3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
4 * 2000, 2001, 2002, 2003, by Larry Wall and others
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
12 * "It all comes from here, the stench and the peril." --Frodo
16 * This file is the lexer for Perl. It's closely linked to the
19 * The main routine is yylex(), which returns the next token.
23 #define PERL_IN_TOKE_C
26 #define yychar PL_yychar
27 #define yylval PL_yylval
29 static char ident_too_long[] = "Identifier too long";
30 static char c_without_g[] = "Use of /c modifier is meaningless without /g";
31 static char c_in_subst[] = "Use of /c modifier is meaningless in s///";
33 static void restore_rsfp(pTHX_ void *f);
34 #ifndef PERL_NO_UTF16_FILTER
35 static I32 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen);
36 static I32 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen);
39 #define XFAKEBRACK 128
42 #ifdef USE_UTF8_SCRIPTS
43 # define UTF (!IN_BYTES)
45 # define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
48 /* In variables named $^X, these are the legal values for X.
49 * 1999-02-27 mjd-perl-patch@plover.com */
50 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
52 /* On MacOS, respect nonbreaking spaces */
53 #ifdef MACOS_TRADITIONAL
54 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
56 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
59 /* LEX_* are values for PL_lex_state, the state of the lexer.
60 * They are arranged oddly so that the guard on the switch statement
61 * can get by with a single comparison (if the compiler is smart enough).
64 /* #define LEX_NOTPARSING 11 is done in perl.h. */
67 #define LEX_INTERPNORMAL 9
68 #define LEX_INTERPCASEMOD 8
69 #define LEX_INTERPPUSH 7
70 #define LEX_INTERPSTART 6
71 #define LEX_INTERPEND 5
72 #define LEX_INTERPENDMAYBE 4
73 #define LEX_INTERPCONCAT 3
74 #define LEX_INTERPCONST 2
75 #define LEX_FORMLINE 1
76 #define LEX_KNOWNEXT 0
84 # define YYMAXLEVEL 100
86 YYSTYPE* yylval_pointer[YYMAXLEVEL];
87 int* yychar_pointer[YYMAXLEVEL];
91 # define yylval (*yylval_pointer[yyactlevel])
92 # define yychar (*yychar_pointer[yyactlevel])
93 # define PERL_YYLEX_PARAM yylval_pointer[yyactlevel],yychar_pointer[yyactlevel]
95 # define yylex() Perl_yylex_r(aTHX_ yylval_pointer[yyactlevel],yychar_pointer[yyactlevel])
100 /* CLINE is a macro that ensures PL_copline has a sane value */
105 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
108 * Convenience functions to return different tokens and prime the
109 * lexer for the next token. They all take an argument.
111 * TOKEN : generic token (used for '(', DOLSHARP, etc)
112 * OPERATOR : generic operator
113 * AOPERATOR : assignment operator
114 * PREBLOCK : beginning the block after an if, while, foreach, ...
115 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
116 * PREREF : *EXPR where EXPR is not a simple identifier
117 * TERM : expression term
118 * LOOPX : loop exiting command (goto, last, dump, etc)
119 * FTST : file test operator
120 * FUN0 : zero-argument function
121 * FUN1 : not used, except for not, which isn't a UNIOP
122 * BOop : bitwise or or xor
124 * SHop : shift operator
125 * PWop : power operator
126 * PMop : pattern-matching operator
127 * Aop : addition-level operator
128 * Mop : multiplication-level operator
129 * Eop : equality-testing operator
130 * Rop : relational operator <= != gt
132 * Also see LOP and lop() below.
135 /* Note that REPORT() and REPORT2() will be expressions that supply
136 * their own trailing comma, not suitable for statements as such. */
137 #ifdef DEBUGGING /* Serve -DT. */
138 # define REPORT(x,retval) tokereport(x,s,(int)retval),
139 # define REPORT2(x,retval) tokereport(x,s, yylval.ival),
141 # define REPORT(x,retval)
142 # define REPORT2(x,retval)
145 #define TOKEN(retval) return (REPORT2("token",retval) PL_bufptr = s,(int)retval)
146 #define OPERATOR(retval) return (REPORT2("operator",retval) PL_expect = XTERM, PL_bufptr = s,(int)retval)
147 #define AOPERATOR(retval) return ao((REPORT2("aop",retval) PL_expect = XTERM, PL_bufptr = s,(int)retval))
148 #define PREBLOCK(retval) return (REPORT2("preblock",retval) PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
149 #define PRETERMBLOCK(retval) return (REPORT2("pretermblock",retval) PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
150 #define PREREF(retval) return (REPORT2("preref",retval) PL_expect = XREF,PL_bufptr = s,(int)retval)
151 #define TERM(retval) return (CLINE, REPORT2("term",retval) PL_expect = XOPERATOR, PL_bufptr = s,(int)retval)
152 #define LOOPX(f) return(yylval.ival=f, REPORT("loopx",f) PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
153 #define FTST(f) return(yylval.ival=f, REPORT("ftst",f) PL_expect = XTERMORDORDOR,PL_bufptr = s,(int)UNIOP)
154 #define FUN0(f) return(yylval.ival = f, REPORT("fun0",f) PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
155 #define FUN1(f) return(yylval.ival = f, REPORT("fun1",f) PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
156 #define BOop(f) return ao((yylval.ival=f, REPORT("bitorop",f) PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
157 #define BAop(f) return ao((yylval.ival=f, REPORT("bitandop",f) PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
158 #define SHop(f) return ao((yylval.ival=f, REPORT("shiftop",f) PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
159 #define PWop(f) return ao((yylval.ival=f, REPORT("powop",f) PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
160 #define PMop(f) return(yylval.ival=f, REPORT("matchop",f) PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
161 #define Aop(f) return ao((yylval.ival=f, REPORT("add",f) PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
162 #define Mop(f) return ao((yylval.ival=f, REPORT("mul",f) PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
163 #define Eop(f) return(yylval.ival=f, REPORT("eq",f) PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
164 #define Rop(f) return(yylval.ival=f, REPORT("rel",f) PL_expect = XTERM,PL_bufptr = s,(int)RELOP)
166 /* This bit of chicanery makes a unary function followed by
167 * a parenthesis into a function with one argument, highest precedence.
168 * The UNIDOR macro is for unary functions that can be followed by the //
169 * operator (such as C<shift // 0>).
171 #define UNI2(f,x) return(yylval.ival = f, \
175 PL_last_uni = PL_oldbufptr, \
176 PL_last_lop_op = f, \
177 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
178 #define UNI(f) UNI2(f,XTERM)
179 #define UNIDOR(f) UNI2(f,XTERMORDORDOR)
181 #define UNIBRACK(f) return(yylval.ival = f, \
184 PL_last_uni = PL_oldbufptr, \
185 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
187 /* grandfather return to old style */
188 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
193 S_tokereport(pTHX_ char *thing, char* s, I32 rv)
196 SV* report = newSVpv(thing, 0);
197 Perl_sv_catpvf(aTHX_ report, ":line %d:%"IVdf":", CopLINE(PL_curcop),
200 if (s - PL_bufptr > 0)
201 sv_catpvn(report, PL_bufptr, s - PL_bufptr);
203 if (PL_oldbufptr && *PL_oldbufptr)
204 sv_catpv(report, PL_tokenbuf);
206 PerlIO_printf(Perl_debug_log, "### %s\n", SvPV_nolen(report));
215 * This subroutine detects &&=, ||=, and //= and turns an ANDAND, OROR or DORDOR
216 * into an OP_ANDASSIGN, OP_ORASSIGN, or OP_DORASSIGN
220 S_ao(pTHX_ int toketype)
222 if (*PL_bufptr == '=') {
224 if (toketype == ANDAND)
225 yylval.ival = OP_ANDASSIGN;
226 else if (toketype == OROR)
227 yylval.ival = OP_ORASSIGN;
228 else if (toketype == DORDOR)
229 yylval.ival = OP_DORASSIGN;
237 * When Perl expects an operator and finds something else, no_op
238 * prints the warning. It always prints "<something> found where
239 * operator expected. It prints "Missing semicolon on previous line?"
240 * if the surprise occurs at the start of the line. "do you need to
241 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
242 * where the compiler doesn't know if foo is a method call or a function.
243 * It prints "Missing operator before end of line" if there's nothing
244 * after the missing operator, or "... before <...>" if there is something
245 * after the missing operator.
249 S_no_op(pTHX_ char *what, char *s)
251 char *oldbp = PL_bufptr;
252 bool is_first = (PL_oldbufptr == PL_linestart);
258 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
259 if (ckWARN_d(WARN_SYNTAX)) {
261 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
262 "\t(Missing semicolon on previous line?)\n");
263 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
265 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
266 if (t < PL_bufptr && isSPACE(*t))
267 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
268 "\t(Do you need to predeclare %.*s?)\n",
269 t - PL_oldoldbufptr, PL_oldoldbufptr);
273 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
274 "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
282 * Complain about missing quote/regexp/heredoc terminator.
283 * If it's called with (char *)NULL then it cauterizes the line buffer.
284 * If we're in a delimited string and the delimiter is a control
285 * character, it's reformatted into a two-char sequence like ^C.
290 S_missingterm(pTHX_ char *s)
295 char *nl = strrchr(s,'\n');
301 iscntrl(PL_multi_close)
303 PL_multi_close < 32 || PL_multi_close == 127
307 tmpbuf[1] = toCTRL(PL_multi_close);
313 *tmpbuf = (char)PL_multi_close;
317 q = strchr(s,'"') ? '\'' : '"';
318 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
326 Perl_deprecate(pTHX_ char *s)
328 if (ckWARN(WARN_DEPRECATED))
329 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED), "Use of %s is deprecated", s);
333 Perl_deprecate_old(pTHX_ char *s)
335 /* This function should NOT be called for any new deprecated warnings */
336 /* Use Perl_deprecate instead */
338 /* It is here to maintain backward compatibility with the pre-5.8 */
339 /* warnings category hierarchy. The "deprecated" category used to */
340 /* live under the "syntax" category. It is now a top-level category */
341 /* in its own right. */
343 if (ckWARN2(WARN_DEPRECATED, WARN_SYNTAX))
344 Perl_warner(aTHX_ packWARN2(WARN_DEPRECATED, WARN_SYNTAX),
345 "Use of %s is deprecated", s);
350 * Deprecate a comma-less variable list.
356 deprecate_old("comma-less variable list");
360 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
361 * utf16-to-utf8-reversed.
364 #ifdef PERL_CR_FILTER
368 register char *s = SvPVX(sv);
369 register char *e = s + SvCUR(sv);
370 /* outer loop optimized to do nothing if there are no CR-LFs */
372 if (*s++ == '\r' && *s == '\n') {
373 /* hit a CR-LF, need to copy the rest */
374 register char *d = s - 1;
377 if (*s == '\r' && s[1] == '\n')
388 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
390 I32 count = FILTER_READ(idx+1, sv, maxlen);
391 if (count > 0 && !maxlen)
399 * Initialize variables. Uses the Perl save_stack to save its state (for
400 * recursive calls to the parser).
404 Perl_lex_start(pTHX_ SV *line)
409 SAVEI32(PL_lex_dojoin);
410 SAVEI32(PL_lex_brackets);
411 SAVEI32(PL_lex_casemods);
412 SAVEI32(PL_lex_starts);
413 SAVEI32(PL_lex_state);
414 SAVEVPTR(PL_lex_inpat);
415 SAVEI32(PL_lex_inwhat);
416 if (PL_lex_state == LEX_KNOWNEXT) {
417 I32 toke = PL_nexttoke;
418 while (--toke >= 0) {
419 SAVEI32(PL_nexttype[toke]);
420 SAVEVPTR(PL_nextval[toke]);
422 SAVEI32(PL_nexttoke);
424 SAVECOPLINE(PL_curcop);
427 SAVEPPTR(PL_oldbufptr);
428 SAVEPPTR(PL_oldoldbufptr);
429 SAVEPPTR(PL_last_lop);
430 SAVEPPTR(PL_last_uni);
431 SAVEPPTR(PL_linestart);
432 SAVESPTR(PL_linestr);
433 SAVEGENERICPV(PL_lex_brackstack);
434 SAVEGENERICPV(PL_lex_casestack);
435 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
436 SAVESPTR(PL_lex_stuff);
437 SAVEI32(PL_lex_defer);
438 SAVEI32(PL_sublex_info.sub_inwhat);
439 SAVESPTR(PL_lex_repl);
441 SAVEINT(PL_lex_expect);
443 PL_lex_state = LEX_NORMAL;
447 New(899, PL_lex_brackstack, 120, char);
448 New(899, PL_lex_casestack, 12, char);
450 *PL_lex_casestack = '\0';
453 PL_lex_stuff = Nullsv;
454 PL_lex_repl = Nullsv;
458 PL_sublex_info.sub_inwhat = 0;
460 if (SvREADONLY(PL_linestr))
461 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
462 s = SvPV(PL_linestr, len);
463 if (!len || s[len-1] != ';') {
464 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
465 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
466 sv_catpvn(PL_linestr, "\n;", 2);
468 SvTEMP_off(PL_linestr);
469 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
470 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
471 PL_last_lop = PL_last_uni = Nullch;
477 * Finalizer for lexing operations. Must be called when the parser is
478 * done with the lexer.
484 PL_doextract = FALSE;
489 * This subroutine has nothing to do with tilting, whether at windmills
490 * or pinball tables. Its name is short for "increment line". It
491 * increments the current line number in CopLINE(PL_curcop) and checks
492 * to see whether the line starts with a comment of the form
493 * # line 500 "foo.pm"
494 * If so, it sets the current line number and file to the values in the comment.
498 S_incline(pTHX_ char *s)
505 CopLINE_inc(PL_curcop);
508 while (SPACE_OR_TAB(*s)) s++;
509 if (strnEQ(s, "line", 4))
513 if (SPACE_OR_TAB(*s))
517 while (SPACE_OR_TAB(*s)) s++;
523 while (SPACE_OR_TAB(*s))
525 if (*s == '"' && (t = strchr(s+1, '"'))) {
530 for (t = s; !isSPACE(*t); t++) ;
533 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
535 if (*e != '\n' && *e != '\0')
536 return; /* false alarm */
541 CopFILE_free(PL_curcop);
542 CopFILE_set(PL_curcop, s);
545 CopLINE_set(PL_curcop, atoi(n)-1);
550 * Called to gobble the appropriate amount and type of whitespace.
551 * Skips comments as well.
555 S_skipspace(pTHX_ register char *s)
557 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
558 while (s < PL_bufend && SPACE_OR_TAB(*s))
564 SSize_t oldprevlen, oldoldprevlen;
565 SSize_t oldloplen = 0, oldunilen = 0;
566 while (s < PL_bufend && isSPACE(*s)) {
567 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
572 if (s < PL_bufend && *s == '#') {
573 while (s < PL_bufend && *s != '\n')
577 if (PL_in_eval && !PL_rsfp) {
584 /* only continue to recharge the buffer if we're at the end
585 * of the buffer, we're not reading from a source filter, and
586 * we're in normal lexing mode
588 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
589 PL_lex_state == LEX_FORMLINE)
592 /* try to recharge the buffer */
593 if ((s = filter_gets(PL_linestr, PL_rsfp,
594 (prevlen = SvCUR(PL_linestr)))) == Nullch)
596 /* end of file. Add on the -p or -n magic */
597 if (PL_minus_n || PL_minus_p) {
598 sv_setpv(PL_linestr,PL_minus_p ?
599 ";}continue{print or die qq(-p destination: $!\\n)" :
601 sv_catpv(PL_linestr,";}");
602 PL_minus_n = PL_minus_p = 0;
605 sv_setpv(PL_linestr,";");
607 /* reset variables for next time we lex */
608 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
610 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
611 PL_last_lop = PL_last_uni = Nullch;
613 /* Close the filehandle. Could be from -P preprocessor,
614 * STDIN, or a regular file. If we were reading code from
615 * STDIN (because the commandline held no -e or filename)
616 * then we don't close it, we reset it so the code can
617 * read from STDIN too.
620 if (PL_preprocess && !PL_in_eval)
621 (void)PerlProc_pclose(PL_rsfp);
622 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
623 PerlIO_clearerr(PL_rsfp);
625 (void)PerlIO_close(PL_rsfp);
630 /* not at end of file, so we only read another line */
631 /* make corresponding updates to old pointers, for yyerror() */
632 oldprevlen = PL_oldbufptr - PL_bufend;
633 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
635 oldunilen = PL_last_uni - PL_bufend;
637 oldloplen = PL_last_lop - PL_bufend;
638 PL_linestart = PL_bufptr = s + prevlen;
639 PL_bufend = s + SvCUR(PL_linestr);
641 PL_oldbufptr = s + oldprevlen;
642 PL_oldoldbufptr = s + oldoldprevlen;
644 PL_last_uni = s + oldunilen;
646 PL_last_lop = s + oldloplen;
649 /* debugger active and we're not compiling the debugger code,
650 * so store the line into the debugger's array of lines
652 if (PERLDB_LINE && PL_curstash != PL_debstash) {
653 SV *sv = NEWSV(85,0);
655 sv_upgrade(sv, SVt_PVMG);
656 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
659 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
666 * Check the unary operators to ensure there's no ambiguity in how they're
667 * used. An ambiguous piece of code would be:
669 * This doesn't mean rand() + 5. Because rand() is a unary operator,
670 * the +5 is its argument.
679 if (PL_oldoldbufptr != PL_last_uni)
681 while (isSPACE(*PL_last_uni))
683 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
684 if ((t = strchr(s, '(')) && t < PL_bufptr)
686 if (ckWARN_d(WARN_AMBIGUOUS)){
689 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
690 "Warning: Use of \"%s\" without parentheses is ambiguous",
697 * LOP : macro to build a list operator. Its behaviour has been replaced
698 * with a subroutine, S_lop() for which LOP is just another name.
701 #define LOP(f,x) return lop(f,x,s)
705 * Build a list operator (or something that might be one). The rules:
706 * - if we have a next token, then it's a list operator [why?]
707 * - if the next thing is an opening paren, then it's a function
708 * - else it's a list operator
712 S_lop(pTHX_ I32 f, int x, char *s)
719 PL_last_lop = PL_oldbufptr;
720 PL_last_lop_op = (OPCODE)f;
734 * When the lexer realizes it knows the next token (for instance,
735 * it is reordering tokens for the parser) then it can call S_force_next
736 * to know what token to return the next time the lexer is called. Caller
737 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
738 * handles the token correctly.
742 S_force_next(pTHX_ I32 type)
744 PL_nexttype[PL_nexttoke] = type;
746 if (PL_lex_state != LEX_KNOWNEXT) {
747 PL_lex_defer = PL_lex_state;
748 PL_lex_expect = PL_expect;
749 PL_lex_state = LEX_KNOWNEXT;
755 * When the lexer knows the next thing is a word (for instance, it has
756 * just seen -> and it knows that the next char is a word char, then
757 * it calls S_force_word to stick the next word into the PL_next lookahead.
760 * char *start : buffer position (must be within PL_linestr)
761 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
762 * int check_keyword : if true, Perl checks to make sure the word isn't
763 * a keyword (do this if the word is a label, e.g. goto FOO)
764 * int allow_pack : if true, : characters will also be allowed (require,
766 * int allow_initial_tick : used by the "sub" lexer only.
770 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
775 start = skipspace(start);
777 if (isIDFIRST_lazy_if(s,UTF) ||
778 (allow_pack && *s == ':') ||
779 (allow_initial_tick && *s == '\'') )
781 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
782 if (check_keyword && keyword(PL_tokenbuf, len))
784 if (token == METHOD) {
789 PL_expect = XOPERATOR;
792 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
793 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
794 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
795 SvUTF8_on(((SVOP*)PL_nextval[PL_nexttoke].opval)->op_sv);
803 * Called when the lexer wants $foo *foo &foo etc, but the program
804 * text only contains the "foo" portion. The first argument is a pointer
805 * to the "foo", and the second argument is the type symbol to prefix.
806 * Forces the next token to be a "WORD".
807 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
811 S_force_ident(pTHX_ register char *s, int kind)
814 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
815 PL_nextval[PL_nexttoke].opval = o;
818 o->op_private = OPpCONST_ENTERED;
819 /* XXX see note in pp_entereval() for why we forgo typo
820 warnings if the symbol must be introduced in an eval.
822 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
823 kind == '$' ? SVt_PV :
824 kind == '@' ? SVt_PVAV :
825 kind == '%' ? SVt_PVHV :
833 Perl_str_to_version(pTHX_ SV *sv)
838 char *start = SvPVx(sv,len);
839 bool utf = SvUTF8(sv) ? TRUE : FALSE;
840 char *end = start + len;
841 while (start < end) {
845 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
850 retval += ((NV)n)/nshift;
859 * Forces the next token to be a version number.
860 * If the next token appears to be an invalid version number, (e.g. "v2b"),
861 * and if "guessing" is TRUE, then no new token is created (and the caller
862 * must use an alternative parsing method).
866 S_force_version(pTHX_ char *s, int guessing)
868 OP *version = Nullop;
877 while (isDIGIT(*d) || *d == '_' || *d == '.')
879 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
881 s = scan_num(s, &yylval);
882 version = yylval.opval;
883 ver = cSVOPx(version)->op_sv;
884 if (SvPOK(ver) && !SvNIOK(ver)) {
885 (void)SvUPGRADE(ver, SVt_PVNV);
886 SvNVX(ver) = str_to_version(ver);
887 SvNOK_on(ver); /* hint that it is a version */
894 /* NOTE: The parser sees the package name and the VERSION swapped */
895 PL_nextval[PL_nexttoke].opval = version;
903 * Tokenize a quoted string passed in as an SV. It finds the next
904 * chunk, up to end of string or a backslash. It may make a new
905 * SV containing that chunk (if HINT_NEW_STRING is on). It also
910 S_tokeq(pTHX_ SV *sv)
921 s = SvPV_force(sv, len);
922 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
925 while (s < send && *s != '\\')
930 if ( PL_hints & HINT_NEW_STRING ) {
931 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
937 if (s + 1 < send && (s[1] == '\\'))
938 s++; /* all that, just for this */
943 SvCUR_set(sv, d - SvPVX(sv));
945 if ( PL_hints & HINT_NEW_STRING )
946 return new_constant(NULL, 0, "q", sv, pv, "q");
951 * Now come three functions related to double-quote context,
952 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
953 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
954 * interact with PL_lex_state, and create fake ( ... ) argument lists
955 * to handle functions and concatenation.
956 * They assume that whoever calls them will be setting up a fake
957 * join call, because each subthing puts a ',' after it. This lets
960 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
962 * (I'm not sure whether the spurious commas at the end of lcfirst's
963 * arguments and join's arguments are created or not).
968 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
970 * Pattern matching will set PL_lex_op to the pattern-matching op to
971 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
973 * OP_CONST and OP_READLINE are easy--just make the new op and return.
975 * Everything else becomes a FUNC.
977 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
978 * had an OP_CONST or OP_READLINE). This just sets us up for a
979 * call to S_sublex_push().
985 register I32 op_type = yylval.ival;
987 if (op_type == OP_NULL) {
988 yylval.opval = PL_lex_op;
992 if (op_type == OP_CONST || op_type == OP_READLINE) {
993 SV *sv = tokeq(PL_lex_stuff);
995 if (SvTYPE(sv) == SVt_PVIV) {
996 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
1002 nsv = newSVpvn(p, len);
1008 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
1009 PL_lex_stuff = Nullsv;
1010 /* Allow <FH> // "foo" */
1011 if (op_type == OP_READLINE)
1012 PL_expect = XTERMORDORDOR;
1016 PL_sublex_info.super_state = PL_lex_state;
1017 PL_sublex_info.sub_inwhat = op_type;
1018 PL_sublex_info.sub_op = PL_lex_op;
1019 PL_lex_state = LEX_INTERPPUSH;
1023 yylval.opval = PL_lex_op;
1033 * Create a new scope to save the lexing state. The scope will be
1034 * ended in S_sublex_done. Returns a '(', starting the function arguments
1035 * to the uc, lc, etc. found before.
1036 * Sets PL_lex_state to LEX_INTERPCONCAT.
1044 PL_lex_state = PL_sublex_info.super_state;
1045 SAVEI32(PL_lex_dojoin);
1046 SAVEI32(PL_lex_brackets);
1047 SAVEI32(PL_lex_casemods);
1048 SAVEI32(PL_lex_starts);
1049 SAVEI32(PL_lex_state);
1050 SAVEVPTR(PL_lex_inpat);
1051 SAVEI32(PL_lex_inwhat);
1052 SAVECOPLINE(PL_curcop);
1053 SAVEPPTR(PL_bufptr);
1054 SAVEPPTR(PL_bufend);
1055 SAVEPPTR(PL_oldbufptr);
1056 SAVEPPTR(PL_oldoldbufptr);
1057 SAVEPPTR(PL_last_lop);
1058 SAVEPPTR(PL_last_uni);
1059 SAVEPPTR(PL_linestart);
1060 SAVESPTR(PL_linestr);
1061 SAVEGENERICPV(PL_lex_brackstack);
1062 SAVEGENERICPV(PL_lex_casestack);
1064 PL_linestr = PL_lex_stuff;
1065 PL_lex_stuff = Nullsv;
1067 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1068 = SvPVX(PL_linestr);
1069 PL_bufend += SvCUR(PL_linestr);
1070 PL_last_lop = PL_last_uni = Nullch;
1071 SAVEFREESV(PL_linestr);
1073 PL_lex_dojoin = FALSE;
1074 PL_lex_brackets = 0;
1075 New(899, PL_lex_brackstack, 120, char);
1076 New(899, PL_lex_casestack, 12, char);
1077 PL_lex_casemods = 0;
1078 *PL_lex_casestack = '\0';
1080 PL_lex_state = LEX_INTERPCONCAT;
1081 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
1083 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1084 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1085 PL_lex_inpat = PL_sublex_info.sub_op;
1087 PL_lex_inpat = Nullop;
1094 * Restores lexer state after a S_sublex_push.
1100 if (!PL_lex_starts++) {
1101 SV *sv = newSVpvn("",0);
1102 if (SvUTF8(PL_linestr))
1104 PL_expect = XOPERATOR;
1105 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1109 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1110 PL_lex_state = LEX_INTERPCASEMOD;
1114 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1115 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1116 PL_linestr = PL_lex_repl;
1118 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1119 PL_bufend += SvCUR(PL_linestr);
1120 PL_last_lop = PL_last_uni = Nullch;
1121 SAVEFREESV(PL_linestr);
1122 PL_lex_dojoin = FALSE;
1123 PL_lex_brackets = 0;
1124 PL_lex_casemods = 0;
1125 *PL_lex_casestack = '\0';
1127 if (SvEVALED(PL_lex_repl)) {
1128 PL_lex_state = LEX_INTERPNORMAL;
1130 /* we don't clear PL_lex_repl here, so that we can check later
1131 whether this is an evalled subst; that means we rely on the
1132 logic to ensure sublex_done() is called again only via the
1133 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1136 PL_lex_state = LEX_INTERPCONCAT;
1137 PL_lex_repl = Nullsv;
1143 PL_bufend = SvPVX(PL_linestr);
1144 PL_bufend += SvCUR(PL_linestr);
1145 PL_expect = XOPERATOR;
1146 PL_sublex_info.sub_inwhat = 0;
1154 Extracts a pattern, double-quoted string, or transliteration. This
1157 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1158 processing a pattern (PL_lex_inpat is true), a transliteration
1159 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1161 Returns a pointer to the character scanned up to. Iff this is
1162 advanced from the start pointer supplied (ie if anything was
1163 successfully parsed), will leave an OP for the substring scanned
1164 in yylval. Caller must intuit reason for not parsing further
1165 by looking at the next characters herself.
1169 double-quoted style: \r and \n
1170 regexp special ones: \D \s
1172 backrefs: \1 (deprecated in substitution replacements)
1173 case and quoting: \U \Q \E
1174 stops on @ and $, but not for $ as tail anchor
1176 In transliterations:
1177 characters are VERY literal, except for - not at the start or end
1178 of the string, which indicates a range. scan_const expands the
1179 range to the full set of intermediate characters.
1181 In double-quoted strings:
1183 double-quoted style: \r and \n
1185 backrefs: \1 (deprecated)
1186 case and quoting: \U \Q \E
1189 scan_const does *not* construct ops to handle interpolated strings.
1190 It stops processing as soon as it finds an embedded $ or @ variable
1191 and leaves it to the caller to work out what's going on.
1193 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @::foo.
1195 $ in pattern could be $foo or could be tail anchor. Assumption:
1196 it's a tail anchor if $ is the last thing in the string, or if it's
1197 followed by one of ")| \n\t"
1199 \1 (backreferences) are turned into $1
1201 The structure of the code is
1202 while (there's a character to process) {
1203 handle transliteration ranges
1204 skip regexp comments
1205 skip # initiated comments in //x patterns
1206 check for embedded @foo
1207 check for embedded scalars
1209 leave intact backslashes from leave (below)
1210 deprecate \1 in strings and sub replacements
1211 handle string-changing backslashes \l \U \Q \E, etc.
1212 switch (what was escaped) {
1213 handle - in a transliteration (becomes a literal -)
1214 handle \132 octal characters
1215 handle 0x15 hex characters
1216 handle \cV (control V)
1217 handle printf backslashes (\f, \r, \n, etc)
1219 } (end if backslash)
1220 } (end while character to read)
1225 S_scan_const(pTHX_ char *start)
1227 register char *send = PL_bufend; /* end of the constant */
1228 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1229 register char *s = start; /* start of the constant */
1230 register char *d = SvPVX(sv); /* destination for copies */
1231 bool dorange = FALSE; /* are we in a translit range? */
1232 bool didrange = FALSE; /* did we just finish a range? */
1233 I32 has_utf8 = FALSE; /* Output constant is UTF8 */
1234 I32 this_utf8 = UTF; /* The source string is assumed to be UTF8 */
1237 const char *leaveit = /* set of acceptably-backslashed characters */
1239 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1242 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1243 /* If we are doing a trans and we know we want UTF8 set expectation */
1244 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
1245 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1249 while (s < send || dorange) {
1250 /* get transliterations out of the way (they're most literal) */
1251 if (PL_lex_inwhat == OP_TRANS) {
1252 /* expand a range A-Z to the full set of characters. AIE! */
1254 I32 i; /* current expanded character */
1255 I32 min; /* first character in range */
1256 I32 max; /* last character in range */
1259 char *c = (char*)utf8_hop((U8*)d, -1);
1263 *c = (char)UTF_TO_NATIVE(0xff);
1264 /* mark the range as done, and continue */
1270 i = d - SvPVX(sv); /* remember current offset */
1271 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1272 d = SvPVX(sv) + i; /* refresh d after realloc */
1273 d -= 2; /* eat the first char and the - */
1275 min = (U8)*d; /* first char in range */
1276 max = (U8)d[1]; /* last char in range */
1280 "Invalid range \"%c-%c\" in transliteration operator",
1281 (char)min, (char)max);
1285 if ((isLOWER(min) && isLOWER(max)) ||
1286 (isUPPER(min) && isUPPER(max))) {
1288 for (i = min; i <= max; i++)
1290 *d++ = NATIVE_TO_NEED(has_utf8,i);
1292 for (i = min; i <= max; i++)
1294 *d++ = NATIVE_TO_NEED(has_utf8,i);
1299 for (i = min; i <= max; i++)
1302 /* mark the range as done, and continue */
1308 /* range begins (ignore - as first or last char) */
1309 else if (*s == '-' && s+1 < send && s != start) {
1311 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
1314 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
1326 /* if we get here, we're not doing a transliteration */
1328 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1329 except for the last char, which will be done separately. */
1330 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1332 while (s+1 < send && *s != ')')
1333 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1335 else if (s[2] == '{' /* This should match regcomp.c */
1336 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1339 char *regparse = s + (s[2] == '{' ? 3 : 4);
1342 while (count && (c = *regparse)) {
1343 if (c == '\\' && regparse[1])
1351 if (*regparse != ')')
1352 regparse--; /* Leave one char for continuation. */
1353 while (s < regparse)
1354 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1358 /* likewise skip #-initiated comments in //x patterns */
1359 else if (*s == '#' && PL_lex_inpat &&
1360 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1361 while (s+1 < send && *s != '\n')
1362 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1365 /* check for embedded arrays
1366 (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-)
1368 else if (*s == '@' && s[1]
1369 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1])))
1372 /* check for embedded scalars. only stop if we're sure it's a
1375 else if (*s == '$') {
1376 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1378 if (s + 1 < send && !strchr("()| \r\n\t", s[1]))
1379 break; /* in regexp, $ might be tail anchor */
1382 /* End of else if chain - OP_TRANS rejoin rest */
1385 if (*s == '\\' && s+1 < send) {
1388 /* some backslashes we leave behind */
1389 if (*leaveit && *s && strchr(leaveit, *s)) {
1390 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
1391 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1395 /* deprecate \1 in strings and substitution replacements */
1396 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1397 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1399 if (ckWARN(WARN_SYNTAX))
1400 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
1405 /* string-change backslash escapes */
1406 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1411 /* if we get here, it's either a quoted -, or a digit */
1414 /* quoted - in transliterations */
1416 if (PL_lex_inwhat == OP_TRANS) {
1423 if (ckWARN(WARN_MISC) &&
1426 Perl_warner(aTHX_ packWARN(WARN_MISC),
1427 "Unrecognized escape \\%c passed through",
1429 /* default action is to copy the quoted character */
1430 goto default_action;
1433 /* \132 indicates an octal constant */
1434 case '0': case '1': case '2': case '3':
1435 case '4': case '5': case '6': case '7':
1439 uv = grok_oct(s, &len, &flags, NULL);
1442 goto NUM_ESCAPE_INSERT;
1444 /* \x24 indicates a hex constant */
1448 char* e = strchr(s, '}');
1449 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1450 PERL_SCAN_DISALLOW_PREFIX;
1455 yyerror("Missing right brace on \\x{}");
1459 uv = grok_hex(s, &len, &flags, NULL);
1465 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
1466 uv = grok_hex(s, &len, &flags, NULL);
1472 /* Insert oct or hex escaped character.
1473 * There will always enough room in sv since such
1474 * escapes will be longer than any UTF-8 sequence
1475 * they can end up as. */
1477 /* We need to map to chars to ASCII before doing the tests
1480 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(uv))) {
1481 if (!has_utf8 && uv > 255) {
1482 /* Might need to recode whatever we have
1483 * accumulated so far if it contains any
1486 * (Can't we keep track of that and avoid
1487 * this rescan? --jhi)
1491 for (c = (U8 *) SvPVX(sv); c < (U8 *)d; c++) {
1492 if (!NATIVE_IS_INVARIANT(*c)) {
1497 STRLEN offset = d - SvPVX(sv);
1499 d = SvGROW(sv, SvLEN(sv) + hicount + 1) + offset;
1503 while (src >= (U8 *)SvPVX(sv)) {
1504 if (!NATIVE_IS_INVARIANT(*src)) {
1505 U8 ch = NATIVE_TO_ASCII(*src);
1506 *dst-- = (U8)UTF8_EIGHT_BIT_LO(ch);
1507 *dst-- = (U8)UTF8_EIGHT_BIT_HI(ch);
1517 if (has_utf8 || uv > 255) {
1518 d = (char*)uvchr_to_utf8((U8*)d, uv);
1520 if (PL_lex_inwhat == OP_TRANS &&
1521 PL_sublex_info.sub_op) {
1522 PL_sublex_info.sub_op->op_private |=
1523 (PL_lex_repl ? OPpTRANS_FROM_UTF
1536 /* \N{LATIN SMALL LETTER A} is a named character */
1540 char* e = strchr(s, '}');
1546 yyerror("Missing right brace on \\N{}");
1550 if (e > s + 2 && s[1] == 'U' && s[2] == '+') {
1552 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1553 PERL_SCAN_DISALLOW_PREFIX;
1556 uv = grok_hex(s, &len, &flags, NULL);
1558 goto NUM_ESCAPE_INSERT;
1560 res = newSVpvn(s + 1, e - s - 1);
1561 res = new_constant( Nullch, 0, "charnames",
1562 res, Nullsv, "\\N{...}" );
1564 sv_utf8_upgrade(res);
1565 str = SvPV(res,len);
1566 #ifdef EBCDIC_NEVER_MIND
1567 /* charnames uses pack U and that has been
1568 * recently changed to do the below uni->native
1569 * mapping, so this would be redundant (and wrong,
1570 * the code point would be doubly converted).
1571 * But leave this in just in case the pack U change
1572 * gets revoked, but the semantics is still
1573 * desireable for charnames. --jhi */
1575 UV uv = utf8_to_uvchr((U8*)str, 0);
1578 U8 tmpbuf[UTF8_MAXLEN+1], *d;
1580 d = uvchr_to_utf8(tmpbuf, UNI_TO_NATIVE(uv));
1581 sv_setpvn(res, (char *)tmpbuf, d - tmpbuf);
1582 str = SvPV(res, len);
1586 if (!has_utf8 && SvUTF8(res)) {
1587 char *ostart = SvPVX(sv);
1588 SvCUR_set(sv, d - ostart);
1591 sv_utf8_upgrade(sv);
1592 /* this just broke our allocation above... */
1593 SvGROW(sv, (STRLEN)(send - start));
1594 d = SvPVX(sv) + SvCUR(sv);
1597 if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */
1598 char *odest = SvPVX(sv);
1600 SvGROW(sv, (SvLEN(sv) + len - (e - s + 4)));
1601 d = SvPVX(sv) + (d - odest);
1603 Copy(str, d, len, char);
1610 yyerror("Missing braces on \\N{}");
1613 /* \c is a control character */
1622 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
1625 yyerror("Missing control char name in \\c");
1629 /* printf-style backslashes, formfeeds, newlines, etc */
1631 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
1634 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
1637 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
1640 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
1643 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
1646 *d++ = ASCII_TO_NEED(has_utf8,'\033');
1649 *d++ = ASCII_TO_NEED(has_utf8,'\007');
1655 } /* end if (backslash) */
1658 /* If we started with encoded form, or already know we want it
1659 and then encode the next character */
1660 if ((has_utf8 || this_utf8) && !NATIVE_IS_INVARIANT((U8)(*s))) {
1662 UV uv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
1663 STRLEN need = UNISKIP(NATIVE_TO_UNI(uv));
1666 /* encoded value larger than old, need extra space (NOTE: SvCUR() not set here) */
1667 STRLEN off = d - SvPVX(sv);
1668 d = SvGROW(sv, SvLEN(sv) + (need-len)) + off;
1670 d = (char*)uvchr_to_utf8((U8*)d, uv);
1674 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1676 } /* while loop to process each character */
1678 /* terminate the string and set up the sv */
1680 SvCUR_set(sv, d - SvPVX(sv));
1681 if (SvCUR(sv) >= SvLEN(sv))
1682 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
1685 if (PL_encoding && !has_utf8) {
1686 sv_recode_to_utf8(sv, PL_encoding);
1692 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1693 PL_sublex_info.sub_op->op_private |=
1694 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1698 /* shrink the sv if we allocated more than we used */
1699 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1700 SvLEN_set(sv, SvCUR(sv) + 1);
1701 Renew(SvPVX(sv), SvLEN(sv), char);
1704 /* return the substring (via yylval) only if we parsed anything */
1705 if (s > PL_bufptr) {
1706 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1707 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1709 ( PL_lex_inwhat == OP_TRANS
1711 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1714 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1721 * Returns TRUE if there's more to the expression (e.g., a subscript),
1724 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1726 * ->[ and ->{ return TRUE
1727 * { and [ outside a pattern are always subscripts, so return TRUE
1728 * if we're outside a pattern and it's not { or [, then return FALSE
1729 * if we're in a pattern and the first char is a {
1730 * {4,5} (any digits around the comma) returns FALSE
1731 * if we're in a pattern and the first char is a [
1733 * [SOMETHING] has a funky algorithm to decide whether it's a
1734 * character class or not. It has to deal with things like
1735 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1736 * anything else returns TRUE
1739 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1742 S_intuit_more(pTHX_ register char *s)
1744 if (PL_lex_brackets)
1746 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1748 if (*s != '{' && *s != '[')
1753 /* In a pattern, so maybe we have {n,m}. */
1770 /* On the other hand, maybe we have a character class */
1773 if (*s == ']' || *s == '^')
1776 /* this is terrifying, and it works */
1777 int weight = 2; /* let's weigh the evidence */
1779 unsigned char un_char = 255, last_un_char;
1780 char *send = strchr(s,']');
1781 char tmpbuf[sizeof PL_tokenbuf * 4];
1783 if (!send) /* has to be an expression */
1786 Zero(seen,256,char);
1789 else if (isDIGIT(*s)) {
1791 if (isDIGIT(s[1]) && s[2] == ']')
1797 for (; s < send; s++) {
1798 last_un_char = un_char;
1799 un_char = (unsigned char)*s;
1804 weight -= seen[un_char] * 10;
1805 if (isALNUM_lazy_if(s+1,UTF)) {
1806 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1807 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1812 else if (*s == '$' && s[1] &&
1813 strchr("[#!%*<>()-=",s[1])) {
1814 if (/*{*/ strchr("])} =",s[2]))
1823 if (strchr("wds]",s[1]))
1825 else if (seen['\''] || seen['"'])
1827 else if (strchr("rnftbxcav",s[1]))
1829 else if (isDIGIT(s[1])) {
1831 while (s[1] && isDIGIT(s[1]))
1841 if (strchr("aA01! ",last_un_char))
1843 if (strchr("zZ79~",s[1]))
1845 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1846 weight -= 5; /* cope with negative subscript */
1849 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1850 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1855 if (keyword(tmpbuf, d - tmpbuf))
1858 if (un_char == last_un_char + 1)
1860 weight -= seen[un_char];
1865 if (weight >= 0) /* probably a character class */
1875 * Does all the checking to disambiguate
1877 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1878 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1880 * First argument is the stuff after the first token, e.g. "bar".
1882 * Not a method if bar is a filehandle.
1883 * Not a method if foo is a subroutine prototyped to take a filehandle.
1884 * Not a method if it's really "Foo $bar"
1885 * Method if it's "foo $bar"
1886 * Not a method if it's really "print foo $bar"
1887 * Method if it's really "foo package::" (interpreted as package->foo)
1888 * Not a method if bar is known to be a subroutine ("sub bar; foo bar")
1889 * Not a method if bar is a filehandle or package, but is quoted with
1894 S_intuit_method(pTHX_ char *start, GV *gv)
1896 char *s = start + (*start == '$');
1897 char tmpbuf[sizeof PL_tokenbuf];
1905 if ((cv = GvCVu(gv))) {
1906 char *proto = SvPVX(cv);
1916 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1917 /* start is the beginning of the possible filehandle/object,
1918 * and s is the end of it
1919 * tmpbuf is a copy of it
1922 if (*start == '$') {
1923 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1928 return *s == '(' ? FUNCMETH : METHOD;
1930 if (!keyword(tmpbuf, len)) {
1931 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1936 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1937 if (indirgv && GvCVu(indirgv))
1939 /* filehandle or package name makes it a method */
1940 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1942 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1943 return 0; /* no assumptions -- "=>" quotes bearword */
1945 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1946 newSVpvn(tmpbuf,len));
1947 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1951 return *s == '(' ? FUNCMETH : METHOD;
1959 * Return a string of Perl code to load the debugger. If PERL5DB
1960 * is set, it will return the contents of that, otherwise a
1961 * compile-time require of perl5db.pl.
1968 char *pdb = PerlEnv_getenv("PERL5DB");
1972 SETERRNO(0,SS_NORMAL);
1973 return "BEGIN { require 'perl5db.pl' }";
1979 /* Encoded script support. filter_add() effectively inserts a
1980 * 'pre-processing' function into the current source input stream.
1981 * Note that the filter function only applies to the current source file
1982 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1984 * The datasv parameter (which may be NULL) can be used to pass
1985 * private data to this instance of the filter. The filter function
1986 * can recover the SV using the FILTER_DATA macro and use it to
1987 * store private buffers and state information.
1989 * The supplied datasv parameter is upgraded to a PVIO type
1990 * and the IoDIRP/IoANY field is used to store the function pointer,
1991 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1992 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1993 * private use must be set using malloc'd pointers.
1997 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
2002 if (!PL_rsfp_filters)
2003 PL_rsfp_filters = newAV();
2005 datasv = NEWSV(255,0);
2006 if (!SvUPGRADE(datasv, SVt_PVIO))
2007 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
2008 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
2009 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
2010 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
2011 (void*)funcp, SvPV_nolen(datasv)));
2012 av_unshift(PL_rsfp_filters, 1);
2013 av_store(PL_rsfp_filters, 0, datasv) ;
2018 /* Delete most recently added instance of this filter function. */
2020 Perl_filter_del(pTHX_ filter_t funcp)
2023 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", (void*)funcp));
2024 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
2026 /* if filter is on top of stack (usual case) just pop it off */
2027 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
2028 if (IoANY(datasv) == (void *)funcp) {
2029 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
2030 IoANY(datasv) = (void *)NULL;
2031 sv_free(av_pop(PL_rsfp_filters));
2035 /* we need to search for the correct entry and clear it */
2036 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
2040 /* Invoke the n'th filter function for the current rsfp. */
2042 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
2045 /* 0 = read one text line */
2050 if (!PL_rsfp_filters)
2052 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
2053 /* Provide a default input filter to make life easy. */
2054 /* Note that we append to the line. This is handy. */
2055 DEBUG_P(PerlIO_printf(Perl_debug_log,
2056 "filter_read %d: from rsfp\n", idx));
2060 int old_len = SvCUR(buf_sv) ;
2062 /* ensure buf_sv is large enough */
2063 SvGROW(buf_sv, (STRLEN)(old_len + maxlen)) ;
2064 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
2065 if (PerlIO_error(PL_rsfp))
2066 return -1; /* error */
2068 return 0 ; /* end of file */
2070 SvCUR_set(buf_sv, old_len + len) ;
2073 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2074 if (PerlIO_error(PL_rsfp))
2075 return -1; /* error */
2077 return 0 ; /* end of file */
2080 return SvCUR(buf_sv);
2082 /* Skip this filter slot if filter has been deleted */
2083 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
2084 DEBUG_P(PerlIO_printf(Perl_debug_log,
2085 "filter_read %d: skipped (filter deleted)\n",
2087 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
2089 /* Get function pointer hidden within datasv */
2090 funcp = (filter_t)IoANY(datasv);
2091 DEBUG_P(PerlIO_printf(Perl_debug_log,
2092 "filter_read %d: via function %p (%s)\n",
2093 idx, (void*)funcp, SvPV_nolen(datasv)));
2094 /* Call function. The function is expected to */
2095 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2096 /* Return: <0:error, =0:eof, >0:not eof */
2097 return (*funcp)(aTHX_ idx, buf_sv, maxlen);
2101 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2103 #ifdef PERL_CR_FILTER
2104 if (!PL_rsfp_filters) {
2105 filter_add(S_cr_textfilter,NULL);
2108 if (PL_rsfp_filters) {
2111 SvCUR_set(sv, 0); /* start with empty line */
2112 if (FILTER_READ(0, sv, 0) > 0)
2113 return ( SvPVX(sv) ) ;
2118 return (sv_gets(sv, fp, append));
2122 S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2126 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2130 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2131 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2133 return GvHV(gv); /* Foo:: */
2136 /* use constant CLASS => 'MyClass' */
2137 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2139 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2140 pkgname = SvPV_nolen(sv);
2144 return gv_stashpv(pkgname, FALSE);
2148 static char* exp_name[] =
2149 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2150 "ATTRTERM", "TERMBLOCK", "TERMORDORDOR"
2157 Works out what to call the token just pulled out of the input
2158 stream. The yacc parser takes care of taking the ops we return and
2159 stitching them into a tree.
2165 if read an identifier
2166 if we're in a my declaration
2167 croak if they tried to say my($foo::bar)
2168 build the ops for a my() declaration
2169 if it's an access to a my() variable
2170 are we in a sort block?
2171 croak if my($a); $a <=> $b
2172 build ops for access to a my() variable
2173 if in a dq string, and they've said @foo and we can't find @foo
2175 build ops for a bareword
2176 if we already built the token before, use it.
2179 #ifdef USE_PURE_BISON
2181 Perl_yylex_r(pTHX_ YYSTYPE *lvalp, int *lcharp)
2186 yylval_pointer[yyactlevel] = lvalp;
2187 yychar_pointer[yyactlevel] = lcharp;
2188 if (yyactlevel >= YYMAXLEVEL)
2189 Perl_croak(aTHX_ "panic: YYMAXLEVEL");
2191 r = Perl_yylex(aTHX);
2201 #pragma segment Perl_yylex
2213 I32 orig_keyword = 0;
2215 /* check if there's an identifier for us to look at */
2216 if (PL_pending_ident)
2217 return S_pending_ident(aTHX);
2219 /* no identifier pending identification */
2221 switch (PL_lex_state) {
2223 case LEX_NORMAL: /* Some compilers will produce faster */
2224 case LEX_INTERPNORMAL: /* code if we comment these out. */
2228 /* when we've already built the next token, just pull it out of the queue */
2231 yylval = PL_nextval[PL_nexttoke];
2233 PL_lex_state = PL_lex_defer;
2234 PL_expect = PL_lex_expect;
2235 PL_lex_defer = LEX_NORMAL;
2237 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2238 "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr,
2239 (IV)PL_nexttype[PL_nexttoke]); });
2241 return(PL_nexttype[PL_nexttoke]);
2243 /* interpolated case modifiers like \L \U, including \Q and \E.
2244 when we get here, PL_bufptr is at the \
2246 case LEX_INTERPCASEMOD:
2248 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2249 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2251 /* handle \E or end of string */
2252 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2256 if (PL_lex_casemods) {
2257 oldmod = PL_lex_casestack[--PL_lex_casemods];
2258 PL_lex_casestack[PL_lex_casemods] = '\0';
2260 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2262 PL_lex_state = LEX_INTERPCONCAT;
2266 if (PL_bufptr != PL_bufend)
2268 PL_lex_state = LEX_INTERPCONCAT;
2272 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2273 "### Saw case modifier at '%s'\n", PL_bufptr); });
2275 if (s[1] == '\\' && s[2] == 'E') {
2277 PL_lex_state = LEX_INTERPCONCAT;
2281 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2282 tmp = *s, *s = s[2], s[2] = (char)tmp; /* misordered... */
2283 if (strchr("LU", *s) &&
2284 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U'))) {
2285 PL_lex_casestack[--PL_lex_casemods] = '\0';
2288 if (PL_lex_casemods > 10)
2289 Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2290 PL_lex_casestack[PL_lex_casemods++] = *s;
2291 PL_lex_casestack[PL_lex_casemods] = '\0';
2292 PL_lex_state = LEX_INTERPCONCAT;
2293 PL_nextval[PL_nexttoke].ival = 0;
2296 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2298 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2300 PL_nextval[PL_nexttoke].ival = OP_LC;
2302 PL_nextval[PL_nexttoke].ival = OP_UC;
2304 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2306 Perl_croak(aTHX_ "panic: yylex");
2310 if (PL_lex_starts) {
2319 case LEX_INTERPPUSH:
2320 return sublex_push();
2322 case LEX_INTERPSTART:
2323 if (PL_bufptr == PL_bufend)
2324 return sublex_done();
2325 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2326 "### Interpolated variable at '%s'\n", PL_bufptr); });
2328 PL_lex_dojoin = (*PL_bufptr == '@');
2329 PL_lex_state = LEX_INTERPNORMAL;
2330 if (PL_lex_dojoin) {
2331 PL_nextval[PL_nexttoke].ival = 0;
2333 force_ident("\"", '$');
2334 PL_nextval[PL_nexttoke].ival = 0;
2336 PL_nextval[PL_nexttoke].ival = 0;
2338 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2341 if (PL_lex_starts++) {
2347 case LEX_INTERPENDMAYBE:
2348 if (intuit_more(PL_bufptr)) {
2349 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2355 if (PL_lex_dojoin) {
2356 PL_lex_dojoin = FALSE;
2357 PL_lex_state = LEX_INTERPCONCAT;
2360 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2361 && SvEVALED(PL_lex_repl))
2363 if (PL_bufptr != PL_bufend)
2364 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2365 PL_lex_repl = Nullsv;
2368 case LEX_INTERPCONCAT:
2370 if (PL_lex_brackets)
2371 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2373 if (PL_bufptr == PL_bufend)
2374 return sublex_done();
2376 if (SvIVX(PL_linestr) == '\'') {
2377 SV *sv = newSVsv(PL_linestr);
2380 else if ( PL_hints & HINT_NEW_RE )
2381 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2382 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2386 s = scan_const(PL_bufptr);
2388 PL_lex_state = LEX_INTERPCASEMOD;
2390 PL_lex_state = LEX_INTERPSTART;
2393 if (s != PL_bufptr) {
2394 PL_nextval[PL_nexttoke] = yylval;
2397 if (PL_lex_starts++)
2407 PL_lex_state = LEX_NORMAL;
2408 s = scan_formline(PL_bufptr);
2409 if (!PL_lex_formbrack)
2415 PL_oldoldbufptr = PL_oldbufptr;
2418 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2419 exp_name[PL_expect], s);
2425 if (isIDFIRST_lazy_if(s,UTF))
2427 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2430 goto fake_eof; /* emulate EOF on ^D or ^Z */
2435 if (PL_lex_brackets) {
2436 if (PL_lex_formbrack)
2437 yyerror("Format not terminated");
2439 yyerror("Missing right curly or square bracket");
2441 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2442 "### Tokener got EOF\n");
2446 if (s++ < PL_bufend)
2447 goto retry; /* ignore stray nulls */
2450 if (!PL_in_eval && !PL_preambled) {
2451 PL_preambled = TRUE;
2452 sv_setpv(PL_linestr,incl_perldb());
2453 if (SvCUR(PL_linestr))
2454 sv_catpv(PL_linestr,";");
2456 while(AvFILLp(PL_preambleav) >= 0) {
2457 SV *tmpsv = av_shift(PL_preambleav);
2458 sv_catsv(PL_linestr, tmpsv);
2459 sv_catpv(PL_linestr, ";");
2462 sv_free((SV*)PL_preambleav);
2463 PL_preambleav = NULL;
2465 if (PL_minus_n || PL_minus_p) {
2466 sv_catpv(PL_linestr, "LINE: while (<>) {");
2468 sv_catpv(PL_linestr,"chomp;");
2471 if (strchr("/'\"", *PL_splitstr)
2472 && strchr(PL_splitstr + 1, *PL_splitstr))
2473 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
2476 s = "'~#\200\1'"; /* surely one char is unused...*/
2477 while (s[1] && strchr(PL_splitstr, *s)) s++;
2479 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s%c",
2480 "q" + (delim == '\''), delim);
2481 for (s = PL_splitstr; *s; s++) {
2483 sv_catpvn(PL_linestr, "\\", 1);
2484 sv_catpvn(PL_linestr, s, 1);
2486 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2490 sv_catpv(PL_linestr,"our @F=split(' ');");
2493 sv_catpv(PL_linestr, "\n");
2494 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2495 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2496 PL_last_lop = PL_last_uni = Nullch;
2497 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2498 SV *sv = NEWSV(85,0);
2500 sv_upgrade(sv, SVt_PVMG);
2501 sv_setsv(sv,PL_linestr);
2504 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2509 bof = PL_rsfp ? TRUE : FALSE;
2510 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2513 if (PL_preprocess && !PL_in_eval)
2514 (void)PerlProc_pclose(PL_rsfp);
2515 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2516 PerlIO_clearerr(PL_rsfp);
2518 (void)PerlIO_close(PL_rsfp);
2520 PL_doextract = FALSE;
2522 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2523 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2524 sv_catpv(PL_linestr,";}");
2525 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2526 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2527 PL_last_lop = PL_last_uni = Nullch;
2528 PL_minus_n = PL_minus_p = 0;
2531 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2532 PL_last_lop = PL_last_uni = Nullch;
2533 sv_setpv(PL_linestr,"");
2534 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2536 /* if it looks like the start of a BOM, check if it in fact is */
2537 else if (bof && (!*s || *(U8*)s == 0xEF || *(U8*)s >= 0xFE)) {
2538 #ifdef PERLIO_IS_STDIO
2539 # ifdef __GNU_LIBRARY__
2540 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
2541 # define FTELL_FOR_PIPE_IS_BROKEN
2545 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2546 # define FTELL_FOR_PIPE_IS_BROKEN
2551 #ifdef FTELL_FOR_PIPE_IS_BROKEN
2552 /* This loses the possibility to detect the bof
2553 * situation on perl -P when the libc5 is being used.
2554 * Workaround? Maybe attach some extra state to PL_rsfp?
2557 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2559 bof = PerlIO_tell(PL_rsfp) == (Off_t)SvCUR(PL_linestr);
2562 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2563 s = swallow_bom((U8*)s);
2567 /* Incest with pod. */
2568 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2569 sv_setpv(PL_linestr, "");
2570 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2571 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2572 PL_last_lop = PL_last_uni = Nullch;
2573 PL_doextract = FALSE;
2577 } while (PL_doextract);
2578 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2579 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2580 SV *sv = NEWSV(85,0);
2582 sv_upgrade(sv, SVt_PVMG);
2583 sv_setsv(sv,PL_linestr);
2586 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2588 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2589 PL_last_lop = PL_last_uni = Nullch;
2590 if (CopLINE(PL_curcop) == 1) {
2591 while (s < PL_bufend && isSPACE(*s))
2593 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2597 if (*s == '#' && *(s+1) == '!')
2599 #ifdef ALTERNATE_SHEBANG
2601 static char as[] = ALTERNATE_SHEBANG;
2602 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2603 d = s + (sizeof(as) - 1);
2605 #endif /* ALTERNATE_SHEBANG */
2614 while (*d && !isSPACE(*d))
2618 #ifdef ARG_ZERO_IS_SCRIPT
2619 if (ipathend > ipath) {
2621 * HP-UX (at least) sets argv[0] to the script name,
2622 * which makes $^X incorrect. And Digital UNIX and Linux,
2623 * at least, set argv[0] to the basename of the Perl
2624 * interpreter. So, having found "#!", we'll set it right.
2626 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV)); /* $^X */
2627 assert(SvPOK(x) || SvGMAGICAL(x));
2628 if (sv_eq(x, CopFILESV(PL_curcop))) {
2629 sv_setpvn(x, ipath, ipathend - ipath);
2635 char *bstart = SvPV(CopFILESV(PL_curcop),blen);
2636 char *lstart = SvPV(x,llen);
2638 bstart += blen - llen;
2639 if (strnEQ(bstart, lstart, llen) && bstart[-1] == '/') {
2640 sv_setpvn(x, ipath, ipathend - ipath);
2645 TAINT_NOT; /* $^X is always tainted, but that's OK */
2647 #endif /* ARG_ZERO_IS_SCRIPT */
2652 d = instr(s,"perl -");
2654 d = instr(s,"perl");
2656 /* avoid getting into infinite loops when shebang
2657 * line contains "Perl" rather than "perl" */
2659 for (d = ipathend-4; d >= ipath; --d) {
2660 if ((*d == 'p' || *d == 'P')
2661 && !ibcmp(d, "perl", 4))
2671 #ifdef ALTERNATE_SHEBANG
2673 * If the ALTERNATE_SHEBANG on this system starts with a
2674 * character that can be part of a Perl expression, then if
2675 * we see it but not "perl", we're probably looking at the
2676 * start of Perl code, not a request to hand off to some
2677 * other interpreter. Similarly, if "perl" is there, but
2678 * not in the first 'word' of the line, we assume the line
2679 * contains the start of the Perl program.
2681 if (d && *s != '#') {
2683 while (*c && !strchr("; \t\r\n\f\v#", *c))
2686 d = Nullch; /* "perl" not in first word; ignore */
2688 *s = '#'; /* Don't try to parse shebang line */
2690 #endif /* ALTERNATE_SHEBANG */
2691 #ifndef MACOS_TRADITIONAL
2696 !instr(s,"indir") &&
2697 instr(PL_origargv[0],"perl"))
2703 while (s < PL_bufend && isSPACE(*s))
2705 if (s < PL_bufend) {
2706 Newz(899,newargv,PL_origargc+3,char*);
2708 while (s < PL_bufend && !isSPACE(*s))
2711 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2714 newargv = PL_origargv;
2717 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
2719 Perl_croak(aTHX_ "Can't exec %s", ipath);
2723 U32 oldpdb = PL_perldb;
2724 bool oldn = PL_minus_n;
2725 bool oldp = PL_minus_p;
2727 while (*d && !isSPACE(*d)) d++;
2728 while (SPACE_OR_TAB(*d)) d++;
2731 bool switches_done = PL_doswitches;
2733 if (*d == 'M' || *d == 'm') {
2735 while (*d && !isSPACE(*d)) d++;
2736 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2739 d = moreswitches(d);
2741 if (PL_doswitches && !switches_done) {
2742 int argc = PL_origargc;
2743 char **argv = PL_origargv;
2746 } while (argc && argv[0][0] == '-' && argv[0][1]);
2747 init_argv_symbols(argc,argv);
2749 if ((PERLDB_LINE && !oldpdb) ||
2750 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2751 /* if we have already added "LINE: while (<>) {",
2752 we must not do it again */
2754 sv_setpv(PL_linestr, "");
2755 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2756 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2757 PL_last_lop = PL_last_uni = Nullch;
2758 PL_preambled = FALSE;
2760 (void)gv_fetchfile(PL_origfilename);
2763 if (PL_doswitches && !switches_done) {
2764 int argc = PL_origargc;
2765 char **argv = PL_origargv;
2768 } while (argc && argv[0][0] == '-' && argv[0][1]);
2769 init_argv_symbols(argc,argv);
2775 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2777 PL_lex_state = LEX_FORMLINE;
2782 #ifdef PERL_STRICT_CR
2783 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2785 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2787 case ' ': case '\t': case '\f': case 013:
2788 #ifdef MACOS_TRADITIONAL
2795 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2796 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2797 /* handle eval qq[#line 1 "foo"\n ...] */
2798 CopLINE_dec(PL_curcop);
2802 while (s < d && *s != '\n')
2806 else if (s > d) /* Found by Ilya: feed random input to Perl. */
2807 Perl_croak(aTHX_ "panic: input overflow");
2809 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2811 PL_lex_state = LEX_FORMLINE;
2821 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2828 while (s < PL_bufend && SPACE_OR_TAB(*s))
2831 if (strnEQ(s,"=>",2)) {
2832 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2833 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2834 "### Saw unary minus before =>, forcing word '%s'\n", s);
2836 OPERATOR('-'); /* unary minus */
2838 PL_last_uni = PL_oldbufptr;
2840 case 'r': ftst = OP_FTEREAD; break;
2841 case 'w': ftst = OP_FTEWRITE; break;
2842 case 'x': ftst = OP_FTEEXEC; break;
2843 case 'o': ftst = OP_FTEOWNED; break;
2844 case 'R': ftst = OP_FTRREAD; break;
2845 case 'W': ftst = OP_FTRWRITE; break;
2846 case 'X': ftst = OP_FTREXEC; break;
2847 case 'O': ftst = OP_FTROWNED; break;
2848 case 'e': ftst = OP_FTIS; break;
2849 case 'z': ftst = OP_FTZERO; break;
2850 case 's': ftst = OP_FTSIZE; break;
2851 case 'f': ftst = OP_FTFILE; break;
2852 case 'd': ftst = OP_FTDIR; break;
2853 case 'l': ftst = OP_FTLINK; break;
2854 case 'p': ftst = OP_FTPIPE; break;
2855 case 'S': ftst = OP_FTSOCK; break;
2856 case 'u': ftst = OP_FTSUID; break;
2857 case 'g': ftst = OP_FTSGID; break;
2858 case 'k': ftst = OP_FTSVTX; break;
2859 case 'b': ftst = OP_FTBLK; break;
2860 case 'c': ftst = OP_FTCHR; break;
2861 case 't': ftst = OP_FTTTY; break;
2862 case 'T': ftst = OP_FTTEXT; break;
2863 case 'B': ftst = OP_FTBINARY; break;
2864 case 'M': case 'A': case 'C':
2865 gv_fetchpv("\024",TRUE, SVt_PV);
2867 case 'M': ftst = OP_FTMTIME; break;
2868 case 'A': ftst = OP_FTATIME; break;
2869 case 'C': ftst = OP_FTCTIME; break;
2877 PL_last_lop_op = (OPCODE)ftst;
2878 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2879 "### Saw file test %c\n", (int)ftst);
2884 /* Assume it was a minus followed by a one-letter named
2885 * subroutine call (or a -bareword), then. */
2886 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2887 "### '-%c' looked like a file test but was not\n",
2896 if (PL_expect == XOPERATOR)
2901 else if (*s == '>') {
2904 if (isIDFIRST_lazy_if(s,UTF)) {
2905 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2913 if (PL_expect == XOPERATOR)
2916 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2918 OPERATOR('-'); /* unary minus */
2925 if (PL_expect == XOPERATOR)
2930 if (PL_expect == XOPERATOR)
2933 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2939 if (PL_expect != XOPERATOR) {
2940 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2941 PL_expect = XOPERATOR;
2942 force_ident(PL_tokenbuf, '*');
2955 if (PL_expect == XOPERATOR) {
2959 PL_tokenbuf[0] = '%';
2960 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2961 if (!PL_tokenbuf[1]) {
2964 PL_pending_ident = '%';
2983 switch (PL_expect) {
2986 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2988 PL_bufptr = s; /* update in case we back off */
2994 PL_expect = XTERMBLOCK;
2998 while (isIDFIRST_lazy_if(s,UTF)) {
2999 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3000 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
3001 if (tmp < 0) tmp = -tmp;
3017 d = scan_str(d,TRUE,TRUE);
3019 /* MUST advance bufptr here to avoid bogus
3020 "at end of line" context messages from yyerror().
3022 PL_bufptr = s + len;
3023 yyerror("Unterminated attribute parameter in attribute list");
3026 return 0; /* EOF indicator */
3030 SV *sv = newSVpvn(s, len);
3031 sv_catsv(sv, PL_lex_stuff);
3032 attrs = append_elem(OP_LIST, attrs,
3033 newSVOP(OP_CONST, 0, sv));
3034 SvREFCNT_dec(PL_lex_stuff);
3035 PL_lex_stuff = Nullsv;
3038 /* NOTE: any CV attrs applied here need to be part of
3039 the CVf_BUILTIN_ATTRS define in cv.h! */
3040 if (!PL_in_my && len == 6 && strnEQ(s, "lvalue", len))
3041 CvLVALUE_on(PL_compcv);
3042 else if (!PL_in_my && len == 6 && strnEQ(s, "locked", len))
3043 CvLOCKED_on(PL_compcv);
3044 else if (!PL_in_my && len == 6 && strnEQ(s, "method", len))
3045 CvMETHOD_on(PL_compcv);
3046 else if (!PL_in_my && len == 9 && strnEQ(s, "assertion", len))
3047 CvASSERTION_on(PL_compcv);
3048 else if (PL_in_my == KEY_our && len == 6 &&
3049 strnEQ(s, "unique", len))
3051 GvUNIQUE_on(cGVOPx_gv(yylval.opval));
3053 ; /* skip that case to avoid loading attributes.pm */
3055 /* After we've set the flags, it could be argued that
3056 we don't need to do the attributes.pm-based setting
3057 process, and shouldn't bother appending recognized
3058 flags. To experiment with that, uncomment the
3059 following "else". (Note that's already been
3060 uncommented. That keeps the above-applied built-in
3061 attributes from being intercepted (and possibly
3062 rejected) by a package's attribute routines, but is
3063 justified by the performance win for the common case
3064 of applying only built-in attributes.) */
3066 attrs = append_elem(OP_LIST, attrs,
3067 newSVOP(OP_CONST, 0,
3071 if (*s == ':' && s[1] != ':')
3074 break; /* require real whitespace or :'s */
3076 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3077 if (*s != ';' && *s != '}' && *s != tmp && (tmp != '=' || *s != ')')) {
3078 char q = ((*s == '\'') ? '"' : '\'');
3079 /* If here for an expression, and parsed no attrs, back off. */
3080 if (tmp == '=' && !attrs) {
3084 /* MUST advance bufptr here to avoid bogus "at end of line"
3085 context messages from yyerror().
3089 yyerror("Unterminated attribute list");
3091 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
3099 PL_nextval[PL_nexttoke].opval = attrs;
3107 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3108 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3125 if (PL_lex_brackets <= 0)
3126 yyerror("Unmatched right square bracket");
3129 if (PL_lex_state == LEX_INTERPNORMAL) {
3130 if (PL_lex_brackets == 0) {
3131 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3132 PL_lex_state = LEX_INTERPEND;
3139 if (PL_lex_brackets > 100) {
3140 Renew(PL_lex_brackstack, PL_lex_brackets + 10, char);
3142 switch (PL_expect) {
3144 if (PL_lex_formbrack) {
3148 if (PL_oldoldbufptr == PL_last_lop)
3149 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3151 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3152 OPERATOR(HASHBRACK);
3154 while (s < PL_bufend && SPACE_OR_TAB(*s))
3157 PL_tokenbuf[0] = '\0';
3158 if (d < PL_bufend && *d == '-') {
3159 PL_tokenbuf[0] = '-';
3161 while (d < PL_bufend && SPACE_OR_TAB(*d))
3164 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3165 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3167 while (d < PL_bufend && SPACE_OR_TAB(*d))
3170 char minus = (PL_tokenbuf[0] == '-');
3171 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3179 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3184 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3189 if (PL_oldoldbufptr == PL_last_lop)
3190 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3192 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3195 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
3197 /* This hack is to get the ${} in the message. */
3199 yyerror("syntax error");
3202 OPERATOR(HASHBRACK);
3204 /* This hack serves to disambiguate a pair of curlies
3205 * as being a block or an anon hash. Normally, expectation
3206 * determines that, but in cases where we're not in a
3207 * position to expect anything in particular (like inside
3208 * eval"") we have to resolve the ambiguity. This code
3209 * covers the case where the first term in the curlies is a
3210 * quoted string. Most other cases need to be explicitly
3211 * disambiguated by prepending a `+' before the opening
3212 * curly in order to force resolution as an anon hash.
3214 * XXX should probably propagate the outer expectation
3215 * into eval"" to rely less on this hack, but that could
3216 * potentially break current behavior of eval"".
3220 if (*s == '\'' || *s == '"' || *s == '`') {
3221 /* common case: get past first string, handling escapes */
3222 for (t++; t < PL_bufend && *t != *s;)
3223 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3227 else if (*s == 'q') {
3230 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3233 /* skip q//-like construct */
3235 char open, close, term;
3238 while (t < PL_bufend && isSPACE(*t))
3240 /* check for q => */
3241 if (t+1 < PL_bufend && t[0] == '=' && t[1] == '>') {
3242 OPERATOR(HASHBRACK);
3246 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3250 for (t++; t < PL_bufend; t++) {
3251 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3253 else if (*t == open)
3257 for (t++; t < PL_bufend; t++) {
3258 if (*t == '\\' && t+1 < PL_bufend)
3260 else if (*t == close && --brackets <= 0)
3262 else if (*t == open)
3269 /* skip plain q word */
3270 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3273 else if (isALNUM_lazy_if(t,UTF)) {
3275 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3278 while (t < PL_bufend && isSPACE(*t))
3280 /* if comma follows first term, call it an anon hash */
3281 /* XXX it could be a comma expression with loop modifiers */
3282 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3283 || (*t == '=' && t[1] == '>')))
3284 OPERATOR(HASHBRACK);
3285 if (PL_expect == XREF)
3288 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3294 yylval.ival = CopLINE(PL_curcop);
3295 if (isSPACE(*s) || *s == '#')
3296 PL_copline = NOLINE; /* invalidate current command line number */
3301 if (PL_lex_brackets <= 0)
3302 yyerror("Unmatched right curly bracket");
3304 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3305 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3306 PL_lex_formbrack = 0;
3307 if (PL_lex_state == LEX_INTERPNORMAL) {
3308 if (PL_lex_brackets == 0) {
3309 if (PL_expect & XFAKEBRACK) {
3310 PL_expect &= XENUMMASK;
3311 PL_lex_state = LEX_INTERPEND;
3313 return yylex(); /* ignore fake brackets */
3315 if (*s == '-' && s[1] == '>')
3316 PL_lex_state = LEX_INTERPENDMAYBE;
3317 else if (*s != '[' && *s != '{')
3318 PL_lex_state = LEX_INTERPEND;
3321 if (PL_expect & XFAKEBRACK) {
3322 PL_expect &= XENUMMASK;
3324 return yylex(); /* ignore fake brackets */
3334 if (PL_expect == XOPERATOR) {
3335 if (ckWARN(WARN_SEMICOLON)
3336 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3338 CopLINE_dec(PL_curcop);
3339 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
3340 CopLINE_inc(PL_curcop);
3345 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3347 PL_expect = XOPERATOR;
3348 force_ident(PL_tokenbuf, '&');
3352 yylval.ival = (OPpENTERSUB_AMPER<<8);
3371 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3372 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Reversed %c= operator",(int)tmp);
3374 if (PL_expect == XSTATE && isALPHA(tmp) &&
3375 (s == PL_linestart+1 || s[-2] == '\n') )
3377 if (PL_in_eval && !PL_rsfp) {
3382 if (strnEQ(s,"=cut",4)) {
3396 PL_doextract = TRUE;
3399 if (PL_lex_brackets < PL_lex_formbrack) {
3401 #ifdef PERL_STRICT_CR
3402 for (t = s; SPACE_OR_TAB(*t); t++) ;
3404 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3406 if (*t == '\n' || *t == '#') {
3424 if (PL_expect != XOPERATOR) {
3425 if (s[1] != '<' && !strchr(s,'>'))
3428 s = scan_heredoc(s);
3430 s = scan_inputsymbol(s);
3431 TERM(sublex_start());
3436 SHop(OP_LEFT_SHIFT);
3450 SHop(OP_RIGHT_SHIFT);
3459 if (PL_expect == XOPERATOR) {
3460 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3463 return ','; /* grandfather non-comma-format format */
3467 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3468 PL_tokenbuf[0] = '@';
3469 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3470 sizeof PL_tokenbuf - 1, FALSE);
3471 if (PL_expect == XOPERATOR)
3472 no_op("Array length", s);
3473 if (!PL_tokenbuf[1])
3475 PL_expect = XOPERATOR;
3476 PL_pending_ident = '#';
3480 PL_tokenbuf[0] = '$';
3481 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3482 sizeof PL_tokenbuf - 1, FALSE);
3483 if (PL_expect == XOPERATOR)
3485 if (!PL_tokenbuf[1]) {
3487 yyerror("Final $ should be \\$ or $name");
3491 /* This kludge not intended to be bulletproof. */
3492 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3493 yylval.opval = newSVOP(OP_CONST, 0,
3494 newSViv(PL_compiling.cop_arybase));
3495 yylval.opval->op_private = OPpCONST_ARYBASE;
3501 if (PL_lex_state == LEX_NORMAL)
3504 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3507 PL_tokenbuf[0] = '@';
3508 if (ckWARN(WARN_SYNTAX)) {
3510 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3513 PL_bufptr = skipspace(PL_bufptr);
3514 while (t < PL_bufend && *t != ']')
3516 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3517 "Multidimensional syntax %.*s not supported",
3518 (t - PL_bufptr) + 1, PL_bufptr);
3522 else if (*s == '{') {
3523 PL_tokenbuf[0] = '%';
3524 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3525 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3527 char tmpbuf[sizeof PL_tokenbuf];
3529 for (t++; isSPACE(*t); t++) ;
3530 if (isIDFIRST_lazy_if(t,UTF)) {
3531 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3532 for (; isSPACE(*t); t++) ;
3533 if (*t == ';' && get_cv(tmpbuf, FALSE))
3534 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3535 "You need to quote \"%s\"", tmpbuf);
3541 PL_expect = XOPERATOR;
3542 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3543 bool islop = (PL_last_lop == PL_oldoldbufptr);
3544 if (!islop || PL_last_lop_op == OP_GREPSTART)
3545 PL_expect = XOPERATOR;
3546 else if (strchr("$@\"'`q", *s))
3547 PL_expect = XTERM; /* e.g. print $fh "foo" */
3548 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3549 PL_expect = XTERM; /* e.g. print $fh &sub */
3550 else if (isIDFIRST_lazy_if(s,UTF)) {
3551 char tmpbuf[sizeof PL_tokenbuf];
3552 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3553 if ((tmp = keyword(tmpbuf, len))) {
3554 /* binary operators exclude handle interpretations */
3566 PL_expect = XTERM; /* e.g. print $fh length() */
3571 PL_expect = XTERM; /* e.g. print $fh subr() */
3574 else if (isDIGIT(*s))
3575 PL_expect = XTERM; /* e.g. print $fh 3 */
3576 else if (*s == '.' && isDIGIT(s[1]))
3577 PL_expect = XTERM; /* e.g. print $fh .3 */
3578 else if (strchr("?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3579 PL_expect = XTERM; /* e.g. print $fh -1 */
3580 else if (*s == '/' && !isSPACE(s[1]) && s[1] != '=' && s[1] != '/')
3581 PL_expect = XTERM; /* e.g. print $fh /.../
3582 XXX except DORDOR operator */
3583 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3584 PL_expect = XTERM; /* print $fh <<"EOF" */
3586 PL_pending_ident = '$';
3590 if (PL_expect == XOPERATOR)
3592 PL_tokenbuf[0] = '@';
3593 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3594 if (!PL_tokenbuf[1]) {
3597 if (PL_lex_state == LEX_NORMAL)
3599 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3601 PL_tokenbuf[0] = '%';
3603 /* Warn about @ where they meant $. */
3604 if (ckWARN(WARN_SYNTAX)) {
3605 if (*s == '[' || *s == '{') {
3607 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3609 if (*t == '}' || *t == ']') {
3611 PL_bufptr = skipspace(PL_bufptr);
3612 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3613 "Scalar value %.*s better written as $%.*s",
3614 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3619 PL_pending_ident = '@';
3622 case '/': /* may be division, defined-or, or pattern */
3623 if (PL_expect == XTERMORDORDOR && s[1] == '/') {
3627 case '?': /* may either be conditional or pattern */
3628 if(PL_expect == XOPERATOR) {
3636 /* A // operator. */
3646 /* Disable warning on "study /blah/" */
3647 if (PL_oldoldbufptr == PL_last_uni
3648 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3649 || memNE(PL_last_uni, "study", 5)
3650 || isALNUM_lazy_if(PL_last_uni+5,UTF)
3653 s = scan_pat(s,OP_MATCH);
3654 TERM(sublex_start());
3658 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3659 #ifdef PERL_STRICT_CR
3662 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3664 && (s == PL_linestart || s[-1] == '\n') )
3666 PL_lex_formbrack = 0;
3670 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3676 yylval.ival = OPf_SPECIAL;
3682 if (PL_expect != XOPERATOR)
3687 case '0': case '1': case '2': case '3': case '4':
3688 case '5': case '6': case '7': case '8': case '9':
3689 s = scan_num(s, &yylval);
3690 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3691 "### Saw number in '%s'\n", s);
3693 if (PL_expect == XOPERATOR)
3698 s = scan_str(s,FALSE,FALSE);
3699 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3700 "### Saw string before '%s'\n", s);
3702 if (PL_expect == XOPERATOR) {
3703 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3706 return ','; /* grandfather non-comma-format format */
3712 missingterm((char*)0);
3713 yylval.ival = OP_CONST;
3714 TERM(sublex_start());
3717 s = scan_str(s,FALSE,FALSE);
3718 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3719 "### Saw string before '%s'\n", s);
3721 if (PL_expect == XOPERATOR) {
3722 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3725 return ','; /* grandfather non-comma-format format */
3731 missingterm((char*)0);
3732 yylval.ival = OP_CONST;
3733 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3734 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
3735 yylval.ival = OP_STRINGIFY;
3739 TERM(sublex_start());
3742 s = scan_str(s,FALSE,FALSE);
3743 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3744 "### Saw backtick string before '%s'\n", s);
3746 if (PL_expect == XOPERATOR)
3747 no_op("Backticks",s);
3749 missingterm((char*)0);
3750 yylval.ival = OP_BACKTICK;
3752 TERM(sublex_start());
3756 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3757 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),"Can't use \\%c to mean $%c in expression",
3759 if (PL_expect == XOPERATOR)
3760 no_op("Backslash",s);
3764 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3768 while (isDIGIT(*start) || *start == '_')
3770 if (*start == '.' && isDIGIT(start[1])) {
3771 s = scan_num(s, &yylval);
3774 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3775 else if (!isALPHA(*start) && (PL_expect == XTERM
3776 || PL_expect == XREF || PL_expect == XSTATE
3777 || PL_expect == XTERMORDORDOR)) {
3781 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3784 s = scan_num(s, &yylval);
3791 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3831 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3833 /* Some keywords can be followed by any delimiter, including ':' */
3834 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3835 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3836 (PL_tokenbuf[0] == 'q' &&
3837 strchr("qwxr", PL_tokenbuf[1])))));
3839 /* x::* is just a word, unless x is "CORE" */
3840 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3844 while (d < PL_bufend && isSPACE(*d))
3845 d++; /* no comments skipped here, or s### is misparsed */
3847 /* Is this a label? */
3848 if (!tmp && PL_expect == XSTATE
3849 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3851 yylval.pval = savepv(PL_tokenbuf);
3856 /* Check for keywords */
3857 tmp = keyword(PL_tokenbuf, len);
3859 /* Is this a word before a => operator? */
3860 if (*d == '=' && d[1] == '>') {
3862 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3863 yylval.opval->op_private = OPpCONST_BARE;
3864 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
3865 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3869 if (tmp < 0) { /* second-class keyword? */
3870 GV *ogv = Nullgv; /* override (winner) */
3871 GV *hgv = Nullgv; /* hidden (loser) */
3872 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3874 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3877 if (GvIMPORTED_CV(gv))
3879 else if (! CvMETHOD(cv))
3883 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3884 (gv = *gvp) != (GV*)&PL_sv_undef &&
3885 GvCVu(gv) && GvIMPORTED_CV(gv))
3892 tmp = 0; /* overridden by import or by GLOBAL */
3895 && -tmp==KEY_lock /* XXX generalizable kludge */
3897 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3899 tmp = 0; /* any sub overrides "weak" keyword */
3901 else { /* no override */
3903 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
3904 Perl_warner(aTHX_ packWARN(WARN_MISC),
3905 "dump() better written as CORE::dump()");
3909 if (ckWARN(WARN_AMBIGUOUS) && hgv
3910 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3911 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
3912 "Ambiguous call resolved as CORE::%s(), %s",
3913 GvENAME(hgv), "qualify as such or use &");
3920 default: /* not a keyword */
3924 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3926 /* Get the rest if it looks like a package qualifier */
3928 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3930 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3933 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3934 *s == '\'' ? "'" : "::");
3939 if (PL_expect == XOPERATOR) {
3940 if (PL_bufptr == PL_linestart) {
3941 CopLINE_dec(PL_curcop);
3942 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
3943 CopLINE_inc(PL_curcop);
3946 no_op("Bareword",s);
3949 /* Look for a subroutine with this name in current package,
3950 unless name is "Foo::", in which case Foo is a bearword
3951 (and a package name). */
3954 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3956 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3957 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
3958 "Bareword \"%s\" refers to nonexistent package",
3961 PL_tokenbuf[len] = '\0';
3968 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3971 /* if we saw a global override before, get the right name */
3974 sv = newSVpvn("CORE::GLOBAL::",14);
3975 sv_catpv(sv,PL_tokenbuf);
3978 sv = newSVpv(PL_tokenbuf,0);
3980 /* Presume this is going to be a bareword of some sort. */
3983 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3984 yylval.opval->op_private = OPpCONST_BARE;
3985 /* UTF-8 package name? */
3986 if (UTF && !IN_BYTES &&
3987 is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
3990 /* And if "Foo::", then that's what it certainly is. */
3995 /* See if it's the indirect object for a list operator. */
3997 if (PL_oldoldbufptr &&
3998 PL_oldoldbufptr < PL_bufptr &&
3999 (PL_oldoldbufptr == PL_last_lop
4000 || PL_oldoldbufptr == PL_last_uni) &&
4001 /* NO SKIPSPACE BEFORE HERE! */
4002 (PL_expect == XREF ||
4003 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
4005 bool immediate_paren = *s == '(';
4007 /* (Now we can afford to cross potential line boundary.) */
4010 /* Two barewords in a row may indicate method call. */
4012 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
4015 /* If not a declared subroutine, it's an indirect object. */
4016 /* (But it's an indir obj regardless for sort.) */
4018 if ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
4019 ((!gv || !GvCVu(gv)) &&
4020 (PL_last_lop_op != OP_MAPSTART &&
4021 PL_last_lop_op != OP_GREPSTART))))
4023 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
4028 PL_expect = XOPERATOR;
4031 /* Is this a word before a => operator? */
4032 if (*s == '=' && s[1] == '>' && !pkgname) {
4034 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
4035 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
4036 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
4040 /* If followed by a paren, it's certainly a subroutine. */
4043 if (gv && GvCVu(gv)) {
4044 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
4045 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
4050 PL_nextval[PL_nexttoke].opval = yylval.opval;
4051 PL_expect = XOPERATOR;
4057 /* If followed by var or block, call it a method (unless sub) */
4059 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
4060 PL_last_lop = PL_oldbufptr;
4061 PL_last_lop_op = OP_METHOD;
4065 /* If followed by a bareword, see if it looks like indir obj. */
4068 && (isIDFIRST_lazy_if(s,UTF) || *s == '$')
4069 && (tmp = intuit_method(s,gv)))
4072 /* Not a method, so call it a subroutine (if defined) */
4074 if (gv && GvCVu(gv)) {
4076 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
4077 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4078 "Ambiguous use of -%s resolved as -&%s()",
4079 PL_tokenbuf, PL_tokenbuf);
4080 /* Check for a constant sub */
4082 if ((sv = cv_const_sv(cv))) {
4084 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
4085 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
4086 yylval.opval->op_private = 0;
4090 /* Resolve to GV now. */
4091 op_free(yylval.opval);
4092 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
4093 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
4094 PL_last_lop = PL_oldbufptr;
4095 PL_last_lop_op = OP_ENTERSUB;
4096 /* Is there a prototype? */
4099 char *proto = SvPV((SV*)cv, len);
4102 if (strEQ(proto, "$"))
4104 while (*proto == ';')
4106 if (*proto == '&' && *s == '{') {
4107 sv_setpv(PL_subname, PL_curstash ?
4108 "__ANON__" : "__ANON__::__ANON__");
4112 PL_nextval[PL_nexttoke].opval = yylval.opval;
4118 /* Call it a bare word */
4120 if (PL_hints & HINT_STRICT_SUBS)
4121 yylval.opval->op_private |= OPpCONST_STRICT;
4124 if (ckWARN(WARN_RESERVED)) {
4125 if (lastchar != '-') {
4126 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4127 if (!*d && !gv_stashpv(PL_tokenbuf,FALSE))
4128 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
4135 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
4136 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4137 "Operator or semicolon missing before %c%s",
4138 lastchar, PL_tokenbuf);
4139 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4140 "Ambiguous use of %c resolved as operator %c",
4141 lastchar, lastchar);
4147 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4148 newSVpv(CopFILE(PL_curcop),0));
4152 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4153 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4156 case KEY___PACKAGE__:
4157 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4159 ? newSVsv(PL_curstname)
4168 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4169 char *pname = "main";
4170 if (PL_tokenbuf[2] == 'D')
4171 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
4172 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4175 GvIOp(gv) = newIO();
4176 IoIFP(GvIOp(gv)) = PL_rsfp;
4177 #if defined(HAS_FCNTL) && defined(F_SETFD)
4179 int fd = PerlIO_fileno(PL_rsfp);
4180 fcntl(fd,F_SETFD,fd >= 3);
4183 /* Mark this internal pseudo-handle as clean */
4184 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4186 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4187 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4188 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4190 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4191 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4192 /* if the script was opened in binmode, we need to revert
4193 * it to text mode for compatibility; but only iff it has CRs
4194 * XXX this is a questionable hack at best. */
4195 if (PL_bufend-PL_bufptr > 2
4196 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4199 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4200 loc = PerlIO_tell(PL_rsfp);
4201 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4204 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
4206 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4207 #endif /* NETWARE */
4208 #ifdef PERLIO_IS_STDIO /* really? */
4209 # if defined(__BORLANDC__)
4210 /* XXX see note in do_binmode() */
4211 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
4215 PerlIO_seek(PL_rsfp, loc, 0);
4219 #ifdef PERLIO_LAYERS
4222 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
4223 else if (PL_encoding) {
4230 XPUSHs(PL_encoding);
4232 call_method("name", G_SCALAR);
4236 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL,
4237 Perl_form(aTHX_ ":encoding(%"SVf")",
4255 if (PL_expect == XSTATE) {
4262 if (*s == ':' && s[1] == ':') {
4265 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4266 if (!(tmp = keyword(PL_tokenbuf, len)))
4267 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4281 LOP(OP_ACCEPT,XTERM);
4287 LOP(OP_ATAN2,XTERM);
4293 LOP(OP_BINMODE,XTERM);
4296 LOP(OP_BLESS,XTERM);
4305 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4322 if (!PL_cryptseen) {
4323 PL_cryptseen = TRUE;
4327 LOP(OP_CRYPT,XTERM);
4330 LOP(OP_CHMOD,XTERM);
4333 LOP(OP_CHOWN,XTERM);
4336 LOP(OP_CONNECT,XTERM);
4352 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4356 PL_hints |= HINT_BLOCK_SCOPE;
4366 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4367 LOP(OP_DBMOPEN,XTERM);
4373 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4380 yylval.ival = CopLINE(PL_curcop);
4394 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4395 UNIBRACK(OP_ENTEREVAL);
4413 case KEY_endhostent:
4419 case KEY_endservent:
4422 case KEY_endprotoent:
4433 yylval.ival = CopLINE(PL_curcop);
4435 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4437 if ((PL_bufend - p) >= 3 &&
4438 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4440 else if ((PL_bufend - p) >= 4 &&
4441 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4444 if (isIDFIRST_lazy_if(p,UTF)) {
4445 p = scan_ident(p, PL_bufend,
4446 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4450 Perl_croak(aTHX_ "Missing $ on loop variable");
4455 LOP(OP_FORMLINE,XTERM);
4461 LOP(OP_FCNTL,XTERM);
4467 LOP(OP_FLOCK,XTERM);
4476 LOP(OP_GREPSTART, XREF);
4479 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4494 case KEY_getpriority:
4495 LOP(OP_GETPRIORITY,XTERM);
4497 case KEY_getprotobyname:
4500 case KEY_getprotobynumber:
4501 LOP(OP_GPBYNUMBER,XTERM);
4503 case KEY_getprotoent:
4515 case KEY_getpeername:
4516 UNI(OP_GETPEERNAME);
4518 case KEY_gethostbyname:
4521 case KEY_gethostbyaddr:
4522 LOP(OP_GHBYADDR,XTERM);
4524 case KEY_gethostent:
4527 case KEY_getnetbyname:
4530 case KEY_getnetbyaddr:
4531 LOP(OP_GNBYADDR,XTERM);
4536 case KEY_getservbyname:
4537 LOP(OP_GSBYNAME,XTERM);
4539 case KEY_getservbyport:
4540 LOP(OP_GSBYPORT,XTERM);
4542 case KEY_getservent:
4545 case KEY_getsockname:
4546 UNI(OP_GETSOCKNAME);
4548 case KEY_getsockopt:
4549 LOP(OP_GSOCKOPT,XTERM);
4571 yylval.ival = CopLINE(PL_curcop);
4575 LOP(OP_INDEX,XTERM);
4581 LOP(OP_IOCTL,XTERM);
4593 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4625 LOP(OP_LISTEN,XTERM);
4634 s = scan_pat(s,OP_MATCH);
4635 TERM(sublex_start());
4638 LOP(OP_MAPSTART, XREF);
4641 LOP(OP_MKDIR,XTERM);
4644 LOP(OP_MSGCTL,XTERM);
4647 LOP(OP_MSGGET,XTERM);
4650 LOP(OP_MSGRCV,XTERM);
4653 LOP(OP_MSGSND,XTERM);
4659 if (isIDFIRST_lazy_if(s,UTF)) {
4660 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4661 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4663 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4664 if (!PL_in_my_stash) {
4667 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4675 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4682 if (PL_expect != XSTATE)
4683 yyerror("\"no\" not allowed in expression");
4684 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4685 s = force_version(s, FALSE);
4690 if (*s == '(' || (s = skipspace(s), *s == '('))
4697 if (isIDFIRST_lazy_if(s,UTF)) {
4699 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4701 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)
4703 && !(t[0] == '=' && t[1] == '>')
4705 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
4706 "Precedence problem: open %.*s should be open(%.*s)",
4707 d - s, s, d - s, s);
4713 yylval.ival = OP_OR;
4723 LOP(OP_OPEN_DIR,XTERM);
4726 checkcomma(s,PL_tokenbuf,"filehandle");
4730 checkcomma(s,PL_tokenbuf,"filehandle");
4749 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4753 LOP(OP_PIPE_OP,XTERM);
4756 s = scan_str(s,FALSE,FALSE);
4758 missingterm((char*)0);
4759 yylval.ival = OP_CONST;
4760 TERM(sublex_start());
4766 s = scan_str(s,FALSE,FALSE);
4768 missingterm((char*)0);
4770 if (SvCUR(PL_lex_stuff)) {
4773 d = SvPV_force(PL_lex_stuff, len);
4776 for (; isSPACE(*d) && len; --len, ++d) ;
4779 if (!warned && ckWARN(WARN_QW)) {
4780 for (; !isSPACE(*d) && len; --len, ++d) {
4782 Perl_warner(aTHX_ packWARN(WARN_QW),
4783 "Possible attempt to separate words with commas");
4786 else if (*d == '#') {
4787 Perl_warner(aTHX_ packWARN(WARN_QW),
4788 "Possible attempt to put comments in qw() list");
4794 for (; !isSPACE(*d) && len; --len, ++d) ;
4796 sv = newSVpvn(b, d-b);
4797 if (DO_UTF8(PL_lex_stuff))
4799 words = append_elem(OP_LIST, words,
4800 newSVOP(OP_CONST, 0, tokeq(sv)));
4804 PL_nextval[PL_nexttoke].opval = words;
4809 SvREFCNT_dec(PL_lex_stuff);
4810 PL_lex_stuff = Nullsv;
4816 s = scan_str(s,FALSE,FALSE);
4818 missingterm((char*)0);
4819 yylval.ival = OP_STRINGIFY;
4820 if (SvIVX(PL_lex_stuff) == '\'')
4821 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4822 TERM(sublex_start());
4825 s = scan_pat(s,OP_QR);
4826 TERM(sublex_start());
4829 s = scan_str(s,FALSE,FALSE);
4831 missingterm((char*)0);
4832 yylval.ival = OP_BACKTICK;
4834 TERM(sublex_start());
4842 s = force_version(s, FALSE);
4844 else if (*s != 'v' || !isDIGIT(s[1])
4845 || (s = force_version(s, TRUE), *s == 'v'))
4847 *PL_tokenbuf = '\0';
4848 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4849 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4850 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4852 yyerror("<> should be quotes");
4860 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4864 LOP(OP_RENAME,XTERM);
4873 LOP(OP_RINDEX,XTERM);
4883 UNIDOR(OP_READLINE);
4896 LOP(OP_REVERSE,XTERM);
4899 UNIDOR(OP_READLINK);
4907 TERM(sublex_start());
4909 TOKEN(1); /* force error */
4918 LOP(OP_SELECT,XTERM);
4924 LOP(OP_SEMCTL,XTERM);
4927 LOP(OP_SEMGET,XTERM);
4930 LOP(OP_SEMOP,XTERM);
4936 LOP(OP_SETPGRP,XTERM);
4938 case KEY_setpriority:
4939 LOP(OP_SETPRIORITY,XTERM);
4941 case KEY_sethostent:
4947 case KEY_setservent:
4950 case KEY_setprotoent:
4960 LOP(OP_SEEKDIR,XTERM);
4962 case KEY_setsockopt:
4963 LOP(OP_SSOCKOPT,XTERM);
4969 LOP(OP_SHMCTL,XTERM);
4972 LOP(OP_SHMGET,XTERM);
4975 LOP(OP_SHMREAD,XTERM);
4978 LOP(OP_SHMWRITE,XTERM);
4981 LOP(OP_SHUTDOWN,XTERM);
4990 LOP(OP_SOCKET,XTERM);
4992 case KEY_socketpair:
4993 LOP(OP_SOCKPAIR,XTERM);
4996 checkcomma(s,PL_tokenbuf,"subroutine name");
4998 if (*s == ';' || *s == ')') /* probably a close */
4999 Perl_croak(aTHX_ "sort is now a reserved word");
5001 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5005 LOP(OP_SPLIT,XTERM);
5008 LOP(OP_SPRINTF,XTERM);
5011 LOP(OP_SPLICE,XTERM);
5026 LOP(OP_SUBSTR,XTERM);
5032 char tmpbuf[sizeof PL_tokenbuf];
5033 SSize_t tboffset = 0;
5034 expectation attrful;
5035 bool have_name, have_proto, bad_proto;
5040 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
5041 (*s == ':' && s[1] == ':'))
5044 attrful = XATTRBLOCK;
5045 /* remember buffer pos'n for later force_word */
5046 tboffset = s - PL_oldbufptr;
5047 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
5048 if (strchr(tmpbuf, ':'))
5049 sv_setpv(PL_subname, tmpbuf);
5051 sv_setsv(PL_subname,PL_curstname);
5052 sv_catpvn(PL_subname,"::",2);
5053 sv_catpvn(PL_subname,tmpbuf,len);
5060 Perl_croak(aTHX_ "Missing name in \"my sub\"");
5061 PL_expect = XTERMBLOCK;
5062 attrful = XATTRTERM;
5063 sv_setpv(PL_subname,"?");
5067 if (key == KEY_format) {
5069 PL_lex_formbrack = PL_lex_brackets + 1;
5071 (void) force_word(PL_oldbufptr + tboffset, WORD,
5076 /* Look for a prototype */
5080 s = scan_str(s,FALSE,FALSE);
5082 Perl_croak(aTHX_ "Prototype not terminated");
5083 /* strip spaces and check for bad characters */
5084 d = SvPVX(PL_lex_stuff);
5087 for (p = d; *p; ++p) {
5090 if (!strchr("$@%*;[]&\\", *p))
5095 if (bad_proto && ckWARN(WARN_SYNTAX))
5096 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5097 "Illegal character in prototype for %"SVf" : %s",
5099 SvCUR(PL_lex_stuff) = tmp;
5107 if (*s == ':' && s[1] != ':')
5108 PL_expect = attrful;
5109 else if (!have_name && *s != '{' && key == KEY_sub)
5110 Perl_croak(aTHX_ "Illegal declaration of anonymous subroutine");
5113 PL_nextval[PL_nexttoke].opval =
5114 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
5115 PL_lex_stuff = Nullsv;
5119 sv_setpv(PL_subname,
5120 PL_curstash ? "__ANON__" : "__ANON__::__ANON__");
5123 (void) force_word(PL_oldbufptr + tboffset, WORD,
5132 LOP(OP_SYSTEM,XREF);
5135 LOP(OP_SYMLINK,XTERM);
5138 LOP(OP_SYSCALL,XTERM);
5141 LOP(OP_SYSOPEN,XTERM);
5144 LOP(OP_SYSSEEK,XTERM);
5147 LOP(OP_SYSREAD,XTERM);
5150 LOP(OP_SYSWRITE,XTERM);
5154 TERM(sublex_start());
5175 LOP(OP_TRUNCATE,XTERM);
5187 yylval.ival = CopLINE(PL_curcop);
5191 yylval.ival = CopLINE(PL_curcop);
5195 LOP(OP_UNLINK,XTERM);
5201 LOP(OP_UNPACK,XTERM);
5204 LOP(OP_UTIME,XTERM);
5210 LOP(OP_UNSHIFT,XTERM);
5213 if (PL_expect != XSTATE)
5214 yyerror("\"use\" not allowed in expression");
5216 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
5217 s = force_version(s, TRUE);
5218 if (*s == ';' || (s = skipspace(s), *s == ';')) {
5219 PL_nextval[PL_nexttoke].opval = Nullop;
5222 else if (*s == 'v') {
5223 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5224 s = force_version(s, FALSE);
5228 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5229 s = force_version(s, FALSE);
5241 yylval.ival = CopLINE(PL_curcop);
5245 PL_hints |= HINT_BLOCK_SCOPE;
5252 LOP(OP_WAITPID,XTERM);
5261 ctl_l[0] = toCTRL('L');
5263 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5266 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5271 if (PL_expect == XOPERATOR)
5277 yylval.ival = OP_XOR;
5282 TERM(sublex_start());
5287 #pragma segment Main
5291 S_pending_ident(pTHX)
5294 register I32 tmp = 0;
5295 /* pit holds the identifier we read and pending_ident is reset */
5296 char pit = PL_pending_ident;
5297 PL_pending_ident = 0;
5299 DEBUG_T({ PerlIO_printf(Perl_debug_log,
5300 "### Tokener saw identifier '%s'\n", PL_tokenbuf); });
5302 /* if we're in a my(), we can't allow dynamics here.
5303 $foo'bar has already been turned into $foo::bar, so
5304 just check for colons.
5306 if it's a legal name, the OP is a PADANY.
5309 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
5310 if (strchr(PL_tokenbuf,':'))
5311 yyerror(Perl_form(aTHX_ "No package name allowed for "
5312 "variable %s in \"our\"",
5314 tmp = allocmy(PL_tokenbuf);
5317 if (strchr(PL_tokenbuf,':'))
5318 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
5320 yylval.opval = newOP(OP_PADANY, 0);
5321 yylval.opval->op_targ = allocmy(PL_tokenbuf);
5327 build the ops for accesses to a my() variable.
5329 Deny my($a) or my($b) in a sort block, *if* $a or $b is
5330 then used in a comparison. This catches most, but not
5331 all cases. For instance, it catches
5332 sort { my($a); $a <=> $b }
5334 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
5335 (although why you'd do that is anyone's guess).
5338 if (!strchr(PL_tokenbuf,':')) {
5340 tmp = pad_findmy(PL_tokenbuf);
5341 if (tmp != NOT_IN_PAD) {
5342 /* might be an "our" variable" */
5343 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
5344 /* build ops for a bareword */
5345 SV *sym = newSVpv(HvNAME(PAD_COMPNAME_OURSTASH(tmp)), 0);
5346 sv_catpvn(sym, "::", 2);
5347 sv_catpv(sym, PL_tokenbuf+1);
5348 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
5349 yylval.opval->op_private = OPpCONST_ENTERED;
5350 gv_fetchpv(SvPVX(sym),
5352 ? (GV_ADDMULTI | GV_ADDINEVAL)
5355 ((PL_tokenbuf[0] == '$') ? SVt_PV
5356 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5361 /* if it's a sort block and they're naming $a or $b */
5362 if (PL_last_lop_op == OP_SORT &&
5363 PL_tokenbuf[0] == '$' &&
5364 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
5367 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
5368 d < PL_bufend && *d != '\n';
5371 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
5372 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
5378 yylval.opval = newOP(OP_PADANY, 0);
5379 yylval.opval->op_targ = tmp;
5385 Whine if they've said @foo in a doublequoted string,
5386 and @foo isn't a variable we can find in the symbol
5389 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
5390 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
5391 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
5392 && ckWARN(WARN_AMBIGUOUS))
5394 /* Downgraded from fatal to warning 20000522 mjd */
5395 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5396 "Possible unintended interpolation of %s in string",
5401 /* build ops for a bareword */
5402 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
5403 yylval.opval->op_private = OPpCONST_ENTERED;
5404 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
5405 ((PL_tokenbuf[0] == '$') ? SVt_PV
5406 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5412 Perl_keyword(pTHX_ register char *d, I32 len)
5417 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5418 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5419 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5420 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5421 if (strEQ(d,"__END__")) return KEY___END__;
5425 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5430 if (strEQ(d,"and")) return -KEY_and;
5431 if (strEQ(d,"abs")) return -KEY_abs;
5434 if (strEQ(d,"alarm")) return -KEY_alarm;
5435 if (strEQ(d,"atan2")) return -KEY_atan2;
5438 if (strEQ(d,"accept")) return -KEY_accept;
5443 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5446 if (strEQ(d,"bless")) return -KEY_bless;
5447 if (strEQ(d,"bind")) return -KEY_bind;
5448 if (strEQ(d,"binmode")) return -KEY_binmode;
5451 if (strEQ(d,"CORE")) return -KEY_CORE;
5452 if (strEQ(d,"CHECK")) return KEY_CHECK;
5457 if (strEQ(d,"cmp")) return -KEY_cmp;
5458 if (strEQ(d,"chr")) return -KEY_chr;
5459 if (strEQ(d,"cos")) return -KEY_cos;
5462 if (strEQ(d,"chop")) return -KEY_chop;
5465 if (strEQ(d,"close")) return -KEY_close;
5466 if (strEQ(d,"chdir")) return -KEY_chdir;
5467 if (strEQ(d,"chomp")) return -KEY_chomp;
5468 if (strEQ(d,"chmod")) return -KEY_chmod;
5469 if (strEQ(d,"chown")) return -KEY_chown;
5470 if (strEQ(d,"crypt")) return -KEY_crypt;
5473 if (strEQ(d,"chroot")) return -KEY_chroot;
5474 if (strEQ(d,"caller")) return -KEY_caller;
5477 if (strEQ(d,"connect")) return -KEY_connect;
5480 if (strEQ(d,"closedir")) return -KEY_closedir;
5481 if (strEQ(d,"continue")) return -KEY_continue;
5486 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5491 if (strEQ(d,"do")) return KEY_do;
5494 if (strEQ(d,"die")) return -KEY_die;
5497 if (strEQ(d,"dump")) return -KEY_dump;
5500 if (strEQ(d,"delete")) return KEY_delete;
5503 if (strEQ(d,"defined")) return KEY_defined;
5504 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5507 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5512 if (strEQ(d,"END")) return KEY_END;
5517 if (strEQ(d,"eq")) return -KEY_eq;
5520 if (strEQ(d,"eof")) return -KEY_eof;
5521 if (strEQ(d,"err")) return -KEY_err;
5522 if (strEQ(d,"exp")) return -KEY_exp;
5525 if (strEQ(d,"else")) return KEY_else;
5526 if (strEQ(d,"exit")) return -KEY_exit;
5527 if (strEQ(d,"eval")) return KEY_eval;
5528 if (strEQ(d,"exec")) return -KEY_exec;
5529 if (strEQ(d,"each")) return -KEY_each;
5532 if (strEQ(d,"elsif")) return KEY_elsif;
5535 if (strEQ(d,"exists")) return KEY_exists;
5536 if (strEQ(d,"elseif") && ckWARN_d(WARN_SYNTAX))
5537 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5538 "elseif should be elsif");
5541 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5542 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5545 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5548 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5549 if (strEQ(d,"endservent")) return -KEY_endservent;
5552 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5559 if (strEQ(d,"for")) return KEY_for;
5562 if (strEQ(d,"fork")) return -KEY_fork;
5565 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5566 if (strEQ(d,"flock")) return -KEY_flock;
5569 if (strEQ(d,"format")) return KEY_format;
5570 if (strEQ(d,"fileno")) return -KEY_fileno;
5573 if (strEQ(d,"foreach")) return KEY_foreach;
5576 if (strEQ(d,"formline")) return -KEY_formline;
5581 if (strnEQ(d,"get",3)) {
5586 if (strEQ(d,"ppid")) return -KEY_getppid;
5587 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5590 if (strEQ(d,"pwent")) return -KEY_getpwent;
5591 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5592 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5595 if (strEQ(d,"peername")) return -KEY_getpeername;
5596 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5597 if (strEQ(d,"priority")) return -KEY_getpriority;
5600 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5603 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5607 else if (*d == 'h') {
5608 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5609 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5610 if (strEQ(d,"hostent")) return -KEY_gethostent;
5612 else if (*d == 'n') {
5613 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5614 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5615 if (strEQ(d,"netent")) return -KEY_getnetent;
5617 else if (*d == 's') {
5618 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5619 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5620 if (strEQ(d,"servent")) return -KEY_getservent;
5621 if (strEQ(d,"sockname")) return -KEY_getsockname;
5622 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5624 else if (*d == 'g') {
5625 if (strEQ(d,"grent")) return -KEY_getgrent;
5626 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5627 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5629 else if (*d == 'l') {
5630 if (strEQ(d,"login")) return -KEY_getlogin;
5632 else if (strEQ(d,"c")) return -KEY_getc;
5637 if (strEQ(d,"gt")) return -KEY_gt;
5638 if (strEQ(d,"ge")) return -KEY_ge;
5641 if (strEQ(d,"grep")) return KEY_grep;
5642 if (strEQ(d,"goto")) return KEY_goto;
5643 if (strEQ(d,"glob")) return KEY_glob;
5646 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5651 if (strEQ(d,"hex")) return -KEY_hex;
5654 if (strEQ(d,"INIT")) return KEY_INIT;
5659 if (strEQ(d,"if")) return KEY_if;
5662 if (strEQ(d,"int")) return -KEY_int;
5665 if (strEQ(d,"index")) return -KEY_index;
5666 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5671 if (strEQ(d,"join")) return -KEY_join;
5675 if (strEQ(d,"keys")) return -KEY_keys;
5676 if (strEQ(d,"kill")) return -KEY_kill;
5682 if (strEQ(d,"lt")) return -KEY_lt;
5683 if (strEQ(d,"le")) return -KEY_le;
5684 if (strEQ(d,"lc")) return -KEY_lc;
5687 if (strEQ(d,"log")) return -KEY_log;
5690 if (strEQ(d,"last")) return KEY_last;
5691 if (strEQ(d,"link")) return -KEY_link;
5692 if (strEQ(d,"lock")) return -KEY_lock;
5695 if (strEQ(d,"local")) return KEY_local;
5696 if (strEQ(d,"lstat")) return -KEY_lstat;
5699 if (strEQ(d,"length")) return -KEY_length;
5700 if (strEQ(d,"listen")) return -KEY_listen;
5703 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5706 if (strEQ(d,"localtime")) return -KEY_localtime;
5712 case 1: return KEY_m;
5714 if (strEQ(d,"my")) return KEY_my;
5717 if (strEQ(d,"map")) return KEY_map;
5720 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5723 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5724 if (strEQ(d,"msgget")) return -KEY_msgget;
5725 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5726 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5731 if (strEQ(d,"next")) return KEY_next;
5732 if (strEQ(d,"ne")) return -KEY_ne;
5733 if (strEQ(d,"not")) return -KEY_not;
5734 if (strEQ(d,"no")) return KEY_no;
5739 if (strEQ(d,"or")) return -KEY_or;
5742 if (strEQ(d,"ord")) return -KEY_ord;
5743 if (strEQ(d,"oct")) return -KEY_oct;
5744 if (strEQ(d,"our")) return KEY_our;
5747 if (strEQ(d,"open")) return -KEY_open;
5750 if (strEQ(d,"opendir")) return -KEY_opendir;
5757 if (strEQ(d,"pop")) return -KEY_pop;
5758 if (strEQ(d,"pos")) return KEY_pos;
5761 if (strEQ(d,"push")) return -KEY_push;
5762 if (strEQ(d,"pack")) return -KEY_pack;
5763 if (strEQ(d,"pipe")) return -KEY_pipe;
5766 if (strEQ(d,"print")) return KEY_print;
5769 if (strEQ(d,"printf")) return KEY_printf;
5772 if (strEQ(d,"package")) return KEY_package;
5775 if (strEQ(d,"prototype")) return KEY_prototype;
5780 if (strEQ(d,"q")) return KEY_q;
5781 if (strEQ(d,"qr")) return KEY_qr;
5782 if (strEQ(d,"qq")) return KEY_qq;
5783 if (strEQ(d,"qw")) return KEY_qw;
5784 if (strEQ(d,"qx")) return KEY_qx;
5786 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5791 if (strEQ(d,"ref")) return -KEY_ref;
5794 if (strEQ(d,"read")) return -KEY_read;
5795 if (strEQ(d,"rand")) return -KEY_rand;
5796 if (strEQ(d,"recv")) return -KEY_recv;
5797 if (strEQ(d,"redo")) return KEY_redo;
5800 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5801 if (strEQ(d,"reset")) return -KEY_reset;
5804 if (strEQ(d,"return")) return KEY_return;
5805 if (strEQ(d,"rename")) return -KEY_rename;
5806 if (strEQ(d,"rindex")) return -KEY_rindex;
5809 if (strEQ(d,"require")) return KEY_require;
5810 if (strEQ(d,"reverse")) return -KEY_reverse;
5811 if (strEQ(d,"readdir")) return -KEY_readdir;
5814 if (strEQ(d,"readlink")) return -KEY_readlink;
5815 if (strEQ(d,"readline")) return -KEY_readline;
5816 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5819 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5825 case 0: return KEY_s;
5827 if (strEQ(d,"scalar")) return KEY_scalar;
5832 if (strEQ(d,"seek")) return -KEY_seek;
5833 if (strEQ(d,"send")) return -KEY_send;
5836 if (strEQ(d,"semop")) return -KEY_semop;
5839 if (strEQ(d,"select")) return -KEY_select;
5840 if (strEQ(d,"semctl")) return -KEY_semctl;
5841 if (strEQ(d,"semget")) return -KEY_semget;
5844 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5845 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5848 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5849 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5852 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5855 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5856 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5857 if (strEQ(d,"setservent")) return -KEY_setservent;
5860 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5861 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5868 if (strEQ(d,"shift")) return -KEY_shift;
5871 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5872 if (strEQ(d,"shmget")) return -KEY_shmget;
5875 if (strEQ(d,"shmread")) return -KEY_shmread;
5878 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5879 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5884 if (strEQ(d,"sin")) return -KEY_sin;
5887 if (strEQ(d,"sleep")) return -KEY_sleep;
5890 if (strEQ(d,"sort")) return KEY_sort;
5891 if (strEQ(d,"socket")) return -KEY_socket;
5892 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5895 if (strEQ(d,"split")) return KEY_split;
5896 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5897 if (strEQ(d,"splice")) return -KEY_splice;
5900 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5903 if (strEQ(d,"srand")) return -KEY_srand;
5906 if (strEQ(d,"stat")) return -KEY_stat;
5907 if (strEQ(d,"study")) return KEY_study;
5910 if (strEQ(d,"substr")) return -KEY_substr;
5911 if (strEQ(d,"sub")) return KEY_sub;
5916 if (strEQ(d,"system")) return -KEY_system;
5919 if (strEQ(d,"symlink")) return -KEY_symlink;
5920 if (strEQ(d,"syscall")) return -KEY_syscall;
5921 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5922 if (strEQ(d,"sysread")) return -KEY_sysread;
5923 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5926 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5935 if (strEQ(d,"tr")) return KEY_tr;
5938 if (strEQ(d,"tie")) return KEY_tie;
5941 if (strEQ(d,"tell")) return -KEY_tell;
5942 if (strEQ(d,"tied")) return KEY_tied;
5943 if (strEQ(d,"time")) return -KEY_time;
5946 if (strEQ(d,"times")) return -KEY_times;
5949 if (strEQ(d,"telldir")) return -KEY_telldir;
5952 if (strEQ(d,"truncate")) return -KEY_truncate;
5959 if (strEQ(d,"uc")) return -KEY_uc;
5962 if (strEQ(d,"use")) return KEY_use;
5965 if (strEQ(d,"undef")) return KEY_undef;
5966 if (strEQ(d,"until")) return KEY_until;
5967 if (strEQ(d,"untie")) return KEY_untie;
5968 if (strEQ(d,"utime")) return -KEY_utime;
5969 if (strEQ(d,"umask")) return -KEY_umask;
5972 if (strEQ(d,"unless")) return KEY_unless;
5973 if (strEQ(d,"unpack")) return -KEY_unpack;
5974 if (strEQ(d,"unlink")) return -KEY_unlink;
5977 if (strEQ(d,"unshift")) return -KEY_unshift;
5978 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5983 if (strEQ(d,"values")) return -KEY_values;
5984 if (strEQ(d,"vec")) return -KEY_vec;
5989 if (strEQ(d,"warn")) return -KEY_warn;
5990 if (strEQ(d,"wait")) return -KEY_wait;
5993 if (strEQ(d,"while")) return KEY_while;
5994 if (strEQ(d,"write")) return -KEY_write;
5997 if (strEQ(d,"waitpid")) return -KEY_waitpid;
6000 if (strEQ(d,"wantarray")) return -KEY_wantarray;
6005 if (len == 1) return -KEY_x;
6006 if (strEQ(d,"xor")) return -KEY_xor;
6009 if (len == 1) return KEY_y;
6018 S_checkcomma(pTHX_ register char *s, char *name, char *what)
6022 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
6023 if (ckWARN(WARN_SYNTAX)) {
6025 for (w = s+2; *w && level; w++) {
6032 for (; *w && isSPACE(*w); w++) ;
6033 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
6034 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
6035 "%s (...) interpreted as function",name);
6038 while (s < PL_bufend && isSPACE(*s))
6042 while (s < PL_bufend && isSPACE(*s))
6044 if (isIDFIRST_lazy_if(s,UTF)) {
6046 while (isALNUM_lazy_if(s,UTF))
6048 while (s < PL_bufend && isSPACE(*s))
6053 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
6057 Perl_croak(aTHX_ "No comma allowed after %s", what);
6062 /* Either returns sv, or mortalizes sv and returns a new SV*.
6063 Best used as sv=new_constant(..., sv, ...).
6064 If s, pv are NULL, calls subroutine with one argument,
6065 and type is used with error messages only. */
6068 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
6072 HV *table = GvHV(PL_hintgv); /* ^H */
6076 const char *why1, *why2, *why3;
6078 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
6081 why2 = strEQ(key,"charnames")
6082 ? "(possibly a missing \"use charnames ...\")"
6084 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
6085 (type ? type: "undef"), why2);
6087 /* This is convoluted and evil ("goto considered harmful")
6088 * but I do not understand the intricacies of all the different
6089 * failure modes of %^H in here. The goal here is to make
6090 * the most probable error message user-friendly. --jhi */
6095 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
6096 (type ? type: "undef"), why1, why2, why3);
6098 yyerror(SvPVX(msg));
6102 cvp = hv_fetch(table, key, strlen(key), FALSE);
6103 if (!cvp || !SvOK(*cvp)) {
6106 why3 = "} is not defined";
6109 sv_2mortal(sv); /* Parent created it permanently */
6112 pv = sv_2mortal(newSVpvn(s, len));
6114 typesv = sv_2mortal(newSVpv(type, 0));
6116 typesv = &PL_sv_undef;
6118 PUSHSTACKi(PERLSI_OVERLOAD);
6130 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
6134 /* Check the eval first */
6135 if (!PL_in_eval && SvTRUE(ERRSV)) {
6137 sv_catpv(ERRSV, "Propagated");
6138 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
6140 res = SvREFCNT_inc(sv);
6144 (void)SvREFCNT_inc(res);
6153 why1 = "Call to &{$^H{";
6155 why3 = "}} did not return a defined value";
6164 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
6166 register char *d = dest;
6167 register char *e = d + destlen - 3; /* two-character token, ending NUL */
6170 Perl_croak(aTHX_ ident_too_long);
6171 if (isALNUM(*s)) /* UTF handled below */
6173 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
6178 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
6182 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6183 char *t = s + UTF8SKIP(s);
6184 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6186 if (d + (t - s) > e)
6187 Perl_croak(aTHX_ ident_too_long);
6188 Copy(s, d, t - s, char);
6201 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
6211 e = d + destlen - 3; /* two-character token, ending NUL */
6213 while (isDIGIT(*s)) {
6215 Perl_croak(aTHX_ ident_too_long);
6222 Perl_croak(aTHX_ ident_too_long);
6223 if (isALNUM(*s)) /* UTF handled below */
6225 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
6230 else if (*s == ':' && s[1] == ':') {
6234 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6235 char *t = s + UTF8SKIP(s);
6236 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6238 if (d + (t - s) > e)
6239 Perl_croak(aTHX_ ident_too_long);
6240 Copy(s, d, t - s, char);
6251 if (PL_lex_state != LEX_NORMAL)
6252 PL_lex_state = LEX_INTERPENDMAYBE;
6255 if (*s == '$' && s[1] &&
6256 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
6269 if (*d == '^' && *s && isCONTROLVAR(*s)) {
6274 if (isSPACE(s[-1])) {
6277 if (!SPACE_OR_TAB(ch)) {
6283 if (isIDFIRST_lazy_if(d,UTF)) {
6287 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
6289 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
6292 Copy(s, d, e - s, char);
6297 while ((isALNUM(*s) || *s == ':') && d < e)
6300 Perl_croak(aTHX_ ident_too_long);
6303 while (s < send && SPACE_OR_TAB(*s)) s++;
6304 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
6305 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
6306 const char *brack = *s == '[' ? "[...]" : "{...}";
6307 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
6308 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
6309 funny, dest, brack, funny, dest, brack);
6312 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
6316 /* Handle extended ${^Foo} variables
6317 * 1999-02-27 mjd-perl-patch@plover.com */
6318 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
6322 while (isALNUM(*s) && d < e) {
6326 Perl_croak(aTHX_ ident_too_long);
6331 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
6332 PL_lex_state = LEX_INTERPEND;
6337 if (PL_lex_state == LEX_NORMAL) {
6338 if (ckWARN(WARN_AMBIGUOUS) &&
6339 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
6341 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
6342 "Ambiguous use of %c{%s} resolved to %c%s",
6343 funny, dest, funny, dest);
6348 s = bracket; /* let the parser handle it */
6352 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
6353 PL_lex_state = LEX_INTERPEND;
6358 Perl_pmflag(pTHX_ U32* pmfl, int ch)
6363 *pmfl |= PMf_GLOBAL;
6365 *pmfl |= PMf_CONTINUE;
6369 *pmfl |= PMf_MULTILINE;
6371 *pmfl |= PMf_SINGLELINE;
6373 *pmfl |= PMf_EXTENDED;
6377 S_scan_pat(pTHX_ char *start, I32 type)
6382 s = scan_str(start,FALSE,FALSE);
6384 Perl_croak(aTHX_ "Search pattern not terminated");
6386 pm = (PMOP*)newPMOP(type, 0);
6387 if (PL_multi_open == '?')
6388 pm->op_pmflags |= PMf_ONCE;
6390 while (*s && strchr("iomsx", *s))
6391 pmflag(&pm->op_pmflags,*s++);
6394 while (*s && strchr("iogcmsx", *s))
6395 pmflag(&pm->op_pmflags,*s++);
6397 /* issue a warning if /c is specified,but /g is not */
6398 if (ckWARN(WARN_REGEXP) &&
6399 (pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL))
6401 Perl_warner(aTHX_ packWARN(WARN_REGEXP), c_without_g);
6404 pm->op_pmpermflags = pm->op_pmflags;
6406 PL_lex_op = (OP*)pm;
6407 yylval.ival = OP_MATCH;
6412 S_scan_subst(pTHX_ char *start)
6419 yylval.ival = OP_NULL;
6421 s = scan_str(start,FALSE,FALSE);
6424 Perl_croak(aTHX_ "Substitution pattern not terminated");
6426 if (s[-1] == PL_multi_open)
6429 first_start = PL_multi_start;
6430 s = scan_str(s,FALSE,FALSE);
6433 SvREFCNT_dec(PL_lex_stuff);
6434 PL_lex_stuff = Nullsv;
6436 Perl_croak(aTHX_ "Substitution replacement not terminated");
6438 PL_multi_start = first_start; /* so whole substitution is taken together */
6440 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6446 else if (strchr("iogcmsx", *s))
6447 pmflag(&pm->op_pmflags,*s++);
6452 /* /c is not meaningful with s/// */
6453 if (ckWARN(WARN_REGEXP) && (pm->op_pmflags & PMf_CONTINUE))
6455 Perl_warner(aTHX_ packWARN(WARN_REGEXP), c_in_subst);
6460 PL_sublex_info.super_bufptr = s;
6461 PL_sublex_info.super_bufend = PL_bufend;
6463 pm->op_pmflags |= PMf_EVAL;
6464 repl = newSVpvn("",0);
6466 sv_catpv(repl, es ? "eval " : "do ");
6467 sv_catpvn(repl, "{ ", 2);
6468 sv_catsv(repl, PL_lex_repl);
6469 sv_catpvn(repl, " };", 2);
6471 SvREFCNT_dec(PL_lex_repl);
6475 pm->op_pmpermflags = pm->op_pmflags;
6476 PL_lex_op = (OP*)pm;
6477 yylval.ival = OP_SUBST;
6482 S_scan_trans(pTHX_ char *start)
6491 yylval.ival = OP_NULL;
6493 s = scan_str(start,FALSE,FALSE);
6495 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6496 if (s[-1] == PL_multi_open)
6499 s = scan_str(s,FALSE,FALSE);
6502 SvREFCNT_dec(PL_lex_stuff);
6503 PL_lex_stuff = Nullsv;
6505 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6508 complement = del = squash = 0;
6509 while (strchr("cds", *s)) {
6511 complement = OPpTRANS_COMPLEMENT;
6513 del = OPpTRANS_DELETE;
6515 squash = OPpTRANS_SQUASH;
6519 New(803, tbl, complement&&!del?258:256, short);
6520 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6521 o->op_private = del|squash|complement|
6522 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
6523 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
6526 yylval.ival = OP_TRANS;
6531 S_scan_heredoc(pTHX_ register char *s)
6534 I32 op_type = OP_SCALAR;
6541 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6545 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6548 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6549 if (*peek && strchr("`'\"",*peek)) {
6552 s = delimcpy(d, e, s, PL_bufend, term, &len);
6562 if (!isALNUM_lazy_if(s,UTF))
6563 deprecate_old("bare << to mean <<\"\"");
6564 for (; isALNUM_lazy_if(s,UTF); s++) {
6569 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6570 Perl_croak(aTHX_ "Delimiter for here document is too long");
6573 len = d - PL_tokenbuf;
6574 #ifndef PERL_STRICT_CR
6575 d = strchr(s, '\r');
6579 while (s < PL_bufend) {
6585 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6594 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6599 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6600 herewas = newSVpvn(s,PL_bufend-s);
6602 s--, herewas = newSVpvn(s,d-s);
6603 s += SvCUR(herewas);
6605 tmpstr = NEWSV(87,79);
6606 sv_upgrade(tmpstr, SVt_PVIV);
6611 else if (term == '`') {
6612 op_type = OP_BACKTICK;
6613 SvIVX(tmpstr) = '\\';
6617 PL_multi_start = CopLINE(PL_curcop);
6618 PL_multi_open = PL_multi_close = '<';
6619 term = *PL_tokenbuf;
6620 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6621 char *bufptr = PL_sublex_info.super_bufptr;
6622 char *bufend = PL_sublex_info.super_bufend;
6623 char *olds = s - SvCUR(herewas);
6624 s = strchr(bufptr, '\n');
6628 while (s < bufend &&
6629 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6631 CopLINE_inc(PL_curcop);
6634 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6635 missingterm(PL_tokenbuf);
6637 sv_setpvn(herewas,bufptr,d-bufptr+1);
6638 sv_setpvn(tmpstr,d+1,s-d);
6640 sv_catpvn(herewas,s,bufend-s);
6641 (void)strcpy(bufptr,SvPVX(herewas));
6648 while (s < PL_bufend &&
6649 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6651 CopLINE_inc(PL_curcop);
6653 if (s >= PL_bufend) {
6654 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6655 missingterm(PL_tokenbuf);
6657 sv_setpvn(tmpstr,d+1,s-d);
6659 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6661 sv_catpvn(herewas,s,PL_bufend-s);
6662 sv_setsv(PL_linestr,herewas);
6663 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6664 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6665 PL_last_lop = PL_last_uni = Nullch;
6668 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6669 while (s >= PL_bufend) { /* multiple line string? */
6671 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6672 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6673 missingterm(PL_tokenbuf);
6675 CopLINE_inc(PL_curcop);
6676 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6677 PL_last_lop = PL_last_uni = Nullch;
6678 #ifndef PERL_STRICT_CR
6679 if (PL_bufend - PL_linestart >= 2) {
6680 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6681 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6683 PL_bufend[-2] = '\n';
6685 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6687 else if (PL_bufend[-1] == '\r')
6688 PL_bufend[-1] = '\n';
6690 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6691 PL_bufend[-1] = '\n';
6693 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6694 SV *sv = NEWSV(88,0);
6696 sv_upgrade(sv, SVt_PVMG);
6697 sv_setsv(sv,PL_linestr);
6700 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6702 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6705 sv_catsv(PL_linestr,herewas);
6706 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6710 sv_catsv(tmpstr,PL_linestr);
6715 PL_multi_end = CopLINE(PL_curcop);
6716 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6717 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6718 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6720 SvREFCNT_dec(herewas);
6722 if (UTF && is_utf8_string((U8*)SvPVX(tmpstr), SvCUR(tmpstr)))
6724 else if (PL_encoding)
6725 sv_recode_to_utf8(tmpstr, PL_encoding);
6727 PL_lex_stuff = tmpstr;
6728 yylval.ival = op_type;
6733 takes: current position in input buffer
6734 returns: new position in input buffer
6735 side-effects: yylval and lex_op are set.
6740 <FH> read from filehandle
6741 <pkg::FH> read from package qualified filehandle
6742 <pkg'FH> read from package qualified filehandle
6743 <$fh> read from filehandle in $fh
6749 S_scan_inputsymbol(pTHX_ char *start)
6751 register char *s = start; /* current position in buffer */
6757 d = PL_tokenbuf; /* start of temp holding space */
6758 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6759 end = strchr(s, '\n');
6762 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6764 /* die if we didn't have space for the contents of the <>,
6765 or if it didn't end, or if we see a newline
6768 if (len >= sizeof PL_tokenbuf)
6769 Perl_croak(aTHX_ "Excessively long <> operator");
6771 Perl_croak(aTHX_ "Unterminated <> operator");
6776 Remember, only scalar variables are interpreted as filehandles by
6777 this code. Anything more complex (e.g., <$fh{$num}>) will be
6778 treated as a glob() call.
6779 This code makes use of the fact that except for the $ at the front,
6780 a scalar variable and a filehandle look the same.
6782 if (*d == '$' && d[1]) d++;
6784 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6785 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6788 /* If we've tried to read what we allow filehandles to look like, and
6789 there's still text left, then it must be a glob() and not a getline.
6790 Use scan_str to pull out the stuff between the <> and treat it
6791 as nothing more than a string.
6794 if (d - PL_tokenbuf != len) {
6795 yylval.ival = OP_GLOB;
6797 s = scan_str(start,FALSE,FALSE);
6799 Perl_croak(aTHX_ "Glob not terminated");
6803 bool readline_overriden = FALSE;
6804 GV *gv_readline = Nullgv;
6806 /* we're in a filehandle read situation */
6809 /* turn <> into <ARGV> */
6811 (void)strcpy(d,"ARGV");
6813 /* Check whether readline() is overriden */
6814 if (((gv_readline = gv_fetchpv("readline", FALSE, SVt_PVCV))
6815 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
6817 ((gvp = (GV**)hv_fetch(PL_globalstash, "readline", 8, FALSE))
6818 && (gv_readline = *gvp) != (GV*)&PL_sv_undef
6819 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
6820 readline_overriden = TRUE;
6822 /* if <$fh>, create the ops to turn the variable into a
6828 /* try to find it in the pad for this block, otherwise find
6829 add symbol table ops
6831 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6832 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
6833 SV *sym = sv_2mortal(
6834 newSVpv(HvNAME(PAD_COMPNAME_OURSTASH(tmp)),0));
6835 sv_catpvn(sym, "::", 2);
6841 OP *o = newOP(OP_PADSV, 0);
6843 PL_lex_op = readline_overriden
6844 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6845 append_elem(OP_LIST, o,
6846 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
6847 : (OP*)newUNOP(OP_READLINE, 0, o);
6856 ? (GV_ADDMULTI | GV_ADDINEVAL)
6859 PL_lex_op = readline_overriden
6860 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6861 append_elem(OP_LIST,
6862 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
6863 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
6864 : (OP*)newUNOP(OP_READLINE, 0,
6865 newUNOP(OP_RV2SV, 0,
6866 newGVOP(OP_GV, 0, gv)));
6868 if (!readline_overriden)
6869 PL_lex_op->op_flags |= OPf_SPECIAL;
6870 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6871 yylval.ival = OP_NULL;
6874 /* If it's none of the above, it must be a literal filehandle
6875 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6877 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6878 PL_lex_op = readline_overriden
6879 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6880 append_elem(OP_LIST,
6881 newGVOP(OP_GV, 0, gv),
6882 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
6883 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6884 yylval.ival = OP_NULL;
6893 takes: start position in buffer
6894 keep_quoted preserve \ on the embedded delimiter(s)
6895 keep_delims preserve the delimiters around the string
6896 returns: position to continue reading from buffer
6897 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6898 updates the read buffer.
6900 This subroutine pulls a string out of the input. It is called for:
6901 q single quotes q(literal text)
6902 ' single quotes 'literal text'
6903 qq double quotes qq(interpolate $here please)
6904 " double quotes "interpolate $here please"
6905 qx backticks qx(/bin/ls -l)
6906 ` backticks `/bin/ls -l`
6907 qw quote words @EXPORT_OK = qw( func() $spam )
6908 m// regexp match m/this/
6909 s/// regexp substitute s/this/that/
6910 tr/// string transliterate tr/this/that/
6911 y/// string transliterate y/this/that/
6912 ($*@) sub prototypes sub foo ($)
6913 (stuff) sub attr parameters sub foo : attr(stuff)
6914 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6916 In most of these cases (all but <>, patterns and transliterate)
6917 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6918 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6919 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6922 It skips whitespace before the string starts, and treats the first
6923 character as the delimiter. If the delimiter is one of ([{< then
6924 the corresponding "close" character )]}> is used as the closing
6925 delimiter. It allows quoting of delimiters, and if the string has
6926 balanced delimiters ([{<>}]) it allows nesting.
6928 On success, the SV with the resulting string is put into lex_stuff or,
6929 if that is already non-NULL, into lex_repl. The second case occurs only
6930 when parsing the RHS of the special constructs s/// and tr/// (y///).
6931 For convenience, the terminating delimiter character is stuffed into
6936 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6938 SV *sv; /* scalar value: string */
6939 char *tmps; /* temp string, used for delimiter matching */
6940 register char *s = start; /* current position in the buffer */
6941 register char term; /* terminating character */
6942 register char *to; /* current position in the sv's data */
6943 I32 brackets = 1; /* bracket nesting level */
6944 bool has_utf8 = FALSE; /* is there any utf8 content? */
6945 I32 termcode; /* terminating char. code */
6946 U8 termstr[UTF8_MAXLEN]; /* terminating string */
6947 STRLEN termlen; /* length of terminating string */
6948 char *last = NULL; /* last position for nesting bracket */
6950 /* skip space before the delimiter */
6954 /* mark where we are, in case we need to report errors */
6957 /* after skipping whitespace, the next character is the terminator */
6960 termcode = termstr[0] = term;
6964 termcode = utf8_to_uvchr((U8*)s, &termlen);
6965 Copy(s, termstr, termlen, U8);
6966 if (!UTF8_IS_INVARIANT(term))
6970 /* mark where we are */
6971 PL_multi_start = CopLINE(PL_curcop);
6972 PL_multi_open = term;
6974 /* find corresponding closing delimiter */
6975 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6976 termcode = termstr[0] = term = tmps[5];
6978 PL_multi_close = term;
6980 /* create a new SV to hold the contents. 87 is leak category, I'm
6981 assuming. 79 is the SV's initial length. What a random number. */
6983 sv_upgrade(sv, SVt_PVIV);
6984 SvIVX(sv) = termcode;
6985 (void)SvPOK_only(sv); /* validate pointer */
6987 /* move past delimiter and try to read a complete string */
6989 sv_catpvn(sv, s, termlen);
6992 if (PL_encoding && !UTF) {
6996 int offset = s - SvPVX(PL_linestr);
6997 bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
6998 &offset, (char*)termstr, termlen);
6999 char *ns = SvPVX(PL_linestr) + offset;
7000 char *svlast = SvEND(sv) - 1;
7002 for (; s < ns; s++) {
7003 if (*s == '\n' && !PL_rsfp)
7004 CopLINE_inc(PL_curcop);
7007 goto read_more_line;
7009 /* handle quoted delimiters */
7010 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
7012 for (t = svlast-2; t >= SvPVX(sv) && *t == '\\';)
7014 if ((svlast-1 - t) % 2) {
7018 SvCUR_set(sv, SvCUR(sv) - 1);
7023 if (PL_multi_open == PL_multi_close) {
7030 for (w = t = last; t < svlast; w++, t++) {
7031 /* At here, all closes are "was quoted" one,
7032 so we don't check PL_multi_close. */
7034 if (!keep_quoted && *(t+1) == PL_multi_open)
7039 else if (*t == PL_multi_open)
7047 SvCUR_set(sv, w - SvPVX(sv));
7050 if (--brackets <= 0)
7056 SvCUR_set(sv, SvCUR(sv) - 1);
7062 /* extend sv if need be */
7063 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
7064 /* set 'to' to the next character in the sv's string */
7065 to = SvPVX(sv)+SvCUR(sv);
7067 /* if open delimiter is the close delimiter read unbridle */
7068 if (PL_multi_open == PL_multi_close) {
7069 for (; s < PL_bufend; s++,to++) {
7070 /* embedded newlines increment the current line number */
7071 if (*s == '\n' && !PL_rsfp)
7072 CopLINE_inc(PL_curcop);
7073 /* handle quoted delimiters */
7074 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
7075 if (!keep_quoted && s[1] == term)
7077 /* any other quotes are simply copied straight through */
7081 /* terminate when run out of buffer (the for() condition), or
7082 have found the terminator */
7083 else if (*s == term) {
7086 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
7089 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
7095 /* if the terminator isn't the same as the start character (e.g.,
7096 matched brackets), we have to allow more in the quoting, and
7097 be prepared for nested brackets.
7100 /* read until we run out of string, or we find the terminator */
7101 for (; s < PL_bufend; s++,to++) {
7102 /* embedded newlines increment the line count */
7103 if (*s == '\n' && !PL_rsfp)
7104 CopLINE_inc(PL_curcop);
7105 /* backslashes can escape the open or closing characters */
7106 if (*s == '\\' && s+1 < PL_bufend) {
7108 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
7113 /* allow nested opens and closes */
7114 else if (*s == PL_multi_close && --brackets <= 0)
7116 else if (*s == PL_multi_open)
7118 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
7123 /* terminate the copied string and update the sv's end-of-string */
7125 SvCUR_set(sv, to - SvPVX(sv));
7128 * this next chunk reads more into the buffer if we're not done yet
7132 break; /* handle case where we are done yet :-) */
7134 #ifndef PERL_STRICT_CR
7135 if (to - SvPVX(sv) >= 2) {
7136 if ((to[-2] == '\r' && to[-1] == '\n') ||
7137 (to[-2] == '\n' && to[-1] == '\r'))
7141 SvCUR_set(sv, to - SvPVX(sv));
7143 else if (to[-1] == '\r')
7146 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
7151 /* if we're out of file, or a read fails, bail and reset the current
7152 line marker so we can report where the unterminated string began
7155 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
7157 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
7160 /* we read a line, so increment our line counter */
7161 CopLINE_inc(PL_curcop);
7163 /* update debugger info */
7164 if (PERLDB_LINE && PL_curstash != PL_debstash) {
7165 SV *sv = NEWSV(88,0);
7167 sv_upgrade(sv, SVt_PVMG);
7168 sv_setsv(sv,PL_linestr);
7171 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
7174 /* having changed the buffer, we must update PL_bufend */
7175 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7176 PL_last_lop = PL_last_uni = Nullch;
7179 /* at this point, we have successfully read the delimited string */
7181 if (!PL_encoding || UTF) {
7183 sv_catpvn(sv, s, termlen);
7186 if (has_utf8 || PL_encoding)
7189 PL_multi_end = CopLINE(PL_curcop);
7191 /* if we allocated too much space, give some back */
7192 if (SvCUR(sv) + 5 < SvLEN(sv)) {
7193 SvLEN_set(sv, SvCUR(sv) + 1);
7194 Renew(SvPVX(sv), SvLEN(sv), char);
7197 /* decide whether this is the first or second quoted string we've read
7210 takes: pointer to position in buffer
7211 returns: pointer to new position in buffer
7212 side-effects: builds ops for the constant in yylval.op
7214 Read a number in any of the formats that Perl accepts:
7216 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
7217 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
7220 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
7222 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
7225 If it reads a number without a decimal point or an exponent, it will
7226 try converting the number to an integer and see if it can do so
7227 without loss of precision.
7231 Perl_scan_num(pTHX_ char *start, YYSTYPE* lvalp)
7233 register char *s = start; /* current position in buffer */
7234 register char *d; /* destination in temp buffer */
7235 register char *e; /* end of temp buffer */
7236 NV nv; /* number read, as a double */
7237 SV *sv = Nullsv; /* place to put the converted number */
7238 bool floatit; /* boolean: int or float? */
7239 char *lastub = 0; /* position of last underbar */
7240 static char number_too_long[] = "Number too long";
7242 /* We use the first character to decide what type of number this is */
7246 Perl_croak(aTHX_ "panic: scan_num");
7248 /* if it starts with a 0, it could be an octal number, a decimal in
7249 0.13 disguise, or a hexadecimal number, or a binary number. */
7253 u holds the "number so far"
7254 shift the power of 2 of the base
7255 (hex == 4, octal == 3, binary == 1)
7256 overflowed was the number more than we can hold?
7258 Shift is used when we add a digit. It also serves as an "are
7259 we in octal/hex/binary?" indicator to disallow hex characters
7265 bool overflowed = FALSE;
7266 bool just_zero = TRUE; /* just plain 0 or binary number? */
7267 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
7268 static char* bases[5] = { "", "binary", "", "octal",
7270 static char* Bases[5] = { "", "Binary", "", "Octal",
7272 static char *maxima[5] = { "",
7273 "0b11111111111111111111111111111111",
7277 char *base, *Base, *max;
7284 } else if (s[1] == 'b') {
7289 /* check for a decimal in disguise */
7290 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
7292 /* so it must be octal */
7299 if (ckWARN(WARN_SYNTAX))
7300 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7301 "Misplaced _ in number");
7305 base = bases[shift];
7306 Base = Bases[shift];
7307 max = maxima[shift];
7309 /* read the rest of the number */
7311 /* x is used in the overflow test,
7312 b is the digit we're adding on. */
7317 /* if we don't mention it, we're done */
7321 /* _ are ignored -- but warned about if consecutive */
7323 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7324 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7325 "Misplaced _ in number");
7329 /* 8 and 9 are not octal */
7332 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
7336 case '2': case '3': case '4':
7337 case '5': case '6': case '7':
7339 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
7343 b = *s++ & 15; /* ASCII digit -> value of digit */
7347 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
7348 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
7349 /* make sure they said 0x */
7354 /* Prepare to put the digit we have onto the end
7355 of the number so far. We check for overflows.
7361 x = u << shift; /* make room for the digit */
7363 if ((x >> shift) != u
7364 && !(PL_hints & HINT_NEW_BINARY)) {
7367 if (ckWARN_d(WARN_OVERFLOW))
7368 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
7369 "Integer overflow in %s number",
7372 u = x | b; /* add the digit to the end */
7375 n *= nvshift[shift];
7376 /* If an NV has not enough bits in its
7377 * mantissa to represent an UV this summing of
7378 * small low-order numbers is a waste of time
7379 * (because the NV cannot preserve the
7380 * low-order bits anyway): we could just
7381 * remember when did we overflow and in the
7382 * end just multiply n by the right
7390 /* if we get here, we had success: make a scalar value from
7395 /* final misplaced underbar check */
7397 if (ckWARN(WARN_SYNTAX))
7398 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
7403 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
7404 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
7405 "%s number > %s non-portable",
7411 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
7412 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
7413 "%s number > %s non-portable",
7418 if (just_zero && (PL_hints & HINT_NEW_INTEGER))
7419 sv = new_constant(start, s - start, "integer",
7421 else if (PL_hints & HINT_NEW_BINARY)
7422 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
7427 handle decimal numbers.
7428 we're also sent here when we read a 0 as the first digit
7430 case '1': case '2': case '3': case '4': case '5':
7431 case '6': case '7': case '8': case '9': case '.':
7434 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
7437 /* read next group of digits and _ and copy into d */
7438 while (isDIGIT(*s) || *s == '_') {
7439 /* skip underscores, checking for misplaced ones
7443 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7444 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7445 "Misplaced _ in number");
7449 /* check for end of fixed-length buffer */
7451 Perl_croak(aTHX_ number_too_long);
7452 /* if we're ok, copy the character */
7457 /* final misplaced underbar check */
7458 if (lastub && s == lastub + 1) {
7459 if (ckWARN(WARN_SYNTAX))
7460 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
7463 /* read a decimal portion if there is one. avoid
7464 3..5 being interpreted as the number 3. followed
7467 if (*s == '.' && s[1] != '.') {
7472 if (ckWARN(WARN_SYNTAX))
7473 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7474 "Misplaced _ in number");
7478 /* copy, ignoring underbars, until we run out of digits.
7480 for (; isDIGIT(*s) || *s == '_'; s++) {
7481 /* fixed length buffer check */
7483 Perl_croak(aTHX_ number_too_long);
7485 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7486 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7487 "Misplaced _ in number");
7493 /* fractional part ending in underbar? */
7495 if (ckWARN(WARN_SYNTAX))
7496 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7497 "Misplaced _ in number");
7499 if (*s == '.' && isDIGIT(s[1])) {
7500 /* oops, it's really a v-string, but without the "v" */
7506 /* read exponent part, if present */
7507 if (*s && strchr("eE",*s) && strchr("+-0123456789_", s[1])) {
7511 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
7512 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
7514 /* stray preinitial _ */
7516 if (ckWARN(WARN_SYNTAX))
7517 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7518 "Misplaced _ in number");
7522 /* allow positive or negative exponent */
7523 if (*s == '+' || *s == '-')
7526 /* stray initial _ */
7528 if (ckWARN(WARN_SYNTAX))
7529 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7530 "Misplaced _ in number");
7534 /* read digits of exponent */
7535 while (isDIGIT(*s) || *s == '_') {
7538 Perl_croak(aTHX_ number_too_long);
7542 if (ckWARN(WARN_SYNTAX) &&
7543 ((lastub && s == lastub + 1) ||
7544 (!isDIGIT(s[1]) && s[1] != '_')))
7545 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7546 "Misplaced _ in number");
7553 /* make an sv from the string */
7557 We try to do an integer conversion first if no characters
7558 indicating "float" have been found.
7563 int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
7565 if (flags == IS_NUMBER_IN_UV) {
7567 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7570 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
7571 if (uv <= (UV) IV_MIN)
7572 sv_setiv(sv, -(IV)uv);
7579 /* terminate the string */
7581 nv = Atof(PL_tokenbuf);
7585 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7586 (PL_hints & HINT_NEW_INTEGER) )
7587 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7588 (floatit ? "float" : "integer"),
7592 /* if it starts with a v, it could be a v-string */
7595 sv = NEWSV(92,5); /* preallocate storage space */
7596 s = scan_vstring(s,sv);
7600 /* make the op for the constant and return */
7603 lvalp->opval = newSVOP(OP_CONST, 0, sv);
7605 lvalp->opval = Nullop;
7611 S_scan_formline(pTHX_ register char *s)
7615 SV *stuff = newSVpvn("",0);
7616 bool needargs = FALSE;
7622 #ifdef PERL_STRICT_CR
7623 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7625 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7627 if (*t == '\n' || t == PL_bufend) {
7632 if (PL_in_eval && !PL_rsfp) {
7633 eol = memchr(s,'\n',PL_bufend-s);
7638 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7640 for (t = s; t < eol; t++) {
7641 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7643 goto enough; /* ~~ must be first line in formline */
7645 if (*t == '@' || *t == '^')
7649 sv_catpvn(stuff, s, eol-s);
7650 #ifndef PERL_STRICT_CR
7651 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7652 char *end = SvPVX(stuff) + SvCUR(stuff);
7664 s = filter_gets(PL_linestr, PL_rsfp, 0);
7665 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7666 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7667 PL_last_lop = PL_last_uni = Nullch;
7679 PL_lex_state = LEX_NORMAL;
7680 PL_nextval[PL_nexttoke].ival = 0;
7684 PL_lex_state = LEX_FORMLINE;
7686 if (UTF && is_utf8_string((U8*)SvPVX(stuff), SvCUR(stuff)))
7688 else if (PL_encoding)
7689 sv_recode_to_utf8(stuff, PL_encoding);
7691 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7693 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7697 SvREFCNT_dec(stuff);
7699 PL_lex_formbrack = 0;
7710 PL_cshlen = strlen(PL_cshname);
7715 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7717 I32 oldsavestack_ix = PL_savestack_ix;
7718 CV* outsidecv = PL_compcv;
7721 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7723 SAVEI32(PL_subline);
7724 save_item(PL_subname);
7725 SAVESPTR(PL_compcv);
7727 PL_compcv = (CV*)NEWSV(1104,0);
7728 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7729 CvFLAGS(PL_compcv) |= flags;
7731 PL_subline = CopLINE(PL_curcop);
7732 CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
7733 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7734 CvOUTSIDE_SEQ(PL_compcv) = PL_cop_seqmax;
7736 return oldsavestack_ix;
7740 #pragma segment Perl_yylex
7743 Perl_yywarn(pTHX_ char *s)
7745 PL_in_eval |= EVAL_WARNONLY;
7747 PL_in_eval &= ~EVAL_WARNONLY;
7752 Perl_yyerror(pTHX_ char *s)
7755 char *context = NULL;
7759 if (!yychar || (yychar == ';' && !PL_rsfp))
7761 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7762 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7765 The code below is removed for NetWare because it abends/crashes on NetWare
7766 when the script has error such as not having the closing quotes like:
7768 Checking of white spaces is anyway done in NetWare code.
7771 while (isSPACE(*PL_oldoldbufptr))
7774 context = PL_oldoldbufptr;
7775 contlen = PL_bufptr - PL_oldoldbufptr;
7777 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7778 PL_oldbufptr != PL_bufptr) {
7781 The code below is removed for NetWare because it abends/crashes on NetWare
7782 when the script has error such as not having the closing quotes like:
7784 Checking of white spaces is anyway done in NetWare code.
7787 while (isSPACE(*PL_oldbufptr))
7790 context = PL_oldbufptr;
7791 contlen = PL_bufptr - PL_oldbufptr;
7793 else if (yychar > 255)
7794 where = "next token ???";
7795 #ifdef USE_PURE_BISON
7796 /* GNU Bison sets the value -2 */
7797 else if (yychar == -2) {
7799 else if ((yychar & 127) == 127) {
7801 if (PL_lex_state == LEX_NORMAL ||
7802 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7803 where = "at end of line";
7804 else if (PL_lex_inpat)
7805 where = "within pattern";
7807 where = "within string";
7810 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7812 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7813 else if (isPRINT_LC(yychar))
7814 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7816 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7817 where = SvPVX(where_sv);
7819 msg = sv_2mortal(newSVpv(s, 0));
7820 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7821 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7823 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7825 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7826 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7827 Perl_sv_catpvf(aTHX_ msg,
7828 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7829 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7832 if (PL_in_eval & EVAL_WARNONLY && ckWARN_d(WARN_SYNTAX))
7833 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "%"SVf, msg);
7836 if (PL_error_count >= 10) {
7837 if (PL_in_eval && SvCUR(ERRSV))
7838 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
7839 ERRSV, OutCopFILE(PL_curcop));
7841 Perl_croak(aTHX_ "%s has too many errors.\n",
7842 OutCopFILE(PL_curcop));
7845 PL_in_my_stash = Nullhv;
7849 #pragma segment Main
7853 S_swallow_bom(pTHX_ U8 *s)
7856 slen = SvCUR(PL_linestr);
7860 /* UTF-16 little-endian */
7861 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
7862 Perl_croak(aTHX_ "Unsupported script encoding");
7863 #ifndef PERL_NO_UTF16_FILTER
7864 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-LE script encoding\n"));
7866 if (PL_bufend > (char*)s) {
7870 filter_add(utf16rev_textfilter, NULL);
7871 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7872 PL_bufend = (char*)utf16_to_utf8_reversed(s, news,
7873 PL_bufend - (char*)s - 1,
7875 Copy(news, s, newlen, U8);
7876 SvCUR_set(PL_linestr, newlen);
7877 PL_bufend = SvPVX(PL_linestr) + newlen;
7878 news[newlen++] = '\0';
7882 Perl_croak(aTHX_ "Unsupported script encoding");
7887 if (s[1] == 0xFF) { /* UTF-16 big-endian */
7888 #ifndef PERL_NO_UTF16_FILTER
7889 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding\n"));
7891 if (PL_bufend > (char *)s) {
7895 filter_add(utf16_textfilter, NULL);
7896 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7897 PL_bufend = (char*)utf16_to_utf8(s, news,
7898 PL_bufend - (char*)s,
7900 Copy(news, s, newlen, U8);
7901 SvCUR_set(PL_linestr, newlen);
7902 PL_bufend = SvPVX(PL_linestr) + newlen;
7903 news[newlen++] = '\0';
7907 Perl_croak(aTHX_ "Unsupported script encoding");
7912 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
7913 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-8 script encoding\n"));
7918 if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */
7919 s[2] == 0xFE && s[3] == 0xFF)
7921 Perl_croak(aTHX_ "Unsupported script encoding");
7929 * Restore a source filter.
7933 restore_rsfp(pTHX_ void *f)
7935 PerlIO *fp = (PerlIO*)f;
7937 if (PL_rsfp == PerlIO_stdin())
7938 PerlIO_clearerr(PL_rsfp);
7939 else if (PL_rsfp && (PL_rsfp != fp))
7940 PerlIO_close(PL_rsfp);
7944 #ifndef PERL_NO_UTF16_FILTER
7946 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
7948 I32 count = FILTER_READ(idx+1, sv, maxlen);
7953 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7954 if (!*SvPV_nolen(sv))
7955 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7958 tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7959 sv_usepvn(sv, (char*)tmps, tend - tmps);
7965 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
7967 I32 count = FILTER_READ(idx+1, sv, maxlen);
7972 if (!*SvPV_nolen(sv))
7973 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7976 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7977 tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7978 sv_usepvn(sv, (char*)tmps, tend - tmps);
7985 Returns a pointer to the next character after the parsed
7986 vstring, as well as updating the passed in sv.
7988 Function must be called like
7991 s = scan_vstring(s,sv);
7993 The sv should already be large enough to store the vstring
7994 passed in, for performance reasons.
7999 Perl_scan_vstring(pTHX_ char *s, SV *sv)
8003 if (*pos == 'v') pos++; /* get past 'v' */
8004 while (pos < PL_bufend && (isDIGIT(*pos) || *pos == '_'))
8007 /* this may not be a v-string if followed by => */
8009 while (next < PL_bufend && isSPACE(*next))
8011 if ((PL_bufend - next) >= 2 && *next == '=' && next[1] == '>' ) {
8012 /* return string not v-string */
8013 sv_setpvn(sv,(char *)s,pos-s);
8018 if (!isALPHA(*pos)) {
8020 U8 tmpbuf[UTF8_MAXLEN+1];
8023 if (*s == 'v') s++; /* get past 'v' */
8025 sv_setpvn(sv, "", 0);
8030 /* this is atoi() that tolerates underscores */
8033 while (--end >= s) {
8038 rev += (*end - '0') * mult;
8040 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
8041 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
8042 "Integer overflow in decimal number");
8046 if (rev > 0x7FFFFFFF)
8047 Perl_croak(aTHX_ "In EBCDIC the v-string components cannot exceed 2147483647");
8049 /* Append native character for the rev point */
8050 tmpend = uvchr_to_utf8(tmpbuf, rev);
8051 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
8052 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev)))
8054 if (pos + 1 < PL_bufend && *pos == '.' && isDIGIT(pos[1]))
8060 while (pos < PL_bufend && (isDIGIT(*pos) || *pos == '_'))
8064 sv_magic(sv,NULL,PERL_MAGIC_vstring,(const char*)start, pos-start);