3 * Copyright (c) 1991-1999, 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 UTF (PL_hints & HINT_UTF8)
34 * Note: we try to be careful never to call the isXXX_utf8() functions
35 * unless we're pretty sure we've seen the beginning of a UTF-8 character
36 * (that is, the two high bits are set). Otherwise we risk loading in the
37 * heavy-duty SWASHINIT and SWASHGET routines unnecessarily.
39 #define isIDFIRST_lazy(p) ((!UTF || (*((U8*)p) < 0xc0)) \
41 : isIDFIRST_utf8((U8*)p))
42 #define isALNUM_lazy(p) ((!UTF || (*((U8*)p) < 0xc0)) \
44 : isALNUM_utf8((U8*)p))
46 /* In variables name $^X, these are the legal values for X.
47 * 1999-02-27 mjd-perl-patch@plover.com */
48 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
50 /* LEX_* are values for PL_lex_state, the state of the lexer.
51 * They are arranged oddly so that the guard on the switch statement
52 * can get by with a single comparison (if the compiler is smart enough).
55 /* #define LEX_NOTPARSING 11 is done in perl.h. */
58 #define LEX_INTERPNORMAL 9
59 #define LEX_INTERPCASEMOD 8
60 #define LEX_INTERPPUSH 7
61 #define LEX_INTERPSTART 6
62 #define LEX_INTERPEND 5
63 #define LEX_INTERPENDMAYBE 4
64 #define LEX_INTERPCONCAT 3
65 #define LEX_INTERPCONST 2
66 #define LEX_FORMLINE 1
67 #define LEX_KNOWNEXT 0
76 /* XXX If this causes problems, set i_unistd=undef in the hint file. */
78 # include <unistd.h> /* Needed for execv() */
87 YYSTYPE* yylval_pointer = NULL;
88 int* yychar_pointer = NULL;
91 # define yylval (*yylval_pointer)
92 # define yychar (*yychar_pointer)
93 # define PERL_YYLEX_PARAM yylval_pointer,yychar_pointer
95 # define yylex() Perl_yylex(aTHX_ yylval_pointer, yychar_pointer)
100 /* CLINE is a macro that ensures PL_copline has a sane value */
105 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
108 * Convenience functions to return different tokens and prime the
109 * lexer for the next token. They all take an argument.
111 * TOKEN : generic token (used for '(', DOLSHARP, etc)
112 * OPERATOR : generic operator
113 * AOPERATOR : assignment operator
114 * PREBLOCK : beginning the block after an if, while, foreach, ...
115 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
116 * PREREF : *EXPR where EXPR is not a simple identifier
117 * TERM : expression term
118 * LOOPX : loop exiting command (goto, last, dump, etc)
119 * FTST : file test operator
120 * FUN0 : zero-argument function
121 * FUN1 : not used, except for not, which isn't a UNIOP
122 * BOop : bitwise or or xor
124 * SHop : shift operator
125 * PWop : power operator
126 * PMop : pattern-matching operator
127 * Aop : addition-level operator
128 * Mop : multiplication-level operator
129 * Eop : equality-testing operator
130 * Rop : relational operator <= != gt
132 * Also see LOP and lop() below.
135 #define TOKEN(retval) return (PL_bufptr = s,(int)retval)
136 #define OPERATOR(retval) return (PL_expect = XTERM,PL_bufptr = s,(int)retval)
137 #define AOPERATOR(retval) return ao((PL_expect = XTERM,PL_bufptr = s,(int)retval))
138 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
139 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
140 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s,(int)retval)
141 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR,PL_bufptr = s,(int)retval)
142 #define LOOPX(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
143 #define FTST(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)UNIOP)
144 #define FUN0(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
145 #define FUN1(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
146 #define BOop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
147 #define BAop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
148 #define SHop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
149 #define PWop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
150 #define PMop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
151 #define Aop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
152 #define Mop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
153 #define Eop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
154 #define Rop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)RELOP)
156 /* This bit of chicanery makes a unary function followed by
157 * a parenthesis into a function with one argument, highest precedence.
159 #define UNI(f) return(yylval.ival = f, \
162 PL_last_uni = PL_oldbufptr, \
163 PL_last_lop_op = f, \
164 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
166 #define UNIBRACK(f) return(yylval.ival = f, \
168 PL_last_uni = PL_oldbufptr, \
169 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
171 /* grandfather return to old style */
172 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
177 * This subroutine detects &&= and ||= and turns an ANDAND or OROR
178 * into an OP_ANDASSIGN or OP_ORASSIGN
182 S_ao(pTHX_ int toketype)
184 if (*PL_bufptr == '=') {
186 if (toketype == ANDAND)
187 yylval.ival = OP_ANDASSIGN;
188 else if (toketype == OROR)
189 yylval.ival = OP_ORASSIGN;
197 * When Perl expects an operator and finds something else, no_op
198 * prints the warning. It always prints "<something> found where
199 * operator expected. It prints "Missing semicolon on previous line?"
200 * if the surprise occurs at the start of the line. "do you need to
201 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
202 * where the compiler doesn't know if foo is a method call or a function.
203 * It prints "Missing operator before end of line" if there's nothing
204 * after the missing operator, or "... before <...>" if there is something
205 * after the missing operator.
209 S_no_op(pTHX_ char *what, char *s)
211 char *oldbp = PL_bufptr;
212 bool is_first = (PL_oldbufptr == PL_linestart);
220 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
222 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
223 else if (PL_oldoldbufptr && isIDFIRST_lazy(PL_oldoldbufptr)) {
225 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy(t) || *t == ':'); t++) ;
226 if (t < PL_bufptr && isSPACE(*t))
227 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
228 t - PL_oldoldbufptr, PL_oldoldbufptr);
231 Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
237 * Complain about missing quote/regexp/heredoc terminator.
238 * If it's called with (char *)NULL then it cauterizes the line buffer.
239 * If we're in a delimited string and the delimiter is a control
240 * character, it's reformatted into a two-char sequence like ^C.
245 S_missingterm(pTHX_ char *s)
250 char *nl = strrchr(s,'\n');
256 iscntrl(PL_multi_close)
258 PL_multi_close < 32 || PL_multi_close == 127
262 tmpbuf[1] = toCTRL(PL_multi_close);
268 *tmpbuf = PL_multi_close;
272 q = strchr(s,'"') ? '\'' : '"';
273 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
281 Perl_deprecate(pTHX_ char *s)
284 if (ckWARN(WARN_DEPRECATED))
285 Perl_warner(aTHX_ WARN_DEPRECATED, "Use of %s is deprecated", s);
290 * Deprecate a comma-less variable list.
296 deprecate("comma-less variable list");
300 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
301 * utf16-to-utf8-reversed.
307 S_win32_textfilter(pTHX_ int idx, SV *sv, int maxlen)
309 I32 count = FILTER_READ(idx+1, sv, maxlen);
310 if (count > 0 && !maxlen)
311 win32_strip_return(sv);
317 S_utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
319 I32 count = FILTER_READ(idx+1, sv, maxlen);
323 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
324 tend = utf16_to_utf8((U16*)SvPVX(sv), tmps, SvCUR(sv));
325 sv_usepvn(sv, (char*)tmps, tend - tmps);
332 S_utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
334 I32 count = FILTER_READ(idx+1, sv, maxlen);
338 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
339 tend = utf16_to_utf8_reversed((U16*)SvPVX(sv), tmps, SvCUR(sv));
340 sv_usepvn(sv, (char*)tmps, tend - tmps);
348 * Initialize variables. Uses the Perl save_stack to save its state (for
349 * recursive calls to the parser).
353 Perl_lex_start(pTHX_ SV *line)
359 SAVEI32(PL_lex_dojoin);
360 SAVEI32(PL_lex_brackets);
361 SAVEI32(PL_lex_fakebrack);
362 SAVEI32(PL_lex_casemods);
363 SAVEI32(PL_lex_starts);
364 SAVEI32(PL_lex_state);
365 SAVEVPTR(PL_lex_inpat);
366 SAVEI32(PL_lex_inwhat);
367 SAVECOPLINE(PL_curcop);
370 SAVEPPTR(PL_oldbufptr);
371 SAVEPPTR(PL_oldoldbufptr);
372 SAVEPPTR(PL_linestart);
373 SAVESPTR(PL_linestr);
374 SAVEPPTR(PL_lex_brackstack);
375 SAVEPPTR(PL_lex_casestack);
376 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
377 SAVESPTR(PL_lex_stuff);
378 SAVEI32(PL_lex_defer);
379 SAVEI32(PL_sublex_info.sub_inwhat);
380 SAVESPTR(PL_lex_repl);
382 SAVEINT(PL_lex_expect);
384 PL_lex_state = LEX_NORMAL;
388 PL_lex_fakebrack = 0;
389 New(899, PL_lex_brackstack, 120, char);
390 New(899, PL_lex_casestack, 12, char);
391 SAVEFREEPV(PL_lex_brackstack);
392 SAVEFREEPV(PL_lex_casestack);
394 *PL_lex_casestack = '\0';
397 PL_lex_stuff = Nullsv;
398 PL_lex_repl = Nullsv;
401 PL_sublex_info.sub_inwhat = 0;
403 if (SvREADONLY(PL_linestr))
404 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
405 s = SvPV(PL_linestr, len);
406 if (len && s[len-1] != ';') {
407 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
408 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
409 sv_catpvn(PL_linestr, "\n;", 2);
411 SvTEMP_off(PL_linestr);
412 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
413 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
415 PL_rs = newSVpvn("\n", 1);
421 * Finalizer for lexing operations. Must be called when the parser is
422 * done with the lexer.
428 PL_doextract = FALSE;
433 * This subroutine has nothing to do with tilting, whether at windmills
434 * or pinball tables. Its name is short for "increment line". It
435 * increments the current line number in CopLINE(PL_curcop) and checks
436 * to see whether the line starts with a comment of the form
437 * # line 500 "foo.pm"
438 * If so, it sets the current line number and file to the values in the comment.
442 S_incline(pTHX_ char *s)
450 CopLINE_inc(PL_curcop);
453 while (*s == ' ' || *s == '\t') s++;
454 if (strnEQ(s, "line ", 5)) {
463 while (*s == ' ' || *s == '\t')
465 if (*s == '"' && (t = strchr(s+1, '"')))
469 return; /* false alarm */
470 for (t = s; !isSPACE(*t); t++) ;
475 CopFILE_set(PL_curcop, s);
477 CopFILE_set(PL_curcop, PL_origfilename);
479 CopLINE_set(PL_curcop, atoi(n)-1);
484 * Called to gobble the appropriate amount and type of whitespace.
485 * Skips comments as well.
489 S_skipspace(pTHX_ register char *s)
492 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
493 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
499 SSize_t oldprevlen, oldoldprevlen;
500 SSize_t oldloplen, oldunilen;
501 while (s < PL_bufend && isSPACE(*s)) {
502 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
507 if (s < PL_bufend && *s == '#') {
508 while (s < PL_bufend && *s != '\n')
512 if (PL_in_eval && !PL_rsfp) {
519 /* only continue to recharge the buffer if we're at the end
520 * of the buffer, we're not reading from a source filter, and
521 * we're in normal lexing mode
523 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
524 PL_lex_state == LEX_FORMLINE)
527 /* try to recharge the buffer */
528 if ((s = filter_gets(PL_linestr, PL_rsfp,
529 (prevlen = SvCUR(PL_linestr)))) == Nullch)
531 /* end of file. Add on the -p or -n magic */
532 if (PL_minus_n || PL_minus_p) {
533 sv_setpv(PL_linestr,PL_minus_p ?
534 ";}continue{print or die qq(-p destination: $!\\n)" :
536 sv_catpv(PL_linestr,";}");
537 PL_minus_n = PL_minus_p = 0;
540 sv_setpv(PL_linestr,";");
542 /* reset variables for next time we lex */
543 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
545 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
547 /* Close the filehandle. Could be from -P preprocessor,
548 * STDIN, or a regular file. If we were reading code from
549 * STDIN (because the commandline held no -e or filename)
550 * then we don't close it, we reset it so the code can
551 * read from STDIN too.
554 if (PL_preprocess && !PL_in_eval)
555 (void)PerlProc_pclose(PL_rsfp);
556 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
557 PerlIO_clearerr(PL_rsfp);
559 (void)PerlIO_close(PL_rsfp);
564 /* not at end of file, so we only read another line */
565 /* make corresponding updates to old pointers, for yyerror() */
566 oldprevlen = PL_oldbufptr - PL_bufend;
567 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
569 oldunilen = PL_last_uni - PL_bufend;
571 oldloplen = PL_last_lop - PL_bufend;
572 PL_linestart = PL_bufptr = s + prevlen;
573 PL_bufend = s + SvCUR(PL_linestr);
575 PL_oldbufptr = s + oldprevlen;
576 PL_oldoldbufptr = s + oldoldprevlen;
578 PL_last_uni = s + oldunilen;
580 PL_last_lop = s + oldloplen;
583 /* debugger active and we're not compiling the debugger code,
584 * so store the line into the debugger's array of lines
586 if (PERLDB_LINE && PL_curstash != PL_debstash) {
587 SV *sv = NEWSV(85,0);
589 sv_upgrade(sv, SVt_PVMG);
590 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
591 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
598 * Check the unary operators to ensure there's no ambiguity in how they're
599 * used. An ambiguous piece of code would be:
601 * This doesn't mean rand() + 5. Because rand() is a unary operator,
602 * the +5 is its argument.
612 if (PL_oldoldbufptr != PL_last_uni)
614 while (isSPACE(*PL_last_uni))
616 for (s = PL_last_uni; isALNUM_lazy(s) || *s == '-'; s++) ;
617 if ((t = strchr(s, '(')) && t < PL_bufptr)
619 if (ckWARN_d(WARN_AMBIGUOUS)){
622 Perl_warner(aTHX_ WARN_AMBIGUOUS,
623 "Warning: Use of \"%s\" without parens is ambiguous",
629 /* workaround to replace the UNI() macro with a function. Only the
630 * hints/uts.sh file mentions this. Other comments elsewhere in the
631 * source indicate Microport Unix might need it too.
637 #define UNI(f) return uni(f,s)
640 S_uni(pTHX_ I32 f, char *s)
645 PL_last_uni = PL_oldbufptr;
656 #endif /* CRIPPLED_CC */
659 * LOP : macro to build a list operator. Its behaviour has been replaced
660 * with a subroutine, S_lop() for which LOP is just another name.
663 #define LOP(f,x) return lop(f,x,s)
667 * Build a list operator (or something that might be one). The rules:
668 * - if we have a next token, then it's a list operator [why?]
669 * - if the next thing is an opening paren, then it's a function
670 * - else it's a list operator
674 S_lop(pTHX_ I32 f, expectation x, char *s)
681 PL_last_lop = PL_oldbufptr;
696 * When the lexer realizes it knows the next token (for instance,
697 * it is reordering tokens for the parser) then it can call S_force_next
698 * to know what token to return the next time the lexer is called. Caller
699 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
700 * handles the token correctly.
704 S_force_next(pTHX_ I32 type)
706 PL_nexttype[PL_nexttoke] = type;
708 if (PL_lex_state != LEX_KNOWNEXT) {
709 PL_lex_defer = PL_lex_state;
710 PL_lex_expect = PL_expect;
711 PL_lex_state = LEX_KNOWNEXT;
717 * When the lexer knows the next thing is a word (for instance, it has
718 * just seen -> and it knows that the next char is a word char, then
719 * it calls S_force_word to stick the next word into the PL_next lookahead.
722 * char *start : buffer position (must be within PL_linestr)
723 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
724 * int check_keyword : if true, Perl checks to make sure the word isn't
725 * a keyword (do this if the word is a label, e.g. goto FOO)
726 * int allow_pack : if true, : characters will also be allowed (require,
728 * int allow_initial_tick : used by the "sub" lexer only.
732 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
737 start = skipspace(start);
739 if (isIDFIRST_lazy(s) ||
740 (allow_pack && *s == ':') ||
741 (allow_initial_tick && *s == '\'') )
743 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
744 if (check_keyword && keyword(PL_tokenbuf, len))
746 if (token == METHOD) {
751 PL_expect = XOPERATOR;
754 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
755 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
763 * Called when the lexer wants $foo *foo &foo etc, but the program
764 * text only contains the "foo" portion. The first argument is a pointer
765 * to the "foo", and the second argument is the type symbol to prefix.
766 * Forces the next token to be a "WORD".
767 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
771 S_force_ident(pTHX_ register char *s, int kind)
774 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
775 PL_nextval[PL_nexttoke].opval = o;
778 dTHR; /* just for in_eval */
779 o->op_private = OPpCONST_ENTERED;
780 /* XXX see note in pp_entereval() for why we forgo typo
781 warnings if the symbol must be introduced in an eval.
783 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
784 kind == '$' ? SVt_PV :
785 kind == '@' ? SVt_PVAV :
786 kind == '%' ? SVt_PVHV :
795 * Forces the next token to be a version number.
799 S_force_version(pTHX_ char *s)
801 OP *version = Nullop;
805 /* default VERSION number -- GBARR */
810 for( d=s, c = 1; isDIGIT(*d) || *d == '_' || (*d == '.' && c--); d++);
811 if((*d == ';' || isSPACE(*d)) && *(skipspace(d)) != ',') {
813 /* real VERSION number -- GBARR */
814 version = yylval.opval;
818 /* NOTE: The parser sees the package name and the VERSION swapped */
819 PL_nextval[PL_nexttoke].opval = version;
827 * Tokenize a quoted string passed in as an SV. It finds the next
828 * chunk, up to end of string or a backslash. It may make a new
829 * SV containing that chunk (if HINT_NEW_STRING is on). It also
834 S_tokeq(pTHX_ SV *sv)
845 s = SvPV_force(sv, len);
849 while (s < send && *s != '\\')
854 if ( PL_hints & HINT_NEW_STRING )
855 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
858 if (s + 1 < send && (s[1] == '\\'))
859 s++; /* all that, just for this */
864 SvCUR_set(sv, d - SvPVX(sv));
866 if ( PL_hints & HINT_NEW_STRING )
867 return new_constant(NULL, 0, "q", sv, pv, "q");
872 * Now come three functions related to double-quote context,
873 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
874 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
875 * interact with PL_lex_state, and create fake ( ... ) argument lists
876 * to handle functions and concatenation.
877 * They assume that whoever calls them will be setting up a fake
878 * join call, because each subthing puts a ',' after it. This lets
881 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
883 * (I'm not sure whether the spurious commas at the end of lcfirst's
884 * arguments and join's arguments are created or not).
889 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
891 * Pattern matching will set PL_lex_op to the pattern-matching op to
892 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
894 * OP_CONST and OP_READLINE are easy--just make the new op and return.
896 * Everything else becomes a FUNC.
898 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
899 * had an OP_CONST or OP_READLINE). This just sets us up for a
900 * call to S_sublex_push().
906 register I32 op_type = yylval.ival;
908 if (op_type == OP_NULL) {
909 yylval.opval = PL_lex_op;
913 if (op_type == OP_CONST || op_type == OP_READLINE) {
914 SV *sv = tokeq(PL_lex_stuff);
916 if (SvTYPE(sv) == SVt_PVIV) {
917 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
923 nsv = newSVpvn(p, len);
927 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
928 PL_lex_stuff = Nullsv;
932 PL_sublex_info.super_state = PL_lex_state;
933 PL_sublex_info.sub_inwhat = op_type;
934 PL_sublex_info.sub_op = PL_lex_op;
935 PL_lex_state = LEX_INTERPPUSH;
939 yylval.opval = PL_lex_op;
949 * Create a new scope to save the lexing state. The scope will be
950 * ended in S_sublex_done. Returns a '(', starting the function arguments
951 * to the uc, lc, etc. found before.
952 * Sets PL_lex_state to LEX_INTERPCONCAT.
961 PL_lex_state = PL_sublex_info.super_state;
962 SAVEI32(PL_lex_dojoin);
963 SAVEI32(PL_lex_brackets);
964 SAVEI32(PL_lex_fakebrack);
965 SAVEI32(PL_lex_casemods);
966 SAVEI32(PL_lex_starts);
967 SAVEI32(PL_lex_state);
968 SAVEVPTR(PL_lex_inpat);
969 SAVEI32(PL_lex_inwhat);
970 SAVECOPLINE(PL_curcop);
972 SAVEPPTR(PL_oldbufptr);
973 SAVEPPTR(PL_oldoldbufptr);
974 SAVEPPTR(PL_linestart);
975 SAVESPTR(PL_linestr);
976 SAVEPPTR(PL_lex_brackstack);
977 SAVEPPTR(PL_lex_casestack);
979 PL_linestr = PL_lex_stuff;
980 PL_lex_stuff = Nullsv;
982 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
984 PL_bufend += SvCUR(PL_linestr);
985 SAVEFREESV(PL_linestr);
987 PL_lex_dojoin = FALSE;
989 PL_lex_fakebrack = 0;
990 New(899, PL_lex_brackstack, 120, char);
991 New(899, PL_lex_casestack, 12, char);
992 SAVEFREEPV(PL_lex_brackstack);
993 SAVEFREEPV(PL_lex_casestack);
995 *PL_lex_casestack = '\0';
997 PL_lex_state = LEX_INTERPCONCAT;
998 CopLINE_set(PL_curcop, PL_multi_start);
1000 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1001 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1002 PL_lex_inpat = PL_sublex_info.sub_op;
1004 PL_lex_inpat = Nullop;
1011 * Restores lexer state after a S_sublex_push.
1017 if (!PL_lex_starts++) {
1018 PL_expect = XOPERATOR;
1019 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn("",0));
1023 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1024 PL_lex_state = LEX_INTERPCASEMOD;
1028 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1029 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1030 PL_linestr = PL_lex_repl;
1032 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1033 PL_bufend += SvCUR(PL_linestr);
1034 SAVEFREESV(PL_linestr);
1035 PL_lex_dojoin = FALSE;
1036 PL_lex_brackets = 0;
1037 PL_lex_fakebrack = 0;
1038 PL_lex_casemods = 0;
1039 *PL_lex_casestack = '\0';
1041 if (SvEVALED(PL_lex_repl)) {
1042 PL_lex_state = LEX_INTERPNORMAL;
1044 /* we don't clear PL_lex_repl here, so that we can check later
1045 whether this is an evalled subst; that means we rely on the
1046 logic to ensure sublex_done() is called again only via the
1047 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1050 PL_lex_state = LEX_INTERPCONCAT;
1051 PL_lex_repl = Nullsv;
1057 PL_bufend = SvPVX(PL_linestr);
1058 PL_bufend += SvCUR(PL_linestr);
1059 PL_expect = XOPERATOR;
1060 PL_sublex_info.sub_inwhat = 0;
1068 Extracts a pattern, double-quoted string, or transliteration. This
1071 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1072 processing a pattern (PL_lex_inpat is true), a transliteration
1073 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1075 Returns a pointer to the character scanned up to. Iff this is
1076 advanced from the start pointer supplied (ie if anything was
1077 successfully parsed), will leave an OP for the substring scanned
1078 in yylval. Caller must intuit reason for not parsing further
1079 by looking at the next characters herself.
1083 double-quoted style: \r and \n
1084 regexp special ones: \D \s
1086 backrefs: \1 (deprecated in substitution replacements)
1087 case and quoting: \U \Q \E
1088 stops on @ and $, but not for $ as tail anchor
1090 In transliterations:
1091 characters are VERY literal, except for - not at the start or end
1092 of the string, which indicates a range. scan_const expands the
1093 range to the full set of intermediate characters.
1095 In double-quoted strings:
1097 double-quoted style: \r and \n
1099 backrefs: \1 (deprecated)
1100 case and quoting: \U \Q \E
1103 scan_const does *not* construct ops to handle interpolated strings.
1104 It stops processing as soon as it finds an embedded $ or @ variable
1105 and leaves it to the caller to work out what's going on.
1107 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1109 $ in pattern could be $foo or could be tail anchor. Assumption:
1110 it's a tail anchor if $ is the last thing in the string, or if it's
1111 followed by one of ")| \n\t"
1113 \1 (backreferences) are turned into $1
1115 The structure of the code is
1116 while (there's a character to process) {
1117 handle transliteration ranges
1118 skip regexp comments
1119 skip # initiated comments in //x patterns
1120 check for embedded @foo
1121 check for embedded scalars
1123 leave intact backslashes from leave (below)
1124 deprecate \1 in strings and sub replacements
1125 handle string-changing backslashes \l \U \Q \E, etc.
1126 switch (what was escaped) {
1127 handle - in a transliteration (becomes a literal -)
1128 handle \132 octal characters
1129 handle 0x15 hex characters
1130 handle \cV (control V)
1131 handle printf backslashes (\f, \r, \n, etc)
1133 } (end if backslash)
1134 } (end while character to read)
1139 S_scan_const(pTHX_ char *start)
1141 register char *send = PL_bufend; /* end of the constant */
1142 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1143 register char *s = start; /* start of the constant */
1144 register char *d = SvPVX(sv); /* destination for copies */
1145 bool dorange = FALSE; /* are we in a translit range? */
1147 I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1148 ? (PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1150 I32 thisutf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1151 ? (PL_sublex_info.sub_op->op_private & (PL_lex_repl ?
1152 OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF))
1154 const char *leaveit = /* set of acceptably-backslashed characters */
1156 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1159 while (s < send || dorange) {
1160 /* get transliterations out of the way (they're most literal) */
1161 if (PL_lex_inwhat == OP_TRANS) {
1162 /* expand a range A-Z to the full set of characters. AIE! */
1164 I32 i; /* current expanded character */
1165 I32 min; /* first character in range */
1166 I32 max; /* last character in range */
1168 i = d - SvPVX(sv); /* remember current offset */
1169 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1170 d = SvPVX(sv) + i; /* refresh d after realloc */
1171 d -= 2; /* eat the first char and the - */
1173 min = (U8)*d; /* first char in range */
1174 max = (U8)d[1]; /* last char in range */
1177 if ((isLOWER(min) && isLOWER(max)) ||
1178 (isUPPER(min) && isUPPER(max))) {
1180 for (i = min; i <= max; i++)
1184 for (i = min; i <= max; i++)
1191 for (i = min; i <= max; i++)
1194 /* mark the range as done, and continue */
1199 /* range begins (ignore - as first or last char) */
1200 else if (*s == '-' && s+1 < send && s != start) {
1202 *d++ = (char)0xff; /* use illegal utf8 byte--see pmtrans */
1211 /* if we get here, we're not doing a transliteration */
1213 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1214 except for the last char, which will be done separately. */
1215 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1217 while (s < send && *s != ')')
1219 } else if (s[2] == '{'
1220 || s[2] == 'p' && s[3] == '{') { /* This should march regcomp.c */
1222 char *regparse = s + (s[2] == '{' ? 3 : 4);
1225 while (count && (c = *regparse)) {
1226 if (c == '\\' && regparse[1])
1234 if (*regparse != ')') {
1235 regparse--; /* Leave one char for continuation. */
1236 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1238 while (s < regparse)
1243 /* likewise skip #-initiated comments in //x patterns */
1244 else if (*s == '#' && PL_lex_inpat &&
1245 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1246 while (s+1 < send && *s != '\n')
1250 /* check for embedded arrays (@foo, @:foo, @'foo, @{foo}, @$foo) */
1251 else if (*s == '@' && s[1] && (isALNUM_lazy(s+1) || strchr(":'{$", s[1])))
1254 /* check for embedded scalars. only stop if we're sure it's a
1257 else if (*s == '$') {
1258 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1260 if (s + 1 < send && !strchr("()| \n\t", s[1]))
1261 break; /* in regexp, $ might be tail anchor */
1264 /* (now in tr/// code again) */
1266 if (*s & 0x80 && thisutf) {
1267 dTHR; /* only for ckWARN */
1268 if (ckWARN(WARN_UTF8)) {
1269 (void)utf8_to_uv((U8*)s, &len); /* could cvt latin-1 to utf8 here... */
1279 if (*s == '\\' && s+1 < send) {
1282 /* some backslashes we leave behind */
1283 if (*leaveit && *s && strchr(leaveit, *s)) {
1289 /* deprecate \1 in strings and substitution replacements */
1290 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1291 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1293 dTHR; /* only for ckWARN */
1294 if (ckWARN(WARN_SYNTAX))
1295 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1300 /* string-change backslash escapes */
1301 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1306 /* if we get here, it's either a quoted -, or a digit */
1309 /* quoted - in transliterations */
1311 if (PL_lex_inwhat == OP_TRANS) {
1319 if (ckWARN(WARN_UNSAFE) && isALPHA(*s))
1320 Perl_warner(aTHX_ WARN_UNSAFE,
1321 "Unrecognized escape \\%c passed through",
1323 /* default action is to copy the quoted character */
1328 /* \132 indicates an octal constant */
1329 case '0': case '1': case '2': case '3':
1330 case '4': case '5': case '6': case '7':
1331 *d++ = (char)scan_oct(s, 3, &len);
1335 /* \x24 indicates a hex constant */
1339 char* e = strchr(s, '}');
1342 yyerror("Missing right brace on \\x{}");
1347 if (ckWARN(WARN_UTF8))
1348 Perl_warner(aTHX_ WARN_UTF8,
1349 "Use of \\x{} without utf8 declaration");
1351 /* note: utf always shorter than hex */
1352 d = (char*)uv_to_utf8((U8*)d,
1353 (UV)scan_hex(s + 1, e - s - 1, &len));
1357 UV uv = (UV)scan_hex(s, 2, &len);
1358 if (utf && PL_lex_inwhat == OP_TRANS &&
1359 utf != (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1361 d = (char*)uv_to_utf8((U8*)d, uv); /* doing a CU or UC */
1364 if (uv >= 127 && UTF) {
1366 if (ckWARN(WARN_UTF8))
1367 Perl_warner(aTHX_ WARN_UTF8,
1368 "\\x%.*s will produce malformed UTF-8 character; use \\x{%.*s} for that",
1377 /* \N{latin small letter a} is a named character */
1381 char* e = strchr(s, '}');
1390 yyerror("Missing right brace on \\N{}");
1394 res = newSVpvn(s + 1, e - s - 1);
1395 res = new_constant( Nullch, 0, "charnames",
1396 res, Nullsv, "\\N{...}" );
1397 str = SvPV(res,len);
1398 if (len > e - s + 4) {
1399 char *odest = SvPVX(sv);
1401 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1402 d = SvPVX(sv) + (d - odest);
1404 Copy(str, d, len, char);
1411 yyerror("Missing braces on \\N{}");
1414 /* \c is a control character */
1428 /* printf-style backslashes, formfeeds, newlines, etc */
1446 *d++ = '\047'; /* CP 1047 */
1449 *d++ = '\057'; /* CP 1047 */
1463 } /* end if (backslash) */
1466 } /* while loop to process each character */
1468 /* terminate the string and set up the sv */
1470 SvCUR_set(sv, d - SvPVX(sv));
1473 /* shrink the sv if we allocated more than we used */
1474 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1475 SvLEN_set(sv, SvCUR(sv) + 1);
1476 Renew(SvPVX(sv), SvLEN(sv), char);
1479 /* return the substring (via yylval) only if we parsed anything */
1480 if (s > PL_bufptr) {
1481 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1482 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1484 ( PL_lex_inwhat == OP_TRANS
1486 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1489 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1496 * Returns TRUE if there's more to the expression (e.g., a subscript),
1499 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1501 * ->[ and ->{ return TRUE
1502 * { and [ outside a pattern are always subscripts, so return TRUE
1503 * if we're outside a pattern and it's not { or [, then return FALSE
1504 * if we're in a pattern and the first char is a {
1505 * {4,5} (any digits around the comma) returns FALSE
1506 * if we're in a pattern and the first char is a [
1508 * [SOMETHING] has a funky algorithm to decide whether it's a
1509 * character class or not. It has to deal with things like
1510 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1511 * anything else returns TRUE
1514 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1517 S_intuit_more(pTHX_ register char *s)
1519 if (PL_lex_brackets)
1521 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1523 if (*s != '{' && *s != '[')
1528 /* In a pattern, so maybe we have {n,m}. */
1545 /* On the other hand, maybe we have a character class */
1548 if (*s == ']' || *s == '^')
1551 /* this is terrifying, and it works */
1552 int weight = 2; /* let's weigh the evidence */
1554 unsigned char un_char = 255, last_un_char;
1555 char *send = strchr(s,']');
1556 char tmpbuf[sizeof PL_tokenbuf * 4];
1558 if (!send) /* has to be an expression */
1561 Zero(seen,256,char);
1564 else if (isDIGIT(*s)) {
1566 if (isDIGIT(s[1]) && s[2] == ']')
1572 for (; s < send; s++) {
1573 last_un_char = un_char;
1574 un_char = (unsigned char)*s;
1579 weight -= seen[un_char] * 10;
1580 if (isALNUM_lazy(s+1)) {
1581 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1582 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1587 else if (*s == '$' && s[1] &&
1588 strchr("[#!%*<>()-=",s[1])) {
1589 if (/*{*/ strchr("])} =",s[2]))
1598 if (strchr("wds]",s[1]))
1600 else if (seen['\''] || seen['"'])
1602 else if (strchr("rnftbxcav",s[1]))
1604 else if (isDIGIT(s[1])) {
1606 while (s[1] && isDIGIT(s[1]))
1616 if (strchr("aA01! ",last_un_char))
1618 if (strchr("zZ79~",s[1]))
1620 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1621 weight -= 5; /* cope with negative subscript */
1624 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1625 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1630 if (keyword(tmpbuf, d - tmpbuf))
1633 if (un_char == last_un_char + 1)
1635 weight -= seen[un_char];
1640 if (weight >= 0) /* probably a character class */
1650 * Does all the checking to disambiguate
1652 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1653 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1655 * First argument is the stuff after the first token, e.g. "bar".
1657 * Not a method if bar is a filehandle.
1658 * Not a method if foo is a subroutine prototyped to take a filehandle.
1659 * Not a method if it's really "Foo $bar"
1660 * Method if it's "foo $bar"
1661 * Not a method if it's really "print foo $bar"
1662 * Method if it's really "foo package::" (interpreted as package->foo)
1663 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1664 * Not a method if bar is a filehandle or package, but is quotd with
1669 S_intuit_method(pTHX_ char *start, GV *gv)
1671 char *s = start + (*start == '$');
1672 char tmpbuf[sizeof PL_tokenbuf];
1680 if ((cv = GvCVu(gv))) {
1681 char *proto = SvPVX(cv);
1691 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1692 /* start is the beginning of the possible filehandle/object,
1693 * and s is the end of it
1694 * tmpbuf is a copy of it
1697 if (*start == '$') {
1698 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1703 return *s == '(' ? FUNCMETH : METHOD;
1705 if (!keyword(tmpbuf, len)) {
1706 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1711 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1712 if (indirgv && GvCVu(indirgv))
1714 /* filehandle or package name makes it a method */
1715 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1717 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1718 return 0; /* no assumptions -- "=>" quotes bearword */
1720 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1721 newSVpvn(tmpbuf,len));
1722 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1726 return *s == '(' ? FUNCMETH : METHOD;
1734 * Return a string of Perl code to load the debugger. If PERL5DB
1735 * is set, it will return the contents of that, otherwise a
1736 * compile-time require of perl5db.pl.
1743 char *pdb = PerlEnv_getenv("PERL5DB");
1747 SETERRNO(0,SS$_NORMAL);
1748 return "BEGIN { require 'perl5db.pl' }";
1754 /* Encoded script support. filter_add() effectively inserts a
1755 * 'pre-processing' function into the current source input stream.
1756 * Note that the filter function only applies to the current source file
1757 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1759 * The datasv parameter (which may be NULL) can be used to pass
1760 * private data to this instance of the filter. The filter function
1761 * can recover the SV using the FILTER_DATA macro and use it to
1762 * store private buffers and state information.
1764 * The supplied datasv parameter is upgraded to a PVIO type
1765 * and the IoDIRP field is used to store the function pointer,
1766 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1767 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1768 * private use must be set using malloc'd pointers.
1772 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1777 if (!PL_rsfp_filters)
1778 PL_rsfp_filters = newAV();
1780 datasv = NEWSV(255,0);
1781 if (!SvUPGRADE(datasv, SVt_PVIO))
1782 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1783 IoDIRP(datasv) = (DIR*)funcp; /* stash funcp into spare field */
1784 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1785 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1786 funcp, SvPV_nolen(datasv)));
1787 av_unshift(PL_rsfp_filters, 1);
1788 av_store(PL_rsfp_filters, 0, datasv) ;
1793 /* Delete most recently added instance of this filter function. */
1795 Perl_filter_del(pTHX_ filter_t funcp)
1798 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1799 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1801 /* if filter is on top of stack (usual case) just pop it off */
1802 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1803 if (IoDIRP(datasv) == (DIR*)funcp) {
1804 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1805 IoDIRP(datasv) = (DIR*)NULL;
1806 sv_free(av_pop(PL_rsfp_filters));
1810 /* we need to search for the correct entry and clear it */
1811 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1815 /* Invoke the n'th filter function for the current rsfp. */
1817 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1820 /* 0 = read one text line */
1825 if (!PL_rsfp_filters)
1827 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1828 /* Provide a default input filter to make life easy. */
1829 /* Note that we append to the line. This is handy. */
1830 DEBUG_P(PerlIO_printf(Perl_debug_log,
1831 "filter_read %d: from rsfp\n", idx));
1835 int old_len = SvCUR(buf_sv) ;
1837 /* ensure buf_sv is large enough */
1838 SvGROW(buf_sv, old_len + maxlen) ;
1839 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1840 if (PerlIO_error(PL_rsfp))
1841 return -1; /* error */
1843 return 0 ; /* end of file */
1845 SvCUR_set(buf_sv, old_len + len) ;
1848 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1849 if (PerlIO_error(PL_rsfp))
1850 return -1; /* error */
1852 return 0 ; /* end of file */
1855 return SvCUR(buf_sv);
1857 /* Skip this filter slot if filter has been deleted */
1858 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1859 DEBUG_P(PerlIO_printf(Perl_debug_log,
1860 "filter_read %d: skipped (filter deleted)\n",
1862 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1864 /* Get function pointer hidden within datasv */
1865 funcp = (filter_t)IoDIRP(datasv);
1866 DEBUG_P(PerlIO_printf(Perl_debug_log,
1867 "filter_read %d: via function %p (%s)\n",
1868 idx, funcp, SvPV_nolen(datasv)));
1869 /* Call function. The function is expected to */
1870 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1871 /* Return: <0:error, =0:eof, >0:not eof */
1872 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1876 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1879 if (!PL_rsfp_filters) {
1880 filter_add(win32_textfilter,NULL);
1883 if (PL_rsfp_filters) {
1886 SvCUR_set(sv, 0); /* start with empty line */
1887 if (FILTER_READ(0, sv, 0) > 0)
1888 return ( SvPVX(sv) ) ;
1893 return (sv_gets(sv, fp, append));
1898 static char* exp_name[] =
1899 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
1900 "ATTRTERM", "TERMBLOCK"
1907 Works out what to call the token just pulled out of the input
1908 stream. The yacc parser takes care of taking the ops we return and
1909 stitching them into a tree.
1915 if read an identifier
1916 if we're in a my declaration
1917 croak if they tried to say my($foo::bar)
1918 build the ops for a my() declaration
1919 if it's an access to a my() variable
1920 are we in a sort block?
1921 croak if my($a); $a <=> $b
1922 build ops for access to a my() variable
1923 if in a dq string, and they've said @foo and we can't find @foo
1925 build ops for a bareword
1926 if we already built the token before, use it.
1930 #ifdef USE_PURE_BISON
1931 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
1944 #ifdef USE_PURE_BISON
1945 yylval_pointer = lvalp;
1946 yychar_pointer = lcharp;
1949 /* check if there's an identifier for us to look at */
1950 if (PL_pending_ident) {
1951 /* pit holds the identifier we read and pending_ident is reset */
1952 char pit = PL_pending_ident;
1953 PL_pending_ident = 0;
1955 /* if we're in a my(), we can't allow dynamics here.
1956 $foo'bar has already been turned into $foo::bar, so
1957 just check for colons.
1959 if it's a legal name, the OP is a PADANY.
1962 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
1963 tmp = pad_allocmy(PL_tokenbuf);
1966 if (strchr(PL_tokenbuf,':'))
1967 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
1969 yylval.opval = newOP(OP_PADANY, 0);
1970 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
1976 build the ops for accesses to a my() variable.
1978 Deny my($a) or my($b) in a sort block, *if* $a or $b is
1979 then used in a comparison. This catches most, but not
1980 all cases. For instance, it catches
1981 sort { my($a); $a <=> $b }
1983 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
1984 (although why you'd do that is anyone's guess).
1987 if (!strchr(PL_tokenbuf,':')) {
1989 /* Check for single character per-thread SVs */
1990 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
1991 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
1992 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
1994 yylval.opval = newOP(OP_THREADSV, 0);
1995 yylval.opval->op_targ = tmp;
1998 #endif /* USE_THREADS */
1999 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2000 /* might be an "our" variable" */
2001 if (SvFLAGS(AvARRAY(PL_comppad_name)[tmp]) & SVpad_OUR) {
2002 /* build ops for a bareword */
2003 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2004 yylval.opval->op_private = OPpCONST_ENTERED;
2005 gv_fetchpv(PL_tokenbuf+1,
2007 ? (GV_ADDMULTI | GV_ADDINEVAL | GV_ADDOUR)
2010 ((PL_tokenbuf[0] == '$') ? SVt_PV
2011 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2016 /* if it's a sort block and they're naming $a or $b */
2017 if (PL_last_lop_op == OP_SORT &&
2018 PL_tokenbuf[0] == '$' &&
2019 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2022 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2023 d < PL_bufend && *d != '\n';
2026 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2027 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2033 yylval.opval = newOP(OP_PADANY, 0);
2034 yylval.opval->op_targ = tmp;
2040 Whine if they've said @foo in a doublequoted string,
2041 and @foo isn't a variable we can find in the symbol
2044 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2045 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2046 if (!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2047 yyerror(Perl_form(aTHX_ "In string, %s now must be written as \\%s",
2048 PL_tokenbuf, PL_tokenbuf));
2051 /* build ops for a bareword */
2052 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2053 yylval.opval->op_private = OPpCONST_ENTERED;
2054 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2055 ((PL_tokenbuf[0] == '$') ? SVt_PV
2056 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2061 /* no identifier pending identification */
2063 switch (PL_lex_state) {
2065 case LEX_NORMAL: /* Some compilers will produce faster */
2066 case LEX_INTERPNORMAL: /* code if we comment these out. */
2070 /* when we've already built the next token, just pull it out of the queue */
2073 yylval = PL_nextval[PL_nexttoke];
2075 PL_lex_state = PL_lex_defer;
2076 PL_expect = PL_lex_expect;
2077 PL_lex_defer = LEX_NORMAL;
2079 return(PL_nexttype[PL_nexttoke]);
2081 /* interpolated case modifiers like \L \U, including \Q and \E.
2082 when we get here, PL_bufptr is at the \
2084 case LEX_INTERPCASEMOD:
2086 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2087 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2089 /* handle \E or end of string */
2090 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2094 if (PL_lex_casemods) {
2095 oldmod = PL_lex_casestack[--PL_lex_casemods];
2096 PL_lex_casestack[PL_lex_casemods] = '\0';
2098 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2100 PL_lex_state = LEX_INTERPCONCAT;
2104 if (PL_bufptr != PL_bufend)
2106 PL_lex_state = LEX_INTERPCONCAT;
2111 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2112 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2113 if (strchr("LU", *s) &&
2114 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2116 PL_lex_casestack[--PL_lex_casemods] = '\0';
2119 if (PL_lex_casemods > 10) {
2120 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2121 if (newlb != PL_lex_casestack) {
2123 PL_lex_casestack = newlb;
2126 PL_lex_casestack[PL_lex_casemods++] = *s;
2127 PL_lex_casestack[PL_lex_casemods] = '\0';
2128 PL_lex_state = LEX_INTERPCONCAT;
2129 PL_nextval[PL_nexttoke].ival = 0;
2132 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2134 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2136 PL_nextval[PL_nexttoke].ival = OP_LC;
2138 PL_nextval[PL_nexttoke].ival = OP_UC;
2140 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2142 Perl_croak(aTHX_ "panic: yylex");
2145 if (PL_lex_starts) {
2154 case LEX_INTERPPUSH:
2155 return sublex_push();
2157 case LEX_INTERPSTART:
2158 if (PL_bufptr == PL_bufend)
2159 return sublex_done();
2161 PL_lex_dojoin = (*PL_bufptr == '@');
2162 PL_lex_state = LEX_INTERPNORMAL;
2163 if (PL_lex_dojoin) {
2164 PL_nextval[PL_nexttoke].ival = 0;
2167 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2168 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2169 force_next(PRIVATEREF);
2171 force_ident("\"", '$');
2172 #endif /* USE_THREADS */
2173 PL_nextval[PL_nexttoke].ival = 0;
2175 PL_nextval[PL_nexttoke].ival = 0;
2177 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2180 if (PL_lex_starts++) {
2186 case LEX_INTERPENDMAYBE:
2187 if (intuit_more(PL_bufptr)) {
2188 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2194 if (PL_lex_dojoin) {
2195 PL_lex_dojoin = FALSE;
2196 PL_lex_state = LEX_INTERPCONCAT;
2199 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2200 && SvEVALED(PL_lex_repl))
2202 if (PL_bufptr != PL_bufend)
2203 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2204 PL_lex_repl = Nullsv;
2207 case LEX_INTERPCONCAT:
2209 if (PL_lex_brackets)
2210 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2212 if (PL_bufptr == PL_bufend)
2213 return sublex_done();
2215 if (SvIVX(PL_linestr) == '\'') {
2216 SV *sv = newSVsv(PL_linestr);
2219 else if ( PL_hints & HINT_NEW_RE )
2220 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2221 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2225 s = scan_const(PL_bufptr);
2227 PL_lex_state = LEX_INTERPCASEMOD;
2229 PL_lex_state = LEX_INTERPSTART;
2232 if (s != PL_bufptr) {
2233 PL_nextval[PL_nexttoke] = yylval;
2236 if (PL_lex_starts++)
2246 PL_lex_state = LEX_NORMAL;
2247 s = scan_formline(PL_bufptr);
2248 if (!PL_lex_formbrack)
2254 PL_oldoldbufptr = PL_oldbufptr;
2257 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2258 exp_name[PL_expect], s);
2264 if (isIDFIRST_lazy(s))
2266 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2269 goto fake_eof; /* emulate EOF on ^D or ^Z */
2274 if (PL_lex_brackets)
2275 yyerror("Missing right curly or square bracket");
2278 if (s++ < PL_bufend)
2279 goto retry; /* ignore stray nulls */
2282 if (!PL_in_eval && !PL_preambled) {
2283 PL_preambled = TRUE;
2284 sv_setpv(PL_linestr,incl_perldb());
2285 if (SvCUR(PL_linestr))
2286 sv_catpv(PL_linestr,";");
2288 while(AvFILLp(PL_preambleav) >= 0) {
2289 SV *tmpsv = av_shift(PL_preambleav);
2290 sv_catsv(PL_linestr, tmpsv);
2291 sv_catpv(PL_linestr, ";");
2294 sv_free((SV*)PL_preambleav);
2295 PL_preambleav = NULL;
2297 if (PL_minus_n || PL_minus_p) {
2298 sv_catpv(PL_linestr, "LINE: while (<>) {");
2300 sv_catpv(PL_linestr,"chomp;");
2302 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2304 GvIMPORTED_AV_on(gv);
2306 if (strchr("/'\"", *PL_splitstr)
2307 && strchr(PL_splitstr + 1, *PL_splitstr))
2308 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2311 s = "'~#\200\1'"; /* surely one char is unused...*/
2312 while (s[1] && strchr(PL_splitstr, *s)) s++;
2314 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2315 "q" + (delim == '\''), delim);
2316 for (s = PL_splitstr; *s; s++) {
2318 sv_catpvn(PL_linestr, "\\", 1);
2319 sv_catpvn(PL_linestr, s, 1);
2321 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2325 sv_catpv(PL_linestr,"@F=split(' ');");
2328 sv_catpv(PL_linestr, "\n");
2329 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2330 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2331 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2332 SV *sv = NEWSV(85,0);
2334 sv_upgrade(sv, SVt_PVMG);
2335 sv_setsv(sv,PL_linestr);
2336 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2341 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2344 if (PL_preprocess && !PL_in_eval)
2345 (void)PerlProc_pclose(PL_rsfp);
2346 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2347 PerlIO_clearerr(PL_rsfp);
2349 (void)PerlIO_close(PL_rsfp);
2351 PL_doextract = FALSE;
2353 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2354 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2355 sv_catpv(PL_linestr,";}");
2356 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2357 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2358 PL_minus_n = PL_minus_p = 0;
2361 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2362 sv_setpv(PL_linestr,"");
2363 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2366 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2367 PL_doextract = FALSE;
2369 /* Incest with pod. */
2370 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2371 sv_setpv(PL_linestr, "");
2372 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2373 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2374 PL_doextract = FALSE;
2378 } while (PL_doextract);
2379 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2380 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2381 SV *sv = NEWSV(85,0);
2383 sv_upgrade(sv, SVt_PVMG);
2384 sv_setsv(sv,PL_linestr);
2385 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2387 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2388 if (CopLINE(PL_curcop) == 1) {
2389 while (s < PL_bufend && isSPACE(*s))
2391 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2395 if (*s == '#' && *(s+1) == '!')
2397 #ifdef ALTERNATE_SHEBANG
2399 static char as[] = ALTERNATE_SHEBANG;
2400 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2401 d = s + (sizeof(as) - 1);
2403 #endif /* ALTERNATE_SHEBANG */
2412 while (*d && !isSPACE(*d))
2416 #ifdef ARG_ZERO_IS_SCRIPT
2417 if (ipathend > ipath) {
2419 * HP-UX (at least) sets argv[0] to the script name,
2420 * which makes $^X incorrect. And Digital UNIX and Linux,
2421 * at least, set argv[0] to the basename of the Perl
2422 * interpreter. So, having found "#!", we'll set it right.
2424 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2425 assert(SvPOK(x) || SvGMAGICAL(x));
2426 if (sv_eq(x, CopFILESV(PL_curcop))) {
2427 sv_setpvn(x, ipath, ipathend - ipath);
2430 TAINT_NOT; /* $^X is always tainted, but that's OK */
2432 #endif /* ARG_ZERO_IS_SCRIPT */
2437 d = instr(s,"perl -");
2439 d = instr(s,"perl");
2441 /* avoid getting into infinite loops when shebang
2442 * line contains "Perl" rather than "perl" */
2444 for (d = ipathend-4; d >= ipath; --d) {
2445 if ((*d == 'p' || *d == 'P')
2446 && !ibcmp(d, "perl", 4))
2456 #ifdef ALTERNATE_SHEBANG
2458 * If the ALTERNATE_SHEBANG on this system starts with a
2459 * character that can be part of a Perl expression, then if
2460 * we see it but not "perl", we're probably looking at the
2461 * start of Perl code, not a request to hand off to some
2462 * other interpreter. Similarly, if "perl" is there, but
2463 * not in the first 'word' of the line, we assume the line
2464 * contains the start of the Perl program.
2466 if (d && *s != '#') {
2468 while (*c && !strchr("; \t\r\n\f\v#", *c))
2471 d = Nullch; /* "perl" not in first word; ignore */
2473 *s = '#'; /* Don't try to parse shebang line */
2475 #endif /* ALTERNATE_SHEBANG */
2480 !instr(s,"indir") &&
2481 instr(PL_origargv[0],"perl"))
2487 while (s < PL_bufend && isSPACE(*s))
2489 if (s < PL_bufend) {
2490 Newz(899,newargv,PL_origargc+3,char*);
2492 while (s < PL_bufend && !isSPACE(*s))
2495 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2498 newargv = PL_origargv;
2500 PerlProc_execv(ipath, newargv);
2501 Perl_croak(aTHX_ "Can't exec %s", ipath);
2504 U32 oldpdb = PL_perldb;
2505 bool oldn = PL_minus_n;
2506 bool oldp = PL_minus_p;
2508 while (*d && !isSPACE(*d)) d++;
2509 while (*d == ' ' || *d == '\t') d++;
2513 if (*d == 'M' || *d == 'm') {
2515 while (*d && !isSPACE(*d)) d++;
2516 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2519 d = moreswitches(d);
2521 if (PERLDB_LINE && !oldpdb ||
2522 ( PL_minus_n || PL_minus_p ) && !(oldn || oldp) )
2523 /* if we have already added "LINE: while (<>) {",
2524 we must not do it again */
2526 sv_setpv(PL_linestr, "");
2527 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2528 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2529 PL_preambled = FALSE;
2531 (void)gv_fetchfile(PL_origfilename);
2538 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2540 PL_lex_state = LEX_FORMLINE;
2545 #ifdef PERL_STRICT_CR
2546 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2548 "(Maybe you didn't strip carriage returns after a network transfer?)\n");
2550 case ' ': case '\t': case '\f': case 013:
2555 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2557 while (s < d && *s != '\n')
2562 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2564 PL_lex_state = LEX_FORMLINE;
2574 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2579 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2582 if (strnEQ(s,"=>",2)) {
2583 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2584 OPERATOR('-'); /* unary minus */
2586 PL_last_uni = PL_oldbufptr;
2587 PL_last_lop_op = OP_FTEREAD; /* good enough */
2589 case 'r': FTST(OP_FTEREAD);
2590 case 'w': FTST(OP_FTEWRITE);
2591 case 'x': FTST(OP_FTEEXEC);
2592 case 'o': FTST(OP_FTEOWNED);
2593 case 'R': FTST(OP_FTRREAD);
2594 case 'W': FTST(OP_FTRWRITE);
2595 case 'X': FTST(OP_FTREXEC);
2596 case 'O': FTST(OP_FTROWNED);
2597 case 'e': FTST(OP_FTIS);
2598 case 'z': FTST(OP_FTZERO);
2599 case 's': FTST(OP_FTSIZE);
2600 case 'f': FTST(OP_FTFILE);
2601 case 'd': FTST(OP_FTDIR);
2602 case 'l': FTST(OP_FTLINK);
2603 case 'p': FTST(OP_FTPIPE);
2604 case 'S': FTST(OP_FTSOCK);
2605 case 'u': FTST(OP_FTSUID);
2606 case 'g': FTST(OP_FTSGID);
2607 case 'k': FTST(OP_FTSVTX);
2608 case 'b': FTST(OP_FTBLK);
2609 case 'c': FTST(OP_FTCHR);
2610 case 't': FTST(OP_FTTTY);
2611 case 'T': FTST(OP_FTTEXT);
2612 case 'B': FTST(OP_FTBINARY);
2613 case 'M': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTMTIME);
2614 case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME);
2615 case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME);
2617 Perl_croak(aTHX_ "Unrecognized file test: -%c", (int)tmp);
2624 if (PL_expect == XOPERATOR)
2629 else if (*s == '>') {
2632 if (isIDFIRST_lazy(s)) {
2633 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2641 if (PL_expect == XOPERATOR)
2644 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2646 OPERATOR('-'); /* unary minus */
2653 if (PL_expect == XOPERATOR)
2658 if (PL_expect == XOPERATOR)
2661 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2667 if (PL_expect != XOPERATOR) {
2668 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2669 PL_expect = XOPERATOR;
2670 force_ident(PL_tokenbuf, '*');
2683 if (PL_expect == XOPERATOR) {
2687 PL_tokenbuf[0] = '%';
2688 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2689 if (!PL_tokenbuf[1]) {
2691 yyerror("Final % should be \\% or %name");
2694 PL_pending_ident = '%';
2713 switch (PL_expect) {
2716 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2718 PL_bufptr = s; /* update in case we back off */
2724 PL_expect = XTERMBLOCK;
2728 while (isIDFIRST_lazy(s)) {
2729 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2731 d = scan_str(d,TRUE,TRUE);
2734 SvREFCNT_dec(PL_lex_stuff);
2735 PL_lex_stuff = Nullsv;
2737 /* MUST advance bufptr here to avoid bogus
2738 "at end of line" context messages from yyerror().
2740 PL_bufptr = s + len;
2741 yyerror("Unterminated attribute parameter in attribute list");
2744 return 0; /* EOF indicator */
2748 SV *sv = newSVpvn(s, len);
2749 sv_catsv(sv, PL_lex_stuff);
2750 attrs = append_elem(OP_LIST, attrs,
2751 newSVOP(OP_CONST, 0, sv));
2752 SvREFCNT_dec(PL_lex_stuff);
2753 PL_lex_stuff = Nullsv;
2756 attrs = append_elem(OP_LIST, attrs,
2757 newSVOP(OP_CONST, 0,
2764 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}' for vi */
2765 if (*s != ';' && *s != tmp) {
2766 char q = ((*s == '\'') ? '"' : '\'');
2767 /* If here for an expression, and parsed no attrs, back off. */
2768 if (tmp == '=' && !attrs) {
2772 /* MUST advance bufptr here to avoid bogus "at end of line"
2773 context messages from yyerror().
2777 yyerror("Unterminated attribute list");
2779 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
2786 PL_nextval[PL_nexttoke].opval = attrs;
2794 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
2795 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
2800 if (CopLINE(PL_curcop) < PL_copline)
2801 PL_copline = CopLINE(PL_curcop);
2812 if (PL_lex_brackets <= 0)
2813 yyerror("Unmatched right square bracket");
2816 if (PL_lex_state == LEX_INTERPNORMAL) {
2817 if (PL_lex_brackets == 0) {
2818 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
2819 PL_lex_state = LEX_INTERPEND;
2826 if (PL_lex_brackets > 100) {
2827 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
2828 if (newlb != PL_lex_brackstack) {
2830 PL_lex_brackstack = newlb;
2833 switch (PL_expect) {
2835 if (PL_lex_formbrack) {
2839 if (PL_oldoldbufptr == PL_last_lop)
2840 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2842 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2843 OPERATOR(HASHBRACK);
2845 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2848 PL_tokenbuf[0] = '\0';
2849 if (d < PL_bufend && *d == '-') {
2850 PL_tokenbuf[0] = '-';
2852 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2855 if (d < PL_bufend && isIDFIRST_lazy(d)) {
2856 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
2858 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2861 char minus = (PL_tokenbuf[0] == '-');
2862 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
2870 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
2875 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2880 if (PL_oldoldbufptr == PL_last_lop)
2881 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2883 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2886 OPERATOR(HASHBRACK);
2887 /* This hack serves to disambiguate a pair of curlies
2888 * as being a block or an anon hash. Normally, expectation
2889 * determines that, but in cases where we're not in a
2890 * position to expect anything in particular (like inside
2891 * eval"") we have to resolve the ambiguity. This code
2892 * covers the case where the first term in the curlies is a
2893 * quoted string. Most other cases need to be explicitly
2894 * disambiguated by prepending a `+' before the opening
2895 * curly in order to force resolution as an anon hash.
2897 * XXX should probably propagate the outer expectation
2898 * into eval"" to rely less on this hack, but that could
2899 * potentially break current behavior of eval"".
2903 if (*s == '\'' || *s == '"' || *s == '`') {
2904 /* common case: get past first string, handling escapes */
2905 for (t++; t < PL_bufend && *t != *s;)
2906 if (*t++ == '\\' && (*t == '\\' || *t == *s))
2910 else if (*s == 'q') {
2913 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
2914 && !isALNUM(*t)))) {
2916 char open, close, term;
2919 while (t < PL_bufend && isSPACE(*t))
2923 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
2927 for (t++; t < PL_bufend; t++) {
2928 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
2930 else if (*t == open)
2934 for (t++; t < PL_bufend; t++) {
2935 if (*t == '\\' && t+1 < PL_bufend)
2937 else if (*t == close && --brackets <= 0)
2939 else if (*t == open)
2945 else if (isIDFIRST_lazy(s)) {
2946 for (t++; t < PL_bufend && isALNUM_lazy(t); t++) ;
2948 while (t < PL_bufend && isSPACE(*t))
2950 /* if comma follows first term, call it an anon hash */
2951 /* XXX it could be a comma expression with loop modifiers */
2952 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
2953 || (*t == '=' && t[1] == '>')))
2954 OPERATOR(HASHBRACK);
2955 if (PL_expect == XREF)
2958 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
2964 yylval.ival = CopLINE(PL_curcop);
2965 if (isSPACE(*s) || *s == '#')
2966 PL_copline = NOLINE; /* invalidate current command line number */
2971 if (PL_lex_brackets <= 0)
2972 yyerror("Unmatched right curly bracket");
2974 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
2975 if (PL_lex_brackets < PL_lex_formbrack)
2976 PL_lex_formbrack = 0;
2977 if (PL_lex_state == LEX_INTERPNORMAL) {
2978 if (PL_lex_brackets == 0) {
2979 if (PL_lex_fakebrack) {
2980 PL_lex_state = LEX_INTERPEND;
2982 return yylex(); /* ignore fake brackets */
2984 if (*s == '-' && s[1] == '>')
2985 PL_lex_state = LEX_INTERPENDMAYBE;
2986 else if (*s != '[' && *s != '{')
2987 PL_lex_state = LEX_INTERPEND;
2990 if (PL_lex_brackets < PL_lex_fakebrack) {
2992 PL_lex_fakebrack = 0;
2993 return yylex(); /* ignore fake brackets */
3003 if (PL_expect == XOPERATOR) {
3004 if (ckWARN(WARN_SEMICOLON) && isIDFIRST_lazy(s) && PL_bufptr == PL_linestart) {
3005 CopLINE_dec(PL_curcop);
3006 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3007 CopLINE_inc(PL_curcop);
3012 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3014 PL_expect = XOPERATOR;
3015 force_ident(PL_tokenbuf, '&');
3019 yylval.ival = (OPpENTERSUB_AMPER<<8);
3038 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3039 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3041 if (PL_expect == XSTATE && isALPHA(tmp) &&
3042 (s == PL_linestart+1 || s[-2] == '\n') )
3044 if (PL_in_eval && !PL_rsfp) {
3049 if (strnEQ(s,"=cut",4)) {
3063 PL_doextract = TRUE;
3066 if (PL_lex_brackets < PL_lex_formbrack) {
3068 #ifdef PERL_STRICT_CR
3069 for (t = s; *t == ' ' || *t == '\t'; t++) ;
3071 for (t = s; *t == ' ' || *t == '\t' || *t == '\r'; t++) ;
3073 if (*t == '\n' || *t == '#') {
3091 if (PL_expect != XOPERATOR) {
3092 if (s[1] != '<' && !strchr(s,'>'))
3095 s = scan_heredoc(s);
3097 s = scan_inputsymbol(s);
3098 TERM(sublex_start());
3103 SHop(OP_LEFT_SHIFT);
3117 SHop(OP_RIGHT_SHIFT);
3126 if (PL_expect == XOPERATOR) {
3127 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3130 return ','; /* grandfather non-comma-format format */
3134 if (s[1] == '#' && (isIDFIRST_lazy(s+2) || strchr("{$:+-", s[2]))) {
3135 PL_tokenbuf[0] = '@';
3136 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3137 sizeof PL_tokenbuf - 1, FALSE);
3138 if (PL_expect == XOPERATOR)
3139 no_op("Array length", s);
3140 if (!PL_tokenbuf[1])
3142 PL_expect = XOPERATOR;
3143 PL_pending_ident = '#';
3147 PL_tokenbuf[0] = '$';
3148 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3149 sizeof PL_tokenbuf - 1, FALSE);
3150 if (PL_expect == XOPERATOR)
3152 if (!PL_tokenbuf[1]) {
3154 yyerror("Final $ should be \\$ or $name");
3158 /* This kludge not intended to be bulletproof. */
3159 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3160 yylval.opval = newSVOP(OP_CONST, 0,
3161 newSViv((IV)PL_compiling.cop_arybase));
3162 yylval.opval->op_private = OPpCONST_ARYBASE;
3168 if (PL_lex_state == LEX_NORMAL)
3171 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3174 PL_tokenbuf[0] = '@';
3175 if (ckWARN(WARN_SYNTAX)) {
3177 isSPACE(*t) || isALNUM_lazy(t) || *t == '$';
3180 PL_bufptr = skipspace(PL_bufptr);
3181 while (t < PL_bufend && *t != ']')
3183 Perl_warner(aTHX_ WARN_SYNTAX,
3184 "Multidimensional syntax %.*s not supported",
3185 (t - PL_bufptr) + 1, PL_bufptr);
3189 else if (*s == '{') {
3190 PL_tokenbuf[0] = '%';
3191 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3192 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3194 char tmpbuf[sizeof PL_tokenbuf];
3196 for (t++; isSPACE(*t); t++) ;
3197 if (isIDFIRST_lazy(t)) {
3198 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3199 for (; isSPACE(*t); t++) ;
3200 if (*t == ';' && get_cv(tmpbuf, FALSE))
3201 Perl_warner(aTHX_ WARN_SYNTAX,
3202 "You need to quote \"%s\"", tmpbuf);
3208 PL_expect = XOPERATOR;
3209 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3210 bool islop = (PL_last_lop == PL_oldoldbufptr);
3211 if (!islop || PL_last_lop_op == OP_GREPSTART)
3212 PL_expect = XOPERATOR;
3213 else if (strchr("$@\"'`q", *s))
3214 PL_expect = XTERM; /* e.g. print $fh "foo" */
3215 else if (strchr("&*<%", *s) && isIDFIRST_lazy(s+1))
3216 PL_expect = XTERM; /* e.g. print $fh &sub */
3217 else if (isIDFIRST_lazy(s)) {
3218 char tmpbuf[sizeof PL_tokenbuf];
3219 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3220 if (tmp = keyword(tmpbuf, len)) {
3221 /* binary operators exclude handle interpretations */
3233 PL_expect = XTERM; /* e.g. print $fh length() */
3238 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3239 if (gv && GvCVu(gv))
3240 PL_expect = XTERM; /* e.g. print $fh subr() */
3243 else if (isDIGIT(*s))
3244 PL_expect = XTERM; /* e.g. print $fh 3 */
3245 else if (*s == '.' && isDIGIT(s[1]))
3246 PL_expect = XTERM; /* e.g. print $fh .3 */
3247 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3248 PL_expect = XTERM; /* e.g. print $fh -1 */
3249 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3250 PL_expect = XTERM; /* print $fh <<"EOF" */
3252 PL_pending_ident = '$';
3256 if (PL_expect == XOPERATOR)
3258 PL_tokenbuf[0] = '@';
3259 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3260 if (!PL_tokenbuf[1]) {
3262 yyerror("Final @ should be \\@ or @name");
3265 if (PL_lex_state == LEX_NORMAL)
3267 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3269 PL_tokenbuf[0] = '%';
3271 /* Warn about @ where they meant $. */
3272 if (ckWARN(WARN_SYNTAX)) {
3273 if (*s == '[' || *s == '{') {
3275 while (*t && (isALNUM_lazy(t) || strchr(" \t$#+-'\"", *t)))
3277 if (*t == '}' || *t == ']') {
3279 PL_bufptr = skipspace(PL_bufptr);
3280 Perl_warner(aTHX_ WARN_SYNTAX,
3281 "Scalar value %.*s better written as $%.*s",
3282 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3287 PL_pending_ident = '@';
3290 case '/': /* may either be division or pattern */
3291 case '?': /* may either be conditional or pattern */
3292 if (PL_expect != XOPERATOR) {
3293 /* Disable warning on "study /blah/" */
3294 if (PL_oldoldbufptr == PL_last_uni
3295 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3296 || memNE(PL_last_uni, "study", 5) || isALNUM_lazy(PL_last_uni+5)))
3298 s = scan_pat(s,OP_MATCH);
3299 TERM(sublex_start());
3307 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3308 #ifdef PERL_STRICT_CR
3311 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3313 && (s == PL_linestart || s[-1] == '\n') )
3315 PL_lex_formbrack = 0;
3319 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3325 yylval.ival = OPf_SPECIAL;
3331 if (PL_expect != XOPERATOR)
3336 case '0': case '1': case '2': case '3': case '4':
3337 case '5': case '6': case '7': case '8': case '9':
3339 if (PL_expect == XOPERATOR)
3344 s = scan_str(s,FALSE,FALSE);
3345 if (PL_expect == XOPERATOR) {
3346 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3349 return ','; /* grandfather non-comma-format format */
3355 missingterm((char*)0);
3356 yylval.ival = OP_CONST;
3357 TERM(sublex_start());
3360 s = scan_str(s,FALSE,FALSE);
3361 if (PL_expect == XOPERATOR) {
3362 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3365 return ','; /* grandfather non-comma-format format */
3371 missingterm((char*)0);
3372 yylval.ival = OP_CONST;
3373 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3374 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3375 yylval.ival = OP_STRINGIFY;
3379 TERM(sublex_start());
3382 s = scan_str(s,FALSE,FALSE);
3383 if (PL_expect == XOPERATOR)
3384 no_op("Backticks",s);
3386 missingterm((char*)0);
3387 yylval.ival = OP_BACKTICK;
3389 TERM(sublex_start());
3393 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3394 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3396 if (PL_expect == XOPERATOR)
3397 no_op("Backslash",s);
3401 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3441 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3443 /* Some keywords can be followed by any delimiter, including ':' */
3444 tmp = (len == 1 && strchr("msyq", PL_tokenbuf[0]) ||
3445 len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3446 (PL_tokenbuf[0] == 'q' &&
3447 strchr("qwxr", PL_tokenbuf[1]))));
3449 /* x::* is just a word, unless x is "CORE" */
3450 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3454 while (d < PL_bufend && isSPACE(*d))
3455 d++; /* no comments skipped here, or s### is misparsed */
3457 /* Is this a label? */
3458 if (!tmp && PL_expect == XSTATE
3459 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3461 yylval.pval = savepv(PL_tokenbuf);
3466 /* Check for keywords */
3467 tmp = keyword(PL_tokenbuf, len);
3469 /* Is this a word before a => operator? */
3470 if (strnEQ(d,"=>",2)) {
3472 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3473 yylval.opval->op_private = OPpCONST_BARE;
3477 if (tmp < 0) { /* second-class keyword? */
3478 GV *ogv = Nullgv; /* override (winner) */
3479 GV *hgv = Nullgv; /* hidden (loser) */
3480 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3482 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3485 if (GvIMPORTED_CV(gv))
3487 else if (! CvMETHOD(cv))
3491 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3492 (gv = *gvp) != (GV*)&PL_sv_undef &&
3493 GvCVu(gv) && GvIMPORTED_CV(gv))
3499 tmp = 0; /* overridden by import or by GLOBAL */
3502 && -tmp==KEY_lock /* XXX generalizable kludge */
3504 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3506 tmp = 0; /* any sub overrides "weak" keyword */
3508 else { /* no override */
3512 if (ckWARN(WARN_AMBIGUOUS) && hgv
3513 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3514 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3515 "Ambiguous call resolved as CORE::%s(), %s",
3516 GvENAME(hgv), "qualify as such or use &");
3523 default: /* not a keyword */
3526 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3528 /* Get the rest if it looks like a package qualifier */
3530 if (*s == '\'' || *s == ':' && s[1] == ':') {
3532 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3535 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3536 *s == '\'' ? "'" : "::");
3540 if (PL_expect == XOPERATOR) {
3541 if (PL_bufptr == PL_linestart) {
3542 CopLINE_dec(PL_curcop);
3543 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3544 CopLINE_inc(PL_curcop);
3547 no_op("Bareword",s);
3550 /* Look for a subroutine with this name in current package,
3551 unless name is "Foo::", in which case Foo is a bearword
3552 (and a package name). */
3555 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3557 if (ckWARN(WARN_UNSAFE) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3558 Perl_warner(aTHX_ WARN_UNSAFE,
3559 "Bareword \"%s\" refers to nonexistent package",
3562 PL_tokenbuf[len] = '\0';
3569 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3572 /* if we saw a global override before, get the right name */
3575 sv = newSVpvn("CORE::GLOBAL::",14);
3576 sv_catpv(sv,PL_tokenbuf);
3579 sv = newSVpv(PL_tokenbuf,0);
3581 /* Presume this is going to be a bareword of some sort. */
3584 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3585 yylval.opval->op_private = OPpCONST_BARE;
3587 /* And if "Foo::", then that's what it certainly is. */
3592 /* See if it's the indirect object for a list operator. */
3594 if (PL_oldoldbufptr &&
3595 PL_oldoldbufptr < PL_bufptr &&
3596 (PL_oldoldbufptr == PL_last_lop
3597 || PL_oldoldbufptr == PL_last_uni) &&
3598 /* NO SKIPSPACE BEFORE HERE! */
3599 (PL_expect == XREF ||
3600 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3602 bool immediate_paren = *s == '(';
3604 /* (Now we can afford to cross potential line boundary.) */
3607 /* Two barewords in a row may indicate method call. */
3609 if ((isIDFIRST_lazy(s) || *s == '$') && (tmp=intuit_method(s,gv)))
3612 /* If not a declared subroutine, it's an indirect object. */
3613 /* (But it's an indir obj regardless for sort.) */
3615 if ((PL_last_lop_op == OP_SORT ||
3616 (!immediate_paren && (!gv || !GvCVu(gv)))) &&
3617 (PL_last_lop_op != OP_MAPSTART &&
3618 PL_last_lop_op != OP_GREPSTART))
3620 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3625 /* If followed by a paren, it's certainly a subroutine. */
3627 PL_expect = XOPERATOR;
3631 if (gv && GvCVu(gv)) {
3632 for (d = s + 1; *d == ' ' || *d == '\t'; d++) ;
3633 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3638 PL_nextval[PL_nexttoke].opval = yylval.opval;
3639 PL_expect = XOPERATOR;
3645 /* If followed by var or block, call it a method (unless sub) */
3647 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3648 PL_last_lop = PL_oldbufptr;
3649 PL_last_lop_op = OP_METHOD;
3653 /* If followed by a bareword, see if it looks like indir obj. */
3655 if ((isIDFIRST_lazy(s) || *s == '$') && (tmp = intuit_method(s,gv)))
3658 /* Not a method, so call it a subroutine (if defined) */
3660 if (gv && GvCVu(gv)) {
3662 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3663 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3664 "Ambiguous use of -%s resolved as -&%s()",
3665 PL_tokenbuf, PL_tokenbuf);
3666 /* Check for a constant sub */
3668 if ((sv = cv_const_sv(cv))) {
3670 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3671 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3672 yylval.opval->op_private = 0;
3676 /* Resolve to GV now. */
3677 op_free(yylval.opval);
3678 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3679 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3680 PL_last_lop = PL_oldbufptr;
3681 PL_last_lop_op = OP_ENTERSUB;
3682 /* Is there a prototype? */
3685 char *proto = SvPV((SV*)cv, len);
3688 if (strEQ(proto, "$"))
3690 if (*proto == '&' && *s == '{') {
3691 sv_setpv(PL_subname,"__ANON__");
3695 PL_nextval[PL_nexttoke].opval = yylval.opval;
3701 /* Call it a bare word */
3703 if (PL_hints & HINT_STRICT_SUBS)
3704 yylval.opval->op_private |= OPpCONST_STRICT;
3707 if (ckWARN(WARN_RESERVED)) {
3708 if (lastchar != '-') {
3709 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
3711 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
3718 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
3719 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3720 "Operator or semicolon missing before %c%s",
3721 lastchar, PL_tokenbuf);
3722 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3723 "Ambiguous use of %c resolved as operator %c",
3724 lastchar, lastchar);
3730 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3731 newSVpv(CopFILE(PL_curcop),0));
3735 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3736 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
3739 case KEY___PACKAGE__:
3740 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3742 ? newSVsv(PL_curstname)
3751 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
3752 char *pname = "main";
3753 if (PL_tokenbuf[2] == 'D')
3754 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
3755 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
3758 GvIOp(gv) = newIO();
3759 IoIFP(GvIOp(gv)) = PL_rsfp;
3760 #if defined(HAS_FCNTL) && defined(F_SETFD)
3762 int fd = PerlIO_fileno(PL_rsfp);
3763 fcntl(fd,F_SETFD,fd >= 3);
3766 /* Mark this internal pseudo-handle as clean */
3767 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
3769 IoTYPE(GvIOp(gv)) = '|';
3770 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
3771 IoTYPE(GvIOp(gv)) = '-';
3773 IoTYPE(GvIOp(gv)) = '<';
3785 if (PL_expect == XSTATE) {
3792 if (*s == ':' && s[1] == ':') {
3795 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3796 tmp = keyword(PL_tokenbuf, len);
3810 LOP(OP_ACCEPT,XTERM);
3816 LOP(OP_ATAN2,XTERM);
3825 LOP(OP_BLESS,XTERM);
3834 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
3851 if (!PL_cryptseen) {
3852 PL_cryptseen = TRUE;
3856 LOP(OP_CRYPT,XTERM);
3859 if (ckWARN(WARN_OCTAL)) {
3860 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
3861 if (*d != '0' && isDIGIT(*d))
3862 Perl_warner(aTHX_ WARN_OCTAL,
3863 "chmod: mode argument is missing initial 0");
3865 LOP(OP_CHMOD,XTERM);
3868 LOP(OP_CHOWN,XTERM);
3871 LOP(OP_CONNECT,XTERM);
3887 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3891 PL_hints |= HINT_BLOCK_SCOPE;
3901 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
3902 LOP(OP_DBMOPEN,XTERM);
3908 s = force_word(s,WORD,TRUE,FALSE,FALSE);
3915 yylval.ival = CopLINE(PL_curcop);
3929 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
3930 UNIBRACK(OP_ENTEREVAL);
3945 case KEY_endhostent:
3951 case KEY_endservent:
3954 case KEY_endprotoent:
3965 yylval.ival = CopLINE(PL_curcop);
3967 if (PL_expect == XSTATE && isIDFIRST_lazy(s)) {
3969 if ((PL_bufend - p) >= 3 &&
3970 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
3972 else if ((PL_bufend - p) >= 4 &&
3973 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
3976 if (isIDFIRST_lazy(p)) {
3977 p = scan_ident(p, PL_bufend,
3978 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3982 Perl_croak(aTHX_ "Missing $ on loop variable");
3987 LOP(OP_FORMLINE,XTERM);
3993 LOP(OP_FCNTL,XTERM);
3999 LOP(OP_FLOCK,XTERM);
4008 LOP(OP_GREPSTART, *s == '(' ? XTERM : XREF);
4011 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4026 case KEY_getpriority:
4027 LOP(OP_GETPRIORITY,XTERM);
4029 case KEY_getprotobyname:
4032 case KEY_getprotobynumber:
4033 LOP(OP_GPBYNUMBER,XTERM);
4035 case KEY_getprotoent:
4047 case KEY_getpeername:
4048 UNI(OP_GETPEERNAME);
4050 case KEY_gethostbyname:
4053 case KEY_gethostbyaddr:
4054 LOP(OP_GHBYADDR,XTERM);
4056 case KEY_gethostent:
4059 case KEY_getnetbyname:
4062 case KEY_getnetbyaddr:
4063 LOP(OP_GNBYADDR,XTERM);
4068 case KEY_getservbyname:
4069 LOP(OP_GSBYNAME,XTERM);
4071 case KEY_getservbyport:
4072 LOP(OP_GSBYPORT,XTERM);
4074 case KEY_getservent:
4077 case KEY_getsockname:
4078 UNI(OP_GETSOCKNAME);
4080 case KEY_getsockopt:
4081 LOP(OP_GSOCKOPT,XTERM);
4103 yylval.ival = CopLINE(PL_curcop);
4107 LOP(OP_INDEX,XTERM);
4113 LOP(OP_IOCTL,XTERM);
4125 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4157 LOP(OP_LISTEN,XTERM);
4166 s = scan_pat(s,OP_MATCH);
4167 TERM(sublex_start());
4170 LOP(OP_MAPSTART, *s == '(' ? XTERM : XREF);
4173 LOP(OP_MKDIR,XTERM);
4176 LOP(OP_MSGCTL,XTERM);
4179 LOP(OP_MSGGET,XTERM);
4182 LOP(OP_MSGRCV,XTERM);
4185 LOP(OP_MSGSND,XTERM);
4191 if (isIDFIRST_lazy(s)) {
4192 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4193 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4195 PL_in_my_stash = gv_stashpv(PL_tokenbuf, FALSE);
4196 if (!PL_in_my_stash) {
4199 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4207 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4214 if (PL_expect != XSTATE)
4215 yyerror("\"no\" not allowed in expression");
4216 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4217 s = force_version(s);
4222 if (*s == '(' || (s = skipspace(s), *s == '('))
4229 if (isIDFIRST_lazy(s)) {
4231 for (d = s; isALNUM_lazy(d); d++) ;
4233 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_AMBIGUOUS))
4234 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4235 "Precedence problem: open %.*s should be open(%.*s)",
4241 yylval.ival = OP_OR;
4251 LOP(OP_OPEN_DIR,XTERM);
4254 checkcomma(s,PL_tokenbuf,"filehandle");
4258 checkcomma(s,PL_tokenbuf,"filehandle");
4277 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4281 LOP(OP_PIPE_OP,XTERM);
4284 s = scan_str(s,FALSE,FALSE);
4286 missingterm((char*)0);
4287 yylval.ival = OP_CONST;
4288 TERM(sublex_start());
4294 s = scan_str(s,FALSE,FALSE);
4296 missingterm((char*)0);
4298 if (SvCUR(PL_lex_stuff)) {
4301 d = SvPV_force(PL_lex_stuff, len);
4303 for (; isSPACE(*d) && len; --len, ++d) ;
4306 if (!warned && ckWARN(WARN_SYNTAX)) {
4307 for (; !isSPACE(*d) && len; --len, ++d) {
4309 Perl_warner(aTHX_ WARN_SYNTAX,
4310 "Possible attempt to separate words with commas");
4313 else if (*d == '#') {
4314 Perl_warner(aTHX_ WARN_SYNTAX,
4315 "Possible attempt to put comments in qw() list");
4321 for (; !isSPACE(*d) && len; --len, ++d) ;
4323 words = append_elem(OP_LIST, words,
4324 newSVOP(OP_CONST, 0, newSVpvn(b, d-b)));
4328 PL_nextval[PL_nexttoke].opval = words;
4333 SvREFCNT_dec(PL_lex_stuff);
4334 PL_lex_stuff = Nullsv;
4339 s = scan_str(s,FALSE,FALSE);
4341 missingterm((char*)0);
4342 yylval.ival = OP_STRINGIFY;
4343 if (SvIVX(PL_lex_stuff) == '\'')
4344 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4345 TERM(sublex_start());
4348 s = scan_pat(s,OP_QR);
4349 TERM(sublex_start());
4352 s = scan_str(s,FALSE,FALSE);
4354 missingterm((char*)0);
4355 yylval.ival = OP_BACKTICK;
4357 TERM(sublex_start());
4363 *PL_tokenbuf = '\0';
4364 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4365 if (isIDFIRST_lazy(PL_tokenbuf))
4366 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4368 yyerror("<> should be quotes");
4375 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4379 LOP(OP_RENAME,XTERM);
4388 LOP(OP_RINDEX,XTERM);
4411 LOP(OP_REVERSE,XTERM);
4422 TERM(sublex_start());
4424 TOKEN(1); /* force error */
4433 LOP(OP_SELECT,XTERM);
4439 LOP(OP_SEMCTL,XTERM);
4442 LOP(OP_SEMGET,XTERM);
4445 LOP(OP_SEMOP,XTERM);
4451 LOP(OP_SETPGRP,XTERM);
4453 case KEY_setpriority:
4454 LOP(OP_SETPRIORITY,XTERM);
4456 case KEY_sethostent:
4462 case KEY_setservent:
4465 case KEY_setprotoent:
4475 LOP(OP_SEEKDIR,XTERM);
4477 case KEY_setsockopt:
4478 LOP(OP_SSOCKOPT,XTERM);
4484 LOP(OP_SHMCTL,XTERM);
4487 LOP(OP_SHMGET,XTERM);
4490 LOP(OP_SHMREAD,XTERM);
4493 LOP(OP_SHMWRITE,XTERM);
4496 LOP(OP_SHUTDOWN,XTERM);
4505 LOP(OP_SOCKET,XTERM);
4507 case KEY_socketpair:
4508 LOP(OP_SOCKPAIR,XTERM);
4511 checkcomma(s,PL_tokenbuf,"subroutine name");
4513 if (*s == ';' || *s == ')') /* probably a close */
4514 Perl_croak(aTHX_ "sort is now a reserved word");
4516 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4520 LOP(OP_SPLIT,XTERM);
4523 LOP(OP_SPRINTF,XTERM);
4526 LOP(OP_SPLICE,XTERM);
4541 LOP(OP_SUBSTR,XTERM);
4547 char tmpbuf[sizeof PL_tokenbuf];
4549 expectation attrful;
4550 bool have_name, have_proto;
4555 if (isIDFIRST_lazy(s) || *s == '\'' ||
4556 (*s == ':' && s[1] == ':'))
4559 attrful = XATTRBLOCK;
4560 /* remember buffer pos'n for later force_word */
4561 tboffset = s - PL_oldbufptr;
4562 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4563 if (strchr(tmpbuf, ':'))
4564 sv_setpv(PL_subname, tmpbuf);
4566 sv_setsv(PL_subname,PL_curstname);
4567 sv_catpvn(PL_subname,"::",2);
4568 sv_catpvn(PL_subname,tmpbuf,len);
4575 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4576 PL_expect = XTERMBLOCK;
4577 attrful = XATTRTERM;
4578 sv_setpv(PL_subname,"?");
4582 if (key == KEY_format) {
4584 PL_lex_formbrack = PL_lex_brackets + 1;
4586 (void) force_word(PL_oldbufptr + tboffset, WORD,
4591 /* Look for a prototype */
4595 s = scan_str(s,FALSE,FALSE);
4598 SvREFCNT_dec(PL_lex_stuff);
4599 PL_lex_stuff = Nullsv;
4600 Perl_croak(aTHX_ "Prototype not terminated");
4603 d = SvPVX(PL_lex_stuff);
4605 for (p = d; *p; ++p) {
4610 SvCUR(PL_lex_stuff) = tmp;
4618 if (*s == ':' && s[1] != ':')
4619 PL_expect = attrful;
4622 PL_nextval[PL_nexttoke].opval =
4623 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4624 PL_lex_stuff = Nullsv;
4628 sv_setpv(PL_subname,"__ANON__");
4631 (void) force_word(PL_oldbufptr + tboffset, WORD,
4640 LOP(OP_SYSTEM,XREF);
4643 LOP(OP_SYMLINK,XTERM);
4646 LOP(OP_SYSCALL,XTERM);
4649 LOP(OP_SYSOPEN,XTERM);
4652 LOP(OP_SYSSEEK,XTERM);
4655 LOP(OP_SYSREAD,XTERM);
4658 LOP(OP_SYSWRITE,XTERM);
4662 TERM(sublex_start());
4683 LOP(OP_TRUNCATE,XTERM);
4695 yylval.ival = CopLINE(PL_curcop);
4699 yylval.ival = CopLINE(PL_curcop);
4703 LOP(OP_UNLINK,XTERM);
4709 LOP(OP_UNPACK,XTERM);
4712 LOP(OP_UTIME,XTERM);
4715 if (ckWARN(WARN_OCTAL)) {
4716 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4717 if (*d != '0' && isDIGIT(*d))
4718 Perl_warner(aTHX_ WARN_OCTAL,
4719 "umask: argument is missing initial 0");
4724 LOP(OP_UNSHIFT,XTERM);
4727 if (PL_expect != XSTATE)
4728 yyerror("\"use\" not allowed in expression");
4731 s = force_version(s);
4732 if(*s == ';' || (s = skipspace(s), *s == ';')) {
4733 PL_nextval[PL_nexttoke].opval = Nullop;
4738 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4739 s = force_version(s);
4751 yylval.ival = CopLINE(PL_curcop);
4755 PL_hints |= HINT_BLOCK_SCOPE;
4762 LOP(OP_WAITPID,XTERM);
4770 static char ctl_l[2];
4772 if (ctl_l[0] == '\0')
4773 ctl_l[0] = toCTRL('L');
4774 gv_fetchpv(ctl_l,TRUE, SVt_PV);
4777 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
4782 if (PL_expect == XOPERATOR)
4788 yylval.ival = OP_XOR;
4793 TERM(sublex_start());
4799 Perl_keyword(pTHX_ register char *d, I32 len)
4804 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
4805 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
4806 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
4807 if (strEQ(d,"__DATA__")) return KEY___DATA__;
4808 if (strEQ(d,"__END__")) return KEY___END__;
4812 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
4817 if (strEQ(d,"and")) return -KEY_and;
4818 if (strEQ(d,"abs")) return -KEY_abs;
4821 if (strEQ(d,"alarm")) return -KEY_alarm;
4822 if (strEQ(d,"atan2")) return -KEY_atan2;
4825 if (strEQ(d,"accept")) return -KEY_accept;
4830 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
4833 if (strEQ(d,"bless")) return -KEY_bless;
4834 if (strEQ(d,"bind")) return -KEY_bind;
4835 if (strEQ(d,"binmode")) return -KEY_binmode;
4838 if (strEQ(d,"CORE")) return -KEY_CORE;
4843 if (strEQ(d,"cmp")) return -KEY_cmp;
4844 if (strEQ(d,"chr")) return -KEY_chr;
4845 if (strEQ(d,"cos")) return -KEY_cos;
4848 if (strEQ(d,"chop")) return KEY_chop;
4851 if (strEQ(d,"close")) return -KEY_close;
4852 if (strEQ(d,"chdir")) return -KEY_chdir;
4853 if (strEQ(d,"chomp")) return KEY_chomp;
4854 if (strEQ(d,"chmod")) return -KEY_chmod;
4855 if (strEQ(d,"chown")) return -KEY_chown;
4856 if (strEQ(d,"crypt")) return -KEY_crypt;
4859 if (strEQ(d,"chroot")) return -KEY_chroot;
4860 if (strEQ(d,"caller")) return -KEY_caller;
4863 if (strEQ(d,"connect")) return -KEY_connect;
4866 if (strEQ(d,"closedir")) return -KEY_closedir;
4867 if (strEQ(d,"continue")) return -KEY_continue;
4872 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
4877 if (strEQ(d,"do")) return KEY_do;
4880 if (strEQ(d,"die")) return -KEY_die;
4883 if (strEQ(d,"dump")) return -KEY_dump;
4886 if (strEQ(d,"delete")) return KEY_delete;
4889 if (strEQ(d,"defined")) return KEY_defined;
4890 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
4893 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
4898 if (strEQ(d,"EQ")) { deprecate(d); return -KEY_eq;}
4899 if (strEQ(d,"END")) return KEY_END;
4904 if (strEQ(d,"eq")) return -KEY_eq;
4907 if (strEQ(d,"eof")) return -KEY_eof;
4908 if (strEQ(d,"exp")) return -KEY_exp;
4911 if (strEQ(d,"else")) return KEY_else;
4912 if (strEQ(d,"exit")) return -KEY_exit;
4913 if (strEQ(d,"eval")) return KEY_eval;
4914 if (strEQ(d,"exec")) return -KEY_exec;
4915 if (strEQ(d,"each")) return KEY_each;
4918 if (strEQ(d,"elsif")) return KEY_elsif;
4921 if (strEQ(d,"exists")) return KEY_exists;
4922 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
4925 if (strEQ(d,"endgrent")) return -KEY_endgrent;
4926 if (strEQ(d,"endpwent")) return -KEY_endpwent;
4929 if (strEQ(d,"endnetent")) return -KEY_endnetent;
4932 if (strEQ(d,"endhostent")) return -KEY_endhostent;
4933 if (strEQ(d,"endservent")) return -KEY_endservent;
4936 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
4943 if (strEQ(d,"for")) return KEY_for;
4946 if (strEQ(d,"fork")) return -KEY_fork;
4949 if (strEQ(d,"fcntl")) return -KEY_fcntl;
4950 if (strEQ(d,"flock")) return -KEY_flock;
4953 if (strEQ(d,"format")) return KEY_format;
4954 if (strEQ(d,"fileno")) return -KEY_fileno;
4957 if (strEQ(d,"foreach")) return KEY_foreach;
4960 if (strEQ(d,"formline")) return -KEY_formline;
4966 if (strEQ(d,"GT")) { deprecate(d); return -KEY_gt;}
4967 if (strEQ(d,"GE")) { deprecate(d); return -KEY_ge;}
4971 if (strnEQ(d,"get",3)) {
4976 if (strEQ(d,"ppid")) return -KEY_getppid;
4977 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
4980 if (strEQ(d,"pwent")) return -KEY_getpwent;
4981 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
4982 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
4985 if (strEQ(d,"peername")) return -KEY_getpeername;
4986 if (strEQ(d,"protoent")) return -KEY_getprotoent;
4987 if (strEQ(d,"priority")) return -KEY_getpriority;
4990 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
4993 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
4997 else if (*d == 'h') {
4998 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
4999 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5000 if (strEQ(d,"hostent")) return -KEY_gethostent;
5002 else if (*d == 'n') {
5003 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5004 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5005 if (strEQ(d,"netent")) return -KEY_getnetent;
5007 else if (*d == 's') {
5008 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5009 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5010 if (strEQ(d,"servent")) return -KEY_getservent;
5011 if (strEQ(d,"sockname")) return -KEY_getsockname;
5012 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5014 else if (*d == 'g') {
5015 if (strEQ(d,"grent")) return -KEY_getgrent;
5016 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5017 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5019 else if (*d == 'l') {
5020 if (strEQ(d,"login")) return -KEY_getlogin;
5022 else if (strEQ(d,"c")) return -KEY_getc;
5027 if (strEQ(d,"gt")) return -KEY_gt;
5028 if (strEQ(d,"ge")) return -KEY_ge;
5031 if (strEQ(d,"grep")) return KEY_grep;
5032 if (strEQ(d,"goto")) return KEY_goto;
5033 if (strEQ(d,"glob")) return KEY_glob;
5036 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5041 if (strEQ(d,"hex")) return -KEY_hex;
5044 if (strEQ(d,"INIT")) return KEY_INIT;
5049 if (strEQ(d,"if")) return KEY_if;
5052 if (strEQ(d,"int")) return -KEY_int;
5055 if (strEQ(d,"index")) return -KEY_index;
5056 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5061 if (strEQ(d,"join")) return -KEY_join;
5065 if (strEQ(d,"keys")) return KEY_keys;
5066 if (strEQ(d,"kill")) return -KEY_kill;
5071 if (strEQ(d,"LT")) { deprecate(d); return -KEY_lt;}
5072 if (strEQ(d,"LE")) { deprecate(d); return -KEY_le;}
5078 if (strEQ(d,"lt")) return -KEY_lt;
5079 if (strEQ(d,"le")) return -KEY_le;
5080 if (strEQ(d,"lc")) return -KEY_lc;
5083 if (strEQ(d,"log")) return -KEY_log;
5086 if (strEQ(d,"last")) return KEY_last;
5087 if (strEQ(d,"link")) return -KEY_link;
5088 if (strEQ(d,"lock")) return -KEY_lock;
5091 if (strEQ(d,"local")) return KEY_local;
5092 if (strEQ(d,"lstat")) return -KEY_lstat;
5095 if (strEQ(d,"length")) return -KEY_length;
5096 if (strEQ(d,"listen")) return -KEY_listen;
5099 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5102 if (strEQ(d,"localtime")) return -KEY_localtime;
5108 case 1: return KEY_m;
5110 if (strEQ(d,"my")) return KEY_my;
5113 if (strEQ(d,"map")) return KEY_map;
5116 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5119 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5120 if (strEQ(d,"msgget")) return -KEY_msgget;
5121 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5122 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5127 if (strEQ(d,"NE")) { deprecate(d); return -KEY_ne;}
5130 if (strEQ(d,"next")) return KEY_next;
5131 if (strEQ(d,"ne")) return -KEY_ne;
5132 if (strEQ(d,"not")) return -KEY_not;
5133 if (strEQ(d,"no")) return KEY_no;
5138 if (strEQ(d,"or")) return -KEY_or;
5141 if (strEQ(d,"ord")) return -KEY_ord;
5142 if (strEQ(d,"oct")) return -KEY_oct;
5143 if (strEQ(d,"our")) return KEY_our;
5146 if (strEQ(d,"open")) return -KEY_open;
5149 if (strEQ(d,"opendir")) return -KEY_opendir;
5156 if (strEQ(d,"pop")) return KEY_pop;
5157 if (strEQ(d,"pos")) return KEY_pos;
5160 if (strEQ(d,"push")) return KEY_push;
5161 if (strEQ(d,"pack")) return -KEY_pack;
5162 if (strEQ(d,"pipe")) return -KEY_pipe;
5165 if (strEQ(d,"print")) return KEY_print;
5168 if (strEQ(d,"printf")) return KEY_printf;
5171 if (strEQ(d,"package")) return KEY_package;
5174 if (strEQ(d,"prototype")) return KEY_prototype;
5179 if (strEQ(d,"q")) return KEY_q;
5180 if (strEQ(d,"qr")) return KEY_qr;
5181 if (strEQ(d,"qq")) return KEY_qq;
5182 if (strEQ(d,"qw")) return KEY_qw;
5183 if (strEQ(d,"qx")) return KEY_qx;
5185 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5190 if (strEQ(d,"ref")) return -KEY_ref;
5193 if (strEQ(d,"read")) return -KEY_read;
5194 if (strEQ(d,"rand")) return -KEY_rand;
5195 if (strEQ(d,"recv")) return -KEY_recv;
5196 if (strEQ(d,"redo")) return KEY_redo;
5199 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5200 if (strEQ(d,"reset")) return -KEY_reset;
5203 if (strEQ(d,"return")) return KEY_return;
5204 if (strEQ(d,"rename")) return -KEY_rename;
5205 if (strEQ(d,"rindex")) return -KEY_rindex;
5208 if (strEQ(d,"require")) return -KEY_require;
5209 if (strEQ(d,"reverse")) return -KEY_reverse;
5210 if (strEQ(d,"readdir")) return -KEY_readdir;
5213 if (strEQ(d,"readlink")) return -KEY_readlink;
5214 if (strEQ(d,"readline")) return -KEY_readline;
5215 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5218 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5223 if (strEQ(d,"STOP")) return KEY_STOP;
5227 case 0: return KEY_s;
5229 if (strEQ(d,"scalar")) return KEY_scalar;
5234 if (strEQ(d,"seek")) return -KEY_seek;
5235 if (strEQ(d,"send")) return -KEY_send;
5238 if (strEQ(d,"semop")) return -KEY_semop;
5241 if (strEQ(d,"select")) return -KEY_select;
5242 if (strEQ(d,"semctl")) return -KEY_semctl;
5243 if (strEQ(d,"semget")) return -KEY_semget;
5246 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5247 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5250 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5251 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5254 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5257 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5258 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5259 if (strEQ(d,"setservent")) return -KEY_setservent;
5262 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5263 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5270 if (strEQ(d,"shift")) return KEY_shift;
5273 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5274 if (strEQ(d,"shmget")) return -KEY_shmget;
5277 if (strEQ(d,"shmread")) return -KEY_shmread;
5280 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5281 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5286 if (strEQ(d,"sin")) return -KEY_sin;
5289 if (strEQ(d,"sleep")) return -KEY_sleep;
5292 if (strEQ(d,"sort")) return KEY_sort;
5293 if (strEQ(d,"socket")) return -KEY_socket;
5294 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5297 if (strEQ(d,"split")) return KEY_split;
5298 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5299 if (strEQ(d,"splice")) return KEY_splice;
5302 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5305 if (strEQ(d,"srand")) return -KEY_srand;
5308 if (strEQ(d,"stat")) return -KEY_stat;
5309 if (strEQ(d,"study")) return KEY_study;
5312 if (strEQ(d,"substr")) return -KEY_substr;
5313 if (strEQ(d,"sub")) return KEY_sub;
5318 if (strEQ(d,"system")) return -KEY_system;
5321 if (strEQ(d,"symlink")) return -KEY_symlink;
5322 if (strEQ(d,"syscall")) return -KEY_syscall;
5323 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5324 if (strEQ(d,"sysread")) return -KEY_sysread;
5325 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5328 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5337 if (strEQ(d,"tr")) return KEY_tr;
5340 if (strEQ(d,"tie")) return KEY_tie;
5343 if (strEQ(d,"tell")) return -KEY_tell;
5344 if (strEQ(d,"tied")) return KEY_tied;
5345 if (strEQ(d,"time")) return -KEY_time;
5348 if (strEQ(d,"times")) return -KEY_times;
5351 if (strEQ(d,"telldir")) return -KEY_telldir;
5354 if (strEQ(d,"truncate")) return -KEY_truncate;
5361 if (strEQ(d,"uc")) return -KEY_uc;
5364 if (strEQ(d,"use")) return KEY_use;
5367 if (strEQ(d,"undef")) return KEY_undef;
5368 if (strEQ(d,"until")) return KEY_until;
5369 if (strEQ(d,"untie")) return KEY_untie;
5370 if (strEQ(d,"utime")) return -KEY_utime;
5371 if (strEQ(d,"umask")) return -KEY_umask;
5374 if (strEQ(d,"unless")) return KEY_unless;
5375 if (strEQ(d,"unpack")) return -KEY_unpack;
5376 if (strEQ(d,"unlink")) return -KEY_unlink;
5379 if (strEQ(d,"unshift")) return KEY_unshift;
5380 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5385 if (strEQ(d,"values")) return -KEY_values;
5386 if (strEQ(d,"vec")) return -KEY_vec;
5391 if (strEQ(d,"warn")) return -KEY_warn;
5392 if (strEQ(d,"wait")) return -KEY_wait;
5395 if (strEQ(d,"while")) return KEY_while;
5396 if (strEQ(d,"write")) return -KEY_write;
5399 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5402 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5407 if (len == 1) return -KEY_x;
5408 if (strEQ(d,"xor")) return -KEY_xor;
5411 if (len == 1) return KEY_y;
5420 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5424 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5425 dTHR; /* only for ckWARN */
5426 if (ckWARN(WARN_SYNTAX)) {
5428 for (w = s+2; *w && level; w++) {
5435 for (; *w && isSPACE(*w); w++) ;
5436 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5437 Perl_warner(aTHX_ WARN_SYNTAX,
5438 "%s (...) interpreted as function",name);
5441 while (s < PL_bufend && isSPACE(*s))
5445 while (s < PL_bufend && isSPACE(*s))
5447 if (isIDFIRST_lazy(s)) {
5449 while (isALNUM_lazy(s))
5451 while (s < PL_bufend && isSPACE(*s))
5456 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5460 Perl_croak(aTHX_ "No comma allowed after %s", what);
5465 /* Either returns sv, or mortalizes sv and returns a new SV*.
5466 Best used as sv=new_constant(..., sv, ...).
5467 If s, pv are NULL, calls subroutine with one argument,
5468 and type is used with error messages only. */
5471 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5475 HV *table = GvHV(PL_hintgv); /* ^H */
5479 const char *why, *why1, *why2;
5481 if (!(PL_hints & HINT_LOCALIZE_HH)) {
5484 why = "%^H is not localized";
5488 msg = Perl_newSVpvf(aTHX_ "constant(%s): %s%s%s",
5489 (type ? type: "undef"), why1, why2, why);
5490 yyerror(SvPVX(msg));
5495 why = "%^H is not defined";
5498 cvp = hv_fetch(table, key, strlen(key), FALSE);
5499 if (!cvp || !SvOK(*cvp)) {
5500 why = "} is not defined";
5505 sv_2mortal(sv); /* Parent created it permanently */
5508 pv = sv_2mortal(newSVpvn(s, len));
5510 typesv = sv_2mortal(newSVpv(type, 0));
5512 typesv = &PL_sv_undef;
5514 PUSHSTACKi(PERLSI_OVERLOAD);
5527 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5531 /* Check the eval first */
5532 if (!PL_in_eval && SvTRUE(ERRSV))
5535 sv_catpv(ERRSV, "Propagated");
5536 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5538 res = SvREFCNT_inc(sv);
5542 (void)SvREFCNT_inc(res);
5551 why = "}} did not return a defined value";
5552 why1 = "Call to &{$^H{";
5562 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5564 register char *d = dest;
5565 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5568 Perl_croak(aTHX_ ident_too_long);
5569 if (isALNUM(*s)) /* UTF handled below */
5571 else if (*s == '\'' && allow_package && isIDFIRST_lazy(s+1)) {
5576 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5580 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5581 char *t = s + UTF8SKIP(s);
5582 while (*t & 0x80 && is_utf8_mark((U8*)t))
5584 if (d + (t - s) > e)
5585 Perl_croak(aTHX_ ident_too_long);
5586 Copy(s, d, t - s, char);
5599 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5606 if (PL_lex_brackets == 0)
5607 PL_lex_fakebrack = 0;
5611 e = d + destlen - 3; /* two-character token, ending NUL */
5613 while (isDIGIT(*s)) {
5615 Perl_croak(aTHX_ ident_too_long);
5622 Perl_croak(aTHX_ ident_too_long);
5623 if (isALNUM(*s)) /* UTF handled below */
5625 else if (*s == '\'' && isIDFIRST_lazy(s+1)) {
5630 else if (*s == ':' && s[1] == ':') {
5634 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5635 char *t = s + UTF8SKIP(s);
5636 while (*t & 0x80 && is_utf8_mark((U8*)t))
5638 if (d + (t - s) > e)
5639 Perl_croak(aTHX_ ident_too_long);
5640 Copy(s, d, t - s, char);
5651 if (PL_lex_state != LEX_NORMAL)
5652 PL_lex_state = LEX_INTERPENDMAYBE;
5655 if (*s == '$' && s[1] &&
5656 (isALNUM_lazy(s+1) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
5669 if (*d == '^' && *s && isCONTROLVAR(*s)) {
5674 if (isSPACE(s[-1])) {
5677 if (ch != ' ' && ch != '\t') {
5683 if (isIDFIRST_lazy(d)) {
5687 while (e < send && isALNUM_lazy(e) || *e == ':') {
5689 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
5692 Copy(s, d, e - s, char);
5697 while ((isALNUM(*s) || *s == ':') && d < e)
5700 Perl_croak(aTHX_ ident_too_long);
5703 while (s < send && (*s == ' ' || *s == '\t')) s++;
5704 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
5705 dTHR; /* only for ckWARN */
5706 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
5707 const char *brack = *s == '[' ? "[...]" : "{...}";
5708 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5709 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
5710 funny, dest, brack, funny, dest, brack);
5712 PL_lex_fakebrack = PL_lex_brackets+1;
5714 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
5718 /* Handle extended ${^Foo} variables
5719 * 1999-02-27 mjd-perl-patch@plover.com */
5720 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
5724 while (isALNUM(*s) && d < e) {
5728 Perl_croak(aTHX_ ident_too_long);
5733 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
5734 PL_lex_state = LEX_INTERPEND;
5737 if (PL_lex_state == LEX_NORMAL) {
5738 dTHR; /* only for ckWARN */
5739 if (ckWARN(WARN_AMBIGUOUS) &&
5740 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
5742 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5743 "Ambiguous use of %c{%s} resolved to %c%s",
5744 funny, dest, funny, dest);
5749 s = bracket; /* let the parser handle it */
5753 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
5754 PL_lex_state = LEX_INTERPEND;
5759 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
5764 *pmfl |= PMf_GLOBAL;
5766 *pmfl |= PMf_CONTINUE;
5770 *pmfl |= PMf_MULTILINE;
5772 *pmfl |= PMf_SINGLELINE;
5774 *pmfl |= PMf_EXTENDED;
5778 S_scan_pat(pTHX_ char *start, I32 type)
5783 s = scan_str(start,FALSE,FALSE);
5786 SvREFCNT_dec(PL_lex_stuff);
5787 PL_lex_stuff = Nullsv;
5788 Perl_croak(aTHX_ "Search pattern not terminated");
5791 pm = (PMOP*)newPMOP(type, 0);
5792 if (PL_multi_open == '?')
5793 pm->op_pmflags |= PMf_ONCE;
5795 while (*s && strchr("iomsx", *s))
5796 pmflag(&pm->op_pmflags,*s++);
5799 while (*s && strchr("iogcmsx", *s))
5800 pmflag(&pm->op_pmflags,*s++);
5802 pm->op_pmpermflags = pm->op_pmflags;
5804 PL_lex_op = (OP*)pm;
5805 yylval.ival = OP_MATCH;
5810 S_scan_subst(pTHX_ char *start)
5817 yylval.ival = OP_NULL;
5819 s = scan_str(start,FALSE,FALSE);
5823 SvREFCNT_dec(PL_lex_stuff);
5824 PL_lex_stuff = Nullsv;
5825 Perl_croak(aTHX_ "Substitution pattern not terminated");
5828 if (s[-1] == PL_multi_open)
5831 first_start = PL_multi_start;
5832 s = scan_str(s,FALSE,FALSE);
5835 SvREFCNT_dec(PL_lex_stuff);
5836 PL_lex_stuff = Nullsv;
5838 SvREFCNT_dec(PL_lex_repl);
5839 PL_lex_repl = Nullsv;
5840 Perl_croak(aTHX_ "Substitution replacement not terminated");
5842 PL_multi_start = first_start; /* so whole substitution is taken together */
5844 pm = (PMOP*)newPMOP(OP_SUBST, 0);
5850 else if (strchr("iogcmsx", *s))
5851 pmflag(&pm->op_pmflags,*s++);
5858 PL_sublex_info.super_bufptr = s;
5859 PL_sublex_info.super_bufend = PL_bufend;
5861 pm->op_pmflags |= PMf_EVAL;
5862 repl = newSVpvn("",0);
5864 sv_catpv(repl, es ? "eval " : "do ");
5865 sv_catpvn(repl, "{ ", 2);
5866 sv_catsv(repl, PL_lex_repl);
5867 sv_catpvn(repl, " };", 2);
5869 SvREFCNT_dec(PL_lex_repl);
5873 pm->op_pmpermflags = pm->op_pmflags;
5874 PL_lex_op = (OP*)pm;
5875 yylval.ival = OP_SUBST;
5880 S_scan_trans(pTHX_ char *start)
5891 yylval.ival = OP_NULL;
5893 s = scan_str(start,FALSE,FALSE);
5896 SvREFCNT_dec(PL_lex_stuff);
5897 PL_lex_stuff = Nullsv;
5898 Perl_croak(aTHX_ "Transliteration pattern not terminated");
5900 if (s[-1] == PL_multi_open)
5903 s = scan_str(s,FALSE,FALSE);
5906 SvREFCNT_dec(PL_lex_stuff);
5907 PL_lex_stuff = Nullsv;
5909 SvREFCNT_dec(PL_lex_repl);
5910 PL_lex_repl = Nullsv;
5911 Perl_croak(aTHX_ "Transliteration replacement not terminated");
5915 o = newSVOP(OP_TRANS, 0, 0);
5916 utf8 = OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF;
5919 New(803,tbl,256,short);
5920 o = newPVOP(OP_TRANS, 0, (char*)tbl);
5924 complement = del = squash = 0;
5925 while (strchr("cdsCU", *s)) {
5927 complement = OPpTRANS_COMPLEMENT;
5929 del = OPpTRANS_DELETE;
5931 squash = OPpTRANS_SQUASH;
5936 utf8 &= ~OPpTRANS_FROM_UTF;
5938 utf8 |= OPpTRANS_FROM_UTF;
5942 utf8 &= ~OPpTRANS_TO_UTF;
5944 utf8 |= OPpTRANS_TO_UTF;
5947 Perl_croak(aTHX_ "Too many /C and /U options");
5952 o->op_private = del|squash|complement|utf8;
5955 yylval.ival = OP_TRANS;
5960 S_scan_heredoc(pTHX_ register char *s)
5964 I32 op_type = OP_SCALAR;
5971 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
5975 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
5978 for (peek = s; *peek == ' ' || *peek == '\t'; peek++) ;
5979 if (*peek && strchr("`'\"",*peek)) {
5982 s = delimcpy(d, e, s, PL_bufend, term, &len);
5992 if (!isALNUM_lazy(s))
5993 deprecate("bare << to mean <<\"\"");
5994 for (; isALNUM_lazy(s); s++) {
5999 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6000 Perl_croak(aTHX_ "Delimiter for here document is too long");
6003 len = d - PL_tokenbuf;
6004 #ifndef PERL_STRICT_CR
6005 d = strchr(s, '\r');
6009 while (s < PL_bufend) {
6015 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6024 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6029 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6030 herewas = newSVpvn(s,PL_bufend-s);
6032 s--, herewas = newSVpvn(s,d-s);
6033 s += SvCUR(herewas);
6035 tmpstr = NEWSV(87,79);
6036 sv_upgrade(tmpstr, SVt_PVIV);
6041 else if (term == '`') {
6042 op_type = OP_BACKTICK;
6043 SvIVX(tmpstr) = '\\';
6047 PL_multi_start = CopLINE(PL_curcop);
6048 PL_multi_open = PL_multi_close = '<';
6049 term = *PL_tokenbuf;
6050 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6051 char *bufptr = PL_sublex_info.super_bufptr;
6052 char *bufend = PL_sublex_info.super_bufend;
6053 char *olds = s - SvCUR(herewas);
6054 s = strchr(bufptr, '\n');
6058 while (s < bufend &&
6059 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6061 CopLINE_inc(PL_curcop);
6064 CopLINE_set(PL_curcop, PL_multi_start);
6065 missingterm(PL_tokenbuf);
6067 sv_setpvn(herewas,bufptr,d-bufptr+1);
6068 sv_setpvn(tmpstr,d+1,s-d);
6070 sv_catpvn(herewas,s,bufend-s);
6071 (void)strcpy(bufptr,SvPVX(herewas));
6078 while (s < PL_bufend &&
6079 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6081 CopLINE_inc(PL_curcop);
6083 if (s >= PL_bufend) {
6084 CopLINE_set(PL_curcop, PL_multi_start);
6085 missingterm(PL_tokenbuf);
6087 sv_setpvn(tmpstr,d+1,s-d);
6089 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6091 sv_catpvn(herewas,s,PL_bufend-s);
6092 sv_setsv(PL_linestr,herewas);
6093 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6094 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6097 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6098 while (s >= PL_bufend) { /* multiple line string? */
6100 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6101 CopLINE_set(PL_curcop, PL_multi_start);
6102 missingterm(PL_tokenbuf);
6104 CopLINE_inc(PL_curcop);
6105 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6106 #ifndef PERL_STRICT_CR
6107 if (PL_bufend - PL_linestart >= 2) {
6108 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6109 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6111 PL_bufend[-2] = '\n';
6113 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6115 else if (PL_bufend[-1] == '\r')
6116 PL_bufend[-1] = '\n';
6118 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6119 PL_bufend[-1] = '\n';
6121 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6122 SV *sv = NEWSV(88,0);
6124 sv_upgrade(sv, SVt_PVMG);
6125 sv_setsv(sv,PL_linestr);
6126 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6128 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6131 sv_catsv(PL_linestr,herewas);
6132 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6136 sv_catsv(tmpstr,PL_linestr);
6141 PL_multi_end = CopLINE(PL_curcop);
6142 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6143 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6144 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6146 SvREFCNT_dec(herewas);
6147 PL_lex_stuff = tmpstr;
6148 yylval.ival = op_type;
6153 takes: current position in input buffer
6154 returns: new position in input buffer
6155 side-effects: yylval and lex_op are set.
6160 <FH> read from filehandle
6161 <pkg::FH> read from package qualified filehandle
6162 <pkg'FH> read from package qualified filehandle
6163 <$fh> read from filehandle in $fh
6169 S_scan_inputsymbol(pTHX_ char *start)
6171 register char *s = start; /* current position in buffer */
6177 d = PL_tokenbuf; /* start of temp holding space */
6178 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6179 end = strchr(s, '\n');
6182 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6184 /* die if we didn't have space for the contents of the <>,
6185 or if it didn't end, or if we see a newline
6188 if (len >= sizeof PL_tokenbuf)
6189 Perl_croak(aTHX_ "Excessively long <> operator");
6191 Perl_croak(aTHX_ "Unterminated <> operator");
6196 Remember, only scalar variables are interpreted as filehandles by
6197 this code. Anything more complex (e.g., <$fh{$num}>) will be
6198 treated as a glob() call.
6199 This code makes use of the fact that except for the $ at the front,
6200 a scalar variable and a filehandle look the same.
6202 if (*d == '$' && d[1]) d++;
6204 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6205 while (*d && (isALNUM_lazy(d) || *d == '\'' || *d == ':'))
6208 /* If we've tried to read what we allow filehandles to look like, and
6209 there's still text left, then it must be a glob() and not a getline.
6210 Use scan_str to pull out the stuff between the <> and treat it
6211 as nothing more than a string.
6214 if (d - PL_tokenbuf != len) {
6215 yylval.ival = OP_GLOB;
6217 s = scan_str(start,FALSE,FALSE);
6219 Perl_croak(aTHX_ "Glob not terminated");
6223 /* we're in a filehandle read situation */
6226 /* turn <> into <ARGV> */
6228 (void)strcpy(d,"ARGV");
6230 /* if <$fh>, create the ops to turn the variable into a
6236 /* try to find it in the pad for this block, otherwise find
6237 add symbol table ops
6239 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6240 OP *o = newOP(OP_PADSV, 0);
6242 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6245 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6246 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6247 newUNOP(OP_RV2SV, 0,
6248 newGVOP(OP_GV, 0, gv)));
6250 PL_lex_op->op_flags |= OPf_SPECIAL;
6251 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6252 yylval.ival = OP_NULL;
6255 /* If it's none of the above, it must be a literal filehandle
6256 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6258 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6259 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6260 yylval.ival = OP_NULL;
6269 takes: start position in buffer
6270 keep_quoted preserve \ on the embedded delimiter(s)
6271 keep_delims preserve the delimiters around the string
6272 returns: position to continue reading from buffer
6273 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6274 updates the read buffer.
6276 This subroutine pulls a string out of the input. It is called for:
6277 q single quotes q(literal text)
6278 ' single quotes 'literal text'
6279 qq double quotes qq(interpolate $here please)
6280 " double quotes "interpolate $here please"
6281 qx backticks qx(/bin/ls -l)
6282 ` backticks `/bin/ls -l`
6283 qw quote words @EXPORT_OK = qw( func() $spam )
6284 m// regexp match m/this/
6285 s/// regexp substitute s/this/that/
6286 tr/// string transliterate tr/this/that/
6287 y/// string transliterate y/this/that/
6288 ($*@) sub prototypes sub foo ($)
6289 (stuff) sub attr parameters sub foo : attr(stuff)
6290 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6292 In most of these cases (all but <>, patterns and transliterate)
6293 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6294 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6295 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6298 It skips whitespace before the string starts, and treats the first
6299 character as the delimiter. If the delimiter is one of ([{< then
6300 the corresponding "close" character )]}> is used as the closing
6301 delimiter. It allows quoting of delimiters, and if the string has
6302 balanced delimiters ([{<>}]) it allows nesting.
6304 The lexer always reads these strings into lex_stuff, except in the
6305 case of the operators which take *two* arguments (s/// and tr///)
6306 when it checks to see if lex_stuff is full (presumably with the 1st
6307 arg to s or tr) and if so puts the string into lex_repl.
6312 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6315 SV *sv; /* scalar value: string */
6316 char *tmps; /* temp string, used for delimiter matching */
6317 register char *s = start; /* current position in the buffer */
6318 register char term; /* terminating character */
6319 register char *to; /* current position in the sv's data */
6320 I32 brackets = 1; /* bracket nesting level */
6322 /* skip space before the delimiter */
6326 /* mark where we are, in case we need to report errors */
6329 /* after skipping whitespace, the next character is the terminator */
6331 /* mark where we are */
6332 PL_multi_start = CopLINE(PL_curcop);
6333 PL_multi_open = term;
6335 /* find corresponding closing delimiter */
6336 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6338 PL_multi_close = term;
6340 /* create a new SV to hold the contents. 87 is leak category, I'm
6341 assuming. 79 is the SV's initial length. What a random number. */
6343 sv_upgrade(sv, SVt_PVIV);
6345 (void)SvPOK_only(sv); /* validate pointer */
6347 /* move past delimiter and try to read a complete string */
6349 sv_catpvn(sv, s, 1);
6352 /* extend sv if need be */
6353 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6354 /* set 'to' to the next character in the sv's string */
6355 to = SvPVX(sv)+SvCUR(sv);
6357 /* if open delimiter is the close delimiter read unbridle */
6358 if (PL_multi_open == PL_multi_close) {
6359 for (; s < PL_bufend; s++,to++) {
6360 /* embedded newlines increment the current line number */
6361 if (*s == '\n' && !PL_rsfp)
6362 CopLINE_inc(PL_curcop);
6363 /* handle quoted delimiters */
6364 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6365 if (!keep_quoted && s[1] == term)
6367 /* any other quotes are simply copied straight through */
6371 /* terminate when run out of buffer (the for() condition), or
6372 have found the terminator */
6373 else if (*s == term)
6379 /* if the terminator isn't the same as the start character (e.g.,
6380 matched brackets), we have to allow more in the quoting, and
6381 be prepared for nested brackets.
6384 /* read until we run out of string, or we find the terminator */
6385 for (; s < PL_bufend; s++,to++) {
6386 /* embedded newlines increment the line count */
6387 if (*s == '\n' && !PL_rsfp)
6388 CopLINE_inc(PL_curcop);
6389 /* backslashes can escape the open or closing characters */
6390 if (*s == '\\' && s+1 < PL_bufend) {
6392 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6397 /* allow nested opens and closes */
6398 else if (*s == PL_multi_close && --brackets <= 0)
6400 else if (*s == PL_multi_open)
6405 /* terminate the copied string and update the sv's end-of-string */
6407 SvCUR_set(sv, to - SvPVX(sv));
6410 * this next chunk reads more into the buffer if we're not done yet
6413 if (s < PL_bufend) break; /* handle case where we are done yet :-) */
6415 #ifndef PERL_STRICT_CR
6416 if (to - SvPVX(sv) >= 2) {
6417 if ((to[-2] == '\r' && to[-1] == '\n') ||
6418 (to[-2] == '\n' && to[-1] == '\r'))
6422 SvCUR_set(sv, to - SvPVX(sv));
6424 else if (to[-1] == '\r')
6427 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6431 /* if we're out of file, or a read fails, bail and reset the current
6432 line marker so we can report where the unterminated string began
6435 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6437 CopLINE_set(PL_curcop, PL_multi_start);
6440 /* we read a line, so increment our line counter */
6441 CopLINE_inc(PL_curcop);
6443 /* update debugger info */
6444 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6445 SV *sv = NEWSV(88,0);
6447 sv_upgrade(sv, SVt_PVMG);
6448 sv_setsv(sv,PL_linestr);
6449 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6452 /* having changed the buffer, we must update PL_bufend */
6453 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6456 /* at this point, we have successfully read the delimited string */
6459 sv_catpvn(sv, s, 1);
6460 PL_multi_end = CopLINE(PL_curcop);
6463 /* if we allocated too much space, give some back */
6464 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6465 SvLEN_set(sv, SvCUR(sv) + 1);
6466 Renew(SvPVX(sv), SvLEN(sv), char);
6469 /* decide whether this is the first or second quoted string we've read
6482 takes: pointer to position in buffer
6483 returns: pointer to new position in buffer
6484 side-effects: builds ops for the constant in yylval.op
6486 Read a number in any of the formats that Perl accepts:
6488 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6489 [\d_]+(\.[\d_]*)?[Ee](\d+)
6491 Underbars (_) are allowed in decimal numbers. If -w is on,
6492 underbars before a decimal point must be at three digit intervals.
6494 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6497 If it reads a number without a decimal point or an exponent, it will
6498 try converting the number to an integer and see if it can do so
6499 without loss of precision.
6503 Perl_scan_num(pTHX_ char *start)
6505 register char *s = start; /* current position in buffer */
6506 register char *d; /* destination in temp buffer */
6507 register char *e; /* end of temp buffer */
6508 IV tryiv; /* used to see if it can be an IV */
6509 NV value; /* number read, as a double */
6510 SV *sv; /* place to put the converted number */
6511 bool floatit; /* boolean: int or float? */
6512 char *lastub = 0; /* position of last underbar */
6513 static char number_too_long[] = "Number too long";
6515 /* We use the first character to decide what type of number this is */
6519 Perl_croak(aTHX_ "panic: scan_num");
6521 /* if it starts with a 0, it could be an octal number, a decimal in
6522 0.13 disguise, or a hexadecimal number, or a binary number.
6527 u holds the "number so far"
6528 shift the power of 2 of the base
6529 (hex == 4, octal == 3, binary == 1)
6530 overflowed was the number more than we can hold?
6532 Shift is used when we add a digit. It also serves as an "are
6533 we in octal/hex/binary?" indicator to disallow hex characters
6540 bool overflowed = FALSE;
6541 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6542 static char* bases[5] = { "", "binary", "", "octal",
6544 static char* Bases[5] = { "", "Binary", "", "Octal",
6546 static char *maxima[5] = { "",
6547 "0b11111111111111111111111111111111",
6551 char *base, *Base, *max;
6557 } else if (s[1] == 'b') {
6561 /* check for a decimal in disguise */
6562 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6564 /* so it must be octal */
6568 base = bases[shift];
6569 Base = Bases[shift];
6570 max = maxima[shift];
6572 /* read the rest of the number */
6574 /* x is used in the overflow test,
6575 b is the digit we're adding on. */
6580 /* if we don't mention it, we're done */
6589 /* 8 and 9 are not octal */
6592 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6596 case '2': case '3': case '4':
6597 case '5': case '6': case '7':
6599 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6603 b = *s++ & 15; /* ASCII digit -> value of digit */
6607 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6608 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6609 /* make sure they said 0x */
6614 /* Prepare to put the digit we have onto the end
6615 of the number so far. We check for overflows.
6620 x = u << shift; /* make room for the digit */
6622 if ((x >> shift) != u
6623 && !(PL_hints & HINT_NEW_BINARY)) {
6627 if (ckWARN_d(WARN_OVERFLOW))
6628 Perl_warner(aTHX_ WARN_OVERFLOW,
6629 "Integer overflow in %s number",
6632 u = x | b; /* add the digit to the end */
6635 n *= nvshift[shift];
6636 /* If an NV has not enough bits in its
6637 * mantissa to represent an UV this summing of
6638 * small low-order numbers is a waste of time
6639 * (because the NV cannot preserve the
6640 * low-order bits anyway): we could just
6641 * remember when did we overflow and in the
6642 * end just multiply n by the right
6650 /* if we get here, we had success: make a scalar value from
6657 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6658 Perl_warner(aTHX_ WARN_PORTABLE,
6659 "%s number > %s non-portable",
6666 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6667 Perl_warner(aTHX_ WARN_PORTABLE,
6668 "%s number > %s non-portable",
6673 if (PL_hints & HINT_NEW_BINARY)
6674 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6679 handle decimal numbers.
6680 we're also sent here when we read a 0 as the first digit
6682 case '1': case '2': case '3': case '4': case '5':
6683 case '6': case '7': case '8': case '9': case '.':
6686 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
6689 /* read next group of digits and _ and copy into d */
6690 while (isDIGIT(*s) || *s == '_') {
6691 /* skip underscores, checking for misplaced ones
6695 dTHR; /* only for ckWARN */
6696 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
6697 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6701 /* check for end of fixed-length buffer */
6703 Perl_croak(aTHX_ number_too_long);
6704 /* if we're ok, copy the character */
6709 /* final misplaced underbar check */
6710 if (lastub && s - lastub != 3) {
6712 if (ckWARN(WARN_SYNTAX))
6713 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6716 /* read a decimal portion if there is one. avoid
6717 3..5 being interpreted as the number 3. followed
6720 if (*s == '.' && s[1] != '.') {
6724 /* copy, ignoring underbars, until we run out of
6725 digits. Note: no misplaced underbar checks!
6727 for (; isDIGIT(*s) || *s == '_'; s++) {
6728 /* fixed length buffer check */
6730 Perl_croak(aTHX_ number_too_long);
6736 /* read exponent part, if present */
6737 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
6741 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
6742 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
6744 /* allow positive or negative exponent */
6745 if (*s == '+' || *s == '-')
6748 /* read digits of exponent (no underbars :-) */
6749 while (isDIGIT(*s)) {
6751 Perl_croak(aTHX_ number_too_long);
6756 /* terminate the string */
6759 /* make an sv from the string */
6762 value = Atof(PL_tokenbuf);
6765 See if we can make do with an integer value without loss of
6766 precision. We use I_V to cast to an int, because some
6767 compilers have issues. Then we try casting it back and see
6768 if it was the same. We only do this if we know we
6769 specifically read an integer.
6771 Note: if floatit is true, then we don't need to do the
6775 if (!floatit && (NV)tryiv == value)
6776 sv_setiv(sv, tryiv);
6778 sv_setnv(sv, value);
6779 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
6780 (PL_hints & HINT_NEW_INTEGER) )
6781 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
6782 (floatit ? "float" : "integer"),
6787 /* make the op for the constant and return */
6789 yylval.opval = newSVOP(OP_CONST, 0, sv);
6795 S_scan_formline(pTHX_ register char *s)
6800 SV *stuff = newSVpvn("",0);
6801 bool needargs = FALSE;
6804 if (*s == '.' || *s == '}') {
6806 #ifdef PERL_STRICT_CR
6807 for (t = s+1;*t == ' ' || *t == '\t'; t++) ;
6809 for (t = s+1;*t == ' ' || *t == '\t' || *t == '\r'; t++) ;
6811 if (*t == '\n' || t == PL_bufend)
6814 if (PL_in_eval && !PL_rsfp) {
6815 eol = strchr(s,'\n');
6820 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6822 for (t = s; t < eol; t++) {
6823 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
6825 goto enough; /* ~~ must be first line in formline */
6827 if (*t == '@' || *t == '^')
6830 sv_catpvn(stuff, s, eol-s);
6834 s = filter_gets(PL_linestr, PL_rsfp, 0);
6835 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
6836 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
6839 yyerror("Format not terminated");
6849 PL_lex_state = LEX_NORMAL;
6850 PL_nextval[PL_nexttoke].ival = 0;
6854 PL_lex_state = LEX_FORMLINE;
6855 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
6857 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
6861 SvREFCNT_dec(stuff);
6862 PL_lex_formbrack = 0;
6873 PL_cshlen = strlen(PL_cshname);
6878 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
6881 I32 oldsavestack_ix = PL_savestack_ix;
6882 CV* outsidecv = PL_compcv;
6886 assert(SvTYPE(PL_compcv) == SVt_PVCV);
6888 SAVEI32(PL_subline);
6889 save_item(PL_subname);
6891 SAVEVPTR(PL_curpad);
6892 SAVESPTR(PL_comppad);
6893 SAVESPTR(PL_comppad_name);
6894 SAVESPTR(PL_compcv);
6895 SAVEI32(PL_comppad_name_fill);
6896 SAVEI32(PL_min_intro_pending);
6897 SAVEI32(PL_max_intro_pending);
6898 SAVEI32(PL_pad_reset_pending);
6900 PL_compcv = (CV*)NEWSV(1104,0);
6901 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
6902 CvFLAGS(PL_compcv) |= flags;
6904 PL_comppad = newAV();
6905 av_push(PL_comppad, Nullsv);
6906 PL_curpad = AvARRAY(PL_comppad);
6907 PL_comppad_name = newAV();
6908 PL_comppad_name_fill = 0;
6909 PL_min_intro_pending = 0;
6911 PL_subline = CopLINE(PL_curcop);
6913 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
6914 PL_curpad[0] = (SV*)newAV();
6915 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
6916 #endif /* USE_THREADS */
6918 comppadlist = newAV();
6919 AvREAL_off(comppadlist);
6920 av_store(comppadlist, 0, (SV*)PL_comppad_name);
6921 av_store(comppadlist, 1, (SV*)PL_comppad);
6923 CvPADLIST(PL_compcv) = comppadlist;
6924 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
6926 CvOWNER(PL_compcv) = 0;
6927 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
6928 MUTEX_INIT(CvMUTEXP(PL_compcv));
6929 #endif /* USE_THREADS */
6931 return oldsavestack_ix;
6935 Perl_yywarn(pTHX_ char *s)
6938 PL_in_eval |= EVAL_WARNONLY;
6940 PL_in_eval &= ~EVAL_WARNONLY;
6945 Perl_yyerror(pTHX_ char *s)
6949 char *context = NULL;
6953 if (!yychar || (yychar == ';' && !PL_rsfp))
6955 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
6956 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
6957 while (isSPACE(*PL_oldoldbufptr))
6959 context = PL_oldoldbufptr;
6960 contlen = PL_bufptr - PL_oldoldbufptr;
6962 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
6963 PL_oldbufptr != PL_bufptr) {
6964 while (isSPACE(*PL_oldbufptr))
6966 context = PL_oldbufptr;
6967 contlen = PL_bufptr - PL_oldbufptr;
6969 else if (yychar > 255)
6970 where = "next token ???";
6971 else if ((yychar & 127) == 127) {
6972 if (PL_lex_state == LEX_NORMAL ||
6973 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
6974 where = "at end of line";
6975 else if (PL_lex_inpat)
6976 where = "within pattern";
6978 where = "within string";
6981 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
6983 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
6984 else if (isPRINT_LC(yychar))
6985 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
6987 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
6988 where = SvPVX(where_sv);
6990 msg = sv_2mortal(newSVpv(s, 0));
6991 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
6992 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
6994 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
6996 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
6997 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
6998 Perl_sv_catpvf(aTHX_ msg,
6999 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7000 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7003 if (PL_in_eval & EVAL_WARNONLY)
7004 Perl_warn(aTHX_ "%_", msg);
7007 if (PL_error_count >= 10)
7008 Perl_croak(aTHX_ "%s has too many errors.\n", CopFILE(PL_curcop));
7010 PL_in_my_stash = Nullhv;
7021 * Restore a source filter.
7025 restore_rsfp(pTHXo_ void *f)
7027 PerlIO *fp = (PerlIO*)f;
7029 if (PL_rsfp == PerlIO_stdin())
7030 PerlIO_clearerr(PL_rsfp);
7031 else if (PL_rsfp && (PL_rsfp != fp))
7032 PerlIO_close(PL_rsfp);