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)
324 S_utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
326 I32 count = FILTER_READ(idx+1, sv, maxlen);
330 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
331 tend = utf16_to_utf8((U16*)SvPVX(sv), tmps, SvCUR(sv));
332 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 (void)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 != ')')
1273 else if (s[2] == '{' /* This should match regcomp.c */
1274 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1277 char *regparse = s + (s[2] == '{' ? 3 : 4);
1280 while (count && (c = *regparse)) {
1281 if (c == '\\' && regparse[1])
1289 if (*regparse != ')') {
1290 regparse--; /* Leave one char for continuation. */
1291 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1293 while (s < regparse)
1298 /* likewise skip #-initiated comments in //x patterns */
1299 else if (*s == '#' && PL_lex_inpat &&
1300 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1301 while (s+1 < send && *s != '\n')
1305 /* check for embedded arrays (@foo, @:foo, @'foo, @{foo}, @$foo) */
1306 else if (*s == '@' && s[1]
1307 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$", s[1])))
1310 /* check for embedded scalars. only stop if we're sure it's a
1313 else if (*s == '$') {
1314 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1316 if (s + 1 < send && !strchr("()| \n\t", s[1]))
1317 break; /* in regexp, $ might be tail anchor */
1320 /* (now in tr/// code again) */
1322 if (*s & 0x80 && thisutf) {
1323 (void)utf8_to_uv((U8*)s, &len);
1325 /* illegal UTF8, make it valid */
1326 char *old_pvx = SvPVX(sv);
1327 /* need space for one extra char (NOTE: SvCUR() not set here) */
1328 d = SvGROW(sv, SvLEN(sv) + 1) + (d - old_pvx);
1329 d = (char*)uv_to_utf8((U8*)d, (U8)*s++);
1340 if (*s == '\\' && s+1 < send) {
1343 /* some backslashes we leave behind */
1344 if (*leaveit && *s && strchr(leaveit, *s)) {
1350 /* deprecate \1 in strings and substitution replacements */
1351 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1352 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1354 dTHR; /* only for ckWARN */
1355 if (ckWARN(WARN_SYNTAX))
1356 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1361 /* string-change backslash escapes */
1362 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1367 /* if we get here, it's either a quoted -, or a digit */
1370 /* quoted - in transliterations */
1372 if (PL_lex_inwhat == OP_TRANS) {
1380 if (ckWARN(WARN_MISC) && isALPHA(*s))
1381 Perl_warner(aTHX_ WARN_MISC,
1382 "Unrecognized escape \\%c passed through",
1384 /* default action is to copy the quoted character */
1389 /* \132 indicates an octal constant */
1390 case '0': case '1': case '2': case '3':
1391 case '4': case '5': case '6': case '7':
1392 uv = (UV)scan_oct(s, 3, &len);
1394 goto NUM_ESCAPE_INSERT;
1396 /* \x24 indicates a hex constant */
1400 char* e = strchr(s, '}');
1402 yyerror("Missing right brace on \\x{}");
1405 uv = (UV)scan_hex(s + 1, e - s - 1, &len);
1409 uv = (UV)scan_hex(s, 2, &len);
1414 /* Insert oct or hex escaped character.
1415 * There will always enough room in sv since such escapes will
1416 * be longer than any utf8 sequence they can end up as
1419 if (!thisutf && !has_utf && uv > 255) {
1420 /* might need to recode whatever we have accumulated so far
1421 * if it contains any hibit chars
1425 for (c = SvPVX(sv); c < d; c++) {
1430 char *old_pvx = SvPVX(sv);
1432 d = SvGROW(sv, SvCUR(sv) + hicount + 1) + (d - old_pvx);
1441 uv_to_utf8((U8*)dst, (U8)*src--);
1451 if (thisutf || uv > 255) {
1452 d = (char*)uv_to_utf8((U8*)d, uv);
1464 /* \N{latin small letter a} is a named character */
1468 char* e = strchr(s, '}');
1474 yyerror("Missing right brace on \\N{}");
1478 res = newSVpvn(s + 1, e - s - 1);
1479 res = new_constant( Nullch, 0, "charnames",
1480 res, Nullsv, "\\N{...}" );
1481 str = SvPV(res,len);
1484 if (len > e - s + 4) {
1485 char *odest = SvPVX(sv);
1487 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1488 d = SvPVX(sv) + (d - odest);
1490 Copy(str, d, len, char);
1497 yyerror("Missing braces on \\N{}");
1500 /* \c is a control character */
1514 /* printf-style backslashes, formfeeds, newlines, etc */
1532 *d++ = '\047'; /* CP 1047 */
1535 *d++ = '\057'; /* CP 1047 */
1549 } /* end if (backslash) */
1552 } /* while loop to process each character */
1554 /* terminate the string and set up the sv */
1556 SvCUR_set(sv, d - SvPVX(sv));
1561 /* shrink the sv if we allocated more than we used */
1562 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1563 SvLEN_set(sv, SvCUR(sv) + 1);
1564 Renew(SvPVX(sv), SvLEN(sv), char);
1567 /* return the substring (via yylval) only if we parsed anything */
1568 if (s > PL_bufptr) {
1569 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1570 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1572 ( PL_lex_inwhat == OP_TRANS
1574 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1577 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1584 * Returns TRUE if there's more to the expression (e.g., a subscript),
1587 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1589 * ->[ and ->{ return TRUE
1590 * { and [ outside a pattern are always subscripts, so return TRUE
1591 * if we're outside a pattern and it's not { or [, then return FALSE
1592 * if we're in a pattern and the first char is a {
1593 * {4,5} (any digits around the comma) returns FALSE
1594 * if we're in a pattern and the first char is a [
1596 * [SOMETHING] has a funky algorithm to decide whether it's a
1597 * character class or not. It has to deal with things like
1598 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1599 * anything else returns TRUE
1602 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1605 S_intuit_more(pTHX_ register char *s)
1607 if (PL_lex_brackets)
1609 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1611 if (*s != '{' && *s != '[')
1616 /* In a pattern, so maybe we have {n,m}. */
1633 /* On the other hand, maybe we have a character class */
1636 if (*s == ']' || *s == '^')
1639 /* this is terrifying, and it works */
1640 int weight = 2; /* let's weigh the evidence */
1642 unsigned char un_char = 255, last_un_char;
1643 char *send = strchr(s,']');
1644 char tmpbuf[sizeof PL_tokenbuf * 4];
1646 if (!send) /* has to be an expression */
1649 Zero(seen,256,char);
1652 else if (isDIGIT(*s)) {
1654 if (isDIGIT(s[1]) && s[2] == ']')
1660 for (; s < send; s++) {
1661 last_un_char = un_char;
1662 un_char = (unsigned char)*s;
1667 weight -= seen[un_char] * 10;
1668 if (isALNUM_lazy_if(s+1,UTF)) {
1669 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1670 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1675 else if (*s == '$' && s[1] &&
1676 strchr("[#!%*<>()-=",s[1])) {
1677 if (/*{*/ strchr("])} =",s[2]))
1686 if (strchr("wds]",s[1]))
1688 else if (seen['\''] || seen['"'])
1690 else if (strchr("rnftbxcav",s[1]))
1692 else if (isDIGIT(s[1])) {
1694 while (s[1] && isDIGIT(s[1]))
1704 if (strchr("aA01! ",last_un_char))
1706 if (strchr("zZ79~",s[1]))
1708 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1709 weight -= 5; /* cope with negative subscript */
1712 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1713 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1718 if (keyword(tmpbuf, d - tmpbuf))
1721 if (un_char == last_un_char + 1)
1723 weight -= seen[un_char];
1728 if (weight >= 0) /* probably a character class */
1738 * Does all the checking to disambiguate
1740 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1741 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1743 * First argument is the stuff after the first token, e.g. "bar".
1745 * Not a method if bar is a filehandle.
1746 * Not a method if foo is a subroutine prototyped to take a filehandle.
1747 * Not a method if it's really "Foo $bar"
1748 * Method if it's "foo $bar"
1749 * Not a method if it's really "print foo $bar"
1750 * Method if it's really "foo package::" (interpreted as package->foo)
1751 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1752 * Not a method if bar is a filehandle or package, but is quoted with
1757 S_intuit_method(pTHX_ char *start, GV *gv)
1759 char *s = start + (*start == '$');
1760 char tmpbuf[sizeof PL_tokenbuf];
1768 if ((cv = GvCVu(gv))) {
1769 char *proto = SvPVX(cv);
1779 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1780 /* start is the beginning of the possible filehandle/object,
1781 * and s is the end of it
1782 * tmpbuf is a copy of it
1785 if (*start == '$') {
1786 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1791 return *s == '(' ? FUNCMETH : METHOD;
1793 if (!keyword(tmpbuf, len)) {
1794 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1799 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1800 if (indirgv && GvCVu(indirgv))
1802 /* filehandle or package name makes it a method */
1803 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1805 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1806 return 0; /* no assumptions -- "=>" quotes bearword */
1808 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1809 newSVpvn(tmpbuf,len));
1810 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1814 return *s == '(' ? FUNCMETH : METHOD;
1822 * Return a string of Perl code to load the debugger. If PERL5DB
1823 * is set, it will return the contents of that, otherwise a
1824 * compile-time require of perl5db.pl.
1831 char *pdb = PerlEnv_getenv("PERL5DB");
1835 SETERRNO(0,SS$_NORMAL);
1836 return "BEGIN { require 'perl5db.pl' }";
1842 /* Encoded script support. filter_add() effectively inserts a
1843 * 'pre-processing' function into the current source input stream.
1844 * Note that the filter function only applies to the current source file
1845 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1847 * The datasv parameter (which may be NULL) can be used to pass
1848 * private data to this instance of the filter. The filter function
1849 * can recover the SV using the FILTER_DATA macro and use it to
1850 * store private buffers and state information.
1852 * The supplied datasv parameter is upgraded to a PVIO type
1853 * and the IoDIRP field is used to store the function pointer,
1854 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1855 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1856 * private use must be set using malloc'd pointers.
1860 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1865 if (!PL_rsfp_filters)
1866 PL_rsfp_filters = newAV();
1868 datasv = NEWSV(255,0);
1869 if (!SvUPGRADE(datasv, SVt_PVIO))
1870 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1871 IoDIRP(datasv) = (DIR*)funcp; /* stash funcp into spare field */
1872 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1873 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1874 funcp, SvPV_nolen(datasv)));
1875 av_unshift(PL_rsfp_filters, 1);
1876 av_store(PL_rsfp_filters, 0, datasv) ;
1881 /* Delete most recently added instance of this filter function. */
1883 Perl_filter_del(pTHX_ filter_t funcp)
1886 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1887 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1889 /* if filter is on top of stack (usual case) just pop it off */
1890 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1891 if (IoDIRP(datasv) == (DIR*)funcp) {
1892 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1893 IoDIRP(datasv) = (DIR*)NULL;
1894 sv_free(av_pop(PL_rsfp_filters));
1898 /* we need to search for the correct entry and clear it */
1899 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1903 /* Invoke the n'th filter function for the current rsfp. */
1905 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1908 /* 0 = read one text line */
1913 if (!PL_rsfp_filters)
1915 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1916 /* Provide a default input filter to make life easy. */
1917 /* Note that we append to the line. This is handy. */
1918 DEBUG_P(PerlIO_printf(Perl_debug_log,
1919 "filter_read %d: from rsfp\n", idx));
1923 int old_len = SvCUR(buf_sv) ;
1925 /* ensure buf_sv is large enough */
1926 SvGROW(buf_sv, old_len + maxlen) ;
1927 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1928 if (PerlIO_error(PL_rsfp))
1929 return -1; /* error */
1931 return 0 ; /* end of file */
1933 SvCUR_set(buf_sv, old_len + len) ;
1936 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1937 if (PerlIO_error(PL_rsfp))
1938 return -1; /* error */
1940 return 0 ; /* end of file */
1943 return SvCUR(buf_sv);
1945 /* Skip this filter slot if filter has been deleted */
1946 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1947 DEBUG_P(PerlIO_printf(Perl_debug_log,
1948 "filter_read %d: skipped (filter deleted)\n",
1950 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1952 /* Get function pointer hidden within datasv */
1953 funcp = (filter_t)IoDIRP(datasv);
1954 DEBUG_P(PerlIO_printf(Perl_debug_log,
1955 "filter_read %d: via function %p (%s)\n",
1956 idx, funcp, SvPV_nolen(datasv)));
1957 /* Call function. The function is expected to */
1958 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1959 /* Return: <0:error, =0:eof, >0:not eof */
1960 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1964 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1966 #ifdef PERL_CR_FILTER
1967 if (!PL_rsfp_filters) {
1968 filter_add(S_cr_textfilter,NULL);
1971 if (PL_rsfp_filters) {
1974 SvCUR_set(sv, 0); /* start with empty line */
1975 if (FILTER_READ(0, sv, 0) > 0)
1976 return ( SvPVX(sv) ) ;
1981 return (sv_gets(sv, fp, append));
1986 static char* exp_name[] =
1987 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
1988 "ATTRTERM", "TERMBLOCK"
1995 Works out what to call the token just pulled out of the input
1996 stream. The yacc parser takes care of taking the ops we return and
1997 stitching them into a tree.
2003 if read an identifier
2004 if we're in a my declaration
2005 croak if they tried to say my($foo::bar)
2006 build the ops for a my() declaration
2007 if it's an access to a my() variable
2008 are we in a sort block?
2009 croak if my($a); $a <=> $b
2010 build ops for access to a my() variable
2011 if in a dq string, and they've said @foo and we can't find @foo
2013 build ops for a bareword
2014 if we already built the token before, use it.
2018 #ifdef USE_PURE_BISON
2019 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
2032 #ifdef USE_PURE_BISON
2033 yylval_pointer = lvalp;
2034 yychar_pointer = lcharp;
2037 /* check if there's an identifier for us to look at */
2038 if (PL_pending_ident) {
2039 /* pit holds the identifier we read and pending_ident is reset */
2040 char pit = PL_pending_ident;
2041 PL_pending_ident = 0;
2043 /* if we're in a my(), we can't allow dynamics here.
2044 $foo'bar has already been turned into $foo::bar, so
2045 just check for colons.
2047 if it's a legal name, the OP is a PADANY.
2050 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
2051 if (strchr(PL_tokenbuf,':'))
2052 yyerror(Perl_form(aTHX_ "No package name allowed for "
2053 "variable %s in \"our\"",
2055 tmp = pad_allocmy(PL_tokenbuf);
2058 if (strchr(PL_tokenbuf,':'))
2059 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
2061 yylval.opval = newOP(OP_PADANY, 0);
2062 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
2068 build the ops for accesses to a my() variable.
2070 Deny my($a) or my($b) in a sort block, *if* $a or $b is
2071 then used in a comparison. This catches most, but not
2072 all cases. For instance, it catches
2073 sort { my($a); $a <=> $b }
2075 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2076 (although why you'd do that is anyone's guess).
2079 if (!strchr(PL_tokenbuf,':')) {
2081 /* Check for single character per-thread SVs */
2082 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2083 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2084 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2086 yylval.opval = newOP(OP_THREADSV, 0);
2087 yylval.opval->op_targ = tmp;
2090 #endif /* USE_THREADS */
2091 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2092 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2093 /* might be an "our" variable" */
2094 if (SvFLAGS(namesv) & SVpad_OUR) {
2095 /* build ops for a bareword */
2096 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2097 sv_catpvn(sym, "::", 2);
2098 sv_catpv(sym, PL_tokenbuf+1);
2099 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2100 yylval.opval->op_private = OPpCONST_ENTERED;
2101 gv_fetchpv(SvPVX(sym),
2103 ? (GV_ADDMULTI | GV_ADDINEVAL)
2106 ((PL_tokenbuf[0] == '$') ? SVt_PV
2107 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2112 /* if it's a sort block and they're naming $a or $b */
2113 if (PL_last_lop_op == OP_SORT &&
2114 PL_tokenbuf[0] == '$' &&
2115 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2118 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2119 d < PL_bufend && *d != '\n';
2122 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2123 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2129 yylval.opval = newOP(OP_PADANY, 0);
2130 yylval.opval->op_targ = tmp;
2136 Whine if they've said @foo in a doublequoted string,
2137 and @foo isn't a variable we can find in the symbol
2140 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2141 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2142 if (!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2143 yyerror(Perl_form(aTHX_ "In string, %s now must be written as \\%s",
2144 PL_tokenbuf, PL_tokenbuf));
2147 /* build ops for a bareword */
2148 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2149 yylval.opval->op_private = OPpCONST_ENTERED;
2150 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2151 ((PL_tokenbuf[0] == '$') ? SVt_PV
2152 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2157 /* no identifier pending identification */
2159 switch (PL_lex_state) {
2161 case LEX_NORMAL: /* Some compilers will produce faster */
2162 case LEX_INTERPNORMAL: /* code if we comment these out. */
2166 /* when we've already built the next token, just pull it out of the queue */
2169 yylval = PL_nextval[PL_nexttoke];
2171 PL_lex_state = PL_lex_defer;
2172 PL_expect = PL_lex_expect;
2173 PL_lex_defer = LEX_NORMAL;
2175 return(PL_nexttype[PL_nexttoke]);
2177 /* interpolated case modifiers like \L \U, including \Q and \E.
2178 when we get here, PL_bufptr is at the \
2180 case LEX_INTERPCASEMOD:
2182 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2183 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2185 /* handle \E or end of string */
2186 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2190 if (PL_lex_casemods) {
2191 oldmod = PL_lex_casestack[--PL_lex_casemods];
2192 PL_lex_casestack[PL_lex_casemods] = '\0';
2194 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2196 PL_lex_state = LEX_INTERPCONCAT;
2200 if (PL_bufptr != PL_bufend)
2202 PL_lex_state = LEX_INTERPCONCAT;
2207 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2208 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2209 if (strchr("LU", *s) &&
2210 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2212 PL_lex_casestack[--PL_lex_casemods] = '\0';
2215 if (PL_lex_casemods > 10) {
2216 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2217 if (newlb != PL_lex_casestack) {
2219 PL_lex_casestack = newlb;
2222 PL_lex_casestack[PL_lex_casemods++] = *s;
2223 PL_lex_casestack[PL_lex_casemods] = '\0';
2224 PL_lex_state = LEX_INTERPCONCAT;
2225 PL_nextval[PL_nexttoke].ival = 0;
2228 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2230 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2232 PL_nextval[PL_nexttoke].ival = OP_LC;
2234 PL_nextval[PL_nexttoke].ival = OP_UC;
2236 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2238 Perl_croak(aTHX_ "panic: yylex");
2241 if (PL_lex_starts) {
2250 case LEX_INTERPPUSH:
2251 return sublex_push();
2253 case LEX_INTERPSTART:
2254 if (PL_bufptr == PL_bufend)
2255 return sublex_done();
2257 PL_lex_dojoin = (*PL_bufptr == '@');
2258 PL_lex_state = LEX_INTERPNORMAL;
2259 if (PL_lex_dojoin) {
2260 PL_nextval[PL_nexttoke].ival = 0;
2263 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2264 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2265 force_next(PRIVATEREF);
2267 force_ident("\"", '$');
2268 #endif /* USE_THREADS */
2269 PL_nextval[PL_nexttoke].ival = 0;
2271 PL_nextval[PL_nexttoke].ival = 0;
2273 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2276 if (PL_lex_starts++) {
2282 case LEX_INTERPENDMAYBE:
2283 if (intuit_more(PL_bufptr)) {
2284 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2290 if (PL_lex_dojoin) {
2291 PL_lex_dojoin = FALSE;
2292 PL_lex_state = LEX_INTERPCONCAT;
2295 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2296 && SvEVALED(PL_lex_repl))
2298 if (PL_bufptr != PL_bufend)
2299 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2300 PL_lex_repl = Nullsv;
2303 case LEX_INTERPCONCAT:
2305 if (PL_lex_brackets)
2306 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2308 if (PL_bufptr == PL_bufend)
2309 return sublex_done();
2311 if (SvIVX(PL_linestr) == '\'') {
2312 SV *sv = newSVsv(PL_linestr);
2315 else if ( PL_hints & HINT_NEW_RE )
2316 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2317 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2321 s = scan_const(PL_bufptr);
2323 PL_lex_state = LEX_INTERPCASEMOD;
2325 PL_lex_state = LEX_INTERPSTART;
2328 if (s != PL_bufptr) {
2329 PL_nextval[PL_nexttoke] = yylval;
2332 if (PL_lex_starts++)
2342 PL_lex_state = LEX_NORMAL;
2343 s = scan_formline(PL_bufptr);
2344 if (!PL_lex_formbrack)
2350 PL_oldoldbufptr = PL_oldbufptr;
2353 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2354 exp_name[PL_expect], s);
2360 if (isIDFIRST_lazy_if(s,UTF))
2362 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2365 goto fake_eof; /* emulate EOF on ^D or ^Z */
2370 if (PL_lex_brackets)
2371 yyerror("Missing right curly or square bracket");
2374 if (s++ < PL_bufend)
2375 goto retry; /* ignore stray nulls */
2378 if (!PL_in_eval && !PL_preambled) {
2379 PL_preambled = TRUE;
2380 sv_setpv(PL_linestr,incl_perldb());
2381 if (SvCUR(PL_linestr))
2382 sv_catpv(PL_linestr,";");
2384 while(AvFILLp(PL_preambleav) >= 0) {
2385 SV *tmpsv = av_shift(PL_preambleav);
2386 sv_catsv(PL_linestr, tmpsv);
2387 sv_catpv(PL_linestr, ";");
2390 sv_free((SV*)PL_preambleav);
2391 PL_preambleav = NULL;
2393 if (PL_minus_n || PL_minus_p) {
2394 sv_catpv(PL_linestr, "LINE: while (<>) {");
2396 sv_catpv(PL_linestr,"chomp;");
2398 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2400 GvIMPORTED_AV_on(gv);
2402 if (strchr("/'\"", *PL_splitstr)
2403 && strchr(PL_splitstr + 1, *PL_splitstr))
2404 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2407 s = "'~#\200\1'"; /* surely one char is unused...*/
2408 while (s[1] && strchr(PL_splitstr, *s)) s++;
2410 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2411 "q" + (delim == '\''), delim);
2412 for (s = PL_splitstr; *s; s++) {
2414 sv_catpvn(PL_linestr, "\\", 1);
2415 sv_catpvn(PL_linestr, s, 1);
2417 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2421 sv_catpv(PL_linestr,"@F=split(' ');");
2424 sv_catpv(PL_linestr, "\n");
2425 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2426 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2427 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2428 SV *sv = NEWSV(85,0);
2430 sv_upgrade(sv, SVt_PVMG);
2431 sv_setsv(sv,PL_linestr);
2432 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2437 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2440 if (PL_preprocess && !PL_in_eval)
2441 (void)PerlProc_pclose(PL_rsfp);
2442 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2443 PerlIO_clearerr(PL_rsfp);
2445 (void)PerlIO_close(PL_rsfp);
2447 PL_doextract = FALSE;
2449 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2450 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2451 sv_catpv(PL_linestr,";}");
2452 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2453 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2454 PL_minus_n = PL_minus_p = 0;
2457 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2458 sv_setpv(PL_linestr,"");
2459 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2462 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2463 PL_doextract = FALSE;
2465 /* Incest with pod. */
2466 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2467 sv_setpv(PL_linestr, "");
2468 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2469 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2470 PL_doextract = FALSE;
2474 } while (PL_doextract);
2475 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2476 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2477 SV *sv = NEWSV(85,0);
2479 sv_upgrade(sv, SVt_PVMG);
2480 sv_setsv(sv,PL_linestr);
2481 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2483 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2484 if (CopLINE(PL_curcop) == 1) {
2485 while (s < PL_bufend && isSPACE(*s))
2487 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2491 if (*s == '#' && *(s+1) == '!')
2493 #ifdef ALTERNATE_SHEBANG
2495 static char as[] = ALTERNATE_SHEBANG;
2496 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2497 d = s + (sizeof(as) - 1);
2499 #endif /* ALTERNATE_SHEBANG */
2508 while (*d && !isSPACE(*d))
2512 #ifdef ARG_ZERO_IS_SCRIPT
2513 if (ipathend > ipath) {
2515 * HP-UX (at least) sets argv[0] to the script name,
2516 * which makes $^X incorrect. And Digital UNIX and Linux,
2517 * at least, set argv[0] to the basename of the Perl
2518 * interpreter. So, having found "#!", we'll set it right.
2520 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2521 assert(SvPOK(x) || SvGMAGICAL(x));
2522 if (sv_eq(x, CopFILESV(PL_curcop))) {
2523 sv_setpvn(x, ipath, ipathend - ipath);
2526 TAINT_NOT; /* $^X is always tainted, but that's OK */
2528 #endif /* ARG_ZERO_IS_SCRIPT */
2533 d = instr(s,"perl -");
2535 d = instr(s,"perl");
2537 /* avoid getting into infinite loops when shebang
2538 * line contains "Perl" rather than "perl" */
2540 for (d = ipathend-4; d >= ipath; --d) {
2541 if ((*d == 'p' || *d == 'P')
2542 && !ibcmp(d, "perl", 4))
2552 #ifdef ALTERNATE_SHEBANG
2554 * If the ALTERNATE_SHEBANG on this system starts with a
2555 * character that can be part of a Perl expression, then if
2556 * we see it but not "perl", we're probably looking at the
2557 * start of Perl code, not a request to hand off to some
2558 * other interpreter. Similarly, if "perl" is there, but
2559 * not in the first 'word' of the line, we assume the line
2560 * contains the start of the Perl program.
2562 if (d && *s != '#') {
2564 while (*c && !strchr("; \t\r\n\f\v#", *c))
2567 d = Nullch; /* "perl" not in first word; ignore */
2569 *s = '#'; /* Don't try to parse shebang line */
2571 #endif /* ALTERNATE_SHEBANG */
2576 !instr(s,"indir") &&
2577 instr(PL_origargv[0],"perl"))
2583 while (s < PL_bufend && isSPACE(*s))
2585 if (s < PL_bufend) {
2586 Newz(899,newargv,PL_origargc+3,char*);
2588 while (s < PL_bufend && !isSPACE(*s))
2591 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2594 newargv = PL_origargv;
2596 PerlProc_execv(ipath, newargv);
2597 Perl_croak(aTHX_ "Can't exec %s", ipath);
2600 U32 oldpdb = PL_perldb;
2601 bool oldn = PL_minus_n;
2602 bool oldp = PL_minus_p;
2604 while (*d && !isSPACE(*d)) d++;
2605 while (*d == ' ' || *d == '\t') d++;
2609 if (*d == 'M' || *d == 'm') {
2611 while (*d && !isSPACE(*d)) d++;
2612 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2615 d = moreswitches(d);
2617 if ((PERLDB_LINE && !oldpdb) ||
2618 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2619 /* if we have already added "LINE: while (<>) {",
2620 we must not do it again */
2622 sv_setpv(PL_linestr, "");
2623 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2624 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2625 PL_preambled = FALSE;
2627 (void)gv_fetchfile(PL_origfilename);
2634 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2636 PL_lex_state = LEX_FORMLINE;
2641 #ifdef PERL_STRICT_CR
2642 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2644 "(Maybe you didn't strip carriage returns after a network transfer?)\n");
2646 case ' ': case '\t': case '\f': case 013:
2651 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2653 while (s < d && *s != '\n')
2658 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2660 PL_lex_state = LEX_FORMLINE;
2670 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2675 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2678 if (strnEQ(s,"=>",2)) {
2679 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2680 OPERATOR('-'); /* unary minus */
2682 PL_last_uni = PL_oldbufptr;
2683 PL_last_lop_op = OP_FTEREAD; /* good enough */
2685 case 'r': FTST(OP_FTEREAD);
2686 case 'w': FTST(OP_FTEWRITE);
2687 case 'x': FTST(OP_FTEEXEC);
2688 case 'o': FTST(OP_FTEOWNED);
2689 case 'R': FTST(OP_FTRREAD);
2690 case 'W': FTST(OP_FTRWRITE);
2691 case 'X': FTST(OP_FTREXEC);
2692 case 'O': FTST(OP_FTROWNED);
2693 case 'e': FTST(OP_FTIS);
2694 case 'z': FTST(OP_FTZERO);
2695 case 's': FTST(OP_FTSIZE);
2696 case 'f': FTST(OP_FTFILE);
2697 case 'd': FTST(OP_FTDIR);
2698 case 'l': FTST(OP_FTLINK);
2699 case 'p': FTST(OP_FTPIPE);
2700 case 'S': FTST(OP_FTSOCK);
2701 case 'u': FTST(OP_FTSUID);
2702 case 'g': FTST(OP_FTSGID);
2703 case 'k': FTST(OP_FTSVTX);
2704 case 'b': FTST(OP_FTBLK);
2705 case 'c': FTST(OP_FTCHR);
2706 case 't': FTST(OP_FTTTY);
2707 case 'T': FTST(OP_FTTEXT);
2708 case 'B': FTST(OP_FTBINARY);
2709 case 'M': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTMTIME);
2710 case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME);
2711 case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME);
2713 Perl_croak(aTHX_ "Unrecognized file test: -%c", (int)tmp);
2720 if (PL_expect == XOPERATOR)
2725 else if (*s == '>') {
2728 if (isIDFIRST_lazy_if(s,UTF)) {
2729 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2737 if (PL_expect == XOPERATOR)
2740 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2742 OPERATOR('-'); /* unary minus */
2749 if (PL_expect == XOPERATOR)
2754 if (PL_expect == XOPERATOR)
2757 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2763 if (PL_expect != XOPERATOR) {
2764 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2765 PL_expect = XOPERATOR;
2766 force_ident(PL_tokenbuf, '*');
2779 if (PL_expect == XOPERATOR) {
2783 PL_tokenbuf[0] = '%';
2784 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2785 if (!PL_tokenbuf[1]) {
2787 yyerror("Final % should be \\% or %name");
2790 PL_pending_ident = '%';
2809 switch (PL_expect) {
2812 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2814 PL_bufptr = s; /* update in case we back off */
2820 PL_expect = XTERMBLOCK;
2824 while (isIDFIRST_lazy_if(s,UTF)) {
2825 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2826 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2827 if (tmp < 0) tmp = -tmp;
2842 d = scan_str(d,TRUE,TRUE);
2845 SvREFCNT_dec(PL_lex_stuff);
2846 PL_lex_stuff = Nullsv;
2848 /* MUST advance bufptr here to avoid bogus
2849 "at end of line" context messages from yyerror().
2851 PL_bufptr = s + len;
2852 yyerror("Unterminated attribute parameter in attribute list");
2855 return 0; /* EOF indicator */
2859 SV *sv = newSVpvn(s, len);
2860 sv_catsv(sv, PL_lex_stuff);
2861 attrs = append_elem(OP_LIST, attrs,
2862 newSVOP(OP_CONST, 0, sv));
2863 SvREFCNT_dec(PL_lex_stuff);
2864 PL_lex_stuff = Nullsv;
2867 attrs = append_elem(OP_LIST, attrs,
2868 newSVOP(OP_CONST, 0,
2872 if (*s == ':' && s[1] != ':')
2875 break; /* require real whitespace or :'s */
2877 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
2878 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
2879 char q = ((*s == '\'') ? '"' : '\'');
2880 /* If here for an expression, and parsed no attrs, back off. */
2881 if (tmp == '=' && !attrs) {
2885 /* MUST advance bufptr here to avoid bogus "at end of line"
2886 context messages from yyerror().
2890 yyerror("Unterminated attribute list");
2892 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
2900 PL_nextval[PL_nexttoke].opval = attrs;
2908 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
2909 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
2914 if (CopLINE(PL_curcop) < PL_copline)
2915 PL_copline = CopLINE(PL_curcop);
2926 if (PL_lex_brackets <= 0)
2927 yyerror("Unmatched right square bracket");
2930 if (PL_lex_state == LEX_INTERPNORMAL) {
2931 if (PL_lex_brackets == 0) {
2932 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
2933 PL_lex_state = LEX_INTERPEND;
2940 if (PL_lex_brackets > 100) {
2941 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
2942 if (newlb != PL_lex_brackstack) {
2944 PL_lex_brackstack = newlb;
2947 switch (PL_expect) {
2949 if (PL_lex_formbrack) {
2953 if (PL_oldoldbufptr == PL_last_lop)
2954 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2956 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2957 OPERATOR(HASHBRACK);
2959 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2962 PL_tokenbuf[0] = '\0';
2963 if (d < PL_bufend && *d == '-') {
2964 PL_tokenbuf[0] = '-';
2966 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2969 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
2970 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
2972 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2975 char minus = (PL_tokenbuf[0] == '-');
2976 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
2984 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
2989 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2994 if (PL_oldoldbufptr == PL_last_lop)
2995 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2997 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3000 OPERATOR(HASHBRACK);
3001 /* This hack serves to disambiguate a pair of curlies
3002 * as being a block or an anon hash. Normally, expectation
3003 * determines that, but in cases where we're not in a
3004 * position to expect anything in particular (like inside
3005 * eval"") we have to resolve the ambiguity. This code
3006 * covers the case where the first term in the curlies is a
3007 * quoted string. Most other cases need to be explicitly
3008 * disambiguated by prepending a `+' before the opening
3009 * curly in order to force resolution as an anon hash.
3011 * XXX should probably propagate the outer expectation
3012 * into eval"" to rely less on this hack, but that could
3013 * potentially break current behavior of eval"".
3017 if (*s == '\'' || *s == '"' || *s == '`') {
3018 /* common case: get past first string, handling escapes */
3019 for (t++; t < PL_bufend && *t != *s;)
3020 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3024 else if (*s == 'q') {
3027 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3031 char open, close, term;
3034 while (t < PL_bufend && isSPACE(*t))
3038 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3042 for (t++; t < PL_bufend; t++) {
3043 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3045 else if (*t == open)
3049 for (t++; t < PL_bufend; t++) {
3050 if (*t == '\\' && t+1 < PL_bufend)
3052 else if (*t == close && --brackets <= 0)
3054 else if (*t == open)
3060 else if (isALNUM_lazy_if(t,UTF)) {
3062 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3065 while (t < PL_bufend && isSPACE(*t))
3067 /* if comma follows first term, call it an anon hash */
3068 /* XXX it could be a comma expression with loop modifiers */
3069 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3070 || (*t == '=' && t[1] == '>')))
3071 OPERATOR(HASHBRACK);
3072 if (PL_expect == XREF)
3075 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3081 yylval.ival = CopLINE(PL_curcop);
3082 if (isSPACE(*s) || *s == '#')
3083 PL_copline = NOLINE; /* invalidate current command line number */
3088 if (PL_lex_brackets <= 0)
3089 yyerror("Unmatched right curly bracket");
3091 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3092 if (PL_lex_brackets < PL_lex_formbrack)
3093 PL_lex_formbrack = 0;
3094 if (PL_lex_state == LEX_INTERPNORMAL) {
3095 if (PL_lex_brackets == 0) {
3096 if (PL_expect & XFAKEBRACK) {
3097 PL_expect &= XENUMMASK;
3098 PL_lex_state = LEX_INTERPEND;
3100 return yylex(); /* ignore fake brackets */
3102 if (*s == '-' && s[1] == '>')
3103 PL_lex_state = LEX_INTERPENDMAYBE;
3104 else if (*s != '[' && *s != '{')
3105 PL_lex_state = LEX_INTERPEND;
3108 if (PL_expect & XFAKEBRACK) {
3109 PL_expect &= XENUMMASK;
3111 return yylex(); /* ignore fake brackets */
3121 if (PL_expect == XOPERATOR) {
3122 if (ckWARN(WARN_SEMICOLON)
3123 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3125 CopLINE_dec(PL_curcop);
3126 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3127 CopLINE_inc(PL_curcop);
3132 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3134 PL_expect = XOPERATOR;
3135 force_ident(PL_tokenbuf, '&');
3139 yylval.ival = (OPpENTERSUB_AMPER<<8);
3158 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3159 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3161 if (PL_expect == XSTATE && isALPHA(tmp) &&
3162 (s == PL_linestart+1 || s[-2] == '\n') )
3164 if (PL_in_eval && !PL_rsfp) {
3169 if (strnEQ(s,"=cut",4)) {
3183 PL_doextract = TRUE;
3186 if (PL_lex_brackets < PL_lex_formbrack) {
3188 #ifdef PERL_STRICT_CR
3189 for (t = s; *t == ' ' || *t == '\t'; t++) ;
3191 for (t = s; *t == ' ' || *t == '\t' || *t == '\r'; t++) ;
3193 if (*t == '\n' || *t == '#') {
3211 if (PL_expect != XOPERATOR) {
3212 if (s[1] != '<' && !strchr(s,'>'))
3215 s = scan_heredoc(s);
3217 s = scan_inputsymbol(s);
3218 TERM(sublex_start());
3223 SHop(OP_LEFT_SHIFT);
3237 SHop(OP_RIGHT_SHIFT);
3246 if (PL_expect == XOPERATOR) {
3247 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3250 return ','; /* grandfather non-comma-format format */
3254 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3255 PL_tokenbuf[0] = '@';
3256 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3257 sizeof PL_tokenbuf - 1, FALSE);
3258 if (PL_expect == XOPERATOR)
3259 no_op("Array length", s);
3260 if (!PL_tokenbuf[1])
3262 PL_expect = XOPERATOR;
3263 PL_pending_ident = '#';
3267 PL_tokenbuf[0] = '$';
3268 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3269 sizeof PL_tokenbuf - 1, FALSE);
3270 if (PL_expect == XOPERATOR)
3272 if (!PL_tokenbuf[1]) {
3274 yyerror("Final $ should be \\$ or $name");
3278 /* This kludge not intended to be bulletproof. */
3279 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3280 yylval.opval = newSVOP(OP_CONST, 0,
3281 newSViv(PL_compiling.cop_arybase));
3282 yylval.opval->op_private = OPpCONST_ARYBASE;
3288 if (PL_lex_state == LEX_NORMAL)
3291 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3294 PL_tokenbuf[0] = '@';
3295 if (ckWARN(WARN_SYNTAX)) {
3297 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3300 PL_bufptr = skipspace(PL_bufptr);
3301 while (t < PL_bufend && *t != ']')
3303 Perl_warner(aTHX_ WARN_SYNTAX,
3304 "Multidimensional syntax %.*s not supported",
3305 (t - PL_bufptr) + 1, PL_bufptr);
3309 else if (*s == '{') {
3310 PL_tokenbuf[0] = '%';
3311 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3312 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3314 char tmpbuf[sizeof PL_tokenbuf];
3316 for (t++; isSPACE(*t); t++) ;
3317 if (isIDFIRST_lazy_if(t,UTF)) {
3318 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3319 for (; isSPACE(*t); t++) ;
3320 if (*t == ';' && get_cv(tmpbuf, FALSE))
3321 Perl_warner(aTHX_ WARN_SYNTAX,
3322 "You need to quote \"%s\"", tmpbuf);
3328 PL_expect = XOPERATOR;
3329 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3330 bool islop = (PL_last_lop == PL_oldoldbufptr);
3331 if (!islop || PL_last_lop_op == OP_GREPSTART)
3332 PL_expect = XOPERATOR;
3333 else if (strchr("$@\"'`q", *s))
3334 PL_expect = XTERM; /* e.g. print $fh "foo" */
3335 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3336 PL_expect = XTERM; /* e.g. print $fh &sub */
3337 else if (isIDFIRST_lazy_if(s,UTF)) {
3338 char tmpbuf[sizeof PL_tokenbuf];
3339 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3340 if ((tmp = keyword(tmpbuf, len))) {
3341 /* binary operators exclude handle interpretations */
3353 PL_expect = XTERM; /* e.g. print $fh length() */
3358 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3359 if (gv && GvCVu(gv))
3360 PL_expect = XTERM; /* e.g. print $fh subr() */
3363 else if (isDIGIT(*s))
3364 PL_expect = XTERM; /* e.g. print $fh 3 */
3365 else if (*s == '.' && isDIGIT(s[1]))
3366 PL_expect = XTERM; /* e.g. print $fh .3 */
3367 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3368 PL_expect = XTERM; /* e.g. print $fh -1 */
3369 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3370 PL_expect = XTERM; /* print $fh <<"EOF" */
3372 PL_pending_ident = '$';
3376 if (PL_expect == XOPERATOR)
3378 PL_tokenbuf[0] = '@';
3379 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3380 if (!PL_tokenbuf[1]) {
3382 yyerror("Final @ should be \\@ or @name");
3385 if (PL_lex_state == LEX_NORMAL)
3387 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3389 PL_tokenbuf[0] = '%';
3391 /* Warn about @ where they meant $. */
3392 if (ckWARN(WARN_SYNTAX)) {
3393 if (*s == '[' || *s == '{') {
3395 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3397 if (*t == '}' || *t == ']') {
3399 PL_bufptr = skipspace(PL_bufptr);
3400 Perl_warner(aTHX_ WARN_SYNTAX,
3401 "Scalar value %.*s better written as $%.*s",
3402 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3407 PL_pending_ident = '@';
3410 case '/': /* may either be division or pattern */
3411 case '?': /* may either be conditional or pattern */
3412 if (PL_expect != XOPERATOR) {
3413 /* Disable warning on "study /blah/" */
3414 if (PL_oldoldbufptr == PL_last_uni
3415 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3416 || memNE(PL_last_uni, "study", 5)
3417 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3419 s = scan_pat(s,OP_MATCH);
3420 TERM(sublex_start());
3428 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3429 #ifdef PERL_STRICT_CR
3432 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3434 && (s == PL_linestart || s[-1] == '\n') )
3436 PL_lex_formbrack = 0;
3440 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3446 yylval.ival = OPf_SPECIAL;
3452 if (PL_expect != XOPERATOR)
3457 case '0': case '1': case '2': case '3': case '4':
3458 case '5': case '6': case '7': case '8': case '9':
3460 if (PL_expect == XOPERATOR)
3465 s = scan_str(s,FALSE,FALSE);
3466 if (PL_expect == XOPERATOR) {
3467 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3470 return ','; /* grandfather non-comma-format format */
3476 missingterm((char*)0);
3477 yylval.ival = OP_CONST;
3478 TERM(sublex_start());
3481 s = scan_str(s,FALSE,FALSE);
3482 if (PL_expect == XOPERATOR) {
3483 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3486 return ','; /* grandfather non-comma-format format */
3492 missingterm((char*)0);
3493 yylval.ival = OP_CONST;
3494 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3495 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3496 yylval.ival = OP_STRINGIFY;
3500 TERM(sublex_start());
3503 s = scan_str(s,FALSE,FALSE);
3504 if (PL_expect == XOPERATOR)
3505 no_op("Backticks",s);
3507 missingterm((char*)0);
3508 yylval.ival = OP_BACKTICK;
3510 TERM(sublex_start());
3514 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3515 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3517 if (PL_expect == XOPERATOR)
3518 no_op("Backslash",s);
3522 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3526 while (isDIGIT(*start) || *start == '_')
3528 if (*start == '.' && isDIGIT(start[1])) {
3532 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3533 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF)) {
3537 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3547 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 if (!(tmp = keyword(PL_tokenbuf, len)))
3964 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
3978 LOP(OP_ACCEPT,XTERM);
3984 LOP(OP_ATAN2,XTERM);
3990 LOP(OP_BINMODE,XTERM);
3993 LOP(OP_BLESS,XTERM);
4002 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4019 if (!PL_cryptseen) {
4020 PL_cryptseen = TRUE;
4024 LOP(OP_CRYPT,XTERM);
4027 if (ckWARN(WARN_CHMOD)) {
4028 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4029 if (*d != '0' && isDIGIT(*d))
4030 Perl_warner(aTHX_ WARN_CHMOD,
4031 "chmod() mode argument is missing initial 0");
4033 LOP(OP_CHMOD,XTERM);
4036 LOP(OP_CHOWN,XTERM);
4039 LOP(OP_CONNECT,XTERM);
4055 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4059 PL_hints |= HINT_BLOCK_SCOPE;
4069 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4070 LOP(OP_DBMOPEN,XTERM);
4076 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4083 yylval.ival = CopLINE(PL_curcop);
4097 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4098 UNIBRACK(OP_ENTEREVAL);
4113 case KEY_endhostent:
4119 case KEY_endservent:
4122 case KEY_endprotoent:
4133 yylval.ival = CopLINE(PL_curcop);
4135 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4137 if ((PL_bufend - p) >= 3 &&
4138 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4140 else if ((PL_bufend - p) >= 4 &&
4141 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4144 if (isIDFIRST_lazy_if(p,UTF)) {
4145 p = scan_ident(p, PL_bufend,
4146 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4150 Perl_croak(aTHX_ "Missing $ on loop variable");
4155 LOP(OP_FORMLINE,XTERM);
4161 LOP(OP_FCNTL,XTERM);
4167 LOP(OP_FLOCK,XTERM);
4176 LOP(OP_GREPSTART, XREF);
4179 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4194 case KEY_getpriority:
4195 LOP(OP_GETPRIORITY,XTERM);
4197 case KEY_getprotobyname:
4200 case KEY_getprotobynumber:
4201 LOP(OP_GPBYNUMBER,XTERM);
4203 case KEY_getprotoent:
4215 case KEY_getpeername:
4216 UNI(OP_GETPEERNAME);
4218 case KEY_gethostbyname:
4221 case KEY_gethostbyaddr:
4222 LOP(OP_GHBYADDR,XTERM);
4224 case KEY_gethostent:
4227 case KEY_getnetbyname:
4230 case KEY_getnetbyaddr:
4231 LOP(OP_GNBYADDR,XTERM);
4236 case KEY_getservbyname:
4237 LOP(OP_GSBYNAME,XTERM);
4239 case KEY_getservbyport:
4240 LOP(OP_GSBYPORT,XTERM);
4242 case KEY_getservent:
4245 case KEY_getsockname:
4246 UNI(OP_GETSOCKNAME);
4248 case KEY_getsockopt:
4249 LOP(OP_GSOCKOPT,XTERM);
4271 yylval.ival = CopLINE(PL_curcop);
4275 LOP(OP_INDEX,XTERM);
4281 LOP(OP_IOCTL,XTERM);
4293 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4325 LOP(OP_LISTEN,XTERM);
4334 s = scan_pat(s,OP_MATCH);
4335 TERM(sublex_start());
4338 LOP(OP_MAPSTART, XREF);
4341 LOP(OP_MKDIR,XTERM);
4344 LOP(OP_MSGCTL,XTERM);
4347 LOP(OP_MSGGET,XTERM);
4350 LOP(OP_MSGRCV,XTERM);
4353 LOP(OP_MSGSND,XTERM);
4359 if (isIDFIRST_lazy_if(s,UTF)) {
4360 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4361 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4363 PL_in_my_stash = gv_stashpv(PL_tokenbuf, FALSE);
4364 if (!PL_in_my_stash) {
4367 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4375 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4382 if (PL_expect != XSTATE)
4383 yyerror("\"no\" not allowed in expression");
4384 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4385 s = force_version(s);
4390 if (*s == '(' || (s = skipspace(s), *s == '('))
4397 if (isIDFIRST_lazy_if(s,UTF)) {
4399 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4401 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4402 Perl_warner(aTHX_ WARN_PRECEDENCE,
4403 "Precedence problem: open %.*s should be open(%.*s)",
4409 yylval.ival = OP_OR;
4419 LOP(OP_OPEN_DIR,XTERM);
4422 checkcomma(s,PL_tokenbuf,"filehandle");
4426 checkcomma(s,PL_tokenbuf,"filehandle");
4445 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4449 LOP(OP_PIPE_OP,XTERM);
4452 s = scan_str(s,FALSE,FALSE);
4454 missingterm((char*)0);
4455 yylval.ival = OP_CONST;
4456 TERM(sublex_start());
4462 s = scan_str(s,FALSE,FALSE);
4464 missingterm((char*)0);
4466 if (SvCUR(PL_lex_stuff)) {
4469 d = SvPV_force(PL_lex_stuff, len);
4471 for (; isSPACE(*d) && len; --len, ++d) ;
4474 if (!warned && ckWARN(WARN_QW)) {
4475 for (; !isSPACE(*d) && len; --len, ++d) {
4477 Perl_warner(aTHX_ WARN_QW,
4478 "Possible attempt to separate words with commas");
4481 else if (*d == '#') {
4482 Perl_warner(aTHX_ WARN_QW,
4483 "Possible attempt to put comments in qw() list");
4489 for (; !isSPACE(*d) && len; --len, ++d) ;
4491 words = append_elem(OP_LIST, words,
4492 newSVOP(OP_CONST, 0, newSVpvn(b, d-b)));
4496 PL_nextval[PL_nexttoke].opval = words;
4501 SvREFCNT_dec(PL_lex_stuff);
4502 PL_lex_stuff = Nullsv;
4507 s = scan_str(s,FALSE,FALSE);
4509 missingterm((char*)0);
4510 yylval.ival = OP_STRINGIFY;
4511 if (SvIVX(PL_lex_stuff) == '\'')
4512 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4513 TERM(sublex_start());
4516 s = scan_pat(s,OP_QR);
4517 TERM(sublex_start());
4520 s = scan_str(s,FALSE,FALSE);
4522 missingterm((char*)0);
4523 yylval.ival = OP_BACKTICK;
4525 TERM(sublex_start());
4532 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4533 s = force_version(s);
4536 *PL_tokenbuf = '\0';
4537 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4538 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4539 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4541 yyerror("<> should be quotes");
4549 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4553 LOP(OP_RENAME,XTERM);
4562 LOP(OP_RINDEX,XTERM);
4585 LOP(OP_REVERSE,XTERM);
4596 TERM(sublex_start());
4598 TOKEN(1); /* force error */
4607 LOP(OP_SELECT,XTERM);
4613 LOP(OP_SEMCTL,XTERM);
4616 LOP(OP_SEMGET,XTERM);
4619 LOP(OP_SEMOP,XTERM);
4625 LOP(OP_SETPGRP,XTERM);
4627 case KEY_setpriority:
4628 LOP(OP_SETPRIORITY,XTERM);
4630 case KEY_sethostent:
4636 case KEY_setservent:
4639 case KEY_setprotoent:
4649 LOP(OP_SEEKDIR,XTERM);
4651 case KEY_setsockopt:
4652 LOP(OP_SSOCKOPT,XTERM);
4658 LOP(OP_SHMCTL,XTERM);
4661 LOP(OP_SHMGET,XTERM);
4664 LOP(OP_SHMREAD,XTERM);
4667 LOP(OP_SHMWRITE,XTERM);
4670 LOP(OP_SHUTDOWN,XTERM);
4679 LOP(OP_SOCKET,XTERM);
4681 case KEY_socketpair:
4682 LOP(OP_SOCKPAIR,XTERM);
4685 checkcomma(s,PL_tokenbuf,"subroutine name");
4687 if (*s == ';' || *s == ')') /* probably a close */
4688 Perl_croak(aTHX_ "sort is now a reserved word");
4690 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4694 LOP(OP_SPLIT,XTERM);
4697 LOP(OP_SPRINTF,XTERM);
4700 LOP(OP_SPLICE,XTERM);
4715 LOP(OP_SUBSTR,XTERM);
4721 char tmpbuf[sizeof PL_tokenbuf];
4723 expectation attrful;
4724 bool have_name, have_proto;
4729 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4730 (*s == ':' && s[1] == ':'))
4733 attrful = XATTRBLOCK;
4734 /* remember buffer pos'n for later force_word */
4735 tboffset = s - PL_oldbufptr;
4736 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4737 if (strchr(tmpbuf, ':'))
4738 sv_setpv(PL_subname, tmpbuf);
4740 sv_setsv(PL_subname,PL_curstname);
4741 sv_catpvn(PL_subname,"::",2);
4742 sv_catpvn(PL_subname,tmpbuf,len);
4749 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4750 PL_expect = XTERMBLOCK;
4751 attrful = XATTRTERM;
4752 sv_setpv(PL_subname,"?");
4756 if (key == KEY_format) {
4758 PL_lex_formbrack = PL_lex_brackets + 1;
4760 (void) force_word(PL_oldbufptr + tboffset, WORD,
4765 /* Look for a prototype */
4769 s = scan_str(s,FALSE,FALSE);
4772 SvREFCNT_dec(PL_lex_stuff);
4773 PL_lex_stuff = Nullsv;
4774 Perl_croak(aTHX_ "Prototype not terminated");
4777 d = SvPVX(PL_lex_stuff);
4779 for (p = d; *p; ++p) {
4784 SvCUR(PL_lex_stuff) = tmp;
4792 if (*s == ':' && s[1] != ':')
4793 PL_expect = attrful;
4796 PL_nextval[PL_nexttoke].opval =
4797 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4798 PL_lex_stuff = Nullsv;
4802 sv_setpv(PL_subname,"__ANON__");
4805 (void) force_word(PL_oldbufptr + tboffset, WORD,
4814 LOP(OP_SYSTEM,XREF);
4817 LOP(OP_SYMLINK,XTERM);
4820 LOP(OP_SYSCALL,XTERM);
4823 LOP(OP_SYSOPEN,XTERM);
4826 LOP(OP_SYSSEEK,XTERM);
4829 LOP(OP_SYSREAD,XTERM);
4832 LOP(OP_SYSWRITE,XTERM);
4836 TERM(sublex_start());
4857 LOP(OP_TRUNCATE,XTERM);
4869 yylval.ival = CopLINE(PL_curcop);
4873 yylval.ival = CopLINE(PL_curcop);
4877 LOP(OP_UNLINK,XTERM);
4883 LOP(OP_UNPACK,XTERM);
4886 LOP(OP_UTIME,XTERM);
4889 if (ckWARN(WARN_UMASK)) {
4890 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4891 if (*d != '0' && isDIGIT(*d))
4892 Perl_warner(aTHX_ WARN_UMASK,
4893 "umask: argument is missing initial 0");
4898 LOP(OP_UNSHIFT,XTERM);
4901 if (PL_expect != XSTATE)
4902 yyerror("\"use\" not allowed in expression");
4904 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4905 s = force_version(s);
4906 if (*s == ';' || (s = skipspace(s), *s == ';')) {
4907 PL_nextval[PL_nexttoke].opval = Nullop;
4912 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4913 s = force_version(s);
4925 yylval.ival = CopLINE(PL_curcop);
4929 PL_hints |= HINT_BLOCK_SCOPE;
4936 LOP(OP_WAITPID,XTERM);
4944 static char ctl_l[2];
4946 if (ctl_l[0] == '\0')
4947 ctl_l[0] = toCTRL('L');
4948 gv_fetchpv(ctl_l,TRUE, SVt_PV);
4951 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
4956 if (PL_expect == XOPERATOR)
4962 yylval.ival = OP_XOR;
4967 TERM(sublex_start());
4973 Perl_keyword(pTHX_ register char *d, I32 len)
4978 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
4979 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
4980 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
4981 if (strEQ(d,"__DATA__")) return KEY___DATA__;
4982 if (strEQ(d,"__END__")) return KEY___END__;
4986 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
4991 if (strEQ(d,"and")) return -KEY_and;
4992 if (strEQ(d,"abs")) return -KEY_abs;
4995 if (strEQ(d,"alarm")) return -KEY_alarm;
4996 if (strEQ(d,"atan2")) return -KEY_atan2;
4999 if (strEQ(d,"accept")) return -KEY_accept;
5004 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5007 if (strEQ(d,"bless")) return -KEY_bless;
5008 if (strEQ(d,"bind")) return -KEY_bind;
5009 if (strEQ(d,"binmode")) return -KEY_binmode;
5012 if (strEQ(d,"CORE")) return -KEY_CORE;
5013 if (strEQ(d,"CHECK")) return KEY_CHECK;
5018 if (strEQ(d,"cmp")) return -KEY_cmp;
5019 if (strEQ(d,"chr")) return -KEY_chr;
5020 if (strEQ(d,"cos")) return -KEY_cos;
5023 if (strEQ(d,"chop")) return KEY_chop;
5026 if (strEQ(d,"close")) return -KEY_close;
5027 if (strEQ(d,"chdir")) return -KEY_chdir;
5028 if (strEQ(d,"chomp")) return KEY_chomp;
5029 if (strEQ(d,"chmod")) return -KEY_chmod;
5030 if (strEQ(d,"chown")) return -KEY_chown;
5031 if (strEQ(d,"crypt")) return -KEY_crypt;
5034 if (strEQ(d,"chroot")) return -KEY_chroot;
5035 if (strEQ(d,"caller")) return -KEY_caller;
5038 if (strEQ(d,"connect")) return -KEY_connect;
5041 if (strEQ(d,"closedir")) return -KEY_closedir;
5042 if (strEQ(d,"continue")) return -KEY_continue;
5047 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5052 if (strEQ(d,"do")) return KEY_do;
5055 if (strEQ(d,"die")) return -KEY_die;
5058 if (strEQ(d,"dump")) return -KEY_dump;
5061 if (strEQ(d,"delete")) return KEY_delete;
5064 if (strEQ(d,"defined")) return KEY_defined;
5065 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5068 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5073 if (strEQ(d,"EQ")) { deprecate(d); return -KEY_eq;}
5074 if (strEQ(d,"END")) return KEY_END;
5079 if (strEQ(d,"eq")) return -KEY_eq;
5082 if (strEQ(d,"eof")) return -KEY_eof;
5083 if (strEQ(d,"exp")) return -KEY_exp;
5086 if (strEQ(d,"else")) return KEY_else;
5087 if (strEQ(d,"exit")) return -KEY_exit;
5088 if (strEQ(d,"eval")) return KEY_eval;
5089 if (strEQ(d,"exec")) return -KEY_exec;
5090 if (strEQ(d,"each")) return KEY_each;
5093 if (strEQ(d,"elsif")) return KEY_elsif;
5096 if (strEQ(d,"exists")) return KEY_exists;
5097 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5100 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5101 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5104 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5107 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5108 if (strEQ(d,"endservent")) return -KEY_endservent;
5111 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5118 if (strEQ(d,"for")) return KEY_for;
5121 if (strEQ(d,"fork")) return -KEY_fork;
5124 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5125 if (strEQ(d,"flock")) return -KEY_flock;
5128 if (strEQ(d,"format")) return KEY_format;
5129 if (strEQ(d,"fileno")) return -KEY_fileno;
5132 if (strEQ(d,"foreach")) return KEY_foreach;
5135 if (strEQ(d,"formline")) return -KEY_formline;
5141 if (strEQ(d,"GT")) { deprecate(d); return -KEY_gt;}
5142 if (strEQ(d,"GE")) { deprecate(d); return -KEY_ge;}
5146 if (strnEQ(d,"get",3)) {
5151 if (strEQ(d,"ppid")) return -KEY_getppid;
5152 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5155 if (strEQ(d,"pwent")) return -KEY_getpwent;
5156 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5157 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5160 if (strEQ(d,"peername")) return -KEY_getpeername;
5161 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5162 if (strEQ(d,"priority")) return -KEY_getpriority;
5165 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5168 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5172 else if (*d == 'h') {
5173 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5174 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5175 if (strEQ(d,"hostent")) return -KEY_gethostent;
5177 else if (*d == 'n') {
5178 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5179 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5180 if (strEQ(d,"netent")) return -KEY_getnetent;
5182 else if (*d == 's') {
5183 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5184 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5185 if (strEQ(d,"servent")) return -KEY_getservent;
5186 if (strEQ(d,"sockname")) return -KEY_getsockname;
5187 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5189 else if (*d == 'g') {
5190 if (strEQ(d,"grent")) return -KEY_getgrent;
5191 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5192 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5194 else if (*d == 'l') {
5195 if (strEQ(d,"login")) return -KEY_getlogin;
5197 else if (strEQ(d,"c")) return -KEY_getc;
5202 if (strEQ(d,"gt")) return -KEY_gt;
5203 if (strEQ(d,"ge")) return -KEY_ge;
5206 if (strEQ(d,"grep")) return KEY_grep;
5207 if (strEQ(d,"goto")) return KEY_goto;
5208 if (strEQ(d,"glob")) return KEY_glob;
5211 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5216 if (strEQ(d,"hex")) return -KEY_hex;
5219 if (strEQ(d,"INIT")) return KEY_INIT;
5224 if (strEQ(d,"if")) return KEY_if;
5227 if (strEQ(d,"int")) return -KEY_int;
5230 if (strEQ(d,"index")) return -KEY_index;
5231 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5236 if (strEQ(d,"join")) return -KEY_join;
5240 if (strEQ(d,"keys")) return KEY_keys;
5241 if (strEQ(d,"kill")) return -KEY_kill;
5246 if (strEQ(d,"LT")) { deprecate(d); return -KEY_lt;}
5247 if (strEQ(d,"LE")) { deprecate(d); return -KEY_le;}
5253 if (strEQ(d,"lt")) return -KEY_lt;
5254 if (strEQ(d,"le")) return -KEY_le;
5255 if (strEQ(d,"lc")) return -KEY_lc;
5258 if (strEQ(d,"log")) return -KEY_log;
5261 if (strEQ(d,"last")) return KEY_last;
5262 if (strEQ(d,"link")) return -KEY_link;
5263 if (strEQ(d,"lock")) return -KEY_lock;
5266 if (strEQ(d,"local")) return KEY_local;
5267 if (strEQ(d,"lstat")) return -KEY_lstat;
5270 if (strEQ(d,"length")) return -KEY_length;
5271 if (strEQ(d,"listen")) return -KEY_listen;
5274 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5277 if (strEQ(d,"localtime")) return -KEY_localtime;
5283 case 1: return KEY_m;
5285 if (strEQ(d,"my")) return KEY_my;
5288 if (strEQ(d,"map")) return KEY_map;
5291 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5294 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5295 if (strEQ(d,"msgget")) return -KEY_msgget;
5296 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5297 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5302 if (strEQ(d,"NE")) { deprecate(d); return -KEY_ne;}
5305 if (strEQ(d,"next")) return KEY_next;
5306 if (strEQ(d,"ne")) return -KEY_ne;
5307 if (strEQ(d,"not")) return -KEY_not;
5308 if (strEQ(d,"no")) return KEY_no;
5313 if (strEQ(d,"or")) return -KEY_or;
5316 if (strEQ(d,"ord")) return -KEY_ord;
5317 if (strEQ(d,"oct")) return -KEY_oct;
5318 if (strEQ(d,"our")) return KEY_our;
5321 if (strEQ(d,"open")) return -KEY_open;
5324 if (strEQ(d,"opendir")) return -KEY_opendir;
5331 if (strEQ(d,"pop")) return KEY_pop;
5332 if (strEQ(d,"pos")) return KEY_pos;
5335 if (strEQ(d,"push")) return KEY_push;
5336 if (strEQ(d,"pack")) return -KEY_pack;
5337 if (strEQ(d,"pipe")) return -KEY_pipe;
5340 if (strEQ(d,"print")) return KEY_print;
5343 if (strEQ(d,"printf")) return KEY_printf;
5346 if (strEQ(d,"package")) return KEY_package;
5349 if (strEQ(d,"prototype")) return KEY_prototype;
5354 if (strEQ(d,"q")) return KEY_q;
5355 if (strEQ(d,"qr")) return KEY_qr;
5356 if (strEQ(d,"qq")) return KEY_qq;
5357 if (strEQ(d,"qw")) return KEY_qw;
5358 if (strEQ(d,"qx")) return KEY_qx;
5360 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5365 if (strEQ(d,"ref")) return -KEY_ref;
5368 if (strEQ(d,"read")) return -KEY_read;
5369 if (strEQ(d,"rand")) return -KEY_rand;
5370 if (strEQ(d,"recv")) return -KEY_recv;
5371 if (strEQ(d,"redo")) return KEY_redo;
5374 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5375 if (strEQ(d,"reset")) return -KEY_reset;
5378 if (strEQ(d,"return")) return KEY_return;
5379 if (strEQ(d,"rename")) return -KEY_rename;
5380 if (strEQ(d,"rindex")) return -KEY_rindex;
5383 if (strEQ(d,"require")) return -KEY_require;
5384 if (strEQ(d,"reverse")) return -KEY_reverse;
5385 if (strEQ(d,"readdir")) return -KEY_readdir;
5388 if (strEQ(d,"readlink")) return -KEY_readlink;
5389 if (strEQ(d,"readline")) return -KEY_readline;
5390 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5393 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5399 case 0: return KEY_s;
5401 if (strEQ(d,"scalar")) return KEY_scalar;
5406 if (strEQ(d,"seek")) return -KEY_seek;
5407 if (strEQ(d,"send")) return -KEY_send;
5410 if (strEQ(d,"semop")) return -KEY_semop;
5413 if (strEQ(d,"select")) return -KEY_select;
5414 if (strEQ(d,"semctl")) return -KEY_semctl;
5415 if (strEQ(d,"semget")) return -KEY_semget;
5418 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5419 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5422 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5423 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5426 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5429 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5430 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5431 if (strEQ(d,"setservent")) return -KEY_setservent;
5434 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5435 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5442 if (strEQ(d,"shift")) return KEY_shift;
5445 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5446 if (strEQ(d,"shmget")) return -KEY_shmget;
5449 if (strEQ(d,"shmread")) return -KEY_shmread;
5452 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5453 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5458 if (strEQ(d,"sin")) return -KEY_sin;
5461 if (strEQ(d,"sleep")) return -KEY_sleep;
5464 if (strEQ(d,"sort")) return KEY_sort;
5465 if (strEQ(d,"socket")) return -KEY_socket;
5466 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5469 if (strEQ(d,"split")) return KEY_split;
5470 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5471 if (strEQ(d,"splice")) return KEY_splice;
5474 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5477 if (strEQ(d,"srand")) return -KEY_srand;
5480 if (strEQ(d,"stat")) return -KEY_stat;
5481 if (strEQ(d,"study")) return KEY_study;
5484 if (strEQ(d,"substr")) return -KEY_substr;
5485 if (strEQ(d,"sub")) return KEY_sub;
5490 if (strEQ(d,"system")) return -KEY_system;
5493 if (strEQ(d,"symlink")) return -KEY_symlink;
5494 if (strEQ(d,"syscall")) return -KEY_syscall;
5495 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5496 if (strEQ(d,"sysread")) return -KEY_sysread;
5497 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5500 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5509 if (strEQ(d,"tr")) return KEY_tr;
5512 if (strEQ(d,"tie")) return KEY_tie;
5515 if (strEQ(d,"tell")) return -KEY_tell;
5516 if (strEQ(d,"tied")) return KEY_tied;
5517 if (strEQ(d,"time")) return -KEY_time;
5520 if (strEQ(d,"times")) return -KEY_times;
5523 if (strEQ(d,"telldir")) return -KEY_telldir;
5526 if (strEQ(d,"truncate")) return -KEY_truncate;
5533 if (strEQ(d,"uc")) return -KEY_uc;
5536 if (strEQ(d,"use")) return KEY_use;
5539 if (strEQ(d,"undef")) return KEY_undef;
5540 if (strEQ(d,"until")) return KEY_until;
5541 if (strEQ(d,"untie")) return KEY_untie;
5542 if (strEQ(d,"utime")) return -KEY_utime;
5543 if (strEQ(d,"umask")) return -KEY_umask;
5546 if (strEQ(d,"unless")) return KEY_unless;
5547 if (strEQ(d,"unpack")) return -KEY_unpack;
5548 if (strEQ(d,"unlink")) return -KEY_unlink;
5551 if (strEQ(d,"unshift")) return KEY_unshift;
5552 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5557 if (strEQ(d,"values")) return -KEY_values;
5558 if (strEQ(d,"vec")) return -KEY_vec;
5563 if (strEQ(d,"warn")) return -KEY_warn;
5564 if (strEQ(d,"wait")) return -KEY_wait;
5567 if (strEQ(d,"while")) return KEY_while;
5568 if (strEQ(d,"write")) return -KEY_write;
5571 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5574 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5579 if (len == 1) return -KEY_x;
5580 if (strEQ(d,"xor")) return -KEY_xor;
5583 if (len == 1) return KEY_y;
5592 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5596 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5597 dTHR; /* only for ckWARN */
5598 if (ckWARN(WARN_SYNTAX)) {
5600 for (w = s+2; *w && level; w++) {
5607 for (; *w && isSPACE(*w); w++) ;
5608 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5609 Perl_warner(aTHX_ WARN_SYNTAX,
5610 "%s (...) interpreted as function",name);
5613 while (s < PL_bufend && isSPACE(*s))
5617 while (s < PL_bufend && isSPACE(*s))
5619 if (isIDFIRST_lazy_if(s,UTF)) {
5621 while (isALNUM_lazy_if(s,UTF))
5623 while (s < PL_bufend && isSPACE(*s))
5628 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5632 Perl_croak(aTHX_ "No comma allowed after %s", what);
5637 /* Either returns sv, or mortalizes sv and returns a new SV*.
5638 Best used as sv=new_constant(..., sv, ...).
5639 If s, pv are NULL, calls subroutine with one argument,
5640 and type is used with error messages only. */
5643 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5647 HV *table = GvHV(PL_hintgv); /* ^H */
5651 const char *why1, *why2, *why3;
5653 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
5656 why1 = "%^H is not consistent";
5657 why2 = strEQ(key,"charnames")
5658 ? " (missing \"use charnames ...\"?)"
5662 msg = Perl_newSVpvf(aTHX_ "constant(%s): %s%s%s",
5663 (type ? type: "undef"), why1, why2, why3);
5664 yyerror(SvPVX(msg));
5668 cvp = hv_fetch(table, key, strlen(key), FALSE);
5669 if (!cvp || !SvOK(*cvp)) {
5672 why3 = "} is not defined";
5675 sv_2mortal(sv); /* Parent created it permanently */
5678 pv = sv_2mortal(newSVpvn(s, len));
5680 typesv = sv_2mortal(newSVpv(type, 0));
5682 typesv = &PL_sv_undef;
5684 PUSHSTACKi(PERLSI_OVERLOAD);
5697 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5701 /* Check the eval first */
5702 if (!PL_in_eval && SvTRUE(ERRSV)) {
5704 sv_catpv(ERRSV, "Propagated");
5705 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5707 res = SvREFCNT_inc(sv);
5711 (void)SvREFCNT_inc(res);
5720 why1 = "Call to &{$^H{";
5722 why3 = "}} did not return a defined value";
5731 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5733 register char *d = dest;
5734 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5737 Perl_croak(aTHX_ ident_too_long);
5738 if (isALNUM(*s)) /* UTF handled below */
5740 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
5745 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5749 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5750 char *t = s + UTF8SKIP(s);
5751 while (*t & 0x80 && is_utf8_mark((U8*)t))
5753 if (d + (t - s) > e)
5754 Perl_croak(aTHX_ ident_too_long);
5755 Copy(s, d, t - s, char);
5768 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5778 e = d + destlen - 3; /* two-character token, ending NUL */
5780 while (isDIGIT(*s)) {
5782 Perl_croak(aTHX_ ident_too_long);
5789 Perl_croak(aTHX_ ident_too_long);
5790 if (isALNUM(*s)) /* UTF handled below */
5792 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
5797 else if (*s == ':' && s[1] == ':') {
5801 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5802 char *t = s + UTF8SKIP(s);
5803 while (*t & 0x80 && is_utf8_mark((U8*)t))
5805 if (d + (t - s) > e)
5806 Perl_croak(aTHX_ ident_too_long);
5807 Copy(s, d, t - s, char);
5818 if (PL_lex_state != LEX_NORMAL)
5819 PL_lex_state = LEX_INTERPENDMAYBE;
5822 if (*s == '$' && s[1] &&
5823 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
5836 if (*d == '^' && *s && isCONTROLVAR(*s)) {
5841 if (isSPACE(s[-1])) {
5844 if (ch != ' ' && ch != '\t') {
5850 if (isIDFIRST_lazy_if(d,UTF)) {
5854 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
5856 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
5859 Copy(s, d, e - s, char);
5864 while ((isALNUM(*s) || *s == ':') && d < e)
5867 Perl_croak(aTHX_ ident_too_long);
5870 while (s < send && (*s == ' ' || *s == '\t')) s++;
5871 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
5872 dTHR; /* only for ckWARN */
5873 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
5874 const char *brack = *s == '[' ? "[...]" : "{...}";
5875 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5876 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
5877 funny, dest, brack, funny, dest, brack);
5880 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
5884 /* Handle extended ${^Foo} variables
5885 * 1999-02-27 mjd-perl-patch@plover.com */
5886 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
5890 while (isALNUM(*s) && d < e) {
5894 Perl_croak(aTHX_ ident_too_long);
5899 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
5900 PL_lex_state = LEX_INTERPEND;
5903 if (PL_lex_state == LEX_NORMAL) {
5904 dTHR; /* only for ckWARN */
5905 if (ckWARN(WARN_AMBIGUOUS) &&
5906 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
5908 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5909 "Ambiguous use of %c{%s} resolved to %c%s",
5910 funny, dest, funny, dest);
5915 s = bracket; /* let the parser handle it */
5919 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
5920 PL_lex_state = LEX_INTERPEND;
5925 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
5930 *pmfl |= PMf_GLOBAL;
5932 *pmfl |= PMf_CONTINUE;
5936 *pmfl |= PMf_MULTILINE;
5938 *pmfl |= PMf_SINGLELINE;
5940 *pmfl |= PMf_EXTENDED;
5944 S_scan_pat(pTHX_ char *start, I32 type)
5949 s = scan_str(start,FALSE,FALSE);
5952 SvREFCNT_dec(PL_lex_stuff);
5953 PL_lex_stuff = Nullsv;
5954 Perl_croak(aTHX_ "Search pattern not terminated");
5957 pm = (PMOP*)newPMOP(type, 0);
5958 if (PL_multi_open == '?')
5959 pm->op_pmflags |= PMf_ONCE;
5961 while (*s && strchr("iomsx", *s))
5962 pmflag(&pm->op_pmflags,*s++);
5965 while (*s && strchr("iogcmsx", *s))
5966 pmflag(&pm->op_pmflags,*s++);
5968 pm->op_pmpermflags = pm->op_pmflags;
5970 PL_lex_op = (OP*)pm;
5971 yylval.ival = OP_MATCH;
5976 S_scan_subst(pTHX_ char *start)
5983 yylval.ival = OP_NULL;
5985 s = scan_str(start,FALSE,FALSE);
5989 SvREFCNT_dec(PL_lex_stuff);
5990 PL_lex_stuff = Nullsv;
5991 Perl_croak(aTHX_ "Substitution pattern not terminated");
5994 if (s[-1] == PL_multi_open)
5997 first_start = PL_multi_start;
5998 s = scan_str(s,FALSE,FALSE);
6001 SvREFCNT_dec(PL_lex_stuff);
6002 PL_lex_stuff = Nullsv;
6004 SvREFCNT_dec(PL_lex_repl);
6005 PL_lex_repl = Nullsv;
6006 Perl_croak(aTHX_ "Substitution replacement not terminated");
6008 PL_multi_start = first_start; /* so whole substitution is taken together */
6010 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6016 else if (strchr("iogcmsx", *s))
6017 pmflag(&pm->op_pmflags,*s++);
6024 PL_sublex_info.super_bufptr = s;
6025 PL_sublex_info.super_bufend = PL_bufend;
6027 pm->op_pmflags |= PMf_EVAL;
6028 repl = newSVpvn("",0);
6030 sv_catpv(repl, es ? "eval " : "do ");
6031 sv_catpvn(repl, "{ ", 2);
6032 sv_catsv(repl, PL_lex_repl);
6033 sv_catpvn(repl, " };", 2);
6035 SvREFCNT_dec(PL_lex_repl);
6039 pm->op_pmpermflags = pm->op_pmflags;
6040 PL_lex_op = (OP*)pm;
6041 yylval.ival = OP_SUBST;
6046 S_scan_trans(pTHX_ char *start)
6057 yylval.ival = OP_NULL;
6059 s = scan_str(start,FALSE,FALSE);
6062 SvREFCNT_dec(PL_lex_stuff);
6063 PL_lex_stuff = Nullsv;
6064 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6066 if (s[-1] == PL_multi_open)
6069 s = scan_str(s,FALSE,FALSE);
6072 SvREFCNT_dec(PL_lex_stuff);
6073 PL_lex_stuff = Nullsv;
6075 SvREFCNT_dec(PL_lex_repl);
6076 PL_lex_repl = Nullsv;
6077 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6081 o = newSVOP(OP_TRANS, 0, 0);
6082 utf8 = OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF;
6085 New(803,tbl,256,short);
6086 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6090 complement = del = squash = 0;
6091 while (strchr("cdsCU", *s)) {
6093 complement = OPpTRANS_COMPLEMENT;
6095 del = OPpTRANS_DELETE;
6097 squash = OPpTRANS_SQUASH;
6102 utf8 &= ~OPpTRANS_FROM_UTF;
6104 utf8 |= OPpTRANS_FROM_UTF;
6108 utf8 &= ~OPpTRANS_TO_UTF;
6110 utf8 |= OPpTRANS_TO_UTF;
6113 Perl_croak(aTHX_ "Too many /C and /U options");
6118 o->op_private = del|squash|complement|utf8;
6121 yylval.ival = OP_TRANS;
6126 S_scan_heredoc(pTHX_ register char *s)
6130 I32 op_type = OP_SCALAR;
6137 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6141 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6144 for (peek = s; *peek == ' ' || *peek == '\t'; peek++) ;
6145 if (*peek && strchr("`'\"",*peek)) {
6148 s = delimcpy(d, e, s, PL_bufend, term, &len);
6158 if (!isALNUM_lazy_if(s,UTF))
6159 deprecate("bare << to mean <<\"\"");
6160 for (; isALNUM_lazy_if(s,UTF); s++) {
6165 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6166 Perl_croak(aTHX_ "Delimiter for here document is too long");
6169 len = d - PL_tokenbuf;
6170 #ifndef PERL_STRICT_CR
6171 d = strchr(s, '\r');
6175 while (s < PL_bufend) {
6181 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6190 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6195 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6196 herewas = newSVpvn(s,PL_bufend-s);
6198 s--, herewas = newSVpvn(s,d-s);
6199 s += SvCUR(herewas);
6201 tmpstr = NEWSV(87,79);
6202 sv_upgrade(tmpstr, SVt_PVIV);
6207 else if (term == '`') {
6208 op_type = OP_BACKTICK;
6209 SvIVX(tmpstr) = '\\';
6213 PL_multi_start = CopLINE(PL_curcop);
6214 PL_multi_open = PL_multi_close = '<';
6215 term = *PL_tokenbuf;
6216 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6217 char *bufptr = PL_sublex_info.super_bufptr;
6218 char *bufend = PL_sublex_info.super_bufend;
6219 char *olds = s - SvCUR(herewas);
6220 s = strchr(bufptr, '\n');
6224 while (s < bufend &&
6225 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6227 CopLINE_inc(PL_curcop);
6230 CopLINE_set(PL_curcop, PL_multi_start);
6231 missingterm(PL_tokenbuf);
6233 sv_setpvn(herewas,bufptr,d-bufptr+1);
6234 sv_setpvn(tmpstr,d+1,s-d);
6236 sv_catpvn(herewas,s,bufend-s);
6237 (void)strcpy(bufptr,SvPVX(herewas));
6244 while (s < PL_bufend &&
6245 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6247 CopLINE_inc(PL_curcop);
6249 if (s >= PL_bufend) {
6250 CopLINE_set(PL_curcop, PL_multi_start);
6251 missingterm(PL_tokenbuf);
6253 sv_setpvn(tmpstr,d+1,s-d);
6255 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6257 sv_catpvn(herewas,s,PL_bufend-s);
6258 sv_setsv(PL_linestr,herewas);
6259 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6260 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6263 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6264 while (s >= PL_bufend) { /* multiple line string? */
6266 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6267 CopLINE_set(PL_curcop, PL_multi_start);
6268 missingterm(PL_tokenbuf);
6270 CopLINE_inc(PL_curcop);
6271 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6272 #ifndef PERL_STRICT_CR
6273 if (PL_bufend - PL_linestart >= 2) {
6274 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6275 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6277 PL_bufend[-2] = '\n';
6279 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6281 else if (PL_bufend[-1] == '\r')
6282 PL_bufend[-1] = '\n';
6284 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6285 PL_bufend[-1] = '\n';
6287 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6288 SV *sv = NEWSV(88,0);
6290 sv_upgrade(sv, SVt_PVMG);
6291 sv_setsv(sv,PL_linestr);
6292 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6294 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6297 sv_catsv(PL_linestr,herewas);
6298 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6302 sv_catsv(tmpstr,PL_linestr);
6307 PL_multi_end = CopLINE(PL_curcop);
6308 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6309 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6310 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6312 SvREFCNT_dec(herewas);
6313 PL_lex_stuff = tmpstr;
6314 yylval.ival = op_type;
6319 takes: current position in input buffer
6320 returns: new position in input buffer
6321 side-effects: yylval and lex_op are set.
6326 <FH> read from filehandle
6327 <pkg::FH> read from package qualified filehandle
6328 <pkg'FH> read from package qualified filehandle
6329 <$fh> read from filehandle in $fh
6335 S_scan_inputsymbol(pTHX_ char *start)
6337 register char *s = start; /* current position in buffer */
6343 d = PL_tokenbuf; /* start of temp holding space */
6344 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6345 end = strchr(s, '\n');
6348 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6350 /* die if we didn't have space for the contents of the <>,
6351 or if it didn't end, or if we see a newline
6354 if (len >= sizeof PL_tokenbuf)
6355 Perl_croak(aTHX_ "Excessively long <> operator");
6357 Perl_croak(aTHX_ "Unterminated <> operator");
6362 Remember, only scalar variables are interpreted as filehandles by
6363 this code. Anything more complex (e.g., <$fh{$num}>) will be
6364 treated as a glob() call.
6365 This code makes use of the fact that except for the $ at the front,
6366 a scalar variable and a filehandle look the same.
6368 if (*d == '$' && d[1]) d++;
6370 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6371 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6374 /* If we've tried to read what we allow filehandles to look like, and
6375 there's still text left, then it must be a glob() and not a getline.
6376 Use scan_str to pull out the stuff between the <> and treat it
6377 as nothing more than a string.
6380 if (d - PL_tokenbuf != len) {
6381 yylval.ival = OP_GLOB;
6383 s = scan_str(start,FALSE,FALSE);
6385 Perl_croak(aTHX_ "Glob not terminated");
6389 /* we're in a filehandle read situation */
6392 /* turn <> into <ARGV> */
6394 (void)strcpy(d,"ARGV");
6396 /* if <$fh>, create the ops to turn the variable into a
6402 /* try to find it in the pad for this block, otherwise find
6403 add symbol table ops
6405 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6406 OP *o = newOP(OP_PADSV, 0);
6408 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6411 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6412 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6413 newUNOP(OP_RV2SV, 0,
6414 newGVOP(OP_GV, 0, gv)));
6416 PL_lex_op->op_flags |= OPf_SPECIAL;
6417 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6418 yylval.ival = OP_NULL;
6421 /* If it's none of the above, it must be a literal filehandle
6422 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6424 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6425 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6426 yylval.ival = OP_NULL;
6435 takes: start position in buffer
6436 keep_quoted preserve \ on the embedded delimiter(s)
6437 keep_delims preserve the delimiters around the string
6438 returns: position to continue reading from buffer
6439 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6440 updates the read buffer.
6442 This subroutine pulls a string out of the input. It is called for:
6443 q single quotes q(literal text)
6444 ' single quotes 'literal text'
6445 qq double quotes qq(interpolate $here please)
6446 " double quotes "interpolate $here please"
6447 qx backticks qx(/bin/ls -l)
6448 ` backticks `/bin/ls -l`
6449 qw quote words @EXPORT_OK = qw( func() $spam )
6450 m// regexp match m/this/
6451 s/// regexp substitute s/this/that/
6452 tr/// string transliterate tr/this/that/
6453 y/// string transliterate y/this/that/
6454 ($*@) sub prototypes sub foo ($)
6455 (stuff) sub attr parameters sub foo : attr(stuff)
6456 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6458 In most of these cases (all but <>, patterns and transliterate)
6459 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6460 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6461 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6464 It skips whitespace before the string starts, and treats the first
6465 character as the delimiter. If the delimiter is one of ([{< then
6466 the corresponding "close" character )]}> is used as the closing
6467 delimiter. It allows quoting of delimiters, and if the string has
6468 balanced delimiters ([{<>}]) it allows nesting.
6470 The lexer always reads these strings into lex_stuff, except in the
6471 case of the operators which take *two* arguments (s/// and tr///)
6472 when it checks to see if lex_stuff is full (presumably with the 1st
6473 arg to s or tr) and if so puts the string into lex_repl.
6478 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6481 SV *sv; /* scalar value: string */
6482 char *tmps; /* temp string, used for delimiter matching */
6483 register char *s = start; /* current position in the buffer */
6484 register char term; /* terminating character */
6485 register char *to; /* current position in the sv's data */
6486 I32 brackets = 1; /* bracket nesting level */
6487 bool has_utf = FALSE; /* is there any utf8 content? */
6489 /* skip space before the delimiter */
6493 /* mark where we are, in case we need to report errors */
6496 /* after skipping whitespace, the next character is the terminator */
6498 if ((term & 0x80) && UTF)
6501 /* mark where we are */
6502 PL_multi_start = CopLINE(PL_curcop);
6503 PL_multi_open = term;
6505 /* find corresponding closing delimiter */
6506 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6508 PL_multi_close = term;
6510 /* create a new SV to hold the contents. 87 is leak category, I'm
6511 assuming. 79 is the SV's initial length. What a random number. */
6513 sv_upgrade(sv, SVt_PVIV);
6515 (void)SvPOK_only(sv); /* validate pointer */
6517 /* move past delimiter and try to read a complete string */
6519 sv_catpvn(sv, s, 1);
6522 /* extend sv if need be */
6523 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6524 /* set 'to' to the next character in the sv's string */
6525 to = SvPVX(sv)+SvCUR(sv);
6527 /* if open delimiter is the close delimiter read unbridle */
6528 if (PL_multi_open == PL_multi_close) {
6529 for (; s < PL_bufend; s++,to++) {
6530 /* embedded newlines increment the current line number */
6531 if (*s == '\n' && !PL_rsfp)
6532 CopLINE_inc(PL_curcop);
6533 /* handle quoted delimiters */
6534 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6535 if (!keep_quoted && s[1] == term)
6537 /* any other quotes are simply copied straight through */
6541 /* terminate when run out of buffer (the for() condition), or
6542 have found the terminator */
6543 else if (*s == term)
6545 else if (!has_utf && (*s & 0x80) && UTF)
6551 /* if the terminator isn't the same as the start character (e.g.,
6552 matched brackets), we have to allow more in the quoting, and
6553 be prepared for nested brackets.
6556 /* read until we run out of string, or we find the terminator */
6557 for (; s < PL_bufend; s++,to++) {
6558 /* embedded newlines increment the line count */
6559 if (*s == '\n' && !PL_rsfp)
6560 CopLINE_inc(PL_curcop);
6561 /* backslashes can escape the open or closing characters */
6562 if (*s == '\\' && s+1 < PL_bufend) {
6564 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6569 /* allow nested opens and closes */
6570 else if (*s == PL_multi_close && --brackets <= 0)
6572 else if (*s == PL_multi_open)
6574 else if (!has_utf && (*s & 0x80) && UTF)
6579 /* terminate the copied string and update the sv's end-of-string */
6581 SvCUR_set(sv, to - SvPVX(sv));
6584 * this next chunk reads more into the buffer if we're not done yet
6588 break; /* handle case where we are done yet :-) */
6590 #ifndef PERL_STRICT_CR
6591 if (to - SvPVX(sv) >= 2) {
6592 if ((to[-2] == '\r' && to[-1] == '\n') ||
6593 (to[-2] == '\n' && to[-1] == '\r'))
6597 SvCUR_set(sv, to - SvPVX(sv));
6599 else if (to[-1] == '\r')
6602 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6606 /* if we're out of file, or a read fails, bail and reset the current
6607 line marker so we can report where the unterminated string began
6610 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6612 CopLINE_set(PL_curcop, PL_multi_start);
6615 /* we read a line, so increment our line counter */
6616 CopLINE_inc(PL_curcop);
6618 /* update debugger info */
6619 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6620 SV *sv = NEWSV(88,0);
6622 sv_upgrade(sv, SVt_PVMG);
6623 sv_setsv(sv,PL_linestr);
6624 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6627 /* having changed the buffer, we must update PL_bufend */
6628 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6631 /* at this point, we have successfully read the delimited string */
6634 sv_catpvn(sv, s, 1);
6637 PL_multi_end = CopLINE(PL_curcop);
6640 /* if we allocated too much space, give some back */
6641 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6642 SvLEN_set(sv, SvCUR(sv) + 1);
6643 Renew(SvPVX(sv), SvLEN(sv), char);
6646 /* decide whether this is the first or second quoted string we've read
6659 takes: pointer to position in buffer
6660 returns: pointer to new position in buffer
6661 side-effects: builds ops for the constant in yylval.op
6663 Read a number in any of the formats that Perl accepts:
6665 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6666 [\d_]+(\.[\d_]*)?[Ee](\d+)
6668 Underbars (_) are allowed in decimal numbers. If -w is on,
6669 underbars before a decimal point must be at three digit intervals.
6671 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6674 If it reads a number without a decimal point or an exponent, it will
6675 try converting the number to an integer and see if it can do so
6676 without loss of precision.
6680 Perl_scan_num(pTHX_ char *start)
6682 register char *s = start; /* current position in buffer */
6683 register char *d; /* destination in temp buffer */
6684 register char *e; /* end of temp buffer */
6685 NV value; /* number read, as a double */
6686 SV *sv = Nullsv; /* place to put the converted number */
6687 bool floatit; /* boolean: int or float? */
6688 char *lastub = 0; /* position of last underbar */
6689 static char number_too_long[] = "Number too long";
6691 /* We use the first character to decide what type of number this is */
6695 Perl_croak(aTHX_ "panic: scan_num");
6697 /* if it starts with a 0, it could be an octal number, a decimal in
6698 0.13 disguise, or a hexadecimal number, or a binary number. */
6702 u holds the "number so far"
6703 shift the power of 2 of the base
6704 (hex == 4, octal == 3, binary == 1)
6705 overflowed was the number more than we can hold?
6707 Shift is used when we add a digit. It also serves as an "are
6708 we in octal/hex/binary?" indicator to disallow hex characters
6715 bool overflowed = FALSE;
6716 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6717 static char* bases[5] = { "", "binary", "", "octal",
6719 static char* Bases[5] = { "", "Binary", "", "Octal",
6721 static char *maxima[5] = { "",
6722 "0b11111111111111111111111111111111",
6726 char *base, *Base, *max;
6732 } else if (s[1] == 'b') {
6736 /* check for a decimal in disguise */
6737 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6739 /* so it must be octal */
6743 base = bases[shift];
6744 Base = Bases[shift];
6745 max = maxima[shift];
6747 /* read the rest of the number */
6749 /* x is used in the overflow test,
6750 b is the digit we're adding on. */
6755 /* if we don't mention it, we're done */
6764 /* 8 and 9 are not octal */
6767 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6771 case '2': case '3': case '4':
6772 case '5': case '6': case '7':
6774 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6778 b = *s++ & 15; /* ASCII digit -> value of digit */
6782 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6783 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6784 /* make sure they said 0x */
6789 /* Prepare to put the digit we have onto the end
6790 of the number so far. We check for overflows.
6795 x = u << shift; /* make room for the digit */
6797 if ((x >> shift) != u
6798 && !(PL_hints & HINT_NEW_BINARY)) {
6802 if (ckWARN_d(WARN_OVERFLOW))
6803 Perl_warner(aTHX_ WARN_OVERFLOW,
6804 "Integer overflow in %s number",
6807 u = x | b; /* add the digit to the end */
6810 n *= nvshift[shift];
6811 /* If an NV has not enough bits in its
6812 * mantissa to represent an UV this summing of
6813 * small low-order numbers is a waste of time
6814 * (because the NV cannot preserve the
6815 * low-order bits anyway): we could just
6816 * remember when did we overflow and in the
6817 * end just multiply n by the right
6825 /* if we get here, we had success: make a scalar value from
6832 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6833 Perl_warner(aTHX_ WARN_PORTABLE,
6834 "%s number > %s non-portable",
6841 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6842 Perl_warner(aTHX_ WARN_PORTABLE,
6843 "%s number > %s non-portable",
6848 if (PL_hints & HINT_NEW_BINARY)
6849 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6854 handle decimal numbers.
6855 we're also sent here when we read a 0 as the first digit
6857 case '1': case '2': case '3': case '4': case '5':
6858 case '6': case '7': case '8': case '9': case '.':
6861 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
6864 /* read next group of digits and _ and copy into d */
6865 while (isDIGIT(*s) || *s == '_') {
6866 /* skip underscores, checking for misplaced ones
6870 dTHR; /* only for ckWARN */
6871 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
6872 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6876 /* check for end of fixed-length buffer */
6878 Perl_croak(aTHX_ number_too_long);
6879 /* if we're ok, copy the character */
6884 /* final misplaced underbar check */
6885 if (lastub && s - lastub != 3) {
6887 if (ckWARN(WARN_SYNTAX))
6888 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6891 /* read a decimal portion if there is one. avoid
6892 3..5 being interpreted as the number 3. followed
6895 if (*s == '.' && s[1] != '.') {
6899 /* copy, ignoring underbars, until we run out of
6900 digits. Note: no misplaced underbar checks!
6902 for (; isDIGIT(*s) || *s == '_'; s++) {
6903 /* fixed length buffer check */
6905 Perl_croak(aTHX_ number_too_long);
6909 if (*s == '.' && isDIGIT(s[1])) {
6910 /* oops, it's really a v-string, but without the "v" */
6916 /* read exponent part, if present */
6917 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
6921 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
6922 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
6924 /* allow positive or negative exponent */
6925 if (*s == '+' || *s == '-')
6928 /* read digits of exponent (no underbars :-) */
6929 while (isDIGIT(*s)) {
6931 Perl_croak(aTHX_ number_too_long);
6936 /* terminate the string */
6939 /* make an sv from the string */
6942 /* unfortunately this monster needs to be on one line or
6943 makedepend will be confused. */
6944 #if (defined(USE_64_BIT_INT) && (!defined(HAS_STRTOLL)|| !defined(HAS_STRTOULL))) || (!defined(USE_64_BIT_INT) && (!defined(HAS_STRTOL) || !defined(HAS_STRTOUL)))
6947 No working strto[u]l[l]. Since atoi() doesn't do range checks,
6948 we need to do this the hard way.
6951 value = Atof(PL_tokenbuf);
6954 See if we can make do with an integer value without loss of
6955 precision. We use I_V to cast to an int, because some
6956 compilers have issues. Then we try casting it back and see
6957 if it was the same. We only do this if we know we
6958 specifically read an integer.
6960 Note: if floatit is true, then we don't need to do the
6964 UV tryuv = U_V(value);
6965 if (!floatit && (NV)tryuv == value) {
6966 if (tryuv <= IV_MAX)
6967 sv_setiv(sv, (IV)tryuv);
6969 sv_setuv(sv, tryuv);
6972 sv_setnv(sv, value);
6976 strtol/strtoll sets errno to ERANGE if the number is too big
6977 for an integer. We try to do an integer conversion first
6978 if no characters indicating "float" have been found.
6985 if (*PL_tokenbuf == '-')
6986 iv = Strtol(PL_tokenbuf, (char**)NULL, 10);
6988 uv = Strtoul(PL_tokenbuf, (char**)NULL, 10);
6990 floatit = TRUE; /* probably just too large */
6991 else if (*PL_tokenbuf == '-')
6997 value = Atof(PL_tokenbuf);
6998 sv_setnv(sv, value);
7001 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7002 (PL_hints & HINT_NEW_INTEGER) )
7003 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7004 (floatit ? "float" : "integer"),
7008 /* if it starts with a v, it could be a v-string */
7014 while (isDIGIT(*pos) || *pos == '_')
7016 if (!isALPHA(*pos)) {
7018 U8 tmpbuf[UTF8_MAXLEN];
7021 s++; /* get past 'v' */
7024 sv_setpvn(sv, "", 0);
7027 if (*s == '0' && isDIGIT(s[1]))
7028 yyerror("Octal number in vector unsupported");
7031 /* this is atoi() that tolerates underscores */
7034 while (--end >= s) {
7039 rev += (*end - '0') * mult;
7041 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
7042 Perl_warner(aTHX_ WARN_OVERFLOW,
7043 "Integer overflow in decimal number");
7046 tmpend = uv_to_utf8(tmpbuf, rev);
7047 utf8 = utf8 || rev > 127;
7048 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
7049 if (*pos == '.' && isDIGIT(pos[1]))
7055 while (isDIGIT(*pos) || *pos == '_')
7063 sv_utf8_downgrade(sv, TRUE);
7070 /* make the op for the constant and return */
7073 yylval.opval = newSVOP(OP_CONST, 0, sv);
7075 yylval.opval = Nullop;
7081 S_scan_formline(pTHX_ register char *s)
7086 SV *stuff = newSVpvn("",0);
7087 bool needargs = FALSE;
7090 if (*s == '.' || *s == '}') {
7092 #ifdef PERL_STRICT_CR
7093 for (t = s+1;*t == ' ' || *t == '\t'; t++) ;
7095 for (t = s+1;*t == ' ' || *t == '\t' || *t == '\r'; t++) ;
7097 if (*t == '\n' || t == PL_bufend)
7100 if (PL_in_eval && !PL_rsfp) {
7101 eol = strchr(s,'\n');
7106 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7108 for (t = s; t < eol; t++) {
7109 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7111 goto enough; /* ~~ must be first line in formline */
7113 if (*t == '@' || *t == '^')
7116 sv_catpvn(stuff, s, eol-s);
7117 #ifndef PERL_STRICT_CR
7118 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7119 char *end = SvPVX(stuff) + SvCUR(stuff);
7128 s = filter_gets(PL_linestr, PL_rsfp, 0);
7129 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7130 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7133 yyerror("Format not terminated");
7143 PL_lex_state = LEX_NORMAL;
7144 PL_nextval[PL_nexttoke].ival = 0;
7148 PL_lex_state = LEX_FORMLINE;
7149 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7151 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7155 SvREFCNT_dec(stuff);
7156 PL_lex_formbrack = 0;
7167 PL_cshlen = strlen(PL_cshname);
7172 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7175 I32 oldsavestack_ix = PL_savestack_ix;
7176 CV* outsidecv = PL_compcv;
7180 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7182 SAVEI32(PL_subline);
7183 save_item(PL_subname);
7186 SAVESPTR(PL_comppad_name);
7187 SAVESPTR(PL_compcv);
7188 SAVEI32(PL_comppad_name_fill);
7189 SAVEI32(PL_min_intro_pending);
7190 SAVEI32(PL_max_intro_pending);
7191 SAVEI32(PL_pad_reset_pending);
7193 PL_compcv = (CV*)NEWSV(1104,0);
7194 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7195 CvFLAGS(PL_compcv) |= flags;
7197 PL_comppad = newAV();
7198 av_push(PL_comppad, Nullsv);
7199 PL_curpad = AvARRAY(PL_comppad);
7200 PL_comppad_name = newAV();
7201 PL_comppad_name_fill = 0;
7202 PL_min_intro_pending = 0;
7204 PL_subline = CopLINE(PL_curcop);
7206 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7207 PL_curpad[0] = (SV*)newAV();
7208 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7209 #endif /* USE_THREADS */
7211 comppadlist = newAV();
7212 AvREAL_off(comppadlist);
7213 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7214 av_store(comppadlist, 1, (SV*)PL_comppad);
7216 CvPADLIST(PL_compcv) = comppadlist;
7217 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7219 CvOWNER(PL_compcv) = 0;
7220 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7221 MUTEX_INIT(CvMUTEXP(PL_compcv));
7222 #endif /* USE_THREADS */
7224 return oldsavestack_ix;
7228 Perl_yywarn(pTHX_ char *s)
7231 PL_in_eval |= EVAL_WARNONLY;
7233 PL_in_eval &= ~EVAL_WARNONLY;
7238 Perl_yyerror(pTHX_ char *s)
7242 char *context = NULL;
7246 if (!yychar || (yychar == ';' && !PL_rsfp))
7248 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7249 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7250 while (isSPACE(*PL_oldoldbufptr))
7252 context = PL_oldoldbufptr;
7253 contlen = PL_bufptr - PL_oldoldbufptr;
7255 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7256 PL_oldbufptr != PL_bufptr) {
7257 while (isSPACE(*PL_oldbufptr))
7259 context = PL_oldbufptr;
7260 contlen = PL_bufptr - PL_oldbufptr;
7262 else if (yychar > 255)
7263 where = "next token ???";
7264 #ifdef USE_PURE_BISON
7265 /* GNU Bison sets the value -2 */
7266 else if (yychar == -2) {
7268 else if ((yychar & 127) == 127) {
7270 if (PL_lex_state == LEX_NORMAL ||
7271 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7272 where = "at end of line";
7273 else if (PL_lex_inpat)
7274 where = "within pattern";
7276 where = "within string";
7279 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7281 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7282 else if (isPRINT_LC(yychar))
7283 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7285 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7286 where = SvPVX(where_sv);
7288 msg = sv_2mortal(newSVpv(s, 0));
7289 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7290 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7292 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7294 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7295 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7296 Perl_sv_catpvf(aTHX_ msg,
7297 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7298 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7301 if (PL_in_eval & EVAL_WARNONLY)
7302 Perl_warn(aTHX_ "%"SVf, msg);
7305 if (PL_error_count >= 10) {
7306 if (PL_in_eval && SvCUR(ERRSV))
7307 Perl_croak(aTHX_ "%_%s has too many errors.\n",
7308 ERRSV, CopFILE(PL_curcop));
7310 Perl_croak(aTHX_ "%s has too many errors.\n",
7311 CopFILE(PL_curcop));
7314 PL_in_my_stash = Nullhv;
7325 * Restore a source filter.
7329 restore_rsfp(pTHXo_ void *f)
7331 PerlIO *fp = (PerlIO*)f;
7333 if (PL_rsfp == PerlIO_stdin())
7334 PerlIO_clearerr(PL_rsfp);
7335 else if (PL_rsfp && (PL_rsfp != fp))
7336 PerlIO_close(PL_rsfp);