3 * Copyright (c) 1991-2002, 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";
29 static char c_without_g[] = "Use of /c modifier is meaningless without /g";
30 static char c_in_subst[] = "Use of /c modifier is meaningless in s///";
32 static void restore_rsfp(pTHX_ void *f);
33 #ifndef PERL_NO_UTF16_FILTER
34 static I32 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen);
35 static I32 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen);
38 #define XFAKEBRACK 128
41 #ifdef USE_UTF8_SCRIPTS
42 # define UTF (!IN_BYTES)
44 # define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
47 /* In variables named $^X, these are the legal values for X.
48 * 1999-02-27 mjd-perl-patch@plover.com */
49 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
51 /* On MacOS, respect nonbreaking spaces */
52 #ifdef MACOS_TRADITIONAL
53 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
55 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
58 /* LEX_* are values for PL_lex_state, the state of the lexer.
59 * They are arranged oddly so that the guard on the switch statement
60 * can get by with a single comparison (if the compiler is smart enough).
63 /* #define LEX_NOTPARSING 11 is done in perl.h. */
66 #define LEX_INTERPNORMAL 9
67 #define LEX_INTERPCASEMOD 8
68 #define LEX_INTERPPUSH 7
69 #define LEX_INTERPSTART 6
70 #define LEX_INTERPEND 5
71 #define LEX_INTERPENDMAYBE 4
72 #define LEX_INTERPCONCAT 3
73 #define LEX_INTERPCONST 2
74 #define LEX_FORMLINE 1
75 #define LEX_KNOWNEXT 0
83 # define YYMAXLEVEL 100
85 YYSTYPE* yylval_pointer[YYMAXLEVEL];
86 int* yychar_pointer[YYMAXLEVEL];
90 # define yylval (*yylval_pointer[yyactlevel])
91 # define yychar (*yychar_pointer[yyactlevel])
92 # define PERL_YYLEX_PARAM yylval_pointer[yyactlevel],yychar_pointer[yyactlevel]
94 # define yylex() Perl_yylex_r(aTHX_ yylval_pointer[yyactlevel],yychar_pointer[yyactlevel])
99 /* CLINE is a macro that ensures PL_copline has a sane value */
104 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
107 * Convenience functions to return different tokens and prime the
108 * lexer for the next token. They all take an argument.
110 * TOKEN : generic token (used for '(', DOLSHARP, etc)
111 * OPERATOR : generic operator
112 * AOPERATOR : assignment operator
113 * PREBLOCK : beginning the block after an if, while, foreach, ...
114 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
115 * PREREF : *EXPR where EXPR is not a simple identifier
116 * TERM : expression term
117 * LOOPX : loop exiting command (goto, last, dump, etc)
118 * FTST : file test operator
119 * FUN0 : zero-argument function
120 * FUN1 : not used, except for not, which isn't a UNIOP
121 * BOop : bitwise or or xor
123 * SHop : shift operator
124 * PWop : power operator
125 * PMop : pattern-matching operator
126 * Aop : addition-level operator
127 * Mop : multiplication-level operator
128 * Eop : equality-testing operator
129 * Rop : relational operator <= != gt
131 * Also see LOP and lop() below.
134 /* Note that REPORT() and REPORT2() will be expressions that supply
135 * their own trailing comma, not suitable for statements as such. */
136 #ifdef DEBUGGING /* Serve -DT. */
137 # define REPORT(x,retval) tokereport(x,s,(int)retval),
138 # define REPORT2(x,retval) tokereport(x,s, yylval.ival),
140 # define REPORT(x,retval)
141 # define REPORT2(x,retval)
144 #define TOKEN(retval) return (REPORT2("token",retval) PL_bufptr = s,(int)retval)
145 #define OPERATOR(retval) return (REPORT2("operator",retval) PL_expect = XTERM, PL_bufptr = s,(int)retval)
146 #define AOPERATOR(retval) return ao((REPORT2("aop",retval) PL_expect = XTERM, PL_bufptr = s,(int)retval))
147 #define PREBLOCK(retval) return (REPORT2("preblock",retval) PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
148 #define PRETERMBLOCK(retval) return (REPORT2("pretermblock",retval) PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
149 #define PREREF(retval) return (REPORT2("preref",retval) PL_expect = XREF,PL_bufptr = s,(int)retval)
150 #define TERM(retval) return (CLINE, REPORT2("term",retval) PL_expect = XOPERATOR, PL_bufptr = s,(int)retval)
151 #define LOOPX(f) return(yylval.ival=f, REPORT("loopx",f) PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
152 #define FTST(f) return(yylval.ival=f, REPORT("ftst",f) PL_expect = XTERMORDORDOR,PL_bufptr = s,(int)UNIOP)
153 #define FUN0(f) return(yylval.ival = f, REPORT("fun0",f) PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
154 #define FUN1(f) return(yylval.ival = f, REPORT("fun1",f) PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
155 #define BOop(f) return ao((yylval.ival=f, REPORT("bitorop",f) PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
156 #define BAop(f) return ao((yylval.ival=f, REPORT("bitandop",f) PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
157 #define SHop(f) return ao((yylval.ival=f, REPORT("shiftop",f) PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
158 #define PWop(f) return ao((yylval.ival=f, REPORT("powop",f) PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
159 #define PMop(f) return(yylval.ival=f, REPORT("matchop",f) PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
160 #define Aop(f) return ao((yylval.ival=f, REPORT("add",f) PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
161 #define Mop(f) return ao((yylval.ival=f, REPORT("mul",f) PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
162 #define Eop(f) return(yylval.ival=f, REPORT("eq",f) PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
163 #define Rop(f) return(yylval.ival=f, REPORT("rel",f) PL_expect = XTERM,PL_bufptr = s,(int)RELOP)
165 /* This bit of chicanery makes a unary function followed by
166 * a parenthesis into a function with one argument, highest precedence.
167 * The UNIDOR macro is for unary functions that can be followed by the //
168 * operator (such as C<shift // 0>).
170 #define UNI2(f,x) return(yylval.ival = f, \
174 PL_last_uni = PL_oldbufptr, \
175 PL_last_lop_op = f, \
176 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
177 #define UNI(f) UNI2(f,XTERM)
178 #define UNIDOR(f) UNI2(f,XTERMORDORDOR)
180 #define UNIBRACK(f) return(yylval.ival = f, \
183 PL_last_uni = PL_oldbufptr, \
184 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
186 /* grandfather return to old style */
187 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
192 S_tokereport(pTHX_ char *thing, char* s, I32 rv)
195 SV* report = newSVpv(thing, 0);
196 Perl_sv_catpvf(aTHX_ report, ":line %d:%"IVdf":", CopLINE(PL_curcop),
199 if (s - PL_bufptr > 0)
200 sv_catpvn(report, PL_bufptr, s - PL_bufptr);
202 if (PL_oldbufptr && *PL_oldbufptr)
203 sv_catpv(report, PL_tokenbuf);
205 PerlIO_printf(Perl_debug_log, "### %s\n", SvPV_nolen(report));
214 * This subroutine detects &&=, ||=, and //= and turns an ANDAND, OROR or DORDOR
215 * into an OP_ANDASSIGN, OP_ORASSIGN, or OP_DORASSIGN
219 S_ao(pTHX_ int toketype)
221 if (*PL_bufptr == '=') {
223 if (toketype == ANDAND)
224 yylval.ival = OP_ANDASSIGN;
225 else if (toketype == OROR)
226 yylval.ival = OP_ORASSIGN;
227 else if (toketype == DORDOR)
228 yylval.ival = OP_DORASSIGN;
236 * When Perl expects an operator and finds something else, no_op
237 * prints the warning. It always prints "<something> found where
238 * operator expected. It prints "Missing semicolon on previous line?"
239 * if the surprise occurs at the start of the line. "do you need to
240 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
241 * where the compiler doesn't know if foo is a method call or a function.
242 * It prints "Missing operator before end of line" if there's nothing
243 * after the missing operator, or "... before <...>" if there is something
244 * after the missing operator.
248 S_no_op(pTHX_ char *what, char *s)
250 char *oldbp = PL_bufptr;
251 bool is_first = (PL_oldbufptr == PL_linestart);
257 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
259 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
260 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
262 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
263 if (t < PL_bufptr && isSPACE(*t))
264 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
265 t - PL_oldoldbufptr, PL_oldoldbufptr);
269 Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
276 * Complain about missing quote/regexp/heredoc terminator.
277 * If it's called with (char *)NULL then it cauterizes the line buffer.
278 * If we're in a delimited string and the delimiter is a control
279 * character, it's reformatted into a two-char sequence like ^C.
284 S_missingterm(pTHX_ char *s)
289 char *nl = strrchr(s,'\n');
295 iscntrl(PL_multi_close)
297 PL_multi_close < 32 || PL_multi_close == 127
301 tmpbuf[1] = toCTRL(PL_multi_close);
307 *tmpbuf = (char)PL_multi_close;
311 q = strchr(s,'"') ? '\'' : '"';
312 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
320 Perl_deprecate(pTHX_ char *s)
322 if (ckWARN(WARN_DEPRECATED))
323 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED), "Use of %s is deprecated", s);
327 Perl_deprecate_old(pTHX_ char *s)
329 /* This function should NOT be called for any new deprecated warnings */
330 /* Use Perl_deprecate instead */
332 /* It is here to maintain backward compatibility with the pre-5.8 */
333 /* warnings category hierarchy. The "deprecated" category used to */
334 /* live under the "syntax" category. It is now a top-level category */
335 /* in its own right. */
337 if (ckWARN2(WARN_DEPRECATED, WARN_SYNTAX))
338 Perl_warner(aTHX_ packWARN2(WARN_DEPRECATED, WARN_SYNTAX),
339 "Use of %s is deprecated", s);
344 * Deprecate a comma-less variable list.
350 deprecate_old("comma-less variable list");
354 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
355 * utf16-to-utf8-reversed.
358 #ifdef PERL_CR_FILTER
362 register char *s = SvPVX(sv);
363 register char *e = s + SvCUR(sv);
364 /* outer loop optimized to do nothing if there are no CR-LFs */
366 if (*s++ == '\r' && *s == '\n') {
367 /* hit a CR-LF, need to copy the rest */
368 register char *d = s - 1;
371 if (*s == '\r' && s[1] == '\n')
382 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
384 I32 count = FILTER_READ(idx+1, sv, maxlen);
385 if (count > 0 && !maxlen)
393 * Initialize variables. Uses the Perl save_stack to save its state (for
394 * recursive calls to the parser).
398 Perl_lex_start(pTHX_ SV *line)
403 SAVEI32(PL_lex_dojoin);
404 SAVEI32(PL_lex_brackets);
405 SAVEI32(PL_lex_casemods);
406 SAVEI32(PL_lex_starts);
407 SAVEI32(PL_lex_state);
408 SAVEVPTR(PL_lex_inpat);
409 SAVEI32(PL_lex_inwhat);
410 if (PL_lex_state == LEX_KNOWNEXT) {
411 I32 toke = PL_nexttoke;
412 while (--toke >= 0) {
413 SAVEI32(PL_nexttype[toke]);
414 SAVEVPTR(PL_nextval[toke]);
416 SAVEI32(PL_nexttoke);
418 SAVECOPLINE(PL_curcop);
421 SAVEPPTR(PL_oldbufptr);
422 SAVEPPTR(PL_oldoldbufptr);
423 SAVEPPTR(PL_last_lop);
424 SAVEPPTR(PL_last_uni);
425 SAVEPPTR(PL_linestart);
426 SAVESPTR(PL_linestr);
427 SAVEPPTR(PL_lex_brackstack);
428 SAVEPPTR(PL_lex_casestack);
429 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
430 SAVESPTR(PL_lex_stuff);
431 SAVEI32(PL_lex_defer);
432 SAVEI32(PL_sublex_info.sub_inwhat);
433 SAVESPTR(PL_lex_repl);
435 SAVEINT(PL_lex_expect);
437 PL_lex_state = LEX_NORMAL;
441 New(899, PL_lex_brackstack, 120, char);
442 New(899, PL_lex_casestack, 12, char);
443 SAVEFREEPV(PL_lex_brackstack);
444 SAVEFREEPV(PL_lex_casestack);
446 *PL_lex_casestack = '\0';
449 PL_lex_stuff = Nullsv;
450 PL_lex_repl = Nullsv;
454 PL_sublex_info.sub_inwhat = 0;
456 if (SvREADONLY(PL_linestr))
457 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
458 s = SvPV(PL_linestr, len);
459 if (!len || s[len-1] != ';') {
460 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
461 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
462 sv_catpvn(PL_linestr, "\n;", 2);
464 SvTEMP_off(PL_linestr);
465 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
466 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
467 PL_last_lop = PL_last_uni = Nullch;
473 * Finalizer for lexing operations. Must be called when the parser is
474 * done with the lexer.
480 PL_doextract = FALSE;
485 * This subroutine has nothing to do with tilting, whether at windmills
486 * or pinball tables. Its name is short for "increment line". It
487 * increments the current line number in CopLINE(PL_curcop) and checks
488 * to see whether the line starts with a comment of the form
489 * # line 500 "foo.pm"
490 * If so, it sets the current line number and file to the values in the comment.
494 S_incline(pTHX_ char *s)
501 CopLINE_inc(PL_curcop);
504 while (SPACE_OR_TAB(*s)) s++;
505 if (strnEQ(s, "line", 4))
509 if (SPACE_OR_TAB(*s))
513 while (SPACE_OR_TAB(*s)) s++;
519 while (SPACE_OR_TAB(*s))
521 if (*s == '"' && (t = strchr(s+1, '"'))) {
526 for (t = s; !isSPACE(*t); t++) ;
529 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
531 if (*e != '\n' && *e != '\0')
532 return; /* false alarm */
537 CopFILE_free(PL_curcop);
538 CopFILE_set(PL_curcop, s);
541 CopLINE_set(PL_curcop, atoi(n)-1);
546 * Called to gobble the appropriate amount and type of whitespace.
547 * Skips comments as well.
551 S_skipspace(pTHX_ register char *s)
553 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
554 while (s < PL_bufend && SPACE_OR_TAB(*s))
560 SSize_t oldprevlen, oldoldprevlen;
561 SSize_t oldloplen = 0, oldunilen = 0;
562 while (s < PL_bufend && isSPACE(*s)) {
563 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
568 if (s < PL_bufend && *s == '#') {
569 while (s < PL_bufend && *s != '\n')
573 if (PL_in_eval && !PL_rsfp) {
580 /* only continue to recharge the buffer if we're at the end
581 * of the buffer, we're not reading from a source filter, and
582 * we're in normal lexing mode
584 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
585 PL_lex_state == LEX_FORMLINE)
588 /* try to recharge the buffer */
589 if ((s = filter_gets(PL_linestr, PL_rsfp,
590 (prevlen = SvCUR(PL_linestr)))) == Nullch)
592 /* end of file. Add on the -p or -n magic */
593 if (PL_minus_n || PL_minus_p) {
594 sv_setpv(PL_linestr,PL_minus_p ?
595 ";}continue{print or die qq(-p destination: $!\\n)" :
597 sv_catpv(PL_linestr,";}");
598 PL_minus_n = PL_minus_p = 0;
601 sv_setpv(PL_linestr,";");
603 /* reset variables for next time we lex */
604 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
606 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
607 PL_last_lop = PL_last_uni = Nullch;
609 /* Close the filehandle. Could be from -P preprocessor,
610 * STDIN, or a regular file. If we were reading code from
611 * STDIN (because the commandline held no -e or filename)
612 * then we don't close it, we reset it so the code can
613 * read from STDIN too.
616 if (PL_preprocess && !PL_in_eval)
617 (void)PerlProc_pclose(PL_rsfp);
618 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
619 PerlIO_clearerr(PL_rsfp);
621 (void)PerlIO_close(PL_rsfp);
626 /* not at end of file, so we only read another line */
627 /* make corresponding updates to old pointers, for yyerror() */
628 oldprevlen = PL_oldbufptr - PL_bufend;
629 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
631 oldunilen = PL_last_uni - PL_bufend;
633 oldloplen = PL_last_lop - PL_bufend;
634 PL_linestart = PL_bufptr = s + prevlen;
635 PL_bufend = s + SvCUR(PL_linestr);
637 PL_oldbufptr = s + oldprevlen;
638 PL_oldoldbufptr = s + oldoldprevlen;
640 PL_last_uni = s + oldunilen;
642 PL_last_lop = s + oldloplen;
645 /* debugger active and we're not compiling the debugger code,
646 * so store the line into the debugger's array of lines
648 if (PERLDB_LINE && PL_curstash != PL_debstash) {
649 SV *sv = NEWSV(85,0);
651 sv_upgrade(sv, SVt_PVMG);
652 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
655 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
662 * Check the unary operators to ensure there's no ambiguity in how they're
663 * used. An ambiguous piece of code would be:
665 * This doesn't mean rand() + 5. Because rand() is a unary operator,
666 * the +5 is its argument.
675 if (PL_oldoldbufptr != PL_last_uni)
677 while (isSPACE(*PL_last_uni))
679 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
680 if ((t = strchr(s, '(')) && t < PL_bufptr)
682 if (ckWARN_d(WARN_AMBIGUOUS)){
685 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
686 "Warning: Use of \"%s\" without parens is ambiguous",
693 * LOP : macro to build a list operator. Its behaviour has been replaced
694 * with a subroutine, S_lop() for which LOP is just another name.
697 #define LOP(f,x) return lop(f,x,s)
701 * Build a list operator (or something that might be one). The rules:
702 * - if we have a next token, then it's a list operator [why?]
703 * - if the next thing is an opening paren, then it's a function
704 * - else it's a list operator
708 S_lop(pTHX_ I32 f, int x, char *s)
715 PL_last_lop = PL_oldbufptr;
716 PL_last_lop_op = (OPCODE)f;
730 * When the lexer realizes it knows the next token (for instance,
731 * it is reordering tokens for the parser) then it can call S_force_next
732 * to know what token to return the next time the lexer is called. Caller
733 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
734 * handles the token correctly.
738 S_force_next(pTHX_ I32 type)
740 PL_nexttype[PL_nexttoke] = type;
742 if (PL_lex_state != LEX_KNOWNEXT) {
743 PL_lex_defer = PL_lex_state;
744 PL_lex_expect = PL_expect;
745 PL_lex_state = LEX_KNOWNEXT;
751 * When the lexer knows the next thing is a word (for instance, it has
752 * just seen -> and it knows that the next char is a word char, then
753 * it calls S_force_word to stick the next word into the PL_next lookahead.
756 * char *start : buffer position (must be within PL_linestr)
757 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
758 * int check_keyword : if true, Perl checks to make sure the word isn't
759 * a keyword (do this if the word is a label, e.g. goto FOO)
760 * int allow_pack : if true, : characters will also be allowed (require,
762 * int allow_initial_tick : used by the "sub" lexer only.
766 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
771 start = skipspace(start);
773 if (isIDFIRST_lazy_if(s,UTF) ||
774 (allow_pack && *s == ':') ||
775 (allow_initial_tick && *s == '\'') )
777 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
778 if (check_keyword && keyword(PL_tokenbuf, len))
780 if (token == METHOD) {
785 PL_expect = XOPERATOR;
788 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
789 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
797 * Called when the lexer wants $foo *foo &foo etc, but the program
798 * text only contains the "foo" portion. The first argument is a pointer
799 * to the "foo", and the second argument is the type symbol to prefix.
800 * Forces the next token to be a "WORD".
801 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
805 S_force_ident(pTHX_ register char *s, int kind)
808 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
809 PL_nextval[PL_nexttoke].opval = o;
812 o->op_private = OPpCONST_ENTERED;
813 /* XXX see note in pp_entereval() for why we forgo typo
814 warnings if the symbol must be introduced in an eval.
816 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
817 kind == '$' ? SVt_PV :
818 kind == '@' ? SVt_PVAV :
819 kind == '%' ? SVt_PVHV :
827 Perl_str_to_version(pTHX_ SV *sv)
832 char *start = SvPVx(sv,len);
833 bool utf = SvUTF8(sv) ? TRUE : FALSE;
834 char *end = start + len;
835 while (start < end) {
839 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
844 retval += ((NV)n)/nshift;
853 * Forces the next token to be a version number.
854 * If the next token appears to be an invalid version number, (e.g. "v2b"),
855 * and if "guessing" is TRUE, then no new token is created (and the caller
856 * must use an alternative parsing method).
860 S_force_version(pTHX_ char *s, int guessing)
862 OP *version = Nullop;
871 while (isDIGIT(*d) || *d == '_' || *d == '.')
873 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
875 s = scan_num(s, &yylval);
876 version = yylval.opval;
877 ver = cSVOPx(version)->op_sv;
878 if (SvPOK(ver) && !SvNIOK(ver)) {
879 (void)SvUPGRADE(ver, SVt_PVNV);
880 SvNVX(ver) = str_to_version(ver);
881 SvNOK_on(ver); /* hint that it is a version */
888 /* NOTE: The parser sees the package name and the VERSION swapped */
889 PL_nextval[PL_nexttoke].opval = version;
897 * Tokenize a quoted string passed in as an SV. It finds the next
898 * chunk, up to end of string or a backslash. It may make a new
899 * SV containing that chunk (if HINT_NEW_STRING is on). It also
904 S_tokeq(pTHX_ SV *sv)
915 s = SvPV_force(sv, len);
916 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
919 while (s < send && *s != '\\')
924 if ( PL_hints & HINT_NEW_STRING ) {
925 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
931 if (s + 1 < send && (s[1] == '\\'))
932 s++; /* all that, just for this */
937 SvCUR_set(sv, d - SvPVX(sv));
939 if ( PL_hints & HINT_NEW_STRING )
940 return new_constant(NULL, 0, "q", sv, pv, "q");
945 * Now come three functions related to double-quote context,
946 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
947 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
948 * interact with PL_lex_state, and create fake ( ... ) argument lists
949 * to handle functions and concatenation.
950 * They assume that whoever calls them will be setting up a fake
951 * join call, because each subthing puts a ',' after it. This lets
954 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
956 * (I'm not sure whether the spurious commas at the end of lcfirst's
957 * arguments and join's arguments are created or not).
962 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
964 * Pattern matching will set PL_lex_op to the pattern-matching op to
965 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
967 * OP_CONST and OP_READLINE are easy--just make the new op and return.
969 * Everything else becomes a FUNC.
971 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
972 * had an OP_CONST or OP_READLINE). This just sets us up for a
973 * call to S_sublex_push().
979 register I32 op_type = yylval.ival;
981 if (op_type == OP_NULL) {
982 yylval.opval = PL_lex_op;
986 if (op_type == OP_CONST || op_type == OP_READLINE) {
987 SV *sv = tokeq(PL_lex_stuff);
989 if (SvTYPE(sv) == SVt_PVIV) {
990 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
996 nsv = newSVpvn(p, len);
1002 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
1003 PL_lex_stuff = Nullsv;
1004 /* Allow <FH> // "foo" */
1005 if (op_type == OP_READLINE)
1006 PL_expect = XTERMORDORDOR;
1010 PL_sublex_info.super_state = PL_lex_state;
1011 PL_sublex_info.sub_inwhat = op_type;
1012 PL_sublex_info.sub_op = PL_lex_op;
1013 PL_lex_state = LEX_INTERPPUSH;
1017 yylval.opval = PL_lex_op;
1027 * Create a new scope to save the lexing state. The scope will be
1028 * ended in S_sublex_done. Returns a '(', starting the function arguments
1029 * to the uc, lc, etc. found before.
1030 * Sets PL_lex_state to LEX_INTERPCONCAT.
1038 PL_lex_state = PL_sublex_info.super_state;
1039 SAVEI32(PL_lex_dojoin);
1040 SAVEI32(PL_lex_brackets);
1041 SAVEI32(PL_lex_casemods);
1042 SAVEI32(PL_lex_starts);
1043 SAVEI32(PL_lex_state);
1044 SAVEVPTR(PL_lex_inpat);
1045 SAVEI32(PL_lex_inwhat);
1046 SAVECOPLINE(PL_curcop);
1047 SAVEPPTR(PL_bufptr);
1048 SAVEPPTR(PL_bufend);
1049 SAVEPPTR(PL_oldbufptr);
1050 SAVEPPTR(PL_oldoldbufptr);
1051 SAVEPPTR(PL_last_lop);
1052 SAVEPPTR(PL_last_uni);
1053 SAVEPPTR(PL_linestart);
1054 SAVESPTR(PL_linestr);
1055 SAVEPPTR(PL_lex_brackstack);
1056 SAVEPPTR(PL_lex_casestack);
1058 PL_linestr = PL_lex_stuff;
1059 PL_lex_stuff = Nullsv;
1061 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1062 = SvPVX(PL_linestr);
1063 PL_bufend += SvCUR(PL_linestr);
1064 PL_last_lop = PL_last_uni = Nullch;
1065 SAVEFREESV(PL_linestr);
1067 PL_lex_dojoin = FALSE;
1068 PL_lex_brackets = 0;
1069 New(899, PL_lex_brackstack, 120, char);
1070 New(899, PL_lex_casestack, 12, char);
1071 SAVEFREEPV(PL_lex_brackstack);
1072 SAVEFREEPV(PL_lex_casestack);
1073 PL_lex_casemods = 0;
1074 *PL_lex_casestack = '\0';
1076 PL_lex_state = LEX_INTERPCONCAT;
1077 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
1079 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1080 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1081 PL_lex_inpat = PL_sublex_info.sub_op;
1083 PL_lex_inpat = Nullop;
1090 * Restores lexer state after a S_sublex_push.
1096 if (!PL_lex_starts++) {
1097 SV *sv = newSVpvn("",0);
1098 if (SvUTF8(PL_linestr))
1100 PL_expect = XOPERATOR;
1101 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1105 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1106 PL_lex_state = LEX_INTERPCASEMOD;
1110 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1111 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1112 PL_linestr = PL_lex_repl;
1114 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1115 PL_bufend += SvCUR(PL_linestr);
1116 PL_last_lop = PL_last_uni = Nullch;
1117 SAVEFREESV(PL_linestr);
1118 PL_lex_dojoin = FALSE;
1119 PL_lex_brackets = 0;
1120 PL_lex_casemods = 0;
1121 *PL_lex_casestack = '\0';
1123 if (SvEVALED(PL_lex_repl)) {
1124 PL_lex_state = LEX_INTERPNORMAL;
1126 /* we don't clear PL_lex_repl here, so that we can check later
1127 whether this is an evalled subst; that means we rely on the
1128 logic to ensure sublex_done() is called again only via the
1129 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1132 PL_lex_state = LEX_INTERPCONCAT;
1133 PL_lex_repl = Nullsv;
1139 PL_bufend = SvPVX(PL_linestr);
1140 PL_bufend += SvCUR(PL_linestr);
1141 PL_expect = XOPERATOR;
1142 PL_sublex_info.sub_inwhat = 0;
1150 Extracts a pattern, double-quoted string, or transliteration. This
1153 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1154 processing a pattern (PL_lex_inpat is true), a transliteration
1155 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1157 Returns a pointer to the character scanned up to. Iff this is
1158 advanced from the start pointer supplied (ie if anything was
1159 successfully parsed), will leave an OP for the substring scanned
1160 in yylval. Caller must intuit reason for not parsing further
1161 by looking at the next characters herself.
1165 double-quoted style: \r and \n
1166 regexp special ones: \D \s
1168 backrefs: \1 (deprecated in substitution replacements)
1169 case and quoting: \U \Q \E
1170 stops on @ and $, but not for $ as tail anchor
1172 In transliterations:
1173 characters are VERY literal, except for - not at the start or end
1174 of the string, which indicates a range. scan_const expands the
1175 range to the full set of intermediate characters.
1177 In double-quoted strings:
1179 double-quoted style: \r and \n
1181 backrefs: \1 (deprecated)
1182 case and quoting: \U \Q \E
1185 scan_const does *not* construct ops to handle interpolated strings.
1186 It stops processing as soon as it finds an embedded $ or @ variable
1187 and leaves it to the caller to work out what's going on.
1189 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @::foo.
1191 $ in pattern could be $foo or could be tail anchor. Assumption:
1192 it's a tail anchor if $ is the last thing in the string, or if it's
1193 followed by one of ")| \n\t"
1195 \1 (backreferences) are turned into $1
1197 The structure of the code is
1198 while (there's a character to process) {
1199 handle transliteration ranges
1200 skip regexp comments
1201 skip # initiated comments in //x patterns
1202 check for embedded @foo
1203 check for embedded scalars
1205 leave intact backslashes from leave (below)
1206 deprecate \1 in strings and sub replacements
1207 handle string-changing backslashes \l \U \Q \E, etc.
1208 switch (what was escaped) {
1209 handle - in a transliteration (becomes a literal -)
1210 handle \132 octal characters
1211 handle 0x15 hex characters
1212 handle \cV (control V)
1213 handle printf backslashes (\f, \r, \n, etc)
1215 } (end if backslash)
1216 } (end while character to read)
1221 S_scan_const(pTHX_ char *start)
1223 register char *send = PL_bufend; /* end of the constant */
1224 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1225 register char *s = start; /* start of the constant */
1226 register char *d = SvPVX(sv); /* destination for copies */
1227 bool dorange = FALSE; /* are we in a translit range? */
1228 bool didrange = FALSE; /* did we just finish a range? */
1229 I32 has_utf8 = FALSE; /* Output constant is UTF8 */
1230 I32 this_utf8 = UTF; /* The source string is assumed to be UTF8 */
1233 const char *leaveit = /* set of acceptably-backslashed characters */
1235 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1238 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1239 /* If we are doing a trans and we know we want UTF8 set expectation */
1240 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
1241 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1245 while (s < send || dorange) {
1246 /* get transliterations out of the way (they're most literal) */
1247 if (PL_lex_inwhat == OP_TRANS) {
1248 /* expand a range A-Z to the full set of characters. AIE! */
1250 I32 i; /* current expanded character */
1251 I32 min; /* first character in range */
1252 I32 max; /* last character in range */
1255 char *c = (char*)utf8_hop((U8*)d, -1);
1259 *c = (char)UTF_TO_NATIVE(0xff);
1260 /* mark the range as done, and continue */
1266 i = d - SvPVX(sv); /* remember current offset */
1267 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1268 d = SvPVX(sv) + i; /* refresh d after realloc */
1269 d -= 2; /* eat the first char and the - */
1271 min = (U8)*d; /* first char in range */
1272 max = (U8)d[1]; /* last char in range */
1276 "Invalid range \"%c-%c\" in transliteration operator",
1277 (char)min, (char)max);
1281 if ((isLOWER(min) && isLOWER(max)) ||
1282 (isUPPER(min) && isUPPER(max))) {
1284 for (i = min; i <= max; i++)
1286 *d++ = NATIVE_TO_NEED(has_utf8,i);
1288 for (i = min; i <= max; i++)
1290 *d++ = NATIVE_TO_NEED(has_utf8,i);
1295 for (i = min; i <= max; i++)
1298 /* mark the range as done, and continue */
1304 /* range begins (ignore - as first or last char) */
1305 else if (*s == '-' && s+1 < send && s != start) {
1307 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
1310 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
1322 /* if we get here, we're not doing a transliteration */
1324 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1325 except for the last char, which will be done separately. */
1326 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1328 while (s < send && *s != ')')
1329 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1331 else if (s[2] == '{' /* This should match regcomp.c */
1332 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1335 char *regparse = s + (s[2] == '{' ? 3 : 4);
1338 while (count && (c = *regparse)) {
1339 if (c == '\\' && regparse[1])
1347 if (*regparse != ')') {
1348 regparse--; /* Leave one char for continuation. */
1349 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1351 while (s < regparse)
1352 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1356 /* likewise skip #-initiated comments in //x patterns */
1357 else if (*s == '#' && PL_lex_inpat &&
1358 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1359 while (s+1 < send && *s != '\n')
1360 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1363 /* check for embedded arrays
1364 (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-)
1366 else if (*s == '@' && s[1]
1367 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1])))
1370 /* check for embedded scalars. only stop if we're sure it's a
1373 else if (*s == '$') {
1374 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1376 if (s + 1 < send && !strchr("()| \r\n\t", s[1]))
1377 break; /* in regexp, $ might be tail anchor */
1380 /* End of else if chain - OP_TRANS rejoin rest */
1383 if (*s == '\\' && s+1 < send) {
1386 /* some backslashes we leave behind */
1387 if (*leaveit && *s && strchr(leaveit, *s)) {
1388 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
1389 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1393 /* deprecate \1 in strings and substitution replacements */
1394 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1395 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1397 if (ckWARN(WARN_SYNTAX))
1398 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
1403 /* string-change backslash escapes */
1404 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1409 /* if we get here, it's either a quoted -, or a digit */
1412 /* quoted - in transliterations */
1414 if (PL_lex_inwhat == OP_TRANS) {
1421 if (ckWARN(WARN_MISC) &&
1424 Perl_warner(aTHX_ packWARN(WARN_MISC),
1425 "Unrecognized escape \\%c passed through",
1427 /* default action is to copy the quoted character */
1428 goto default_action;
1431 /* \132 indicates an octal constant */
1432 case '0': case '1': case '2': case '3':
1433 case '4': case '5': case '6': case '7':
1437 uv = grok_oct(s, &len, &flags, NULL);
1440 goto NUM_ESCAPE_INSERT;
1442 /* \x24 indicates a hex constant */
1446 char* e = strchr(s, '}');
1447 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1448 PERL_SCAN_DISALLOW_PREFIX;
1453 yyerror("Missing right brace on \\x{}");
1457 uv = grok_hex(s, &len, &flags, NULL);
1463 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
1464 uv = grok_hex(s, &len, &flags, NULL);
1470 /* Insert oct or hex escaped character.
1471 * There will always enough room in sv since such
1472 * escapes will be longer than any UTF-8 sequence
1473 * they can end up as. */
1475 /* We need to map to chars to ASCII before doing the tests
1478 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(uv))) {
1479 if (!has_utf8 && uv > 255) {
1480 /* Might need to recode whatever we have
1481 * accumulated so far if it contains any
1484 * (Can't we keep track of that and avoid
1485 * this rescan? --jhi)
1489 for (c = (U8 *) SvPVX(sv); c < (U8 *)d; c++) {
1490 if (!NATIVE_IS_INVARIANT(*c)) {
1495 STRLEN offset = d - SvPVX(sv);
1497 d = SvGROW(sv, SvLEN(sv) + hicount + 1) + offset;
1501 while (src >= (U8 *)SvPVX(sv)) {
1502 if (!NATIVE_IS_INVARIANT(*src)) {
1503 U8 ch = NATIVE_TO_ASCII(*src);
1504 *dst-- = (U8)UTF8_EIGHT_BIT_LO(ch);
1505 *dst-- = (U8)UTF8_EIGHT_BIT_HI(ch);
1515 if (has_utf8 || uv > 255) {
1516 d = (char*)uvchr_to_utf8((U8*)d, uv);
1518 if (PL_lex_inwhat == OP_TRANS &&
1519 PL_sublex_info.sub_op) {
1520 PL_sublex_info.sub_op->op_private |=
1521 (PL_lex_repl ? OPpTRANS_FROM_UTF
1534 /* \N{LATIN SMALL LETTER A} is a named character */
1538 char* e = strchr(s, '}');
1544 yyerror("Missing right brace on \\N{}");
1548 if (e > s + 2 && s[1] == 'U' && s[2] == '+') {
1550 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1551 PERL_SCAN_DISALLOW_PREFIX;
1554 uv = grok_hex(s, &len, &flags, NULL);
1556 goto NUM_ESCAPE_INSERT;
1558 res = newSVpvn(s + 1, e - s - 1);
1559 res = new_constant( Nullch, 0, "charnames",
1560 res, Nullsv, "\\N{...}" );
1562 sv_utf8_upgrade(res);
1563 str = SvPV(res,len);
1564 #ifdef EBCDIC_NEVER_MIND
1565 /* charnames uses pack U and that has been
1566 * recently changed to do the below uni->native
1567 * mapping, so this would be redundant (and wrong,
1568 * the code point would be doubly converted).
1569 * But leave this in just in case the pack U change
1570 * gets revoked, but the semantics is still
1571 * desireable for charnames. --jhi */
1573 UV uv = utf8_to_uvchr((U8*)str, 0);
1576 U8 tmpbuf[UTF8_MAXLEN+1], *d;
1578 d = uvchr_to_utf8(tmpbuf, UNI_TO_NATIVE(uv));
1579 sv_setpvn(res, (char *)tmpbuf, d - tmpbuf);
1580 str = SvPV(res, len);
1584 if (!has_utf8 && SvUTF8(res)) {
1585 char *ostart = SvPVX(sv);
1586 SvCUR_set(sv, d - ostart);
1589 sv_utf8_upgrade(sv);
1590 /* this just broke our allocation above... */
1591 SvGROW(sv, (STRLEN)(send - start));
1592 d = SvPVX(sv) + SvCUR(sv);
1595 if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */
1596 char *odest = SvPVX(sv);
1598 SvGROW(sv, (SvLEN(sv) + len - (e - s + 4)));
1599 d = SvPVX(sv) + (d - odest);
1601 Copy(str, d, len, char);
1608 yyerror("Missing braces on \\N{}");
1611 /* \c is a control character */
1620 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
1624 /* printf-style backslashes, formfeeds, newlines, etc */
1626 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
1629 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
1632 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
1635 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
1638 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
1641 *d++ = ASCII_TO_NEED(has_utf8,'\033');
1644 *d++ = ASCII_TO_NEED(has_utf8,'\007');
1650 } /* end if (backslash) */
1653 /* If we started with encoded form, or already know we want it
1654 and then encode the next character */
1655 if ((has_utf8 || this_utf8) && !NATIVE_IS_INVARIANT((U8)(*s))) {
1657 UV uv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
1658 STRLEN need = UNISKIP(NATIVE_TO_UNI(uv));
1661 /* encoded value larger than old, need extra space (NOTE: SvCUR() not set here) */
1662 STRLEN off = d - SvPVX(sv);
1663 d = SvGROW(sv, SvLEN(sv) + (need-len)) + off;
1665 d = (char*)uvchr_to_utf8((U8*)d, uv);
1669 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1671 } /* while loop to process each character */
1673 /* terminate the string and set up the sv */
1675 SvCUR_set(sv, d - SvPVX(sv));
1676 if (SvCUR(sv) >= SvLEN(sv))
1677 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
1680 if (PL_encoding && !has_utf8) {
1681 sv_recode_to_utf8(sv, PL_encoding);
1686 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1687 PL_sublex_info.sub_op->op_private |=
1688 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1692 /* shrink the sv if we allocated more than we used */
1693 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1694 SvLEN_set(sv, SvCUR(sv) + 1);
1695 Renew(SvPVX(sv), SvLEN(sv), char);
1698 /* return the substring (via yylval) only if we parsed anything */
1699 if (s > PL_bufptr) {
1700 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1701 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1703 ( PL_lex_inwhat == OP_TRANS
1705 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1708 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1715 * Returns TRUE if there's more to the expression (e.g., a subscript),
1718 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1720 * ->[ and ->{ return TRUE
1721 * { and [ outside a pattern are always subscripts, so return TRUE
1722 * if we're outside a pattern and it's not { or [, then return FALSE
1723 * if we're in a pattern and the first char is a {
1724 * {4,5} (any digits around the comma) returns FALSE
1725 * if we're in a pattern and the first char is a [
1727 * [SOMETHING] has a funky algorithm to decide whether it's a
1728 * character class or not. It has to deal with things like
1729 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1730 * anything else returns TRUE
1733 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1736 S_intuit_more(pTHX_ register char *s)
1738 if (PL_lex_brackets)
1740 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1742 if (*s != '{' && *s != '[')
1747 /* In a pattern, so maybe we have {n,m}. */
1764 /* On the other hand, maybe we have a character class */
1767 if (*s == ']' || *s == '^')
1770 /* this is terrifying, and it works */
1771 int weight = 2; /* let's weigh the evidence */
1773 unsigned char un_char = 255, last_un_char;
1774 char *send = strchr(s,']');
1775 char tmpbuf[sizeof PL_tokenbuf * 4];
1777 if (!send) /* has to be an expression */
1780 Zero(seen,256,char);
1783 else if (isDIGIT(*s)) {
1785 if (isDIGIT(s[1]) && s[2] == ']')
1791 for (; s < send; s++) {
1792 last_un_char = un_char;
1793 un_char = (unsigned char)*s;
1798 weight -= seen[un_char] * 10;
1799 if (isALNUM_lazy_if(s+1,UTF)) {
1800 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1801 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1806 else if (*s == '$' && s[1] &&
1807 strchr("[#!%*<>()-=",s[1])) {
1808 if (/*{*/ strchr("])} =",s[2]))
1817 if (strchr("wds]",s[1]))
1819 else if (seen['\''] || seen['"'])
1821 else if (strchr("rnftbxcav",s[1]))
1823 else if (isDIGIT(s[1])) {
1825 while (s[1] && isDIGIT(s[1]))
1835 if (strchr("aA01! ",last_un_char))
1837 if (strchr("zZ79~",s[1]))
1839 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1840 weight -= 5; /* cope with negative subscript */
1843 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1844 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1849 if (keyword(tmpbuf, d - tmpbuf))
1852 if (un_char == last_un_char + 1)
1854 weight -= seen[un_char];
1859 if (weight >= 0) /* probably a character class */
1869 * Does all the checking to disambiguate
1871 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1872 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1874 * First argument is the stuff after the first token, e.g. "bar".
1876 * Not a method if bar is a filehandle.
1877 * Not a method if foo is a subroutine prototyped to take a filehandle.
1878 * Not a method if it's really "Foo $bar"
1879 * Method if it's "foo $bar"
1880 * Not a method if it's really "print foo $bar"
1881 * Method if it's really "foo package::" (interpreted as package->foo)
1882 * Not a method if bar is known to be a subroutine ("sub bar; foo bar")
1883 * Not a method if bar is a filehandle or package, but is quoted with
1888 S_intuit_method(pTHX_ char *start, GV *gv)
1890 char *s = start + (*start == '$');
1891 char tmpbuf[sizeof PL_tokenbuf];
1899 if ((cv = GvCVu(gv))) {
1900 char *proto = SvPVX(cv);
1910 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1911 /* start is the beginning of the possible filehandle/object,
1912 * and s is the end of it
1913 * tmpbuf is a copy of it
1916 if (*start == '$') {
1917 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1922 return *s == '(' ? FUNCMETH : METHOD;
1924 if (!keyword(tmpbuf, len)) {
1925 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1930 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1931 if (indirgv && GvCVu(indirgv))
1933 /* filehandle or package name makes it a method */
1934 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1936 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1937 return 0; /* no assumptions -- "=>" quotes bearword */
1939 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1940 newSVpvn(tmpbuf,len));
1941 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1945 return *s == '(' ? FUNCMETH : METHOD;
1953 * Return a string of Perl code to load the debugger. If PERL5DB
1954 * is set, it will return the contents of that, otherwise a
1955 * compile-time require of perl5db.pl.
1962 char *pdb = PerlEnv_getenv("PERL5DB");
1966 SETERRNO(0,SS_NORMAL);
1967 return "BEGIN { require 'perl5db.pl' }";
1973 /* Encoded script support. filter_add() effectively inserts a
1974 * 'pre-processing' function into the current source input stream.
1975 * Note that the filter function only applies to the current source file
1976 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1978 * The datasv parameter (which may be NULL) can be used to pass
1979 * private data to this instance of the filter. The filter function
1980 * can recover the SV using the FILTER_DATA macro and use it to
1981 * store private buffers and state information.
1983 * The supplied datasv parameter is upgraded to a PVIO type
1984 * and the IoDIRP/IoANY field is used to store the function pointer,
1985 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1986 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1987 * private use must be set using malloc'd pointers.
1991 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1996 if (!PL_rsfp_filters)
1997 PL_rsfp_filters = newAV();
1999 datasv = NEWSV(255,0);
2000 if (!SvUPGRADE(datasv, SVt_PVIO))
2001 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
2002 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
2003 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
2004 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
2005 (void*)funcp, SvPV_nolen(datasv)));
2006 av_unshift(PL_rsfp_filters, 1);
2007 av_store(PL_rsfp_filters, 0, datasv) ;
2012 /* Delete most recently added instance of this filter function. */
2014 Perl_filter_del(pTHX_ filter_t funcp)
2017 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", (void*)funcp));
2018 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
2020 /* if filter is on top of stack (usual case) just pop it off */
2021 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
2022 if (IoANY(datasv) == (void *)funcp) {
2023 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
2024 IoANY(datasv) = (void *)NULL;
2025 sv_free(av_pop(PL_rsfp_filters));
2029 /* we need to search for the correct entry and clear it */
2030 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
2034 /* Invoke the n'th filter function for the current rsfp. */
2036 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
2039 /* 0 = read one text line */
2044 if (!PL_rsfp_filters)
2046 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
2047 /* Provide a default input filter to make life easy. */
2048 /* Note that we append to the line. This is handy. */
2049 DEBUG_P(PerlIO_printf(Perl_debug_log,
2050 "filter_read %d: from rsfp\n", idx));
2054 int old_len = SvCUR(buf_sv) ;
2056 /* ensure buf_sv is large enough */
2057 SvGROW(buf_sv, (STRLEN)(old_len + maxlen)) ;
2058 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
2059 if (PerlIO_error(PL_rsfp))
2060 return -1; /* error */
2062 return 0 ; /* end of file */
2064 SvCUR_set(buf_sv, old_len + len) ;
2067 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2068 if (PerlIO_error(PL_rsfp))
2069 return -1; /* error */
2071 return 0 ; /* end of file */
2074 return SvCUR(buf_sv);
2076 /* Skip this filter slot if filter has been deleted */
2077 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
2078 DEBUG_P(PerlIO_printf(Perl_debug_log,
2079 "filter_read %d: skipped (filter deleted)\n",
2081 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
2083 /* Get function pointer hidden within datasv */
2084 funcp = (filter_t)IoANY(datasv);
2085 DEBUG_P(PerlIO_printf(Perl_debug_log,
2086 "filter_read %d: via function %p (%s)\n",
2087 idx, (void*)funcp, SvPV_nolen(datasv)));
2088 /* Call function. The function is expected to */
2089 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2090 /* Return: <0:error, =0:eof, >0:not eof */
2091 return (*funcp)(aTHX_ idx, buf_sv, maxlen);
2095 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2097 #ifdef PERL_CR_FILTER
2098 if (!PL_rsfp_filters) {
2099 filter_add(S_cr_textfilter,NULL);
2102 if (PL_rsfp_filters) {
2105 SvCUR_set(sv, 0); /* start with empty line */
2106 if (FILTER_READ(0, sv, 0) > 0)
2107 return ( SvPVX(sv) ) ;
2112 return (sv_gets(sv, fp, append));
2116 S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2120 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2124 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2125 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2127 return GvHV(gv); /* Foo:: */
2130 /* use constant CLASS => 'MyClass' */
2131 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2133 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2134 pkgname = SvPV_nolen(sv);
2138 return gv_stashpv(pkgname, FALSE);
2142 static char* exp_name[] =
2143 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2144 "ATTRTERM", "TERMBLOCK"
2151 Works out what to call the token just pulled out of the input
2152 stream. The yacc parser takes care of taking the ops we return and
2153 stitching them into a tree.
2159 if read an identifier
2160 if we're in a my declaration
2161 croak if they tried to say my($foo::bar)
2162 build the ops for a my() declaration
2163 if it's an access to a my() variable
2164 are we in a sort block?
2165 croak if my($a); $a <=> $b
2166 build ops for access to a my() variable
2167 if in a dq string, and they've said @foo and we can't find @foo
2169 build ops for a bareword
2170 if we already built the token before, use it.
2173 #ifdef USE_PURE_BISON
2175 Perl_yylex_r(pTHX_ YYSTYPE *lvalp, int *lcharp)
2180 yylval_pointer[yyactlevel] = lvalp;
2181 yychar_pointer[yyactlevel] = lcharp;
2182 if (yyactlevel >= YYMAXLEVEL)
2183 Perl_croak(aTHX_ "panic: YYMAXLEVEL");
2185 r = Perl_yylex(aTHX);
2195 #pragma segment Perl_yylex
2208 /* check if there's an identifier for us to look at */
2209 if (PL_pending_ident)
2210 return S_pending_ident(aTHX);
2212 /* no identifier pending identification */
2214 switch (PL_lex_state) {
2216 case LEX_NORMAL: /* Some compilers will produce faster */
2217 case LEX_INTERPNORMAL: /* code if we comment these out. */
2221 /* when we've already built the next token, just pull it out of the queue */
2224 yylval = PL_nextval[PL_nexttoke];
2226 PL_lex_state = PL_lex_defer;
2227 PL_expect = PL_lex_expect;
2228 PL_lex_defer = LEX_NORMAL;
2230 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2231 "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr,
2232 (IV)PL_nexttype[PL_nexttoke]); });
2234 return(PL_nexttype[PL_nexttoke]);
2236 /* interpolated case modifiers like \L \U, including \Q and \E.
2237 when we get here, PL_bufptr is at the \
2239 case LEX_INTERPCASEMOD:
2241 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2242 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2244 /* handle \E or end of string */
2245 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2249 if (PL_lex_casemods) {
2250 oldmod = PL_lex_casestack[--PL_lex_casemods];
2251 PL_lex_casestack[PL_lex_casemods] = '\0';
2253 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2255 PL_lex_state = LEX_INTERPCONCAT;
2259 if (PL_bufptr != PL_bufend)
2261 PL_lex_state = LEX_INTERPCONCAT;
2265 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2266 "### Saw case modifier at '%s'\n", PL_bufptr); });
2268 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2269 tmp = *s, *s = s[2], s[2] = (char)tmp; /* misordered... */
2270 if (strchr("LU", *s) &&
2271 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2273 PL_lex_casestack[--PL_lex_casemods] = '\0';
2276 if (PL_lex_casemods > 10) {
2277 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2278 if (newlb != PL_lex_casestack) {
2280 PL_lex_casestack = newlb;
2283 PL_lex_casestack[PL_lex_casemods++] = *s;
2284 PL_lex_casestack[PL_lex_casemods] = '\0';
2285 PL_lex_state = LEX_INTERPCONCAT;
2286 PL_nextval[PL_nexttoke].ival = 0;
2289 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2291 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2293 PL_nextval[PL_nexttoke].ival = OP_LC;
2295 PL_nextval[PL_nexttoke].ival = OP_UC;
2297 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2299 Perl_croak(aTHX_ "panic: yylex");
2302 if (PL_lex_starts) {
2311 case LEX_INTERPPUSH:
2312 return sublex_push();
2314 case LEX_INTERPSTART:
2315 if (PL_bufptr == PL_bufend)
2316 return sublex_done();
2317 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2318 "### Interpolated variable at '%s'\n", PL_bufptr); });
2320 PL_lex_dojoin = (*PL_bufptr == '@');
2321 PL_lex_state = LEX_INTERPNORMAL;
2322 if (PL_lex_dojoin) {
2323 PL_nextval[PL_nexttoke].ival = 0;
2325 force_ident("\"", '$');
2326 PL_nextval[PL_nexttoke].ival = 0;
2328 PL_nextval[PL_nexttoke].ival = 0;
2330 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2333 if (PL_lex_starts++) {
2339 case LEX_INTERPENDMAYBE:
2340 if (intuit_more(PL_bufptr)) {
2341 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2347 if (PL_lex_dojoin) {
2348 PL_lex_dojoin = FALSE;
2349 PL_lex_state = LEX_INTERPCONCAT;
2352 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2353 && SvEVALED(PL_lex_repl))
2355 if (PL_bufptr != PL_bufend)
2356 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2357 PL_lex_repl = Nullsv;
2360 case LEX_INTERPCONCAT:
2362 if (PL_lex_brackets)
2363 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2365 if (PL_bufptr == PL_bufend)
2366 return sublex_done();
2368 if (SvIVX(PL_linestr) == '\'') {
2369 SV *sv = newSVsv(PL_linestr);
2372 else if ( PL_hints & HINT_NEW_RE )
2373 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2374 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2378 s = scan_const(PL_bufptr);
2380 PL_lex_state = LEX_INTERPCASEMOD;
2382 PL_lex_state = LEX_INTERPSTART;
2385 if (s != PL_bufptr) {
2386 PL_nextval[PL_nexttoke] = yylval;
2389 if (PL_lex_starts++)
2399 PL_lex_state = LEX_NORMAL;
2400 s = scan_formline(PL_bufptr);
2401 if (!PL_lex_formbrack)
2407 PL_oldoldbufptr = PL_oldbufptr;
2410 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2411 exp_name[PL_expect], s);
2417 if (isIDFIRST_lazy_if(s,UTF))
2419 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2422 goto fake_eof; /* emulate EOF on ^D or ^Z */
2427 if (PL_lex_brackets)
2428 yyerror("Missing right curly or square bracket");
2429 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2430 "### Tokener got EOF\n");
2434 if (s++ < PL_bufend)
2435 goto retry; /* ignore stray nulls */
2438 if (!PL_in_eval && !PL_preambled) {
2439 PL_preambled = TRUE;
2440 sv_setpv(PL_linestr,incl_perldb());
2441 if (SvCUR(PL_linestr))
2442 sv_catpv(PL_linestr,";");
2444 while(AvFILLp(PL_preambleav) >= 0) {
2445 SV *tmpsv = av_shift(PL_preambleav);
2446 sv_catsv(PL_linestr, tmpsv);
2447 sv_catpv(PL_linestr, ";");
2450 sv_free((SV*)PL_preambleav);
2451 PL_preambleav = NULL;
2453 if (PL_minus_n || PL_minus_p) {
2454 sv_catpv(PL_linestr, "LINE: while (<>) {");
2456 sv_catpv(PL_linestr,"chomp;");
2459 if (strchr("/'\"", *PL_splitstr)
2460 && strchr(PL_splitstr + 1, *PL_splitstr))
2461 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
2464 s = "'~#\200\1'"; /* surely one char is unused...*/
2465 while (s[1] && strchr(PL_splitstr, *s)) s++;
2467 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s%c",
2468 "q" + (delim == '\''), delim);
2469 for (s = PL_splitstr; *s; s++) {
2471 sv_catpvn(PL_linestr, "\\", 1);
2472 sv_catpvn(PL_linestr, s, 1);
2474 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2478 sv_catpv(PL_linestr,"our @F=split(' ');");
2481 sv_catpv(PL_linestr, "\n");
2482 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2483 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2484 PL_last_lop = PL_last_uni = Nullch;
2485 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2486 SV *sv = NEWSV(85,0);
2488 sv_upgrade(sv, SVt_PVMG);
2489 sv_setsv(sv,PL_linestr);
2492 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2497 bof = PL_rsfp ? TRUE : FALSE;
2498 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2501 if (PL_preprocess && !PL_in_eval)
2502 (void)PerlProc_pclose(PL_rsfp);
2503 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2504 PerlIO_clearerr(PL_rsfp);
2506 (void)PerlIO_close(PL_rsfp);
2508 PL_doextract = FALSE;
2510 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2511 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2512 sv_catpv(PL_linestr,";}");
2513 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2514 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2515 PL_last_lop = PL_last_uni = Nullch;
2516 PL_minus_n = PL_minus_p = 0;
2519 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2520 PL_last_lop = PL_last_uni = Nullch;
2521 sv_setpv(PL_linestr,"");
2522 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2524 /* if it looks like the start of a BOM, check if it in fact is */
2525 else if (bof && (!*s || *(U8*)s == 0xEF || *(U8*)s >= 0xFE)) {
2526 #ifdef PERLIO_IS_STDIO
2527 # ifdef __GNU_LIBRARY__
2528 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
2529 # define FTELL_FOR_PIPE_IS_BROKEN
2533 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2534 # define FTELL_FOR_PIPE_IS_BROKEN
2539 #ifdef FTELL_FOR_PIPE_IS_BROKEN
2540 /* This loses the possibility to detect the bof
2541 * situation on perl -P when the libc5 is being used.
2542 * Workaround? Maybe attach some extra state to PL_rsfp?
2545 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2547 bof = PerlIO_tell(PL_rsfp) == (Off_t)SvCUR(PL_linestr);
2550 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2551 s = swallow_bom((U8*)s);
2555 /* Incest with pod. */
2556 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2557 sv_setpv(PL_linestr, "");
2558 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2559 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2560 PL_last_lop = PL_last_uni = Nullch;
2561 PL_doextract = FALSE;
2565 } while (PL_doextract);
2566 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2567 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2568 SV *sv = NEWSV(85,0);
2570 sv_upgrade(sv, SVt_PVMG);
2571 sv_setsv(sv,PL_linestr);
2574 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2576 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2577 PL_last_lop = PL_last_uni = Nullch;
2578 if (CopLINE(PL_curcop) == 1) {
2579 while (s < PL_bufend && isSPACE(*s))
2581 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2585 if (*s == '#' && *(s+1) == '!')
2587 #ifdef ALTERNATE_SHEBANG
2589 static char as[] = ALTERNATE_SHEBANG;
2590 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2591 d = s + (sizeof(as) - 1);
2593 #endif /* ALTERNATE_SHEBANG */
2602 while (*d && !isSPACE(*d))
2606 #ifdef ARG_ZERO_IS_SCRIPT
2607 if (ipathend > ipath) {
2609 * HP-UX (at least) sets argv[0] to the script name,
2610 * which makes $^X incorrect. And Digital UNIX and Linux,
2611 * at least, set argv[0] to the basename of the Perl
2612 * interpreter. So, having found "#!", we'll set it right.
2614 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV)); /* $^X */
2615 assert(SvPOK(x) || SvGMAGICAL(x));
2616 if (sv_eq(x, CopFILESV(PL_curcop))) {
2617 sv_setpvn(x, ipath, ipathend - ipath);
2623 char *bstart = SvPV(CopFILESV(PL_curcop),blen);
2624 char *lstart = SvPV(x,llen);
2626 bstart += blen - llen;
2627 if (strnEQ(bstart, lstart, llen) && bstart[-1] == '/') {
2628 sv_setpvn(x, ipath, ipathend - ipath);
2633 TAINT_NOT; /* $^X is always tainted, but that's OK */
2635 #endif /* ARG_ZERO_IS_SCRIPT */
2640 d = instr(s,"perl -");
2642 d = instr(s,"perl");
2644 /* avoid getting into infinite loops when shebang
2645 * line contains "Perl" rather than "perl" */
2647 for (d = ipathend-4; d >= ipath; --d) {
2648 if ((*d == 'p' || *d == 'P')
2649 && !ibcmp(d, "perl", 4))
2659 #ifdef ALTERNATE_SHEBANG
2661 * If the ALTERNATE_SHEBANG on this system starts with a
2662 * character that can be part of a Perl expression, then if
2663 * we see it but not "perl", we're probably looking at the
2664 * start of Perl code, not a request to hand off to some
2665 * other interpreter. Similarly, if "perl" is there, but
2666 * not in the first 'word' of the line, we assume the line
2667 * contains the start of the Perl program.
2669 if (d && *s != '#') {
2671 while (*c && !strchr("; \t\r\n\f\v#", *c))
2674 d = Nullch; /* "perl" not in first word; ignore */
2676 *s = '#'; /* Don't try to parse shebang line */
2678 #endif /* ALTERNATE_SHEBANG */
2679 #ifndef MACOS_TRADITIONAL
2684 !instr(s,"indir") &&
2685 instr(PL_origargv[0],"perl"))
2691 while (s < PL_bufend && isSPACE(*s))
2693 if (s < PL_bufend) {
2694 Newz(899,newargv,PL_origargc+3,char*);
2696 while (s < PL_bufend && !isSPACE(*s))
2699 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2702 newargv = PL_origargv;
2704 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
2705 Perl_croak(aTHX_ "Can't exec %s", ipath);
2709 U32 oldpdb = PL_perldb;
2710 bool oldn = PL_minus_n;
2711 bool oldp = PL_minus_p;
2713 while (*d && !isSPACE(*d)) d++;
2714 while (SPACE_OR_TAB(*d)) d++;
2717 bool switches_done = PL_doswitches;
2719 if (*d == 'M' || *d == 'm') {
2721 while (*d && !isSPACE(*d)) d++;
2722 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2725 d = moreswitches(d);
2727 if ((PERLDB_LINE && !oldpdb) ||
2728 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2729 /* if we have already added "LINE: while (<>) {",
2730 we must not do it again */
2732 sv_setpv(PL_linestr, "");
2733 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2734 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2735 PL_last_lop = PL_last_uni = Nullch;
2736 PL_preambled = FALSE;
2738 (void)gv_fetchfile(PL_origfilename);
2741 if (PL_doswitches && !switches_done) {
2742 int argc = PL_origargc;
2743 char **argv = PL_origargv;
2746 } while (argc && argv[0][0] == '-' && argv[0][1]);
2747 init_argv_symbols(argc,argv);
2753 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2755 PL_lex_state = LEX_FORMLINE;
2760 #ifdef PERL_STRICT_CR
2761 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2763 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2765 case ' ': case '\t': case '\f': case 013:
2766 #ifdef MACOS_TRADITIONAL
2773 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2774 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2775 /* handle eval qq[#line 1 "foo"\n ...] */
2776 CopLINE_dec(PL_curcop);
2780 while (s < d && *s != '\n')
2784 else if (s > d) /* Found by Ilya: feed random input to Perl. */
2785 Perl_croak(aTHX_ "panic: input overflow");
2787 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2789 PL_lex_state = LEX_FORMLINE;
2799 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2806 while (s < PL_bufend && SPACE_OR_TAB(*s))
2809 if (strnEQ(s,"=>",2)) {
2810 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2811 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2812 "### Saw unary minus before =>, forcing word '%s'\n", s);
2814 OPERATOR('-'); /* unary minus */
2816 PL_last_uni = PL_oldbufptr;
2818 case 'r': ftst = OP_FTEREAD; break;
2819 case 'w': ftst = OP_FTEWRITE; break;
2820 case 'x': ftst = OP_FTEEXEC; break;
2821 case 'o': ftst = OP_FTEOWNED; break;
2822 case 'R': ftst = OP_FTRREAD; break;
2823 case 'W': ftst = OP_FTRWRITE; break;
2824 case 'X': ftst = OP_FTREXEC; break;
2825 case 'O': ftst = OP_FTROWNED; break;
2826 case 'e': ftst = OP_FTIS; break;
2827 case 'z': ftst = OP_FTZERO; break;
2828 case 's': ftst = OP_FTSIZE; break;
2829 case 'f': ftst = OP_FTFILE; break;
2830 case 'd': ftst = OP_FTDIR; break;
2831 case 'l': ftst = OP_FTLINK; break;
2832 case 'p': ftst = OP_FTPIPE; break;
2833 case 'S': ftst = OP_FTSOCK; break;
2834 case 'u': ftst = OP_FTSUID; break;
2835 case 'g': ftst = OP_FTSGID; break;
2836 case 'k': ftst = OP_FTSVTX; break;
2837 case 'b': ftst = OP_FTBLK; break;
2838 case 'c': ftst = OP_FTCHR; break;
2839 case 't': ftst = OP_FTTTY; break;
2840 case 'T': ftst = OP_FTTEXT; break;
2841 case 'B': ftst = OP_FTBINARY; break;
2842 case 'M': case 'A': case 'C':
2843 gv_fetchpv("\024",TRUE, SVt_PV);
2845 case 'M': ftst = OP_FTMTIME; break;
2846 case 'A': ftst = OP_FTATIME; break;
2847 case 'C': ftst = OP_FTCTIME; break;
2855 PL_last_lop_op = (OPCODE)ftst;
2856 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2857 "### Saw file test %c\n", (int)ftst);
2862 /* Assume it was a minus followed by a one-letter named
2863 * subroutine call (or a -bareword), then. */
2864 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2865 "### %c looked like a file test but was not\n",
2874 if (PL_expect == XOPERATOR)
2879 else if (*s == '>') {
2882 if (isIDFIRST_lazy_if(s,UTF)) {
2883 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2891 if (PL_expect == XOPERATOR)
2894 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2896 OPERATOR('-'); /* unary minus */
2903 if (PL_expect == XOPERATOR)
2908 if (PL_expect == XOPERATOR)
2911 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2917 if (PL_expect != XOPERATOR) {
2918 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2919 PL_expect = XOPERATOR;
2920 force_ident(PL_tokenbuf, '*');
2933 if (PL_expect == XOPERATOR) {
2937 PL_tokenbuf[0] = '%';
2938 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2939 if (!PL_tokenbuf[1]) {
2941 yyerror("Final % should be \\% or %name");
2944 PL_pending_ident = '%';
2963 switch (PL_expect) {
2966 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2968 PL_bufptr = s; /* update in case we back off */
2974 PL_expect = XTERMBLOCK;
2978 while (isIDFIRST_lazy_if(s,UTF)) {
2979 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2980 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2981 if (tmp < 0) tmp = -tmp;
2997 d = scan_str(d,TRUE,TRUE);
2999 /* MUST advance bufptr here to avoid bogus
3000 "at end of line" context messages from yyerror().
3002 PL_bufptr = s + len;
3003 yyerror("Unterminated attribute parameter in attribute list");
3006 return 0; /* EOF indicator */
3010 SV *sv = newSVpvn(s, len);
3011 sv_catsv(sv, PL_lex_stuff);
3012 attrs = append_elem(OP_LIST, attrs,
3013 newSVOP(OP_CONST, 0, sv));
3014 SvREFCNT_dec(PL_lex_stuff);
3015 PL_lex_stuff = Nullsv;
3018 /* NOTE: any CV attrs applied here need to be part of
3019 the CVf_BUILTIN_ATTRS define in cv.h! */
3020 if (!PL_in_my && len == 6 && strnEQ(s, "lvalue", len))
3021 CvLVALUE_on(PL_compcv);
3022 else if (!PL_in_my && len == 6 && strnEQ(s, "locked", len))
3023 CvLOCKED_on(PL_compcv);
3024 else if (!PL_in_my && len == 6 && strnEQ(s, "method", len))
3025 CvMETHOD_on(PL_compcv);
3027 else if (PL_in_my == KEY_our && len == 6 &&
3028 strnEQ(s, "unique", len))
3029 GvUNIQUE_on(cGVOPx_gv(yylval.opval));
3031 /* After we've set the flags, it could be argued that
3032 we don't need to do the attributes.pm-based setting
3033 process, and shouldn't bother appending recognized
3034 flags. To experiment with that, uncomment the
3035 following "else". (Note that's already been
3036 uncommented. That keeps the above-applied built-in
3037 attributes from being intercepted (and possibly
3038 rejected) by a package's attribute routines, but is
3039 justified by the performance win for the common case
3040 of applying only built-in attributes.) */
3042 attrs = append_elem(OP_LIST, attrs,
3043 newSVOP(OP_CONST, 0,
3047 if (*s == ':' && s[1] != ':')
3050 break; /* require real whitespace or :'s */
3052 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3053 if (*s != ';' && *s != '}' && *s != tmp && (tmp != '=' || *s != ')')) {
3054 char q = ((*s == '\'') ? '"' : '\'');
3055 /* If here for an expression, and parsed no attrs, back off. */
3056 if (tmp == '=' && !attrs) {
3060 /* MUST advance bufptr here to avoid bogus "at end of line"
3061 context messages from yyerror().
3065 yyerror("Unterminated attribute list");
3067 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
3075 PL_nextval[PL_nexttoke].opval = attrs;
3083 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3084 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3100 if (PL_lex_brackets <= 0)
3101 yyerror("Unmatched right square bracket");
3104 if (PL_lex_state == LEX_INTERPNORMAL) {
3105 if (PL_lex_brackets == 0) {
3106 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3107 PL_lex_state = LEX_INTERPEND;
3114 if (PL_lex_brackets > 100) {
3115 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
3116 if (newlb != PL_lex_brackstack) {
3118 PL_lex_brackstack = newlb;
3121 switch (PL_expect) {
3123 if (PL_lex_formbrack) {
3127 if (PL_oldoldbufptr == PL_last_lop)
3128 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3130 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3131 OPERATOR(HASHBRACK);
3133 while (s < PL_bufend && SPACE_OR_TAB(*s))
3136 PL_tokenbuf[0] = '\0';
3137 if (d < PL_bufend && *d == '-') {
3138 PL_tokenbuf[0] = '-';
3140 while (d < PL_bufend && SPACE_OR_TAB(*d))
3143 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3144 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3146 while (d < PL_bufend && SPACE_OR_TAB(*d))
3149 char minus = (PL_tokenbuf[0] == '-');
3150 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3158 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3163 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3168 if (PL_oldoldbufptr == PL_last_lop)
3169 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3171 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3174 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
3176 /* This hack is to get the ${} in the message. */
3178 yyerror("syntax error");
3181 OPERATOR(HASHBRACK);
3183 /* This hack serves to disambiguate a pair of curlies
3184 * as being a block or an anon hash. Normally, expectation
3185 * determines that, but in cases where we're not in a
3186 * position to expect anything in particular (like inside
3187 * eval"") we have to resolve the ambiguity. This code
3188 * covers the case where the first term in the curlies is a
3189 * quoted string. Most other cases need to be explicitly
3190 * disambiguated by prepending a `+' before the opening
3191 * curly in order to force resolution as an anon hash.
3193 * XXX should probably propagate the outer expectation
3194 * into eval"" to rely less on this hack, but that could
3195 * potentially break current behavior of eval"".
3199 if (*s == '\'' || *s == '"' || *s == '`') {
3200 /* common case: get past first string, handling escapes */
3201 for (t++; t < PL_bufend && *t != *s;)
3202 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3206 else if (*s == 'q') {
3209 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3213 char open, close, term;
3216 while (t < PL_bufend && isSPACE(*t))
3220 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3224 for (t++; t < PL_bufend; t++) {
3225 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3227 else if (*t == open)
3231 for (t++; t < PL_bufend; t++) {
3232 if (*t == '\\' && t+1 < PL_bufend)
3234 else if (*t == close && --brackets <= 0)
3236 else if (*t == open)
3242 else if (isALNUM_lazy_if(t,UTF)) {
3244 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3247 while (t < PL_bufend && isSPACE(*t))
3249 /* if comma follows first term, call it an anon hash */
3250 /* XXX it could be a comma expression with loop modifiers */
3251 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3252 || (*t == '=' && t[1] == '>')))
3253 OPERATOR(HASHBRACK);
3254 if (PL_expect == XREF)
3257 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3263 yylval.ival = CopLINE(PL_curcop);
3264 if (isSPACE(*s) || *s == '#')
3265 PL_copline = NOLINE; /* invalidate current command line number */
3270 if (PL_lex_brackets <= 0)
3271 yyerror("Unmatched right curly bracket");
3273 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3274 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3275 PL_lex_formbrack = 0;
3276 if (PL_lex_state == LEX_INTERPNORMAL) {
3277 if (PL_lex_brackets == 0) {
3278 if (PL_expect & XFAKEBRACK) {
3279 PL_expect &= XENUMMASK;
3280 PL_lex_state = LEX_INTERPEND;
3282 return yylex(); /* ignore fake brackets */
3284 if (*s == '-' && s[1] == '>')
3285 PL_lex_state = LEX_INTERPENDMAYBE;
3286 else if (*s != '[' && *s != '{')
3287 PL_lex_state = LEX_INTERPEND;
3290 if (PL_expect & XFAKEBRACK) {
3291 PL_expect &= XENUMMASK;
3293 return yylex(); /* ignore fake brackets */
3303 if (PL_expect == XOPERATOR) {
3304 if (ckWARN(WARN_SEMICOLON)
3305 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3307 CopLINE_dec(PL_curcop);
3308 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
3309 CopLINE_inc(PL_curcop);
3314 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3316 PL_expect = XOPERATOR;
3317 force_ident(PL_tokenbuf, '&');
3321 yylval.ival = (OPpENTERSUB_AMPER<<8);
3340 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3341 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Reversed %c= operator",(int)tmp);
3343 if (PL_expect == XSTATE && isALPHA(tmp) &&
3344 (s == PL_linestart+1 || s[-2] == '\n') )
3346 if (PL_in_eval && !PL_rsfp) {
3351 if (strnEQ(s,"=cut",4)) {
3365 PL_doextract = TRUE;
3368 if (PL_lex_brackets < PL_lex_formbrack) {
3370 #ifdef PERL_STRICT_CR
3371 for (t = s; SPACE_OR_TAB(*t); t++) ;
3373 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3375 if (*t == '\n' || *t == '#') {
3393 if (PL_expect != XOPERATOR) {
3394 if (s[1] != '<' && !strchr(s,'>'))
3397 s = scan_heredoc(s);
3399 s = scan_inputsymbol(s);
3400 TERM(sublex_start());
3405 SHop(OP_LEFT_SHIFT);
3419 SHop(OP_RIGHT_SHIFT);
3428 if (PL_expect == XOPERATOR) {
3429 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3432 return ','; /* grandfather non-comma-format format */
3436 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3437 PL_tokenbuf[0] = '@';
3438 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3439 sizeof PL_tokenbuf - 1, FALSE);
3440 if (PL_expect == XOPERATOR)
3441 no_op("Array length", s);
3442 if (!PL_tokenbuf[1])
3444 PL_expect = XOPERATOR;
3445 PL_pending_ident = '#';
3449 PL_tokenbuf[0] = '$';
3450 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3451 sizeof PL_tokenbuf - 1, FALSE);
3452 if (PL_expect == XOPERATOR)
3454 if (!PL_tokenbuf[1]) {
3456 yyerror("Final $ should be \\$ or $name");
3460 /* This kludge not intended to be bulletproof. */
3461 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3462 yylval.opval = newSVOP(OP_CONST, 0,
3463 newSViv(PL_compiling.cop_arybase));
3464 yylval.opval->op_private = OPpCONST_ARYBASE;
3470 if (PL_lex_state == LEX_NORMAL)
3473 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3476 PL_tokenbuf[0] = '@';
3477 if (ckWARN(WARN_SYNTAX)) {
3479 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3482 PL_bufptr = skipspace(PL_bufptr);
3483 while (t < PL_bufend && *t != ']')
3485 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3486 "Multidimensional syntax %.*s not supported",
3487 (t - PL_bufptr) + 1, PL_bufptr);
3491 else if (*s == '{') {
3492 PL_tokenbuf[0] = '%';
3493 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3494 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3496 char tmpbuf[sizeof PL_tokenbuf];
3498 for (t++; isSPACE(*t); t++) ;
3499 if (isIDFIRST_lazy_if(t,UTF)) {
3500 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3501 for (; isSPACE(*t); t++) ;
3502 if (*t == ';' && get_cv(tmpbuf, FALSE))
3503 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3504 "You need to quote \"%s\"", tmpbuf);
3510 PL_expect = XOPERATOR;
3511 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3512 bool islop = (PL_last_lop == PL_oldoldbufptr);
3513 if (!islop || PL_last_lop_op == OP_GREPSTART)
3514 PL_expect = XOPERATOR;
3515 else if (strchr("$@\"'`q", *s))
3516 PL_expect = XTERM; /* e.g. print $fh "foo" */
3517 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3518 PL_expect = XTERM; /* e.g. print $fh &sub */
3519 else if (isIDFIRST_lazy_if(s,UTF)) {
3520 char tmpbuf[sizeof PL_tokenbuf];
3521 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3522 if ((tmp = keyword(tmpbuf, len))) {
3523 /* binary operators exclude handle interpretations */
3535 PL_expect = XTERM; /* e.g. print $fh length() */
3540 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3541 if (gv && GvCVu(gv))
3542 PL_expect = XTERM; /* e.g. print $fh subr() */
3545 else if (isDIGIT(*s))
3546 PL_expect = XTERM; /* e.g. print $fh 3 */
3547 else if (*s == '.' && isDIGIT(s[1]))
3548 PL_expect = XTERM; /* e.g. print $fh .3 */
3549 else if (strchr("?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3550 PL_expect = XTERM; /* e.g. print $fh -1 */
3551 else if (*s == '/' && !isSPACE(s[1]) && s[1] != '=' && s[1] != '/')
3552 PL_expect = XTERM; /* e.g. print $fh /.../
3553 XXX except DORDOR operator */
3554 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3555 PL_expect = XTERM; /* print $fh <<"EOF" */
3557 PL_pending_ident = '$';
3561 if (PL_expect == XOPERATOR)
3563 PL_tokenbuf[0] = '@';
3564 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3565 if (!PL_tokenbuf[1]) {
3567 yyerror("Final @ should be \\@ or @name");
3570 if (PL_lex_state == LEX_NORMAL)
3572 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3574 PL_tokenbuf[0] = '%';
3576 /* Warn about @ where they meant $. */
3577 if (ckWARN(WARN_SYNTAX)) {
3578 if (*s == '[' || *s == '{') {
3580 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3582 if (*t == '}' || *t == ']') {
3584 PL_bufptr = skipspace(PL_bufptr);
3585 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3586 "Scalar value %.*s better written as $%.*s",
3587 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3592 PL_pending_ident = '@';
3595 case '/': /* may be division, defined-or, or pattern */
3596 if (PL_expect == XTERMORDORDOR && s[1] == '/') {
3600 case '?': /* may either be conditional or pattern */
3601 if(PL_expect == XOPERATOR) {
3609 /* A // operator. */
3619 /* Disable warning on "study /blah/" */
3620 if (PL_oldoldbufptr == PL_last_uni
3621 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3622 || memNE(PL_last_uni, "study", 5)
3623 || isALNUM_lazy_if(PL_last_uni+5,UTF)
3626 s = scan_pat(s,OP_MATCH);
3627 TERM(sublex_start());
3631 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3632 #ifdef PERL_STRICT_CR
3635 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3637 && (s == PL_linestart || s[-1] == '\n') )
3639 PL_lex_formbrack = 0;
3643 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3649 yylval.ival = OPf_SPECIAL;
3655 if (PL_expect != XOPERATOR)
3660 case '0': case '1': case '2': case '3': case '4':
3661 case '5': case '6': case '7': case '8': case '9':
3662 s = scan_num(s, &yylval);
3663 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3664 "### Saw number in '%s'\n", s);
3666 if (PL_expect == XOPERATOR)
3671 s = scan_str(s,FALSE,FALSE);
3672 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3673 "### Saw string before '%s'\n", s);
3675 if (PL_expect == XOPERATOR) {
3676 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3679 return ','; /* grandfather non-comma-format format */
3685 missingterm((char*)0);
3686 yylval.ival = OP_CONST;
3687 TERM(sublex_start());
3690 s = scan_str(s,FALSE,FALSE);
3691 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3692 "### Saw string before '%s'\n", s);
3694 if (PL_expect == XOPERATOR) {
3695 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3698 return ','; /* grandfather non-comma-format format */
3704 missingterm((char*)0);
3705 yylval.ival = OP_CONST;
3706 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3707 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
3708 yylval.ival = OP_STRINGIFY;
3712 TERM(sublex_start());
3715 s = scan_str(s,FALSE,FALSE);
3716 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3717 "### Saw backtick string before '%s'\n", s);
3719 if (PL_expect == XOPERATOR)
3720 no_op("Backticks",s);
3722 missingterm((char*)0);
3723 yylval.ival = OP_BACKTICK;
3725 TERM(sublex_start());
3729 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3730 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),"Can't use \\%c to mean $%c in expression",
3732 if (PL_expect == XOPERATOR)
3733 no_op("Backslash",s);
3737 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3741 while (isDIGIT(*start) || *start == '_')
3743 if (*start == '.' && isDIGIT(start[1])) {
3744 s = scan_num(s, &yylval);
3747 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3748 else if (!isALPHA(*start) && (PL_expect == XTERM
3749 || PL_expect == XREF || PL_expect == XSTATE
3750 || PL_expect == XTERMORDORDOR)) {
3754 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3757 s = scan_num(s, &yylval);
3764 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3799 I32 orig_keyword = 0;
3804 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3806 /* Some keywords can be followed by any delimiter, including ':' */
3807 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3808 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3809 (PL_tokenbuf[0] == 'q' &&
3810 strchr("qwxr", PL_tokenbuf[1])))));
3812 /* x::* is just a word, unless x is "CORE" */
3813 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3817 while (d < PL_bufend && isSPACE(*d))
3818 d++; /* no comments skipped here, or s### is misparsed */
3820 /* Is this a label? */
3821 if (!tmp && PL_expect == XSTATE
3822 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3824 yylval.pval = savepv(PL_tokenbuf);
3829 /* Check for keywords */
3830 tmp = keyword(PL_tokenbuf, len);
3832 /* Is this a word before a => operator? */
3833 if (*d == '=' && d[1] == '>') {
3835 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3836 yylval.opval->op_private = OPpCONST_BARE;
3837 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
3838 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3842 if (tmp < 0) { /* second-class keyword? */
3843 GV *ogv = Nullgv; /* override (winner) */
3844 GV *hgv = Nullgv; /* hidden (loser) */
3845 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3847 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3850 if (GvIMPORTED_CV(gv))
3852 else if (! CvMETHOD(cv))
3856 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3857 (gv = *gvp) != (GV*)&PL_sv_undef &&
3858 GvCVu(gv) && GvIMPORTED_CV(gv))
3865 tmp = 0; /* overridden by import or by GLOBAL */
3868 && -tmp==KEY_lock /* XXX generalizable kludge */
3870 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3872 tmp = 0; /* any sub overrides "weak" keyword */
3874 else { /* no override */
3876 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
3877 Perl_warner(aTHX_ packWARN(WARN_MISC),
3878 "dump() better written as CORE::dump()");
3882 if (ckWARN(WARN_AMBIGUOUS) && hgv
3883 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3884 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
3885 "Ambiguous call resolved as CORE::%s(), %s",
3886 GvENAME(hgv), "qualify as such or use &");
3893 default: /* not a keyword */
3897 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3899 /* Get the rest if it looks like a package qualifier */
3901 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3903 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3906 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3907 *s == '\'' ? "'" : "::");
3912 if (PL_expect == XOPERATOR) {
3913 if (PL_bufptr == PL_linestart) {
3914 CopLINE_dec(PL_curcop);
3915 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
3916 CopLINE_inc(PL_curcop);
3919 no_op("Bareword",s);
3922 /* Look for a subroutine with this name in current package,
3923 unless name is "Foo::", in which case Foo is a bearword
3924 (and a package name). */
3927 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3929 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3930 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
3931 "Bareword \"%s\" refers to nonexistent package",
3934 PL_tokenbuf[len] = '\0';
3941 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3944 /* if we saw a global override before, get the right name */
3947 sv = newSVpvn("CORE::GLOBAL::",14);
3948 sv_catpv(sv,PL_tokenbuf);
3951 sv = newSVpv(PL_tokenbuf,0);
3953 /* Presume this is going to be a bareword of some sort. */
3956 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3957 yylval.opval->op_private = OPpCONST_BARE;
3958 /* UTF-8 package name? */
3959 if (UTF && !IN_BYTES &&
3960 is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
3963 /* And if "Foo::", then that's what it certainly is. */
3968 /* See if it's the indirect object for a list operator. */
3970 if (PL_oldoldbufptr &&
3971 PL_oldoldbufptr < PL_bufptr &&
3972 (PL_oldoldbufptr == PL_last_lop
3973 || PL_oldoldbufptr == PL_last_uni) &&
3974 /* NO SKIPSPACE BEFORE HERE! */
3975 (PL_expect == XREF ||
3976 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3978 bool immediate_paren = *s == '(';
3980 /* (Now we can afford to cross potential line boundary.) */
3983 /* Two barewords in a row may indicate method call. */
3985 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3988 /* If not a declared subroutine, it's an indirect object. */
3989 /* (But it's an indir obj regardless for sort.) */
3991 if ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
3992 ((!gv || !GvCVu(gv)) &&
3993 (PL_last_lop_op != OP_MAPSTART &&
3994 PL_last_lop_op != OP_GREPSTART))))
3996 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
4001 PL_expect = XOPERATOR;
4004 /* Is this a word before a => operator? */
4005 if (*s == '=' && s[1] == '>' && !pkgname) {
4007 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
4008 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
4009 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
4013 /* If followed by a paren, it's certainly a subroutine. */
4016 if (gv && GvCVu(gv)) {
4017 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
4018 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
4023 PL_nextval[PL_nexttoke].opval = yylval.opval;
4024 PL_expect = XOPERATOR;
4030 /* If followed by var or block, call it a method (unless sub) */
4032 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
4033 PL_last_lop = PL_oldbufptr;
4034 PL_last_lop_op = OP_METHOD;
4038 /* If followed by a bareword, see if it looks like indir obj. */
4041 && (isIDFIRST_lazy_if(s,UTF) || *s == '$')
4042 && (tmp = intuit_method(s,gv)))
4045 /* Not a method, so call it a subroutine (if defined) */
4047 if (gv && GvCVu(gv)) {
4049 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
4050 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4051 "Ambiguous use of -%s resolved as -&%s()",
4052 PL_tokenbuf, PL_tokenbuf);
4053 /* Check for a constant sub */
4055 if ((sv = cv_const_sv(cv))) {
4057 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
4058 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
4059 yylval.opval->op_private = 0;
4063 /* Resolve to GV now. */
4064 op_free(yylval.opval);
4065 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
4066 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
4067 PL_last_lop = PL_oldbufptr;
4068 PL_last_lop_op = OP_ENTERSUB;
4069 /* Is there a prototype? */
4072 char *proto = SvPV((SV*)cv, len);
4075 if (strEQ(proto, "$"))
4077 if (*proto == '&' && *s == '{') {
4078 sv_setpv(PL_subname, PL_curstash ?
4079 "__ANON__" : "__ANON__::__ANON__");
4083 PL_nextval[PL_nexttoke].opval = yylval.opval;
4089 /* Call it a bare word */
4091 if (PL_hints & HINT_STRICT_SUBS)
4092 yylval.opval->op_private |= OPpCONST_STRICT;
4095 if (ckWARN(WARN_RESERVED)) {
4096 if (lastchar != '-') {
4097 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4098 if (!*d && !gv_stashpv(PL_tokenbuf,FALSE))
4099 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
4106 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
4107 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4108 "Operator or semicolon missing before %c%s",
4109 lastchar, PL_tokenbuf);
4110 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4111 "Ambiguous use of %c resolved as operator %c",
4112 lastchar, lastchar);
4118 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4119 newSVpv(CopFILE(PL_curcop),0));
4123 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4124 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4127 case KEY___PACKAGE__:
4128 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4130 ? newSVsv(PL_curstname)
4139 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4140 char *pname = "main";
4141 if (PL_tokenbuf[2] == 'D')
4142 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
4143 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4146 GvIOp(gv) = newIO();
4147 IoIFP(GvIOp(gv)) = PL_rsfp;
4148 #if defined(HAS_FCNTL) && defined(F_SETFD)
4150 int fd = PerlIO_fileno(PL_rsfp);
4151 fcntl(fd,F_SETFD,fd >= 3);
4154 /* Mark this internal pseudo-handle as clean */
4155 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4157 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4158 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4159 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4161 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4162 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4163 /* if the script was opened in binmode, we need to revert
4164 * it to text mode for compatibility; but only iff it has CRs
4165 * XXX this is a questionable hack at best. */
4166 if (PL_bufend-PL_bufptr > 2
4167 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4170 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4171 loc = PerlIO_tell(PL_rsfp);
4172 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4175 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
4177 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4178 #endif /* NETWARE */
4179 #ifdef PERLIO_IS_STDIO /* really? */
4180 # if defined(__BORLANDC__)
4181 /* XXX see note in do_binmode() */
4182 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
4186 PerlIO_seek(PL_rsfp, loc, 0);
4190 #ifdef PERLIO_LAYERS
4191 if (UTF && !IN_BYTES)
4192 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
4205 if (PL_expect == XSTATE) {
4212 if (*s == ':' && s[1] == ':') {
4215 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4216 if (!(tmp = keyword(PL_tokenbuf, len)))
4217 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4231 LOP(OP_ACCEPT,XTERM);
4237 LOP(OP_ATAN2,XTERM);
4243 LOP(OP_BINMODE,XTERM);
4246 LOP(OP_BLESS,XTERM);
4255 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4272 if (!PL_cryptseen) {
4273 PL_cryptseen = TRUE;
4277 LOP(OP_CRYPT,XTERM);
4280 LOP(OP_CHMOD,XTERM);
4283 LOP(OP_CHOWN,XTERM);
4286 LOP(OP_CONNECT,XTERM);
4302 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4306 PL_hints |= HINT_BLOCK_SCOPE;
4316 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4317 LOP(OP_DBMOPEN,XTERM);
4323 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4330 yylval.ival = CopLINE(PL_curcop);
4344 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4345 UNIBRACK(OP_ENTEREVAL);
4363 case KEY_endhostent:
4369 case KEY_endservent:
4372 case KEY_endprotoent:
4383 yylval.ival = CopLINE(PL_curcop);
4385 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4387 if ((PL_bufend - p) >= 3 &&
4388 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4390 else if ((PL_bufend - p) >= 4 &&
4391 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4394 if (isIDFIRST_lazy_if(p,UTF)) {
4395 p = scan_ident(p, PL_bufend,
4396 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4400 Perl_croak(aTHX_ "Missing $ on loop variable");
4405 LOP(OP_FORMLINE,XTERM);
4411 LOP(OP_FCNTL,XTERM);
4417 LOP(OP_FLOCK,XTERM);
4426 LOP(OP_GREPSTART, XREF);
4429 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4444 case KEY_getpriority:
4445 LOP(OP_GETPRIORITY,XTERM);
4447 case KEY_getprotobyname:
4450 case KEY_getprotobynumber:
4451 LOP(OP_GPBYNUMBER,XTERM);
4453 case KEY_getprotoent:
4465 case KEY_getpeername:
4466 UNI(OP_GETPEERNAME);
4468 case KEY_gethostbyname:
4471 case KEY_gethostbyaddr:
4472 LOP(OP_GHBYADDR,XTERM);
4474 case KEY_gethostent:
4477 case KEY_getnetbyname:
4480 case KEY_getnetbyaddr:
4481 LOP(OP_GNBYADDR,XTERM);
4486 case KEY_getservbyname:
4487 LOP(OP_GSBYNAME,XTERM);
4489 case KEY_getservbyport:
4490 LOP(OP_GSBYPORT,XTERM);
4492 case KEY_getservent:
4495 case KEY_getsockname:
4496 UNI(OP_GETSOCKNAME);
4498 case KEY_getsockopt:
4499 LOP(OP_GSOCKOPT,XTERM);
4521 yylval.ival = CopLINE(PL_curcop);
4525 LOP(OP_INDEX,XTERM);
4531 LOP(OP_IOCTL,XTERM);
4543 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4575 LOP(OP_LISTEN,XTERM);
4584 s = scan_pat(s,OP_MATCH);
4585 TERM(sublex_start());
4588 LOP(OP_MAPSTART, XREF);
4591 LOP(OP_MKDIR,XTERM);
4594 LOP(OP_MSGCTL,XTERM);
4597 LOP(OP_MSGGET,XTERM);
4600 LOP(OP_MSGRCV,XTERM);
4603 LOP(OP_MSGSND,XTERM);
4609 if (isIDFIRST_lazy_if(s,UTF)) {
4610 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4611 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4613 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4614 if (!PL_in_my_stash) {
4617 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4625 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4632 if (PL_expect != XSTATE)
4633 yyerror("\"no\" not allowed in expression");
4634 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4635 s = force_version(s, FALSE);
4640 if (*s == '(' || (s = skipspace(s), *s == '('))
4647 if (isIDFIRST_lazy_if(s,UTF)) {
4649 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4651 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4652 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
4653 "Precedence problem: open %.*s should be open(%.*s)",
4659 yylval.ival = OP_OR;
4669 LOP(OP_OPEN_DIR,XTERM);
4672 checkcomma(s,PL_tokenbuf,"filehandle");
4676 checkcomma(s,PL_tokenbuf,"filehandle");
4695 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4699 LOP(OP_PIPE_OP,XTERM);
4702 s = scan_str(s,FALSE,FALSE);
4704 missingterm((char*)0);
4705 yylval.ival = OP_CONST;
4706 TERM(sublex_start());
4712 s = scan_str(s,FALSE,FALSE);
4714 missingterm((char*)0);
4716 if (SvCUR(PL_lex_stuff)) {
4719 d = SvPV_force(PL_lex_stuff, len);
4722 for (; isSPACE(*d) && len; --len, ++d) ;
4725 if (!warned && ckWARN(WARN_QW)) {
4726 for (; !isSPACE(*d) && len; --len, ++d) {
4728 Perl_warner(aTHX_ packWARN(WARN_QW),
4729 "Possible attempt to separate words with commas");
4732 else if (*d == '#') {
4733 Perl_warner(aTHX_ packWARN(WARN_QW),
4734 "Possible attempt to put comments in qw() list");
4740 for (; !isSPACE(*d) && len; --len, ++d) ;
4742 sv = newSVpvn(b, d-b);
4743 if (DO_UTF8(PL_lex_stuff))
4745 words = append_elem(OP_LIST, words,
4746 newSVOP(OP_CONST, 0, tokeq(sv)));
4750 PL_nextval[PL_nexttoke].opval = words;
4755 SvREFCNT_dec(PL_lex_stuff);
4756 PL_lex_stuff = Nullsv;
4762 s = scan_str(s,FALSE,FALSE);
4764 missingterm((char*)0);
4765 yylval.ival = OP_STRINGIFY;
4766 if (SvIVX(PL_lex_stuff) == '\'')
4767 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4768 TERM(sublex_start());
4771 s = scan_pat(s,OP_QR);
4772 TERM(sublex_start());
4775 s = scan_str(s,FALSE,FALSE);
4777 missingterm((char*)0);
4778 yylval.ival = OP_BACKTICK;
4780 TERM(sublex_start());
4788 s = force_version(s, FALSE);
4790 else if (*s != 'v' || !isDIGIT(s[1])
4791 || (s = force_version(s, TRUE), *s == 'v'))
4793 *PL_tokenbuf = '\0';
4794 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4795 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4796 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4798 yyerror("<> should be quotes");
4806 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4810 LOP(OP_RENAME,XTERM);
4819 LOP(OP_RINDEX,XTERM);
4829 UNIDOR(OP_READLINE);
4842 LOP(OP_REVERSE,XTERM);
4845 UNIDOR(OP_READLINK);
4853 TERM(sublex_start());
4855 TOKEN(1); /* force error */
4864 LOP(OP_SELECT,XTERM);
4870 LOP(OP_SEMCTL,XTERM);
4873 LOP(OP_SEMGET,XTERM);
4876 LOP(OP_SEMOP,XTERM);
4882 LOP(OP_SETPGRP,XTERM);
4884 case KEY_setpriority:
4885 LOP(OP_SETPRIORITY,XTERM);
4887 case KEY_sethostent:
4893 case KEY_setservent:
4896 case KEY_setprotoent:
4906 LOP(OP_SEEKDIR,XTERM);
4908 case KEY_setsockopt:
4909 LOP(OP_SSOCKOPT,XTERM);
4915 LOP(OP_SHMCTL,XTERM);
4918 LOP(OP_SHMGET,XTERM);
4921 LOP(OP_SHMREAD,XTERM);
4924 LOP(OP_SHMWRITE,XTERM);
4927 LOP(OP_SHUTDOWN,XTERM);
4936 LOP(OP_SOCKET,XTERM);
4938 case KEY_socketpair:
4939 LOP(OP_SOCKPAIR,XTERM);
4942 checkcomma(s,PL_tokenbuf,"subroutine name");
4944 if (*s == ';' || *s == ')') /* probably a close */
4945 Perl_croak(aTHX_ "sort is now a reserved word");
4947 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4951 LOP(OP_SPLIT,XTERM);
4954 LOP(OP_SPRINTF,XTERM);
4957 LOP(OP_SPLICE,XTERM);
4972 LOP(OP_SUBSTR,XTERM);
4978 char tmpbuf[sizeof PL_tokenbuf];
4979 SSize_t tboffset = 0;
4980 expectation attrful;
4981 bool have_name, have_proto, bad_proto;
4986 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4987 (*s == ':' && s[1] == ':'))
4990 attrful = XATTRBLOCK;
4991 /* remember buffer pos'n for later force_word */
4992 tboffset = s - PL_oldbufptr;
4993 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4994 if (strchr(tmpbuf, ':'))
4995 sv_setpv(PL_subname, tmpbuf);
4997 sv_setsv(PL_subname,PL_curstname);
4998 sv_catpvn(PL_subname,"::",2);
4999 sv_catpvn(PL_subname,tmpbuf,len);
5006 Perl_croak(aTHX_ "Missing name in \"my sub\"");
5007 PL_expect = XTERMBLOCK;
5008 attrful = XATTRTERM;
5009 sv_setpv(PL_subname,"?");
5013 if (key == KEY_format) {
5015 PL_lex_formbrack = PL_lex_brackets + 1;
5017 (void) force_word(PL_oldbufptr + tboffset, WORD,
5022 /* Look for a prototype */
5026 s = scan_str(s,FALSE,FALSE);
5028 Perl_croak(aTHX_ "Prototype not terminated");
5029 /* strip spaces and check for bad characters */
5030 d = SvPVX(PL_lex_stuff);
5033 for (p = d; *p; ++p) {
5036 if (!strchr("$@%*;[]&\\", *p))
5041 if (bad_proto && ckWARN(WARN_SYNTAX))
5042 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5043 "Illegal character in prototype for %s : %s",
5044 SvPVX(PL_subname), d);
5045 SvCUR(PL_lex_stuff) = tmp;
5053 if (*s == ':' && s[1] != ':')
5054 PL_expect = attrful;
5057 PL_nextval[PL_nexttoke].opval =
5058 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
5059 PL_lex_stuff = Nullsv;
5063 sv_setpv(PL_subname,
5064 PL_curstash ? "__ANON__" : "__ANON__::__ANON__");
5067 (void) force_word(PL_oldbufptr + tboffset, WORD,
5076 LOP(OP_SYSTEM,XREF);
5079 LOP(OP_SYMLINK,XTERM);
5082 LOP(OP_SYSCALL,XTERM);
5085 LOP(OP_SYSOPEN,XTERM);
5088 LOP(OP_SYSSEEK,XTERM);
5091 LOP(OP_SYSREAD,XTERM);
5094 LOP(OP_SYSWRITE,XTERM);
5098 TERM(sublex_start());
5119 LOP(OP_TRUNCATE,XTERM);
5131 yylval.ival = CopLINE(PL_curcop);
5135 yylval.ival = CopLINE(PL_curcop);
5139 LOP(OP_UNLINK,XTERM);
5145 LOP(OP_UNPACK,XTERM);
5148 LOP(OP_UTIME,XTERM);
5154 LOP(OP_UNSHIFT,XTERM);
5157 if (PL_expect != XSTATE)
5158 yyerror("\"use\" not allowed in expression");
5160 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
5161 s = force_version(s, TRUE);
5162 if (*s == ';' || (s = skipspace(s), *s == ';')) {
5163 PL_nextval[PL_nexttoke].opval = Nullop;
5166 else if (*s == 'v') {
5167 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5168 s = force_version(s, FALSE);
5172 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5173 s = force_version(s, FALSE);
5185 yylval.ival = CopLINE(PL_curcop);
5189 PL_hints |= HINT_BLOCK_SCOPE;
5196 LOP(OP_WAITPID,XTERM);
5205 ctl_l[0] = toCTRL('L');
5207 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5210 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5215 if (PL_expect == XOPERATOR)
5221 yylval.ival = OP_XOR;
5226 TERM(sublex_start());
5231 #pragma segment Main
5235 S_pending_ident(pTHX)
5239 /* pit holds the identifier we read and pending_ident is reset */
5240 char pit = PL_pending_ident;
5241 PL_pending_ident = 0;
5243 DEBUG_T({ PerlIO_printf(Perl_debug_log,
5244 "### Tokener saw identifier '%s'\n", PL_tokenbuf); });
5246 /* if we're in a my(), we can't allow dynamics here.
5247 $foo'bar has already been turned into $foo::bar, so
5248 just check for colons.
5250 if it's a legal name, the OP is a PADANY.
5253 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
5254 if (strchr(PL_tokenbuf,':'))
5255 yyerror(Perl_form(aTHX_ "No package name allowed for "
5256 "variable %s in \"our\"",
5258 tmp = allocmy(PL_tokenbuf);
5261 if (strchr(PL_tokenbuf,':'))
5262 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
5264 yylval.opval = newOP(OP_PADANY, 0);
5265 yylval.opval->op_targ = allocmy(PL_tokenbuf);
5271 build the ops for accesses to a my() variable.
5273 Deny my($a) or my($b) in a sort block, *if* $a or $b is
5274 then used in a comparison. This catches most, but not
5275 all cases. For instance, it catches
5276 sort { my($a); $a <=> $b }
5278 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
5279 (although why you'd do that is anyone's guess).
5282 if (!strchr(PL_tokenbuf,':')) {
5283 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
5284 /* might be an "our" variable" */
5285 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
5286 /* build ops for a bareword */
5287 SV *sym = newSVpv(HvNAME(PAD_COMPNAME_OURSTASH(tmp)), 0);
5288 sv_catpvn(sym, "::", 2);
5289 sv_catpv(sym, PL_tokenbuf+1);
5290 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
5291 yylval.opval->op_private = OPpCONST_ENTERED;
5292 gv_fetchpv(SvPVX(sym),
5294 ? (GV_ADDMULTI | GV_ADDINEVAL)
5297 ((PL_tokenbuf[0] == '$') ? SVt_PV
5298 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5303 /* if it's a sort block and they're naming $a or $b */
5304 if (PL_last_lop_op == OP_SORT &&
5305 PL_tokenbuf[0] == '$' &&
5306 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
5309 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
5310 d < PL_bufend && *d != '\n';
5313 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
5314 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
5320 yylval.opval = newOP(OP_PADANY, 0);
5321 yylval.opval->op_targ = tmp;
5327 Whine if they've said @foo in a doublequoted string,
5328 and @foo isn't a variable we can find in the symbol
5331 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
5332 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
5333 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
5334 && ckWARN(WARN_AMBIGUOUS))
5336 /* Downgraded from fatal to warning 20000522 mjd */
5337 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5338 "Possible unintended interpolation of %s in string",
5343 /* build ops for a bareword */
5344 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
5345 yylval.opval->op_private = OPpCONST_ENTERED;
5346 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
5347 ((PL_tokenbuf[0] == '$') ? SVt_PV
5348 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5354 Perl_keyword(pTHX_ register char *d, I32 len)
5359 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5360 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5361 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5362 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5363 if (strEQ(d,"__END__")) return KEY___END__;
5367 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5372 if (strEQ(d,"and")) return -KEY_and;
5373 if (strEQ(d,"abs")) return -KEY_abs;
5376 if (strEQ(d,"alarm")) return -KEY_alarm;
5377 if (strEQ(d,"atan2")) return -KEY_atan2;
5380 if (strEQ(d,"accept")) return -KEY_accept;
5385 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5388 if (strEQ(d,"bless")) return -KEY_bless;
5389 if (strEQ(d,"bind")) return -KEY_bind;
5390 if (strEQ(d,"binmode")) return -KEY_binmode;
5393 if (strEQ(d,"CORE")) return -KEY_CORE;
5394 if (strEQ(d,"CHECK")) return KEY_CHECK;
5399 if (strEQ(d,"cmp")) return -KEY_cmp;
5400 if (strEQ(d,"chr")) return -KEY_chr;
5401 if (strEQ(d,"cos")) return -KEY_cos;
5404 if (strEQ(d,"chop")) return -KEY_chop;
5407 if (strEQ(d,"close")) return -KEY_close;
5408 if (strEQ(d,"chdir")) return -KEY_chdir;
5409 if (strEQ(d,"chomp")) return -KEY_chomp;
5410 if (strEQ(d,"chmod")) return -KEY_chmod;
5411 if (strEQ(d,"chown")) return -KEY_chown;
5412 if (strEQ(d,"crypt")) return -KEY_crypt;
5415 if (strEQ(d,"chroot")) return -KEY_chroot;
5416 if (strEQ(d,"caller")) return -KEY_caller;
5419 if (strEQ(d,"connect")) return -KEY_connect;
5422 if (strEQ(d,"closedir")) return -KEY_closedir;
5423 if (strEQ(d,"continue")) return -KEY_continue;
5428 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5433 if (strEQ(d,"do")) return KEY_do;
5436 if (strEQ(d,"die")) return -KEY_die;
5439 if (strEQ(d,"dump")) return -KEY_dump;
5442 if (strEQ(d,"delete")) return KEY_delete;
5445 if (strEQ(d,"defined")) return KEY_defined;
5446 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5449 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5454 if (strEQ(d,"END")) return KEY_END;
5459 if (strEQ(d,"eq")) return -KEY_eq;
5462 if (strEQ(d,"eof")) return -KEY_eof;
5463 if (strEQ(d,"err")) return -KEY_err;
5464 if (strEQ(d,"exp")) return -KEY_exp;
5467 if (strEQ(d,"else")) return KEY_else;
5468 if (strEQ(d,"exit")) return -KEY_exit;
5469 if (strEQ(d,"eval")) return KEY_eval;
5470 if (strEQ(d,"exec")) return -KEY_exec;
5471 if (strEQ(d,"each")) return -KEY_each;
5474 if (strEQ(d,"elsif")) return KEY_elsif;
5477 if (strEQ(d,"exists")) return KEY_exists;
5478 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5481 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5482 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5485 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5488 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5489 if (strEQ(d,"endservent")) return -KEY_endservent;
5492 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5499 if (strEQ(d,"for")) return KEY_for;
5502 if (strEQ(d,"fork")) return -KEY_fork;
5505 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5506 if (strEQ(d,"flock")) return -KEY_flock;
5509 if (strEQ(d,"format")) return KEY_format;
5510 if (strEQ(d,"fileno")) return -KEY_fileno;
5513 if (strEQ(d,"foreach")) return KEY_foreach;
5516 if (strEQ(d,"formline")) return -KEY_formline;
5521 if (strnEQ(d,"get",3)) {
5526 if (strEQ(d,"ppid")) return -KEY_getppid;
5527 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5530 if (strEQ(d,"pwent")) return -KEY_getpwent;
5531 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5532 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5535 if (strEQ(d,"peername")) return -KEY_getpeername;
5536 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5537 if (strEQ(d,"priority")) return -KEY_getpriority;
5540 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5543 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5547 else if (*d == 'h') {
5548 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5549 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5550 if (strEQ(d,"hostent")) return -KEY_gethostent;
5552 else if (*d == 'n') {
5553 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5554 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5555 if (strEQ(d,"netent")) return -KEY_getnetent;
5557 else if (*d == 's') {
5558 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5559 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5560 if (strEQ(d,"servent")) return -KEY_getservent;
5561 if (strEQ(d,"sockname")) return -KEY_getsockname;
5562 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5564 else if (*d == 'g') {
5565 if (strEQ(d,"grent")) return -KEY_getgrent;
5566 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5567 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5569 else if (*d == 'l') {
5570 if (strEQ(d,"login")) return -KEY_getlogin;
5572 else if (strEQ(d,"c")) return -KEY_getc;
5577 if (strEQ(d,"gt")) return -KEY_gt;
5578 if (strEQ(d,"ge")) return -KEY_ge;
5581 if (strEQ(d,"grep")) return KEY_grep;
5582 if (strEQ(d,"goto")) return KEY_goto;
5583 if (strEQ(d,"glob")) return KEY_glob;
5586 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5591 if (strEQ(d,"hex")) return -KEY_hex;
5594 if (strEQ(d,"INIT")) return KEY_INIT;
5599 if (strEQ(d,"if")) return KEY_if;
5602 if (strEQ(d,"int")) return -KEY_int;
5605 if (strEQ(d,"index")) return -KEY_index;
5606 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5611 if (strEQ(d,"join")) return -KEY_join;
5615 if (strEQ(d,"keys")) return -KEY_keys;
5616 if (strEQ(d,"kill")) return -KEY_kill;
5622 if (strEQ(d,"lt")) return -KEY_lt;
5623 if (strEQ(d,"le")) return -KEY_le;
5624 if (strEQ(d,"lc")) return -KEY_lc;
5627 if (strEQ(d,"log")) return -KEY_log;
5630 if (strEQ(d,"last")) return KEY_last;
5631 if (strEQ(d,"link")) return -KEY_link;
5632 if (strEQ(d,"lock")) return -KEY_lock;
5635 if (strEQ(d,"local")) return KEY_local;
5636 if (strEQ(d,"lstat")) return -KEY_lstat;
5639 if (strEQ(d,"length")) return -KEY_length;
5640 if (strEQ(d,"listen")) return -KEY_listen;
5643 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5646 if (strEQ(d,"localtime")) return -KEY_localtime;
5652 case 1: return KEY_m;
5654 if (strEQ(d,"my")) return KEY_my;
5657 if (strEQ(d,"map")) return KEY_map;
5660 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5663 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5664 if (strEQ(d,"msgget")) return -KEY_msgget;
5665 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5666 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5671 if (strEQ(d,"next")) return KEY_next;
5672 if (strEQ(d,"ne")) return -KEY_ne;
5673 if (strEQ(d,"not")) return -KEY_not;
5674 if (strEQ(d,"no")) return KEY_no;
5679 if (strEQ(d,"or")) return -KEY_or;
5682 if (strEQ(d,"ord")) return -KEY_ord;
5683 if (strEQ(d,"oct")) return -KEY_oct;
5684 if (strEQ(d,"our")) return KEY_our;
5687 if (strEQ(d,"open")) return -KEY_open;
5690 if (strEQ(d,"opendir")) return -KEY_opendir;
5697 if (strEQ(d,"pop")) return -KEY_pop;
5698 if (strEQ(d,"pos")) return KEY_pos;
5701 if (strEQ(d,"push")) return -KEY_push;
5702 if (strEQ(d,"pack")) return -KEY_pack;
5703 if (strEQ(d,"pipe")) return -KEY_pipe;
5706 if (strEQ(d,"print")) return KEY_print;
5709 if (strEQ(d,"printf")) return KEY_printf;
5712 if (strEQ(d,"package")) return KEY_package;
5715 if (strEQ(d,"prototype")) return KEY_prototype;
5720 if (strEQ(d,"q")) return KEY_q;
5721 if (strEQ(d,"qr")) return KEY_qr;
5722 if (strEQ(d,"qq")) return KEY_qq;
5723 if (strEQ(d,"qw")) return KEY_qw;
5724 if (strEQ(d,"qx")) return KEY_qx;
5726 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5731 if (strEQ(d,"ref")) return -KEY_ref;
5734 if (strEQ(d,"read")) return -KEY_read;
5735 if (strEQ(d,"rand")) return -KEY_rand;
5736 if (strEQ(d,"recv")) return -KEY_recv;
5737 if (strEQ(d,"redo")) return KEY_redo;
5740 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5741 if (strEQ(d,"reset")) return -KEY_reset;
5744 if (strEQ(d,"return")) return KEY_return;
5745 if (strEQ(d,"rename")) return -KEY_rename;
5746 if (strEQ(d,"rindex")) return -KEY_rindex;
5749 if (strEQ(d,"require")) return KEY_require;
5750 if (strEQ(d,"reverse")) return -KEY_reverse;
5751 if (strEQ(d,"readdir")) return -KEY_readdir;
5754 if (strEQ(d,"readlink")) return -KEY_readlink;
5755 if (strEQ(d,"readline")) return -KEY_readline;
5756 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5759 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5765 case 0: return KEY_s;
5767 if (strEQ(d,"scalar")) return KEY_scalar;
5772 if (strEQ(d,"seek")) return -KEY_seek;
5773 if (strEQ(d,"send")) return -KEY_send;
5776 if (strEQ(d,"semop")) return -KEY_semop;
5779 if (strEQ(d,"select")) return -KEY_select;
5780 if (strEQ(d,"semctl")) return -KEY_semctl;
5781 if (strEQ(d,"semget")) return -KEY_semget;
5784 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5785 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5788 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5789 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5792 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5795 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5796 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5797 if (strEQ(d,"setservent")) return -KEY_setservent;
5800 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5801 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5808 if (strEQ(d,"shift")) return -KEY_shift;
5811 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5812 if (strEQ(d,"shmget")) return -KEY_shmget;
5815 if (strEQ(d,"shmread")) return -KEY_shmread;
5818 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5819 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5824 if (strEQ(d,"sin")) return -KEY_sin;
5827 if (strEQ(d,"sleep")) return -KEY_sleep;
5830 if (strEQ(d,"sort")) return KEY_sort;
5831 if (strEQ(d,"socket")) return -KEY_socket;
5832 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5835 if (strEQ(d,"split")) return KEY_split;
5836 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5837 if (strEQ(d,"splice")) return -KEY_splice;
5840 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5843 if (strEQ(d,"srand")) return -KEY_srand;
5846 if (strEQ(d,"stat")) return -KEY_stat;
5847 if (strEQ(d,"study")) return KEY_study;
5850 if (strEQ(d,"substr")) return -KEY_substr;
5851 if (strEQ(d,"sub")) return KEY_sub;
5856 if (strEQ(d,"system")) return -KEY_system;
5859 if (strEQ(d,"symlink")) return -KEY_symlink;
5860 if (strEQ(d,"syscall")) return -KEY_syscall;
5861 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5862 if (strEQ(d,"sysread")) return -KEY_sysread;
5863 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5866 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5875 if (strEQ(d,"tr")) return KEY_tr;
5878 if (strEQ(d,"tie")) return KEY_tie;
5881 if (strEQ(d,"tell")) return -KEY_tell;
5882 if (strEQ(d,"tied")) return KEY_tied;
5883 if (strEQ(d,"time")) return -KEY_time;
5886 if (strEQ(d,"times")) return -KEY_times;
5889 if (strEQ(d,"telldir")) return -KEY_telldir;
5892 if (strEQ(d,"truncate")) return -KEY_truncate;
5899 if (strEQ(d,"uc")) return -KEY_uc;
5902 if (strEQ(d,"use")) return KEY_use;
5905 if (strEQ(d,"undef")) return KEY_undef;
5906 if (strEQ(d,"until")) return KEY_until;
5907 if (strEQ(d,"untie")) return KEY_untie;
5908 if (strEQ(d,"utime")) return -KEY_utime;
5909 if (strEQ(d,"umask")) return -KEY_umask;
5912 if (strEQ(d,"unless")) return KEY_unless;
5913 if (strEQ(d,"unpack")) return -KEY_unpack;
5914 if (strEQ(d,"unlink")) return -KEY_unlink;
5917 if (strEQ(d,"unshift")) return -KEY_unshift;
5918 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5923 if (strEQ(d,"values")) return -KEY_values;
5924 if (strEQ(d,"vec")) return -KEY_vec;
5929 if (strEQ(d,"warn")) return -KEY_warn;
5930 if (strEQ(d,"wait")) return -KEY_wait;
5933 if (strEQ(d,"while")) return KEY_while;
5934 if (strEQ(d,"write")) return -KEY_write;
5937 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5940 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5945 if (len == 1) return -KEY_x;
5946 if (strEQ(d,"xor")) return -KEY_xor;
5949 if (len == 1) return KEY_y;
5958 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5962 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5963 if (ckWARN(WARN_SYNTAX)) {
5965 for (w = s+2; *w && level; w++) {
5972 for (; *w && isSPACE(*w); w++) ;
5973 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5974 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5975 "%s (...) interpreted as function",name);
5978 while (s < PL_bufend && isSPACE(*s))
5982 while (s < PL_bufend && isSPACE(*s))
5984 if (isIDFIRST_lazy_if(s,UTF)) {
5986 while (isALNUM_lazy_if(s,UTF))
5988 while (s < PL_bufend && isSPACE(*s))
5993 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5997 Perl_croak(aTHX_ "No comma allowed after %s", what);
6002 /* Either returns sv, or mortalizes sv and returns a new SV*.
6003 Best used as sv=new_constant(..., sv, ...).
6004 If s, pv are NULL, calls subroutine with one argument,
6005 and type is used with error messages only. */
6008 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
6012 HV *table = GvHV(PL_hintgv); /* ^H */
6016 const char *why1, *why2, *why3;
6018 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
6021 why2 = strEQ(key,"charnames")
6022 ? "(possibly a missing \"use charnames ...\")"
6024 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
6025 (type ? type: "undef"), why2);
6027 /* This is convoluted and evil ("goto considered harmful")
6028 * but I do not understand the intricacies of all the different
6029 * failure modes of %^H in here. The goal here is to make
6030 * the most probable error message user-friendly. --jhi */
6035 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
6036 (type ? type: "undef"), why1, why2, why3);
6038 yyerror(SvPVX(msg));
6042 cvp = hv_fetch(table, key, strlen(key), FALSE);
6043 if (!cvp || !SvOK(*cvp)) {
6046 why3 = "} is not defined";
6049 sv_2mortal(sv); /* Parent created it permanently */
6052 pv = sv_2mortal(newSVpvn(s, len));
6054 typesv = sv_2mortal(newSVpv(type, 0));
6056 typesv = &PL_sv_undef;
6058 PUSHSTACKi(PERLSI_OVERLOAD);
6070 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
6074 /* Check the eval first */
6075 if (!PL_in_eval && SvTRUE(ERRSV)) {
6077 sv_catpv(ERRSV, "Propagated");
6078 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
6080 res = SvREFCNT_inc(sv);
6084 (void)SvREFCNT_inc(res);
6093 why1 = "Call to &{$^H{";
6095 why3 = "}} did not return a defined value";
6104 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
6106 register char *d = dest;
6107 register char *e = d + destlen - 3; /* two-character token, ending NUL */
6110 Perl_croak(aTHX_ ident_too_long);
6111 if (isALNUM(*s)) /* UTF handled below */
6113 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
6118 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
6122 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6123 char *t = s + UTF8SKIP(s);
6124 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6126 if (d + (t - s) > e)
6127 Perl_croak(aTHX_ ident_too_long);
6128 Copy(s, d, t - s, char);
6141 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
6151 e = d + destlen - 3; /* two-character token, ending NUL */
6153 while (isDIGIT(*s)) {
6155 Perl_croak(aTHX_ ident_too_long);
6162 Perl_croak(aTHX_ ident_too_long);
6163 if (isALNUM(*s)) /* UTF handled below */
6165 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
6170 else if (*s == ':' && s[1] == ':') {
6174 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6175 char *t = s + UTF8SKIP(s);
6176 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6178 if (d + (t - s) > e)
6179 Perl_croak(aTHX_ ident_too_long);
6180 Copy(s, d, t - s, char);
6191 if (PL_lex_state != LEX_NORMAL)
6192 PL_lex_state = LEX_INTERPENDMAYBE;
6195 if (*s == '$' && s[1] &&
6196 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
6209 if (*d == '^' && *s && isCONTROLVAR(*s)) {
6214 if (isSPACE(s[-1])) {
6217 if (!SPACE_OR_TAB(ch)) {
6223 if (isIDFIRST_lazy_if(d,UTF)) {
6227 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
6229 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
6232 Copy(s, d, e - s, char);
6237 while ((isALNUM(*s) || *s == ':') && d < e)
6240 Perl_croak(aTHX_ ident_too_long);
6243 while (s < send && SPACE_OR_TAB(*s)) s++;
6244 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
6245 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
6246 const char *brack = *s == '[' ? "[...]" : "{...}";
6247 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
6248 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
6249 funny, dest, brack, funny, dest, brack);
6252 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
6256 /* Handle extended ${^Foo} variables
6257 * 1999-02-27 mjd-perl-patch@plover.com */
6258 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
6262 while (isALNUM(*s) && d < e) {
6266 Perl_croak(aTHX_ ident_too_long);
6271 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
6272 PL_lex_state = LEX_INTERPEND;
6275 if (PL_lex_state == LEX_NORMAL) {
6276 if (ckWARN(WARN_AMBIGUOUS) &&
6277 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
6279 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
6280 "Ambiguous use of %c{%s} resolved to %c%s",
6281 funny, dest, funny, dest);
6286 s = bracket; /* let the parser handle it */
6290 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
6291 PL_lex_state = LEX_INTERPEND;
6296 Perl_pmflag(pTHX_ U32* pmfl, int ch)
6301 *pmfl |= PMf_GLOBAL;
6303 *pmfl |= PMf_CONTINUE;
6307 *pmfl |= PMf_MULTILINE;
6309 *pmfl |= PMf_SINGLELINE;
6311 *pmfl |= PMf_EXTENDED;
6315 S_scan_pat(pTHX_ char *start, I32 type)
6320 s = scan_str(start,FALSE,FALSE);
6322 Perl_croak(aTHX_ "Search pattern not terminated");
6324 pm = (PMOP*)newPMOP(type, 0);
6325 if (PL_multi_open == '?')
6326 pm->op_pmflags |= PMf_ONCE;
6328 while (*s && strchr("iomsx", *s))
6329 pmflag(&pm->op_pmflags,*s++);
6332 while (*s && strchr("iogcmsx", *s))
6333 pmflag(&pm->op_pmflags,*s++);
6335 /* issue a warning if /c is specified,but /g is not */
6336 if (ckWARN(WARN_REGEXP) &&
6337 (pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL))
6339 Perl_warner(aTHX_ packWARN(WARN_REGEXP), c_without_g);
6342 pm->op_pmpermflags = pm->op_pmflags;
6344 PL_lex_op = (OP*)pm;
6345 yylval.ival = OP_MATCH;
6350 S_scan_subst(pTHX_ char *start)
6357 yylval.ival = OP_NULL;
6359 s = scan_str(start,FALSE,FALSE);
6362 Perl_croak(aTHX_ "Substitution pattern not terminated");
6364 if (s[-1] == PL_multi_open)
6367 first_start = PL_multi_start;
6368 s = scan_str(s,FALSE,FALSE);
6371 SvREFCNT_dec(PL_lex_stuff);
6372 PL_lex_stuff = Nullsv;
6374 Perl_croak(aTHX_ "Substitution replacement not terminated");
6376 PL_multi_start = first_start; /* so whole substitution is taken together */
6378 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6384 else if (strchr("iogcmsx", *s))
6385 pmflag(&pm->op_pmflags,*s++);
6390 /* /c is not meaningful with s/// */
6391 if (ckWARN(WARN_REGEXP) && (pm->op_pmflags & PMf_CONTINUE))
6393 Perl_warner(aTHX_ packWARN(WARN_REGEXP), c_in_subst);
6398 PL_sublex_info.super_bufptr = s;
6399 PL_sublex_info.super_bufend = PL_bufend;
6401 pm->op_pmflags |= PMf_EVAL;
6402 repl = newSVpvn("",0);
6404 sv_catpv(repl, es ? "eval " : "do ");
6405 sv_catpvn(repl, "{ ", 2);
6406 sv_catsv(repl, PL_lex_repl);
6407 sv_catpvn(repl, " };", 2);
6409 SvREFCNT_dec(PL_lex_repl);
6413 pm->op_pmpermflags = pm->op_pmflags;
6414 PL_lex_op = (OP*)pm;
6415 yylval.ival = OP_SUBST;
6420 S_scan_trans(pTHX_ char *start)
6429 yylval.ival = OP_NULL;
6431 s = scan_str(start,FALSE,FALSE);
6433 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6434 if (s[-1] == PL_multi_open)
6437 s = scan_str(s,FALSE,FALSE);
6440 SvREFCNT_dec(PL_lex_stuff);
6441 PL_lex_stuff = Nullsv;
6443 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6446 complement = del = squash = 0;
6447 while (strchr("cds", *s)) {
6449 complement = OPpTRANS_COMPLEMENT;
6451 del = OPpTRANS_DELETE;
6453 squash = OPpTRANS_SQUASH;
6457 New(803, tbl, complement&&!del?258:256, short);
6458 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6459 o->op_private = del|squash|complement|
6460 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
6461 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
6464 yylval.ival = OP_TRANS;
6469 S_scan_heredoc(pTHX_ register char *s)
6472 I32 op_type = OP_SCALAR;
6479 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6483 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6486 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6487 if (*peek && strchr("`'\"",*peek)) {
6490 s = delimcpy(d, e, s, PL_bufend, term, &len);
6500 if (!isALNUM_lazy_if(s,UTF))
6501 deprecate_old("bare << to mean <<\"\"");
6502 for (; isALNUM_lazy_if(s,UTF); s++) {
6507 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6508 Perl_croak(aTHX_ "Delimiter for here document is too long");
6511 len = d - PL_tokenbuf;
6512 #ifndef PERL_STRICT_CR
6513 d = strchr(s, '\r');
6517 while (s < PL_bufend) {
6523 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6532 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6537 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6538 herewas = newSVpvn(s,PL_bufend-s);
6540 s--, herewas = newSVpvn(s,d-s);
6541 s += SvCUR(herewas);
6543 tmpstr = NEWSV(87,79);
6544 sv_upgrade(tmpstr, SVt_PVIV);
6549 else if (term == '`') {
6550 op_type = OP_BACKTICK;
6551 SvIVX(tmpstr) = '\\';
6555 PL_multi_start = CopLINE(PL_curcop);
6556 PL_multi_open = PL_multi_close = '<';
6557 term = *PL_tokenbuf;
6558 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6559 char *bufptr = PL_sublex_info.super_bufptr;
6560 char *bufend = PL_sublex_info.super_bufend;
6561 char *olds = s - SvCUR(herewas);
6562 s = strchr(bufptr, '\n');
6566 while (s < bufend &&
6567 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6569 CopLINE_inc(PL_curcop);
6572 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6573 missingterm(PL_tokenbuf);
6575 sv_setpvn(herewas,bufptr,d-bufptr+1);
6576 sv_setpvn(tmpstr,d+1,s-d);
6578 sv_catpvn(herewas,s,bufend-s);
6579 (void)strcpy(bufptr,SvPVX(herewas));
6586 while (s < PL_bufend &&
6587 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6589 CopLINE_inc(PL_curcop);
6591 if (s >= PL_bufend) {
6592 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6593 missingterm(PL_tokenbuf);
6595 sv_setpvn(tmpstr,d+1,s-d);
6597 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6599 sv_catpvn(herewas,s,PL_bufend-s);
6600 sv_setsv(PL_linestr,herewas);
6601 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6602 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6603 PL_last_lop = PL_last_uni = Nullch;
6606 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6607 while (s >= PL_bufend) { /* multiple line string? */
6609 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6610 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6611 missingterm(PL_tokenbuf);
6613 CopLINE_inc(PL_curcop);
6614 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6615 PL_last_lop = PL_last_uni = Nullch;
6616 #ifndef PERL_STRICT_CR
6617 if (PL_bufend - PL_linestart >= 2) {
6618 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6619 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6621 PL_bufend[-2] = '\n';
6623 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6625 else if (PL_bufend[-1] == '\r')
6626 PL_bufend[-1] = '\n';
6628 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6629 PL_bufend[-1] = '\n';
6631 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6632 SV *sv = NEWSV(88,0);
6634 sv_upgrade(sv, SVt_PVMG);
6635 sv_setsv(sv,PL_linestr);
6638 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6640 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6643 sv_catsv(PL_linestr,herewas);
6644 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6648 sv_catsv(tmpstr,PL_linestr);
6653 PL_multi_end = CopLINE(PL_curcop);
6654 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6655 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6656 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6658 SvREFCNT_dec(herewas);
6659 if (UTF && !IN_BYTES && is_utf8_string((U8*)SvPVX(tmpstr), SvCUR(tmpstr)))
6661 PL_lex_stuff = tmpstr;
6662 yylval.ival = op_type;
6667 takes: current position in input buffer
6668 returns: new position in input buffer
6669 side-effects: yylval and lex_op are set.
6674 <FH> read from filehandle
6675 <pkg::FH> read from package qualified filehandle
6676 <pkg'FH> read from package qualified filehandle
6677 <$fh> read from filehandle in $fh
6683 S_scan_inputsymbol(pTHX_ char *start)
6685 register char *s = start; /* current position in buffer */
6691 d = PL_tokenbuf; /* start of temp holding space */
6692 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6693 end = strchr(s, '\n');
6696 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6698 /* die if we didn't have space for the contents of the <>,
6699 or if it didn't end, or if we see a newline
6702 if (len >= sizeof PL_tokenbuf)
6703 Perl_croak(aTHX_ "Excessively long <> operator");
6705 Perl_croak(aTHX_ "Unterminated <> operator");
6710 Remember, only scalar variables are interpreted as filehandles by
6711 this code. Anything more complex (e.g., <$fh{$num}>) will be
6712 treated as a glob() call.
6713 This code makes use of the fact that except for the $ at the front,
6714 a scalar variable and a filehandle look the same.
6716 if (*d == '$' && d[1]) d++;
6718 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6719 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6722 /* If we've tried to read what we allow filehandles to look like, and
6723 there's still text left, then it must be a glob() and not a getline.
6724 Use scan_str to pull out the stuff between the <> and treat it
6725 as nothing more than a string.
6728 if (d - PL_tokenbuf != len) {
6729 yylval.ival = OP_GLOB;
6731 s = scan_str(start,FALSE,FALSE);
6733 Perl_croak(aTHX_ "Glob not terminated");
6737 bool readline_overriden = FALSE;
6738 GV *gv_readline = Nullgv;
6740 /* we're in a filehandle read situation */
6743 /* turn <> into <ARGV> */
6745 (void)strcpy(d,"ARGV");
6747 /* Check whether readline() is overriden */
6748 if (((gv_readline = gv_fetchpv("readline", FALSE, SVt_PVCV))
6749 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
6751 ((gvp = (GV**)hv_fetch(PL_globalstash, "readline", 8, FALSE))
6752 && (gv_readline = *gvp) != (GV*)&PL_sv_undef
6753 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
6754 readline_overriden = TRUE;
6756 /* if <$fh>, create the ops to turn the variable into a
6762 /* try to find it in the pad for this block, otherwise find
6763 add symbol table ops
6765 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6766 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
6767 SV *sym = sv_2mortal(
6768 newSVpv(HvNAME(PAD_COMPNAME_OURSTASH(tmp)),0));
6769 sv_catpvn(sym, "::", 2);
6775 OP *o = newOP(OP_PADSV, 0);
6777 PL_lex_op = readline_overriden
6778 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6779 append_elem(OP_LIST, o,
6780 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
6781 : (OP*)newUNOP(OP_READLINE, 0, o);
6790 ? (GV_ADDMULTI | GV_ADDINEVAL)
6793 PL_lex_op = readline_overriden
6794 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6795 append_elem(OP_LIST,
6796 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
6797 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
6798 : (OP*)newUNOP(OP_READLINE, 0,
6799 newUNOP(OP_RV2SV, 0,
6800 newGVOP(OP_GV, 0, gv)));
6802 if (!readline_overriden)
6803 PL_lex_op->op_flags |= OPf_SPECIAL;
6804 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6805 yylval.ival = OP_NULL;
6808 /* If it's none of the above, it must be a literal filehandle
6809 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6811 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6812 PL_lex_op = readline_overriden
6813 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6814 append_elem(OP_LIST,
6815 newGVOP(OP_GV, 0, gv),
6816 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
6817 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6818 yylval.ival = OP_NULL;
6827 takes: start position in buffer
6828 keep_quoted preserve \ on the embedded delimiter(s)
6829 keep_delims preserve the delimiters around the string
6830 returns: position to continue reading from buffer
6831 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6832 updates the read buffer.
6834 This subroutine pulls a string out of the input. It is called for:
6835 q single quotes q(literal text)
6836 ' single quotes 'literal text'
6837 qq double quotes qq(interpolate $here please)
6838 " double quotes "interpolate $here please"
6839 qx backticks qx(/bin/ls -l)
6840 ` backticks `/bin/ls -l`
6841 qw quote words @EXPORT_OK = qw( func() $spam )
6842 m// regexp match m/this/
6843 s/// regexp substitute s/this/that/
6844 tr/// string transliterate tr/this/that/
6845 y/// string transliterate y/this/that/
6846 ($*@) sub prototypes sub foo ($)
6847 (stuff) sub attr parameters sub foo : attr(stuff)
6848 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6850 In most of these cases (all but <>, patterns and transliterate)
6851 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6852 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6853 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6856 It skips whitespace before the string starts, and treats the first
6857 character as the delimiter. If the delimiter is one of ([{< then
6858 the corresponding "close" character )]}> is used as the closing
6859 delimiter. It allows quoting of delimiters, and if the string has
6860 balanced delimiters ([{<>}]) it allows nesting.
6862 On success, the SV with the resulting string is put into lex_stuff or,
6863 if that is already non-NULL, into lex_repl. The second case occurs only
6864 when parsing the RHS of the special constructs s/// and tr/// (y///).
6865 For convenience, the terminating delimiter character is stuffed into
6870 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6872 SV *sv; /* scalar value: string */
6873 char *tmps; /* temp string, used for delimiter matching */
6874 register char *s = start; /* current position in the buffer */
6875 register char term; /* terminating character */
6876 register char *to; /* current position in the sv's data */
6877 I32 brackets = 1; /* bracket nesting level */
6878 bool has_utf8 = FALSE; /* is there any utf8 content? */
6880 /* skip space before the delimiter */
6884 /* mark where we are, in case we need to report errors */
6887 /* after skipping whitespace, the next character is the terminator */
6889 if (!UTF8_IS_INVARIANT((U8)term) && UTF)
6892 /* mark where we are */
6893 PL_multi_start = CopLINE(PL_curcop);
6894 PL_multi_open = term;
6896 /* find corresponding closing delimiter */
6897 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6899 PL_multi_close = term;
6901 /* create a new SV to hold the contents. 87 is leak category, I'm
6902 assuming. 79 is the SV's initial length. What a random number. */
6904 sv_upgrade(sv, SVt_PVIV);
6906 (void)SvPOK_only(sv); /* validate pointer */
6908 /* move past delimiter and try to read a complete string */
6910 sv_catpvn(sv, s, 1);
6913 /* extend sv if need be */
6914 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6915 /* set 'to' to the next character in the sv's string */
6916 to = SvPVX(sv)+SvCUR(sv);
6918 /* if open delimiter is the close delimiter read unbridle */
6919 if (PL_multi_open == PL_multi_close) {
6920 for (; s < PL_bufend; s++,to++) {
6921 /* embedded newlines increment the current line number */
6922 if (*s == '\n' && !PL_rsfp)
6923 CopLINE_inc(PL_curcop);
6924 /* handle quoted delimiters */
6925 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6926 if (!keep_quoted && s[1] == term)
6928 /* any other quotes are simply copied straight through */
6932 /* terminate when run out of buffer (the for() condition), or
6933 have found the terminator */
6934 else if (*s == term)
6936 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
6942 /* if the terminator isn't the same as the start character (e.g.,
6943 matched brackets), we have to allow more in the quoting, and
6944 be prepared for nested brackets.
6947 /* read until we run out of string, or we find the terminator */
6948 for (; s < PL_bufend; s++,to++) {
6949 /* embedded newlines increment the line count */
6950 if (*s == '\n' && !PL_rsfp)
6951 CopLINE_inc(PL_curcop);
6952 /* backslashes can escape the open or closing characters */
6953 if (*s == '\\' && s+1 < PL_bufend) {
6955 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6960 /* allow nested opens and closes */
6961 else if (*s == PL_multi_close && --brackets <= 0)
6963 else if (*s == PL_multi_open)
6965 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
6970 /* terminate the copied string and update the sv's end-of-string */
6972 SvCUR_set(sv, to - SvPVX(sv));
6975 * this next chunk reads more into the buffer if we're not done yet
6979 break; /* handle case where we are done yet :-) */
6981 #ifndef PERL_STRICT_CR
6982 if (to - SvPVX(sv) >= 2) {
6983 if ((to[-2] == '\r' && to[-1] == '\n') ||
6984 (to[-2] == '\n' && to[-1] == '\r'))
6988 SvCUR_set(sv, to - SvPVX(sv));
6990 else if (to[-1] == '\r')
6993 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6997 /* if we're out of file, or a read fails, bail and reset the current
6998 line marker so we can report where the unterminated string began
7001 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
7003 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
7006 /* we read a line, so increment our line counter */
7007 CopLINE_inc(PL_curcop);
7009 /* update debugger info */
7010 if (PERLDB_LINE && PL_curstash != PL_debstash) {
7011 SV *sv = NEWSV(88,0);
7013 sv_upgrade(sv, SVt_PVMG);
7014 sv_setsv(sv,PL_linestr);
7017 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
7020 /* having changed the buffer, we must update PL_bufend */
7021 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7022 PL_last_lop = PL_last_uni = Nullch;
7025 /* at this point, we have successfully read the delimited string */
7028 sv_catpvn(sv, s, 1);
7031 PL_multi_end = CopLINE(PL_curcop);
7034 /* if we allocated too much space, give some back */
7035 if (SvCUR(sv) + 5 < SvLEN(sv)) {
7036 SvLEN_set(sv, SvCUR(sv) + 1);
7037 Renew(SvPVX(sv), SvLEN(sv), char);
7040 /* decide whether this is the first or second quoted string we've read
7053 takes: pointer to position in buffer
7054 returns: pointer to new position in buffer
7055 side-effects: builds ops for the constant in yylval.op
7057 Read a number in any of the formats that Perl accepts:
7059 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
7060 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
7063 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
7065 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
7068 If it reads a number without a decimal point or an exponent, it will
7069 try converting the number to an integer and see if it can do so
7070 without loss of precision.
7074 Perl_scan_num(pTHX_ char *start, YYSTYPE* lvalp)
7076 register char *s = start; /* current position in buffer */
7077 register char *d; /* destination in temp buffer */
7078 register char *e; /* end of temp buffer */
7079 NV nv; /* number read, as a double */
7080 SV *sv = Nullsv; /* place to put the converted number */
7081 bool floatit; /* boolean: int or float? */
7082 char *lastub = 0; /* position of last underbar */
7083 static char number_too_long[] = "Number too long";
7085 /* We use the first character to decide what type of number this is */
7089 Perl_croak(aTHX_ "panic: scan_num");
7091 /* if it starts with a 0, it could be an octal number, a decimal in
7092 0.13 disguise, or a hexadecimal number, or a binary number. */
7096 u holds the "number so far"
7097 shift the power of 2 of the base
7098 (hex == 4, octal == 3, binary == 1)
7099 overflowed was the number more than we can hold?
7101 Shift is used when we add a digit. It also serves as an "are
7102 we in octal/hex/binary?" indicator to disallow hex characters
7108 bool overflowed = FALSE;
7109 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
7110 static char* bases[5] = { "", "binary", "", "octal",
7112 static char* Bases[5] = { "", "Binary", "", "Octal",
7114 static char *maxima[5] = { "",
7115 "0b11111111111111111111111111111111",
7119 char *base, *Base, *max;
7125 } else if (s[1] == 'b') {
7129 /* check for a decimal in disguise */
7130 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
7132 /* so it must be octal */
7139 if (ckWARN(WARN_SYNTAX))
7140 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7141 "Misplaced _ in number");
7145 base = bases[shift];
7146 Base = Bases[shift];
7147 max = maxima[shift];
7149 /* read the rest of the number */
7151 /* x is used in the overflow test,
7152 b is the digit we're adding on. */
7157 /* if we don't mention it, we're done */
7161 /* _ are ignored -- but warned about if consecutive */
7163 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7164 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7165 "Misplaced _ in number");
7169 /* 8 and 9 are not octal */
7172 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
7176 case '2': case '3': case '4':
7177 case '5': case '6': case '7':
7179 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
7183 b = *s++ & 15; /* ASCII digit -> value of digit */
7187 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
7188 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
7189 /* make sure they said 0x */
7194 /* Prepare to put the digit we have onto the end
7195 of the number so far. We check for overflows.
7200 x = u << shift; /* make room for the digit */
7202 if ((x >> shift) != u
7203 && !(PL_hints & HINT_NEW_BINARY)) {
7206 if (ckWARN_d(WARN_OVERFLOW))
7207 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
7208 "Integer overflow in %s number",
7211 u = x | b; /* add the digit to the end */
7214 n *= nvshift[shift];
7215 /* If an NV has not enough bits in its
7216 * mantissa to represent an UV this summing of
7217 * small low-order numbers is a waste of time
7218 * (because the NV cannot preserve the
7219 * low-order bits anyway): we could just
7220 * remember when did we overflow and in the
7221 * end just multiply n by the right
7229 /* if we get here, we had success: make a scalar value from
7234 /* final misplaced underbar check */
7236 if (ckWARN(WARN_SYNTAX))
7237 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
7242 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
7243 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
7244 "%s number > %s non-portable",
7250 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
7251 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
7252 "%s number > %s non-portable",
7257 if (PL_hints & HINT_NEW_BINARY)
7258 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
7263 handle decimal numbers.
7264 we're also sent here when we read a 0 as the first digit
7266 case '1': case '2': case '3': case '4': case '5':
7267 case '6': case '7': case '8': case '9': case '.':
7270 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
7273 /* read next group of digits and _ and copy into d */
7274 while (isDIGIT(*s) || *s == '_') {
7275 /* skip underscores, checking for misplaced ones
7279 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7280 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7281 "Misplaced _ in number");
7285 /* check for end of fixed-length buffer */
7287 Perl_croak(aTHX_ number_too_long);
7288 /* if we're ok, copy the character */
7293 /* final misplaced underbar check */
7294 if (lastub && s == lastub + 1) {
7295 if (ckWARN(WARN_SYNTAX))
7296 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
7299 /* read a decimal portion if there is one. avoid
7300 3..5 being interpreted as the number 3. followed
7303 if (*s == '.' && s[1] != '.') {
7308 if (ckWARN(WARN_SYNTAX))
7309 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7310 "Misplaced _ in number");
7314 /* copy, ignoring underbars, until we run out of digits.
7316 for (; isDIGIT(*s) || *s == '_'; s++) {
7317 /* fixed length buffer check */
7319 Perl_croak(aTHX_ number_too_long);
7321 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7322 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7323 "Misplaced _ in number");
7329 /* fractional part ending in underbar? */
7331 if (ckWARN(WARN_SYNTAX))
7332 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7333 "Misplaced _ in number");
7335 if (*s == '.' && isDIGIT(s[1])) {
7336 /* oops, it's really a v-string, but without the "v" */
7342 /* read exponent part, if present */
7343 if (*s && strchr("eE",*s) && strchr("+-0123456789_", s[1])) {
7347 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
7348 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
7350 /* stray preinitial _ */
7352 if (ckWARN(WARN_SYNTAX))
7353 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7354 "Misplaced _ in number");
7358 /* allow positive or negative exponent */
7359 if (*s == '+' || *s == '-')
7362 /* stray initial _ */
7364 if (ckWARN(WARN_SYNTAX))
7365 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7366 "Misplaced _ in number");
7370 /* read digits of exponent */
7371 while (isDIGIT(*s) || *s == '_') {
7374 Perl_croak(aTHX_ number_too_long);
7378 if (ckWARN(WARN_SYNTAX) &&
7379 ((lastub && s == lastub + 1) ||
7380 (!isDIGIT(s[1]) && s[1] != '_')))
7381 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7382 "Misplaced _ in number");
7389 /* make an sv from the string */
7393 We try to do an integer conversion first if no characters
7394 indicating "float" have been found.
7399 int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
7401 if (flags == IS_NUMBER_IN_UV) {
7403 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7406 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
7407 if (uv <= (UV) IV_MIN)
7408 sv_setiv(sv, -(IV)uv);
7415 /* terminate the string */
7417 nv = Atof(PL_tokenbuf);
7421 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7422 (PL_hints & HINT_NEW_INTEGER) )
7423 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7424 (floatit ? "float" : "integer"),
7428 /* if it starts with a v, it could be a v-string */
7431 sv = NEWSV(92,5); /* preallocate storage space */
7432 s = scan_vstring(s,sv);
7436 /* make the op for the constant and return */
7439 lvalp->opval = newSVOP(OP_CONST, 0, sv);
7441 lvalp->opval = Nullop;
7447 S_scan_formline(pTHX_ register char *s)
7451 SV *stuff = newSVpvn("",0);
7452 bool needargs = FALSE;
7455 if (*s == '.' || *s == /*{*/'}') {
7457 #ifdef PERL_STRICT_CR
7458 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7460 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7462 if (*t == '\n' || t == PL_bufend)
7465 if (PL_in_eval && !PL_rsfp) {
7466 eol = strchr(s,'\n');
7471 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7473 for (t = s; t < eol; t++) {
7474 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7476 goto enough; /* ~~ must be first line in formline */
7478 if (*t == '@' || *t == '^')
7482 sv_catpvn(stuff, s, eol-s);
7483 #ifndef PERL_STRICT_CR
7484 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7485 char *end = SvPVX(stuff) + SvCUR(stuff);
7497 s = filter_gets(PL_linestr, PL_rsfp, 0);
7498 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7499 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7500 PL_last_lop = PL_last_uni = Nullch;
7503 yyerror("Format not terminated");
7513 PL_lex_state = LEX_NORMAL;
7514 PL_nextval[PL_nexttoke].ival = 0;
7518 PL_lex_state = LEX_FORMLINE;
7519 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7521 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7525 SvREFCNT_dec(stuff);
7526 PL_lex_formbrack = 0;
7537 PL_cshlen = strlen(PL_cshname);
7542 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7544 I32 oldsavestack_ix = PL_savestack_ix;
7545 CV* outsidecv = PL_compcv;
7548 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7550 SAVEI32(PL_subline);
7551 save_item(PL_subname);
7552 SAVESPTR(PL_compcv);
7554 PL_compcv = (CV*)NEWSV(1104,0);
7555 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7556 CvFLAGS(PL_compcv) |= flags;
7558 PL_subline = CopLINE(PL_curcop);
7559 CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
7560 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7562 return oldsavestack_ix;
7566 #pragma segment Perl_yylex
7569 Perl_yywarn(pTHX_ char *s)
7571 PL_in_eval |= EVAL_WARNONLY;
7573 PL_in_eval &= ~EVAL_WARNONLY;
7578 Perl_yyerror(pTHX_ char *s)
7581 char *context = NULL;
7585 if (!yychar || (yychar == ';' && !PL_rsfp))
7587 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7588 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7591 The code below is removed for NetWare because it abends/crashes on NetWare
7592 when the script has error such as not having the closing quotes like:
7594 Checking of white spaces is anyway done in NetWare code.
7597 while (isSPACE(*PL_oldoldbufptr))
7600 context = PL_oldoldbufptr;
7601 contlen = PL_bufptr - PL_oldoldbufptr;
7603 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7604 PL_oldbufptr != PL_bufptr) {
7607 The code below is removed for NetWare because it abends/crashes on NetWare
7608 when the script has error such as not having the closing quotes like:
7610 Checking of white spaces is anyway done in NetWare code.
7613 while (isSPACE(*PL_oldbufptr))
7616 context = PL_oldbufptr;
7617 contlen = PL_bufptr - PL_oldbufptr;
7619 else if (yychar > 255)
7620 where = "next token ???";
7621 #ifdef USE_PURE_BISON
7622 /* GNU Bison sets the value -2 */
7623 else if (yychar == -2) {
7625 else if ((yychar & 127) == 127) {
7627 if (PL_lex_state == LEX_NORMAL ||
7628 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7629 where = "at end of line";
7630 else if (PL_lex_inpat)
7631 where = "within pattern";
7633 where = "within string";
7636 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7638 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7639 else if (isPRINT_LC(yychar))
7640 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7642 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7643 where = SvPVX(where_sv);
7645 msg = sv_2mortal(newSVpv(s, 0));
7646 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7647 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7649 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7651 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7652 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7653 Perl_sv_catpvf(aTHX_ msg,
7654 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7655 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7658 if (PL_in_eval & EVAL_WARNONLY)
7659 Perl_warn(aTHX_ "%"SVf, msg);
7662 if (PL_error_count >= 10) {
7663 if (PL_in_eval && SvCUR(ERRSV))
7664 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
7665 ERRSV, OutCopFILE(PL_curcop));
7667 Perl_croak(aTHX_ "%s has too many errors.\n",
7668 OutCopFILE(PL_curcop));
7671 PL_in_my_stash = Nullhv;
7675 #pragma segment Main
7679 S_swallow_bom(pTHX_ U8 *s)
7682 slen = SvCUR(PL_linestr);
7686 /* UTF-16 little-endian */
7687 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
7688 Perl_croak(aTHX_ "Unsupported script encoding");
7689 #ifndef PERL_NO_UTF16_FILTER
7690 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-LE script encoding\n"));
7692 if (PL_bufend > (char*)s) {
7696 filter_add(utf16rev_textfilter, NULL);
7697 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7698 PL_bufend = (char*)utf16_to_utf8_reversed(s, news,
7699 PL_bufend - (char*)s - 1,
7701 Copy(news, s, newlen, U8);
7702 SvCUR_set(PL_linestr, newlen);
7703 PL_bufend = SvPVX(PL_linestr) + newlen;
7704 news[newlen++] = '\0';
7708 Perl_croak(aTHX_ "Unsupported script encoding");
7713 if (s[1] == 0xFF) { /* UTF-16 big-endian */
7714 #ifndef PERL_NO_UTF16_FILTER
7715 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding\n"));
7717 if (PL_bufend > (char *)s) {
7721 filter_add(utf16_textfilter, NULL);
7722 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7723 PL_bufend = (char*)utf16_to_utf8(s, news,
7724 PL_bufend - (char*)s,
7726 Copy(news, s, newlen, U8);
7727 SvCUR_set(PL_linestr, newlen);
7728 PL_bufend = SvPVX(PL_linestr) + newlen;
7729 news[newlen++] = '\0';
7733 Perl_croak(aTHX_ "Unsupported script encoding");
7738 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
7739 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-8 script encoding\n"));
7744 if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */
7745 s[2] == 0xFE && s[3] == 0xFF)
7747 Perl_croak(aTHX_ "Unsupported script encoding");
7755 * Restore a source filter.
7759 restore_rsfp(pTHX_ void *f)
7761 PerlIO *fp = (PerlIO*)f;
7763 if (PL_rsfp == PerlIO_stdin())
7764 PerlIO_clearerr(PL_rsfp);
7765 else if (PL_rsfp && (PL_rsfp != fp))
7766 PerlIO_close(PL_rsfp);
7770 #ifndef PERL_NO_UTF16_FILTER
7772 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
7774 I32 count = FILTER_READ(idx+1, sv, maxlen);
7779 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7780 if (!*SvPV_nolen(sv))
7781 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7784 tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7785 sv_usepvn(sv, (char*)tmps, tend - tmps);
7791 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
7793 I32 count = FILTER_READ(idx+1, sv, maxlen);
7798 if (!*SvPV_nolen(sv))
7799 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7802 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7803 tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7804 sv_usepvn(sv, (char*)tmps, tend - tmps);