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);
361 SAVECOPLINE(PL_curcop);
364 SAVEPPTR(PL_oldbufptr);
365 SAVEPPTR(PL_oldoldbufptr);
366 SAVEPPTR(PL_linestart);
367 SAVESPTR(PL_linestr);
368 SAVEPPTR(PL_lex_brackstack);
369 SAVEPPTR(PL_lex_casestack);
370 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
371 SAVESPTR(PL_lex_stuff);
372 SAVEI32(PL_lex_defer);
373 SAVEI32(PL_sublex_info.sub_inwhat);
374 SAVESPTR(PL_lex_repl);
376 SAVEINT(PL_lex_expect);
378 PL_lex_state = LEX_NORMAL;
382 New(899, PL_lex_brackstack, 120, char);
383 New(899, PL_lex_casestack, 12, char);
384 SAVEFREEPV(PL_lex_brackstack);
385 SAVEFREEPV(PL_lex_casestack);
387 *PL_lex_casestack = '\0';
390 PL_lex_stuff = Nullsv;
391 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_chk((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_chk((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);
1493 sv_utf8_upgrade(sv);
1494 /* this just broke our allocation above... */
1495 SvGROW(sv, send - start);
1496 d = SvPVX(sv) + SvCUR(sv);
1499 if (len > e - s + 4) {
1500 char *odest = SvPVX(sv);
1502 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1503 d = SvPVX(sv) + (d - odest);
1505 Copy(str, d, len, char);
1512 yyerror("Missing braces on \\N{}");
1515 /* \c is a control character */
1530 /* printf-style backslashes, formfeeds, newlines, etc */
1548 *d++ = '\047'; /* CP 1047 */
1551 *d++ = '\057'; /* CP 1047 */
1565 } /* end if (backslash) */
1568 } /* while loop to process each character */
1570 /* terminate the string and set up the sv */
1572 SvCUR_set(sv, d - SvPVX(sv));
1577 /* shrink the sv if we allocated more than we used */
1578 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1579 SvLEN_set(sv, SvCUR(sv) + 1);
1580 Renew(SvPVX(sv), SvLEN(sv), char);
1583 /* return the substring (via yylval) only if we parsed anything */
1584 if (s > PL_bufptr) {
1585 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1586 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1588 ( PL_lex_inwhat == OP_TRANS
1590 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1593 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1600 * Returns TRUE if there's more to the expression (e.g., a subscript),
1603 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1605 * ->[ and ->{ return TRUE
1606 * { and [ outside a pattern are always subscripts, so return TRUE
1607 * if we're outside a pattern and it's not { or [, then return FALSE
1608 * if we're in a pattern and the first char is a {
1609 * {4,5} (any digits around the comma) returns FALSE
1610 * if we're in a pattern and the first char is a [
1612 * [SOMETHING] has a funky algorithm to decide whether it's a
1613 * character class or not. It has to deal with things like
1614 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1615 * anything else returns TRUE
1618 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1621 S_intuit_more(pTHX_ register char *s)
1623 if (PL_lex_brackets)
1625 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1627 if (*s != '{' && *s != '[')
1632 /* In a pattern, so maybe we have {n,m}. */
1649 /* On the other hand, maybe we have a character class */
1652 if (*s == ']' || *s == '^')
1655 /* this is terrifying, and it works */
1656 int weight = 2; /* let's weigh the evidence */
1658 unsigned char un_char = 255, last_un_char;
1659 char *send = strchr(s,']');
1660 char tmpbuf[sizeof PL_tokenbuf * 4];
1662 if (!send) /* has to be an expression */
1665 Zero(seen,256,char);
1668 else if (isDIGIT(*s)) {
1670 if (isDIGIT(s[1]) && s[2] == ']')
1676 for (; s < send; s++) {
1677 last_un_char = un_char;
1678 un_char = (unsigned char)*s;
1683 weight -= seen[un_char] * 10;
1684 if (isALNUM_lazy_if(s+1,UTF)) {
1685 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1686 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1691 else if (*s == '$' && s[1] &&
1692 strchr("[#!%*<>()-=",s[1])) {
1693 if (/*{*/ strchr("])} =",s[2]))
1702 if (strchr("wds]",s[1]))
1704 else if (seen['\''] || seen['"'])
1706 else if (strchr("rnftbxcav",s[1]))
1708 else if (isDIGIT(s[1])) {
1710 while (s[1] && isDIGIT(s[1]))
1720 if (strchr("aA01! ",last_un_char))
1722 if (strchr("zZ79~",s[1]))
1724 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1725 weight -= 5; /* cope with negative subscript */
1728 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1729 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1734 if (keyword(tmpbuf, d - tmpbuf))
1737 if (un_char == last_un_char + 1)
1739 weight -= seen[un_char];
1744 if (weight >= 0) /* probably a character class */
1754 * Does all the checking to disambiguate
1756 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1757 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1759 * First argument is the stuff after the first token, e.g. "bar".
1761 * Not a method if bar is a filehandle.
1762 * Not a method if foo is a subroutine prototyped to take a filehandle.
1763 * Not a method if it's really "Foo $bar"
1764 * Method if it's "foo $bar"
1765 * Not a method if it's really "print foo $bar"
1766 * Method if it's really "foo package::" (interpreted as package->foo)
1767 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1768 * Not a method if bar is a filehandle or package, but is quoted with
1773 S_intuit_method(pTHX_ char *start, GV *gv)
1775 char *s = start + (*start == '$');
1776 char tmpbuf[sizeof PL_tokenbuf];
1784 if ((cv = GvCVu(gv))) {
1785 char *proto = SvPVX(cv);
1795 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1796 /* start is the beginning of the possible filehandle/object,
1797 * and s is the end of it
1798 * tmpbuf is a copy of it
1801 if (*start == '$') {
1802 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1807 return *s == '(' ? FUNCMETH : METHOD;
1809 if (!keyword(tmpbuf, len)) {
1810 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1815 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1816 if (indirgv && GvCVu(indirgv))
1818 /* filehandle or package name makes it a method */
1819 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1821 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1822 return 0; /* no assumptions -- "=>" quotes bearword */
1824 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1825 newSVpvn(tmpbuf,len));
1826 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1830 return *s == '(' ? FUNCMETH : METHOD;
1838 * Return a string of Perl code to load the debugger. If PERL5DB
1839 * is set, it will return the contents of that, otherwise a
1840 * compile-time require of perl5db.pl.
1847 char *pdb = PerlEnv_getenv("PERL5DB");
1851 SETERRNO(0,SS$_NORMAL);
1852 return "BEGIN { require 'perl5db.pl' }";
1858 /* Encoded script support. filter_add() effectively inserts a
1859 * 'pre-processing' function into the current source input stream.
1860 * Note that the filter function only applies to the current source file
1861 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1863 * The datasv parameter (which may be NULL) can be used to pass
1864 * private data to this instance of the filter. The filter function
1865 * can recover the SV using the FILTER_DATA macro and use it to
1866 * store private buffers and state information.
1868 * The supplied datasv parameter is upgraded to a PVIO type
1869 * and the IoDIRP/IoANY field is used to store the function pointer,
1870 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1871 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1872 * private use must be set using malloc'd pointers.
1876 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1881 if (!PL_rsfp_filters)
1882 PL_rsfp_filters = newAV();
1884 datasv = NEWSV(255,0);
1885 if (!SvUPGRADE(datasv, SVt_PVIO))
1886 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1887 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
1888 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1889 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1890 funcp, SvPV_nolen(datasv)));
1891 av_unshift(PL_rsfp_filters, 1);
1892 av_store(PL_rsfp_filters, 0, datasv) ;
1897 /* Delete most recently added instance of this filter function. */
1899 Perl_filter_del(pTHX_ filter_t funcp)
1902 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1903 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1905 /* if filter is on top of stack (usual case) just pop it off */
1906 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1907 if (IoANY(datasv) == (void *)funcp) {
1908 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1909 IoANY(datasv) = (void *)NULL;
1910 sv_free(av_pop(PL_rsfp_filters));
1914 /* we need to search for the correct entry and clear it */
1915 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1919 /* Invoke the n'th filter function for the current rsfp. */
1921 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1924 /* 0 = read one text line */
1929 if (!PL_rsfp_filters)
1931 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1932 /* Provide a default input filter to make life easy. */
1933 /* Note that we append to the line. This is handy. */
1934 DEBUG_P(PerlIO_printf(Perl_debug_log,
1935 "filter_read %d: from rsfp\n", idx));
1939 int old_len = SvCUR(buf_sv) ;
1941 /* ensure buf_sv is large enough */
1942 SvGROW(buf_sv, old_len + maxlen) ;
1943 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1944 if (PerlIO_error(PL_rsfp))
1945 return -1; /* error */
1947 return 0 ; /* end of file */
1949 SvCUR_set(buf_sv, old_len + len) ;
1952 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1953 if (PerlIO_error(PL_rsfp))
1954 return -1; /* error */
1956 return 0 ; /* end of file */
1959 return SvCUR(buf_sv);
1961 /* Skip this filter slot if filter has been deleted */
1962 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1963 DEBUG_P(PerlIO_printf(Perl_debug_log,
1964 "filter_read %d: skipped (filter deleted)\n",
1966 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1968 /* Get function pointer hidden within datasv */
1969 funcp = (filter_t)IoANY(datasv);
1970 DEBUG_P(PerlIO_printf(Perl_debug_log,
1971 "filter_read %d: via function %p (%s)\n",
1972 idx, funcp, SvPV_nolen(datasv)));
1973 /* Call function. The function is expected to */
1974 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1975 /* Return: <0:error, =0:eof, >0:not eof */
1976 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1980 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1982 #ifdef PERL_CR_FILTER
1983 if (!PL_rsfp_filters) {
1984 filter_add(S_cr_textfilter,NULL);
1987 if (PL_rsfp_filters) {
1990 SvCUR_set(sv, 0); /* start with empty line */
1991 if (FILTER_READ(0, sv, 0) > 0)
1992 return ( SvPVX(sv) ) ;
1997 return (sv_gets(sv, fp, append));
2001 S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2005 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2009 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2010 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2012 return GvHV(gv); /* Foo:: */
2015 /* use constant CLASS => 'MyClass' */
2016 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2018 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2019 pkgname = SvPV_nolen(sv);
2023 return gv_stashpv(pkgname, FALSE);
2027 static char* exp_name[] =
2028 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2029 "ATTRTERM", "TERMBLOCK"
2036 Works out what to call the token just pulled out of the input
2037 stream. The yacc parser takes care of taking the ops we return and
2038 stitching them into a tree.
2044 if read an identifier
2045 if we're in a my declaration
2046 croak if they tried to say my($foo::bar)
2047 build the ops for a my() declaration
2048 if it's an access to a my() variable
2049 are we in a sort block?
2050 croak if my($a); $a <=> $b
2051 build ops for access to a my() variable
2052 if in a dq string, and they've said @foo and we can't find @foo
2054 build ops for a bareword
2055 if we already built the token before, use it.
2059 #pragma segment Perl_yylex
2062 #ifdef USE_PURE_BISON
2063 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
2076 #ifdef USE_PURE_BISON
2077 yylval_pointer = lvalp;
2078 yychar_pointer = lcharp;
2081 /* check if there's an identifier for us to look at */
2082 if (PL_pending_ident) {
2083 /* pit holds the identifier we read and pending_ident is reset */
2084 char pit = PL_pending_ident;
2085 PL_pending_ident = 0;
2087 /* if we're in a my(), we can't allow dynamics here.
2088 $foo'bar has already been turned into $foo::bar, so
2089 just check for colons.
2091 if it's a legal name, the OP is a PADANY.
2094 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
2095 if (strchr(PL_tokenbuf,':'))
2096 yyerror(Perl_form(aTHX_ "No package name allowed for "
2097 "variable %s in \"our\"",
2099 tmp = pad_allocmy(PL_tokenbuf);
2102 if (strchr(PL_tokenbuf,':'))
2103 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
2105 yylval.opval = newOP(OP_PADANY, 0);
2106 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
2112 build the ops for accesses to a my() variable.
2114 Deny my($a) or my($b) in a sort block, *if* $a or $b is
2115 then used in a comparison. This catches most, but not
2116 all cases. For instance, it catches
2117 sort { my($a); $a <=> $b }
2119 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2120 (although why you'd do that is anyone's guess).
2123 if (!strchr(PL_tokenbuf,':')) {
2125 /* Check for single character per-thread SVs */
2126 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2127 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2128 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2130 yylval.opval = newOP(OP_THREADSV, 0);
2131 yylval.opval->op_targ = tmp;
2134 #endif /* USE_THREADS */
2135 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2136 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2137 /* might be an "our" variable" */
2138 if (SvFLAGS(namesv) & SVpad_OUR) {
2139 /* build ops for a bareword */
2140 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2141 sv_catpvn(sym, "::", 2);
2142 sv_catpv(sym, PL_tokenbuf+1);
2143 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2144 yylval.opval->op_private = OPpCONST_ENTERED;
2145 gv_fetchpv(SvPVX(sym),
2147 ? (GV_ADDMULTI | GV_ADDINEVAL)
2150 ((PL_tokenbuf[0] == '$') ? SVt_PV
2151 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2156 /* if it's a sort block and they're naming $a or $b */
2157 if (PL_last_lop_op == OP_SORT &&
2158 PL_tokenbuf[0] == '$' &&
2159 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2162 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2163 d < PL_bufend && *d != '\n';
2166 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2167 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2173 yylval.opval = newOP(OP_PADANY, 0);
2174 yylval.opval->op_targ = tmp;
2180 Whine if they've said @foo in a doublequoted string,
2181 and @foo isn't a variable we can find in the symbol
2184 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2185 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2186 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2187 && ckWARN(WARN_AMBIGUOUS))
2189 /* Downgraded from fatal to warning 20000522 mjd */
2190 Perl_warner(aTHX_ WARN_AMBIGUOUS,
2191 "Possible unintended interpolation of %s in string",
2196 /* build ops for a bareword */
2197 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2198 yylval.opval->op_private = OPpCONST_ENTERED;
2199 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2200 ((PL_tokenbuf[0] == '$') ? SVt_PV
2201 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2206 /* no identifier pending identification */
2208 switch (PL_lex_state) {
2210 case LEX_NORMAL: /* Some compilers will produce faster */
2211 case LEX_INTERPNORMAL: /* code if we comment these out. */
2215 /* when we've already built the next token, just pull it out of the queue */
2218 yylval = PL_nextval[PL_nexttoke];
2220 PL_lex_state = PL_lex_defer;
2221 PL_expect = PL_lex_expect;
2222 PL_lex_defer = LEX_NORMAL;
2224 return(PL_nexttype[PL_nexttoke]);
2226 /* interpolated case modifiers like \L \U, including \Q and \E.
2227 when we get here, PL_bufptr is at the \
2229 case LEX_INTERPCASEMOD:
2231 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2232 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2234 /* handle \E or end of string */
2235 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2239 if (PL_lex_casemods) {
2240 oldmod = PL_lex_casestack[--PL_lex_casemods];
2241 PL_lex_casestack[PL_lex_casemods] = '\0';
2243 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2245 PL_lex_state = LEX_INTERPCONCAT;
2249 if (PL_bufptr != PL_bufend)
2251 PL_lex_state = LEX_INTERPCONCAT;
2256 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2257 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2258 if (strchr("LU", *s) &&
2259 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2261 PL_lex_casestack[--PL_lex_casemods] = '\0';
2264 if (PL_lex_casemods > 10) {
2265 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2266 if (newlb != PL_lex_casestack) {
2268 PL_lex_casestack = newlb;
2271 PL_lex_casestack[PL_lex_casemods++] = *s;
2272 PL_lex_casestack[PL_lex_casemods] = '\0';
2273 PL_lex_state = LEX_INTERPCONCAT;
2274 PL_nextval[PL_nexttoke].ival = 0;
2277 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2279 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2281 PL_nextval[PL_nexttoke].ival = OP_LC;
2283 PL_nextval[PL_nexttoke].ival = OP_UC;
2285 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2287 Perl_croak(aTHX_ "panic: yylex");
2290 if (PL_lex_starts) {
2299 case LEX_INTERPPUSH:
2300 return sublex_push();
2302 case LEX_INTERPSTART:
2303 if (PL_bufptr == PL_bufend)
2304 return sublex_done();
2306 PL_lex_dojoin = (*PL_bufptr == '@');
2307 PL_lex_state = LEX_INTERPNORMAL;
2308 if (PL_lex_dojoin) {
2309 PL_nextval[PL_nexttoke].ival = 0;
2312 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2313 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2314 force_next(PRIVATEREF);
2316 force_ident("\"", '$');
2317 #endif /* USE_THREADS */
2318 PL_nextval[PL_nexttoke].ival = 0;
2320 PL_nextval[PL_nexttoke].ival = 0;
2322 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2325 if (PL_lex_starts++) {
2331 case LEX_INTERPENDMAYBE:
2332 if (intuit_more(PL_bufptr)) {
2333 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2339 if (PL_lex_dojoin) {
2340 PL_lex_dojoin = FALSE;
2341 PL_lex_state = LEX_INTERPCONCAT;
2344 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2345 && SvEVALED(PL_lex_repl))
2347 if (PL_bufptr != PL_bufend)
2348 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2349 PL_lex_repl = Nullsv;
2352 case LEX_INTERPCONCAT:
2354 if (PL_lex_brackets)
2355 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2357 if (PL_bufptr == PL_bufend)
2358 return sublex_done();
2360 if (SvIVX(PL_linestr) == '\'') {
2361 SV *sv = newSVsv(PL_linestr);
2364 else if ( PL_hints & HINT_NEW_RE )
2365 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2366 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2370 s = scan_const(PL_bufptr);
2372 PL_lex_state = LEX_INTERPCASEMOD;
2374 PL_lex_state = LEX_INTERPSTART;
2377 if (s != PL_bufptr) {
2378 PL_nextval[PL_nexttoke] = yylval;
2381 if (PL_lex_starts++)
2391 PL_lex_state = LEX_NORMAL;
2392 s = scan_formline(PL_bufptr);
2393 if (!PL_lex_formbrack)
2399 PL_oldoldbufptr = PL_oldbufptr;
2402 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2403 exp_name[PL_expect], s);
2409 if (isIDFIRST_lazy_if(s,UTF))
2411 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2414 goto fake_eof; /* emulate EOF on ^D or ^Z */
2419 if (PL_lex_brackets)
2420 yyerror("Missing right curly or square bracket");
2423 if (s++ < PL_bufend)
2424 goto retry; /* ignore stray nulls */
2427 if (!PL_in_eval && !PL_preambled) {
2428 PL_preambled = TRUE;
2429 sv_setpv(PL_linestr,incl_perldb());
2430 if (SvCUR(PL_linestr))
2431 sv_catpv(PL_linestr,";");
2433 while(AvFILLp(PL_preambleav) >= 0) {
2434 SV *tmpsv = av_shift(PL_preambleav);
2435 sv_catsv(PL_linestr, tmpsv);
2436 sv_catpv(PL_linestr, ";");
2439 sv_free((SV*)PL_preambleav);
2440 PL_preambleav = NULL;
2442 if (PL_minus_n || PL_minus_p) {
2443 sv_catpv(PL_linestr, "LINE: while (<>) {");
2445 sv_catpv(PL_linestr,"chomp;");
2447 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2449 GvIMPORTED_AV_on(gv);
2451 if (strchr("/'\"", *PL_splitstr)
2452 && strchr(PL_splitstr + 1, *PL_splitstr))
2453 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2456 s = "'~#\200\1'"; /* surely one char is unused...*/
2457 while (s[1] && strchr(PL_splitstr, *s)) s++;
2459 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2460 "q" + (delim == '\''), delim);
2461 for (s = PL_splitstr; *s; s++) {
2463 sv_catpvn(PL_linestr, "\\", 1);
2464 sv_catpvn(PL_linestr, s, 1);
2466 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2470 sv_catpv(PL_linestr,"@F=split(' ');");
2473 sv_catpv(PL_linestr, "\n");
2474 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2475 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2476 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2477 SV *sv = NEWSV(85,0);
2479 sv_upgrade(sv, SVt_PVMG);
2480 sv_setsv(sv,PL_linestr);
2481 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2487 bof = PL_rsfp && (PerlIO_tell(PL_rsfp) == 0); /* *Before* read! */
2488 s = filter_gets(PL_linestr, PL_rsfp, 0);
2492 if (PL_preprocess && !PL_in_eval)
2493 (void)PerlProc_pclose(PL_rsfp);
2494 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2495 PerlIO_clearerr(PL_rsfp);
2497 (void)PerlIO_close(PL_rsfp);
2499 PL_doextract = FALSE;
2501 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2502 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2503 sv_catpv(PL_linestr,";}");
2504 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2505 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2506 PL_minus_n = PL_minus_p = 0;
2509 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2510 sv_setpv(PL_linestr,"");
2511 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2513 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2514 s = swallow_bom((U8*)s);
2517 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2518 PL_doextract = FALSE;
2520 /* Incest with pod. */
2521 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2522 sv_setpv(PL_linestr, "");
2523 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2524 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2525 PL_doextract = FALSE;
2529 } while (PL_doextract);
2530 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2531 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2532 SV *sv = NEWSV(85,0);
2534 sv_upgrade(sv, SVt_PVMG);
2535 sv_setsv(sv,PL_linestr);
2536 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2538 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2539 if (CopLINE(PL_curcop) == 1) {
2540 while (s < PL_bufend && isSPACE(*s))
2542 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2546 if (*s == '#' && *(s+1) == '!')
2548 #ifdef ALTERNATE_SHEBANG
2550 static char as[] = ALTERNATE_SHEBANG;
2551 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2552 d = s + (sizeof(as) - 1);
2554 #endif /* ALTERNATE_SHEBANG */
2563 while (*d && !isSPACE(*d))
2567 #ifdef ARG_ZERO_IS_SCRIPT
2568 if (ipathend > ipath) {
2570 * HP-UX (at least) sets argv[0] to the script name,
2571 * which makes $^X incorrect. And Digital UNIX and Linux,
2572 * at least, set argv[0] to the basename of the Perl
2573 * interpreter. So, having found "#!", we'll set it right.
2575 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2576 assert(SvPOK(x) || SvGMAGICAL(x));
2577 if (sv_eq(x, CopFILESV(PL_curcop))) {
2578 sv_setpvn(x, ipath, ipathend - ipath);
2581 TAINT_NOT; /* $^X is always tainted, but that's OK */
2583 #endif /* ARG_ZERO_IS_SCRIPT */
2588 d = instr(s,"perl -");
2590 d = instr(s,"perl");
2592 /* avoid getting into infinite loops when shebang
2593 * line contains "Perl" rather than "perl" */
2595 for (d = ipathend-4; d >= ipath; --d) {
2596 if ((*d == 'p' || *d == 'P')
2597 && !ibcmp(d, "perl", 4))
2607 #ifdef ALTERNATE_SHEBANG
2609 * If the ALTERNATE_SHEBANG on this system starts with a
2610 * character that can be part of a Perl expression, then if
2611 * we see it but not "perl", we're probably looking at the
2612 * start of Perl code, not a request to hand off to some
2613 * other interpreter. Similarly, if "perl" is there, but
2614 * not in the first 'word' of the line, we assume the line
2615 * contains the start of the Perl program.
2617 if (d && *s != '#') {
2619 while (*c && !strchr("; \t\r\n\f\v#", *c))
2622 d = Nullch; /* "perl" not in first word; ignore */
2624 *s = '#'; /* Don't try to parse shebang line */
2626 #endif /* ALTERNATE_SHEBANG */
2627 #ifndef MACOS_TRADITIONAL
2632 !instr(s,"indir") &&
2633 instr(PL_origargv[0],"perl"))
2639 while (s < PL_bufend && isSPACE(*s))
2641 if (s < PL_bufend) {
2642 Newz(899,newargv,PL_origargc+3,char*);
2644 while (s < PL_bufend && !isSPACE(*s))
2647 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2650 newargv = PL_origargv;
2652 PerlProc_execv(ipath, newargv);
2653 Perl_croak(aTHX_ "Can't exec %s", ipath);
2657 U32 oldpdb = PL_perldb;
2658 bool oldn = PL_minus_n;
2659 bool oldp = PL_minus_p;
2661 while (*d && !isSPACE(*d)) d++;
2662 while (SPACE_OR_TAB(*d)) d++;
2666 if (*d == 'M' || *d == 'm') {
2668 while (*d && !isSPACE(*d)) d++;
2669 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2672 d = moreswitches(d);
2674 if ((PERLDB_LINE && !oldpdb) ||
2675 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2676 /* if we have already added "LINE: while (<>) {",
2677 we must not do it again */
2679 sv_setpv(PL_linestr, "");
2680 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2681 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2682 PL_preambled = FALSE;
2684 (void)gv_fetchfile(PL_origfilename);
2691 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2693 PL_lex_state = LEX_FORMLINE;
2698 #ifdef PERL_STRICT_CR
2699 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2701 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2703 case ' ': case '\t': case '\f': case 013:
2704 #ifdef MACOS_TRADITIONAL
2711 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2712 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2713 /* handle eval qq[#line 1 "foo"\n ...] */
2714 CopLINE_dec(PL_curcop);
2718 while (s < d && *s != '\n')
2723 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2725 PL_lex_state = LEX_FORMLINE;
2735 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2740 while (s < PL_bufend && SPACE_OR_TAB(*s))
2743 if (strnEQ(s,"=>",2)) {
2744 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2745 OPERATOR('-'); /* unary minus */
2747 PL_last_uni = PL_oldbufptr;
2748 PL_last_lop_op = OP_FTEREAD; /* good enough */
2750 case 'r': FTST(OP_FTEREAD);
2751 case 'w': FTST(OP_FTEWRITE);
2752 case 'x': FTST(OP_FTEEXEC);
2753 case 'o': FTST(OP_FTEOWNED);
2754 case 'R': FTST(OP_FTRREAD);
2755 case 'W': FTST(OP_FTRWRITE);
2756 case 'X': FTST(OP_FTREXEC);
2757 case 'O': FTST(OP_FTROWNED);
2758 case 'e': FTST(OP_FTIS);
2759 case 'z': FTST(OP_FTZERO);
2760 case 's': FTST(OP_FTSIZE);
2761 case 'f': FTST(OP_FTFILE);
2762 case 'd': FTST(OP_FTDIR);
2763 case 'l': FTST(OP_FTLINK);
2764 case 'p': FTST(OP_FTPIPE);
2765 case 'S': FTST(OP_FTSOCK);
2766 case 'u': FTST(OP_FTSUID);
2767 case 'g': FTST(OP_FTSGID);
2768 case 'k': FTST(OP_FTSVTX);
2769 case 'b': FTST(OP_FTBLK);
2770 case 'c': FTST(OP_FTCHR);
2771 case 't': FTST(OP_FTTTY);
2772 case 'T': FTST(OP_FTTEXT);
2773 case 'B': FTST(OP_FTBINARY);
2774 case 'M': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTMTIME);
2775 case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME);
2776 case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME);
2778 Perl_croak(aTHX_ "Unrecognized file test: -%c", (int)tmp);
2785 if (PL_expect == XOPERATOR)
2790 else if (*s == '>') {
2793 if (isIDFIRST_lazy_if(s,UTF)) {
2794 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2802 if (PL_expect == XOPERATOR)
2805 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2807 OPERATOR('-'); /* unary minus */
2814 if (PL_expect == XOPERATOR)
2819 if (PL_expect == XOPERATOR)
2822 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2828 if (PL_expect != XOPERATOR) {
2829 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2830 PL_expect = XOPERATOR;
2831 force_ident(PL_tokenbuf, '*');
2844 if (PL_expect == XOPERATOR) {
2848 PL_tokenbuf[0] = '%';
2849 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2850 if (!PL_tokenbuf[1]) {
2852 yyerror("Final % should be \\% or %name");
2855 PL_pending_ident = '%';
2874 switch (PL_expect) {
2877 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2879 PL_bufptr = s; /* update in case we back off */
2885 PL_expect = XTERMBLOCK;
2889 while (isIDFIRST_lazy_if(s,UTF)) {
2890 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2891 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2892 if (tmp < 0) tmp = -tmp;
2907 d = scan_str(d,TRUE,TRUE);
2910 SvREFCNT_dec(PL_lex_stuff);
2911 PL_lex_stuff = Nullsv;
2913 /* MUST advance bufptr here to avoid bogus
2914 "at end of line" context messages from yyerror().
2916 PL_bufptr = s + len;
2917 yyerror("Unterminated attribute parameter in attribute list");
2920 return 0; /* EOF indicator */
2924 SV *sv = newSVpvn(s, len);
2925 sv_catsv(sv, PL_lex_stuff);
2926 attrs = append_elem(OP_LIST, attrs,
2927 newSVOP(OP_CONST, 0, sv));
2928 SvREFCNT_dec(PL_lex_stuff);
2929 PL_lex_stuff = Nullsv;
2932 attrs = append_elem(OP_LIST, attrs,
2933 newSVOP(OP_CONST, 0,
2937 if (*s == ':' && s[1] != ':')
2940 break; /* require real whitespace or :'s */
2942 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
2943 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
2944 char q = ((*s == '\'') ? '"' : '\'');
2945 /* If here for an expression, and parsed no attrs, back off. */
2946 if (tmp == '=' && !attrs) {
2950 /* MUST advance bufptr here to avoid bogus "at end of line"
2951 context messages from yyerror().
2955 yyerror("Unterminated attribute list");
2957 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
2965 PL_nextval[PL_nexttoke].opval = attrs;
2973 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
2974 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
2990 if (PL_lex_brackets <= 0)
2991 yyerror("Unmatched right square bracket");
2994 if (PL_lex_state == LEX_INTERPNORMAL) {
2995 if (PL_lex_brackets == 0) {
2996 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
2997 PL_lex_state = LEX_INTERPEND;
3004 if (PL_lex_brackets > 100) {
3005 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
3006 if (newlb != PL_lex_brackstack) {
3008 PL_lex_brackstack = newlb;
3011 switch (PL_expect) {
3013 if (PL_lex_formbrack) {
3017 if (PL_oldoldbufptr == PL_last_lop)
3018 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3020 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3021 OPERATOR(HASHBRACK);
3023 while (s < PL_bufend && SPACE_OR_TAB(*s))
3026 PL_tokenbuf[0] = '\0';
3027 if (d < PL_bufend && *d == '-') {
3028 PL_tokenbuf[0] = '-';
3030 while (d < PL_bufend && SPACE_OR_TAB(*d))
3033 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3034 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3036 while (d < PL_bufend && SPACE_OR_TAB(*d))
3039 char minus = (PL_tokenbuf[0] == '-');
3040 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3048 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3053 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3058 if (PL_oldoldbufptr == PL_last_lop)
3059 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3061 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3064 OPERATOR(HASHBRACK);
3065 /* This hack serves to disambiguate a pair of curlies
3066 * as being a block or an anon hash. Normally, expectation
3067 * determines that, but in cases where we're not in a
3068 * position to expect anything in particular (like inside
3069 * eval"") we have to resolve the ambiguity. This code
3070 * covers the case where the first term in the curlies is a
3071 * quoted string. Most other cases need to be explicitly
3072 * disambiguated by prepending a `+' before the opening
3073 * curly in order to force resolution as an anon hash.
3075 * XXX should probably propagate the outer expectation
3076 * into eval"" to rely less on this hack, but that could
3077 * potentially break current behavior of eval"".
3081 if (*s == '\'' || *s == '"' || *s == '`') {
3082 /* common case: get past first string, handling escapes */
3083 for (t++; t < PL_bufend && *t != *s;)
3084 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3088 else if (*s == 'q') {
3091 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3095 char open, close, term;
3098 while (t < PL_bufend && isSPACE(*t))
3102 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3106 for (t++; t < PL_bufend; t++) {
3107 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3109 else if (*t == open)
3113 for (t++; t < PL_bufend; t++) {
3114 if (*t == '\\' && t+1 < PL_bufend)
3116 else if (*t == close && --brackets <= 0)
3118 else if (*t == open)
3124 else if (isALNUM_lazy_if(t,UTF)) {
3126 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3129 while (t < PL_bufend && isSPACE(*t))
3131 /* if comma follows first term, call it an anon hash */
3132 /* XXX it could be a comma expression with loop modifiers */
3133 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3134 || (*t == '=' && t[1] == '>')))
3135 OPERATOR(HASHBRACK);
3136 if (PL_expect == XREF)
3139 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3145 yylval.ival = CopLINE(PL_curcop);
3146 if (isSPACE(*s) || *s == '#')
3147 PL_copline = NOLINE; /* invalidate current command line number */
3152 if (PL_lex_brackets <= 0)
3153 yyerror("Unmatched right curly bracket");
3155 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3156 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3157 PL_lex_formbrack = 0;
3158 if (PL_lex_state == LEX_INTERPNORMAL) {
3159 if (PL_lex_brackets == 0) {
3160 if (PL_expect & XFAKEBRACK) {
3161 PL_expect &= XENUMMASK;
3162 PL_lex_state = LEX_INTERPEND;
3164 return yylex(); /* ignore fake brackets */
3166 if (*s == '-' && s[1] == '>')
3167 PL_lex_state = LEX_INTERPENDMAYBE;
3168 else if (*s != '[' && *s != '{')
3169 PL_lex_state = LEX_INTERPEND;
3172 if (PL_expect & XFAKEBRACK) {
3173 PL_expect &= XENUMMASK;
3175 return yylex(); /* ignore fake brackets */
3185 if (PL_expect == XOPERATOR) {
3186 if (ckWARN(WARN_SEMICOLON)
3187 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3189 CopLINE_dec(PL_curcop);
3190 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3191 CopLINE_inc(PL_curcop);
3196 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3198 PL_expect = XOPERATOR;
3199 force_ident(PL_tokenbuf, '&');
3203 yylval.ival = (OPpENTERSUB_AMPER<<8);
3222 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3223 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3225 if (PL_expect == XSTATE && isALPHA(tmp) &&
3226 (s == PL_linestart+1 || s[-2] == '\n') )
3228 if (PL_in_eval && !PL_rsfp) {
3233 if (strnEQ(s,"=cut",4)) {
3247 PL_doextract = TRUE;
3250 if (PL_lex_brackets < PL_lex_formbrack) {
3252 #ifdef PERL_STRICT_CR
3253 for (t = s; SPACE_OR_TAB(*t); t++) ;
3255 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3257 if (*t == '\n' || *t == '#') {
3275 if (PL_expect != XOPERATOR) {
3276 if (s[1] != '<' && !strchr(s,'>'))
3279 s = scan_heredoc(s);
3281 s = scan_inputsymbol(s);
3282 TERM(sublex_start());
3287 SHop(OP_LEFT_SHIFT);
3301 SHop(OP_RIGHT_SHIFT);
3310 if (PL_expect == XOPERATOR) {
3311 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3314 return ','; /* grandfather non-comma-format format */
3318 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3319 PL_tokenbuf[0] = '@';
3320 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3321 sizeof PL_tokenbuf - 1, FALSE);
3322 if (PL_expect == XOPERATOR)
3323 no_op("Array length", s);
3324 if (!PL_tokenbuf[1])
3326 PL_expect = XOPERATOR;
3327 PL_pending_ident = '#';
3331 PL_tokenbuf[0] = '$';
3332 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3333 sizeof PL_tokenbuf - 1, FALSE);
3334 if (PL_expect == XOPERATOR)
3336 if (!PL_tokenbuf[1]) {
3338 yyerror("Final $ should be \\$ or $name");
3342 /* This kludge not intended to be bulletproof. */
3343 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3344 yylval.opval = newSVOP(OP_CONST, 0,
3345 newSViv(PL_compiling.cop_arybase));
3346 yylval.opval->op_private = OPpCONST_ARYBASE;
3352 if (PL_lex_state == LEX_NORMAL)
3355 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3358 PL_tokenbuf[0] = '@';
3359 if (ckWARN(WARN_SYNTAX)) {
3361 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3364 PL_bufptr = skipspace(PL_bufptr);
3365 while (t < PL_bufend && *t != ']')
3367 Perl_warner(aTHX_ WARN_SYNTAX,
3368 "Multidimensional syntax %.*s not supported",
3369 (t - PL_bufptr) + 1, PL_bufptr);
3373 else if (*s == '{') {
3374 PL_tokenbuf[0] = '%';
3375 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3376 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3378 char tmpbuf[sizeof PL_tokenbuf];
3380 for (t++; isSPACE(*t); t++) ;
3381 if (isIDFIRST_lazy_if(t,UTF)) {
3382 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3383 for (; isSPACE(*t); t++) ;
3384 if (*t == ';' && get_cv(tmpbuf, FALSE))
3385 Perl_warner(aTHX_ WARN_SYNTAX,
3386 "You need to quote \"%s\"", tmpbuf);
3392 PL_expect = XOPERATOR;
3393 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3394 bool islop = (PL_last_lop == PL_oldoldbufptr);
3395 if (!islop || PL_last_lop_op == OP_GREPSTART)
3396 PL_expect = XOPERATOR;
3397 else if (strchr("$@\"'`q", *s))
3398 PL_expect = XTERM; /* e.g. print $fh "foo" */
3399 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3400 PL_expect = XTERM; /* e.g. print $fh &sub */
3401 else if (isIDFIRST_lazy_if(s,UTF)) {
3402 char tmpbuf[sizeof PL_tokenbuf];
3403 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3404 if ((tmp = keyword(tmpbuf, len))) {
3405 /* binary operators exclude handle interpretations */
3417 PL_expect = XTERM; /* e.g. print $fh length() */
3422 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3423 if (gv && GvCVu(gv))
3424 PL_expect = XTERM; /* e.g. print $fh subr() */
3427 else if (isDIGIT(*s))
3428 PL_expect = XTERM; /* e.g. print $fh 3 */
3429 else if (*s == '.' && isDIGIT(s[1]))
3430 PL_expect = XTERM; /* e.g. print $fh .3 */
3431 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3432 PL_expect = XTERM; /* e.g. print $fh -1 */
3433 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3434 PL_expect = XTERM; /* print $fh <<"EOF" */
3436 PL_pending_ident = '$';
3440 if (PL_expect == XOPERATOR)
3442 PL_tokenbuf[0] = '@';
3443 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3444 if (!PL_tokenbuf[1]) {
3446 yyerror("Final @ should be \\@ or @name");
3449 if (PL_lex_state == LEX_NORMAL)
3451 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3453 PL_tokenbuf[0] = '%';
3455 /* Warn about @ where they meant $. */
3456 if (ckWARN(WARN_SYNTAX)) {
3457 if (*s == '[' || *s == '{') {
3459 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3461 if (*t == '}' || *t == ']') {
3463 PL_bufptr = skipspace(PL_bufptr);
3464 Perl_warner(aTHX_ WARN_SYNTAX,
3465 "Scalar value %.*s better written as $%.*s",
3466 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3471 PL_pending_ident = '@';
3474 case '/': /* may either be division or pattern */
3475 case '?': /* may either be conditional or pattern */
3476 if (PL_expect != XOPERATOR) {
3477 /* Disable warning on "study /blah/" */
3478 if (PL_oldoldbufptr == PL_last_uni
3479 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3480 || memNE(PL_last_uni, "study", 5)
3481 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3483 s = scan_pat(s,OP_MATCH);
3484 TERM(sublex_start());
3492 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3493 #ifdef PERL_STRICT_CR
3496 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3498 && (s == PL_linestart || s[-1] == '\n') )
3500 PL_lex_formbrack = 0;
3504 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3510 yylval.ival = OPf_SPECIAL;
3516 if (PL_expect != XOPERATOR)
3521 case '0': case '1': case '2': case '3': case '4':
3522 case '5': case '6': case '7': case '8': case '9':
3524 if (PL_expect == XOPERATOR)
3529 s = scan_str(s,FALSE,FALSE);
3530 if (PL_expect == XOPERATOR) {
3531 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3534 return ','; /* grandfather non-comma-format format */
3540 missingterm((char*)0);
3541 yylval.ival = OP_CONST;
3542 TERM(sublex_start());
3545 s = scan_str(s,FALSE,FALSE);
3546 if (PL_expect == XOPERATOR) {
3547 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3550 return ','; /* grandfather non-comma-format format */
3556 missingterm((char*)0);
3557 yylval.ival = OP_CONST;
3558 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3559 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3560 yylval.ival = OP_STRINGIFY;
3564 TERM(sublex_start());
3567 s = scan_str(s,FALSE,FALSE);
3568 if (PL_expect == XOPERATOR)
3569 no_op("Backticks",s);
3571 missingterm((char*)0);
3572 yylval.ival = OP_BACKTICK;
3574 TERM(sublex_start());
3578 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3579 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3581 if (PL_expect == XOPERATOR)
3582 no_op("Backslash",s);
3586 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3590 while (isDIGIT(*start) || *start == '_')
3592 if (*start == '.' && isDIGIT(start[1])) {
3596 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3597 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF)) {
3601 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3611 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3650 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3652 /* Some keywords can be followed by any delimiter, including ':' */
3653 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3654 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3655 (PL_tokenbuf[0] == 'q' &&
3656 strchr("qwxr", PL_tokenbuf[1])))));
3658 /* x::* is just a word, unless x is "CORE" */
3659 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3663 while (d < PL_bufend && isSPACE(*d))
3664 d++; /* no comments skipped here, or s### is misparsed */
3666 /* Is this a label? */
3667 if (!tmp && PL_expect == XSTATE
3668 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3670 yylval.pval = savepv(PL_tokenbuf);
3675 /* Check for keywords */
3676 tmp = keyword(PL_tokenbuf, len);
3678 /* Is this a word before a => operator? */
3679 if (*d == '=' && d[1] == '>') {
3681 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3682 yylval.opval->op_private = OPpCONST_BARE;
3686 if (tmp < 0) { /* second-class keyword? */
3687 GV *ogv = Nullgv; /* override (winner) */
3688 GV *hgv = Nullgv; /* hidden (loser) */
3689 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3691 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3694 if (GvIMPORTED_CV(gv))
3696 else if (! CvMETHOD(cv))
3700 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3701 (gv = *gvp) != (GV*)&PL_sv_undef &&
3702 GvCVu(gv) && GvIMPORTED_CV(gv))
3708 tmp = 0; /* overridden by import or by GLOBAL */
3711 && -tmp==KEY_lock /* XXX generalizable kludge */
3713 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3715 tmp = 0; /* any sub overrides "weak" keyword */
3717 else { /* no override */
3721 if (ckWARN(WARN_AMBIGUOUS) && hgv
3722 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3723 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3724 "Ambiguous call resolved as CORE::%s(), %s",
3725 GvENAME(hgv), "qualify as such or use &");
3732 default: /* not a keyword */
3735 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3737 /* Get the rest if it looks like a package qualifier */
3739 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3741 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3744 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3745 *s == '\'' ? "'" : "::");
3749 if (PL_expect == XOPERATOR) {
3750 if (PL_bufptr == PL_linestart) {
3751 CopLINE_dec(PL_curcop);
3752 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3753 CopLINE_inc(PL_curcop);
3756 no_op("Bareword",s);
3759 /* Look for a subroutine with this name in current package,
3760 unless name is "Foo::", in which case Foo is a bearword
3761 (and a package name). */
3764 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3766 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3767 Perl_warner(aTHX_ WARN_BAREWORD,
3768 "Bareword \"%s\" refers to nonexistent package",
3771 PL_tokenbuf[len] = '\0';
3778 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3781 /* if we saw a global override before, get the right name */
3784 sv = newSVpvn("CORE::GLOBAL::",14);
3785 sv_catpv(sv,PL_tokenbuf);
3788 sv = newSVpv(PL_tokenbuf,0);
3790 /* Presume this is going to be a bareword of some sort. */
3793 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3794 yylval.opval->op_private = OPpCONST_BARE;
3796 /* And if "Foo::", then that's what it certainly is. */
3801 /* See if it's the indirect object for a list operator. */
3803 if (PL_oldoldbufptr &&
3804 PL_oldoldbufptr < PL_bufptr &&
3805 (PL_oldoldbufptr == PL_last_lop
3806 || PL_oldoldbufptr == PL_last_uni) &&
3807 /* NO SKIPSPACE BEFORE HERE! */
3808 (PL_expect == XREF ||
3809 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3811 bool immediate_paren = *s == '(';
3813 /* (Now we can afford to cross potential line boundary.) */
3816 /* Two barewords in a row may indicate method call. */
3818 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3821 /* If not a declared subroutine, it's an indirect object. */
3822 /* (But it's an indir obj regardless for sort.) */
3824 if ((PL_last_lop_op == OP_SORT ||
3825 (!immediate_paren && (!gv || !GvCVu(gv)))) &&
3826 (PL_last_lop_op != OP_MAPSTART &&
3827 PL_last_lop_op != OP_GREPSTART))
3829 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3835 PL_expect = XOPERATOR;
3838 /* Is this a word before a => operator? */
3839 if (*s == '=' && s[1] == '>') {
3841 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
3845 /* If followed by a paren, it's certainly a subroutine. */
3848 if (gv && GvCVu(gv)) {
3849 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
3850 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3855 PL_nextval[PL_nexttoke].opval = yylval.opval;
3856 PL_expect = XOPERATOR;
3862 /* If followed by var or block, call it a method (unless sub) */
3864 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3865 PL_last_lop = PL_oldbufptr;
3866 PL_last_lop_op = OP_METHOD;
3870 /* If followed by a bareword, see if it looks like indir obj. */
3872 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3875 /* Not a method, so call it a subroutine (if defined) */
3877 if (gv && GvCVu(gv)) {
3879 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3880 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3881 "Ambiguous use of -%s resolved as -&%s()",
3882 PL_tokenbuf, PL_tokenbuf);
3883 /* Check for a constant sub */
3885 if ((sv = cv_const_sv(cv))) {
3887 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3888 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3889 yylval.opval->op_private = 0;
3893 /* Resolve to GV now. */
3894 op_free(yylval.opval);
3895 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3896 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3897 PL_last_lop = PL_oldbufptr;
3898 PL_last_lop_op = OP_ENTERSUB;
3899 /* Is there a prototype? */
3902 char *proto = SvPV((SV*)cv, len);
3905 if (strEQ(proto, "$"))
3907 if (*proto == '&' && *s == '{') {
3908 sv_setpv(PL_subname,"__ANON__");
3912 PL_nextval[PL_nexttoke].opval = yylval.opval;
3918 /* Call it a bare word */
3920 if (PL_hints & HINT_STRICT_SUBS)
3921 yylval.opval->op_private |= OPpCONST_STRICT;
3924 if (ckWARN(WARN_RESERVED)) {
3925 if (lastchar != '-') {
3926 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
3928 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
3935 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
3936 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3937 "Operator or semicolon missing before %c%s",
3938 lastchar, PL_tokenbuf);
3939 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3940 "Ambiguous use of %c resolved as operator %c",
3941 lastchar, lastchar);
3947 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3948 newSVpv(CopFILE(PL_curcop),0));
3952 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3953 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
3956 case KEY___PACKAGE__:
3957 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3959 ? newSVsv(PL_curstname)
3968 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
3969 char *pname = "main";
3970 if (PL_tokenbuf[2] == 'D')
3971 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
3972 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
3975 GvIOp(gv) = newIO();
3976 IoIFP(GvIOp(gv)) = PL_rsfp;
3977 #if defined(HAS_FCNTL) && defined(F_SETFD)
3979 int fd = PerlIO_fileno(PL_rsfp);
3980 fcntl(fd,F_SETFD,fd >= 3);
3983 /* Mark this internal pseudo-handle as clean */
3984 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
3986 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
3987 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
3988 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
3990 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
3991 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
3992 /* if the script was opened in binmode, we need to revert
3993 * it to text mode for compatibility; but only iff it has CRs
3994 * XXX this is a questionable hack at best. */
3995 if (PL_bufend-PL_bufptr > 2
3996 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
3999 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4000 loc = PerlIO_tell(PL_rsfp);
4001 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4003 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4004 #if defined(__BORLANDC__)
4005 /* XXX see note in do_binmode() */
4006 ((FILE*)PL_rsfp)->flags |= _F_BIN;
4009 PerlIO_seek(PL_rsfp, loc, 0);
4024 if (PL_expect == XSTATE) {
4031 if (*s == ':' && s[1] == ':') {
4034 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4035 if (!(tmp = keyword(PL_tokenbuf, len)))
4036 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4050 LOP(OP_ACCEPT,XTERM);
4056 LOP(OP_ATAN2,XTERM);
4062 LOP(OP_BINMODE,XTERM);
4065 LOP(OP_BLESS,XTERM);
4074 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4091 if (!PL_cryptseen) {
4092 PL_cryptseen = TRUE;
4096 LOP(OP_CRYPT,XTERM);
4099 if (ckWARN(WARN_CHMOD)) {
4100 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4101 if (*d != '0' && isDIGIT(*d))
4102 Perl_warner(aTHX_ WARN_CHMOD,
4103 "chmod() mode argument is missing initial 0");
4105 LOP(OP_CHMOD,XTERM);
4108 LOP(OP_CHOWN,XTERM);
4111 LOP(OP_CONNECT,XTERM);
4127 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4131 PL_hints |= HINT_BLOCK_SCOPE;
4141 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4142 LOP(OP_DBMOPEN,XTERM);
4148 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4155 yylval.ival = CopLINE(PL_curcop);
4169 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4170 UNIBRACK(OP_ENTEREVAL);
4185 case KEY_endhostent:
4191 case KEY_endservent:
4194 case KEY_endprotoent:
4205 yylval.ival = CopLINE(PL_curcop);
4207 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4209 if ((PL_bufend - p) >= 3 &&
4210 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4212 else if ((PL_bufend - p) >= 4 &&
4213 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4216 if (isIDFIRST_lazy_if(p,UTF)) {
4217 p = scan_ident(p, PL_bufend,
4218 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4222 Perl_croak(aTHX_ "Missing $ on loop variable");
4227 LOP(OP_FORMLINE,XTERM);
4233 LOP(OP_FCNTL,XTERM);
4239 LOP(OP_FLOCK,XTERM);
4248 LOP(OP_GREPSTART, XREF);
4251 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4266 case KEY_getpriority:
4267 LOP(OP_GETPRIORITY,XTERM);
4269 case KEY_getprotobyname:
4272 case KEY_getprotobynumber:
4273 LOP(OP_GPBYNUMBER,XTERM);
4275 case KEY_getprotoent:
4287 case KEY_getpeername:
4288 UNI(OP_GETPEERNAME);
4290 case KEY_gethostbyname:
4293 case KEY_gethostbyaddr:
4294 LOP(OP_GHBYADDR,XTERM);
4296 case KEY_gethostent:
4299 case KEY_getnetbyname:
4302 case KEY_getnetbyaddr:
4303 LOP(OP_GNBYADDR,XTERM);
4308 case KEY_getservbyname:
4309 LOP(OP_GSBYNAME,XTERM);
4311 case KEY_getservbyport:
4312 LOP(OP_GSBYPORT,XTERM);
4314 case KEY_getservent:
4317 case KEY_getsockname:
4318 UNI(OP_GETSOCKNAME);
4320 case KEY_getsockopt:
4321 LOP(OP_GSOCKOPT,XTERM);
4343 yylval.ival = CopLINE(PL_curcop);
4347 LOP(OP_INDEX,XTERM);
4353 LOP(OP_IOCTL,XTERM);
4365 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4397 LOP(OP_LISTEN,XTERM);
4406 s = scan_pat(s,OP_MATCH);
4407 TERM(sublex_start());
4410 LOP(OP_MAPSTART, XREF);
4413 LOP(OP_MKDIR,XTERM);
4416 LOP(OP_MSGCTL,XTERM);
4419 LOP(OP_MSGGET,XTERM);
4422 LOP(OP_MSGRCV,XTERM);
4425 LOP(OP_MSGSND,XTERM);
4431 if (isIDFIRST_lazy_if(s,UTF)) {
4432 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4433 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4435 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4436 if (!PL_in_my_stash) {
4439 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4447 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4454 if (PL_expect != XSTATE)
4455 yyerror("\"no\" not allowed in expression");
4456 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4457 s = force_version(s);
4462 if (*s == '(' || (s = skipspace(s), *s == '('))
4469 if (isIDFIRST_lazy_if(s,UTF)) {
4471 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4473 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4474 Perl_warner(aTHX_ WARN_PRECEDENCE,
4475 "Precedence problem: open %.*s should be open(%.*s)",
4481 yylval.ival = OP_OR;
4491 LOP(OP_OPEN_DIR,XTERM);
4494 checkcomma(s,PL_tokenbuf,"filehandle");
4498 checkcomma(s,PL_tokenbuf,"filehandle");
4517 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4521 LOP(OP_PIPE_OP,XTERM);
4524 s = scan_str(s,FALSE,FALSE);
4526 missingterm((char*)0);
4527 yylval.ival = OP_CONST;
4528 TERM(sublex_start());
4534 s = scan_str(s,FALSE,FALSE);
4536 missingterm((char*)0);
4538 if (SvCUR(PL_lex_stuff)) {
4541 d = SvPV_force(PL_lex_stuff, len);
4543 for (; isSPACE(*d) && len; --len, ++d) ;
4546 if (!warned && ckWARN(WARN_QW)) {
4547 for (; !isSPACE(*d) && len; --len, ++d) {
4549 Perl_warner(aTHX_ WARN_QW,
4550 "Possible attempt to separate words with commas");
4553 else if (*d == '#') {
4554 Perl_warner(aTHX_ WARN_QW,
4555 "Possible attempt to put comments in qw() list");
4561 for (; !isSPACE(*d) && len; --len, ++d) ;
4563 words = append_elem(OP_LIST, words,
4564 newSVOP(OP_CONST, 0, tokeq(newSVpvn(b, d-b))));
4568 PL_nextval[PL_nexttoke].opval = words;
4573 SvREFCNT_dec(PL_lex_stuff);
4574 PL_lex_stuff = Nullsv;
4579 s = scan_str(s,FALSE,FALSE);
4581 missingterm((char*)0);
4582 yylval.ival = OP_STRINGIFY;
4583 if (SvIVX(PL_lex_stuff) == '\'')
4584 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4585 TERM(sublex_start());
4588 s = scan_pat(s,OP_QR);
4589 TERM(sublex_start());
4592 s = scan_str(s,FALSE,FALSE);
4594 missingterm((char*)0);
4595 yylval.ival = OP_BACKTICK;
4597 TERM(sublex_start());
4604 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4605 s = force_version(s);
4608 *PL_tokenbuf = '\0';
4609 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4610 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4611 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4613 yyerror("<> should be quotes");
4621 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4625 LOP(OP_RENAME,XTERM);
4634 LOP(OP_RINDEX,XTERM);
4657 LOP(OP_REVERSE,XTERM);
4668 TERM(sublex_start());
4670 TOKEN(1); /* force error */
4679 LOP(OP_SELECT,XTERM);
4685 LOP(OP_SEMCTL,XTERM);
4688 LOP(OP_SEMGET,XTERM);
4691 LOP(OP_SEMOP,XTERM);
4697 LOP(OP_SETPGRP,XTERM);
4699 case KEY_setpriority:
4700 LOP(OP_SETPRIORITY,XTERM);
4702 case KEY_sethostent:
4708 case KEY_setservent:
4711 case KEY_setprotoent:
4721 LOP(OP_SEEKDIR,XTERM);
4723 case KEY_setsockopt:
4724 LOP(OP_SSOCKOPT,XTERM);
4730 LOP(OP_SHMCTL,XTERM);
4733 LOP(OP_SHMGET,XTERM);
4736 LOP(OP_SHMREAD,XTERM);
4739 LOP(OP_SHMWRITE,XTERM);
4742 LOP(OP_SHUTDOWN,XTERM);
4751 LOP(OP_SOCKET,XTERM);
4753 case KEY_socketpair:
4754 LOP(OP_SOCKPAIR,XTERM);
4757 checkcomma(s,PL_tokenbuf,"subroutine name");
4759 if (*s == ';' || *s == ')') /* probably a close */
4760 Perl_croak(aTHX_ "sort is now a reserved word");
4762 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4766 LOP(OP_SPLIT,XTERM);
4769 LOP(OP_SPRINTF,XTERM);
4772 LOP(OP_SPLICE,XTERM);
4787 LOP(OP_SUBSTR,XTERM);
4793 char tmpbuf[sizeof PL_tokenbuf];
4795 expectation attrful;
4796 bool have_name, have_proto;
4801 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4802 (*s == ':' && s[1] == ':'))
4805 attrful = XATTRBLOCK;
4806 /* remember buffer pos'n for later force_word */
4807 tboffset = s - PL_oldbufptr;
4808 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4809 if (strchr(tmpbuf, ':'))
4810 sv_setpv(PL_subname, tmpbuf);
4812 sv_setsv(PL_subname,PL_curstname);
4813 sv_catpvn(PL_subname,"::",2);
4814 sv_catpvn(PL_subname,tmpbuf,len);
4821 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4822 PL_expect = XTERMBLOCK;
4823 attrful = XATTRTERM;
4824 sv_setpv(PL_subname,"?");
4828 if (key == KEY_format) {
4830 PL_lex_formbrack = PL_lex_brackets + 1;
4832 (void) force_word(PL_oldbufptr + tboffset, WORD,
4837 /* Look for a prototype */
4841 s = scan_str(s,FALSE,FALSE);
4844 SvREFCNT_dec(PL_lex_stuff);
4845 PL_lex_stuff = Nullsv;
4846 Perl_croak(aTHX_ "Prototype not terminated");
4849 d = SvPVX(PL_lex_stuff);
4851 for (p = d; *p; ++p) {
4856 SvCUR(PL_lex_stuff) = tmp;
4864 if (*s == ':' && s[1] != ':')
4865 PL_expect = attrful;
4868 PL_nextval[PL_nexttoke].opval =
4869 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4870 PL_lex_stuff = Nullsv;
4874 sv_setpv(PL_subname,"__ANON__");
4877 (void) force_word(PL_oldbufptr + tboffset, WORD,
4886 LOP(OP_SYSTEM,XREF);
4889 LOP(OP_SYMLINK,XTERM);
4892 LOP(OP_SYSCALL,XTERM);
4895 LOP(OP_SYSOPEN,XTERM);
4898 LOP(OP_SYSSEEK,XTERM);
4901 LOP(OP_SYSREAD,XTERM);
4904 LOP(OP_SYSWRITE,XTERM);
4908 TERM(sublex_start());
4929 LOP(OP_TRUNCATE,XTERM);
4941 yylval.ival = CopLINE(PL_curcop);
4945 yylval.ival = CopLINE(PL_curcop);
4949 LOP(OP_UNLINK,XTERM);
4955 LOP(OP_UNPACK,XTERM);
4958 LOP(OP_UTIME,XTERM);
4961 if (ckWARN(WARN_UMASK)) {
4962 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4963 if (*d != '0' && isDIGIT(*d))
4964 Perl_warner(aTHX_ WARN_UMASK,
4965 "umask: argument is missing initial 0");
4970 LOP(OP_UNSHIFT,XTERM);
4973 if (PL_expect != XSTATE)
4974 yyerror("\"use\" not allowed in expression");
4976 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4977 s = force_version(s);
4978 if (*s == ';' || (s = skipspace(s), *s == ';')) {
4979 PL_nextval[PL_nexttoke].opval = Nullop;
4984 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4985 s = force_version(s);
4997 yylval.ival = CopLINE(PL_curcop);
5001 PL_hints |= HINT_BLOCK_SCOPE;
5008 LOP(OP_WAITPID,XTERM);
5016 static char ctl_l[2];
5018 if (ctl_l[0] == '\0')
5019 ctl_l[0] = toCTRL('L');
5020 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5023 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5028 if (PL_expect == XOPERATOR)
5034 yylval.ival = OP_XOR;
5039 TERM(sublex_start());
5044 #pragma segment Main
5048 Perl_keyword(pTHX_ register char *d, I32 len)
5053 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5054 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5055 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5056 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5057 if (strEQ(d,"__END__")) return KEY___END__;
5061 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5066 if (strEQ(d,"and")) return -KEY_and;
5067 if (strEQ(d,"abs")) return -KEY_abs;
5070 if (strEQ(d,"alarm")) return -KEY_alarm;
5071 if (strEQ(d,"atan2")) return -KEY_atan2;
5074 if (strEQ(d,"accept")) return -KEY_accept;
5079 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5082 if (strEQ(d,"bless")) return -KEY_bless;
5083 if (strEQ(d,"bind")) return -KEY_bind;
5084 if (strEQ(d,"binmode")) return -KEY_binmode;
5087 if (strEQ(d,"CORE")) return -KEY_CORE;
5088 if (strEQ(d,"CHECK")) return KEY_CHECK;
5093 if (strEQ(d,"cmp")) return -KEY_cmp;
5094 if (strEQ(d,"chr")) return -KEY_chr;
5095 if (strEQ(d,"cos")) return -KEY_cos;
5098 if (strEQ(d,"chop")) return -KEY_chop;
5101 if (strEQ(d,"close")) return -KEY_close;
5102 if (strEQ(d,"chdir")) return -KEY_chdir;
5103 if (strEQ(d,"chomp")) return -KEY_chomp;
5104 if (strEQ(d,"chmod")) return -KEY_chmod;
5105 if (strEQ(d,"chown")) return -KEY_chown;
5106 if (strEQ(d,"crypt")) return -KEY_crypt;
5109 if (strEQ(d,"chroot")) return -KEY_chroot;
5110 if (strEQ(d,"caller")) return -KEY_caller;
5113 if (strEQ(d,"connect")) return -KEY_connect;
5116 if (strEQ(d,"closedir")) return -KEY_closedir;
5117 if (strEQ(d,"continue")) return -KEY_continue;
5122 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5127 if (strEQ(d,"do")) return KEY_do;
5130 if (strEQ(d,"die")) return -KEY_die;
5133 if (strEQ(d,"dump")) return -KEY_dump;
5136 if (strEQ(d,"delete")) return KEY_delete;
5139 if (strEQ(d,"defined")) return KEY_defined;
5140 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5143 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5148 if (strEQ(d,"END")) return KEY_END;
5153 if (strEQ(d,"eq")) return -KEY_eq;
5156 if (strEQ(d,"eof")) return -KEY_eof;
5157 if (strEQ(d,"exp")) return -KEY_exp;
5160 if (strEQ(d,"else")) return KEY_else;
5161 if (strEQ(d,"exit")) return -KEY_exit;
5162 if (strEQ(d,"eval")) return KEY_eval;
5163 if (strEQ(d,"exec")) return -KEY_exec;
5164 if (strEQ(d,"each")) return KEY_each;
5167 if (strEQ(d,"elsif")) return KEY_elsif;
5170 if (strEQ(d,"exists")) return KEY_exists;
5171 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5174 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5175 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5178 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5181 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5182 if (strEQ(d,"endservent")) return -KEY_endservent;
5185 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5192 if (strEQ(d,"for")) return KEY_for;
5195 if (strEQ(d,"fork")) return -KEY_fork;
5198 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5199 if (strEQ(d,"flock")) return -KEY_flock;
5202 if (strEQ(d,"format")) return KEY_format;
5203 if (strEQ(d,"fileno")) return -KEY_fileno;
5206 if (strEQ(d,"foreach")) return KEY_foreach;
5209 if (strEQ(d,"formline")) return -KEY_formline;
5214 if (strnEQ(d,"get",3)) {
5219 if (strEQ(d,"ppid")) return -KEY_getppid;
5220 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5223 if (strEQ(d,"pwent")) return -KEY_getpwent;
5224 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5225 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5228 if (strEQ(d,"peername")) return -KEY_getpeername;
5229 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5230 if (strEQ(d,"priority")) return -KEY_getpriority;
5233 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5236 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5240 else if (*d == 'h') {
5241 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5242 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5243 if (strEQ(d,"hostent")) return -KEY_gethostent;
5245 else if (*d == 'n') {
5246 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5247 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5248 if (strEQ(d,"netent")) return -KEY_getnetent;
5250 else if (*d == 's') {
5251 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5252 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5253 if (strEQ(d,"servent")) return -KEY_getservent;
5254 if (strEQ(d,"sockname")) return -KEY_getsockname;
5255 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5257 else if (*d == 'g') {
5258 if (strEQ(d,"grent")) return -KEY_getgrent;
5259 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5260 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5262 else if (*d == 'l') {
5263 if (strEQ(d,"login")) return -KEY_getlogin;
5265 else if (strEQ(d,"c")) return -KEY_getc;
5270 if (strEQ(d,"gt")) return -KEY_gt;
5271 if (strEQ(d,"ge")) return -KEY_ge;
5274 if (strEQ(d,"grep")) return KEY_grep;
5275 if (strEQ(d,"goto")) return KEY_goto;
5276 if (strEQ(d,"glob")) return KEY_glob;
5279 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5284 if (strEQ(d,"hex")) return -KEY_hex;
5287 if (strEQ(d,"INIT")) return KEY_INIT;
5292 if (strEQ(d,"if")) return KEY_if;
5295 if (strEQ(d,"int")) return -KEY_int;
5298 if (strEQ(d,"index")) return -KEY_index;
5299 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5304 if (strEQ(d,"join")) return -KEY_join;
5308 if (strEQ(d,"keys")) return KEY_keys;
5309 if (strEQ(d,"kill")) return -KEY_kill;
5315 if (strEQ(d,"lt")) return -KEY_lt;
5316 if (strEQ(d,"le")) return -KEY_le;
5317 if (strEQ(d,"lc")) return -KEY_lc;
5320 if (strEQ(d,"log")) return -KEY_log;
5323 if (strEQ(d,"last")) return KEY_last;
5324 if (strEQ(d,"link")) return -KEY_link;
5325 if (strEQ(d,"lock")) return -KEY_lock;
5328 if (strEQ(d,"local")) return KEY_local;
5329 if (strEQ(d,"lstat")) return -KEY_lstat;
5332 if (strEQ(d,"length")) return -KEY_length;
5333 if (strEQ(d,"listen")) return -KEY_listen;
5336 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5339 if (strEQ(d,"localtime")) return -KEY_localtime;
5345 case 1: return KEY_m;
5347 if (strEQ(d,"my")) return KEY_my;
5350 if (strEQ(d,"map")) return KEY_map;
5353 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5356 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5357 if (strEQ(d,"msgget")) return -KEY_msgget;
5358 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5359 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5364 if (strEQ(d,"next")) return KEY_next;
5365 if (strEQ(d,"ne")) return -KEY_ne;
5366 if (strEQ(d,"not")) return -KEY_not;
5367 if (strEQ(d,"no")) return KEY_no;
5372 if (strEQ(d,"or")) return -KEY_or;
5375 if (strEQ(d,"ord")) return -KEY_ord;
5376 if (strEQ(d,"oct")) return -KEY_oct;
5377 if (strEQ(d,"our")) return KEY_our;
5380 if (strEQ(d,"open")) return -KEY_open;
5383 if (strEQ(d,"opendir")) return -KEY_opendir;
5390 if (strEQ(d,"pop")) return KEY_pop;
5391 if (strEQ(d,"pos")) return KEY_pos;
5394 if (strEQ(d,"push")) return KEY_push;
5395 if (strEQ(d,"pack")) return -KEY_pack;
5396 if (strEQ(d,"pipe")) return -KEY_pipe;
5399 if (strEQ(d,"print")) return KEY_print;
5402 if (strEQ(d,"printf")) return KEY_printf;
5405 if (strEQ(d,"package")) return KEY_package;
5408 if (strEQ(d,"prototype")) return KEY_prototype;
5413 if (strEQ(d,"q")) return KEY_q;
5414 if (strEQ(d,"qr")) return KEY_qr;
5415 if (strEQ(d,"qq")) return KEY_qq;
5416 if (strEQ(d,"qw")) return KEY_qw;
5417 if (strEQ(d,"qx")) return KEY_qx;
5419 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5424 if (strEQ(d,"ref")) return -KEY_ref;
5427 if (strEQ(d,"read")) return -KEY_read;
5428 if (strEQ(d,"rand")) return -KEY_rand;
5429 if (strEQ(d,"recv")) return -KEY_recv;
5430 if (strEQ(d,"redo")) return KEY_redo;
5433 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5434 if (strEQ(d,"reset")) return -KEY_reset;
5437 if (strEQ(d,"return")) return KEY_return;
5438 if (strEQ(d,"rename")) return -KEY_rename;
5439 if (strEQ(d,"rindex")) return -KEY_rindex;
5442 if (strEQ(d,"require")) return -KEY_require;
5443 if (strEQ(d,"reverse")) return -KEY_reverse;
5444 if (strEQ(d,"readdir")) return -KEY_readdir;
5447 if (strEQ(d,"readlink")) return -KEY_readlink;
5448 if (strEQ(d,"readline")) return -KEY_readline;
5449 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5452 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5458 case 0: return KEY_s;
5460 if (strEQ(d,"scalar")) return KEY_scalar;
5465 if (strEQ(d,"seek")) return -KEY_seek;
5466 if (strEQ(d,"send")) return -KEY_send;
5469 if (strEQ(d,"semop")) return -KEY_semop;
5472 if (strEQ(d,"select")) return -KEY_select;
5473 if (strEQ(d,"semctl")) return -KEY_semctl;
5474 if (strEQ(d,"semget")) return -KEY_semget;
5477 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5478 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5481 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5482 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5485 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5488 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5489 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5490 if (strEQ(d,"setservent")) return -KEY_setservent;
5493 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5494 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5501 if (strEQ(d,"shift")) return KEY_shift;
5504 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5505 if (strEQ(d,"shmget")) return -KEY_shmget;
5508 if (strEQ(d,"shmread")) return -KEY_shmread;
5511 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5512 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5517 if (strEQ(d,"sin")) return -KEY_sin;
5520 if (strEQ(d,"sleep")) return -KEY_sleep;
5523 if (strEQ(d,"sort")) return KEY_sort;
5524 if (strEQ(d,"socket")) return -KEY_socket;
5525 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5528 if (strEQ(d,"split")) return KEY_split;
5529 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5530 if (strEQ(d,"splice")) return KEY_splice;
5533 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5536 if (strEQ(d,"srand")) return -KEY_srand;
5539 if (strEQ(d,"stat")) return -KEY_stat;
5540 if (strEQ(d,"study")) return KEY_study;
5543 if (strEQ(d,"substr")) return -KEY_substr;
5544 if (strEQ(d,"sub")) return KEY_sub;
5549 if (strEQ(d,"system")) return -KEY_system;
5552 if (strEQ(d,"symlink")) return -KEY_symlink;
5553 if (strEQ(d,"syscall")) return -KEY_syscall;
5554 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5555 if (strEQ(d,"sysread")) return -KEY_sysread;
5556 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5559 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5568 if (strEQ(d,"tr")) return KEY_tr;
5571 if (strEQ(d,"tie")) return KEY_tie;
5574 if (strEQ(d,"tell")) return -KEY_tell;
5575 if (strEQ(d,"tied")) return KEY_tied;
5576 if (strEQ(d,"time")) return -KEY_time;
5579 if (strEQ(d,"times")) return -KEY_times;
5582 if (strEQ(d,"telldir")) return -KEY_telldir;
5585 if (strEQ(d,"truncate")) return -KEY_truncate;
5592 if (strEQ(d,"uc")) return -KEY_uc;
5595 if (strEQ(d,"use")) return KEY_use;
5598 if (strEQ(d,"undef")) return KEY_undef;
5599 if (strEQ(d,"until")) return KEY_until;
5600 if (strEQ(d,"untie")) return KEY_untie;
5601 if (strEQ(d,"utime")) return -KEY_utime;
5602 if (strEQ(d,"umask")) return -KEY_umask;
5605 if (strEQ(d,"unless")) return KEY_unless;
5606 if (strEQ(d,"unpack")) return -KEY_unpack;
5607 if (strEQ(d,"unlink")) return -KEY_unlink;
5610 if (strEQ(d,"unshift")) return KEY_unshift;
5611 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5616 if (strEQ(d,"values")) return -KEY_values;
5617 if (strEQ(d,"vec")) return -KEY_vec;
5622 if (strEQ(d,"warn")) return -KEY_warn;
5623 if (strEQ(d,"wait")) return -KEY_wait;
5626 if (strEQ(d,"while")) return KEY_while;
5627 if (strEQ(d,"write")) return -KEY_write;
5630 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5633 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5638 if (len == 1) return -KEY_x;
5639 if (strEQ(d,"xor")) return -KEY_xor;
5642 if (len == 1) return KEY_y;
5651 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5655 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5656 dTHR; /* only for ckWARN */
5657 if (ckWARN(WARN_SYNTAX)) {
5659 for (w = s+2; *w && level; w++) {
5666 for (; *w && isSPACE(*w); w++) ;
5667 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5668 Perl_warner(aTHX_ WARN_SYNTAX,
5669 "%s (...) interpreted as function",name);
5672 while (s < PL_bufend && isSPACE(*s))
5676 while (s < PL_bufend && isSPACE(*s))
5678 if (isIDFIRST_lazy_if(s,UTF)) {
5680 while (isALNUM_lazy_if(s,UTF))
5682 while (s < PL_bufend && isSPACE(*s))
5687 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5691 Perl_croak(aTHX_ "No comma allowed after %s", what);
5696 /* Either returns sv, or mortalizes sv and returns a new SV*.
5697 Best used as sv=new_constant(..., sv, ...).
5698 If s, pv are NULL, calls subroutine with one argument,
5699 and type is used with error messages only. */
5702 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5706 HV *table = GvHV(PL_hintgv); /* ^H */
5710 const char *why1, *why2, *why3;
5712 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
5715 why1 = "%^H is not consistent";
5716 why2 = strEQ(key,"charnames")
5717 ? " (missing \"use charnames ...\"?)"
5721 msg = Perl_newSVpvf(aTHX_ "constant(%s): %s%s%s",
5722 (type ? type: "undef"), why1, why2, why3);
5723 yyerror(SvPVX(msg));
5727 cvp = hv_fetch(table, key, strlen(key), FALSE);
5728 if (!cvp || !SvOK(*cvp)) {
5731 why3 = "} is not defined";
5734 sv_2mortal(sv); /* Parent created it permanently */
5737 pv = sv_2mortal(newSVpvn(s, len));
5739 typesv = sv_2mortal(newSVpv(type, 0));
5741 typesv = &PL_sv_undef;
5743 PUSHSTACKi(PERLSI_OVERLOAD);
5755 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5759 /* Check the eval first */
5760 if (!PL_in_eval && SvTRUE(ERRSV)) {
5762 sv_catpv(ERRSV, "Propagated");
5763 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5765 res = SvREFCNT_inc(sv);
5769 (void)SvREFCNT_inc(res);
5778 why1 = "Call to &{$^H{";
5780 why3 = "}} did not return a defined value";
5789 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5791 register char *d = dest;
5792 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5795 Perl_croak(aTHX_ ident_too_long);
5796 if (isALNUM(*s)) /* UTF handled below */
5798 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
5803 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5807 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5808 char *t = s + UTF8SKIP(s);
5809 while (*t & 0x80 && is_utf8_mark((U8*)t))
5811 if (d + (t - s) > e)
5812 Perl_croak(aTHX_ ident_too_long);
5813 Copy(s, d, t - s, char);
5826 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5836 e = d + destlen - 3; /* two-character token, ending NUL */
5838 while (isDIGIT(*s)) {
5840 Perl_croak(aTHX_ ident_too_long);
5847 Perl_croak(aTHX_ ident_too_long);
5848 if (isALNUM(*s)) /* UTF handled below */
5850 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
5855 else if (*s == ':' && s[1] == ':') {
5859 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5860 char *t = s + UTF8SKIP(s);
5861 while (*t & 0x80 && is_utf8_mark((U8*)t))
5863 if (d + (t - s) > e)
5864 Perl_croak(aTHX_ ident_too_long);
5865 Copy(s, d, t - s, char);
5876 if (PL_lex_state != LEX_NORMAL)
5877 PL_lex_state = LEX_INTERPENDMAYBE;
5880 if (*s == '$' && s[1] &&
5881 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
5894 if (*d == '^' && *s && isCONTROLVAR(*s)) {
5899 if (isSPACE(s[-1])) {
5902 if (!SPACE_OR_TAB(ch)) {
5908 if (isIDFIRST_lazy_if(d,UTF)) {
5912 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
5914 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
5917 Copy(s, d, e - s, char);
5922 while ((isALNUM(*s) || *s == ':') && d < e)
5925 Perl_croak(aTHX_ ident_too_long);
5928 while (s < send && SPACE_OR_TAB(*s)) s++;
5929 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
5930 dTHR; /* only for ckWARN */
5931 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
5932 const char *brack = *s == '[' ? "[...]" : "{...}";
5933 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5934 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
5935 funny, dest, brack, funny, dest, brack);
5938 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
5942 /* Handle extended ${^Foo} variables
5943 * 1999-02-27 mjd-perl-patch@plover.com */
5944 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
5948 while (isALNUM(*s) && d < e) {
5952 Perl_croak(aTHX_ ident_too_long);
5957 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
5958 PL_lex_state = LEX_INTERPEND;
5961 if (PL_lex_state == LEX_NORMAL) {
5962 dTHR; /* only for ckWARN */
5963 if (ckWARN(WARN_AMBIGUOUS) &&
5964 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
5966 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5967 "Ambiguous use of %c{%s} resolved to %c%s",
5968 funny, dest, funny, dest);
5973 s = bracket; /* let the parser handle it */
5977 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
5978 PL_lex_state = LEX_INTERPEND;
5983 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
5988 *pmfl |= PMf_GLOBAL;
5990 *pmfl |= PMf_CONTINUE;
5994 *pmfl |= PMf_MULTILINE;
5996 *pmfl |= PMf_SINGLELINE;
5998 *pmfl |= PMf_EXTENDED;
6002 S_scan_pat(pTHX_ char *start, I32 type)
6007 s = scan_str(start,FALSE,FALSE);
6010 SvREFCNT_dec(PL_lex_stuff);
6011 PL_lex_stuff = Nullsv;
6012 Perl_croak(aTHX_ "Search pattern not terminated");
6015 pm = (PMOP*)newPMOP(type, 0);
6016 if (PL_multi_open == '?')
6017 pm->op_pmflags |= PMf_ONCE;
6019 while (*s && strchr("iomsx", *s))
6020 pmflag(&pm->op_pmflags,*s++);
6023 while (*s && strchr("iogcmsx", *s))
6024 pmflag(&pm->op_pmflags,*s++);
6026 pm->op_pmpermflags = pm->op_pmflags;
6028 PL_lex_op = (OP*)pm;
6029 yylval.ival = OP_MATCH;
6034 S_scan_subst(pTHX_ char *start)
6041 yylval.ival = OP_NULL;
6043 s = scan_str(start,FALSE,FALSE);
6047 SvREFCNT_dec(PL_lex_stuff);
6048 PL_lex_stuff = Nullsv;
6049 Perl_croak(aTHX_ "Substitution pattern not terminated");
6052 if (s[-1] == PL_multi_open)
6055 first_start = PL_multi_start;
6056 s = scan_str(s,FALSE,FALSE);
6059 SvREFCNT_dec(PL_lex_stuff);
6060 PL_lex_stuff = Nullsv;
6062 SvREFCNT_dec(PL_lex_repl);
6063 PL_lex_repl = Nullsv;
6064 Perl_croak(aTHX_ "Substitution replacement not terminated");
6066 PL_multi_start = first_start; /* so whole substitution is taken together */
6068 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6074 else if (strchr("iogcmsx", *s))
6075 pmflag(&pm->op_pmflags,*s++);
6082 PL_sublex_info.super_bufptr = s;
6083 PL_sublex_info.super_bufend = PL_bufend;
6085 pm->op_pmflags |= PMf_EVAL;
6086 repl = newSVpvn("",0);
6088 sv_catpv(repl, es ? "eval " : "do ");
6089 sv_catpvn(repl, "{ ", 2);
6090 sv_catsv(repl, PL_lex_repl);
6091 sv_catpvn(repl, " };", 2);
6093 SvREFCNT_dec(PL_lex_repl);
6097 pm->op_pmpermflags = pm->op_pmflags;
6098 PL_lex_op = (OP*)pm;
6099 yylval.ival = OP_SUBST;
6104 S_scan_trans(pTHX_ char *start)
6115 yylval.ival = OP_NULL;
6117 s = scan_str(start,FALSE,FALSE);
6120 SvREFCNT_dec(PL_lex_stuff);
6121 PL_lex_stuff = Nullsv;
6122 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6124 if (s[-1] == PL_multi_open)
6127 s = scan_str(s,FALSE,FALSE);
6130 SvREFCNT_dec(PL_lex_stuff);
6131 PL_lex_stuff = Nullsv;
6133 SvREFCNT_dec(PL_lex_repl);
6134 PL_lex_repl = Nullsv;
6135 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6138 New(803,tbl,256,short);
6139 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6141 complement = del = squash = 0;
6142 while (strchr("cds", *s)) {
6144 complement = OPpTRANS_COMPLEMENT;
6146 del = OPpTRANS_DELETE;
6148 squash = OPpTRANS_SQUASH;
6151 o->op_private = del|squash|complement;
6154 yylval.ival = OP_TRANS;
6159 S_scan_heredoc(pTHX_ register char *s)
6163 I32 op_type = OP_SCALAR;
6170 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6174 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6177 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6178 if (*peek && strchr("`'\"",*peek)) {
6181 s = delimcpy(d, e, s, PL_bufend, term, &len);
6191 if (!isALNUM_lazy_if(s,UTF))
6192 deprecate("bare << to mean <<\"\"");
6193 for (; isALNUM_lazy_if(s,UTF); s++) {
6198 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6199 Perl_croak(aTHX_ "Delimiter for here document is too long");
6202 len = d - PL_tokenbuf;
6203 #ifndef PERL_STRICT_CR
6204 d = strchr(s, '\r');
6208 while (s < PL_bufend) {
6214 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6223 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6228 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6229 herewas = newSVpvn(s,PL_bufend-s);
6231 s--, herewas = newSVpvn(s,d-s);
6232 s += SvCUR(herewas);
6234 tmpstr = NEWSV(87,79);
6235 sv_upgrade(tmpstr, SVt_PVIV);
6240 else if (term == '`') {
6241 op_type = OP_BACKTICK;
6242 SvIVX(tmpstr) = '\\';
6246 PL_multi_start = CopLINE(PL_curcop);
6247 PL_multi_open = PL_multi_close = '<';
6248 term = *PL_tokenbuf;
6249 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6250 char *bufptr = PL_sublex_info.super_bufptr;
6251 char *bufend = PL_sublex_info.super_bufend;
6252 char *olds = s - SvCUR(herewas);
6253 s = strchr(bufptr, '\n');
6257 while (s < bufend &&
6258 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6260 CopLINE_inc(PL_curcop);
6263 CopLINE_set(PL_curcop, PL_multi_start);
6264 missingterm(PL_tokenbuf);
6266 sv_setpvn(herewas,bufptr,d-bufptr+1);
6267 sv_setpvn(tmpstr,d+1,s-d);
6269 sv_catpvn(herewas,s,bufend-s);
6270 (void)strcpy(bufptr,SvPVX(herewas));
6277 while (s < PL_bufend &&
6278 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6280 CopLINE_inc(PL_curcop);
6282 if (s >= PL_bufend) {
6283 CopLINE_set(PL_curcop, PL_multi_start);
6284 missingterm(PL_tokenbuf);
6286 sv_setpvn(tmpstr,d+1,s-d);
6288 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6290 sv_catpvn(herewas,s,PL_bufend-s);
6291 sv_setsv(PL_linestr,herewas);
6292 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6293 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6296 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6297 while (s >= PL_bufend) { /* multiple line string? */
6299 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6300 CopLINE_set(PL_curcop, PL_multi_start);
6301 missingterm(PL_tokenbuf);
6303 CopLINE_inc(PL_curcop);
6304 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6305 #ifndef PERL_STRICT_CR
6306 if (PL_bufend - PL_linestart >= 2) {
6307 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6308 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6310 PL_bufend[-2] = '\n';
6312 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6314 else if (PL_bufend[-1] == '\r')
6315 PL_bufend[-1] = '\n';
6317 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6318 PL_bufend[-1] = '\n';
6320 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6321 SV *sv = NEWSV(88,0);
6323 sv_upgrade(sv, SVt_PVMG);
6324 sv_setsv(sv,PL_linestr);
6325 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6327 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6330 sv_catsv(PL_linestr,herewas);
6331 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6335 sv_catsv(tmpstr,PL_linestr);
6340 PL_multi_end = CopLINE(PL_curcop);
6341 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6342 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6343 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6345 SvREFCNT_dec(herewas);
6346 PL_lex_stuff = tmpstr;
6347 yylval.ival = op_type;
6352 takes: current position in input buffer
6353 returns: new position in input buffer
6354 side-effects: yylval and lex_op are set.
6359 <FH> read from filehandle
6360 <pkg::FH> read from package qualified filehandle
6361 <pkg'FH> read from package qualified filehandle
6362 <$fh> read from filehandle in $fh
6368 S_scan_inputsymbol(pTHX_ char *start)
6370 register char *s = start; /* current position in buffer */
6376 d = PL_tokenbuf; /* start of temp holding space */
6377 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6378 end = strchr(s, '\n');
6381 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6383 /* die if we didn't have space for the contents of the <>,
6384 or if it didn't end, or if we see a newline
6387 if (len >= sizeof PL_tokenbuf)
6388 Perl_croak(aTHX_ "Excessively long <> operator");
6390 Perl_croak(aTHX_ "Unterminated <> operator");
6395 Remember, only scalar variables are interpreted as filehandles by
6396 this code. Anything more complex (e.g., <$fh{$num}>) will be
6397 treated as a glob() call.
6398 This code makes use of the fact that except for the $ at the front,
6399 a scalar variable and a filehandle look the same.
6401 if (*d == '$' && d[1]) d++;
6403 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6404 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6407 /* If we've tried to read what we allow filehandles to look like, and
6408 there's still text left, then it must be a glob() and not a getline.
6409 Use scan_str to pull out the stuff between the <> and treat it
6410 as nothing more than a string.
6413 if (d - PL_tokenbuf != len) {
6414 yylval.ival = OP_GLOB;
6416 s = scan_str(start,FALSE,FALSE);
6418 Perl_croak(aTHX_ "Glob not terminated");
6422 /* we're in a filehandle read situation */
6425 /* turn <> into <ARGV> */
6427 (void)strcpy(d,"ARGV");
6429 /* if <$fh>, create the ops to turn the variable into a
6435 /* try to find it in the pad for this block, otherwise find
6436 add symbol table ops
6438 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6439 OP *o = newOP(OP_PADSV, 0);
6441 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6444 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6445 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6446 newUNOP(OP_RV2SV, 0,
6447 newGVOP(OP_GV, 0, gv)));
6449 PL_lex_op->op_flags |= OPf_SPECIAL;
6450 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6451 yylval.ival = OP_NULL;
6454 /* If it's none of the above, it must be a literal filehandle
6455 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6457 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6458 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6459 yylval.ival = OP_NULL;
6468 takes: start position in buffer
6469 keep_quoted preserve \ on the embedded delimiter(s)
6470 keep_delims preserve the delimiters around the string
6471 returns: position to continue reading from buffer
6472 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6473 updates the read buffer.
6475 This subroutine pulls a string out of the input. It is called for:
6476 q single quotes q(literal text)
6477 ' single quotes 'literal text'
6478 qq double quotes qq(interpolate $here please)
6479 " double quotes "interpolate $here please"
6480 qx backticks qx(/bin/ls -l)
6481 ` backticks `/bin/ls -l`
6482 qw quote words @EXPORT_OK = qw( func() $spam )
6483 m// regexp match m/this/
6484 s/// regexp substitute s/this/that/
6485 tr/// string transliterate tr/this/that/
6486 y/// string transliterate y/this/that/
6487 ($*@) sub prototypes sub foo ($)
6488 (stuff) sub attr parameters sub foo : attr(stuff)
6489 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6491 In most of these cases (all but <>, patterns and transliterate)
6492 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6493 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6494 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6497 It skips whitespace before the string starts, and treats the first
6498 character as the delimiter. If the delimiter is one of ([{< then
6499 the corresponding "close" character )]}> is used as the closing
6500 delimiter. It allows quoting of delimiters, and if the string has
6501 balanced delimiters ([{<>}]) it allows nesting.
6503 The lexer always reads these strings into lex_stuff, except in the
6504 case of the operators which take *two* arguments (s/// and tr///)
6505 when it checks to see if lex_stuff is full (presumably with the 1st
6506 arg to s or tr) and if so puts the string into lex_repl.
6511 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6514 SV *sv; /* scalar value: string */
6515 char *tmps; /* temp string, used for delimiter matching */
6516 register char *s = start; /* current position in the buffer */
6517 register char term; /* terminating character */
6518 register char *to; /* current position in the sv's data */
6519 I32 brackets = 1; /* bracket nesting level */
6520 bool has_utf = FALSE; /* is there any utf8 content? */
6522 /* skip space before the delimiter */
6526 /* mark where we are, in case we need to report errors */
6529 /* after skipping whitespace, the next character is the terminator */
6531 if ((term & 0x80) && UTF)
6534 /* mark where we are */
6535 PL_multi_start = CopLINE(PL_curcop);
6536 PL_multi_open = term;
6538 /* find corresponding closing delimiter */
6539 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6541 PL_multi_close = term;
6543 /* create a new SV to hold the contents. 87 is leak category, I'm
6544 assuming. 79 is the SV's initial length. What a random number. */
6546 sv_upgrade(sv, SVt_PVIV);
6548 (void)SvPOK_only(sv); /* validate pointer */
6550 /* move past delimiter and try to read a complete string */
6552 sv_catpvn(sv, s, 1);
6555 /* extend sv if need be */
6556 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6557 /* set 'to' to the next character in the sv's string */
6558 to = SvPVX(sv)+SvCUR(sv);
6560 /* if open delimiter is the close delimiter read unbridle */
6561 if (PL_multi_open == PL_multi_close) {
6562 for (; s < PL_bufend; s++,to++) {
6563 /* embedded newlines increment the current line number */
6564 if (*s == '\n' && !PL_rsfp)
6565 CopLINE_inc(PL_curcop);
6566 /* handle quoted delimiters */
6567 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6568 if (!keep_quoted && s[1] == term)
6570 /* any other quotes are simply copied straight through */
6574 /* terminate when run out of buffer (the for() condition), or
6575 have found the terminator */
6576 else if (*s == term)
6578 else if (!has_utf && (*s & 0x80) && UTF)
6584 /* if the terminator isn't the same as the start character (e.g.,
6585 matched brackets), we have to allow more in the quoting, and
6586 be prepared for nested brackets.
6589 /* read until we run out of string, or we find the terminator */
6590 for (; s < PL_bufend; s++,to++) {
6591 /* embedded newlines increment the line count */
6592 if (*s == '\n' && !PL_rsfp)
6593 CopLINE_inc(PL_curcop);
6594 /* backslashes can escape the open or closing characters */
6595 if (*s == '\\' && s+1 < PL_bufend) {
6597 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6602 /* allow nested opens and closes */
6603 else if (*s == PL_multi_close && --brackets <= 0)
6605 else if (*s == PL_multi_open)
6607 else if (!has_utf && (*s & 0x80) && UTF)
6612 /* terminate the copied string and update the sv's end-of-string */
6614 SvCUR_set(sv, to - SvPVX(sv));
6617 * this next chunk reads more into the buffer if we're not done yet
6621 break; /* handle case where we are done yet :-) */
6623 #ifndef PERL_STRICT_CR
6624 if (to - SvPVX(sv) >= 2) {
6625 if ((to[-2] == '\r' && to[-1] == '\n') ||
6626 (to[-2] == '\n' && to[-1] == '\r'))
6630 SvCUR_set(sv, to - SvPVX(sv));
6632 else if (to[-1] == '\r')
6635 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6639 /* if we're out of file, or a read fails, bail and reset the current
6640 line marker so we can report where the unterminated string began
6643 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6645 CopLINE_set(PL_curcop, PL_multi_start);
6648 /* we read a line, so increment our line counter */
6649 CopLINE_inc(PL_curcop);
6651 /* update debugger info */
6652 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6653 SV *sv = NEWSV(88,0);
6655 sv_upgrade(sv, SVt_PVMG);
6656 sv_setsv(sv,PL_linestr);
6657 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6660 /* having changed the buffer, we must update PL_bufend */
6661 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6664 /* at this point, we have successfully read the delimited string */
6667 sv_catpvn(sv, s, 1);
6670 PL_multi_end = CopLINE(PL_curcop);
6673 /* if we allocated too much space, give some back */
6674 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6675 SvLEN_set(sv, SvCUR(sv) + 1);
6676 Renew(SvPVX(sv), SvLEN(sv), char);
6679 /* decide whether this is the first or second quoted string we've read
6692 takes: pointer to position in buffer
6693 returns: pointer to new position in buffer
6694 side-effects: builds ops for the constant in yylval.op
6696 Read a number in any of the formats that Perl accepts:
6698 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6699 [\d_]+(\.[\d_]*)?[Ee](\d+)
6701 Underbars (_) are allowed in decimal numbers. If -w is on,
6702 underbars before a decimal point must be at three digit intervals.
6704 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6707 If it reads a number without a decimal point or an exponent, it will
6708 try converting the number to an integer and see if it can do so
6709 without loss of precision.
6713 Perl_scan_num(pTHX_ char *start)
6715 register char *s = start; /* current position in buffer */
6716 register char *d; /* destination in temp buffer */
6717 register char *e; /* end of temp buffer */
6718 NV nv; /* number read, as a double */
6719 SV *sv = Nullsv; /* place to put the converted number */
6720 bool floatit; /* boolean: int or float? */
6721 char *lastub = 0; /* position of last underbar */
6722 static char number_too_long[] = "Number too long";
6724 /* We use the first character to decide what type of number this is */
6728 Perl_croak(aTHX_ "panic: scan_num");
6730 /* if it starts with a 0, it could be an octal number, a decimal in
6731 0.13 disguise, or a hexadecimal number, or a binary number. */
6735 u holds the "number so far"
6736 shift the power of 2 of the base
6737 (hex == 4, octal == 3, binary == 1)
6738 overflowed was the number more than we can hold?
6740 Shift is used when we add a digit. It also serves as an "are
6741 we in octal/hex/binary?" indicator to disallow hex characters
6748 bool overflowed = FALSE;
6749 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6750 static char* bases[5] = { "", "binary", "", "octal",
6752 static char* Bases[5] = { "", "Binary", "", "Octal",
6754 static char *maxima[5] = { "",
6755 "0b11111111111111111111111111111111",
6759 char *base, *Base, *max;
6765 } else if (s[1] == 'b') {
6769 /* check for a decimal in disguise */
6770 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6772 /* so it must be octal */
6776 base = bases[shift];
6777 Base = Bases[shift];
6778 max = maxima[shift];
6780 /* read the rest of the number */
6782 /* x is used in the overflow test,
6783 b is the digit we're adding on. */
6788 /* if we don't mention it, we're done */
6797 /* 8 and 9 are not octal */
6800 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6804 case '2': case '3': case '4':
6805 case '5': case '6': case '7':
6807 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6811 b = *s++ & 15; /* ASCII digit -> value of digit */
6815 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6816 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6817 /* make sure they said 0x */
6822 /* Prepare to put the digit we have onto the end
6823 of the number so far. We check for overflows.
6828 x = u << shift; /* make room for the digit */
6830 if ((x >> shift) != u
6831 && !(PL_hints & HINT_NEW_BINARY)) {
6835 if (ckWARN_d(WARN_OVERFLOW))
6836 Perl_warner(aTHX_ WARN_OVERFLOW,
6837 "Integer overflow in %s number",
6840 u = x | b; /* add the digit to the end */
6843 n *= nvshift[shift];
6844 /* If an NV has not enough bits in its
6845 * mantissa to represent an UV this summing of
6846 * small low-order numbers is a waste of time
6847 * (because the NV cannot preserve the
6848 * low-order bits anyway): we could just
6849 * remember when did we overflow and in the
6850 * end just multiply n by the right
6858 /* if we get here, we had success: make a scalar value from
6865 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6866 Perl_warner(aTHX_ WARN_PORTABLE,
6867 "%s number > %s non-portable",
6874 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6875 Perl_warner(aTHX_ WARN_PORTABLE,
6876 "%s number > %s non-portable",
6881 if (PL_hints & HINT_NEW_BINARY)
6882 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6887 handle decimal numbers.
6888 we're also sent here when we read a 0 as the first digit
6890 case '1': case '2': case '3': case '4': case '5':
6891 case '6': case '7': case '8': case '9': case '.':
6894 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
6897 /* read next group of digits and _ and copy into d */
6898 while (isDIGIT(*s) || *s == '_') {
6899 /* skip underscores, checking for misplaced ones
6903 dTHR; /* only for ckWARN */
6904 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
6905 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6909 /* check for end of fixed-length buffer */
6911 Perl_croak(aTHX_ number_too_long);
6912 /* if we're ok, copy the character */
6917 /* final misplaced underbar check */
6918 if (lastub && s - lastub != 3) {
6920 if (ckWARN(WARN_SYNTAX))
6921 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6924 /* read a decimal portion if there is one. avoid
6925 3..5 being interpreted as the number 3. followed
6928 if (*s == '.' && s[1] != '.') {
6932 /* copy, ignoring underbars, until we run out of
6933 digits. Note: no misplaced underbar checks!
6935 for (; isDIGIT(*s) || *s == '_'; s++) {
6936 /* fixed length buffer check */
6938 Perl_croak(aTHX_ number_too_long);
6942 if (*s == '.' && isDIGIT(s[1])) {
6943 /* oops, it's really a v-string, but without the "v" */
6949 /* read exponent part, if present */
6950 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
6954 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
6955 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
6957 /* allow positive or negative exponent */
6958 if (*s == '+' || *s == '-')
6961 /* read digits of exponent (no underbars :-) */
6962 while (isDIGIT(*s)) {
6964 Perl_croak(aTHX_ number_too_long);
6969 /* terminate the string */
6972 /* make an sv from the string */
6975 #if defined(Strtol) && defined(Strtoul)
6978 strtol/strtoll sets errno to ERANGE if the number is too big
6979 for an integer. We try to do an integer conversion first
6980 if no characters indicating "float" have been found.
6987 if (*PL_tokenbuf == '-')
6988 iv = Strtol(PL_tokenbuf, (char**)NULL, 10);
6990 uv = Strtoul(PL_tokenbuf, (char**)NULL, 10);
6992 floatit = TRUE; /* Probably just too large. */
6993 else if (*PL_tokenbuf == '-')
6995 else if (uv <= IV_MAX)
6996 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7001 nv = Atof(PL_tokenbuf);
7006 No working strtou?ll?.
7008 Unfortunately atol() doesn't do range checks (returning
7009 LONG_MIN/LONG_MAX, and setting errno to ERANGE on overflows)
7010 everywhere [1], so we cannot use use atol() (or atoll()).
7011 If we could, they would be used, as Atol(), very much like
7012 Strtol() and Strtoul() are used above.
7014 [1] XXX Configure test needed to check for atol()
7015 (and atoll()) overflow behaviour XXX
7019 We need to do this the hard way. */
7021 nv = Atof(PL_tokenbuf);
7023 /* See if we can make do with an integer value without loss of
7024 precision. We use U_V to cast to a UV, because some
7025 compilers have issues. Then we try casting it back and see
7026 if it was the same [1]. We only do this if we know we
7027 specifically read an integer. If floatit is true, then we
7028 don't need to do the conversion at all.
7030 [1] Note that this is lossy if our NVs cannot preserve our
7031 UVs. There are metaconfig defines NV_PRESERVES_UV (a boolean)
7032 and NV_PRESERVES_UV_BITS (a number), but in general we really
7033 do hope all such potentially lossy platforms have strtou?ll?
7034 to do a lossless IV/UV conversion.
7036 Maybe could do some tricks with DBL_DIG, LDBL_DIG and
7037 DBL_MANT_DIG and LDBL_MANT_DIG (these are already available
7038 as NV_DIG and NV_MANT_DIG)?
7044 if (!floatit && (NV)uv == nv) {
7046 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7054 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7055 (PL_hints & HINT_NEW_INTEGER) )
7056 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7057 (floatit ? "float" : "integer"),
7061 /* if it starts with a v, it could be a v-string */
7067 while (isDIGIT(*pos) || *pos == '_')
7069 if (!isALPHA(*pos)) {
7071 U8 tmpbuf[UTF8_MAXLEN];
7074 s++; /* get past 'v' */
7077 sv_setpvn(sv, "", 0);
7080 if (*s == '0' && isDIGIT(s[1]))
7081 yyerror("Octal number in vector unsupported");
7084 /* this is atoi() that tolerates underscores */
7087 while (--end >= s) {
7092 rev += (*end - '0') * mult;
7094 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
7095 Perl_warner(aTHX_ WARN_OVERFLOW,
7096 "Integer overflow in decimal number");
7099 tmpend = uv_to_utf8(tmpbuf, rev);
7100 utf8 = utf8 || rev > 127;
7101 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
7102 if (*pos == '.' && isDIGIT(pos[1]))
7108 while (isDIGIT(*pos) || *pos == '_')
7116 sv_utf8_downgrade(sv, TRUE);
7123 /* make the op for the constant and return */
7126 yylval.opval = newSVOP(OP_CONST, 0, sv);
7128 yylval.opval = Nullop;
7134 S_scan_formline(pTHX_ register char *s)
7139 SV *stuff = newSVpvn("",0);
7140 bool needargs = FALSE;
7143 if (*s == '.' || *s == /*{*/'}') {
7145 #ifdef PERL_STRICT_CR
7146 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7148 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7150 if (*t == '\n' || t == PL_bufend)
7153 if (PL_in_eval && !PL_rsfp) {
7154 eol = strchr(s,'\n');
7159 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7161 for (t = s; t < eol; t++) {
7162 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7164 goto enough; /* ~~ must be first line in formline */
7166 if (*t == '@' || *t == '^')
7169 sv_catpvn(stuff, s, eol-s);
7170 #ifndef PERL_STRICT_CR
7171 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7172 char *end = SvPVX(stuff) + SvCUR(stuff);
7181 s = filter_gets(PL_linestr, PL_rsfp, 0);
7182 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7183 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7186 yyerror("Format not terminated");
7196 PL_lex_state = LEX_NORMAL;
7197 PL_nextval[PL_nexttoke].ival = 0;
7201 PL_lex_state = LEX_FORMLINE;
7202 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7204 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7208 SvREFCNT_dec(stuff);
7209 PL_lex_formbrack = 0;
7220 PL_cshlen = strlen(PL_cshname);
7225 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7228 I32 oldsavestack_ix = PL_savestack_ix;
7229 CV* outsidecv = PL_compcv;
7233 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7235 SAVEI32(PL_subline);
7236 save_item(PL_subname);
7239 SAVESPTR(PL_comppad_name);
7240 SAVESPTR(PL_compcv);
7241 SAVEI32(PL_comppad_name_fill);
7242 SAVEI32(PL_min_intro_pending);
7243 SAVEI32(PL_max_intro_pending);
7244 SAVEI32(PL_pad_reset_pending);
7246 PL_compcv = (CV*)NEWSV(1104,0);
7247 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7248 CvFLAGS(PL_compcv) |= flags;
7250 PL_comppad = newAV();
7251 av_push(PL_comppad, Nullsv);
7252 PL_curpad = AvARRAY(PL_comppad);
7253 PL_comppad_name = newAV();
7254 PL_comppad_name_fill = 0;
7255 PL_min_intro_pending = 0;
7257 PL_subline = CopLINE(PL_curcop);
7259 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7260 PL_curpad[0] = (SV*)newAV();
7261 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7262 #endif /* USE_THREADS */
7264 comppadlist = newAV();
7265 AvREAL_off(comppadlist);
7266 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7267 av_store(comppadlist, 1, (SV*)PL_comppad);
7269 CvPADLIST(PL_compcv) = comppadlist;
7270 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7272 CvOWNER(PL_compcv) = 0;
7273 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7274 MUTEX_INIT(CvMUTEXP(PL_compcv));
7275 #endif /* USE_THREADS */
7277 return oldsavestack_ix;
7281 Perl_yywarn(pTHX_ char *s)
7284 PL_in_eval |= EVAL_WARNONLY;
7286 PL_in_eval &= ~EVAL_WARNONLY;
7291 Perl_yyerror(pTHX_ char *s)
7295 char *context = NULL;
7299 if (!yychar || (yychar == ';' && !PL_rsfp))
7301 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7302 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7303 while (isSPACE(*PL_oldoldbufptr))
7305 context = PL_oldoldbufptr;
7306 contlen = PL_bufptr - PL_oldoldbufptr;
7308 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7309 PL_oldbufptr != PL_bufptr) {
7310 while (isSPACE(*PL_oldbufptr))
7312 context = PL_oldbufptr;
7313 contlen = PL_bufptr - PL_oldbufptr;
7315 else if (yychar > 255)
7316 where = "next token ???";
7317 #ifdef USE_PURE_BISON
7318 /* GNU Bison sets the value -2 */
7319 else if (yychar == -2) {
7321 else if ((yychar & 127) == 127) {
7323 if (PL_lex_state == LEX_NORMAL ||
7324 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7325 where = "at end of line";
7326 else if (PL_lex_inpat)
7327 where = "within pattern";
7329 where = "within string";
7332 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7334 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7335 else if (isPRINT_LC(yychar))
7336 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7338 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7339 where = SvPVX(where_sv);
7341 msg = sv_2mortal(newSVpv(s, 0));
7342 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7343 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7345 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7347 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7348 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7349 Perl_sv_catpvf(aTHX_ msg,
7350 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7351 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7354 if (PL_in_eval & EVAL_WARNONLY)
7355 Perl_warn(aTHX_ "%"SVf, msg);
7358 if (PL_error_count >= 10) {
7359 if (PL_in_eval && SvCUR(ERRSV))
7360 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
7361 ERRSV, CopFILE(PL_curcop));
7363 Perl_croak(aTHX_ "%s has too many errors.\n",
7364 CopFILE(PL_curcop));
7367 PL_in_my_stash = Nullhv;
7372 S_swallow_bom(pTHX_ U8 *s)
7375 slen = SvCUR(PL_linestr);
7379 /* UTF-16 little-endian */
7380 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
7381 Perl_croak(aTHX_ "Unsupported script encoding");
7382 #ifndef PERL_NO_UTF16_FILTER
7383 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-LE script encoding\n"));
7385 if (PL_bufend > (char*)s) {
7389 filter_add(utf16rev_textfilter, NULL);
7390 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7391 PL_bufend = (char*)utf16_to_utf8_reversed(s, news,
7392 PL_bufend - (char*)s - 1,
7394 Copy(news, s, newlen, U8);
7395 SvCUR_set(PL_linestr, newlen);
7396 PL_bufend = SvPVX(PL_linestr) + newlen;
7397 news[newlen++] = '\0';
7401 Perl_croak(aTHX_ "Unsupported script encoding");
7406 if (s[1] == 0xFF) { /* UTF-16 big-endian */
7407 #ifndef PERL_NO_UTF16_FILTER
7408 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding\n"));
7410 if (PL_bufend > (char *)s) {
7414 filter_add(utf16_textfilter, NULL);
7415 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7416 PL_bufend = (char*)utf16_to_utf8(s, news,
7417 PL_bufend - (char*)s,
7419 Copy(news, s, newlen, U8);
7420 SvCUR_set(PL_linestr, newlen);
7421 PL_bufend = SvPVX(PL_linestr) + newlen;
7422 news[newlen++] = '\0';
7426 Perl_croak(aTHX_ "Unsupported script encoding");
7431 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
7432 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-8 script encoding\n"));
7437 if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */
7438 s[2] == 0xFE && s[3] == 0xFF)
7440 Perl_croak(aTHX_ "Unsupported script encoding");
7452 * Restore a source filter.
7456 restore_rsfp(pTHXo_ void *f)
7458 PerlIO *fp = (PerlIO*)f;
7460 if (PL_rsfp == PerlIO_stdin())
7461 PerlIO_clearerr(PL_rsfp);
7462 else if (PL_rsfp && (PL_rsfp != fp))
7463 PerlIO_close(PL_rsfp);
7467 #ifndef PERL_NO_UTF16_FILTER
7469 utf16_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7471 I32 count = FILTER_READ(idx+1, sv, maxlen);
7476 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7477 if (!*SvPV_nolen(sv))
7478 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7481 tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7482 sv_usepvn(sv, (char*)tmps, tend - tmps);
7488 utf16rev_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7490 I32 count = FILTER_READ(idx+1, sv, maxlen);
7495 if (!*SvPV_nolen(sv))
7496 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7499 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7500 tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7501 sv_usepvn(sv, (char*)tmps, tend - tmps);