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);
214 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
216 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
217 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
219 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
220 if (t < PL_bufptr && isSPACE(*t))
221 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
222 t - PL_oldoldbufptr, PL_oldoldbufptr);
226 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, 0);
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",
1222 (char)min, (char)max);
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, 0);
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 /* this just broke our allocation above... */
1494 SvGROW(sv, send - start);
1495 d = SvPVX(sv) + SvCUR(sv);
1498 if (len > e - s + 4) {
1499 char *odest = SvPVX(sv);
1501 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1502 d = SvPVX(sv) + (d - odest);
1504 Copy(str, d, len, char);
1511 yyerror("Missing braces on \\N{}");
1514 /* \c is a control character */
1529 /* printf-style backslashes, formfeeds, newlines, etc */
1547 *d++ = '\047'; /* CP 1047 */
1550 *d++ = '\057'; /* CP 1047 */
1564 } /* end if (backslash) */
1567 } /* while loop to process each character */
1569 /* terminate the string and set up the sv */
1571 SvCUR_set(sv, d - SvPVX(sv));
1576 /* shrink the sv if we allocated more than we used */
1577 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1578 SvLEN_set(sv, SvCUR(sv) + 1);
1579 Renew(SvPVX(sv), SvLEN(sv), char);
1582 /* return the substring (via yylval) only if we parsed anything */
1583 if (s > PL_bufptr) {
1584 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1585 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1587 ( PL_lex_inwhat == OP_TRANS
1589 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1592 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1599 * Returns TRUE if there's more to the expression (e.g., a subscript),
1602 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1604 * ->[ and ->{ return TRUE
1605 * { and [ outside a pattern are always subscripts, so return TRUE
1606 * if we're outside a pattern and it's not { or [, then return FALSE
1607 * if we're in a pattern and the first char is a {
1608 * {4,5} (any digits around the comma) returns FALSE
1609 * if we're in a pattern and the first char is a [
1611 * [SOMETHING] has a funky algorithm to decide whether it's a
1612 * character class or not. It has to deal with things like
1613 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1614 * anything else returns TRUE
1617 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1620 S_intuit_more(pTHX_ register char *s)
1622 if (PL_lex_brackets)
1624 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1626 if (*s != '{' && *s != '[')
1631 /* In a pattern, so maybe we have {n,m}. */
1648 /* On the other hand, maybe we have a character class */
1651 if (*s == ']' || *s == '^')
1654 /* this is terrifying, and it works */
1655 int weight = 2; /* let's weigh the evidence */
1657 unsigned char un_char = 255, last_un_char;
1658 char *send = strchr(s,']');
1659 char tmpbuf[sizeof PL_tokenbuf * 4];
1661 if (!send) /* has to be an expression */
1664 Zero(seen,256,char);
1667 else if (isDIGIT(*s)) {
1669 if (isDIGIT(s[1]) && s[2] == ']')
1675 for (; s < send; s++) {
1676 last_un_char = un_char;
1677 un_char = (unsigned char)*s;
1682 weight -= seen[un_char] * 10;
1683 if (isALNUM_lazy_if(s+1,UTF)) {
1684 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1685 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1690 else if (*s == '$' && s[1] &&
1691 strchr("[#!%*<>()-=",s[1])) {
1692 if (/*{*/ strchr("])} =",s[2]))
1701 if (strchr("wds]",s[1]))
1703 else if (seen['\''] || seen['"'])
1705 else if (strchr("rnftbxcav",s[1]))
1707 else if (isDIGIT(s[1])) {
1709 while (s[1] && isDIGIT(s[1]))
1719 if (strchr("aA01! ",last_un_char))
1721 if (strchr("zZ79~",s[1]))
1723 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1724 weight -= 5; /* cope with negative subscript */
1727 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1728 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1733 if (keyword(tmpbuf, d - tmpbuf))
1736 if (un_char == last_un_char + 1)
1738 weight -= seen[un_char];
1743 if (weight >= 0) /* probably a character class */
1753 * Does all the checking to disambiguate
1755 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1756 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1758 * First argument is the stuff after the first token, e.g. "bar".
1760 * Not a method if bar is a filehandle.
1761 * Not a method if foo is a subroutine prototyped to take a filehandle.
1762 * Not a method if it's really "Foo $bar"
1763 * Method if it's "foo $bar"
1764 * Not a method if it's really "print foo $bar"
1765 * Method if it's really "foo package::" (interpreted as package->foo)
1766 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1767 * Not a method if bar is a filehandle or package, but is quoted with
1772 S_intuit_method(pTHX_ char *start, GV *gv)
1774 char *s = start + (*start == '$');
1775 char tmpbuf[sizeof PL_tokenbuf];
1783 if ((cv = GvCVu(gv))) {
1784 char *proto = SvPVX(cv);
1794 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1795 /* start is the beginning of the possible filehandle/object,
1796 * and s is the end of it
1797 * tmpbuf is a copy of it
1800 if (*start == '$') {
1801 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1806 return *s == '(' ? FUNCMETH : METHOD;
1808 if (!keyword(tmpbuf, len)) {
1809 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1814 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1815 if (indirgv && GvCVu(indirgv))
1817 /* filehandle or package name makes it a method */
1818 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1820 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1821 return 0; /* no assumptions -- "=>" quotes bearword */
1823 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1824 newSVpvn(tmpbuf,len));
1825 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1829 return *s == '(' ? FUNCMETH : METHOD;
1837 * Return a string of Perl code to load the debugger. If PERL5DB
1838 * is set, it will return the contents of that, otherwise a
1839 * compile-time require of perl5db.pl.
1846 char *pdb = PerlEnv_getenv("PERL5DB");
1850 SETERRNO(0,SS$_NORMAL);
1851 return "BEGIN { require 'perl5db.pl' }";
1857 /* Encoded script support. filter_add() effectively inserts a
1858 * 'pre-processing' function into the current source input stream.
1859 * Note that the filter function only applies to the current source file
1860 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1862 * The datasv parameter (which may be NULL) can be used to pass
1863 * private data to this instance of the filter. The filter function
1864 * can recover the SV using the FILTER_DATA macro and use it to
1865 * store private buffers and state information.
1867 * The supplied datasv parameter is upgraded to a PVIO type
1868 * and the IoDIRP/IoANY field is used to store the function pointer,
1869 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1870 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1871 * private use must be set using malloc'd pointers.
1875 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1880 if (!PL_rsfp_filters)
1881 PL_rsfp_filters = newAV();
1883 datasv = NEWSV(255,0);
1884 if (!SvUPGRADE(datasv, SVt_PVIO))
1885 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1886 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
1887 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1888 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1889 funcp, SvPV_nolen(datasv)));
1890 av_unshift(PL_rsfp_filters, 1);
1891 av_store(PL_rsfp_filters, 0, datasv) ;
1896 /* Delete most recently added instance of this filter function. */
1898 Perl_filter_del(pTHX_ filter_t funcp)
1901 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1902 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1904 /* if filter is on top of stack (usual case) just pop it off */
1905 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1906 if (IoANY(datasv) == (void *)funcp) {
1907 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1908 IoANY(datasv) = (void *)NULL;
1909 sv_free(av_pop(PL_rsfp_filters));
1913 /* we need to search for the correct entry and clear it */
1914 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1918 /* Invoke the n'th filter function for the current rsfp. */
1920 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1923 /* 0 = read one text line */
1928 if (!PL_rsfp_filters)
1930 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1931 /* Provide a default input filter to make life easy. */
1932 /* Note that we append to the line. This is handy. */
1933 DEBUG_P(PerlIO_printf(Perl_debug_log,
1934 "filter_read %d: from rsfp\n", idx));
1938 int old_len = SvCUR(buf_sv) ;
1940 /* ensure buf_sv is large enough */
1941 SvGROW(buf_sv, old_len + maxlen) ;
1942 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1943 if (PerlIO_error(PL_rsfp))
1944 return -1; /* error */
1946 return 0 ; /* end of file */
1948 SvCUR_set(buf_sv, old_len + len) ;
1951 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1952 if (PerlIO_error(PL_rsfp))
1953 return -1; /* error */
1955 return 0 ; /* end of file */
1958 return SvCUR(buf_sv);
1960 /* Skip this filter slot if filter has been deleted */
1961 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1962 DEBUG_P(PerlIO_printf(Perl_debug_log,
1963 "filter_read %d: skipped (filter deleted)\n",
1965 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1967 /* Get function pointer hidden within datasv */
1968 funcp = (filter_t)IoANY(datasv);
1969 DEBUG_P(PerlIO_printf(Perl_debug_log,
1970 "filter_read %d: via function %p (%s)\n",
1971 idx, funcp, SvPV_nolen(datasv)));
1972 /* Call function. The function is expected to */
1973 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1974 /* Return: <0:error, =0:eof, >0:not eof */
1975 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1979 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1981 #ifdef PERL_CR_FILTER
1982 if (!PL_rsfp_filters) {
1983 filter_add(S_cr_textfilter,NULL);
1986 if (PL_rsfp_filters) {
1989 SvCUR_set(sv, 0); /* start with empty line */
1990 if (FILTER_READ(0, sv, 0) > 0)
1991 return ( SvPVX(sv) ) ;
1996 return (sv_gets(sv, fp, append));
2000 S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2004 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2008 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2009 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2011 return GvHV(gv); /* Foo:: */
2014 /* use constant CLASS => 'MyClass' */
2015 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2017 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2018 pkgname = SvPV_nolen(sv);
2022 return gv_stashpv(pkgname, FALSE);
2026 static char* exp_name[] =
2027 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2028 "ATTRTERM", "TERMBLOCK"
2035 Works out what to call the token just pulled out of the input
2036 stream. The yacc parser takes care of taking the ops we return and
2037 stitching them into a tree.
2043 if read an identifier
2044 if we're in a my declaration
2045 croak if they tried to say my($foo::bar)
2046 build the ops for a my() declaration
2047 if it's an access to a my() variable
2048 are we in a sort block?
2049 croak if my($a); $a <=> $b
2050 build ops for access to a my() variable
2051 if in a dq string, and they've said @foo and we can't find @foo
2053 build ops for a bareword
2054 if we already built the token before, use it.
2058 #pragma segment Perl_yylex
2061 #ifdef USE_PURE_BISON
2062 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
2075 #ifdef USE_PURE_BISON
2076 yylval_pointer = lvalp;
2077 yychar_pointer = lcharp;
2080 /* check if there's an identifier for us to look at */
2081 if (PL_pending_ident) {
2082 /* pit holds the identifier we read and pending_ident is reset */
2083 char pit = PL_pending_ident;
2084 PL_pending_ident = 0;
2086 /* if we're in a my(), we can't allow dynamics here.
2087 $foo'bar has already been turned into $foo::bar, so
2088 just check for colons.
2090 if it's a legal name, the OP is a PADANY.
2093 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
2094 if (strchr(PL_tokenbuf,':'))
2095 yyerror(Perl_form(aTHX_ "No package name allowed for "
2096 "variable %s in \"our\"",
2098 tmp = pad_allocmy(PL_tokenbuf);
2101 if (strchr(PL_tokenbuf,':'))
2102 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
2104 yylval.opval = newOP(OP_PADANY, 0);
2105 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
2111 build the ops for accesses to a my() variable.
2113 Deny my($a) or my($b) in a sort block, *if* $a or $b is
2114 then used in a comparison. This catches most, but not
2115 all cases. For instance, it catches
2116 sort { my($a); $a <=> $b }
2118 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2119 (although why you'd do that is anyone's guess).
2122 if (!strchr(PL_tokenbuf,':')) {
2124 /* Check for single character per-thread SVs */
2125 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2126 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2127 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2129 yylval.opval = newOP(OP_THREADSV, 0);
2130 yylval.opval->op_targ = tmp;
2133 #endif /* USE_THREADS */
2134 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2135 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2136 /* might be an "our" variable" */
2137 if (SvFLAGS(namesv) & SVpad_OUR) {
2138 /* build ops for a bareword */
2139 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2140 sv_catpvn(sym, "::", 2);
2141 sv_catpv(sym, PL_tokenbuf+1);
2142 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2143 yylval.opval->op_private = OPpCONST_ENTERED;
2144 gv_fetchpv(SvPVX(sym),
2146 ? (GV_ADDMULTI | GV_ADDINEVAL)
2149 ((PL_tokenbuf[0] == '$') ? SVt_PV
2150 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2155 /* if it's a sort block and they're naming $a or $b */
2156 if (PL_last_lop_op == OP_SORT &&
2157 PL_tokenbuf[0] == '$' &&
2158 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2161 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2162 d < PL_bufend && *d != '\n';
2165 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2166 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2172 yylval.opval = newOP(OP_PADANY, 0);
2173 yylval.opval->op_targ = tmp;
2179 Whine if they've said @foo in a doublequoted string,
2180 and @foo isn't a variable we can find in the symbol
2183 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2184 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2185 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2186 && ckWARN(WARN_AMBIGUOUS))
2188 /* Downgraded from fatal to warning 20000522 mjd */
2189 Perl_warner(aTHX_ WARN_AMBIGUOUS,
2190 "Possible unintended interpolation of %s in string",
2195 /* build ops for a bareword */
2196 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2197 yylval.opval->op_private = OPpCONST_ENTERED;
2198 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2199 ((PL_tokenbuf[0] == '$') ? SVt_PV
2200 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2205 /* no identifier pending identification */
2207 switch (PL_lex_state) {
2209 case LEX_NORMAL: /* Some compilers will produce faster */
2210 case LEX_INTERPNORMAL: /* code if we comment these out. */
2214 /* when we've already built the next token, just pull it out of the queue */
2217 yylval = PL_nextval[PL_nexttoke];
2219 PL_lex_state = PL_lex_defer;
2220 PL_expect = PL_lex_expect;
2221 PL_lex_defer = LEX_NORMAL;
2223 return(PL_nexttype[PL_nexttoke]);
2225 /* interpolated case modifiers like \L \U, including \Q and \E.
2226 when we get here, PL_bufptr is at the \
2228 case LEX_INTERPCASEMOD:
2230 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2231 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2233 /* handle \E or end of string */
2234 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2238 if (PL_lex_casemods) {
2239 oldmod = PL_lex_casestack[--PL_lex_casemods];
2240 PL_lex_casestack[PL_lex_casemods] = '\0';
2242 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2244 PL_lex_state = LEX_INTERPCONCAT;
2248 if (PL_bufptr != PL_bufend)
2250 PL_lex_state = LEX_INTERPCONCAT;
2255 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2256 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2257 if (strchr("LU", *s) &&
2258 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2260 PL_lex_casestack[--PL_lex_casemods] = '\0';
2263 if (PL_lex_casemods > 10) {
2264 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2265 if (newlb != PL_lex_casestack) {
2267 PL_lex_casestack = newlb;
2270 PL_lex_casestack[PL_lex_casemods++] = *s;
2271 PL_lex_casestack[PL_lex_casemods] = '\0';
2272 PL_lex_state = LEX_INTERPCONCAT;
2273 PL_nextval[PL_nexttoke].ival = 0;
2276 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2278 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2280 PL_nextval[PL_nexttoke].ival = OP_LC;
2282 PL_nextval[PL_nexttoke].ival = OP_UC;
2284 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2286 Perl_croak(aTHX_ "panic: yylex");
2289 if (PL_lex_starts) {
2298 case LEX_INTERPPUSH:
2299 return sublex_push();
2301 case LEX_INTERPSTART:
2302 if (PL_bufptr == PL_bufend)
2303 return sublex_done();
2305 PL_lex_dojoin = (*PL_bufptr == '@');
2306 PL_lex_state = LEX_INTERPNORMAL;
2307 if (PL_lex_dojoin) {
2308 PL_nextval[PL_nexttoke].ival = 0;
2311 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2312 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2313 force_next(PRIVATEREF);
2315 force_ident("\"", '$');
2316 #endif /* USE_THREADS */
2317 PL_nextval[PL_nexttoke].ival = 0;
2319 PL_nextval[PL_nexttoke].ival = 0;
2321 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2324 if (PL_lex_starts++) {
2330 case LEX_INTERPENDMAYBE:
2331 if (intuit_more(PL_bufptr)) {
2332 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2338 if (PL_lex_dojoin) {
2339 PL_lex_dojoin = FALSE;
2340 PL_lex_state = LEX_INTERPCONCAT;
2343 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2344 && SvEVALED(PL_lex_repl))
2346 if (PL_bufptr != PL_bufend)
2347 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2348 PL_lex_repl = Nullsv;
2351 case LEX_INTERPCONCAT:
2353 if (PL_lex_brackets)
2354 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2356 if (PL_bufptr == PL_bufend)
2357 return sublex_done();
2359 if (SvIVX(PL_linestr) == '\'') {
2360 SV *sv = newSVsv(PL_linestr);
2363 else if ( PL_hints & HINT_NEW_RE )
2364 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2365 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2369 s = scan_const(PL_bufptr);
2371 PL_lex_state = LEX_INTERPCASEMOD;
2373 PL_lex_state = LEX_INTERPSTART;
2376 if (s != PL_bufptr) {
2377 PL_nextval[PL_nexttoke] = yylval;
2380 if (PL_lex_starts++)
2390 PL_lex_state = LEX_NORMAL;
2391 s = scan_formline(PL_bufptr);
2392 if (!PL_lex_formbrack)
2398 PL_oldoldbufptr = PL_oldbufptr;
2401 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2402 exp_name[PL_expect], s);
2408 if (isIDFIRST_lazy_if(s,UTF))
2410 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2413 goto fake_eof; /* emulate EOF on ^D or ^Z */
2418 if (PL_lex_brackets)
2419 yyerror("Missing right curly or square bracket");
2422 if (s++ < PL_bufend)
2423 goto retry; /* ignore stray nulls */
2426 if (!PL_in_eval && !PL_preambled) {
2427 PL_preambled = TRUE;
2428 sv_setpv(PL_linestr,incl_perldb());
2429 if (SvCUR(PL_linestr))
2430 sv_catpv(PL_linestr,";");
2432 while(AvFILLp(PL_preambleav) >= 0) {
2433 SV *tmpsv = av_shift(PL_preambleav);
2434 sv_catsv(PL_linestr, tmpsv);
2435 sv_catpv(PL_linestr, ";");
2438 sv_free((SV*)PL_preambleav);
2439 PL_preambleav = NULL;
2441 if (PL_minus_n || PL_minus_p) {
2442 sv_catpv(PL_linestr, "LINE: while (<>) {");
2444 sv_catpv(PL_linestr,"chomp;");
2446 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2448 GvIMPORTED_AV_on(gv);
2450 if (strchr("/'\"", *PL_splitstr)
2451 && strchr(PL_splitstr + 1, *PL_splitstr))
2452 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2455 s = "'~#\200\1'"; /* surely one char is unused...*/
2456 while (s[1] && strchr(PL_splitstr, *s)) s++;
2458 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2459 "q" + (delim == '\''), delim);
2460 for (s = PL_splitstr; *s; s++) {
2462 sv_catpvn(PL_linestr, "\\", 1);
2463 sv_catpvn(PL_linestr, s, 1);
2465 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2469 sv_catpv(PL_linestr,"@F=split(' ');");
2472 sv_catpv(PL_linestr, "\n");
2473 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2474 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2475 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2476 SV *sv = NEWSV(85,0);
2478 sv_upgrade(sv, SVt_PVMG);
2479 sv_setsv(sv,PL_linestr);
2480 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2486 bof = PL_rsfp && (PerlIO_tell(PL_rsfp) == 0); /* *Before* read! */
2487 s = filter_gets(PL_linestr, PL_rsfp, 0);
2491 if (PL_preprocess && !PL_in_eval)
2492 (void)PerlProc_pclose(PL_rsfp);
2493 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2494 PerlIO_clearerr(PL_rsfp);
2496 (void)PerlIO_close(PL_rsfp);
2498 PL_doextract = FALSE;
2500 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2501 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2502 sv_catpv(PL_linestr,";}");
2503 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2504 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2505 PL_minus_n = PL_minus_p = 0;
2508 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2509 sv_setpv(PL_linestr,"");
2510 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2512 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2513 s = swallow_bom((U8*)s);
2516 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2517 PL_doextract = FALSE;
2519 /* Incest with pod. */
2520 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2521 sv_setpv(PL_linestr, "");
2522 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2523 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2524 PL_doextract = FALSE;
2528 } while (PL_doextract);
2529 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2530 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2531 SV *sv = NEWSV(85,0);
2533 sv_upgrade(sv, SVt_PVMG);
2534 sv_setsv(sv,PL_linestr);
2535 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2537 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2538 if (CopLINE(PL_curcop) == 1) {
2539 while (s < PL_bufend && isSPACE(*s))
2541 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2545 if (*s == '#' && *(s+1) == '!')
2547 #ifdef ALTERNATE_SHEBANG
2549 static char as[] = ALTERNATE_SHEBANG;
2550 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2551 d = s + (sizeof(as) - 1);
2553 #endif /* ALTERNATE_SHEBANG */
2562 while (*d && !isSPACE(*d))
2566 #ifdef ARG_ZERO_IS_SCRIPT
2567 if (ipathend > ipath) {
2569 * HP-UX (at least) sets argv[0] to the script name,
2570 * which makes $^X incorrect. And Digital UNIX and Linux,
2571 * at least, set argv[0] to the basename of the Perl
2572 * interpreter. So, having found "#!", we'll set it right.
2574 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2575 assert(SvPOK(x) || SvGMAGICAL(x));
2576 if (sv_eq(x, CopFILESV(PL_curcop))) {
2577 sv_setpvn(x, ipath, ipathend - ipath);
2580 TAINT_NOT; /* $^X is always tainted, but that's OK */
2582 #endif /* ARG_ZERO_IS_SCRIPT */
2587 d = instr(s,"perl -");
2589 d = instr(s,"perl");
2591 /* avoid getting into infinite loops when shebang
2592 * line contains "Perl" rather than "perl" */
2594 for (d = ipathend-4; d >= ipath; --d) {
2595 if ((*d == 'p' || *d == 'P')
2596 && !ibcmp(d, "perl", 4))
2606 #ifdef ALTERNATE_SHEBANG
2608 * If the ALTERNATE_SHEBANG on this system starts with a
2609 * character that can be part of a Perl expression, then if
2610 * we see it but not "perl", we're probably looking at the
2611 * start of Perl code, not a request to hand off to some
2612 * other interpreter. Similarly, if "perl" is there, but
2613 * not in the first 'word' of the line, we assume the line
2614 * contains the start of the Perl program.
2616 if (d && *s != '#') {
2618 while (*c && !strchr("; \t\r\n\f\v#", *c))
2621 d = Nullch; /* "perl" not in first word; ignore */
2623 *s = '#'; /* Don't try to parse shebang line */
2625 #endif /* ALTERNATE_SHEBANG */
2626 #ifndef MACOS_TRADITIONAL
2631 !instr(s,"indir") &&
2632 instr(PL_origargv[0],"perl"))
2638 while (s < PL_bufend && isSPACE(*s))
2640 if (s < PL_bufend) {
2641 Newz(899,newargv,PL_origargc+3,char*);
2643 while (s < PL_bufend && !isSPACE(*s))
2646 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2649 newargv = PL_origargv;
2651 PerlProc_execv(ipath, newargv);
2652 Perl_croak(aTHX_ "Can't exec %s", ipath);
2656 U32 oldpdb = PL_perldb;
2657 bool oldn = PL_minus_n;
2658 bool oldp = PL_minus_p;
2660 while (*d && !isSPACE(*d)) d++;
2661 while (SPACE_OR_TAB(*d)) d++;
2665 if (*d == 'M' || *d == 'm') {
2667 while (*d && !isSPACE(*d)) d++;
2668 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2671 d = moreswitches(d);
2673 if ((PERLDB_LINE && !oldpdb) ||
2674 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2675 /* if we have already added "LINE: while (<>) {",
2676 we must not do it again */
2678 sv_setpv(PL_linestr, "");
2679 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2680 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2681 PL_preambled = FALSE;
2683 (void)gv_fetchfile(PL_origfilename);
2690 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2692 PL_lex_state = LEX_FORMLINE;
2697 #ifdef PERL_STRICT_CR
2698 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2700 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2702 case ' ': case '\t': case '\f': case 013:
2703 #ifdef MACOS_TRADITIONAL
2710 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2711 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2712 /* handle eval qq[#line 1 "foo"\n ...] */
2713 CopLINE_dec(PL_curcop);
2717 while (s < d && *s != '\n')
2722 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2724 PL_lex_state = LEX_FORMLINE;
2734 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2739 while (s < PL_bufend && SPACE_OR_TAB(*s))
2742 if (strnEQ(s,"=>",2)) {
2743 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2744 OPERATOR('-'); /* unary minus */
2746 PL_last_uni = PL_oldbufptr;
2747 PL_last_lop_op = OP_FTEREAD; /* good enough */
2749 case 'r': FTST(OP_FTEREAD);
2750 case 'w': FTST(OP_FTEWRITE);
2751 case 'x': FTST(OP_FTEEXEC);
2752 case 'o': FTST(OP_FTEOWNED);
2753 case 'R': FTST(OP_FTRREAD);
2754 case 'W': FTST(OP_FTRWRITE);
2755 case 'X': FTST(OP_FTREXEC);
2756 case 'O': FTST(OP_FTROWNED);
2757 case 'e': FTST(OP_FTIS);
2758 case 'z': FTST(OP_FTZERO);
2759 case 's': FTST(OP_FTSIZE);
2760 case 'f': FTST(OP_FTFILE);
2761 case 'd': FTST(OP_FTDIR);
2762 case 'l': FTST(OP_FTLINK);
2763 case 'p': FTST(OP_FTPIPE);
2764 case 'S': FTST(OP_FTSOCK);
2765 case 'u': FTST(OP_FTSUID);
2766 case 'g': FTST(OP_FTSGID);
2767 case 'k': FTST(OP_FTSVTX);
2768 case 'b': FTST(OP_FTBLK);
2769 case 'c': FTST(OP_FTCHR);
2770 case 't': FTST(OP_FTTTY);
2771 case 'T': FTST(OP_FTTEXT);
2772 case 'B': FTST(OP_FTBINARY);
2773 case 'M': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTMTIME);
2774 case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME);
2775 case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME);
2777 Perl_croak(aTHX_ "Unrecognized file test: -%c", (int)tmp);
2784 if (PL_expect == XOPERATOR)
2789 else if (*s == '>') {
2792 if (isIDFIRST_lazy_if(s,UTF)) {
2793 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2801 if (PL_expect == XOPERATOR)
2804 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2806 OPERATOR('-'); /* unary minus */
2813 if (PL_expect == XOPERATOR)
2818 if (PL_expect == XOPERATOR)
2821 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2827 if (PL_expect != XOPERATOR) {
2828 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2829 PL_expect = XOPERATOR;
2830 force_ident(PL_tokenbuf, '*');
2843 if (PL_expect == XOPERATOR) {
2847 PL_tokenbuf[0] = '%';
2848 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2849 if (!PL_tokenbuf[1]) {
2851 yyerror("Final % should be \\% or %name");
2854 PL_pending_ident = '%';
2873 switch (PL_expect) {
2876 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2878 PL_bufptr = s; /* update in case we back off */
2884 PL_expect = XTERMBLOCK;
2888 while (isIDFIRST_lazy_if(s,UTF)) {
2889 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2890 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2891 if (tmp < 0) tmp = -tmp;
2906 d = scan_str(d,TRUE,TRUE);
2909 SvREFCNT_dec(PL_lex_stuff);
2910 PL_lex_stuff = Nullsv;
2912 /* MUST advance bufptr here to avoid bogus
2913 "at end of line" context messages from yyerror().
2915 PL_bufptr = s + len;
2916 yyerror("Unterminated attribute parameter in attribute list");
2919 return 0; /* EOF indicator */
2923 SV *sv = newSVpvn(s, len);
2924 sv_catsv(sv, PL_lex_stuff);
2925 attrs = append_elem(OP_LIST, attrs,
2926 newSVOP(OP_CONST, 0, sv));
2927 SvREFCNT_dec(PL_lex_stuff);
2928 PL_lex_stuff = Nullsv;
2931 attrs = append_elem(OP_LIST, attrs,
2932 newSVOP(OP_CONST, 0,
2936 if (*s == ':' && s[1] != ':')
2939 break; /* require real whitespace or :'s */
2941 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
2942 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
2943 char q = ((*s == '\'') ? '"' : '\'');
2944 /* If here for an expression, and parsed no attrs, back off. */
2945 if (tmp == '=' && !attrs) {
2949 /* MUST advance bufptr here to avoid bogus "at end of line"
2950 context messages from yyerror().
2954 yyerror("Unterminated attribute list");
2956 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
2964 PL_nextval[PL_nexttoke].opval = attrs;
2972 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
2973 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
2989 if (PL_lex_brackets <= 0)
2990 yyerror("Unmatched right square bracket");
2993 if (PL_lex_state == LEX_INTERPNORMAL) {
2994 if (PL_lex_brackets == 0) {
2995 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
2996 PL_lex_state = LEX_INTERPEND;
3003 if (PL_lex_brackets > 100) {
3004 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
3005 if (newlb != PL_lex_brackstack) {
3007 PL_lex_brackstack = newlb;
3010 switch (PL_expect) {
3012 if (PL_lex_formbrack) {
3016 if (PL_oldoldbufptr == PL_last_lop)
3017 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3019 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3020 OPERATOR(HASHBRACK);
3022 while (s < PL_bufend && SPACE_OR_TAB(*s))
3025 PL_tokenbuf[0] = '\0';
3026 if (d < PL_bufend && *d == '-') {
3027 PL_tokenbuf[0] = '-';
3029 while (d < PL_bufend && SPACE_OR_TAB(*d))
3032 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3033 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3035 while (d < PL_bufend && SPACE_OR_TAB(*d))
3038 char minus = (PL_tokenbuf[0] == '-');
3039 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3047 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3052 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3057 if (PL_oldoldbufptr == PL_last_lop)
3058 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3060 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3063 OPERATOR(HASHBRACK);
3064 /* This hack serves to disambiguate a pair of curlies
3065 * as being a block or an anon hash. Normally, expectation
3066 * determines that, but in cases where we're not in a
3067 * position to expect anything in particular (like inside
3068 * eval"") we have to resolve the ambiguity. This code
3069 * covers the case where the first term in the curlies is a
3070 * quoted string. Most other cases need to be explicitly
3071 * disambiguated by prepending a `+' before the opening
3072 * curly in order to force resolution as an anon hash.
3074 * XXX should probably propagate the outer expectation
3075 * into eval"" to rely less on this hack, but that could
3076 * potentially break current behavior of eval"".
3080 if (*s == '\'' || *s == '"' || *s == '`') {
3081 /* common case: get past first string, handling escapes */
3082 for (t++; t < PL_bufend && *t != *s;)
3083 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3087 else if (*s == 'q') {
3090 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3094 char open, close, term;
3097 while (t < PL_bufend && isSPACE(*t))
3101 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3105 for (t++; t < PL_bufend; t++) {
3106 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3108 else if (*t == open)
3112 for (t++; t < PL_bufend; t++) {
3113 if (*t == '\\' && t+1 < PL_bufend)
3115 else if (*t == close && --brackets <= 0)
3117 else if (*t == open)
3123 else if (isALNUM_lazy_if(t,UTF)) {
3125 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3128 while (t < PL_bufend && isSPACE(*t))
3130 /* if comma follows first term, call it an anon hash */
3131 /* XXX it could be a comma expression with loop modifiers */
3132 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3133 || (*t == '=' && t[1] == '>')))
3134 OPERATOR(HASHBRACK);
3135 if (PL_expect == XREF)
3138 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3144 yylval.ival = CopLINE(PL_curcop);
3145 if (isSPACE(*s) || *s == '#')
3146 PL_copline = NOLINE; /* invalidate current command line number */
3151 if (PL_lex_brackets <= 0)
3152 yyerror("Unmatched right curly bracket");
3154 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3155 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3156 PL_lex_formbrack = 0;
3157 if (PL_lex_state == LEX_INTERPNORMAL) {
3158 if (PL_lex_brackets == 0) {
3159 if (PL_expect & XFAKEBRACK) {
3160 PL_expect &= XENUMMASK;
3161 PL_lex_state = LEX_INTERPEND;
3163 return yylex(); /* ignore fake brackets */
3165 if (*s == '-' && s[1] == '>')
3166 PL_lex_state = LEX_INTERPENDMAYBE;
3167 else if (*s != '[' && *s != '{')
3168 PL_lex_state = LEX_INTERPEND;
3171 if (PL_expect & XFAKEBRACK) {
3172 PL_expect &= XENUMMASK;
3174 return yylex(); /* ignore fake brackets */
3184 if (PL_expect == XOPERATOR) {
3185 if (ckWARN(WARN_SEMICOLON)
3186 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3188 CopLINE_dec(PL_curcop);
3189 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3190 CopLINE_inc(PL_curcop);
3195 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3197 PL_expect = XOPERATOR;
3198 force_ident(PL_tokenbuf, '&');
3202 yylval.ival = (OPpENTERSUB_AMPER<<8);
3221 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3222 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3224 if (PL_expect == XSTATE && isALPHA(tmp) &&
3225 (s == PL_linestart+1 || s[-2] == '\n') )
3227 if (PL_in_eval && !PL_rsfp) {
3232 if (strnEQ(s,"=cut",4)) {
3246 PL_doextract = TRUE;
3249 if (PL_lex_brackets < PL_lex_formbrack) {
3251 #ifdef PERL_STRICT_CR
3252 for (t = s; SPACE_OR_TAB(*t); t++) ;
3254 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3256 if (*t == '\n' || *t == '#') {
3274 if (PL_expect != XOPERATOR) {
3275 if (s[1] != '<' && !strchr(s,'>'))
3278 s = scan_heredoc(s);
3280 s = scan_inputsymbol(s);
3281 TERM(sublex_start());
3286 SHop(OP_LEFT_SHIFT);
3300 SHop(OP_RIGHT_SHIFT);
3309 if (PL_expect == XOPERATOR) {
3310 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3313 return ','; /* grandfather non-comma-format format */
3317 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3318 PL_tokenbuf[0] = '@';
3319 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3320 sizeof PL_tokenbuf - 1, FALSE);
3321 if (PL_expect == XOPERATOR)
3322 no_op("Array length", s);
3323 if (!PL_tokenbuf[1])
3325 PL_expect = XOPERATOR;
3326 PL_pending_ident = '#';
3330 PL_tokenbuf[0] = '$';
3331 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3332 sizeof PL_tokenbuf - 1, FALSE);
3333 if (PL_expect == XOPERATOR)
3335 if (!PL_tokenbuf[1]) {
3337 yyerror("Final $ should be \\$ or $name");
3341 /* This kludge not intended to be bulletproof. */
3342 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3343 yylval.opval = newSVOP(OP_CONST, 0,
3344 newSViv(PL_compiling.cop_arybase));
3345 yylval.opval->op_private = OPpCONST_ARYBASE;
3351 if (PL_lex_state == LEX_NORMAL)
3354 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3357 PL_tokenbuf[0] = '@';
3358 if (ckWARN(WARN_SYNTAX)) {
3360 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3363 PL_bufptr = skipspace(PL_bufptr);
3364 while (t < PL_bufend && *t != ']')
3366 Perl_warner(aTHX_ WARN_SYNTAX,
3367 "Multidimensional syntax %.*s not supported",
3368 (t - PL_bufptr) + 1, PL_bufptr);
3372 else if (*s == '{') {
3373 PL_tokenbuf[0] = '%';
3374 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3375 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3377 char tmpbuf[sizeof PL_tokenbuf];
3379 for (t++; isSPACE(*t); t++) ;
3380 if (isIDFIRST_lazy_if(t,UTF)) {
3381 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3382 for (; isSPACE(*t); t++) ;
3383 if (*t == ';' && get_cv(tmpbuf, FALSE))
3384 Perl_warner(aTHX_ WARN_SYNTAX,
3385 "You need to quote \"%s\"", tmpbuf);
3391 PL_expect = XOPERATOR;
3392 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3393 bool islop = (PL_last_lop == PL_oldoldbufptr);
3394 if (!islop || PL_last_lop_op == OP_GREPSTART)
3395 PL_expect = XOPERATOR;
3396 else if (strchr("$@\"'`q", *s))
3397 PL_expect = XTERM; /* e.g. print $fh "foo" */
3398 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3399 PL_expect = XTERM; /* e.g. print $fh &sub */
3400 else if (isIDFIRST_lazy_if(s,UTF)) {
3401 char tmpbuf[sizeof PL_tokenbuf];
3402 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3403 if ((tmp = keyword(tmpbuf, len))) {
3404 /* binary operators exclude handle interpretations */
3416 PL_expect = XTERM; /* e.g. print $fh length() */
3421 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3422 if (gv && GvCVu(gv))
3423 PL_expect = XTERM; /* e.g. print $fh subr() */
3426 else if (isDIGIT(*s))
3427 PL_expect = XTERM; /* e.g. print $fh 3 */
3428 else if (*s == '.' && isDIGIT(s[1]))
3429 PL_expect = XTERM; /* e.g. print $fh .3 */
3430 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3431 PL_expect = XTERM; /* e.g. print $fh -1 */
3432 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3433 PL_expect = XTERM; /* print $fh <<"EOF" */
3435 PL_pending_ident = '$';
3439 if (PL_expect == XOPERATOR)
3441 PL_tokenbuf[0] = '@';
3442 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3443 if (!PL_tokenbuf[1]) {
3445 yyerror("Final @ should be \\@ or @name");
3448 if (PL_lex_state == LEX_NORMAL)
3450 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3452 PL_tokenbuf[0] = '%';
3454 /* Warn about @ where they meant $. */
3455 if (ckWARN(WARN_SYNTAX)) {
3456 if (*s == '[' || *s == '{') {
3458 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3460 if (*t == '}' || *t == ']') {
3462 PL_bufptr = skipspace(PL_bufptr);
3463 Perl_warner(aTHX_ WARN_SYNTAX,
3464 "Scalar value %.*s better written as $%.*s",
3465 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3470 PL_pending_ident = '@';
3473 case '/': /* may either be division or pattern */
3474 case '?': /* may either be conditional or pattern */
3475 if (PL_expect != XOPERATOR) {
3476 /* Disable warning on "study /blah/" */
3477 if (PL_oldoldbufptr == PL_last_uni
3478 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3479 || memNE(PL_last_uni, "study", 5)
3480 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3482 s = scan_pat(s,OP_MATCH);
3483 TERM(sublex_start());
3491 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3492 #ifdef PERL_STRICT_CR
3495 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3497 && (s == PL_linestart || s[-1] == '\n') )
3499 PL_lex_formbrack = 0;
3503 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3509 yylval.ival = OPf_SPECIAL;
3515 if (PL_expect != XOPERATOR)
3520 case '0': case '1': case '2': case '3': case '4':
3521 case '5': case '6': case '7': case '8': case '9':
3523 if (PL_expect == XOPERATOR)
3528 s = scan_str(s,FALSE,FALSE);
3529 if (PL_expect == XOPERATOR) {
3530 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3533 return ','; /* grandfather non-comma-format format */
3539 missingterm((char*)0);
3540 yylval.ival = OP_CONST;
3541 TERM(sublex_start());
3544 s = scan_str(s,FALSE,FALSE);
3545 if (PL_expect == XOPERATOR) {
3546 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3549 return ','; /* grandfather non-comma-format format */
3555 missingterm((char*)0);
3556 yylval.ival = OP_CONST;
3557 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3558 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3559 yylval.ival = OP_STRINGIFY;
3563 TERM(sublex_start());
3566 s = scan_str(s,FALSE,FALSE);
3567 if (PL_expect == XOPERATOR)
3568 no_op("Backticks",s);
3570 missingterm((char*)0);
3571 yylval.ival = OP_BACKTICK;
3573 TERM(sublex_start());
3577 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3578 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3580 if (PL_expect == XOPERATOR)
3581 no_op("Backslash",s);
3585 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3589 while (isDIGIT(*start) || *start == '_')
3591 if (*start == '.' && isDIGIT(start[1])) {
3595 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3596 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF)) {
3600 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3610 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3649 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3651 /* Some keywords can be followed by any delimiter, including ':' */
3652 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3653 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3654 (PL_tokenbuf[0] == 'q' &&
3655 strchr("qwxr", PL_tokenbuf[1])))));
3657 /* x::* is just a word, unless x is "CORE" */
3658 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3662 while (d < PL_bufend && isSPACE(*d))
3663 d++; /* no comments skipped here, or s### is misparsed */
3665 /* Is this a label? */
3666 if (!tmp && PL_expect == XSTATE
3667 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3669 yylval.pval = savepv(PL_tokenbuf);
3674 /* Check for keywords */
3675 tmp = keyword(PL_tokenbuf, len);
3677 /* Is this a word before a => operator? */
3678 if (*d == '=' && d[1] == '>') {
3680 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3681 yylval.opval->op_private = OPpCONST_BARE;
3685 if (tmp < 0) { /* second-class keyword? */
3686 GV *ogv = Nullgv; /* override (winner) */
3687 GV *hgv = Nullgv; /* hidden (loser) */
3688 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3690 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3693 if (GvIMPORTED_CV(gv))
3695 else if (! CvMETHOD(cv))
3699 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3700 (gv = *gvp) != (GV*)&PL_sv_undef &&
3701 GvCVu(gv) && GvIMPORTED_CV(gv))
3707 tmp = 0; /* overridden by import or by GLOBAL */
3710 && -tmp==KEY_lock /* XXX generalizable kludge */
3712 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3714 tmp = 0; /* any sub overrides "weak" keyword */
3716 else { /* no override */
3720 if (ckWARN(WARN_AMBIGUOUS) && hgv
3721 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3722 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3723 "Ambiguous call resolved as CORE::%s(), %s",
3724 GvENAME(hgv), "qualify as such or use &");
3731 default: /* not a keyword */
3734 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3736 /* Get the rest if it looks like a package qualifier */
3738 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3740 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3743 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3744 *s == '\'' ? "'" : "::");
3748 if (PL_expect == XOPERATOR) {
3749 if (PL_bufptr == PL_linestart) {
3750 CopLINE_dec(PL_curcop);
3751 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3752 CopLINE_inc(PL_curcop);
3755 no_op("Bareword",s);
3758 /* Look for a subroutine with this name in current package,
3759 unless name is "Foo::", in which case Foo is a bearword
3760 (and a package name). */
3763 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3765 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3766 Perl_warner(aTHX_ WARN_BAREWORD,
3767 "Bareword \"%s\" refers to nonexistent package",
3770 PL_tokenbuf[len] = '\0';
3777 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3780 /* if we saw a global override before, get the right name */
3783 sv = newSVpvn("CORE::GLOBAL::",14);
3784 sv_catpv(sv,PL_tokenbuf);
3787 sv = newSVpv(PL_tokenbuf,0);
3789 /* Presume this is going to be a bareword of some sort. */
3792 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3793 yylval.opval->op_private = OPpCONST_BARE;
3795 /* And if "Foo::", then that's what it certainly is. */
3800 /* See if it's the indirect object for a list operator. */
3802 if (PL_oldoldbufptr &&
3803 PL_oldoldbufptr < PL_bufptr &&
3804 (PL_oldoldbufptr == PL_last_lop
3805 || PL_oldoldbufptr == PL_last_uni) &&
3806 /* NO SKIPSPACE BEFORE HERE! */
3807 (PL_expect == XREF ||
3808 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3810 bool immediate_paren = *s == '(';
3812 /* (Now we can afford to cross potential line boundary.) */
3815 /* Two barewords in a row may indicate method call. */
3817 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3820 /* If not a declared subroutine, it's an indirect object. */
3821 /* (But it's an indir obj regardless for sort.) */
3823 if ((PL_last_lop_op == OP_SORT ||
3824 (!immediate_paren && (!gv || !GvCVu(gv)))) &&
3825 (PL_last_lop_op != OP_MAPSTART &&
3826 PL_last_lop_op != OP_GREPSTART))
3828 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3834 PL_expect = XOPERATOR;
3837 /* Is this a word before a => operator? */
3838 if (*s == '=' && s[1] == '>') {
3840 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
3844 /* If followed by a paren, it's certainly a subroutine. */
3847 if (gv && GvCVu(gv)) {
3848 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
3849 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3854 PL_nextval[PL_nexttoke].opval = yylval.opval;
3855 PL_expect = XOPERATOR;
3861 /* If followed by var or block, call it a method (unless sub) */
3863 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3864 PL_last_lop = PL_oldbufptr;
3865 PL_last_lop_op = OP_METHOD;
3869 /* If followed by a bareword, see if it looks like indir obj. */
3871 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3874 /* Not a method, so call it a subroutine (if defined) */
3876 if (gv && GvCVu(gv)) {
3878 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3879 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3880 "Ambiguous use of -%s resolved as -&%s()",
3881 PL_tokenbuf, PL_tokenbuf);
3882 /* Check for a constant sub */
3884 if ((sv = cv_const_sv(cv))) {
3886 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3887 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3888 yylval.opval->op_private = 0;
3892 /* Resolve to GV now. */
3893 op_free(yylval.opval);
3894 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3895 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3896 PL_last_lop = PL_oldbufptr;
3897 PL_last_lop_op = OP_ENTERSUB;
3898 /* Is there a prototype? */
3901 char *proto = SvPV((SV*)cv, len);
3904 if (strEQ(proto, "$"))
3906 if (*proto == '&' && *s == '{') {
3907 sv_setpv(PL_subname,"__ANON__");
3911 PL_nextval[PL_nexttoke].opval = yylval.opval;
3917 /* Call it a bare word */
3919 if (PL_hints & HINT_STRICT_SUBS)
3920 yylval.opval->op_private |= OPpCONST_STRICT;
3923 if (ckWARN(WARN_RESERVED)) {
3924 if (lastchar != '-') {
3925 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
3927 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
3934 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
3935 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3936 "Operator or semicolon missing before %c%s",
3937 lastchar, PL_tokenbuf);
3938 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3939 "Ambiguous use of %c resolved as operator %c",
3940 lastchar, lastchar);
3946 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3947 newSVpv(CopFILE(PL_curcop),0));
3951 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3952 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
3955 case KEY___PACKAGE__:
3956 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3958 ? newSVsv(PL_curstname)
3967 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
3968 char *pname = "main";
3969 if (PL_tokenbuf[2] == 'D')
3970 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
3971 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
3974 GvIOp(gv) = newIO();
3975 IoIFP(GvIOp(gv)) = PL_rsfp;
3976 #if defined(HAS_FCNTL) && defined(F_SETFD)
3978 int fd = PerlIO_fileno(PL_rsfp);
3979 fcntl(fd,F_SETFD,fd >= 3);
3982 /* Mark this internal pseudo-handle as clean */
3983 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
3985 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
3986 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
3987 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
3989 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
3990 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
3991 /* if the script was opened in binmode, we need to revert
3992 * it to text mode for compatibility; but only iff it has CRs
3993 * XXX this is a questionable hack at best. */
3994 if (PL_bufend-PL_bufptr > 2
3995 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
3998 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
3999 loc = PerlIO_tell(PL_rsfp);
4000 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4002 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4003 #if defined(__BORLANDC__)
4004 /* XXX see note in do_binmode() */
4005 ((FILE*)PL_rsfp)->flags |= _F_BIN;
4008 PerlIO_seek(PL_rsfp, loc, 0);
4023 if (PL_expect == XSTATE) {
4030 if (*s == ':' && s[1] == ':') {
4033 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4034 if (!(tmp = keyword(PL_tokenbuf, len)))
4035 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4049 LOP(OP_ACCEPT,XTERM);
4055 LOP(OP_ATAN2,XTERM);
4061 LOP(OP_BINMODE,XTERM);
4064 LOP(OP_BLESS,XTERM);
4073 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4090 if (!PL_cryptseen) {
4091 PL_cryptseen = TRUE;
4095 LOP(OP_CRYPT,XTERM);
4098 if (ckWARN(WARN_CHMOD)) {
4099 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4100 if (*d != '0' && isDIGIT(*d))
4101 Perl_warner(aTHX_ WARN_CHMOD,
4102 "chmod() mode argument is missing initial 0");
4104 LOP(OP_CHMOD,XTERM);
4107 LOP(OP_CHOWN,XTERM);
4110 LOP(OP_CONNECT,XTERM);
4126 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4130 PL_hints |= HINT_BLOCK_SCOPE;
4140 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4141 LOP(OP_DBMOPEN,XTERM);
4147 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4154 yylval.ival = CopLINE(PL_curcop);
4168 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4169 UNIBRACK(OP_ENTEREVAL);
4184 case KEY_endhostent:
4190 case KEY_endservent:
4193 case KEY_endprotoent:
4204 yylval.ival = CopLINE(PL_curcop);
4206 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4208 if ((PL_bufend - p) >= 3 &&
4209 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4211 else if ((PL_bufend - p) >= 4 &&
4212 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4215 if (isIDFIRST_lazy_if(p,UTF)) {
4216 p = scan_ident(p, PL_bufend,
4217 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4221 Perl_croak(aTHX_ "Missing $ on loop variable");
4226 LOP(OP_FORMLINE,XTERM);
4232 LOP(OP_FCNTL,XTERM);
4238 LOP(OP_FLOCK,XTERM);
4247 LOP(OP_GREPSTART, XREF);
4250 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4265 case KEY_getpriority:
4266 LOP(OP_GETPRIORITY,XTERM);
4268 case KEY_getprotobyname:
4271 case KEY_getprotobynumber:
4272 LOP(OP_GPBYNUMBER,XTERM);
4274 case KEY_getprotoent:
4286 case KEY_getpeername:
4287 UNI(OP_GETPEERNAME);
4289 case KEY_gethostbyname:
4292 case KEY_gethostbyaddr:
4293 LOP(OP_GHBYADDR,XTERM);
4295 case KEY_gethostent:
4298 case KEY_getnetbyname:
4301 case KEY_getnetbyaddr:
4302 LOP(OP_GNBYADDR,XTERM);
4307 case KEY_getservbyname:
4308 LOP(OP_GSBYNAME,XTERM);
4310 case KEY_getservbyport:
4311 LOP(OP_GSBYPORT,XTERM);
4313 case KEY_getservent:
4316 case KEY_getsockname:
4317 UNI(OP_GETSOCKNAME);
4319 case KEY_getsockopt:
4320 LOP(OP_GSOCKOPT,XTERM);
4342 yylval.ival = CopLINE(PL_curcop);
4346 LOP(OP_INDEX,XTERM);
4352 LOP(OP_IOCTL,XTERM);
4364 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4396 LOP(OP_LISTEN,XTERM);
4405 s = scan_pat(s,OP_MATCH);
4406 TERM(sublex_start());
4409 LOP(OP_MAPSTART, XREF);
4412 LOP(OP_MKDIR,XTERM);
4415 LOP(OP_MSGCTL,XTERM);
4418 LOP(OP_MSGGET,XTERM);
4421 LOP(OP_MSGRCV,XTERM);
4424 LOP(OP_MSGSND,XTERM);
4430 if (isIDFIRST_lazy_if(s,UTF)) {
4431 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4432 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4434 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4435 if (!PL_in_my_stash) {
4438 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4446 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4453 if (PL_expect != XSTATE)
4454 yyerror("\"no\" not allowed in expression");
4455 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4456 s = force_version(s);
4461 if (*s == '(' || (s = skipspace(s), *s == '('))
4468 if (isIDFIRST_lazy_if(s,UTF)) {
4470 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4472 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4473 Perl_warner(aTHX_ WARN_PRECEDENCE,
4474 "Precedence problem: open %.*s should be open(%.*s)",
4480 yylval.ival = OP_OR;
4490 LOP(OP_OPEN_DIR,XTERM);
4493 checkcomma(s,PL_tokenbuf,"filehandle");
4497 checkcomma(s,PL_tokenbuf,"filehandle");
4516 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4520 LOP(OP_PIPE_OP,XTERM);
4523 s = scan_str(s,FALSE,FALSE);
4525 missingterm((char*)0);
4526 yylval.ival = OP_CONST;
4527 TERM(sublex_start());
4533 s = scan_str(s,FALSE,FALSE);
4535 missingterm((char*)0);
4537 if (SvCUR(PL_lex_stuff)) {
4540 d = SvPV_force(PL_lex_stuff, len);
4542 for (; isSPACE(*d) && len; --len, ++d) ;
4545 if (!warned && ckWARN(WARN_QW)) {
4546 for (; !isSPACE(*d) && len; --len, ++d) {
4548 Perl_warner(aTHX_ WARN_QW,
4549 "Possible attempt to separate words with commas");
4552 else if (*d == '#') {
4553 Perl_warner(aTHX_ WARN_QW,
4554 "Possible attempt to put comments in qw() list");
4560 for (; !isSPACE(*d) && len; --len, ++d) ;
4562 words = append_elem(OP_LIST, words,
4563 newSVOP(OP_CONST, 0, tokeq(newSVpvn(b, d-b))));
4567 PL_nextval[PL_nexttoke].opval = words;
4572 SvREFCNT_dec(PL_lex_stuff);
4573 PL_lex_stuff = Nullsv;
4578 s = scan_str(s,FALSE,FALSE);
4580 missingterm((char*)0);
4581 yylval.ival = OP_STRINGIFY;
4582 if (SvIVX(PL_lex_stuff) == '\'')
4583 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4584 TERM(sublex_start());
4587 s = scan_pat(s,OP_QR);
4588 TERM(sublex_start());
4591 s = scan_str(s,FALSE,FALSE);
4593 missingterm((char*)0);
4594 yylval.ival = OP_BACKTICK;
4596 TERM(sublex_start());
4603 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4604 s = force_version(s);
4607 *PL_tokenbuf = '\0';
4608 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4609 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4610 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4612 yyerror("<> should be quotes");
4620 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4624 LOP(OP_RENAME,XTERM);
4633 LOP(OP_RINDEX,XTERM);
4656 LOP(OP_REVERSE,XTERM);
4667 TERM(sublex_start());
4669 TOKEN(1); /* force error */
4678 LOP(OP_SELECT,XTERM);
4684 LOP(OP_SEMCTL,XTERM);
4687 LOP(OP_SEMGET,XTERM);
4690 LOP(OP_SEMOP,XTERM);
4696 LOP(OP_SETPGRP,XTERM);
4698 case KEY_setpriority:
4699 LOP(OP_SETPRIORITY,XTERM);
4701 case KEY_sethostent:
4707 case KEY_setservent:
4710 case KEY_setprotoent:
4720 LOP(OP_SEEKDIR,XTERM);
4722 case KEY_setsockopt:
4723 LOP(OP_SSOCKOPT,XTERM);
4729 LOP(OP_SHMCTL,XTERM);
4732 LOP(OP_SHMGET,XTERM);
4735 LOP(OP_SHMREAD,XTERM);
4738 LOP(OP_SHMWRITE,XTERM);
4741 LOP(OP_SHUTDOWN,XTERM);
4750 LOP(OP_SOCKET,XTERM);
4752 case KEY_socketpair:
4753 LOP(OP_SOCKPAIR,XTERM);
4756 checkcomma(s,PL_tokenbuf,"subroutine name");
4758 if (*s == ';' || *s == ')') /* probably a close */
4759 Perl_croak(aTHX_ "sort is now a reserved word");
4761 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4765 LOP(OP_SPLIT,XTERM);
4768 LOP(OP_SPRINTF,XTERM);
4771 LOP(OP_SPLICE,XTERM);
4786 LOP(OP_SUBSTR,XTERM);
4792 char tmpbuf[sizeof PL_tokenbuf];
4794 expectation attrful;
4795 bool have_name, have_proto;
4800 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4801 (*s == ':' && s[1] == ':'))
4804 attrful = XATTRBLOCK;
4805 /* remember buffer pos'n for later force_word */
4806 tboffset = s - PL_oldbufptr;
4807 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4808 if (strchr(tmpbuf, ':'))
4809 sv_setpv(PL_subname, tmpbuf);
4811 sv_setsv(PL_subname,PL_curstname);
4812 sv_catpvn(PL_subname,"::",2);
4813 sv_catpvn(PL_subname,tmpbuf,len);
4820 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4821 PL_expect = XTERMBLOCK;
4822 attrful = XATTRTERM;
4823 sv_setpv(PL_subname,"?");
4827 if (key == KEY_format) {
4829 PL_lex_formbrack = PL_lex_brackets + 1;
4831 (void) force_word(PL_oldbufptr + tboffset, WORD,
4836 /* Look for a prototype */
4840 s = scan_str(s,FALSE,FALSE);
4843 SvREFCNT_dec(PL_lex_stuff);
4844 PL_lex_stuff = Nullsv;
4845 Perl_croak(aTHX_ "Prototype not terminated");
4848 d = SvPVX(PL_lex_stuff);
4850 for (p = d; *p; ++p) {
4855 SvCUR(PL_lex_stuff) = tmp;
4863 if (*s == ':' && s[1] != ':')
4864 PL_expect = attrful;
4867 PL_nextval[PL_nexttoke].opval =
4868 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4869 PL_lex_stuff = Nullsv;
4873 sv_setpv(PL_subname,"__ANON__");
4876 (void) force_word(PL_oldbufptr + tboffset, WORD,
4885 LOP(OP_SYSTEM,XREF);
4888 LOP(OP_SYMLINK,XTERM);
4891 LOP(OP_SYSCALL,XTERM);
4894 LOP(OP_SYSOPEN,XTERM);
4897 LOP(OP_SYSSEEK,XTERM);
4900 LOP(OP_SYSREAD,XTERM);
4903 LOP(OP_SYSWRITE,XTERM);
4907 TERM(sublex_start());
4928 LOP(OP_TRUNCATE,XTERM);
4940 yylval.ival = CopLINE(PL_curcop);
4944 yylval.ival = CopLINE(PL_curcop);
4948 LOP(OP_UNLINK,XTERM);
4954 LOP(OP_UNPACK,XTERM);
4957 LOP(OP_UTIME,XTERM);
4960 if (ckWARN(WARN_UMASK)) {
4961 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4962 if (*d != '0' && isDIGIT(*d))
4963 Perl_warner(aTHX_ WARN_UMASK,
4964 "umask: argument is missing initial 0");
4969 LOP(OP_UNSHIFT,XTERM);
4972 if (PL_expect != XSTATE)
4973 yyerror("\"use\" not allowed in expression");
4975 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4976 s = force_version(s);
4977 if (*s == ';' || (s = skipspace(s), *s == ';')) {
4978 PL_nextval[PL_nexttoke].opval = Nullop;
4983 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4984 s = force_version(s);
4996 yylval.ival = CopLINE(PL_curcop);
5000 PL_hints |= HINT_BLOCK_SCOPE;
5007 LOP(OP_WAITPID,XTERM);
5015 static char ctl_l[2];
5017 if (ctl_l[0] == '\0')
5018 ctl_l[0] = toCTRL('L');
5019 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5022 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5027 if (PL_expect == XOPERATOR)
5033 yylval.ival = OP_XOR;
5038 TERM(sublex_start());
5043 #pragma segment Main
5047 Perl_keyword(pTHX_ register char *d, I32 len)
5052 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5053 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5054 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5055 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5056 if (strEQ(d,"__END__")) return KEY___END__;
5060 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5065 if (strEQ(d,"and")) return -KEY_and;
5066 if (strEQ(d,"abs")) return -KEY_abs;
5069 if (strEQ(d,"alarm")) return -KEY_alarm;
5070 if (strEQ(d,"atan2")) return -KEY_atan2;
5073 if (strEQ(d,"accept")) return -KEY_accept;
5078 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5081 if (strEQ(d,"bless")) return -KEY_bless;
5082 if (strEQ(d,"bind")) return -KEY_bind;
5083 if (strEQ(d,"binmode")) return -KEY_binmode;
5086 if (strEQ(d,"CORE")) return -KEY_CORE;
5087 if (strEQ(d,"CHECK")) return KEY_CHECK;
5092 if (strEQ(d,"cmp")) return -KEY_cmp;
5093 if (strEQ(d,"chr")) return -KEY_chr;
5094 if (strEQ(d,"cos")) return -KEY_cos;
5097 if (strEQ(d,"chop")) return -KEY_chop;
5100 if (strEQ(d,"close")) return -KEY_close;
5101 if (strEQ(d,"chdir")) return -KEY_chdir;
5102 if (strEQ(d,"chomp")) return -KEY_chomp;
5103 if (strEQ(d,"chmod")) return -KEY_chmod;
5104 if (strEQ(d,"chown")) return -KEY_chown;
5105 if (strEQ(d,"crypt")) return -KEY_crypt;
5108 if (strEQ(d,"chroot")) return -KEY_chroot;
5109 if (strEQ(d,"caller")) return -KEY_caller;
5112 if (strEQ(d,"connect")) return -KEY_connect;
5115 if (strEQ(d,"closedir")) return -KEY_closedir;
5116 if (strEQ(d,"continue")) return -KEY_continue;
5121 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5126 if (strEQ(d,"do")) return KEY_do;
5129 if (strEQ(d,"die")) return -KEY_die;
5132 if (strEQ(d,"dump")) return -KEY_dump;
5135 if (strEQ(d,"delete")) return KEY_delete;
5138 if (strEQ(d,"defined")) return KEY_defined;
5139 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5142 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5147 if (strEQ(d,"END")) return KEY_END;
5152 if (strEQ(d,"eq")) return -KEY_eq;
5155 if (strEQ(d,"eof")) return -KEY_eof;
5156 if (strEQ(d,"exp")) return -KEY_exp;
5159 if (strEQ(d,"else")) return KEY_else;
5160 if (strEQ(d,"exit")) return -KEY_exit;
5161 if (strEQ(d,"eval")) return KEY_eval;
5162 if (strEQ(d,"exec")) return -KEY_exec;
5163 if (strEQ(d,"each")) return KEY_each;
5166 if (strEQ(d,"elsif")) return KEY_elsif;
5169 if (strEQ(d,"exists")) return KEY_exists;
5170 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5173 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5174 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5177 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5180 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5181 if (strEQ(d,"endservent")) return -KEY_endservent;
5184 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5191 if (strEQ(d,"for")) return KEY_for;
5194 if (strEQ(d,"fork")) return -KEY_fork;
5197 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5198 if (strEQ(d,"flock")) return -KEY_flock;
5201 if (strEQ(d,"format")) return KEY_format;
5202 if (strEQ(d,"fileno")) return -KEY_fileno;
5205 if (strEQ(d,"foreach")) return KEY_foreach;
5208 if (strEQ(d,"formline")) return -KEY_formline;
5213 if (strnEQ(d,"get",3)) {
5218 if (strEQ(d,"ppid")) return -KEY_getppid;
5219 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5222 if (strEQ(d,"pwent")) return -KEY_getpwent;
5223 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5224 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5227 if (strEQ(d,"peername")) return -KEY_getpeername;
5228 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5229 if (strEQ(d,"priority")) return -KEY_getpriority;
5232 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5235 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5239 else if (*d == 'h') {
5240 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5241 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5242 if (strEQ(d,"hostent")) return -KEY_gethostent;
5244 else if (*d == 'n') {
5245 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5246 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5247 if (strEQ(d,"netent")) return -KEY_getnetent;
5249 else if (*d == 's') {
5250 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5251 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5252 if (strEQ(d,"servent")) return -KEY_getservent;
5253 if (strEQ(d,"sockname")) return -KEY_getsockname;
5254 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5256 else if (*d == 'g') {
5257 if (strEQ(d,"grent")) return -KEY_getgrent;
5258 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5259 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5261 else if (*d == 'l') {
5262 if (strEQ(d,"login")) return -KEY_getlogin;
5264 else if (strEQ(d,"c")) return -KEY_getc;
5269 if (strEQ(d,"gt")) return -KEY_gt;
5270 if (strEQ(d,"ge")) return -KEY_ge;
5273 if (strEQ(d,"grep")) return KEY_grep;
5274 if (strEQ(d,"goto")) return KEY_goto;
5275 if (strEQ(d,"glob")) return KEY_glob;
5278 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5283 if (strEQ(d,"hex")) return -KEY_hex;
5286 if (strEQ(d,"INIT")) return KEY_INIT;
5291 if (strEQ(d,"if")) return KEY_if;
5294 if (strEQ(d,"int")) return -KEY_int;
5297 if (strEQ(d,"index")) return -KEY_index;
5298 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5303 if (strEQ(d,"join")) return -KEY_join;
5307 if (strEQ(d,"keys")) return KEY_keys;
5308 if (strEQ(d,"kill")) return -KEY_kill;
5314 if (strEQ(d,"lt")) return -KEY_lt;
5315 if (strEQ(d,"le")) return -KEY_le;
5316 if (strEQ(d,"lc")) return -KEY_lc;
5319 if (strEQ(d,"log")) return -KEY_log;
5322 if (strEQ(d,"last")) return KEY_last;
5323 if (strEQ(d,"link")) return -KEY_link;
5324 if (strEQ(d,"lock")) return -KEY_lock;
5327 if (strEQ(d,"local")) return KEY_local;
5328 if (strEQ(d,"lstat")) return -KEY_lstat;
5331 if (strEQ(d,"length")) return -KEY_length;
5332 if (strEQ(d,"listen")) return -KEY_listen;
5335 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5338 if (strEQ(d,"localtime")) return -KEY_localtime;
5344 case 1: return KEY_m;
5346 if (strEQ(d,"my")) return KEY_my;
5349 if (strEQ(d,"map")) return KEY_map;
5352 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5355 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5356 if (strEQ(d,"msgget")) return -KEY_msgget;
5357 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5358 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5363 if (strEQ(d,"next")) return KEY_next;
5364 if (strEQ(d,"ne")) return -KEY_ne;
5365 if (strEQ(d,"not")) return -KEY_not;
5366 if (strEQ(d,"no")) return KEY_no;
5371 if (strEQ(d,"or")) return -KEY_or;
5374 if (strEQ(d,"ord")) return -KEY_ord;
5375 if (strEQ(d,"oct")) return -KEY_oct;
5376 if (strEQ(d,"our")) return KEY_our;
5379 if (strEQ(d,"open")) return -KEY_open;
5382 if (strEQ(d,"opendir")) return -KEY_opendir;
5389 if (strEQ(d,"pop")) return KEY_pop;
5390 if (strEQ(d,"pos")) return KEY_pos;
5393 if (strEQ(d,"push")) return KEY_push;
5394 if (strEQ(d,"pack")) return -KEY_pack;
5395 if (strEQ(d,"pipe")) return -KEY_pipe;
5398 if (strEQ(d,"print")) return KEY_print;
5401 if (strEQ(d,"printf")) return KEY_printf;
5404 if (strEQ(d,"package")) return KEY_package;
5407 if (strEQ(d,"prototype")) return KEY_prototype;
5412 if (strEQ(d,"q")) return KEY_q;
5413 if (strEQ(d,"qr")) return KEY_qr;
5414 if (strEQ(d,"qq")) return KEY_qq;
5415 if (strEQ(d,"qw")) return KEY_qw;
5416 if (strEQ(d,"qx")) return KEY_qx;
5418 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5423 if (strEQ(d,"ref")) return -KEY_ref;
5426 if (strEQ(d,"read")) return -KEY_read;
5427 if (strEQ(d,"rand")) return -KEY_rand;
5428 if (strEQ(d,"recv")) return -KEY_recv;
5429 if (strEQ(d,"redo")) return KEY_redo;
5432 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5433 if (strEQ(d,"reset")) return -KEY_reset;
5436 if (strEQ(d,"return")) return KEY_return;
5437 if (strEQ(d,"rename")) return -KEY_rename;
5438 if (strEQ(d,"rindex")) return -KEY_rindex;
5441 if (strEQ(d,"require")) return -KEY_require;
5442 if (strEQ(d,"reverse")) return -KEY_reverse;
5443 if (strEQ(d,"readdir")) return -KEY_readdir;
5446 if (strEQ(d,"readlink")) return -KEY_readlink;
5447 if (strEQ(d,"readline")) return -KEY_readline;
5448 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5451 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5457 case 0: return KEY_s;
5459 if (strEQ(d,"scalar")) return KEY_scalar;
5464 if (strEQ(d,"seek")) return -KEY_seek;
5465 if (strEQ(d,"send")) return -KEY_send;
5468 if (strEQ(d,"semop")) return -KEY_semop;
5471 if (strEQ(d,"select")) return -KEY_select;
5472 if (strEQ(d,"semctl")) return -KEY_semctl;
5473 if (strEQ(d,"semget")) return -KEY_semget;
5476 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5477 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5480 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5481 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5484 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5487 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5488 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5489 if (strEQ(d,"setservent")) return -KEY_setservent;
5492 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5493 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5500 if (strEQ(d,"shift")) return KEY_shift;
5503 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5504 if (strEQ(d,"shmget")) return -KEY_shmget;
5507 if (strEQ(d,"shmread")) return -KEY_shmread;
5510 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5511 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5516 if (strEQ(d,"sin")) return -KEY_sin;
5519 if (strEQ(d,"sleep")) return -KEY_sleep;
5522 if (strEQ(d,"sort")) return KEY_sort;
5523 if (strEQ(d,"socket")) return -KEY_socket;
5524 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5527 if (strEQ(d,"split")) return KEY_split;
5528 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5529 if (strEQ(d,"splice")) return KEY_splice;
5532 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5535 if (strEQ(d,"srand")) return -KEY_srand;
5538 if (strEQ(d,"stat")) return -KEY_stat;
5539 if (strEQ(d,"study")) return KEY_study;
5542 if (strEQ(d,"substr")) return -KEY_substr;
5543 if (strEQ(d,"sub")) return KEY_sub;
5548 if (strEQ(d,"system")) return -KEY_system;
5551 if (strEQ(d,"symlink")) return -KEY_symlink;
5552 if (strEQ(d,"syscall")) return -KEY_syscall;
5553 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5554 if (strEQ(d,"sysread")) return -KEY_sysread;
5555 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5558 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5567 if (strEQ(d,"tr")) return KEY_tr;
5570 if (strEQ(d,"tie")) return KEY_tie;
5573 if (strEQ(d,"tell")) return -KEY_tell;
5574 if (strEQ(d,"tied")) return KEY_tied;
5575 if (strEQ(d,"time")) return -KEY_time;
5578 if (strEQ(d,"times")) return -KEY_times;
5581 if (strEQ(d,"telldir")) return -KEY_telldir;
5584 if (strEQ(d,"truncate")) return -KEY_truncate;
5591 if (strEQ(d,"uc")) return -KEY_uc;
5594 if (strEQ(d,"use")) return KEY_use;
5597 if (strEQ(d,"undef")) return KEY_undef;
5598 if (strEQ(d,"until")) return KEY_until;
5599 if (strEQ(d,"untie")) return KEY_untie;
5600 if (strEQ(d,"utime")) return -KEY_utime;
5601 if (strEQ(d,"umask")) return -KEY_umask;
5604 if (strEQ(d,"unless")) return KEY_unless;
5605 if (strEQ(d,"unpack")) return -KEY_unpack;
5606 if (strEQ(d,"unlink")) return -KEY_unlink;
5609 if (strEQ(d,"unshift")) return KEY_unshift;
5610 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5615 if (strEQ(d,"values")) return -KEY_values;
5616 if (strEQ(d,"vec")) return -KEY_vec;
5621 if (strEQ(d,"warn")) return -KEY_warn;
5622 if (strEQ(d,"wait")) return -KEY_wait;
5625 if (strEQ(d,"while")) return KEY_while;
5626 if (strEQ(d,"write")) return -KEY_write;
5629 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5632 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5637 if (len == 1) return -KEY_x;
5638 if (strEQ(d,"xor")) return -KEY_xor;
5641 if (len == 1) return KEY_y;
5650 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5654 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5655 dTHR; /* only for ckWARN */
5656 if (ckWARN(WARN_SYNTAX)) {
5658 for (w = s+2; *w && level; w++) {
5665 for (; *w && isSPACE(*w); w++) ;
5666 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5667 Perl_warner(aTHX_ WARN_SYNTAX,
5668 "%s (...) interpreted as function",name);
5671 while (s < PL_bufend && isSPACE(*s))
5675 while (s < PL_bufend && isSPACE(*s))
5677 if (isIDFIRST_lazy_if(s,UTF)) {
5679 while (isALNUM_lazy_if(s,UTF))
5681 while (s < PL_bufend && isSPACE(*s))
5686 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5690 Perl_croak(aTHX_ "No comma allowed after %s", what);
5695 /* Either returns sv, or mortalizes sv and returns a new SV*.
5696 Best used as sv=new_constant(..., sv, ...).
5697 If s, pv are NULL, calls subroutine with one argument,
5698 and type is used with error messages only. */
5701 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5705 HV *table = GvHV(PL_hintgv); /* ^H */
5709 const char *why1, *why2, *why3;
5711 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
5714 why1 = "%^H is not consistent";
5715 why2 = strEQ(key,"charnames")
5716 ? " (missing \"use charnames ...\"?)"
5720 msg = Perl_newSVpvf(aTHX_ "constant(%s): %s%s%s",
5721 (type ? type: "undef"), why1, why2, why3);
5722 yyerror(SvPVX(msg));
5726 cvp = hv_fetch(table, key, strlen(key), FALSE);
5727 if (!cvp || !SvOK(*cvp)) {
5730 why3 = "} is not defined";
5733 sv_2mortal(sv); /* Parent created it permanently */
5736 pv = sv_2mortal(newSVpvn(s, len));
5738 typesv = sv_2mortal(newSVpv(type, 0));
5740 typesv = &PL_sv_undef;
5742 PUSHSTACKi(PERLSI_OVERLOAD);
5754 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5758 /* Check the eval first */
5759 if (!PL_in_eval && SvTRUE(ERRSV)) {
5761 sv_catpv(ERRSV, "Propagated");
5762 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5764 res = SvREFCNT_inc(sv);
5768 (void)SvREFCNT_inc(res);
5777 why1 = "Call to &{$^H{";
5779 why3 = "}} did not return a defined value";
5788 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5790 register char *d = dest;
5791 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5794 Perl_croak(aTHX_ ident_too_long);
5795 if (isALNUM(*s)) /* UTF handled below */
5797 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
5802 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5806 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5807 char *t = s + UTF8SKIP(s);
5808 while (*t & 0x80 && is_utf8_mark((U8*)t))
5810 if (d + (t - s) > e)
5811 Perl_croak(aTHX_ ident_too_long);
5812 Copy(s, d, t - s, char);
5825 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5835 e = d + destlen - 3; /* two-character token, ending NUL */
5837 while (isDIGIT(*s)) {
5839 Perl_croak(aTHX_ ident_too_long);
5846 Perl_croak(aTHX_ ident_too_long);
5847 if (isALNUM(*s)) /* UTF handled below */
5849 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
5854 else if (*s == ':' && s[1] == ':') {
5858 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5859 char *t = s + UTF8SKIP(s);
5860 while (*t & 0x80 && is_utf8_mark((U8*)t))
5862 if (d + (t - s) > e)
5863 Perl_croak(aTHX_ ident_too_long);
5864 Copy(s, d, t - s, char);
5875 if (PL_lex_state != LEX_NORMAL)
5876 PL_lex_state = LEX_INTERPENDMAYBE;
5879 if (*s == '$' && s[1] &&
5880 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
5893 if (*d == '^' && *s && isCONTROLVAR(*s)) {
5898 if (isSPACE(s[-1])) {
5901 if (!SPACE_OR_TAB(ch)) {
5907 if (isIDFIRST_lazy_if(d,UTF)) {
5911 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
5913 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
5916 Copy(s, d, e - s, char);
5921 while ((isALNUM(*s) || *s == ':') && d < e)
5924 Perl_croak(aTHX_ ident_too_long);
5927 while (s < send && SPACE_OR_TAB(*s)) s++;
5928 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
5929 dTHR; /* only for ckWARN */
5930 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
5931 const char *brack = *s == '[' ? "[...]" : "{...}";
5932 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5933 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
5934 funny, dest, brack, funny, dest, brack);
5937 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
5941 /* Handle extended ${^Foo} variables
5942 * 1999-02-27 mjd-perl-patch@plover.com */
5943 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
5947 while (isALNUM(*s) && d < e) {
5951 Perl_croak(aTHX_ ident_too_long);
5956 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
5957 PL_lex_state = LEX_INTERPEND;
5960 if (PL_lex_state == LEX_NORMAL) {
5961 dTHR; /* only for ckWARN */
5962 if (ckWARN(WARN_AMBIGUOUS) &&
5963 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
5965 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5966 "Ambiguous use of %c{%s} resolved to %c%s",
5967 funny, dest, funny, dest);
5972 s = bracket; /* let the parser handle it */
5976 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
5977 PL_lex_state = LEX_INTERPEND;
5982 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
5987 *pmfl |= PMf_GLOBAL;
5989 *pmfl |= PMf_CONTINUE;
5993 *pmfl |= PMf_MULTILINE;
5995 *pmfl |= PMf_SINGLELINE;
5997 *pmfl |= PMf_EXTENDED;
6001 S_scan_pat(pTHX_ char *start, I32 type)
6006 s = scan_str(start,FALSE,FALSE);
6009 SvREFCNT_dec(PL_lex_stuff);
6010 PL_lex_stuff = Nullsv;
6011 Perl_croak(aTHX_ "Search pattern not terminated");
6014 pm = (PMOP*)newPMOP(type, 0);
6015 if (PL_multi_open == '?')
6016 pm->op_pmflags |= PMf_ONCE;
6018 while (*s && strchr("iomsx", *s))
6019 pmflag(&pm->op_pmflags,*s++);
6022 while (*s && strchr("iogcmsx", *s))
6023 pmflag(&pm->op_pmflags,*s++);
6025 pm->op_pmpermflags = pm->op_pmflags;
6027 PL_lex_op = (OP*)pm;
6028 yylval.ival = OP_MATCH;
6033 S_scan_subst(pTHX_ char *start)
6040 yylval.ival = OP_NULL;
6042 s = scan_str(start,FALSE,FALSE);
6046 SvREFCNT_dec(PL_lex_stuff);
6047 PL_lex_stuff = Nullsv;
6048 Perl_croak(aTHX_ "Substitution pattern not terminated");
6051 if (s[-1] == PL_multi_open)
6054 first_start = PL_multi_start;
6055 s = scan_str(s,FALSE,FALSE);
6058 SvREFCNT_dec(PL_lex_stuff);
6059 PL_lex_stuff = Nullsv;
6061 SvREFCNT_dec(PL_lex_repl);
6062 PL_lex_repl = Nullsv;
6063 Perl_croak(aTHX_ "Substitution replacement not terminated");
6065 PL_multi_start = first_start; /* so whole substitution is taken together */
6067 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6073 else if (strchr("iogcmsx", *s))
6074 pmflag(&pm->op_pmflags,*s++);
6081 PL_sublex_info.super_bufptr = s;
6082 PL_sublex_info.super_bufend = PL_bufend;
6084 pm->op_pmflags |= PMf_EVAL;
6085 repl = newSVpvn("",0);
6087 sv_catpv(repl, es ? "eval " : "do ");
6088 sv_catpvn(repl, "{ ", 2);
6089 sv_catsv(repl, PL_lex_repl);
6090 sv_catpvn(repl, " };", 2);
6092 SvREFCNT_dec(PL_lex_repl);
6096 pm->op_pmpermflags = pm->op_pmflags;
6097 PL_lex_op = (OP*)pm;
6098 yylval.ival = OP_SUBST;
6103 S_scan_trans(pTHX_ char *start)
6114 yylval.ival = OP_NULL;
6116 s = scan_str(start,FALSE,FALSE);
6119 SvREFCNT_dec(PL_lex_stuff);
6120 PL_lex_stuff = Nullsv;
6121 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6123 if (s[-1] == PL_multi_open)
6126 s = scan_str(s,FALSE,FALSE);
6129 SvREFCNT_dec(PL_lex_stuff);
6130 PL_lex_stuff = Nullsv;
6132 SvREFCNT_dec(PL_lex_repl);
6133 PL_lex_repl = Nullsv;
6134 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6137 New(803,tbl,256,short);
6138 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6140 complement = del = squash = 0;
6141 while (strchr("cds", *s)) {
6143 complement = OPpTRANS_COMPLEMENT;
6145 del = OPpTRANS_DELETE;
6147 squash = OPpTRANS_SQUASH;
6150 o->op_private = del|squash|complement;
6153 yylval.ival = OP_TRANS;
6158 S_scan_heredoc(pTHX_ register char *s)
6162 I32 op_type = OP_SCALAR;
6169 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6173 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6176 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6177 if (*peek && strchr("`'\"",*peek)) {
6180 s = delimcpy(d, e, s, PL_bufend, term, &len);
6190 if (!isALNUM_lazy_if(s,UTF))
6191 deprecate("bare << to mean <<\"\"");
6192 for (; isALNUM_lazy_if(s,UTF); s++) {
6197 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6198 Perl_croak(aTHX_ "Delimiter for here document is too long");
6201 len = d - PL_tokenbuf;
6202 #ifndef PERL_STRICT_CR
6203 d = strchr(s, '\r');
6207 while (s < PL_bufend) {
6213 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6222 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6227 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6228 herewas = newSVpvn(s,PL_bufend-s);
6230 s--, herewas = newSVpvn(s,d-s);
6231 s += SvCUR(herewas);
6233 tmpstr = NEWSV(87,79);
6234 sv_upgrade(tmpstr, SVt_PVIV);
6239 else if (term == '`') {
6240 op_type = OP_BACKTICK;
6241 SvIVX(tmpstr) = '\\';
6245 PL_multi_start = CopLINE(PL_curcop);
6246 PL_multi_open = PL_multi_close = '<';
6247 term = *PL_tokenbuf;
6248 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6249 char *bufptr = PL_sublex_info.super_bufptr;
6250 char *bufend = PL_sublex_info.super_bufend;
6251 char *olds = s - SvCUR(herewas);
6252 s = strchr(bufptr, '\n');
6256 while (s < bufend &&
6257 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6259 CopLINE_inc(PL_curcop);
6262 CopLINE_set(PL_curcop, PL_multi_start);
6263 missingterm(PL_tokenbuf);
6265 sv_setpvn(herewas,bufptr,d-bufptr+1);
6266 sv_setpvn(tmpstr,d+1,s-d);
6268 sv_catpvn(herewas,s,bufend-s);
6269 (void)strcpy(bufptr,SvPVX(herewas));
6276 while (s < PL_bufend &&
6277 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6279 CopLINE_inc(PL_curcop);
6281 if (s >= PL_bufend) {
6282 CopLINE_set(PL_curcop, PL_multi_start);
6283 missingterm(PL_tokenbuf);
6285 sv_setpvn(tmpstr,d+1,s-d);
6287 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6289 sv_catpvn(herewas,s,PL_bufend-s);
6290 sv_setsv(PL_linestr,herewas);
6291 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6292 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6295 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6296 while (s >= PL_bufend) { /* multiple line string? */
6298 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6299 CopLINE_set(PL_curcop, PL_multi_start);
6300 missingterm(PL_tokenbuf);
6302 CopLINE_inc(PL_curcop);
6303 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6304 #ifndef PERL_STRICT_CR
6305 if (PL_bufend - PL_linestart >= 2) {
6306 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6307 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6309 PL_bufend[-2] = '\n';
6311 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6313 else if (PL_bufend[-1] == '\r')
6314 PL_bufend[-1] = '\n';
6316 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6317 PL_bufend[-1] = '\n';
6319 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6320 SV *sv = NEWSV(88,0);
6322 sv_upgrade(sv, SVt_PVMG);
6323 sv_setsv(sv,PL_linestr);
6324 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6326 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6329 sv_catsv(PL_linestr,herewas);
6330 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6334 sv_catsv(tmpstr,PL_linestr);
6339 PL_multi_end = CopLINE(PL_curcop);
6340 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6341 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6342 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6344 SvREFCNT_dec(herewas);
6345 PL_lex_stuff = tmpstr;
6346 yylval.ival = op_type;
6351 takes: current position in input buffer
6352 returns: new position in input buffer
6353 side-effects: yylval and lex_op are set.
6358 <FH> read from filehandle
6359 <pkg::FH> read from package qualified filehandle
6360 <pkg'FH> read from package qualified filehandle
6361 <$fh> read from filehandle in $fh
6367 S_scan_inputsymbol(pTHX_ char *start)
6369 register char *s = start; /* current position in buffer */
6375 d = PL_tokenbuf; /* start of temp holding space */
6376 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6377 end = strchr(s, '\n');
6380 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6382 /* die if we didn't have space for the contents of the <>,
6383 or if it didn't end, or if we see a newline
6386 if (len >= sizeof PL_tokenbuf)
6387 Perl_croak(aTHX_ "Excessively long <> operator");
6389 Perl_croak(aTHX_ "Unterminated <> operator");
6394 Remember, only scalar variables are interpreted as filehandles by
6395 this code. Anything more complex (e.g., <$fh{$num}>) will be
6396 treated as a glob() call.
6397 This code makes use of the fact that except for the $ at the front,
6398 a scalar variable and a filehandle look the same.
6400 if (*d == '$' && d[1]) d++;
6402 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6403 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6406 /* If we've tried to read what we allow filehandles to look like, and
6407 there's still text left, then it must be a glob() and not a getline.
6408 Use scan_str to pull out the stuff between the <> and treat it
6409 as nothing more than a string.
6412 if (d - PL_tokenbuf != len) {
6413 yylval.ival = OP_GLOB;
6415 s = scan_str(start,FALSE,FALSE);
6417 Perl_croak(aTHX_ "Glob not terminated");
6421 /* we're in a filehandle read situation */
6424 /* turn <> into <ARGV> */
6426 (void)strcpy(d,"ARGV");
6428 /* if <$fh>, create the ops to turn the variable into a
6434 /* try to find it in the pad for this block, otherwise find
6435 add symbol table ops
6437 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6438 OP *o = newOP(OP_PADSV, 0);
6440 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6443 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6444 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6445 newUNOP(OP_RV2SV, 0,
6446 newGVOP(OP_GV, 0, gv)));
6448 PL_lex_op->op_flags |= OPf_SPECIAL;
6449 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6450 yylval.ival = OP_NULL;
6453 /* If it's none of the above, it must be a literal filehandle
6454 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6456 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6457 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6458 yylval.ival = OP_NULL;
6467 takes: start position in buffer
6468 keep_quoted preserve \ on the embedded delimiter(s)
6469 keep_delims preserve the delimiters around the string
6470 returns: position to continue reading from buffer
6471 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6472 updates the read buffer.
6474 This subroutine pulls a string out of the input. It is called for:
6475 q single quotes q(literal text)
6476 ' single quotes 'literal text'
6477 qq double quotes qq(interpolate $here please)
6478 " double quotes "interpolate $here please"
6479 qx backticks qx(/bin/ls -l)
6480 ` backticks `/bin/ls -l`
6481 qw quote words @EXPORT_OK = qw( func() $spam )
6482 m// regexp match m/this/
6483 s/// regexp substitute s/this/that/
6484 tr/// string transliterate tr/this/that/
6485 y/// string transliterate y/this/that/
6486 ($*@) sub prototypes sub foo ($)
6487 (stuff) sub attr parameters sub foo : attr(stuff)
6488 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6490 In most of these cases (all but <>, patterns and transliterate)
6491 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6492 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6493 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6496 It skips whitespace before the string starts, and treats the first
6497 character as the delimiter. If the delimiter is one of ([{< then
6498 the corresponding "close" character )]}> is used as the closing
6499 delimiter. It allows quoting of delimiters, and if the string has
6500 balanced delimiters ([{<>}]) it allows nesting.
6502 The lexer always reads these strings into lex_stuff, except in the
6503 case of the operators which take *two* arguments (s/// and tr///)
6504 when it checks to see if lex_stuff is full (presumably with the 1st
6505 arg to s or tr) and if so puts the string into lex_repl.
6510 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6513 SV *sv; /* scalar value: string */
6514 char *tmps; /* temp string, used for delimiter matching */
6515 register char *s = start; /* current position in the buffer */
6516 register char term; /* terminating character */
6517 register char *to; /* current position in the sv's data */
6518 I32 brackets = 1; /* bracket nesting level */
6519 bool has_utf = FALSE; /* is there any utf8 content? */
6521 /* skip space before the delimiter */
6525 /* mark where we are, in case we need to report errors */
6528 /* after skipping whitespace, the next character is the terminator */
6530 if ((term & 0x80) && UTF)
6533 /* mark where we are */
6534 PL_multi_start = CopLINE(PL_curcop);
6535 PL_multi_open = term;
6537 /* find corresponding closing delimiter */
6538 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6540 PL_multi_close = term;
6542 /* create a new SV to hold the contents. 87 is leak category, I'm
6543 assuming. 79 is the SV's initial length. What a random number. */
6545 sv_upgrade(sv, SVt_PVIV);
6547 (void)SvPOK_only(sv); /* validate pointer */
6549 /* move past delimiter and try to read a complete string */
6551 sv_catpvn(sv, s, 1);
6554 /* extend sv if need be */
6555 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6556 /* set 'to' to the next character in the sv's string */
6557 to = SvPVX(sv)+SvCUR(sv);
6559 /* if open delimiter is the close delimiter read unbridle */
6560 if (PL_multi_open == PL_multi_close) {
6561 for (; s < PL_bufend; s++,to++) {
6562 /* embedded newlines increment the current line number */
6563 if (*s == '\n' && !PL_rsfp)
6564 CopLINE_inc(PL_curcop);
6565 /* handle quoted delimiters */
6566 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6567 if (!keep_quoted && s[1] == term)
6569 /* any other quotes are simply copied straight through */
6573 /* terminate when run out of buffer (the for() condition), or
6574 have found the terminator */
6575 else if (*s == term)
6577 else if (!has_utf && (*s & 0x80) && UTF)
6583 /* if the terminator isn't the same as the start character (e.g.,
6584 matched brackets), we have to allow more in the quoting, and
6585 be prepared for nested brackets.
6588 /* read until we run out of string, or we find the terminator */
6589 for (; s < PL_bufend; s++,to++) {
6590 /* embedded newlines increment the line count */
6591 if (*s == '\n' && !PL_rsfp)
6592 CopLINE_inc(PL_curcop);
6593 /* backslashes can escape the open or closing characters */
6594 if (*s == '\\' && s+1 < PL_bufend) {
6596 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6601 /* allow nested opens and closes */
6602 else if (*s == PL_multi_close && --brackets <= 0)
6604 else if (*s == PL_multi_open)
6606 else if (!has_utf && (*s & 0x80) && UTF)
6611 /* terminate the copied string and update the sv's end-of-string */
6613 SvCUR_set(sv, to - SvPVX(sv));
6616 * this next chunk reads more into the buffer if we're not done yet
6620 break; /* handle case where we are done yet :-) */
6622 #ifndef PERL_STRICT_CR
6623 if (to - SvPVX(sv) >= 2) {
6624 if ((to[-2] == '\r' && to[-1] == '\n') ||
6625 (to[-2] == '\n' && to[-1] == '\r'))
6629 SvCUR_set(sv, to - SvPVX(sv));
6631 else if (to[-1] == '\r')
6634 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6638 /* if we're out of file, or a read fails, bail and reset the current
6639 line marker so we can report where the unterminated string began
6642 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6644 CopLINE_set(PL_curcop, PL_multi_start);
6647 /* we read a line, so increment our line counter */
6648 CopLINE_inc(PL_curcop);
6650 /* update debugger info */
6651 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6652 SV *sv = NEWSV(88,0);
6654 sv_upgrade(sv, SVt_PVMG);
6655 sv_setsv(sv,PL_linestr);
6656 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6659 /* having changed the buffer, we must update PL_bufend */
6660 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6663 /* at this point, we have successfully read the delimited string */
6666 sv_catpvn(sv, s, 1);
6669 PL_multi_end = CopLINE(PL_curcop);
6672 /* if we allocated too much space, give some back */
6673 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6674 SvLEN_set(sv, SvCUR(sv) + 1);
6675 Renew(SvPVX(sv), SvLEN(sv), char);
6678 /* decide whether this is the first or second quoted string we've read
6691 takes: pointer to position in buffer
6692 returns: pointer to new position in buffer
6693 side-effects: builds ops for the constant in yylval.op
6695 Read a number in any of the formats that Perl accepts:
6697 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6698 [\d_]+(\.[\d_]*)?[Ee](\d+)
6700 Underbars (_) are allowed in decimal numbers. If -w is on,
6701 underbars before a decimal point must be at three digit intervals.
6703 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6706 If it reads a number without a decimal point or an exponent, it will
6707 try converting the number to an integer and see if it can do so
6708 without loss of precision.
6712 Perl_scan_num(pTHX_ char *start)
6714 register char *s = start; /* current position in buffer */
6715 register char *d; /* destination in temp buffer */
6716 register char *e; /* end of temp buffer */
6717 NV nv; /* number read, as a double */
6718 SV *sv = Nullsv; /* place to put the converted number */
6719 bool floatit; /* boolean: int or float? */
6720 char *lastub = 0; /* position of last underbar */
6721 static char number_too_long[] = "Number too long";
6723 /* We use the first character to decide what type of number this is */
6727 Perl_croak(aTHX_ "panic: scan_num");
6729 /* if it starts with a 0, it could be an octal number, a decimal in
6730 0.13 disguise, or a hexadecimal number, or a binary number. */
6734 u holds the "number so far"
6735 shift the power of 2 of the base
6736 (hex == 4, octal == 3, binary == 1)
6737 overflowed was the number more than we can hold?
6739 Shift is used when we add a digit. It also serves as an "are
6740 we in octal/hex/binary?" indicator to disallow hex characters
6747 bool overflowed = FALSE;
6748 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6749 static char* bases[5] = { "", "binary", "", "octal",
6751 static char* Bases[5] = { "", "Binary", "", "Octal",
6753 static char *maxima[5] = { "",
6754 "0b11111111111111111111111111111111",
6758 char *base, *Base, *max;
6764 } else if (s[1] == 'b') {
6768 /* check for a decimal in disguise */
6769 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6771 /* so it must be octal */
6775 base = bases[shift];
6776 Base = Bases[shift];
6777 max = maxima[shift];
6779 /* read the rest of the number */
6781 /* x is used in the overflow test,
6782 b is the digit we're adding on. */
6787 /* if we don't mention it, we're done */
6796 /* 8 and 9 are not octal */
6799 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6803 case '2': case '3': case '4':
6804 case '5': case '6': case '7':
6806 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6810 b = *s++ & 15; /* ASCII digit -> value of digit */
6814 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6815 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6816 /* make sure they said 0x */
6821 /* Prepare to put the digit we have onto the end
6822 of the number so far. We check for overflows.
6827 x = u << shift; /* make room for the digit */
6829 if ((x >> shift) != u
6830 && !(PL_hints & HINT_NEW_BINARY)) {
6834 if (ckWARN_d(WARN_OVERFLOW))
6835 Perl_warner(aTHX_ WARN_OVERFLOW,
6836 "Integer overflow in %s number",
6839 u = x | b; /* add the digit to the end */
6842 n *= nvshift[shift];
6843 /* If an NV has not enough bits in its
6844 * mantissa to represent an UV this summing of
6845 * small low-order numbers is a waste of time
6846 * (because the NV cannot preserve the
6847 * low-order bits anyway): we could just
6848 * remember when did we overflow and in the
6849 * end just multiply n by the right
6857 /* if we get here, we had success: make a scalar value from
6864 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6865 Perl_warner(aTHX_ WARN_PORTABLE,
6866 "%s number > %s non-portable",
6873 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6874 Perl_warner(aTHX_ WARN_PORTABLE,
6875 "%s number > %s non-portable",
6880 if (PL_hints & HINT_NEW_BINARY)
6881 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6886 handle decimal numbers.
6887 we're also sent here when we read a 0 as the first digit
6889 case '1': case '2': case '3': case '4': case '5':
6890 case '6': case '7': case '8': case '9': case '.':
6893 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
6896 /* read next group of digits and _ and copy into d */
6897 while (isDIGIT(*s) || *s == '_') {
6898 /* skip underscores, checking for misplaced ones
6902 dTHR; /* only for ckWARN */
6903 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
6904 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6908 /* check for end of fixed-length buffer */
6910 Perl_croak(aTHX_ number_too_long);
6911 /* if we're ok, copy the character */
6916 /* final misplaced underbar check */
6917 if (lastub && s - lastub != 3) {
6919 if (ckWARN(WARN_SYNTAX))
6920 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6923 /* read a decimal portion if there is one. avoid
6924 3..5 being interpreted as the number 3. followed
6927 if (*s == '.' && s[1] != '.') {
6931 /* copy, ignoring underbars, until we run out of
6932 digits. Note: no misplaced underbar checks!
6934 for (; isDIGIT(*s) || *s == '_'; s++) {
6935 /* fixed length buffer check */
6937 Perl_croak(aTHX_ number_too_long);
6941 if (*s == '.' && isDIGIT(s[1])) {
6942 /* oops, it's really a v-string, but without the "v" */
6948 /* read exponent part, if present */
6949 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
6953 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
6954 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
6956 /* allow positive or negative exponent */
6957 if (*s == '+' || *s == '-')
6960 /* read digits of exponent (no underbars :-) */
6961 while (isDIGIT(*s)) {
6963 Perl_croak(aTHX_ number_too_long);
6968 /* terminate the string */
6971 /* make an sv from the string */
6974 #if defined(Strtol) && defined(Strtoul)
6977 strtol/strtoll sets errno to ERANGE if the number is too big
6978 for an integer. We try to do an integer conversion first
6979 if no characters indicating "float" have been found.
6986 if (*PL_tokenbuf == '-')
6987 iv = Strtol(PL_tokenbuf, (char**)NULL, 10);
6989 uv = Strtoul(PL_tokenbuf, (char**)NULL, 10);
6991 floatit = TRUE; /* Probably just too large. */
6992 else if (*PL_tokenbuf == '-')
6994 else if (uv <= IV_MAX)
6995 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7000 nv = Atof(PL_tokenbuf);
7005 No working strtou?ll?.
7007 Unfortunately atol() doesn't do range checks (returning
7008 LONG_MIN/LONG_MAX, and setting errno to ERANGE on overflows)
7009 everywhere [1], so we cannot use use atol() (or atoll()).
7010 If we could, they would be used, as Atol(), very much like
7011 Strtol() and Strtoul() are used above.
7013 [1] XXX Configure test needed to check for atol()
7014 (and atoll()) overflow behaviour XXX
7018 We need to do this the hard way. */
7020 nv = Atof(PL_tokenbuf);
7022 /* See if we can make do with an integer value without loss of
7023 precision. We use U_V to cast to a UV, because some
7024 compilers have issues. Then we try casting it back and see
7025 if it was the same [1]. We only do this if we know we
7026 specifically read an integer. If floatit is true, then we
7027 don't need to do the conversion at all.
7029 [1] Note that this is lossy if our NVs cannot preserve our
7030 UVs. There are metaconfig defines NV_PRESERVES_UV (a boolean)
7031 and NV_PRESERVES_UV_BITS (a number), but in general we really
7032 do hope all such potentially lossy platforms have strtou?ll?
7033 to do a lossless IV/UV conversion.
7035 Maybe could do some tricks with DBL_DIG, LDBL_DIG and
7036 DBL_MANT_DIG and LDBL_MANT_DIG (these are already available
7037 as NV_DIG and NV_MANT_DIG)?
7043 if (!floatit && (NV)uv == nv) {
7045 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7053 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7054 (PL_hints & HINT_NEW_INTEGER) )
7055 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7056 (floatit ? "float" : "integer"),
7060 /* if it starts with a v, it could be a v-string */
7066 while (isDIGIT(*pos) || *pos == '_')
7068 if (!isALPHA(*pos)) {
7070 U8 tmpbuf[UTF8_MAXLEN];
7073 s++; /* get past 'v' */
7076 sv_setpvn(sv, "", 0);
7079 if (*s == '0' && isDIGIT(s[1]))
7080 yyerror("Octal number in vector unsupported");
7083 /* this is atoi() that tolerates underscores */
7086 while (--end >= s) {
7091 rev += (*end - '0') * mult;
7093 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
7094 Perl_warner(aTHX_ WARN_OVERFLOW,
7095 "Integer overflow in decimal number");
7098 tmpend = uv_to_utf8(tmpbuf, rev);
7099 utf8 = utf8 || rev > 127;
7100 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
7101 if (*pos == '.' && isDIGIT(pos[1]))
7107 while (isDIGIT(*pos) || *pos == '_')
7115 sv_utf8_downgrade(sv, TRUE);
7122 /* make the op for the constant and return */
7125 yylval.opval = newSVOP(OP_CONST, 0, sv);
7127 yylval.opval = Nullop;
7133 S_scan_formline(pTHX_ register char *s)
7138 SV *stuff = newSVpvn("",0);
7139 bool needargs = FALSE;
7142 if (*s == '.' || *s == /*{*/'}') {
7144 #ifdef PERL_STRICT_CR
7145 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7147 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7149 if (*t == '\n' || t == PL_bufend)
7152 if (PL_in_eval && !PL_rsfp) {
7153 eol = strchr(s,'\n');
7158 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7160 for (t = s; t < eol; t++) {
7161 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7163 goto enough; /* ~~ must be first line in formline */
7165 if (*t == '@' || *t == '^')
7168 sv_catpvn(stuff, s, eol-s);
7169 #ifndef PERL_STRICT_CR
7170 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7171 char *end = SvPVX(stuff) + SvCUR(stuff);
7180 s = filter_gets(PL_linestr, PL_rsfp, 0);
7181 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7182 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7185 yyerror("Format not terminated");
7195 PL_lex_state = LEX_NORMAL;
7196 PL_nextval[PL_nexttoke].ival = 0;
7200 PL_lex_state = LEX_FORMLINE;
7201 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7203 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7207 SvREFCNT_dec(stuff);
7208 PL_lex_formbrack = 0;
7219 PL_cshlen = strlen(PL_cshname);
7224 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7227 I32 oldsavestack_ix = PL_savestack_ix;
7228 CV* outsidecv = PL_compcv;
7232 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7234 SAVEI32(PL_subline);
7235 save_item(PL_subname);
7238 SAVESPTR(PL_comppad_name);
7239 SAVESPTR(PL_compcv);
7240 SAVEI32(PL_comppad_name_fill);
7241 SAVEI32(PL_min_intro_pending);
7242 SAVEI32(PL_max_intro_pending);
7243 SAVEI32(PL_pad_reset_pending);
7245 PL_compcv = (CV*)NEWSV(1104,0);
7246 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7247 CvFLAGS(PL_compcv) |= flags;
7249 PL_comppad = newAV();
7250 av_push(PL_comppad, Nullsv);
7251 PL_curpad = AvARRAY(PL_comppad);
7252 PL_comppad_name = newAV();
7253 PL_comppad_name_fill = 0;
7254 PL_min_intro_pending = 0;
7256 PL_subline = CopLINE(PL_curcop);
7258 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7259 PL_curpad[0] = (SV*)newAV();
7260 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7261 #endif /* USE_THREADS */
7263 comppadlist = newAV();
7264 AvREAL_off(comppadlist);
7265 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7266 av_store(comppadlist, 1, (SV*)PL_comppad);
7268 CvPADLIST(PL_compcv) = comppadlist;
7269 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7271 CvOWNER(PL_compcv) = 0;
7272 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7273 MUTEX_INIT(CvMUTEXP(PL_compcv));
7274 #endif /* USE_THREADS */
7276 return oldsavestack_ix;
7280 Perl_yywarn(pTHX_ char *s)
7283 PL_in_eval |= EVAL_WARNONLY;
7285 PL_in_eval &= ~EVAL_WARNONLY;
7290 Perl_yyerror(pTHX_ char *s)
7294 char *context = NULL;
7298 if (!yychar || (yychar == ';' && !PL_rsfp))
7300 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7301 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7302 while (isSPACE(*PL_oldoldbufptr))
7304 context = PL_oldoldbufptr;
7305 contlen = PL_bufptr - PL_oldoldbufptr;
7307 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7308 PL_oldbufptr != PL_bufptr) {
7309 while (isSPACE(*PL_oldbufptr))
7311 context = PL_oldbufptr;
7312 contlen = PL_bufptr - PL_oldbufptr;
7314 else if (yychar > 255)
7315 where = "next token ???";
7316 #ifdef USE_PURE_BISON
7317 /* GNU Bison sets the value -2 */
7318 else if (yychar == -2) {
7320 else if ((yychar & 127) == 127) {
7322 if (PL_lex_state == LEX_NORMAL ||
7323 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7324 where = "at end of line";
7325 else if (PL_lex_inpat)
7326 where = "within pattern";
7328 where = "within string";
7331 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7333 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7334 else if (isPRINT_LC(yychar))
7335 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7337 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7338 where = SvPVX(where_sv);
7340 msg = sv_2mortal(newSVpv(s, 0));
7341 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7342 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7344 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7346 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7347 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7348 Perl_sv_catpvf(aTHX_ msg,
7349 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7350 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7353 if (PL_in_eval & EVAL_WARNONLY)
7354 Perl_warn(aTHX_ "%"SVf, msg);
7357 if (PL_error_count >= 10) {
7358 if (PL_in_eval && SvCUR(ERRSV))
7359 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
7360 ERRSV, CopFILE(PL_curcop));
7362 Perl_croak(aTHX_ "%s has too many errors.\n",
7363 CopFILE(PL_curcop));
7366 PL_in_my_stash = Nullhv;
7371 S_swallow_bom(pTHX_ U8 *s)
7374 slen = SvCUR(PL_linestr);
7378 /* UTF-16 little-endian */
7379 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
7380 Perl_croak(aTHX_ "Unsupported script encoding");
7381 #ifndef PERL_NO_UTF16_FILTER
7382 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-LE script encoding\n"));
7384 if (PL_bufend > (char*)s) {
7388 filter_add(utf16rev_textfilter, NULL);
7389 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7390 PL_bufend = (char*)utf16_to_utf8_reversed(s, news,
7391 PL_bufend - (char*)s - 1,
7393 Copy(news, s, newlen, U8);
7394 SvCUR_set(PL_linestr, newlen);
7395 PL_bufend = SvPVX(PL_linestr) + newlen;
7396 news[newlen++] = '\0';
7400 Perl_croak(aTHX_ "Unsupported script encoding");
7405 if (s[1] == 0xFF) { /* UTF-16 big-endian */
7406 #ifndef PERL_NO_UTF16_FILTER
7407 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding\n"));
7409 if (PL_bufend > (char *)s) {
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(s, news,
7416 PL_bufend - (char*)s,
7418 Copy(news, s, newlen, U8);
7419 SvCUR_set(PL_linestr, newlen);
7420 PL_bufend = SvPVX(PL_linestr) + newlen;
7421 news[newlen++] = '\0';
7425 Perl_croak(aTHX_ "Unsupported script encoding");
7430 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
7431 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-8 script encoding\n"));
7436 if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */
7437 s[2] == 0xFE && s[3] == 0xFF)
7439 Perl_croak(aTHX_ "Unsupported script encoding");
7451 * Restore a source filter.
7455 restore_rsfp(pTHXo_ void *f)
7457 PerlIO *fp = (PerlIO*)f;
7459 if (PL_rsfp == PerlIO_stdin())
7460 PerlIO_clearerr(PL_rsfp);
7461 else if (PL_rsfp && (PL_rsfp != fp))
7462 PerlIO_close(PL_rsfp);
7466 #ifndef PERL_NO_UTF16_FILTER
7468 utf16_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7470 I32 count = FILTER_READ(idx+1, sv, maxlen);
7475 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7476 if (!*SvPV_nolen(sv))
7477 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7480 tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7481 sv_usepvn(sv, (char*)tmps, tend - tmps);
7487 utf16rev_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7489 I32 count = FILTER_READ(idx+1, sv, maxlen);
7494 if (!*SvPV_nolen(sv))
7495 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7498 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7499 tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7500 sv_usepvn(sv, (char*)tmps, tend - tmps);