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() */
84 #define YYMAXLEVEL 100
86 YYSTYPE* yylval_pointer[YYMAXLEVEL];
87 int* yychar_pointer[YYMAXLEVEL];
91 # define yylval (*yylval_pointer[yyactlevel])
92 # define yychar (*yychar_pointer[yyactlevel])
93 # define PERL_YYLEX_PARAM yylval_pointer[yyactlevel],yychar_pointer[yyactlevel]
95 # define yylex() Perl_yylex(aTHX_ yylval_pointer[yyactlevel],yychar_pointer[yyactlevel])
100 /* CLINE is a macro that ensures PL_copline has a sane value */
105 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
108 * Convenience functions to return different tokens and prime the
109 * lexer for the next token. They all take an argument.
111 * TOKEN : generic token (used for '(', DOLSHARP, etc)
112 * OPERATOR : generic operator
113 * AOPERATOR : assignment operator
114 * PREBLOCK : beginning the block after an if, while, foreach, ...
115 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
116 * PREREF : *EXPR where EXPR is not a simple identifier
117 * TERM : expression term
118 * LOOPX : loop exiting command (goto, last, dump, etc)
119 * FTST : file test operator
120 * FUN0 : zero-argument function
121 * FUN1 : not used, except for not, which isn't a UNIOP
122 * BOop : bitwise or or xor
124 * SHop : shift operator
125 * PWop : power operator
126 * PMop : pattern-matching operator
127 * Aop : addition-level operator
128 * Mop : multiplication-level operator
129 * Eop : equality-testing operator
130 * Rop : relational operator <= != gt
132 * Also see LOP and lop() below.
135 #define TOKEN(retval) return (PL_bufptr = s,(int)retval)
136 #define OPERATOR(retval) return (PL_expect = XTERM,PL_bufptr = s,(int)retval)
137 #define AOPERATOR(retval) return ao((PL_expect = XTERM,PL_bufptr = s,(int)retval))
138 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
139 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
140 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s,(int)retval)
141 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR,PL_bufptr = s,(int)retval)
142 #define LOOPX(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
143 #define FTST(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)UNIOP)
144 #define FUN0(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
145 #define FUN1(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
146 #define BOop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
147 #define BAop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
148 #define SHop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
149 #define PWop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
150 #define PMop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
151 #define Aop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
152 #define Mop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
153 #define Eop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
154 #define Rop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)RELOP)
156 /* This bit of chicanery makes a unary function followed by
157 * a parenthesis into a function with one argument, highest precedence.
159 #define UNI(f) return(yylval.ival = f, \
162 PL_last_uni = PL_oldbufptr, \
163 PL_last_lop_op = f, \
164 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
166 #define UNIBRACK(f) return(yylval.ival = f, \
168 PL_last_uni = PL_oldbufptr, \
169 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
171 /* grandfather return to old style */
172 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
177 * This subroutine detects &&= and ||= and turns an ANDAND or OROR
178 * into an OP_ANDASSIGN or OP_ORASSIGN
182 S_ao(pTHX_ int toketype)
184 if (*PL_bufptr == '=') {
186 if (toketype == ANDAND)
187 yylval.ival = OP_ANDASSIGN;
188 else if (toketype == OROR)
189 yylval.ival = OP_ORASSIGN;
197 * When Perl expects an operator and finds something else, no_op
198 * prints the warning. It always prints "<something> found where
199 * operator expected. It prints "Missing semicolon on previous line?"
200 * if the surprise occurs at the start of the line. "do you need to
201 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
202 * where the compiler doesn't know if foo is a method call or a function.
203 * It prints "Missing operator before end of line" if there's nothing
204 * after the missing operator, or "... before <...>" if there is something
205 * after the missing operator.
209 S_no_op(pTHX_ char *what, char *s)
211 char *oldbp = PL_bufptr;
212 bool is_first = (PL_oldbufptr == PL_linestart);
218 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
220 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
221 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
223 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
224 if (t < PL_bufptr && isSPACE(*t))
225 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
226 t - PL_oldoldbufptr, PL_oldoldbufptr);
230 Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
237 * Complain about missing quote/regexp/heredoc terminator.
238 * If it's called with (char *)NULL then it cauterizes the line buffer.
239 * If we're in a delimited string and the delimiter is a control
240 * character, it's reformatted into a two-char sequence like ^C.
245 S_missingterm(pTHX_ char *s)
250 char *nl = strrchr(s,'\n');
256 iscntrl(PL_multi_close)
258 PL_multi_close < 32 || PL_multi_close == 127
262 tmpbuf[1] = toCTRL(PL_multi_close);
268 *tmpbuf = PL_multi_close;
272 q = strchr(s,'"') ? '\'' : '"';
273 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
281 Perl_deprecate(pTHX_ char *s)
284 if (ckWARN(WARN_DEPRECATED))
285 Perl_warner(aTHX_ WARN_DEPRECATED, "Use of %s is deprecated", s);
290 * Deprecate a comma-less variable list.
296 deprecate("comma-less variable list");
300 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
301 * utf16-to-utf8-reversed.
304 #ifdef PERL_CR_FILTER
308 register char *s = SvPVX(sv);
309 register char *e = s + SvCUR(sv);
310 /* outer loop optimized to do nothing if there are no CR-LFs */
312 if (*s++ == '\r' && *s == '\n') {
313 /* hit a CR-LF, need to copy the rest */
314 register char *d = s - 1;
317 if (*s == '\r' && s[1] == '\n')
328 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
330 I32 count = FILTER_READ(idx+1, sv, maxlen);
331 if (count > 0 && !maxlen)
339 * Initialize variables. Uses the Perl save_stack to save its state (for
340 * recursive calls to the parser).
344 Perl_lex_start(pTHX_ SV *line)
350 SAVEI32(PL_lex_dojoin);
351 SAVEI32(PL_lex_brackets);
352 SAVEI32(PL_lex_casemods);
353 SAVEI32(PL_lex_starts);
354 SAVEI32(PL_lex_state);
355 SAVEVPTR(PL_lex_inpat);
356 SAVEI32(PL_lex_inwhat);
357 if (PL_lex_state == LEX_KNOWNEXT) {
358 I32 toke = PL_nexttoke;
359 while (--toke >= 0) {
360 SAVEI32(PL_nexttype[toke]);
361 SAVEVPTR(PL_nextval[toke]);
363 SAVEI32(PL_nexttoke);
365 SAVECOPLINE(PL_curcop);
368 SAVEPPTR(PL_oldbufptr);
369 SAVEPPTR(PL_oldoldbufptr);
370 SAVEPPTR(PL_linestart);
371 SAVESPTR(PL_linestr);
372 SAVEPPTR(PL_lex_brackstack);
373 SAVEPPTR(PL_lex_casestack);
374 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
375 SAVESPTR(PL_lex_stuff);
376 SAVEI32(PL_lex_defer);
377 SAVEI32(PL_sublex_info.sub_inwhat);
378 SAVESPTR(PL_lex_repl);
380 SAVEINT(PL_lex_expect);
382 PL_lex_state = LEX_NORMAL;
386 New(899, PL_lex_brackstack, 120, char);
387 New(899, PL_lex_casestack, 12, char);
388 SAVEFREEPV(PL_lex_brackstack);
389 SAVEFREEPV(PL_lex_casestack);
391 *PL_lex_casestack = '\0';
394 PL_lex_stuff = Nullsv;
395 PL_lex_repl = Nullsv;
399 PL_sublex_info.sub_inwhat = 0;
401 if (SvREADONLY(PL_linestr))
402 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
403 s = SvPV(PL_linestr, len);
404 if (len && s[len-1] != ';') {
405 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
406 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
407 sv_catpvn(PL_linestr, "\n;", 2);
409 SvTEMP_off(PL_linestr);
410 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
411 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
413 PL_rs = newSVpvn("\n", 1);
419 * Finalizer for lexing operations. Must be called when the parser is
420 * done with the lexer.
426 PL_doextract = FALSE;
431 * This subroutine has nothing to do with tilting, whether at windmills
432 * or pinball tables. Its name is short for "increment line". It
433 * increments the current line number in CopLINE(PL_curcop) and checks
434 * to see whether the line starts with a comment of the form
435 * # line 500 "foo.pm"
436 * If so, it sets the current line number and file to the values in the comment.
440 S_incline(pTHX_ char *s)
448 CopLINE_inc(PL_curcop);
451 while (SPACE_OR_TAB(*s)) s++;
452 if (strnEQ(s, "line", 4))
456 if (*s == ' ' || *s == '\t')
460 while (SPACE_OR_TAB(*s)) s++;
466 while (SPACE_OR_TAB(*s))
468 if (*s == '"' && (t = strchr(s+1, '"'))) {
473 for (t = s; !isSPACE(*t); t++) ;
476 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
478 if (*e != '\n' && *e != '\0')
479 return; /* false alarm */
485 Safefree(CopFILE(PL_curcop));
487 SvREFCNT_dec(CopFILEGV(PL_curcop));
489 CopFILE_set(PL_curcop, s);
492 CopLINE_set(PL_curcop, atoi(n)-1);
497 * Called to gobble the appropriate amount and type of whitespace.
498 * Skips comments as well.
502 S_skipspace(pTHX_ register char *s)
505 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
506 while (s < PL_bufend && SPACE_OR_TAB(*s))
512 SSize_t oldprevlen, oldoldprevlen;
513 SSize_t oldloplen, oldunilen;
514 while (s < PL_bufend && isSPACE(*s)) {
515 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
520 if (s < PL_bufend && *s == '#') {
521 while (s < PL_bufend && *s != '\n')
525 if (PL_in_eval && !PL_rsfp) {
532 /* only continue to recharge the buffer if we're at the end
533 * of the buffer, we're not reading from a source filter, and
534 * we're in normal lexing mode
536 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
537 PL_lex_state == LEX_FORMLINE)
540 /* try to recharge the buffer */
541 if ((s = filter_gets(PL_linestr, PL_rsfp,
542 (prevlen = SvCUR(PL_linestr)))) == Nullch)
544 /* end of file. Add on the -p or -n magic */
545 if (PL_minus_n || PL_minus_p) {
546 sv_setpv(PL_linestr,PL_minus_p ?
547 ";}continue{print or die qq(-p destination: $!\\n)" :
549 sv_catpv(PL_linestr,";}");
550 PL_minus_n = PL_minus_p = 0;
553 sv_setpv(PL_linestr,";");
555 /* reset variables for next time we lex */
556 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
558 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
560 /* Close the filehandle. Could be from -P preprocessor,
561 * STDIN, or a regular file. If we were reading code from
562 * STDIN (because the commandline held no -e or filename)
563 * then we don't close it, we reset it so the code can
564 * read from STDIN too.
567 if (PL_preprocess && !PL_in_eval)
568 (void)PerlProc_pclose(PL_rsfp);
569 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
570 PerlIO_clearerr(PL_rsfp);
572 (void)PerlIO_close(PL_rsfp);
577 /* not at end of file, so we only read another line */
578 /* make corresponding updates to old pointers, for yyerror() */
579 oldprevlen = PL_oldbufptr - PL_bufend;
580 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
582 oldunilen = PL_last_uni - PL_bufend;
584 oldloplen = PL_last_lop - PL_bufend;
585 PL_linestart = PL_bufptr = s + prevlen;
586 PL_bufend = s + SvCUR(PL_linestr);
588 PL_oldbufptr = s + oldprevlen;
589 PL_oldoldbufptr = s + oldoldprevlen;
591 PL_last_uni = s + oldunilen;
593 PL_last_lop = s + oldloplen;
596 /* debugger active and we're not compiling the debugger code,
597 * so store the line into the debugger's array of lines
599 if (PERLDB_LINE && PL_curstash != PL_debstash) {
600 SV *sv = NEWSV(85,0);
602 sv_upgrade(sv, SVt_PVMG);
603 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
604 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
611 * Check the unary operators to ensure there's no ambiguity in how they're
612 * used. An ambiguous piece of code would be:
614 * This doesn't mean rand() + 5. Because rand() is a unary operator,
615 * the +5 is its argument.
625 if (PL_oldoldbufptr != PL_last_uni)
627 while (isSPACE(*PL_last_uni))
629 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
630 if ((t = strchr(s, '(')) && t < PL_bufptr)
632 if (ckWARN_d(WARN_AMBIGUOUS)){
635 Perl_warner(aTHX_ WARN_AMBIGUOUS,
636 "Warning: Use of \"%s\" without parens is ambiguous",
642 /* workaround to replace the UNI() macro with a function. Only the
643 * hints/uts.sh file mentions this. Other comments elsewhere in the
644 * source indicate Microport Unix might need it too.
650 #define UNI(f) return uni(f,s)
653 S_uni(pTHX_ I32 f, char *s)
658 PL_last_uni = PL_oldbufptr;
669 #endif /* CRIPPLED_CC */
672 * LOP : macro to build a list operator. Its behaviour has been replaced
673 * with a subroutine, S_lop() for which LOP is just another name.
676 #define LOP(f,x) return lop(f,x,s)
680 * Build a list operator (or something that might be one). The rules:
681 * - if we have a next token, then it's a list operator [why?]
682 * - if the next thing is an opening paren, then it's a function
683 * - else it's a list operator
687 S_lop(pTHX_ I32 f, int x, char *s)
694 PL_last_lop = PL_oldbufptr;
709 * When the lexer realizes it knows the next token (for instance,
710 * it is reordering tokens for the parser) then it can call S_force_next
711 * to know what token to return the next time the lexer is called. Caller
712 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
713 * handles the token correctly.
717 S_force_next(pTHX_ I32 type)
719 PL_nexttype[PL_nexttoke] = type;
721 if (PL_lex_state != LEX_KNOWNEXT) {
722 PL_lex_defer = PL_lex_state;
723 PL_lex_expect = PL_expect;
724 PL_lex_state = LEX_KNOWNEXT;
730 * When the lexer knows the next thing is a word (for instance, it has
731 * just seen -> and it knows that the next char is a word char, then
732 * it calls S_force_word to stick the next word into the PL_next lookahead.
735 * char *start : buffer position (must be within PL_linestr)
736 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
737 * int check_keyword : if true, Perl checks to make sure the word isn't
738 * a keyword (do this if the word is a label, e.g. goto FOO)
739 * int allow_pack : if true, : characters will also be allowed (require,
741 * int allow_initial_tick : used by the "sub" lexer only.
745 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
750 start = skipspace(start);
752 if (isIDFIRST_lazy_if(s,UTF) ||
753 (allow_pack && *s == ':') ||
754 (allow_initial_tick && *s == '\'') )
756 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
757 if (check_keyword && keyword(PL_tokenbuf, len))
759 if (token == METHOD) {
764 PL_expect = XOPERATOR;
767 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
768 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
776 * Called when the lexer wants $foo *foo &foo etc, but the program
777 * text only contains the "foo" portion. The first argument is a pointer
778 * to the "foo", and the second argument is the type symbol to prefix.
779 * Forces the next token to be a "WORD".
780 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
784 S_force_ident(pTHX_ register char *s, int kind)
787 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
788 PL_nextval[PL_nexttoke].opval = o;
791 dTHR; /* just for in_eval */
792 o->op_private = OPpCONST_ENTERED;
793 /* XXX see note in pp_entereval() for why we forgo typo
794 warnings if the symbol must be introduced in an eval.
796 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
797 kind == '$' ? SVt_PV :
798 kind == '@' ? SVt_PVAV :
799 kind == '%' ? SVt_PVHV :
807 Perl_str_to_version(pTHX_ SV *sv)
812 char *start = SvPVx(sv,len);
813 bool utf = SvUTF8(sv) ? TRUE : FALSE;
814 char *end = start + len;
815 while (start < end) {
819 n = utf8_to_uv_chk((U8*)start, &skip, 0);
824 retval += ((NV)n)/nshift;
833 * Forces the next token to be a version number.
837 S_force_version(pTHX_ char *s)
839 OP *version = Nullop;
848 for (; isDIGIT(*d) || *d == '_' || *d == '.'; d++);
849 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
852 version = yylval.opval;
853 ver = cSVOPx(version)->op_sv;
854 if (SvPOK(ver) && !SvNIOK(ver)) {
855 (void)SvUPGRADE(ver, SVt_PVNV);
856 SvNVX(ver) = str_to_version(ver);
857 SvNOK_on(ver); /* hint that it is a version */
862 /* NOTE: The parser sees the package name and the VERSION swapped */
863 PL_nextval[PL_nexttoke].opval = version;
871 * Tokenize a quoted string passed in as an SV. It finds the next
872 * chunk, up to end of string or a backslash. It may make a new
873 * SV containing that chunk (if HINT_NEW_STRING is on). It also
878 S_tokeq(pTHX_ SV *sv)
889 s = SvPV_force(sv, len);
890 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
893 while (s < send && *s != '\\')
898 if ( PL_hints & HINT_NEW_STRING )
899 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
902 if (s + 1 < send && (s[1] == '\\'))
903 s++; /* all that, just for this */
908 SvCUR_set(sv, d - SvPVX(sv));
910 if ( PL_hints & HINT_NEW_STRING )
911 return new_constant(NULL, 0, "q", sv, pv, "q");
916 * Now come three functions related to double-quote context,
917 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
918 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
919 * interact with PL_lex_state, and create fake ( ... ) argument lists
920 * to handle functions and concatenation.
921 * They assume that whoever calls them will be setting up a fake
922 * join call, because each subthing puts a ',' after it. This lets
925 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
927 * (I'm not sure whether the spurious commas at the end of lcfirst's
928 * arguments and join's arguments are created or not).
933 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
935 * Pattern matching will set PL_lex_op to the pattern-matching op to
936 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
938 * OP_CONST and OP_READLINE are easy--just make the new op and return.
940 * Everything else becomes a FUNC.
942 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
943 * had an OP_CONST or OP_READLINE). This just sets us up for a
944 * call to S_sublex_push().
950 register I32 op_type = yylval.ival;
952 if (op_type == OP_NULL) {
953 yylval.opval = PL_lex_op;
957 if (op_type == OP_CONST || op_type == OP_READLINE) {
958 SV *sv = tokeq(PL_lex_stuff);
960 if (SvTYPE(sv) == SVt_PVIV) {
961 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
967 nsv = newSVpvn(p, len);
973 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
974 PL_lex_stuff = Nullsv;
978 PL_sublex_info.super_state = PL_lex_state;
979 PL_sublex_info.sub_inwhat = op_type;
980 PL_sublex_info.sub_op = PL_lex_op;
981 PL_lex_state = LEX_INTERPPUSH;
985 yylval.opval = PL_lex_op;
995 * Create a new scope to save the lexing state. The scope will be
996 * ended in S_sublex_done. Returns a '(', starting the function arguments
997 * to the uc, lc, etc. found before.
998 * Sets PL_lex_state to LEX_INTERPCONCAT.
1007 PL_lex_state = PL_sublex_info.super_state;
1008 SAVEI32(PL_lex_dojoin);
1009 SAVEI32(PL_lex_brackets);
1010 SAVEI32(PL_lex_casemods);
1011 SAVEI32(PL_lex_starts);
1012 SAVEI32(PL_lex_state);
1013 SAVEVPTR(PL_lex_inpat);
1014 SAVEI32(PL_lex_inwhat);
1015 SAVECOPLINE(PL_curcop);
1016 SAVEPPTR(PL_bufptr);
1017 SAVEPPTR(PL_oldbufptr);
1018 SAVEPPTR(PL_oldoldbufptr);
1019 SAVEPPTR(PL_linestart);
1020 SAVESPTR(PL_linestr);
1021 SAVEPPTR(PL_lex_brackstack);
1022 SAVEPPTR(PL_lex_casestack);
1024 PL_linestr = PL_lex_stuff;
1025 PL_lex_stuff = Nullsv;
1027 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1028 = SvPVX(PL_linestr);
1029 PL_bufend += SvCUR(PL_linestr);
1030 SAVEFREESV(PL_linestr);
1032 PL_lex_dojoin = FALSE;
1033 PL_lex_brackets = 0;
1034 New(899, PL_lex_brackstack, 120, char);
1035 New(899, PL_lex_casestack, 12, char);
1036 SAVEFREEPV(PL_lex_brackstack);
1037 SAVEFREEPV(PL_lex_casestack);
1038 PL_lex_casemods = 0;
1039 *PL_lex_casestack = '\0';
1041 PL_lex_state = LEX_INTERPCONCAT;
1042 CopLINE_set(PL_curcop, PL_multi_start);
1044 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1045 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1046 PL_lex_inpat = PL_sublex_info.sub_op;
1048 PL_lex_inpat = Nullop;
1055 * Restores lexer state after a S_sublex_push.
1061 if (!PL_lex_starts++) {
1062 PL_expect = XOPERATOR;
1063 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn("",0));
1067 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1068 PL_lex_state = LEX_INTERPCASEMOD;
1072 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1073 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1074 PL_linestr = PL_lex_repl;
1076 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1077 PL_bufend += SvCUR(PL_linestr);
1078 SAVEFREESV(PL_linestr);
1079 PL_lex_dojoin = FALSE;
1080 PL_lex_brackets = 0;
1081 PL_lex_casemods = 0;
1082 *PL_lex_casestack = '\0';
1084 if (SvEVALED(PL_lex_repl)) {
1085 PL_lex_state = LEX_INTERPNORMAL;
1087 /* we don't clear PL_lex_repl here, so that we can check later
1088 whether this is an evalled subst; that means we rely on the
1089 logic to ensure sublex_done() is called again only via the
1090 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1093 PL_lex_state = LEX_INTERPCONCAT;
1094 PL_lex_repl = Nullsv;
1100 PL_bufend = SvPVX(PL_linestr);
1101 PL_bufend += SvCUR(PL_linestr);
1102 PL_expect = XOPERATOR;
1103 PL_sublex_info.sub_inwhat = 0;
1111 Extracts a pattern, double-quoted string, or transliteration. This
1114 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1115 processing a pattern (PL_lex_inpat is true), a transliteration
1116 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1118 Returns a pointer to the character scanned up to. Iff this is
1119 advanced from the start pointer supplied (ie if anything was
1120 successfully parsed), will leave an OP for the substring scanned
1121 in yylval. Caller must intuit reason for not parsing further
1122 by looking at the next characters herself.
1126 double-quoted style: \r and \n
1127 regexp special ones: \D \s
1129 backrefs: \1 (deprecated in substitution replacements)
1130 case and quoting: \U \Q \E
1131 stops on @ and $, but not for $ as tail anchor
1133 In transliterations:
1134 characters are VERY literal, except for - not at the start or end
1135 of the string, which indicates a range. scan_const expands the
1136 range to the full set of intermediate characters.
1138 In double-quoted strings:
1140 double-quoted style: \r and \n
1142 backrefs: \1 (deprecated)
1143 case and quoting: \U \Q \E
1146 scan_const does *not* construct ops to handle interpolated strings.
1147 It stops processing as soon as it finds an embedded $ or @ variable
1148 and leaves it to the caller to work out what's going on.
1150 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1152 $ in pattern could be $foo or could be tail anchor. Assumption:
1153 it's a tail anchor if $ is the last thing in the string, or if it's
1154 followed by one of ")| \n\t"
1156 \1 (backreferences) are turned into $1
1158 The structure of the code is
1159 while (there's a character to process) {
1160 handle transliteration ranges
1161 skip regexp comments
1162 skip # initiated comments in //x patterns
1163 check for embedded @foo
1164 check for embedded scalars
1166 leave intact backslashes from leave (below)
1167 deprecate \1 in strings and sub replacements
1168 handle string-changing backslashes \l \U \Q \E, etc.
1169 switch (what was escaped) {
1170 handle - in a transliteration (becomes a literal -)
1171 handle \132 octal characters
1172 handle 0x15 hex characters
1173 handle \cV (control V)
1174 handle printf backslashes (\f, \r, \n, etc)
1176 } (end if backslash)
1177 } (end while character to read)
1182 S_scan_const(pTHX_ char *start)
1184 register char *send = PL_bufend; /* end of the constant */
1185 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1186 register char *s = start; /* start of the constant */
1187 register char *d = SvPVX(sv); /* destination for copies */
1188 bool dorange = FALSE; /* are we in a translit range? */
1189 bool didrange = FALSE; /* did we just finish a range? */
1190 bool has_utf = FALSE; /* embedded \x{} */
1194 I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1195 ? (PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1197 I32 thisutf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1198 ? (PL_sublex_info.sub_op->op_private & (PL_lex_repl ?
1199 OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF))
1201 const char *leaveit = /* set of acceptably-backslashed characters */
1203 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1206 while (s < send || dorange) {
1207 /* get transliterations out of the way (they're most literal) */
1208 if (PL_lex_inwhat == OP_TRANS) {
1209 /* expand a range A-Z to the full set of characters. AIE! */
1211 I32 i; /* current expanded character */
1212 I32 min; /* first character in range */
1213 I32 max; /* last character in range */
1215 i = d - SvPVX(sv); /* remember current offset */
1216 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1217 d = SvPVX(sv) + i; /* refresh d after realloc */
1218 d -= 2; /* eat the first char and the - */
1220 min = (U8)*d; /* first char in range */
1221 max = (U8)d[1]; /* last char in range */
1225 "Invalid [] range \"%c-%c\" in transliteration operator",
1226 (char)min, (char)max);
1230 if ((isLOWER(min) && isLOWER(max)) ||
1231 (isUPPER(min) && isUPPER(max))) {
1233 for (i = min; i <= max; i++)
1237 for (i = min; i <= max; i++)
1244 for (i = min; i <= max; i++)
1247 /* mark the range as done, and continue */
1253 /* range begins (ignore - as first or last char) */
1254 else if (*s == '-' && s+1 < send && s != start) {
1256 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
1259 *d++ = (char)0xff; /* use illegal utf8 byte--see pmtrans */
1271 /* if we get here, we're not doing a transliteration */
1273 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1274 except for the last char, which will be done separately. */
1275 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1277 while (s < send && *s != ')')
1280 else if (s[2] == '{' /* This should match regcomp.c */
1281 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1284 char *regparse = s + (s[2] == '{' ? 3 : 4);
1287 while (count && (c = *regparse)) {
1288 if (c == '\\' && regparse[1])
1296 if (*regparse != ')') {
1297 regparse--; /* Leave one char for continuation. */
1298 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1300 while (s < regparse)
1305 /* likewise skip #-initiated comments in //x patterns */
1306 else if (*s == '#' && PL_lex_inpat &&
1307 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1308 while (s+1 < send && *s != '\n')
1312 /* check for embedded arrays
1313 (@foo, @:foo, @'foo, @{foo}, @$foo, @+, @-)
1315 else if (*s == '@' && s[1]
1316 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1])))
1319 /* check for embedded scalars. only stop if we're sure it's a
1322 else if (*s == '$') {
1323 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1325 if (s + 1 < send && !strchr("()| \n\t", s[1]))
1326 break; /* in regexp, $ might be tail anchor */
1329 /* (now in tr/// code again) */
1331 if (*s & 0x80 && thisutf) {
1332 (void)utf8_to_uv_chk((U8*)s, &len, 0);
1334 /* illegal UTF8, make it valid */
1335 char *old_pvx = SvPVX(sv);
1336 /* need space for one extra char (NOTE: SvCUR() not set here) */
1337 d = SvGROW(sv, SvLEN(sv) + 1) + (d - old_pvx);
1338 d = (char*)uv_to_utf8((U8*)d, (U8)*s++);
1349 if (*s == '\\' && s+1 < send) {
1352 /* some backslashes we leave behind */
1353 if (*leaveit && *s && strchr(leaveit, *s)) {
1359 /* deprecate \1 in strings and substitution replacements */
1360 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1361 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1363 dTHR; /* only for ckWARN */
1364 if (ckWARN(WARN_SYNTAX))
1365 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1370 /* string-change backslash escapes */
1371 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1376 /* if we get here, it's either a quoted -, or a digit */
1379 /* quoted - in transliterations */
1381 if (PL_lex_inwhat == OP_TRANS) {
1389 if (ckWARN(WARN_MISC) && isALNUM(*s))
1390 Perl_warner(aTHX_ WARN_MISC,
1391 "Unrecognized escape \\%c passed through",
1393 /* default action is to copy the quoted character */
1398 /* \132 indicates an octal constant */
1399 case '0': case '1': case '2': case '3':
1400 case '4': case '5': case '6': case '7':
1401 len = 0; /* disallow underscores */
1402 uv = (UV)scan_oct(s, 3, &len);
1404 goto NUM_ESCAPE_INSERT;
1406 /* \x24 indicates a hex constant */
1410 char* e = strchr(s, '}');
1412 yyerror("Missing right brace on \\x{}");
1415 len = 1; /* allow underscores */
1416 uv = (UV)scan_hex(s + 1, e - s - 1, &len);
1420 len = 0; /* disallow underscores */
1421 uv = (UV)scan_hex(s, 2, &len);
1426 /* Insert oct or hex escaped character.
1427 * There will always enough room in sv since such escapes will
1428 * be longer than any utf8 sequence they can end up as
1431 if (!thisutf && !has_utf && uv > 255) {
1432 /* might need to recode whatever we have accumulated so far
1433 * if it contains any hibit chars
1437 for (c = SvPVX(sv); c < d; c++) {
1442 char *old_pvx = SvPVX(sv);
1444 d = SvGROW(sv, SvCUR(sv) + hicount + 1) + (d - old_pvx);
1453 uv_to_utf8((U8*)dst, (U8)*src--);
1463 if (thisutf || uv > 255) {
1464 d = (char*)uv_to_utf8((U8*)d, uv);
1476 /* \N{latin small letter a} is a named character */
1480 char* e = strchr(s, '}');
1486 yyerror("Missing right brace on \\N{}");
1490 res = newSVpvn(s + 1, e - s - 1);
1491 res = new_constant( Nullch, 0, "charnames",
1492 res, Nullsv, "\\N{...}" );
1493 str = SvPV(res,len);
1494 if (!has_utf && SvUTF8(res)) {
1495 char *ostart = SvPVX(sv);
1496 SvCUR_set(sv, d - ostart);
1499 sv_utf8_upgrade(sv);
1500 /* this just broke our allocation above... */
1501 SvGROW(sv, send - start);
1502 d = SvPVX(sv) + SvCUR(sv);
1505 if (len > e - s + 4) {
1506 char *odest = SvPVX(sv);
1508 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1509 d = SvPVX(sv) + (d - odest);
1511 Copy(str, d, len, char);
1518 yyerror("Missing braces on \\N{}");
1521 /* \c is a control character */
1536 /* printf-style backslashes, formfeeds, newlines, etc */
1554 *d++ = '\047'; /* CP 1047 */
1557 *d++ = '\057'; /* CP 1047 */
1571 } /* end if (backslash) */
1574 } /* while loop to process each character */
1576 /* terminate the string and set up the sv */
1578 SvCUR_set(sv, d - SvPVX(sv));
1583 /* shrink the sv if we allocated more than we used */
1584 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1585 SvLEN_set(sv, SvCUR(sv) + 1);
1586 Renew(SvPVX(sv), SvLEN(sv), char);
1589 /* return the substring (via yylval) only if we parsed anything */
1590 if (s > PL_bufptr) {
1591 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1592 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1594 ( PL_lex_inwhat == OP_TRANS
1596 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1599 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1606 * Returns TRUE if there's more to the expression (e.g., a subscript),
1609 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1611 * ->[ and ->{ return TRUE
1612 * { and [ outside a pattern are always subscripts, so return TRUE
1613 * if we're outside a pattern and it's not { or [, then return FALSE
1614 * if we're in a pattern and the first char is a {
1615 * {4,5} (any digits around the comma) returns FALSE
1616 * if we're in a pattern and the first char is a [
1618 * [SOMETHING] has a funky algorithm to decide whether it's a
1619 * character class or not. It has to deal with things like
1620 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1621 * anything else returns TRUE
1624 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1627 S_intuit_more(pTHX_ register char *s)
1629 if (PL_lex_brackets)
1631 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1633 if (*s != '{' && *s != '[')
1638 /* In a pattern, so maybe we have {n,m}. */
1655 /* On the other hand, maybe we have a character class */
1658 if (*s == ']' || *s == '^')
1661 /* this is terrifying, and it works */
1662 int weight = 2; /* let's weigh the evidence */
1664 unsigned char un_char = 255, last_un_char;
1665 char *send = strchr(s,']');
1666 char tmpbuf[sizeof PL_tokenbuf * 4];
1668 if (!send) /* has to be an expression */
1671 Zero(seen,256,char);
1674 else if (isDIGIT(*s)) {
1676 if (isDIGIT(s[1]) && s[2] == ']')
1682 for (; s < send; s++) {
1683 last_un_char = un_char;
1684 un_char = (unsigned char)*s;
1689 weight -= seen[un_char] * 10;
1690 if (isALNUM_lazy_if(s+1,UTF)) {
1691 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1692 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1697 else if (*s == '$' && s[1] &&
1698 strchr("[#!%*<>()-=",s[1])) {
1699 if (/*{*/ strchr("])} =",s[2]))
1708 if (strchr("wds]",s[1]))
1710 else if (seen['\''] || seen['"'])
1712 else if (strchr("rnftbxcav",s[1]))
1714 else if (isDIGIT(s[1])) {
1716 while (s[1] && isDIGIT(s[1]))
1726 if (strchr("aA01! ",last_un_char))
1728 if (strchr("zZ79~",s[1]))
1730 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1731 weight -= 5; /* cope with negative subscript */
1734 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1735 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1740 if (keyword(tmpbuf, d - tmpbuf))
1743 if (un_char == last_un_char + 1)
1745 weight -= seen[un_char];
1750 if (weight >= 0) /* probably a character class */
1760 * Does all the checking to disambiguate
1762 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1763 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1765 * First argument is the stuff after the first token, e.g. "bar".
1767 * Not a method if bar is a filehandle.
1768 * Not a method if foo is a subroutine prototyped to take a filehandle.
1769 * Not a method if it's really "Foo $bar"
1770 * Method if it's "foo $bar"
1771 * Not a method if it's really "print foo $bar"
1772 * Method if it's really "foo package::" (interpreted as package->foo)
1773 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1774 * Not a method if bar is a filehandle or package, but is quoted with
1779 S_intuit_method(pTHX_ char *start, GV *gv)
1781 char *s = start + (*start == '$');
1782 char tmpbuf[sizeof PL_tokenbuf];
1790 if ((cv = GvCVu(gv))) {
1791 char *proto = SvPVX(cv);
1801 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1802 /* start is the beginning of the possible filehandle/object,
1803 * and s is the end of it
1804 * tmpbuf is a copy of it
1807 if (*start == '$') {
1808 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1813 return *s == '(' ? FUNCMETH : METHOD;
1815 if (!keyword(tmpbuf, len)) {
1816 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1821 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1822 if (indirgv && GvCVu(indirgv))
1824 /* filehandle or package name makes it a method */
1825 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1827 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1828 return 0; /* no assumptions -- "=>" quotes bearword */
1830 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1831 newSVpvn(tmpbuf,len));
1832 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1836 return *s == '(' ? FUNCMETH : METHOD;
1844 * Return a string of Perl code to load the debugger. If PERL5DB
1845 * is set, it will return the contents of that, otherwise a
1846 * compile-time require of perl5db.pl.
1853 char *pdb = PerlEnv_getenv("PERL5DB");
1857 SETERRNO(0,SS$_NORMAL);
1858 return "BEGIN { require 'perl5db.pl' }";
1864 /* Encoded script support. filter_add() effectively inserts a
1865 * 'pre-processing' function into the current source input stream.
1866 * Note that the filter function only applies to the current source file
1867 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1869 * The datasv parameter (which may be NULL) can be used to pass
1870 * private data to this instance of the filter. The filter function
1871 * can recover the SV using the FILTER_DATA macro and use it to
1872 * store private buffers and state information.
1874 * The supplied datasv parameter is upgraded to a PVIO type
1875 * and the IoDIRP/IoANY field is used to store the function pointer,
1876 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1877 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1878 * private use must be set using malloc'd pointers.
1882 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1887 if (!PL_rsfp_filters)
1888 PL_rsfp_filters = newAV();
1890 datasv = NEWSV(255,0);
1891 if (!SvUPGRADE(datasv, SVt_PVIO))
1892 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1893 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
1894 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1895 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1896 funcp, SvPV_nolen(datasv)));
1897 av_unshift(PL_rsfp_filters, 1);
1898 av_store(PL_rsfp_filters, 0, datasv) ;
1903 /* Delete most recently added instance of this filter function. */
1905 Perl_filter_del(pTHX_ filter_t funcp)
1908 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1909 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1911 /* if filter is on top of stack (usual case) just pop it off */
1912 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1913 if (IoANY(datasv) == (void *)funcp) {
1914 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1915 IoANY(datasv) = (void *)NULL;
1916 sv_free(av_pop(PL_rsfp_filters));
1920 /* we need to search for the correct entry and clear it */
1921 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1925 /* Invoke the n'th filter function for the current rsfp. */
1927 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1930 /* 0 = read one text line */
1935 if (!PL_rsfp_filters)
1937 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1938 /* Provide a default input filter to make life easy. */
1939 /* Note that we append to the line. This is handy. */
1940 DEBUG_P(PerlIO_printf(Perl_debug_log,
1941 "filter_read %d: from rsfp\n", idx));
1945 int old_len = SvCUR(buf_sv) ;
1947 /* ensure buf_sv is large enough */
1948 SvGROW(buf_sv, old_len + maxlen) ;
1949 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1950 if (PerlIO_error(PL_rsfp))
1951 return -1; /* error */
1953 return 0 ; /* end of file */
1955 SvCUR_set(buf_sv, old_len + len) ;
1958 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1959 if (PerlIO_error(PL_rsfp))
1960 return -1; /* error */
1962 return 0 ; /* end of file */
1965 return SvCUR(buf_sv);
1967 /* Skip this filter slot if filter has been deleted */
1968 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1969 DEBUG_P(PerlIO_printf(Perl_debug_log,
1970 "filter_read %d: skipped (filter deleted)\n",
1972 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1974 /* Get function pointer hidden within datasv */
1975 funcp = (filter_t)IoANY(datasv);
1976 DEBUG_P(PerlIO_printf(Perl_debug_log,
1977 "filter_read %d: via function %p (%s)\n",
1978 idx, funcp, SvPV_nolen(datasv)));
1979 /* Call function. The function is expected to */
1980 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1981 /* Return: <0:error, =0:eof, >0:not eof */
1982 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1986 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1988 #ifdef PERL_CR_FILTER
1989 if (!PL_rsfp_filters) {
1990 filter_add(S_cr_textfilter,NULL);
1993 if (PL_rsfp_filters) {
1996 SvCUR_set(sv, 0); /* start with empty line */
1997 if (FILTER_READ(0, sv, 0) > 0)
1998 return ( SvPVX(sv) ) ;
2003 return (sv_gets(sv, fp, append));
2007 S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2011 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2015 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2016 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2018 return GvHV(gv); /* Foo:: */
2021 /* use constant CLASS => 'MyClass' */
2022 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2024 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2025 pkgname = SvPV_nolen(sv);
2029 return gv_stashpv(pkgname, FALSE);
2033 static char* exp_name[] =
2034 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2035 "ATTRTERM", "TERMBLOCK"
2042 Works out what to call the token just pulled out of the input
2043 stream. The yacc parser takes care of taking the ops we return and
2044 stitching them into a tree.
2050 if read an identifier
2051 if we're in a my declaration
2052 croak if they tried to say my($foo::bar)
2053 build the ops for a my() declaration
2054 if it's an access to a my() variable
2055 are we in a sort block?
2056 croak if my($a); $a <=> $b
2057 build ops for access to a my() variable
2058 if in a dq string, and they've said @foo and we can't find @foo
2060 build ops for a bareword
2061 if we already built the token before, use it.
2065 #pragma segment Perl_yylex
2068 #ifdef USE_PURE_BISON
2069 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
2077 #ifdef USE_PURE_BISON
2078 /* increment level and store the argument pointers */
2080 if (yyactlevel >= YYMAXLEVEL) {
2081 /* What to do ??? */
2083 yylval_pointer[yyactlevel] = lvalp;
2084 yychar_pointer[yyactlevel] = lcharp;
2085 /* Save last pointer at the bottom */
2086 yylval_pointer[0] = lvalp;
2087 yychar_pointer[0] = lcharp;
2092 #ifdef USE_PURE_BISON
2100 S_syylex(pTHX) /* need to be separate from yylex for reentrancy */
2110 /* check if there's an identifier for us to look at */
2111 if (PL_pending_ident) {
2112 /* pit holds the identifier we read and pending_ident is reset */
2113 char pit = PL_pending_ident;
2114 PL_pending_ident = 0;
2116 /* if we're in a my(), we can't allow dynamics here.
2117 $foo'bar has already been turned into $foo::bar, so
2118 just check for colons.
2120 if it's a legal name, the OP is a PADANY.
2123 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
2124 if (strchr(PL_tokenbuf,':'))
2125 yyerror(Perl_form(aTHX_ "No package name allowed for "
2126 "variable %s in \"our\"",
2128 tmp = pad_allocmy(PL_tokenbuf);
2131 if (strchr(PL_tokenbuf,':'))
2132 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
2134 yylval.opval = newOP(OP_PADANY, 0);
2135 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
2141 build the ops for accesses to a my() variable.
2143 Deny my($a) or my($b) in a sort block, *if* $a or $b is
2144 then used in a comparison. This catches most, but not
2145 all cases. For instance, it catches
2146 sort { my($a); $a <=> $b }
2148 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2149 (although why you'd do that is anyone's guess).
2152 if (!strchr(PL_tokenbuf,':')) {
2154 /* Check for single character per-thread SVs */
2155 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2156 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2157 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2159 yylval.opval = newOP(OP_THREADSV, 0);
2160 yylval.opval->op_targ = tmp;
2163 #endif /* USE_THREADS */
2164 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2165 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2166 /* might be an "our" variable" */
2167 if (SvFLAGS(namesv) & SVpad_OUR) {
2168 /* build ops for a bareword */
2169 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2170 sv_catpvn(sym, "::", 2);
2171 sv_catpv(sym, PL_tokenbuf+1);
2172 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2173 yylval.opval->op_private = OPpCONST_ENTERED;
2174 gv_fetchpv(SvPVX(sym),
2176 ? (GV_ADDMULTI | GV_ADDINEVAL)
2179 ((PL_tokenbuf[0] == '$') ? SVt_PV
2180 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2185 /* if it's a sort block and they're naming $a or $b */
2186 if (PL_last_lop_op == OP_SORT &&
2187 PL_tokenbuf[0] == '$' &&
2188 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2191 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2192 d < PL_bufend && *d != '\n';
2195 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2196 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2202 yylval.opval = newOP(OP_PADANY, 0);
2203 yylval.opval->op_targ = tmp;
2209 Whine if they've said @foo in a doublequoted string,
2210 and @foo isn't a variable we can find in the symbol
2213 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2214 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2215 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2216 && ckWARN(WARN_AMBIGUOUS))
2218 /* Downgraded from fatal to warning 20000522 mjd */
2219 Perl_warner(aTHX_ WARN_AMBIGUOUS,
2220 "Possible unintended interpolation of %s in string",
2225 /* build ops for a bareword */
2226 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2227 yylval.opval->op_private = OPpCONST_ENTERED;
2228 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2229 ((PL_tokenbuf[0] == '$') ? SVt_PV
2230 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2235 /* no identifier pending identification */
2237 switch (PL_lex_state) {
2239 case LEX_NORMAL: /* Some compilers will produce faster */
2240 case LEX_INTERPNORMAL: /* code if we comment these out. */
2244 /* when we've already built the next token, just pull it out of the queue */
2247 yylval = PL_nextval[PL_nexttoke];
2249 PL_lex_state = PL_lex_defer;
2250 PL_expect = PL_lex_expect;
2251 PL_lex_defer = LEX_NORMAL;
2253 return(PL_nexttype[PL_nexttoke]);
2255 /* interpolated case modifiers like \L \U, including \Q and \E.
2256 when we get here, PL_bufptr is at the \
2258 case LEX_INTERPCASEMOD:
2260 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2261 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2263 /* handle \E or end of string */
2264 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2268 if (PL_lex_casemods) {
2269 oldmod = PL_lex_casestack[--PL_lex_casemods];
2270 PL_lex_casestack[PL_lex_casemods] = '\0';
2272 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2274 PL_lex_state = LEX_INTERPCONCAT;
2278 if (PL_bufptr != PL_bufend)
2280 PL_lex_state = LEX_INTERPCONCAT;
2285 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2286 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2287 if (strchr("LU", *s) &&
2288 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2290 PL_lex_casestack[--PL_lex_casemods] = '\0';
2293 if (PL_lex_casemods > 10) {
2294 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2295 if (newlb != PL_lex_casestack) {
2297 PL_lex_casestack = newlb;
2300 PL_lex_casestack[PL_lex_casemods++] = *s;
2301 PL_lex_casestack[PL_lex_casemods] = '\0';
2302 PL_lex_state = LEX_INTERPCONCAT;
2303 PL_nextval[PL_nexttoke].ival = 0;
2306 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2308 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2310 PL_nextval[PL_nexttoke].ival = OP_LC;
2312 PL_nextval[PL_nexttoke].ival = OP_UC;
2314 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2316 Perl_croak(aTHX_ "panic: yylex");
2319 if (PL_lex_starts) {
2328 case LEX_INTERPPUSH:
2329 return sublex_push();
2331 case LEX_INTERPSTART:
2332 if (PL_bufptr == PL_bufend)
2333 return sublex_done();
2335 PL_lex_dojoin = (*PL_bufptr == '@');
2336 PL_lex_state = LEX_INTERPNORMAL;
2337 if (PL_lex_dojoin) {
2338 PL_nextval[PL_nexttoke].ival = 0;
2341 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2342 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2343 force_next(PRIVATEREF);
2345 force_ident("\"", '$');
2346 #endif /* USE_THREADS */
2347 PL_nextval[PL_nexttoke].ival = 0;
2349 PL_nextval[PL_nexttoke].ival = 0;
2351 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2354 if (PL_lex_starts++) {
2360 case LEX_INTERPENDMAYBE:
2361 if (intuit_more(PL_bufptr)) {
2362 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2368 if (PL_lex_dojoin) {
2369 PL_lex_dojoin = FALSE;
2370 PL_lex_state = LEX_INTERPCONCAT;
2373 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2374 && SvEVALED(PL_lex_repl))
2376 if (PL_bufptr != PL_bufend)
2377 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2378 PL_lex_repl = Nullsv;
2381 case LEX_INTERPCONCAT:
2383 if (PL_lex_brackets)
2384 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2386 if (PL_bufptr == PL_bufend)
2387 return sublex_done();
2389 if (SvIVX(PL_linestr) == '\'') {
2390 SV *sv = newSVsv(PL_linestr);
2393 else if ( PL_hints & HINT_NEW_RE )
2394 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2395 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2399 s = scan_const(PL_bufptr);
2401 PL_lex_state = LEX_INTERPCASEMOD;
2403 PL_lex_state = LEX_INTERPSTART;
2406 if (s != PL_bufptr) {
2407 PL_nextval[PL_nexttoke] = yylval;
2410 if (PL_lex_starts++)
2420 PL_lex_state = LEX_NORMAL;
2421 s = scan_formline(PL_bufptr);
2422 if (!PL_lex_formbrack)
2428 PL_oldoldbufptr = PL_oldbufptr;
2431 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2432 exp_name[PL_expect], s);
2438 if (isIDFIRST_lazy_if(s,UTF))
2440 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2443 goto fake_eof; /* emulate EOF on ^D or ^Z */
2448 if (PL_lex_brackets)
2449 yyerror("Missing right curly or square bracket");
2452 if (s++ < PL_bufend)
2453 goto retry; /* ignore stray nulls */
2456 if (!PL_in_eval && !PL_preambled) {
2457 PL_preambled = TRUE;
2458 sv_setpv(PL_linestr,incl_perldb());
2459 if (SvCUR(PL_linestr))
2460 sv_catpv(PL_linestr,";");
2462 while(AvFILLp(PL_preambleav) >= 0) {
2463 SV *tmpsv = av_shift(PL_preambleav);
2464 sv_catsv(PL_linestr, tmpsv);
2465 sv_catpv(PL_linestr, ";");
2468 sv_free((SV*)PL_preambleav);
2469 PL_preambleav = NULL;
2471 if (PL_minus_n || PL_minus_p) {
2472 sv_catpv(PL_linestr, "LINE: while (<>) {");
2474 sv_catpv(PL_linestr,"chomp;");
2476 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2478 GvIMPORTED_AV_on(gv);
2480 if (strchr("/'\"", *PL_splitstr)
2481 && strchr(PL_splitstr + 1, *PL_splitstr))
2482 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2485 s = "'~#\200\1'"; /* surely one char is unused...*/
2486 while (s[1] && strchr(PL_splitstr, *s)) s++;
2488 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2489 "q" + (delim == '\''), delim);
2490 for (s = PL_splitstr; *s; s++) {
2492 sv_catpvn(PL_linestr, "\\", 1);
2493 sv_catpvn(PL_linestr, s, 1);
2495 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2499 sv_catpv(PL_linestr,"@F=split(' ');");
2502 sv_catpv(PL_linestr, "\n");
2503 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2504 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2505 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2506 SV *sv = NEWSV(85,0);
2508 sv_upgrade(sv, SVt_PVMG);
2509 sv_setsv(sv,PL_linestr);
2510 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2516 bof = PL_rsfp && (PerlIO_tell(PL_rsfp) == 0); /* *Before* read! */
2517 s = filter_gets(PL_linestr, PL_rsfp, 0);
2521 if (PL_preprocess && !PL_in_eval)
2522 (void)PerlProc_pclose(PL_rsfp);
2523 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2524 PerlIO_clearerr(PL_rsfp);
2526 (void)PerlIO_close(PL_rsfp);
2528 PL_doextract = FALSE;
2530 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2531 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2532 sv_catpv(PL_linestr,";}");
2533 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2534 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2535 PL_minus_n = PL_minus_p = 0;
2538 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2539 sv_setpv(PL_linestr,"");
2540 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2542 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2543 s = swallow_bom((U8*)s);
2546 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2547 PL_doextract = FALSE;
2549 /* Incest with pod. */
2550 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2551 sv_setpv(PL_linestr, "");
2552 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2553 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2554 PL_doextract = FALSE;
2558 } while (PL_doextract);
2559 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2560 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2561 SV *sv = NEWSV(85,0);
2563 sv_upgrade(sv, SVt_PVMG);
2564 sv_setsv(sv,PL_linestr);
2565 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2567 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2568 if (CopLINE(PL_curcop) == 1) {
2569 while (s < PL_bufend && isSPACE(*s))
2571 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2575 if (*s == '#' && *(s+1) == '!')
2577 #ifdef ALTERNATE_SHEBANG
2579 static char as[] = ALTERNATE_SHEBANG;
2580 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2581 d = s + (sizeof(as) - 1);
2583 #endif /* ALTERNATE_SHEBANG */
2592 while (*d && !isSPACE(*d))
2596 #ifdef ARG_ZERO_IS_SCRIPT
2597 if (ipathend > ipath) {
2599 * HP-UX (at least) sets argv[0] to the script name,
2600 * which makes $^X incorrect. And Digital UNIX and Linux,
2601 * at least, set argv[0] to the basename of the Perl
2602 * interpreter. So, having found "#!", we'll set it right.
2604 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2605 assert(SvPOK(x) || SvGMAGICAL(x));
2606 if (sv_eq(x, CopFILESV(PL_curcop))) {
2607 sv_setpvn(x, ipath, ipathend - ipath);
2610 TAINT_NOT; /* $^X is always tainted, but that's OK */
2612 #endif /* ARG_ZERO_IS_SCRIPT */
2617 d = instr(s,"perl -");
2619 d = instr(s,"perl");
2621 /* avoid getting into infinite loops when shebang
2622 * line contains "Perl" rather than "perl" */
2624 for (d = ipathend-4; d >= ipath; --d) {
2625 if ((*d == 'p' || *d == 'P')
2626 && !ibcmp(d, "perl", 4))
2636 #ifdef ALTERNATE_SHEBANG
2638 * If the ALTERNATE_SHEBANG on this system starts with a
2639 * character that can be part of a Perl expression, then if
2640 * we see it but not "perl", we're probably looking at the
2641 * start of Perl code, not a request to hand off to some
2642 * other interpreter. Similarly, if "perl" is there, but
2643 * not in the first 'word' of the line, we assume the line
2644 * contains the start of the Perl program.
2646 if (d && *s != '#') {
2648 while (*c && !strchr("; \t\r\n\f\v#", *c))
2651 d = Nullch; /* "perl" not in first word; ignore */
2653 *s = '#'; /* Don't try to parse shebang line */
2655 #endif /* ALTERNATE_SHEBANG */
2656 #ifndef MACOS_TRADITIONAL
2661 !instr(s,"indir") &&
2662 instr(PL_origargv[0],"perl"))
2668 while (s < PL_bufend && isSPACE(*s))
2670 if (s < PL_bufend) {
2671 Newz(899,newargv,PL_origargc+3,char*);
2673 while (s < PL_bufend && !isSPACE(*s))
2676 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2679 newargv = PL_origargv;
2681 PerlProc_execv(ipath, newargv);
2682 Perl_croak(aTHX_ "Can't exec %s", ipath);
2686 U32 oldpdb = PL_perldb;
2687 bool oldn = PL_minus_n;
2688 bool oldp = PL_minus_p;
2690 while (*d && !isSPACE(*d)) d++;
2691 while (SPACE_OR_TAB(*d)) d++;
2695 if (*d == 'M' || *d == 'm') {
2697 while (*d && !isSPACE(*d)) d++;
2698 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2701 d = moreswitches(d);
2703 if ((PERLDB_LINE && !oldpdb) ||
2704 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2705 /* if we have already added "LINE: while (<>) {",
2706 we must not do it again */
2708 sv_setpv(PL_linestr, "");
2709 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2710 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2711 PL_preambled = FALSE;
2713 (void)gv_fetchfile(PL_origfilename);
2720 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2722 PL_lex_state = LEX_FORMLINE;
2727 #ifdef PERL_STRICT_CR
2728 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2730 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2732 case ' ': case '\t': case '\f': case 013:
2733 #ifdef MACOS_TRADITIONAL
2740 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2741 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2742 /* handle eval qq[#line 1 "foo"\n ...] */
2743 CopLINE_dec(PL_curcop);
2747 while (s < d && *s != '\n')
2752 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2754 PL_lex_state = LEX_FORMLINE;
2764 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2769 while (s < PL_bufend && SPACE_OR_TAB(*s))
2772 if (strnEQ(s,"=>",2)) {
2773 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2774 OPERATOR('-'); /* unary minus */
2776 PL_last_uni = PL_oldbufptr;
2777 PL_last_lop_op = OP_FTEREAD; /* good enough */
2779 case 'r': FTST(OP_FTEREAD);
2780 case 'w': FTST(OP_FTEWRITE);
2781 case 'x': FTST(OP_FTEEXEC);
2782 case 'o': FTST(OP_FTEOWNED);
2783 case 'R': FTST(OP_FTRREAD);
2784 case 'W': FTST(OP_FTRWRITE);
2785 case 'X': FTST(OP_FTREXEC);
2786 case 'O': FTST(OP_FTROWNED);
2787 case 'e': FTST(OP_FTIS);
2788 case 'z': FTST(OP_FTZERO);
2789 case 's': FTST(OP_FTSIZE);
2790 case 'f': FTST(OP_FTFILE);
2791 case 'd': FTST(OP_FTDIR);
2792 case 'l': FTST(OP_FTLINK);
2793 case 'p': FTST(OP_FTPIPE);
2794 case 'S': FTST(OP_FTSOCK);
2795 case 'u': FTST(OP_FTSUID);
2796 case 'g': FTST(OP_FTSGID);
2797 case 'k': FTST(OP_FTSVTX);
2798 case 'b': FTST(OP_FTBLK);
2799 case 'c': FTST(OP_FTCHR);
2800 case 't': FTST(OP_FTTTY);
2801 case 'T': FTST(OP_FTTEXT);
2802 case 'B': FTST(OP_FTBINARY);
2803 case 'M': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTMTIME);
2804 case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME);
2805 case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME);
2807 Perl_croak(aTHX_ "Unrecognized file test: -%c", (int)tmp);
2814 if (PL_expect == XOPERATOR)
2819 else if (*s == '>') {
2822 if (isIDFIRST_lazy_if(s,UTF)) {
2823 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2831 if (PL_expect == XOPERATOR)
2834 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2836 OPERATOR('-'); /* unary minus */
2843 if (PL_expect == XOPERATOR)
2848 if (PL_expect == XOPERATOR)
2851 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2857 if (PL_expect != XOPERATOR) {
2858 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2859 PL_expect = XOPERATOR;
2860 force_ident(PL_tokenbuf, '*');
2873 if (PL_expect == XOPERATOR) {
2877 PL_tokenbuf[0] = '%';
2878 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2879 if (!PL_tokenbuf[1]) {
2881 yyerror("Final % should be \\% or %name");
2884 PL_pending_ident = '%';
2903 switch (PL_expect) {
2906 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2908 PL_bufptr = s; /* update in case we back off */
2914 PL_expect = XTERMBLOCK;
2918 while (isIDFIRST_lazy_if(s,UTF)) {
2919 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2920 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2921 if (tmp < 0) tmp = -tmp;
2936 d = scan_str(d,TRUE,TRUE);
2939 SvREFCNT_dec(PL_lex_stuff);
2940 PL_lex_stuff = Nullsv;
2942 /* MUST advance bufptr here to avoid bogus
2943 "at end of line" context messages from yyerror().
2945 PL_bufptr = s + len;
2946 yyerror("Unterminated attribute parameter in attribute list");
2949 return 0; /* EOF indicator */
2953 SV *sv = newSVpvn(s, len);
2954 sv_catsv(sv, PL_lex_stuff);
2955 attrs = append_elem(OP_LIST, attrs,
2956 newSVOP(OP_CONST, 0, sv));
2957 SvREFCNT_dec(PL_lex_stuff);
2958 PL_lex_stuff = Nullsv;
2961 attrs = append_elem(OP_LIST, attrs,
2962 newSVOP(OP_CONST, 0,
2966 if (*s == ':' && s[1] != ':')
2969 break; /* require real whitespace or :'s */
2971 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
2972 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
2973 char q = ((*s == '\'') ? '"' : '\'');
2974 /* If here for an expression, and parsed no attrs, back off. */
2975 if (tmp == '=' && !attrs) {
2979 /* MUST advance bufptr here to avoid bogus "at end of line"
2980 context messages from yyerror().
2984 yyerror("Unterminated attribute list");
2986 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
2994 PL_nextval[PL_nexttoke].opval = attrs;
3002 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3003 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3019 if (PL_lex_brackets <= 0)
3020 yyerror("Unmatched right square bracket");
3023 if (PL_lex_state == LEX_INTERPNORMAL) {
3024 if (PL_lex_brackets == 0) {
3025 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3026 PL_lex_state = LEX_INTERPEND;
3033 if (PL_lex_brackets > 100) {
3034 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
3035 if (newlb != PL_lex_brackstack) {
3037 PL_lex_brackstack = newlb;
3040 switch (PL_expect) {
3042 if (PL_lex_formbrack) {
3046 if (PL_oldoldbufptr == PL_last_lop)
3047 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3049 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3050 OPERATOR(HASHBRACK);
3052 while (s < PL_bufend && SPACE_OR_TAB(*s))
3055 PL_tokenbuf[0] = '\0';
3056 if (d < PL_bufend && *d == '-') {
3057 PL_tokenbuf[0] = '-';
3059 while (d < PL_bufend && SPACE_OR_TAB(*d))
3062 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3063 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3065 while (d < PL_bufend && SPACE_OR_TAB(*d))
3068 char minus = (PL_tokenbuf[0] == '-');
3069 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3077 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3082 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3087 if (PL_oldoldbufptr == PL_last_lop)
3088 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3090 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3093 OPERATOR(HASHBRACK);
3094 /* This hack serves to disambiguate a pair of curlies
3095 * as being a block or an anon hash. Normally, expectation
3096 * determines that, but in cases where we're not in a
3097 * position to expect anything in particular (like inside
3098 * eval"") we have to resolve the ambiguity. This code
3099 * covers the case where the first term in the curlies is a
3100 * quoted string. Most other cases need to be explicitly
3101 * disambiguated by prepending a `+' before the opening
3102 * curly in order to force resolution as an anon hash.
3104 * XXX should probably propagate the outer expectation
3105 * into eval"" to rely less on this hack, but that could
3106 * potentially break current behavior of eval"".
3110 if (*s == '\'' || *s == '"' || *s == '`') {
3111 /* common case: get past first string, handling escapes */
3112 for (t++; t < PL_bufend && *t != *s;)
3113 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3117 else if (*s == 'q') {
3120 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3124 char open, close, term;
3127 while (t < PL_bufend && isSPACE(*t))
3131 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3135 for (t++; t < PL_bufend; t++) {
3136 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3138 else if (*t == open)
3142 for (t++; t < PL_bufend; t++) {
3143 if (*t == '\\' && t+1 < PL_bufend)
3145 else if (*t == close && --brackets <= 0)
3147 else if (*t == open)
3153 else if (isALNUM_lazy_if(t,UTF)) {
3155 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3158 while (t < PL_bufend && isSPACE(*t))
3160 /* if comma follows first term, call it an anon hash */
3161 /* XXX it could be a comma expression with loop modifiers */
3162 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3163 || (*t == '=' && t[1] == '>')))
3164 OPERATOR(HASHBRACK);
3165 if (PL_expect == XREF)
3168 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3174 yylval.ival = CopLINE(PL_curcop);
3175 if (isSPACE(*s) || *s == '#')
3176 PL_copline = NOLINE; /* invalidate current command line number */
3181 if (PL_lex_brackets <= 0)
3182 yyerror("Unmatched right curly bracket");
3184 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3185 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3186 PL_lex_formbrack = 0;
3187 if (PL_lex_state == LEX_INTERPNORMAL) {
3188 if (PL_lex_brackets == 0) {
3189 if (PL_expect & XFAKEBRACK) {
3190 PL_expect &= XENUMMASK;
3191 PL_lex_state = LEX_INTERPEND;
3193 return yylex(); /* ignore fake brackets */
3195 if (*s == '-' && s[1] == '>')
3196 PL_lex_state = LEX_INTERPENDMAYBE;
3197 else if (*s != '[' && *s != '{')
3198 PL_lex_state = LEX_INTERPEND;
3201 if (PL_expect & XFAKEBRACK) {
3202 PL_expect &= XENUMMASK;
3204 return yylex(); /* ignore fake brackets */
3214 if (PL_expect == XOPERATOR) {
3215 if (ckWARN(WARN_SEMICOLON)
3216 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3218 CopLINE_dec(PL_curcop);
3219 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3220 CopLINE_inc(PL_curcop);
3225 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3227 PL_expect = XOPERATOR;
3228 force_ident(PL_tokenbuf, '&');
3232 yylval.ival = (OPpENTERSUB_AMPER<<8);
3251 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3252 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3254 if (PL_expect == XSTATE && isALPHA(tmp) &&
3255 (s == PL_linestart+1 || s[-2] == '\n') )
3257 if (PL_in_eval && !PL_rsfp) {
3262 if (strnEQ(s,"=cut",4)) {
3276 PL_doextract = TRUE;
3279 if (PL_lex_brackets < PL_lex_formbrack) {
3281 #ifdef PERL_STRICT_CR
3282 for (t = s; SPACE_OR_TAB(*t); t++) ;
3284 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3286 if (*t == '\n' || *t == '#') {
3304 if (PL_expect != XOPERATOR) {
3305 if (s[1] != '<' && !strchr(s,'>'))
3308 s = scan_heredoc(s);
3310 s = scan_inputsymbol(s);
3311 TERM(sublex_start());
3316 SHop(OP_LEFT_SHIFT);
3330 SHop(OP_RIGHT_SHIFT);
3339 if (PL_expect == XOPERATOR) {
3340 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3343 return ','; /* grandfather non-comma-format format */
3347 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3348 PL_tokenbuf[0] = '@';
3349 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3350 sizeof PL_tokenbuf - 1, FALSE);
3351 if (PL_expect == XOPERATOR)
3352 no_op("Array length", s);
3353 if (!PL_tokenbuf[1])
3355 PL_expect = XOPERATOR;
3356 PL_pending_ident = '#';
3360 PL_tokenbuf[0] = '$';
3361 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3362 sizeof PL_tokenbuf - 1, FALSE);
3363 if (PL_expect == XOPERATOR)
3365 if (!PL_tokenbuf[1]) {
3367 yyerror("Final $ should be \\$ or $name");
3371 /* This kludge not intended to be bulletproof. */
3372 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3373 yylval.opval = newSVOP(OP_CONST, 0,
3374 newSViv(PL_compiling.cop_arybase));
3375 yylval.opval->op_private = OPpCONST_ARYBASE;
3381 if (PL_lex_state == LEX_NORMAL)
3384 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3387 PL_tokenbuf[0] = '@';
3388 if (ckWARN(WARN_SYNTAX)) {
3390 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3393 PL_bufptr = skipspace(PL_bufptr);
3394 while (t < PL_bufend && *t != ']')
3396 Perl_warner(aTHX_ WARN_SYNTAX,
3397 "Multidimensional syntax %.*s not supported",
3398 (t - PL_bufptr) + 1, PL_bufptr);
3402 else if (*s == '{') {
3403 PL_tokenbuf[0] = '%';
3404 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3405 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3407 char tmpbuf[sizeof PL_tokenbuf];
3409 for (t++; isSPACE(*t); t++) ;
3410 if (isIDFIRST_lazy_if(t,UTF)) {
3411 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3412 for (; isSPACE(*t); t++) ;
3413 if (*t == ';' && get_cv(tmpbuf, FALSE))
3414 Perl_warner(aTHX_ WARN_SYNTAX,
3415 "You need to quote \"%s\"", tmpbuf);
3421 PL_expect = XOPERATOR;
3422 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3423 bool islop = (PL_last_lop == PL_oldoldbufptr);
3424 if (!islop || PL_last_lop_op == OP_GREPSTART)
3425 PL_expect = XOPERATOR;
3426 else if (strchr("$@\"'`q", *s))
3427 PL_expect = XTERM; /* e.g. print $fh "foo" */
3428 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3429 PL_expect = XTERM; /* e.g. print $fh &sub */
3430 else if (isIDFIRST_lazy_if(s,UTF)) {
3431 char tmpbuf[sizeof PL_tokenbuf];
3432 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3433 if ((tmp = keyword(tmpbuf, len))) {
3434 /* binary operators exclude handle interpretations */
3446 PL_expect = XTERM; /* e.g. print $fh length() */
3451 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3452 if (gv && GvCVu(gv))
3453 PL_expect = XTERM; /* e.g. print $fh subr() */
3456 else if (isDIGIT(*s))
3457 PL_expect = XTERM; /* e.g. print $fh 3 */
3458 else if (*s == '.' && isDIGIT(s[1]))
3459 PL_expect = XTERM; /* e.g. print $fh .3 */
3460 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3461 PL_expect = XTERM; /* e.g. print $fh -1 */
3462 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3463 PL_expect = XTERM; /* print $fh <<"EOF" */
3465 PL_pending_ident = '$';
3469 if (PL_expect == XOPERATOR)
3471 PL_tokenbuf[0] = '@';
3472 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3473 if (!PL_tokenbuf[1]) {
3475 yyerror("Final @ should be \\@ or @name");
3478 if (PL_lex_state == LEX_NORMAL)
3480 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3482 PL_tokenbuf[0] = '%';
3484 /* Warn about @ where they meant $. */
3485 if (ckWARN(WARN_SYNTAX)) {
3486 if (*s == '[' || *s == '{') {
3488 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3490 if (*t == '}' || *t == ']') {
3492 PL_bufptr = skipspace(PL_bufptr);
3493 Perl_warner(aTHX_ WARN_SYNTAX,
3494 "Scalar value %.*s better written as $%.*s",
3495 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3500 PL_pending_ident = '@';
3503 case '/': /* may either be division or pattern */
3504 case '?': /* may either be conditional or pattern */
3505 if (PL_expect != XOPERATOR) {
3506 /* Disable warning on "study /blah/" */
3507 if (PL_oldoldbufptr == PL_last_uni
3508 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3509 || memNE(PL_last_uni, "study", 5)
3510 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3512 s = scan_pat(s,OP_MATCH);
3513 TERM(sublex_start());
3521 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3522 #ifdef PERL_STRICT_CR
3525 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3527 && (s == PL_linestart || s[-1] == '\n') )
3529 PL_lex_formbrack = 0;
3533 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3539 yylval.ival = OPf_SPECIAL;
3545 if (PL_expect != XOPERATOR)
3550 case '0': case '1': case '2': case '3': case '4':
3551 case '5': case '6': case '7': case '8': case '9':
3553 if (PL_expect == XOPERATOR)
3558 s = scan_str(s,FALSE,FALSE);
3559 if (PL_expect == XOPERATOR) {
3560 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3563 return ','; /* grandfather non-comma-format format */
3569 missingterm((char*)0);
3570 yylval.ival = OP_CONST;
3571 TERM(sublex_start());
3574 s = scan_str(s,FALSE,FALSE);
3575 if (PL_expect == XOPERATOR) {
3576 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3579 return ','; /* grandfather non-comma-format format */
3585 missingterm((char*)0);
3586 yylval.ival = OP_CONST;
3587 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3588 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3589 yylval.ival = OP_STRINGIFY;
3593 TERM(sublex_start());
3596 s = scan_str(s,FALSE,FALSE);
3597 if (PL_expect == XOPERATOR)
3598 no_op("Backticks",s);
3600 missingterm((char*)0);
3601 yylval.ival = OP_BACKTICK;
3603 TERM(sublex_start());
3607 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3608 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3610 if (PL_expect == XOPERATOR)
3611 no_op("Backslash",s);
3615 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3619 while (isDIGIT(*start) || *start == '_')
3621 if (*start == '.' && isDIGIT(start[1])) {
3625 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3626 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF)) {
3630 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3640 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3679 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3681 /* Some keywords can be followed by any delimiter, including ':' */
3682 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3683 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3684 (PL_tokenbuf[0] == 'q' &&
3685 strchr("qwxr", PL_tokenbuf[1])))));
3687 /* x::* is just a word, unless x is "CORE" */
3688 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3692 while (d < PL_bufend && isSPACE(*d))
3693 d++; /* no comments skipped here, or s### is misparsed */
3695 /* Is this a label? */
3696 if (!tmp && PL_expect == XSTATE
3697 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3699 yylval.pval = savepv(PL_tokenbuf);
3704 /* Check for keywords */
3705 tmp = keyword(PL_tokenbuf, len);
3707 /* Is this a word before a => operator? */
3708 if (*d == '=' && d[1] == '>') {
3710 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3711 yylval.opval->op_private = OPpCONST_BARE;
3715 if (tmp < 0) { /* second-class keyword? */
3716 GV *ogv = Nullgv; /* override (winner) */
3717 GV *hgv = Nullgv; /* hidden (loser) */
3718 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3720 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3723 if (GvIMPORTED_CV(gv))
3725 else if (! CvMETHOD(cv))
3729 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3730 (gv = *gvp) != (GV*)&PL_sv_undef &&
3731 GvCVu(gv) && GvIMPORTED_CV(gv))
3737 tmp = 0; /* overridden by import or by GLOBAL */
3740 && -tmp==KEY_lock /* XXX generalizable kludge */
3742 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3744 tmp = 0; /* any sub overrides "weak" keyword */
3746 else { /* no override */
3750 if (ckWARN(WARN_AMBIGUOUS) && hgv
3751 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3752 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3753 "Ambiguous call resolved as CORE::%s(), %s",
3754 GvENAME(hgv), "qualify as such or use &");
3761 default: /* not a keyword */
3764 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3766 /* Get the rest if it looks like a package qualifier */
3768 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3770 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3773 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3774 *s == '\'' ? "'" : "::");
3778 if (PL_expect == XOPERATOR) {
3779 if (PL_bufptr == PL_linestart) {
3780 CopLINE_dec(PL_curcop);
3781 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3782 CopLINE_inc(PL_curcop);
3785 no_op("Bareword",s);
3788 /* Look for a subroutine with this name in current package,
3789 unless name is "Foo::", in which case Foo is a bearword
3790 (and a package name). */
3793 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3795 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3796 Perl_warner(aTHX_ WARN_BAREWORD,
3797 "Bareword \"%s\" refers to nonexistent package",
3800 PL_tokenbuf[len] = '\0';
3807 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3810 /* if we saw a global override before, get the right name */
3813 sv = newSVpvn("CORE::GLOBAL::",14);
3814 sv_catpv(sv,PL_tokenbuf);
3817 sv = newSVpv(PL_tokenbuf,0);
3819 /* Presume this is going to be a bareword of some sort. */
3822 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3823 yylval.opval->op_private = OPpCONST_BARE;
3825 /* And if "Foo::", then that's what it certainly is. */
3830 /* See if it's the indirect object for a list operator. */
3832 if (PL_oldoldbufptr &&
3833 PL_oldoldbufptr < PL_bufptr &&
3834 (PL_oldoldbufptr == PL_last_lop
3835 || PL_oldoldbufptr == PL_last_uni) &&
3836 /* NO SKIPSPACE BEFORE HERE! */
3837 (PL_expect == XREF ||
3838 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3840 bool immediate_paren = *s == '(';
3842 /* (Now we can afford to cross potential line boundary.) */
3845 /* Two barewords in a row may indicate method call. */
3847 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3850 /* If not a declared subroutine, it's an indirect object. */
3851 /* (But it's an indir obj regardless for sort.) */
3853 if ((PL_last_lop_op == OP_SORT ||
3854 (!immediate_paren && (!gv || !GvCVu(gv)))) &&
3855 (PL_last_lop_op != OP_MAPSTART &&
3856 PL_last_lop_op != OP_GREPSTART))
3858 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3864 PL_expect = XOPERATOR;
3867 /* Is this a word before a => operator? */
3868 if (*s == '=' && s[1] == '>') {
3870 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
3874 /* If followed by a paren, it's certainly a subroutine. */
3877 if (gv && GvCVu(gv)) {
3878 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
3879 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3884 PL_nextval[PL_nexttoke].opval = yylval.opval;
3885 PL_expect = XOPERATOR;
3891 /* If followed by var or block, call it a method (unless sub) */
3893 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3894 PL_last_lop = PL_oldbufptr;
3895 PL_last_lop_op = OP_METHOD;
3899 /* If followed by a bareword, see if it looks like indir obj. */
3901 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3904 /* Not a method, so call it a subroutine (if defined) */
3906 if (gv && GvCVu(gv)) {
3908 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3909 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3910 "Ambiguous use of -%s resolved as -&%s()",
3911 PL_tokenbuf, PL_tokenbuf);
3912 /* Check for a constant sub */
3914 if ((sv = cv_const_sv(cv))) {
3916 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3917 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3918 yylval.opval->op_private = 0;
3922 /* Resolve to GV now. */
3923 op_free(yylval.opval);
3924 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3925 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3926 PL_last_lop = PL_oldbufptr;
3927 PL_last_lop_op = OP_ENTERSUB;
3928 /* Is there a prototype? */
3931 char *proto = SvPV((SV*)cv, len);
3934 if (strEQ(proto, "$"))
3936 if (*proto == '&' && *s == '{') {
3937 sv_setpv(PL_subname,"__ANON__");
3941 PL_nextval[PL_nexttoke].opval = yylval.opval;
3947 /* Call it a bare word */
3949 if (PL_hints & HINT_STRICT_SUBS)
3950 yylval.opval->op_private |= OPpCONST_STRICT;
3953 if (ckWARN(WARN_RESERVED)) {
3954 if (lastchar != '-') {
3955 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
3957 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
3964 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
3965 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3966 "Operator or semicolon missing before %c%s",
3967 lastchar, PL_tokenbuf);
3968 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3969 "Ambiguous use of %c resolved as operator %c",
3970 lastchar, lastchar);
3976 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3977 newSVpv(CopFILE(PL_curcop),0));
3981 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3982 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
3985 case KEY___PACKAGE__:
3986 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3988 ? newSVsv(PL_curstname)
3997 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
3998 char *pname = "main";
3999 if (PL_tokenbuf[2] == 'D')
4000 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
4001 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4004 GvIOp(gv) = newIO();
4005 IoIFP(GvIOp(gv)) = PL_rsfp;
4006 #if defined(HAS_FCNTL) && defined(F_SETFD)
4008 int fd = PerlIO_fileno(PL_rsfp);
4009 fcntl(fd,F_SETFD,fd >= 3);
4012 /* Mark this internal pseudo-handle as clean */
4013 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4015 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4016 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4017 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4019 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4020 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4021 /* if the script was opened in binmode, we need to revert
4022 * it to text mode for compatibility; but only iff it has CRs
4023 * XXX this is a questionable hack at best. */
4024 if (PL_bufend-PL_bufptr > 2
4025 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4028 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4029 loc = PerlIO_tell(PL_rsfp);
4030 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4032 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4033 #if defined(__BORLANDC__)
4034 /* XXX see note in do_binmode() */
4035 ((FILE*)PL_rsfp)->flags |= _F_BIN;
4038 PerlIO_seek(PL_rsfp, loc, 0);
4053 if (PL_expect == XSTATE) {
4060 if (*s == ':' && s[1] == ':') {
4063 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4064 if (!(tmp = keyword(PL_tokenbuf, len)))
4065 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4079 LOP(OP_ACCEPT,XTERM);
4085 LOP(OP_ATAN2,XTERM);
4091 LOP(OP_BINMODE,XTERM);
4094 LOP(OP_BLESS,XTERM);
4103 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4120 if (!PL_cryptseen) {
4121 PL_cryptseen = TRUE;
4125 LOP(OP_CRYPT,XTERM);
4128 if (ckWARN(WARN_CHMOD)) {
4129 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4130 if (*d != '0' && isDIGIT(*d))
4131 Perl_warner(aTHX_ WARN_CHMOD,
4132 "chmod() mode argument is missing initial 0");
4134 LOP(OP_CHMOD,XTERM);
4137 LOP(OP_CHOWN,XTERM);
4140 LOP(OP_CONNECT,XTERM);
4156 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4160 PL_hints |= HINT_BLOCK_SCOPE;
4170 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4171 LOP(OP_DBMOPEN,XTERM);
4177 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4184 yylval.ival = CopLINE(PL_curcop);
4198 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4199 UNIBRACK(OP_ENTEREVAL);
4214 case KEY_endhostent:
4220 case KEY_endservent:
4223 case KEY_endprotoent:
4234 yylval.ival = CopLINE(PL_curcop);
4236 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4238 if ((PL_bufend - p) >= 3 &&
4239 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4241 else if ((PL_bufend - p) >= 4 &&
4242 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4245 if (isIDFIRST_lazy_if(p,UTF)) {
4246 p = scan_ident(p, PL_bufend,
4247 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4251 Perl_croak(aTHX_ "Missing $ on loop variable");
4256 LOP(OP_FORMLINE,XTERM);
4262 LOP(OP_FCNTL,XTERM);
4268 LOP(OP_FLOCK,XTERM);
4277 LOP(OP_GREPSTART, XREF);
4280 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4295 case KEY_getpriority:
4296 LOP(OP_GETPRIORITY,XTERM);
4298 case KEY_getprotobyname:
4301 case KEY_getprotobynumber:
4302 LOP(OP_GPBYNUMBER,XTERM);
4304 case KEY_getprotoent:
4316 case KEY_getpeername:
4317 UNI(OP_GETPEERNAME);
4319 case KEY_gethostbyname:
4322 case KEY_gethostbyaddr:
4323 LOP(OP_GHBYADDR,XTERM);
4325 case KEY_gethostent:
4328 case KEY_getnetbyname:
4331 case KEY_getnetbyaddr:
4332 LOP(OP_GNBYADDR,XTERM);
4337 case KEY_getservbyname:
4338 LOP(OP_GSBYNAME,XTERM);
4340 case KEY_getservbyport:
4341 LOP(OP_GSBYPORT,XTERM);
4343 case KEY_getservent:
4346 case KEY_getsockname:
4347 UNI(OP_GETSOCKNAME);
4349 case KEY_getsockopt:
4350 LOP(OP_GSOCKOPT,XTERM);
4372 yylval.ival = CopLINE(PL_curcop);
4376 LOP(OP_INDEX,XTERM);
4382 LOP(OP_IOCTL,XTERM);
4394 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4426 LOP(OP_LISTEN,XTERM);
4435 s = scan_pat(s,OP_MATCH);
4436 TERM(sublex_start());
4439 LOP(OP_MAPSTART, XREF);
4442 LOP(OP_MKDIR,XTERM);
4445 LOP(OP_MSGCTL,XTERM);
4448 LOP(OP_MSGGET,XTERM);
4451 LOP(OP_MSGRCV,XTERM);
4454 LOP(OP_MSGSND,XTERM);
4460 if (isIDFIRST_lazy_if(s,UTF)) {
4461 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4462 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4464 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4465 if (!PL_in_my_stash) {
4468 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4476 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4483 if (PL_expect != XSTATE)
4484 yyerror("\"no\" not allowed in expression");
4485 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4486 s = force_version(s);
4491 if (*s == '(' || (s = skipspace(s), *s == '('))
4498 if (isIDFIRST_lazy_if(s,UTF)) {
4500 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4502 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4503 Perl_warner(aTHX_ WARN_PRECEDENCE,
4504 "Precedence problem: open %.*s should be open(%.*s)",
4510 yylval.ival = OP_OR;
4520 LOP(OP_OPEN_DIR,XTERM);
4523 checkcomma(s,PL_tokenbuf,"filehandle");
4527 checkcomma(s,PL_tokenbuf,"filehandle");
4546 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4550 LOP(OP_PIPE_OP,XTERM);
4553 s = scan_str(s,FALSE,FALSE);
4555 missingterm((char*)0);
4556 yylval.ival = OP_CONST;
4557 TERM(sublex_start());
4563 s = scan_str(s,FALSE,FALSE);
4565 missingterm((char*)0);
4567 if (SvCUR(PL_lex_stuff)) {
4570 d = SvPV_force(PL_lex_stuff, len);
4572 for (; isSPACE(*d) && len; --len, ++d) ;
4575 if (!warned && ckWARN(WARN_QW)) {
4576 for (; !isSPACE(*d) && len; --len, ++d) {
4578 Perl_warner(aTHX_ WARN_QW,
4579 "Possible attempt to separate words with commas");
4582 else if (*d == '#') {
4583 Perl_warner(aTHX_ WARN_QW,
4584 "Possible attempt to put comments in qw() list");
4590 for (; !isSPACE(*d) && len; --len, ++d) ;
4592 words = append_elem(OP_LIST, words,
4593 newSVOP(OP_CONST, 0, tokeq(newSVpvn(b, d-b))));
4597 PL_nextval[PL_nexttoke].opval = words;
4602 SvREFCNT_dec(PL_lex_stuff);
4603 PL_lex_stuff = Nullsv;
4608 s = scan_str(s,FALSE,FALSE);
4610 missingterm((char*)0);
4611 yylval.ival = OP_STRINGIFY;
4612 if (SvIVX(PL_lex_stuff) == '\'')
4613 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4614 TERM(sublex_start());
4617 s = scan_pat(s,OP_QR);
4618 TERM(sublex_start());
4621 s = scan_str(s,FALSE,FALSE);
4623 missingterm((char*)0);
4624 yylval.ival = OP_BACKTICK;
4626 TERM(sublex_start());
4633 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4634 s = force_version(s);
4637 *PL_tokenbuf = '\0';
4638 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4639 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4640 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4642 yyerror("<> should be quotes");
4650 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4654 LOP(OP_RENAME,XTERM);
4663 LOP(OP_RINDEX,XTERM);
4686 LOP(OP_REVERSE,XTERM);
4697 TERM(sublex_start());
4699 TOKEN(1); /* force error */
4708 LOP(OP_SELECT,XTERM);
4714 LOP(OP_SEMCTL,XTERM);
4717 LOP(OP_SEMGET,XTERM);
4720 LOP(OP_SEMOP,XTERM);
4726 LOP(OP_SETPGRP,XTERM);
4728 case KEY_setpriority:
4729 LOP(OP_SETPRIORITY,XTERM);
4731 case KEY_sethostent:
4737 case KEY_setservent:
4740 case KEY_setprotoent:
4750 LOP(OP_SEEKDIR,XTERM);
4752 case KEY_setsockopt:
4753 LOP(OP_SSOCKOPT,XTERM);
4759 LOP(OP_SHMCTL,XTERM);
4762 LOP(OP_SHMGET,XTERM);
4765 LOP(OP_SHMREAD,XTERM);
4768 LOP(OP_SHMWRITE,XTERM);
4771 LOP(OP_SHUTDOWN,XTERM);
4780 LOP(OP_SOCKET,XTERM);
4782 case KEY_socketpair:
4783 LOP(OP_SOCKPAIR,XTERM);
4786 checkcomma(s,PL_tokenbuf,"subroutine name");
4788 if (*s == ';' || *s == ')') /* probably a close */
4789 Perl_croak(aTHX_ "sort is now a reserved word");
4791 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4795 LOP(OP_SPLIT,XTERM);
4798 LOP(OP_SPRINTF,XTERM);
4801 LOP(OP_SPLICE,XTERM);
4816 LOP(OP_SUBSTR,XTERM);
4822 char tmpbuf[sizeof PL_tokenbuf];
4824 expectation attrful;
4825 bool have_name, have_proto;
4830 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4831 (*s == ':' && s[1] == ':'))
4834 attrful = XATTRBLOCK;
4835 /* remember buffer pos'n for later force_word */
4836 tboffset = s - PL_oldbufptr;
4837 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4838 if (strchr(tmpbuf, ':'))
4839 sv_setpv(PL_subname, tmpbuf);
4841 sv_setsv(PL_subname,PL_curstname);
4842 sv_catpvn(PL_subname,"::",2);
4843 sv_catpvn(PL_subname,tmpbuf,len);
4850 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4851 PL_expect = XTERMBLOCK;
4852 attrful = XATTRTERM;
4853 sv_setpv(PL_subname,"?");
4857 if (key == KEY_format) {
4859 PL_lex_formbrack = PL_lex_brackets + 1;
4861 (void) force_word(PL_oldbufptr + tboffset, WORD,
4866 /* Look for a prototype */
4870 s = scan_str(s,FALSE,FALSE);
4873 SvREFCNT_dec(PL_lex_stuff);
4874 PL_lex_stuff = Nullsv;
4875 Perl_croak(aTHX_ "Prototype not terminated");
4878 d = SvPVX(PL_lex_stuff);
4880 for (p = d; *p; ++p) {
4885 SvCUR(PL_lex_stuff) = tmp;
4893 if (*s == ':' && s[1] != ':')
4894 PL_expect = attrful;
4897 PL_nextval[PL_nexttoke].opval =
4898 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4899 PL_lex_stuff = Nullsv;
4903 sv_setpv(PL_subname,"__ANON__");
4906 (void) force_word(PL_oldbufptr + tboffset, WORD,
4915 LOP(OP_SYSTEM,XREF);
4918 LOP(OP_SYMLINK,XTERM);
4921 LOP(OP_SYSCALL,XTERM);
4924 LOP(OP_SYSOPEN,XTERM);
4927 LOP(OP_SYSSEEK,XTERM);
4930 LOP(OP_SYSREAD,XTERM);
4933 LOP(OP_SYSWRITE,XTERM);
4937 TERM(sublex_start());
4958 LOP(OP_TRUNCATE,XTERM);
4970 yylval.ival = CopLINE(PL_curcop);
4974 yylval.ival = CopLINE(PL_curcop);
4978 LOP(OP_UNLINK,XTERM);
4984 LOP(OP_UNPACK,XTERM);
4987 LOP(OP_UTIME,XTERM);
4990 if (ckWARN(WARN_UMASK)) {
4991 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4992 if (*d != '0' && isDIGIT(*d))
4993 Perl_warner(aTHX_ WARN_UMASK,
4994 "umask: argument is missing initial 0");
4999 LOP(OP_UNSHIFT,XTERM);
5002 if (PL_expect != XSTATE)
5003 yyerror("\"use\" not allowed in expression");
5005 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
5006 s = force_version(s);
5007 if (*s == ';' || (s = skipspace(s), *s == ';')) {
5008 PL_nextval[PL_nexttoke].opval = Nullop;
5013 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5014 s = force_version(s);
5026 yylval.ival = CopLINE(PL_curcop);
5030 PL_hints |= HINT_BLOCK_SCOPE;
5037 LOP(OP_WAITPID,XTERM);
5045 static char ctl_l[2];
5047 if (ctl_l[0] == '\0')
5048 ctl_l[0] = toCTRL('L');
5049 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5052 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5057 if (PL_expect == XOPERATOR)
5063 yylval.ival = OP_XOR;
5068 TERM(sublex_start());
5073 #pragma segment Main
5077 Perl_keyword(pTHX_ register char *d, I32 len)
5082 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5083 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5084 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5085 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5086 if (strEQ(d,"__END__")) return KEY___END__;
5090 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5095 if (strEQ(d,"and")) return -KEY_and;
5096 if (strEQ(d,"abs")) return -KEY_abs;
5099 if (strEQ(d,"alarm")) return -KEY_alarm;
5100 if (strEQ(d,"atan2")) return -KEY_atan2;
5103 if (strEQ(d,"accept")) return -KEY_accept;
5108 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5111 if (strEQ(d,"bless")) return -KEY_bless;
5112 if (strEQ(d,"bind")) return -KEY_bind;
5113 if (strEQ(d,"binmode")) return -KEY_binmode;
5116 if (strEQ(d,"CORE")) return -KEY_CORE;
5117 if (strEQ(d,"CHECK")) return KEY_CHECK;
5122 if (strEQ(d,"cmp")) return -KEY_cmp;
5123 if (strEQ(d,"chr")) return -KEY_chr;
5124 if (strEQ(d,"cos")) return -KEY_cos;
5127 if (strEQ(d,"chop")) return -KEY_chop;
5130 if (strEQ(d,"close")) return -KEY_close;
5131 if (strEQ(d,"chdir")) return -KEY_chdir;
5132 if (strEQ(d,"chomp")) return -KEY_chomp;
5133 if (strEQ(d,"chmod")) return -KEY_chmod;
5134 if (strEQ(d,"chown")) return -KEY_chown;
5135 if (strEQ(d,"crypt")) return -KEY_crypt;
5138 if (strEQ(d,"chroot")) return -KEY_chroot;
5139 if (strEQ(d,"caller")) return -KEY_caller;
5142 if (strEQ(d,"connect")) return -KEY_connect;
5145 if (strEQ(d,"closedir")) return -KEY_closedir;
5146 if (strEQ(d,"continue")) return -KEY_continue;
5151 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5156 if (strEQ(d,"do")) return KEY_do;
5159 if (strEQ(d,"die")) return -KEY_die;
5162 if (strEQ(d,"dump")) return -KEY_dump;
5165 if (strEQ(d,"delete")) return KEY_delete;
5168 if (strEQ(d,"defined")) return KEY_defined;
5169 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5172 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5177 if (strEQ(d,"END")) return KEY_END;
5182 if (strEQ(d,"eq")) return -KEY_eq;
5185 if (strEQ(d,"eof")) return -KEY_eof;
5186 if (strEQ(d,"exp")) return -KEY_exp;
5189 if (strEQ(d,"else")) return KEY_else;
5190 if (strEQ(d,"exit")) return -KEY_exit;
5191 if (strEQ(d,"eval")) return KEY_eval;
5192 if (strEQ(d,"exec")) return -KEY_exec;
5193 if (strEQ(d,"each")) return KEY_each;
5196 if (strEQ(d,"elsif")) return KEY_elsif;
5199 if (strEQ(d,"exists")) return KEY_exists;
5200 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5203 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5204 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5207 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5210 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5211 if (strEQ(d,"endservent")) return -KEY_endservent;
5214 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5221 if (strEQ(d,"for")) return KEY_for;
5224 if (strEQ(d,"fork")) return -KEY_fork;
5227 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5228 if (strEQ(d,"flock")) return -KEY_flock;
5231 if (strEQ(d,"format")) return KEY_format;
5232 if (strEQ(d,"fileno")) return -KEY_fileno;
5235 if (strEQ(d,"foreach")) return KEY_foreach;
5238 if (strEQ(d,"formline")) return -KEY_formline;
5243 if (strnEQ(d,"get",3)) {
5248 if (strEQ(d,"ppid")) return -KEY_getppid;
5249 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5252 if (strEQ(d,"pwent")) return -KEY_getpwent;
5253 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5254 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5257 if (strEQ(d,"peername")) return -KEY_getpeername;
5258 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5259 if (strEQ(d,"priority")) return -KEY_getpriority;
5262 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5265 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5269 else if (*d == 'h') {
5270 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5271 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5272 if (strEQ(d,"hostent")) return -KEY_gethostent;
5274 else if (*d == 'n') {
5275 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5276 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5277 if (strEQ(d,"netent")) return -KEY_getnetent;
5279 else if (*d == 's') {
5280 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5281 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5282 if (strEQ(d,"servent")) return -KEY_getservent;
5283 if (strEQ(d,"sockname")) return -KEY_getsockname;
5284 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5286 else if (*d == 'g') {
5287 if (strEQ(d,"grent")) return -KEY_getgrent;
5288 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5289 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5291 else if (*d == 'l') {
5292 if (strEQ(d,"login")) return -KEY_getlogin;
5294 else if (strEQ(d,"c")) return -KEY_getc;
5299 if (strEQ(d,"gt")) return -KEY_gt;
5300 if (strEQ(d,"ge")) return -KEY_ge;
5303 if (strEQ(d,"grep")) return KEY_grep;
5304 if (strEQ(d,"goto")) return KEY_goto;
5305 if (strEQ(d,"glob")) return KEY_glob;
5308 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5313 if (strEQ(d,"hex")) return -KEY_hex;
5316 if (strEQ(d,"INIT")) return KEY_INIT;
5321 if (strEQ(d,"if")) return KEY_if;
5324 if (strEQ(d,"int")) return -KEY_int;
5327 if (strEQ(d,"index")) return -KEY_index;
5328 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5333 if (strEQ(d,"join")) return -KEY_join;
5337 if (strEQ(d,"keys")) return KEY_keys;
5338 if (strEQ(d,"kill")) return -KEY_kill;
5344 if (strEQ(d,"lt")) return -KEY_lt;
5345 if (strEQ(d,"le")) return -KEY_le;
5346 if (strEQ(d,"lc")) return -KEY_lc;
5349 if (strEQ(d,"log")) return -KEY_log;
5352 if (strEQ(d,"last")) return KEY_last;
5353 if (strEQ(d,"link")) return -KEY_link;
5354 if (strEQ(d,"lock")) return -KEY_lock;
5357 if (strEQ(d,"local")) return KEY_local;
5358 if (strEQ(d,"lstat")) return -KEY_lstat;
5361 if (strEQ(d,"length")) return -KEY_length;
5362 if (strEQ(d,"listen")) return -KEY_listen;
5365 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5368 if (strEQ(d,"localtime")) return -KEY_localtime;
5374 case 1: return KEY_m;
5376 if (strEQ(d,"my")) return KEY_my;
5379 if (strEQ(d,"map")) return KEY_map;
5382 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5385 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5386 if (strEQ(d,"msgget")) return -KEY_msgget;
5387 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5388 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5393 if (strEQ(d,"next")) return KEY_next;
5394 if (strEQ(d,"ne")) return -KEY_ne;
5395 if (strEQ(d,"not")) return -KEY_not;
5396 if (strEQ(d,"no")) return KEY_no;
5401 if (strEQ(d,"or")) return -KEY_or;
5404 if (strEQ(d,"ord")) return -KEY_ord;
5405 if (strEQ(d,"oct")) return -KEY_oct;
5406 if (strEQ(d,"our")) return KEY_our;
5409 if (strEQ(d,"open")) return -KEY_open;
5412 if (strEQ(d,"opendir")) return -KEY_opendir;
5419 if (strEQ(d,"pop")) return KEY_pop;
5420 if (strEQ(d,"pos")) return KEY_pos;
5423 if (strEQ(d,"push")) return KEY_push;
5424 if (strEQ(d,"pack")) return -KEY_pack;
5425 if (strEQ(d,"pipe")) return -KEY_pipe;
5428 if (strEQ(d,"print")) return KEY_print;
5431 if (strEQ(d,"printf")) return KEY_printf;
5434 if (strEQ(d,"package")) return KEY_package;
5437 if (strEQ(d,"prototype")) return KEY_prototype;
5442 if (strEQ(d,"q")) return KEY_q;
5443 if (strEQ(d,"qr")) return KEY_qr;
5444 if (strEQ(d,"qq")) return KEY_qq;
5445 if (strEQ(d,"qw")) return KEY_qw;
5446 if (strEQ(d,"qx")) return KEY_qx;
5448 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5453 if (strEQ(d,"ref")) return -KEY_ref;
5456 if (strEQ(d,"read")) return -KEY_read;
5457 if (strEQ(d,"rand")) return -KEY_rand;
5458 if (strEQ(d,"recv")) return -KEY_recv;
5459 if (strEQ(d,"redo")) return KEY_redo;
5462 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5463 if (strEQ(d,"reset")) return -KEY_reset;
5466 if (strEQ(d,"return")) return KEY_return;
5467 if (strEQ(d,"rename")) return -KEY_rename;
5468 if (strEQ(d,"rindex")) return -KEY_rindex;
5471 if (strEQ(d,"require")) return -KEY_require;
5472 if (strEQ(d,"reverse")) return -KEY_reverse;
5473 if (strEQ(d,"readdir")) return -KEY_readdir;
5476 if (strEQ(d,"readlink")) return -KEY_readlink;
5477 if (strEQ(d,"readline")) return -KEY_readline;
5478 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5481 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5487 case 0: return KEY_s;
5489 if (strEQ(d,"scalar")) return KEY_scalar;
5494 if (strEQ(d,"seek")) return -KEY_seek;
5495 if (strEQ(d,"send")) return -KEY_send;
5498 if (strEQ(d,"semop")) return -KEY_semop;
5501 if (strEQ(d,"select")) return -KEY_select;
5502 if (strEQ(d,"semctl")) return -KEY_semctl;
5503 if (strEQ(d,"semget")) return -KEY_semget;
5506 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5507 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5510 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5511 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5514 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5517 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5518 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5519 if (strEQ(d,"setservent")) return -KEY_setservent;
5522 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5523 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5530 if (strEQ(d,"shift")) return KEY_shift;
5533 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5534 if (strEQ(d,"shmget")) return -KEY_shmget;
5537 if (strEQ(d,"shmread")) return -KEY_shmread;
5540 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5541 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5546 if (strEQ(d,"sin")) return -KEY_sin;
5549 if (strEQ(d,"sleep")) return -KEY_sleep;
5552 if (strEQ(d,"sort")) return KEY_sort;
5553 if (strEQ(d,"socket")) return -KEY_socket;
5554 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5557 if (strEQ(d,"split")) return KEY_split;
5558 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5559 if (strEQ(d,"splice")) return KEY_splice;
5562 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5565 if (strEQ(d,"srand")) return -KEY_srand;
5568 if (strEQ(d,"stat")) return -KEY_stat;
5569 if (strEQ(d,"study")) return KEY_study;
5572 if (strEQ(d,"substr")) return -KEY_substr;
5573 if (strEQ(d,"sub")) return KEY_sub;
5578 if (strEQ(d,"system")) return -KEY_system;
5581 if (strEQ(d,"symlink")) return -KEY_symlink;
5582 if (strEQ(d,"syscall")) return -KEY_syscall;
5583 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5584 if (strEQ(d,"sysread")) return -KEY_sysread;
5585 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5588 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5597 if (strEQ(d,"tr")) return KEY_tr;
5600 if (strEQ(d,"tie")) return KEY_tie;
5603 if (strEQ(d,"tell")) return -KEY_tell;
5604 if (strEQ(d,"tied")) return KEY_tied;
5605 if (strEQ(d,"time")) return -KEY_time;
5608 if (strEQ(d,"times")) return -KEY_times;
5611 if (strEQ(d,"telldir")) return -KEY_telldir;
5614 if (strEQ(d,"truncate")) return -KEY_truncate;
5621 if (strEQ(d,"uc")) return -KEY_uc;
5624 if (strEQ(d,"use")) return KEY_use;
5627 if (strEQ(d,"undef")) return KEY_undef;
5628 if (strEQ(d,"until")) return KEY_until;
5629 if (strEQ(d,"untie")) return KEY_untie;
5630 if (strEQ(d,"utime")) return -KEY_utime;
5631 if (strEQ(d,"umask")) return -KEY_umask;
5634 if (strEQ(d,"unless")) return KEY_unless;
5635 if (strEQ(d,"unpack")) return -KEY_unpack;
5636 if (strEQ(d,"unlink")) return -KEY_unlink;
5639 if (strEQ(d,"unshift")) return KEY_unshift;
5640 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5645 if (strEQ(d,"values")) return -KEY_values;
5646 if (strEQ(d,"vec")) return -KEY_vec;
5651 if (strEQ(d,"warn")) return -KEY_warn;
5652 if (strEQ(d,"wait")) return -KEY_wait;
5655 if (strEQ(d,"while")) return KEY_while;
5656 if (strEQ(d,"write")) return -KEY_write;
5659 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5662 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5667 if (len == 1) return -KEY_x;
5668 if (strEQ(d,"xor")) return -KEY_xor;
5671 if (len == 1) return KEY_y;
5680 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5684 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5685 dTHR; /* only for ckWARN */
5686 if (ckWARN(WARN_SYNTAX)) {
5688 for (w = s+2; *w && level; w++) {
5695 for (; *w && isSPACE(*w); w++) ;
5696 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5697 Perl_warner(aTHX_ WARN_SYNTAX,
5698 "%s (...) interpreted as function",name);
5701 while (s < PL_bufend && isSPACE(*s))
5705 while (s < PL_bufend && isSPACE(*s))
5707 if (isIDFIRST_lazy_if(s,UTF)) {
5709 while (isALNUM_lazy_if(s,UTF))
5711 while (s < PL_bufend && isSPACE(*s))
5716 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5720 Perl_croak(aTHX_ "No comma allowed after %s", what);
5725 /* Either returns sv, or mortalizes sv and returns a new SV*.
5726 Best used as sv=new_constant(..., sv, ...).
5727 If s, pv are NULL, calls subroutine with one argument,
5728 and type is used with error messages only. */
5731 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5735 HV *table = GvHV(PL_hintgv); /* ^H */
5739 const char *why1, *why2, *why3;
5741 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
5744 why1 = "%^H is not consistent";
5745 why2 = strEQ(key,"charnames")
5746 ? " (missing \"use charnames ...\"?)"
5750 msg = Perl_newSVpvf(aTHX_ "constant(%s): %s%s%s",
5751 (type ? type: "undef"), why1, why2, why3);
5752 yyerror(SvPVX(msg));
5756 cvp = hv_fetch(table, key, strlen(key), FALSE);
5757 if (!cvp || !SvOK(*cvp)) {
5760 why3 = "} is not defined";
5763 sv_2mortal(sv); /* Parent created it permanently */
5766 pv = sv_2mortal(newSVpvn(s, len));
5768 typesv = sv_2mortal(newSVpv(type, 0));
5770 typesv = &PL_sv_undef;
5772 PUSHSTACKi(PERLSI_OVERLOAD);
5784 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5788 /* Check the eval first */
5789 if (!PL_in_eval && SvTRUE(ERRSV)) {
5791 sv_catpv(ERRSV, "Propagated");
5792 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5794 res = SvREFCNT_inc(sv);
5798 (void)SvREFCNT_inc(res);
5807 why1 = "Call to &{$^H{";
5809 why3 = "}} did not return a defined value";
5818 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5820 register char *d = dest;
5821 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5824 Perl_croak(aTHX_ ident_too_long);
5825 if (isALNUM(*s)) /* UTF handled below */
5827 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
5832 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5836 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5837 char *t = s + UTF8SKIP(s);
5838 while (*t & 0x80 && is_utf8_mark((U8*)t))
5840 if (d + (t - s) > e)
5841 Perl_croak(aTHX_ ident_too_long);
5842 Copy(s, d, t - s, char);
5855 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5865 e = d + destlen - 3; /* two-character token, ending NUL */
5867 while (isDIGIT(*s)) {
5869 Perl_croak(aTHX_ ident_too_long);
5876 Perl_croak(aTHX_ ident_too_long);
5877 if (isALNUM(*s)) /* UTF handled below */
5879 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
5884 else if (*s == ':' && s[1] == ':') {
5888 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5889 char *t = s + UTF8SKIP(s);
5890 while (*t & 0x80 && is_utf8_mark((U8*)t))
5892 if (d + (t - s) > e)
5893 Perl_croak(aTHX_ ident_too_long);
5894 Copy(s, d, t - s, char);
5905 if (PL_lex_state != LEX_NORMAL)
5906 PL_lex_state = LEX_INTERPENDMAYBE;
5909 if (*s == '$' && s[1] &&
5910 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
5923 if (*d == '^' && *s && isCONTROLVAR(*s)) {
5928 if (isSPACE(s[-1])) {
5931 if (!SPACE_OR_TAB(ch)) {
5937 if (isIDFIRST_lazy_if(d,UTF)) {
5941 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
5943 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
5946 Copy(s, d, e - s, char);
5951 while ((isALNUM(*s) || *s == ':') && d < e)
5954 Perl_croak(aTHX_ ident_too_long);
5957 while (s < send && SPACE_OR_TAB(*s)) s++;
5958 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
5959 dTHR; /* only for ckWARN */
5960 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
5961 const char *brack = *s == '[' ? "[...]" : "{...}";
5962 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5963 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
5964 funny, dest, brack, funny, dest, brack);
5967 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
5971 /* Handle extended ${^Foo} variables
5972 * 1999-02-27 mjd-perl-patch@plover.com */
5973 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
5977 while (isALNUM(*s) && d < e) {
5981 Perl_croak(aTHX_ ident_too_long);
5986 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
5987 PL_lex_state = LEX_INTERPEND;
5990 if (PL_lex_state == LEX_NORMAL) {
5991 dTHR; /* only for ckWARN */
5992 if (ckWARN(WARN_AMBIGUOUS) &&
5993 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
5995 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5996 "Ambiguous use of %c{%s} resolved to %c%s",
5997 funny, dest, funny, dest);
6002 s = bracket; /* let the parser handle it */
6006 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
6007 PL_lex_state = LEX_INTERPEND;
6012 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
6017 *pmfl |= PMf_GLOBAL;
6019 *pmfl |= PMf_CONTINUE;
6023 *pmfl |= PMf_MULTILINE;
6025 *pmfl |= PMf_SINGLELINE;
6027 *pmfl |= PMf_EXTENDED;
6031 S_scan_pat(pTHX_ char *start, I32 type)
6036 s = scan_str(start,FALSE,FALSE);
6039 SvREFCNT_dec(PL_lex_stuff);
6040 PL_lex_stuff = Nullsv;
6041 Perl_croak(aTHX_ "Search pattern not terminated");
6044 pm = (PMOP*)newPMOP(type, 0);
6045 if (PL_multi_open == '?')
6046 pm->op_pmflags |= PMf_ONCE;
6048 while (*s && strchr("iomsx", *s))
6049 pmflag(&pm->op_pmflags,*s++);
6052 while (*s && strchr("iogcmsx", *s))
6053 pmflag(&pm->op_pmflags,*s++);
6055 pm->op_pmpermflags = pm->op_pmflags;
6057 PL_lex_op = (OP*)pm;
6058 yylval.ival = OP_MATCH;
6063 S_scan_subst(pTHX_ char *start)
6070 yylval.ival = OP_NULL;
6072 s = scan_str(start,FALSE,FALSE);
6076 SvREFCNT_dec(PL_lex_stuff);
6077 PL_lex_stuff = Nullsv;
6078 Perl_croak(aTHX_ "Substitution pattern not terminated");
6081 if (s[-1] == PL_multi_open)
6084 first_start = PL_multi_start;
6085 s = scan_str(s,FALSE,FALSE);
6088 SvREFCNT_dec(PL_lex_stuff);
6089 PL_lex_stuff = Nullsv;
6091 SvREFCNT_dec(PL_lex_repl);
6092 PL_lex_repl = Nullsv;
6093 Perl_croak(aTHX_ "Substitution replacement not terminated");
6095 PL_multi_start = first_start; /* so whole substitution is taken together */
6097 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6103 else if (strchr("iogcmsx", *s))
6104 pmflag(&pm->op_pmflags,*s++);
6111 PL_sublex_info.super_bufptr = s;
6112 PL_sublex_info.super_bufend = PL_bufend;
6114 pm->op_pmflags |= PMf_EVAL;
6115 repl = newSVpvn("",0);
6117 sv_catpv(repl, es ? "eval " : "do ");
6118 sv_catpvn(repl, "{ ", 2);
6119 sv_catsv(repl, PL_lex_repl);
6120 sv_catpvn(repl, " };", 2);
6122 SvREFCNT_dec(PL_lex_repl);
6126 pm->op_pmpermflags = pm->op_pmflags;
6127 PL_lex_op = (OP*)pm;
6128 yylval.ival = OP_SUBST;
6133 S_scan_trans(pTHX_ char *start)
6144 yylval.ival = OP_NULL;
6146 s = scan_str(start,FALSE,FALSE);
6149 SvREFCNT_dec(PL_lex_stuff);
6150 PL_lex_stuff = Nullsv;
6151 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6153 if (s[-1] == PL_multi_open)
6156 s = scan_str(s,FALSE,FALSE);
6159 SvREFCNT_dec(PL_lex_stuff);
6160 PL_lex_stuff = Nullsv;
6162 SvREFCNT_dec(PL_lex_repl);
6163 PL_lex_repl = Nullsv;
6164 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6167 New(803,tbl,256,short);
6168 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6170 complement = del = squash = 0;
6171 while (strchr("cds", *s)) {
6173 complement = OPpTRANS_COMPLEMENT;
6175 del = OPpTRANS_DELETE;
6177 squash = OPpTRANS_SQUASH;
6180 o->op_private = del|squash|complement;
6183 yylval.ival = OP_TRANS;
6188 S_scan_heredoc(pTHX_ register char *s)
6192 I32 op_type = OP_SCALAR;
6199 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6203 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6206 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6207 if (*peek && strchr("`'\"",*peek)) {
6210 s = delimcpy(d, e, s, PL_bufend, term, &len);
6220 if (!isALNUM_lazy_if(s,UTF))
6221 deprecate("bare << to mean <<\"\"");
6222 for (; isALNUM_lazy_if(s,UTF); s++) {
6227 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6228 Perl_croak(aTHX_ "Delimiter for here document is too long");
6231 len = d - PL_tokenbuf;
6232 #ifndef PERL_STRICT_CR
6233 d = strchr(s, '\r');
6237 while (s < PL_bufend) {
6243 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6252 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6257 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6258 herewas = newSVpvn(s,PL_bufend-s);
6260 s--, herewas = newSVpvn(s,d-s);
6261 s += SvCUR(herewas);
6263 tmpstr = NEWSV(87,79);
6264 sv_upgrade(tmpstr, SVt_PVIV);
6269 else if (term == '`') {
6270 op_type = OP_BACKTICK;
6271 SvIVX(tmpstr) = '\\';
6275 PL_multi_start = CopLINE(PL_curcop);
6276 PL_multi_open = PL_multi_close = '<';
6277 term = *PL_tokenbuf;
6278 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6279 char *bufptr = PL_sublex_info.super_bufptr;
6280 char *bufend = PL_sublex_info.super_bufend;
6281 char *olds = s - SvCUR(herewas);
6282 s = strchr(bufptr, '\n');
6286 while (s < bufend &&
6287 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6289 CopLINE_inc(PL_curcop);
6292 CopLINE_set(PL_curcop, PL_multi_start);
6293 missingterm(PL_tokenbuf);
6295 sv_setpvn(herewas,bufptr,d-bufptr+1);
6296 sv_setpvn(tmpstr,d+1,s-d);
6298 sv_catpvn(herewas,s,bufend-s);
6299 (void)strcpy(bufptr,SvPVX(herewas));
6306 while (s < PL_bufend &&
6307 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6309 CopLINE_inc(PL_curcop);
6311 if (s >= PL_bufend) {
6312 CopLINE_set(PL_curcop, PL_multi_start);
6313 missingterm(PL_tokenbuf);
6315 sv_setpvn(tmpstr,d+1,s-d);
6317 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6319 sv_catpvn(herewas,s,PL_bufend-s);
6320 sv_setsv(PL_linestr,herewas);
6321 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6322 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6325 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6326 while (s >= PL_bufend) { /* multiple line string? */
6328 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6329 CopLINE_set(PL_curcop, PL_multi_start);
6330 missingterm(PL_tokenbuf);
6332 CopLINE_inc(PL_curcop);
6333 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6334 #ifndef PERL_STRICT_CR
6335 if (PL_bufend - PL_linestart >= 2) {
6336 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6337 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6339 PL_bufend[-2] = '\n';
6341 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6343 else if (PL_bufend[-1] == '\r')
6344 PL_bufend[-1] = '\n';
6346 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6347 PL_bufend[-1] = '\n';
6349 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6350 SV *sv = NEWSV(88,0);
6352 sv_upgrade(sv, SVt_PVMG);
6353 sv_setsv(sv,PL_linestr);
6354 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6356 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6359 sv_catsv(PL_linestr,herewas);
6360 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6364 sv_catsv(tmpstr,PL_linestr);
6369 PL_multi_end = CopLINE(PL_curcop);
6370 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6371 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6372 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6374 SvREFCNT_dec(herewas);
6375 PL_lex_stuff = tmpstr;
6376 yylval.ival = op_type;
6381 takes: current position in input buffer
6382 returns: new position in input buffer
6383 side-effects: yylval and lex_op are set.
6388 <FH> read from filehandle
6389 <pkg::FH> read from package qualified filehandle
6390 <pkg'FH> read from package qualified filehandle
6391 <$fh> read from filehandle in $fh
6397 S_scan_inputsymbol(pTHX_ char *start)
6399 register char *s = start; /* current position in buffer */
6405 d = PL_tokenbuf; /* start of temp holding space */
6406 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6407 end = strchr(s, '\n');
6410 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6412 /* die if we didn't have space for the contents of the <>,
6413 or if it didn't end, or if we see a newline
6416 if (len >= sizeof PL_tokenbuf)
6417 Perl_croak(aTHX_ "Excessively long <> operator");
6419 Perl_croak(aTHX_ "Unterminated <> operator");
6424 Remember, only scalar variables are interpreted as filehandles by
6425 this code. Anything more complex (e.g., <$fh{$num}>) will be
6426 treated as a glob() call.
6427 This code makes use of the fact that except for the $ at the front,
6428 a scalar variable and a filehandle look the same.
6430 if (*d == '$' && d[1]) d++;
6432 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6433 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6436 /* If we've tried to read what we allow filehandles to look like, and
6437 there's still text left, then it must be a glob() and not a getline.
6438 Use scan_str to pull out the stuff between the <> and treat it
6439 as nothing more than a string.
6442 if (d - PL_tokenbuf != len) {
6443 yylval.ival = OP_GLOB;
6445 s = scan_str(start,FALSE,FALSE);
6447 Perl_croak(aTHX_ "Glob not terminated");
6451 /* we're in a filehandle read situation */
6454 /* turn <> into <ARGV> */
6456 (void)strcpy(d,"ARGV");
6458 /* if <$fh>, create the ops to turn the variable into a
6464 /* try to find it in the pad for this block, otherwise find
6465 add symbol table ops
6467 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6468 OP *o = newOP(OP_PADSV, 0);
6470 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6473 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6474 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6475 newUNOP(OP_RV2SV, 0,
6476 newGVOP(OP_GV, 0, gv)));
6478 PL_lex_op->op_flags |= OPf_SPECIAL;
6479 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6480 yylval.ival = OP_NULL;
6483 /* If it's none of the above, it must be a literal filehandle
6484 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6486 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6487 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6488 yylval.ival = OP_NULL;
6497 takes: start position in buffer
6498 keep_quoted preserve \ on the embedded delimiter(s)
6499 keep_delims preserve the delimiters around the string
6500 returns: position to continue reading from buffer
6501 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6502 updates the read buffer.
6504 This subroutine pulls a string out of the input. It is called for:
6505 q single quotes q(literal text)
6506 ' single quotes 'literal text'
6507 qq double quotes qq(interpolate $here please)
6508 " double quotes "interpolate $here please"
6509 qx backticks qx(/bin/ls -l)
6510 ` backticks `/bin/ls -l`
6511 qw quote words @EXPORT_OK = qw( func() $spam )
6512 m// regexp match m/this/
6513 s/// regexp substitute s/this/that/
6514 tr/// string transliterate tr/this/that/
6515 y/// string transliterate y/this/that/
6516 ($*@) sub prototypes sub foo ($)
6517 (stuff) sub attr parameters sub foo : attr(stuff)
6518 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6520 In most of these cases (all but <>, patterns and transliterate)
6521 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6522 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6523 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6526 It skips whitespace before the string starts, and treats the first
6527 character as the delimiter. If the delimiter is one of ([{< then
6528 the corresponding "close" character )]}> is used as the closing
6529 delimiter. It allows quoting of delimiters, and if the string has
6530 balanced delimiters ([{<>}]) it allows nesting.
6532 The lexer always reads these strings into lex_stuff, except in the
6533 case of the operators which take *two* arguments (s/// and tr///)
6534 when it checks to see if lex_stuff is full (presumably with the 1st
6535 arg to s or tr) and if so puts the string into lex_repl.
6540 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6543 SV *sv; /* scalar value: string */
6544 char *tmps; /* temp string, used for delimiter matching */
6545 register char *s = start; /* current position in the buffer */
6546 register char term; /* terminating character */
6547 register char *to; /* current position in the sv's data */
6548 I32 brackets = 1; /* bracket nesting level */
6549 bool has_utf = FALSE; /* is there any utf8 content? */
6551 /* skip space before the delimiter */
6555 /* mark where we are, in case we need to report errors */
6558 /* after skipping whitespace, the next character is the terminator */
6560 if ((term & 0x80) && UTF)
6563 /* mark where we are */
6564 PL_multi_start = CopLINE(PL_curcop);
6565 PL_multi_open = term;
6567 /* find corresponding closing delimiter */
6568 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6570 PL_multi_close = term;
6572 /* create a new SV to hold the contents. 87 is leak category, I'm
6573 assuming. 79 is the SV's initial length. What a random number. */
6575 sv_upgrade(sv, SVt_PVIV);
6577 (void)SvPOK_only(sv); /* validate pointer */
6579 /* move past delimiter and try to read a complete string */
6581 sv_catpvn(sv, s, 1);
6584 /* extend sv if need be */
6585 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6586 /* set 'to' to the next character in the sv's string */
6587 to = SvPVX(sv)+SvCUR(sv);
6589 /* if open delimiter is the close delimiter read unbridle */
6590 if (PL_multi_open == PL_multi_close) {
6591 for (; s < PL_bufend; s++,to++) {
6592 /* embedded newlines increment the current line number */
6593 if (*s == '\n' && !PL_rsfp)
6594 CopLINE_inc(PL_curcop);
6595 /* handle quoted delimiters */
6596 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6597 if (!keep_quoted && s[1] == term)
6599 /* any other quotes are simply copied straight through */
6603 /* terminate when run out of buffer (the for() condition), or
6604 have found the terminator */
6605 else if (*s == term)
6607 else if (!has_utf && (*s & 0x80) && UTF)
6613 /* if the terminator isn't the same as the start character (e.g.,
6614 matched brackets), we have to allow more in the quoting, and
6615 be prepared for nested brackets.
6618 /* read until we run out of string, or we find the terminator */
6619 for (; s < PL_bufend; s++,to++) {
6620 /* embedded newlines increment the line count */
6621 if (*s == '\n' && !PL_rsfp)
6622 CopLINE_inc(PL_curcop);
6623 /* backslashes can escape the open or closing characters */
6624 if (*s == '\\' && s+1 < PL_bufend) {
6626 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6631 /* allow nested opens and closes */
6632 else if (*s == PL_multi_close && --brackets <= 0)
6634 else if (*s == PL_multi_open)
6636 else if (!has_utf && (*s & 0x80) && UTF)
6641 /* terminate the copied string and update the sv's end-of-string */
6643 SvCUR_set(sv, to - SvPVX(sv));
6646 * this next chunk reads more into the buffer if we're not done yet
6650 break; /* handle case where we are done yet :-) */
6652 #ifndef PERL_STRICT_CR
6653 if (to - SvPVX(sv) >= 2) {
6654 if ((to[-2] == '\r' && to[-1] == '\n') ||
6655 (to[-2] == '\n' && to[-1] == '\r'))
6659 SvCUR_set(sv, to - SvPVX(sv));
6661 else if (to[-1] == '\r')
6664 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6668 /* if we're out of file, or a read fails, bail and reset the current
6669 line marker so we can report where the unterminated string began
6672 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6674 CopLINE_set(PL_curcop, PL_multi_start);
6677 /* we read a line, so increment our line counter */
6678 CopLINE_inc(PL_curcop);
6680 /* update debugger info */
6681 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6682 SV *sv = NEWSV(88,0);
6684 sv_upgrade(sv, SVt_PVMG);
6685 sv_setsv(sv,PL_linestr);
6686 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6689 /* having changed the buffer, we must update PL_bufend */
6690 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6693 /* at this point, we have successfully read the delimited string */
6696 sv_catpvn(sv, s, 1);
6699 PL_multi_end = CopLINE(PL_curcop);
6702 /* if we allocated too much space, give some back */
6703 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6704 SvLEN_set(sv, SvCUR(sv) + 1);
6705 Renew(SvPVX(sv), SvLEN(sv), char);
6708 /* decide whether this is the first or second quoted string we've read
6721 takes: pointer to position in buffer
6722 returns: pointer to new position in buffer
6723 side-effects: builds ops for the constant in yylval.op
6725 Read a number in any of the formats that Perl accepts:
6727 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6728 [\d_]+(\.[\d_]*)?[Ee](\d+)
6730 Underbars (_) are allowed in decimal numbers. If -w is on,
6731 underbars before a decimal point must be at three digit intervals.
6733 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6736 If it reads a number without a decimal point or an exponent, it will
6737 try converting the number to an integer and see if it can do so
6738 without loss of precision.
6742 Perl_scan_num(pTHX_ char *start)
6744 register char *s = start; /* current position in buffer */
6745 register char *d; /* destination in temp buffer */
6746 register char *e; /* end of temp buffer */
6747 NV nv; /* number read, as a double */
6748 SV *sv = Nullsv; /* place to put the converted number */
6749 bool floatit; /* boolean: int or float? */
6750 char *lastub = 0; /* position of last underbar */
6751 static char number_too_long[] = "Number too long";
6753 /* We use the first character to decide what type of number this is */
6757 Perl_croak(aTHX_ "panic: scan_num");
6759 /* if it starts with a 0, it could be an octal number, a decimal in
6760 0.13 disguise, or a hexadecimal number, or a binary number. */
6764 u holds the "number so far"
6765 shift the power of 2 of the base
6766 (hex == 4, octal == 3, binary == 1)
6767 overflowed was the number more than we can hold?
6769 Shift is used when we add a digit. It also serves as an "are
6770 we in octal/hex/binary?" indicator to disallow hex characters
6777 bool overflowed = FALSE;
6778 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6779 static char* bases[5] = { "", "binary", "", "octal",
6781 static char* Bases[5] = { "", "Binary", "", "Octal",
6783 static char *maxima[5] = { "",
6784 "0b11111111111111111111111111111111",
6788 char *base, *Base, *max;
6794 } else if (s[1] == 'b') {
6798 /* check for a decimal in disguise */
6799 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6801 /* so it must be octal */
6805 base = bases[shift];
6806 Base = Bases[shift];
6807 max = maxima[shift];
6809 /* read the rest of the number */
6811 /* x is used in the overflow test,
6812 b is the digit we're adding on. */
6817 /* if we don't mention it, we're done */
6826 /* 8 and 9 are not octal */
6829 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6833 case '2': case '3': case '4':
6834 case '5': case '6': case '7':
6836 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6840 b = *s++ & 15; /* ASCII digit -> value of digit */
6844 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6845 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6846 /* make sure they said 0x */
6851 /* Prepare to put the digit we have onto the end
6852 of the number so far. We check for overflows.
6857 x = u << shift; /* make room for the digit */
6859 if ((x >> shift) != u
6860 && !(PL_hints & HINT_NEW_BINARY)) {
6864 if (ckWARN_d(WARN_OVERFLOW))
6865 Perl_warner(aTHX_ WARN_OVERFLOW,
6866 "Integer overflow in %s number",
6869 u = x | b; /* add the digit to the end */
6872 n *= nvshift[shift];
6873 /* If an NV has not enough bits in its
6874 * mantissa to represent an UV this summing of
6875 * small low-order numbers is a waste of time
6876 * (because the NV cannot preserve the
6877 * low-order bits anyway): we could just
6878 * remember when did we overflow and in the
6879 * end just multiply n by the right
6887 /* if we get here, we had success: make a scalar value from
6894 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6895 Perl_warner(aTHX_ WARN_PORTABLE,
6896 "%s number > %s non-portable",
6903 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6904 Perl_warner(aTHX_ WARN_PORTABLE,
6905 "%s number > %s non-portable",
6910 if (PL_hints & HINT_NEW_BINARY)
6911 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6916 handle decimal numbers.
6917 we're also sent here when we read a 0 as the first digit
6919 case '1': case '2': case '3': case '4': case '5':
6920 case '6': case '7': case '8': case '9': case '.':
6923 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
6926 /* read next group of digits and _ and copy into d */
6927 while (isDIGIT(*s) || *s == '_') {
6928 /* skip underscores, checking for misplaced ones
6932 dTHR; /* only for ckWARN */
6933 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
6934 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6938 /* check for end of fixed-length buffer */
6940 Perl_croak(aTHX_ number_too_long);
6941 /* if we're ok, copy the character */
6946 /* final misplaced underbar check */
6947 if (lastub && s - lastub != 3) {
6949 if (ckWARN(WARN_SYNTAX))
6950 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6953 /* read a decimal portion if there is one. avoid
6954 3..5 being interpreted as the number 3. followed
6957 if (*s == '.' && s[1] != '.') {
6961 /* copy, ignoring underbars, until we run out of
6962 digits. Note: no misplaced underbar checks!
6964 for (; isDIGIT(*s) || *s == '_'; s++) {
6965 /* fixed length buffer check */
6967 Perl_croak(aTHX_ number_too_long);
6971 if (*s == '.' && isDIGIT(s[1])) {
6972 /* oops, it's really a v-string, but without the "v" */
6978 /* read exponent part, if present */
6979 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
6983 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
6984 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
6986 /* allow positive or negative exponent */
6987 if (*s == '+' || *s == '-')
6990 /* read digits of exponent (no underbars :-) */
6991 while (isDIGIT(*s)) {
6993 Perl_croak(aTHX_ number_too_long);
6998 /* terminate the string */
7001 /* make an sv from the string */
7004 #if defined(Strtol) && defined(Strtoul)
7007 strtol/strtoll sets errno to ERANGE if the number is too big
7008 for an integer. We try to do an integer conversion first
7009 if no characters indicating "float" have been found.
7016 if (*PL_tokenbuf == '-')
7017 iv = Strtol(PL_tokenbuf, (char**)NULL, 10);
7019 uv = Strtoul(PL_tokenbuf, (char**)NULL, 10);
7021 floatit = TRUE; /* Probably just too large. */
7022 else if (*PL_tokenbuf == '-')
7024 else if (uv <= IV_MAX)
7025 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7030 nv = Atof(PL_tokenbuf);
7035 No working strtou?ll?.
7037 Unfortunately atol() doesn't do range checks (returning
7038 LONG_MIN/LONG_MAX, and setting errno to ERANGE on overflows)
7039 everywhere [1], so we cannot use use atol() (or atoll()).
7040 If we could, they would be used, as Atol(), very much like
7041 Strtol() and Strtoul() are used above.
7043 [1] XXX Configure test needed to check for atol()
7044 (and atoll()) overflow behaviour XXX
7048 We need to do this the hard way. */
7050 nv = Atof(PL_tokenbuf);
7052 /* See if we can make do with an integer value without loss of
7053 precision. We use U_V to cast to a UV, because some
7054 compilers have issues. Then we try casting it back and see
7055 if it was the same [1]. We only do this if we know we
7056 specifically read an integer. If floatit is true, then we
7057 don't need to do the conversion at all.
7059 [1] Note that this is lossy if our NVs cannot preserve our
7060 UVs. There are metaconfig defines NV_PRESERVES_UV (a boolean)
7061 and NV_PRESERVES_UV_BITS (a number), but in general we really
7062 do hope all such potentially lossy platforms have strtou?ll?
7063 to do a lossless IV/UV conversion.
7065 Maybe could do some tricks with DBL_DIG, LDBL_DIG and
7066 DBL_MANT_DIG and LDBL_MANT_DIG (these are already available
7067 as NV_DIG and NV_MANT_DIG)?
7073 if (!floatit && (NV)uv == nv) {
7075 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7083 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7084 (PL_hints & HINT_NEW_INTEGER) )
7085 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7086 (floatit ? "float" : "integer"),
7090 /* if it starts with a v, it could be a v-string */
7096 while (isDIGIT(*pos) || *pos == '_')
7098 if (!isALPHA(*pos)) {
7100 U8 tmpbuf[UTF8_MAXLEN];
7103 s++; /* get past 'v' */
7106 sv_setpvn(sv, "", 0);
7109 if (*s == '0' && isDIGIT(s[1]))
7110 yyerror("Octal number in vector unsupported");
7113 /* this is atoi() that tolerates underscores */
7116 while (--end >= s) {
7121 rev += (*end - '0') * mult;
7123 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
7124 Perl_warner(aTHX_ WARN_OVERFLOW,
7125 "Integer overflow in decimal number");
7128 tmpend = uv_to_utf8(tmpbuf, rev);
7129 utf8 = utf8 || rev > 127;
7130 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
7131 if (*pos == '.' && isDIGIT(pos[1]))
7137 while (isDIGIT(*pos) || *pos == '_')
7145 sv_utf8_downgrade(sv, TRUE);
7152 /* make the op for the constant and return */
7155 yylval.opval = newSVOP(OP_CONST, 0, sv);
7157 yylval.opval = Nullop;
7163 S_scan_formline(pTHX_ register char *s)
7168 SV *stuff = newSVpvn("",0);
7169 bool needargs = FALSE;
7172 if (*s == '.' || *s == /*{*/'}') {
7174 #ifdef PERL_STRICT_CR
7175 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7177 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7179 if (*t == '\n' || t == PL_bufend)
7182 if (PL_in_eval && !PL_rsfp) {
7183 eol = strchr(s,'\n');
7188 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7190 for (t = s; t < eol; t++) {
7191 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7193 goto enough; /* ~~ must be first line in formline */
7195 if (*t == '@' || *t == '^')
7198 sv_catpvn(stuff, s, eol-s);
7199 #ifndef PERL_STRICT_CR
7200 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7201 char *end = SvPVX(stuff) + SvCUR(stuff);
7210 s = filter_gets(PL_linestr, PL_rsfp, 0);
7211 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7212 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7215 yyerror("Format not terminated");
7225 PL_lex_state = LEX_NORMAL;
7226 PL_nextval[PL_nexttoke].ival = 0;
7230 PL_lex_state = LEX_FORMLINE;
7231 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7233 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7237 SvREFCNT_dec(stuff);
7238 PL_lex_formbrack = 0;
7249 PL_cshlen = strlen(PL_cshname);
7254 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7257 I32 oldsavestack_ix = PL_savestack_ix;
7258 CV* outsidecv = PL_compcv;
7262 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7264 SAVEI32(PL_subline);
7265 save_item(PL_subname);
7268 SAVESPTR(PL_comppad_name);
7269 SAVESPTR(PL_compcv);
7270 SAVEI32(PL_comppad_name_fill);
7271 SAVEI32(PL_min_intro_pending);
7272 SAVEI32(PL_max_intro_pending);
7273 SAVEI32(PL_pad_reset_pending);
7275 PL_compcv = (CV*)NEWSV(1104,0);
7276 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7277 CvFLAGS(PL_compcv) |= flags;
7279 PL_comppad = newAV();
7280 av_push(PL_comppad, Nullsv);
7281 PL_curpad = AvARRAY(PL_comppad);
7282 PL_comppad_name = newAV();
7283 PL_comppad_name_fill = 0;
7284 PL_min_intro_pending = 0;
7286 PL_subline = CopLINE(PL_curcop);
7288 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7289 PL_curpad[0] = (SV*)newAV();
7290 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7291 #endif /* USE_THREADS */
7293 comppadlist = newAV();
7294 AvREAL_off(comppadlist);
7295 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7296 av_store(comppadlist, 1, (SV*)PL_comppad);
7298 CvPADLIST(PL_compcv) = comppadlist;
7299 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7301 CvOWNER(PL_compcv) = 0;
7302 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7303 MUTEX_INIT(CvMUTEXP(PL_compcv));
7304 #endif /* USE_THREADS */
7306 return oldsavestack_ix;
7310 Perl_yywarn(pTHX_ char *s)
7313 PL_in_eval |= EVAL_WARNONLY;
7315 PL_in_eval &= ~EVAL_WARNONLY;
7320 Perl_yyerror(pTHX_ char *s)
7324 char *context = NULL;
7328 if (!yychar || (yychar == ';' && !PL_rsfp))
7330 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7331 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7332 while (isSPACE(*PL_oldoldbufptr))
7334 context = PL_oldoldbufptr;
7335 contlen = PL_bufptr - PL_oldoldbufptr;
7337 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7338 PL_oldbufptr != PL_bufptr) {
7339 while (isSPACE(*PL_oldbufptr))
7341 context = PL_oldbufptr;
7342 contlen = PL_bufptr - PL_oldbufptr;
7344 else if (yychar > 255)
7345 where = "next token ???";
7346 #ifdef USE_PURE_BISON
7347 /* GNU Bison sets the value -2 */
7348 else if (yychar == -2) {
7350 else if ((yychar & 127) == 127) {
7352 if (PL_lex_state == LEX_NORMAL ||
7353 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7354 where = "at end of line";
7355 else if (PL_lex_inpat)
7356 where = "within pattern";
7358 where = "within string";
7361 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7363 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7364 else if (isPRINT_LC(yychar))
7365 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7367 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7368 where = SvPVX(where_sv);
7370 msg = sv_2mortal(newSVpv(s, 0));
7371 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7372 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7374 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7376 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7377 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7378 Perl_sv_catpvf(aTHX_ msg,
7379 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7380 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7383 if (PL_in_eval & EVAL_WARNONLY)
7384 Perl_warn(aTHX_ "%"SVf, msg);
7387 if (PL_error_count >= 10) {
7388 if (PL_in_eval && SvCUR(ERRSV))
7389 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
7390 ERRSV, CopFILE(PL_curcop));
7392 Perl_croak(aTHX_ "%s has too many errors.\n",
7393 CopFILE(PL_curcop));
7396 PL_in_my_stash = Nullhv;
7401 S_swallow_bom(pTHX_ U8 *s)
7404 slen = SvCUR(PL_linestr);
7408 /* UTF-16 little-endian */
7409 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
7410 Perl_croak(aTHX_ "Unsupported script encoding");
7411 #ifndef PERL_NO_UTF16_FILTER
7412 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-LE script encoding\n"));
7414 if (PL_bufend > (char*)s) {
7418 filter_add(utf16rev_textfilter, NULL);
7419 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7420 PL_bufend = (char*)utf16_to_utf8_reversed(s, news,
7421 PL_bufend - (char*)s - 1,
7423 Copy(news, s, newlen, U8);
7424 SvCUR_set(PL_linestr, newlen);
7425 PL_bufend = SvPVX(PL_linestr) + newlen;
7426 news[newlen++] = '\0';
7430 Perl_croak(aTHX_ "Unsupported script encoding");
7435 if (s[1] == 0xFF) { /* UTF-16 big-endian */
7436 #ifndef PERL_NO_UTF16_FILTER
7437 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding\n"));
7439 if (PL_bufend > (char *)s) {
7443 filter_add(utf16_textfilter, NULL);
7444 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7445 PL_bufend = (char*)utf16_to_utf8(s, news,
7446 PL_bufend - (char*)s,
7448 Copy(news, s, newlen, U8);
7449 SvCUR_set(PL_linestr, newlen);
7450 PL_bufend = SvPVX(PL_linestr) + newlen;
7451 news[newlen++] = '\0';
7455 Perl_croak(aTHX_ "Unsupported script encoding");
7460 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
7461 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-8 script encoding\n"));
7466 if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */
7467 s[2] == 0xFE && s[3] == 0xFF)
7469 Perl_croak(aTHX_ "Unsupported script encoding");
7481 * Restore a source filter.
7485 restore_rsfp(pTHXo_ void *f)
7487 PerlIO *fp = (PerlIO*)f;
7489 if (PL_rsfp == PerlIO_stdin())
7490 PerlIO_clearerr(PL_rsfp);
7491 else if (PL_rsfp && (PL_rsfp != fp))
7492 PerlIO_close(PL_rsfp);
7496 #ifndef PERL_NO_UTF16_FILTER
7498 utf16_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7500 I32 count = FILTER_READ(idx+1, sv, maxlen);
7505 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7506 if (!*SvPV_nolen(sv))
7507 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7510 tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7511 sv_usepvn(sv, (char*)tmps, tend - tmps);
7517 utf16rev_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7519 I32 count = FILTER_READ(idx+1, sv, maxlen);
7524 if (!*SvPV_nolen(sv))
7525 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7528 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7529 tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7530 sv_usepvn(sv, (char*)tmps, tend - tmps);