3 * Copyright (c) 1991-2000, Larry Wall
5 * You may distribute under the terms of either the GNU General Public
6 * License or the Artistic License, as specified in the README file.
11 * "It all comes from here, the stench and the peril." --Frodo
15 * This file is the lexer for Perl. It's closely linked to the
18 * The main routine is yylex(), which returns the next token.
22 #define PERL_IN_TOKE_C
25 #define yychar PL_yychar
26 #define yylval PL_yylval
28 static char ident_too_long[] = "Identifier too long";
30 static void restore_rsfp(pTHXo_ void *f);
32 #define XFAKEBRACK 128
35 /*#define UTF (SvUTF8(PL_linestr) && !(PL_hints & HINT_BYTE))*/
36 #define UTF (PL_hints & HINT_UTF8)
38 /* In variables name $^X, these are the legal values for X.
39 * 1999-02-27 mjd-perl-patch@plover.com */
40 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
42 /* LEX_* are values for PL_lex_state, the state of the lexer.
43 * They are arranged oddly so that the guard on the switch statement
44 * can get by with a single comparison (if the compiler is smart enough).
47 /* #define LEX_NOTPARSING 11 is done in perl.h. */
50 #define LEX_INTERPNORMAL 9
51 #define LEX_INTERPCASEMOD 8
52 #define LEX_INTERPPUSH 7
53 #define LEX_INTERPSTART 6
54 #define LEX_INTERPEND 5
55 #define LEX_INTERPENDMAYBE 4
56 #define LEX_INTERPCONCAT 3
57 #define LEX_INTERPCONST 2
58 #define LEX_FORMLINE 1
59 #define LEX_KNOWNEXT 0
61 /* XXX If this causes problems, set i_unistd=undef in the hint file. */
63 # include <unistd.h> /* Needed for execv() */
72 YYSTYPE* yylval_pointer = NULL;
73 int* yychar_pointer = NULL;
76 # define yylval (*yylval_pointer)
77 # define yychar (*yychar_pointer)
78 # define PERL_YYLEX_PARAM yylval_pointer,yychar_pointer
80 # define yylex() Perl_yylex(aTHX_ yylval_pointer, yychar_pointer)
85 /* CLINE is a macro that ensures PL_copline has a sane value */
90 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
93 * Convenience functions to return different tokens and prime the
94 * lexer for the next token. They all take an argument.
96 * TOKEN : generic token (used for '(', DOLSHARP, etc)
97 * OPERATOR : generic operator
98 * AOPERATOR : assignment operator
99 * PREBLOCK : beginning the block after an if, while, foreach, ...
100 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
101 * PREREF : *EXPR where EXPR is not a simple identifier
102 * TERM : expression term
103 * LOOPX : loop exiting command (goto, last, dump, etc)
104 * FTST : file test operator
105 * FUN0 : zero-argument function
106 * FUN1 : not used, except for not, which isn't a UNIOP
107 * BOop : bitwise or or xor
109 * SHop : shift operator
110 * PWop : power operator
111 * PMop : pattern-matching operator
112 * Aop : addition-level operator
113 * Mop : multiplication-level operator
114 * Eop : equality-testing operator
115 * Rop : relational operator <= != gt
117 * Also see LOP and lop() below.
120 #define TOKEN(retval) return (PL_bufptr = s,(int)retval)
121 #define OPERATOR(retval) return (PL_expect = XTERM,PL_bufptr = s,(int)retval)
122 #define AOPERATOR(retval) return ao((PL_expect = XTERM,PL_bufptr = s,(int)retval))
123 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
124 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
125 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s,(int)retval)
126 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR,PL_bufptr = s,(int)retval)
127 #define LOOPX(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
128 #define FTST(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)UNIOP)
129 #define FUN0(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
130 #define FUN1(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
131 #define BOop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
132 #define BAop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
133 #define SHop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
134 #define PWop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
135 #define PMop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
136 #define Aop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
137 #define Mop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
138 #define Eop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
139 #define Rop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)RELOP)
141 /* This bit of chicanery makes a unary function followed by
142 * a parenthesis into a function with one argument, highest precedence.
144 #define UNI(f) return(yylval.ival = f, \
147 PL_last_uni = PL_oldbufptr, \
148 PL_last_lop_op = f, \
149 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
151 #define UNIBRACK(f) return(yylval.ival = f, \
153 PL_last_uni = PL_oldbufptr, \
154 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
156 /* grandfather return to old style */
157 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
162 * This subroutine detects &&= and ||= and turns an ANDAND or OROR
163 * into an OP_ANDASSIGN or OP_ORASSIGN
167 S_ao(pTHX_ int toketype)
169 if (*PL_bufptr == '=') {
171 if (toketype == ANDAND)
172 yylval.ival = OP_ANDASSIGN;
173 else if (toketype == OROR)
174 yylval.ival = OP_ORASSIGN;
182 * When Perl expects an operator and finds something else, no_op
183 * prints the warning. It always prints "<something> found where
184 * operator expected. It prints "Missing semicolon on previous line?"
185 * if the surprise occurs at the start of the line. "do you need to
186 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
187 * where the compiler doesn't know if foo is a method call or a function.
188 * It prints "Missing operator before end of line" if there's nothing
189 * after the missing operator, or "... before <...>" if there is something
190 * after the missing operator.
194 S_no_op(pTHX_ char *what, char *s)
196 char *oldbp = PL_bufptr;
197 bool is_first = (PL_oldbufptr == PL_linestart);
205 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
207 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
208 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
210 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
211 if (t < PL_bufptr && isSPACE(*t))
212 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
213 t - PL_oldoldbufptr, PL_oldoldbufptr);
216 Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
222 * Complain about missing quote/regexp/heredoc terminator.
223 * If it's called with (char *)NULL then it cauterizes the line buffer.
224 * If we're in a delimited string and the delimiter is a control
225 * character, it's reformatted into a two-char sequence like ^C.
230 S_missingterm(pTHX_ char *s)
235 char *nl = strrchr(s,'\n');
241 iscntrl(PL_multi_close)
243 PL_multi_close < 32 || PL_multi_close == 127
247 tmpbuf[1] = toCTRL(PL_multi_close);
253 *tmpbuf = PL_multi_close;
257 q = strchr(s,'"') ? '\'' : '"';
258 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
266 Perl_deprecate(pTHX_ char *s)
269 if (ckWARN(WARN_DEPRECATED))
270 Perl_warner(aTHX_ WARN_DEPRECATED, "Use of %s is deprecated", s);
275 * Deprecate a comma-less variable list.
281 deprecate("comma-less variable list");
285 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
286 * utf16-to-utf8-reversed.
289 #ifdef PERL_CR_FILTER
293 register char *s = SvPVX(sv);
294 register char *e = s + SvCUR(sv);
295 /* outer loop optimized to do nothing if there are no CR-LFs */
297 if (*s++ == '\r' && *s == '\n') {
298 /* hit a CR-LF, need to copy the rest */
299 register char *d = s - 1;
302 if (*s == '\r' && s[1] == '\n')
313 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
315 I32 count = FILTER_READ(idx+1, sv, maxlen);
316 if (count > 0 && !maxlen)
323 S_utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
325 I32 count = FILTER_READ(idx+1, sv, maxlen);
329 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
330 tend = utf16_to_utf8((U16*)SvPVX(sv), tmps, SvCUR(sv));
331 sv_usepvn(sv, (char*)tmps, tend - tmps);
338 S_utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
340 I32 count = FILTER_READ(idx+1, sv, maxlen);
344 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
345 tend = utf16_to_utf8_reversed((U16*)SvPVX(sv), tmps, SvCUR(sv));
346 sv_usepvn(sv, (char*)tmps, tend - tmps);
354 * Initialize variables. Uses the Perl save_stack to save its state (for
355 * recursive calls to the parser).
359 Perl_lex_start(pTHX_ SV *line)
365 SAVEI32(PL_lex_dojoin);
366 SAVEI32(PL_lex_brackets);
367 SAVEI32(PL_lex_casemods);
368 SAVEI32(PL_lex_starts);
369 SAVEI32(PL_lex_state);
370 SAVEVPTR(PL_lex_inpat);
371 SAVEI32(PL_lex_inwhat);
372 if (PL_lex_state == LEX_KNOWNEXT) {
373 I32 toke = PL_nexttoke;
374 while (--toke >= 0) {
375 SAVEI32(PL_nexttype[toke]);
376 SAVEVPTR(PL_nextval[toke]);
378 SAVEI32(PL_nexttoke);
381 SAVECOPLINE(PL_curcop);
384 SAVEPPTR(PL_oldbufptr);
385 SAVEPPTR(PL_oldoldbufptr);
386 SAVEPPTR(PL_linestart);
387 SAVESPTR(PL_linestr);
388 SAVEPPTR(PL_lex_brackstack);
389 SAVEPPTR(PL_lex_casestack);
390 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
391 SAVESPTR(PL_lex_stuff);
392 SAVEI32(PL_lex_defer);
393 SAVEI32(PL_sublex_info.sub_inwhat);
394 SAVESPTR(PL_lex_repl);
396 SAVEINT(PL_lex_expect);
398 PL_lex_state = LEX_NORMAL;
402 New(899, PL_lex_brackstack, 120, char);
403 New(899, PL_lex_casestack, 12, char);
404 SAVEFREEPV(PL_lex_brackstack);
405 SAVEFREEPV(PL_lex_casestack);
407 *PL_lex_casestack = '\0';
410 PL_lex_stuff = Nullsv;
411 PL_lex_repl = Nullsv;
414 PL_sublex_info.sub_inwhat = 0;
416 if (SvREADONLY(PL_linestr))
417 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
418 s = SvPV(PL_linestr, len);
419 if (len && s[len-1] != ';') {
420 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
421 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
422 sv_catpvn(PL_linestr, "\n;", 2);
424 SvTEMP_off(PL_linestr);
425 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
426 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
428 PL_rs = newSVpvn("\n", 1);
434 * Finalizer for lexing operations. Must be called when the parser is
435 * done with the lexer.
441 PL_doextract = FALSE;
446 * This subroutine has nothing to do with tilting, whether at windmills
447 * or pinball tables. Its name is short for "increment line". It
448 * increments the current line number in CopLINE(PL_curcop) and checks
449 * to see whether the line starts with a comment of the form
450 * # line 500 "foo.pm"
451 * If so, it sets the current line number and file to the values in the comment.
455 S_incline(pTHX_ char *s)
463 CopLINE_inc(PL_curcop);
466 while (*s == ' ' || *s == '\t') s++;
467 if (strnEQ(s, "line", 4))
471 if (*s == ' ' || *s == '\t')
475 while (*s == ' ' || *s == '\t') s++;
481 while (*s == ' ' || *s == '\t')
483 if (*s == '"' && (t = strchr(s+1, '"'))) {
488 for (t = s; !isSPACE(*t); t++) ;
491 while (*e == ' ' || *e == '\t' || *e == '\r' || *e == '\f')
493 if (*e != '\n' && *e != '\0')
494 return; /* false alarm */
499 CopFILE_set(PL_curcop, s);
501 CopLINE_set(PL_curcop, atoi(n)-1);
506 * Called to gobble the appropriate amount and type of whitespace.
507 * Skips comments as well.
511 S_skipspace(pTHX_ register char *s)
514 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
515 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
521 SSize_t oldprevlen, oldoldprevlen;
522 SSize_t oldloplen, oldunilen;
523 while (s < PL_bufend && isSPACE(*s)) {
524 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
529 if (s < PL_bufend && *s == '#') {
530 while (s < PL_bufend && *s != '\n')
534 if (PL_in_eval && !PL_rsfp) {
541 /* only continue to recharge the buffer if we're at the end
542 * of the buffer, we're not reading from a source filter, and
543 * we're in normal lexing mode
545 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
546 PL_lex_state == LEX_FORMLINE)
549 /* try to recharge the buffer */
550 if ((s = filter_gets(PL_linestr, PL_rsfp,
551 (prevlen = SvCUR(PL_linestr)))) == Nullch)
553 /* end of file. Add on the -p or -n magic */
554 if (PL_minus_n || PL_minus_p) {
555 sv_setpv(PL_linestr,PL_minus_p ?
556 ";}continue{print or die qq(-p destination: $!\\n)" :
558 sv_catpv(PL_linestr,";}");
559 PL_minus_n = PL_minus_p = 0;
562 sv_setpv(PL_linestr,";");
564 /* reset variables for next time we lex */
565 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
567 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
569 /* Close the filehandle. Could be from -P preprocessor,
570 * STDIN, or a regular file. If we were reading code from
571 * STDIN (because the commandline held no -e or filename)
572 * then we don't close it, we reset it so the code can
573 * read from STDIN too.
576 if (PL_preprocess && !PL_in_eval)
577 (void)PerlProc_pclose(PL_rsfp);
578 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
579 PerlIO_clearerr(PL_rsfp);
581 (void)PerlIO_close(PL_rsfp);
586 /* not at end of file, so we only read another line */
587 /* make corresponding updates to old pointers, for yyerror() */
588 oldprevlen = PL_oldbufptr - PL_bufend;
589 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
591 oldunilen = PL_last_uni - PL_bufend;
593 oldloplen = PL_last_lop - PL_bufend;
594 PL_linestart = PL_bufptr = s + prevlen;
595 PL_bufend = s + SvCUR(PL_linestr);
597 PL_oldbufptr = s + oldprevlen;
598 PL_oldoldbufptr = s + oldoldprevlen;
600 PL_last_uni = s + oldunilen;
602 PL_last_lop = s + oldloplen;
605 /* debugger active and we're not compiling the debugger code,
606 * so store the line into the debugger's array of lines
608 if (PERLDB_LINE && PL_curstash != PL_debstash) {
609 SV *sv = NEWSV(85,0);
611 sv_upgrade(sv, SVt_PVMG);
612 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
613 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
620 * Check the unary operators to ensure there's no ambiguity in how they're
621 * used. An ambiguous piece of code would be:
623 * This doesn't mean rand() + 5. Because rand() is a unary operator,
624 * the +5 is its argument.
634 if (PL_oldoldbufptr != PL_last_uni)
636 while (isSPACE(*PL_last_uni))
638 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
639 if ((t = strchr(s, '(')) && t < PL_bufptr)
641 if (ckWARN_d(WARN_AMBIGUOUS)){
644 Perl_warner(aTHX_ WARN_AMBIGUOUS,
645 "Warning: Use of \"%s\" without parens is ambiguous",
651 /* workaround to replace the UNI() macro with a function. Only the
652 * hints/uts.sh file mentions this. Other comments elsewhere in the
653 * source indicate Microport Unix might need it too.
659 #define UNI(f) return uni(f,s)
662 S_uni(pTHX_ I32 f, char *s)
667 PL_last_uni = PL_oldbufptr;
678 #endif /* CRIPPLED_CC */
681 * LOP : macro to build a list operator. Its behaviour has been replaced
682 * with a subroutine, S_lop() for which LOP is just another name.
685 #define LOP(f,x) return lop(f,x,s)
689 * Build a list operator (or something that might be one). The rules:
690 * - if we have a next token, then it's a list operator [why?]
691 * - if the next thing is an opening paren, then it's a function
692 * - else it's a list operator
696 S_lop(pTHX_ I32 f, int x, char *s)
703 PL_last_lop = PL_oldbufptr;
718 * When the lexer realizes it knows the next token (for instance,
719 * it is reordering tokens for the parser) then it can call S_force_next
720 * to know what token to return the next time the lexer is called. Caller
721 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
722 * handles the token correctly.
726 S_force_next(pTHX_ I32 type)
728 PL_nexttype[PL_nexttoke] = type;
730 if (PL_lex_state != LEX_KNOWNEXT) {
731 PL_lex_defer = PL_lex_state;
732 PL_lex_expect = PL_expect;
733 PL_lex_state = LEX_KNOWNEXT;
739 * When the lexer knows the next thing is a word (for instance, it has
740 * just seen -> and it knows that the next char is a word char, then
741 * it calls S_force_word to stick the next word into the PL_next lookahead.
744 * char *start : buffer position (must be within PL_linestr)
745 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
746 * int check_keyword : if true, Perl checks to make sure the word isn't
747 * a keyword (do this if the word is a label, e.g. goto FOO)
748 * int allow_pack : if true, : characters will also be allowed (require,
750 * int allow_initial_tick : used by the "sub" lexer only.
754 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
759 start = skipspace(start);
761 if (isIDFIRST_lazy_if(s,UTF) ||
762 (allow_pack && *s == ':') ||
763 (allow_initial_tick && *s == '\'') )
765 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
766 if (check_keyword && keyword(PL_tokenbuf, len))
768 if (token == METHOD) {
773 PL_expect = XOPERATOR;
776 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
777 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
785 * Called when the lexer wants $foo *foo &foo etc, but the program
786 * text only contains the "foo" portion. The first argument is a pointer
787 * to the "foo", and the second argument is the type symbol to prefix.
788 * Forces the next token to be a "WORD".
789 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
793 S_force_ident(pTHX_ register char *s, int kind)
796 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
797 PL_nextval[PL_nexttoke].opval = o;
800 dTHR; /* just for in_eval */
801 o->op_private = OPpCONST_ENTERED;
802 /* XXX see note in pp_entereval() for why we forgo typo
803 warnings if the symbol must be introduced in an eval.
805 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
806 kind == '$' ? SVt_PV :
807 kind == '@' ? SVt_PVAV :
808 kind == '%' ? SVt_PVHV :
816 Perl_str_to_version(pTHX_ SV *sv)
821 char *start = SvPVx(sv,len);
822 bool utf = SvUTF8(sv);
823 char *end = start + len;
824 while (start < end) {
828 n = utf8_to_uv((U8*)start, &skip);
833 retval += ((NV)n)/nshift;
842 * Forces the next token to be a version number.
846 S_force_version(pTHX_ char *s)
848 OP *version = Nullop;
857 for (; isDIGIT(*d) || *d == '_' || *d == '.'; d++);
858 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
861 version = yylval.opval;
862 ver = cSVOPx(version)->op_sv;
863 if (SvPOK(ver) && !SvNIOK(ver)) {
864 SvUPGRADE(ver, SVt_PVNV);
865 SvNVX(ver) = str_to_version(ver);
866 SvNOK_on(ver); /* hint that it is a version */
871 /* NOTE: The parser sees the package name and the VERSION swapped */
872 PL_nextval[PL_nexttoke].opval = version;
880 * Tokenize a quoted string passed in as an SV. It finds the next
881 * chunk, up to end of string or a backslash. It may make a new
882 * SV containing that chunk (if HINT_NEW_STRING is on). It also
887 S_tokeq(pTHX_ SV *sv)
898 s = SvPV_force(sv, len);
902 while (s < send && *s != '\\')
907 if ( PL_hints & HINT_NEW_STRING )
908 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
911 if (s + 1 < send && (s[1] == '\\'))
912 s++; /* all that, just for this */
917 SvCUR_set(sv, d - SvPVX(sv));
919 if ( PL_hints & HINT_NEW_STRING )
920 return new_constant(NULL, 0, "q", sv, pv, "q");
925 * Now come three functions related to double-quote context,
926 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
927 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
928 * interact with PL_lex_state, and create fake ( ... ) argument lists
929 * to handle functions and concatenation.
930 * They assume that whoever calls them will be setting up a fake
931 * join call, because each subthing puts a ',' after it. This lets
934 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
936 * (I'm not sure whether the spurious commas at the end of lcfirst's
937 * arguments and join's arguments are created or not).
942 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
944 * Pattern matching will set PL_lex_op to the pattern-matching op to
945 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
947 * OP_CONST and OP_READLINE are easy--just make the new op and return.
949 * Everything else becomes a FUNC.
951 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
952 * had an OP_CONST or OP_READLINE). This just sets us up for a
953 * call to S_sublex_push().
959 register I32 op_type = yylval.ival;
961 if (op_type == OP_NULL) {
962 yylval.opval = PL_lex_op;
966 if (op_type == OP_CONST || op_type == OP_READLINE) {
967 SV *sv = tokeq(PL_lex_stuff);
969 if (SvTYPE(sv) == SVt_PVIV) {
970 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
976 nsv = newSVpvn(p, len);
980 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
981 PL_lex_stuff = Nullsv;
985 PL_sublex_info.super_state = PL_lex_state;
986 PL_sublex_info.sub_inwhat = op_type;
987 PL_sublex_info.sub_op = PL_lex_op;
988 PL_lex_state = LEX_INTERPPUSH;
992 yylval.opval = PL_lex_op;
1002 * Create a new scope to save the lexing state. The scope will be
1003 * ended in S_sublex_done. Returns a '(', starting the function arguments
1004 * to the uc, lc, etc. found before.
1005 * Sets PL_lex_state to LEX_INTERPCONCAT.
1014 PL_lex_state = PL_sublex_info.super_state;
1015 SAVEI32(PL_lex_dojoin);
1016 SAVEI32(PL_lex_brackets);
1017 SAVEI32(PL_lex_casemods);
1018 SAVEI32(PL_lex_starts);
1019 SAVEI32(PL_lex_state);
1020 SAVEVPTR(PL_lex_inpat);
1021 SAVEI32(PL_lex_inwhat);
1022 SAVECOPLINE(PL_curcop);
1023 SAVEPPTR(PL_bufptr);
1024 SAVEPPTR(PL_oldbufptr);
1025 SAVEPPTR(PL_oldoldbufptr);
1026 SAVEPPTR(PL_linestart);
1027 SAVESPTR(PL_linestr);
1028 SAVEPPTR(PL_lex_brackstack);
1029 SAVEPPTR(PL_lex_casestack);
1031 PL_linestr = PL_lex_stuff;
1032 PL_lex_stuff = Nullsv;
1034 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1035 = SvPVX(PL_linestr);
1036 PL_bufend += SvCUR(PL_linestr);
1037 SAVEFREESV(PL_linestr);
1039 PL_lex_dojoin = FALSE;
1040 PL_lex_brackets = 0;
1041 New(899, PL_lex_brackstack, 120, char);
1042 New(899, PL_lex_casestack, 12, char);
1043 SAVEFREEPV(PL_lex_brackstack);
1044 SAVEFREEPV(PL_lex_casestack);
1045 PL_lex_casemods = 0;
1046 *PL_lex_casestack = '\0';
1048 PL_lex_state = LEX_INTERPCONCAT;
1049 CopLINE_set(PL_curcop, PL_multi_start);
1051 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1052 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1053 PL_lex_inpat = PL_sublex_info.sub_op;
1055 PL_lex_inpat = Nullop;
1062 * Restores lexer state after a S_sublex_push.
1068 if (!PL_lex_starts++) {
1069 PL_expect = XOPERATOR;
1070 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn("",0));
1074 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1075 PL_lex_state = LEX_INTERPCASEMOD;
1079 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1080 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1081 PL_linestr = PL_lex_repl;
1083 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1084 PL_bufend += SvCUR(PL_linestr);
1085 SAVEFREESV(PL_linestr);
1086 PL_lex_dojoin = FALSE;
1087 PL_lex_brackets = 0;
1088 PL_lex_casemods = 0;
1089 *PL_lex_casestack = '\0';
1091 if (SvEVALED(PL_lex_repl)) {
1092 PL_lex_state = LEX_INTERPNORMAL;
1094 /* we don't clear PL_lex_repl here, so that we can check later
1095 whether this is an evalled subst; that means we rely on the
1096 logic to ensure sublex_done() is called again only via the
1097 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1100 PL_lex_state = LEX_INTERPCONCAT;
1101 PL_lex_repl = Nullsv;
1107 PL_bufend = SvPVX(PL_linestr);
1108 PL_bufend += SvCUR(PL_linestr);
1109 PL_expect = XOPERATOR;
1110 PL_sublex_info.sub_inwhat = 0;
1118 Extracts a pattern, double-quoted string, or transliteration. This
1121 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1122 processing a pattern (PL_lex_inpat is true), a transliteration
1123 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1125 Returns a pointer to the character scanned up to. Iff this is
1126 advanced from the start pointer supplied (ie if anything was
1127 successfully parsed), will leave an OP for the substring scanned
1128 in yylval. Caller must intuit reason for not parsing further
1129 by looking at the next characters herself.
1133 double-quoted style: \r and \n
1134 regexp special ones: \D \s
1136 backrefs: \1 (deprecated in substitution replacements)
1137 case and quoting: \U \Q \E
1138 stops on @ and $, but not for $ as tail anchor
1140 In transliterations:
1141 characters are VERY literal, except for - not at the start or end
1142 of the string, which indicates a range. scan_const expands the
1143 range to the full set of intermediate characters.
1145 In double-quoted strings:
1147 double-quoted style: \r and \n
1149 backrefs: \1 (deprecated)
1150 case and quoting: \U \Q \E
1153 scan_const does *not* construct ops to handle interpolated strings.
1154 It stops processing as soon as it finds an embedded $ or @ variable
1155 and leaves it to the caller to work out what's going on.
1157 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1159 $ in pattern could be $foo or could be tail anchor. Assumption:
1160 it's a tail anchor if $ is the last thing in the string, or if it's
1161 followed by one of ")| \n\t"
1163 \1 (backreferences) are turned into $1
1165 The structure of the code is
1166 while (there's a character to process) {
1167 handle transliteration ranges
1168 skip regexp comments
1169 skip # initiated comments in //x patterns
1170 check for embedded @foo
1171 check for embedded scalars
1173 leave intact backslashes from leave (below)
1174 deprecate \1 in strings and sub replacements
1175 handle string-changing backslashes \l \U \Q \E, etc.
1176 switch (what was escaped) {
1177 handle - in a transliteration (becomes a literal -)
1178 handle \132 octal characters
1179 handle 0x15 hex characters
1180 handle \cV (control V)
1181 handle printf backslashes (\f, \r, \n, etc)
1183 } (end if backslash)
1184 } (end while character to read)
1189 S_scan_const(pTHX_ char *start)
1191 register char *send = PL_bufend; /* end of the constant */
1192 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1193 register char *s = start; /* start of the constant */
1194 register char *d = SvPVX(sv); /* destination for copies */
1195 bool dorange = FALSE; /* are we in a translit range? */
1196 bool has_utf = FALSE; /* embedded \x{} */
1200 I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1201 ? (PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1203 I32 thisutf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1204 ? (PL_sublex_info.sub_op->op_private & (PL_lex_repl ?
1205 OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF))
1207 const char *leaveit = /* set of acceptably-backslashed characters */
1209 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1212 while (s < send || dorange) {
1213 /* get transliterations out of the way (they're most literal) */
1214 if (PL_lex_inwhat == OP_TRANS) {
1215 /* expand a range A-Z to the full set of characters. AIE! */
1217 I32 i; /* current expanded character */
1218 I32 min; /* first character in range */
1219 I32 max; /* last character in range */
1221 i = d - SvPVX(sv); /* remember current offset */
1222 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1223 d = SvPVX(sv) + i; /* refresh d after realloc */
1224 d -= 2; /* eat the first char and the - */
1226 min = (U8)*d; /* first char in range */
1227 max = (U8)d[1]; /* last char in range */
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 */
1252 /* range begins (ignore - as first or last char) */
1253 else if (*s == '-' && s+1 < send && s != start) {
1255 *d++ = (char)0xff; /* use illegal utf8 byte--see pmtrans */
1264 /* if we get here, we're not doing a transliteration */
1266 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1267 except for the last char, which will be done separately. */
1268 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1270 while (s < send && *s != ')')
1272 } else if (s[2] == '{' /* This should match regcomp.c */
1273 || (s[2] == 'p' || s[2] == '?') && s[3] == '{') {
1275 char *regparse = s + (s[2] == '{' ? 3 : 4);
1278 while (count && (c = *regparse)) {
1279 if (c == '\\' && regparse[1])
1287 if (*regparse != ')') {
1288 regparse--; /* Leave one char for continuation. */
1289 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1291 while (s < regparse)
1296 /* likewise skip #-initiated comments in //x patterns */
1297 else if (*s == '#' && PL_lex_inpat &&
1298 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1299 while (s+1 < send && *s != '\n')
1303 /* check for embedded arrays (@foo, @:foo, @'foo, @{foo}, @$foo) */
1304 else if (*s == '@' && s[1]
1305 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$", s[1])))
1308 /* check for embedded scalars. only stop if we're sure it's a
1311 else if (*s == '$') {
1312 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1314 if (s + 1 < send && !strchr("()| \n\t", s[1]))
1315 break; /* in regexp, $ might be tail anchor */
1318 /* (now in tr/// code again) */
1320 if (*s & 0x80 && thisutf) {
1321 (void)utf8_to_uv((U8*)s, &len);
1323 /* illegal UTF8, make it valid */
1324 char *old_pvx = SvPVX(sv);
1325 /* need space for one extra char (NOTE: SvCUR() not set here) */
1326 d = SvGROW(sv, SvLEN(sv) + 1) + (d - old_pvx);
1327 d = (char*)uv_to_utf8((U8*)d, (U8)*s++);
1338 if (*s == '\\' && s+1 < send) {
1341 /* some backslashes we leave behind */
1342 if (*leaveit && *s && strchr(leaveit, *s)) {
1348 /* deprecate \1 in strings and substitution replacements */
1349 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1350 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1352 dTHR; /* only for ckWARN */
1353 if (ckWARN(WARN_SYNTAX))
1354 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1359 /* string-change backslash escapes */
1360 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1365 /* if we get here, it's either a quoted -, or a digit */
1368 /* quoted - in transliterations */
1370 if (PL_lex_inwhat == OP_TRANS) {
1378 if (ckWARN(WARN_MISC) && isALPHA(*s))
1379 Perl_warner(aTHX_ WARN_MISC,
1380 "Unrecognized escape \\%c passed through",
1382 /* default action is to copy the quoted character */
1387 /* \132 indicates an octal constant */
1388 case '0': case '1': case '2': case '3':
1389 case '4': case '5': case '6': case '7':
1390 uv = (UV)scan_oct(s, 3, &len);
1392 goto NUM_ESCAPE_INSERT;
1394 /* \x24 indicates a hex constant */
1398 char* e = strchr(s, '}');
1400 yyerror("Missing right brace on \\x{}");
1403 uv = (UV)scan_hex(s + 1, e - s - 1, &len);
1407 uv = (UV)scan_hex(s, 2, &len);
1412 /* Insert oct or hex escaped character.
1413 * There will always enough room in sv since such escapes will
1414 * be longer than any utf8 sequence they can end up as
1417 if (!thisutf && !has_utf && uv > 255) {
1418 /* might need to recode whatever we have accumulated so far
1419 * if it contains any hibit chars
1423 for (c = SvPVX(sv); c < d; c++) {
1428 char *old_pvx = SvPVX(sv);
1430 d = SvGROW(sv, SvCUR(sv) + hicount + 1) + (d - old_pvx);
1439 uv_to_utf8((U8*)dst, (U8)*src--);
1449 if (thisutf || uv > 255) {
1450 d = (char*)uv_to_utf8((U8*)d, uv);
1462 /* \N{latin small letter a} is a named character */
1466 char* e = strchr(s, '}');
1475 yyerror("Missing right brace on \\N{}");
1479 res = newSVpvn(s + 1, e - s - 1);
1480 res = new_constant( Nullch, 0, "charnames",
1481 res, Nullsv, "\\N{...}" );
1482 str = SvPV(res,len);
1483 if (len > e - s + 4) {
1484 char *odest = SvPVX(sv);
1486 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1487 d = SvPVX(sv) + (d - odest);
1489 Copy(str, d, len, char);
1496 yyerror("Missing braces on \\N{}");
1499 /* \c is a control character */
1513 /* printf-style backslashes, formfeeds, newlines, etc */
1531 *d++ = '\047'; /* CP 1047 */
1534 *d++ = '\057'; /* CP 1047 */
1548 } /* end if (backslash) */
1551 } /* while loop to process each character */
1553 /* terminate the string and set up the sv */
1555 SvCUR_set(sv, d - SvPVX(sv));
1560 /* shrink the sv if we allocated more than we used */
1561 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1562 SvLEN_set(sv, SvCUR(sv) + 1);
1563 Renew(SvPVX(sv), SvLEN(sv), char);
1566 /* return the substring (via yylval) only if we parsed anything */
1567 if (s > PL_bufptr) {
1568 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1569 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1571 ( PL_lex_inwhat == OP_TRANS
1573 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1576 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1583 * Returns TRUE if there's more to the expression (e.g., a subscript),
1586 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1588 * ->[ and ->{ return TRUE
1589 * { and [ outside a pattern are always subscripts, so return TRUE
1590 * if we're outside a pattern and it's not { or [, then return FALSE
1591 * if we're in a pattern and the first char is a {
1592 * {4,5} (any digits around the comma) returns FALSE
1593 * if we're in a pattern and the first char is a [
1595 * [SOMETHING] has a funky algorithm to decide whether it's a
1596 * character class or not. It has to deal with things like
1597 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1598 * anything else returns TRUE
1601 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1604 S_intuit_more(pTHX_ register char *s)
1606 if (PL_lex_brackets)
1608 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1610 if (*s != '{' && *s != '[')
1615 /* In a pattern, so maybe we have {n,m}. */
1632 /* On the other hand, maybe we have a character class */
1635 if (*s == ']' || *s == '^')
1638 /* this is terrifying, and it works */
1639 int weight = 2; /* let's weigh the evidence */
1641 unsigned char un_char = 255, last_un_char;
1642 char *send = strchr(s,']');
1643 char tmpbuf[sizeof PL_tokenbuf * 4];
1645 if (!send) /* has to be an expression */
1648 Zero(seen,256,char);
1651 else if (isDIGIT(*s)) {
1653 if (isDIGIT(s[1]) && s[2] == ']')
1659 for (; s < send; s++) {
1660 last_un_char = un_char;
1661 un_char = (unsigned char)*s;
1666 weight -= seen[un_char] * 10;
1667 if (isALNUM_lazy_if(s+1,UTF)) {
1668 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1669 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1674 else if (*s == '$' && s[1] &&
1675 strchr("[#!%*<>()-=",s[1])) {
1676 if (/*{*/ strchr("])} =",s[2]))
1685 if (strchr("wds]",s[1]))
1687 else if (seen['\''] || seen['"'])
1689 else if (strchr("rnftbxcav",s[1]))
1691 else if (isDIGIT(s[1])) {
1693 while (s[1] && isDIGIT(s[1]))
1703 if (strchr("aA01! ",last_un_char))
1705 if (strchr("zZ79~",s[1]))
1707 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1708 weight -= 5; /* cope with negative subscript */
1711 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1712 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1717 if (keyword(tmpbuf, d - tmpbuf))
1720 if (un_char == last_un_char + 1)
1722 weight -= seen[un_char];
1727 if (weight >= 0) /* probably a character class */
1737 * Does all the checking to disambiguate
1739 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1740 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1742 * First argument is the stuff after the first token, e.g. "bar".
1744 * Not a method if bar is a filehandle.
1745 * Not a method if foo is a subroutine prototyped to take a filehandle.
1746 * Not a method if it's really "Foo $bar"
1747 * Method if it's "foo $bar"
1748 * Not a method if it's really "print foo $bar"
1749 * Method if it's really "foo package::" (interpreted as package->foo)
1750 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1751 * Not a method if bar is a filehandle or package, but is quoted with
1756 S_intuit_method(pTHX_ char *start, GV *gv)
1758 char *s = start + (*start == '$');
1759 char tmpbuf[sizeof PL_tokenbuf];
1767 if ((cv = GvCVu(gv))) {
1768 char *proto = SvPVX(cv);
1778 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1779 /* start is the beginning of the possible filehandle/object,
1780 * and s is the end of it
1781 * tmpbuf is a copy of it
1784 if (*start == '$') {
1785 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1790 return *s == '(' ? FUNCMETH : METHOD;
1792 if (!keyword(tmpbuf, len)) {
1793 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1798 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1799 if (indirgv && GvCVu(indirgv))
1801 /* filehandle or package name makes it a method */
1802 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1804 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1805 return 0; /* no assumptions -- "=>" quotes bearword */
1807 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1808 newSVpvn(tmpbuf,len));
1809 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1813 return *s == '(' ? FUNCMETH : METHOD;
1821 * Return a string of Perl code to load the debugger. If PERL5DB
1822 * is set, it will return the contents of that, otherwise a
1823 * compile-time require of perl5db.pl.
1830 char *pdb = PerlEnv_getenv("PERL5DB");
1834 SETERRNO(0,SS$_NORMAL);
1835 return "BEGIN { require 'perl5db.pl' }";
1841 /* Encoded script support. filter_add() effectively inserts a
1842 * 'pre-processing' function into the current source input stream.
1843 * Note that the filter function only applies to the current source file
1844 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1846 * The datasv parameter (which may be NULL) can be used to pass
1847 * private data to this instance of the filter. The filter function
1848 * can recover the SV using the FILTER_DATA macro and use it to
1849 * store private buffers and state information.
1851 * The supplied datasv parameter is upgraded to a PVIO type
1852 * and the IoDIRP field is used to store the function pointer,
1853 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1854 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1855 * private use must be set using malloc'd pointers.
1859 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1864 if (!PL_rsfp_filters)
1865 PL_rsfp_filters = newAV();
1867 datasv = NEWSV(255,0);
1868 if (!SvUPGRADE(datasv, SVt_PVIO))
1869 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1870 IoDIRP(datasv) = (DIR*)funcp; /* stash funcp into spare field */
1871 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1872 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1873 funcp, SvPV_nolen(datasv)));
1874 av_unshift(PL_rsfp_filters, 1);
1875 av_store(PL_rsfp_filters, 0, datasv) ;
1880 /* Delete most recently added instance of this filter function. */
1882 Perl_filter_del(pTHX_ filter_t funcp)
1885 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1886 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1888 /* if filter is on top of stack (usual case) just pop it off */
1889 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1890 if (IoDIRP(datasv) == (DIR*)funcp) {
1891 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1892 IoDIRP(datasv) = (DIR*)NULL;
1893 sv_free(av_pop(PL_rsfp_filters));
1897 /* we need to search for the correct entry and clear it */
1898 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1902 /* Invoke the n'th filter function for the current rsfp. */
1904 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1907 /* 0 = read one text line */
1912 if (!PL_rsfp_filters)
1914 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1915 /* Provide a default input filter to make life easy. */
1916 /* Note that we append to the line. This is handy. */
1917 DEBUG_P(PerlIO_printf(Perl_debug_log,
1918 "filter_read %d: from rsfp\n", idx));
1922 int old_len = SvCUR(buf_sv) ;
1924 /* ensure buf_sv is large enough */
1925 SvGROW(buf_sv, old_len + maxlen) ;
1926 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1927 if (PerlIO_error(PL_rsfp))
1928 return -1; /* error */
1930 return 0 ; /* end of file */
1932 SvCUR_set(buf_sv, old_len + len) ;
1935 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1936 if (PerlIO_error(PL_rsfp))
1937 return -1; /* error */
1939 return 0 ; /* end of file */
1942 return SvCUR(buf_sv);
1944 /* Skip this filter slot if filter has been deleted */
1945 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1946 DEBUG_P(PerlIO_printf(Perl_debug_log,
1947 "filter_read %d: skipped (filter deleted)\n",
1949 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1951 /* Get function pointer hidden within datasv */
1952 funcp = (filter_t)IoDIRP(datasv);
1953 DEBUG_P(PerlIO_printf(Perl_debug_log,
1954 "filter_read %d: via function %p (%s)\n",
1955 idx, funcp, SvPV_nolen(datasv)));
1956 /* Call function. The function is expected to */
1957 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1958 /* Return: <0:error, =0:eof, >0:not eof */
1959 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1963 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1965 #ifdef PERL_CR_FILTER
1966 if (!PL_rsfp_filters) {
1967 filter_add(S_cr_textfilter,NULL);
1970 if (PL_rsfp_filters) {
1973 SvCUR_set(sv, 0); /* start with empty line */
1974 if (FILTER_READ(0, sv, 0) > 0)
1975 return ( SvPVX(sv) ) ;
1980 return (sv_gets(sv, fp, append));
1985 static char* exp_name[] =
1986 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
1987 "ATTRTERM", "TERMBLOCK"
1994 Works out what to call the token just pulled out of the input
1995 stream. The yacc parser takes care of taking the ops we return and
1996 stitching them into a tree.
2002 if read an identifier
2003 if we're in a my declaration
2004 croak if they tried to say my($foo::bar)
2005 build the ops for a my() declaration
2006 if it's an access to a my() variable
2007 are we in a sort block?
2008 croak if my($a); $a <=> $b
2009 build ops for access to a my() variable
2010 if in a dq string, and they've said @foo and we can't find @foo
2012 build ops for a bareword
2013 if we already built the token before, use it.
2017 #ifdef USE_PURE_BISON
2018 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
2031 #ifdef USE_PURE_BISON
2032 yylval_pointer = lvalp;
2033 yychar_pointer = lcharp;
2036 /* check if there's an identifier for us to look at */
2037 if (PL_pending_ident) {
2038 /* pit holds the identifier we read and pending_ident is reset */
2039 char pit = PL_pending_ident;
2040 PL_pending_ident = 0;
2042 /* if we're in a my(), we can't allow dynamics here.
2043 $foo'bar has already been turned into $foo::bar, so
2044 just check for colons.
2046 if it's a legal name, the OP is a PADANY.
2049 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
2050 if (strchr(PL_tokenbuf,':'))
2051 yyerror(Perl_form(aTHX_ "No package name allowed for "
2052 "variable %s in \"our\"",
2054 tmp = pad_allocmy(PL_tokenbuf);
2057 if (strchr(PL_tokenbuf,':'))
2058 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
2060 yylval.opval = newOP(OP_PADANY, 0);
2061 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
2067 build the ops for accesses to a my() variable.
2069 Deny my($a) or my($b) in a sort block, *if* $a or $b is
2070 then used in a comparison. This catches most, but not
2071 all cases. For instance, it catches
2072 sort { my($a); $a <=> $b }
2074 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2075 (although why you'd do that is anyone's guess).
2078 if (!strchr(PL_tokenbuf,':')) {
2080 /* Check for single character per-thread SVs */
2081 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2082 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2083 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2085 yylval.opval = newOP(OP_THREADSV, 0);
2086 yylval.opval->op_targ = tmp;
2089 #endif /* USE_THREADS */
2090 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2091 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2092 /* might be an "our" variable" */
2093 if (SvFLAGS(namesv) & SVpad_OUR) {
2094 /* build ops for a bareword */
2095 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2096 sv_catpvn(sym, "::", 2);
2097 sv_catpv(sym, PL_tokenbuf+1);
2098 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2099 yylval.opval->op_private = OPpCONST_ENTERED;
2100 gv_fetchpv(SvPVX(sym),
2102 ? (GV_ADDMULTI | GV_ADDINEVAL)
2105 ((PL_tokenbuf[0] == '$') ? SVt_PV
2106 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2111 /* if it's a sort block and they're naming $a or $b */
2112 if (PL_last_lop_op == OP_SORT &&
2113 PL_tokenbuf[0] == '$' &&
2114 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2117 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2118 d < PL_bufend && *d != '\n';
2121 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2122 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2128 yylval.opval = newOP(OP_PADANY, 0);
2129 yylval.opval->op_targ = tmp;
2135 Whine if they've said @foo in a doublequoted string,
2136 and @foo isn't a variable we can find in the symbol
2139 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2140 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2141 if (!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2142 yyerror(Perl_form(aTHX_ "In string, %s now must be written as \\%s",
2143 PL_tokenbuf, PL_tokenbuf));
2146 /* build ops for a bareword */
2147 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2148 yylval.opval->op_private = OPpCONST_ENTERED;
2149 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2150 ((PL_tokenbuf[0] == '$') ? SVt_PV
2151 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2156 /* no identifier pending identification */
2158 switch (PL_lex_state) {
2160 case LEX_NORMAL: /* Some compilers will produce faster */
2161 case LEX_INTERPNORMAL: /* code if we comment these out. */
2165 /* when we've already built the next token, just pull it out of the queue */
2168 yylval = PL_nextval[PL_nexttoke];
2170 PL_lex_state = PL_lex_defer;
2171 PL_expect = PL_lex_expect;
2172 PL_lex_defer = LEX_NORMAL;
2174 return(PL_nexttype[PL_nexttoke]);
2176 /* interpolated case modifiers like \L \U, including \Q and \E.
2177 when we get here, PL_bufptr is at the \
2179 case LEX_INTERPCASEMOD:
2181 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2182 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2184 /* handle \E or end of string */
2185 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2189 if (PL_lex_casemods) {
2190 oldmod = PL_lex_casestack[--PL_lex_casemods];
2191 PL_lex_casestack[PL_lex_casemods] = '\0';
2193 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2195 PL_lex_state = LEX_INTERPCONCAT;
2199 if (PL_bufptr != PL_bufend)
2201 PL_lex_state = LEX_INTERPCONCAT;
2206 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2207 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2208 if (strchr("LU", *s) &&
2209 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2211 PL_lex_casestack[--PL_lex_casemods] = '\0';
2214 if (PL_lex_casemods > 10) {
2215 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2216 if (newlb != PL_lex_casestack) {
2218 PL_lex_casestack = newlb;
2221 PL_lex_casestack[PL_lex_casemods++] = *s;
2222 PL_lex_casestack[PL_lex_casemods] = '\0';
2223 PL_lex_state = LEX_INTERPCONCAT;
2224 PL_nextval[PL_nexttoke].ival = 0;
2227 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2229 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2231 PL_nextval[PL_nexttoke].ival = OP_LC;
2233 PL_nextval[PL_nexttoke].ival = OP_UC;
2235 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2237 Perl_croak(aTHX_ "panic: yylex");
2240 if (PL_lex_starts) {
2249 case LEX_INTERPPUSH:
2250 return sublex_push();
2252 case LEX_INTERPSTART:
2253 if (PL_bufptr == PL_bufend)
2254 return sublex_done();
2256 PL_lex_dojoin = (*PL_bufptr == '@');
2257 PL_lex_state = LEX_INTERPNORMAL;
2258 if (PL_lex_dojoin) {
2259 PL_nextval[PL_nexttoke].ival = 0;
2262 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2263 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2264 force_next(PRIVATEREF);
2266 force_ident("\"", '$');
2267 #endif /* USE_THREADS */
2268 PL_nextval[PL_nexttoke].ival = 0;
2270 PL_nextval[PL_nexttoke].ival = 0;
2272 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2275 if (PL_lex_starts++) {
2281 case LEX_INTERPENDMAYBE:
2282 if (intuit_more(PL_bufptr)) {
2283 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2289 if (PL_lex_dojoin) {
2290 PL_lex_dojoin = FALSE;
2291 PL_lex_state = LEX_INTERPCONCAT;
2294 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2295 && SvEVALED(PL_lex_repl))
2297 if (PL_bufptr != PL_bufend)
2298 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2299 PL_lex_repl = Nullsv;
2302 case LEX_INTERPCONCAT:
2304 if (PL_lex_brackets)
2305 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2307 if (PL_bufptr == PL_bufend)
2308 return sublex_done();
2310 if (SvIVX(PL_linestr) == '\'') {
2311 SV *sv = newSVsv(PL_linestr);
2314 else if ( PL_hints & HINT_NEW_RE )
2315 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2316 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2320 s = scan_const(PL_bufptr);
2322 PL_lex_state = LEX_INTERPCASEMOD;
2324 PL_lex_state = LEX_INTERPSTART;
2327 if (s != PL_bufptr) {
2328 PL_nextval[PL_nexttoke] = yylval;
2331 if (PL_lex_starts++)
2341 PL_lex_state = LEX_NORMAL;
2342 s = scan_formline(PL_bufptr);
2343 if (!PL_lex_formbrack)
2349 PL_oldoldbufptr = PL_oldbufptr;
2352 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2353 exp_name[PL_expect], s);
2359 if (isIDFIRST_lazy_if(s,UTF))
2361 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2364 goto fake_eof; /* emulate EOF on ^D or ^Z */
2369 if (PL_lex_brackets)
2370 yyerror("Missing right curly or square bracket");
2373 if (s++ < PL_bufend)
2374 goto retry; /* ignore stray nulls */
2377 if (!PL_in_eval && !PL_preambled) {
2378 PL_preambled = TRUE;
2379 sv_setpv(PL_linestr,incl_perldb());
2380 if (SvCUR(PL_linestr))
2381 sv_catpv(PL_linestr,";");
2383 while(AvFILLp(PL_preambleav) >= 0) {
2384 SV *tmpsv = av_shift(PL_preambleav);
2385 sv_catsv(PL_linestr, tmpsv);
2386 sv_catpv(PL_linestr, ";");
2389 sv_free((SV*)PL_preambleav);
2390 PL_preambleav = NULL;
2392 if (PL_minus_n || PL_minus_p) {
2393 sv_catpv(PL_linestr, "LINE: while (<>) {");
2395 sv_catpv(PL_linestr,"chomp;");
2397 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2399 GvIMPORTED_AV_on(gv);
2401 if (strchr("/'\"", *PL_splitstr)
2402 && strchr(PL_splitstr + 1, *PL_splitstr))
2403 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2406 s = "'~#\200\1'"; /* surely one char is unused...*/
2407 while (s[1] && strchr(PL_splitstr, *s)) s++;
2409 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2410 "q" + (delim == '\''), delim);
2411 for (s = PL_splitstr; *s; s++) {
2413 sv_catpvn(PL_linestr, "\\", 1);
2414 sv_catpvn(PL_linestr, s, 1);
2416 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2420 sv_catpv(PL_linestr,"@F=split(' ');");
2423 sv_catpv(PL_linestr, "\n");
2424 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2425 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2426 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2427 SV *sv = NEWSV(85,0);
2429 sv_upgrade(sv, SVt_PVMG);
2430 sv_setsv(sv,PL_linestr);
2431 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2436 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2439 if (PL_preprocess && !PL_in_eval)
2440 (void)PerlProc_pclose(PL_rsfp);
2441 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2442 PerlIO_clearerr(PL_rsfp);
2444 (void)PerlIO_close(PL_rsfp);
2446 PL_doextract = FALSE;
2448 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2449 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2450 sv_catpv(PL_linestr,";}");
2451 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2452 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2453 PL_minus_n = PL_minus_p = 0;
2456 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2457 sv_setpv(PL_linestr,"");
2458 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2461 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2462 PL_doextract = FALSE;
2464 /* Incest with pod. */
2465 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2466 sv_setpv(PL_linestr, "");
2467 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2468 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2469 PL_doextract = FALSE;
2473 } while (PL_doextract);
2474 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2475 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2476 SV *sv = NEWSV(85,0);
2478 sv_upgrade(sv, SVt_PVMG);
2479 sv_setsv(sv,PL_linestr);
2480 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2482 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2483 if (CopLINE(PL_curcop) == 1) {
2484 while (s < PL_bufend && isSPACE(*s))
2486 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2490 if (*s == '#' && *(s+1) == '!')
2492 #ifdef ALTERNATE_SHEBANG
2494 static char as[] = ALTERNATE_SHEBANG;
2495 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2496 d = s + (sizeof(as) - 1);
2498 #endif /* ALTERNATE_SHEBANG */
2507 while (*d && !isSPACE(*d))
2511 #ifdef ARG_ZERO_IS_SCRIPT
2512 if (ipathend > ipath) {
2514 * HP-UX (at least) sets argv[0] to the script name,
2515 * which makes $^X incorrect. And Digital UNIX and Linux,
2516 * at least, set argv[0] to the basename of the Perl
2517 * interpreter. So, having found "#!", we'll set it right.
2519 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2520 assert(SvPOK(x) || SvGMAGICAL(x));
2521 if (sv_eq(x, CopFILESV(PL_curcop))) {
2522 sv_setpvn(x, ipath, ipathend - ipath);
2525 TAINT_NOT; /* $^X is always tainted, but that's OK */
2527 #endif /* ARG_ZERO_IS_SCRIPT */
2532 d = instr(s,"perl -");
2534 d = instr(s,"perl");
2536 /* avoid getting into infinite loops when shebang
2537 * line contains "Perl" rather than "perl" */
2539 for (d = ipathend-4; d >= ipath; --d) {
2540 if ((*d == 'p' || *d == 'P')
2541 && !ibcmp(d, "perl", 4))
2551 #ifdef ALTERNATE_SHEBANG
2553 * If the ALTERNATE_SHEBANG on this system starts with a
2554 * character that can be part of a Perl expression, then if
2555 * we see it but not "perl", we're probably looking at the
2556 * start of Perl code, not a request to hand off to some
2557 * other interpreter. Similarly, if "perl" is there, but
2558 * not in the first 'word' of the line, we assume the line
2559 * contains the start of the Perl program.
2561 if (d && *s != '#') {
2563 while (*c && !strchr("; \t\r\n\f\v#", *c))
2566 d = Nullch; /* "perl" not in first word; ignore */
2568 *s = '#'; /* Don't try to parse shebang line */
2570 #endif /* ALTERNATE_SHEBANG */
2575 !instr(s,"indir") &&
2576 instr(PL_origargv[0],"perl"))
2582 while (s < PL_bufend && isSPACE(*s))
2584 if (s < PL_bufend) {
2585 Newz(899,newargv,PL_origargc+3,char*);
2587 while (s < PL_bufend && !isSPACE(*s))
2590 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2593 newargv = PL_origargv;
2595 PerlProc_execv(ipath, newargv);
2596 Perl_croak(aTHX_ "Can't exec %s", ipath);
2599 U32 oldpdb = PL_perldb;
2600 bool oldn = PL_minus_n;
2601 bool oldp = PL_minus_p;
2603 while (*d && !isSPACE(*d)) d++;
2604 while (*d == ' ' || *d == '\t') d++;
2608 if (*d == 'M' || *d == 'm') {
2610 while (*d && !isSPACE(*d)) d++;
2611 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2614 d = moreswitches(d);
2616 if (PERLDB_LINE && !oldpdb ||
2617 ( PL_minus_n || PL_minus_p ) && !(oldn || oldp) )
2618 /* if we have already added "LINE: while (<>) {",
2619 we must not do it again */
2621 sv_setpv(PL_linestr, "");
2622 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2623 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2624 PL_preambled = FALSE;
2626 (void)gv_fetchfile(PL_origfilename);
2633 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2635 PL_lex_state = LEX_FORMLINE;
2640 #ifdef PERL_STRICT_CR
2641 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2643 "(Maybe you didn't strip carriage returns after a network transfer?)\n");
2645 case ' ': case '\t': case '\f': case 013:
2650 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2652 while (s < d && *s != '\n')
2657 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2659 PL_lex_state = LEX_FORMLINE;
2669 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2674 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2677 if (strnEQ(s,"=>",2)) {
2678 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2679 OPERATOR('-'); /* unary minus */
2681 PL_last_uni = PL_oldbufptr;
2682 PL_last_lop_op = OP_FTEREAD; /* good enough */
2684 case 'r': FTST(OP_FTEREAD);
2685 case 'w': FTST(OP_FTEWRITE);
2686 case 'x': FTST(OP_FTEEXEC);
2687 case 'o': FTST(OP_FTEOWNED);
2688 case 'R': FTST(OP_FTRREAD);
2689 case 'W': FTST(OP_FTRWRITE);
2690 case 'X': FTST(OP_FTREXEC);
2691 case 'O': FTST(OP_FTROWNED);
2692 case 'e': FTST(OP_FTIS);
2693 case 'z': FTST(OP_FTZERO);
2694 case 's': FTST(OP_FTSIZE);
2695 case 'f': FTST(OP_FTFILE);
2696 case 'd': FTST(OP_FTDIR);
2697 case 'l': FTST(OP_FTLINK);
2698 case 'p': FTST(OP_FTPIPE);
2699 case 'S': FTST(OP_FTSOCK);
2700 case 'u': FTST(OP_FTSUID);
2701 case 'g': FTST(OP_FTSGID);
2702 case 'k': FTST(OP_FTSVTX);
2703 case 'b': FTST(OP_FTBLK);
2704 case 'c': FTST(OP_FTCHR);
2705 case 't': FTST(OP_FTTTY);
2706 case 'T': FTST(OP_FTTEXT);
2707 case 'B': FTST(OP_FTBINARY);
2708 case 'M': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTMTIME);
2709 case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME);
2710 case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME);
2712 Perl_croak(aTHX_ "Unrecognized file test: -%c", (int)tmp);
2719 if (PL_expect == XOPERATOR)
2724 else if (*s == '>') {
2727 if (isIDFIRST_lazy_if(s,UTF)) {
2728 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2736 if (PL_expect == XOPERATOR)
2739 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2741 OPERATOR('-'); /* unary minus */
2748 if (PL_expect == XOPERATOR)
2753 if (PL_expect == XOPERATOR)
2756 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2762 if (PL_expect != XOPERATOR) {
2763 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2764 PL_expect = XOPERATOR;
2765 force_ident(PL_tokenbuf, '*');
2778 if (PL_expect == XOPERATOR) {
2782 PL_tokenbuf[0] = '%';
2783 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2784 if (!PL_tokenbuf[1]) {
2786 yyerror("Final % should be \\% or %name");
2789 PL_pending_ident = '%';
2808 switch (PL_expect) {
2811 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2813 PL_bufptr = s; /* update in case we back off */
2819 PL_expect = XTERMBLOCK;
2823 while (isIDFIRST_lazy_if(s,UTF)) {
2824 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2825 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2826 if (tmp < 0) tmp = -tmp;
2841 d = scan_str(d,TRUE,TRUE);
2844 SvREFCNT_dec(PL_lex_stuff);
2845 PL_lex_stuff = Nullsv;
2847 /* MUST advance bufptr here to avoid bogus
2848 "at end of line" context messages from yyerror().
2850 PL_bufptr = s + len;
2851 yyerror("Unterminated attribute parameter in attribute list");
2854 return 0; /* EOF indicator */
2858 SV *sv = newSVpvn(s, len);
2859 sv_catsv(sv, PL_lex_stuff);
2860 attrs = append_elem(OP_LIST, attrs,
2861 newSVOP(OP_CONST, 0, sv));
2862 SvREFCNT_dec(PL_lex_stuff);
2863 PL_lex_stuff = Nullsv;
2866 attrs = append_elem(OP_LIST, attrs,
2867 newSVOP(OP_CONST, 0,
2871 if (*s == ':' && s[1] != ':')
2874 break; /* require real whitespace or :'s */
2876 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
2877 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
2878 char q = ((*s == '\'') ? '"' : '\'');
2879 /* If here for an expression, and parsed no attrs, back off. */
2880 if (tmp == '=' && !attrs) {
2884 /* MUST advance bufptr here to avoid bogus "at end of line"
2885 context messages from yyerror().
2889 yyerror("Unterminated attribute list");
2891 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
2899 PL_nextval[PL_nexttoke].opval = attrs;
2907 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
2908 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
2913 if (CopLINE(PL_curcop) < PL_copline)
2914 PL_copline = CopLINE(PL_curcop);
2925 if (PL_lex_brackets <= 0)
2926 yyerror("Unmatched right square bracket");
2929 if (PL_lex_state == LEX_INTERPNORMAL) {
2930 if (PL_lex_brackets == 0) {
2931 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
2932 PL_lex_state = LEX_INTERPEND;
2939 if (PL_lex_brackets > 100) {
2940 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
2941 if (newlb != PL_lex_brackstack) {
2943 PL_lex_brackstack = newlb;
2946 switch (PL_expect) {
2948 if (PL_lex_formbrack) {
2952 if (PL_oldoldbufptr == PL_last_lop)
2953 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2955 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2956 OPERATOR(HASHBRACK);
2958 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2961 PL_tokenbuf[0] = '\0';
2962 if (d < PL_bufend && *d == '-') {
2963 PL_tokenbuf[0] = '-';
2965 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2968 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
2969 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
2971 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2974 char minus = (PL_tokenbuf[0] == '-');
2975 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
2983 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
2988 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2993 if (PL_oldoldbufptr == PL_last_lop)
2994 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2996 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2999 OPERATOR(HASHBRACK);
3000 /* This hack serves to disambiguate a pair of curlies
3001 * as being a block or an anon hash. Normally, expectation
3002 * determines that, but in cases where we're not in a
3003 * position to expect anything in particular (like inside
3004 * eval"") we have to resolve the ambiguity. This code
3005 * covers the case where the first term in the curlies is a
3006 * quoted string. Most other cases need to be explicitly
3007 * disambiguated by prepending a `+' before the opening
3008 * curly in order to force resolution as an anon hash.
3010 * XXX should probably propagate the outer expectation
3011 * into eval"" to rely less on this hack, but that could
3012 * potentially break current behavior of eval"".
3016 if (*s == '\'' || *s == '"' || *s == '`') {
3017 /* common case: get past first string, handling escapes */
3018 for (t++; t < PL_bufend && *t != *s;)
3019 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3023 else if (*s == 'q') {
3026 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3030 char open, close, term;
3033 while (t < PL_bufend && isSPACE(*t))
3037 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3041 for (t++; t < PL_bufend; t++) {
3042 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3044 else if (*t == open)
3048 for (t++; t < PL_bufend; t++) {
3049 if (*t == '\\' && t+1 < PL_bufend)
3051 else if (*t == close && --brackets <= 0)
3053 else if (*t == open)
3059 else if (isALNUM_lazy_if(t,UTF)) {
3061 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3064 while (t < PL_bufend && isSPACE(*t))
3066 /* if comma follows first term, call it an anon hash */
3067 /* XXX it could be a comma expression with loop modifiers */
3068 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3069 || (*t == '=' && t[1] == '>')))
3070 OPERATOR(HASHBRACK);
3071 if (PL_expect == XREF)
3074 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3080 yylval.ival = CopLINE(PL_curcop);
3081 if (isSPACE(*s) || *s == '#')
3082 PL_copline = NOLINE; /* invalidate current command line number */
3087 if (PL_lex_brackets <= 0)
3088 yyerror("Unmatched right curly bracket");
3090 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3091 if (PL_lex_brackets < PL_lex_formbrack)
3092 PL_lex_formbrack = 0;
3093 if (PL_lex_state == LEX_INTERPNORMAL) {
3094 if (PL_lex_brackets == 0) {
3095 if (PL_expect & XFAKEBRACK) {
3096 PL_expect &= XENUMMASK;
3097 PL_lex_state = LEX_INTERPEND;
3099 return yylex(); /* ignore fake brackets */
3101 if (*s == '-' && s[1] == '>')
3102 PL_lex_state = LEX_INTERPENDMAYBE;
3103 else if (*s != '[' && *s != '{')
3104 PL_lex_state = LEX_INTERPEND;
3107 if (PL_expect & XFAKEBRACK) {
3108 PL_expect &= XENUMMASK;
3110 return yylex(); /* ignore fake brackets */
3120 if (PL_expect == XOPERATOR) {
3121 if (ckWARN(WARN_SEMICOLON)
3122 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3124 CopLINE_dec(PL_curcop);
3125 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3126 CopLINE_inc(PL_curcop);
3131 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3133 PL_expect = XOPERATOR;
3134 force_ident(PL_tokenbuf, '&');
3138 yylval.ival = (OPpENTERSUB_AMPER<<8);
3157 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3158 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3160 if (PL_expect == XSTATE && isALPHA(tmp) &&
3161 (s == PL_linestart+1 || s[-2] == '\n') )
3163 if (PL_in_eval && !PL_rsfp) {
3168 if (strnEQ(s,"=cut",4)) {
3182 PL_doextract = TRUE;
3185 if (PL_lex_brackets < PL_lex_formbrack) {
3187 #ifdef PERL_STRICT_CR
3188 for (t = s; *t == ' ' || *t == '\t'; t++) ;
3190 for (t = s; *t == ' ' || *t == '\t' || *t == '\r'; t++) ;
3192 if (*t == '\n' || *t == '#') {
3210 if (PL_expect != XOPERATOR) {
3211 if (s[1] != '<' && !strchr(s,'>'))
3214 s = scan_heredoc(s);
3216 s = scan_inputsymbol(s);
3217 TERM(sublex_start());
3222 SHop(OP_LEFT_SHIFT);
3236 SHop(OP_RIGHT_SHIFT);
3245 if (PL_expect == XOPERATOR) {
3246 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3249 return ','; /* grandfather non-comma-format format */
3253 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3254 PL_tokenbuf[0] = '@';
3255 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3256 sizeof PL_tokenbuf - 1, FALSE);
3257 if (PL_expect == XOPERATOR)
3258 no_op("Array length", s);
3259 if (!PL_tokenbuf[1])
3261 PL_expect = XOPERATOR;
3262 PL_pending_ident = '#';
3266 PL_tokenbuf[0] = '$';
3267 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3268 sizeof PL_tokenbuf - 1, FALSE);
3269 if (PL_expect == XOPERATOR)
3271 if (!PL_tokenbuf[1]) {
3273 yyerror("Final $ should be \\$ or $name");
3277 /* This kludge not intended to be bulletproof. */
3278 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3279 yylval.opval = newSVOP(OP_CONST, 0,
3280 newSViv((IV)PL_compiling.cop_arybase));
3281 yylval.opval->op_private = OPpCONST_ARYBASE;
3287 if (PL_lex_state == LEX_NORMAL)
3290 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3293 PL_tokenbuf[0] = '@';
3294 if (ckWARN(WARN_SYNTAX)) {
3296 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3299 PL_bufptr = skipspace(PL_bufptr);
3300 while (t < PL_bufend && *t != ']')
3302 Perl_warner(aTHX_ WARN_SYNTAX,
3303 "Multidimensional syntax %.*s not supported",
3304 (t - PL_bufptr) + 1, PL_bufptr);
3308 else if (*s == '{') {
3309 PL_tokenbuf[0] = '%';
3310 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3311 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3313 char tmpbuf[sizeof PL_tokenbuf];
3315 for (t++; isSPACE(*t); t++) ;
3316 if (isIDFIRST_lazy_if(t,UTF)) {
3317 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3318 for (; isSPACE(*t); t++) ;
3319 if (*t == ';' && get_cv(tmpbuf, FALSE))
3320 Perl_warner(aTHX_ WARN_SYNTAX,
3321 "You need to quote \"%s\"", tmpbuf);
3327 PL_expect = XOPERATOR;
3328 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3329 bool islop = (PL_last_lop == PL_oldoldbufptr);
3330 if (!islop || PL_last_lop_op == OP_GREPSTART)
3331 PL_expect = XOPERATOR;
3332 else if (strchr("$@\"'`q", *s))
3333 PL_expect = XTERM; /* e.g. print $fh "foo" */
3334 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3335 PL_expect = XTERM; /* e.g. print $fh &sub */
3336 else if (isIDFIRST_lazy_if(s,UTF)) {
3337 char tmpbuf[sizeof PL_tokenbuf];
3338 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3339 if (tmp = keyword(tmpbuf, len)) {
3340 /* binary operators exclude handle interpretations */
3352 PL_expect = XTERM; /* e.g. print $fh length() */
3357 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3358 if (gv && GvCVu(gv))
3359 PL_expect = XTERM; /* e.g. print $fh subr() */
3362 else if (isDIGIT(*s))
3363 PL_expect = XTERM; /* e.g. print $fh 3 */
3364 else if (*s == '.' && isDIGIT(s[1]))
3365 PL_expect = XTERM; /* e.g. print $fh .3 */
3366 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3367 PL_expect = XTERM; /* e.g. print $fh -1 */
3368 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3369 PL_expect = XTERM; /* print $fh <<"EOF" */
3371 PL_pending_ident = '$';
3375 if (PL_expect == XOPERATOR)
3377 PL_tokenbuf[0] = '@';
3378 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3379 if (!PL_tokenbuf[1]) {
3381 yyerror("Final @ should be \\@ or @name");
3384 if (PL_lex_state == LEX_NORMAL)
3386 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3388 PL_tokenbuf[0] = '%';
3390 /* Warn about @ where they meant $. */
3391 if (ckWARN(WARN_SYNTAX)) {
3392 if (*s == '[' || *s == '{') {
3394 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3396 if (*t == '}' || *t == ']') {
3398 PL_bufptr = skipspace(PL_bufptr);
3399 Perl_warner(aTHX_ WARN_SYNTAX,
3400 "Scalar value %.*s better written as $%.*s",
3401 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3406 PL_pending_ident = '@';
3409 case '/': /* may either be division or pattern */
3410 case '?': /* may either be conditional or pattern */
3411 if (PL_expect != XOPERATOR) {
3412 /* Disable warning on "study /blah/" */
3413 if (PL_oldoldbufptr == PL_last_uni
3414 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3415 || memNE(PL_last_uni, "study", 5)
3416 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3418 s = scan_pat(s,OP_MATCH);
3419 TERM(sublex_start());
3427 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3428 #ifdef PERL_STRICT_CR
3431 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3433 && (s == PL_linestart || s[-1] == '\n') )
3435 PL_lex_formbrack = 0;
3439 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3445 yylval.ival = OPf_SPECIAL;
3451 if (PL_expect != XOPERATOR)
3456 case '0': case '1': case '2': case '3': case '4':
3457 case '5': case '6': case '7': case '8': case '9':
3459 if (PL_expect == XOPERATOR)
3464 s = scan_str(s,FALSE,FALSE);
3465 if (PL_expect == XOPERATOR) {
3466 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3469 return ','; /* grandfather non-comma-format format */
3475 missingterm((char*)0);
3476 yylval.ival = OP_CONST;
3477 TERM(sublex_start());
3480 s = scan_str(s,FALSE,FALSE);
3481 if (PL_expect == XOPERATOR) {
3482 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3485 return ','; /* grandfather non-comma-format format */
3491 missingterm((char*)0);
3492 yylval.ival = OP_CONST;
3493 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3494 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3495 yylval.ival = OP_STRINGIFY;
3499 TERM(sublex_start());
3502 s = scan_str(s,FALSE,FALSE);
3503 if (PL_expect == XOPERATOR)
3504 no_op("Backticks",s);
3506 missingterm((char*)0);
3507 yylval.ival = OP_BACKTICK;
3509 TERM(sublex_start());
3513 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3514 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3516 if (PL_expect == XOPERATOR)
3517 no_op("Backslash",s);
3521 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3525 while (isDIGIT(*start) || *start == '_')
3527 if (*start == '.' && isDIGIT(start[1])) {
3531 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3532 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF)) {
3536 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3546 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3586 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3588 /* Some keywords can be followed by any delimiter, including ':' */
3589 tmp = (len == 1 && strchr("msyq", PL_tokenbuf[0]) ||
3590 len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3591 (PL_tokenbuf[0] == 'q' &&
3592 strchr("qwxr", PL_tokenbuf[1]))));
3594 /* x::* is just a word, unless x is "CORE" */
3595 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3599 while (d < PL_bufend && isSPACE(*d))
3600 d++; /* no comments skipped here, or s### is misparsed */
3602 /* Is this a label? */
3603 if (!tmp && PL_expect == XSTATE
3604 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3606 yylval.pval = savepv(PL_tokenbuf);
3611 /* Check for keywords */
3612 tmp = keyword(PL_tokenbuf, len);
3614 /* Is this a word before a => operator? */
3615 if (strnEQ(d,"=>",2)) {
3617 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3618 yylval.opval->op_private = OPpCONST_BARE;
3622 if (tmp < 0) { /* second-class keyword? */
3623 GV *ogv = Nullgv; /* override (winner) */
3624 GV *hgv = Nullgv; /* hidden (loser) */
3625 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3627 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3630 if (GvIMPORTED_CV(gv))
3632 else if (! CvMETHOD(cv))
3636 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3637 (gv = *gvp) != (GV*)&PL_sv_undef &&
3638 GvCVu(gv) && GvIMPORTED_CV(gv))
3644 tmp = 0; /* overridden by import or by GLOBAL */
3647 && -tmp==KEY_lock /* XXX generalizable kludge */
3649 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3651 tmp = 0; /* any sub overrides "weak" keyword */
3653 else { /* no override */
3657 if (ckWARN(WARN_AMBIGUOUS) && hgv
3658 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3659 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3660 "Ambiguous call resolved as CORE::%s(), %s",
3661 GvENAME(hgv), "qualify as such or use &");
3668 default: /* not a keyword */
3671 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3673 /* Get the rest if it looks like a package qualifier */
3675 if (*s == '\'' || *s == ':' && s[1] == ':') {
3677 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3680 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3681 *s == '\'' ? "'" : "::");
3685 if (PL_expect == XOPERATOR) {
3686 if (PL_bufptr == PL_linestart) {
3687 CopLINE_dec(PL_curcop);
3688 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3689 CopLINE_inc(PL_curcop);
3692 no_op("Bareword",s);
3695 /* Look for a subroutine with this name in current package,
3696 unless name is "Foo::", in which case Foo is a bearword
3697 (and a package name). */
3700 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3702 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3703 Perl_warner(aTHX_ WARN_BAREWORD,
3704 "Bareword \"%s\" refers to nonexistent package",
3707 PL_tokenbuf[len] = '\0';
3714 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3717 /* if we saw a global override before, get the right name */
3720 sv = newSVpvn("CORE::GLOBAL::",14);
3721 sv_catpv(sv,PL_tokenbuf);
3724 sv = newSVpv(PL_tokenbuf,0);
3726 /* Presume this is going to be a bareword of some sort. */
3729 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3730 yylval.opval->op_private = OPpCONST_BARE;
3732 /* And if "Foo::", then that's what it certainly is. */
3737 /* See if it's the indirect object for a list operator. */
3739 if (PL_oldoldbufptr &&
3740 PL_oldoldbufptr < PL_bufptr &&
3741 (PL_oldoldbufptr == PL_last_lop
3742 || PL_oldoldbufptr == PL_last_uni) &&
3743 /* NO SKIPSPACE BEFORE HERE! */
3744 (PL_expect == XREF ||
3745 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3747 bool immediate_paren = *s == '(';
3749 /* (Now we can afford to cross potential line boundary.) */
3752 /* Two barewords in a row may indicate method call. */
3754 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3757 /* If not a declared subroutine, it's an indirect object. */
3758 /* (But it's an indir obj regardless for sort.) */
3760 if ((PL_last_lop_op == OP_SORT ||
3761 (!immediate_paren && (!gv || !GvCVu(gv)))) &&
3762 (PL_last_lop_op != OP_MAPSTART &&
3763 PL_last_lop_op != OP_GREPSTART))
3765 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3770 /* If followed by a paren, it's certainly a subroutine. */
3772 PL_expect = XOPERATOR;
3776 if (gv && GvCVu(gv)) {
3777 for (d = s + 1; *d == ' ' || *d == '\t'; d++) ;
3778 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3783 PL_nextval[PL_nexttoke].opval = yylval.opval;
3784 PL_expect = XOPERATOR;
3790 /* If followed by var or block, call it a method (unless sub) */
3792 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3793 PL_last_lop = PL_oldbufptr;
3794 PL_last_lop_op = OP_METHOD;
3798 /* If followed by a bareword, see if it looks like indir obj. */
3800 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3803 /* Not a method, so call it a subroutine (if defined) */
3805 if (gv && GvCVu(gv)) {
3807 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3808 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3809 "Ambiguous use of -%s resolved as -&%s()",
3810 PL_tokenbuf, PL_tokenbuf);
3811 /* Check for a constant sub */
3813 if ((sv = cv_const_sv(cv))) {
3815 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3816 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3817 yylval.opval->op_private = 0;
3821 /* Resolve to GV now. */
3822 op_free(yylval.opval);
3823 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3824 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3825 PL_last_lop = PL_oldbufptr;
3826 PL_last_lop_op = OP_ENTERSUB;
3827 /* Is there a prototype? */
3830 char *proto = SvPV((SV*)cv, len);
3833 if (strEQ(proto, "$"))
3835 if (*proto == '&' && *s == '{') {
3836 sv_setpv(PL_subname,"__ANON__");
3840 PL_nextval[PL_nexttoke].opval = yylval.opval;
3846 /* Call it a bare word */
3848 if (PL_hints & HINT_STRICT_SUBS)
3849 yylval.opval->op_private |= OPpCONST_STRICT;
3852 if (ckWARN(WARN_RESERVED)) {
3853 if (lastchar != '-') {
3854 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
3856 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
3863 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
3864 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3865 "Operator or semicolon missing before %c%s",
3866 lastchar, PL_tokenbuf);
3867 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3868 "Ambiguous use of %c resolved as operator %c",
3869 lastchar, lastchar);
3875 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3876 newSVpv(CopFILE(PL_curcop),0));
3880 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3881 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
3884 case KEY___PACKAGE__:
3885 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3887 ? newSVsv(PL_curstname)
3896 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
3897 char *pname = "main";
3898 if (PL_tokenbuf[2] == 'D')
3899 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
3900 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
3903 GvIOp(gv) = newIO();
3904 IoIFP(GvIOp(gv)) = PL_rsfp;
3905 #if defined(HAS_FCNTL) && defined(F_SETFD)
3907 int fd = PerlIO_fileno(PL_rsfp);
3908 fcntl(fd,F_SETFD,fd >= 3);
3911 /* Mark this internal pseudo-handle as clean */
3912 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
3914 IoTYPE(GvIOp(gv)) = '|';
3915 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
3916 IoTYPE(GvIOp(gv)) = '-';
3918 IoTYPE(GvIOp(gv)) = '<';
3919 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
3920 /* if the script was opened in binmode, we need to revert
3921 * it to text mode for compatibility; but only iff it has CRs
3922 * XXX this is a questionable hack at best. */
3923 if (PL_bufend-PL_bufptr > 2
3924 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
3927 if (IoTYPE(GvIOp(gv)) == '<') {
3928 loc = PerlIO_tell(PL_rsfp);
3929 (void)PerlIO_seek(PL_rsfp, 0L, 0);
3931 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
3932 #if defined(__BORLANDC__)
3933 /* XXX see note in do_binmode() */
3934 ((FILE*)PL_rsfp)->flags |= _F_BIN;
3937 PerlIO_seek(PL_rsfp, loc, 0);
3952 if (PL_expect == XSTATE) {
3959 if (*s == ':' && s[1] == ':') {
3962 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3963 tmp = keyword(PL_tokenbuf, len);
3977 LOP(OP_ACCEPT,XTERM);
3983 LOP(OP_ATAN2,XTERM);
3992 LOP(OP_BLESS,XTERM);
4001 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4018 if (!PL_cryptseen) {
4019 PL_cryptseen = TRUE;
4023 LOP(OP_CRYPT,XTERM);
4026 if (ckWARN(WARN_CHMOD)) {
4027 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4028 if (*d != '0' && isDIGIT(*d))
4029 Perl_warner(aTHX_ WARN_CHMOD,
4030 "chmod() mode argument is missing initial 0");
4032 LOP(OP_CHMOD,XTERM);
4035 LOP(OP_CHOWN,XTERM);
4038 LOP(OP_CONNECT,XTERM);
4054 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4058 PL_hints |= HINT_BLOCK_SCOPE;
4068 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4069 LOP(OP_DBMOPEN,XTERM);
4075 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4082 yylval.ival = CopLINE(PL_curcop);
4096 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4097 UNIBRACK(OP_ENTEREVAL);
4112 case KEY_endhostent:
4118 case KEY_endservent:
4121 case KEY_endprotoent:
4132 yylval.ival = CopLINE(PL_curcop);
4134 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4136 if ((PL_bufend - p) >= 3 &&
4137 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4139 else if ((PL_bufend - p) >= 4 &&
4140 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4143 if (isIDFIRST_lazy_if(p,UTF)) {
4144 p = scan_ident(p, PL_bufend,
4145 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4149 Perl_croak(aTHX_ "Missing $ on loop variable");
4154 LOP(OP_FORMLINE,XTERM);
4160 LOP(OP_FCNTL,XTERM);
4166 LOP(OP_FLOCK,XTERM);
4175 LOP(OP_GREPSTART, XREF);
4178 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4193 case KEY_getpriority:
4194 LOP(OP_GETPRIORITY,XTERM);
4196 case KEY_getprotobyname:
4199 case KEY_getprotobynumber:
4200 LOP(OP_GPBYNUMBER,XTERM);
4202 case KEY_getprotoent:
4214 case KEY_getpeername:
4215 UNI(OP_GETPEERNAME);
4217 case KEY_gethostbyname:
4220 case KEY_gethostbyaddr:
4221 LOP(OP_GHBYADDR,XTERM);
4223 case KEY_gethostent:
4226 case KEY_getnetbyname:
4229 case KEY_getnetbyaddr:
4230 LOP(OP_GNBYADDR,XTERM);
4235 case KEY_getservbyname:
4236 LOP(OP_GSBYNAME,XTERM);
4238 case KEY_getservbyport:
4239 LOP(OP_GSBYPORT,XTERM);
4241 case KEY_getservent:
4244 case KEY_getsockname:
4245 UNI(OP_GETSOCKNAME);
4247 case KEY_getsockopt:
4248 LOP(OP_GSOCKOPT,XTERM);
4270 yylval.ival = CopLINE(PL_curcop);
4274 LOP(OP_INDEX,XTERM);
4280 LOP(OP_IOCTL,XTERM);
4292 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4324 LOP(OP_LISTEN,XTERM);
4333 s = scan_pat(s,OP_MATCH);
4334 TERM(sublex_start());
4337 LOP(OP_MAPSTART, XREF);
4340 LOP(OP_MKDIR,XTERM);
4343 LOP(OP_MSGCTL,XTERM);
4346 LOP(OP_MSGGET,XTERM);
4349 LOP(OP_MSGRCV,XTERM);
4352 LOP(OP_MSGSND,XTERM);
4358 if (isIDFIRST_lazy_if(s,UTF)) {
4359 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4360 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4362 PL_in_my_stash = gv_stashpv(PL_tokenbuf, FALSE);
4363 if (!PL_in_my_stash) {
4366 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4374 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4381 if (PL_expect != XSTATE)
4382 yyerror("\"no\" not allowed in expression");
4383 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4384 s = force_version(s);
4389 if (*s == '(' || (s = skipspace(s), *s == '('))
4396 if (isIDFIRST_lazy_if(s,UTF)) {
4398 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4400 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4401 Perl_warner(aTHX_ WARN_PRECEDENCE,
4402 "Precedence problem: open %.*s should be open(%.*s)",
4408 yylval.ival = OP_OR;
4418 LOP(OP_OPEN_DIR,XTERM);
4421 checkcomma(s,PL_tokenbuf,"filehandle");
4425 checkcomma(s,PL_tokenbuf,"filehandle");
4444 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4448 LOP(OP_PIPE_OP,XTERM);
4451 s = scan_str(s,FALSE,FALSE);
4453 missingterm((char*)0);
4454 yylval.ival = OP_CONST;
4455 TERM(sublex_start());
4461 s = scan_str(s,FALSE,FALSE);
4463 missingterm((char*)0);
4465 if (SvCUR(PL_lex_stuff)) {
4468 d = SvPV_force(PL_lex_stuff, len);
4470 for (; isSPACE(*d) && len; --len, ++d) ;
4473 if (!warned && ckWARN(WARN_QW)) {
4474 for (; !isSPACE(*d) && len; --len, ++d) {
4476 Perl_warner(aTHX_ WARN_QW,
4477 "Possible attempt to separate words with commas");
4480 else if (*d == '#') {
4481 Perl_warner(aTHX_ WARN_QW,
4482 "Possible attempt to put comments in qw() list");
4488 for (; !isSPACE(*d) && len; --len, ++d) ;
4490 words = append_elem(OP_LIST, words,
4491 newSVOP(OP_CONST, 0, newSVpvn(b, d-b)));
4495 PL_nextval[PL_nexttoke].opval = words;
4500 SvREFCNT_dec(PL_lex_stuff);
4501 PL_lex_stuff = Nullsv;
4506 s = scan_str(s,FALSE,FALSE);
4508 missingterm((char*)0);
4509 yylval.ival = OP_STRINGIFY;
4510 if (SvIVX(PL_lex_stuff) == '\'')
4511 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4512 TERM(sublex_start());
4515 s = scan_pat(s,OP_QR);
4516 TERM(sublex_start());
4519 s = scan_str(s,FALSE,FALSE);
4521 missingterm((char*)0);
4522 yylval.ival = OP_BACKTICK;
4524 TERM(sublex_start());
4531 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4532 s = force_version(s);
4535 *PL_tokenbuf = '\0';
4536 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4537 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4538 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4540 yyerror("<> should be quotes");
4548 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4552 LOP(OP_RENAME,XTERM);
4561 LOP(OP_RINDEX,XTERM);
4584 LOP(OP_REVERSE,XTERM);
4595 TERM(sublex_start());
4597 TOKEN(1); /* force error */
4606 LOP(OP_SELECT,XTERM);
4612 LOP(OP_SEMCTL,XTERM);
4615 LOP(OP_SEMGET,XTERM);
4618 LOP(OP_SEMOP,XTERM);
4624 LOP(OP_SETPGRP,XTERM);
4626 case KEY_setpriority:
4627 LOP(OP_SETPRIORITY,XTERM);
4629 case KEY_sethostent:
4635 case KEY_setservent:
4638 case KEY_setprotoent:
4648 LOP(OP_SEEKDIR,XTERM);
4650 case KEY_setsockopt:
4651 LOP(OP_SSOCKOPT,XTERM);
4657 LOP(OP_SHMCTL,XTERM);
4660 LOP(OP_SHMGET,XTERM);
4663 LOP(OP_SHMREAD,XTERM);
4666 LOP(OP_SHMWRITE,XTERM);
4669 LOP(OP_SHUTDOWN,XTERM);
4678 LOP(OP_SOCKET,XTERM);
4680 case KEY_socketpair:
4681 LOP(OP_SOCKPAIR,XTERM);
4684 checkcomma(s,PL_tokenbuf,"subroutine name");
4686 if (*s == ';' || *s == ')') /* probably a close */
4687 Perl_croak(aTHX_ "sort is now a reserved word");
4689 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4693 LOP(OP_SPLIT,XTERM);
4696 LOP(OP_SPRINTF,XTERM);
4699 LOP(OP_SPLICE,XTERM);
4714 LOP(OP_SUBSTR,XTERM);
4720 char tmpbuf[sizeof PL_tokenbuf];
4722 expectation attrful;
4723 bool have_name, have_proto;
4728 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4729 (*s == ':' && s[1] == ':'))
4732 attrful = XATTRBLOCK;
4733 /* remember buffer pos'n for later force_word */
4734 tboffset = s - PL_oldbufptr;
4735 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4736 if (strchr(tmpbuf, ':'))
4737 sv_setpv(PL_subname, tmpbuf);
4739 sv_setsv(PL_subname,PL_curstname);
4740 sv_catpvn(PL_subname,"::",2);
4741 sv_catpvn(PL_subname,tmpbuf,len);
4748 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4749 PL_expect = XTERMBLOCK;
4750 attrful = XATTRTERM;
4751 sv_setpv(PL_subname,"?");
4755 if (key == KEY_format) {
4757 PL_lex_formbrack = PL_lex_brackets + 1;
4759 (void) force_word(PL_oldbufptr + tboffset, WORD,
4764 /* Look for a prototype */
4768 s = scan_str(s,FALSE,FALSE);
4771 SvREFCNT_dec(PL_lex_stuff);
4772 PL_lex_stuff = Nullsv;
4773 Perl_croak(aTHX_ "Prototype not terminated");
4776 d = SvPVX(PL_lex_stuff);
4778 for (p = d; *p; ++p) {
4783 SvCUR(PL_lex_stuff) = tmp;
4791 if (*s == ':' && s[1] != ':')
4792 PL_expect = attrful;
4795 PL_nextval[PL_nexttoke].opval =
4796 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4797 PL_lex_stuff = Nullsv;
4801 sv_setpv(PL_subname,"__ANON__");
4804 (void) force_word(PL_oldbufptr + tboffset, WORD,
4813 LOP(OP_SYSTEM,XREF);
4816 LOP(OP_SYMLINK,XTERM);
4819 LOP(OP_SYSCALL,XTERM);
4822 LOP(OP_SYSOPEN,XTERM);
4825 LOP(OP_SYSSEEK,XTERM);
4828 LOP(OP_SYSREAD,XTERM);
4831 LOP(OP_SYSWRITE,XTERM);
4835 TERM(sublex_start());
4856 LOP(OP_TRUNCATE,XTERM);
4868 yylval.ival = CopLINE(PL_curcop);
4872 yylval.ival = CopLINE(PL_curcop);
4876 LOP(OP_UNLINK,XTERM);
4882 LOP(OP_UNPACK,XTERM);
4885 LOP(OP_UTIME,XTERM);
4888 if (ckWARN(WARN_UMASK)) {
4889 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4890 if (*d != '0' && isDIGIT(*d))
4891 Perl_warner(aTHX_ WARN_UMASK,
4892 "umask: argument is missing initial 0");
4897 LOP(OP_UNSHIFT,XTERM);
4900 if (PL_expect != XSTATE)
4901 yyerror("\"use\" not allowed in expression");
4903 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4904 s = force_version(s);
4905 if (*s == ';' || (s = skipspace(s), *s == ';')) {
4906 PL_nextval[PL_nexttoke].opval = Nullop;
4911 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4912 s = force_version(s);
4924 yylval.ival = CopLINE(PL_curcop);
4928 PL_hints |= HINT_BLOCK_SCOPE;
4935 LOP(OP_WAITPID,XTERM);
4943 static char ctl_l[2];
4945 if (ctl_l[0] == '\0')
4946 ctl_l[0] = toCTRL('L');
4947 gv_fetchpv(ctl_l,TRUE, SVt_PV);
4950 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
4955 if (PL_expect == XOPERATOR)
4961 yylval.ival = OP_XOR;
4966 TERM(sublex_start());
4972 Perl_keyword(pTHX_ register char *d, I32 len)
4977 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
4978 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
4979 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
4980 if (strEQ(d,"__DATA__")) return KEY___DATA__;
4981 if (strEQ(d,"__END__")) return KEY___END__;
4985 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
4990 if (strEQ(d,"and")) return -KEY_and;
4991 if (strEQ(d,"abs")) return -KEY_abs;
4994 if (strEQ(d,"alarm")) return -KEY_alarm;
4995 if (strEQ(d,"atan2")) return -KEY_atan2;
4998 if (strEQ(d,"accept")) return -KEY_accept;
5003 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5006 if (strEQ(d,"bless")) return -KEY_bless;
5007 if (strEQ(d,"bind")) return -KEY_bind;
5008 if (strEQ(d,"binmode")) return -KEY_binmode;
5011 if (strEQ(d,"CORE")) return -KEY_CORE;
5012 if (strEQ(d,"CHECK")) return KEY_CHECK;
5017 if (strEQ(d,"cmp")) return -KEY_cmp;
5018 if (strEQ(d,"chr")) return -KEY_chr;
5019 if (strEQ(d,"cos")) return -KEY_cos;
5022 if (strEQ(d,"chop")) return KEY_chop;
5025 if (strEQ(d,"close")) return -KEY_close;
5026 if (strEQ(d,"chdir")) return -KEY_chdir;
5027 if (strEQ(d,"chomp")) return KEY_chomp;
5028 if (strEQ(d,"chmod")) return -KEY_chmod;
5029 if (strEQ(d,"chown")) return -KEY_chown;
5030 if (strEQ(d,"crypt")) return -KEY_crypt;
5033 if (strEQ(d,"chroot")) return -KEY_chroot;
5034 if (strEQ(d,"caller")) return -KEY_caller;
5037 if (strEQ(d,"connect")) return -KEY_connect;
5040 if (strEQ(d,"closedir")) return -KEY_closedir;
5041 if (strEQ(d,"continue")) return -KEY_continue;
5046 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5051 if (strEQ(d,"do")) return KEY_do;
5054 if (strEQ(d,"die")) return -KEY_die;
5057 if (strEQ(d,"dump")) return -KEY_dump;
5060 if (strEQ(d,"delete")) return KEY_delete;
5063 if (strEQ(d,"defined")) return KEY_defined;
5064 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5067 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5072 if (strEQ(d,"EQ")) { deprecate(d); return -KEY_eq;}
5073 if (strEQ(d,"END")) return KEY_END;
5078 if (strEQ(d,"eq")) return -KEY_eq;
5081 if (strEQ(d,"eof")) return -KEY_eof;
5082 if (strEQ(d,"exp")) return -KEY_exp;
5085 if (strEQ(d,"else")) return KEY_else;
5086 if (strEQ(d,"exit")) return -KEY_exit;
5087 if (strEQ(d,"eval")) return KEY_eval;
5088 if (strEQ(d,"exec")) return -KEY_exec;
5089 if (strEQ(d,"each")) return KEY_each;
5092 if (strEQ(d,"elsif")) return KEY_elsif;
5095 if (strEQ(d,"exists")) return KEY_exists;
5096 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5099 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5100 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5103 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5106 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5107 if (strEQ(d,"endservent")) return -KEY_endservent;
5110 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5117 if (strEQ(d,"for")) return KEY_for;
5120 if (strEQ(d,"fork")) return -KEY_fork;
5123 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5124 if (strEQ(d,"flock")) return -KEY_flock;
5127 if (strEQ(d,"format")) return KEY_format;
5128 if (strEQ(d,"fileno")) return -KEY_fileno;
5131 if (strEQ(d,"foreach")) return KEY_foreach;
5134 if (strEQ(d,"formline")) return -KEY_formline;
5140 if (strEQ(d,"GT")) { deprecate(d); return -KEY_gt;}
5141 if (strEQ(d,"GE")) { deprecate(d); return -KEY_ge;}
5145 if (strnEQ(d,"get",3)) {
5150 if (strEQ(d,"ppid")) return -KEY_getppid;
5151 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5154 if (strEQ(d,"pwent")) return -KEY_getpwent;
5155 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5156 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5159 if (strEQ(d,"peername")) return -KEY_getpeername;
5160 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5161 if (strEQ(d,"priority")) return -KEY_getpriority;
5164 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5167 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5171 else if (*d == 'h') {
5172 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5173 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5174 if (strEQ(d,"hostent")) return -KEY_gethostent;
5176 else if (*d == 'n') {
5177 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5178 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5179 if (strEQ(d,"netent")) return -KEY_getnetent;
5181 else if (*d == 's') {
5182 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5183 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5184 if (strEQ(d,"servent")) return -KEY_getservent;
5185 if (strEQ(d,"sockname")) return -KEY_getsockname;
5186 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5188 else if (*d == 'g') {
5189 if (strEQ(d,"grent")) return -KEY_getgrent;
5190 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5191 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5193 else if (*d == 'l') {
5194 if (strEQ(d,"login")) return -KEY_getlogin;
5196 else if (strEQ(d,"c")) return -KEY_getc;
5201 if (strEQ(d,"gt")) return -KEY_gt;
5202 if (strEQ(d,"ge")) return -KEY_ge;
5205 if (strEQ(d,"grep")) return KEY_grep;
5206 if (strEQ(d,"goto")) return KEY_goto;
5207 if (strEQ(d,"glob")) return KEY_glob;
5210 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5215 if (strEQ(d,"hex")) return -KEY_hex;
5218 if (strEQ(d,"INIT")) return KEY_INIT;
5223 if (strEQ(d,"if")) return KEY_if;
5226 if (strEQ(d,"int")) return -KEY_int;
5229 if (strEQ(d,"index")) return -KEY_index;
5230 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5235 if (strEQ(d,"join")) return -KEY_join;
5239 if (strEQ(d,"keys")) return KEY_keys;
5240 if (strEQ(d,"kill")) return -KEY_kill;
5245 if (strEQ(d,"LT")) { deprecate(d); return -KEY_lt;}
5246 if (strEQ(d,"LE")) { deprecate(d); return -KEY_le;}
5252 if (strEQ(d,"lt")) return -KEY_lt;
5253 if (strEQ(d,"le")) return -KEY_le;
5254 if (strEQ(d,"lc")) return -KEY_lc;
5257 if (strEQ(d,"log")) return -KEY_log;
5260 if (strEQ(d,"last")) return KEY_last;
5261 if (strEQ(d,"link")) return -KEY_link;
5262 if (strEQ(d,"lock")) return -KEY_lock;
5265 if (strEQ(d,"local")) return KEY_local;
5266 if (strEQ(d,"lstat")) return -KEY_lstat;
5269 if (strEQ(d,"length")) return -KEY_length;
5270 if (strEQ(d,"listen")) return -KEY_listen;
5273 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5276 if (strEQ(d,"localtime")) return -KEY_localtime;
5282 case 1: return KEY_m;
5284 if (strEQ(d,"my")) return KEY_my;
5287 if (strEQ(d,"map")) return KEY_map;
5290 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5293 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5294 if (strEQ(d,"msgget")) return -KEY_msgget;
5295 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5296 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5301 if (strEQ(d,"NE")) { deprecate(d); return -KEY_ne;}
5304 if (strEQ(d,"next")) return KEY_next;
5305 if (strEQ(d,"ne")) return -KEY_ne;
5306 if (strEQ(d,"not")) return -KEY_not;
5307 if (strEQ(d,"no")) return KEY_no;
5312 if (strEQ(d,"or")) return -KEY_or;
5315 if (strEQ(d,"ord")) return -KEY_ord;
5316 if (strEQ(d,"oct")) return -KEY_oct;
5317 if (strEQ(d,"our")) return KEY_our;
5320 if (strEQ(d,"open")) return -KEY_open;
5323 if (strEQ(d,"opendir")) return -KEY_opendir;
5330 if (strEQ(d,"pop")) return KEY_pop;
5331 if (strEQ(d,"pos")) return KEY_pos;
5334 if (strEQ(d,"push")) return KEY_push;
5335 if (strEQ(d,"pack")) return -KEY_pack;
5336 if (strEQ(d,"pipe")) return -KEY_pipe;
5339 if (strEQ(d,"print")) return KEY_print;
5342 if (strEQ(d,"printf")) return KEY_printf;
5345 if (strEQ(d,"package")) return KEY_package;
5348 if (strEQ(d,"prototype")) return KEY_prototype;
5353 if (strEQ(d,"q")) return KEY_q;
5354 if (strEQ(d,"qr")) return KEY_qr;
5355 if (strEQ(d,"qq")) return KEY_qq;
5356 if (strEQ(d,"qw")) return KEY_qw;
5357 if (strEQ(d,"qx")) return KEY_qx;
5359 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5364 if (strEQ(d,"ref")) return -KEY_ref;
5367 if (strEQ(d,"read")) return -KEY_read;
5368 if (strEQ(d,"rand")) return -KEY_rand;
5369 if (strEQ(d,"recv")) return -KEY_recv;
5370 if (strEQ(d,"redo")) return KEY_redo;
5373 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5374 if (strEQ(d,"reset")) return -KEY_reset;
5377 if (strEQ(d,"return")) return KEY_return;
5378 if (strEQ(d,"rename")) return -KEY_rename;
5379 if (strEQ(d,"rindex")) return -KEY_rindex;
5382 if (strEQ(d,"require")) return -KEY_require;
5383 if (strEQ(d,"reverse")) return -KEY_reverse;
5384 if (strEQ(d,"readdir")) return -KEY_readdir;
5387 if (strEQ(d,"readlink")) return -KEY_readlink;
5388 if (strEQ(d,"readline")) return -KEY_readline;
5389 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5392 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5398 case 0: return KEY_s;
5400 if (strEQ(d,"scalar")) return KEY_scalar;
5405 if (strEQ(d,"seek")) return -KEY_seek;
5406 if (strEQ(d,"send")) return -KEY_send;
5409 if (strEQ(d,"semop")) return -KEY_semop;
5412 if (strEQ(d,"select")) return -KEY_select;
5413 if (strEQ(d,"semctl")) return -KEY_semctl;
5414 if (strEQ(d,"semget")) return -KEY_semget;
5417 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5418 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5421 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5422 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5425 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5428 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5429 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5430 if (strEQ(d,"setservent")) return -KEY_setservent;
5433 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5434 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5441 if (strEQ(d,"shift")) return KEY_shift;
5444 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5445 if (strEQ(d,"shmget")) return -KEY_shmget;
5448 if (strEQ(d,"shmread")) return -KEY_shmread;
5451 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5452 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5457 if (strEQ(d,"sin")) return -KEY_sin;
5460 if (strEQ(d,"sleep")) return -KEY_sleep;
5463 if (strEQ(d,"sort")) return KEY_sort;
5464 if (strEQ(d,"socket")) return -KEY_socket;
5465 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5468 if (strEQ(d,"split")) return KEY_split;
5469 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5470 if (strEQ(d,"splice")) return KEY_splice;
5473 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5476 if (strEQ(d,"srand")) return -KEY_srand;
5479 if (strEQ(d,"stat")) return -KEY_stat;
5480 if (strEQ(d,"study")) return KEY_study;
5483 if (strEQ(d,"substr")) return -KEY_substr;
5484 if (strEQ(d,"sub")) return KEY_sub;
5489 if (strEQ(d,"system")) return -KEY_system;
5492 if (strEQ(d,"symlink")) return -KEY_symlink;
5493 if (strEQ(d,"syscall")) return -KEY_syscall;
5494 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5495 if (strEQ(d,"sysread")) return -KEY_sysread;
5496 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5499 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5508 if (strEQ(d,"tr")) return KEY_tr;
5511 if (strEQ(d,"tie")) return KEY_tie;
5514 if (strEQ(d,"tell")) return -KEY_tell;
5515 if (strEQ(d,"tied")) return KEY_tied;
5516 if (strEQ(d,"time")) return -KEY_time;
5519 if (strEQ(d,"times")) return -KEY_times;
5522 if (strEQ(d,"telldir")) return -KEY_telldir;
5525 if (strEQ(d,"truncate")) return -KEY_truncate;
5532 if (strEQ(d,"uc")) return -KEY_uc;
5535 if (strEQ(d,"use")) return KEY_use;
5538 if (strEQ(d,"undef")) return KEY_undef;
5539 if (strEQ(d,"until")) return KEY_until;
5540 if (strEQ(d,"untie")) return KEY_untie;
5541 if (strEQ(d,"utime")) return -KEY_utime;
5542 if (strEQ(d,"umask")) return -KEY_umask;
5545 if (strEQ(d,"unless")) return KEY_unless;
5546 if (strEQ(d,"unpack")) return -KEY_unpack;
5547 if (strEQ(d,"unlink")) return -KEY_unlink;
5550 if (strEQ(d,"unshift")) return KEY_unshift;
5551 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5556 if (strEQ(d,"values")) return -KEY_values;
5557 if (strEQ(d,"vec")) return -KEY_vec;
5562 if (strEQ(d,"warn")) return -KEY_warn;
5563 if (strEQ(d,"wait")) return -KEY_wait;
5566 if (strEQ(d,"while")) return KEY_while;
5567 if (strEQ(d,"write")) return -KEY_write;
5570 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5573 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5578 if (len == 1) return -KEY_x;
5579 if (strEQ(d,"xor")) return -KEY_xor;
5582 if (len == 1) return KEY_y;
5591 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5595 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5596 dTHR; /* only for ckWARN */
5597 if (ckWARN(WARN_SYNTAX)) {
5599 for (w = s+2; *w && level; w++) {
5606 for (; *w && isSPACE(*w); w++) ;
5607 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5608 Perl_warner(aTHX_ WARN_SYNTAX,
5609 "%s (...) interpreted as function",name);
5612 while (s < PL_bufend && isSPACE(*s))
5616 while (s < PL_bufend && isSPACE(*s))
5618 if (isIDFIRST_lazy_if(s,UTF)) {
5620 while (isALNUM_lazy_if(s,UTF))
5622 while (s < PL_bufend && isSPACE(*s))
5627 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5631 Perl_croak(aTHX_ "No comma allowed after %s", what);
5636 /* Either returns sv, or mortalizes sv and returns a new SV*.
5637 Best used as sv=new_constant(..., sv, ...).
5638 If s, pv are NULL, calls subroutine with one argument,
5639 and type is used with error messages only. */
5642 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5646 HV *table = GvHV(PL_hintgv); /* ^H */
5650 const char *why, *why1, *why2;
5652 if (!(PL_hints & HINT_LOCALIZE_HH)) {
5655 why = "%^H is not localized";
5659 msg = Perl_newSVpvf(aTHX_ "constant(%s): %s%s%s",
5660 (type ? type: "undef"), why1, why2, why);
5661 yyerror(SvPVX(msg));
5666 why = "%^H is not defined";
5669 cvp = hv_fetch(table, key, strlen(key), FALSE);
5670 if (!cvp || !SvOK(*cvp)) {
5671 why = "} is not defined";
5676 sv_2mortal(sv); /* Parent created it permanently */
5679 pv = sv_2mortal(newSVpvn(s, len));
5681 typesv = sv_2mortal(newSVpv(type, 0));
5683 typesv = &PL_sv_undef;
5685 PUSHSTACKi(PERLSI_OVERLOAD);
5698 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5702 /* Check the eval first */
5703 if (!PL_in_eval && SvTRUE(ERRSV)) {
5705 sv_catpv(ERRSV, "Propagated");
5706 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5708 res = SvREFCNT_inc(sv);
5712 (void)SvREFCNT_inc(res);
5721 why = "}} did not return a defined value";
5722 why1 = "Call to &{$^H{";
5732 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5734 register char *d = dest;
5735 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5738 Perl_croak(aTHX_ ident_too_long);
5739 if (isALNUM(*s)) /* UTF handled below */
5741 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
5746 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5750 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5751 char *t = s + UTF8SKIP(s);
5752 while (*t & 0x80 && is_utf8_mark((U8*)t))
5754 if (d + (t - s) > e)
5755 Perl_croak(aTHX_ ident_too_long);
5756 Copy(s, d, t - s, char);
5769 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5779 e = d + destlen - 3; /* two-character token, ending NUL */
5781 while (isDIGIT(*s)) {
5783 Perl_croak(aTHX_ ident_too_long);
5790 Perl_croak(aTHX_ ident_too_long);
5791 if (isALNUM(*s)) /* UTF handled below */
5793 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
5798 else if (*s == ':' && s[1] == ':') {
5802 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5803 char *t = s + UTF8SKIP(s);
5804 while (*t & 0x80 && is_utf8_mark((U8*)t))
5806 if (d + (t - s) > e)
5807 Perl_croak(aTHX_ ident_too_long);
5808 Copy(s, d, t - s, char);
5819 if (PL_lex_state != LEX_NORMAL)
5820 PL_lex_state = LEX_INTERPENDMAYBE;
5823 if (*s == '$' && s[1] &&
5824 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
5837 if (*d == '^' && *s && isCONTROLVAR(*s)) {
5842 if (isSPACE(s[-1])) {
5845 if (ch != ' ' && ch != '\t') {
5851 if (isIDFIRST_lazy_if(d,UTF)) {
5855 while (e < send && isALNUM_lazy_if(e,UTF) || *e == ':') {
5857 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
5860 Copy(s, d, e - s, char);
5865 while ((isALNUM(*s) || *s == ':') && d < e)
5868 Perl_croak(aTHX_ ident_too_long);
5871 while (s < send && (*s == ' ' || *s == '\t')) s++;
5872 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
5873 dTHR; /* only for ckWARN */
5874 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
5875 const char *brack = *s == '[' ? "[...]" : "{...}";
5876 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5877 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
5878 funny, dest, brack, funny, dest, brack);
5881 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
5885 /* Handle extended ${^Foo} variables
5886 * 1999-02-27 mjd-perl-patch@plover.com */
5887 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
5891 while (isALNUM(*s) && d < e) {
5895 Perl_croak(aTHX_ ident_too_long);
5900 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
5901 PL_lex_state = LEX_INTERPEND;
5904 if (PL_lex_state == LEX_NORMAL) {
5905 dTHR; /* only for ckWARN */
5906 if (ckWARN(WARN_AMBIGUOUS) &&
5907 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
5909 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5910 "Ambiguous use of %c{%s} resolved to %c%s",
5911 funny, dest, funny, dest);
5916 s = bracket; /* let the parser handle it */
5920 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
5921 PL_lex_state = LEX_INTERPEND;
5926 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
5931 *pmfl |= PMf_GLOBAL;
5933 *pmfl |= PMf_CONTINUE;
5937 *pmfl |= PMf_MULTILINE;
5939 *pmfl |= PMf_SINGLELINE;
5941 *pmfl |= PMf_EXTENDED;
5945 S_scan_pat(pTHX_ char *start, I32 type)
5950 s = scan_str(start,FALSE,FALSE);
5953 SvREFCNT_dec(PL_lex_stuff);
5954 PL_lex_stuff = Nullsv;
5955 Perl_croak(aTHX_ "Search pattern not terminated");
5958 pm = (PMOP*)newPMOP(type, 0);
5959 if (PL_multi_open == '?')
5960 pm->op_pmflags |= PMf_ONCE;
5962 while (*s && strchr("iomsx", *s))
5963 pmflag(&pm->op_pmflags,*s++);
5966 while (*s && strchr("iogcmsx", *s))
5967 pmflag(&pm->op_pmflags,*s++);
5969 pm->op_pmpermflags = pm->op_pmflags;
5971 PL_lex_op = (OP*)pm;
5972 yylval.ival = OP_MATCH;
5977 S_scan_subst(pTHX_ char *start)
5984 yylval.ival = OP_NULL;
5986 s = scan_str(start,FALSE,FALSE);
5990 SvREFCNT_dec(PL_lex_stuff);
5991 PL_lex_stuff = Nullsv;
5992 Perl_croak(aTHX_ "Substitution pattern not terminated");
5995 if (s[-1] == PL_multi_open)
5998 first_start = PL_multi_start;
5999 s = scan_str(s,FALSE,FALSE);
6002 SvREFCNT_dec(PL_lex_stuff);
6003 PL_lex_stuff = Nullsv;
6005 SvREFCNT_dec(PL_lex_repl);
6006 PL_lex_repl = Nullsv;
6007 Perl_croak(aTHX_ "Substitution replacement not terminated");
6009 PL_multi_start = first_start; /* so whole substitution is taken together */
6011 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6017 else if (strchr("iogcmsx", *s))
6018 pmflag(&pm->op_pmflags,*s++);
6025 PL_sublex_info.super_bufptr = s;
6026 PL_sublex_info.super_bufend = PL_bufend;
6028 pm->op_pmflags |= PMf_EVAL;
6029 repl = newSVpvn("",0);
6031 sv_catpv(repl, es ? "eval " : "do ");
6032 sv_catpvn(repl, "{ ", 2);
6033 sv_catsv(repl, PL_lex_repl);
6034 sv_catpvn(repl, " };", 2);
6036 SvREFCNT_dec(PL_lex_repl);
6040 pm->op_pmpermflags = pm->op_pmflags;
6041 PL_lex_op = (OP*)pm;
6042 yylval.ival = OP_SUBST;
6047 S_scan_trans(pTHX_ char *start)
6058 yylval.ival = OP_NULL;
6060 s = scan_str(start,FALSE,FALSE);
6063 SvREFCNT_dec(PL_lex_stuff);
6064 PL_lex_stuff = Nullsv;
6065 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6067 if (s[-1] == PL_multi_open)
6070 s = scan_str(s,FALSE,FALSE);
6073 SvREFCNT_dec(PL_lex_stuff);
6074 PL_lex_stuff = Nullsv;
6076 SvREFCNT_dec(PL_lex_repl);
6077 PL_lex_repl = Nullsv;
6078 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6082 o = newSVOP(OP_TRANS, 0, 0);
6083 utf8 = OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF;
6086 New(803,tbl,256,short);
6087 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6091 complement = del = squash = 0;
6092 while (strchr("cdsCU", *s)) {
6094 complement = OPpTRANS_COMPLEMENT;
6096 del = OPpTRANS_DELETE;
6098 squash = OPpTRANS_SQUASH;
6103 utf8 &= ~OPpTRANS_FROM_UTF;
6105 utf8 |= OPpTRANS_FROM_UTF;
6109 utf8 &= ~OPpTRANS_TO_UTF;
6111 utf8 |= OPpTRANS_TO_UTF;
6114 Perl_croak(aTHX_ "Too many /C and /U options");
6119 o->op_private = del|squash|complement|utf8;
6122 yylval.ival = OP_TRANS;
6127 S_scan_heredoc(pTHX_ register char *s)
6131 I32 op_type = OP_SCALAR;
6138 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6142 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6145 for (peek = s; *peek == ' ' || *peek == '\t'; peek++) ;
6146 if (*peek && strchr("`'\"",*peek)) {
6149 s = delimcpy(d, e, s, PL_bufend, term, &len);
6159 if (!isALNUM_lazy_if(s,UTF))
6160 deprecate("bare << to mean <<\"\"");
6161 for (; isALNUM_lazy_if(s,UTF); s++) {
6166 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6167 Perl_croak(aTHX_ "Delimiter for here document is too long");
6170 len = d - PL_tokenbuf;
6171 #ifndef PERL_STRICT_CR
6172 d = strchr(s, '\r');
6176 while (s < PL_bufend) {
6182 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6191 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6196 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6197 herewas = newSVpvn(s,PL_bufend-s);
6199 s--, herewas = newSVpvn(s,d-s);
6200 s += SvCUR(herewas);
6202 tmpstr = NEWSV(87,79);
6203 sv_upgrade(tmpstr, SVt_PVIV);
6208 else if (term == '`') {
6209 op_type = OP_BACKTICK;
6210 SvIVX(tmpstr) = '\\';
6214 PL_multi_start = CopLINE(PL_curcop);
6215 PL_multi_open = PL_multi_close = '<';
6216 term = *PL_tokenbuf;
6217 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6218 char *bufptr = PL_sublex_info.super_bufptr;
6219 char *bufend = PL_sublex_info.super_bufend;
6220 char *olds = s - SvCUR(herewas);
6221 s = strchr(bufptr, '\n');
6225 while (s < bufend &&
6226 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6228 CopLINE_inc(PL_curcop);
6231 CopLINE_set(PL_curcop, PL_multi_start);
6232 missingterm(PL_tokenbuf);
6234 sv_setpvn(herewas,bufptr,d-bufptr+1);
6235 sv_setpvn(tmpstr,d+1,s-d);
6237 sv_catpvn(herewas,s,bufend-s);
6238 (void)strcpy(bufptr,SvPVX(herewas));
6245 while (s < PL_bufend &&
6246 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6248 CopLINE_inc(PL_curcop);
6250 if (s >= PL_bufend) {
6251 CopLINE_set(PL_curcop, PL_multi_start);
6252 missingterm(PL_tokenbuf);
6254 sv_setpvn(tmpstr,d+1,s-d);
6256 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6258 sv_catpvn(herewas,s,PL_bufend-s);
6259 sv_setsv(PL_linestr,herewas);
6260 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6261 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6264 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6265 while (s >= PL_bufend) { /* multiple line string? */
6267 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6268 CopLINE_set(PL_curcop, PL_multi_start);
6269 missingterm(PL_tokenbuf);
6271 CopLINE_inc(PL_curcop);
6272 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6273 #ifndef PERL_STRICT_CR
6274 if (PL_bufend - PL_linestart >= 2) {
6275 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6276 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6278 PL_bufend[-2] = '\n';
6280 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6282 else if (PL_bufend[-1] == '\r')
6283 PL_bufend[-1] = '\n';
6285 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6286 PL_bufend[-1] = '\n';
6288 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6289 SV *sv = NEWSV(88,0);
6291 sv_upgrade(sv, SVt_PVMG);
6292 sv_setsv(sv,PL_linestr);
6293 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6295 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6298 sv_catsv(PL_linestr,herewas);
6299 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6303 sv_catsv(tmpstr,PL_linestr);
6308 PL_multi_end = CopLINE(PL_curcop);
6309 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6310 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6311 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6313 SvREFCNT_dec(herewas);
6314 PL_lex_stuff = tmpstr;
6315 yylval.ival = op_type;
6320 takes: current position in input buffer
6321 returns: new position in input buffer
6322 side-effects: yylval and lex_op are set.
6327 <FH> read from filehandle
6328 <pkg::FH> read from package qualified filehandle
6329 <pkg'FH> read from package qualified filehandle
6330 <$fh> read from filehandle in $fh
6336 S_scan_inputsymbol(pTHX_ char *start)
6338 register char *s = start; /* current position in buffer */
6344 d = PL_tokenbuf; /* start of temp holding space */
6345 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6346 end = strchr(s, '\n');
6349 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6351 /* die if we didn't have space for the contents of the <>,
6352 or if it didn't end, or if we see a newline
6355 if (len >= sizeof PL_tokenbuf)
6356 Perl_croak(aTHX_ "Excessively long <> operator");
6358 Perl_croak(aTHX_ "Unterminated <> operator");
6363 Remember, only scalar variables are interpreted as filehandles by
6364 this code. Anything more complex (e.g., <$fh{$num}>) will be
6365 treated as a glob() call.
6366 This code makes use of the fact that except for the $ at the front,
6367 a scalar variable and a filehandle look the same.
6369 if (*d == '$' && d[1]) d++;
6371 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6372 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6375 /* If we've tried to read what we allow filehandles to look like, and
6376 there's still text left, then it must be a glob() and not a getline.
6377 Use scan_str to pull out the stuff between the <> and treat it
6378 as nothing more than a string.
6381 if (d - PL_tokenbuf != len) {
6382 yylval.ival = OP_GLOB;
6384 s = scan_str(start,FALSE,FALSE);
6386 Perl_croak(aTHX_ "Glob not terminated");
6390 /* we're in a filehandle read situation */
6393 /* turn <> into <ARGV> */
6395 (void)strcpy(d,"ARGV");
6397 /* if <$fh>, create the ops to turn the variable into a
6403 /* try to find it in the pad for this block, otherwise find
6404 add symbol table ops
6406 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6407 OP *o = newOP(OP_PADSV, 0);
6409 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6412 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6413 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6414 newUNOP(OP_RV2SV, 0,
6415 newGVOP(OP_GV, 0, gv)));
6417 PL_lex_op->op_flags |= OPf_SPECIAL;
6418 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6419 yylval.ival = OP_NULL;
6422 /* If it's none of the above, it must be a literal filehandle
6423 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6425 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6426 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6427 yylval.ival = OP_NULL;
6436 takes: start position in buffer
6437 keep_quoted preserve \ on the embedded delimiter(s)
6438 keep_delims preserve the delimiters around the string
6439 returns: position to continue reading from buffer
6440 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6441 updates the read buffer.
6443 This subroutine pulls a string out of the input. It is called for:
6444 q single quotes q(literal text)
6445 ' single quotes 'literal text'
6446 qq double quotes qq(interpolate $here please)
6447 " double quotes "interpolate $here please"
6448 qx backticks qx(/bin/ls -l)
6449 ` backticks `/bin/ls -l`
6450 qw quote words @EXPORT_OK = qw( func() $spam )
6451 m// regexp match m/this/
6452 s/// regexp substitute s/this/that/
6453 tr/// string transliterate tr/this/that/
6454 y/// string transliterate y/this/that/
6455 ($*@) sub prototypes sub foo ($)
6456 (stuff) sub attr parameters sub foo : attr(stuff)
6457 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6459 In most of these cases (all but <>, patterns and transliterate)
6460 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6461 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6462 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6465 It skips whitespace before the string starts, and treats the first
6466 character as the delimiter. If the delimiter is one of ([{< then
6467 the corresponding "close" character )]}> is used as the closing
6468 delimiter. It allows quoting of delimiters, and if the string has
6469 balanced delimiters ([{<>}]) it allows nesting.
6471 The lexer always reads these strings into lex_stuff, except in the
6472 case of the operators which take *two* arguments (s/// and tr///)
6473 when it checks to see if lex_stuff is full (presumably with the 1st
6474 arg to s or tr) and if so puts the string into lex_repl.
6479 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6482 SV *sv; /* scalar value: string */
6483 char *tmps; /* temp string, used for delimiter matching */
6484 register char *s = start; /* current position in the buffer */
6485 register char term; /* terminating character */
6486 register char *to; /* current position in the sv's data */
6487 I32 brackets = 1; /* bracket nesting level */
6488 bool has_utf = FALSE; /* is there any utf8 content? */
6490 /* skip space before the delimiter */
6494 /* mark where we are, in case we need to report errors */
6497 /* after skipping whitespace, the next character is the terminator */
6499 if ((term & 0x80) && UTF)
6502 /* mark where we are */
6503 PL_multi_start = CopLINE(PL_curcop);
6504 PL_multi_open = term;
6506 /* find corresponding closing delimiter */
6507 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6509 PL_multi_close = term;
6511 /* create a new SV to hold the contents. 87 is leak category, I'm
6512 assuming. 79 is the SV's initial length. What a random number. */
6514 sv_upgrade(sv, SVt_PVIV);
6516 (void)SvPOK_only(sv); /* validate pointer */
6518 /* move past delimiter and try to read a complete string */
6520 sv_catpvn(sv, s, 1);
6523 /* extend sv if need be */
6524 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6525 /* set 'to' to the next character in the sv's string */
6526 to = SvPVX(sv)+SvCUR(sv);
6528 /* if open delimiter is the close delimiter read unbridle */
6529 if (PL_multi_open == PL_multi_close) {
6530 for (; s < PL_bufend; s++,to++) {
6531 /* embedded newlines increment the current line number */
6532 if (*s == '\n' && !PL_rsfp)
6533 CopLINE_inc(PL_curcop);
6534 /* handle quoted delimiters */
6535 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6536 if (!keep_quoted && s[1] == term)
6538 /* any other quotes are simply copied straight through */
6542 /* terminate when run out of buffer (the for() condition), or
6543 have found the terminator */
6544 else if (*s == term)
6546 else if (!has_utf && (*s & 0x80) && UTF)
6552 /* if the terminator isn't the same as the start character (e.g.,
6553 matched brackets), we have to allow more in the quoting, and
6554 be prepared for nested brackets.
6557 /* read until we run out of string, or we find the terminator */
6558 for (; s < PL_bufend; s++,to++) {
6559 /* embedded newlines increment the line count */
6560 if (*s == '\n' && !PL_rsfp)
6561 CopLINE_inc(PL_curcop);
6562 /* backslashes can escape the open or closing characters */
6563 if (*s == '\\' && s+1 < PL_bufend) {
6565 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6570 /* allow nested opens and closes */
6571 else if (*s == PL_multi_close && --brackets <= 0)
6573 else if (*s == PL_multi_open)
6575 else if (!has_utf && (*s & 0x80) && UTF)
6580 /* terminate the copied string and update the sv's end-of-string */
6582 SvCUR_set(sv, to - SvPVX(sv));
6585 * this next chunk reads more into the buffer if we're not done yet
6589 break; /* handle case where we are done yet :-) */
6591 #ifndef PERL_STRICT_CR
6592 if (to - SvPVX(sv) >= 2) {
6593 if ((to[-2] == '\r' && to[-1] == '\n') ||
6594 (to[-2] == '\n' && to[-1] == '\r'))
6598 SvCUR_set(sv, to - SvPVX(sv));
6600 else if (to[-1] == '\r')
6603 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6607 /* if we're out of file, or a read fails, bail and reset the current
6608 line marker so we can report where the unterminated string began
6611 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6613 CopLINE_set(PL_curcop, PL_multi_start);
6616 /* we read a line, so increment our line counter */
6617 CopLINE_inc(PL_curcop);
6619 /* update debugger info */
6620 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6621 SV *sv = NEWSV(88,0);
6623 sv_upgrade(sv, SVt_PVMG);
6624 sv_setsv(sv,PL_linestr);
6625 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6628 /* having changed the buffer, we must update PL_bufend */
6629 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6632 /* at this point, we have successfully read the delimited string */
6635 sv_catpvn(sv, s, 1);
6638 PL_multi_end = CopLINE(PL_curcop);
6641 /* if we allocated too much space, give some back */
6642 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6643 SvLEN_set(sv, SvCUR(sv) + 1);
6644 Renew(SvPVX(sv), SvLEN(sv), char);
6647 /* decide whether this is the first or second quoted string we've read
6660 takes: pointer to position in buffer
6661 returns: pointer to new position in buffer
6662 side-effects: builds ops for the constant in yylval.op
6664 Read a number in any of the formats that Perl accepts:
6666 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6667 [\d_]+(\.[\d_]*)?[Ee](\d+)
6669 Underbars (_) are allowed in decimal numbers. If -w is on,
6670 underbars before a decimal point must be at three digit intervals.
6672 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6675 If it reads a number without a decimal point or an exponent, it will
6676 try converting the number to an integer and see if it can do so
6677 without loss of precision.
6681 Perl_scan_num(pTHX_ char *start)
6683 register char *s = start; /* current position in buffer */
6684 register char *d; /* destination in temp buffer */
6685 register char *e; /* end of temp buffer */
6686 UV tryuv; /* used to see if it can be an UV */
6687 NV value; /* number read, as a double */
6688 SV *sv = Nullsv; /* place to put the converted number */
6689 bool floatit; /* boolean: int or float? */
6690 char *lastub = 0; /* position of last underbar */
6691 static char number_too_long[] = "Number too long";
6693 /* We use the first character to decide what type of number this is */
6697 Perl_croak(aTHX_ "panic: scan_num");
6699 /* if it starts with a 0, it could be an octal number, a decimal in
6700 0.13 disguise, or a hexadecimal number, or a binary number. */
6704 u holds the "number so far"
6705 shift the power of 2 of the base
6706 (hex == 4, octal == 3, binary == 1)
6707 overflowed was the number more than we can hold?
6709 Shift is used when we add a digit. It also serves as an "are
6710 we in octal/hex/binary?" indicator to disallow hex characters
6717 bool overflowed = FALSE;
6718 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6719 static char* bases[5] = { "", "binary", "", "octal",
6721 static char* Bases[5] = { "", "Binary", "", "Octal",
6723 static char *maxima[5] = { "",
6724 "0b11111111111111111111111111111111",
6728 char *base, *Base, *max;
6734 } else if (s[1] == 'b') {
6738 /* check for a decimal in disguise */
6739 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6741 /* so it must be octal */
6745 base = bases[shift];
6746 Base = Bases[shift];
6747 max = maxima[shift];
6749 /* read the rest of the number */
6751 /* x is used in the overflow test,
6752 b is the digit we're adding on. */
6757 /* if we don't mention it, we're done */
6766 /* 8 and 9 are not octal */
6769 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6773 case '2': case '3': case '4':
6774 case '5': case '6': case '7':
6776 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6780 b = *s++ & 15; /* ASCII digit -> value of digit */
6784 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6785 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6786 /* make sure they said 0x */
6791 /* Prepare to put the digit we have onto the end
6792 of the number so far. We check for overflows.
6797 x = u << shift; /* make room for the digit */
6799 if ((x >> shift) != u
6800 && !(PL_hints & HINT_NEW_BINARY)) {
6804 if (ckWARN_d(WARN_OVERFLOW))
6805 Perl_warner(aTHX_ WARN_OVERFLOW,
6806 "Integer overflow in %s number",
6809 u = x | b; /* add the digit to the end */
6812 n *= nvshift[shift];
6813 /* If an NV has not enough bits in its
6814 * mantissa to represent an UV this summing of
6815 * small low-order numbers is a waste of time
6816 * (because the NV cannot preserve the
6817 * low-order bits anyway): we could just
6818 * remember when did we overflow and in the
6819 * end just multiply n by the right
6827 /* if we get here, we had success: make a scalar value from
6834 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6835 Perl_warner(aTHX_ WARN_PORTABLE,
6836 "%s number > %s non-portable",
6843 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6844 Perl_warner(aTHX_ WARN_PORTABLE,
6845 "%s number > %s non-portable",
6850 if (PL_hints & HINT_NEW_BINARY)
6851 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6856 handle decimal numbers.
6857 we're also sent here when we read a 0 as the first digit
6859 case '1': case '2': case '3': case '4': case '5':
6860 case '6': case '7': case '8': case '9': case '.':
6863 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
6866 /* read next group of digits and _ and copy into d */
6867 while (isDIGIT(*s) || *s == '_') {
6868 /* skip underscores, checking for misplaced ones
6872 dTHR; /* only for ckWARN */
6873 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
6874 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6878 /* check for end of fixed-length buffer */
6880 Perl_croak(aTHX_ number_too_long);
6881 /* if we're ok, copy the character */
6886 /* final misplaced underbar check */
6887 if (lastub && s - lastub != 3) {
6889 if (ckWARN(WARN_SYNTAX))
6890 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6893 /* read a decimal portion if there is one. avoid
6894 3..5 being interpreted as the number 3. followed
6897 if (*s == '.' && s[1] != '.') {
6901 /* copy, ignoring underbars, until we run out of
6902 digits. Note: no misplaced underbar checks!
6904 for (; isDIGIT(*s) || *s == '_'; s++) {
6905 /* fixed length buffer check */
6907 Perl_croak(aTHX_ number_too_long);
6911 if (*s == '.' && isDIGIT(s[1])) {
6912 /* oops, it's really a v-string, but without the "v" */
6918 /* read exponent part, if present */
6919 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
6923 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
6924 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
6926 /* allow positive or negative exponent */
6927 if (*s == '+' || *s == '-')
6930 /* read digits of exponent (no underbars :-) */
6931 while (isDIGIT(*s)) {
6933 Perl_croak(aTHX_ number_too_long);
6938 /* terminate the string */
6941 /* make an sv from the string */
6944 value = Atof(PL_tokenbuf);
6947 See if we can make do with an integer value without loss of
6948 precision. We use I_V to cast to an int, because some
6949 compilers have issues. Then we try casting it back and see
6950 if it was the same. We only do this if we know we
6951 specifically read an integer.
6953 Note: if floatit is true, then we don't need to do the
6957 if (!floatit && (NV)tryuv == value)
6958 sv_setuv(sv, tryuv);
6960 sv_setnv(sv, value);
6961 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
6962 (PL_hints & HINT_NEW_INTEGER) )
6963 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
6964 (floatit ? "float" : "integer"),
6967 /* if it starts with a v, it could be a version number */
6973 while (isDIGIT(*pos) || *pos == '_')
6975 if (!isALPHA(*pos)) {
6977 U8 tmpbuf[UTF8_MAXLEN];
6980 s++; /* get past 'v' */
6983 sv_setpvn(sv, "", 0);
6986 if (*s == '0' && isDIGIT(s[1]))
6987 yyerror("Octal number in vector unsupported");
6990 /* this is atoi() that tolerates underscores */
6993 while (--end >= s) {
6998 rev += (*end - '0') * mult;
7000 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
7001 Perl_warner(aTHX_ WARN_OVERFLOW,
7002 "Integer overflow in decimal number");
7005 tmpend = uv_to_utf8(tmpbuf, rev);
7006 utf8 = utf8 || rev > 127;
7007 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
7008 if (*pos == '.' && isDIGIT(pos[1]))
7014 while (isDIGIT(*pos) || *pos == '_')
7022 sv_utf8_downgrade(sv, TRUE);
7029 /* make the op for the constant and return */
7032 yylval.opval = newSVOP(OP_CONST, 0, sv);
7034 yylval.opval = Nullop;
7040 S_scan_formline(pTHX_ register char *s)
7045 SV *stuff = newSVpvn("",0);
7046 bool needargs = FALSE;
7049 if (*s == '.' || *s == '}') {
7051 #ifdef PERL_STRICT_CR
7052 for (t = s+1;*t == ' ' || *t == '\t'; t++) ;
7054 for (t = s+1;*t == ' ' || *t == '\t' || *t == '\r'; t++) ;
7056 if (*t == '\n' || t == PL_bufend)
7059 if (PL_in_eval && !PL_rsfp) {
7060 eol = strchr(s,'\n');
7065 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7067 for (t = s; t < eol; t++) {
7068 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7070 goto enough; /* ~~ must be first line in formline */
7072 if (*t == '@' || *t == '^')
7075 sv_catpvn(stuff, s, eol-s);
7076 #ifndef PERL_STRICT_CR
7077 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7078 char *end = SvPVX(stuff) + SvCUR(stuff);
7087 s = filter_gets(PL_linestr, PL_rsfp, 0);
7088 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7089 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7092 yyerror("Format not terminated");
7102 PL_lex_state = LEX_NORMAL;
7103 PL_nextval[PL_nexttoke].ival = 0;
7107 PL_lex_state = LEX_FORMLINE;
7108 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7110 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7114 SvREFCNT_dec(stuff);
7115 PL_lex_formbrack = 0;
7126 PL_cshlen = strlen(PL_cshname);
7131 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7134 I32 oldsavestack_ix = PL_savestack_ix;
7135 CV* outsidecv = PL_compcv;
7139 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7141 SAVEI32(PL_subline);
7142 save_item(PL_subname);
7145 SAVESPTR(PL_comppad_name);
7146 SAVESPTR(PL_compcv);
7147 SAVEI32(PL_comppad_name_fill);
7148 SAVEI32(PL_min_intro_pending);
7149 SAVEI32(PL_max_intro_pending);
7150 SAVEI32(PL_pad_reset_pending);
7152 PL_compcv = (CV*)NEWSV(1104,0);
7153 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7154 CvFLAGS(PL_compcv) |= flags;
7156 PL_comppad = newAV();
7157 av_push(PL_comppad, Nullsv);
7158 PL_curpad = AvARRAY(PL_comppad);
7159 PL_comppad_name = newAV();
7160 PL_comppad_name_fill = 0;
7161 PL_min_intro_pending = 0;
7163 PL_subline = CopLINE(PL_curcop);
7165 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7166 PL_curpad[0] = (SV*)newAV();
7167 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7168 #endif /* USE_THREADS */
7170 comppadlist = newAV();
7171 AvREAL_off(comppadlist);
7172 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7173 av_store(comppadlist, 1, (SV*)PL_comppad);
7175 CvPADLIST(PL_compcv) = comppadlist;
7176 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7178 CvOWNER(PL_compcv) = 0;
7179 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7180 MUTEX_INIT(CvMUTEXP(PL_compcv));
7181 #endif /* USE_THREADS */
7183 return oldsavestack_ix;
7187 Perl_yywarn(pTHX_ char *s)
7190 PL_in_eval |= EVAL_WARNONLY;
7192 PL_in_eval &= ~EVAL_WARNONLY;
7197 Perl_yyerror(pTHX_ char *s)
7201 char *context = NULL;
7205 if (!yychar || (yychar == ';' && !PL_rsfp))
7207 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7208 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7209 while (isSPACE(*PL_oldoldbufptr))
7211 context = PL_oldoldbufptr;
7212 contlen = PL_bufptr - PL_oldoldbufptr;
7214 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7215 PL_oldbufptr != PL_bufptr) {
7216 while (isSPACE(*PL_oldbufptr))
7218 context = PL_oldbufptr;
7219 contlen = PL_bufptr - PL_oldbufptr;
7221 else if (yychar > 255)
7222 where = "next token ???";
7223 #ifdef USE_PURE_BISON
7224 /* GNU Bison sets the value -2 */
7225 else if (yychar == -2) {
7227 else if ((yychar & 127) == 127) {
7229 if (PL_lex_state == LEX_NORMAL ||
7230 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7231 where = "at end of line";
7232 else if (PL_lex_inpat)
7233 where = "within pattern";
7235 where = "within string";
7238 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7240 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7241 else if (isPRINT_LC(yychar))
7242 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7244 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7245 where = SvPVX(where_sv);
7247 msg = sv_2mortal(newSVpv(s, 0));
7248 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7249 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7251 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7253 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7254 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7255 Perl_sv_catpvf(aTHX_ msg,
7256 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7257 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7260 if (PL_in_eval & EVAL_WARNONLY)
7261 Perl_warn(aTHX_ "%"SVf, msg);
7264 if (PL_error_count >= 10)
7265 Perl_croak(aTHX_ "%s has too many errors.\n", CopFILE(PL_curcop));
7267 PL_in_my_stash = Nullhv;
7278 * Restore a source filter.
7282 restore_rsfp(pTHXo_ void *f)
7284 PerlIO *fp = (PerlIO*)f;
7286 if (PL_rsfp == PerlIO_stdin())
7287 PerlIO_clearerr(PL_rsfp);
7288 else if (PL_rsfp && (PL_rsfp != fp))
7289 PerlIO_close(PL_rsfp);