminated <> operator");
6417 Remember, only scalar variables are interpreted as filehandles by
6418 this code. Anything more complex (e.g., <$fh{$num}>) will be
6419 treated as a glob() call.
6420 This code makes use of the fact that except for the $ at the front,
6421 a scalar variable and a filehandle look the same.
6423 if (*d == '$' && d[1]) d++;
6425 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6426 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6429 /* If we've tried to read what we allow filehandles to look like, and
6430 there's still text left, then it must be a glob() and not a getline.
6431 Use scan_str to pull out the stuff between the <> and treat it
6432 as nothing more than a string.
6435 if (d - PL_tokenbuf != len) {
6436 yylval.ival = OP_GLOB;
6438 s = scan_str(start,FALSE,FALSE);
6440 Perl_croak(aTHX_ "Glob not terminated");
6444 /* we're in a filehandle read situation */
6447 /* turn <> into <ARGV> */
6449 (void)strcpy(d,"ARGV");
6451 /* if <$fh>, create the ops to turn the variable into a
6457 /* try to find it in the pad for this block, otherwise find
6458 add symbol table ops
6460 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6461 OP *o = newOP(OP_PADSV, 0);
6463 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6466 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6467 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6468 newUNOP(OP_RV2SV, 0,
6469 newGVOP(OP_GV, 0, gv)));
6471 PL_lex_op->op_flags |= OPf_SPECIAL;
6472 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6473 yylval.ival = OP_NULL;
6476 /* If it's none of the above, it must be a literal filehandle
6477 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6479 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6480 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6481 yylval.ival = OP_NULL;
6490 takes: start position in buffer
6491 keep_quoted preserve \ on the embedded delimiter(s)
6492 keep_delims preserve the delimiters around the string
6493 returns: position to continue reading from buffer
6494 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6495 updates the read buffer.
6497 This subroutine pulls a string out of the input. It is called for:
6498 q single quotes q(literal text)
6499 ' single quotes 'literal text'
6500 qq double quotes qq(interpolate $here please)
6501 " double quotes "interpolate $here please"
6502 qx backticks qx(/bin/ls -l)
6503 ` backticks `/bin/ls -l`
6504 qw quote words @EXPORT_OK = qw( func() $spam )
6505 m// regexp match m/this/
6506 s/// regexp substitute s/this/that/
6507 tr/// string transliterate tr/this/that/
6508 y/// string transliterate y/this/that/
6509 ($*@) sub prototypes sub foo ($)
6510 (stuff) sub attr parameters sub foo : attr(stuff)
6511 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6513 In most of these cases (all but <>, patterns and transliterate)
6514 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6515 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6516 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6519 It skips whitespace before the string starts, and treats the first
6520 character as the delimiter. If the delimiter is one of ([{< then
6521 the corresponding "close" character )]}> is used as the closing
6522 delimiter. It allows quoting of delimiters, and if the string has
6523 balanced delimiters ([{<>}]) it allows nesting.
6525 The lexer always reads these strings into lex_stuff, except in the
6526 case of the operators which take *two* arguments (s/// and tr///)
6527 when it checks to see if lex_stuff is full (presumably with the 1st
6528 arg to s or tr) and if so puts the string into lex_repl.
6533 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6536 SV *sv; /* scalar value: string */
6537 char *tmps; /* temp string, used for delimiter matching */
6538 register char *s = start; /* current position in the buffer */
6539 register char term; /* terminating character */
6540 register char *to; /* current position in the sv's data */
6541 I32 brackets = 1; /* bracket nesting level */
6542 bool has_utf = FALSE; /* is there any utf8 content? */
6544 /* skip space before the delimiter */
6548 /* mark where we are, in case we need to report errors */
6551 /* after skipping whitespace, the next character is the terminator */
6553 if ((term & 0x80) && UTF)
6556 /* mark where we are */
6557 PL_multi_start = CopLINE(PL_curcop);
6558 PL_multi_open = term;
6560 /* find corresponding closing delimiter */
6561 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6563 PL_multi_close = term;
6565 /* create a new SV to hold the contents. 87 is leak category, I'm
6566 assuming. 79 is the SV's initial length. What a random number. */
6568 sv_upgrade(sv, SVt_PVIV);
6570 (void)SvPOK_only(sv); /* validate pointer */
6572 /* move past delimiter and try to read a complete string */
6574 sv_catpvn(sv, s, 1);
6577 /* extend sv if need be */
6578 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6579 /* set 'to' to the next character in the sv's string */
6580 to = SvPVX(sv)+SvCUR(sv);
6582 /* if open delimiter is the close delimiter read unbridle */
6583 if (PL_multi_open == PL_multi_close) {
6584 for (; s < PL_bufend; s++,to++) {
6585 /* embedded newlines increment the current line number */
6586 if (*s == '\n' && !PL_rsfp)
6587 CopLINE_inc(PL_curcop);
6588 /* handle quoted delimiters */
6589 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6590 if (!keep_quoted && s[1] == term)
6592 /* any other quotes are simply copied straight through */
6596 /* terminate when run out of buffer (the for() condition), or
6597 have found the terminator */
6598 else if (*s == term)
6600 else if (!has_utf && (*s & 0x80) && UTF)
6606 /* if the terminator isn't the same as the start character (e.g.,
6607 matched brackets), we have to allow more in the quoting, and
6608 be prepared for nested brackets.
6611 /* read until we run out of string, or we find the terminator */
6612 for (; s < PL_bufend; s++,to++) {
6613 /* embedded newlines increment the line count */
6614 if (*s == '\n' && !PL_rsfp)
6615 CopLINE_inc(PL_curcop);
6616 /* backslashes can escape the open or closing characters */
6617 if (*s == '\\' && s+1 < PL_bufend) {
6619 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6624 /* allow nested opens and closes */
6625 else if (*s == PL_multi_close && --brackets <= 0)
6627 else if (*s == PL_multi_open)
6629 else if (!has_utf && (*s & 0x80) && UTF)
6634 /* terminate the copied string and update the sv's end-of-string */
6636 SvCUR_set(sv, to - SvPVX(sv));
6639 * this next chunk reads more into the buffer if we're not done yet
6643 break; /* handle case where we are done yet :-) */
6645 #ifndef PERL_STRICT_CR
6646 if (to - SvPVX(sv) >= 2) {
6647 if ((to[-2] == '\r' && to[-1] == '\n') ||
6648 (to[-2] == '\n' && to[-1] == '\r'))
6652 SvCUR_set(sv, to - SvPVX(sv));
6654 else if (to[-1] == '\r')
6657 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6661 /* if we're out of file, or a read fails, bail and reset the current
6662 line marker so we can report where the unterminated string began
6665 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6667 CopLINE_set(PL_curcop, PL_multi_start);
6670 /* we read a line, so increment our line counter */
6671 CopLINE_inc(PL_curcop);
6673 /* update debugger info */
6674 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6675 SV *sv = NEWSV(88,0);
6677 sv_upgrade(sv, SVt_PVMG);
6678 sv_setsv(sv,PL_linestr);
6679 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6682 /* having changed the buffer, we must update PL_bufend */
6683 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6686 /* at this point, we have successfully read the delimited string */
6689 sv_catpvn(sv, s, 1);
6692 PL_multi_end = CopLINE(PL_curcop);
6695 /* if we allocated too much space, give some back */
6696 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6697 SvLEN_set(sv, SvCUR(sv) + 1);
6698 Renew(SvPVX(sv), SvLEN(sv), char);
6701 /* decide whether this is the first or second quoted string we've read
6714 takes: pointer to position in buffer
6715 returns: pointer to new position in buffer
6716 side-effects: builds ops for the constant in yylval.op
6718 Read a number in any of the formats that Perl accepts:
6720 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6721 [\d_]+(\.[\d_]*)?[Ee](\d+)
6723 Underbars (_) are allowed in decimal numbers. If -w is on,
6724 underbars before a decimal point must be at three digit intervals.
6726 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6729 If it reads a number without a decimal point or an exponent, it will
6730 try converting the number to an integer and see if it can do so
6731 without loss of precision.
6735 Perl_scan_num(pTHX_ char *start)
6737 register char *s = start; /* current position in buffer */
6738 register char *d; /* destination in temp buffer */
6739 register char *e; /* end of temp buffer */
6740 NV nv; /* number read, as a double */
6741 SV *sv = Nullsv; /* place to put the converted number */
6742 bool floatit; /* boolean: int or float? */
6743 char *lastub = 0; /* position of last underbar */
6744 static char number_too_long[] = "Number too long";
6746 /* We use the first character to decide what type of number this is */
6750 Perl_croak(aTHX_ "panic: scan_num");
6752 /* if it starts with a 0, it could be an octal number, a decimal in
6753 0.13 disguise, or a hexadecimal number, or a binary number. */
6757 u holds the "number so far"
6758 shift the power of 2 of the base
6759 (hex == 4, octal == 3, binary == 1)
6760 overflowed was the number more than we can hold?
6762 Shift is used when we add a digit. It also serves as an "are
6763 we in octal/hex/binary?" indicator to disallow hex characters
6770 bool overflowed = FALSE;
6771 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6772 static char* bases[5] = { "", "binary", "", "octal",
6774 static char* Bases[5] = { "", "Binary", "", "Octal",
6776 static char *maxima[5] = { "",
6777 "0b11111111111111111111111111111111",
6781 char *base, *Base, *max;
6787 } else if (s[1] == 'b') {
6791 /* check for a decimal in disguise */
6792 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6794 /* so it must be octal */
6798 base = bases[shift];
6799 Base = Bases[shift];
6800 max = maxima[shift];
6802 /* read the rest of the number */
6804 /* x is used in the overflow test,
6805 b is the digit we're adding on. */
6810 /* if we don't mention it, we're done */
6819 /* 8 and 9 are not octal */
6822 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6826 case '2': case '3': case '4':
6827 case '5': case '6': case '7':
6829 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6833 b = *s++ & 15; /* ASCII digit -> value of digit */
6837 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6838 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6839 /* make sure they said 0x */
6844 /* Prepare to put the digit we have onto the end
6845 of the number so far. We check for overflows.
6850 x = u << shift; /* make room for the digit */
6852 if ((x >> shift) != u
6853 && !(PL_hints & HINT_NEW_BINARY)) {
6857 if (ckWARN_d(WARN_OVERFLOW))
6858 Perl_warner(aTHX_ WARN_OVERFLOW,
6859 "Integer overflow in %s number",
6862 u = x | b; /* add the digit to the end */
6865 n *= nvshift[shift];
6866 /* If an NV has not enough bits in its
6867 * mantissa to represent an UV this summing of
6868 * small low-order numbers is a waste of time
6869 * (because the NV cannot preserve the
6870 * low-order bits anyway): we could just
6871 * remember when did we overflow and in the
6872 * end just multiply n by the right
6880 /* if we get here, we had success: make a scalar value from
6887 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6888 Perl_warner(aTHX_ WARN_PORTABLE,
6889 "%s number > %s non-portable",
6896 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6897 Perl_warner(aTHX_ WARN_PORTABLE,
6898 "%s number > %s non-portable",
6903 if (PL_hints & HINT_NEW_BINARY)
6904 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6909 handle decimal numbers.
6910 we're also sent here when we read a 0 as the first digit
6912 case '1': case '2': case '3': case '4': case '5':
6913 case '6': case '7': case '8': case '9': case '.':
6916 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
6919 /* read next group of digits and _ and copy into d */
6920 while (isDIGIT(*s) || *s == '_') {
6921 /* skip underscores, checking for misplaced ones
6925 dTHR; /* only for ckWARN */
6926 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
6927 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6931 /* check for end of fixed-length buffer */
6933 Perl_croak(aTHX_ number_too_long);
6934 /* if we're ok, copy the character */
6939 /* final misplaced underbar check */
6940 if (lastub && s - lastub != 3) {
6942 if (ckWARN(WARN_SYNTAX))
6943 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6946 /* read a decimal portion if there is one. avoid
6947 3..5 being interpreted as the number 3. followed
6950 if (*s == '.' && s[1] != '.') {
6954 /* copy, ignoring underbars, until we run out of
6955 digits. Note: no misplaced underbar checks!
6957 for (; isDIGIT(*s) || *s == '_'; s++) {
6958 /* fixed length buffer check */
6960 Perl_croak(aTHX_ number_too_long);
6964 if (*s == '.' && isDIGIT(s[1])) {
6965 /* oops, it's really a v-string, but without the "v" */
6971 /* read exponent part, if present */
6972 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
6976 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
6977 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
6979 /* allow positive or negative exponent */
6980 if (*s == '+' || *s == '-')
6983 /* read digits of exponent (no underbars :-) */
6984 while (isDIGIT(*s)) {
6986 Perl_croak(aTHX_ number_too_long);
6991 /* terminate the string */
6994 /* make an sv from the string */
6997 #if defined(Strtol) && defined(Strtoul)
7000 strtol/strtoll sets errno to ERANGE if the number is too big
7001 for an integer. We try to do an integer conversion first
7002 if no characters indicating "float" have been found.
7009 if (*PL_tokenbuf == '-')
7010 iv = Strtol(PL_tokenbuf, (char**)NULL, 10);
7012 uv = Strtoul(PL_tokenbuf, (char**)NULL, 10);
7014 floatit = TRUE; /* Probably just too large. */
7015 else if (*PL_tokenbuf == '-')
7017 else if (uv <= IV_MAX)
7018 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7023 nv = Atof(PL_tokenbuf);
7028 No working strtou?ll?.
7030 Unfortunately atol() doesn't do range checks (returning
7031 LONG_MIN/LONG_MAX, and setting errno to ERANGE on overflows)
7032 everywhere [1], so we cannot use use atol() (or atoll()).
7033 If we could, they would be used, as Atol(), very much like
7034 Strtol() and Strtoul() are used above.
7036 [1] XXX Configure test needed to check for atol()
7037 (and atoll()) overflow behaviour XXX
7041 We need to do this the hard way. */
7043 nv = Atof(PL_tokenbuf);
7045 /* See if we can make do with an integer value without loss of
7046 precision. We use U_V to cast to a UV, because some
7047 compilers have issues. Then we try casting it back and see
7048 if it was the same [1]. We only do this if we know we
7049 specifically read an integer. If floatit is true, then we
7050 don't need to do the conversion at all.
7052 [1] Note that this is lossy if our NVs cannot preserve our
7053 UVs. There are metaconfig defines NV_PRESERVES_UV (a boolean)
7054 and NV_PRESERVES_UV_BITS (a number), but in general we really
7055 do hope all such potentially lossy platforms have strtou?ll?
7056 to do a lossless IV/UV conversion.
7058 Maybe could do some tricks with DBL_DIG, LDBL_DIG and
7059 DBL_MANT_DIG and LDBL_MANT_DIG (these are already available
7060 as NV_DIG and NV_MANT_DIG)?
7066 if (!floatit && (NV)uv == nv) {
7068 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7076 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7077 (PL_hints & HINT_NEW_INTEGER) )
7078 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7079 (floatit ? "float" : "integer"),
7083 /* if it starts with a v, it could be a v-string */
7089 while (isDIGIT(*pos) || *pos == '_')
7091 if (!isALPHA(*pos)) {
7093 U8 tmpbuf[UTF8_MAXLEN];
7096 s++; /* get past 'v' */
7099 sv_setpvn(sv, "", 0);
7102 if (*s == '0' && isDIGIT(s[1]))
7103 yyerror("Octal number in vector unsupported");
7106 /* this is atoi() that tolerates underscores */
7109 while (--end >= s) {
7114 rev += (*end - '0') * mult;
7116 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
7117 Perl_warner(aTHX_ WARN_OVERFLOW,
7118 "Integer overflow in decimal number");
7121 tmpend = uv_to_utf8(tmpbuf, rev);
7122 utf8 = utf8 || rev > 127;
7123 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
7124 if (*pos == '.' && isDIGIT(pos[1]))
7130 while (isDIGIT(*pos) || *pos == '_')
7138 sv_utf8_downgrade(sv, TRUE);
7145 /* make the op for the constant and return */
7148 yylval.opval = newSVOP(OP_CONST, 0, sv);
7150 yylval.opval = Nullop;
7156 S_scan_formline(pTHX_ register char *s)
7161 SV *stuff = newSVpvn("",0);
7162 bool needargs = FALSE;
7165 if (*s == '.' || *s == '}') {
7167 #ifdef PERL_STRICT_CR
7168 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7170 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7172 if (*t == '\n' || t == PL_bufend)
7175 if (PL_in_eval && !PL_rsfp) {
7176 eol = strchr(s,'\n');
7181 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7183 for (t = s; t < eol; t++) {
7184 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7186 goto enough; /* ~~ must be first line in formline */
7188 if (*t == '@' || *t == '^')
7191 sv_catpvn(stuff, s, eol-s);
7192 #ifndef PERL_STRICT_CR
7193 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7194 char *end = SvPVX(stuff) + SvCUR(stuff);
7203 s = filter_gets(PL_linestr, PL_rsfp, 0);
7204 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7205 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7208 yyerror("Format not terminated");
7218 PL_lex_state = LEX_NORMAL;
7219 PL_nextval[PL_nexttoke].ival = 0;
7223 PL_lex_state = LEX_FORMLINE;
7224 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7226 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7230 SvREFCNT_dec(stuff);
7231 PL_lex_formbrack = 0;
7242 PL_cshlen = strlen(PL_cshname);
7247 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7250 I32 oldsavestack_ix = PL_savestack_ix;
7251 CV* outsidecv = PL_compcv;
7255 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7257 SAVEI32(PL_subline);
7258 save_item(PL_subname);
7261 SAVESPTR(PL_comppad_name);
7262 SAVESPTR(PL_compcv);
7263 SAVEI32(PL_comppad_name_fill);
7264 SAVEI32(PL_min_intro_pending);
7265 SAVEI32(PL_max_intro_pending);
7266 SAVEI32(PL_pad_reset_pending);
7268 PL_compcv = (CV*)NEWSV(1104,0);
7269 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7270 CvFLAGS(PL_compcv) |= flags;
7272 PL_comppad = newAV();
7273 av_push(PL_comppad, Nullsv);
7274 PL_curpad = AvARRAY(PL_comppad);
7275 PL_comppad_name = newAV();
7276 PL_comppad_name_fill = 0;
7277 PL_min_intro_pending = 0;
7279 PL_subline = CopLINE(PL_curcop);
7281 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7282 PL_curpad[0] = (SV*)newAV();
7283 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7284 #endif /* USE_THREADS */
7286 comppadlist = newAV();
7287 AvREAL_off(comppadlist);
7288 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7289 av_store(comppadlist, 1, (SV*)PL_comppad);
7291 CvPADLIST(PL_compcv) = comppadlist;
7292 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7294 CvOWNER(PL_compcv) = 0;
7295 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7296 MUTEX_INIT(CvMUTEXP(PL_compcv));
7297 #endif /* USE_THREADS */
7299 return oldsavestack_ix;
7303 Perl_yywarn(pTHX_ char *s)
7306 PL_in_eval |= EVAL_WARNONLY;
7308 PL_in_eval &= ~EVAL_WARNONLY;
7313 Perl_yyerror(pTHX_ char *s)
7317 char *context = NULL;
7321 if (!yychar || (yychar == ';' && !PL_rsfp))
7323 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7324 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7325 while (isSPACE(*PL_oldoldbufptr))
7327 context = PL_oldoldbufptr;
7328 contlen = PL_bufptr - PL_oldoldbufptr;
7330 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7331 PL_oldbufptr != PL_bufptr) {
7332 while (isSPACE(*PL_oldbufptr))
7334 context = PL_oldbufptr;
7335 contlen = PL_bufptr - PL_oldbufptr;
7337 else if (yychar > 255)
7338 where = "next token ???";
7339 #ifdef USE_PURE_BISON
7340 /* GNU Bison sets the value -2 */
7341 else if (yychar == -2) {
7343 else if ((yychar & 127) == 127) {
7345 if (PL_lex_state == LEX_NORMAL ||
7346 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7347 where = "at end of line";
7348 else if (PL_lex_inpat)
7349 where = "within pattern";
7351 where = "within string";
7354 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7356 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7357 else if (isPRINT_LC(yychar))
7358 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7360 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7361 where = SvPVX(where_sv);
7363 msg = sv_2mortal(newSVpv(s, 0));
7364 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7365 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7367 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7369 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7370 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7371 Perl_sv_catpvf(aTHX_ msg,
7372 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7373 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7376 if (PL_in_eval & EVAL_WARNONLY)
7377 Perl_warn(aTHX_ "%"SVf, msg);
7380 if (PL_error_count >= 10) {
7381 if (PL_in_eval && SvCUR(ERRSV))
7382 Perl_croak(aTHX_ "%_%s has too many errors.\n",
7383 ERRSV, CopFILE(PL_curcop));
7385 Perl_croak(aTHX_ "%s has too many errors.\n",
7386 CopFILE(PL_curcop));
7389 PL_in_my_stash = Nullhv;
7400 * Restore a source filter.
7404 restore_rsfp(pTHXo_ void *f)
7406 PerlIO *fp = (PerlIO*)f;
7408 if (PL_rsfp == PerlIO_stdin())
7409 PerlIO_clearerr(PL_rsfp);
7410 else if (PL_rsfp && (PL_rsfp != fp))
7411 PerlIO_close(PL_rsfp);
7416 S_swallow_bom(pTHX_ char *s) {
7418 slen = SvCUR(PL_linestr);
7421 if ((s[1] & 255) == 254) {
7422 /* UTF-16 little-endian */
7423 #ifdef PERL_UTF16_FILTER
7427 if (*s == 0 && s[1] == 0) /* UTF-32 little-endian */
7428 Perl_croak(aTHX_ "Unsupported script encoding");
7429 #ifdef PERL_UTF16_FILTER
7430 filter_add(S_utf16rev_textfilter, NULL);
7431 New(898, news, (PL_bufend - s) * 3 / 2 + 1, U8);
7432 PL_bufend = utf16_to_utf8((U16*)s, news, PL_bufend - s);
7435 Perl_croak(aTHX_ "Unsupported script encoding");
7441 if ((s[1] & 255) == 255) { /* UTF-16 big-endian */
7442 #ifdef PERL_UTF16_FILTER
7444 filter_add(S_utf16_textfilter, NULL);
7445 New(898, news, (PL_bufend - s) * 3 / 2 + 1, U8);
7446 PL_bufend = utf16_to_utf8((U16*)s, news, PL_bufend - s);
7449 Perl_croak(aTHX_ "Unsupported script encoding");
7455 if ( slen>2 && (s[1] & 255) == 187 && (s[2] & 255) == 191) {
7460 if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */
7461 s[2] & 255 == 254 && s[3] & 255 == 255)
7462 Perl_croak(aTHX_ "Unsupported script encoding");