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)
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);
993 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
994 PL_lex_stuff = Nullsv;
998 PL_sublex_info.super_state = PL_lex_state;
999 PL_sublex_info.sub_inwhat = op_type;
1000 PL_sublex_info.sub_op = PL_lex_op;
1001 PL_lex_state = LEX_INTERPPUSH;
1005 yylval.opval = PL_lex_op;
1015 * Create a new scope to save the lexing state. The scope will be
1016 * ended in S_sublex_done. Returns a '(', starting the function arguments
1017 * to the uc, lc, etc. found before.
1018 * Sets PL_lex_state to LEX_INTERPCONCAT.
1027 PL_lex_state = PL_sublex_info.super_state;
1028 SAVEI32(PL_lex_dojoin);
1029 SAVEI32(PL_lex_brackets);
1030 SAVEI32(PL_lex_casemods);
1031 SAVEI32(PL_lex_starts);
1032 SAVEI32(PL_lex_state);
1033 SAVEVPTR(PL_lex_inpat);
1034 SAVEI32(PL_lex_inwhat);
1035 SAVECOPLINE(PL_curcop);
1036 SAVEPPTR(PL_bufptr);
1037 SAVEPPTR(PL_oldbufptr);
1038 SAVEPPTR(PL_oldoldbufptr);
1039 SAVEPPTR(PL_linestart);
1040 SAVESPTR(PL_linestr);
1041 SAVEPPTR(PL_lex_brackstack);
1042 SAVEPPTR(PL_lex_casestack);
1044 PL_linestr = PL_lex_stuff;
1045 PL_lex_stuff = Nullsv;
1047 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1048 = SvPVX(PL_linestr);
1049 PL_bufend += SvCUR(PL_linestr);
1050 SAVEFREESV(PL_linestr);
1052 PL_lex_dojoin = FALSE;
1053 PL_lex_brackets = 0;
1054 New(899, PL_lex_brackstack, 120, char);
1055 New(899, PL_lex_casestack, 12, char);
1056 SAVEFREEPV(PL_lex_brackstack);
1057 SAVEFREEPV(PL_lex_casestack);
1058 PL_lex_casemods = 0;
1059 *PL_lex_casestack = '\0';
1061 PL_lex_state = LEX_INTERPCONCAT;
1062 CopLINE_set(PL_curcop, PL_multi_start);
1064 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1065 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1066 PL_lex_inpat = PL_sublex_info.sub_op;
1068 PL_lex_inpat = Nullop;
1075 * Restores lexer state after a S_sublex_push.
1081 if (!PL_lex_starts++) {
1082 PL_expect = XOPERATOR;
1083 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn("",0));
1087 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1088 PL_lex_state = LEX_INTERPCASEMOD;
1092 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1093 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1094 PL_linestr = PL_lex_repl;
1096 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1097 PL_bufend += SvCUR(PL_linestr);
1098 SAVEFREESV(PL_linestr);
1099 PL_lex_dojoin = FALSE;
1100 PL_lex_brackets = 0;
1101 PL_lex_casemods = 0;
1102 *PL_lex_casestack = '\0';
1104 if (SvEVALED(PL_lex_repl)) {
1105 PL_lex_state = LEX_INTERPNORMAL;
1107 /* we don't clear PL_lex_repl here, so that we can check later
1108 whether this is an evalled subst; that means we rely on the
1109 logic to ensure sublex_done() is called again only via the
1110 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1113 PL_lex_state = LEX_INTERPCONCAT;
1114 PL_lex_repl = Nullsv;
1120 PL_bufend = SvPVX(PL_linestr);
1121 PL_bufend += SvCUR(PL_linestr);
1122 PL_expect = XOPERATOR;
1123 PL_sublex_info.sub_inwhat = 0;
1131 Extracts a pattern, double-quoted string, or transliteration. This
1134 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1135 processing a pattern (PL_lex_inpat is true), a transliteration
1136 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1138 Returns a pointer to the character scanned up to. Iff this is
1139 advanced from the start pointer supplied (ie if anything was
1140 successfully parsed), will leave an OP for the substring scanned
1141 in yylval. Caller must intuit reason for not parsing further
1142 by looking at the next characters herself.
1146 double-quoted style: \r and \n
1147 regexp special ones: \D \s
1149 backrefs: \1 (deprecated in substitution replacements)
1150 case and quoting: \U \Q \E
1151 stops on @ and $, but not for $ as tail anchor
1153 In transliterations:
1154 characters are VERY literal, except for - not at the start or end
1155 of the string, which indicates a range. scan_const expands the
1156 range to the full set of intermediate characters.
1158 In double-quoted strings:
1160 double-quoted style: \r and \n
1162 backrefs: \1 (deprecated)
1163 case and quoting: \U \Q \E
1166 scan_const does *not* construct ops to handle interpolated strings.
1167 It stops processing as soon as it finds an embedded $ or @ variable
1168 and leaves it to the caller to work out what's going on.
1170 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1172 $ in pattern could be $foo or could be tail anchor. Assumption:
1173 it's a tail anchor if $ is the last thing in the string, or if it's
1174 followed by one of ")| \n\t"
1176 \1 (backreferences) are turned into $1
1178 The structure of the code is
1179 while (there's a character to process) {
1180 handle transliteration ranges
1181 skip regexp comments
1182 skip # initiated comments in //x patterns
1183 check for embedded @foo
1184 check for embedded scalars
1186 leave intact backslashes from leave (below)
1187 deprecate \1 in strings and sub replacements
1188 handle string-changing backslashes \l \U \Q \E, etc.
1189 switch (what was escaped) {
1190 handle - in a transliteration (becomes a literal -)
1191 handle \132 octal characters
1192 handle 0x15 hex characters
1193 handle \cV (control V)
1194 handle printf backslashes (\f, \r, \n, etc)
1196 } (end if backslash)
1197 } (end while character to read)
1202 S_scan_const(pTHX_ char *start)
1204 register char *send = PL_bufend; /* end of the constant */
1205 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1206 register char *s = start; /* start of the constant */
1207 register char *d = SvPVX(sv); /* destination for copies */
1208 bool dorange = FALSE; /* are we in a translit range? */
1209 bool has_utf = FALSE; /* embedded \x{} */
1213 I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1214 ? (PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1216 I32 thisutf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1217 ? (PL_sublex_info.sub_op->op_private & (PL_lex_repl ?
1218 OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF))
1220 const char *leaveit = /* set of acceptably-backslashed characters */
1222 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1225 while (s < send || dorange) {
1226 /* get transliterations out of the way (they're most literal) */
1227 if (PL_lex_inwhat == OP_TRANS) {
1228 /* expand a range A-Z to the full set of characters. AIE! */
1230 I32 i; /* current expanded character */
1231 I32 min; /* first character in range */
1232 I32 max; /* last character in range */
1234 i = d - SvPVX(sv); /* remember current offset */
1235 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1236 d = SvPVX(sv) + i; /* refresh d after realloc */
1237 d -= 2; /* eat the first char and the - */
1239 min = (U8)*d; /* first char in range */
1240 max = (U8)d[1]; /* last char in range */
1243 if ((isLOWER(min) && isLOWER(max)) ||
1244 (isUPPER(min) && isUPPER(max))) {
1246 for (i = min; i <= max; i++)
1250 for (i = min; i <= max; i++)
1257 for (i = min; i <= max; i++)
1260 /* mark the range as done, and continue */
1265 /* range begins (ignore - as first or last char) */
1266 else if (*s == '-' && s+1 < send && s != start) {
1268 *d++ = (char)0xff; /* use illegal utf8 byte--see pmtrans */
1277 /* if we get here, we're not doing a transliteration */
1279 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1280 except for the last char, which will be done separately. */
1281 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1283 while (s < send && *s != ')')
1286 else if (s[2] == '{' /* This should match regcomp.c */
1287 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1290 char *regparse = s + (s[2] == '{' ? 3 : 4);
1293 while (count && (c = *regparse)) {
1294 if (c == '\\' && regparse[1])
1302 if (*regparse != ')') {
1303 regparse--; /* Leave one char for continuation. */
1304 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1306 while (s < regparse)
1311 /* likewise skip #-initiated comments in //x patterns */
1312 else if (*s == '#' && PL_lex_inpat &&
1313 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1314 while (s+1 < send && *s != '\n')
1318 /* check for embedded arrays (@foo, @:foo, @'foo, @{foo}, @$foo) */
1319 else if (*s == '@' && s[1]
1320 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$", s[1])))
1323 /* check for embedded scalars. only stop if we're sure it's a
1326 else if (*s == '$') {
1327 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1329 if (s + 1 < send && !strchr("()| \n\t", s[1]))
1330 break; /* in regexp, $ might be tail anchor */
1333 /* (now in tr/// code again) */
1335 if (*s & 0x80 && thisutf) {
1336 (void)utf8_to_uv((U8*)s, &len);
1338 /* illegal UTF8, make it valid */
1339 char *old_pvx = SvPVX(sv);
1340 /* need space for one extra char (NOTE: SvCUR() not set here) */
1341 d = SvGROW(sv, SvLEN(sv) + 1) + (d - old_pvx);
1342 d = (char*)uv_to_utf8((U8*)d, (U8)*s++);
1353 if (*s == '\\' && s+1 < send) {
1356 /* some backslashes we leave behind */
1357 if (*leaveit && *s && strchr(leaveit, *s)) {
1363 /* deprecate \1 in strings and substitution replacements */
1364 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1365 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1367 dTHR; /* only for ckWARN */
1368 if (ckWARN(WARN_SYNTAX))
1369 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1374 /* string-change backslash escapes */
1375 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1380 /* if we get here, it's either a quoted -, or a digit */
1383 /* quoted - in transliterations */
1385 if (PL_lex_inwhat == OP_TRANS) {
1393 if (ckWARN(WARN_MISC) && isALNUM(*s))
1394 Perl_warner(aTHX_ WARN_MISC,
1395 "Unrecognized escape \\%c passed through",
1397 /* default action is to copy the quoted character */
1402 /* \132 indicates an octal constant */
1403 case '0': case '1': case '2': case '3':
1404 case '4': case '5': case '6': case '7':
1405 len = 0; /* disallow underscores */
1406 uv = (UV)scan_oct(s, 3, &len);
1408 goto NUM_ESCAPE_INSERT;
1410 /* \x24 indicates a hex constant */
1414 char* e = strchr(s, '}');
1416 yyerror("Missing right brace on \\x{}");
1419 len = 1; /* allow underscores */
1420 uv = (UV)scan_hex(s + 1, e - s - 1, &len);
1424 len = 0; /* disallow underscores */
1425 uv = (UV)scan_hex(s, 2, &len);
1430 /* Insert oct or hex escaped character.
1431 * There will always enough room in sv since such escapes will
1432 * be longer than any utf8 sequence they can end up as
1435 if (!thisutf && !has_utf && uv > 255) {
1436 /* might need to recode whatever we have accumulated so far
1437 * if it contains any hibit chars
1441 for (c = SvPVX(sv); c < d; c++) {
1446 char *old_pvx = SvPVX(sv);
1448 d = SvGROW(sv, SvCUR(sv) + hicount + 1) + (d - old_pvx);
1457 uv_to_utf8((U8*)dst, (U8)*src--);
1467 if (thisutf || uv > 255) {
1468 d = (char*)uv_to_utf8((U8*)d, uv);
1480 /* \N{latin small letter a} is a named character */
1484 char* e = strchr(s, '}');
1490 yyerror("Missing right brace on \\N{}");
1494 res = newSVpvn(s + 1, e - s - 1);
1495 res = new_constant( Nullch, 0, "charnames",
1496 res, Nullsv, "\\N{...}" );
1497 str = SvPV(res,len);
1498 if (!has_utf && SvUTF8(res)) {
1499 char *ostart = SvPVX(sv);
1500 SvCUR_set(sv, d - ostart);
1502 sv_utf8_upgrade(sv);
1503 d = SvPVX(sv) + SvCUR(sv);
1506 if (len > e - s + 4) {
1507 char *odest = SvPVX(sv);
1509 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1510 d = SvPVX(sv) + (d - odest);
1512 Copy(str, d, len, char);
1519 yyerror("Missing braces on \\N{}");
1522 /* \c is a control character */
1537 /* printf-style backslashes, formfeeds, newlines, etc */
1555 *d++ = '\047'; /* CP 1047 */
1558 *d++ = '\057'; /* CP 1047 */
1572 } /* end if (backslash) */
1575 } /* while loop to process each character */
1577 /* terminate the string and set up the sv */
1579 SvCUR_set(sv, d - SvPVX(sv));
1584 /* shrink the sv if we allocated more than we used */
1585 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1586 SvLEN_set(sv, SvCUR(sv) + 1);
1587 Renew(SvPVX(sv), SvLEN(sv), char);
1590 /* return the substring (via yylval) only if we parsed anything */
1591 if (s > PL_bufptr) {
1592 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1593 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1595 ( PL_lex_inwhat == OP_TRANS
1597 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1600 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1607 * Returns TRUE if there's more to the expression (e.g., a subscript),
1610 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1612 * ->[ and ->{ return TRUE
1613 * { and [ outside a pattern are always subscripts, so return TRUE
1614 * if we're outside a pattern and it's not { or [, then return FALSE
1615 * if we're in a pattern and the first char is a {
1616 * {4,5} (any digits around the comma) returns FALSE
1617 * if we're in a pattern and the first char is a [
1619 * [SOMETHING] has a funky algorithm to decide whether it's a
1620 * character class or not. It has to deal with things like
1621 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1622 * anything else returns TRUE
1625 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1628 S_intuit_more(pTHX_ register char *s)
1630 if (PL_lex_brackets)
1632 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1634 if (*s != '{' && *s != '[')
1639 /* In a pattern, so maybe we have {n,m}. */
1656 /* On the other hand, maybe we have a character class */
1659 if (*s == ']' || *s == '^')
1662 /* this is terrifying, and it works */
1663 int weight = 2; /* let's weigh the evidence */
1665 unsigned char un_char = 255, last_un_char;
1666 char *send = strchr(s,']');
1667 char tmpbuf[sizeof PL_tokenbuf * 4];
1669 if (!send) /* has to be an expression */
1672 Zero(seen,256,char);
1675 else if (isDIGIT(*s)) {
1677 if (isDIGIT(s[1]) && s[2] == ']')
1683 for (; s < send; s++) {
1684 last_un_char = un_char;
1685 un_char = (unsigned char)*s;
1690 weight -= seen[un_char] * 10;
1691 if (isALNUM_lazy_if(s+1,UTF)) {
1692 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1693 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1698 else if (*s == '$' && s[1] &&
1699 strchr("[#!%*<>()-=",s[1])) {
1700 if (/*{*/ strchr("])} =",s[2]))
1709 if (strchr("wds]",s[1]))
1711 else if (seen['\''] || seen['"'])
1713 else if (strchr("rnftbxcav",s[1]))
1715 else if (isDIGIT(s[1])) {
1717 while (s[1] && isDIGIT(s[1]))
1727 if (strchr("aA01! ",last_un_char))
1729 if (strchr("zZ79~",s[1]))
1731 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1732 weight -= 5; /* cope with negative subscript */
1735 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1736 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1741 if (keyword(tmpbuf, d - tmpbuf))
1744 if (un_char == last_un_char + 1)
1746 weight -= seen[un_char];
1751 if (weight >= 0) /* probably a character class */
1761 * Does all the checking to disambiguate
1763 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1764 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1766 * First argument is the stuff after the first token, e.g. "bar".
1768 * Not a method if bar is a filehandle.
1769 * Not a method if foo is a subroutine prototyped to take a filehandle.
1770 * Not a method if it's really "Foo $bar"
1771 * Method if it's "foo $bar"
1772 * Not a method if it's really "print foo $bar"
1773 * Method if it's really "foo package::" (interpreted as package->foo)
1774 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1775 * Not a method if bar is a filehandle or package, but is quoted with
1780 S_intuit_method(pTHX_ char *start, GV *gv)
1782 char *s = start + (*start == '$');
1783 char tmpbuf[sizeof PL_tokenbuf];
1791 if ((cv = GvCVu(gv))) {
1792 char *proto = SvPVX(cv);
1802 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1803 /* start is the beginning of the possible filehandle/object,
1804 * and s is the end of it
1805 * tmpbuf is a copy of it
1808 if (*start == '$') {
1809 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1814 return *s == '(' ? FUNCMETH : METHOD;
1816 if (!keyword(tmpbuf, len)) {
1817 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1822 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1823 if (indirgv && GvCVu(indirgv))
1825 /* filehandle or package name makes it a method */
1826 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1828 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1829 return 0; /* no assumptions -- "=>" quotes bearword */
1831 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1832 newSVpvn(tmpbuf,len));
1833 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1837 return *s == '(' ? FUNCMETH : METHOD;
1845 * Return a string of Perl code to load the debugger. If PERL5DB
1846 * is set, it will return the contents of that, otherwise a
1847 * compile-time require of perl5db.pl.
1854 char *pdb = PerlEnv_getenv("PERL5DB");
1858 SETERRNO(0,SS$_NORMAL);
1859 return "BEGIN { require 'perl5db.pl' }";
1865 /* Encoded script support. filter_add() effectively inserts a
1866 * 'pre-processing' function into the current source input stream.
1867 * Note that the filter function only applies to the current source file
1868 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1870 * The datasv parameter (which may be NULL) can be used to pass
1871 * private data to this instance of the filter. The filter function
1872 * can recover the SV using the FILTER_DATA macro and use it to
1873 * store private buffers and state information.
1875 * The supplied datasv parameter is upgraded to a PVIO type
1876 * and the IoDIRP field is used to store the function pointer,
1877 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1878 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1879 * private use must be set using malloc'd pointers.
1883 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1888 if (!PL_rsfp_filters)
1889 PL_rsfp_filters = newAV();
1891 datasv = NEWSV(255,0);
1892 if (!SvUPGRADE(datasv, SVt_PVIO))
1893 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1894 IoDIRP(datasv) = (DIR*)funcp; /* stash funcp into spare field */
1895 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1896 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1897 funcp, SvPV_nolen(datasv)));
1898 av_unshift(PL_rsfp_filters, 1);
1899 av_store(PL_rsfp_filters, 0, datasv) ;
1904 /* Delete most recently added instance of this filter function. */
1906 Perl_filter_del(pTHX_ filter_t funcp)
1909 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1910 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1912 /* if filter is on top of stack (usual case) just pop it off */
1913 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1914 if (IoDIRP(datasv) == (DIR*)funcp) {
1915 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1916 IoDIRP(datasv) = (DIR*)NULL;
1917 sv_free(av_pop(PL_rsfp_filters));
1921 /* we need to search for the correct entry and clear it */
1922 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1926 /* Invoke the n'th filter function for the current rsfp. */
1928 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1931 /* 0 = read one text line */
1936 if (!PL_rsfp_filters)
1938 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1939 /* Provide a default input filter to make life easy. */
1940 /* Note that we append to the line. This is handy. */
1941 DEBUG_P(PerlIO_printf(Perl_debug_log,
1942 "filter_read %d: from rsfp\n", idx));
1946 int old_len = SvCUR(buf_sv) ;
1948 /* ensure buf_sv is large enough */
1949 SvGROW(buf_sv, old_len + maxlen) ;
1950 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1951 if (PerlIO_error(PL_rsfp))
1952 return -1; /* error */
1954 return 0 ; /* end of file */
1956 SvCUR_set(buf_sv, old_len + len) ;
1959 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1960 if (PerlIO_error(PL_rsfp))
1961 return -1; /* error */
1963 return 0 ; /* end of file */
1966 return SvCUR(buf_sv);
1968 /* Skip this filter slot if filter has been deleted */
1969 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1970 DEBUG_P(PerlIO_printf(Perl_debug_log,
1971 "filter_read %d: skipped (filter deleted)\n",
1973 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1975 /* Get function pointer hidden within datasv */
1976 funcp = (filter_t)IoDIRP(datasv);
1977 DEBUG_P(PerlIO_printf(Perl_debug_log,
1978 "filter_read %d: via function %p (%s)\n",
1979 idx, funcp, SvPV_nolen(datasv)));
1980 /* Call function. The function is expected to */
1981 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1982 /* Return: <0:error, =0:eof, >0:not eof */
1983 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1987 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1989 #ifdef PERL_CR_FILTER
1990 if (!PL_rsfp_filters) {
1991 filter_add(S_cr_textfilter,NULL);
1994 if (PL_rsfp_filters) {
1997 SvCUR_set(sv, 0); /* start with empty line */
1998 if (FILTER_READ(0, sv, 0) > 0)
1999 return ( SvPVX(sv) ) ;
2004 return (sv_gets(sv, fp, append));
2009 static char* exp_name[] =
2010 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2011 "ATTRTERM", "TERMBLOCK"
2018 Works out what to call the token just pulled out of the input
2019 stream. The yacc parser takes care of taking the ops we return and
2020 stitching them into a tree.
2026 if read an identifier
2027 if we're in a my declaration
2028 croak if they tried to say my($foo::bar)
2029 build the ops for a my() declaration
2030 if it's an access to a my() variable
2031 are we in a sort block?
2032 croak if my($a); $a <=> $b
2033 build ops for access to a my() variable
2034 if in a dq string, and they've said @foo and we can't find @foo
2036 build ops for a bareword
2037 if we already built the token before, use it.
2041 #pragma segment Perl_yylex
2044 #ifdef USE_PURE_BISON
2045 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
2058 #ifdef USE_PURE_BISON
2059 yylval_pointer = lvalp;
2060 yychar_pointer = lcharp;
2063 /* check if there's an identifier for us to look at */
2064 if (PL_pending_ident) {
2065 /* pit holds the identifier we read and pending_ident is reset */
2066 char pit = PL_pending_ident;
2067 PL_pending_ident = 0;
2069 /* if we're in a my(), we can't allow dynamics here.
2070 $foo'bar has already been turned into $foo::bar, so
2071 just check for colons.
2073 if it's a legal name, the OP is a PADANY.
2076 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
2077 if (strchr(PL_tokenbuf,':'))
2078 yyerror(Perl_form(aTHX_ "No package name allowed for "
2079 "variable %s in \"our\"",
2081 tmp = pad_allocmy(PL_tokenbuf);
2084 if (strchr(PL_tokenbuf,':'))
2085 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
2087 yylval.opval = newOP(OP_PADANY, 0);
2088 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
2094 build the ops for accesses to a my() variable.
2096 Deny my($a) or my($b) in a sort block, *if* $a or $b is
2097 then used in a comparison. This catches most, but not
2098 all cases. For instance, it catches
2099 sort { my($a); $a <=> $b }
2101 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2102 (although why you'd do that is anyone's guess).
2105 if (!strchr(PL_tokenbuf,':')) {
2107 /* Check for single character per-thread SVs */
2108 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2109 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2110 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2112 yylval.opval = newOP(OP_THREADSV, 0);
2113 yylval.opval->op_targ = tmp;
2116 #endif /* USE_THREADS */
2117 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2118 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2119 /* might be an "our" variable" */
2120 if (SvFLAGS(namesv) & SVpad_OUR) {
2121 /* build ops for a bareword */
2122 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2123 sv_catpvn(sym, "::", 2);
2124 sv_catpv(sym, PL_tokenbuf+1);
2125 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2126 yylval.opval->op_private = OPpCONST_ENTERED;
2127 gv_fetchpv(SvPVX(sym),
2129 ? (GV_ADDMULTI | GV_ADDINEVAL)
2132 ((PL_tokenbuf[0] == '$') ? SVt_PV
2133 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2138 /* if it's a sort block and they're naming $a or $b */
2139 if (PL_last_lop_op == OP_SORT &&
2140 PL_tokenbuf[0] == '$' &&
2141 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2144 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2145 d < PL_bufend && *d != '\n';
2148 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2149 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2155 yylval.opval = newOP(OP_PADANY, 0);
2156 yylval.opval->op_targ = tmp;
2162 Whine if they've said @foo in a doublequoted string,
2163 and @foo isn't a variable we can find in the symbol
2166 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2167 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2168 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2169 && ckWARN(WARN_AMBIGUOUS))
2171 /* Downgraded from fatal to warning 20000522 mjd */
2172 Perl_warner(aTHX_ WARN_AMBIGUOUS,
2173 "Possible unintended interpolation of %s in string",
2178 /* build ops for a bareword */
2179 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2180 yylval.opval->op_private = OPpCONST_ENTERED;
2181 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2182 ((PL_tokenbuf[0] == '$') ? SVt_PV
2183 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2188 /* no identifier pending identification */
2190 switch (PL_lex_state) {
2192 case LEX_NORMAL: /* Some compilers will produce faster */
2193 case LEX_INTERPNORMAL: /* code if we comment these out. */
2197 /* when we've already built the next token, just pull it out of the queue */
2200 yylval = PL_nextval[PL_nexttoke];
2202 PL_lex_state = PL_lex_defer;
2203 PL_expect = PL_lex_expect;
2204 PL_lex_defer = LEX_NORMAL;
2206 return(PL_nexttype[PL_nexttoke]);
2208 /* interpolated case modifiers like \L \U, including \Q and \E.
2209 when we get here, PL_bufptr is at the \
2211 case LEX_INTERPCASEMOD:
2213 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2214 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2216 /* handle \E or end of string */
2217 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2221 if (PL_lex_casemods) {
2222 oldmod = PL_lex_casestack[--PL_lex_casemods];
2223 PL_lex_casestack[PL_lex_casemods] = '\0';
2225 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2227 PL_lex_state = LEX_INTERPCONCAT;
2231 if (PL_bufptr != PL_bufend)
2233 PL_lex_state = LEX_INTERPCONCAT;
2238 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2239 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2240 if (strchr("LU", *s) &&
2241 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2243 PL_lex_casestack[--PL_lex_casemods] = '\0';
2246 if (PL_lex_casemods > 10) {
2247 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2248 if (newlb != PL_lex_casestack) {
2250 PL_lex_casestack = newlb;
2253 PL_lex_casestack[PL_lex_casemods++] = *s;
2254 PL_lex_casestack[PL_lex_casemods] = '\0';
2255 PL_lex_state = LEX_INTERPCONCAT;
2256 PL_nextval[PL_nexttoke].ival = 0;
2259 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2261 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2263 PL_nextval[PL_nexttoke].ival = OP_LC;
2265 PL_nextval[PL_nexttoke].ival = OP_UC;
2267 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2269 Perl_croak(aTHX_ "panic: yylex");
2272 if (PL_lex_starts) {
2281 case LEX_INTERPPUSH:
2282 return sublex_push();
2284 case LEX_INTERPSTART:
2285 if (PL_bufptr == PL_bufend)
2286 return sublex_done();
2288 PL_lex_dojoin = (*PL_bufptr == '@');
2289 PL_lex_state = LEX_INTERPNORMAL;
2290 if (PL_lex_dojoin) {
2291 PL_nextval[PL_nexttoke].ival = 0;
2294 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2295 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2296 force_next(PRIVATEREF);
2298 force_ident("\"", '$');
2299 #endif /* USE_THREADS */
2300 PL_nextval[PL_nexttoke].ival = 0;
2302 PL_nextval[PL_nexttoke].ival = 0;
2304 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2307 if (PL_lex_starts++) {
2313 case LEX_INTERPENDMAYBE:
2314 if (intuit_more(PL_bufptr)) {
2315 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2321 if (PL_lex_dojoin) {
2322 PL_lex_dojoin = FALSE;
2323 PL_lex_state = LEX_INTERPCONCAT;
2326 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2327 && SvEVALED(PL_lex_repl))
2329 if (PL_bufptr != PL_bufend)
2330 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2331 PL_lex_repl = Nullsv;
2334 case LEX_INTERPCONCAT:
2336 if (PL_lex_brackets)
2337 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2339 if (PL_bufptr == PL_bufend)
2340 return sublex_done();
2342 if (SvIVX(PL_linestr) == '\'') {
2343 SV *sv = newSVsv(PL_linestr);
2346 else if ( PL_hints & HINT_NEW_RE )
2347 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2348 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2352 s = scan_const(PL_bufptr);
2354 PL_lex_state = LEX_INTERPCASEMOD;
2356 PL_lex_state = LEX_INTERPSTART;
2359 if (s != PL_bufptr) {
2360 PL_nextval[PL_nexttoke] = yylval;
2363 if (PL_lex_starts++)
2373 PL_lex_state = LEX_NORMAL;
2374 s = scan_formline(PL_bufptr);
2375 if (!PL_lex_formbrack)
2381 PL_oldoldbufptr = PL_oldbufptr;
2384 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2385 exp_name[PL_expect], s);
2391 if (isIDFIRST_lazy_if(s,UTF))
2393 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2396 goto fake_eof; /* emulate EOF on ^D or ^Z */
2401 if (PL_lex_brackets)
2402 yyerror("Missing right curly or square bracket");
2405 if (s++ < PL_bufend)
2406 goto retry; /* ignore stray nulls */
2409 if (!PL_in_eval && !PL_preambled) {
2410 PL_preambled = TRUE;
2411 sv_setpv(PL_linestr,incl_perldb());
2412 if (SvCUR(PL_linestr))
2413 sv_catpv(PL_linestr,";");
2415 while(AvFILLp(PL_preambleav) >= 0) {
2416 SV *tmpsv = av_shift(PL_preambleav);
2417 sv_catsv(PL_linestr, tmpsv);
2418 sv_catpv(PL_linestr, ";");
2421 sv_free((SV*)PL_preambleav);
2422 PL_preambleav = NULL;
2424 if (PL_minus_n || PL_minus_p) {
2425 sv_catpv(PL_linestr, "LINE: while (<>) {");
2427 sv_catpv(PL_linestr,"chomp;");
2429 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2431 GvIMPORTED_AV_on(gv);
2433 if (strchr("/'\"", *PL_splitstr)
2434 && strchr(PL_splitstr + 1, *PL_splitstr))
2435 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2438 s = "'~#\200\1'"; /* surely one char is unused...*/
2439 while (s[1] && strchr(PL_splitstr, *s)) s++;
2441 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2442 "q" + (delim == '\''), delim);
2443 for (s = PL_splitstr; *s; s++) {
2445 sv_catpvn(PL_linestr, "\\", 1);
2446 sv_catpvn(PL_linestr, s, 1);
2448 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2452 sv_catpv(PL_linestr,"@F=split(' ');");
2455 sv_catpv(PL_linestr, "\n");
2456 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2457 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2458 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2459 SV *sv = NEWSV(85,0);
2461 sv_upgrade(sv, SVt_PVMG);
2462 sv_setsv(sv,PL_linestr);
2463 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2468 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2471 if (PL_preprocess && !PL_in_eval)
2472 (void)PerlProc_pclose(PL_rsfp);
2473 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2474 PerlIO_clearerr(PL_rsfp);
2476 (void)PerlIO_close(PL_rsfp);
2478 PL_doextract = FALSE;
2480 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2481 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2482 sv_catpv(PL_linestr,";}");
2483 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2484 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2485 PL_minus_n = PL_minus_p = 0;
2488 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2489 sv_setpv(PL_linestr,"");
2490 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2493 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2494 PL_doextract = FALSE;
2496 /* Incest with pod. */
2497 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2498 sv_setpv(PL_linestr, "");
2499 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2500 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2501 PL_doextract = FALSE;
2505 } while (PL_doextract);
2506 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2507 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2508 SV *sv = NEWSV(85,0);
2510 sv_upgrade(sv, SVt_PVMG);
2511 sv_setsv(sv,PL_linestr);
2512 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2514 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2515 if (CopLINE(PL_curcop) == 1) {
2516 while (s < PL_bufend && isSPACE(*s))
2518 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2522 if (*s == '#' && *(s+1) == '!')
2524 #ifdef ALTERNATE_SHEBANG
2526 static char as[] = ALTERNATE_SHEBANG;
2527 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2528 d = s + (sizeof(as) - 1);
2530 #endif /* ALTERNATE_SHEBANG */
2539 while (*d && !isSPACE(*d))
2543 #ifdef ARG_ZERO_IS_SCRIPT
2544 if (ipathend > ipath) {
2546 * HP-UX (at least) sets argv[0] to the script name,
2547 * which makes $^X incorrect. And Digital UNIX and Linux,
2548 * at least, set argv[0] to the basename of the Perl
2549 * interpreter. So, having found "#!", we'll set it right.
2551 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2552 assert(SvPOK(x) || SvGMAGICAL(x));
2553 if (sv_eq(x, CopFILESV(PL_curcop))) {
2554 sv_setpvn(x, ipath, ipathend - ipath);
2557 TAINT_NOT; /* $^X is always tainted, but that's OK */
2559 #endif /* ARG_ZERO_IS_SCRIPT */
2564 d = instr(s,"perl -");
2566 d = instr(s,"perl");
2568 /* avoid getting into infinite loops when shebang
2569 * line contains "Perl" rather than "perl" */
2571 for (d = ipathend-4; d >= ipath; --d) {
2572 if ((*d == 'p' || *d == 'P')
2573 && !ibcmp(d, "perl", 4))
2583 #ifdef ALTERNATE_SHEBANG
2585 * If the ALTERNATE_SHEBANG on this system starts with a
2586 * character that can be part of a Perl expression, then if
2587 * we see it but not "perl", we're probably looking at the
2588 * start of Perl code, not a request to hand off to some
2589 * other interpreter. Similarly, if "perl" is there, but
2590 * not in the first 'word' of the line, we assume the line
2591 * contains the start of the Perl program.
2593 if (d && *s != '#') {
2595 while (*c && !strchr("; \t\r\n\f\v#", *c))
2598 d = Nullch; /* "perl" not in first word; ignore */
2600 *s = '#'; /* Don't try to parse shebang line */
2602 #endif /* ALTERNATE_SHEBANG */
2603 #ifndef MACOS_TRADITIONAL
2608 !instr(s,"indir") &&
2609 instr(PL_origargv[0],"perl"))
2615 while (s < PL_bufend && isSPACE(*s))
2617 if (s < PL_bufend) {
2618 Newz(899,newargv,PL_origargc+3,char*);
2620 while (s < PL_bufend && !isSPACE(*s))
2623 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2626 newargv = PL_origargv;
2628 PerlProc_execv(ipath, newargv);
2629 Perl_croak(aTHX_ "Can't exec %s", ipath);
2633 U32 oldpdb = PL_perldb;
2634 bool oldn = PL_minus_n;
2635 bool oldp = PL_minus_p;
2637 while (*d && !isSPACE(*d)) d++;
2638 while (SPACE_OR_TAB(*d)) d++;
2642 if (*d == 'M' || *d == 'm') {
2644 while (*d && !isSPACE(*d)) d++;
2645 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2648 d = moreswitches(d);
2650 if ((PERLDB_LINE && !oldpdb) ||
2651 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2652 /* if we have already added "LINE: while (<>) {",
2653 we must not do it again */
2655 sv_setpv(PL_linestr, "");
2656 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2657 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2658 PL_preambled = FALSE;
2660 (void)gv_fetchfile(PL_origfilename);
2667 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2669 PL_lex_state = LEX_FORMLINE;
2674 #ifdef PERL_STRICT_CR
2675 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2677 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2679 case ' ': case '\t': case '\f': case 013:
2680 #ifdef MACOS_TRADITIONAL
2687 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2688 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2689 /* handle eval qq[#line 1 "foo"\n ...] */
2690 CopLINE_dec(PL_curcop);
2694 while (s < d && *s != '\n')
2699 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2701 PL_lex_state = LEX_FORMLINE;
2711 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2716 while (s < PL_bufend && SPACE_OR_TAB(*s))
2719 if (strnEQ(s,"=>",2)) {
2720 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2721 OPERATOR('-'); /* unary minus */
2723 PL_last_uni = PL_oldbufptr;
2724 PL_last_lop_op = OP_FTEREAD; /* good enough */
2726 case 'r': FTST(OP_FTEREAD);
2727 case 'w': FTST(OP_FTEWRITE);
2728 case 'x': FTST(OP_FTEEXEC);
2729 case 'o': FTST(OP_FTEOWNED);
2730 case 'R': FTST(OP_FTRREAD);
2731 case 'W': FTST(OP_FTRWRITE);
2732 case 'X': FTST(OP_FTREXEC);
2733 case 'O': FTST(OP_FTROWNED);
2734 case 'e': FTST(OP_FTIS);
2735 case 'z': FTST(OP_FTZERO);
2736 case 's': FTST(OP_FTSIZE);
2737 case 'f': FTST(OP_FTFILE);
2738 case 'd': FTST(OP_FTDIR);
2739 case 'l': FTST(OP_FTLINK);
2740 case 'p': FTST(OP_FTPIPE);
2741 case 'S': FTST(OP_FTSOCK);
2742 case 'u': FTST(OP_FTSUID);
2743 case 'g': FTST(OP_FTSGID);
2744 case 'k': FTST(OP_FTSVTX);
2745 case 'b': FTST(OP_FTBLK);
2746 case 'c': FTST(OP_FTCHR);
2747 case 't': FTST(OP_FTTTY);
2748 case 'T': FTST(OP_FTTEXT);
2749 case 'B': FTST(OP_FTBINARY);
2750 case 'M': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTMTIME);
2751 case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME);
2752 case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME);
2754 Perl_croak(aTHX_ "Unrecognized file test: -%c", (int)tmp);
2761 if (PL_expect == XOPERATOR)
2766 else if (*s == '>') {
2769 if (isIDFIRST_lazy_if(s,UTF)) {
2770 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2778 if (PL_expect == XOPERATOR)
2781 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2783 OPERATOR('-'); /* unary minus */
2790 if (PL_expect == XOPERATOR)
2795 if (PL_expect == XOPERATOR)
2798 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2804 if (PL_expect != XOPERATOR) {
2805 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2806 PL_expect = XOPERATOR;
2807 force_ident(PL_tokenbuf, '*');
2820 if (PL_expect == XOPERATOR) {
2824 PL_tokenbuf[0] = '%';
2825 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2826 if (!PL_tokenbuf[1]) {
2828 yyerror("Final % should be \\% or %name");
2831 PL_pending_ident = '%';
2850 switch (PL_expect) {
2853 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2855 PL_bufptr = s; /* update in case we back off */
2861 PL_expect = XTERMBLOCK;
2865 while (isIDFIRST_lazy_if(s,UTF)) {
2866 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2867 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2868 if (tmp < 0) tmp = -tmp;
2883 d = scan_str(d,TRUE,TRUE);
2886 SvREFCNT_dec(PL_lex_stuff);
2887 PL_lex_stuff = Nullsv;
2889 /* MUST advance bufptr here to avoid bogus
2890 "at end of line" context messages from yyerror().
2892 PL_bufptr = s + len;
2893 yyerror("Unterminated attribute parameter in attribute list");
2896 return 0; /* EOF indicator */
2900 SV *sv = newSVpvn(s, len);
2901 sv_catsv(sv, PL_lex_stuff);
2902 attrs = append_elem(OP_LIST, attrs,
2903 newSVOP(OP_CONST, 0, sv));
2904 SvREFCNT_dec(PL_lex_stuff);
2905 PL_lex_stuff = Nullsv;
2908 attrs = append_elem(OP_LIST, attrs,
2909 newSVOP(OP_CONST, 0,
2913 if (*s == ':' && s[1] != ':')
2916 break; /* require real whitespace or :'s */
2918 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
2919 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
2920 char q = ((*s == '\'') ? '"' : '\'');
2921 /* If here for an expression, and parsed no attrs, back off. */
2922 if (tmp == '=' && !attrs) {
2926 /* MUST advance bufptr here to avoid bogus "at end of line"
2927 context messages from yyerror().
2931 yyerror("Unterminated attribute list");
2933 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
2941 PL_nextval[PL_nexttoke].opval = attrs;
2949 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
2950 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
2966 if (PL_lex_brackets <= 0)
2967 yyerror("Unmatched right square bracket");
2970 if (PL_lex_state == LEX_INTERPNORMAL) {
2971 if (PL_lex_brackets == 0) {
2972 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
2973 PL_lex_state = LEX_INTERPEND;
2980 if (PL_lex_brackets > 100) {
2981 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
2982 if (newlb != PL_lex_brackstack) {
2984 PL_lex_brackstack = newlb;
2987 switch (PL_expect) {
2989 if (PL_lex_formbrack) {
2993 if (PL_oldoldbufptr == PL_last_lop)
2994 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2996 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2997 OPERATOR(HASHBRACK);
2999 while (s < PL_bufend && SPACE_OR_TAB(*s))
3002 PL_tokenbuf[0] = '\0';
3003 if (d < PL_bufend && *d == '-') {
3004 PL_tokenbuf[0] = '-';
3006 while (d < PL_bufend && SPACE_OR_TAB(*d))
3009 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3010 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3012 while (d < PL_bufend && SPACE_OR_TAB(*d))
3015 char minus = (PL_tokenbuf[0] == '-');
3016 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3024 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3029 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3034 if (PL_oldoldbufptr == PL_last_lop)
3035 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3037 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3040 OPERATOR(HASHBRACK);
3041 /* This hack serves to disambiguate a pair of curlies
3042 * as being a block or an anon hash. Normally, expectation
3043 * determines that, but in cases where we're not in a
3044 * position to expect anything in particular (like inside
3045 * eval"") we have to resolve the ambiguity. This code
3046 * covers the case where the first term in the curlies is a
3047 * quoted string. Most other cases need to be explicitly
3048 * disambiguated by prepending a `+' before the opening
3049 * curly in order to force resolution as an anon hash.
3051 * XXX should probably propagate the outer expectation
3052 * into eval"" to rely less on this hack, but that could
3053 * potentially break current behavior of eval"".
3057 if (*s == '\'' || *s == '"' || *s == '`') {
3058 /* common case: get past first string, handling escapes */
3059 for (t++; t < PL_bufend && *t != *s;)
3060 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3064 else if (*s == 'q') {
3067 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3071 char open, close, term;
3074 while (t < PL_bufend && isSPACE(*t))
3078 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3082 for (t++; t < PL_bufend; t++) {
3083 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3085 else if (*t == open)
3089 for (t++; t < PL_bufend; t++) {
3090 if (*t == '\\' && t+1 < PL_bufend)
3092 else if (*t == close && --brackets <= 0)
3094 else if (*t == open)
3100 else if (isALNUM_lazy_if(t,UTF)) {
3102 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3105 while (t < PL_bufend && isSPACE(*t))
3107 /* if comma follows first term, call it an anon hash */
3108 /* XXX it could be a comma expression with loop modifiers */
3109 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3110 || (*t == '=' && t[1] == '>')))
3111 OPERATOR(HASHBRACK);
3112 if (PL_expect == XREF)
3115 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3121 yylval.ival = CopLINE(PL_curcop);
3122 if (isSPACE(*s) || *s == '#')
3123 PL_copline = NOLINE; /* invalidate current command line number */
3128 if (PL_lex_brackets <= 0)
3129 yyerror("Unmatched right curly bracket");
3131 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3132 if (PL_lex_brackets < PL_lex_formbrack)
3133 PL_lex_formbrack = 0;
3134 if (PL_lex_state == LEX_INTERPNORMAL) {
3135 if (PL_lex_brackets == 0) {
3136 if (PL_expect & XFAKEBRACK) {
3137 PL_expect &= XENUMMASK;
3138 PL_lex_state = LEX_INTERPEND;
3140 return yylex(); /* ignore fake brackets */
3142 if (*s == '-' && s[1] == '>')
3143 PL_lex_state = LEX_INTERPENDMAYBE;
3144 else if (*s != '[' && *s != '{')
3145 PL_lex_state = LEX_INTERPEND;
3148 if (PL_expect & XFAKEBRACK) {
3149 PL_expect &= XENUMMASK;
3151 return yylex(); /* ignore fake brackets */
3161 if (PL_expect == XOPERATOR) {
3162 if (ckWARN(WARN_SEMICOLON)
3163 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3165 CopLINE_dec(PL_curcop);
3166 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3167 CopLINE_inc(PL_curcop);
3172 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3174 PL_expect = XOPERATOR;
3175 force_ident(PL_tokenbuf, '&');
3179 yylval.ival = (OPpENTERSUB_AMPER<<8);
3198 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3199 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3201 if (PL_expect == XSTATE && isALPHA(tmp) &&
3202 (s == PL_linestart+1 || s[-2] == '\n') )
3204 if (PL_in_eval && !PL_rsfp) {
3209 if (strnEQ(s,"=cut",4)) {
3223 PL_doextract = TRUE;
3226 if (PL_lex_brackets < PL_lex_formbrack) {
3228 #ifdef PERL_STRICT_CR
3229 for (t = s; SPACE_OR_TAB(*t); t++) ;
3231 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3233 if (*t == '\n' || *t == '#') {
3251 if (PL_expect != XOPERATOR) {
3252 if (s[1] != '<' && !strchr(s,'>'))
3255 s = scan_heredoc(s);
3257 s = scan_inputsymbol(s);
3258 TERM(sublex_start());
3263 SHop(OP_LEFT_SHIFT);
3277 SHop(OP_RIGHT_SHIFT);
3286 if (PL_expect == XOPERATOR) {
3287 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3290 return ','; /* grandfather non-comma-format format */
3294 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3295 PL_tokenbuf[0] = '@';
3296 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3297 sizeof PL_tokenbuf - 1, FALSE);
3298 if (PL_expect == XOPERATOR)
3299 no_op("Array length", s);
3300 if (!PL_tokenbuf[1])
3302 PL_expect = XOPERATOR;
3303 PL_pending_ident = '#';
3307 PL_tokenbuf[0] = '$';
3308 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3309 sizeof PL_tokenbuf - 1, FALSE);
3310 if (PL_expect == XOPERATOR)
3312 if (!PL_tokenbuf[1]) {
3314 yyerror("Final $ should be \\$ or $name");
3318 /* This kludge not intended to be bulletproof. */
3319 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3320 yylval.opval = newSVOP(OP_CONST, 0,
3321 newSViv(PL_compiling.cop_arybase));
3322 yylval.opval->op_private = OPpCONST_ARYBASE;
3328 if (PL_lex_state == LEX_NORMAL)
3331 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3334 PL_tokenbuf[0] = '@';
3335 if (ckWARN(WARN_SYNTAX)) {
3337 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3340 PL_bufptr = skipspace(PL_bufptr);
3341 while (t < PL_bufend && *t != ']')
3343 Perl_warner(aTHX_ WARN_SYNTAX,
3344 "Multidimensional syntax %.*s not supported",
3345 (t - PL_bufptr) + 1, PL_bufptr);
3349 else if (*s == '{') {
3350 PL_tokenbuf[0] = '%';
3351 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3352 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3354 char tmpbuf[sizeof PL_tokenbuf];
3356 for (t++; isSPACE(*t); t++) ;
3357 if (isIDFIRST_lazy_if(t,UTF)) {
3358 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3359 for (; isSPACE(*t); t++) ;
3360 if (*t == ';' && get_cv(tmpbuf, FALSE))
3361 Perl_warner(aTHX_ WARN_SYNTAX,
3362 "You need to quote \"%s\"", tmpbuf);
3368 PL_expect = XOPERATOR;
3369 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3370 bool islop = (PL_last_lop == PL_oldoldbufptr);
3371 if (!islop || PL_last_lop_op == OP_GREPSTART)
3372 PL_expect = XOPERATOR;
3373 else if (strchr("$@\"'`q", *s))
3374 PL_expect = XTERM; /* e.g. print $fh "foo" */
3375 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3376 PL_expect = XTERM; /* e.g. print $fh &sub */
3377 else if (isIDFIRST_lazy_if(s,UTF)) {
3378 char tmpbuf[sizeof PL_tokenbuf];
3379 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3380 if ((tmp = keyword(tmpbuf, len))) {
3381 /* binary operators exclude handle interpretations */
3393 PL_expect = XTERM; /* e.g. print $fh length() */
3398 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3399 if (gv && GvCVu(gv))
3400 PL_expect = XTERM; /* e.g. print $fh subr() */
3403 else if (isDIGIT(*s))
3404 PL_expect = XTERM; /* e.g. print $fh 3 */
3405 else if (*s == '.' && isDIGIT(s[1]))
3406 PL_expect = XTERM; /* e.g. print $fh .3 */
3407 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3408 PL_expect = XTERM; /* e.g. print $fh -1 */
3409 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3410 PL_expect = XTERM; /* print $fh <<"EOF" */
3412 PL_pending_ident = '$';
3416 if (PL_expect == XOPERATOR)
3418 PL_tokenbuf[0] = '@';
3419 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3420 if (!PL_tokenbuf[1]) {
3422 yyerror("Final @ should be \\@ or @name");
3425 if (PL_lex_state == LEX_NORMAL)
3427 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3429 PL_tokenbuf[0] = '%';
3431 /* Warn about @ where they meant $. */
3432 if (ckWARN(WARN_SYNTAX)) {
3433 if (*s == '[' || *s == '{') {
3435 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3437 if (*t == '}' || *t == ']') {
3439 PL_bufptr = skipspace(PL_bufptr);
3440 Perl_warner(aTHX_ WARN_SYNTAX,
3441 "Scalar value %.*s better written as $%.*s",
3442 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3447 PL_pending_ident = '@';
3450 case '/': /* may either be division or pattern */
3451 case '?': /* may either be conditional or pattern */
3452 if (PL_expect != XOPERATOR) {
3453 /* Disable warning on "study /blah/" */
3454 if (PL_oldoldbufptr == PL_last_uni
3455 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3456 || memNE(PL_last_uni, "study", 5)
3457 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3459 s = scan_pat(s,OP_MATCH);
3460 TERM(sublex_start());
3468 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3469 #ifdef PERL_STRICT_CR
3472 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3474 && (s == PL_linestart || s[-1] == '\n') )
3476 PL_lex_formbrack = 0;
3480 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3486 yylval.ival = OPf_SPECIAL;
3492 if (PL_expect != XOPERATOR)
3497 case '0': case '1': case '2': case '3': case '4':
3498 case '5': case '6': case '7': case '8': case '9':
3500 if (PL_expect == XOPERATOR)
3505 s = scan_str(s,FALSE,FALSE);
3506 if (PL_expect == XOPERATOR) {
3507 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3510 return ','; /* grandfather non-comma-format format */
3516 missingterm((char*)0);
3517 yylval.ival = OP_CONST;
3518 TERM(sublex_start());
3521 s = scan_str(s,FALSE,FALSE);
3522 if (PL_expect == XOPERATOR) {
3523 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3526 return ','; /* grandfather non-comma-format format */
3532 missingterm((char*)0);
3533 yylval.ival = OP_CONST;
3534 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3535 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3536 yylval.ival = OP_STRINGIFY;
3540 TERM(sublex_start());
3543 s = scan_str(s,FALSE,FALSE);
3544 if (PL_expect == XOPERATOR)
3545 no_op("Backticks",s);
3547 missingterm((char*)0);
3548 yylval.ival = OP_BACKTICK;
3550 TERM(sublex_start());
3554 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3555 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3557 if (PL_expect == XOPERATOR)
3558 no_op("Backslash",s);
3562 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3566 while (isDIGIT(*start) || *start == '_')
3568 if (*start == '.' && isDIGIT(start[1])) {
3572 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3573 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF)) {
3577 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3587 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3626 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3628 /* Some keywords can be followed by any delimiter, including ':' */
3629 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3630 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3631 (PL_tokenbuf[0] == 'q' &&
3632 strchr("qwxr", PL_tokenbuf[1])))));
3634 /* x::* is just a word, unless x is "CORE" */
3635 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3639 while (d < PL_bufend && isSPACE(*d))
3640 d++; /* no comments skipped here, or s### is misparsed */
3642 /* Is this a label? */
3643 if (!tmp && PL_expect == XSTATE
3644 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3646 yylval.pval = savepv(PL_tokenbuf);
3651 /* Check for keywords */
3652 tmp = keyword(PL_tokenbuf, len);
3654 /* Is this a word before a => operator? */
3655 if (*d == '=' && d[1] == '>') {
3657 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3658 yylval.opval->op_private = OPpCONST_BARE;
3662 if (tmp < 0) { /* second-class keyword? */
3663 GV *ogv = Nullgv; /* override (winner) */
3664 GV *hgv = Nullgv; /* hidden (loser) */
3665 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3667 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3670 if (GvIMPORTED_CV(gv))
3672 else if (! CvMETHOD(cv))
3676 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3677 (gv = *gvp) != (GV*)&PL_sv_undef &&
3678 GvCVu(gv) && GvIMPORTED_CV(gv))
3684 tmp = 0; /* overridden by import or by GLOBAL */
3687 && -tmp==KEY_lock /* XXX generalizable kludge */
3689 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3691 tmp = 0; /* any sub overrides "weak" keyword */
3693 else { /* no override */
3697 if (ckWARN(WARN_AMBIGUOUS) && hgv
3698 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3699 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3700 "Ambiguous call resolved as CORE::%s(), %s",
3701 GvENAME(hgv), "qualify as such or use &");
3708 default: /* not a keyword */
3711 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3713 /* Get the rest if it looks like a package qualifier */
3715 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3717 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3720 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3721 *s == '\'' ? "'" : "::");
3725 if (PL_expect == XOPERATOR) {
3726 if (PL_bufptr == PL_linestart) {
3727 CopLINE_dec(PL_curcop);
3728 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3729 CopLINE_inc(PL_curcop);
3732 no_op("Bareword",s);
3735 /* Look for a subroutine with this name in current package,
3736 unless name is "Foo::", in which case Foo is a bearword
3737 (and a package name). */
3740 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3742 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3743 Perl_warner(aTHX_ WARN_BAREWORD,
3744 "Bareword \"%s\" refers to nonexistent package",
3747 PL_tokenbuf[len] = '\0';
3754 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3757 /* if we saw a global override before, get the right name */
3760 sv = newSVpvn("CORE::GLOBAL::",14);
3761 sv_catpv(sv,PL_tokenbuf);
3764 sv = newSVpv(PL_tokenbuf,0);
3766 /* Presume this is going to be a bareword of some sort. */
3769 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3770 yylval.opval->op_private = OPpCONST_BARE;
3772 /* And if "Foo::", then that's what it certainly is. */
3777 /* See if it's the indirect object for a list operator. */
3779 if (PL_oldoldbufptr &&
3780 PL_oldoldbufptr < PL_bufptr &&
3781 (PL_oldoldbufptr == PL_last_lop
3782 || PL_oldoldbufptr == PL_last_uni) &&
3783 /* NO SKIPSPACE BEFORE HERE! */
3784 (PL_expect == XREF ||
3785 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3787 bool immediate_paren = *s == '(';
3789 /* (Now we can afford to cross potential line boundary.) */
3792 /* Two barewords in a row may indicate method call. */
3794 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3797 /* If not a declared subroutine, it's an indirect object. */
3798 /* (But it's an indir obj regardless for sort.) */
3800 if ((PL_last_lop_op == OP_SORT ||
3801 (!immediate_paren && (!gv || !GvCVu(gv)))) &&
3802 (PL_last_lop_op != OP_MAPSTART &&
3803 PL_last_lop_op != OP_GREPSTART))
3805 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3811 PL_expect = XOPERATOR;
3814 /* Is this a word before a => operator? */
3815 if (*s == '=' && s[1] == '>') {
3817 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
3821 /* If followed by a paren, it's certainly a subroutine. */
3824 if (gv && GvCVu(gv)) {
3825 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
3826 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3831 PL_nextval[PL_nexttoke].opval = yylval.opval;
3832 PL_expect = XOPERATOR;
3838 /* If followed by var or block, call it a method (unless sub) */
3840 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3841 PL_last_lop = PL_oldbufptr;
3842 PL_last_lop_op = OP_METHOD;
3846 /* If followed by a bareword, see if it looks like indir obj. */
3848 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3851 /* Not a method, so call it a subroutine (if defined) */
3853 if (gv && GvCVu(gv)) {
3855 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3856 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3857 "Ambiguous use of -%s resolved as -&%s()",
3858 PL_tokenbuf, PL_tokenbuf);
3859 /* Check for a constant sub */
3861 if ((sv = cv_const_sv(cv))) {
3863 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3864 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3865 yylval.opval->op_private = 0;
3869 /* Resolve to GV now. */
3870 op_free(yylval.opval);
3871 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3872 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3873 PL_last_lop = PL_oldbufptr;
3874 PL_last_lop_op = OP_ENTERSUB;
3875 /* Is there a prototype? */
3878 char *proto = SvPV((SV*)cv, len);
3881 if (strEQ(proto, "$"))
3883 if (*proto == '&' && *s == '{') {
3884 sv_setpv(PL_subname,"__ANON__");
3888 PL_nextval[PL_nexttoke].opval = yylval.opval;
3894 /* Call it a bare word */
3896 if (PL_hints & HINT_STRICT_SUBS)
3897 yylval.opval->op_private |= OPpCONST_STRICT;
3900 if (ckWARN(WARN_RESERVED)) {
3901 if (lastchar != '-') {
3902 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
3904 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
3911 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
3912 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3913 "Operator or semicolon missing before %c%s",
3914 lastchar, PL_tokenbuf);
3915 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3916 "Ambiguous use of %c resolved as operator %c",
3917 lastchar, lastchar);
3923 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3924 newSVpv(CopFILE(PL_curcop),0));
3928 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3929 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
3932 case KEY___PACKAGE__:
3933 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3935 ? newSVsv(PL_curstname)
3944 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
3945 char *pname = "main";
3946 if (PL_tokenbuf[2] == 'D')
3947 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
3948 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
3951 GvIOp(gv) = newIO();
3952 IoIFP(GvIOp(gv)) = PL_rsfp;
3953 #if defined(HAS_FCNTL) && defined(F_SETFD)
3955 int fd = PerlIO_fileno(PL_rsfp);
3956 fcntl(fd,F_SETFD,fd >= 3);
3959 /* Mark this internal pseudo-handle as clean */
3960 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
3962 IoTYPE(GvIOp(gv)) = '|';
3963 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
3964 IoTYPE(GvIOp(gv)) = '-';
3966 IoTYPE(GvIOp(gv)) = '<';
3967 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
3968 /* if the script was opened in binmode, we need to revert
3969 * it to text mode for compatibility; but only iff it has CRs
3970 * XXX this is a questionable hack at best. */
3971 if (PL_bufend-PL_bufptr > 2
3972 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
3975 if (IoTYPE(GvIOp(gv)) == '<') {
3976 loc = PerlIO_tell(PL_rsfp);
3977 (void)PerlIO_seek(PL_rsfp, 0L, 0);
3979 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
3980 #if defined(__BORLANDC__)
3981 /* XXX see note in do_binmode() */
3982 ((FILE*)PL_rsfp)->flags |= _F_BIN;
3985 PerlIO_seek(PL_rsfp, loc, 0);
4000 if (PL_expect == XSTATE) {
4007 if (*s == ':' && s[1] == ':') {
4010 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4011 if (!(tmp = keyword(PL_tokenbuf, len)))
4012 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4026 LOP(OP_ACCEPT,XTERM);
4032 LOP(OP_ATAN2,XTERM);
4038 LOP(OP_BINMODE,XTERM);
4041 LOP(OP_BLESS,XTERM);
4050 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4067 if (!PL_cryptseen) {
4068 PL_cryptseen = TRUE;
4072 LOP(OP_CRYPT,XTERM);
4075 if (ckWARN(WARN_CHMOD)) {
4076 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4077 if (*d != '0' && isDIGIT(*d))
4078 Perl_warner(aTHX_ WARN_CHMOD,
4079 "chmod() mode argument is missing initial 0");
4081 LOP(OP_CHMOD,XTERM);
4084 LOP(OP_CHOWN,XTERM);
4087 LOP(OP_CONNECT,XTERM);
4103 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4107 PL_hints |= HINT_BLOCK_SCOPE;
4117 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4118 LOP(OP_DBMOPEN,XTERM);
4124 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4131 yylval.ival = CopLINE(PL_curcop);
4145 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4146 UNIBRACK(OP_ENTEREVAL);
4161 case KEY_endhostent:
4167 case KEY_endservent:
4170 case KEY_endprotoent:
4181 yylval.ival = CopLINE(PL_curcop);
4183 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4185 if ((PL_bufend - p) >= 3 &&
4186 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4188 else if ((PL_bufend - p) >= 4 &&
4189 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4192 if (isIDFIRST_lazy_if(p,UTF)) {
4193 p = scan_ident(p, PL_bufend,
4194 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4198 Perl_croak(aTHX_ "Missing $ on loop variable");
4203 LOP(OP_FORMLINE,XTERM);
4209 LOP(OP_FCNTL,XTERM);
4215 LOP(OP_FLOCK,XTERM);
4224 LOP(OP_GREPSTART, XREF);
4227 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4242 case KEY_getpriority:
4243 LOP(OP_GETPRIORITY,XTERM);
4245 case KEY_getprotobyname:
4248 case KEY_getprotobynumber:
4249 LOP(OP_GPBYNUMBER,XTERM);
4251 case KEY_getprotoent:
4263 case KEY_getpeername:
4264 UNI(OP_GETPEERNAME);
4266 case KEY_gethostbyname:
4269 case KEY_gethostbyaddr:
4270 LOP(OP_GHBYADDR,XTERM);
4272 case KEY_gethostent:
4275 case KEY_getnetbyname:
4278 case KEY_getnetbyaddr:
4279 LOP(OP_GNBYADDR,XTERM);
4284 case KEY_getservbyname:
4285 LOP(OP_GSBYNAME,XTERM);
4287 case KEY_getservbyport:
4288 LOP(OP_GSBYPORT,XTERM);
4290 case KEY_getservent:
4293 case KEY_getsockname:
4294 UNI(OP_GETSOCKNAME);
4296 case KEY_getsockopt:
4297 LOP(OP_GSOCKOPT,XTERM);
4319 yylval.ival = CopLINE(PL_curcop);
4323 LOP(OP_INDEX,XTERM);
4329 LOP(OP_IOCTL,XTERM);
4341 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4373 LOP(OP_LISTEN,XTERM);
4382 s = scan_pat(s,OP_MATCH);
4383 TERM(sublex_start());
4386 LOP(OP_MAPSTART, XREF);
4389 LOP(OP_MKDIR,XTERM);
4392 LOP(OP_MSGCTL,XTERM);
4395 LOP(OP_MSGGET,XTERM);
4398 LOP(OP_MSGRCV,XTERM);
4401 LOP(OP_MSGSND,XTERM);
4407 if (isIDFIRST_lazy_if(s,UTF)) {
4408 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4409 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4411 PL_in_my_stash = gv_stashpv(PL_tokenbuf, FALSE);
4412 if (!PL_in_my_stash) {
4415 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4423 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4430 if (PL_expect != XSTATE)
4431 yyerror("\"no\" not allowed in expression");
4432 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4433 s = force_version(s);
4438 if (*s == '(' || (s = skipspace(s), *s == '('))
4445 if (isIDFIRST_lazy_if(s,UTF)) {
4447 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4449 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4450 Perl_warner(aTHX_ WARN_PRECEDENCE,
4451 "Precedence problem: open %.*s should be open(%.*s)",
4457 yylval.ival = OP_OR;
4467 LOP(OP_OPEN_DIR,XTERM);
4470 checkcomma(s,PL_tokenbuf,"filehandle");
4474 checkcomma(s,PL_tokenbuf,"filehandle");
4493 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4497 LOP(OP_PIPE_OP,XTERM);
4500 s = scan_str(s,FALSE,FALSE);
4502 missingterm((char*)0);
4503 yylval.ival = OP_CONST;
4504 TERM(sublex_start());
4510 s = scan_str(s,FALSE,FALSE);
4512 missingterm((char*)0);
4514 if (SvCUR(PL_lex_stuff)) {
4517 d = SvPV_force(PL_lex_stuff, len);
4519 for (; isSPACE(*d) && len; --len, ++d) ;
4522 if (!warned && ckWARN(WARN_QW)) {
4523 for (; !isSPACE(*d) && len; --len, ++d) {
4525 Perl_warner(aTHX_ WARN_QW,
4526 "Possible attempt to separate words with commas");
4529 else if (*d == '#') {
4530 Perl_warner(aTHX_ WARN_QW,
4531 "Possible attempt to put comments in qw() list");
4537 for (; !isSPACE(*d) && len; --len, ++d) ;
4539 words = append_elem(OP_LIST, words,
4540 newSVOP(OP_CONST, 0, tokeq(newSVpvn(b, d-b))));
4544 PL_nextval[PL_nexttoke].opval = words;
4549 SvREFCNT_dec(PL_lex_stuff);
4550 PL_lex_stuff = Nullsv;
4555 s = scan_str(s,FALSE,FALSE);
4557 missingterm((char*)0);
4558 yylval.ival = OP_STRINGIFY;
4559 if (SvIVX(PL_lex_stuff) == '\'')
4560 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4561 TERM(sublex_start());
4564 s = scan_pat(s,OP_QR);
4565 TERM(sublex_start());
4568 s = scan_str(s,FALSE,FALSE);
4570 missingterm((char*)0);
4571 yylval.ival = OP_BACKTICK;
4573 TERM(sublex_start());
4580 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4581 s = force_version(s);
4584 *PL_tokenbuf = '\0';
4585 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4586 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4587 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4589 yyerror("<> should be quotes");
4597 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4601 LOP(OP_RENAME,XTERM);
4610 LOP(OP_RINDEX,XTERM);
4633 LOP(OP_REVERSE,XTERM);
4644 TERM(sublex_start());
4646 TOKEN(1); /* force error */
4655 LOP(OP_SELECT,XTERM);
4661 LOP(OP_SEMCTL,XTERM);
4664 LOP(OP_SEMGET,XTERM);
4667 LOP(OP_SEMOP,XTERM);
4673 LOP(OP_SETPGRP,XTERM);
4675 case KEY_setpriority:
4676 LOP(OP_SETPRIORITY,XTERM);
4678 case KEY_sethostent:
4684 case KEY_setservent:
4687 case KEY_setprotoent:
4697 LOP(OP_SEEKDIR,XTERM);
4699 case KEY_setsockopt:
4700 LOP(OP_SSOCKOPT,XTERM);
4706 LOP(OP_SHMCTL,XTERM);
4709 LOP(OP_SHMGET,XTERM);
4712 LOP(OP_SHMREAD,XTERM);
4715 LOP(OP_SHMWRITE,XTERM);
4718 LOP(OP_SHUTDOWN,XTERM);
4727 LOP(OP_SOCKET,XTERM);
4729 case KEY_socketpair:
4730 LOP(OP_SOCKPAIR,XTERM);
4733 checkcomma(s,PL_tokenbuf,"subroutine name");
4735 if (*s == ';' || *s == ')') /* probably a close */
4736 Perl_croak(aTHX_ "sort is now a reserved word");
4738 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4742 LOP(OP_SPLIT,XTERM);
4745 LOP(OP_SPRINTF,XTERM);
4748 LOP(OP_SPLICE,XTERM);
4763 LOP(OP_SUBSTR,XTERM);
4769 char tmpbuf[sizeof PL_tokenbuf];
4771 expectation attrful;
4772 bool have_name, have_proto;
4777 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4778 (*s == ':' && s[1] == ':'))
4781 attrful = XATTRBLOCK;
4782 /* remember buffer pos'n for later force_word */
4783 tboffset = s - PL_oldbufptr;
4784 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4785 if (strchr(tmpbuf, ':'))
4786 sv_setpv(PL_subname, tmpbuf);
4788 sv_setsv(PL_subname,PL_curstname);
4789 sv_catpvn(PL_subname,"::",2);
4790 sv_catpvn(PL_subname,tmpbuf,len);
4797 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4798 PL_expect = XTERMBLOCK;
4799 attrful = XATTRTERM;
4800 sv_setpv(PL_subname,"?");
4804 if (key == KEY_format) {
4806 PL_lex_formbrack = PL_lex_brackets + 1;
4808 (void) force_word(PL_oldbufptr + tboffset, WORD,
4813 /* Look for a prototype */
4817 s = scan_str(s,FALSE,FALSE);
4820 SvREFCNT_dec(PL_lex_stuff);
4821 PL_lex_stuff = Nullsv;
4822 Perl_croak(aTHX_ "Prototype not terminated");
4825 d = SvPVX(PL_lex_stuff);
4827 for (p = d; *p; ++p) {
4832 SvCUR(PL_lex_stuff) = tmp;
4840 if (*s == ':' && s[1] != ':')
4841 PL_expect = attrful;
4844 PL_nextval[PL_nexttoke].opval =
4845 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4846 PL_lex_stuff = Nullsv;
4850 sv_setpv(PL_subname,"__ANON__");
4853 (void) force_word(PL_oldbufptr + tboffset, WORD,
4862 LOP(OP_SYSTEM,XREF);
4865 LOP(OP_SYMLINK,XTERM);
4868 LOP(OP_SYSCALL,XTERM);
4871 LOP(OP_SYSOPEN,XTERM);
4874 LOP(OP_SYSSEEK,XTERM);
4877 LOP(OP_SYSREAD,XTERM);
4880 LOP(OP_SYSWRITE,XTERM);
4884 TERM(sublex_start());
4905 LOP(OP_TRUNCATE,XTERM);
4917 yylval.ival = CopLINE(PL_curcop);
4921 yylval.ival = CopLINE(PL_curcop);
4925 LOP(OP_UNLINK,XTERM);
4931 LOP(OP_UNPACK,XTERM);
4934 LOP(OP_UTIME,XTERM);
4937 if (ckWARN(WARN_UMASK)) {
4938 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4939 if (*d != '0' && isDIGIT(*d))
4940 Perl_warner(aTHX_ WARN_UMASK,
4941 "umask: argument is missing initial 0");
4946 LOP(OP_UNSHIFT,XTERM);
4949 if (PL_expect != XSTATE)
4950 yyerror("\"use\" not allowed in expression");
4952 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4953 s = force_version(s);
4954 if (*s == ';' || (s = skipspace(s), *s == ';')) {
4955 PL_nextval[PL_nexttoke].opval = Nullop;
4960 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4961 s = force_version(s);
4973 yylval.ival = CopLINE(PL_curcop);
4977 PL_hints |= HINT_BLOCK_SCOPE;
4984 LOP(OP_WAITPID,XTERM);
4992 static char ctl_l[2];
4994 if (ctl_l[0] == '\0')
4995 ctl_l[0] = toCTRL('L');
4996 gv_fetchpv(ctl_l,TRUE, SVt_PV);
4999 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5004 if (PL_expect == XOPERATOR)
5010 yylval.ival = OP_XOR;
5015 TERM(sublex_start());
5020 #pragma segment Main
5024 Perl_keyword(pTHX_ register char *d, I32 len)
5029 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5030 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5031 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5032 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5033 if (strEQ(d,"__END__")) return KEY___END__;
5037 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5042 if (strEQ(d,"and")) return -KEY_and;
5043 if (strEQ(d,"abs")) return -KEY_abs;
5046 if (strEQ(d,"alarm")) return -KEY_alarm;
5047 if (strEQ(d,"atan2")) return -KEY_atan2;
5050 if (strEQ(d,"accept")) return -KEY_accept;
5055 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5058 if (strEQ(d,"bless")) return -KEY_bless;
5059 if (strEQ(d,"bind")) return -KEY_bind;
5060 if (strEQ(d,"binmode")) return -KEY_binmode;
5063 if (strEQ(d,"CORE")) return -KEY_CORE;
5064 if (strEQ(d,"CHECK")) return KEY_CHECK;
5069 if (strEQ(d,"cmp")) return -KEY_cmp;
5070 if (strEQ(d,"chr")) return -KEY_chr;
5071 if (strEQ(d,"cos")) return -KEY_cos;
5074 if (strEQ(d,"chop")) return KEY_chop;
5077 if (strEQ(d,"close")) return -KEY_close;
5078 if (strEQ(d,"chdir")) return -KEY_chdir;
5079 if (strEQ(d,"chomp")) return KEY_chomp;
5080 if (strEQ(d,"chmod")) return -KEY_chmod;
5081 if (strEQ(d,"chown")) return -KEY_chown;
5082 if (strEQ(d,"crypt")) return -KEY_crypt;
5085 if (strEQ(d,"chroot")) return -KEY_chroot;
5086 if (strEQ(d,"caller")) return -KEY_caller;
5089 if (strEQ(d,"connect")) return -KEY_connect;
5092 if (strEQ(d,"closedir")) return -KEY_closedir;
5093 if (strEQ(d,"continue")) return -KEY_continue;
5098 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5103 if (strEQ(d,"do")) return KEY_do;
5106 if (strEQ(d,"die")) return -KEY_die;
5109 if (strEQ(d,"dump")) return -KEY_dump;
5112 if (strEQ(d,"delete")) return KEY_delete;
5115 if (strEQ(d,"defined")) return KEY_defined;
5116 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5119 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5124 if (strEQ(d,"EQ")) { deprecate(d); return -KEY_eq;}
5125 if (strEQ(d,"END")) return KEY_END;
5130 if (strEQ(d,"eq")) return -KEY_eq;
5133 if (strEQ(d,"eof")) return -KEY_eof;
5134 if (strEQ(d,"exp")) return -KEY_exp;
5137 if (strEQ(d,"else")) return KEY_else;
5138 if (strEQ(d,"exit")) return -KEY_exit;
5139 if (strEQ(d,"eval")) return KEY_eval;
5140 if (strEQ(d,"exec")) return -KEY_exec;
5141 if (strEQ(d,"each")) return KEY_each;
5144 if (strEQ(d,"elsif")) return KEY_elsif;
5147 if (strEQ(d,"exists")) return KEY_exists;
5148 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5151 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5152 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5155 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5158 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5159 if (strEQ(d,"endservent")) return -KEY_endservent;
5162 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5169 if (strEQ(d,"for")) return KEY_for;
5172 if (strEQ(d,"fork")) return -KEY_fork;
5175 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5176 if (strEQ(d,"flock")) return -KEY_flock;
5179 if (strEQ(d,"format")) return KEY_format;
5180 if (strEQ(d,"fileno")) return -KEY_fileno;
5183 if (strEQ(d,"foreach")) return KEY_foreach;
5186 if (strEQ(d,"formline")) return -KEY_formline;
5192 if (strEQ(d,"GT")) { deprecate(d); return -KEY_gt;}
5193 if (strEQ(d,"GE")) { deprecate(d); return -KEY_ge;}
5197 if (strnEQ(d,"get",3)) {
5202 if (strEQ(d,"ppid")) return -KEY_getppid;
5203 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5206 if (strEQ(d,"pwent")) return -KEY_getpwent;
5207 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5208 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5211 if (strEQ(d,"peername")) return -KEY_getpeername;
5212 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5213 if (strEQ(d,"priority")) return -KEY_getpriority;
5216 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5219 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5223 else if (*d == 'h') {
5224 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5225 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5226 if (strEQ(d,"hostent")) return -KEY_gethostent;
5228 else if (*d == 'n') {
5229 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5230 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5231 if (strEQ(d,"netent")) return -KEY_getnetent;
5233 else if (*d == 's') {
5234 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5235 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5236 if (strEQ(d,"servent")) return -KEY_getservent;
5237 if (strEQ(d,"sockname")) return -KEY_getsockname;
5238 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5240 else if (*d == 'g') {
5241 if (strEQ(d,"grent")) return -KEY_getgrent;
5242 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5243 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5245 else if (*d == 'l') {
5246 if (strEQ(d,"login")) return -KEY_getlogin;
5248 else if (strEQ(d,"c")) return -KEY_getc;
5253 if (strEQ(d,"gt")) return -KEY_gt;
5254 if (strEQ(d,"ge")) return -KEY_ge;
5257 if (strEQ(d,"grep")) return KEY_grep;
5258 if (strEQ(d,"goto")) return KEY_goto;
5259 if (strEQ(d,"glob")) return KEY_glob;
5262 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5267 if (strEQ(d,"hex")) return -KEY_hex;
5270 if (strEQ(d,"INIT")) return KEY_INIT;
5275 if (strEQ(d,"if")) return KEY_if;
5278 if (strEQ(d,"int")) return -KEY_int;
5281 if (strEQ(d,"index")) return -KEY_index;
5282 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5287 if (strEQ(d,"join")) return -KEY_join;
5291 if (strEQ(d,"keys")) return KEY_keys;
5292 if (strEQ(d,"kill")) return -KEY_kill;
5297 if (strEQ(d,"LT")) { deprecate(d); return -KEY_lt;}
5298 if (strEQ(d,"LE")) { deprecate(d); return -KEY_le;}
5304 if (strEQ(d,"lt")) return -KEY_lt;
5305 if (strEQ(d,"le")) return -KEY_le;
5306 if (strEQ(d,"lc")) return -KEY_lc;
5309 if (strEQ(d,"log")) return -KEY_log;
5312 if (strEQ(d,"last")) return KEY_last;
5313 if (strEQ(d,"link")) return -KEY_link;
5314 if (strEQ(d,"lock")) return -KEY_lock;
5317 if (strEQ(d,"local")) return KEY_local;
5318 if (strEQ(d,"lstat")) return -KEY_lstat;
5321 if (strEQ(d,"length")) return -KEY_length;
5322 if (strEQ(d,"listen")) return -KEY_listen;
5325 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5328 if (strEQ(d,"localtime")) return -KEY_localtime;
5334 case 1: return KEY_m;
5336 if (strEQ(d,"my")) return KEY_my;
5339 if (strEQ(d,"map")) return KEY_map;
5342 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5345 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5346 if (strEQ(d,"msgget")) return -KEY_msgget;
5347 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5348 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5353 if (strEQ(d,"NE")) { deprecate(d); return -KEY_ne;}
5356 if (strEQ(d,"next")) return KEY_next;
5357 if (strEQ(d,"ne")) return -KEY_ne;
5358 if (strEQ(d,"not")) return -KEY_not;
5359 if (strEQ(d,"no")) return KEY_no;
5364 if (strEQ(d,"or")) return -KEY_or;
5367 if (strEQ(d,"ord")) return -KEY_ord;
5368 if (strEQ(d,"oct")) return -KEY_oct;
5369 if (strEQ(d,"our")) return KEY_our;
5372 if (strEQ(d,"open")) return -KEY_open;
5375 if (strEQ(d,"opendir")) return -KEY_opendir;
5382 if (strEQ(d,"pop")) return KEY_pop;
5383 if (strEQ(d,"pos")) return KEY_pos;
5386 if (strEQ(d,"push")) return KEY_push;
5387 if (strEQ(d,"pack")) return -KEY_pack;
5388 if (strEQ(d,"pipe")) return -KEY_pipe;
5391 if (strEQ(d,"print")) return KEY_print;
5394 if (strEQ(d,"printf")) return KEY_printf;
5397 if (strEQ(d,"package")) return KEY_package;
5400 if (strEQ(d,"prototype")) return KEY_prototype;
5405 if (strEQ(d,"q")) return KEY_q;
5406 if (strEQ(d,"qr")) return KEY_qr;
5407 if (strEQ(d,"qq")) return KEY_qq;
5408 if (strEQ(d,"qw")) return KEY_qw;
5409 if (strEQ(d,"qx")) return KEY_qx;
5411 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5416 if (strEQ(d,"ref")) return -KEY_ref;
5419 if (strEQ(d,"read")) return -KEY_read;
5420 if (strEQ(d,"rand")) return -KEY_rand;
5421 if (strEQ(d,"recv")) return -KEY_recv;
5422 if (strEQ(d,"redo")) return KEY_redo;
5425 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5426 if (strEQ(d,"reset")) return -KEY_reset;
5429 if (strEQ(d,"return")) return KEY_return;
5430 if (strEQ(d,"rename")) return -KEY_rename;
5431 if (strEQ(d,"rindex")) return -KEY_rindex;
5434 if (strEQ(d,"require")) return -KEY_require;
5435 if (strEQ(d,"reverse")) return -KEY_reverse;
5436 if (strEQ(d,"readdir")) return -KEY_readdir;
5439 if (strEQ(d,"readlink")) return -KEY_readlink;
5440 if (strEQ(d,"readline")) return -KEY_readline;
5441 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5444 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5450 case 0: return KEY_s;
5452 if (strEQ(d,"scalar")) return KEY_scalar;
5457 if (strEQ(d,"seek")) return -KEY_seek;
5458 if (strEQ(d,"send")) return -KEY_send;
5461 if (strEQ(d,"semop")) return -KEY_semop;
5464 if (strEQ(d,"select")) return -KEY_select;
5465 if (strEQ(d,"semctl")) return -KEY_semctl;
5466 if (strEQ(d,"semget")) return -KEY_semget;
5469 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5470 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5473 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5474 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5477 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5480 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5481 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5482 if (strEQ(d,"setservent")) return -KEY_setservent;
5485 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5486 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5493 if (strEQ(d,"shift")) return KEY_shift;
5496 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5497 if (strEQ(d,"shmget")) return -KEY_shmget;
5500 if (strEQ(d,"shmread")) return -KEY_shmread;
5503 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5504 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5509 if (strEQ(d,"sin")) return -KEY_sin;
5512 if (strEQ(d,"sleep")) return -KEY_sleep;
5515 if (strEQ(d,"sort")) return KEY_sort;
5516 if (strEQ(d,"socket")) return -KEY_socket;
5517 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5520 if (strEQ(d,"split")) return KEY_split;
5521 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5522 if (strEQ(d,"splice")) return KEY_splice;
5525 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5528 if (strEQ(d,"srand")) return -KEY_srand;
5531 if (strEQ(d,"stat")) return -KEY_stat;
5532 if (strEQ(d,"study")) return KEY_study;
5535 if (strEQ(d,"substr")) return -KEY_substr;
5536 if (strEQ(d,"sub")) return KEY_sub;
5541 if (strEQ(d,"system")) return -KEY_system;
5544 if (strEQ(d,"symlink")) return -KEY_symlink;
5545 if (strEQ(d,"syscall")) return -KEY_syscall;
5546 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5547 if (strEQ(d,"sysread")) return -KEY_sysread;
5548 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5551 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5560 if (strEQ(d,"tr")) return KEY_tr;
5563 if (strEQ(d,"tie")) return KEY_tie;
5566 if (strEQ(d,"tell")) return -KEY_tell;
5567 if (strEQ(d,"tied")) return KEY_tied;
5568 if (strEQ(d,"time")) return -KEY_time;
5571 if (strEQ(d,"times")) return -KEY_times;
5574 if (strEQ(d,"telldir")) return -KEY_telldir;
5577 if (strEQ(d,"truncate")) return -KEY_truncate;
5584 if (strEQ(d,"uc")) return -KEY_uc;
5587 if (strEQ(d,"use")) return KEY_use;
5590 if (strEQ(d,"undef")) return KEY_undef;
5591 if (strEQ(d,"until")) return KEY_until;
5592 if (strEQ(d,"untie")) return KEY_untie;
5593 if (strEQ(d,"utime")) return -KEY_utime;
5594 if (strEQ(d,"umask")) return -KEY_umask;
5597 if (strEQ(d,"unless")) return KEY_unless;
5598 if (strEQ(d,"unpack")) return -KEY_unpack;
5599 if (strEQ(d,"unlink")) return -KEY_unlink;
5602 if (strEQ(d,"unshift")) return KEY_unshift;
5603 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5608 if (strEQ(d,"values")) return -KEY_values;
5609 if (strEQ(d,"vec")) return -KEY_vec;
5614 if (strEQ(d,"warn")) return -KEY_warn;
5615 if (strEQ(d,"wait")) return -KEY_wait;
5618 if (strEQ(d,"while")) return KEY_while;
5619 if (strEQ(d,"write")) return -KEY_write;
5622 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5625 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5630 if (len == 1) return -KEY_x;
5631 if (strEQ(d,"xor")) return -KEY_xor;
5634 if (len == 1) return KEY_y;
5643 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5647 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5648 dTHR; /* only for ckWARN */
5649 if (ckWARN(WARN_SYNTAX)) {
5651 for (w = s+2; *w && level; w++) {
5658 for (; *w && isSPACE(*w); w++) ;
5659 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5660 Perl_warner(aTHX_ WARN_SYNTAX,
5661 "%s (...) interpreted as function",name);
5664 while (s < PL_bufend && isSPACE(*s))
5668 while (s < PL_bufend && isSPACE(*s))
5670 if (isIDFIRST_lazy_if(s,UTF)) {
5672 while (isALNUM_lazy_if(s,UTF))
5674 while (s < PL_bufend && isSPACE(*s))
5679 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5683 Perl_croak(aTHX_ "No comma allowed after %s", what);
5688 /* Either returns sv, or mortalizes sv and returns a new SV*.
5689 Best used as sv=new_constant(..., sv, ...).
5690 If s, pv are NULL, calls subroutine with one argument,
5691 and type is used with error messages only. */
5694 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5698 HV *table = GvHV(PL_hintgv); /* ^H */
5702 const char *why1, *why2, *why3;
5704 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
5707 why1 = "%^H is not consistent";
5708 why2 = strEQ(key,"charnames")
5709 ? " (missing \"use charnames ...\"?)"
5713 msg = Perl_newSVpvf(aTHX_ "constant(%s): %s%s%s",
5714 (type ? type: "undef"), why1, why2, why3);
5715 yyerror(SvPVX(msg));
5719 cvp = hv_fetch(table, key, strlen(key), FALSE);
5720 if (!cvp || !SvOK(*cvp)) {
5723 why3 = "} is not defined";
5726 sv_2mortal(sv); /* Parent created it permanently */
5729 pv = sv_2mortal(newSVpvn(s, len));
5731 typesv = sv_2mortal(newSVpv(type, 0));
5733 typesv = &PL_sv_undef;
5735 PUSHSTACKi(PERLSI_OVERLOAD);
5748 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5752 /* Check the eval first */
5753 if (!PL_in_eval && SvTRUE(ERRSV)) {
5755 sv_catpv(ERRSV, "Propagated");
5756 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5758 res = SvREFCNT_inc(sv);
5762 (void)SvREFCNT_inc(res);
5771 why1 = "Call to &{$^H{";
5773 why3 = "}} did not return a defined value";
5782 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5784 register char *d = dest;
5785 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5788 Perl_croak(aTHX_ ident_too_long);
5789 if (isALNUM(*s)) /* UTF handled below */
5791 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
5796 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5800 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5801 char *t = s + UTF8SKIP(s);
5802 while (*t & 0x80 && is_utf8_mark((U8*)t))
5804 if (d + (t - s) > e)
5805 Perl_croak(aTHX_ ident_too_long);
5806 Copy(s, d, t - s, char);
5819 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5829 e = d + destlen - 3; /* two-character token, ending NUL */
5831 while (isDIGIT(*s)) {
5833 Perl_croak(aTHX_ ident_too_long);
5840 Perl_croak(aTHX_ ident_too_long);
5841 if (isALNUM(*s)) /* UTF handled below */
5843 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
5848 else if (*s == ':' && s[1] == ':') {
5852 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5853 char *t = s + UTF8SKIP(s);
5854 while (*t & 0x80 && is_utf8_mark((U8*)t))
5856 if (d + (t - s) > e)
5857 Perl_croak(aTHX_ ident_too_long);
5858 Copy(s, d, t - s, char);
5869 if (PL_lex_state != LEX_NORMAL)
5870 PL_lex_state = LEX_INTERPENDMAYBE;
5873 if (*s == '$' && s[1] &&
5874 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
5887 if (*d == '^' && *s && isCONTROLVAR(*s)) {
5892 if (isSPACE(s[-1])) {
5895 if (!SPACE_OR_TAB(ch)) {
5901 if (isIDFIRST_lazy_if(d,UTF)) {
5905 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
5907 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
5910 Copy(s, d, e - s, char);
5915 while ((isALNUM(*s) || *s == ':') && d < e)
5918 Perl_croak(aTHX_ ident_too_long);
5921 while (s < send && SPACE_OR_TAB(*s)) s++;
5922 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
5923 dTHR; /* only for ckWARN */
5924 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
5925 const char *brack = *s == '[' ? "[...]" : "{...}";
5926 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5927 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
5928 funny, dest, brack, funny, dest, brack);
5931 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
5935 /* Handle extended ${^Foo} variables
5936 * 1999-02-27 mjd-perl-patch@plover.com */
5937 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
5941 while (isALNUM(*s) && d < e) {
5945 Perl_croak(aTHX_ ident_too_long);
5950 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
5951 PL_lex_state = LEX_INTERPEND;
5954 if (PL_lex_state == LEX_NORMAL) {
5955 dTHR; /* only for ckWARN */
5956 if (ckWARN(WARN_AMBIGUOUS) &&
5957 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
5959 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5960 "Ambiguous use of %c{%s} resolved to %c%s",
5961 funny, dest, funny, dest);
5966 s = bracket; /* let the parser handle it */
5970 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
5971 PL_lex_state = LEX_INTERPEND;
5976 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
5981 *pmfl |= PMf_GLOBAL;
5983 *pmfl |= PMf_CONTINUE;
5987 *pmfl |= PMf_MULTILINE;
5989 *pmfl |= PMf_SINGLELINE;
5991 *pmfl |= PMf_EXTENDED;
5995 S_scan_pat(pTHX_ char *start, I32 type)
6000 s = scan_str(start,FALSE,FALSE);
6003 SvREFCNT_dec(PL_lex_stuff);
6004 PL_lex_stuff = Nullsv;
6005 Perl_croak(aTHX_ "Search pattern not terminated");
6008 pm = (PMOP*)newPMOP(type, 0);
6009 if (PL_multi_open == '?')
6010 pm->op_pmflags |= PMf_ONCE;
6012 while (*s && strchr("iomsx", *s))
6013 pmflag(&pm->op_pmflags,*s++);
6016 while (*s && strchr("iogcmsx", *s))
6017 pmflag(&pm->op_pmflags,*s++);
6019 pm->op_pmpermflags = pm->op_pmflags;
6021 PL_lex_op = (OP*)pm;
6022 yylval.ival = OP_MATCH;
6027 S_scan_subst(pTHX_ char *start)
6034 yylval.ival = OP_NULL;
6036 s = scan_str(start,FALSE,FALSE);
6040 SvREFCNT_dec(PL_lex_stuff);
6041 PL_lex_stuff = Nullsv;
6042 Perl_croak(aTHX_ "Substitution pattern not terminated");
6045 if (s[-1] == PL_multi_open)
6048 first_start = PL_multi_start;
6049 s = scan_str(s,FALSE,FALSE);
6052 SvREFCNT_dec(PL_lex_stuff);
6053 PL_lex_stuff = Nullsv;
6055 SvREFCNT_dec(PL_lex_repl);
6056 PL_lex_repl = Nullsv;
6057 Perl_croak(aTHX_ "Substitution replacement not terminated");
6059 PL_multi_start = first_start; /* so whole substitution is taken together */
6061 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6067 else if (strchr("iogcmsx", *s))
6068 pmflag(&pm->op_pmflags,*s++);
6075 PL_sublex_info.super_bufptr = s;
6076 PL_sublex_info.super_bufend = PL_bufend;
6078 pm->op_pmflags |= PMf_EVAL;
6079 repl = newSVpvn("",0);
6081 sv_catpv(repl, es ? "eval " : "do ");
6082 sv_catpvn(repl, "{ ", 2);
6083 sv_catsv(repl, PL_lex_repl);
6084 sv_catpvn(repl, " };", 2);
6086 SvREFCNT_dec(PL_lex_repl);
6090 pm->op_pmpermflags = pm->op_pmflags;
6091 PL_lex_op = (OP*)pm;
6092 yylval.ival = OP_SUBST;
6097 S_scan_trans(pTHX_ char *start)
6108 yylval.ival = OP_NULL;
6110 s = scan_str(start,FALSE,FALSE);
6113 SvREFCNT_dec(PL_lex_stuff);
6114 PL_lex_stuff = Nullsv;
6115 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6117 if (s[-1] == PL_multi_open)
6120 s = scan_str(s,FALSE,FALSE);
6123 SvREFCNT_dec(PL_lex_stuff);
6124 PL_lex_stuff = Nullsv;
6126 SvREFCNT_dec(PL_lex_repl);
6127 PL_lex_repl = Nullsv;
6128 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6132 o = newSVOP(OP_TRANS, 0, 0);
6133 utf8 = OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF;
6136 New(803,tbl,256,short);
6137 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6141 complement = del = squash = 0;
6142 while (strchr("cdsCU", *s)) {
6144 complement = OPpTRANS_COMPLEMENT;
6146 del = OPpTRANS_DELETE;
6148 squash = OPpTRANS_SQUASH;
6153 utf8 &= ~OPpTRANS_FROM_UTF;
6155 utf8 |= OPpTRANS_FROM_UTF;
6159 utf8 &= ~OPpTRANS_TO_UTF;
6161 utf8 |= OPpTRANS_TO_UTF;
6164 Perl_croak(aTHX_ "Too many /C and /U options");
6169 o->op_private = del|squash|complement|utf8;
6172 yylval.ival = OP_TRANS;
6177 S_scan_heredoc(pTHX_ register char *s)
6181 I32 op_type = OP_SCALAR;
6188 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6192 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6195 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6196 if (*peek && strchr("`'\"",*peek)) {
6199 s = delimcpy(d, e, s, PL_bufend, term, &len);
6209 if (!isALNUM_lazy_if(s,UTF))
6210 deprecate("bare << to mean <<\"\"");
6211 for (; isALNUM_lazy_if(s,UTF); s++) {
6216 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6217 Perl_croak(aTHX_ "Delimiter for here document is too long");
6220 len = d - PL_tokenbuf;
6221 #ifndef PERL_STRICT_CR
6222 d = strchr(s, '\r');
6226 while (s < PL_bufend) {
6232 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6241 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6246 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6247 herewas = newSVpvn(s,PL_bufend-s);
6249 s--, herewas = newSVpvn(s,d-s);
6250 s += SvCUR(herewas);
6252 tmpstr = NEWSV(87,79);
6253 sv_upgrade(tmpstr, SVt_PVIV);
6258 else if (term == '`') {
6259 op_type = OP_BACKTICK;
6260 SvIVX(tmpstr) = '\\';
6264 PL_multi_start = CopLINE(PL_curcop);
6265 PL_multi_open = PL_multi_close = '<';
6266 term = *PL_tokenbuf;
6267 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6268 char *bufptr = PL_sublex_info.super_bufptr;
6269 char *bufend = PL_sublex_info.super_bufend;
6270 char *olds = s - SvCUR(herewas);
6271 s = strchr(bufptr, '\n');
6275 while (s < bufend &&
6276 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6278 CopLINE_inc(PL_curcop);
6281 CopLINE_set(PL_curcop, PL_multi_start);
6282 missingterm(PL_tokenbuf);
6284 sv_setpvn(herewas,bufptr,d-bufptr+1);
6285 sv_setpvn(tmpstr,d+1,s-d);
6287 sv_catpvn(herewas,s,bufend-s);
6288 (void)strcpy(bufptr,SvPVX(herewas));
6295 while (s < PL_bufend &&
6296 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6298 CopLINE_inc(PL_curcop);
6300 if (s >= PL_bufend) {
6301 CopLINE_set(PL_curcop, PL_multi_start);
6302 missingterm(PL_tokenbuf);
6304 sv_setpvn(tmpstr,d+1,s-d);
6306 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6308 sv_catpvn(herewas,s,PL_bufend-s);
6309 sv_setsv(PL_linestr,herewas);
6310 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6311 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6314 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6315 while (s >= PL_bufend) { /* multiple line string? */
6317 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6318 CopLINE_set(PL_curcop, PL_multi_start);
6319 missingterm(PL_tokenbuf);
6321 CopLINE_inc(PL_curcop);
6322 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6323 #ifndef PERL_STRICT_CR
6324 if (PL_bufend - PL_linestart >= 2) {
6325 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6326 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6328 PL_bufend[-2] = '\n';
6330 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6332 else if (PL_bufend[-1] == '\r')
6333 PL_bufend[-1] = '\n';
6335 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6336 PL_bufend[-1] = '\n';
6338 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6339 SV *sv = NEWSV(88,0);
6341 sv_upgrade(sv, SVt_PVMG);
6342 sv_setsv(sv,PL_linestr);
6343 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6345 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6348 sv_catsv(PL_linestr,herewas);
6349 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6353 sv_catsv(tmpstr,PL_linestr);
6358 PL_multi_end = CopLINE(PL_curcop);
6359 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6360 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6361 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6363 SvREFCNT_dec(herewas);
6364 PL_lex_stuff = tmpstr;
6365 yylval.ival = op_type;
6370 takes: current position in input buffer
6371 returns: new position in input buffer
6372 side-effects: yylval and lex_op are set.
6377 <FH> read from filehandle
6378 <pkg::FH> read from package qualified filehandle
6379 <pkg'FH> read from package qualified filehandle
6380 <$fh> read from filehandle in $fh
6386 S_scan_inputsymbol(pTHX_ char *start)
6388 register char *s = start; /* current position in buffer */
6394 d = PL_tokenbuf; /* start of temp holding space */
6395 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6396 end = strchr(s, '\n');
6399 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6401 /* die if we didn't have space for the contents of the <>,
6402 or if it didn't end, or if we see a newline
6405 if (len >= sizeof PL_tokenbuf)
6406 Perl_croak(aTHX_ "Excessively long <> operator");
6408 Perl_croak(aTHX_ "Unterminated <> operator");
6413 Remember, only scalar variables are interpreted as filehandles by
6414 this code. Anything more complex (e.g., <$fh{$num}>) will be
6415 treated as a glob() call.
6416 This code makes use of the fact that except for the $ at the front,
6417 a scalar variable and a filehandle look the same.
6419 if (*d == '$' && d[1]) d++;
6421 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6422 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6425 /* If we've tried to read what we allow filehandles to look like, and
6426 there's still text left, then it must be a glob() and not a getline.
6427 Use scan_str to pull out the stuff between the <> and treat it
6428 as nothing more than a string.
6431 if (d - PL_tokenbuf != len) {
6432 yylval.ival = OP_GLOB;
6434 s = scan_str(start,FALSE,FALSE);
6436 Perl_croak(aTHX_ "Glob not terminated");
6440 /* we're in a filehandle read situation */
6443 /* turn <> into <ARGV> */
6445 (void)strcpy(d,"ARGV");
6447 /* if <$fh>, create the ops to turn the variable into a
6453 /* try to find it in the pad for this block, otherwise find
6454 add symbol table ops
6456 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6457 OP *o = newOP(OP_PADSV, 0);
6459 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6462 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6463 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6464 newUNOP(OP_RV2SV, 0,
6465 newGVOP(OP_GV, 0, gv)));
6467 PL_lex_op->op_flags |= OPf_SPECIAL;
6468 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6469 yylval.ival = OP_NULL;
6472 /* If it's none of the above, it must be a literal filehandle
6473 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6475 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6476 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6477 yylval.ival = OP_NULL;
6486 takes: start position in buffer
6487 keep_quoted preserve \ on the embedded delimiter(s)
6488 keep_delims preserve the delimiters around the string
6489 returns: position to continue reading from buffer
6490 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6491 updates the read buffer.
6493 This subroutine pulls a string out of the input. It is called for:
6494 q single quotes q(literal text)
6495 ' single quotes 'literal text'
6496 qq double quotes qq(interpolate $here please)
6497 " double quotes "interpolate $here please"
6498 qx backticks qx(/bin/ls -l)
6499 ` backticks `/bin/ls -l`
6500 qw quote words @EXPORT_OK = qw( func() $spam )
6501 m// regexp match m/this/
6502 s/// regexp substitute s/this/that/
6503 tr/// string transliterate tr/this/that/
6504 y/// string transliterate y/this/that/
6505 ($*@) sub prototypes sub foo ($)
6506 (stuff) sub attr parameters sub foo : attr(stuff)
6507 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6509 In most of these cases (all but <>, patterns and transliterate)
6510 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6511 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6512 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6515 It skips whitespace before the string starts, and treats the first
6516 character as the delimiter. If the delimiter is one of ([{< then
6517 the corresponding "close" character )]}> is used as the closing
6518 delimiter. It allows quoting of delimiters, and if the string has
6519 balanced delimiters ([{<>}]) it allows nesting.
6521 The lexer always reads these strings into lex_stuff, except in the
6522 case of the operators which take *two* arguments (s/// and tr///)
6523 when it checks to see if lex_stuff is full (presumably with the 1st
6524 arg to s or tr) and if so puts the string into lex_repl.
6529 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6532 SV *sv; /* scalar value: string */
6533 char *tmps; /* temp string, used for delimiter matching */
6534 register char *s = start; /* current position in the buffer */
6535 register char term; /* terminating character */
6536 register char *to; /* current position in the sv's data */
6537 I32 brackets = 1; /* bracket nesting level */
6538 bool has_utf = FALSE; /* is there any utf8 content? */
6540 /* skip space before the delimiter */
6544 /* mark where we are, in case we need to report errors */
6547 /* after skipping whitespace, the next character is the terminator */
6549 if ((term & 0x80) && UTF)
6552 /* mark where we are */
6553 PL_multi_start = CopLINE(PL_curcop);
6554 PL_multi_open = term;
6556 /* find corresponding closing delimiter */
6557 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6559 PL_multi_close = term;
6561 /* create a new SV to hold the contents. 87 is leak category, I'm
6562 assuming. 79 is the SV's initial length. What a random number. */
6564 sv_upgrade(sv, SVt_PVIV);
6566 (void)SvPOK_only(sv); /* validate pointer */
6568 /* move past delimiter and try to read a complete string */
6570 sv_catpvn(sv, s, 1);
6573 /* extend sv if need be */
6574 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6575 /* set 'to' to the next character in the sv's string */
6576 to = SvPVX(sv)+SvCUR(sv);
6578 /* if open delimiter is the close delimiter read unbridle */
6579 if (PL_multi_open == PL_multi_close) {
6580 for (; s < PL_bufend; s++,to++) {
6581 /* embedded newlines increment the current line number */
6582 if (*s == '\n' && !PL_rsfp)
6583 CopLINE_inc(PL_curcop);
6584 /* handle quoted delimiters */
6585 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6586 if (!keep_quoted && s[1] == term)
6588 /* any other quotes are simply copied straight through */
6592 /* terminate when run out of buffer (the for() condition), or
6593 have found the terminator */
6594 else if (*s == term)
6596 else if (!has_utf && (*s & 0x80) && UTF)
6602 /* if the terminator isn't the same as the start character (e.g.,
6603 matched brackets), we have to allow more in the quoting, and
6604 be prepared for nested brackets.
6607 /* read until we run out of string, or we find the terminator */
6608 for (; s < PL_bufend; s++,to++) {
6609 /* embedded newlines increment the line count */
6610 if (*s == '\n' && !PL_rsfp)
6611 CopLINE_inc(PL_curcop);
6612 /* backslashes can escape the open or closing characters */
6613 if (*s == '\\' && s+1 < PL_bufend) {
6615 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6620 /* allow nested opens and closes */
6621 else if (*s == PL_multi_close && --brackets <= 0)
6623 else if (*s == PL_multi_open)
6625 else if (!has_utf && (*s & 0x80) && UTF)
6630 /* terminate the copied string and update the sv's end-of-string */
6632 SvCUR_set(sv, to - SvPVX(sv));
6635 * this next chunk reads more into the buffer if we're not done yet
6639 break; /* handle case where we are done yet :-) */
6641 #ifndef PERL_STRICT_CR
6642 if (to - SvPVX(sv) >= 2) {
6643 if ((to[-2] == '\r' && to[-1] == '\n') ||
6644 (to[-2] == '\n' && to[-1] == '\r'))
6648 SvCUR_set(sv, to - SvPVX(sv));
6650 else if (to[-1] == '\r')
6653 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6657 /* if we're out of file, or a read fails, bail and reset the current
6658 line marker so we can report where the unterminated string began
6661 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6663 CopLINE_set(PL_curcop, PL_multi_start);
6666 /* we read a line, so increment our line counter */
6667 CopLINE_inc(PL_curcop);
6669 /* update debugger info */
6670 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6671 SV *sv = NEWSV(88,0);
6673 sv_upgrade(sv, SVt_PVMG);
6674 sv_setsv(sv,PL_linestr);
6675 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6678 /* having changed the buffer, we must update PL_bufend */
6679 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6682 /* at this point, we have successfully read the delimited string */
6685 sv_catpvn(sv, s, 1);
6688 PL_multi_end = CopLINE(PL_curcop);
6691 /* if we allocated too much space, give some back */
6692 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6693 SvLEN_set(sv, SvCUR(sv) + 1);
6694 Renew(SvPVX(sv), SvLEN(sv), char);
6697 /* decide whether this is the first or second quoted string we've read
6710 takes: pointer to position in buffer
6711 returns: pointer to new position in buffer
6712 side-effects: builds ops for the constant in yylval.op
6714 Read a number in any of the formats that Perl accepts:
6716 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6717 [\d_]+(\.[\d_]*)?[Ee](\d+)
6719 Underbars (_) are allowed in decimal numbers. If -w is on,
6720 underbars before a decimal point must be at three digit intervals.
6722 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6725 If it reads a number without a decimal point or an exponent, it will
6726 try converting the number to an integer and see if it can do so
6727 without loss of precision.
6731 Perl_scan_num(pTHX_ char *start)
6733 register char *s = start; /* current position in buffer */
6734 register char *d; /* destination in temp buffer */
6735 register char *e; /* end of temp buffer */
6736 NV nv; /* number read, as a double */
6737 SV *sv = Nullsv; /* place to put the converted number */
6738 bool floatit; /* boolean: int or float? */
6739 char *lastub = 0; /* position of last underbar */
6740 static char number_too_long[] = "Number too long";
6742 /* We use the first character to decide what type of number this is */
6746 Perl_croak(aTHX_ "panic: scan_num");
6748 /* if it starts with a 0, it could be an octal number, a decimal in
6749 0.13 disguise, or a hexadecimal number, or a binary number. */
6753 u holds the "number so far"
6754 shift the power of 2 of the base
6755 (hex == 4, octal == 3, binary == 1)
6756 overflowed was the number more than we can hold?
6758 Shift is used when we add a digit. It also serves as an "are
6759 we in octal/hex/binary?" indicator to disallow hex characters
6766 bool overflowed = FALSE;
6767 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6768 static char* bases[5] = { "", "binary", "", "octal",
6770 static char* Bases[5] = { "", "Binary", "", "Octal",
6772 static char *maxima[5] = { "",
6773 "0b11111111111111111111111111111111",
6777 char *base, *Base, *max;
6783 } else if (s[1] == 'b') {
6787 /* check for a decimal in disguise */
6788 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6790 /* so it must be octal */
6794 base = bases[shift];
6795 Base = Bases[shift];
6796 max = maxima[shift];
6798 /* read the rest of the number */
6800 /* x is used in the overflow test,
6801 b is the digit we're adding on. */
6806 /* if we don't mention it, we're done */
6815 /* 8 and 9 are not octal */
6818 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6822 case '2': case '3': case '4':
6823 case '5': case '6': case '7':
6825 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6829 b = *s++ & 15; /* ASCII digit -> value of digit */
6833 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6834 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6835 /* make sure they said 0x */
6840 /* Prepare to put the digit we have onto the end
6841 of the number so far. We check for overflows.
6846 x = u << shift; /* make room for the digit */
6848 if ((x >> shift) != u
6849 && !(PL_hints & HINT_NEW_BINARY)) {
6853 if (ckWARN_d(WARN_OVERFLOW))
6854 Perl_warner(aTHX_ WARN_OVERFLOW,
6855 "Integer overflow in %s number",
6858 u = x | b; /* add the digit to the end */
6861 n *= nvshift[shift];
6862 /* If an NV has not enough bits in its
6863 * mantissa to represent an UV this summing of
6864 * small low-order numbers is a waste of time
6865 * (because the NV cannot preserve the
6866 * low-order bits anyway): we could just
6867 * remember when did we overflow and in the
6868 * end just multiply n by the right
6876 /* if we get here, we had success: make a scalar value from
6883 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6884 Perl_warner(aTHX_ WARN_PORTABLE,
6885 "%s number > %s non-portable",
6892 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6893 Perl_warner(aTHX_ WARN_PORTABLE,
6894 "%s number > %s non-portable",
6899 if (PL_hints & HINT_NEW_BINARY)
6900 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6905 handle decimal numbers.
6906 we're also sent here when we read a 0 as the first digit
6908 case '1': case '2': case '3': case '4': case '5':
6909 case '6': case '7': case '8': case '9': case '.':
6912 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
6915 /* read next group of digits and _ and copy into d */
6916 while (isDIGIT(*s) || *s == '_') {
6917 /* skip underscores, checking for misplaced ones
6921 dTHR; /* only for ckWARN */
6922 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
6923 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6927 /* check for end of fixed-length buffer */
6929 Perl_croak(aTHX_ number_too_long);
6930 /* if we're ok, copy the character */
6935 /* final misplaced underbar check */
6936 if (lastub && s - lastub != 3) {
6938 if (ckWARN(WARN_SYNTAX))
6939 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6942 /* read a decimal portion if there is one. avoid
6943 3..5 being interpreted as the number 3. followed
6946 if (*s == '.' && s[1] != '.') {
6950 /* copy, ignoring underbars, until we run out of
6951 digits. Note: no misplaced underbar checks!
6953 for (; isDIGIT(*s) || *s == '_'; s++) {
6954 /* fixed length buffer check */
6956 Perl_croak(aTHX_ number_too_long);
6960 if (*s == '.' && isDIGIT(s[1])) {
6961 /* oops, it's really a v-string, but without the "v" */
6967 /* read exponent part, if present */
6968 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
6972 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
6973 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
6975 /* allow positive or negative exponent */
6976 if (*s == '+' || *s == '-')
6979 /* read digits of exponent (no underbars :-) */
6980 while (isDIGIT(*s)) {
6982 Perl_croak(aTHX_ number_too_long);
6987 /* terminate the string */
6990 /* make an sv from the string */
6993 #if defined(Strtol) && defined(Strtoul)
6996 strtol/strtoll sets errno to ERANGE if the number is too big
6997 for an integer. We try to do an integer conversion first
6998 if no characters indicating "float" have been found.
7005 if (*PL_tokenbuf == '-')
7006 iv = Strtol(PL_tokenbuf, (char**)NULL, 10);
7008 uv = Strtoul(PL_tokenbuf, (char**)NULL, 10);
7010 floatit = TRUE; /* Probably just too large. */
7011 else if (*PL_tokenbuf == '-')
7013 else if (uv <= IV_MAX)
7014 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7019 nv = Atof(PL_tokenbuf);
7024 No working strtou?ll?.
7026 Unfortunately atol() doesn't do range checks (returning
7027 LONG_MIN/LONG_MAX, and setting errno to ERANGE on overflows)
7028 everywhere [1], so we cannot use use atol() (or atoll()).
7029 If we could, they would be used, as Atol(), very much like
7030 Strtol() and Strtoul() are used above.
7032 [1] XXX Configure test needed to check for atol()
7033 (and atoll()) overflow behaviour XXX
7037 We need to do this the hard way. */
7039 nv = Atof(PL_tokenbuf);
7041 /* See if we can make do with an integer value without loss of
7042 precision. We use U_V to cast to a UV, because some
7043 compilers have issues. Then we try casting it back and see
7044 if it was the same [1]. We only do this if we know we
7045 specifically read an integer. If floatit is true, then we
7046 don't need to do the conversion at all.
7048 [1] Note that this is lossy if our NVs cannot preserve our
7049 UVs. There are metaconfig defines NV_PRESERVES_UV (a boolean)
7050 and NV_PRESERVES_UV_BITS (a number), but in general we really
7051 do hope all such potentially lossy platforms have strtou?ll?
7052 to do a lossless IV/UV conversion.
7054 Maybe could do some tricks with DBL_DIG, LDBL_DIG and
7055 DBL_MANT_DIG and LDBL_MANT_DIG (these are already available
7056 as NV_DIG and NV_MANT_DIG)?
7062 if (!floatit && (NV)uv == nv) {
7064 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7072 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7073 (PL_hints & HINT_NEW_INTEGER) )
7074 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7075 (floatit ? "float" : "integer"),
7079 /* if it starts with a v, it could be a v-string */
7085 while (isDIGIT(*pos) || *pos == '_')
7087 if (!isALPHA(*pos)) {
7089 U8 tmpbuf[UTF8_MAXLEN];
7092 s++; /* get past 'v' */
7095 sv_setpvn(sv, "", 0);
7098 if (*s == '0' && isDIGIT(s[1]))
7099 yyerror("Octal number in vector unsupported");
7102 /* this is atoi() that tolerates underscores */
7105 while (--end >= s) {
7110 rev += (*end - '0') * mult;
7112 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
7113 Perl_warner(aTHX_ WARN_OVERFLOW,
7114 "Integer overflow in decimal number");
7117 tmpend = uv_to_utf8(tmpbuf, rev);
7118 utf8 = utf8 || rev > 127;
7119 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
7120 if (*pos == '.' && isDIGIT(pos[1]))
7126 while (isDIGIT(*pos) || *pos == '_')
7134 sv_utf8_downgrade(sv, TRUE);
7141 /* make the op for the constant and return */
7144 yylval.opval = newSVOP(OP_CONST, 0, sv);
7146 yylval.opval = Nullop;
7152 S_scan_formline(pTHX_ register char *s)
7157 SV *stuff = newSVpvn("",0);
7158 bool needargs = FALSE;
7161 if (*s == '.' || *s == '}') {
7163 #ifdef PERL_STRICT_CR
7164 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7166 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7168 if (*t == '\n' || t == PL_bufend)
7171 if (PL_in_eval && !PL_rsfp) {
7172 eol = strchr(s,'\n');
7177 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7179 for (t = s; t < eol; t++) {
7180 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7182 goto enough; /* ~~ must be first line in formline */
7184 if (*t == '@' || *t == '^')
7187 sv_catpvn(stuff, s, eol-s);
7188 #ifndef PERL_STRICT_CR
7189 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7190 char *end = SvPVX(stuff) + SvCUR(stuff);
7199 s = filter_gets(PL_linestr, PL_rsfp, 0);
7200 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7201 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7204 yyerror("Format not terminated");
7214 PL_lex_state = LEX_NORMAL;
7215 PL_nextval[PL_nexttoke].ival = 0;
7219 PL_lex_state = LEX_FORMLINE;
7220 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7222 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7226 SvREFCNT_dec(stuff);
7227 PL_lex_formbrack = 0;
7238 PL_cshlen = strlen(PL_cshname);
7243 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7246 I32 oldsavestack_ix = PL_savestack_ix;
7247 CV* outsidecv = PL_compcv;
7251 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7253 SAVEI32(PL_subline);
7254 save_item(PL_subname);
7257 SAVESPTR(PL_comppad_name);
7258 SAVESPTR(PL_compcv);
7259 SAVEI32(PL_comppad_name_fill);
7260 SAVEI32(PL_min_intro_pending);
7261 SAVEI32(PL_max_intro_pending);
7262 SAVEI32(PL_pad_reset_pending);
7264 PL_compcv = (CV*)NEWSV(1104,0);
7265 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7266 CvFLAGS(PL_compcv) |= flags;
7268 PL_comppad = newAV();
7269 av_push(PL_comppad, Nullsv);
7270 PL_curpad = AvARRAY(PL_comppad);
7271 PL_comppad_name = newAV();
7272 PL_comppad_name_fill = 0;
7273 PL_min_intro_pending = 0;
7275 PL_subline = CopLINE(PL_curcop);
7277 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7278 PL_curpad[0] = (SV*)newAV();
7279 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7280 #endif /* USE_THREADS */
7282 comppadlist = newAV();
7283 AvREAL_off(comppadlist);
7284 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7285 av_store(comppadlist, 1, (SV*)PL_comppad);
7287 CvPADLIST(PL_compcv) = comppadlist;
7288 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7290 CvOWNER(PL_compcv) = 0;
7291 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7292 MUTEX_INIT(CvMUTEXP(PL_compcv));
7293 #endif /* USE_THREADS */
7295 return oldsavestack_ix;
7299 Perl_yywarn(pTHX_ char *s)
7302 PL_in_eval |= EVAL_WARNONLY;
7304 PL_in_eval &= ~EVAL_WARNONLY;
7309 Perl_yyerror(pTHX_ char *s)
7313 char *context = NULL;
7317 if (!yychar || (yychar == ';' && !PL_rsfp))
7319 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7320 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7321 while (isSPACE(*PL_oldoldbufptr))
7323 context = PL_oldoldbufptr;
7324 contlen = PL_bufptr - PL_oldoldbufptr;
7326 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7327 PL_oldbufptr != PL_bufptr) {
7328 while (isSPACE(*PL_oldbufptr))
7330 context = PL_oldbufptr;
7331 contlen = PL_bufptr - PL_oldbufptr;
7333 else if (yychar > 255)
7334 where = "next token ???";
7335 #ifdef USE_PURE_BISON
7336 /* GNU Bison sets the value -2 */
7337 else if (yychar == -2) {
7339 else if ((yychar & 127) == 127) {
7341 if (PL_lex_state == LEX_NORMAL ||
7342 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7343 where = "at end of line";
7344 else if (PL_lex_inpat)
7345 where = "within pattern";
7347 where = "within string";
7350 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7352 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7353 else if (isPRINT_LC(yychar))
7354 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7356 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7357 where = SvPVX(where_sv);
7359 msg = sv_2mortal(newSVpv(s, 0));
7360 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7361 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7363 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7365 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7366 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7367 Perl_sv_catpvf(aTHX_ msg,
7368 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7369 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7372 if (PL_in_eval & EVAL_WARNONLY)
7373 Perl_warn(aTHX_ "%"SVf, msg);
7376 if (PL_error_count >= 10) {
7377 if (PL_in_eval && SvCUR(ERRSV))
7378 Perl_croak(aTHX_ "%_%s has too many errors.\n",
7379 ERRSV, CopFILE(PL_curcop));
7381 Perl_croak(aTHX_ "%s has too many errors.\n",
7382 CopFILE(PL_curcop));
7385 PL_in_my_stash = Nullhv;
7396 * Restore a source filter.
7400 restore_rsfp(pTHXo_ void *f)
7402 PerlIO *fp = (PerlIO*)f;
7404 if (PL_rsfp == PerlIO_stdin())
7405 PerlIO_clearerr(PL_rsfp);
7406 else if (PL_rsfp && (PL_rsfp != fp))
7407 PerlIO_close(PL_rsfp);