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);
31 #ifndef PERL_NO_UTF16_FILTER
32 static I32 utf16_textfilter(pTHXo_ int idx, SV *sv, int maxlen);
33 static I32 utf16rev_textfilter(pTHXo_ int idx, SV *sv, int maxlen);
36 #define XFAKEBRACK 128
39 /*#define UTF (SvUTF8(PL_linestr) && !(PL_hints & HINT_BYTE))*/
40 #define UTF (PL_hints & HINT_UTF8)
42 /* In variables name $^X, these are the legal values for X.
43 * 1999-02-27 mjd-perl-patch@plover.com */
44 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
46 /* On MacOS, respect nonbreaking spaces */
47 #ifdef MACOS_TRADITIONAL
48 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
50 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
53 /* LEX_* are values for PL_lex_state, the state of the lexer.
54 * They are arranged oddly so that the guard on the switch statement
55 * can get by with a single comparison (if the compiler is smart enough).
58 /* #define LEX_NOTPARSING 11 is done in perl.h. */
61 #define LEX_INTERPNORMAL 9
62 #define LEX_INTERPCASEMOD 8
63 #define LEX_INTERPPUSH 7
64 #define LEX_INTERPSTART 6
65 #define LEX_INTERPEND 5
66 #define LEX_INTERPENDMAYBE 4
67 #define LEX_INTERPCONCAT 3
68 #define LEX_INTERPCONST 2
69 #define LEX_FORMLINE 1
70 #define LEX_KNOWNEXT 0
72 /* XXX If this causes problems, set i_unistd=undef in the hint file. */
74 # include <unistd.h> /* Needed for execv() */
83 YYSTYPE* yylval_pointer = NULL;
84 int* yychar_pointer = NULL;
87 # define yylval (*yylval_pointer)
88 # define yychar (*yychar_pointer)
89 # define PERL_YYLEX_PARAM yylval_pointer,yychar_pointer
91 # define yylex() Perl_yylex(aTHX_ yylval_pointer, yychar_pointer)
96 /* CLINE is a macro that ensures PL_copline has a sane value */
101 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
104 * Convenience functions to return different tokens and prime the
105 * lexer for the next token. They all take an argument.
107 * TOKEN : generic token (used for '(', DOLSHARP, etc)
108 * OPERATOR : generic operator
109 * AOPERATOR : assignment operator
110 * PREBLOCK : beginning the block after an if, while, foreach, ...
111 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
112 * PREREF : *EXPR where EXPR is not a simple identifier
113 * TERM : expression term
114 * LOOPX : loop exiting command (goto, last, dump, etc)
115 * FTST : file test operator
116 * FUN0 : zero-argument function
117 * FUN1 : not used, except for not, which isn't a UNIOP
118 * BOop : bitwise or or xor
120 * SHop : shift operator
121 * PWop : power operator
122 * PMop : pattern-matching operator
123 * Aop : addition-level operator
124 * Mop : multiplication-level operator
125 * Eop : equality-testing operator
126 * Rop : relational operator <= != gt
128 * Also see LOP and lop() below.
131 #define TOKEN(retval) return (PL_bufptr = s,(int)retval)
132 #define OPERATOR(retval) return (PL_expect = XTERM,PL_bufptr = s,(int)retval)
133 #define AOPERATOR(retval) return ao((PL_expect = XTERM,PL_bufptr = s,(int)retval))
134 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
135 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
136 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s,(int)retval)
137 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR,PL_bufptr = s,(int)retval)
138 #define LOOPX(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
139 #define FTST(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)UNIOP)
140 #define FUN0(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
141 #define FUN1(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
142 #define BOop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
143 #define BAop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
144 #define SHop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
145 #define PWop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
146 #define PMop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
147 #define Aop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
148 #define Mop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
149 #define Eop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
150 #define Rop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)RELOP)
152 /* This bit of chicanery makes a unary function followed by
153 * a parenthesis into a function with one argument, highest precedence.
155 #define UNI(f) return(yylval.ival = f, \
158 PL_last_uni = PL_oldbufptr, \
159 PL_last_lop_op = f, \
160 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
162 #define UNIBRACK(f) return(yylval.ival = f, \
164 PL_last_uni = PL_oldbufptr, \
165 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
167 /* grandfather return to old style */
168 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
173 * This subroutine detects &&= and ||= and turns an ANDAND or OROR
174 * into an OP_ANDASSIGN or OP_ORASSIGN
178 S_ao(pTHX_ int toketype)
180 if (*PL_bufptr == '=') {
182 if (toketype == ANDAND)
183 yylval.ival = OP_ANDASSIGN;
184 else if (toketype == OROR)
185 yylval.ival = OP_ORASSIGN;
193 * When Perl expects an operator and finds something else, no_op
194 * prints the warning. It always prints "<something> found where
195 * operator expected. It prints "Missing semicolon on previous line?"
196 * if the surprise occurs at the start of the line. "do you need to
197 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
198 * where the compiler doesn't know if foo is a method call or a function.
199 * It prints "Missing operator before end of line" if there's nothing
200 * after the missing operator, or "... before <...>" if there is something
201 * after the missing operator.
205 S_no_op(pTHX_ char *what, char *s)
207 char *oldbp = PL_bufptr;
208 bool is_first = (PL_oldbufptr == PL_linestart);
216 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
218 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
219 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
221 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
222 if (t < PL_bufptr && isSPACE(*t))
223 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
224 t - PL_oldoldbufptr, PL_oldoldbufptr);
227 Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
233 * Complain about missing quote/regexp/heredoc terminator.
234 * If it's called with (char *)NULL then it cauterizes the line buffer.
235 * If we're in a delimited string and the delimiter is a control
236 * character, it's reformatted into a two-char sequence like ^C.
241 S_missingterm(pTHX_ char *s)
246 char *nl = strrchr(s,'\n');
252 iscntrl(PL_multi_close)
254 PL_multi_close < 32 || PL_multi_close == 127
258 tmpbuf[1] = toCTRL(PL_multi_close);
264 *tmpbuf = PL_multi_close;
268 q = strchr(s,'"') ? '\'' : '"';
269 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
277 Perl_deprecate(pTHX_ char *s)
280 if (ckWARN(WARN_DEPRECATED))
281 Perl_warner(aTHX_ WARN_DEPRECATED, "Use of %s is deprecated", s);
286 * Deprecate a comma-less variable list.
292 deprecate("comma-less variable list");
296 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
297 * utf16-to-utf8-reversed.
300 #ifdef PERL_CR_FILTER
304 register char *s = SvPVX(sv);
305 register char *e = s + SvCUR(sv);
306 /* outer loop optimized to do nothing if there are no CR-LFs */
308 if (*s++ == '\r' && *s == '\n') {
309 /* hit a CR-LF, need to copy the rest */
310 register char *d = s - 1;
313 if (*s == '\r' && s[1] == '\n')
324 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
326 I32 count = FILTER_READ(idx+1, sv, maxlen);
327 if (count > 0 && !maxlen)
335 * Initialize variables. Uses the Perl save_stack to save its state (for
336 * recursive calls to the parser).
340 Perl_lex_start(pTHX_ SV *line)
346 SAVEI32(PL_lex_dojoin);
347 SAVEI32(PL_lex_brackets);
348 SAVEI32(PL_lex_casemods);
349 SAVEI32(PL_lex_starts);
350 SAVEI32(PL_lex_state);
351 SAVEVPTR(PL_lex_inpat);
352 SAVEI32(PL_lex_inwhat);
353 if (PL_lex_state == LEX_KNOWNEXT) {
354 I32 toke = PL_nexttoke;
355 while (--toke >= 0) {
356 SAVEI32(PL_nexttype[toke]);
357 SAVEVPTR(PL_nextval[toke]);
359 SAVEI32(PL_nexttoke);
362 SAVECOPLINE(PL_curcop);
365 SAVEPPTR(PL_oldbufptr);
366 SAVEPPTR(PL_oldoldbufptr);
367 SAVEPPTR(PL_linestart);
368 SAVESPTR(PL_linestr);
369 SAVEPPTR(PL_lex_brackstack);
370 SAVEPPTR(PL_lex_casestack);
371 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
372 SAVESPTR(PL_lex_stuff);
373 SAVEI32(PL_lex_defer);
374 SAVEI32(PL_sublex_info.sub_inwhat);
375 SAVESPTR(PL_lex_repl);
377 SAVEINT(PL_lex_expect);
379 PL_lex_state = LEX_NORMAL;
383 New(899, PL_lex_brackstack, 120, char);
384 New(899, PL_lex_casestack, 12, char);
385 SAVEFREEPV(PL_lex_brackstack);
386 SAVEFREEPV(PL_lex_casestack);
388 *PL_lex_casestack = '\0';
391 PL_lex_stuff = Nullsv;
392 PL_lex_repl = Nullsv;
395 PL_sublex_info.sub_inwhat = 0;
397 if (SvREADONLY(PL_linestr))
398 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
399 s = SvPV(PL_linestr, len);
400 if (len && s[len-1] != ';') {
401 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
402 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
403 sv_catpvn(PL_linestr, "\n;", 2);
405 SvTEMP_off(PL_linestr);
406 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
407 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
409 PL_rs = newSVpvn("\n", 1);
415 * Finalizer for lexing operations. Must be called when the parser is
416 * done with the lexer.
422 PL_doextract = FALSE;
427 * This subroutine has nothing to do with tilting, whether at windmills
428 * or pinball tables. Its name is short for "increment line". It
429 * increments the current line number in CopLINE(PL_curcop) and checks
430 * to see whether the line starts with a comment of the form
431 * # line 500 "foo.pm"
432 * If so, it sets the current line number and file to the values in the comment.
436 S_incline(pTHX_ char *s)
444 CopLINE_inc(PL_curcop);
447 while (SPACE_OR_TAB(*s)) s++;
448 if (strnEQ(s, "line", 4))
452 if (*s == ' ' || *s == '\t')
456 while (SPACE_OR_TAB(*s)) s++;
462 while (SPACE_OR_TAB(*s))
464 if (*s == '"' && (t = strchr(s+1, '"'))) {
469 for (t = s; !isSPACE(*t); t++) ;
472 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
474 if (*e != '\n' && *e != '\0')
475 return; /* false alarm */
481 Safefree(CopFILE(PL_curcop));
483 SvREFCNT_dec(CopFILEGV(PL_curcop));
485 CopFILE_set(PL_curcop, s);
488 CopLINE_set(PL_curcop, atoi(n)-1);
493 * Called to gobble the appropriate amount and type of whitespace.
494 * Skips comments as well.
498 S_skipspace(pTHX_ register char *s)
501 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
502 while (s < PL_bufend && SPACE_OR_TAB(*s))
508 SSize_t oldprevlen, oldoldprevlen;
509 SSize_t oldloplen, oldunilen;
510 while (s < PL_bufend && isSPACE(*s)) {
511 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
516 if (s < PL_bufend && *s == '#') {
517 while (s < PL_bufend && *s != '\n')
521 if (PL_in_eval && !PL_rsfp) {
528 /* only continue to recharge the buffer if we're at the end
529 * of the buffer, we're not reading from a source filter, and
530 * we're in normal lexing mode
532 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
533 PL_lex_state == LEX_FORMLINE)
536 /* try to recharge the buffer */
537 if ((s = filter_gets(PL_linestr, PL_rsfp,
538 (prevlen = SvCUR(PL_linestr)))) == Nullch)
540 /* end of file. Add on the -p or -n magic */
541 if (PL_minus_n || PL_minus_p) {
542 sv_setpv(PL_linestr,PL_minus_p ?
543 ";}continue{print or die qq(-p destination: $!\\n)" :
545 sv_catpv(PL_linestr,";}");
546 PL_minus_n = PL_minus_p = 0;
549 sv_setpv(PL_linestr,";");
551 /* reset variables for next time we lex */
552 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
554 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
556 /* Close the filehandle. Could be from -P preprocessor,
557 * STDIN, or a regular file. If we were reading code from
558 * STDIN (because the commandline held no -e or filename)
559 * then we don't close it, we reset it so the code can
560 * read from STDIN too.
563 if (PL_preprocess && !PL_in_eval)
564 (void)PerlProc_pclose(PL_rsfp);
565 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
566 PerlIO_clearerr(PL_rsfp);
568 (void)PerlIO_close(PL_rsfp);
573 /* not at end of file, so we only read another line */
574 /* make corresponding updates to old pointers, for yyerror() */
575 oldprevlen = PL_oldbufptr - PL_bufend;
576 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
578 oldunilen = PL_last_uni - PL_bufend;
580 oldloplen = PL_last_lop - PL_bufend;
581 PL_linestart = PL_bufptr = s + prevlen;
582 PL_bufend = s + SvCUR(PL_linestr);
584 PL_oldbufptr = s + oldprevlen;
585 PL_oldoldbufptr = s + oldoldprevlen;
587 PL_last_uni = s + oldunilen;
589 PL_last_lop = s + oldloplen;
592 /* debugger active and we're not compiling the debugger code,
593 * so store the line into the debugger's array of lines
595 if (PERLDB_LINE && PL_curstash != PL_debstash) {
596 SV *sv = NEWSV(85,0);
598 sv_upgrade(sv, SVt_PVMG);
599 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
600 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
607 * Check the unary operators to ensure there's no ambiguity in how they're
608 * used. An ambiguous piece of code would be:
610 * This doesn't mean rand() + 5. Because rand() is a unary operator,
611 * the +5 is its argument.
621 if (PL_oldoldbufptr != PL_last_uni)
623 while (isSPACE(*PL_last_uni))
625 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
626 if ((t = strchr(s, '(')) && t < PL_bufptr)
628 if (ckWARN_d(WARN_AMBIGUOUS)){
631 Perl_warner(aTHX_ WARN_AMBIGUOUS,
632 "Warning: Use of \"%s\" without parens is ambiguous",
638 /* workaround to replace the UNI() macro with a function. Only the
639 * hints/uts.sh file mentions this. Other comments elsewhere in the
640 * source indicate Microport Unix might need it too.
646 #define UNI(f) return uni(f,s)
649 S_uni(pTHX_ I32 f, char *s)
654 PL_last_uni = PL_oldbufptr;
665 #endif /* CRIPPLED_CC */
668 * LOP : macro to build a list operator. Its behaviour has been replaced
669 * with a subroutine, S_lop() for which LOP is just another name.
672 #define LOP(f,x) return lop(f,x,s)
676 * Build a list operator (or something that might be one). The rules:
677 * - if we have a next token, then it's a list operator [why?]
678 * - if the next thing is an opening paren, then it's a function
679 * - else it's a list operator
683 S_lop(pTHX_ I32 f, int x, char *s)
690 PL_last_lop = PL_oldbufptr;
705 * When the lexer realizes it knows the next token (for instance,
706 * it is reordering tokens for the parser) then it can call S_force_next
707 * to know what token to return the next time the lexer is called. Caller
708 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
709 * handles the token correctly.
713 S_force_next(pTHX_ I32 type)
715 PL_nexttype[PL_nexttoke] = type;
717 if (PL_lex_state != LEX_KNOWNEXT) {
718 PL_lex_defer = PL_lex_state;
719 PL_lex_expect = PL_expect;
720 PL_lex_state = LEX_KNOWNEXT;
726 * When the lexer knows the next thing is a word (for instance, it has
727 * just seen -> and it knows that the next char is a word char, then
728 * it calls S_force_word to stick the next word into the PL_next lookahead.
731 * char *start : buffer position (must be within PL_linestr)
732 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
733 * int check_keyword : if true, Perl checks to make sure the word isn't
734 * a keyword (do this if the word is a label, e.g. goto FOO)
735 * int allow_pack : if true, : characters will also be allowed (require,
737 * int allow_initial_tick : used by the "sub" lexer only.
741 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
746 start = skipspace(start);
748 if (isIDFIRST_lazy_if(s,UTF) ||
749 (allow_pack && *s == ':') ||
750 (allow_initial_tick && *s == '\'') )
752 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
753 if (check_keyword && keyword(PL_tokenbuf, len))
755 if (token == METHOD) {
760 PL_expect = XOPERATOR;
763 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
764 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
772 * Called when the lexer wants $foo *foo &foo etc, but the program
773 * text only contains the "foo" portion. The first argument is a pointer
774 * to the "foo", and the second argument is the type symbol to prefix.
775 * Forces the next token to be a "WORD".
776 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
780 S_force_ident(pTHX_ register char *s, int kind)
783 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
784 PL_nextval[PL_nexttoke].opval = o;
787 dTHR; /* just for in_eval */
788 o->op_private = OPpCONST_ENTERED;
789 /* XXX see note in pp_entereval() for why we forgo typo
790 warnings if the symbol must be introduced in an eval.
792 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
793 kind == '$' ? SVt_PV :
794 kind == '@' ? SVt_PVAV :
795 kind == '%' ? SVt_PVHV :
803 Perl_str_to_version(pTHX_ SV *sv)
808 char *start = SvPVx(sv,len);
809 bool utf = SvUTF8(sv) ? TRUE : FALSE;
810 char *end = start + len;
811 while (start < end) {
815 n = utf8_to_uv((U8*)start, &skip);
820 retval += ((NV)n)/nshift;
829 * Forces the next token to be a version number.
833 S_force_version(pTHX_ char *s)
835 OP *version = Nullop;
844 for (; isDIGIT(*d) || *d == '_' || *d == '.'; d++);
845 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
848 version = yylval.opval;
849 ver = cSVOPx(version)->op_sv;
850 if (SvPOK(ver) && !SvNIOK(ver)) {
851 (void)SvUPGRADE(ver, SVt_PVNV);
852 SvNVX(ver) = str_to_version(ver);
853 SvNOK_on(ver); /* hint that it is a version */
858 /* NOTE: The parser sees the package name and the VERSION swapped */
859 PL_nextval[PL_nexttoke].opval = version;
867 * Tokenize a quoted string passed in as an SV. It finds the next
868 * chunk, up to end of string or a backslash. It may make a new
869 * SV containing that chunk (if HINT_NEW_STRING is on). It also
874 S_tokeq(pTHX_ SV *sv)
885 s = SvPV_force(sv, len);
886 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
889 while (s < send && *s != '\\')
894 if ( PL_hints & HINT_NEW_STRING )
895 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
898 if (s + 1 < send && (s[1] == '\\'))
899 s++; /* all that, just for this */
904 SvCUR_set(sv, d - SvPVX(sv));
906 if ( PL_hints & HINT_NEW_STRING )
907 return new_constant(NULL, 0, "q", sv, pv, "q");
912 * Now come three functions related to double-quote context,
913 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
914 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
915 * interact with PL_lex_state, and create fake ( ... ) argument lists
916 * to handle functions and concatenation.
917 * They assume that whoever calls them will be setting up a fake
918 * join call, because each subthing puts a ',' after it. This lets
921 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
923 * (I'm not sure whether the spurious commas at the end of lcfirst's
924 * arguments and join's arguments are created or not).
929 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
931 * Pattern matching will set PL_lex_op to the pattern-matching op to
932 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
934 * OP_CONST and OP_READLINE are easy--just make the new op and return.
936 * Everything else becomes a FUNC.
938 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
939 * had an OP_CONST or OP_READLINE). This just sets us up for a
940 * call to S_sublex_push().
946 register I32 op_type = yylval.ival;
948 if (op_type == OP_NULL) {
949 yylval.opval = PL_lex_op;
953 if (op_type == OP_CONST || op_type == OP_READLINE) {
954 SV *sv = tokeq(PL_lex_stuff);
956 if (SvTYPE(sv) == SVt_PVIV) {
957 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
963 nsv = newSVpvn(p, len);
969 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
970 PL_lex_stuff = Nullsv;
974 PL_sublex_info.super_state = PL_lex_state;
975 PL_sublex_info.sub_inwhat = op_type;
976 PL_sublex_info.sub_op = PL_lex_op;
977 PL_lex_state = LEX_INTERPPUSH;
981 yylval.opval = PL_lex_op;
991 * Create a new scope to save the lexing state. The scope will be
992 * ended in S_sublex_done. Returns a '(', starting the function arguments
993 * to the uc, lc, etc. found before.
994 * Sets PL_lex_state to LEX_INTERPCONCAT.
1003 PL_lex_state = PL_sublex_info.super_state;
1004 SAVEI32(PL_lex_dojoin);
1005 SAVEI32(PL_lex_brackets);
1006 SAVEI32(PL_lex_casemods);
1007 SAVEI32(PL_lex_starts);
1008 SAVEI32(PL_lex_state);
1009 SAVEVPTR(PL_lex_inpat);
1010 SAVEI32(PL_lex_inwhat);
1011 SAVECOPLINE(PL_curcop);
1012 SAVEPPTR(PL_bufptr);
1013 SAVEPPTR(PL_oldbufptr);
1014 SAVEPPTR(PL_oldoldbufptr);
1015 SAVEPPTR(PL_linestart);
1016 SAVESPTR(PL_linestr);
1017 SAVEPPTR(PL_lex_brackstack);
1018 SAVEPPTR(PL_lex_casestack);
1020 PL_linestr = PL_lex_stuff;
1021 PL_lex_stuff = Nullsv;
1023 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1024 = SvPVX(PL_linestr);
1025 PL_bufend += SvCUR(PL_linestr);
1026 SAVEFREESV(PL_linestr);
1028 PL_lex_dojoin = FALSE;
1029 PL_lex_brackets = 0;
1030 New(899, PL_lex_brackstack, 120, char);
1031 New(899, PL_lex_casestack, 12, char);
1032 SAVEFREEPV(PL_lex_brackstack);
1033 SAVEFREEPV(PL_lex_casestack);
1034 PL_lex_casemods = 0;
1035 *PL_lex_casestack = '\0';
1037 PL_lex_state = LEX_INTERPCONCAT;
1038 CopLINE_set(PL_curcop, PL_multi_start);
1040 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1041 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1042 PL_lex_inpat = PL_sublex_info.sub_op;
1044 PL_lex_inpat = Nullop;
1051 * Restores lexer state after a S_sublex_push.
1057 if (!PL_lex_starts++) {
1058 PL_expect = XOPERATOR;
1059 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn("",0));
1063 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1064 PL_lex_state = LEX_INTERPCASEMOD;
1068 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1069 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1070 PL_linestr = PL_lex_repl;
1072 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1073 PL_bufend += SvCUR(PL_linestr);
1074 SAVEFREESV(PL_linestr);
1075 PL_lex_dojoin = FALSE;
1076 PL_lex_brackets = 0;
1077 PL_lex_casemods = 0;
1078 *PL_lex_casestack = '\0';
1080 if (SvEVALED(PL_lex_repl)) {
1081 PL_lex_state = LEX_INTERPNORMAL;
1083 /* we don't clear PL_lex_repl here, so that we can check later
1084 whether this is an evalled subst; that means we rely on the
1085 logic to ensure sublex_done() is called again only via the
1086 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1089 PL_lex_state = LEX_INTERPCONCAT;
1090 PL_lex_repl = Nullsv;
1096 PL_bufend = SvPVX(PL_linestr);
1097 PL_bufend += SvCUR(PL_linestr);
1098 PL_expect = XOPERATOR;
1099 PL_sublex_info.sub_inwhat = 0;
1107 Extracts a pattern, double-quoted string, or transliteration. This
1110 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1111 processing a pattern (PL_lex_inpat is true), a transliteration
1112 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1114 Returns a pointer to the character scanned up to. Iff this is
1115 advanced from the start pointer supplied (ie if anything was
1116 successfully parsed), will leave an OP for the substring scanned
1117 in yylval. Caller must intuit reason for not parsing further
1118 by looking at the next characters herself.
1122 double-quoted style: \r and \n
1123 regexp special ones: \D \s
1125 backrefs: \1 (deprecated in substitution replacements)
1126 case and quoting: \U \Q \E
1127 stops on @ and $, but not for $ as tail anchor
1129 In transliterations:
1130 characters are VERY literal, except for - not at the start or end
1131 of the string, which indicates a range. scan_const expands the
1132 range to the full set of intermediate characters.
1134 In double-quoted strings:
1136 double-quoted style: \r and \n
1138 backrefs: \1 (deprecated)
1139 case and quoting: \U \Q \E
1142 scan_const does *not* construct ops to handle interpolated strings.
1143 It stops processing as soon as it finds an embedded $ or @ variable
1144 and leaves it to the caller to work out what's going on.
1146 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1148 $ in pattern could be $foo or could be tail anchor. Assumption:
1149 it's a tail anchor if $ is the last thing in the string, or if it's
1150 followed by one of ")| \n\t"
1152 \1 (backreferences) are turned into $1
1154 The structure of the code is
1155 while (there's a character to process) {
1156 handle transliteration ranges
1157 skip regexp comments
1158 skip # initiated comments in //x patterns
1159 check for embedded @foo
1160 check for embedded scalars
1162 leave intact backslashes from leave (below)
1163 deprecate \1 in strings and sub replacements
1164 handle string-changing backslashes \l \U \Q \E, etc.
1165 switch (what was escaped) {
1166 handle - in a transliteration (becomes a literal -)
1167 handle \132 octal characters
1168 handle 0x15 hex characters
1169 handle \cV (control V)
1170 handle printf backslashes (\f, \r, \n, etc)
1172 } (end if backslash)
1173 } (end while character to read)
1178 S_scan_const(pTHX_ char *start)
1180 register char *send = PL_bufend; /* end of the constant */
1181 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1182 register char *s = start; /* start of the constant */
1183 register char *d = SvPVX(sv); /* destination for copies */
1184 bool dorange = FALSE; /* are we in a translit range? */
1185 bool didrange = FALSE; /* did we just finish a range? */
1186 bool has_utf = FALSE; /* embedded \x{} */
1190 I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1191 ? (PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1193 I32 thisutf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1194 ? (PL_sublex_info.sub_op->op_private & (PL_lex_repl ?
1195 OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF))
1197 const char *leaveit = /* set of acceptably-backslashed characters */
1199 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1202 while (s < send || dorange) {
1203 /* get transliterations out of the way (they're most literal) */
1204 if (PL_lex_inwhat == OP_TRANS) {
1205 /* expand a range A-Z to the full set of characters. AIE! */
1207 I32 i; /* current expanded character */
1208 I32 min; /* first character in range */
1209 I32 max; /* last character in range */
1211 i = d - SvPVX(sv); /* remember current offset */
1212 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1213 d = SvPVX(sv) + i; /* refresh d after realloc */
1214 d -= 2; /* eat the first char and the - */
1216 min = (U8)*d; /* first char in range */
1217 max = (U8)d[1]; /* last char in range */
1221 "Invalid [] range \"%c-%c\" in transliteration operator",
1226 if ((isLOWER(min) && isLOWER(max)) ||
1227 (isUPPER(min) && isUPPER(max))) {
1229 for (i = min; i <= max; i++)
1233 for (i = min; i <= max; i++)
1240 for (i = min; i <= max; i++)
1243 /* mark the range as done, and continue */
1249 /* range begins (ignore - as first or last char) */
1250 else if (*s == '-' && s+1 < send && s != start) {
1252 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
1255 *d++ = (char)0xff; /* use illegal utf8 byte--see pmtrans */
1267 /* if we get here, we're not doing a transliteration */
1269 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1270 except for the last char, which will be done separately. */
1271 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1273 while (s < send && *s != ')')
1276 else if (s[2] == '{' /* This should match regcomp.c */
1277 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1280 char *regparse = s + (s[2] == '{' ? 3 : 4);
1283 while (count && (c = *regparse)) {
1284 if (c == '\\' && regparse[1])
1292 if (*regparse != ')') {
1293 regparse--; /* Leave one char for continuation. */
1294 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1296 while (s < regparse)
1301 /* likewise skip #-initiated comments in //x patterns */
1302 else if (*s == '#' && PL_lex_inpat &&
1303 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1304 while (s+1 < send && *s != '\n')
1308 /* check for embedded arrays (@foo, @:foo, @'foo, @{foo}, @$foo) */
1309 else if (*s == '@' && s[1]
1310 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$", s[1])))
1313 /* check for embedded scalars. only stop if we're sure it's a
1316 else if (*s == '$') {
1317 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1319 if (s + 1 < send && !strchr("()| \n\t", s[1]))
1320 break; /* in regexp, $ might be tail anchor */
1323 /* (now in tr/// code again) */
1325 if (*s & 0x80 && thisutf) {
1326 (void)utf8_to_uv((U8*)s, &len);
1328 /* illegal UTF8, make it valid */
1329 char *old_pvx = SvPVX(sv);
1330 /* need space for one extra char (NOTE: SvCUR() not set here) */
1331 d = SvGROW(sv, SvLEN(sv) + 1) + (d - old_pvx);
1332 d = (char*)uv_to_utf8((U8*)d, (U8)*s++);
1343 if (*s == '\\' && s+1 < send) {
1346 /* some backslashes we leave behind */
1347 if (*leaveit && *s && strchr(leaveit, *s)) {
1353 /* deprecate \1 in strings and substitution replacements */
1354 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1355 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1357 dTHR; /* only for ckWARN */
1358 if (ckWARN(WARN_SYNTAX))
1359 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1364 /* string-change backslash escapes */
1365 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1370 /* if we get here, it's either a quoted -, or a digit */
1373 /* quoted - in transliterations */
1375 if (PL_lex_inwhat == OP_TRANS) {
1383 if (ckWARN(WARN_MISC) && isALNUM(*s))
1384 Perl_warner(aTHX_ WARN_MISC,
1385 "Unrecognized escape \\%c passed through",
1387 /* default action is to copy the quoted character */
1392 /* \132 indicates an octal constant */
1393 case '0': case '1': case '2': case '3':
1394 case '4': case '5': case '6': case '7':
1395 len = 0; /* disallow underscores */
1396 uv = (UV)scan_oct(s, 3, &len);
1398 goto NUM_ESCAPE_INSERT;
1400 /* \x24 indicates a hex constant */
1404 char* e = strchr(s, '}');
1406 yyerror("Missing right brace on \\x{}");
1409 len = 1; /* allow underscores */
1410 uv = (UV)scan_hex(s + 1, e - s - 1, &len);
1414 len = 0; /* disallow underscores */
1415 uv = (UV)scan_hex(s, 2, &len);
1420 /* Insert oct or hex escaped character.
1421 * There will always enough room in sv since such escapes will
1422 * be longer than any utf8 sequence they can end up as
1425 if (!thisutf && !has_utf && uv > 255) {
1426 /* might need to recode whatever we have accumulated so far
1427 * if it contains any hibit chars
1431 for (c = SvPVX(sv); c < d; c++) {
1436 char *old_pvx = SvPVX(sv);
1438 d = SvGROW(sv, SvCUR(sv) + hicount + 1) + (d - old_pvx);
1447 uv_to_utf8((U8*)dst, (U8)*src--);
1457 if (thisutf || uv > 255) {
1458 d = (char*)uv_to_utf8((U8*)d, uv);
1470 /* \N{latin small letter a} is a named character */
1474 char* e = strchr(s, '}');
1480 yyerror("Missing right brace on \\N{}");
1484 res = newSVpvn(s + 1, e - s - 1);
1485 res = new_constant( Nullch, 0, "charnames",
1486 res, Nullsv, "\\N{...}" );
1487 str = SvPV(res,len);
1488 if (!has_utf && SvUTF8(res)) {
1489 char *ostart = SvPVX(sv);
1490 SvCUR_set(sv, d - ostart);
1492 sv_utf8_upgrade(sv);
1493 d = SvPVX(sv) + SvCUR(sv);
1496 if (len > e - s + 4) {
1497 char *odest = SvPVX(sv);
1499 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1500 d = SvPVX(sv) + (d - odest);
1502 Copy(str, d, len, char);
1509 yyerror("Missing braces on \\N{}");
1512 /* \c is a control character */
1527 /* printf-style backslashes, formfeeds, newlines, etc */
1545 *d++ = '\047'; /* CP 1047 */
1548 *d++ = '\057'; /* CP 1047 */
1562 } /* end if (backslash) */
1565 } /* while loop to process each character */
1567 /* terminate the string and set up the sv */
1569 SvCUR_set(sv, d - SvPVX(sv));
1574 /* shrink the sv if we allocated more than we used */
1575 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1576 SvLEN_set(sv, SvCUR(sv) + 1);
1577 Renew(SvPVX(sv), SvLEN(sv), char);
1580 /* return the substring (via yylval) only if we parsed anything */
1581 if (s > PL_bufptr) {
1582 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1583 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1585 ( PL_lex_inwhat == OP_TRANS
1587 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1590 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1597 * Returns TRUE if there's more to the expression (e.g., a subscript),
1600 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1602 * ->[ and ->{ return TRUE
1603 * { and [ outside a pattern are always subscripts, so return TRUE
1604 * if we're outside a pattern and it's not { or [, then return FALSE
1605 * if we're in a pattern and the first char is a {
1606 * {4,5} (any digits around the comma) returns FALSE
1607 * if we're in a pattern and the first char is a [
1609 * [SOMETHING] has a funky algorithm to decide whether it's a
1610 * character class or not. It has to deal with things like
1611 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1612 * anything else returns TRUE
1615 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1618 S_intuit_more(pTHX_ register char *s)
1620 if (PL_lex_brackets)
1622 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1624 if (*s != '{' && *s != '[')
1629 /* In a pattern, so maybe we have {n,m}. */
1646 /* On the other hand, maybe we have a character class */
1649 if (*s == ']' || *s == '^')
1652 /* this is terrifying, and it works */
1653 int weight = 2; /* let's weigh the evidence */
1655 unsigned char un_char = 255, last_un_char;
1656 char *send = strchr(s,']');
1657 char tmpbuf[sizeof PL_tokenbuf * 4];
1659 if (!send) /* has to be an expression */
1662 Zero(seen,256,char);
1665 else if (isDIGIT(*s)) {
1667 if (isDIGIT(s[1]) && s[2] == ']')
1673 for (; s < send; s++) {
1674 last_un_char = un_char;
1675 un_char = (unsigned char)*s;
1680 weight -= seen[un_char] * 10;
1681 if (isALNUM_lazy_if(s+1,UTF)) {
1682 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1683 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1688 else if (*s == '$' && s[1] &&
1689 strchr("[#!%*<>()-=",s[1])) {
1690 if (/*{*/ strchr("])} =",s[2]))
1699 if (strchr("wds]",s[1]))
1701 else if (seen['\''] || seen['"'])
1703 else if (strchr("rnftbxcav",s[1]))
1705 else if (isDIGIT(s[1])) {
1707 while (s[1] && isDIGIT(s[1]))
1717 if (strchr("aA01! ",last_un_char))
1719 if (strchr("zZ79~",s[1]))
1721 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1722 weight -= 5; /* cope with negative subscript */
1725 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1726 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1731 if (keyword(tmpbuf, d - tmpbuf))
1734 if (un_char == last_un_char + 1)
1736 weight -= seen[un_char];
1741 if (weight >= 0) /* probably a character class */
1751 * Does all the checking to disambiguate
1753 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1754 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1756 * First argument is the stuff after the first token, e.g. "bar".
1758 * Not a method if bar is a filehandle.
1759 * Not a method if foo is a subroutine prototyped to take a filehandle.
1760 * Not a method if it's really "Foo $bar"
1761 * Method if it's "foo $bar"
1762 * Not a method if it's really "print foo $bar"
1763 * Method if it's really "foo package::" (interpreted as package->foo)
1764 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1765 * Not a method if bar is a filehandle or package, but is quoted with
1770 S_intuit_method(pTHX_ char *start, GV *gv)
1772 char *s = start + (*start == '$');
1773 char tmpbuf[sizeof PL_tokenbuf];
1781 if ((cv = GvCVu(gv))) {
1782 char *proto = SvPVX(cv);
1792 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1793 /* start is the beginning of the possible filehandle/object,
1794 * and s is the end of it
1795 * tmpbuf is a copy of it
1798 if (*start == '$') {
1799 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1804 return *s == '(' ? FUNCMETH : METHOD;
1806 if (!keyword(tmpbuf, len)) {
1807 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1812 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1813 if (indirgv && GvCVu(indirgv))
1815 /* filehandle or package name makes it a method */
1816 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1818 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1819 return 0; /* no assumptions -- "=>" quotes bearword */
1821 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1822 newSVpvn(tmpbuf,len));
1823 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1827 return *s == '(' ? FUNCMETH : METHOD;
1835 * Return a string of Perl code to load the debugger. If PERL5DB
1836 * is set, it will return the contents of that, otherwise a
1837 * compile-time require of perl5db.pl.
1844 char *pdb = PerlEnv_getenv("PERL5DB");
1848 SETERRNO(0,SS$_NORMAL);
1849 return "BEGIN { require 'perl5db.pl' }";
1855 /* Encoded script support. filter_add() effectively inserts a
1856 * 'pre-processing' function into the current source input stream.
1857 * Note that the filter function only applies to the current source file
1858 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1860 * The datasv parameter (which may be NULL) can be used to pass
1861 * private data to this instance of the filter. The filter function
1862 * can recover the SV using the FILTER_DATA macro and use it to
1863 * store private buffers and state information.
1865 * The supplied datasv parameter is upgraded to a PVIO type
1866 * and the IoDIRP/IoANY field is used to store the function pointer,
1867 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1868 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1869 * private use must be set using malloc'd pointers.
1873 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1878 if (!PL_rsfp_filters)
1879 PL_rsfp_filters = newAV();
1881 datasv = NEWSV(255,0);
1882 if (!SvUPGRADE(datasv, SVt_PVIO))
1883 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1884 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
1885 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1886 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1887 funcp, SvPV_nolen(datasv)));
1888 av_unshift(PL_rsfp_filters, 1);
1889 av_store(PL_rsfp_filters, 0, datasv) ;
1894 /* Delete most recently added instance of this filter function. */
1896 Perl_filter_del(pTHX_ filter_t funcp)
1899 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1900 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1902 /* if filter is on top of stack (usual case) just pop it off */
1903 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1904 if (IoANY(datasv) == (void *)funcp) {
1905 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1906 IoANY(datasv) = (void *)NULL;
1907 sv_free(av_pop(PL_rsfp_filters));
1911 /* we need to search for the correct entry and clear it */
1912 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1916 /* Invoke the n'th filter function for the current rsfp. */
1918 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1921 /* 0 = read one text line */
1926 if (!PL_rsfp_filters)
1928 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1929 /* Provide a default input filter to make life easy. */
1930 /* Note that we append to the line. This is handy. */
1931 DEBUG_P(PerlIO_printf(Perl_debug_log,
1932 "filter_read %d: from rsfp\n", idx));
1936 int old_len = SvCUR(buf_sv) ;
1938 /* ensure buf_sv is large enough */
1939 SvGROW(buf_sv, old_len + maxlen) ;
1940 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1941 if (PerlIO_error(PL_rsfp))
1942 return -1; /* error */
1944 return 0 ; /* end of file */
1946 SvCUR_set(buf_sv, old_len + len) ;
1949 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1950 if (PerlIO_error(PL_rsfp))
1951 return -1; /* error */
1953 return 0 ; /* end of file */
1956 return SvCUR(buf_sv);
1958 /* Skip this filter slot if filter has been deleted */
1959 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1960 DEBUG_P(PerlIO_printf(Perl_debug_log,
1961 "filter_read %d: skipped (filter deleted)\n",
1963 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1965 /* Get function pointer hidden within datasv */
1966 funcp = (filter_t)IoANY(datasv);
1967 DEBUG_P(PerlIO_printf(Perl_debug_log,
1968 "filter_read %d: via function %p (%s)\n",
1969 idx, funcp, SvPV_nolen(datasv)));
1970 /* Call function. The function is expected to */
1971 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1972 /* Return: <0:error, =0:eof, >0:not eof */
1973 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1977 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1979 #ifdef PERL_CR_FILTER
1980 if (!PL_rsfp_filters) {
1981 filter_add(S_cr_textfilter,NULL);
1984 if (PL_rsfp_filters) {
1987 SvCUR_set(sv, 0); /* start with empty line */
1988 if (FILTER_READ(0, sv, 0) > 0)
1989 return ( SvPVX(sv) ) ;
1994 return (sv_gets(sv, fp, append));
1998 S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2002 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2006 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2007 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2009 return GvHV(gv); /* Foo:: */
2012 /* use constant CLASS => 'MyClass' */
2013 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2015 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2016 pkgname = SvPV_nolen(sv);
2020 return gv_stashpv(pkgname, FALSE);
2024 static char* exp_name[] =
2025 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2026 "ATTRTERM", "TERMBLOCK"
2033 Works out what to call the token just pulled out of the input
2034 stream. The yacc parser takes care of taking the ops we return and
2035 stitching them into a tree.
2041 if read an identifier
2042 if we're in a my declaration
2043 croak if they tried to say my($foo::bar)
2044 build the ops for a my() declaration
2045 if it's an access to a my() variable
2046 are we in a sort block?
2047 croak if my($a); $a <=> $b
2048 build ops for access to a my() variable
2049 if in a dq string, and they've said @foo and we can't find @foo
2051 build ops for a bareword
2052 if we already built the token before, use it.
2056 #pragma segment Perl_yylex
2059 #ifdef USE_PURE_BISON
2060 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
2073 #ifdef USE_PURE_BISON
2074 yylval_pointer = lvalp;
2075 yychar_pointer = lcharp;
2078 /* check if there's an identifier for us to look at */
2079 if (PL_pending_ident) {
2080 /* pit holds the identifier we read and pending_ident is reset */
2081 char pit = PL_pending_ident;
2082 PL_pending_ident = 0;
2084 /* if we're in a my(), we can't allow dynamics here.
2085 $foo'bar has already been turned into $foo::bar, so
2086 just check for colons.
2088 if it's a legal name, the OP is a PADANY.
2091 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
2092 if (strchr(PL_tokenbuf,':'))
2093 yyerror(Perl_form(aTHX_ "No package name allowed for "
2094 "variable %s in \"our\"",
2096 tmp = pad_allocmy(PL_tokenbuf);
2099 if (strchr(PL_tokenbuf,':'))
2100 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
2102 yylval.opval = newOP(OP_PADANY, 0);
2103 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
2109 build the ops for accesses to a my() variable.
2111 Deny my($a) or my($b) in a sort block, *if* $a or $b is
2112 then used in a comparison. This catches most, but not
2113 all cases. For instance, it catches
2114 sort { my($a); $a <=> $b }
2116 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2117 (although why you'd do that is anyone's guess).
2120 if (!strchr(PL_tokenbuf,':')) {
2122 /* Check for single character per-thread SVs */
2123 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2124 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2125 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2127 yylval.opval = newOP(OP_THREADSV, 0);
2128 yylval.opval->op_targ = tmp;
2131 #endif /* USE_THREADS */
2132 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2133 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2134 /* might be an "our" variable" */
2135 if (SvFLAGS(namesv) & SVpad_OUR) {
2136 /* build ops for a bareword */
2137 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2138 sv_catpvn(sym, "::", 2);
2139 sv_catpv(sym, PL_tokenbuf+1);
2140 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2141 yylval.opval->op_private = OPpCONST_ENTERED;
2142 gv_fetchpv(SvPVX(sym),
2144 ? (GV_ADDMULTI | GV_ADDINEVAL)
2147 ((PL_tokenbuf[0] == '$') ? SVt_PV
2148 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2153 /* if it's a sort block and they're naming $a or $b */
2154 if (PL_last_lop_op == OP_SORT &&
2155 PL_tokenbuf[0] == '$' &&
2156 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2159 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2160 d < PL_bufend && *d != '\n';
2163 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2164 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2170 yylval.opval = newOP(OP_PADANY, 0);
2171 yylval.opval->op_targ = tmp;
2177 Whine if they've said @foo in a doublequoted string,
2178 and @foo isn't a variable we can find in the symbol
2181 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2182 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2183 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2184 && ckWARN(WARN_AMBIGUOUS))
2186 /* Downgraded from fatal to warning 20000522 mjd */
2187 Perl_warner(aTHX_ WARN_AMBIGUOUS,
2188 "Possible unintended interpolation of %s in string",
2193 /* build ops for a bareword */
2194 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2195 yylval.opval->op_private = OPpCONST_ENTERED;
2196 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2197 ((PL_tokenbuf[0] == '$') ? SVt_PV
2198 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2203 /* no identifier pending identification */
2205 switch (PL_lex_state) {
2207 case LEX_NORMAL: /* Some compilers will produce faster */
2208 case LEX_INTERPNORMAL: /* code if we comment these out. */
2212 /* when we've already built the next token, just pull it out of the queue */
2215 yylval = PL_nextval[PL_nexttoke];
2217 PL_lex_state = PL_lex_defer;
2218 PL_expect = PL_lex_expect;
2219 PL_lex_defer = LEX_NORMAL;
2221 return(PL_nexttype[PL_nexttoke]);
2223 /* interpolated case modifiers like \L \U, including \Q and \E.
2224 when we get here, PL_bufptr is at the \
2226 case LEX_INTERPCASEMOD:
2228 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2229 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2231 /* handle \E or end of string */
2232 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2236 if (PL_lex_casemods) {
2237 oldmod = PL_lex_casestack[--PL_lex_casemods];
2238 PL_lex_casestack[PL_lex_casemods] = '\0';
2240 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2242 PL_lex_state = LEX_INTERPCONCAT;
2246 if (PL_bufptr != PL_bufend)
2248 PL_lex_state = LEX_INTERPCONCAT;
2253 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2254 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2255 if (strchr("LU", *s) &&
2256 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2258 PL_lex_casestack[--PL_lex_casemods] = '\0';
2261 if (PL_lex_casemods > 10) {
2262 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2263 if (newlb != PL_lex_casestack) {
2265 PL_lex_casestack = newlb;
2268 PL_lex_casestack[PL_lex_casemods++] = *s;
2269 PL_lex_casestack[PL_lex_casemods] = '\0';
2270 PL_lex_state = LEX_INTERPCONCAT;
2271 PL_nextval[PL_nexttoke].ival = 0;
2274 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2276 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2278 PL_nextval[PL_nexttoke].ival = OP_LC;
2280 PL_nextval[PL_nexttoke].ival = OP_UC;
2282 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2284 Perl_croak(aTHX_ "panic: yylex");
2287 if (PL_lex_starts) {
2296 case LEX_INTERPPUSH:
2297 return sublex_push();
2299 case LEX_INTERPSTART:
2300 if (PL_bufptr == PL_bufend)
2301 return sublex_done();
2303 PL_lex_dojoin = (*PL_bufptr == '@');
2304 PL_lex_state = LEX_INTERPNORMAL;
2305 if (PL_lex_dojoin) {
2306 PL_nextval[PL_nexttoke].ival = 0;
2309 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2310 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2311 force_next(PRIVATEREF);
2313 force_ident("\"", '$');
2314 #endif /* USE_THREADS */
2315 PL_nextval[PL_nexttoke].ival = 0;
2317 PL_nextval[PL_nexttoke].ival = 0;
2319 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2322 if (PL_lex_starts++) {
2328 case LEX_INTERPENDMAYBE:
2329 if (intuit_more(PL_bufptr)) {
2330 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2336 if (PL_lex_dojoin) {
2337 PL_lex_dojoin = FALSE;
2338 PL_lex_state = LEX_INTERPCONCAT;
2341 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2342 && SvEVALED(PL_lex_repl))
2344 if (PL_bufptr != PL_bufend)
2345 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2346 PL_lex_repl = Nullsv;
2349 case LEX_INTERPCONCAT:
2351 if (PL_lex_brackets)
2352 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2354 if (PL_bufptr == PL_bufend)
2355 return sublex_done();
2357 if (SvIVX(PL_linestr) == '\'') {
2358 SV *sv = newSVsv(PL_linestr);
2361 else if ( PL_hints & HINT_NEW_RE )
2362 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2363 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2367 s = scan_const(PL_bufptr);
2369 PL_lex_state = LEX_INTERPCASEMOD;
2371 PL_lex_state = LEX_INTERPSTART;
2374 if (s != PL_bufptr) {
2375 PL_nextval[PL_nexttoke] = yylval;
2378 if (PL_lex_starts++)
2388 PL_lex_state = LEX_NORMAL;
2389 s = scan_formline(PL_bufptr);
2390 if (!PL_lex_formbrack)
2396 PL_oldoldbufptr = PL_oldbufptr;
2399 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2400 exp_name[PL_expect], s);
2406 if (isIDFIRST_lazy_if(s,UTF))
2408 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2411 goto fake_eof; /* emulate EOF on ^D or ^Z */
2416 if (PL_lex_brackets)
2417 yyerror("Missing right curly or square bracket");
2420 if (s++ < PL_bufend)
2421 goto retry; /* ignore stray nulls */
2424 if (!PL_in_eval && !PL_preambled) {
2425 PL_preambled = TRUE;
2426 sv_setpv(PL_linestr,incl_perldb());
2427 if (SvCUR(PL_linestr))
2428 sv_catpv(PL_linestr,";");
2430 while(AvFILLp(PL_preambleav) >= 0) {
2431 SV *tmpsv = av_shift(PL_preambleav);
2432 sv_catsv(PL_linestr, tmpsv);
2433 sv_catpv(PL_linestr, ";");
2436 sv_free((SV*)PL_preambleav);
2437 PL_preambleav = NULL;
2439 if (PL_minus_n || PL_minus_p) {
2440 sv_catpv(PL_linestr, "LINE: while (<>) {");
2442 sv_catpv(PL_linestr,"chomp;");
2444 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2446 GvIMPORTED_AV_on(gv);
2448 if (strchr("/'\"", *PL_splitstr)
2449 && strchr(PL_splitstr + 1, *PL_splitstr))
2450 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2453 s = "'~#\200\1'"; /* surely one char is unused...*/
2454 while (s[1] && strchr(PL_splitstr, *s)) s++;
2456 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2457 "q" + (delim == '\''), delim);
2458 for (s = PL_splitstr; *s; s++) {
2460 sv_catpvn(PL_linestr, "\\", 1);
2461 sv_catpvn(PL_linestr, s, 1);
2463 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2467 sv_catpv(PL_linestr,"@F=split(' ');");
2470 sv_catpv(PL_linestr, "\n");
2471 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2472 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2473 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2474 SV *sv = NEWSV(85,0);
2476 sv_upgrade(sv, SVt_PVMG);
2477 sv_setsv(sv,PL_linestr);
2478 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2484 bof = PL_rsfp && (PerlIO_tell(PL_rsfp) == 0); /* *Before* read! */
2485 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2488 if (PL_preprocess && !PL_in_eval)
2489 (void)PerlProc_pclose(PL_rsfp);
2490 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2491 PerlIO_clearerr(PL_rsfp);
2493 (void)PerlIO_close(PL_rsfp);
2495 PL_doextract = FALSE;
2497 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2498 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2499 sv_catpv(PL_linestr,";}");
2500 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2501 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2502 PL_minus_n = PL_minus_p = 0;
2505 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2506 sv_setpv(PL_linestr,"");
2507 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2510 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2511 PL_doextract = FALSE;
2513 /* Incest with pod. */
2514 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2515 sv_setpv(PL_linestr, "");
2516 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2517 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2518 PL_doextract = FALSE;
2522 s = swallow_bom((U8*)s);
2524 } while (PL_doextract);
2525 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2526 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2527 SV *sv = NEWSV(85,0);
2529 sv_upgrade(sv, SVt_PVMG);
2530 sv_setsv(sv,PL_linestr);
2531 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2533 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2534 if (CopLINE(PL_curcop) == 1) {
2535 while (s < PL_bufend && isSPACE(*s))
2537 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2541 if (*s == '#' && *(s+1) == '!')
2543 #ifdef ALTERNATE_SHEBANG
2545 static char as[] = ALTERNATE_SHEBANG;
2546 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2547 d = s + (sizeof(as) - 1);
2549 #endif /* ALTERNATE_SHEBANG */
2558 while (*d && !isSPACE(*d))
2562 #ifdef ARG_ZERO_IS_SCRIPT
2563 if (ipathend > ipath) {
2565 * HP-UX (at least) sets argv[0] to the script name,
2566 * which makes $^X incorrect. And Digital UNIX and Linux,
2567 * at least, set argv[0] to the basename of the Perl
2568 * interpreter. So, having found "#!", we'll set it right.
2570 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2571 assert(SvPOK(x) || SvGMAGICAL(x));
2572 if (sv_eq(x, CopFILESV(PL_curcop))) {
2573 sv_setpvn(x, ipath, ipathend - ipath);
2576 TAINT_NOT; /* $^X is always tainted, but that's OK */
2578 #endif /* ARG_ZERO_IS_SCRIPT */
2583 d = instr(s,"perl -");
2585 d = instr(s,"perl");
2587 /* avoid getting into infinite loops when shebang
2588 * line contains "Perl" rather than "perl" */
2590 for (d = ipathend-4; d >= ipath; --d) {
2591 if ((*d == 'p' || *d == 'P')
2592 && !ibcmp(d, "perl", 4))
2602 #ifdef ALTERNATE_SHEBANG
2604 * If the ALTERNATE_SHEBANG on this system starts with a
2605 * character that can be part of a Perl expression, then if
2606 * we see it but not "perl", we're probably looking at the
2607 * start of Perl code, not a request to hand off to some
2608 * other interpreter. Similarly, if "perl" is there, but
2609 * not in the first 'word' of the line, we assume the line
2610 * contains the start of the Perl program.
2612 if (d && *s != '#') {
2614 while (*c && !strchr("; \t\r\n\f\v#", *c))
2617 d = Nullch; /* "perl" not in first word; ignore */
2619 *s = '#'; /* Don't try to parse shebang line */
2621 #endif /* ALTERNATE_SHEBANG */
2622 #ifndef MACOS_TRADITIONAL
2627 !instr(s,"indir") &&
2628 instr(PL_origargv[0],"perl"))
2634 while (s < PL_bufend && isSPACE(*s))
2636 if (s < PL_bufend) {
2637 Newz(899,newargv,PL_origargc+3,char*);
2639 while (s < PL_bufend && !isSPACE(*s))
2642 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2645 newargv = PL_origargv;
2647 PerlProc_execv(ipath, newargv);
2648 Perl_croak(aTHX_ "Can't exec %s", ipath);
2652 U32 oldpdb = PL_perldb;
2653 bool oldn = PL_minus_n;
2654 bool oldp = PL_minus_p;
2656 while (*d && !isSPACE(*d)) d++;
2657 while (SPACE_OR_TAB(*d)) d++;
2661 if (*d == 'M' || *d == 'm') {
2663 while (*d && !isSPACE(*d)) d++;
2664 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2667 d = moreswitches(d);
2669 if ((PERLDB_LINE && !oldpdb) ||
2670 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2671 /* if we have already added "LINE: while (<>) {",
2672 we must not do it again */
2674 sv_setpv(PL_linestr, "");
2675 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2676 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2677 PL_preambled = FALSE;
2679 (void)gv_fetchfile(PL_origfilename);
2686 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2688 PL_lex_state = LEX_FORMLINE;
2693 #ifdef PERL_STRICT_CR
2694 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2696 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2698 case ' ': case '\t': case '\f': case 013:
2699 #ifdef MACOS_TRADITIONAL
2706 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2707 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2708 /* handle eval qq[#line 1 "foo"\n ...] */
2709 CopLINE_dec(PL_curcop);
2713 while (s < d && *s != '\n')
2718 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2720 PL_lex_state = LEX_FORMLINE;
2730 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2735 while (s < PL_bufend && SPACE_OR_TAB(*s))
2738 if (strnEQ(s,"=>",2)) {
2739 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2740 OPERATOR('-'); /* unary minus */
2742 PL_last_uni = PL_oldbufptr;
2743 PL_last_lop_op = OP_FTEREAD; /* good enough */
2745 case 'r': FTST(OP_FTEREAD);
2746 case 'w': FTST(OP_FTEWRITE);
2747 case 'x': FTST(OP_FTEEXEC);
2748 case 'o': FTST(OP_FTEOWNED);
2749 case 'R': FTST(OP_FTRREAD);
2750 case 'W': FTST(OP_FTRWRITE);
2751 case 'X': FTST(OP_FTREXEC);
2752 case 'O': FTST(OP_FTROWNED);
2753 case 'e': FTST(OP_FTIS);
2754 case 'z': FTST(OP_FTZERO);
2755 case 's': FTST(OP_FTSIZE);
2756 case 'f': FTST(OP_FTFILE);
2757 case 'd': FTST(OP_FTDIR);
2758 case 'l': FTST(OP_FTLINK);
2759 case 'p': FTST(OP_FTPIPE);
2760 case 'S': FTST(OP_FTSOCK);
2761 case 'u': FTST(OP_FTSUID);
2762 case 'g': FTST(OP_FTSGID);
2763 case 'k': FTST(OP_FTSVTX);
2764 case 'b': FTST(OP_FTBLK);
2765 case 'c': FTST(OP_FTCHR);
2766 case 't': FTST(OP_FTTTY);
2767 case 'T': FTST(OP_FTTEXT);
2768 case 'B': FTST(OP_FTBINARY);
2769 case 'M': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTMTIME);
2770 case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME);
2771 case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME);
2773 Perl_croak(aTHX_ "Unrecognized file test: -%c", (int)tmp);
2780 if (PL_expect == XOPERATOR)
2785 else if (*s == '>') {
2788 if (isIDFIRST_lazy_if(s,UTF)) {
2789 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2797 if (PL_expect == XOPERATOR)
2800 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2802 OPERATOR('-'); /* unary minus */
2809 if (PL_expect == XOPERATOR)
2814 if (PL_expect == XOPERATOR)
2817 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2823 if (PL_expect != XOPERATOR) {
2824 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2825 PL_expect = XOPERATOR;
2826 force_ident(PL_tokenbuf, '*');
2839 if (PL_expect == XOPERATOR) {
2843 PL_tokenbuf[0] = '%';
2844 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2845 if (!PL_tokenbuf[1]) {
2847 yyerror("Final % should be \\% or %name");
2850 PL_pending_ident = '%';
2869 switch (PL_expect) {
2872 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2874 PL_bufptr = s; /* update in case we back off */
2880 PL_expect = XTERMBLOCK;
2884 while (isIDFIRST_lazy_if(s,UTF)) {
2885 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2886 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2887 if (tmp < 0) tmp = -tmp;
2902 d = scan_str(d,TRUE,TRUE);
2905 SvREFCNT_dec(PL_lex_stuff);
2906 PL_lex_stuff = Nullsv;
2908 /* MUST advance bufptr here to avoid bogus
2909 "at end of line" context messages from yyerror().
2911 PL_bufptr = s + len;
2912 yyerror("Unterminated attribute parameter in attribute list");
2915 return 0; /* EOF indicator */
2919 SV *sv = newSVpvn(s, len);
2920 sv_catsv(sv, PL_lex_stuff);
2921 attrs = append_elem(OP_LIST, attrs,
2922 newSVOP(OP_CONST, 0, sv));
2923 SvREFCNT_dec(PL_lex_stuff);
2924 PL_lex_stuff = Nullsv;
2927 attrs = append_elem(OP_LIST, attrs,
2928 newSVOP(OP_CONST, 0,
2932 if (*s == ':' && s[1] != ':')
2935 break; /* require real whitespace or :'s */
2937 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
2938 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
2939 char q = ((*s == '\'') ? '"' : '\'');
2940 /* If here for an expression, and parsed no attrs, back off. */
2941 if (tmp == '=' && !attrs) {
2945 /* MUST advance bufptr here to avoid bogus "at end of line"
2946 context messages from yyerror().
2950 yyerror("Unterminated attribute list");
2952 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
2960 PL_nextval[PL_nexttoke].opval = attrs;
2968 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
2969 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
2985 if (PL_lex_brackets <= 0)
2986 yyerror("Unmatched right square bracket");
2989 if (PL_lex_state == LEX_INTERPNORMAL) {
2990 if (PL_lex_brackets == 0) {
2991 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
2992 PL_lex_state = LEX_INTERPEND;
2999 if (PL_lex_brackets > 100) {
3000 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
3001 if (newlb != PL_lex_brackstack) {
3003 PL_lex_brackstack = newlb;
3006 switch (PL_expect) {
3008 if (PL_lex_formbrack) {
3012 if (PL_oldoldbufptr == PL_last_lop)
3013 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3015 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3016 OPERATOR(HASHBRACK);
3018 while (s < PL_bufend && SPACE_OR_TAB(*s))
3021 PL_tokenbuf[0] = '\0';
3022 if (d < PL_bufend && *d == '-') {
3023 PL_tokenbuf[0] = '-';
3025 while (d < PL_bufend && SPACE_OR_TAB(*d))
3028 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3029 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3031 while (d < PL_bufend && SPACE_OR_TAB(*d))
3034 char minus = (PL_tokenbuf[0] == '-');
3035 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3043 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3048 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3053 if (PL_oldoldbufptr == PL_last_lop)
3054 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3056 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3059 OPERATOR(HASHBRACK);
3060 /* This hack serves to disambiguate a pair of curlies
3061 * as being a block or an anon hash. Normally, expectation
3062 * determines that, but in cases where we're not in a
3063 * position to expect anything in particular (like inside
3064 * eval"") we have to resolve the ambiguity. This code
3065 * covers the case where the first term in the curlies is a
3066 * quoted string. Most other cases need to be explicitly
3067 * disambiguated by prepending a `+' before the opening
3068 * curly in order to force resolution as an anon hash.
3070 * XXX should probably propagate the outer expectation
3071 * into eval"" to rely less on this hack, but that could
3072 * potentially break current behavior of eval"".
3076 if (*s == '\'' || *s == '"' || *s == '`') {
3077 /* common case: get past first string, handling escapes */
3078 for (t++; t < PL_bufend && *t != *s;)
3079 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3083 else if (*s == 'q') {
3086 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3090 char open, close, term;
3093 while (t < PL_bufend && isSPACE(*t))
3097 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3101 for (t++; t < PL_bufend; t++) {
3102 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3104 else if (*t == open)
3108 for (t++; t < PL_bufend; t++) {
3109 if (*t == '\\' && t+1 < PL_bufend)
3111 else if (*t == close && --brackets <= 0)
3113 else if (*t == open)
3119 else if (isALNUM_lazy_if(t,UTF)) {
3121 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3124 while (t < PL_bufend && isSPACE(*t))
3126 /* if comma follows first term, call it an anon hash */
3127 /* XXX it could be a comma expression with loop modifiers */
3128 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3129 || (*t == '=' && t[1] == '>')))
3130 OPERATOR(HASHBRACK);
3131 if (PL_expect == XREF)
3134 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3140 yylval.ival = CopLINE(PL_curcop);
3141 if (isSPACE(*s) || *s == '#')
3142 PL_copline = NOLINE; /* invalidate current command line number */
3147 if (PL_lex_brackets <= 0)
3148 yyerror("Unmatched right curly bracket");
3150 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3151 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3152 PL_lex_formbrack = 0;
3153 if (PL_lex_state == LEX_INTERPNORMAL) {
3154 if (PL_lex_brackets == 0) {
3155 if (PL_expect & XFAKEBRACK) {
3156 PL_expect &= XENUMMASK;
3157 PL_lex_state = LEX_INTERPEND;
3159 return yylex(); /* ignore fake brackets */
3161 if (*s == '-' && s[1] == '>')
3162 PL_lex_state = LEX_INTERPENDMAYBE;
3163 else if (*s != '[' && *s != '{')
3164 PL_lex_state = LEX_INTERPEND;
3167 if (PL_expect & XFAKEBRACK) {
3168 PL_expect &= XENUMMASK;
3170 return yylex(); /* ignore fake brackets */
3180 if (PL_expect == XOPERATOR) {
3181 if (ckWARN(WARN_SEMICOLON)
3182 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3184 CopLINE_dec(PL_curcop);
3185 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3186 CopLINE_inc(PL_curcop);
3191 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3193 PL_expect = XOPERATOR;
3194 force_ident(PL_tokenbuf, '&');
3198 yylval.ival = (OPpENTERSUB_AMPER<<8);
3217 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3218 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3220 if (PL_expect == XSTATE && isALPHA(tmp) &&
3221 (s == PL_linestart+1 || s[-2] == '\n') )
3223 if (PL_in_eval && !PL_rsfp) {
3228 if (strnEQ(s,"=cut",4)) {
3242 PL_doextract = TRUE;
3245 if (PL_lex_brackets < PL_lex_formbrack) {
3247 #ifdef PERL_STRICT_CR
3248 for (t = s; SPACE_OR_TAB(*t); t++) ;
3250 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3252 if (*t == '\n' || *t == '#') {
3270 if (PL_expect != XOPERATOR) {
3271 if (s[1] != '<' && !strchr(s,'>'))
3274 s = scan_heredoc(s);
3276 s = scan_inputsymbol(s);
3277 TERM(sublex_start());
3282 SHop(OP_LEFT_SHIFT);
3296 SHop(OP_RIGHT_SHIFT);
3305 if (PL_expect == XOPERATOR) {
3306 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3309 return ','; /* grandfather non-comma-format format */
3313 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3314 PL_tokenbuf[0] = '@';
3315 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3316 sizeof PL_tokenbuf - 1, FALSE);
3317 if (PL_expect == XOPERATOR)
3318 no_op("Array length", s);
3319 if (!PL_tokenbuf[1])
3321 PL_expect = XOPERATOR;
3322 PL_pending_ident = '#';
3326 PL_tokenbuf[0] = '$';
3327 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3328 sizeof PL_tokenbuf - 1, FALSE);
3329 if (PL_expect == XOPERATOR)
3331 if (!PL_tokenbuf[1]) {
3333 yyerror("Final $ should be \\$ or $name");
3337 /* This kludge not intended to be bulletproof. */
3338 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3339 yylval.opval = newSVOP(OP_CONST, 0,
3340 newSViv(PL_compiling.cop_arybase));
3341 yylval.opval->op_private = OPpCONST_ARYBASE;
3347 if (PL_lex_state == LEX_NORMAL)
3350 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3353 PL_tokenbuf[0] = '@';
3354 if (ckWARN(WARN_SYNTAX)) {
3356 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3359 PL_bufptr = skipspace(PL_bufptr);
3360 while (t < PL_bufend && *t != ']')
3362 Perl_warner(aTHX_ WARN_SYNTAX,
3363 "Multidimensional syntax %.*s not supported",
3364 (t - PL_bufptr) + 1, PL_bufptr);
3368 else if (*s == '{') {
3369 PL_tokenbuf[0] = '%';
3370 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3371 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3373 char tmpbuf[sizeof PL_tokenbuf];
3375 for (t++; isSPACE(*t); t++) ;
3376 if (isIDFIRST_lazy_if(t,UTF)) {
3377 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3378 for (; isSPACE(*t); t++) ;
3379 if (*t == ';' && get_cv(tmpbuf, FALSE))
3380 Perl_warner(aTHX_ WARN_SYNTAX,
3381 "You need to quote \"%s\"", tmpbuf);
3387 PL_expect = XOPERATOR;
3388 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3389 bool islop = (PL_last_lop == PL_oldoldbufptr);
3390 if (!islop || PL_last_lop_op == OP_GREPSTART)
3391 PL_expect = XOPERATOR;
3392 else if (strchr("$@\"'`q", *s))
3393 PL_expect = XTERM; /* e.g. print $fh "foo" */
3394 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3395 PL_expect = XTERM; /* e.g. print $fh &sub */
3396 else if (isIDFIRST_lazy_if(s,UTF)) {
3397 char tmpbuf[sizeof PL_tokenbuf];
3398 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3399 if ((tmp = keyword(tmpbuf, len))) {
3400 /* binary operators exclude handle interpretations */
3412 PL_expect = XTERM; /* e.g. print $fh length() */
3417 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3418 if (gv && GvCVu(gv))
3419 PL_expect = XTERM; /* e.g. print $fh subr() */
3422 else if (isDIGIT(*s))
3423 PL_expect = XTERM; /* e.g. print $fh 3 */
3424 else if (*s == '.' && isDIGIT(s[1]))
3425 PL_expect = XTERM; /* e.g. print $fh .3 */
3426 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3427 PL_expect = XTERM; /* e.g. print $fh -1 */
3428 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3429 PL_expect = XTERM; /* print $fh <<"EOF" */
3431 PL_pending_ident = '$';
3435 if (PL_expect == XOPERATOR)
3437 PL_tokenbuf[0] = '@';
3438 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3439 if (!PL_tokenbuf[1]) {
3441 yyerror("Final @ should be \\@ or @name");
3444 if (PL_lex_state == LEX_NORMAL)
3446 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3448 PL_tokenbuf[0] = '%';
3450 /* Warn about @ where they meant $. */
3451 if (ckWARN(WARN_SYNTAX)) {
3452 if (*s == '[' || *s == '{') {
3454 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3456 if (*t == '}' || *t == ']') {
3458 PL_bufptr = skipspace(PL_bufptr);
3459 Perl_warner(aTHX_ WARN_SYNTAX,
3460 "Scalar value %.*s better written as $%.*s",
3461 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3466 PL_pending_ident = '@';
3469 case '/': /* may either be division or pattern */
3470 case '?': /* may either be conditional or pattern */
3471 if (PL_expect != XOPERATOR) {
3472 /* Disable warning on "study /blah/" */
3473 if (PL_oldoldbufptr == PL_last_uni
3474 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3475 || memNE(PL_last_uni, "study", 5)
3476 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3478 s = scan_pat(s,OP_MATCH);
3479 TERM(sublex_start());
3487 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3488 #ifdef PERL_STRICT_CR
3491 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3493 && (s == PL_linestart || s[-1] == '\n') )
3495 PL_lex_formbrack = 0;
3499 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3505 yylval.ival = OPf_SPECIAL;
3511 if (PL_expect != XOPERATOR)
3516 case '0': case '1': case '2': case '3': case '4':
3517 case '5': case '6': case '7': case '8': case '9':
3519 if (PL_expect == XOPERATOR)
3524 s = scan_str(s,FALSE,FALSE);
3525 if (PL_expect == XOPERATOR) {
3526 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3529 return ','; /* grandfather non-comma-format format */
3535 missingterm((char*)0);
3536 yylval.ival = OP_CONST;
3537 TERM(sublex_start());
3540 s = scan_str(s,FALSE,FALSE);
3541 if (PL_expect == XOPERATOR) {
3542 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3545 return ','; /* grandfather non-comma-format format */
3551 missingterm((char*)0);
3552 yylval.ival = OP_CONST;
3553 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3554 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3555 yylval.ival = OP_STRINGIFY;
3559 TERM(sublex_start());
3562 s = scan_str(s,FALSE,FALSE);
3563 if (PL_expect == XOPERATOR)
3564 no_op("Backticks",s);
3566 missingterm((char*)0);
3567 yylval.ival = OP_BACKTICK;
3569 TERM(sublex_start());
3573 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3574 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3576 if (PL_expect == XOPERATOR)
3577 no_op("Backslash",s);
3581 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3585 while (isDIGIT(*start) || *start == '_')
3587 if (*start == '.' && isDIGIT(start[1])) {
3591 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3592 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF)) {
3596 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3606 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3645 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3647 /* Some keywords can be followed by any delimiter, including ':' */
3648 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3649 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3650 (PL_tokenbuf[0] == 'q' &&
3651 strchr("qwxr", PL_tokenbuf[1])))));
3653 /* x::* is just a word, unless x is "CORE" */
3654 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3658 while (d < PL_bufend && isSPACE(*d))
3659 d++; /* no comments skipped here, or s### is misparsed */
3661 /* Is this a label? */
3662 if (!tmp && PL_expect == XSTATE
3663 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3665 yylval.pval = savepv(PL_tokenbuf);
3670 /* Check for keywords */
3671 tmp = keyword(PL_tokenbuf, len);
3673 /* Is this a word before a => operator? */
3674 if (*d == '=' && d[1] == '>') {
3676 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3677 yylval.opval->op_private = OPpCONST_BARE;
3681 if (tmp < 0) { /* second-class keyword? */
3682 GV *ogv = Nullgv; /* override (winner) */
3683 GV *hgv = Nullgv; /* hidden (loser) */
3684 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3686 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3689 if (GvIMPORTED_CV(gv))
3691 else if (! CvMETHOD(cv))
3695 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3696 (gv = *gvp) != (GV*)&PL_sv_undef &&
3697 GvCVu(gv) && GvIMPORTED_CV(gv))
3703 tmp = 0; /* overridden by import or by GLOBAL */
3706 && -tmp==KEY_lock /* XXX generalizable kludge */
3708 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3710 tmp = 0; /* any sub overrides "weak" keyword */
3712 else { /* no override */
3716 if (ckWARN(WARN_AMBIGUOUS) && hgv
3717 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3718 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3719 "Ambiguous call resolved as CORE::%s(), %s",
3720 GvENAME(hgv), "qualify as such or use &");
3727 default: /* not a keyword */
3730 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3732 /* Get the rest if it looks like a package qualifier */
3734 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3736 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3739 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3740 *s == '\'' ? "'" : "::");
3744 if (PL_expect == XOPERATOR) {
3745 if (PL_bufptr == PL_linestart) {
3746 CopLINE_dec(PL_curcop);
3747 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3748 CopLINE_inc(PL_curcop);
3751 no_op("Bareword",s);
3754 /* Look for a subroutine with this name in current package,
3755 unless name is "Foo::", in which case Foo is a bearword
3756 (and a package name). */
3759 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3761 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3762 Perl_warner(aTHX_ WARN_BAREWORD,
3763 "Bareword \"%s\" refers to nonexistent package",
3766 PL_tokenbuf[len] = '\0';
3773 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3776 /* if we saw a global override before, get the right name */
3779 sv = newSVpvn("CORE::GLOBAL::",14);
3780 sv_catpv(sv,PL_tokenbuf);
3783 sv = newSVpv(PL_tokenbuf,0);
3785 /* Presume this is going to be a bareword of some sort. */
3788 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3789 yylval.opval->op_private = OPpCONST_BARE;
3791 /* And if "Foo::", then that's what it certainly is. */
3796 /* See if it's the indirect object for a list operator. */
3798 if (PL_oldoldbufptr &&
3799 PL_oldoldbufptr < PL_bufptr &&
3800 (PL_oldoldbufptr == PL_last_lop
3801 || PL_oldoldbufptr == PL_last_uni) &&
3802 /* NO SKIPSPACE BEFORE HERE! */
3803 (PL_expect == XREF ||
3804 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3806 bool immediate_paren = *s == '(';
3808 /* (Now we can afford to cross potential line boundary.) */
3811 /* Two barewords in a row may indicate method call. */
3813 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3816 /* If not a declared subroutine, it's an indirect object. */
3817 /* (But it's an indir obj regardless for sort.) */
3819 if ((PL_last_lop_op == OP_SORT ||
3820 (!immediate_paren && (!gv || !GvCVu(gv)))) &&
3821 (PL_last_lop_op != OP_MAPSTART &&
3822 PL_last_lop_op != OP_GREPSTART))
3824 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3830 PL_expect = XOPERATOR;
3833 /* Is this a word before a => operator? */
3834 if (*s == '=' && s[1] == '>') {
3836 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
3840 /* If followed by a paren, it's certainly a subroutine. */
3843 if (gv && GvCVu(gv)) {
3844 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
3845 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3850 PL_nextval[PL_nexttoke].opval = yylval.opval;
3851 PL_expect = XOPERATOR;
3857 /* If followed by var or block, call it a method (unless sub) */
3859 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3860 PL_last_lop = PL_oldbufptr;
3861 PL_last_lop_op = OP_METHOD;
3865 /* If followed by a bareword, see if it looks like indir obj. */
3867 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3870 /* Not a method, so call it a subroutine (if defined) */
3872 if (gv && GvCVu(gv)) {
3874 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3875 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3876 "Ambiguous use of -%s resolved as -&%s()",
3877 PL_tokenbuf, PL_tokenbuf);
3878 /* Check for a constant sub */
3880 if ((sv = cv_const_sv(cv))) {
3882 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3883 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3884 yylval.opval->op_private = 0;
3888 /* Resolve to GV now. */
3889 op_free(yylval.opval);
3890 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3891 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3892 PL_last_lop = PL_oldbufptr;
3893 PL_last_lop_op = OP_ENTERSUB;
3894 /* Is there a prototype? */
3897 char *proto = SvPV((SV*)cv, len);
3900 if (strEQ(proto, "$"))
3902 if (*proto == '&' && *s == '{') {
3903 sv_setpv(PL_subname,"__ANON__");
3907 PL_nextval[PL_nexttoke].opval = yylval.opval;
3913 /* Call it a bare word */
3915 if (PL_hints & HINT_STRICT_SUBS)
3916 yylval.opval->op_private |= OPpCONST_STRICT;
3919 if (ckWARN(WARN_RESERVED)) {
3920 if (lastchar != '-') {
3921 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
3923 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
3930 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
3931 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3932 "Operator or semicolon missing before %c%s",
3933 lastchar, PL_tokenbuf);
3934 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3935 "Ambiguous use of %c resolved as operator %c",
3936 lastchar, lastchar);
3942 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3943 newSVpv(CopFILE(PL_curcop),0));
3947 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3948 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
3951 case KEY___PACKAGE__:
3952 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3954 ? newSVsv(PL_curstname)
3963 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
3964 char *pname = "main";
3965 if (PL_tokenbuf[2] == 'D')
3966 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
3967 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
3970 GvIOp(gv) = newIO();
3971 IoIFP(GvIOp(gv)) = PL_rsfp;
3972 #if defined(HAS_FCNTL) && defined(F_SETFD)
3974 int fd = PerlIO_fileno(PL_rsfp);
3975 fcntl(fd,F_SETFD,fd >= 3);
3978 /* Mark this internal pseudo-handle as clean */
3979 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
3981 IoTYPE(GvIOp(gv)) = '|';
3982 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
3983 IoTYPE(GvIOp(gv)) = '-';
3985 IoTYPE(GvIOp(gv)) = '<';
3986 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
3987 /* if the script was opened in binmode, we need to revert
3988 * it to text mode for compatibility; but only iff it has CRs
3989 * XXX this is a questionable hack at best. */
3990 if (PL_bufend-PL_bufptr > 2
3991 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
3994 if (IoTYPE(GvIOp(gv)) == '<') {
3995 loc = PerlIO_tell(PL_rsfp);
3996 (void)PerlIO_seek(PL_rsfp, 0L, 0);
3998 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
3999 #if defined(__BORLANDC__)
4000 /* XXX see note in do_binmode() */
4001 ((FILE*)PL_rsfp)->flags |= _F_BIN;
4004 PerlIO_seek(PL_rsfp, loc, 0);
4019 if (PL_expect == XSTATE) {
4026 if (*s == ':' && s[1] == ':') {
4029 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4030 if (!(tmp = keyword(PL_tokenbuf, len)))
4031 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4045 LOP(OP_ACCEPT,XTERM);
4051 LOP(OP_ATAN2,XTERM);
4057 LOP(OP_BINMODE,XTERM);
4060 LOP(OP_BLESS,XTERM);
4069 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4086 if (!PL_cryptseen) {
4087 PL_cryptseen = TRUE;
4091 LOP(OP_CRYPT,XTERM);
4094 if (ckWARN(WARN_CHMOD)) {
4095 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4096 if (*d != '0' && isDIGIT(*d))
4097 Perl_warner(aTHX_ WARN_CHMOD,
4098 "chmod() mode argument is missing initial 0");
4100 LOP(OP_CHMOD,XTERM);
4103 LOP(OP_CHOWN,XTERM);
4106 LOP(OP_CONNECT,XTERM);
4122 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4126 PL_hints |= HINT_BLOCK_SCOPE;
4136 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4137 LOP(OP_DBMOPEN,XTERM);
4143 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4150 yylval.ival = CopLINE(PL_curcop);
4164 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4165 UNIBRACK(OP_ENTEREVAL);
4180 case KEY_endhostent:
4186 case KEY_endservent:
4189 case KEY_endprotoent:
4200 yylval.ival = CopLINE(PL_curcop);
4202 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4204 if ((PL_bufend - p) >= 3 &&
4205 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4207 else if ((PL_bufend - p) >= 4 &&
4208 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4211 if (isIDFIRST_lazy_if(p,UTF)) {
4212 p = scan_ident(p, PL_bufend,
4213 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4217 Perl_croak(aTHX_ "Missing $ on loop variable");
4222 LOP(OP_FORMLINE,XTERM);
4228 LOP(OP_FCNTL,XTERM);
4234 LOP(OP_FLOCK,XTERM);
4243 LOP(OP_GREPSTART, XREF);
4246 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4261 case KEY_getpriority:
4262 LOP(OP_GETPRIORITY,XTERM);
4264 case KEY_getprotobyname:
4267 case KEY_getprotobynumber:
4268 LOP(OP_GPBYNUMBER,XTERM);
4270 case KEY_getprotoent:
4282 case KEY_getpeername:
4283 UNI(OP_GETPEERNAME);
4285 case KEY_gethostbyname:
4288 case KEY_gethostbyaddr:
4289 LOP(OP_GHBYADDR,XTERM);
4291 case KEY_gethostent:
4294 case KEY_getnetbyname:
4297 case KEY_getnetbyaddr:
4298 LOP(OP_GNBYADDR,XTERM);
4303 case KEY_getservbyname:
4304 LOP(OP_GSBYNAME,XTERM);
4306 case KEY_getservbyport:
4307 LOP(OP_GSBYPORT,XTERM);
4309 case KEY_getservent:
4312 case KEY_getsockname:
4313 UNI(OP_GETSOCKNAME);
4315 case KEY_getsockopt:
4316 LOP(OP_GSOCKOPT,XTERM);
4338 yylval.ival = CopLINE(PL_curcop);
4342 LOP(OP_INDEX,XTERM);
4348 LOP(OP_IOCTL,XTERM);
4360 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4392 LOP(OP_LISTEN,XTERM);
4401 s = scan_pat(s,OP_MATCH);
4402 TERM(sublex_start());
4405 LOP(OP_MAPSTART, XREF);
4408 LOP(OP_MKDIR,XTERM);
4411 LOP(OP_MSGCTL,XTERM);
4414 LOP(OP_MSGGET,XTERM);
4417 LOP(OP_MSGRCV,XTERM);
4420 LOP(OP_MSGSND,XTERM);
4426 if (isIDFIRST_lazy_if(s,UTF)) {
4427 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4428 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4430 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4431 if (!PL_in_my_stash) {
4434 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4442 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4449 if (PL_expect != XSTATE)
4450 yyerror("\"no\" not allowed in expression");
4451 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4452 s = force_version(s);
4457 if (*s == '(' || (s = skipspace(s), *s == '('))
4464 if (isIDFIRST_lazy_if(s,UTF)) {
4466 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4468 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4469 Perl_warner(aTHX_ WARN_PRECEDENCE,
4470 "Precedence problem: open %.*s should be open(%.*s)",
4476 yylval.ival = OP_OR;
4486 LOP(OP_OPEN_DIR,XTERM);
4489 checkcomma(s,PL_tokenbuf,"filehandle");
4493 checkcomma(s,PL_tokenbuf,"filehandle");
4512 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4516 LOP(OP_PIPE_OP,XTERM);
4519 s = scan_str(s,FALSE,FALSE);
4521 missingterm((char*)0);
4522 yylval.ival = OP_CONST;
4523 TERM(sublex_start());
4529 s = scan_str(s,FALSE,FALSE);
4531 missingterm((char*)0);
4533 if (SvCUR(PL_lex_stuff)) {
4536 d = SvPV_force(PL_lex_stuff, len);
4538 for (; isSPACE(*d) && len; --len, ++d) ;
4541 if (!warned && ckWARN(WARN_QW)) {
4542 for (; !isSPACE(*d) && len; --len, ++d) {
4544 Perl_warner(aTHX_ WARN_QW,
4545 "Possible attempt to separate words with commas");
4548 else if (*d == '#') {
4549 Perl_warner(aTHX_ WARN_QW,
4550 "Possible attempt to put comments in qw() list");
4556 for (; !isSPACE(*d) && len; --len, ++d) ;
4558 words = append_elem(OP_LIST, words,
4559 newSVOP(OP_CONST, 0, tokeq(newSVpvn(b, d-b))));
4563 PL_nextval[PL_nexttoke].opval = words;
4568 SvREFCNT_dec(PL_lex_stuff);
4569 PL_lex_stuff = Nullsv;
4574 s = scan_str(s,FALSE,FALSE);
4576 missingterm((char*)0);
4577 yylval.ival = OP_STRINGIFY;
4578 if (SvIVX(PL_lex_stuff) == '\'')
4579 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4580 TERM(sublex_start());
4583 s = scan_pat(s,OP_QR);
4584 TERM(sublex_start());
4587 s = scan_str(s,FALSE,FALSE);
4589 missingterm((char*)0);
4590 yylval.ival = OP_BACKTICK;
4592 TERM(sublex_start());
4599 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4600 s = force_version(s);
4603 *PL_tokenbuf = '\0';
4604 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4605 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4606 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4608 yyerror("<> should be quotes");
4616 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4620 LOP(OP_RENAME,XTERM);
4629 LOP(OP_RINDEX,XTERM);
4652 LOP(OP_REVERSE,XTERM);
4663 TERM(sublex_start());
4665 TOKEN(1); /* force error */
4674 LOP(OP_SELECT,XTERM);
4680 LOP(OP_SEMCTL,XTERM);
4683 LOP(OP_SEMGET,XTERM);
4686 LOP(OP_SEMOP,XTERM);
4692 LOP(OP_SETPGRP,XTERM);
4694 case KEY_setpriority:
4695 LOP(OP_SETPRIORITY,XTERM);
4697 case KEY_sethostent:
4703 case KEY_setservent:
4706 case KEY_setprotoent:
4716 LOP(OP_SEEKDIR,XTERM);
4718 case KEY_setsockopt:
4719 LOP(OP_SSOCKOPT,XTERM);
4725 LOP(OP_SHMCTL,XTERM);
4728 LOP(OP_SHMGET,XTERM);
4731 LOP(OP_SHMREAD,XTERM);
4734 LOP(OP_SHMWRITE,XTERM);
4737 LOP(OP_SHUTDOWN,XTERM);
4746 LOP(OP_SOCKET,XTERM);
4748 case KEY_socketpair:
4749 LOP(OP_SOCKPAIR,XTERM);
4752 checkcomma(s,PL_tokenbuf,"subroutine name");
4754 if (*s == ';' || *s == ')') /* probably a close */
4755 Perl_croak(aTHX_ "sort is now a reserved word");
4757 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4761 LOP(OP_SPLIT,XTERM);
4764 LOP(OP_SPRINTF,XTERM);
4767 LOP(OP_SPLICE,XTERM);
4782 LOP(OP_SUBSTR,XTERM);
4788 char tmpbuf[sizeof PL_tokenbuf];
4790 expectation attrful;
4791 bool have_name, have_proto;
4796 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4797 (*s == ':' && s[1] == ':'))
4800 attrful = XATTRBLOCK;
4801 /* remember buffer pos'n for later force_word */
4802 tboffset = s - PL_oldbufptr;
4803 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4804 if (strchr(tmpbuf, ':'))
4805 sv_setpv(PL_subname, tmpbuf);
4807 sv_setsv(PL_subname,PL_curstname);
4808 sv_catpvn(PL_subname,"::",2);
4809 sv_catpvn(PL_subname,tmpbuf,len);
4816 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4817 PL_expect = XTERMBLOCK;
4818 attrful = XATTRTERM;
4819 sv_setpv(PL_subname,"?");
4823 if (key == KEY_format) {
4825 PL_lex_formbrack = PL_lex_brackets + 1;
4827 (void) force_word(PL_oldbufptr + tboffset, WORD,
4832 /* Look for a prototype */
4836 s = scan_str(s,FALSE,FALSE);
4839 SvREFCNT_dec(PL_lex_stuff);
4840 PL_lex_stuff = Nullsv;
4841 Perl_croak(aTHX_ "Prototype not terminated");
4844 d = SvPVX(PL_lex_stuff);
4846 for (p = d; *p; ++p) {
4851 SvCUR(PL_lex_stuff) = tmp;
4859 if (*s == ':' && s[1] != ':')
4860 PL_expect = attrful;
4863 PL_nextval[PL_nexttoke].opval =
4864 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4865 PL_lex_stuff = Nullsv;
4869 sv_setpv(PL_subname,"__ANON__");
4872 (void) force_word(PL_oldbufptr + tboffset, WORD,
4881 LOP(OP_SYSTEM,XREF);
4884 LOP(OP_SYMLINK,XTERM);
4887 LOP(OP_SYSCALL,XTERM);
4890 LOP(OP_SYSOPEN,XTERM);
4893 LOP(OP_SYSSEEK,XTERM);
4896 LOP(OP_SYSREAD,XTERM);
4899 LOP(OP_SYSWRITE,XTERM);
4903 TERM(sublex_start());
4924 LOP(OP_TRUNCATE,XTERM);
4936 yylval.ival = CopLINE(PL_curcop);
4940 yylval.ival = CopLINE(PL_curcop);
4944 LOP(OP_UNLINK,XTERM);
4950 LOP(OP_UNPACK,XTERM);
4953 LOP(OP_UTIME,XTERM);
4956 if (ckWARN(WARN_UMASK)) {
4957 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4958 if (*d != '0' && isDIGIT(*d))
4959 Perl_warner(aTHX_ WARN_UMASK,
4960 "umask: argument is missing initial 0");
4965 LOP(OP_UNSHIFT,XTERM);
4968 if (PL_expect != XSTATE)
4969 yyerror("\"use\" not allowed in expression");
4971 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4972 s = force_version(s);
4973 if (*s == ';' || (s = skipspace(s), *s == ';')) {
4974 PL_nextval[PL_nexttoke].opval = Nullop;
4979 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4980 s = force_version(s);
4992 yylval.ival = CopLINE(PL_curcop);
4996 PL_hints |= HINT_BLOCK_SCOPE;
5003 LOP(OP_WAITPID,XTERM);
5011 static char ctl_l[2];
5013 if (ctl_l[0] == '\0')
5014 ctl_l[0] = toCTRL('L');
5015 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5018 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5023 if (PL_expect == XOPERATOR)
5029 yylval.ival = OP_XOR;
5034 TERM(sublex_start());
5039 #pragma segment Main
5043 Perl_keyword(pTHX_ register char *d, I32 len)
5048 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5049 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5050 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5051 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5052 if (strEQ(d,"__END__")) return KEY___END__;
5056 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5061 if (strEQ(d,"and")) return -KEY_and;
5062 if (strEQ(d,"abs")) return -KEY_abs;
5065 if (strEQ(d,"alarm")) return -KEY_alarm;
5066 if (strEQ(d,"atan2")) return -KEY_atan2;
5069 if (strEQ(d,"accept")) return -KEY_accept;
5074 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5077 if (strEQ(d,"bless")) return -KEY_bless;
5078 if (strEQ(d,"bind")) return -KEY_bind;
5079 if (strEQ(d,"binmode")) return -KEY_binmode;
5082 if (strEQ(d,"CORE")) return -KEY_CORE;
5083 if (strEQ(d,"CHECK")) return KEY_CHECK;
5088 if (strEQ(d,"cmp")) return -KEY_cmp;
5089 if (strEQ(d,"chr")) return -KEY_chr;
5090 if (strEQ(d,"cos")) return -KEY_cos;
5093 if (strEQ(d,"chop")) return KEY_chop;
5096 if (strEQ(d,"close")) return -KEY_close;
5097 if (strEQ(d,"chdir")) return -KEY_chdir;
5098 if (strEQ(d,"chomp")) return KEY_chomp;
5099 if (strEQ(d,"chmod")) return -KEY_chmod;
5100 if (strEQ(d,"chown")) return -KEY_chown;
5101 if (strEQ(d,"crypt")) return -KEY_crypt;
5104 if (strEQ(d,"chroot")) return -KEY_chroot;
5105 if (strEQ(d,"caller")) return -KEY_caller;
5108 if (strEQ(d,"connect")) return -KEY_connect;
5111 if (strEQ(d,"closedir")) return -KEY_closedir;
5112 if (strEQ(d,"continue")) return -KEY_continue;
5117 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5122 if (strEQ(d,"do")) return KEY_do;
5125 if (strEQ(d,"die")) return -KEY_die;
5128 if (strEQ(d,"dump")) return -KEY_dump;
5131 if (strEQ(d,"delete")) return KEY_delete;
5134 if (strEQ(d,"defined")) return KEY_defined;
5135 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5138 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5143 if (strEQ(d,"EQ")) { deprecate(d); return -KEY_eq;}
5144 if (strEQ(d,"END")) return KEY_END;
5149 if (strEQ(d,"eq")) return -KEY_eq;
5152 if (strEQ(d,"eof")) return -KEY_eof;
5153 if (strEQ(d,"exp")) return -KEY_exp;
5156 if (strEQ(d,"else")) return KEY_else;
5157 if (strEQ(d,"exit")) return -KEY_exit;
5158 if (strEQ(d,"eval")) return KEY_eval;
5159 if (strEQ(d,"exec")) return -KEY_exec;
5160 if (strEQ(d,"each")) return KEY_each;
5163 if (strEQ(d,"elsif")) return KEY_elsif;
5166 if (strEQ(d,"exists")) return KEY_exists;
5167 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5170 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5171 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5174 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5177 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5178 if (strEQ(d,"endservent")) return -KEY_endservent;
5181 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5188 if (strEQ(d,"for")) return KEY_for;
5191 if (strEQ(d,"fork")) return -KEY_fork;
5194 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5195 if (strEQ(d,"flock")) return -KEY_flock;
5198 if (strEQ(d,"format")) return KEY_format;
5199 if (strEQ(d,"fileno")) return -KEY_fileno;
5202 if (strEQ(d,"foreach")) return KEY_foreach;
5205 if (strEQ(d,"formline")) return -KEY_formline;
5211 if (strEQ(d,"GT")) { deprecate(d); return -KEY_gt;}
5212 if (strEQ(d,"GE")) { deprecate(d); return -KEY_ge;}
5216 if (strnEQ(d,"get",3)) {
5221 if (strEQ(d,"ppid")) return -KEY_getppid;
5222 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5225 if (strEQ(d,"pwent")) return -KEY_getpwent;
5226 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5227 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5230 if (strEQ(d,"peername")) return -KEY_getpeername;
5231 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5232 if (strEQ(d,"priority")) return -KEY_getpriority;
5235 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5238 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5242 else if (*d == 'h') {
5243 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5244 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5245 if (strEQ(d,"hostent")) return -KEY_gethostent;
5247 else if (*d == 'n') {
5248 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5249 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5250 if (strEQ(d,"netent")) return -KEY_getnetent;
5252 else if (*d == 's') {
5253 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5254 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5255 if (strEQ(d,"servent")) return -KEY_getservent;
5256 if (strEQ(d,"sockname")) return -KEY_getsockname;
5257 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5259 else if (*d == 'g') {
5260 if (strEQ(d,"grent")) return -KEY_getgrent;
5261 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5262 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5264 else if (*d == 'l') {
5265 if (strEQ(d,"login")) return -KEY_getlogin;
5267 else if (strEQ(d,"c")) return -KEY_getc;
5272 if (strEQ(d,"gt")) return -KEY_gt;
5273 if (strEQ(d,"ge")) return -KEY_ge;
5276 if (strEQ(d,"grep")) return KEY_grep;
5277 if (strEQ(d,"goto")) return KEY_goto;
5278 if (strEQ(d,"glob")) return KEY_glob;
5281 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5286 if (strEQ(d,"hex")) return -KEY_hex;
5289 if (strEQ(d,"INIT")) return KEY_INIT;
5294 if (strEQ(d,"if")) return KEY_if;
5297 if (strEQ(d,"int")) return -KEY_int;
5300 if (strEQ(d,"index")) return -KEY_index;
5301 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5306 if (strEQ(d,"join")) return -KEY_join;
5310 if (strEQ(d,"keys")) return KEY_keys;
5311 if (strEQ(d,"kill")) return -KEY_kill;
5316 if (strEQ(d,"LT")) { deprecate(d); return -KEY_lt;}
5317 if (strEQ(d,"LE")) { deprecate(d); return -KEY_le;}
5323 if (strEQ(d,"lt")) return -KEY_lt;
5324 if (strEQ(d,"le")) return -KEY_le;
5325 if (strEQ(d,"lc")) return -KEY_lc;
5328 if (strEQ(d,"log")) return -KEY_log;
5331 if (strEQ(d,"last")) return KEY_last;
5332 if (strEQ(d,"link")) return -KEY_link;
5333 if (strEQ(d,"lock")) return -KEY_lock;
5336 if (strEQ(d,"local")) return KEY_local;
5337 if (strEQ(d,"lstat")) return -KEY_lstat;
5340 if (strEQ(d,"length")) return -KEY_length;
5341 if (strEQ(d,"listen")) return -KEY_listen;
5344 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5347 if (strEQ(d,"localtime")) return -KEY_localtime;
5353 case 1: return KEY_m;
5355 if (strEQ(d,"my")) return KEY_my;
5358 if (strEQ(d,"map")) return KEY_map;
5361 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5364 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5365 if (strEQ(d,"msgget")) return -KEY_msgget;
5366 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5367 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5372 if (strEQ(d,"NE")) { deprecate(d); return -KEY_ne;}
5375 if (strEQ(d,"next")) return KEY_next;
5376 if (strEQ(d,"ne")) return -KEY_ne;
5377 if (strEQ(d,"not")) return -KEY_not;
5378 if (strEQ(d,"no")) return KEY_no;
5383 if (strEQ(d,"or")) return -KEY_or;
5386 if (strEQ(d,"ord")) return -KEY_ord;
5387 if (strEQ(d,"oct")) return -KEY_oct;
5388 if (strEQ(d,"our")) return KEY_our;
5391 if (strEQ(d,"open")) return -KEY_open;
5394 if (strEQ(d,"opendir")) return -KEY_opendir;
5401 if (strEQ(d,"pop")) return KEY_pop;
5402 if (strEQ(d,"pos")) return KEY_pos;
5405 if (strEQ(d,"push")) return KEY_push;
5406 if (strEQ(d,"pack")) return -KEY_pack;
5407 if (strEQ(d,"pipe")) return -KEY_pipe;
5410 if (strEQ(d,"print")) return KEY_print;
5413 if (strEQ(d,"printf")) return KEY_printf;
5416 if (strEQ(d,"package")) return KEY_package;
5419 if (strEQ(d,"prototype")) return KEY_prototype;
5424 if (strEQ(d,"q")) return KEY_q;
5425 if (strEQ(d,"qr")) return KEY_qr;
5426 if (strEQ(d,"qq")) return KEY_qq;
5427 if (strEQ(d,"qw")) return KEY_qw;
5428 if (strEQ(d,"qx")) return KEY_qx;
5430 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5435 if (strEQ(d,"ref")) return -KEY_ref;
5438 if (strEQ(d,"read")) return -KEY_read;
5439 if (strEQ(d,"rand")) return -KEY_rand;
5440 if (strEQ(d,"recv")) return -KEY_recv;
5441 if (strEQ(d,"redo")) return KEY_redo;
5444 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5445 if (strEQ(d,"reset")) return -KEY_reset;
5448 if (strEQ(d,"return")) return KEY_return;
5449 if (strEQ(d,"rename")) return -KEY_rename;
5450 if (strEQ(d,"rindex")) return -KEY_rindex;
5453 if (strEQ(d,"require")) return -KEY_require;
5454 if (strEQ(d,"reverse")) return -KEY_reverse;
5455 if (strEQ(d,"readdir")) return -KEY_readdir;
5458 if (strEQ(d,"readlink")) return -KEY_readlink;
5459 if (strEQ(d,"readline")) return -KEY_readline;
5460 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5463 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5469 case 0: return KEY_s;
5471 if (strEQ(d,"scalar")) return KEY_scalar;
5476 if (strEQ(d,"seek")) return -KEY_seek;
5477 if (strEQ(d,"send")) return -KEY_send;
5480 if (strEQ(d,"semop")) return -KEY_semop;
5483 if (strEQ(d,"select")) return -KEY_select;
5484 if (strEQ(d,"semctl")) return -KEY_semctl;
5485 if (strEQ(d,"semget")) return -KEY_semget;
5488 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5489 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5492 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5493 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5496 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5499 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5500 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5501 if (strEQ(d,"setservent")) return -KEY_setservent;
5504 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5505 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5512 if (strEQ(d,"shift")) return KEY_shift;
5515 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5516 if (strEQ(d,"shmget")) return -KEY_shmget;
5519 if (strEQ(d,"shmread")) return -KEY_shmread;
5522 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5523 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5528 if (strEQ(d,"sin")) return -KEY_sin;
5531 if (strEQ(d,"sleep")) return -KEY_sleep;
5534 if (strEQ(d,"sort")) return KEY_sort;
5535 if (strEQ(d,"socket")) return -KEY_socket;
5536 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5539 if (strEQ(d,"split")) return KEY_split;
5540 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5541 if (strEQ(d,"splice")) return KEY_splice;
5544 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5547 if (strEQ(d,"srand")) return -KEY_srand;
5550 if (strEQ(d,"stat")) return -KEY_stat;
5551 if (strEQ(d,"study")) return KEY_study;
5554 if (strEQ(d,"substr")) return -KEY_substr;
5555 if (strEQ(d,"sub")) return KEY_sub;
5560 if (strEQ(d,"system")) return -KEY_system;
5563 if (strEQ(d,"symlink")) return -KEY_symlink;
5564 if (strEQ(d,"syscall")) return -KEY_syscall;
5565 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5566 if (strEQ(d,"sysread")) return -KEY_sysread;
5567 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5570 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5579 if (strEQ(d,"tr")) return KEY_tr;
5582 if (strEQ(d,"tie")) return KEY_tie;
5585 if (strEQ(d,"tell")) return -KEY_tell;
5586 if (strEQ(d,"tied")) return KEY_tied;
5587 if (strEQ(d,"time")) return -KEY_time;
5590 if (strEQ(d,"times")) return -KEY_times;
5593 if (strEQ(d,"telldir")) return -KEY_telldir;
5596 if (strEQ(d,"truncate")) return -KEY_truncate;
5603 if (strEQ(d,"uc")) return -KEY_uc;
5606 if (strEQ(d,"use")) return KEY_use;
5609 if (strEQ(d,"undef")) return KEY_undef;
5610 if (strEQ(d,"until")) return KEY_until;
5611 if (strEQ(d,"untie")) return KEY_untie;
5612 if (strEQ(d,"utime")) return -KEY_utime;
5613 if (strEQ(d,"umask")) return -KEY_umask;
5616 if (strEQ(d,"unless")) return KEY_unless;
5617 if (strEQ(d,"unpack")) return -KEY_unpack;
5618 if (strEQ(d,"unlink")) return -KEY_unlink;
5621 if (strEQ(d,"unshift")) return KEY_unshift;
5622 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5627 if (strEQ(d,"values")) return -KEY_values;
5628 if (strEQ(d,"vec")) return -KEY_vec;
5633 if (strEQ(d,"warn")) return -KEY_warn;
5634 if (strEQ(d,"wait")) return -KEY_wait;
5637 if (strEQ(d,"while")) return KEY_while;
5638 if (strEQ(d,"write")) return -KEY_write;
5641 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5644 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5649 if (len == 1) return -KEY_x;
5650 if (strEQ(d,"xor")) return -KEY_xor;
5653 if (len == 1) return KEY_y;
5662 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5666 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5667 dTHR; /* only for ckWARN */
5668 if (ckWARN(WARN_SYNTAX)) {
5670 for (w = s+2; *w && level; w++) {
5677 for (; *w && isSPACE(*w); w++) ;
5678 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5679 Perl_warner(aTHX_ WARN_SYNTAX,
5680 "%s (...) interpreted as function",name);
5683 while (s < PL_bufend && isSPACE(*s))
5687 while (s < PL_bufend && isSPACE(*s))
5689 if (isIDFIRST_lazy_if(s,UTF)) {
5691 while (isALNUM_lazy_if(s,UTF))
5693 while (s < PL_bufend && isSPACE(*s))
5698 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5702 Perl_croak(aTHX_ "No comma allowed after %s", what);
5707 /* Either returns sv, or mortalizes sv and returns a new SV*.
5708 Best used as sv=new_constant(..., sv, ...).
5709 If s, pv are NULL, calls subroutine with one argument,
5710 and type is used with error messages only. */
5713 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5717 HV *table = GvHV(PL_hintgv); /* ^H */
5721 const char *why1, *why2, *why3;
5723 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
5726 why1 = "%^H is not consistent";
5727 why2 = strEQ(key,"charnames")
5728 ? " (missing \"use charnames ...\"?)"
5732 msg = Perl_newSVpvf(aTHX_ "constant(%s): %s%s%s",
5733 (type ? type: "undef"), why1, why2, why3);
5734 yyerror(SvPVX(msg));
5738 cvp = hv_fetch(table, key, strlen(key), FALSE);
5739 if (!cvp || !SvOK(*cvp)) {
5742 why3 = "} is not defined";
5745 sv_2mortal(sv); /* Parent created it permanently */
5748 pv = sv_2mortal(newSVpvn(s, len));
5750 typesv = sv_2mortal(newSVpv(type, 0));
5752 typesv = &PL_sv_undef;
5754 PUSHSTACKi(PERLSI_OVERLOAD);
5767 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5771 /* Check the eval first */
5772 if (!PL_in_eval && SvTRUE(ERRSV)) {
5774 sv_catpv(ERRSV, "Propagated");
5775 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5777 res = SvREFCNT_inc(sv);
5781 (void)SvREFCNT_inc(res);
5790 why1 = "Call to &{$^H{";
5792 why3 = "}} did not return a defined value";
5801 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5803 register char *d = dest;
5804 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5807 Perl_croak(aTHX_ ident_too_long);
5808 if (isALNUM(*s)) /* UTF handled below */
5810 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
5815 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5819 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5820 char *t = s + UTF8SKIP(s);
5821 while (*t & 0x80 && is_utf8_mark((U8*)t))
5823 if (d + (t - s) > e)
5824 Perl_croak(aTHX_ ident_too_long);
5825 Copy(s, d, t - s, char);
5838 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5848 e = d + destlen - 3; /* two-character token, ending NUL */
5850 while (isDIGIT(*s)) {
5852 Perl_croak(aTHX_ ident_too_long);
5859 Perl_croak(aTHX_ ident_too_long);
5860 if (isALNUM(*s)) /* UTF handled below */
5862 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
5867 else if (*s == ':' && s[1] == ':') {
5871 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5872 char *t = s + UTF8SKIP(s);
5873 while (*t & 0x80 && is_utf8_mark((U8*)t))
5875 if (d + (t - s) > e)
5876 Perl_croak(aTHX_ ident_too_long);
5877 Copy(s, d, t - s, char);
5888 if (PL_lex_state != LEX_NORMAL)
5889 PL_lex_state = LEX_INTERPENDMAYBE;
5892 if (*s == '$' && s[1] &&
5893 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
5906 if (*d == '^' && *s && isCONTROLVAR(*s)) {
5911 if (isSPACE(s[-1])) {
5914 if (!SPACE_OR_TAB(ch)) {
5920 if (isIDFIRST_lazy_if(d,UTF)) {
5924 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
5926 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
5929 Copy(s, d, e - s, char);
5934 while ((isALNUM(*s) || *s == ':') && d < e)
5937 Perl_croak(aTHX_ ident_too_long);
5940 while (s < send && SPACE_OR_TAB(*s)) s++;
5941 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
5942 dTHR; /* only for ckWARN */
5943 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
5944 const char *brack = *s == '[' ? "[...]" : "{...}";
5945 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5946 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
5947 funny, dest, brack, funny, dest, brack);
5950 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
5954 /* Handle extended ${^Foo} variables
5955 * 1999-02-27 mjd-perl-patch@plover.com */
5956 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
5960 while (isALNUM(*s) && d < e) {
5964 Perl_croak(aTHX_ ident_too_long);
5969 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
5970 PL_lex_state = LEX_INTERPEND;
5973 if (PL_lex_state == LEX_NORMAL) {
5974 dTHR; /* only for ckWARN */
5975 if (ckWARN(WARN_AMBIGUOUS) &&
5976 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
5978 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5979 "Ambiguous use of %c{%s} resolved to %c%s",
5980 funny, dest, funny, dest);
5985 s = bracket; /* let the parser handle it */
5989 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
5990 PL_lex_state = LEX_INTERPEND;
5995 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
6000 *pmfl |= PMf_GLOBAL;
6002 *pmfl |= PMf_CONTINUE;
6006 *pmfl |= PMf_MULTILINE;
6008 *pmfl |= PMf_SINGLELINE;
6010 *pmfl |= PMf_EXTENDED;
6014 S_scan_pat(pTHX_ char *start, I32 type)
6019 s = scan_str(start,FALSE,FALSE);
6022 SvREFCNT_dec(PL_lex_stuff);
6023 PL_lex_stuff = Nullsv;
6024 Perl_croak(aTHX_ "Search pattern not terminated");
6027 pm = (PMOP*)newPMOP(type, 0);
6028 if (PL_multi_open == '?')
6029 pm->op_pmflags |= PMf_ONCE;
6031 while (*s && strchr("iomsx", *s))
6032 pmflag(&pm->op_pmflags,*s++);
6035 while (*s && strchr("iogcmsx", *s))
6036 pmflag(&pm->op_pmflags,*s++);
6038 pm->op_pmpermflags = pm->op_pmflags;
6040 PL_lex_op = (OP*)pm;
6041 yylval.ival = OP_MATCH;
6046 S_scan_subst(pTHX_ char *start)
6053 yylval.ival = OP_NULL;
6055 s = scan_str(start,FALSE,FALSE);
6059 SvREFCNT_dec(PL_lex_stuff);
6060 PL_lex_stuff = Nullsv;
6061 Perl_croak(aTHX_ "Substitution pattern not terminated");
6064 if (s[-1] == PL_multi_open)
6067 first_start = PL_multi_start;
6068 s = scan_str(s,FALSE,FALSE);
6071 SvREFCNT_dec(PL_lex_stuff);
6072 PL_lex_stuff = Nullsv;
6074 SvREFCNT_dec(PL_lex_repl);
6075 PL_lex_repl = Nullsv;
6076 Perl_croak(aTHX_ "Substitution replacement not terminated");
6078 PL_multi_start = first_start; /* so whole substitution is taken together */
6080 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6086 else if (strchr("iogcmsx", *s))
6087 pmflag(&pm->op_pmflags,*s++);
6094 PL_sublex_info.super_bufptr = s;
6095 PL_sublex_info.super_bufend = PL_bufend;
6097 pm->op_pmflags |= PMf_EVAL;
6098 repl = newSVpvn("",0);
6100 sv_catpv(repl, es ? "eval " : "do ");
6101 sv_catpvn(repl, "{ ", 2);
6102 sv_catsv(repl, PL_lex_repl);
6103 sv_catpvn(repl, " };", 2);
6105 SvREFCNT_dec(PL_lex_repl);
6109 pm->op_pmpermflags = pm->op_pmflags;
6110 PL_lex_op = (OP*)pm;
6111 yylval.ival = OP_SUBST;
6116 S_scan_trans(pTHX_ char *start)
6127 yylval.ival = OP_NULL;
6129 s = scan_str(start,FALSE,FALSE);
6132 SvREFCNT_dec(PL_lex_stuff);
6133 PL_lex_stuff = Nullsv;
6134 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6136 if (s[-1] == PL_multi_open)
6139 s = scan_str(s,FALSE,FALSE);
6142 SvREFCNT_dec(PL_lex_stuff);
6143 PL_lex_stuff = Nullsv;
6145 SvREFCNT_dec(PL_lex_repl);
6146 PL_lex_repl = Nullsv;
6147 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6150 New(803,tbl,256,short);
6151 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6153 complement = del = squash = 0;
6154 while (strchr("cds", *s)) {
6156 complement = OPpTRANS_COMPLEMENT;
6158 del = OPpTRANS_DELETE;
6160 squash = OPpTRANS_SQUASH;
6163 o->op_private = del|squash|complement;
6166 yylval.ival = OP_TRANS;
6171 S_scan_heredoc(pTHX_ register char *s)
6175 I32 op_type = OP_SCALAR;
6182 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6186 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6189 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6190 if (*peek && strchr("`'\"",*peek)) {
6193 s = delimcpy(d, e, s, PL_bufend, term, &len);
6203 if (!isALNUM_lazy_if(s,UTF))
6204 deprecate("bare << to mean <<\"\"");
6205 for (; isALNUM_lazy_if(s,UTF); s++) {
6210 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6211 Perl_croak(aTHX_ "Delimiter for here document is too long");
6214 len = d - PL_tokenbuf;
6215 #ifndef PERL_STRICT_CR
6216 d = strchr(s, '\r');
6220 while (s < PL_bufend) {
6226 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6235 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6240 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6241 herewas = newSVpvn(s,PL_bufend-s);
6243 s--, herewas = newSVpvn(s,d-s);
6244 s += SvCUR(herewas);
6246 tmpstr = NEWSV(87,79);
6247 sv_upgrade(tmpstr, SVt_PVIV);
6252 else if (term == '`') {
6253 op_type = OP_BACKTICK;
6254 SvIVX(tmpstr) = '\\';
6258 PL_multi_start = CopLINE(PL_curcop);
6259 PL_multi_open = PL_multi_close = '<';
6260 term = *PL_tokenbuf;
6261 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6262 char *bufptr = PL_sublex_info.super_bufptr;
6263 char *bufend = PL_sublex_info.super_bufend;
6264 char *olds = s - SvCUR(herewas);
6265 s = strchr(bufptr, '\n');
6269 while (s < bufend &&
6270 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6272 CopLINE_inc(PL_curcop);
6275 CopLINE_set(PL_curcop, PL_multi_start);
6276 missingterm(PL_tokenbuf);
6278 sv_setpvn(herewas,bufptr,d-bufptr+1);
6279 sv_setpvn(tmpstr,d+1,s-d);
6281 sv_catpvn(herewas,s,bufend-s);
6282 (void)strcpy(bufptr,SvPVX(herewas));
6289 while (s < PL_bufend &&
6290 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6292 CopLINE_inc(PL_curcop);
6294 if (s >= PL_bufend) {
6295 CopLINE_set(PL_curcop, PL_multi_start);
6296 missingterm(PL_tokenbuf);
6298 sv_setpvn(tmpstr,d+1,s-d);
6300 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6302 sv_catpvn(herewas,s,PL_bufend-s);
6303 sv_setsv(PL_linestr,herewas);
6304 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6305 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6308 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6309 while (s >= PL_bufend) { /* multiple line string? */
6311 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6312 CopLINE_set(PL_curcop, PL_multi_start);
6313 missingterm(PL_tokenbuf);
6315 CopLINE_inc(PL_curcop);
6316 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6317 #ifndef PERL_STRICT_CR
6318 if (PL_bufend - PL_linestart >= 2) {
6319 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6320 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6322 PL_bufend[-2] = '\n';
6324 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6326 else if (PL_bufend[-1] == '\r')
6327 PL_bufend[-1] = '\n';
6329 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6330 PL_bufend[-1] = '\n';
6332 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6333 SV *sv = NEWSV(88,0);
6335 sv_upgrade(sv, SVt_PVMG);
6336 sv_setsv(sv,PL_linestr);
6337 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6339 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6342 sv_catsv(PL_linestr,herewas);
6343 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6347 sv_catsv(tmpstr,PL_linestr);
6352 PL_multi_end = CopLINE(PL_curcop);
6353 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6354 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6355 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6357 SvREFCNT_dec(herewas);
6358 PL_lex_stuff = tmpstr;
6359 yylval.ival = op_type;
6364 takes: current position in input buffer
6365 returns: new position in input buffer
6366 side-effects: yylval and lex_op are set.
6371 <FH> read from filehandle
6372 <pkg::FH> read from package qualified filehandle
6373 <pkg'FH> read from package qualified filehandle
6374 <$fh> read from filehandle in $fh
6380 S_scan_inputsymbol(pTHX_ char *start)
6382 register char *s = start; /* current position in buffer */
6388 d = PL_tokenbuf; /* start of temp holding space */
6389 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6390 end = strchr(s, '\n');
6393 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6395 /* die if we didn't have space for the contents of the <>,
6396 or if it didn't end, or if we see a newline
6399 if (len >= sizeof PL_tokenbuf)
6400 Perl_croak(aTHX_ "Excessively long <> operator");
6402 Perl_croak(aTHX_ "Unterminated <> operator");
6407 Remember, only scalar variables are interpreted as filehandles by
6408 this code. Anything more complex (e.g., <$fh{$num}>) will be
6409 treated as a glob() call.
6410 This code makes use of the fact that except for the $ at the front,
6411 a scalar variable and a filehandle look the same.
6413 if (*d == '$' && d[1]) d++;
6415 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6416 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6419 /* If we've tried to read what we allow filehandles to look like, and
6420 there's still text left, then it must be a glob() and not a getline.
6421 Use scan_str to pull out the stuff between the <> and treat it
6422 as nothing more than a string.
6425 if (d - PL_tokenbuf != len) {
6426 yylval.ival = OP_GLOB;
6428 s = scan_str(start,FALSE,FALSE);
6430 Perl_croak(aTHX_ "Glob not terminated");
6434 /* we're in a filehandle read situation */
6437 /* turn <> into <ARGV> */
6439 (void)strcpy(d,"ARGV");
6441 /* if <$fh>, create the ops to turn the variable into a
6447 /* try to find it in the pad for this block, otherwise find
6448 add symbol table ops
6450 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6451 OP *o = newOP(OP_PADSV, 0);
6453 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6456 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6457 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6458 newUNOP(OP_RV2SV, 0,
6459 newGVOP(OP_GV, 0, gv)));
6461 PL_lex_op->op_flags |= OPf_SPECIAL;
6462 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6463 yylval.ival = OP_NULL;
6466 /* If it's none of the above, it must be a literal filehandle
6467 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6469 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6470 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6471 yylval.ival = OP_NULL;
6480 takes: start position in buffer
6481 keep_quoted preserve \ on the embedded delimiter(s)
6482 keep_delims preserve the delimiters around the string
6483 returns: position to continue reading from buffer
6484 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6485 updates the read buffer.
6487 This subroutine pulls a string out of the input. It is called for:
6488 q single quotes q(literal text)
6489 ' single quotes 'literal text'
6490 qq double quotes qq(interpolate $here please)
6491 " double quotes "interpolate $here please"
6492 qx backticks qx(/bin/ls -l)
6493 ` backticks `/bin/ls -l`
6494 qw quote words @EXPORT_OK = qw( func() $spam )
6495 m// regexp match m/this/
6496 s/// regexp substitute s/this/that/
6497 tr/// string transliterate tr/this/that/
6498 y/// string transliterate y/this/that/
6499 ($*@) sub prototypes sub foo ($)
6500 (stuff) sub attr parameters sub foo : attr(stuff)
6501 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6503 In most of these cases (all but <>, patterns and transliterate)
6504 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6505 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6506 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6509 It skips whitespace before the string starts, and treats the first
6510 character as the delimiter. If the delimiter is one of ([{< then
6511 the corresponding "close" character )]}> is used as the closing
6512 delimiter. It allows quoting of delimiters, and if the string has
6513 balanced delimiters ([{<>}]) it allows nesting.
6515 The lexer always reads these strings into lex_stuff, except in the
6516 case of the operators which take *two* arguments (s/// and tr///)
6517 when it checks to see if lex_stuff is full (presumably with the 1st
6518 arg to s or tr) and if so puts the string into lex_repl.
6523 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6526 SV *sv; /* scalar value: string */
6527 char *tmps; /* temp string, used for delimiter matching */
6528 register char *s = start; /* current position in the buffer */
6529 register char term; /* terminating character */
6530 register char *to; /* current position in the sv's data */
6531 I32 brackets = 1; /* bracket nesting level */
6532 bool has_utf = FALSE; /* is there any utf8 content? */
6534 /* skip space before the delimiter */
6538 /* mark where we are, in case we need to report errors */
6541 /* after skipping whitespace, the next character is the terminator */
6543 if ((term & 0x80) && UTF)
6546 /* mark where we are */
6547 PL_multi_start = CopLINE(PL_curcop);
6548 PL_multi_open = term;
6550 /* find corresponding closing delimiter */
6551 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6553 PL_multi_close = term;
6555 /* create a new SV to hold the contents. 87 is leak category, I'm
6556 assuming. 79 is the SV's initial length. What a random number. */
6558 sv_upgrade(sv, SVt_PVIV);
6560 (void)SvPOK_only(sv); /* validate pointer */
6562 /* move past delimiter and try to read a complete string */
6564 sv_catpvn(sv, s, 1);
6567 /* extend sv if need be */
6568 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6569 /* set 'to' to the next character in the sv's string */
6570 to = SvPVX(sv)+SvCUR(sv);
6572 /* if open delimiter is the close delimiter read unbridle */
6573 if (PL_multi_open == PL_multi_close) {
6574 for (; s < PL_bufend; s++,to++) {
6575 /* embedded newlines increment the current line number */
6576 if (*s == '\n' && !PL_rsfp)
6577 CopLINE_inc(PL_curcop);
6578 /* handle quoted delimiters */
6579 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6580 if (!keep_quoted && s[1] == term)
6582 /* any other quotes are simply copied straight through */
6586 /* terminate when run out of buffer (the for() condition), or
6587 have found the terminator */
6588 else if (*s == term)
6590 else if (!has_utf && (*s & 0x80) && UTF)
6596 /* if the terminator isn't the same as the start character (e.g.,
6597 matched brackets), we have to allow more in the quoting, and
6598 be prepared for nested brackets.
6601 /* read until we run out of string, or we find the terminator */
6602 for (; s < PL_bufend; s++,to++) {
6603 /* embedded newlines increment the line count */
6604 if (*s == '\n' && !PL_rsfp)
6605 CopLINE_inc(PL_curcop);
6606 /* backslashes can escape the open or closing characters */
6607 if (*s == '\\' && s+1 < PL_bufend) {
6609 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6614 /* allow nested opens and closes */
6615 else if (*s == PL_multi_close && --brackets <= 0)
6617 else if (*s == PL_multi_open)
6619 else if (!has_utf && (*s & 0x80) && UTF)
6624 /* terminate the copied string and update the sv's end-of-string */
6626 SvCUR_set(sv, to - SvPVX(sv));
6629 * this next chunk reads more into the buffer if we're not done yet
6633 break; /* handle case where we are done yet :-) */
6635 #ifndef PERL_STRICT_CR
6636 if (to - SvPVX(sv) >= 2) {
6637 if ((to[-2] == '\r' && to[-1] == '\n') ||
6638 (to[-2] == '\n' && to[-1] == '\r'))
6642 SvCUR_set(sv, to - SvPVX(sv));
6644 else if (to[-1] == '\r')
6647 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6651 /* if we're out of file, or a read fails, bail and reset the current
6652 line marker so we can report where the unterminated string began
6655 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6657 CopLINE_set(PL_curcop, PL_multi_start);
6660 /* we read a line, so increment our line counter */
6661 CopLINE_inc(PL_curcop);
6663 /* update debugger info */
6664 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6665 SV *sv = NEWSV(88,0);
6667 sv_upgrade(sv, SVt_PVMG);
6668 sv_setsv(sv,PL_linestr);
6669 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6672 /* having changed the buffer, we must update PL_bufend */
6673 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6676 /* at this point, we have successfully read the delimited string */
6679 sv_catpvn(sv, s, 1);
6682 PL_multi_end = CopLINE(PL_curcop);
6685 /* if we allocated too much space, give some back */
6686 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6687 SvLEN_set(sv, SvCUR(sv) + 1);
6688 Renew(SvPVX(sv), SvLEN(sv), char);
6691 /* decide whether this is the first or second quoted string we've read
6704 takes: pointer to position in buffer
6705 returns: pointer to new position in buffer
6706 side-effects: builds ops for the constant in yylval.op
6708 Read a number in any of the formats that Perl accepts:
6710 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6711 [\d_]+(\.[\d_]*)?[Ee](\d+)
6713 Underbars (_) are allowed in decimal numbers. If -w is on,
6714 underbars before a decimal point must be at three digit intervals.
6716 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6719 If it reads a number without a decimal point or an exponent, it will
6720 try converting the number to an integer and see if it can do so
6721 without loss of precision.
6725 Perl_scan_num(pTHX_ char *start)
6727 register char *s = start; /* current position in buffer */
6728 register char *d; /* destination in temp buffer */
6729 register char *e; /* end of temp buffer */
6730 NV nv; /* number read, as a double */
6731 SV *sv = Nullsv; /* place to put the converted number */
6732 bool floatit; /* boolean: int or float? */
6733 char *lastub = 0; /* position of last underbar */
6734 static char number_too_long[] = "Number too long";
6736 /* We use the first character to decide what type of number this is */
6740 Perl_croak(aTHX_ "panic: scan_num");
6742 /* if it starts with a 0, it could be an octal number, a decimal in
6743 0.13 disguise, or a hexadecimal number, or a binary number. */
6747 u holds the "number so far"
6748 shift the power of 2 of the base
6749 (hex == 4, octal == 3, binary == 1)
6750 overflowed was the number more than we can hold?
6752 Shift is used when we add a digit. It also serves as an "are
6753 we in octal/hex/binary?" indicator to disallow hex characters
6760 bool overflowed = FALSE;
6761 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6762 static char* bases[5] = { "", "binary", "", "octal",
6764 static char* Bases[5] = { "", "Binary", "", "Octal",
6766 static char *maxima[5] = { "",
6767 "0b11111111111111111111111111111111",
6771 char *base, *Base, *max;
6777 } else if (s[1] == 'b') {
6781 /* check for a decimal in disguise */
6782 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6784 /* so it must be octal */
6788 base = bases[shift];
6789 Base = Bases[shift];
6790 max = maxima[shift];
6792 /* read the rest of the number */
6794 /* x is used in the overflow test,
6795 b is the digit we're adding on. */
6800 /* if we don't mention it, we're done */
6809 /* 8 and 9 are not octal */
6812 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6816 case '2': case '3': case '4':
6817 case '5': case '6': case '7':
6819 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6823 b = *s++ & 15; /* ASCII digit -> value of digit */
6827 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6828 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6829 /* make sure they said 0x */
6834 /* Prepare to put the digit we have onto the end
6835 of the number so far. We check for overflows.
6840 x = u << shift; /* make room for the digit */
6842 if ((x >> shift) != u
6843 && !(PL_hints & HINT_NEW_BINARY)) {
6847 if (ckWARN_d(WARN_OVERFLOW))
6848 Perl_warner(aTHX_ WARN_OVERFLOW,
6849 "Integer overflow in %s number",
6852 u = x | b; /* add the digit to the end */
6855 n *= nvshift[shift];
6856 /* If an NV has not enough bits in its
6857 * mantissa to represent an UV this summing of
6858 * small low-order numbers is a waste of time
6859 * (because the NV cannot preserve the
6860 * low-order bits anyway): we could just
6861 * remember when did we overflow and in the
6862 * end just multiply n by the right
6870 /* if we get here, we had success: make a scalar value from
6877 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6878 Perl_warner(aTHX_ WARN_PORTABLE,
6879 "%s number > %s non-portable",
6886 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6887 Perl_warner(aTHX_ WARN_PORTABLE,
6888 "%s number > %s non-portable",
6893 if (PL_hints & HINT_NEW_BINARY)
6894 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6899 handle decimal numbers.
6900 we're also sent here when we read a 0 as the first digit
6902 case '1': case '2': case '3': case '4': case '5':
6903 case '6': case '7': case '8': case '9': case '.':
6906 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
6909 /* read next group of digits and _ and copy into d */
6910 while (isDIGIT(*s) || *s == '_') {
6911 /* skip underscores, checking for misplaced ones
6915 dTHR; /* only for ckWARN */
6916 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
6917 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6921 /* check for end of fixed-length buffer */
6923 Perl_croak(aTHX_ number_too_long);
6924 /* if we're ok, copy the character */
6929 /* final misplaced underbar check */
6930 if (lastub && s - lastub != 3) {
6932 if (ckWARN(WARN_SYNTAX))
6933 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6936 /* read a decimal portion if there is one. avoid
6937 3..5 being interpreted as the number 3. followed
6940 if (*s == '.' && s[1] != '.') {
6944 /* copy, ignoring underbars, until we run out of
6945 digits. Note: no misplaced underbar checks!
6947 for (; isDIGIT(*s) || *s == '_'; s++) {
6948 /* fixed length buffer check */
6950 Perl_croak(aTHX_ number_too_long);
6954 if (*s == '.' && isDIGIT(s[1])) {
6955 /* oops, it's really a v-string, but without the "v" */
6961 /* read exponent part, if present */
6962 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
6966 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
6967 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
6969 /* allow positive or negative exponent */
6970 if (*s == '+' || *s == '-')
6973 /* read digits of exponent (no underbars :-) */
6974 while (isDIGIT(*s)) {
6976 Perl_croak(aTHX_ number_too_long);
6981 /* terminate the string */
6984 /* make an sv from the string */
6987 #if defined(Strtol) && defined(Strtoul)
6990 strtol/strtoll sets errno to ERANGE if the number is too big
6991 for an integer. We try to do an integer conversion first
6992 if no characters indicating "float" have been found.
6999 if (*PL_tokenbuf == '-')
7000 iv = Strtol(PL_tokenbuf, (char**)NULL, 10);
7002 uv = Strtoul(PL_tokenbuf, (char**)NULL, 10);
7004 floatit = TRUE; /* Probably just too large. */
7005 else if (*PL_tokenbuf == '-')
7007 else if (uv <= IV_MAX)
7008 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7013 nv = Atof(PL_tokenbuf);
7018 No working strtou?ll?.
7020 Unfortunately atol() doesn't do range checks (returning
7021 LONG_MIN/LONG_MAX, and setting errno to ERANGE on overflows)
7022 everywhere [1], so we cannot use use atol() (or atoll()).
7023 If we could, they would be used, as Atol(), very much like
7024 Strtol() and Strtoul() are used above.
7026 [1] XXX Configure test needed to check for atol()
7027 (and atoll()) overflow behaviour XXX
7031 We need to do this the hard way. */
7033 nv = Atof(PL_tokenbuf);
7035 /* See if we can make do with an integer value without loss of
7036 precision. We use U_V to cast to a UV, because some
7037 compilers have issues. Then we try casting it back and see
7038 if it was the same [1]. We only do this if we know we
7039 specifically read an integer. If floatit is true, then we
7040 don't need to do the conversion at all.
7042 [1] Note that this is lossy if our NVs cannot preserve our
7043 UVs. There are metaconfig defines NV_PRESERVES_UV (a boolean)
7044 and NV_PRESERVES_UV_BITS (a number), but in general we really
7045 do hope all such potentially lossy platforms have strtou?ll?
7046 to do a lossless IV/UV conversion.
7048 Maybe could do some tricks with DBL_DIG, LDBL_DIG and
7049 DBL_MANT_DIG and LDBL_MANT_DIG (these are already available
7050 as NV_DIG and NV_MANT_DIG)?
7056 if (!floatit && (NV)uv == nv) {
7058 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7066 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7067 (PL_hints & HINT_NEW_INTEGER) )
7068 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7069 (floatit ? "float" : "integer"),
7073 /* if it starts with a v, it could be a v-string */
7079 while (isDIGIT(*pos) || *pos == '_')
7081 if (!isALPHA(*pos)) {
7083 U8 tmpbuf[UTF8_MAXLEN];
7086 s++; /* get past 'v' */
7089 sv_setpvn(sv, "", 0);
7092 if (*s == '0' && isDIGIT(s[1]))
7093 yyerror("Octal number in vector unsupported");
7096 /* this is atoi() that tolerates underscores */
7099 while (--end >= s) {
7104 rev += (*end - '0') * mult;
7106 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
7107 Perl_warner(aTHX_ WARN_OVERFLOW,
7108 "Integer overflow in decimal number");
7111 tmpend = uv_to_utf8(tmpbuf, rev);
7112 utf8 = utf8 || rev > 127;
7113 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
7114 if (*pos == '.' && isDIGIT(pos[1]))
7120 while (isDIGIT(*pos) || *pos == '_')
7128 sv_utf8_downgrade(sv, TRUE);
7135 /* make the op for the constant and return */
7138 yylval.opval = newSVOP(OP_CONST, 0, sv);
7140 yylval.opval = Nullop;
7146 S_scan_formline(pTHX_ register char *s)
7151 SV *stuff = newSVpvn("",0);
7152 bool needargs = FALSE;
7155 if (*s == '.' || *s == /*{*/'}') {
7157 #ifdef PERL_STRICT_CR
7158 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7160 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7162 if (*t == '\n' || t == PL_bufend)
7165 if (PL_in_eval && !PL_rsfp) {
7166 eol = strchr(s,'\n');
7171 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7173 for (t = s; t < eol; t++) {
7174 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7176 goto enough; /* ~~ must be first line in formline */
7178 if (*t == '@' || *t == '^')
7181 sv_catpvn(stuff, s, eol-s);
7182 #ifndef PERL_STRICT_CR
7183 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7184 char *end = SvPVX(stuff) + SvCUR(stuff);
7193 s = filter_gets(PL_linestr, PL_rsfp, 0);
7194 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7195 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7198 yyerror("Format not terminated");
7208 PL_lex_state = LEX_NORMAL;
7209 PL_nextval[PL_nexttoke].ival = 0;
7213 PL_lex_state = LEX_FORMLINE;
7214 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7216 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7220 SvREFCNT_dec(stuff);
7221 PL_lex_formbrack = 0;
7232 PL_cshlen = strlen(PL_cshname);
7237 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7240 I32 oldsavestack_ix = PL_savestack_ix;
7241 CV* outsidecv = PL_compcv;
7245 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7247 SAVEI32(PL_subline);
7248 save_item(PL_subname);
7251 SAVESPTR(PL_comppad_name);
7252 SAVESPTR(PL_compcv);
7253 SAVEI32(PL_comppad_name_fill);
7254 SAVEI32(PL_min_intro_pending);
7255 SAVEI32(PL_max_intro_pending);
7256 SAVEI32(PL_pad_reset_pending);
7258 PL_compcv = (CV*)NEWSV(1104,0);
7259 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7260 CvFLAGS(PL_compcv) |= flags;
7262 PL_comppad = newAV();
7263 av_push(PL_comppad, Nullsv);
7264 PL_curpad = AvARRAY(PL_comppad);
7265 PL_comppad_name = newAV();
7266 PL_comppad_name_fill = 0;
7267 PL_min_intro_pending = 0;
7269 PL_subline = CopLINE(PL_curcop);
7271 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7272 PL_curpad[0] = (SV*)newAV();
7273 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7274 #endif /* USE_THREADS */
7276 comppadlist = newAV();
7277 AvREAL_off(comppadlist);
7278 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7279 av_store(comppadlist, 1, (SV*)PL_comppad);
7281 CvPADLIST(PL_compcv) = comppadlist;
7282 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7284 CvOWNER(PL_compcv) = 0;
7285 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7286 MUTEX_INIT(CvMUTEXP(PL_compcv));
7287 #endif /* USE_THREADS */
7289 return oldsavestack_ix;
7293 Perl_yywarn(pTHX_ char *s)
7296 PL_in_eval |= EVAL_WARNONLY;
7298 PL_in_eval &= ~EVAL_WARNONLY;
7303 Perl_yyerror(pTHX_ char *s)
7307 char *context = NULL;
7311 if (!yychar || (yychar == ';' && !PL_rsfp))
7313 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7314 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7315 while (isSPACE(*PL_oldoldbufptr))
7317 context = PL_oldoldbufptr;
7318 contlen = PL_bufptr - PL_oldoldbufptr;
7320 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7321 PL_oldbufptr != PL_bufptr) {
7322 while (isSPACE(*PL_oldbufptr))
7324 context = PL_oldbufptr;
7325 contlen = PL_bufptr - PL_oldbufptr;
7327 else if (yychar > 255)
7328 where = "next token ???";
7329 #ifdef USE_PURE_BISON
7330 /* GNU Bison sets the value -2 */
7331 else if (yychar == -2) {
7333 else if ((yychar & 127) == 127) {
7335 if (PL_lex_state == LEX_NORMAL ||
7336 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7337 where = "at end of line";
7338 else if (PL_lex_inpat)
7339 where = "within pattern";
7341 where = "within string";
7344 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7346 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7347 else if (isPRINT_LC(yychar))
7348 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7350 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7351 where = SvPVX(where_sv);
7353 msg = sv_2mortal(newSVpv(s, 0));
7354 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7355 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7357 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7359 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7360 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7361 Perl_sv_catpvf(aTHX_ msg,
7362 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7363 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7366 if (PL_in_eval & EVAL_WARNONLY)
7367 Perl_warn(aTHX_ "%"SVf, msg);
7370 if (PL_error_count >= 10) {
7371 if (PL_in_eval && SvCUR(ERRSV))
7372 Perl_croak(aTHX_ "%_%s has too many errors.\n",
7373 ERRSV, CopFILE(PL_curcop));
7375 Perl_croak(aTHX_ "%s has too many errors.\n",
7376 CopFILE(PL_curcop));
7379 PL_in_my_stash = Nullhv;
7384 S_swallow_bom(pTHX_ U8 *s)
7387 slen = SvCUR(PL_linestr);
7391 /* UTF-16 little-endian */
7392 #ifndef PERL_NO_UTF16_FILTER
7395 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
7396 Perl_croak(aTHX_ "Unsupported script encoding");
7397 #ifndef PERL_NO_UTF16_FILTER
7399 filter_add(utf16rev_textfilter, NULL);
7400 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7401 PL_bufend = (char*)utf16_to_utf8((U16*)s, news,
7402 PL_bufend - (char*)s);
7405 Perl_croak(aTHX_ "Unsupported script encoding");
7410 if (s[1] == 0xFF) { /* UTF-16 big-endian */
7411 #ifndef PERL_NO_UTF16_FILTER
7413 filter_add(utf16_textfilter, NULL);
7414 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7415 PL_bufend = (char*)utf16_to_utf8((U16*)s, news,
7416 PL_bufend - (char*)s);
7419 Perl_croak(aTHX_ "Unsupported script encoding");
7424 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
7429 if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */
7430 s[2] == 0xFE && s[3] == 0xFF)
7432 Perl_croak(aTHX_ "Unsupported script encoding");
7444 * Restore a source filter.
7448 restore_rsfp(pTHXo_ void *f)
7450 PerlIO *fp = (PerlIO*)f;
7452 if (PL_rsfp == PerlIO_stdin())
7453 PerlIO_clearerr(PL_rsfp);
7454 else if (PL_rsfp && (PL_rsfp != fp))
7455 PerlIO_close(PL_rsfp);
7459 #ifndef PERL_NO_UTF16_FILTER
7461 utf16_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7463 I32 count = FILTER_READ(idx+1, sv, maxlen);
7467 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7468 tend = utf16_to_utf8((U16*)SvPVX(sv), tmps, SvCUR(sv));
7469 sv_usepvn(sv, (char*)tmps, tend - tmps);
7475 utf16rev_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7477 I32 count = FILTER_READ(idx+1, sv, maxlen);
7481 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7482 tend = utf16_to_utf8_reversed((U16*)SvPVX(sv), tmps, SvCUR(sv));
7483 sv_usepvn(sv, (char*)tmps, tend - tmps);