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));
1623 yyerror("Missing control char name in \\c");
1627 /* printf-style backslashes, formfeeds, newlines, etc */
1629 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
1632 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
1635 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
1638 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
1641 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
1644 *d++ = ASCII_TO_NEED(has_utf8,'\033');
1647 *d++ = ASCII_TO_NEED(has_utf8,'\007');
1653 } /* end if (backslash) */
1656 /* If we started with encoded form, or already know we want it
1657 and then encode the next character */
1658 if ((has_utf8 || this_utf8) && !NATIVE_IS_INVARIANT((U8)(*s))) {
1660 UV uv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
1661 STRLEN need = UNISKIP(NATIVE_TO_UNI(uv));
1664 /* encoded value larger than old, need extra space (NOTE: SvCUR() not set here) */
1665 STRLEN off = d - SvPVX(sv);
1666 d = SvGROW(sv, SvLEN(sv) + (need-len)) + off;
1668 d = (char*)uvchr_to_utf8((U8*)d, uv);
1672 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1674 } /* while loop to process each character */
1676 /* terminate the string and set up the sv */
1678 SvCUR_set(sv, d - SvPVX(sv));
1679 if (SvCUR(sv) >= SvLEN(sv))
1680 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
1683 if (PL_encoding && !has_utf8) {
1684 sv_recode_to_utf8(sv, PL_encoding);
1690 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1691 PL_sublex_info.sub_op->op_private |=
1692 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1696 /* shrink the sv if we allocated more than we used */
1697 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1698 SvLEN_set(sv, SvCUR(sv) + 1);
1699 Renew(SvPVX(sv), SvLEN(sv), char);
1702 /* return the substring (via yylval) only if we parsed anything */
1703 if (s > PL_bufptr) {
1704 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1705 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1707 ( PL_lex_inwhat == OP_TRANS
1709 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1712 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1719 * Returns TRUE if there's more to the expression (e.g., a subscript),
1722 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1724 * ->[ and ->{ return TRUE
1725 * { and [ outside a pattern are always subscripts, so return TRUE
1726 * if we're outside a pattern and it's not { or [, then return FALSE
1727 * if we're in a pattern and the first char is a {
1728 * {4,5} (any digits around the comma) returns FALSE
1729 * if we're in a pattern and the first char is a [
1731 * [SOMETHING] has a funky algorithm to decide whether it's a
1732 * character class or not. It has to deal with things like
1733 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1734 * anything else returns TRUE
1737 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1740 S_intuit_more(pTHX_ register char *s)
1742 if (PL_lex_brackets)
1744 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1746 if (*s != '{' && *s != '[')
1751 /* In a pattern, so maybe we have {n,m}. */
1768 /* On the other hand, maybe we have a character class */
1771 if (*s == ']' || *s == '^')
1774 /* this is terrifying, and it works */
1775 int weight = 2; /* let's weigh the evidence */
1777 unsigned char un_char = 255, last_un_char;
1778 char *send = strchr(s,']');
1779 char tmpbuf[sizeof PL_tokenbuf * 4];
1781 if (!send) /* has to be an expression */
1784 Zero(seen,256,char);
1787 else if (isDIGIT(*s)) {
1789 if (isDIGIT(s[1]) && s[2] == ']')
1795 for (; s < send; s++) {
1796 last_un_char = un_char;
1797 un_char = (unsigned char)*s;
1802 weight -= seen[un_char] * 10;
1803 if (isALNUM_lazy_if(s+1,UTF)) {
1804 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1805 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1810 else if (*s == '$' && s[1] &&
1811 strchr("[#!%*<>()-=",s[1])) {
1812 if (/*{*/ strchr("])} =",s[2]))
1821 if (strchr("wds]",s[1]))
1823 else if (seen['\''] || seen['"'])
1825 else if (strchr("rnftbxcav",s[1]))
1827 else if (isDIGIT(s[1])) {
1829 while (s[1] && isDIGIT(s[1]))
1839 if (strchr("aA01! ",last_un_char))
1841 if (strchr("zZ79~",s[1]))
1843 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1844 weight -= 5; /* cope with negative subscript */
1847 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1848 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1853 if (keyword(tmpbuf, d - tmpbuf))
1856 if (un_char == last_un_char + 1)
1858 weight -= seen[un_char];
1863 if (weight >= 0) /* probably a character class */
1873 * Does all the checking to disambiguate
1875 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1876 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1878 * First argument is the stuff after the first token, e.g. "bar".
1880 * Not a method if bar is a filehandle.
1881 * Not a method if foo is a subroutine prototyped to take a filehandle.
1882 * Not a method if it's really "Foo $bar"
1883 * Method if it's "foo $bar"
1884 * Not a method if it's really "print foo $bar"
1885 * Method if it's really "foo package::" (interpreted as package->foo)
1886 * Not a method if bar is known to be a subroutine ("sub bar; foo bar")
1887 * Not a method if bar is a filehandle or package, but is quoted with
1892 S_intuit_method(pTHX_ char *start, GV *gv)
1894 char *s = start + (*start == '$');
1895 char tmpbuf[sizeof PL_tokenbuf];
1903 if ((cv = GvCVu(gv))) {
1904 char *proto = SvPVX(cv);
1914 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1915 /* start is the beginning of the possible filehandle/object,
1916 * and s is the end of it
1917 * tmpbuf is a copy of it
1920 if (*start == '$') {
1921 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1926 return *s == '(' ? FUNCMETH : METHOD;
1928 if (!keyword(tmpbuf, len)) {
1929 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1934 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1935 if (indirgv && GvCVu(indirgv))
1937 /* filehandle or package name makes it a method */
1938 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1940 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1941 return 0; /* no assumptions -- "=>" quotes bearword */
1943 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1944 newSVpvn(tmpbuf,len));
1945 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1949 return *s == '(' ? FUNCMETH : METHOD;
1957 * Return a string of Perl code to load the debugger. If PERL5DB
1958 * is set, it will return the contents of that, otherwise a
1959 * compile-time require of perl5db.pl.
1966 char *pdb = PerlEnv_getenv("PERL5DB");
1970 SETERRNO(0,SS_NORMAL);
1971 return "BEGIN { require 'perl5db.pl' }";
1977 /* Encoded script support. filter_add() effectively inserts a
1978 * 'pre-processing' function into the current source input stream.
1979 * Note that the filter function only applies to the current source file
1980 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1982 * The datasv parameter (which may be NULL) can be used to pass
1983 * private data to this instance of the filter. The filter function
1984 * can recover the SV using the FILTER_DATA macro and use it to
1985 * store private buffers and state information.
1987 * The supplied datasv parameter is upgraded to a PVIO type
1988 * and the IoDIRP/IoANY field is used to store the function pointer,
1989 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1990 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1991 * private use must be set using malloc'd pointers.
1995 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
2000 if (!PL_rsfp_filters)
2001 PL_rsfp_filters = newAV();
2003 datasv = NEWSV(255,0);
2004 if (!SvUPGRADE(datasv, SVt_PVIO))
2005 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
2006 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
2007 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
2008 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
2009 (void*)funcp, SvPV_nolen(datasv)));
2010 av_unshift(PL_rsfp_filters, 1);
2011 av_store(PL_rsfp_filters, 0, datasv) ;
2016 /* Delete most recently added instance of this filter function. */
2018 Perl_filter_del(pTHX_ filter_t funcp)
2021 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", (void*)funcp));
2022 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
2024 /* if filter is on top of stack (usual case) just pop it off */
2025 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
2026 if (IoANY(datasv) == (void *)funcp) {
2027 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
2028 IoANY(datasv) = (void *)NULL;
2029 sv_free(av_pop(PL_rsfp_filters));
2033 /* we need to search for the correct entry and clear it */
2034 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
2038 /* Invoke the n'th filter function for the current rsfp. */
2040 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
2043 /* 0 = read one text line */
2048 if (!PL_rsfp_filters)
2050 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
2051 /* Provide a default input filter to make life easy. */
2052 /* Note that we append to the line. This is handy. */
2053 DEBUG_P(PerlIO_printf(Perl_debug_log,
2054 "filter_read %d: from rsfp\n", idx));
2058 int old_len = SvCUR(buf_sv) ;
2060 /* ensure buf_sv is large enough */
2061 SvGROW(buf_sv, (STRLEN)(old_len + maxlen)) ;
2062 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
2063 if (PerlIO_error(PL_rsfp))
2064 return -1; /* error */
2066 return 0 ; /* end of file */
2068 SvCUR_set(buf_sv, old_len + len) ;
2071 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2072 if (PerlIO_error(PL_rsfp))
2073 return -1; /* error */
2075 return 0 ; /* end of file */
2078 return SvCUR(buf_sv);
2080 /* Skip this filter slot if filter has been deleted */
2081 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
2082 DEBUG_P(PerlIO_printf(Perl_debug_log,
2083 "filter_read %d: skipped (filter deleted)\n",
2085 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
2087 /* Get function pointer hidden within datasv */
2088 funcp = (filter_t)IoANY(datasv);
2089 DEBUG_P(PerlIO_printf(Perl_debug_log,
2090 "filter_read %d: via function %p (%s)\n",
2091 idx, (void*)funcp, SvPV_nolen(datasv)));
2092 /* Call function. The function is expected to */
2093 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2094 /* Return: <0:error, =0:eof, >0:not eof */
2095 return (*funcp)(aTHX_ idx, buf_sv, maxlen);
2099 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2101 #ifdef PERL_CR_FILTER
2102 if (!PL_rsfp_filters) {
2103 filter_add(S_cr_textfilter,NULL);
2106 if (PL_rsfp_filters) {
2109 SvCUR_set(sv, 0); /* start with empty line */
2110 if (FILTER_READ(0, sv, 0) > 0)
2111 return ( SvPVX(sv) ) ;
2116 return (sv_gets(sv, fp, append));
2120 S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2124 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2128 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2129 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2131 return GvHV(gv); /* Foo:: */
2134 /* use constant CLASS => 'MyClass' */
2135 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2137 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2138 pkgname = SvPV_nolen(sv);
2142 return gv_stashpv(pkgname, FALSE);
2146 static char* exp_name[] =
2147 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2148 "ATTRTERM", "TERMBLOCK"
2155 Works out what to call the token just pulled out of the input
2156 stream. The yacc parser takes care of taking the ops we return and
2157 stitching them into a tree.
2163 if read an identifier
2164 if we're in a my declaration
2165 croak if they tried to say my($foo::bar)
2166 build the ops for a my() declaration
2167 if it's an access to a my() variable
2168 are we in a sort block?
2169 croak if my($a); $a <=> $b
2170 build ops for access to a my() variable
2171 if in a dq string, and they've said @foo and we can't find @foo
2173 build ops for a bareword
2174 if we already built the token before, use it.
2177 #ifdef USE_PURE_BISON
2179 Perl_yylex_r(pTHX_ YYSTYPE *lvalp, int *lcharp)
2184 yylval_pointer[yyactlevel] = lvalp;
2185 yychar_pointer[yyactlevel] = lcharp;
2186 if (yyactlevel >= YYMAXLEVEL)
2187 Perl_croak(aTHX_ "panic: YYMAXLEVEL");
2189 r = Perl_yylex(aTHX);
2199 #pragma segment Perl_yylex
2211 I32 orig_keyword = 0;
2213 /* check if there's an identifier for us to look at */
2214 if (PL_pending_ident)
2215 return S_pending_ident(aTHX);
2217 /* no identifier pending identification */
2219 switch (PL_lex_state) {
2221 case LEX_NORMAL: /* Some compilers will produce faster */
2222 case LEX_INTERPNORMAL: /* code if we comment these out. */
2226 /* when we've already built the next token, just pull it out of the queue */
2229 yylval = PL_nextval[PL_nexttoke];
2231 PL_lex_state = PL_lex_defer;
2232 PL_expect = PL_lex_expect;
2233 PL_lex_defer = LEX_NORMAL;
2235 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2236 "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr,
2237 (IV)PL_nexttype[PL_nexttoke]); });
2239 return(PL_nexttype[PL_nexttoke]);
2241 /* interpolated case modifiers like \L \U, including \Q and \E.
2242 when we get here, PL_bufptr is at the \
2244 case LEX_INTERPCASEMOD:
2246 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2247 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2249 /* handle \E or end of string */
2250 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2254 if (PL_lex_casemods) {
2255 oldmod = PL_lex_casestack[--PL_lex_casemods];
2256 PL_lex_casestack[PL_lex_casemods] = '\0';
2258 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2260 PL_lex_state = LEX_INTERPCONCAT;
2264 if (PL_bufptr != PL_bufend)
2266 PL_lex_state = LEX_INTERPCONCAT;
2270 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2271 "### Saw case modifier at '%s'\n", PL_bufptr); });
2273 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2274 tmp = *s, *s = s[2], s[2] = (char)tmp; /* misordered... */
2275 if (strchr("LU", *s) &&
2276 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2278 PL_lex_casestack[--PL_lex_casemods] = '\0';
2281 if (PL_lex_casemods > 10) {
2282 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2283 if (newlb != PL_lex_casestack) {
2285 PL_lex_casestack = newlb;
2288 PL_lex_casestack[PL_lex_casemods++] = *s;
2289 PL_lex_casestack[PL_lex_casemods] = '\0';
2290 PL_lex_state = LEX_INTERPCONCAT;
2291 PL_nextval[PL_nexttoke].ival = 0;
2294 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2296 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2298 PL_nextval[PL_nexttoke].ival = OP_LC;
2300 PL_nextval[PL_nexttoke].ival = OP_UC;
2302 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2304 Perl_croak(aTHX_ "panic: yylex");
2307 if (PL_lex_starts) {
2316 case LEX_INTERPPUSH:
2317 return sublex_push();
2319 case LEX_INTERPSTART:
2320 if (PL_bufptr == PL_bufend)
2321 return sublex_done();
2322 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2323 "### Interpolated variable at '%s'\n", PL_bufptr); });
2325 PL_lex_dojoin = (*PL_bufptr == '@');
2326 PL_lex_state = LEX_INTERPNORMAL;
2327 if (PL_lex_dojoin) {
2328 PL_nextval[PL_nexttoke].ival = 0;
2330 force_ident("\"", '$');
2331 PL_nextval[PL_nexttoke].ival = 0;
2333 PL_nextval[PL_nexttoke].ival = 0;
2335 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2338 if (PL_lex_starts++) {
2344 case LEX_INTERPENDMAYBE:
2345 if (intuit_more(PL_bufptr)) {
2346 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2352 if (PL_lex_dojoin) {
2353 PL_lex_dojoin = FALSE;
2354 PL_lex_state = LEX_INTERPCONCAT;
2357 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2358 && SvEVALED(PL_lex_repl))
2360 if (PL_bufptr != PL_bufend)
2361 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2362 PL_lex_repl = Nullsv;
2365 case LEX_INTERPCONCAT:
2367 if (PL_lex_brackets)
2368 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2370 if (PL_bufptr == PL_bufend)
2371 return sublex_done();
2373 if (SvIVX(PL_linestr) == '\'') {
2374 SV *sv = newSVsv(PL_linestr);
2377 else if ( PL_hints & HINT_NEW_RE )
2378 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2379 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2383 s = scan_const(PL_bufptr);
2385 PL_lex_state = LEX_INTERPCASEMOD;
2387 PL_lex_state = LEX_INTERPSTART;
2390 if (s != PL_bufptr) {
2391 PL_nextval[PL_nexttoke] = yylval;
2394 if (PL_lex_starts++)
2404 PL_lex_state = LEX_NORMAL;
2405 s = scan_formline(PL_bufptr);
2406 if (!PL_lex_formbrack)
2412 PL_oldoldbufptr = PL_oldbufptr;
2415 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2416 exp_name[PL_expect], s);
2422 if (isIDFIRST_lazy_if(s,UTF))
2424 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2427 goto fake_eof; /* emulate EOF on ^D or ^Z */
2432 if (PL_lex_brackets)
2433 yyerror("Missing right curly or square bracket");
2434 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2435 "### Tokener got EOF\n");
2439 if (s++ < PL_bufend)
2440 goto retry; /* ignore stray nulls */
2443 if (!PL_in_eval && !PL_preambled) {
2444 PL_preambled = TRUE;
2445 sv_setpv(PL_linestr,incl_perldb());
2446 if (SvCUR(PL_linestr))
2447 sv_catpv(PL_linestr,";");
2449 while(AvFILLp(PL_preambleav) >= 0) {
2450 SV *tmpsv = av_shift(PL_preambleav);
2451 sv_catsv(PL_linestr, tmpsv);
2452 sv_catpv(PL_linestr, ";");
2455 sv_free((SV*)PL_preambleav);
2456 PL_preambleav = NULL;
2458 if (PL_minus_n || PL_minus_p) {
2459 sv_catpv(PL_linestr, "LINE: while (<>) {");
2461 sv_catpv(PL_linestr,"chomp;");
2464 if (strchr("/'\"", *PL_splitstr)
2465 && strchr(PL_splitstr + 1, *PL_splitstr))
2466 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
2469 s = "'~#\200\1'"; /* surely one char is unused...*/
2470 while (s[1] && strchr(PL_splitstr, *s)) s++;
2472 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s%c",
2473 "q" + (delim == '\''), delim);
2474 for (s = PL_splitstr; *s; s++) {
2476 sv_catpvn(PL_linestr, "\\", 1);
2477 sv_catpvn(PL_linestr, s, 1);
2479 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2483 sv_catpv(PL_linestr,"our @F=split(' ');");
2486 sv_catpv(PL_linestr, "\n");
2487 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2488 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2489 PL_last_lop = PL_last_uni = Nullch;
2490 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2491 SV *sv = NEWSV(85,0);
2493 sv_upgrade(sv, SVt_PVMG);
2494 sv_setsv(sv,PL_linestr);
2497 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2502 bof = PL_rsfp ? TRUE : FALSE;
2503 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2506 if (PL_preprocess && !PL_in_eval)
2507 (void)PerlProc_pclose(PL_rsfp);
2508 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2509 PerlIO_clearerr(PL_rsfp);
2511 (void)PerlIO_close(PL_rsfp);
2513 PL_doextract = FALSE;
2515 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2516 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2517 sv_catpv(PL_linestr,";}");
2518 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2519 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2520 PL_last_lop = PL_last_uni = Nullch;
2521 PL_minus_n = PL_minus_p = 0;
2524 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2525 PL_last_lop = PL_last_uni = Nullch;
2526 sv_setpv(PL_linestr,"");
2527 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2529 /* if it looks like the start of a BOM, check if it in fact is */
2530 else if (bof && (!*s || *(U8*)s == 0xEF || *(U8*)s >= 0xFE)) {
2531 #ifdef PERLIO_IS_STDIO
2532 # ifdef __GNU_LIBRARY__
2533 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
2534 # define FTELL_FOR_PIPE_IS_BROKEN
2538 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2539 # define FTELL_FOR_PIPE_IS_BROKEN
2544 #ifdef FTELL_FOR_PIPE_IS_BROKEN
2545 /* This loses the possibility to detect the bof
2546 * situation on perl -P when the libc5 is being used.
2547 * Workaround? Maybe attach some extra state to PL_rsfp?
2550 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2552 bof = PerlIO_tell(PL_rsfp) == (Off_t)SvCUR(PL_linestr);
2555 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2556 s = swallow_bom((U8*)s);
2560 /* Incest with pod. */
2561 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2562 sv_setpv(PL_linestr, "");
2563 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2564 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2565 PL_last_lop = PL_last_uni = Nullch;
2566 PL_doextract = FALSE;
2570 } while (PL_doextract);
2571 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2572 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2573 SV *sv = NEWSV(85,0);
2575 sv_upgrade(sv, SVt_PVMG);
2576 sv_setsv(sv,PL_linestr);
2579 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2581 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2582 PL_last_lop = PL_last_uni = Nullch;
2583 if (CopLINE(PL_curcop) == 1) {
2584 while (s < PL_bufend && isSPACE(*s))
2586 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2590 if (*s == '#' && *(s+1) == '!')
2592 #ifdef ALTERNATE_SHEBANG
2594 static char as[] = ALTERNATE_SHEBANG;
2595 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2596 d = s + (sizeof(as) - 1);
2598 #endif /* ALTERNATE_SHEBANG */
2607 while (*d && !isSPACE(*d))
2611 #ifdef ARG_ZERO_IS_SCRIPT
2612 if (ipathend > ipath) {
2614 * HP-UX (at least) sets argv[0] to the script name,
2615 * which makes $^X incorrect. And Digital UNIX and Linux,
2616 * at least, set argv[0] to the basename of the Perl
2617 * interpreter. So, having found "#!", we'll set it right.
2619 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV)); /* $^X */
2620 assert(SvPOK(x) || SvGMAGICAL(x));
2621 if (sv_eq(x, CopFILESV(PL_curcop))) {
2622 sv_setpvn(x, ipath, ipathend - ipath);
2628 char *bstart = SvPV(CopFILESV(PL_curcop),blen);
2629 char *lstart = SvPV(x,llen);
2631 bstart += blen - llen;
2632 if (strnEQ(bstart, lstart, llen) && bstart[-1] == '/') {
2633 sv_setpvn(x, ipath, ipathend - ipath);
2638 TAINT_NOT; /* $^X is always tainted, but that's OK */
2640 #endif /* ARG_ZERO_IS_SCRIPT */
2645 d = instr(s,"perl -");
2647 d = instr(s,"perl");
2649 /* avoid getting into infinite loops when shebang
2650 * line contains "Perl" rather than "perl" */
2652 for (d = ipathend-4; d >= ipath; --d) {
2653 if ((*d == 'p' || *d == 'P')
2654 && !ibcmp(d, "perl", 4))
2664 #ifdef ALTERNATE_SHEBANG
2666 * If the ALTERNATE_SHEBANG on this system starts with a
2667 * character that can be part of a Perl expression, then if
2668 * we see it but not "perl", we're probably looking at the
2669 * start of Perl code, not a request to hand off to some
2670 * other interpreter. Similarly, if "perl" is there, but
2671 * not in the first 'word' of the line, we assume the line
2672 * contains the start of the Perl program.
2674 if (d && *s != '#') {
2676 while (*c && !strchr("; \t\r\n\f\v#", *c))
2679 d = Nullch; /* "perl" not in first word; ignore */
2681 *s = '#'; /* Don't try to parse shebang line */
2683 #endif /* ALTERNATE_SHEBANG */
2684 #ifndef MACOS_TRADITIONAL
2689 !instr(s,"indir") &&
2690 instr(PL_origargv[0],"perl"))
2696 while (s < PL_bufend && isSPACE(*s))
2698 if (s < PL_bufend) {
2699 Newz(899,newargv,PL_origargc+3,char*);
2701 while (s < PL_bufend && !isSPACE(*s))
2704 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2707 newargv = PL_origargv;
2709 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
2710 Perl_croak(aTHX_ "Can't exec %s", ipath);
2714 U32 oldpdb = PL_perldb;
2715 bool oldn = PL_minus_n;
2716 bool oldp = PL_minus_p;
2718 while (*d && !isSPACE(*d)) d++;
2719 while (SPACE_OR_TAB(*d)) d++;
2722 bool switches_done = PL_doswitches;
2724 if (*d == 'M' || *d == 'm') {
2726 while (*d && !isSPACE(*d)) d++;
2727 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2730 d = moreswitches(d);
2732 if ((PERLDB_LINE && !oldpdb) ||
2733 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2734 /* if we have already added "LINE: while (<>) {",
2735 we must not do it again */
2737 sv_setpv(PL_linestr, "");
2738 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2739 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2740 PL_last_lop = PL_last_uni = Nullch;
2741 PL_preambled = FALSE;
2743 (void)gv_fetchfile(PL_origfilename);
2746 if (PL_doswitches && !switches_done) {
2747 int argc = PL_origargc;
2748 char **argv = PL_origargv;
2751 } while (argc && argv[0][0] == '-' && argv[0][1]);
2752 init_argv_symbols(argc,argv);
2758 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2760 PL_lex_state = LEX_FORMLINE;
2765 #ifdef PERL_STRICT_CR
2766 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2768 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2770 case ' ': case '\t': case '\f': case 013:
2771 #ifdef MACOS_TRADITIONAL
2778 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2779 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2780 /* handle eval qq[#line 1 "foo"\n ...] */
2781 CopLINE_dec(PL_curcop);
2785 while (s < d && *s != '\n')
2789 else if (s > d) /* Found by Ilya: feed random input to Perl. */
2790 Perl_croak(aTHX_ "panic: input overflow");
2792 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2794 PL_lex_state = LEX_FORMLINE;
2804 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2811 while (s < PL_bufend && SPACE_OR_TAB(*s))
2814 if (strnEQ(s,"=>",2)) {
2815 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2816 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2817 "### Saw unary minus before =>, forcing word '%s'\n", s);
2819 OPERATOR('-'); /* unary minus */
2821 PL_last_uni = PL_oldbufptr;
2823 case 'r': ftst = OP_FTEREAD; break;
2824 case 'w': ftst = OP_FTEWRITE; break;
2825 case 'x': ftst = OP_FTEEXEC; break;
2826 case 'o': ftst = OP_FTEOWNED; break;
2827 case 'R': ftst = OP_FTRREAD; break;
2828 case 'W': ftst = OP_FTRWRITE; break;
2829 case 'X': ftst = OP_FTREXEC; break;
2830 case 'O': ftst = OP_FTROWNED; break;
2831 case 'e': ftst = OP_FTIS; break;
2832 case 'z': ftst = OP_FTZERO; break;
2833 case 's': ftst = OP_FTSIZE; break;
2834 case 'f': ftst = OP_FTFILE; break;
2835 case 'd': ftst = OP_FTDIR; break;
2836 case 'l': ftst = OP_FTLINK; break;
2837 case 'p': ftst = OP_FTPIPE; break;
2838 case 'S': ftst = OP_FTSOCK; break;
2839 case 'u': ftst = OP_FTSUID; break;
2840 case 'g': ftst = OP_FTSGID; break;
2841 case 'k': ftst = OP_FTSVTX; break;
2842 case 'b': ftst = OP_FTBLK; break;
2843 case 'c': ftst = OP_FTCHR; break;
2844 case 't': ftst = OP_FTTTY; break;
2845 case 'T': ftst = OP_FTTEXT; break;
2846 case 'B': ftst = OP_FTBINARY; break;
2847 case 'M': case 'A': case 'C':
2848 gv_fetchpv("\024",TRUE, SVt_PV);
2850 case 'M': ftst = OP_FTMTIME; break;
2851 case 'A': ftst = OP_FTATIME; break;
2852 case 'C': ftst = OP_FTCTIME; break;
2860 PL_last_lop_op = (OPCODE)ftst;
2861 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2862 "### Saw file test %c\n", (int)ftst);
2867 /* Assume it was a minus followed by a one-letter named
2868 * subroutine call (or a -bareword), then. */
2869 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2870 "### %c looked like a file test but was not\n",
2879 if (PL_expect == XOPERATOR)
2884 else if (*s == '>') {
2887 if (isIDFIRST_lazy_if(s,UTF)) {
2888 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2896 if (PL_expect == XOPERATOR)
2899 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2901 OPERATOR('-'); /* unary minus */
2908 if (PL_expect == XOPERATOR)
2913 if (PL_expect == XOPERATOR)
2916 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2922 if (PL_expect != XOPERATOR) {
2923 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2924 PL_expect = XOPERATOR;
2925 force_ident(PL_tokenbuf, '*');
2938 if (PL_expect == XOPERATOR) {
2942 PL_tokenbuf[0] = '%';
2943 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2944 if (!PL_tokenbuf[1]) {
2946 yyerror("Final % should be \\% or %name");
2949 PL_pending_ident = '%';
2968 switch (PL_expect) {
2971 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2973 PL_bufptr = s; /* update in case we back off */
2979 PL_expect = XTERMBLOCK;
2983 while (isIDFIRST_lazy_if(s,UTF)) {
2984 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2985 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2986 if (tmp < 0) tmp = -tmp;
3002 d = scan_str(d,TRUE,TRUE);
3004 /* MUST advance bufptr here to avoid bogus
3005 "at end of line" context messages from yyerror().
3007 PL_bufptr = s + len;
3008 yyerror("Unterminated attribute parameter in attribute list");
3011 return 0; /* EOF indicator */
3015 SV *sv = newSVpvn(s, len);
3016 sv_catsv(sv, PL_lex_stuff);
3017 attrs = append_elem(OP_LIST, attrs,
3018 newSVOP(OP_CONST, 0, sv));
3019 SvREFCNT_dec(PL_lex_stuff);
3020 PL_lex_stuff = Nullsv;
3023 /* NOTE: any CV attrs applied here need to be part of
3024 the CVf_BUILTIN_ATTRS define in cv.h! */
3025 if (!PL_in_my && len == 6 && strnEQ(s, "lvalue", len))
3026 CvLVALUE_on(PL_compcv);
3027 else if (!PL_in_my && len == 6 && strnEQ(s, "locked", len))
3028 CvLOCKED_on(PL_compcv);
3029 else if (!PL_in_my && len == 6 && strnEQ(s, "method", len))
3030 CvMETHOD_on(PL_compcv);
3032 else if (PL_in_my == KEY_our && len == 6 &&
3033 strnEQ(s, "unique", len))
3034 GvUNIQUE_on(cGVOPx_gv(yylval.opval));
3036 /* After we've set the flags, it could be argued that
3037 we don't need to do the attributes.pm-based setting
3038 process, and shouldn't bother appending recognized
3039 flags. To experiment with that, uncomment the
3040 following "else". (Note that's already been
3041 uncommented. That keeps the above-applied built-in
3042 attributes from being intercepted (and possibly
3043 rejected) by a package's attribute routines, but is
3044 justified by the performance win for the common case
3045 of applying only built-in attributes.) */
3047 attrs = append_elem(OP_LIST, attrs,
3048 newSVOP(OP_CONST, 0,
3052 if (*s == ':' && s[1] != ':')
3055 break; /* require real whitespace or :'s */
3057 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3058 if (*s != ';' && *s != '}' && *s != tmp && (tmp != '=' || *s != ')')) {
3059 char q = ((*s == '\'') ? '"' : '\'');
3060 /* If here for an expression, and parsed no attrs, back off. */
3061 if (tmp == '=' && !attrs) {
3065 /* MUST advance bufptr here to avoid bogus "at end of line"
3066 context messages from yyerror().
3070 yyerror("Unterminated attribute list");
3072 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
3080 PL_nextval[PL_nexttoke].opval = attrs;
3088 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3089 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3105 if (PL_lex_brackets <= 0)
3106 yyerror("Unmatched right square bracket");
3109 if (PL_lex_state == LEX_INTERPNORMAL) {
3110 if (PL_lex_brackets == 0) {
3111 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3112 PL_lex_state = LEX_INTERPEND;
3119 if (PL_lex_brackets > 100) {
3120 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
3121 if (newlb != PL_lex_brackstack) {
3123 PL_lex_brackstack = newlb;
3126 switch (PL_expect) {
3128 if (PL_lex_formbrack) {
3132 if (PL_oldoldbufptr == PL_last_lop)
3133 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3135 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3136 OPERATOR(HASHBRACK);
3138 while (s < PL_bufend && SPACE_OR_TAB(*s))
3141 PL_tokenbuf[0] = '\0';
3142 if (d < PL_bufend && *d == '-') {
3143 PL_tokenbuf[0] = '-';
3145 while (d < PL_bufend && SPACE_OR_TAB(*d))
3148 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3149 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3151 while (d < PL_bufend && SPACE_OR_TAB(*d))
3154 char minus = (PL_tokenbuf[0] == '-');
3155 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3163 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3168 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3173 if (PL_oldoldbufptr == PL_last_lop)
3174 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3176 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3179 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
3181 /* This hack is to get the ${} in the message. */
3183 yyerror("syntax error");
3186 OPERATOR(HASHBRACK);
3188 /* This hack serves to disambiguate a pair of curlies
3189 * as being a block or an anon hash. Normally, expectation
3190 * determines that, but in cases where we're not in a
3191 * position to expect anything in particular (like inside
3192 * eval"") we have to resolve the ambiguity. This code
3193 * covers the case where the first term in the curlies is a
3194 * quoted string. Most other cases need to be explicitly
3195 * disambiguated by prepending a `+' before the opening
3196 * curly in order to force resolution as an anon hash.
3198 * XXX should probably propagate the outer expectation
3199 * into eval"" to rely less on this hack, but that could
3200 * potentially break current behavior of eval"".
3204 if (*s == '\'' || *s == '"' || *s == '`') {
3205 /* common case: get past first string, handling escapes */
3206 for (t++; t < PL_bufend && *t != *s;)
3207 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3211 else if (*s == 'q') {
3214 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3218 char open, close, term;
3221 while (t < PL_bufend && isSPACE(*t))
3225 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3229 for (t++; t < PL_bufend; t++) {
3230 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3232 else if (*t == open)
3236 for (t++; t < PL_bufend; t++) {
3237 if (*t == '\\' && t+1 < PL_bufend)
3239 else if (*t == close && --brackets <= 0)
3241 else if (*t == open)
3247 else if (isALNUM_lazy_if(t,UTF)) {
3249 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3252 while (t < PL_bufend && isSPACE(*t))
3254 /* if comma follows first term, call it an anon hash */
3255 /* XXX it could be a comma expression with loop modifiers */
3256 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3257 || (*t == '=' && t[1] == '>')))
3258 OPERATOR(HASHBRACK);
3259 if (PL_expect == XREF)
3262 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3268 yylval.ival = CopLINE(PL_curcop);
3269 if (isSPACE(*s) || *s == '#')
3270 PL_copline = NOLINE; /* invalidate current command line number */
3275 if (PL_lex_brackets <= 0)
3276 yyerror("Unmatched right curly bracket");
3278 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3279 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3280 PL_lex_formbrack = 0;
3281 if (PL_lex_state == LEX_INTERPNORMAL) {
3282 if (PL_lex_brackets == 0) {
3283 if (PL_expect & XFAKEBRACK) {
3284 PL_expect &= XENUMMASK;
3285 PL_lex_state = LEX_INTERPEND;
3287 return yylex(); /* ignore fake brackets */
3289 if (*s == '-' && s[1] == '>')
3290 PL_lex_state = LEX_INTERPENDMAYBE;
3291 else if (*s != '[' && *s != '{')
3292 PL_lex_state = LEX_INTERPEND;
3295 if (PL_expect & XFAKEBRACK) {
3296 PL_expect &= XENUMMASK;
3298 return yylex(); /* ignore fake brackets */
3308 if (PL_expect == XOPERATOR) {
3309 if (ckWARN(WARN_SEMICOLON)
3310 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3312 CopLINE_dec(PL_curcop);
3313 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
3314 CopLINE_inc(PL_curcop);
3319 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3321 PL_expect = XOPERATOR;
3322 force_ident(PL_tokenbuf, '&');
3326 yylval.ival = (OPpENTERSUB_AMPER<<8);
3345 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3346 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Reversed %c= operator",(int)tmp);
3348 if (PL_expect == XSTATE && isALPHA(tmp) &&
3349 (s == PL_linestart+1 || s[-2] == '\n') )
3351 if (PL_in_eval && !PL_rsfp) {
3356 if (strnEQ(s,"=cut",4)) {
3370 PL_doextract = TRUE;
3373 if (PL_lex_brackets < PL_lex_formbrack) {
3375 #ifdef PERL_STRICT_CR
3376 for (t = s; SPACE_OR_TAB(*t); t++) ;
3378 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3380 if (*t == '\n' || *t == '#') {
3398 if (PL_expect != XOPERATOR) {
3399 if (s[1] != '<' && !strchr(s,'>'))
3402 s = scan_heredoc(s);
3404 s = scan_inputsymbol(s);
3405 TERM(sublex_start());
3410 SHop(OP_LEFT_SHIFT);
3424 SHop(OP_RIGHT_SHIFT);
3433 if (PL_expect == XOPERATOR) {
3434 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3437 return ','; /* grandfather non-comma-format format */
3441 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3442 PL_tokenbuf[0] = '@';
3443 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3444 sizeof PL_tokenbuf - 1, FALSE);
3445 if (PL_expect == XOPERATOR)
3446 no_op("Array length", s);
3447 if (!PL_tokenbuf[1])
3449 PL_expect = XOPERATOR;
3450 PL_pending_ident = '#';
3454 PL_tokenbuf[0] = '$';
3455 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3456 sizeof PL_tokenbuf - 1, FALSE);
3457 if (PL_expect == XOPERATOR)
3459 if (!PL_tokenbuf[1]) {
3461 yyerror("Final $ should be \\$ or $name");
3465 /* This kludge not intended to be bulletproof. */
3466 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3467 yylval.opval = newSVOP(OP_CONST, 0,
3468 newSViv(PL_compiling.cop_arybase));
3469 yylval.opval->op_private = OPpCONST_ARYBASE;
3475 if (PL_lex_state == LEX_NORMAL)
3478 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3481 PL_tokenbuf[0] = '@';
3482 if (ckWARN(WARN_SYNTAX)) {
3484 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3487 PL_bufptr = skipspace(PL_bufptr);
3488 while (t < PL_bufend && *t != ']')
3490 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3491 "Multidimensional syntax %.*s not supported",
3492 (t - PL_bufptr) + 1, PL_bufptr);
3496 else if (*s == '{') {
3497 PL_tokenbuf[0] = '%';
3498 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3499 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3501 char tmpbuf[sizeof PL_tokenbuf];
3503 for (t++; isSPACE(*t); t++) ;
3504 if (isIDFIRST_lazy_if(t,UTF)) {
3505 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3506 for (; isSPACE(*t); t++) ;
3507 if (*t == ';' && get_cv(tmpbuf, FALSE))
3508 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3509 "You need to quote \"%s\"", tmpbuf);
3515 PL_expect = XOPERATOR;
3516 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3517 bool islop = (PL_last_lop == PL_oldoldbufptr);
3518 if (!islop || PL_last_lop_op == OP_GREPSTART)
3519 PL_expect = XOPERATOR;
3520 else if (strchr("$@\"'`q", *s))
3521 PL_expect = XTERM; /* e.g. print $fh "foo" */
3522 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3523 PL_expect = XTERM; /* e.g. print $fh &sub */
3524 else if (isIDFIRST_lazy_if(s,UTF)) {
3525 char tmpbuf[sizeof PL_tokenbuf];
3526 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3527 if ((tmp = keyword(tmpbuf, len))) {
3528 /* binary operators exclude handle interpretations */
3540 PL_expect = XTERM; /* e.g. print $fh length() */
3545 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3546 if (gv && GvCVu(gv))
3547 PL_expect = XTERM; /* e.g. print $fh subr() */
3550 else if (isDIGIT(*s))
3551 PL_expect = XTERM; /* e.g. print $fh 3 */
3552 else if (*s == '.' && isDIGIT(s[1]))
3553 PL_expect = XTERM; /* e.g. print $fh .3 */
3554 else if (strchr("?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3555 PL_expect = XTERM; /* e.g. print $fh -1 */
3556 else if (*s == '/' && !isSPACE(s[1]) && s[1] != '=' && s[1] != '/')
3557 PL_expect = XTERM; /* e.g. print $fh /.../
3558 XXX except DORDOR operator */
3559 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3560 PL_expect = XTERM; /* print $fh <<"EOF" */
3562 PL_pending_ident = '$';
3566 if (PL_expect == XOPERATOR)
3568 PL_tokenbuf[0] = '@';
3569 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3570 if (!PL_tokenbuf[1]) {
3572 yyerror("Final @ should be \\@ or @name");
3575 if (PL_lex_state == LEX_NORMAL)
3577 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3579 PL_tokenbuf[0] = '%';
3581 /* Warn about @ where they meant $. */
3582 if (ckWARN(WARN_SYNTAX)) {
3583 if (*s == '[' || *s == '{') {
3585 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3587 if (*t == '}' || *t == ']') {
3589 PL_bufptr = skipspace(PL_bufptr);
3590 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3591 "Scalar value %.*s better written as $%.*s",
3592 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3597 PL_pending_ident = '@';
3600 case '/': /* may be division, defined-or, or pattern */
3601 if (PL_expect == XTERMORDORDOR && s[1] == '/') {
3605 case '?': /* may either be conditional or pattern */
3606 if(PL_expect == XOPERATOR) {
3614 /* A // operator. */
3624 /* Disable warning on "study /blah/" */
3625 if (PL_oldoldbufptr == PL_last_uni
3626 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3627 || memNE(PL_last_uni, "study", 5)
3628 || isALNUM_lazy_if(PL_last_uni+5,UTF)
3631 s = scan_pat(s,OP_MATCH);
3632 TERM(sublex_start());
3636 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3637 #ifdef PERL_STRICT_CR
3640 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3642 && (s == PL_linestart || s[-1] == '\n') )
3644 PL_lex_formbrack = 0;
3648 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3654 yylval.ival = OPf_SPECIAL;
3660 if (PL_expect != XOPERATOR)
3665 case '0': case '1': case '2': case '3': case '4':
3666 case '5': case '6': case '7': case '8': case '9':
3667 s = scan_num(s, &yylval);
3668 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3669 "### Saw number in '%s'\n", s);
3671 if (PL_expect == XOPERATOR)
3676 s = scan_str(s,FALSE,FALSE);
3677 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3678 "### Saw string before '%s'\n", s);
3680 if (PL_expect == XOPERATOR) {
3681 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3684 return ','; /* grandfather non-comma-format format */
3690 missingterm((char*)0);
3691 yylval.ival = OP_CONST;
3692 TERM(sublex_start());
3695 s = scan_str(s,FALSE,FALSE);
3696 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3697 "### Saw string before '%s'\n", s);
3699 if (PL_expect == XOPERATOR) {
3700 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3703 return ','; /* grandfather non-comma-format format */
3709 missingterm((char*)0);
3710 yylval.ival = OP_CONST;
3711 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3712 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
3713 yylval.ival = OP_STRINGIFY;
3717 TERM(sublex_start());
3720 s = scan_str(s,FALSE,FALSE);
3721 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3722 "### Saw backtick string before '%s'\n", s);
3724 if (PL_expect == XOPERATOR)
3725 no_op("Backticks",s);
3727 missingterm((char*)0);
3728 yylval.ival = OP_BACKTICK;
3730 TERM(sublex_start());
3734 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3735 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),"Can't use \\%c to mean $%c in expression",
3737 if (PL_expect == XOPERATOR)
3738 no_op("Backslash",s);
3742 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3746 while (isDIGIT(*start) || *start == '_')
3748 if (*start == '.' && isDIGIT(start[1])) {
3749 s = scan_num(s, &yylval);
3752 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3753 else if (!isALPHA(*start) && (PL_expect == XTERM
3754 || PL_expect == XREF || PL_expect == XSTATE
3755 || PL_expect == XTERMORDORDOR)) {
3759 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3762 s = scan_num(s, &yylval);
3769 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3809 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3811 /* Some keywords can be followed by any delimiter, including ':' */
3812 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3813 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3814 (PL_tokenbuf[0] == 'q' &&
3815 strchr("qwxr", PL_tokenbuf[1])))));
3817 /* x::* is just a word, unless x is "CORE" */
3818 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3822 while (d < PL_bufend && isSPACE(*d))
3823 d++; /* no comments skipped here, or s### is misparsed */
3825 /* Is this a label? */
3826 if (!tmp && PL_expect == XSTATE
3827 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3829 yylval.pval = savepv(PL_tokenbuf);
3834 /* Check for keywords */
3835 tmp = keyword(PL_tokenbuf, len);
3837 /* Is this a word before a => operator? */
3838 if (*d == '=' && d[1] == '>') {
3840 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3841 yylval.opval->op_private = OPpCONST_BARE;
3842 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
3843 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3847 if (tmp < 0) { /* second-class keyword? */
3848 GV *ogv = Nullgv; /* override (winner) */
3849 GV *hgv = Nullgv; /* hidden (loser) */
3850 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3852 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3855 if (GvIMPORTED_CV(gv))
3857 else if (! CvMETHOD(cv))
3861 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3862 (gv = *gvp) != (GV*)&PL_sv_undef &&
3863 GvCVu(gv) && GvIMPORTED_CV(gv))
3870 tmp = 0; /* overridden by import or by GLOBAL */
3873 && -tmp==KEY_lock /* XXX generalizable kludge */
3875 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3877 tmp = 0; /* any sub overrides "weak" keyword */
3879 else { /* no override */
3881 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
3882 Perl_warner(aTHX_ packWARN(WARN_MISC),
3883 "dump() better written as CORE::dump()");
3887 if (ckWARN(WARN_AMBIGUOUS) && hgv
3888 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3889 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
3890 "Ambiguous call resolved as CORE::%s(), %s",
3891 GvENAME(hgv), "qualify as such or use &");
3898 default: /* not a keyword */
3902 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3904 /* Get the rest if it looks like a package qualifier */
3906 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3908 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3911 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3912 *s == '\'' ? "'" : "::");
3917 if (PL_expect == XOPERATOR) {
3918 if (PL_bufptr == PL_linestart) {
3919 CopLINE_dec(PL_curcop);
3920 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
3921 CopLINE_inc(PL_curcop);
3924 no_op("Bareword",s);
3927 /* Look for a subroutine with this name in current package,
3928 unless name is "Foo::", in which case Foo is a bearword
3929 (and a package name). */
3932 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3934 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3935 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
3936 "Bareword \"%s\" refers to nonexistent package",
3939 PL_tokenbuf[len] = '\0';
3946 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3949 /* if we saw a global override before, get the right name */
3952 sv = newSVpvn("CORE::GLOBAL::",14);
3953 sv_catpv(sv,PL_tokenbuf);
3956 sv = newSVpv(PL_tokenbuf,0);
3958 /* Presume this is going to be a bareword of some sort. */
3961 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3962 yylval.opval->op_private = OPpCONST_BARE;
3963 /* UTF-8 package name? */
3964 if (UTF && !IN_BYTES &&
3965 is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
3968 /* And if "Foo::", then that's what it certainly is. */
3973 /* See if it's the indirect object for a list operator. */
3975 if (PL_oldoldbufptr &&
3976 PL_oldoldbufptr < PL_bufptr &&
3977 (PL_oldoldbufptr == PL_last_lop
3978 || PL_oldoldbufptr == PL_last_uni) &&
3979 /* NO SKIPSPACE BEFORE HERE! */
3980 (PL_expect == XREF ||
3981 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3983 bool immediate_paren = *s == '(';
3985 /* (Now we can afford to cross potential line boundary.) */
3988 /* Two barewords in a row may indicate method call. */
3990 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3993 /* If not a declared subroutine, it's an indirect object. */
3994 /* (But it's an indir obj regardless for sort.) */
3996 if ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
3997 ((!gv || !GvCVu(gv)) &&
3998 (PL_last_lop_op != OP_MAPSTART &&
3999 PL_last_lop_op != OP_GREPSTART))))
4001 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
4006 PL_expect = XOPERATOR;
4009 /* Is this a word before a => operator? */
4010 if (*s == '=' && s[1] == '>' && !pkgname) {
4012 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
4013 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
4014 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
4018 /* If followed by a paren, it's certainly a subroutine. */
4021 if (gv && GvCVu(gv)) {
4022 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
4023 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
4028 PL_nextval[PL_nexttoke].opval = yylval.opval;
4029 PL_expect = XOPERATOR;
4035 /* If followed by var or block, call it a method (unless sub) */
4037 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
4038 PL_last_lop = PL_oldbufptr;
4039 PL_last_lop_op = OP_METHOD;
4043 /* If followed by a bareword, see if it looks like indir obj. */
4046 && (isIDFIRST_lazy_if(s,UTF) || *s == '$')
4047 && (tmp = intuit_method(s,gv)))
4050 /* Not a method, so call it a subroutine (if defined) */
4052 if (gv && GvCVu(gv)) {
4054 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
4055 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4056 "Ambiguous use of -%s resolved as -&%s()",
4057 PL_tokenbuf, PL_tokenbuf);
4058 /* Check for a constant sub */
4060 if ((sv = cv_const_sv(cv))) {
4062 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
4063 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
4064 yylval.opval->op_private = 0;
4068 /* Resolve to GV now. */
4069 op_free(yylval.opval);
4070 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
4071 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
4072 PL_last_lop = PL_oldbufptr;
4073 PL_last_lop_op = OP_ENTERSUB;
4074 /* Is there a prototype? */
4077 char *proto = SvPV((SV*)cv, len);
4080 if (strEQ(proto, "$"))
4082 if (*proto == '&' && *s == '{') {
4083 sv_setpv(PL_subname, PL_curstash ?
4084 "__ANON__" : "__ANON__::__ANON__");
4088 PL_nextval[PL_nexttoke].opval = yylval.opval;
4094 /* Call it a bare word */
4096 if (PL_hints & HINT_STRICT_SUBS)
4097 yylval.opval->op_private |= OPpCONST_STRICT;
4100 if (ckWARN(WARN_RESERVED)) {
4101 if (lastchar != '-') {
4102 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4103 if (!*d && !gv_stashpv(PL_tokenbuf,FALSE))
4104 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
4111 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
4112 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4113 "Operator or semicolon missing before %c%s",
4114 lastchar, PL_tokenbuf);
4115 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4116 "Ambiguous use of %c resolved as operator %c",
4117 lastchar, lastchar);
4123 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4124 newSVpv(CopFILE(PL_curcop),0));
4128 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4129 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4132 case KEY___PACKAGE__:
4133 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4135 ? newSVsv(PL_curstname)
4144 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4145 char *pname = "main";
4146 if (PL_tokenbuf[2] == 'D')
4147 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
4148 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4151 GvIOp(gv) = newIO();
4152 IoIFP(GvIOp(gv)) = PL_rsfp;
4153 #if defined(HAS_FCNTL) && defined(F_SETFD)
4155 int fd = PerlIO_fileno(PL_rsfp);
4156 fcntl(fd,F_SETFD,fd >= 3);
4159 /* Mark this internal pseudo-handle as clean */
4160 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4162 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4163 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4164 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4166 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4167 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4168 /* if the script was opened in binmode, we need to revert
4169 * it to text mode for compatibility; but only iff it has CRs
4170 * XXX this is a questionable hack at best. */
4171 if (PL_bufend-PL_bufptr > 2
4172 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4175 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4176 loc = PerlIO_tell(PL_rsfp);
4177 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4180 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
4182 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4183 #endif /* NETWARE */
4184 #ifdef PERLIO_IS_STDIO /* really? */
4185 # if defined(__BORLANDC__)
4186 /* XXX see note in do_binmode() */
4187 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
4191 PerlIO_seek(PL_rsfp, loc, 0);
4195 #ifdef PERLIO_LAYERS
4196 if (UTF && !IN_BYTES)
4197 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
4210 if (PL_expect == XSTATE) {
4217 if (*s == ':' && s[1] == ':') {
4220 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4221 if (!(tmp = keyword(PL_tokenbuf, len)))
4222 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4236 LOP(OP_ACCEPT,XTERM);
4242 LOP(OP_ATAN2,XTERM);
4248 LOP(OP_BINMODE,XTERM);
4251 LOP(OP_BLESS,XTERM);
4260 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4277 if (!PL_cryptseen) {
4278 PL_cryptseen = TRUE;
4282 LOP(OP_CRYPT,XTERM);
4285 LOP(OP_CHMOD,XTERM);
4288 LOP(OP_CHOWN,XTERM);
4291 LOP(OP_CONNECT,XTERM);
4307 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4311 PL_hints |= HINT_BLOCK_SCOPE;
4321 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4322 LOP(OP_DBMOPEN,XTERM);
4328 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4335 yylval.ival = CopLINE(PL_curcop);
4349 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4350 UNIBRACK(OP_ENTEREVAL);
4368 case KEY_endhostent:
4374 case KEY_endservent:
4377 case KEY_endprotoent:
4388 yylval.ival = CopLINE(PL_curcop);
4390 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4392 if ((PL_bufend - p) >= 3 &&
4393 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4395 else if ((PL_bufend - p) >= 4 &&
4396 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4399 if (isIDFIRST_lazy_if(p,UTF)) {
4400 p = scan_ident(p, PL_bufend,
4401 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4405 Perl_croak(aTHX_ "Missing $ on loop variable");
4410 LOP(OP_FORMLINE,XTERM);
4416 LOP(OP_FCNTL,XTERM);
4422 LOP(OP_FLOCK,XTERM);
4431 LOP(OP_GREPSTART, XREF);
4434 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4449 case KEY_getpriority:
4450 LOP(OP_GETPRIORITY,XTERM);
4452 case KEY_getprotobyname:
4455 case KEY_getprotobynumber:
4456 LOP(OP_GPBYNUMBER,XTERM);
4458 case KEY_getprotoent:
4470 case KEY_getpeername:
4471 UNI(OP_GETPEERNAME);
4473 case KEY_gethostbyname:
4476 case KEY_gethostbyaddr:
4477 LOP(OP_GHBYADDR,XTERM);
4479 case KEY_gethostent:
4482 case KEY_getnetbyname:
4485 case KEY_getnetbyaddr:
4486 LOP(OP_GNBYADDR,XTERM);
4491 case KEY_getservbyname:
4492 LOP(OP_GSBYNAME,XTERM);
4494 case KEY_getservbyport:
4495 LOP(OP_GSBYPORT,XTERM);
4497 case KEY_getservent:
4500 case KEY_getsockname:
4501 UNI(OP_GETSOCKNAME);
4503 case KEY_getsockopt:
4504 LOP(OP_GSOCKOPT,XTERM);
4526 yylval.ival = CopLINE(PL_curcop);
4530 LOP(OP_INDEX,XTERM);
4536 LOP(OP_IOCTL,XTERM);
4548 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4580 LOP(OP_LISTEN,XTERM);
4589 s = scan_pat(s,OP_MATCH);
4590 TERM(sublex_start());
4593 LOP(OP_MAPSTART, XREF);
4596 LOP(OP_MKDIR,XTERM);
4599 LOP(OP_MSGCTL,XTERM);
4602 LOP(OP_MSGGET,XTERM);
4605 LOP(OP_MSGRCV,XTERM);
4608 LOP(OP_MSGSND,XTERM);
4614 if (isIDFIRST_lazy_if(s,UTF)) {
4615 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4616 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4618 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4619 if (!PL_in_my_stash) {
4622 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4630 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4637 if (PL_expect != XSTATE)
4638 yyerror("\"no\" not allowed in expression");
4639 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4640 s = force_version(s, FALSE);
4645 if (*s == '(' || (s = skipspace(s), *s == '('))
4652 if (isIDFIRST_lazy_if(s,UTF)) {
4654 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4656 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)
4658 && !(t[0] == '=' && t[1] == '>')
4660 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
4661 "Precedence problem: open %.*s should be open(%.*s)",
4662 d - s, s, d - s, s);
4668 yylval.ival = OP_OR;
4678 LOP(OP_OPEN_DIR,XTERM);
4681 checkcomma(s,PL_tokenbuf,"filehandle");
4685 checkcomma(s,PL_tokenbuf,"filehandle");
4704 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4708 LOP(OP_PIPE_OP,XTERM);
4711 s = scan_str(s,FALSE,FALSE);
4713 missingterm((char*)0);
4714 yylval.ival = OP_CONST;
4715 TERM(sublex_start());
4721 s = scan_str(s,FALSE,FALSE);
4723 missingterm((char*)0);
4725 if (SvCUR(PL_lex_stuff)) {
4728 d = SvPV_force(PL_lex_stuff, len);
4731 for (; isSPACE(*d) && len; --len, ++d) ;
4734 if (!warned && ckWARN(WARN_QW)) {
4735 for (; !isSPACE(*d) && len; --len, ++d) {
4737 Perl_warner(aTHX_ packWARN(WARN_QW),
4738 "Possible attempt to separate words with commas");
4741 else if (*d == '#') {
4742 Perl_warner(aTHX_ packWARN(WARN_QW),
4743 "Possible attempt to put comments in qw() list");
4749 for (; !isSPACE(*d) && len; --len, ++d) ;
4751 sv = newSVpvn(b, d-b);
4752 if (DO_UTF8(PL_lex_stuff))
4754 words = append_elem(OP_LIST, words,
4755 newSVOP(OP_CONST, 0, tokeq(sv)));
4759 PL_nextval[PL_nexttoke].opval = words;
4764 SvREFCNT_dec(PL_lex_stuff);
4765 PL_lex_stuff = Nullsv;
4771 s = scan_str(s,FALSE,FALSE);
4773 missingterm((char*)0);
4774 yylval.ival = OP_STRINGIFY;
4775 if (SvIVX(PL_lex_stuff) == '\'')
4776 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4777 TERM(sublex_start());
4780 s = scan_pat(s,OP_QR);
4781 TERM(sublex_start());
4784 s = scan_str(s,FALSE,FALSE);
4786 missingterm((char*)0);
4787 yylval.ival = OP_BACKTICK;
4789 TERM(sublex_start());
4797 s = force_version(s, FALSE);
4799 else if (*s != 'v' || !isDIGIT(s[1])
4800 || (s = force_version(s, TRUE), *s == 'v'))
4802 *PL_tokenbuf = '\0';
4803 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4804 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4805 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4807 yyerror("<> should be quotes");
4815 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4819 LOP(OP_RENAME,XTERM);
4828 LOP(OP_RINDEX,XTERM);
4838 UNIDOR(OP_READLINE);
4851 LOP(OP_REVERSE,XTERM);
4854 UNIDOR(OP_READLINK);
4862 TERM(sublex_start());
4864 TOKEN(1); /* force error */
4873 LOP(OP_SELECT,XTERM);
4879 LOP(OP_SEMCTL,XTERM);
4882 LOP(OP_SEMGET,XTERM);
4885 LOP(OP_SEMOP,XTERM);
4891 LOP(OP_SETPGRP,XTERM);
4893 case KEY_setpriority:
4894 LOP(OP_SETPRIORITY,XTERM);
4896 case KEY_sethostent:
4902 case KEY_setservent:
4905 case KEY_setprotoent:
4915 LOP(OP_SEEKDIR,XTERM);
4917 case KEY_setsockopt:
4918 LOP(OP_SSOCKOPT,XTERM);
4924 LOP(OP_SHMCTL,XTERM);
4927 LOP(OP_SHMGET,XTERM);
4930 LOP(OP_SHMREAD,XTERM);
4933 LOP(OP_SHMWRITE,XTERM);
4936 LOP(OP_SHUTDOWN,XTERM);
4945 LOP(OP_SOCKET,XTERM);
4947 case KEY_socketpair:
4948 LOP(OP_SOCKPAIR,XTERM);
4951 checkcomma(s,PL_tokenbuf,"subroutine name");
4953 if (*s == ';' || *s == ')') /* probably a close */
4954 Perl_croak(aTHX_ "sort is now a reserved word");
4956 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4960 LOP(OP_SPLIT,XTERM);
4963 LOP(OP_SPRINTF,XTERM);
4966 LOP(OP_SPLICE,XTERM);
4981 LOP(OP_SUBSTR,XTERM);
4987 char tmpbuf[sizeof PL_tokenbuf];
4988 SSize_t tboffset = 0;
4989 expectation attrful;
4990 bool have_name, have_proto, bad_proto;
4995 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4996 (*s == ':' && s[1] == ':'))
4999 attrful = XATTRBLOCK;
5000 /* remember buffer pos'n for later force_word */
5001 tboffset = s - PL_oldbufptr;
5002 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
5003 if (strchr(tmpbuf, ':'))
5004 sv_setpv(PL_subname, tmpbuf);
5006 sv_setsv(PL_subname,PL_curstname);
5007 sv_catpvn(PL_subname,"::",2);
5008 sv_catpvn(PL_subname,tmpbuf,len);
5015 Perl_croak(aTHX_ "Missing name in \"my sub\"");
5016 PL_expect = XTERMBLOCK;
5017 attrful = XATTRTERM;
5018 sv_setpv(PL_subname,"?");
5022 if (key == KEY_format) {
5024 PL_lex_formbrack = PL_lex_brackets + 1;
5026 (void) force_word(PL_oldbufptr + tboffset, WORD,
5031 /* Look for a prototype */
5035 s = scan_str(s,FALSE,FALSE);
5037 Perl_croak(aTHX_ "Prototype not terminated");
5038 /* strip spaces and check for bad characters */
5039 d = SvPVX(PL_lex_stuff);
5042 for (p = d; *p; ++p) {
5045 if (!strchr("$@%*;[]&\\", *p))
5050 if (bad_proto && ckWARN(WARN_SYNTAX))
5051 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5052 "Illegal character in prototype for %s : %s",
5053 SvPVX(PL_subname), d);
5054 SvCUR(PL_lex_stuff) = tmp;
5062 if (*s == ':' && s[1] != ':')
5063 PL_expect = attrful;
5064 else if (!have_name && *s != '{' && key == KEY_sub)
5065 Perl_croak(aTHX_ "Illegal declaration of anonymous subroutine");
5068 PL_nextval[PL_nexttoke].opval =
5069 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
5070 PL_lex_stuff = Nullsv;
5074 sv_setpv(PL_subname,
5075 PL_curstash ? "__ANON__" : "__ANON__::__ANON__");
5078 (void) force_word(PL_oldbufptr + tboffset, WORD,
5087 LOP(OP_SYSTEM,XREF);
5090 LOP(OP_SYMLINK,XTERM);
5093 LOP(OP_SYSCALL,XTERM);
5096 LOP(OP_SYSOPEN,XTERM);
5099 LOP(OP_SYSSEEK,XTERM);
5102 LOP(OP_SYSREAD,XTERM);
5105 LOP(OP_SYSWRITE,XTERM);
5109 TERM(sublex_start());
5130 LOP(OP_TRUNCATE,XTERM);
5142 yylval.ival = CopLINE(PL_curcop);
5146 yylval.ival = CopLINE(PL_curcop);
5150 LOP(OP_UNLINK,XTERM);
5156 LOP(OP_UNPACK,XTERM);
5159 LOP(OP_UTIME,XTERM);
5165 LOP(OP_UNSHIFT,XTERM);
5168 if (PL_expect != XSTATE)
5169 yyerror("\"use\" not allowed in expression");
5171 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
5172 s = force_version(s, TRUE);
5173 if (*s == ';' || (s = skipspace(s), *s == ';')) {
5174 PL_nextval[PL_nexttoke].opval = Nullop;
5177 else if (*s == 'v') {
5178 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5179 s = force_version(s, FALSE);
5183 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5184 s = force_version(s, FALSE);
5196 yylval.ival = CopLINE(PL_curcop);
5200 PL_hints |= HINT_BLOCK_SCOPE;
5207 LOP(OP_WAITPID,XTERM);
5216 ctl_l[0] = toCTRL('L');
5218 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5221 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5226 if (PL_expect == XOPERATOR)
5232 yylval.ival = OP_XOR;
5237 TERM(sublex_start());
5242 #pragma segment Main
5246 S_pending_ident(pTHX)
5250 /* pit holds the identifier we read and pending_ident is reset */
5251 char pit = PL_pending_ident;
5252 PL_pending_ident = 0;
5254 DEBUG_T({ PerlIO_printf(Perl_debug_log,
5255 "### Tokener saw identifier '%s'\n", PL_tokenbuf); });
5257 /* if we're in a my(), we can't allow dynamics here.
5258 $foo'bar has already been turned into $foo::bar, so
5259 just check for colons.
5261 if it's a legal name, the OP is a PADANY.
5264 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
5265 if (strchr(PL_tokenbuf,':'))
5266 yyerror(Perl_form(aTHX_ "No package name allowed for "
5267 "variable %s in \"our\"",
5269 tmp = allocmy(PL_tokenbuf);
5272 if (strchr(PL_tokenbuf,':'))
5273 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
5275 yylval.opval = newOP(OP_PADANY, 0);
5276 yylval.opval->op_targ = allocmy(PL_tokenbuf);
5282 build the ops for accesses to a my() variable.
5284 Deny my($a) or my($b) in a sort block, *if* $a or $b is
5285 then used in a comparison. This catches most, but not
5286 all cases. For instance, it catches
5287 sort { my($a); $a <=> $b }
5289 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
5290 (although why you'd do that is anyone's guess).
5293 if (!strchr(PL_tokenbuf,':')) {
5294 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
5295 /* might be an "our" variable" */
5296 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
5297 /* build ops for a bareword */
5298 SV *sym = newSVpv(HvNAME(PAD_COMPNAME_OURSTASH(tmp)), 0);
5299 sv_catpvn(sym, "::", 2);
5300 sv_catpv(sym, PL_tokenbuf+1);
5301 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
5302 yylval.opval->op_private = OPpCONST_ENTERED;
5303 gv_fetchpv(SvPVX(sym),
5305 ? (GV_ADDMULTI | GV_ADDINEVAL)
5308 ((PL_tokenbuf[0] == '$') ? SVt_PV
5309 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5314 /* if it's a sort block and they're naming $a or $b */
5315 if (PL_last_lop_op == OP_SORT &&
5316 PL_tokenbuf[0] == '$' &&
5317 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
5320 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
5321 d < PL_bufend && *d != '\n';
5324 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
5325 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
5331 yylval.opval = newOP(OP_PADANY, 0);
5332 yylval.opval->op_targ = tmp;
5338 Whine if they've said @foo in a doublequoted string,
5339 and @foo isn't a variable we can find in the symbol
5342 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
5343 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
5344 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
5345 && ckWARN(WARN_AMBIGUOUS))
5347 /* Downgraded from fatal to warning 20000522 mjd */
5348 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5349 "Possible unintended interpolation of %s in string",
5354 /* build ops for a bareword */
5355 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
5356 yylval.opval->op_private = OPpCONST_ENTERED;
5357 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
5358 ((PL_tokenbuf[0] == '$') ? SVt_PV
5359 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5365 Perl_keyword(pTHX_ register char *d, I32 len)
5370 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5371 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5372 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5373 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5374 if (strEQ(d,"__END__")) return KEY___END__;
5378 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5383 if (strEQ(d,"and")) return -KEY_and;
5384 if (strEQ(d,"abs")) return -KEY_abs;
5387 if (strEQ(d,"alarm")) return -KEY_alarm;
5388 if (strEQ(d,"atan2")) return -KEY_atan2;
5391 if (strEQ(d,"accept")) return -KEY_accept;
5396 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5399 if (strEQ(d,"bless")) return -KEY_bless;
5400 if (strEQ(d,"bind")) return -KEY_bind;
5401 if (strEQ(d,"binmode")) return -KEY_binmode;
5404 if (strEQ(d,"CORE")) return -KEY_CORE;
5405 if (strEQ(d,"CHECK")) return KEY_CHECK;
5410 if (strEQ(d,"cmp")) return -KEY_cmp;
5411 if (strEQ(d,"chr")) return -KEY_chr;
5412 if (strEQ(d,"cos")) return -KEY_cos;
5415 if (strEQ(d,"chop")) return -KEY_chop;
5418 if (strEQ(d,"close")) return -KEY_close;
5419 if (strEQ(d,"chdir")) return -KEY_chdir;
5420 if (strEQ(d,"chomp")) return -KEY_chomp;
5421 if (strEQ(d,"chmod")) return -KEY_chmod;
5422 if (strEQ(d,"chown")) return -KEY_chown;
5423 if (strEQ(d,"crypt")) return -KEY_crypt;
5426 if (strEQ(d,"chroot")) return -KEY_chroot;
5427 if (strEQ(d,"caller")) return -KEY_caller;
5430 if (strEQ(d,"connect")) return -KEY_connect;
5433 if (strEQ(d,"closedir")) return -KEY_closedir;
5434 if (strEQ(d,"continue")) return -KEY_continue;
5439 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5444 if (strEQ(d,"do")) return KEY_do;
5447 if (strEQ(d,"die")) return -KEY_die;
5450 if (strEQ(d,"dump")) return -KEY_dump;
5453 if (strEQ(d,"delete")) return KEY_delete;
5456 if (strEQ(d,"defined")) return KEY_defined;
5457 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5460 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5465 if (strEQ(d,"END")) return KEY_END;
5470 if (strEQ(d,"eq")) return -KEY_eq;
5473 if (strEQ(d,"eof")) return -KEY_eof;
5474 if (strEQ(d,"err")) return -KEY_err;
5475 if (strEQ(d,"exp")) return -KEY_exp;
5478 if (strEQ(d,"else")) return KEY_else;
5479 if (strEQ(d,"exit")) return -KEY_exit;
5480 if (strEQ(d,"eval")) return KEY_eval;
5481 if (strEQ(d,"exec")) return -KEY_exec;
5482 if (strEQ(d,"each")) return -KEY_each;
5485 if (strEQ(d,"elsif")) return KEY_elsif;
5488 if (strEQ(d,"exists")) return KEY_exists;
5489 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5492 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5493 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5496 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5499 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5500 if (strEQ(d,"endservent")) return -KEY_endservent;
5503 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5510 if (strEQ(d,"for")) return KEY_for;
5513 if (strEQ(d,"fork")) return -KEY_fork;
5516 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5517 if (strEQ(d,"flock")) return -KEY_flock;
5520 if (strEQ(d,"format")) return KEY_format;
5521 if (strEQ(d,"fileno")) return -KEY_fileno;
5524 if (strEQ(d,"foreach")) return KEY_foreach;
5527 if (strEQ(d,"formline")) return -KEY_formline;
5532 if (strnEQ(d,"get",3)) {
5537 if (strEQ(d,"ppid")) return -KEY_getppid;
5538 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5541 if (strEQ(d,"pwent")) return -KEY_getpwent;
5542 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5543 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5546 if (strEQ(d,"peername")) return -KEY_getpeername;
5547 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5548 if (strEQ(d,"priority")) return -KEY_getpriority;
5551 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5554 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5558 else if (*d == 'h') {
5559 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5560 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5561 if (strEQ(d,"hostent")) return -KEY_gethostent;
5563 else if (*d == 'n') {
5564 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5565 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5566 if (strEQ(d,"netent")) return -KEY_getnetent;
5568 else if (*d == 's') {
5569 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5570 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5571 if (strEQ(d,"servent")) return -KEY_getservent;
5572 if (strEQ(d,"sockname")) return -KEY_getsockname;
5573 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5575 else if (*d == 'g') {
5576 if (strEQ(d,"grent")) return -KEY_getgrent;
5577 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5578 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5580 else if (*d == 'l') {
5581 if (strEQ(d,"login")) return -KEY_getlogin;
5583 else if (strEQ(d,"c")) return -KEY_getc;
5588 if (strEQ(d,"gt")) return -KEY_gt;
5589 if (strEQ(d,"ge")) return -KEY_ge;
5592 if (strEQ(d,"grep")) return KEY_grep;
5593 if (strEQ(d,"goto")) return KEY_goto;
5594 if (strEQ(d,"glob")) return KEY_glob;
5597 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5602 if (strEQ(d,"hex")) return -KEY_hex;
5605 if (strEQ(d,"INIT")) return KEY_INIT;
5610 if (strEQ(d,"if")) return KEY_if;
5613 if (strEQ(d,"int")) return -KEY_int;
5616 if (strEQ(d,"index")) return -KEY_index;
5617 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5622 if (strEQ(d,"join")) return -KEY_join;
5626 if (strEQ(d,"keys")) return -KEY_keys;
5627 if (strEQ(d,"kill")) return -KEY_kill;
5633 if (strEQ(d,"lt")) return -KEY_lt;
5634 if (strEQ(d,"le")) return -KEY_le;
5635 if (strEQ(d,"lc")) return -KEY_lc;
5638 if (strEQ(d,"log")) return -KEY_log;
5641 if (strEQ(d,"last")) return KEY_last;
5642 if (strEQ(d,"link")) return -KEY_link;
5643 if (strEQ(d,"lock")) return -KEY_lock;
5646 if (strEQ(d,"local")) return KEY_local;
5647 if (strEQ(d,"lstat")) return -KEY_lstat;
5650 if (strEQ(d,"length")) return -KEY_length;
5651 if (strEQ(d,"listen")) return -KEY_listen;
5654 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5657 if (strEQ(d,"localtime")) return -KEY_localtime;
5663 case 1: return KEY_m;
5665 if (strEQ(d,"my")) return KEY_my;
5668 if (strEQ(d,"map")) return KEY_map;
5671 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5674 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5675 if (strEQ(d,"msgget")) return -KEY_msgget;
5676 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5677 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5682 if (strEQ(d,"next")) return KEY_next;
5683 if (strEQ(d,"ne")) return -KEY_ne;
5684 if (strEQ(d,"not")) return -KEY_not;
5685 if (strEQ(d,"no")) return KEY_no;
5690 if (strEQ(d,"or")) return -KEY_or;
5693 if (strEQ(d,"ord")) return -KEY_ord;
5694 if (strEQ(d,"oct")) return -KEY_oct;
5695 if (strEQ(d,"our")) return KEY_our;
5698 if (strEQ(d,"open")) return -KEY_open;
5701 if (strEQ(d,"opendir")) return -KEY_opendir;
5708 if (strEQ(d,"pop")) return -KEY_pop;
5709 if (strEQ(d,"pos")) return KEY_pos;
5712 if (strEQ(d,"push")) return -KEY_push;
5713 if (strEQ(d,"pack")) return -KEY_pack;
5714 if (strEQ(d,"pipe")) return -KEY_pipe;
5717 if (strEQ(d,"print")) return KEY_print;
5720 if (strEQ(d,"printf")) return KEY_printf;
5723 if (strEQ(d,"package")) return KEY_package;
5726 if (strEQ(d,"prototype")) return KEY_prototype;
5731 if (strEQ(d,"q")) return KEY_q;
5732 if (strEQ(d,"qr")) return KEY_qr;
5733 if (strEQ(d,"qq")) return KEY_qq;
5734 if (strEQ(d,"qw")) return KEY_qw;
5735 if (strEQ(d,"qx")) return KEY_qx;
5737 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5742 if (strEQ(d,"ref")) return -KEY_ref;
5745 if (strEQ(d,"read")) return -KEY_read;
5746 if (strEQ(d,"rand")) return -KEY_rand;
5747 if (strEQ(d,"recv")) return -KEY_recv;
5748 if (strEQ(d,"redo")) return KEY_redo;
5751 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5752 if (strEQ(d,"reset")) return -KEY_reset;
5755 if (strEQ(d,"return")) return KEY_return;
5756 if (strEQ(d,"rename")) return -KEY_rename;
5757 if (strEQ(d,"rindex")) return -KEY_rindex;
5760 if (strEQ(d,"require")) return KEY_require;
5761 if (strEQ(d,"reverse")) return -KEY_reverse;
5762 if (strEQ(d,"readdir")) return -KEY_readdir;
5765 if (strEQ(d,"readlink")) return -KEY_readlink;
5766 if (strEQ(d,"readline")) return -KEY_readline;
5767 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5770 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5776 case 0: return KEY_s;
5778 if (strEQ(d,"scalar")) return KEY_scalar;
5783 if (strEQ(d,"seek")) return -KEY_seek;
5784 if (strEQ(d,"send")) return -KEY_send;
5787 if (strEQ(d,"semop")) return -KEY_semop;
5790 if (strEQ(d,"select")) return -KEY_select;
5791 if (strEQ(d,"semctl")) return -KEY_semctl;
5792 if (strEQ(d,"semget")) return -KEY_semget;
5795 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5796 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5799 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5800 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5803 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5806 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5807 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5808 if (strEQ(d,"setservent")) return -KEY_setservent;
5811 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5812 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5819 if (strEQ(d,"shift")) return -KEY_shift;
5822 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5823 if (strEQ(d,"shmget")) return -KEY_shmget;
5826 if (strEQ(d,"shmread")) return -KEY_shmread;
5829 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5830 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5835 if (strEQ(d,"sin")) return -KEY_sin;
5838 if (strEQ(d,"sleep")) return -KEY_sleep;
5841 if (strEQ(d,"sort")) return KEY_sort;
5842 if (strEQ(d,"socket")) return -KEY_socket;
5843 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5846 if (strEQ(d,"split")) return KEY_split;
5847 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5848 if (strEQ(d,"splice")) return -KEY_splice;
5851 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5854 if (strEQ(d,"srand")) return -KEY_srand;
5857 if (strEQ(d,"stat")) return -KEY_stat;
5858 if (strEQ(d,"study")) return KEY_study;
5861 if (strEQ(d,"substr")) return -KEY_substr;
5862 if (strEQ(d,"sub")) return KEY_sub;
5867 if (strEQ(d,"system")) return -KEY_system;
5870 if (strEQ(d,"symlink")) return -KEY_symlink;
5871 if (strEQ(d,"syscall")) return -KEY_syscall;
5872 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5873 if (strEQ(d,"sysread")) return -KEY_sysread;
5874 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5877 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5886 if (strEQ(d,"tr")) return KEY_tr;
5889 if (strEQ(d,"tie")) return KEY_tie;
5892 if (strEQ(d,"tell")) return -KEY_tell;
5893 if (strEQ(d,"tied")) return KEY_tied;
5894 if (strEQ(d,"time")) return -KEY_time;
5897 if (strEQ(d,"times")) return -KEY_times;
5900 if (strEQ(d,"telldir")) return -KEY_telldir;
5903 if (strEQ(d,"truncate")) return -KEY_truncate;
5910 if (strEQ(d,"uc")) return -KEY_uc;
5913 if (strEQ(d,"use")) return KEY_use;
5916 if (strEQ(d,"undef")) return KEY_undef;
5917 if (strEQ(d,"until")) return KEY_until;
5918 if (strEQ(d,"untie")) return KEY_untie;
5919 if (strEQ(d,"utime")) return -KEY_utime;
5920 if (strEQ(d,"umask")) return -KEY_umask;
5923 if (strEQ(d,"unless")) return KEY_unless;
5924 if (strEQ(d,"unpack")) return -KEY_unpack;
5925 if (strEQ(d,"unlink")) return -KEY_unlink;
5928 if (strEQ(d,"unshift")) return -KEY_unshift;
5929 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5934 if (strEQ(d,"values")) return -KEY_values;
5935 if (strEQ(d,"vec")) return -KEY_vec;
5940 if (strEQ(d,"warn")) return -KEY_warn;
5941 if (strEQ(d,"wait")) return -KEY_wait;
5944 if (strEQ(d,"while")) return KEY_while;
5945 if (strEQ(d,"write")) return -KEY_write;
5948 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5951 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5956 if (len == 1) return -KEY_x;
5957 if (strEQ(d,"xor")) return -KEY_xor;
5960 if (len == 1) return KEY_y;
5969 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5973 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5974 if (ckWARN(WARN_SYNTAX)) {
5976 for (w = s+2; *w && level; w++) {
5983 for (; *w && isSPACE(*w); w++) ;
5984 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5985 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5986 "%s (...) interpreted as function",name);
5989 while (s < PL_bufend && isSPACE(*s))
5993 while (s < PL_bufend && isSPACE(*s))
5995 if (isIDFIRST_lazy_if(s,UTF)) {
5997 while (isALNUM_lazy_if(s,UTF))
5999 while (s < PL_bufend && isSPACE(*s))
6004 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
6008 Perl_croak(aTHX_ "No comma allowed after %s", what);
6013 /* Either returns sv, or mortalizes sv and returns a new SV*.
6014 Best used as sv=new_constant(..., sv, ...).
6015 If s, pv are NULL, calls subroutine with one argument,
6016 and type is used with error messages only. */
6019 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
6023 HV *table = GvHV(PL_hintgv); /* ^H */
6027 const char *why1, *why2, *why3;
6029 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
6032 why2 = strEQ(key,"charnames")
6033 ? "(possibly a missing \"use charnames ...\")"
6035 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
6036 (type ? type: "undef"), why2);
6038 /* This is convoluted and evil ("goto considered harmful")
6039 * but I do not understand the intricacies of all the different
6040 * failure modes of %^H in here. The goal here is to make
6041 * the most probable error message user-friendly. --jhi */
6046 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
6047 (type ? type: "undef"), why1, why2, why3);
6049 yyerror(SvPVX(msg));
6053 cvp = hv_fetch(table, key, strlen(key), FALSE);
6054 if (!cvp || !SvOK(*cvp)) {
6057 why3 = "} is not defined";
6060 sv_2mortal(sv); /* Parent created it permanently */
6063 pv = sv_2mortal(newSVpvn(s, len));
6065 typesv = sv_2mortal(newSVpv(type, 0));
6067 typesv = &PL_sv_undef;
6069 PUSHSTACKi(PERLSI_OVERLOAD);
6081 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
6085 /* Check the eval first */
6086 if (!PL_in_eval && SvTRUE(ERRSV)) {
6088 sv_catpv(ERRSV, "Propagated");
6089 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
6091 res = SvREFCNT_inc(sv);
6095 (void)SvREFCNT_inc(res);
6104 why1 = "Call to &{$^H{";
6106 why3 = "}} did not return a defined value";
6115 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
6117 register char *d = dest;
6118 register char *e = d + destlen - 3; /* two-character token, ending NUL */
6121 Perl_croak(aTHX_ ident_too_long);
6122 if (isALNUM(*s)) /* UTF handled below */
6124 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
6129 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
6133 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6134 char *t = s + UTF8SKIP(s);
6135 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6137 if (d + (t - s) > e)
6138 Perl_croak(aTHX_ ident_too_long);
6139 Copy(s, d, t - s, char);
6152 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
6162 e = d + destlen - 3; /* two-character token, ending NUL */
6164 while (isDIGIT(*s)) {
6166 Perl_croak(aTHX_ ident_too_long);
6173 Perl_croak(aTHX_ ident_too_long);
6174 if (isALNUM(*s)) /* UTF handled below */
6176 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
6181 else if (*s == ':' && s[1] == ':') {
6185 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6186 char *t = s + UTF8SKIP(s);
6187 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6189 if (d + (t - s) > e)
6190 Perl_croak(aTHX_ ident_too_long);
6191 Copy(s, d, t - s, char);
6202 if (PL_lex_state != LEX_NORMAL)
6203 PL_lex_state = LEX_INTERPENDMAYBE;
6206 if (*s == '$' && s[1] &&
6207 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
6220 if (*d == '^' && *s && isCONTROLVAR(*s)) {
6225 if (isSPACE(s[-1])) {
6228 if (!SPACE_OR_TAB(ch)) {
6234 if (isIDFIRST_lazy_if(d,UTF)) {
6238 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
6240 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
6243 Copy(s, d, e - s, char);
6248 while ((isALNUM(*s) || *s == ':') && d < e)
6251 Perl_croak(aTHX_ ident_too_long);
6254 while (s < send && SPACE_OR_TAB(*s)) s++;
6255 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
6256 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
6257 const char *brack = *s == '[' ? "[...]" : "{...}";
6258 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
6259 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
6260 funny, dest, brack, funny, dest, brack);
6263 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
6267 /* Handle extended ${^Foo} variables
6268 * 1999-02-27 mjd-perl-patch@plover.com */
6269 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
6273 while (isALNUM(*s) && d < e) {
6277 Perl_croak(aTHX_ ident_too_long);
6282 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
6283 PL_lex_state = LEX_INTERPEND;
6286 if (PL_lex_state == LEX_NORMAL) {
6287 if (ckWARN(WARN_AMBIGUOUS) &&
6288 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
6290 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
6291 "Ambiguous use of %c{%s} resolved to %c%s",
6292 funny, dest, funny, dest);
6297 s = bracket; /* let the parser handle it */
6301 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
6302 PL_lex_state = LEX_INTERPEND;
6307 Perl_pmflag(pTHX_ U32* pmfl, int ch)
6312 *pmfl |= PMf_GLOBAL;
6314 *pmfl |= PMf_CONTINUE;
6318 *pmfl |= PMf_MULTILINE;
6320 *pmfl |= PMf_SINGLELINE;
6322 *pmfl |= PMf_EXTENDED;
6326 S_scan_pat(pTHX_ char *start, I32 type)
6331 s = scan_str(start,FALSE,FALSE);
6333 Perl_croak(aTHX_ "Search pattern not terminated");
6335 pm = (PMOP*)newPMOP(type, 0);
6336 if (PL_multi_open == '?')
6337 pm->op_pmflags |= PMf_ONCE;
6339 while (*s && strchr("iomsx", *s))
6340 pmflag(&pm->op_pmflags,*s++);
6343 while (*s && strchr("iogcmsx", *s))
6344 pmflag(&pm->op_pmflags,*s++);
6346 /* issue a warning if /c is specified,but /g is not */
6347 if (ckWARN(WARN_REGEXP) &&
6348 (pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL))
6350 Perl_warner(aTHX_ packWARN(WARN_REGEXP), c_without_g);
6353 pm->op_pmpermflags = pm->op_pmflags;
6355 PL_lex_op = (OP*)pm;
6356 yylval.ival = OP_MATCH;
6361 S_scan_subst(pTHX_ char *start)
6368 yylval.ival = OP_NULL;
6370 s = scan_str(start,FALSE,FALSE);
6373 Perl_croak(aTHX_ "Substitution pattern not terminated");
6375 if (s[-1] == PL_multi_open)
6378 first_start = PL_multi_start;
6379 s = scan_str(s,FALSE,FALSE);
6382 SvREFCNT_dec(PL_lex_stuff);
6383 PL_lex_stuff = Nullsv;
6385 Perl_croak(aTHX_ "Substitution replacement not terminated");
6387 PL_multi_start = first_start; /* so whole substitution is taken together */
6389 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6395 else if (strchr("iogcmsx", *s))
6396 pmflag(&pm->op_pmflags,*s++);
6401 /* /c is not meaningful with s/// */
6402 if (ckWARN(WARN_REGEXP) && (pm->op_pmflags & PMf_CONTINUE))
6404 Perl_warner(aTHX_ packWARN(WARN_REGEXP), c_in_subst);
6409 PL_sublex_info.super_bufptr = s;
6410 PL_sublex_info.super_bufend = PL_bufend;
6412 pm->op_pmflags |= PMf_EVAL;
6413 repl = newSVpvn("",0);
6415 sv_catpv(repl, es ? "eval " : "do ");
6416 sv_catpvn(repl, "{ ", 2);
6417 sv_catsv(repl, PL_lex_repl);
6418 sv_catpvn(repl, " };", 2);
6420 SvREFCNT_dec(PL_lex_repl);
6424 pm->op_pmpermflags = pm->op_pmflags;
6425 PL_lex_op = (OP*)pm;
6426 yylval.ival = OP_SUBST;
6431 S_scan_trans(pTHX_ char *start)
6440 yylval.ival = OP_NULL;
6442 s = scan_str(start,FALSE,FALSE);
6444 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6445 if (s[-1] == PL_multi_open)
6448 s = scan_str(s,FALSE,FALSE);
6451 SvREFCNT_dec(PL_lex_stuff);
6452 PL_lex_stuff = Nullsv;
6454 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6457 complement = del = squash = 0;
6458 while (strchr("cds", *s)) {
6460 complement = OPpTRANS_COMPLEMENT;
6462 del = OPpTRANS_DELETE;
6464 squash = OPpTRANS_SQUASH;
6468 New(803, tbl, complement&&!del?258:256, short);
6469 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6470 o->op_private = del|squash|complement|
6471 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
6472 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
6475 yylval.ival = OP_TRANS;
6480 S_scan_heredoc(pTHX_ register char *s)
6483 I32 op_type = OP_SCALAR;
6490 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6494 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6497 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6498 if (*peek && strchr("`'\"",*peek)) {
6501 s = delimcpy(d, e, s, PL_bufend, term, &len);
6511 if (!isALNUM_lazy_if(s,UTF))
6512 deprecate_old("bare << to mean <<\"\"");
6513 for (; isALNUM_lazy_if(s,UTF); s++) {
6518 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6519 Perl_croak(aTHX_ "Delimiter for here document is too long");
6522 len = d - PL_tokenbuf;
6523 #ifndef PERL_STRICT_CR
6524 d = strchr(s, '\r');
6528 while (s < PL_bufend) {
6534 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6543 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6548 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6549 herewas = newSVpvn(s,PL_bufend-s);
6551 s--, herewas = newSVpvn(s,d-s);
6552 s += SvCUR(herewas);
6554 tmpstr = NEWSV(87,79);
6555 sv_upgrade(tmpstr, SVt_PVIV);
6560 else if (term == '`') {
6561 op_type = OP_BACKTICK;
6562 SvIVX(tmpstr) = '\\';
6566 PL_multi_start = CopLINE(PL_curcop);
6567 PL_multi_open = PL_multi_close = '<';
6568 term = *PL_tokenbuf;
6569 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6570 char *bufptr = PL_sublex_info.super_bufptr;
6571 char *bufend = PL_sublex_info.super_bufend;
6572 char *olds = s - SvCUR(herewas);
6573 s = strchr(bufptr, '\n');
6577 while (s < bufend &&
6578 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6580 CopLINE_inc(PL_curcop);
6583 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6584 missingterm(PL_tokenbuf);
6586 sv_setpvn(herewas,bufptr,d-bufptr+1);
6587 sv_setpvn(tmpstr,d+1,s-d);
6589 sv_catpvn(herewas,s,bufend-s);
6590 (void)strcpy(bufptr,SvPVX(herewas));
6597 while (s < PL_bufend &&
6598 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6600 CopLINE_inc(PL_curcop);
6602 if (s >= PL_bufend) {
6603 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6604 missingterm(PL_tokenbuf);
6606 sv_setpvn(tmpstr,d+1,s-d);
6608 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6610 sv_catpvn(herewas,s,PL_bufend-s);
6611 sv_setsv(PL_linestr,herewas);
6612 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6613 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6614 PL_last_lop = PL_last_uni = Nullch;
6617 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6618 while (s >= PL_bufend) { /* multiple line string? */
6620 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6621 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6622 missingterm(PL_tokenbuf);
6624 CopLINE_inc(PL_curcop);
6625 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6626 PL_last_lop = PL_last_uni = Nullch;
6627 #ifndef PERL_STRICT_CR
6628 if (PL_bufend - PL_linestart >= 2) {
6629 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6630 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6632 PL_bufend[-2] = '\n';
6634 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6636 else if (PL_bufend[-1] == '\r')
6637 PL_bufend[-1] = '\n';
6639 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6640 PL_bufend[-1] = '\n';
6642 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6643 SV *sv = NEWSV(88,0);
6645 sv_upgrade(sv, SVt_PVMG);
6646 sv_setsv(sv,PL_linestr);
6649 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6651 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6654 sv_catsv(PL_linestr,herewas);
6655 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6659 sv_catsv(tmpstr,PL_linestr);
6664 PL_multi_end = CopLINE(PL_curcop);
6665 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6666 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6667 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6669 SvREFCNT_dec(herewas);
6670 if (UTF && !IN_BYTES && is_utf8_string((U8*)SvPVX(tmpstr), SvCUR(tmpstr)))
6672 PL_lex_stuff = tmpstr;
6673 yylval.ival = op_type;
6678 takes: current position in input buffer
6679 returns: new position in input buffer
6680 side-effects: yylval and lex_op are set.
6685 <FH> read from filehandle
6686 <pkg::FH> read from package qualified filehandle
6687 <pkg'FH> read from package qualified filehandle
6688 <$fh> read from filehandle in $fh
6694 S_scan_inputsymbol(pTHX_ char *start)
6696 register char *s = start; /* current position in buffer */
6702 d = PL_tokenbuf; /* start of temp holding space */
6703 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6704 end = strchr(s, '\n');
6707 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6709 /* die if we didn't have space for the contents of the <>,
6710 or if it didn't end, or if we see a newline
6713 if (len >= sizeof PL_tokenbuf)
6714 Perl_croak(aTHX_ "Excessively long <> operator");
6716 Perl_croak(aTHX_ "Unterminated <> operator");
6721 Remember, only scalar variables are interpreted as filehandles by
6722 this code. Anything more complex (e.g., <$fh{$num}>) will be
6723 treated as a glob() call.
6724 This code makes use of the fact that except for the $ at the front,
6725 a scalar variable and a filehandle look the same.
6727 if (*d == '$' && d[1]) d++;
6729 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6730 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6733 /* If we've tried to read what we allow filehandles to look like, and
6734 there's still text left, then it must be a glob() and not a getline.
6735 Use scan_str to pull out the stuff between the <> and treat it
6736 as nothing more than a string.
6739 if (d - PL_tokenbuf != len) {
6740 yylval.ival = OP_GLOB;
6742 s = scan_str(start,FALSE,FALSE);
6744 Perl_croak(aTHX_ "Glob not terminated");
6748 bool readline_overriden = FALSE;
6749 GV *gv_readline = Nullgv;
6751 /* we're in a filehandle read situation */
6754 /* turn <> into <ARGV> */
6756 (void)strcpy(d,"ARGV");
6758 /* Check whether readline() is overriden */
6759 if (((gv_readline = gv_fetchpv("readline", FALSE, SVt_PVCV))
6760 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
6762 ((gvp = (GV**)hv_fetch(PL_globalstash, "readline", 8, FALSE))
6763 && (gv_readline = *gvp) != (GV*)&PL_sv_undef
6764 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
6765 readline_overriden = TRUE;
6767 /* if <$fh>, create the ops to turn the variable into a
6773 /* try to find it in the pad for this block, otherwise find
6774 add symbol table ops
6776 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6777 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
6778 SV *sym = sv_2mortal(
6779 newSVpv(HvNAME(PAD_COMPNAME_OURSTASH(tmp)),0));
6780 sv_catpvn(sym, "::", 2);
6786 OP *o = newOP(OP_PADSV, 0);
6788 PL_lex_op = readline_overriden
6789 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6790 append_elem(OP_LIST, o,
6791 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
6792 : (OP*)newUNOP(OP_READLINE, 0, o);
6801 ? (GV_ADDMULTI | GV_ADDINEVAL)
6804 PL_lex_op = readline_overriden
6805 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6806 append_elem(OP_LIST,
6807 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
6808 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
6809 : (OP*)newUNOP(OP_READLINE, 0,
6810 newUNOP(OP_RV2SV, 0,
6811 newGVOP(OP_GV, 0, gv)));
6813 if (!readline_overriden)
6814 PL_lex_op->op_flags |= OPf_SPECIAL;
6815 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6816 yylval.ival = OP_NULL;
6819 /* If it's none of the above, it must be a literal filehandle
6820 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6822 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6823 PL_lex_op = readline_overriden
6824 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6825 append_elem(OP_LIST,
6826 newGVOP(OP_GV, 0, gv),
6827 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
6828 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6829 yylval.ival = OP_NULL;
6838 takes: start position in buffer
6839 keep_quoted preserve \ on the embedded delimiter(s)
6840 keep_delims preserve the delimiters around the string
6841 returns: position to continue reading from buffer
6842 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6843 updates the read buffer.
6845 This subroutine pulls a string out of the input. It is called for:
6846 q single quotes q(literal text)
6847 ' single quotes 'literal text'
6848 qq double quotes qq(interpolate $here please)
6849 " double quotes "interpolate $here please"
6850 qx backticks qx(/bin/ls -l)
6851 ` backticks `/bin/ls -l`
6852 qw quote words @EXPORT_OK = qw( func() $spam )
6853 m// regexp match m/this/
6854 s/// regexp substitute s/this/that/
6855 tr/// string transliterate tr/this/that/
6856 y/// string transliterate y/this/that/
6857 ($*@) sub prototypes sub foo ($)
6858 (stuff) sub attr parameters sub foo : attr(stuff)
6859 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6861 In most of these cases (all but <>, patterns and transliterate)
6862 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6863 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6864 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6867 It skips whitespace before the string starts, and treats the first
6868 character as the delimiter. If the delimiter is one of ([{< then
6869 the corresponding "close" character )]}> is used as the closing
6870 delimiter. It allows quoting of delimiters, and if the string has
6871 balanced delimiters ([{<>}]) it allows nesting.
6873 On success, the SV with the resulting string is put into lex_stuff or,
6874 if that is already non-NULL, into lex_repl. The second case occurs only
6875 when parsing the RHS of the special constructs s/// and tr/// (y///).
6876 For convenience, the terminating delimiter character is stuffed into
6881 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6883 SV *sv; /* scalar value: string */
6884 char *tmps; /* temp string, used for delimiter matching */
6885 register char *s = start; /* current position in the buffer */
6886 register char term; /* terminating character */
6887 register char *to; /* current position in the sv's data */
6888 I32 brackets = 1; /* bracket nesting level */
6889 bool has_utf8 = FALSE; /* is there any utf8 content? */
6891 /* skip space before the delimiter */
6895 /* mark where we are, in case we need to report errors */
6898 /* after skipping whitespace, the next character is the terminator */
6900 if (!UTF8_IS_INVARIANT((U8)term) && UTF)
6903 /* mark where we are */
6904 PL_multi_start = CopLINE(PL_curcop);
6905 PL_multi_open = term;
6907 /* find corresponding closing delimiter */
6908 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6910 PL_multi_close = term;
6912 /* create a new SV to hold the contents. 87 is leak category, I'm
6913 assuming. 79 is the SV's initial length. What a random number. */
6915 sv_upgrade(sv, SVt_PVIV);
6917 (void)SvPOK_only(sv); /* validate pointer */
6919 /* move past delimiter and try to read a complete string */
6921 sv_catpvn(sv, s, 1);
6924 /* extend sv if need be */
6925 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6926 /* set 'to' to the next character in the sv's string */
6927 to = SvPVX(sv)+SvCUR(sv);
6929 /* if open delimiter is the close delimiter read unbridle */
6930 if (PL_multi_open == PL_multi_close) {
6931 for (; s < PL_bufend; s++,to++) {
6932 /* embedded newlines increment the current line number */
6933 if (*s == '\n' && !PL_rsfp)
6934 CopLINE_inc(PL_curcop);
6935 /* handle quoted delimiters */
6936 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6937 if (!keep_quoted && s[1] == term)
6939 /* any other quotes are simply copied straight through */
6943 /* terminate when run out of buffer (the for() condition), or
6944 have found the terminator */
6945 else if (*s == term)
6947 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
6953 /* if the terminator isn't the same as the start character (e.g.,
6954 matched brackets), we have to allow more in the quoting, and
6955 be prepared for nested brackets.
6958 /* read until we run out of string, or we find the terminator */
6959 for (; s < PL_bufend; s++,to++) {
6960 /* embedded newlines increment the line count */
6961 if (*s == '\n' && !PL_rsfp)
6962 CopLINE_inc(PL_curcop);
6963 /* backslashes can escape the open or closing characters */
6964 if (*s == '\\' && s+1 < PL_bufend) {
6966 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6971 /* allow nested opens and closes */
6972 else if (*s == PL_multi_close && --brackets <= 0)
6974 else if (*s == PL_multi_open)
6976 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
6981 /* terminate the copied string and update the sv's end-of-string */
6983 SvCUR_set(sv, to - SvPVX(sv));
6986 * this next chunk reads more into the buffer if we're not done yet
6990 break; /* handle case where we are done yet :-) */
6992 #ifndef PERL_STRICT_CR
6993 if (to - SvPVX(sv) >= 2) {
6994 if ((to[-2] == '\r' && to[-1] == '\n') ||
6995 (to[-2] == '\n' && to[-1] == '\r'))
6999 SvCUR_set(sv, to - SvPVX(sv));
7001 else if (to[-1] == '\r')
7004 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
7008 /* if we're out of file, or a read fails, bail and reset the current
7009 line marker so we can report where the unterminated string began
7012 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
7014 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
7017 /* we read a line, so increment our line counter */
7018 CopLINE_inc(PL_curcop);
7020 /* update debugger info */
7021 if (PERLDB_LINE && PL_curstash != PL_debstash) {
7022 SV *sv = NEWSV(88,0);
7024 sv_upgrade(sv, SVt_PVMG);
7025 sv_setsv(sv,PL_linestr);
7028 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
7031 /* having changed the buffer, we must update PL_bufend */
7032 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7033 PL_last_lop = PL_last_uni = Nullch;
7036 /* at this point, we have successfully read the delimited string */
7039 sv_catpvn(sv, s, 1);
7042 else if (PL_encoding)
7043 sv_recode_to_utf8(sv, PL_encoding);
7045 PL_multi_end = CopLINE(PL_curcop);
7048 /* if we allocated too much space, give some back */
7049 if (SvCUR(sv) + 5 < SvLEN(sv)) {
7050 SvLEN_set(sv, SvCUR(sv) + 1);
7051 Renew(SvPVX(sv), SvLEN(sv), char);
7054 /* decide whether this is the first or second quoted string we've read
7067 takes: pointer to position in buffer
7068 returns: pointer to new position in buffer
7069 side-effects: builds ops for the constant in yylval.op
7071 Read a number in any of the formats that Perl accepts:
7073 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
7074 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
7077 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
7079 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
7082 If it reads a number without a decimal point or an exponent, it will
7083 try converting the number to an integer and see if it can do so
7084 without loss of precision.
7088 Perl_scan_num(pTHX_ char *start, YYSTYPE* lvalp)
7090 register char *s = start; /* current position in buffer */
7091 register char *d; /* destination in temp buffer */
7092 register char *e; /* end of temp buffer */
7093 NV nv; /* number read, as a double */
7094 SV *sv = Nullsv; /* place to put the converted number */
7095 bool floatit; /* boolean: int or float? */
7096 char *lastub = 0; /* position of last underbar */
7097 static char number_too_long[] = "Number too long";
7099 /* We use the first character to decide what type of number this is */
7103 Perl_croak(aTHX_ "panic: scan_num");
7105 /* if it starts with a 0, it could be an octal number, a decimal in
7106 0.13 disguise, or a hexadecimal number, or a binary number. */
7110 u holds the "number so far"
7111 shift the power of 2 of the base
7112 (hex == 4, octal == 3, binary == 1)
7113 overflowed was the number more than we can hold?
7115 Shift is used when we add a digit. It also serves as an "are
7116 we in octal/hex/binary?" indicator to disallow hex characters
7122 bool overflowed = FALSE;
7123 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
7124 static char* bases[5] = { "", "binary", "", "octal",
7126 static char* Bases[5] = { "", "Binary", "", "Octal",
7128 static char *maxima[5] = { "",
7129 "0b11111111111111111111111111111111",
7133 char *base, *Base, *max;
7139 } else if (s[1] == 'b') {
7143 /* check for a decimal in disguise */
7144 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
7146 /* so it must be octal */
7153 if (ckWARN(WARN_SYNTAX))
7154 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7155 "Misplaced _ in number");
7159 base = bases[shift];
7160 Base = Bases[shift];
7161 max = maxima[shift];
7163 /* read the rest of the number */
7165 /* x is used in the overflow test,
7166 b is the digit we're adding on. */
7171 /* if we don't mention it, we're done */
7175 /* _ are ignored -- but warned about if consecutive */
7177 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7178 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7179 "Misplaced _ in number");
7183 /* 8 and 9 are not octal */
7186 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
7190 case '2': case '3': case '4':
7191 case '5': case '6': case '7':
7193 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
7197 b = *s++ & 15; /* ASCII digit -> value of digit */
7201 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
7202 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
7203 /* make sure they said 0x */
7208 /* Prepare to put the digit we have onto the end
7209 of the number so far. We check for overflows.
7214 x = u << shift; /* make room for the digit */
7216 if ((x >> shift) != u
7217 && !(PL_hints & HINT_NEW_BINARY)) {
7220 if (ckWARN_d(WARN_OVERFLOW))
7221 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
7222 "Integer overflow in %s number",
7225 u = x | b; /* add the digit to the end */
7228 n *= nvshift[shift];
7229 /* If an NV has not enough bits in its
7230 * mantissa to represent an UV this summing of
7231 * small low-order numbers is a waste of time
7232 * (because the NV cannot preserve the
7233 * low-order bits anyway): we could just
7234 * remember when did we overflow and in the
7235 * end just multiply n by the right
7243 /* if we get here, we had success: make a scalar value from
7248 /* final misplaced underbar check */
7250 if (ckWARN(WARN_SYNTAX))
7251 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
7256 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
7257 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
7258 "%s number > %s non-portable",
7264 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
7265 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
7266 "%s number > %s non-portable",
7271 if (PL_hints & HINT_NEW_BINARY)
7272 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
7277 handle decimal numbers.
7278 we're also sent here when we read a 0 as the first digit
7280 case '1': case '2': case '3': case '4': case '5':
7281 case '6': case '7': case '8': case '9': case '.':
7284 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
7287 /* read next group of digits and _ and copy into d */
7288 while (isDIGIT(*s) || *s == '_') {
7289 /* skip underscores, checking for misplaced ones
7293 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7294 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7295 "Misplaced _ in number");
7299 /* check for end of fixed-length buffer */
7301 Perl_croak(aTHX_ number_too_long);
7302 /* if we're ok, copy the character */
7307 /* final misplaced underbar check */
7308 if (lastub && s == lastub + 1) {
7309 if (ckWARN(WARN_SYNTAX))
7310 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
7313 /* read a decimal portion if there is one. avoid
7314 3..5 being interpreted as the number 3. followed
7317 if (*s == '.' && s[1] != '.') {
7322 if (ckWARN(WARN_SYNTAX))
7323 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7324 "Misplaced _ in number");
7328 /* copy, ignoring underbars, until we run out of digits.
7330 for (; isDIGIT(*s) || *s == '_'; s++) {
7331 /* fixed length buffer check */
7333 Perl_croak(aTHX_ number_too_long);
7335 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7336 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7337 "Misplaced _ in number");
7343 /* fractional part ending in underbar? */
7345 if (ckWARN(WARN_SYNTAX))
7346 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7347 "Misplaced _ in number");
7349 if (*s == '.' && isDIGIT(s[1])) {
7350 /* oops, it's really a v-string, but without the "v" */
7356 /* read exponent part, if present */
7357 if (*s && strchr("eE",*s) && strchr("+-0123456789_", s[1])) {
7361 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
7362 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
7364 /* stray preinitial _ */
7366 if (ckWARN(WARN_SYNTAX))
7367 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7368 "Misplaced _ in number");
7372 /* allow positive or negative exponent */
7373 if (*s == '+' || *s == '-')
7376 /* stray initial _ */
7378 if (ckWARN(WARN_SYNTAX))
7379 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7380 "Misplaced _ in number");
7384 /* read digits of exponent */
7385 while (isDIGIT(*s) || *s == '_') {
7388 Perl_croak(aTHX_ number_too_long);
7392 if (ckWARN(WARN_SYNTAX) &&
7393 ((lastub && s == lastub + 1) ||
7394 (!isDIGIT(s[1]) && s[1] != '_')))
7395 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7396 "Misplaced _ in number");
7403 /* make an sv from the string */
7407 We try to do an integer conversion first if no characters
7408 indicating "float" have been found.
7413 int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
7415 if (flags == IS_NUMBER_IN_UV) {
7417 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7420 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
7421 if (uv <= (UV) IV_MIN)
7422 sv_setiv(sv, -(IV)uv);
7429 /* terminate the string */
7431 nv = Atof(PL_tokenbuf);
7435 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7436 (PL_hints & HINT_NEW_INTEGER) )
7437 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7438 (floatit ? "float" : "integer"),
7442 /* if it starts with a v, it could be a v-string */
7445 sv = NEWSV(92,5); /* preallocate storage space */
7446 s = scan_vstring(s,sv);
7450 /* make the op for the constant and return */
7453 lvalp->opval = newSVOP(OP_CONST, 0, sv);
7455 lvalp->opval = Nullop;
7461 S_scan_formline(pTHX_ register char *s)
7465 SV *stuff = newSVpvn("",0);
7466 bool needargs = FALSE;
7469 if (*s == '.' || *s == /*{*/'}') {
7471 #ifdef PERL_STRICT_CR
7472 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7474 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7476 if (*t == '\n' || t == PL_bufend)
7479 if (PL_in_eval && !PL_rsfp) {
7480 eol = strchr(s,'\n');
7485 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7487 for (t = s; t < eol; t++) {
7488 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7490 goto enough; /* ~~ must be first line in formline */
7492 if (*t == '@' || *t == '^')
7496 sv_catpvn(stuff, s, eol-s);
7497 #ifndef PERL_STRICT_CR
7498 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7499 char *end = SvPVX(stuff) + SvCUR(stuff);
7511 s = filter_gets(PL_linestr, PL_rsfp, 0);
7512 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7513 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7514 PL_last_lop = PL_last_uni = Nullch;
7517 yyerror("Format not terminated");
7527 PL_lex_state = LEX_NORMAL;
7528 PL_nextval[PL_nexttoke].ival = 0;
7532 PL_lex_state = LEX_FORMLINE;
7533 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7535 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7539 SvREFCNT_dec(stuff);
7540 PL_lex_formbrack = 0;
7551 PL_cshlen = strlen(PL_cshname);
7556 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7558 I32 oldsavestack_ix = PL_savestack_ix;
7559 CV* outsidecv = PL_compcv;
7562 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7564 SAVEI32(PL_subline);
7565 save_item(PL_subname);
7566 SAVESPTR(PL_compcv);
7568 PL_compcv = (CV*)NEWSV(1104,0);
7569 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7570 CvFLAGS(PL_compcv) |= flags;
7572 PL_subline = CopLINE(PL_curcop);
7573 CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
7574 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7575 CvOUTSIDE_SEQ(PL_compcv) = PL_cop_seqmax;
7577 return oldsavestack_ix;
7581 #pragma segment Perl_yylex
7584 Perl_yywarn(pTHX_ char *s)
7586 PL_in_eval |= EVAL_WARNONLY;
7588 PL_in_eval &= ~EVAL_WARNONLY;
7593 Perl_yyerror(pTHX_ char *s)
7596 char *context = NULL;
7600 if (!yychar || (yychar == ';' && !PL_rsfp))
7602 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7603 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7606 The code below is removed for NetWare because it abends/crashes on NetWare
7607 when the script has error such as not having the closing quotes like:
7609 Checking of white spaces is anyway done in NetWare code.
7612 while (isSPACE(*PL_oldoldbufptr))
7615 context = PL_oldoldbufptr;
7616 contlen = PL_bufptr - PL_oldoldbufptr;
7618 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7619 PL_oldbufptr != PL_bufptr) {
7622 The code below is removed for NetWare because it abends/crashes on NetWare
7623 when the script has error such as not having the closing quotes like:
7625 Checking of white spaces is anyway done in NetWare code.
7628 while (isSPACE(*PL_oldbufptr))
7631 context = PL_oldbufptr;
7632 contlen = PL_bufptr - PL_oldbufptr;
7634 else if (yychar > 255)
7635 where = "next token ???";
7636 #ifdef USE_PURE_BISON
7637 /* GNU Bison sets the value -2 */
7638 else if (yychar == -2) {
7640 else if ((yychar & 127) == 127) {
7642 if (PL_lex_state == LEX_NORMAL ||
7643 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7644 where = "at end of line";
7645 else if (PL_lex_inpat)
7646 where = "within pattern";
7648 where = "within string";
7651 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7653 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7654 else if (isPRINT_LC(yychar))
7655 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7657 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7658 where = SvPVX(where_sv);
7660 msg = sv_2mortal(newSVpv(s, 0));
7661 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7662 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7664 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7666 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7667 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7668 Perl_sv_catpvf(aTHX_ msg,
7669 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7670 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7673 if (PL_in_eval & EVAL_WARNONLY)
7674 Perl_warn(aTHX_ "%"SVf, msg);
7677 if (PL_error_count >= 10) {
7678 if (PL_in_eval && SvCUR(ERRSV))
7679 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
7680 ERRSV, OutCopFILE(PL_curcop));
7682 Perl_croak(aTHX_ "%s has too many errors.\n",
7683 OutCopFILE(PL_curcop));
7686 PL_in_my_stash = Nullhv;
7690 #pragma segment Main
7694 S_swallow_bom(pTHX_ U8 *s)
7697 slen = SvCUR(PL_linestr);
7701 /* UTF-16 little-endian */
7702 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
7703 Perl_croak(aTHX_ "Unsupported script encoding");
7704 #ifndef PERL_NO_UTF16_FILTER
7705 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-LE script encoding\n"));
7707 if (PL_bufend > (char*)s) {
7711 filter_add(utf16rev_textfilter, NULL);
7712 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7713 PL_bufend = (char*)utf16_to_utf8_reversed(s, news,
7714 PL_bufend - (char*)s - 1,
7716 Copy(news, s, newlen, U8);
7717 SvCUR_set(PL_linestr, newlen);
7718 PL_bufend = SvPVX(PL_linestr) + newlen;
7719 news[newlen++] = '\0';
7723 Perl_croak(aTHX_ "Unsupported script encoding");
7728 if (s[1] == 0xFF) { /* UTF-16 big-endian */
7729 #ifndef PERL_NO_UTF16_FILTER
7730 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding\n"));
7732 if (PL_bufend > (char *)s) {
7736 filter_add(utf16_textfilter, NULL);
7737 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7738 PL_bufend = (char*)utf16_to_utf8(s, news,
7739 PL_bufend - (char*)s,
7741 Copy(news, s, newlen, U8);
7742 SvCUR_set(PL_linestr, newlen);
7743 PL_bufend = SvPVX(PL_linestr) + newlen;
7744 news[newlen++] = '\0';
7748 Perl_croak(aTHX_ "Unsupported script encoding");
7753 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
7754 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-8 script encoding\n"));
7759 if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */
7760 s[2] == 0xFE && s[3] == 0xFF)
7762 Perl_croak(aTHX_ "Unsupported script encoding");
7770 * Restore a source filter.
7774 restore_rsfp(pTHX_ void *f)
7776 PerlIO *fp = (PerlIO*)f;
7778 if (PL_rsfp == PerlIO_stdin())
7779 PerlIO_clearerr(PL_rsfp);
7780 else if (PL_rsfp && (PL_rsfp != fp))
7781 PerlIO_close(PL_rsfp);
7785 #ifndef PERL_NO_UTF16_FILTER
7787 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
7789 I32 count = FILTER_READ(idx+1, sv, maxlen);
7794 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7795 if (!*SvPV_nolen(sv))
7796 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7799 tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7800 sv_usepvn(sv, (char*)tmps, tend - tmps);
7806 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
7808 I32 count = FILTER_READ(idx+1, sv, maxlen);
7813 if (!*SvPV_nolen(sv))
7814 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7817 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7818 tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7819 sv_usepvn(sv, (char*)tmps, tend - tmps);