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));
260 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
261 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
263 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
264 if (t < PL_bufptr && isSPACE(*t))
265 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
266 t - PL_oldoldbufptr, PL_oldoldbufptr);
270 Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
277 * Complain about missing quote/regexp/heredoc terminator.
278 * If it's called with (char *)NULL then it cauterizes the line buffer.
279 * If we're in a delimited string and the delimiter is a control
280 * character, it's reformatted into a two-char sequence like ^C.
285 S_missingterm(pTHX_ char *s)
290 char *nl = strrchr(s,'\n');
296 iscntrl(PL_multi_close)
298 PL_multi_close < 32 || PL_multi_close == 127
302 tmpbuf[1] = toCTRL(PL_multi_close);
308 *tmpbuf = (char)PL_multi_close;
312 q = strchr(s,'"') ? '\'' : '"';
313 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
321 Perl_deprecate(pTHX_ char *s)
323 if (ckWARN(WARN_DEPRECATED))
324 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED), "Use of %s is deprecated", s);
328 Perl_deprecate_old(pTHX_ char *s)
330 /* This function should NOT be called for any new deprecated warnings */
331 /* Use Perl_deprecate instead */
333 /* It is here to maintain backward compatibility with the pre-5.8 */
334 /* warnings category hierarchy. The "deprecated" category used to */
335 /* live under the "syntax" category. It is now a top-level category */
336 /* in its own right. */
338 if (ckWARN2(WARN_DEPRECATED, WARN_SYNTAX))
339 Perl_warner(aTHX_ packWARN2(WARN_DEPRECATED, WARN_SYNTAX),
340 "Use of %s is deprecated", s);
345 * Deprecate a comma-less variable list.
351 deprecate_old("comma-less variable list");
355 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
356 * utf16-to-utf8-reversed.
359 #ifdef PERL_CR_FILTER
363 register char *s = SvPVX(sv);
364 register char *e = s + SvCUR(sv);
365 /* outer loop optimized to do nothing if there are no CR-LFs */
367 if (*s++ == '\r' && *s == '\n') {
368 /* hit a CR-LF, need to copy the rest */
369 register char *d = s - 1;
372 if (*s == '\r' && s[1] == '\n')
383 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
385 I32 count = FILTER_READ(idx+1, sv, maxlen);
386 if (count > 0 && !maxlen)
394 * Initialize variables. Uses the Perl save_stack to save its state (for
395 * recursive calls to the parser).
399 Perl_lex_start(pTHX_ SV *line)
404 SAVEI32(PL_lex_dojoin);
405 SAVEI32(PL_lex_brackets);
406 SAVEI32(PL_lex_casemods);
407 SAVEI32(PL_lex_starts);
408 SAVEI32(PL_lex_state);
409 SAVEVPTR(PL_lex_inpat);
410 SAVEI32(PL_lex_inwhat);
411 if (PL_lex_state == LEX_KNOWNEXT) {
412 I32 toke = PL_nexttoke;
413 while (--toke >= 0) {
414 SAVEI32(PL_nexttype[toke]);
415 SAVEVPTR(PL_nextval[toke]);
417 SAVEI32(PL_nexttoke);
419 SAVECOPLINE(PL_curcop);
422 SAVEPPTR(PL_oldbufptr);
423 SAVEPPTR(PL_oldoldbufptr);
424 SAVEPPTR(PL_last_lop);
425 SAVEPPTR(PL_last_uni);
426 SAVEPPTR(PL_linestart);
427 SAVESPTR(PL_linestr);
428 SAVEGENERICPV(PL_lex_brackstack);
429 SAVEGENERICPV(PL_lex_casestack);
430 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
431 SAVESPTR(PL_lex_stuff);
432 SAVEI32(PL_lex_defer);
433 SAVEI32(PL_sublex_info.sub_inwhat);
434 SAVESPTR(PL_lex_repl);
436 SAVEINT(PL_lex_expect);
438 PL_lex_state = LEX_NORMAL;
442 New(899, PL_lex_brackstack, 120, char);
443 New(899, PL_lex_casestack, 12, char);
445 *PL_lex_casestack = '\0';
448 PL_lex_stuff = Nullsv;
449 PL_lex_repl = Nullsv;
453 PL_sublex_info.sub_inwhat = 0;
455 if (SvREADONLY(PL_linestr))
456 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
457 s = SvPV(PL_linestr, len);
458 if (!len || s[len-1] != ';') {
459 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
460 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
461 sv_catpvn(PL_linestr, "\n;", 2);
463 SvTEMP_off(PL_linestr);
464 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
465 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
466 PL_last_lop = PL_last_uni = Nullch;
472 * Finalizer for lexing operations. Must be called when the parser is
473 * done with the lexer.
479 PL_doextract = FALSE;
484 * This subroutine has nothing to do with tilting, whether at windmills
485 * or pinball tables. Its name is short for "increment line". It
486 * increments the current line number in CopLINE(PL_curcop) and checks
487 * to see whether the line starts with a comment of the form
488 * # line 500 "foo.pm"
489 * If so, it sets the current line number and file to the values in the comment.
493 S_incline(pTHX_ char *s)
500 CopLINE_inc(PL_curcop);
503 while (SPACE_OR_TAB(*s)) s++;
504 if (strnEQ(s, "line", 4))
508 if (SPACE_OR_TAB(*s))
512 while (SPACE_OR_TAB(*s)) s++;
518 while (SPACE_OR_TAB(*s))
520 if (*s == '"' && (t = strchr(s+1, '"'))) {
525 for (t = s; !isSPACE(*t); t++) ;
528 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
530 if (*e != '\n' && *e != '\0')
531 return; /* false alarm */
536 CopFILE_free(PL_curcop);
537 CopFILE_set(PL_curcop, s);
540 CopLINE_set(PL_curcop, atoi(n)-1);
545 * Called to gobble the appropriate amount and type of whitespace.
546 * Skips comments as well.
550 S_skipspace(pTHX_ register char *s)
552 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
553 while (s < PL_bufend && SPACE_OR_TAB(*s))
559 SSize_t oldprevlen, oldoldprevlen;
560 SSize_t oldloplen = 0, oldunilen = 0;
561 while (s < PL_bufend && isSPACE(*s)) {
562 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
567 if (s < PL_bufend && *s == '#') {
568 while (s < PL_bufend && *s != '\n')
572 if (PL_in_eval && !PL_rsfp) {
579 /* only continue to recharge the buffer if we're at the end
580 * of the buffer, we're not reading from a source filter, and
581 * we're in normal lexing mode
583 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
584 PL_lex_state == LEX_FORMLINE)
587 /* try to recharge the buffer */
588 if ((s = filter_gets(PL_linestr, PL_rsfp,
589 (prevlen = SvCUR(PL_linestr)))) == Nullch)
591 /* end of file. Add on the -p or -n magic */
592 if (PL_minus_n || PL_minus_p) {
593 sv_setpv(PL_linestr,PL_minus_p ?
594 ";}continue{print or die qq(-p destination: $!\\n)" :
596 sv_catpv(PL_linestr,";}");
597 PL_minus_n = PL_minus_p = 0;
600 sv_setpv(PL_linestr,";");
602 /* reset variables for next time we lex */
603 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
605 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
606 PL_last_lop = PL_last_uni = Nullch;
608 /* Close the filehandle. Could be from -P preprocessor,
609 * STDIN, or a regular file. If we were reading code from
610 * STDIN (because the commandline held no -e or filename)
611 * then we don't close it, we reset it so the code can
612 * read from STDIN too.
615 if (PL_preprocess && !PL_in_eval)
616 (void)PerlProc_pclose(PL_rsfp);
617 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
618 PerlIO_clearerr(PL_rsfp);
620 (void)PerlIO_close(PL_rsfp);
625 /* not at end of file, so we only read another line */
626 /* make corresponding updates to old pointers, for yyerror() */
627 oldprevlen = PL_oldbufptr - PL_bufend;
628 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
630 oldunilen = PL_last_uni - PL_bufend;
632 oldloplen = PL_last_lop - PL_bufend;
633 PL_linestart = PL_bufptr = s + prevlen;
634 PL_bufend = s + SvCUR(PL_linestr);
636 PL_oldbufptr = s + oldprevlen;
637 PL_oldoldbufptr = s + oldoldprevlen;
639 PL_last_uni = s + oldunilen;
641 PL_last_lop = s + oldloplen;
644 /* debugger active and we're not compiling the debugger code,
645 * so store the line into the debugger's array of lines
647 if (PERLDB_LINE && PL_curstash != PL_debstash) {
648 SV *sv = NEWSV(85,0);
650 sv_upgrade(sv, SVt_PVMG);
651 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
654 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
661 * Check the unary operators to ensure there's no ambiguity in how they're
662 * used. An ambiguous piece of code would be:
664 * This doesn't mean rand() + 5. Because rand() is a unary operator,
665 * the +5 is its argument.
674 if (PL_oldoldbufptr != PL_last_uni)
676 while (isSPACE(*PL_last_uni))
678 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
679 if ((t = strchr(s, '(')) && t < PL_bufptr)
681 if (ckWARN_d(WARN_AMBIGUOUS)){
684 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
685 "Warning: Use of \"%s\" without parentheses is ambiguous",
692 * LOP : macro to build a list operator. Its behaviour has been replaced
693 * with a subroutine, S_lop() for which LOP is just another name.
696 #define LOP(f,x) return lop(f,x,s)
700 * Build a list operator (or something that might be one). The rules:
701 * - if we have a next token, then it's a list operator [why?]
702 * - if the next thing is an opening paren, then it's a function
703 * - else it's a list operator
707 S_lop(pTHX_ I32 f, int x, char *s)
714 PL_last_lop = PL_oldbufptr;
715 PL_last_lop_op = (OPCODE)f;
729 * When the lexer realizes it knows the next token (for instance,
730 * it is reordering tokens for the parser) then it can call S_force_next
731 * to know what token to return the next time the lexer is called. Caller
732 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
733 * handles the token correctly.
737 S_force_next(pTHX_ I32 type)
739 PL_nexttype[PL_nexttoke] = type;
741 if (PL_lex_state != LEX_KNOWNEXT) {
742 PL_lex_defer = PL_lex_state;
743 PL_lex_expect = PL_expect;
744 PL_lex_state = LEX_KNOWNEXT;
750 * When the lexer knows the next thing is a word (for instance, it has
751 * just seen -> and it knows that the next char is a word char, then
752 * it calls S_force_word to stick the next word into the PL_next lookahead.
755 * char *start : buffer position (must be within PL_linestr)
756 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
757 * int check_keyword : if true, Perl checks to make sure the word isn't
758 * a keyword (do this if the word is a label, e.g. goto FOO)
759 * int allow_pack : if true, : characters will also be allowed (require,
761 * int allow_initial_tick : used by the "sub" lexer only.
765 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
770 start = skipspace(start);
772 if (isIDFIRST_lazy_if(s,UTF) ||
773 (allow_pack && *s == ':') ||
774 (allow_initial_tick && *s == '\'') )
776 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
777 if (check_keyword && keyword(PL_tokenbuf, len))
779 if (token == METHOD) {
784 PL_expect = XOPERATOR;
787 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
788 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
796 * Called when the lexer wants $foo *foo &foo etc, but the program
797 * text only contains the "foo" portion. The first argument is a pointer
798 * to the "foo", and the second argument is the type symbol to prefix.
799 * Forces the next token to be a "WORD".
800 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
804 S_force_ident(pTHX_ register char *s, int kind)
807 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
808 PL_nextval[PL_nexttoke].opval = o;
811 o->op_private = OPpCONST_ENTERED;
812 /* XXX see note in pp_entereval() for why we forgo typo
813 warnings if the symbol must be introduced in an eval.
815 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
816 kind == '$' ? SVt_PV :
817 kind == '@' ? SVt_PVAV :
818 kind == '%' ? SVt_PVHV :
826 Perl_str_to_version(pTHX_ SV *sv)
831 char *start = SvPVx(sv,len);
832 bool utf = SvUTF8(sv) ? TRUE : FALSE;
833 char *end = start + len;
834 while (start < end) {
838 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
843 retval += ((NV)n)/nshift;
852 * Forces the next token to be a version number.
853 * If the next token appears to be an invalid version number, (e.g. "v2b"),
854 * and if "guessing" is TRUE, then no new token is created (and the caller
855 * must use an alternative parsing method).
859 S_force_version(pTHX_ char *s, int guessing)
861 OP *version = Nullop;
870 while (isDIGIT(*d) || *d == '_' || *d == '.')
872 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
874 s = scan_num(s, &yylval);
875 version = yylval.opval;
876 ver = cSVOPx(version)->op_sv;
877 if (SvPOK(ver) && !SvNIOK(ver)) {
878 (void)SvUPGRADE(ver, SVt_PVNV);
879 SvNVX(ver) = str_to_version(ver);
880 SvNOK_on(ver); /* hint that it is a version */
887 /* NOTE: The parser sees the package name and the VERSION swapped */
888 PL_nextval[PL_nexttoke].opval = version;
896 * Tokenize a quoted string passed in as an SV. It finds the next
897 * chunk, up to end of string or a backslash. It may make a new
898 * SV containing that chunk (if HINT_NEW_STRING is on). It also
903 S_tokeq(pTHX_ SV *sv)
914 s = SvPV_force(sv, len);
915 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
918 while (s < send && *s != '\\')
923 if ( PL_hints & HINT_NEW_STRING ) {
924 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
930 if (s + 1 < send && (s[1] == '\\'))
931 s++; /* all that, just for this */
936 SvCUR_set(sv, d - SvPVX(sv));
938 if ( PL_hints & HINT_NEW_STRING )
939 return new_constant(NULL, 0, "q", sv, pv, "q");
944 * Now come three functions related to double-quote context,
945 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
946 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
947 * interact with PL_lex_state, and create fake ( ... ) argument lists
948 * to handle functions and concatenation.
949 * They assume that whoever calls them will be setting up a fake
950 * join call, because each subthing puts a ',' after it. This lets
953 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
955 * (I'm not sure whether the spurious commas at the end of lcfirst's
956 * arguments and join's arguments are created or not).
961 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
963 * Pattern matching will set PL_lex_op to the pattern-matching op to
964 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
966 * OP_CONST and OP_READLINE are easy--just make the new op and return.
968 * Everything else becomes a FUNC.
970 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
971 * had an OP_CONST or OP_READLINE). This just sets us up for a
972 * call to S_sublex_push().
978 register I32 op_type = yylval.ival;
980 if (op_type == OP_NULL) {
981 yylval.opval = PL_lex_op;
985 if (op_type == OP_CONST || op_type == OP_READLINE) {
986 SV *sv = tokeq(PL_lex_stuff);
988 if (SvTYPE(sv) == SVt_PVIV) {
989 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
995 nsv = newSVpvn(p, len);
1001 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
1002 PL_lex_stuff = Nullsv;
1003 /* Allow <FH> // "foo" */
1004 if (op_type == OP_READLINE)
1005 PL_expect = XTERMORDORDOR;
1009 PL_sublex_info.super_state = PL_lex_state;
1010 PL_sublex_info.sub_inwhat = op_type;
1011 PL_sublex_info.sub_op = PL_lex_op;
1012 PL_lex_state = LEX_INTERPPUSH;
1016 yylval.opval = PL_lex_op;
1026 * Create a new scope to save the lexing state. The scope will be
1027 * ended in S_sublex_done. Returns a '(', starting the function arguments
1028 * to the uc, lc, etc. found before.
1029 * Sets PL_lex_state to LEX_INTERPCONCAT.
1037 PL_lex_state = PL_sublex_info.super_state;
1038 SAVEI32(PL_lex_dojoin);
1039 SAVEI32(PL_lex_brackets);
1040 SAVEI32(PL_lex_casemods);
1041 SAVEI32(PL_lex_starts);
1042 SAVEI32(PL_lex_state);
1043 SAVEVPTR(PL_lex_inpat);
1044 SAVEI32(PL_lex_inwhat);
1045 SAVECOPLINE(PL_curcop);
1046 SAVEPPTR(PL_bufptr);
1047 SAVEPPTR(PL_bufend);
1048 SAVEPPTR(PL_oldbufptr);
1049 SAVEPPTR(PL_oldoldbufptr);
1050 SAVEPPTR(PL_last_lop);
1051 SAVEPPTR(PL_last_uni);
1052 SAVEPPTR(PL_linestart);
1053 SAVESPTR(PL_linestr);
1054 SAVEGENERICPV(PL_lex_brackstack);
1055 SAVEGENERICPV(PL_lex_casestack);
1057 PL_linestr = PL_lex_stuff;
1058 PL_lex_stuff = Nullsv;
1060 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1061 = SvPVX(PL_linestr);
1062 PL_bufend += SvCUR(PL_linestr);
1063 PL_last_lop = PL_last_uni = Nullch;
1064 SAVEFREESV(PL_linestr);
1066 PL_lex_dojoin = FALSE;
1067 PL_lex_brackets = 0;
1068 New(899, PL_lex_brackstack, 120, char);
1069 New(899, PL_lex_casestack, 12, char);
1070 PL_lex_casemods = 0;
1071 *PL_lex_casestack = '\0';
1073 PL_lex_state = LEX_INTERPCONCAT;
1074 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
1076 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1077 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1078 PL_lex_inpat = PL_sublex_info.sub_op;
1080 PL_lex_inpat = Nullop;
1087 * Restores lexer state after a S_sublex_push.
1093 if (!PL_lex_starts++) {
1094 SV *sv = newSVpvn("",0);
1095 if (SvUTF8(PL_linestr))
1097 PL_expect = XOPERATOR;
1098 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1102 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1103 PL_lex_state = LEX_INTERPCASEMOD;
1107 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1108 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1109 PL_linestr = PL_lex_repl;
1111 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1112 PL_bufend += SvCUR(PL_linestr);
1113 PL_last_lop = PL_last_uni = Nullch;
1114 SAVEFREESV(PL_linestr);
1115 PL_lex_dojoin = FALSE;
1116 PL_lex_brackets = 0;
1117 PL_lex_casemods = 0;
1118 *PL_lex_casestack = '\0';
1120 if (SvEVALED(PL_lex_repl)) {
1121 PL_lex_state = LEX_INTERPNORMAL;
1123 /* we don't clear PL_lex_repl here, so that we can check later
1124 whether this is an evalled subst; that means we rely on the
1125 logic to ensure sublex_done() is called again only via the
1126 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1129 PL_lex_state = LEX_INTERPCONCAT;
1130 PL_lex_repl = Nullsv;
1136 PL_bufend = SvPVX(PL_linestr);
1137 PL_bufend += SvCUR(PL_linestr);
1138 PL_expect = XOPERATOR;
1139 PL_sublex_info.sub_inwhat = 0;
1147 Extracts a pattern, double-quoted string, or transliteration. This
1150 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1151 processing a pattern (PL_lex_inpat is true), a transliteration
1152 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1154 Returns a pointer to the character scanned up to. Iff this is
1155 advanced from the start pointer supplied (ie if anything was
1156 successfully parsed), will leave an OP for the substring scanned
1157 in yylval. Caller must intuit reason for not parsing further
1158 by looking at the next characters herself.
1162 double-quoted style: \r and \n
1163 regexp special ones: \D \s
1165 backrefs: \1 (deprecated in substitution replacements)
1166 case and quoting: \U \Q \E
1167 stops on @ and $, but not for $ as tail anchor
1169 In transliterations:
1170 characters are VERY literal, except for - not at the start or end
1171 of the string, which indicates a range. scan_const expands the
1172 range to the full set of intermediate characters.
1174 In double-quoted strings:
1176 double-quoted style: \r and \n
1178 backrefs: \1 (deprecated)
1179 case and quoting: \U \Q \E
1182 scan_const does *not* construct ops to handle interpolated strings.
1183 It stops processing as soon as it finds an embedded $ or @ variable
1184 and leaves it to the caller to work out what's going on.
1186 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @::foo.
1188 $ in pattern could be $foo or could be tail anchor. Assumption:
1189 it's a tail anchor if $ is the last thing in the string, or if it's
1190 followed by one of ")| \n\t"
1192 \1 (backreferences) are turned into $1
1194 The structure of the code is
1195 while (there's a character to process) {
1196 handle transliteration ranges
1197 skip regexp comments
1198 skip # initiated comments in //x patterns
1199 check for embedded @foo
1200 check for embedded scalars
1202 leave intact backslashes from leave (below)
1203 deprecate \1 in strings and sub replacements
1204 handle string-changing backslashes \l \U \Q \E, etc.
1205 switch (what was escaped) {
1206 handle - in a transliteration (becomes a literal -)
1207 handle \132 octal characters
1208 handle 0x15 hex characters
1209 handle \cV (control V)
1210 handle printf backslashes (\f, \r, \n, etc)
1212 } (end if backslash)
1213 } (end while character to read)
1218 S_scan_const(pTHX_ char *start)
1220 register char *send = PL_bufend; /* end of the constant */
1221 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1222 register char *s = start; /* start of the constant */
1223 register char *d = SvPVX(sv); /* destination for copies */
1224 bool dorange = FALSE; /* are we in a translit range? */
1225 bool didrange = FALSE; /* did we just finish a range? */
1226 I32 has_utf8 = FALSE; /* Output constant is UTF8 */
1227 I32 this_utf8 = UTF; /* The source string is assumed to be UTF8 */
1230 const char *leaveit = /* set of acceptably-backslashed characters */
1232 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1235 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1236 /* If we are doing a trans and we know we want UTF8 set expectation */
1237 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
1238 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1242 while (s < send || dorange) {
1243 /* get transliterations out of the way (they're most literal) */
1244 if (PL_lex_inwhat == OP_TRANS) {
1245 /* expand a range A-Z to the full set of characters. AIE! */
1247 I32 i; /* current expanded character */
1248 I32 min; /* first character in range */
1249 I32 max; /* last character in range */
1252 char *c = (char*)utf8_hop((U8*)d, -1);
1256 *c = (char)UTF_TO_NATIVE(0xff);
1257 /* mark the range as done, and continue */
1263 i = d - SvPVX(sv); /* remember current offset */
1264 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1265 d = SvPVX(sv) + i; /* refresh d after realloc */
1266 d -= 2; /* eat the first char and the - */
1268 min = (U8)*d; /* first char in range */
1269 max = (U8)d[1]; /* last char in range */
1273 "Invalid range \"%c-%c\" in transliteration operator",
1274 (char)min, (char)max);
1278 if ((isLOWER(min) && isLOWER(max)) ||
1279 (isUPPER(min) && isUPPER(max))) {
1281 for (i = min; i <= max; i++)
1283 *d++ = NATIVE_TO_NEED(has_utf8,i);
1285 for (i = min; i <= max; i++)
1287 *d++ = NATIVE_TO_NEED(has_utf8,i);
1292 for (i = min; i <= max; i++)
1295 /* mark the range as done, and continue */
1301 /* range begins (ignore - as first or last char) */
1302 else if (*s == '-' && s+1 < send && s != start) {
1304 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
1307 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
1319 /* if we get here, we're not doing a transliteration */
1321 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1322 except for the last char, which will be done separately. */
1323 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1325 while (s < send && *s != ')')
1326 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1328 else if (s[2] == '{' /* This should match regcomp.c */
1329 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1332 char *regparse = s + (s[2] == '{' ? 3 : 4);
1335 while (count && (c = *regparse)) {
1336 if (c == '\\' && regparse[1])
1344 if (*regparse != ')') {
1345 regparse--; /* Leave one char for continuation. */
1346 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1348 while (s < regparse)
1349 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1353 /* likewise skip #-initiated comments in //x patterns */
1354 else if (*s == '#' && PL_lex_inpat &&
1355 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1356 while (s+1 < send && *s != '\n')
1357 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1360 /* check for embedded arrays
1361 (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-)
1363 else if (*s == '@' && s[1]
1364 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1])))
1367 /* check for embedded scalars. only stop if we're sure it's a
1370 else if (*s == '$') {
1371 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1373 if (s + 1 < send && !strchr("()| \r\n\t", s[1]))
1374 break; /* in regexp, $ might be tail anchor */
1377 /* End of else if chain - OP_TRANS rejoin rest */
1380 if (*s == '\\' && s+1 < send) {
1383 /* some backslashes we leave behind */
1384 if (*leaveit && *s && strchr(leaveit, *s)) {
1385 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
1386 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1390 /* deprecate \1 in strings and substitution replacements */
1391 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1392 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1394 if (ckWARN(WARN_SYNTAX))
1395 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
1400 /* string-change backslash escapes */
1401 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1406 /* if we get here, it's either a quoted -, or a digit */
1409 /* quoted - in transliterations */
1411 if (PL_lex_inwhat == OP_TRANS) {
1418 if (ckWARN(WARN_MISC) &&
1421 Perl_warner(aTHX_ packWARN(WARN_MISC),
1422 "Unrecognized escape \\%c passed through",
1424 /* default action is to copy the quoted character */
1425 goto default_action;
1428 /* \132 indicates an octal constant */
1429 case '0': case '1': case '2': case '3':
1430 case '4': case '5': case '6': case '7':
1434 uv = grok_oct(s, &len, &flags, NULL);
1437 goto NUM_ESCAPE_INSERT;
1439 /* \x24 indicates a hex constant */
1443 char* e = strchr(s, '}');
1444 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1445 PERL_SCAN_DISALLOW_PREFIX;
1450 yyerror("Missing right brace on \\x{}");
1454 uv = grok_hex(s, &len, &flags, NULL);
1460 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
1461 uv = grok_hex(s, &len, &flags, NULL);
1467 /* Insert oct or hex escaped character.
1468 * There will always enough room in sv since such
1469 * escapes will be longer than any UTF-8 sequence
1470 * they can end up as. */
1472 /* We need to map to chars to ASCII before doing the tests
1475 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(uv))) {
1476 if (!has_utf8 && uv > 255) {
1477 /* Might need to recode whatever we have
1478 * accumulated so far if it contains any
1481 * (Can't we keep track of that and avoid
1482 * this rescan? --jhi)
1486 for (c = (U8 *) SvPVX(sv); c < (U8 *)d; c++) {
1487 if (!NATIVE_IS_INVARIANT(*c)) {
1492 STRLEN offset = d - SvPVX(sv);
1494 d = SvGROW(sv, SvLEN(sv) + hicount + 1) + offset;
1498 while (src >= (U8 *)SvPVX(sv)) {
1499 if (!NATIVE_IS_INVARIANT(*src)) {
1500 U8 ch = NATIVE_TO_ASCII(*src);
1501 *dst-- = (U8)UTF8_EIGHT_BIT_LO(ch);
1502 *dst-- = (U8)UTF8_EIGHT_BIT_HI(ch);
1512 if (has_utf8 || uv > 255) {
1513 d = (char*)uvchr_to_utf8((U8*)d, uv);
1515 if (PL_lex_inwhat == OP_TRANS &&
1516 PL_sublex_info.sub_op) {
1517 PL_sublex_info.sub_op->op_private |=
1518 (PL_lex_repl ? OPpTRANS_FROM_UTF
1531 /* \N{LATIN SMALL LETTER A} is a named character */
1535 char* e = strchr(s, '}');
1541 yyerror("Missing right brace on \\N{}");
1545 if (e > s + 2 && s[1] == 'U' && s[2] == '+') {
1547 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1548 PERL_SCAN_DISALLOW_PREFIX;
1551 uv = grok_hex(s, &len, &flags, NULL);
1553 goto NUM_ESCAPE_INSERT;
1555 res = newSVpvn(s + 1, e - s - 1);
1556 res = new_constant( Nullch, 0, "charnames",
1557 res, Nullsv, "\\N{...}" );
1559 sv_utf8_upgrade(res);
1560 str = SvPV(res,len);
1561 #ifdef EBCDIC_NEVER_MIND
1562 /* charnames uses pack U and that has been
1563 * recently changed to do the below uni->native
1564 * mapping, so this would be redundant (and wrong,
1565 * the code point would be doubly converted).
1566 * But leave this in just in case the pack U change
1567 * gets revoked, but the semantics is still
1568 * desireable for charnames. --jhi */
1570 UV uv = utf8_to_uvchr((U8*)str, 0);
1573 U8 tmpbuf[UTF8_MAXLEN+1], *d;
1575 d = uvchr_to_utf8(tmpbuf, UNI_TO_NATIVE(uv));
1576 sv_setpvn(res, (char *)tmpbuf, d - tmpbuf);
1577 str = SvPV(res, len);
1581 if (!has_utf8 && SvUTF8(res)) {
1582 char *ostart = SvPVX(sv);
1583 SvCUR_set(sv, d - ostart);
1586 sv_utf8_upgrade(sv);
1587 /* this just broke our allocation above... */
1588 SvGROW(sv, (STRLEN)(send - start));
1589 d = SvPVX(sv) + SvCUR(sv);
1592 if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */
1593 char *odest = SvPVX(sv);
1595 SvGROW(sv, (SvLEN(sv) + len - (e - s + 4)));
1596 d = SvPVX(sv) + (d - odest);
1598 Copy(str, d, len, char);
1605 yyerror("Missing braces on \\N{}");
1608 /* \c is a control character */
1617 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
1620 yyerror("Missing control char name in \\c");
1624 /* printf-style backslashes, formfeeds, newlines, etc */
1626 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
1629 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
1632 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
1635 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
1638 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
1641 *d++ = ASCII_TO_NEED(has_utf8,'\033');
1644 *d++ = ASCII_TO_NEED(has_utf8,'\007');
1650 } /* end if (backslash) */
1653 /* If we started with encoded form, or already know we want it
1654 and then encode the next character */
1655 if ((has_utf8 || this_utf8) && !NATIVE_IS_INVARIANT((U8)(*s))) {
1657 UV uv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
1658 STRLEN need = UNISKIP(NATIVE_TO_UNI(uv));
1661 /* encoded value larger than old, need extra space (NOTE: SvCUR() not set here) */
1662 STRLEN off = d - SvPVX(sv);
1663 d = SvGROW(sv, SvLEN(sv) + (need-len)) + off;
1665 d = (char*)uvchr_to_utf8((U8*)d, uv);
1669 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1671 } /* while loop to process each character */
1673 /* terminate the string and set up the sv */
1675 SvCUR_set(sv, d - SvPVX(sv));
1676 if (SvCUR(sv) >= SvLEN(sv))
1677 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
1680 if (PL_encoding && !has_utf8) {
1681 sv_recode_to_utf8(sv, PL_encoding);
1687 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1688 PL_sublex_info.sub_op->op_private |=
1689 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1693 /* shrink the sv if we allocated more than we used */
1694 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1695 SvLEN_set(sv, SvCUR(sv) + 1);
1696 Renew(SvPVX(sv), SvLEN(sv), char);
1699 /* return the substring (via yylval) only if we parsed anything */
1700 if (s > PL_bufptr) {
1701 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1702 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1704 ( PL_lex_inwhat == OP_TRANS
1706 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1709 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1716 * Returns TRUE if there's more to the expression (e.g., a subscript),
1719 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1721 * ->[ and ->{ return TRUE
1722 * { and [ outside a pattern are always subscripts, so return TRUE
1723 * if we're outside a pattern and it's not { or [, then return FALSE
1724 * if we're in a pattern and the first char is a {
1725 * {4,5} (any digits around the comma) returns FALSE
1726 * if we're in a pattern and the first char is a [
1728 * [SOMETHING] has a funky algorithm to decide whether it's a
1729 * character class or not. It has to deal with things like
1730 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1731 * anything else returns TRUE
1734 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1737 S_intuit_more(pTHX_ register char *s)
1739 if (PL_lex_brackets)
1741 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1743 if (*s != '{' && *s != '[')
1748 /* In a pattern, so maybe we have {n,m}. */
1765 /* On the other hand, maybe we have a character class */
1768 if (*s == ']' || *s == '^')
1771 /* this is terrifying, and it works */
1772 int weight = 2; /* let's weigh the evidence */
1774 unsigned char un_char = 255, last_un_char;
1775 char *send = strchr(s,']');
1776 char tmpbuf[sizeof PL_tokenbuf * 4];
1778 if (!send) /* has to be an expression */
1781 Zero(seen,256,char);
1784 else if (isDIGIT(*s)) {
1786 if (isDIGIT(s[1]) && s[2] == ']')
1792 for (; s < send; s++) {
1793 last_un_char = un_char;
1794 un_char = (unsigned char)*s;
1799 weight -= seen[un_char] * 10;
1800 if (isALNUM_lazy_if(s+1,UTF)) {
1801 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1802 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1807 else if (*s == '$' && s[1] &&
1808 strchr("[#!%*<>()-=",s[1])) {
1809 if (/*{*/ strchr("])} =",s[2]))
1818 if (strchr("wds]",s[1]))
1820 else if (seen['\''] || seen['"'])
1822 else if (strchr("rnftbxcav",s[1]))
1824 else if (isDIGIT(s[1])) {
1826 while (s[1] && isDIGIT(s[1]))
1836 if (strchr("aA01! ",last_un_char))
1838 if (strchr("zZ79~",s[1]))
1840 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1841 weight -= 5; /* cope with negative subscript */
1844 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1845 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1850 if (keyword(tmpbuf, d - tmpbuf))
1853 if (un_char == last_un_char + 1)
1855 weight -= seen[un_char];
1860 if (weight >= 0) /* probably a character class */
1870 * Does all the checking to disambiguate
1872 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1873 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1875 * First argument is the stuff after the first token, e.g. "bar".
1877 * Not a method if bar is a filehandle.
1878 * Not a method if foo is a subroutine prototyped to take a filehandle.
1879 * Not a method if it's really "Foo $bar"
1880 * Method if it's "foo $bar"
1881 * Not a method if it's really "print foo $bar"
1882 * Method if it's really "foo package::" (interpreted as package->foo)
1883 * Not a method if bar is known to be a subroutine ("sub bar; foo bar")
1884 * Not a method if bar is a filehandle or package, but is quoted with
1889 S_intuit_method(pTHX_ char *start, GV *gv)
1891 char *s = start + (*start == '$');
1892 char tmpbuf[sizeof PL_tokenbuf];
1900 if ((cv = GvCVu(gv))) {
1901 char *proto = SvPVX(cv);
1911 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1912 /* start is the beginning of the possible filehandle/object,
1913 * and s is the end of it
1914 * tmpbuf is a copy of it
1917 if (*start == '$') {
1918 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1923 return *s == '(' ? FUNCMETH : METHOD;
1925 if (!keyword(tmpbuf, len)) {
1926 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1931 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1932 if (indirgv && GvCVu(indirgv))
1934 /* filehandle or package name makes it a method */
1935 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1937 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1938 return 0; /* no assumptions -- "=>" quotes bearword */
1940 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1941 newSVpvn(tmpbuf,len));
1942 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1946 return *s == '(' ? FUNCMETH : METHOD;
1954 * Return a string of Perl code to load the debugger. If PERL5DB
1955 * is set, it will return the contents of that, otherwise a
1956 * compile-time require of perl5db.pl.
1963 char *pdb = PerlEnv_getenv("PERL5DB");
1967 SETERRNO(0,SS_NORMAL);
1968 return "BEGIN { require 'perl5db.pl' }";
1974 /* Encoded script support. filter_add() effectively inserts a
1975 * 'pre-processing' function into the current source input stream.
1976 * Note that the filter function only applies to the current source file
1977 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1979 * The datasv parameter (which may be NULL) can be used to pass
1980 * private data to this instance of the filter. The filter function
1981 * can recover the SV using the FILTER_DATA macro and use it to
1982 * store private buffers and state information.
1984 * The supplied datasv parameter is upgraded to a PVIO type
1985 * and the IoDIRP/IoANY field is used to store the function pointer,
1986 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1987 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1988 * private use must be set using malloc'd pointers.
1992 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1997 if (!PL_rsfp_filters)
1998 PL_rsfp_filters = newAV();
2000 datasv = NEWSV(255,0);
2001 if (!SvUPGRADE(datasv, SVt_PVIO))
2002 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
2003 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
2004 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
2005 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
2006 (void*)funcp, SvPV_nolen(datasv)));
2007 av_unshift(PL_rsfp_filters, 1);
2008 av_store(PL_rsfp_filters, 0, datasv) ;
2013 /* Delete most recently added instance of this filter function. */
2015 Perl_filter_del(pTHX_ filter_t funcp)
2018 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", (void*)funcp));
2019 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
2021 /* if filter is on top of stack (usual case) just pop it off */
2022 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
2023 if (IoANY(datasv) == (void *)funcp) {
2024 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
2025 IoANY(datasv) = (void *)NULL;
2026 sv_free(av_pop(PL_rsfp_filters));
2030 /* we need to search for the correct entry and clear it */
2031 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
2035 /* Invoke the n'th filter function for the current rsfp. */
2037 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
2040 /* 0 = read one text line */
2045 if (!PL_rsfp_filters)
2047 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
2048 /* Provide a default input filter to make life easy. */
2049 /* Note that we append to the line. This is handy. */
2050 DEBUG_P(PerlIO_printf(Perl_debug_log,
2051 "filter_read %d: from rsfp\n", idx));
2055 int old_len = SvCUR(buf_sv) ;
2057 /* ensure buf_sv is large enough */
2058 SvGROW(buf_sv, (STRLEN)(old_len + maxlen)) ;
2059 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
2060 if (PerlIO_error(PL_rsfp))
2061 return -1; /* error */
2063 return 0 ; /* end of file */
2065 SvCUR_set(buf_sv, old_len + len) ;
2068 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2069 if (PerlIO_error(PL_rsfp))
2070 return -1; /* error */
2072 return 0 ; /* end of file */
2075 return SvCUR(buf_sv);
2077 /* Skip this filter slot if filter has been deleted */
2078 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
2079 DEBUG_P(PerlIO_printf(Perl_debug_log,
2080 "filter_read %d: skipped (filter deleted)\n",
2082 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
2084 /* Get function pointer hidden within datasv */
2085 funcp = (filter_t)IoANY(datasv);
2086 DEBUG_P(PerlIO_printf(Perl_debug_log,
2087 "filter_read %d: via function %p (%s)\n",
2088 idx, (void*)funcp, SvPV_nolen(datasv)));
2089 /* Call function. The function is expected to */
2090 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2091 /* Return: <0:error, =0:eof, >0:not eof */
2092 return (*funcp)(aTHX_ idx, buf_sv, maxlen);
2096 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2098 #ifdef PERL_CR_FILTER
2099 if (!PL_rsfp_filters) {
2100 filter_add(S_cr_textfilter,NULL);
2103 if (PL_rsfp_filters) {
2106 SvCUR_set(sv, 0); /* start with empty line */
2107 if (FILTER_READ(0, sv, 0) > 0)
2108 return ( SvPVX(sv) ) ;
2113 return (sv_gets(sv, fp, append));
2117 S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2121 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2125 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2126 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2128 return GvHV(gv); /* Foo:: */
2131 /* use constant CLASS => 'MyClass' */
2132 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2134 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2135 pkgname = SvPV_nolen(sv);
2139 return gv_stashpv(pkgname, FALSE);
2143 static char* exp_name[] =
2144 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2145 "ATTRTERM", "TERMBLOCK", "TERMORDORDOR"
2152 Works out what to call the token just pulled out of the input
2153 stream. The yacc parser takes care of taking the ops we return and
2154 stitching them into a tree.
2160 if read an identifier
2161 if we're in a my declaration
2162 croak if they tried to say my($foo::bar)
2163 build the ops for a my() declaration
2164 if it's an access to a my() variable
2165 are we in a sort block?
2166 croak if my($a); $a <=> $b
2167 build ops for access to a my() variable
2168 if in a dq string, and they've said @foo and we can't find @foo
2170 build ops for a bareword
2171 if we already built the token before, use it.
2174 #ifdef USE_PURE_BISON
2176 Perl_yylex_r(pTHX_ YYSTYPE *lvalp, int *lcharp)
2181 yylval_pointer[yyactlevel] = lvalp;
2182 yychar_pointer[yyactlevel] = lcharp;
2183 if (yyactlevel >= YYMAXLEVEL)
2184 Perl_croak(aTHX_ "panic: YYMAXLEVEL");
2186 r = Perl_yylex(aTHX);
2196 #pragma segment Perl_yylex
2208 I32 orig_keyword = 0;
2210 /* check if there's an identifier for us to look at */
2211 if (PL_pending_ident)
2212 return S_pending_ident(aTHX);
2214 /* no identifier pending identification */
2216 switch (PL_lex_state) {
2218 case LEX_NORMAL: /* Some compilers will produce faster */
2219 case LEX_INTERPNORMAL: /* code if we comment these out. */
2223 /* when we've already built the next token, just pull it out of the queue */
2226 yylval = PL_nextval[PL_nexttoke];
2228 PL_lex_state = PL_lex_defer;
2229 PL_expect = PL_lex_expect;
2230 PL_lex_defer = LEX_NORMAL;
2232 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2233 "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr,
2234 (IV)PL_nexttype[PL_nexttoke]); });
2236 return(PL_nexttype[PL_nexttoke]);
2238 /* interpolated case modifiers like \L \U, including \Q and \E.
2239 when we get here, PL_bufptr is at the \
2241 case LEX_INTERPCASEMOD:
2243 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2244 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2246 /* handle \E or end of string */
2247 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2251 if (PL_lex_casemods) {
2252 oldmod = PL_lex_casestack[--PL_lex_casemods];
2253 PL_lex_casestack[PL_lex_casemods] = '\0';
2255 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2257 PL_lex_state = LEX_INTERPCONCAT;
2261 if (PL_bufptr != PL_bufend)
2263 PL_lex_state = LEX_INTERPCONCAT;
2267 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2268 "### Saw case modifier at '%s'\n", PL_bufptr); });
2270 if (s[1] == '\\' && s[2] == 'E') {
2272 PL_lex_state = LEX_INTERPCONCAT;
2276 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2277 tmp = *s, *s = s[2], s[2] = (char)tmp; /* misordered... */
2278 if (strchr("LU", *s) &&
2279 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U'))) {
2280 PL_lex_casestack[--PL_lex_casemods] = '\0';
2283 if (PL_lex_casemods > 10)
2284 Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2285 PL_lex_casestack[PL_lex_casemods++] = *s;
2286 PL_lex_casestack[PL_lex_casemods] = '\0';
2287 PL_lex_state = LEX_INTERPCONCAT;
2288 PL_nextval[PL_nexttoke].ival = 0;
2291 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2293 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2295 PL_nextval[PL_nexttoke].ival = OP_LC;
2297 PL_nextval[PL_nexttoke].ival = OP_UC;
2299 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2301 Perl_croak(aTHX_ "panic: yylex");
2305 if (PL_lex_starts) {
2314 case LEX_INTERPPUSH:
2315 return sublex_push();
2317 case LEX_INTERPSTART:
2318 if (PL_bufptr == PL_bufend)
2319 return sublex_done();
2320 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2321 "### Interpolated variable at '%s'\n", PL_bufptr); });
2323 PL_lex_dojoin = (*PL_bufptr == '@');
2324 PL_lex_state = LEX_INTERPNORMAL;
2325 if (PL_lex_dojoin) {
2326 PL_nextval[PL_nexttoke].ival = 0;
2328 force_ident("\"", '$');
2329 PL_nextval[PL_nexttoke].ival = 0;
2331 PL_nextval[PL_nexttoke].ival = 0;
2333 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2336 if (PL_lex_starts++) {
2342 case LEX_INTERPENDMAYBE:
2343 if (intuit_more(PL_bufptr)) {
2344 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2350 if (PL_lex_dojoin) {
2351 PL_lex_dojoin = FALSE;
2352 PL_lex_state = LEX_INTERPCONCAT;
2355 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2356 && SvEVALED(PL_lex_repl))
2358 if (PL_bufptr != PL_bufend)
2359 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2360 PL_lex_repl = Nullsv;
2363 case LEX_INTERPCONCAT:
2365 if (PL_lex_brackets)
2366 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2368 if (PL_bufptr == PL_bufend)
2369 return sublex_done();
2371 if (SvIVX(PL_linestr) == '\'') {
2372 SV *sv = newSVsv(PL_linestr);
2375 else if ( PL_hints & HINT_NEW_RE )
2376 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2377 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2381 s = scan_const(PL_bufptr);
2383 PL_lex_state = LEX_INTERPCASEMOD;
2385 PL_lex_state = LEX_INTERPSTART;
2388 if (s != PL_bufptr) {
2389 PL_nextval[PL_nexttoke] = yylval;
2392 if (PL_lex_starts++)
2402 PL_lex_state = LEX_NORMAL;
2403 s = scan_formline(PL_bufptr);
2404 if (!PL_lex_formbrack)
2410 PL_oldoldbufptr = PL_oldbufptr;
2413 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2414 exp_name[PL_expect], s);
2420 if (isIDFIRST_lazy_if(s,UTF))
2422 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2425 goto fake_eof; /* emulate EOF on ^D or ^Z */
2430 if (PL_lex_brackets)
2431 yyerror("Missing right curly or square bracket");
2432 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2433 "### Tokener got EOF\n");
2437 if (s++ < PL_bufend)
2438 goto retry; /* ignore stray nulls */
2441 if (!PL_in_eval && !PL_preambled) {
2442 PL_preambled = TRUE;
2443 sv_setpv(PL_linestr,incl_perldb());
2444 if (SvCUR(PL_linestr))
2445 sv_catpv(PL_linestr,";");
2447 while(AvFILLp(PL_preambleav) >= 0) {
2448 SV *tmpsv = av_shift(PL_preambleav);
2449 sv_catsv(PL_linestr, tmpsv);
2450 sv_catpv(PL_linestr, ";");
2453 sv_free((SV*)PL_preambleav);
2454 PL_preambleav = NULL;
2456 if (PL_minus_n || PL_minus_p) {
2457 sv_catpv(PL_linestr, "LINE: while (<>) {");
2459 sv_catpv(PL_linestr,"chomp;");
2462 if (strchr("/'\"", *PL_splitstr)
2463 && strchr(PL_splitstr + 1, *PL_splitstr))
2464 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
2467 s = "'~#\200\1'"; /* surely one char is unused...*/
2468 while (s[1] && strchr(PL_splitstr, *s)) s++;
2470 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s%c",
2471 "q" + (delim == '\''), delim);
2472 for (s = PL_splitstr; *s; s++) {
2474 sv_catpvn(PL_linestr, "\\", 1);
2475 sv_catpvn(PL_linestr, s, 1);
2477 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2481 sv_catpv(PL_linestr,"our @F=split(' ');");
2484 sv_catpv(PL_linestr, "\n");
2485 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2486 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2487 PL_last_lop = PL_last_uni = Nullch;
2488 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2489 SV *sv = NEWSV(85,0);
2491 sv_upgrade(sv, SVt_PVMG);
2492 sv_setsv(sv,PL_linestr);
2495 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2500 bof = PL_rsfp ? TRUE : FALSE;
2501 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2504 if (PL_preprocess && !PL_in_eval)
2505 (void)PerlProc_pclose(PL_rsfp);
2506 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2507 PerlIO_clearerr(PL_rsfp);
2509 (void)PerlIO_close(PL_rsfp);
2511 PL_doextract = FALSE;
2513 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2514 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2515 sv_catpv(PL_linestr,";}");
2516 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2517 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2518 PL_last_lop = PL_last_uni = Nullch;
2519 PL_minus_n = PL_minus_p = 0;
2522 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2523 PL_last_lop = PL_last_uni = Nullch;
2524 sv_setpv(PL_linestr,"");
2525 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2527 /* if it looks like the start of a BOM, check if it in fact is */
2528 else if (bof && (!*s || *(U8*)s == 0xEF || *(U8*)s >= 0xFE)) {
2529 #ifdef PERLIO_IS_STDIO
2530 # ifdef __GNU_LIBRARY__
2531 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
2532 # define FTELL_FOR_PIPE_IS_BROKEN
2536 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2537 # define FTELL_FOR_PIPE_IS_BROKEN
2542 #ifdef FTELL_FOR_PIPE_IS_BROKEN
2543 /* This loses the possibility to detect the bof
2544 * situation on perl -P when the libc5 is being used.
2545 * Workaround? Maybe attach some extra state to PL_rsfp?
2548 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2550 bof = PerlIO_tell(PL_rsfp) == (Off_t)SvCUR(PL_linestr);
2553 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2554 s = swallow_bom((U8*)s);
2558 /* Incest with pod. */
2559 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2560 sv_setpv(PL_linestr, "");
2561 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2562 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2563 PL_last_lop = PL_last_uni = Nullch;
2564 PL_doextract = FALSE;
2568 } while (PL_doextract);
2569 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2570 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2571 SV *sv = NEWSV(85,0);
2573 sv_upgrade(sv, SVt_PVMG);
2574 sv_setsv(sv,PL_linestr);
2577 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2579 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2580 PL_last_lop = PL_last_uni = Nullch;
2581 if (CopLINE(PL_curcop) == 1) {
2582 while (s < PL_bufend && isSPACE(*s))
2584 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2588 if (*s == '#' && *(s+1) == '!')
2590 #ifdef ALTERNATE_SHEBANG
2592 static char as[] = ALTERNATE_SHEBANG;
2593 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2594 d = s + (sizeof(as) - 1);
2596 #endif /* ALTERNATE_SHEBANG */
2605 while (*d && !isSPACE(*d))
2609 #ifdef ARG_ZERO_IS_SCRIPT
2610 if (ipathend > ipath) {
2612 * HP-UX (at least) sets argv[0] to the script name,
2613 * which makes $^X incorrect. And Digital UNIX and Linux,
2614 * at least, set argv[0] to the basename of the Perl
2615 * interpreter. So, having found "#!", we'll set it right.
2617 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV)); /* $^X */
2618 assert(SvPOK(x) || SvGMAGICAL(x));
2619 if (sv_eq(x, CopFILESV(PL_curcop))) {
2620 sv_setpvn(x, ipath, ipathend - ipath);
2626 char *bstart = SvPV(CopFILESV(PL_curcop),blen);
2627 char *lstart = SvPV(x,llen);
2629 bstart += blen - llen;
2630 if (strnEQ(bstart, lstart, llen) && bstart[-1] == '/') {
2631 sv_setpvn(x, ipath, ipathend - ipath);
2636 TAINT_NOT; /* $^X is always tainted, but that's OK */
2638 #endif /* ARG_ZERO_IS_SCRIPT */
2643 d = instr(s,"perl -");
2645 d = instr(s,"perl");
2647 /* avoid getting into infinite loops when shebang
2648 * line contains "Perl" rather than "perl" */
2650 for (d = ipathend-4; d >= ipath; --d) {
2651 if ((*d == 'p' || *d == 'P')
2652 && !ibcmp(d, "perl", 4))
2662 #ifdef ALTERNATE_SHEBANG
2664 * If the ALTERNATE_SHEBANG on this system starts with a
2665 * character that can be part of a Perl expression, then if
2666 * we see it but not "perl", we're probably looking at the
2667 * start of Perl code, not a request to hand off to some
2668 * other interpreter. Similarly, if "perl" is there, but
2669 * not in the first 'word' of the line, we assume the line
2670 * contains the start of the Perl program.
2672 if (d && *s != '#') {
2674 while (*c && !strchr("; \t\r\n\f\v#", *c))
2677 d = Nullch; /* "perl" not in first word; ignore */
2679 *s = '#'; /* Don't try to parse shebang line */
2681 #endif /* ALTERNATE_SHEBANG */
2682 #ifndef MACOS_TRADITIONAL
2687 !instr(s,"indir") &&
2688 instr(PL_origargv[0],"perl"))
2694 while (s < PL_bufend && isSPACE(*s))
2696 if (s < PL_bufend) {
2697 Newz(899,newargv,PL_origargc+3,char*);
2699 while (s < PL_bufend && !isSPACE(*s))
2702 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2705 newargv = PL_origargv;
2708 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
2710 Perl_croak(aTHX_ "Can't exec %s", ipath);
2714 U32 oldpdb = PL_perldb;
2715 bool oldn = PL_minus_n;
2716 bool oldp = PL_minus_p;
2718 while (*d && !isSPACE(*d)) d++;
2719 while (SPACE_OR_TAB(*d)) d++;
2722 bool switches_done = PL_doswitches;
2724 if (*d == 'M' || *d == 'm') {
2726 while (*d && !isSPACE(*d)) d++;
2727 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2730 d = moreswitches(d);
2732 if (PL_doswitches && !switches_done) {
2733 int argc = PL_origargc;
2734 char **argv = PL_origargv;
2737 } while (argc && argv[0][0] == '-' && argv[0][1]);
2738 init_argv_symbols(argc,argv);
2740 if ((PERLDB_LINE && !oldpdb) ||
2741 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2742 /* if we have already added "LINE: while (<>) {",
2743 we must not do it again */
2745 sv_setpv(PL_linestr, "");
2746 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2747 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2748 PL_last_lop = PL_last_uni = Nullch;
2749 PL_preambled = FALSE;
2751 (void)gv_fetchfile(PL_origfilename);
2754 if (PL_doswitches && !switches_done) {
2755 int argc = PL_origargc;
2756 char **argv = PL_origargv;
2759 } while (argc && argv[0][0] == '-' && argv[0][1]);
2760 init_argv_symbols(argc,argv);
2766 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2768 PL_lex_state = LEX_FORMLINE;
2773 #ifdef PERL_STRICT_CR
2774 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2776 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2778 case ' ': case '\t': case '\f': case 013:
2779 #ifdef MACOS_TRADITIONAL
2786 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2787 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2788 /* handle eval qq[#line 1 "foo"\n ...] */
2789 CopLINE_dec(PL_curcop);
2793 while (s < d && *s != '\n')
2797 else if (s > d) /* Found by Ilya: feed random input to Perl. */
2798 Perl_croak(aTHX_ "panic: input overflow");
2800 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2802 PL_lex_state = LEX_FORMLINE;
2812 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2819 while (s < PL_bufend && SPACE_OR_TAB(*s))
2822 if (strnEQ(s,"=>",2)) {
2823 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2824 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2825 "### Saw unary minus before =>, forcing word '%s'\n", s);
2827 OPERATOR('-'); /* unary minus */
2829 PL_last_uni = PL_oldbufptr;
2831 case 'r': ftst = OP_FTEREAD; break;
2832 case 'w': ftst = OP_FTEWRITE; break;
2833 case 'x': ftst = OP_FTEEXEC; break;
2834 case 'o': ftst = OP_FTEOWNED; break;
2835 case 'R': ftst = OP_FTRREAD; break;
2836 case 'W': ftst = OP_FTRWRITE; break;
2837 case 'X': ftst = OP_FTREXEC; break;
2838 case 'O': ftst = OP_FTROWNED; break;
2839 case 'e': ftst = OP_FTIS; break;
2840 case 'z': ftst = OP_FTZERO; break;
2841 case 's': ftst = OP_FTSIZE; break;
2842 case 'f': ftst = OP_FTFILE; break;
2843 case 'd': ftst = OP_FTDIR; break;
2844 case 'l': ftst = OP_FTLINK; break;
2845 case 'p': ftst = OP_FTPIPE; break;
2846 case 'S': ftst = OP_FTSOCK; break;
2847 case 'u': ftst = OP_FTSUID; break;
2848 case 'g': ftst = OP_FTSGID; break;
2849 case 'k': ftst = OP_FTSVTX; break;
2850 case 'b': ftst = OP_FTBLK; break;
2851 case 'c': ftst = OP_FTCHR; break;
2852 case 't': ftst = OP_FTTTY; break;
2853 case 'T': ftst = OP_FTTEXT; break;
2854 case 'B': ftst = OP_FTBINARY; break;
2855 case 'M': case 'A': case 'C':
2856 gv_fetchpv("\024",TRUE, SVt_PV);
2858 case 'M': ftst = OP_FTMTIME; break;
2859 case 'A': ftst = OP_FTATIME; break;
2860 case 'C': ftst = OP_FTCTIME; break;
2868 PL_last_lop_op = (OPCODE)ftst;
2869 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2870 "### Saw file test %c\n", (int)ftst);
2875 /* Assume it was a minus followed by a one-letter named
2876 * subroutine call (or a -bareword), then. */
2877 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2878 "### %c looked like a file test but was not\n",
2887 if (PL_expect == XOPERATOR)
2892 else if (*s == '>') {
2895 if (isIDFIRST_lazy_if(s,UTF)) {
2896 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2904 if (PL_expect == XOPERATOR)
2907 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2909 OPERATOR('-'); /* unary minus */
2916 if (PL_expect == XOPERATOR)
2921 if (PL_expect == XOPERATOR)
2924 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2930 if (PL_expect != XOPERATOR) {
2931 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2932 PL_expect = XOPERATOR;
2933 force_ident(PL_tokenbuf, '*');
2946 if (PL_expect == XOPERATOR) {
2950 PL_tokenbuf[0] = '%';
2951 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2952 if (!PL_tokenbuf[1]) {
2955 PL_pending_ident = '%';
2974 switch (PL_expect) {
2977 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2979 PL_bufptr = s; /* update in case we back off */
2985 PL_expect = XTERMBLOCK;
2989 while (isIDFIRST_lazy_if(s,UTF)) {
2990 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2991 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2992 if (tmp < 0) tmp = -tmp;
3008 d = scan_str(d,TRUE,TRUE);
3010 /* MUST advance bufptr here to avoid bogus
3011 "at end of line" context messages from yyerror().
3013 PL_bufptr = s + len;
3014 yyerror("Unterminated attribute parameter in attribute list");
3017 return 0; /* EOF indicator */
3021 SV *sv = newSVpvn(s, len);
3022 sv_catsv(sv, PL_lex_stuff);
3023 attrs = append_elem(OP_LIST, attrs,
3024 newSVOP(OP_CONST, 0, sv));
3025 SvREFCNT_dec(PL_lex_stuff);
3026 PL_lex_stuff = Nullsv;
3029 /* NOTE: any CV attrs applied here need to be part of
3030 the CVf_BUILTIN_ATTRS define in cv.h! */
3031 if (!PL_in_my && len == 6 && strnEQ(s, "lvalue", len))
3032 CvLVALUE_on(PL_compcv);
3033 else if (!PL_in_my && len == 6 && strnEQ(s, "locked", len))
3034 CvLOCKED_on(PL_compcv);
3035 else if (!PL_in_my && len == 6 && strnEQ(s, "method", len))
3036 CvMETHOD_on(PL_compcv);
3037 else if (!PL_in_my && len == 9 && strnEQ(s, "assertion", len))
3038 CvASSERTION_on(PL_compcv);
3040 else if (PL_in_my == KEY_our && len == 6 &&
3041 strnEQ(s, "unique", len))
3042 GvUNIQUE_on(cGVOPx_gv(yylval.opval));
3044 /* After we've set the flags, it could be argued that
3045 we don't need to do the attributes.pm-based setting
3046 process, and shouldn't bother appending recognized
3047 flags. To experiment with that, uncomment the
3048 following "else". (Note that's already been
3049 uncommented. That keeps the above-applied built-in
3050 attributes from being intercepted (and possibly
3051 rejected) by a package's attribute routines, but is
3052 justified by the performance win for the common case
3053 of applying only built-in attributes.) */
3055 attrs = append_elem(OP_LIST, attrs,
3056 newSVOP(OP_CONST, 0,
3060 if (*s == ':' && s[1] != ':')
3063 break; /* require real whitespace or :'s */
3065 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3066 if (*s != ';' && *s != '}' && *s != tmp && (tmp != '=' || *s != ')')) {
3067 char q = ((*s == '\'') ? '"' : '\'');
3068 /* If here for an expression, and parsed no attrs, back off. */
3069 if (tmp == '=' && !attrs) {
3073 /* MUST advance bufptr here to avoid bogus "at end of line"
3074 context messages from yyerror().
3078 yyerror("Unterminated attribute list");
3080 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
3088 PL_nextval[PL_nexttoke].opval = attrs;
3096 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3097 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3114 if (PL_lex_brackets <= 0)
3115 yyerror("Unmatched right square bracket");
3118 if (PL_lex_state == LEX_INTERPNORMAL) {
3119 if (PL_lex_brackets == 0) {
3120 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3121 PL_lex_state = LEX_INTERPEND;
3128 if (PL_lex_brackets > 100) {
3129 Renew(PL_lex_brackstack, PL_lex_brackets + 10, char);
3131 switch (PL_expect) {
3133 if (PL_lex_formbrack) {
3137 if (PL_oldoldbufptr == PL_last_lop)
3138 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3140 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3141 OPERATOR(HASHBRACK);
3143 while (s < PL_bufend && SPACE_OR_TAB(*s))
3146 PL_tokenbuf[0] = '\0';
3147 if (d < PL_bufend && *d == '-') {
3148 PL_tokenbuf[0] = '-';
3150 while (d < PL_bufend && SPACE_OR_TAB(*d))
3153 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3154 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3156 while (d < PL_bufend && SPACE_OR_TAB(*d))
3159 char minus = (PL_tokenbuf[0] == '-');
3160 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3168 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3173 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3178 if (PL_oldoldbufptr == PL_last_lop)
3179 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3181 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3184 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
3186 /* This hack is to get the ${} in the message. */
3188 yyerror("syntax error");
3191 OPERATOR(HASHBRACK);
3193 /* This hack serves to disambiguate a pair of curlies
3194 * as being a block or an anon hash. Normally, expectation
3195 * determines that, but in cases where we're not in a
3196 * position to expect anything in particular (like inside
3197 * eval"") we have to resolve the ambiguity. This code
3198 * covers the case where the first term in the curlies is a
3199 * quoted string. Most other cases need to be explicitly
3200 * disambiguated by prepending a `+' before the opening
3201 * curly in order to force resolution as an anon hash.
3203 * XXX should probably propagate the outer expectation
3204 * into eval"" to rely less on this hack, but that could
3205 * potentially break current behavior of eval"".
3209 if (*s == '\'' || *s == '"' || *s == '`') {
3210 /* common case: get past first string, handling escapes */
3211 for (t++; t < PL_bufend && *t != *s;)
3212 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3216 else if (*s == 'q') {
3219 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3222 /* skip q//-like construct */
3224 char open, close, term;
3227 while (t < PL_bufend && isSPACE(*t))
3229 /* check for q => */
3230 if (t+1 < PL_bufend && t[0] == '=' && t[1] == '>') {
3231 OPERATOR(HASHBRACK);
3235 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3239 for (t++; t < PL_bufend; t++) {
3240 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3242 else if (*t == open)
3246 for (t++; t < PL_bufend; t++) {
3247 if (*t == '\\' && t+1 < PL_bufend)
3249 else if (*t == close && --brackets <= 0)
3251 else if (*t == open)
3258 /* skip plain q word */
3259 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3262 else if (isALNUM_lazy_if(t,UTF)) {
3264 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3267 while (t < PL_bufend && isSPACE(*t))
3269 /* if comma follows first term, call it an anon hash */
3270 /* XXX it could be a comma expression with loop modifiers */
3271 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3272 || (*t == '=' && t[1] == '>')))
3273 OPERATOR(HASHBRACK);
3274 if (PL_expect == XREF)
3277 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3283 yylval.ival = CopLINE(PL_curcop);
3284 if (isSPACE(*s) || *s == '#')
3285 PL_copline = NOLINE; /* invalidate current command line number */
3290 if (PL_lex_brackets <= 0)
3291 yyerror("Unmatched right curly bracket");
3293 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3294 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3295 PL_lex_formbrack = 0;
3296 if (PL_lex_state == LEX_INTERPNORMAL) {
3297 if (PL_lex_brackets == 0) {
3298 if (PL_expect & XFAKEBRACK) {
3299 PL_expect &= XENUMMASK;
3300 PL_lex_state = LEX_INTERPEND;
3302 return yylex(); /* ignore fake brackets */
3304 if (*s == '-' && s[1] == '>')
3305 PL_lex_state = LEX_INTERPENDMAYBE;
3306 else if (*s != '[' && *s != '{')
3307 PL_lex_state = LEX_INTERPEND;
3310 if (PL_expect & XFAKEBRACK) {
3311 PL_expect &= XENUMMASK;
3313 return yylex(); /* ignore fake brackets */
3323 if (PL_expect == XOPERATOR) {
3324 if (ckWARN(WARN_SEMICOLON)
3325 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3327 CopLINE_dec(PL_curcop);
3328 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
3329 CopLINE_inc(PL_curcop);
3334 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3336 PL_expect = XOPERATOR;
3337 force_ident(PL_tokenbuf, '&');
3341 yylval.ival = (OPpENTERSUB_AMPER<<8);
3360 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3361 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Reversed %c= operator",(int)tmp);
3363 if (PL_expect == XSTATE && isALPHA(tmp) &&
3364 (s == PL_linestart+1 || s[-2] == '\n') )
3366 if (PL_in_eval && !PL_rsfp) {
3371 if (strnEQ(s,"=cut",4)) {
3385 PL_doextract = TRUE;
3388 if (PL_lex_brackets < PL_lex_formbrack) {
3390 #ifdef PERL_STRICT_CR
3391 for (t = s; SPACE_OR_TAB(*t); t++) ;
3393 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3395 if (*t == '\n' || *t == '#') {
3413 if (PL_expect != XOPERATOR) {
3414 if (s[1] != '<' && !strchr(s,'>'))
3417 s = scan_heredoc(s);
3419 s = scan_inputsymbol(s);
3420 TERM(sublex_start());
3425 SHop(OP_LEFT_SHIFT);
3439 SHop(OP_RIGHT_SHIFT);
3448 if (PL_expect == XOPERATOR) {
3449 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3452 return ','; /* grandfather non-comma-format format */
3456 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3457 PL_tokenbuf[0] = '@';
3458 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3459 sizeof PL_tokenbuf - 1, FALSE);
3460 if (PL_expect == XOPERATOR)
3461 no_op("Array length", s);
3462 if (!PL_tokenbuf[1])
3464 PL_expect = XOPERATOR;
3465 PL_pending_ident = '#';
3469 PL_tokenbuf[0] = '$';
3470 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3471 sizeof PL_tokenbuf - 1, FALSE);
3472 if (PL_expect == XOPERATOR)
3474 if (!PL_tokenbuf[1]) {
3476 yyerror("Final $ should be \\$ or $name");
3480 /* This kludge not intended to be bulletproof. */
3481 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3482 yylval.opval = newSVOP(OP_CONST, 0,
3483 newSViv(PL_compiling.cop_arybase));
3484 yylval.opval->op_private = OPpCONST_ARYBASE;
3490 if (PL_lex_state == LEX_NORMAL)
3493 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3496 PL_tokenbuf[0] = '@';
3497 if (ckWARN(WARN_SYNTAX)) {
3499 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3502 PL_bufptr = skipspace(PL_bufptr);
3503 while (t < PL_bufend && *t != ']')
3505 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3506 "Multidimensional syntax %.*s not supported",
3507 (t - PL_bufptr) + 1, PL_bufptr);
3511 else if (*s == '{') {
3512 PL_tokenbuf[0] = '%';
3513 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3514 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3516 char tmpbuf[sizeof PL_tokenbuf];
3518 for (t++; isSPACE(*t); t++) ;
3519 if (isIDFIRST_lazy_if(t,UTF)) {
3520 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3521 for (; isSPACE(*t); t++) ;
3522 if (*t == ';' && get_cv(tmpbuf, FALSE))
3523 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3524 "You need to quote \"%s\"", tmpbuf);
3530 PL_expect = XOPERATOR;
3531 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3532 bool islop = (PL_last_lop == PL_oldoldbufptr);
3533 if (!islop || PL_last_lop_op == OP_GREPSTART)
3534 PL_expect = XOPERATOR;
3535 else if (strchr("$@\"'`q", *s))
3536 PL_expect = XTERM; /* e.g. print $fh "foo" */
3537 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3538 PL_expect = XTERM; /* e.g. print $fh &sub */
3539 else if (isIDFIRST_lazy_if(s,UTF)) {
3540 char tmpbuf[sizeof PL_tokenbuf];
3541 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3542 if ((tmp = keyword(tmpbuf, len))) {
3543 /* binary operators exclude handle interpretations */
3555 PL_expect = XTERM; /* e.g. print $fh length() */
3560 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3561 if (gv && GvCVu(gv))
3562 PL_expect = XTERM; /* e.g. print $fh subr() */
3565 else if (isDIGIT(*s))
3566 PL_expect = XTERM; /* e.g. print $fh 3 */
3567 else if (*s == '.' && isDIGIT(s[1]))
3568 PL_expect = XTERM; /* e.g. print $fh .3 */
3569 else if (strchr("?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3570 PL_expect = XTERM; /* e.g. print $fh -1 */
3571 else if (*s == '/' && !isSPACE(s[1]) && s[1] != '=' && s[1] != '/')
3572 PL_expect = XTERM; /* e.g. print $fh /.../
3573 XXX except DORDOR operator */
3574 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3575 PL_expect = XTERM; /* print $fh <<"EOF" */
3577 PL_pending_ident = '$';
3581 if (PL_expect == XOPERATOR)
3583 PL_tokenbuf[0] = '@';
3584 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3585 if (!PL_tokenbuf[1]) {
3588 if (PL_lex_state == LEX_NORMAL)
3590 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3592 PL_tokenbuf[0] = '%';
3594 /* Warn about @ where they meant $. */
3595 if (ckWARN(WARN_SYNTAX)) {
3596 if (*s == '[' || *s == '{') {
3598 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3600 if (*t == '}' || *t == ']') {
3602 PL_bufptr = skipspace(PL_bufptr);
3603 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3604 "Scalar value %.*s better written as $%.*s",
3605 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3610 PL_pending_ident = '@';
3613 case '/': /* may be division, defined-or, or pattern */
3614 if (PL_expect == XTERMORDORDOR && s[1] == '/') {
3618 case '?': /* may either be conditional or pattern */
3619 if(PL_expect == XOPERATOR) {
3627 /* A // operator. */
3637 /* Disable warning on "study /blah/" */
3638 if (PL_oldoldbufptr == PL_last_uni
3639 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3640 || memNE(PL_last_uni, "study", 5)
3641 || isALNUM_lazy_if(PL_last_uni+5,UTF)
3644 s = scan_pat(s,OP_MATCH);
3645 TERM(sublex_start());
3649 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3650 #ifdef PERL_STRICT_CR
3653 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3655 && (s == PL_linestart || s[-1] == '\n') )
3657 PL_lex_formbrack = 0;
3661 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3667 yylval.ival = OPf_SPECIAL;
3673 if (PL_expect != XOPERATOR)
3678 case '0': case '1': case '2': case '3': case '4':
3679 case '5': case '6': case '7': case '8': case '9':
3680 s = scan_num(s, &yylval);
3681 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3682 "### Saw number in '%s'\n", s);
3684 if (PL_expect == XOPERATOR)
3689 s = scan_str(s,FALSE,FALSE);
3690 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3691 "### Saw string before '%s'\n", s);
3693 if (PL_expect == XOPERATOR) {
3694 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3697 return ','; /* grandfather non-comma-format format */
3703 missingterm((char*)0);
3704 yylval.ival = OP_CONST;
3705 TERM(sublex_start());
3708 s = scan_str(s,FALSE,FALSE);
3709 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3710 "### Saw string before '%s'\n", s);
3712 if (PL_expect == XOPERATOR) {
3713 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3716 return ','; /* grandfather non-comma-format format */
3722 missingterm((char*)0);
3723 yylval.ival = OP_CONST;
3724 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3725 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
3726 yylval.ival = OP_STRINGIFY;
3730 TERM(sublex_start());
3733 s = scan_str(s,FALSE,FALSE);
3734 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3735 "### Saw backtick string before '%s'\n", s);
3737 if (PL_expect == XOPERATOR)
3738 no_op("Backticks",s);
3740 missingterm((char*)0);
3741 yylval.ival = OP_BACKTICK;
3743 TERM(sublex_start());
3747 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3748 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),"Can't use \\%c to mean $%c in expression",
3750 if (PL_expect == XOPERATOR)
3751 no_op("Backslash",s);
3755 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3759 while (isDIGIT(*start) || *start == '_')
3761 if (*start == '.' && isDIGIT(start[1])) {
3762 s = scan_num(s, &yylval);
3765 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3766 else if (!isALPHA(*start) && (PL_expect == XTERM
3767 || PL_expect == XREF || PL_expect == XSTATE
3768 || PL_expect == XTERMORDORDOR)) {
3772 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3775 s = scan_num(s, &yylval);
3782 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3822 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3824 /* Some keywords can be followed by any delimiter, including ':' */
3825 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3826 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3827 (PL_tokenbuf[0] == 'q' &&
3828 strchr("qwxr", PL_tokenbuf[1])))));
3830 /* x::* is just a word, unless x is "CORE" */
3831 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3835 while (d < PL_bufend && isSPACE(*d))
3836 d++; /* no comments skipped here, or s### is misparsed */
3838 /* Is this a label? */
3839 if (!tmp && PL_expect == XSTATE
3840 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3842 yylval.pval = savepv(PL_tokenbuf);
3847 /* Check for keywords */
3848 tmp = keyword(PL_tokenbuf, len);
3850 /* Is this a word before a => operator? */
3851 if (*d == '=' && d[1] == '>') {
3853 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3854 yylval.opval->op_private = OPpCONST_BARE;
3855 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
3856 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3860 if (tmp < 0) { /* second-class keyword? */
3861 GV *ogv = Nullgv; /* override (winner) */
3862 GV *hgv = Nullgv; /* hidden (loser) */
3863 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3865 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3868 if (GvIMPORTED_CV(gv))
3870 else if (! CvMETHOD(cv))
3874 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3875 (gv = *gvp) != (GV*)&PL_sv_undef &&
3876 GvCVu(gv) && GvIMPORTED_CV(gv))
3883 tmp = 0; /* overridden by import or by GLOBAL */
3886 && -tmp==KEY_lock /* XXX generalizable kludge */
3888 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3890 tmp = 0; /* any sub overrides "weak" keyword */
3892 else { /* no override */
3894 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
3895 Perl_warner(aTHX_ packWARN(WARN_MISC),
3896 "dump() better written as CORE::dump()");
3900 if (ckWARN(WARN_AMBIGUOUS) && hgv
3901 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3902 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
3903 "Ambiguous call resolved as CORE::%s(), %s",
3904 GvENAME(hgv), "qualify as such or use &");
3911 default: /* not a keyword */
3915 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3917 /* Get the rest if it looks like a package qualifier */
3919 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3921 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3924 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3925 *s == '\'' ? "'" : "::");
3930 if (PL_expect == XOPERATOR) {
3931 if (PL_bufptr == PL_linestart) {
3932 CopLINE_dec(PL_curcop);
3933 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
3934 CopLINE_inc(PL_curcop);
3937 no_op("Bareword",s);
3940 /* Look for a subroutine with this name in current package,
3941 unless name is "Foo::", in which case Foo is a bearword
3942 (and a package name). */
3945 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3947 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3948 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
3949 "Bareword \"%s\" refers to nonexistent package",
3952 PL_tokenbuf[len] = '\0';
3959 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3962 /* if we saw a global override before, get the right name */
3965 sv = newSVpvn("CORE::GLOBAL::",14);
3966 sv_catpv(sv,PL_tokenbuf);
3969 sv = newSVpv(PL_tokenbuf,0);
3971 /* Presume this is going to be a bareword of some sort. */
3974 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3975 yylval.opval->op_private = OPpCONST_BARE;
3976 /* UTF-8 package name? */
3977 if (UTF && !IN_BYTES &&
3978 is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
3981 /* And if "Foo::", then that's what it certainly is. */
3986 /* See if it's the indirect object for a list operator. */
3988 if (PL_oldoldbufptr &&
3989 PL_oldoldbufptr < PL_bufptr &&
3990 (PL_oldoldbufptr == PL_last_lop
3991 || PL_oldoldbufptr == PL_last_uni) &&
3992 /* NO SKIPSPACE BEFORE HERE! */
3993 (PL_expect == XREF ||
3994 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3996 bool immediate_paren = *s == '(';
3998 /* (Now we can afford to cross potential line boundary.) */
4001 /* Two barewords in a row may indicate method call. */
4003 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
4006 /* If not a declared subroutine, it's an indirect object. */
4007 /* (But it's an indir obj regardless for sort.) */
4009 if ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
4010 ((!gv || !GvCVu(gv)) &&
4011 (PL_last_lop_op != OP_MAPSTART &&
4012 PL_last_lop_op != OP_GREPSTART))))
4014 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
4019 PL_expect = XOPERATOR;
4022 /* Is this a word before a => operator? */
4023 if (*s == '=' && s[1] == '>' && !pkgname) {
4025 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
4026 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
4027 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
4031 /* If followed by a paren, it's certainly a subroutine. */
4034 if (gv && GvCVu(gv)) {
4035 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
4036 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
4041 PL_nextval[PL_nexttoke].opval = yylval.opval;
4042 PL_expect = XOPERATOR;
4048 /* If followed by var or block, call it a method (unless sub) */
4050 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
4051 PL_last_lop = PL_oldbufptr;
4052 PL_last_lop_op = OP_METHOD;
4056 /* If followed by a bareword, see if it looks like indir obj. */
4059 && (isIDFIRST_lazy_if(s,UTF) || *s == '$')
4060 && (tmp = intuit_method(s,gv)))
4063 /* Not a method, so call it a subroutine (if defined) */
4065 if (gv && GvCVu(gv)) {
4067 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
4068 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4069 "Ambiguous use of -%s resolved as -&%s()",
4070 PL_tokenbuf, PL_tokenbuf);
4071 /* Check for a constant sub */
4073 if ((sv = cv_const_sv(cv))) {
4075 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
4076 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
4077 yylval.opval->op_private = 0;
4081 /* Resolve to GV now. */
4082 op_free(yylval.opval);
4083 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
4084 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
4085 PL_last_lop = PL_oldbufptr;
4086 PL_last_lop_op = OP_ENTERSUB;
4087 /* Is there a prototype? */
4090 char *proto = SvPV((SV*)cv, len);
4093 if (strEQ(proto, "$"))
4095 while (*proto == ';')
4097 if (*proto == '&' && *s == '{') {
4098 sv_setpv(PL_subname, PL_curstash ?
4099 "__ANON__" : "__ANON__::__ANON__");
4103 PL_nextval[PL_nexttoke].opval = yylval.opval;
4109 /* Call it a bare word */
4111 if (PL_hints & HINT_STRICT_SUBS)
4112 yylval.opval->op_private |= OPpCONST_STRICT;
4115 if (ckWARN(WARN_RESERVED)) {
4116 if (lastchar != '-') {
4117 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4118 if (!*d && !gv_stashpv(PL_tokenbuf,FALSE))
4119 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
4126 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
4127 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4128 "Operator or semicolon missing before %c%s",
4129 lastchar, PL_tokenbuf);
4130 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4131 "Ambiguous use of %c resolved as operator %c",
4132 lastchar, lastchar);
4138 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4139 newSVpv(CopFILE(PL_curcop),0));
4143 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4144 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4147 case KEY___PACKAGE__:
4148 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4150 ? newSVsv(PL_curstname)
4159 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4160 char *pname = "main";
4161 if (PL_tokenbuf[2] == 'D')
4162 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
4163 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4166 GvIOp(gv) = newIO();
4167 IoIFP(GvIOp(gv)) = PL_rsfp;
4168 #if defined(HAS_FCNTL) && defined(F_SETFD)
4170 int fd = PerlIO_fileno(PL_rsfp);
4171 fcntl(fd,F_SETFD,fd >= 3);
4174 /* Mark this internal pseudo-handle as clean */
4175 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4177 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4178 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4179 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4181 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4182 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4183 /* if the script was opened in binmode, we need to revert
4184 * it to text mode for compatibility; but only iff it has CRs
4185 * XXX this is a questionable hack at best. */
4186 if (PL_bufend-PL_bufptr > 2
4187 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4190 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4191 loc = PerlIO_tell(PL_rsfp);
4192 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4195 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
4197 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4198 #endif /* NETWARE */
4199 #ifdef PERLIO_IS_STDIO /* really? */
4200 # if defined(__BORLANDC__)
4201 /* XXX see note in do_binmode() */
4202 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
4206 PerlIO_seek(PL_rsfp, loc, 0);
4210 #ifdef PERLIO_LAYERS
4213 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
4214 else if (PL_encoding) {
4221 XPUSHs(PL_encoding);
4223 call_method("name", G_SCALAR);
4227 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL,
4228 Perl_form(aTHX_ ":encoding(%"SVf")",
4246 if (PL_expect == XSTATE) {
4253 if (*s == ':' && s[1] == ':') {
4256 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4257 if (!(tmp = keyword(PL_tokenbuf, len)))
4258 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4272 LOP(OP_ACCEPT,XTERM);
4278 LOP(OP_ATAN2,XTERM);
4284 LOP(OP_BINMODE,XTERM);
4287 LOP(OP_BLESS,XTERM);
4296 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4313 if (!PL_cryptseen) {
4314 PL_cryptseen = TRUE;
4318 LOP(OP_CRYPT,XTERM);
4321 LOP(OP_CHMOD,XTERM);
4324 LOP(OP_CHOWN,XTERM);
4327 LOP(OP_CONNECT,XTERM);
4343 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4347 PL_hints |= HINT_BLOCK_SCOPE;
4357 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4358 LOP(OP_DBMOPEN,XTERM);
4364 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4371 yylval.ival = CopLINE(PL_curcop);
4385 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4386 UNIBRACK(OP_ENTEREVAL);
4404 case KEY_endhostent:
4410 case KEY_endservent:
4413 case KEY_endprotoent:
4424 yylval.ival = CopLINE(PL_curcop);
4426 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4428 if ((PL_bufend - p) >= 3 &&
4429 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4431 else if ((PL_bufend - p) >= 4 &&
4432 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4435 if (isIDFIRST_lazy_if(p,UTF)) {
4436 p = scan_ident(p, PL_bufend,
4437 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4441 Perl_croak(aTHX_ "Missing $ on loop variable");
4446 LOP(OP_FORMLINE,XTERM);
4452 LOP(OP_FCNTL,XTERM);
4458 LOP(OP_FLOCK,XTERM);
4467 LOP(OP_GREPSTART, XREF);
4470 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4485 case KEY_getpriority:
4486 LOP(OP_GETPRIORITY,XTERM);
4488 case KEY_getprotobyname:
4491 case KEY_getprotobynumber:
4492 LOP(OP_GPBYNUMBER,XTERM);
4494 case KEY_getprotoent:
4506 case KEY_getpeername:
4507 UNI(OP_GETPEERNAME);
4509 case KEY_gethostbyname:
4512 case KEY_gethostbyaddr:
4513 LOP(OP_GHBYADDR,XTERM);
4515 case KEY_gethostent:
4518 case KEY_getnetbyname:
4521 case KEY_getnetbyaddr:
4522 LOP(OP_GNBYADDR,XTERM);
4527 case KEY_getservbyname:
4528 LOP(OP_GSBYNAME,XTERM);
4530 case KEY_getservbyport:
4531 LOP(OP_GSBYPORT,XTERM);
4533 case KEY_getservent:
4536 case KEY_getsockname:
4537 UNI(OP_GETSOCKNAME);
4539 case KEY_getsockopt:
4540 LOP(OP_GSOCKOPT,XTERM);
4562 yylval.ival = CopLINE(PL_curcop);
4566 LOP(OP_INDEX,XTERM);
4572 LOP(OP_IOCTL,XTERM);
4584 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4616 LOP(OP_LISTEN,XTERM);
4625 s = scan_pat(s,OP_MATCH);
4626 TERM(sublex_start());
4629 LOP(OP_MAPSTART, XREF);
4632 LOP(OP_MKDIR,XTERM);
4635 LOP(OP_MSGCTL,XTERM);
4638 LOP(OP_MSGGET,XTERM);
4641 LOP(OP_MSGRCV,XTERM);
4644 LOP(OP_MSGSND,XTERM);
4650 if (isIDFIRST_lazy_if(s,UTF)) {
4651 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4652 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4654 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4655 if (!PL_in_my_stash) {
4658 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4666 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4673 if (PL_expect != XSTATE)
4674 yyerror("\"no\" not allowed in expression");
4675 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4676 s = force_version(s, FALSE);
4681 if (*s == '(' || (s = skipspace(s), *s == '('))
4688 if (isIDFIRST_lazy_if(s,UTF)) {
4690 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4692 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)
4694 && !(t[0] == '=' && t[1] == '>')
4696 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
4697 "Precedence problem: open %.*s should be open(%.*s)",
4698 d - s, s, d - s, s);
4704 yylval.ival = OP_OR;
4714 LOP(OP_OPEN_DIR,XTERM);
4717 checkcomma(s,PL_tokenbuf,"filehandle");
4721 checkcomma(s,PL_tokenbuf,"filehandle");
4740 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4744 LOP(OP_PIPE_OP,XTERM);
4747 s = scan_str(s,FALSE,FALSE);
4749 missingterm((char*)0);
4750 yylval.ival = OP_CONST;
4751 TERM(sublex_start());
4757 s = scan_str(s,FALSE,FALSE);
4759 missingterm((char*)0);
4761 if (SvCUR(PL_lex_stuff)) {
4764 d = SvPV_force(PL_lex_stuff, len);
4767 for (; isSPACE(*d) && len; --len, ++d) ;
4770 if (!warned && ckWARN(WARN_QW)) {
4771 for (; !isSPACE(*d) && len; --len, ++d) {
4773 Perl_warner(aTHX_ packWARN(WARN_QW),
4774 "Possible attempt to separate words with commas");
4777 else if (*d == '#') {
4778 Perl_warner(aTHX_ packWARN(WARN_QW),
4779 "Possible attempt to put comments in qw() list");
4785 for (; !isSPACE(*d) && len; --len, ++d) ;
4787 sv = newSVpvn(b, d-b);
4788 if (DO_UTF8(PL_lex_stuff))
4790 words = append_elem(OP_LIST, words,
4791 newSVOP(OP_CONST, 0, tokeq(sv)));
4795 PL_nextval[PL_nexttoke].opval = words;
4800 SvREFCNT_dec(PL_lex_stuff);
4801 PL_lex_stuff = Nullsv;
4807 s = scan_str(s,FALSE,FALSE);
4809 missingterm((char*)0);
4810 yylval.ival = OP_STRINGIFY;
4811 if (SvIVX(PL_lex_stuff) == '\'')
4812 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4813 TERM(sublex_start());
4816 s = scan_pat(s,OP_QR);
4817 TERM(sublex_start());
4820 s = scan_str(s,FALSE,FALSE);
4822 missingterm((char*)0);
4823 yylval.ival = OP_BACKTICK;
4825 TERM(sublex_start());
4833 s = force_version(s, FALSE);
4835 else if (*s != 'v' || !isDIGIT(s[1])
4836 || (s = force_version(s, TRUE), *s == 'v'))
4838 *PL_tokenbuf = '\0';
4839 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4840 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4841 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4843 yyerror("<> should be quotes");
4851 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4855 LOP(OP_RENAME,XTERM);
4864 LOP(OP_RINDEX,XTERM);
4874 UNIDOR(OP_READLINE);
4887 LOP(OP_REVERSE,XTERM);
4890 UNIDOR(OP_READLINK);
4898 TERM(sublex_start());
4900 TOKEN(1); /* force error */
4909 LOP(OP_SELECT,XTERM);
4915 LOP(OP_SEMCTL,XTERM);
4918 LOP(OP_SEMGET,XTERM);
4921 LOP(OP_SEMOP,XTERM);
4927 LOP(OP_SETPGRP,XTERM);
4929 case KEY_setpriority:
4930 LOP(OP_SETPRIORITY,XTERM);
4932 case KEY_sethostent:
4938 case KEY_setservent:
4941 case KEY_setprotoent:
4951 LOP(OP_SEEKDIR,XTERM);
4953 case KEY_setsockopt:
4954 LOP(OP_SSOCKOPT,XTERM);
4960 LOP(OP_SHMCTL,XTERM);
4963 LOP(OP_SHMGET,XTERM);
4966 LOP(OP_SHMREAD,XTERM);
4969 LOP(OP_SHMWRITE,XTERM);
4972 LOP(OP_SHUTDOWN,XTERM);
4981 LOP(OP_SOCKET,XTERM);
4983 case KEY_socketpair:
4984 LOP(OP_SOCKPAIR,XTERM);
4987 checkcomma(s,PL_tokenbuf,"subroutine name");
4989 if (*s == ';' || *s == ')') /* probably a close */
4990 Perl_croak(aTHX_ "sort is now a reserved word");
4992 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4996 LOP(OP_SPLIT,XTERM);
4999 LOP(OP_SPRINTF,XTERM);
5002 LOP(OP_SPLICE,XTERM);
5017 LOP(OP_SUBSTR,XTERM);
5023 char tmpbuf[sizeof PL_tokenbuf];
5024 SSize_t tboffset = 0;
5025 expectation attrful;
5026 bool have_name, have_proto, bad_proto;
5031 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
5032 (*s == ':' && s[1] == ':'))
5035 attrful = XATTRBLOCK;
5036 /* remember buffer pos'n for later force_word */
5037 tboffset = s - PL_oldbufptr;
5038 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
5039 if (strchr(tmpbuf, ':'))
5040 sv_setpv(PL_subname, tmpbuf);
5042 sv_setsv(PL_subname,PL_curstname);
5043 sv_catpvn(PL_subname,"::",2);
5044 sv_catpvn(PL_subname,tmpbuf,len);
5051 Perl_croak(aTHX_ "Missing name in \"my sub\"");
5052 PL_expect = XTERMBLOCK;
5053 attrful = XATTRTERM;
5054 sv_setpv(PL_subname,"?");
5058 if (key == KEY_format) {
5060 PL_lex_formbrack = PL_lex_brackets + 1;
5062 (void) force_word(PL_oldbufptr + tboffset, WORD,
5067 /* Look for a prototype */
5071 s = scan_str(s,FALSE,FALSE);
5073 Perl_croak(aTHX_ "Prototype not terminated");
5074 /* strip spaces and check for bad characters */
5075 d = SvPVX(PL_lex_stuff);
5078 for (p = d; *p; ++p) {
5081 if (!strchr("$@%*;[]&\\", *p))
5086 if (bad_proto && ckWARN(WARN_SYNTAX))
5087 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5088 "Illegal character in prototype for %"SVf" : %s",
5090 SvCUR(PL_lex_stuff) = tmp;
5098 if (*s == ':' && s[1] != ':')
5099 PL_expect = attrful;
5100 else if (!have_name && *s != '{' && key == KEY_sub)
5101 Perl_croak(aTHX_ "Illegal declaration of anonymous subroutine");
5104 PL_nextval[PL_nexttoke].opval =
5105 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
5106 PL_lex_stuff = Nullsv;
5110 sv_setpv(PL_subname,
5111 PL_curstash ? "__ANON__" : "__ANON__::__ANON__");
5114 (void) force_word(PL_oldbufptr + tboffset, WORD,
5123 LOP(OP_SYSTEM,XREF);
5126 LOP(OP_SYMLINK,XTERM);
5129 LOP(OP_SYSCALL,XTERM);
5132 LOP(OP_SYSOPEN,XTERM);
5135 LOP(OP_SYSSEEK,XTERM);
5138 LOP(OP_SYSREAD,XTERM);
5141 LOP(OP_SYSWRITE,XTERM);
5145 TERM(sublex_start());
5166 LOP(OP_TRUNCATE,XTERM);
5178 yylval.ival = CopLINE(PL_curcop);
5182 yylval.ival = CopLINE(PL_curcop);
5186 LOP(OP_UNLINK,XTERM);
5192 LOP(OP_UNPACK,XTERM);
5195 LOP(OP_UTIME,XTERM);
5201 LOP(OP_UNSHIFT,XTERM);
5204 if (PL_expect != XSTATE)
5205 yyerror("\"use\" not allowed in expression");
5207 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
5208 s = force_version(s, TRUE);
5209 if (*s == ';' || (s = skipspace(s), *s == ';')) {
5210 PL_nextval[PL_nexttoke].opval = Nullop;
5213 else if (*s == 'v') {
5214 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5215 s = force_version(s, FALSE);
5219 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5220 s = force_version(s, FALSE);
5232 yylval.ival = CopLINE(PL_curcop);
5236 PL_hints |= HINT_BLOCK_SCOPE;
5243 LOP(OP_WAITPID,XTERM);
5252 ctl_l[0] = toCTRL('L');
5254 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5257 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5262 if (PL_expect == XOPERATOR)
5268 yylval.ival = OP_XOR;
5273 TERM(sublex_start());
5278 #pragma segment Main
5282 S_pending_ident(pTHX)
5285 register I32 tmp = 0;
5286 /* pit holds the identifier we read and pending_ident is reset */
5287 char pit = PL_pending_ident;
5288 PL_pending_ident = 0;
5290 DEBUG_T({ PerlIO_printf(Perl_debug_log,
5291 "### Tokener saw identifier '%s'\n", PL_tokenbuf); });
5293 /* if we're in a my(), we can't allow dynamics here.
5294 $foo'bar has already been turned into $foo::bar, so
5295 just check for colons.
5297 if it's a legal name, the OP is a PADANY.
5300 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
5301 if (strchr(PL_tokenbuf,':'))
5302 yyerror(Perl_form(aTHX_ "No package name allowed for "
5303 "variable %s in \"our\"",
5305 tmp = allocmy(PL_tokenbuf);
5308 if (strchr(PL_tokenbuf,':'))
5309 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
5311 yylval.opval = newOP(OP_PADANY, 0);
5312 yylval.opval->op_targ = allocmy(PL_tokenbuf);
5318 build the ops for accesses to a my() variable.
5320 Deny my($a) or my($b) in a sort block, *if* $a or $b is
5321 then used in a comparison. This catches most, but not
5322 all cases. For instance, it catches
5323 sort { my($a); $a <=> $b }
5325 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
5326 (although why you'd do that is anyone's guess).
5329 if (!strchr(PL_tokenbuf,':')) {
5331 tmp = pad_findmy(PL_tokenbuf);
5332 if (tmp != NOT_IN_PAD) {
5333 /* might be an "our" variable" */
5334 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
5335 /* build ops for a bareword */
5336 SV *sym = newSVpv(HvNAME(PAD_COMPNAME_OURSTASH(tmp)), 0);
5337 sv_catpvn(sym, "::", 2);
5338 sv_catpv(sym, PL_tokenbuf+1);
5339 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
5340 yylval.opval->op_private = OPpCONST_ENTERED;
5341 gv_fetchpv(SvPVX(sym),
5343 ? (GV_ADDMULTI | GV_ADDINEVAL)
5346 ((PL_tokenbuf[0] == '$') ? SVt_PV
5347 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5352 /* if it's a sort block and they're naming $a or $b */
5353 if (PL_last_lop_op == OP_SORT &&
5354 PL_tokenbuf[0] == '$' &&
5355 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
5358 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
5359 d < PL_bufend && *d != '\n';
5362 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
5363 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
5369 yylval.opval = newOP(OP_PADANY, 0);
5370 yylval.opval->op_targ = tmp;
5376 Whine if they've said @foo in a doublequoted string,
5377 and @foo isn't a variable we can find in the symbol
5380 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
5381 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
5382 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
5383 && ckWARN(WARN_AMBIGUOUS))
5385 /* Downgraded from fatal to warning 20000522 mjd */
5386 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5387 "Possible unintended interpolation of %s in string",
5392 /* build ops for a bareword */
5393 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
5394 yylval.opval->op_private = OPpCONST_ENTERED;
5395 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
5396 ((PL_tokenbuf[0] == '$') ? SVt_PV
5397 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5403 Perl_keyword(pTHX_ register char *d, I32 len)
5408 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5409 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5410 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5411 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5412 if (strEQ(d,"__END__")) return KEY___END__;
5416 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5421 if (strEQ(d,"and")) return -KEY_and;
5422 if (strEQ(d,"abs")) return -KEY_abs;
5425 if (strEQ(d,"alarm")) return -KEY_alarm;
5426 if (strEQ(d,"atan2")) return -KEY_atan2;
5429 if (strEQ(d,"accept")) return -KEY_accept;
5434 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5437 if (strEQ(d,"bless")) return -KEY_bless;
5438 if (strEQ(d,"bind")) return -KEY_bind;
5439 if (strEQ(d,"binmode")) return -KEY_binmode;
5442 if (strEQ(d,"CORE")) return -KEY_CORE;
5443 if (strEQ(d,"CHECK")) return KEY_CHECK;
5448 if (strEQ(d,"cmp")) return -KEY_cmp;
5449 if (strEQ(d,"chr")) return -KEY_chr;
5450 if (strEQ(d,"cos")) return -KEY_cos;
5453 if (strEQ(d,"chop")) return -KEY_chop;
5456 if (strEQ(d,"close")) return -KEY_close;
5457 if (strEQ(d,"chdir")) return -KEY_chdir;
5458 if (strEQ(d,"chomp")) return -KEY_chomp;
5459 if (strEQ(d,"chmod")) return -KEY_chmod;
5460 if (strEQ(d,"chown")) return -KEY_chown;
5461 if (strEQ(d,"crypt")) return -KEY_crypt;
5464 if (strEQ(d,"chroot")) return -KEY_chroot;
5465 if (strEQ(d,"caller")) return -KEY_caller;
5468 if (strEQ(d,"connect")) return -KEY_connect;
5471 if (strEQ(d,"closedir")) return -KEY_closedir;
5472 if (strEQ(d,"continue")) return -KEY_continue;
5477 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5482 if (strEQ(d,"do")) return KEY_do;
5485 if (strEQ(d,"die")) return -KEY_die;
5488 if (strEQ(d,"dump")) return -KEY_dump;
5491 if (strEQ(d,"delete")) return KEY_delete;
5494 if (strEQ(d,"defined")) return KEY_defined;
5495 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5498 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5503 if (strEQ(d,"END")) return KEY_END;
5508 if (strEQ(d,"eq")) return -KEY_eq;
5511 if (strEQ(d,"eof")) return -KEY_eof;
5512 if (strEQ(d,"err")) return -KEY_err;
5513 if (strEQ(d,"exp")) return -KEY_exp;
5516 if (strEQ(d,"else")) return KEY_else;
5517 if (strEQ(d,"exit")) return -KEY_exit;
5518 if (strEQ(d,"eval")) return KEY_eval;
5519 if (strEQ(d,"exec")) return -KEY_exec;
5520 if (strEQ(d,"each")) return -KEY_each;
5523 if (strEQ(d,"elsif")) return KEY_elsif;
5526 if (strEQ(d,"exists")) return KEY_exists;
5527 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5530 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5531 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5534 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5537 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5538 if (strEQ(d,"endservent")) return -KEY_endservent;
5541 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5548 if (strEQ(d,"for")) return KEY_for;
5551 if (strEQ(d,"fork")) return -KEY_fork;
5554 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5555 if (strEQ(d,"flock")) return -KEY_flock;
5558 if (strEQ(d,"format")) return KEY_format;
5559 if (strEQ(d,"fileno")) return -KEY_fileno;
5562 if (strEQ(d,"foreach")) return KEY_foreach;
5565 if (strEQ(d,"formline")) return -KEY_formline;
5570 if (strnEQ(d,"get",3)) {
5575 if (strEQ(d,"ppid")) return -KEY_getppid;
5576 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5579 if (strEQ(d,"pwent")) return -KEY_getpwent;
5580 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5581 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5584 if (strEQ(d,"peername")) return -KEY_getpeername;
5585 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5586 if (strEQ(d,"priority")) return -KEY_getpriority;
5589 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5592 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5596 else if (*d == 'h') {
5597 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5598 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5599 if (strEQ(d,"hostent")) return -KEY_gethostent;
5601 else if (*d == 'n') {
5602 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5603 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5604 if (strEQ(d,"netent")) return -KEY_getnetent;
5606 else if (*d == 's') {
5607 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5608 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5609 if (strEQ(d,"servent")) return -KEY_getservent;
5610 if (strEQ(d,"sockname")) return -KEY_getsockname;
5611 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5613 else if (*d == 'g') {
5614 if (strEQ(d,"grent")) return -KEY_getgrent;
5615 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5616 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5618 else if (*d == 'l') {
5619 if (strEQ(d,"login")) return -KEY_getlogin;
5621 else if (strEQ(d,"c")) return -KEY_getc;
5626 if (strEQ(d,"gt")) return -KEY_gt;
5627 if (strEQ(d,"ge")) return -KEY_ge;
5630 if (strEQ(d,"grep")) return KEY_grep;
5631 if (strEQ(d,"goto")) return KEY_goto;
5632 if (strEQ(d,"glob")) return KEY_glob;
5635 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5640 if (strEQ(d,"hex")) return -KEY_hex;
5643 if (strEQ(d,"INIT")) return KEY_INIT;
5648 if (strEQ(d,"if")) return KEY_if;
5651 if (strEQ(d,"int")) return -KEY_int;
5654 if (strEQ(d,"index")) return -KEY_index;
5655 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5660 if (strEQ(d,"join")) return -KEY_join;
5664 if (strEQ(d,"keys")) return -KEY_keys;
5665 if (strEQ(d,"kill")) return -KEY_kill;
5671 if (strEQ(d,"lt")) return -KEY_lt;
5672 if (strEQ(d,"le")) return -KEY_le;
5673 if (strEQ(d,"lc")) return -KEY_lc;
5676 if (strEQ(d,"log")) return -KEY_log;
5679 if (strEQ(d,"last")) return KEY_last;
5680 if (strEQ(d,"link")) return -KEY_link;
5681 if (strEQ(d,"lock")) return -KEY_lock;
5684 if (strEQ(d,"local")) return KEY_local;
5685 if (strEQ(d,"lstat")) return -KEY_lstat;
5688 if (strEQ(d,"length")) return -KEY_length;
5689 if (strEQ(d,"listen")) return -KEY_listen;
5692 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5695 if (strEQ(d,"localtime")) return -KEY_localtime;
5701 case 1: return KEY_m;
5703 if (strEQ(d,"my")) return KEY_my;
5706 if (strEQ(d,"map")) return KEY_map;
5709 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5712 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5713 if (strEQ(d,"msgget")) return -KEY_msgget;
5714 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5715 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5720 if (strEQ(d,"next")) return KEY_next;
5721 if (strEQ(d,"ne")) return -KEY_ne;
5722 if (strEQ(d,"not")) return -KEY_not;
5723 if (strEQ(d,"no")) return KEY_no;
5728 if (strEQ(d,"or")) return -KEY_or;
5731 if (strEQ(d,"ord")) return -KEY_ord;
5732 if (strEQ(d,"oct")) return -KEY_oct;
5733 if (strEQ(d,"our")) return KEY_our;
5736 if (strEQ(d,"open")) return -KEY_open;
5739 if (strEQ(d,"opendir")) return -KEY_opendir;
5746 if (strEQ(d,"pop")) return -KEY_pop;
5747 if (strEQ(d,"pos")) return KEY_pos;
5750 if (strEQ(d,"push")) return -KEY_push;
5751 if (strEQ(d,"pack")) return -KEY_pack;
5752 if (strEQ(d,"pipe")) return -KEY_pipe;
5755 if (strEQ(d,"print")) return KEY_print;
5758 if (strEQ(d,"printf")) return KEY_printf;
5761 if (strEQ(d,"package")) return KEY_package;
5764 if (strEQ(d,"prototype")) return KEY_prototype;
5769 if (strEQ(d,"q")) return KEY_q;
5770 if (strEQ(d,"qr")) return KEY_qr;
5771 if (strEQ(d,"qq")) return KEY_qq;
5772 if (strEQ(d,"qw")) return KEY_qw;
5773 if (strEQ(d,"qx")) return KEY_qx;
5775 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5780 if (strEQ(d,"ref")) return -KEY_ref;
5783 if (strEQ(d,"read")) return -KEY_read;
5784 if (strEQ(d,"rand")) return -KEY_rand;
5785 if (strEQ(d,"recv")) return -KEY_recv;
5786 if (strEQ(d,"redo")) return KEY_redo;
5789 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5790 if (strEQ(d,"reset")) return -KEY_reset;
5793 if (strEQ(d,"return")) return KEY_return;
5794 if (strEQ(d,"rename")) return -KEY_rename;
5795 if (strEQ(d,"rindex")) return -KEY_rindex;
5798 if (strEQ(d,"require")) return KEY_require;
5799 if (strEQ(d,"reverse")) return -KEY_reverse;
5800 if (strEQ(d,"readdir")) return -KEY_readdir;
5803 if (strEQ(d,"readlink")) return -KEY_readlink;
5804 if (strEQ(d,"readline")) return -KEY_readline;
5805 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5808 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5814 case 0: return KEY_s;
5816 if (strEQ(d,"scalar")) return KEY_scalar;
5821 if (strEQ(d,"seek")) return -KEY_seek;
5822 if (strEQ(d,"send")) return -KEY_send;
5825 if (strEQ(d,"semop")) return -KEY_semop;
5828 if (strEQ(d,"select")) return -KEY_select;
5829 if (strEQ(d,"semctl")) return -KEY_semctl;
5830 if (strEQ(d,"semget")) return -KEY_semget;
5833 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5834 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5837 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5838 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5841 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5844 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5845 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5846 if (strEQ(d,"setservent")) return -KEY_setservent;
5849 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5850 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5857 if (strEQ(d,"shift")) return -KEY_shift;
5860 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5861 if (strEQ(d,"shmget")) return -KEY_shmget;
5864 if (strEQ(d,"shmread")) return -KEY_shmread;
5867 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5868 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5873 if (strEQ(d,"sin")) return -KEY_sin;
5876 if (strEQ(d,"sleep")) return -KEY_sleep;
5879 if (strEQ(d,"sort")) return KEY_sort;
5880 if (strEQ(d,"socket")) return -KEY_socket;
5881 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5884 if (strEQ(d,"split")) return KEY_split;
5885 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5886 if (strEQ(d,"splice")) return -KEY_splice;
5889 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5892 if (strEQ(d,"srand")) return -KEY_srand;
5895 if (strEQ(d,"stat")) return -KEY_stat;
5896 if (strEQ(d,"study")) return KEY_study;
5899 if (strEQ(d,"substr")) return -KEY_substr;
5900 if (strEQ(d,"sub")) return KEY_sub;
5905 if (strEQ(d,"system")) return -KEY_system;
5908 if (strEQ(d,"symlink")) return -KEY_symlink;
5909 if (strEQ(d,"syscall")) return -KEY_syscall;
5910 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5911 if (strEQ(d,"sysread")) return -KEY_sysread;
5912 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5915 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5924 if (strEQ(d,"tr")) return KEY_tr;
5927 if (strEQ(d,"tie")) return KEY_tie;
5930 if (strEQ(d,"tell")) return -KEY_tell;
5931 if (strEQ(d,"tied")) return KEY_tied;
5932 if (strEQ(d,"time")) return -KEY_time;
5935 if (strEQ(d,"times")) return -KEY_times;
5938 if (strEQ(d,"telldir")) return -KEY_telldir;
5941 if (strEQ(d,"truncate")) return -KEY_truncate;
5948 if (strEQ(d,"uc")) return -KEY_uc;
5951 if (strEQ(d,"use")) return KEY_use;
5954 if (strEQ(d,"undef")) return KEY_undef;
5955 if (strEQ(d,"until")) return KEY_until;
5956 if (strEQ(d,"untie")) return KEY_untie;
5957 if (strEQ(d,"utime")) return -KEY_utime;
5958 if (strEQ(d,"umask")) return -KEY_umask;
5961 if (strEQ(d,"unless")) return KEY_unless;
5962 if (strEQ(d,"unpack")) return -KEY_unpack;
5963 if (strEQ(d,"unlink")) return -KEY_unlink;
5966 if (strEQ(d,"unshift")) return -KEY_unshift;
5967 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5972 if (strEQ(d,"values")) return -KEY_values;
5973 if (strEQ(d,"vec")) return -KEY_vec;
5978 if (strEQ(d,"warn")) return -KEY_warn;
5979 if (strEQ(d,"wait")) return -KEY_wait;
5982 if (strEQ(d,"while")) return KEY_while;
5983 if (strEQ(d,"write")) return -KEY_write;
5986 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5989 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5994 if (len == 1) return -KEY_x;
5995 if (strEQ(d,"xor")) return -KEY_xor;
5998 if (len == 1) return KEY_y;
6007 S_checkcomma(pTHX_ register char *s, char *name, char *what)
6011 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
6012 if (ckWARN(WARN_SYNTAX)) {
6014 for (w = s+2; *w && level; w++) {
6021 for (; *w && isSPACE(*w); w++) ;
6022 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
6023 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
6024 "%s (...) interpreted as function",name);
6027 while (s < PL_bufend && isSPACE(*s))
6031 while (s < PL_bufend && isSPACE(*s))
6033 if (isIDFIRST_lazy_if(s,UTF)) {
6035 while (isALNUM_lazy_if(s,UTF))
6037 while (s < PL_bufend && isSPACE(*s))
6042 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
6046 Perl_croak(aTHX_ "No comma allowed after %s", what);
6051 /* Either returns sv, or mortalizes sv and returns a new SV*.
6052 Best used as sv=new_constant(..., sv, ...).
6053 If s, pv are NULL, calls subroutine with one argument,
6054 and type is used with error messages only. */
6057 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
6061 HV *table = GvHV(PL_hintgv); /* ^H */
6065 const char *why1, *why2, *why3;
6067 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
6070 why2 = strEQ(key,"charnames")
6071 ? "(possibly a missing \"use charnames ...\")"
6073 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
6074 (type ? type: "undef"), why2);
6076 /* This is convoluted and evil ("goto considered harmful")
6077 * but I do not understand the intricacies of all the different
6078 * failure modes of %^H in here. The goal here is to make
6079 * the most probable error message user-friendly. --jhi */
6084 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
6085 (type ? type: "undef"), why1, why2, why3);
6087 yyerror(SvPVX(msg));
6091 cvp = hv_fetch(table, key, strlen(key), FALSE);
6092 if (!cvp || !SvOK(*cvp)) {
6095 why3 = "} is not defined";
6098 sv_2mortal(sv); /* Parent created it permanently */
6101 pv = sv_2mortal(newSVpvn(s, len));
6103 typesv = sv_2mortal(newSVpv(type, 0));
6105 typesv = &PL_sv_undef;
6107 PUSHSTACKi(PERLSI_OVERLOAD);
6119 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
6123 /* Check the eval first */
6124 if (!PL_in_eval && SvTRUE(ERRSV)) {
6126 sv_catpv(ERRSV, "Propagated");
6127 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
6129 res = SvREFCNT_inc(sv);
6133 (void)SvREFCNT_inc(res);
6142 why1 = "Call to &{$^H{";
6144 why3 = "}} did not return a defined value";
6153 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
6155 register char *d = dest;
6156 register char *e = d + destlen - 3; /* two-character token, ending NUL */
6159 Perl_croak(aTHX_ ident_too_long);
6160 if (isALNUM(*s)) /* UTF handled below */
6162 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
6167 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
6171 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6172 char *t = s + UTF8SKIP(s);
6173 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6175 if (d + (t - s) > e)
6176 Perl_croak(aTHX_ ident_too_long);
6177 Copy(s, d, t - s, char);
6190 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
6200 e = d + destlen - 3; /* two-character token, ending NUL */
6202 while (isDIGIT(*s)) {
6204 Perl_croak(aTHX_ ident_too_long);
6211 Perl_croak(aTHX_ ident_too_long);
6212 if (isALNUM(*s)) /* UTF handled below */
6214 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
6219 else if (*s == ':' && s[1] == ':') {
6223 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6224 char *t = s + UTF8SKIP(s);
6225 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6227 if (d + (t - s) > e)
6228 Perl_croak(aTHX_ ident_too_long);
6229 Copy(s, d, t - s, char);
6240 if (PL_lex_state != LEX_NORMAL)
6241 PL_lex_state = LEX_INTERPENDMAYBE;
6244 if (*s == '$' && s[1] &&
6245 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
6258 if (*d == '^' && *s && isCONTROLVAR(*s)) {
6263 if (isSPACE(s[-1])) {
6266 if (!SPACE_OR_TAB(ch)) {
6272 if (isIDFIRST_lazy_if(d,UTF)) {
6276 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
6278 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
6281 Copy(s, d, e - s, char);
6286 while ((isALNUM(*s) || *s == ':') && d < e)
6289 Perl_croak(aTHX_ ident_too_long);
6292 while (s < send && SPACE_OR_TAB(*s)) s++;
6293 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
6294 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
6295 const char *brack = *s == '[' ? "[...]" : "{...}";
6296 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
6297 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
6298 funny, dest, brack, funny, dest, brack);
6301 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
6305 /* Handle extended ${^Foo} variables
6306 * 1999-02-27 mjd-perl-patch@plover.com */
6307 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
6311 while (isALNUM(*s) && d < e) {
6315 Perl_croak(aTHX_ ident_too_long);
6320 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
6321 PL_lex_state = LEX_INTERPEND;
6326 if (PL_lex_state == LEX_NORMAL) {
6327 if (ckWARN(WARN_AMBIGUOUS) &&
6328 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
6330 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
6331 "Ambiguous use of %c{%s} resolved to %c%s",
6332 funny, dest, funny, dest);
6337 s = bracket; /* let the parser handle it */
6341 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
6342 PL_lex_state = LEX_INTERPEND;
6347 Perl_pmflag(pTHX_ U32* pmfl, int ch)
6352 *pmfl |= PMf_GLOBAL;
6354 *pmfl |= PMf_CONTINUE;
6358 *pmfl |= PMf_MULTILINE;
6360 *pmfl |= PMf_SINGLELINE;
6362 *pmfl |= PMf_EXTENDED;
6366 S_scan_pat(pTHX_ char *start, I32 type)
6371 s = scan_str(start,FALSE,FALSE);
6373 Perl_croak(aTHX_ "Search pattern not terminated");
6375 pm = (PMOP*)newPMOP(type, 0);
6376 if (PL_multi_open == '?')
6377 pm->op_pmflags |= PMf_ONCE;
6379 while (*s && strchr("iomsx", *s))
6380 pmflag(&pm->op_pmflags,*s++);
6383 while (*s && strchr("iogcmsx", *s))
6384 pmflag(&pm->op_pmflags,*s++);
6386 /* issue a warning if /c is specified,but /g is not */
6387 if (ckWARN(WARN_REGEXP) &&
6388 (pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL))
6390 Perl_warner(aTHX_ packWARN(WARN_REGEXP), c_without_g);
6393 pm->op_pmpermflags = pm->op_pmflags;
6395 PL_lex_op = (OP*)pm;
6396 yylval.ival = OP_MATCH;
6401 S_scan_subst(pTHX_ char *start)
6408 yylval.ival = OP_NULL;
6410 s = scan_str(start,FALSE,FALSE);
6413 Perl_croak(aTHX_ "Substitution pattern not terminated");
6415 if (s[-1] == PL_multi_open)
6418 first_start = PL_multi_start;
6419 s = scan_str(s,FALSE,FALSE);
6422 SvREFCNT_dec(PL_lex_stuff);
6423 PL_lex_stuff = Nullsv;
6425 Perl_croak(aTHX_ "Substitution replacement not terminated");
6427 PL_multi_start = first_start; /* so whole substitution is taken together */
6429 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6435 else if (strchr("iogcmsx", *s))
6436 pmflag(&pm->op_pmflags,*s++);
6441 /* /c is not meaningful with s/// */
6442 if (ckWARN(WARN_REGEXP) && (pm->op_pmflags & PMf_CONTINUE))
6444 Perl_warner(aTHX_ packWARN(WARN_REGEXP), c_in_subst);
6449 PL_sublex_info.super_bufptr = s;
6450 PL_sublex_info.super_bufend = PL_bufend;
6452 pm->op_pmflags |= PMf_EVAL;
6453 repl = newSVpvn("",0);
6455 sv_catpv(repl, es ? "eval " : "do ");
6456 sv_catpvn(repl, "{ ", 2);
6457 sv_catsv(repl, PL_lex_repl);
6458 sv_catpvn(repl, " };", 2);
6460 SvREFCNT_dec(PL_lex_repl);
6464 pm->op_pmpermflags = pm->op_pmflags;
6465 PL_lex_op = (OP*)pm;
6466 yylval.ival = OP_SUBST;
6471 S_scan_trans(pTHX_ char *start)
6480 yylval.ival = OP_NULL;
6482 s = scan_str(start,FALSE,FALSE);
6484 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6485 if (s[-1] == PL_multi_open)
6488 s = scan_str(s,FALSE,FALSE);
6491 SvREFCNT_dec(PL_lex_stuff);
6492 PL_lex_stuff = Nullsv;
6494 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6497 complement = del = squash = 0;
6498 while (strchr("cds", *s)) {
6500 complement = OPpTRANS_COMPLEMENT;
6502 del = OPpTRANS_DELETE;
6504 squash = OPpTRANS_SQUASH;
6508 New(803, tbl, complement&&!del?258:256, short);
6509 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6510 o->op_private = del|squash|complement|
6511 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
6512 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
6515 yylval.ival = OP_TRANS;
6520 S_scan_heredoc(pTHX_ register char *s)
6523 I32 op_type = OP_SCALAR;
6530 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6534 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6537 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6538 if (*peek && strchr("`'\"",*peek)) {
6541 s = delimcpy(d, e, s, PL_bufend, term, &len);
6551 if (!isALNUM_lazy_if(s,UTF))
6552 deprecate_old("bare << to mean <<\"\"");
6553 for (; isALNUM_lazy_if(s,UTF); s++) {
6558 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6559 Perl_croak(aTHX_ "Delimiter for here document is too long");
6562 len = d - PL_tokenbuf;
6563 #ifndef PERL_STRICT_CR
6564 d = strchr(s, '\r');
6568 while (s < PL_bufend) {
6574 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6583 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6588 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6589 herewas = newSVpvn(s,PL_bufend-s);
6591 s--, herewas = newSVpvn(s,d-s);
6592 s += SvCUR(herewas);
6594 tmpstr = NEWSV(87,79);
6595 sv_upgrade(tmpstr, SVt_PVIV);
6600 else if (term == '`') {
6601 op_type = OP_BACKTICK;
6602 SvIVX(tmpstr) = '\\';
6606 PL_multi_start = CopLINE(PL_curcop);
6607 PL_multi_open = PL_multi_close = '<';
6608 term = *PL_tokenbuf;
6609 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6610 char *bufptr = PL_sublex_info.super_bufptr;
6611 char *bufend = PL_sublex_info.super_bufend;
6612 char *olds = s - SvCUR(herewas);
6613 s = strchr(bufptr, '\n');
6617 while (s < bufend &&
6618 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6620 CopLINE_inc(PL_curcop);
6623 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6624 missingterm(PL_tokenbuf);
6626 sv_setpvn(herewas,bufptr,d-bufptr+1);
6627 sv_setpvn(tmpstr,d+1,s-d);
6629 sv_catpvn(herewas,s,bufend-s);
6630 (void)strcpy(bufptr,SvPVX(herewas));
6637 while (s < PL_bufend &&
6638 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6640 CopLINE_inc(PL_curcop);
6642 if (s >= PL_bufend) {
6643 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6644 missingterm(PL_tokenbuf);
6646 sv_setpvn(tmpstr,d+1,s-d);
6648 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6650 sv_catpvn(herewas,s,PL_bufend-s);
6651 sv_setsv(PL_linestr,herewas);
6652 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6653 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6654 PL_last_lop = PL_last_uni = Nullch;
6657 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6658 while (s >= PL_bufend) { /* multiple line string? */
6660 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6661 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6662 missingterm(PL_tokenbuf);
6664 CopLINE_inc(PL_curcop);
6665 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6666 PL_last_lop = PL_last_uni = Nullch;
6667 #ifndef PERL_STRICT_CR
6668 if (PL_bufend - PL_linestart >= 2) {
6669 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6670 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6672 PL_bufend[-2] = '\n';
6674 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6676 else if (PL_bufend[-1] == '\r')
6677 PL_bufend[-1] = '\n';
6679 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6680 PL_bufend[-1] = '\n';
6682 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6683 SV *sv = NEWSV(88,0);
6685 sv_upgrade(sv, SVt_PVMG);
6686 sv_setsv(sv,PL_linestr);
6689 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6691 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6694 sv_catsv(PL_linestr,herewas);
6695 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6699 sv_catsv(tmpstr,PL_linestr);
6704 PL_multi_end = CopLINE(PL_curcop);
6705 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6706 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6707 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6709 SvREFCNT_dec(herewas);
6711 if (UTF && is_utf8_string((U8*)SvPVX(tmpstr), SvCUR(tmpstr)))
6713 else if (PL_encoding)
6714 sv_recode_to_utf8(tmpstr, PL_encoding);
6716 PL_lex_stuff = tmpstr;
6717 yylval.ival = op_type;
6722 takes: current position in input buffer
6723 returns: new position in input buffer
6724 side-effects: yylval and lex_op are set.
6729 <FH> read from filehandle
6730 <pkg::FH> read from package qualified filehandle
6731 <pkg'FH> read from package qualified filehandle
6732 <$fh> read from filehandle in $fh
6738 S_scan_inputsymbol(pTHX_ char *start)
6740 register char *s = start; /* current position in buffer */
6746 d = PL_tokenbuf; /* start of temp holding space */
6747 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6748 end = strchr(s, '\n');
6751 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6753 /* die if we didn't have space for the contents of the <>,
6754 or if it didn't end, or if we see a newline
6757 if (len >= sizeof PL_tokenbuf)
6758 Perl_croak(aTHX_ "Excessively long <> operator");
6760 Perl_croak(aTHX_ "Unterminated <> operator");
6765 Remember, only scalar variables are interpreted as filehandles by
6766 this code. Anything more complex (e.g., <$fh{$num}>) will be
6767 treated as a glob() call.
6768 This code makes use of the fact that except for the $ at the front,
6769 a scalar variable and a filehandle look the same.
6771 if (*d == '$' && d[1]) d++;
6773 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6774 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6777 /* If we've tried to read what we allow filehandles to look like, and
6778 there's still text left, then it must be a glob() and not a getline.
6779 Use scan_str to pull out the stuff between the <> and treat it
6780 as nothing more than a string.
6783 if (d - PL_tokenbuf != len) {
6784 yylval.ival = OP_GLOB;
6786 s = scan_str(start,FALSE,FALSE);
6788 Perl_croak(aTHX_ "Glob not terminated");
6792 bool readline_overriden = FALSE;
6793 GV *gv_readline = Nullgv;
6795 /* we're in a filehandle read situation */
6798 /* turn <> into <ARGV> */
6800 (void)strcpy(d,"ARGV");
6802 /* Check whether readline() is overriden */
6803 if (((gv_readline = gv_fetchpv("readline", FALSE, SVt_PVCV))
6804 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
6806 ((gvp = (GV**)hv_fetch(PL_globalstash, "readline", 8, FALSE))
6807 && (gv_readline = *gvp) != (GV*)&PL_sv_undef
6808 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
6809 readline_overriden = TRUE;
6811 /* if <$fh>, create the ops to turn the variable into a
6817 /* try to find it in the pad for this block, otherwise find
6818 add symbol table ops
6820 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6821 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
6822 SV *sym = sv_2mortal(
6823 newSVpv(HvNAME(PAD_COMPNAME_OURSTASH(tmp)),0));
6824 sv_catpvn(sym, "::", 2);
6830 OP *o = newOP(OP_PADSV, 0);
6832 PL_lex_op = readline_overriden
6833 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6834 append_elem(OP_LIST, o,
6835 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
6836 : (OP*)newUNOP(OP_READLINE, 0, o);
6845 ? (GV_ADDMULTI | GV_ADDINEVAL)
6848 PL_lex_op = readline_overriden
6849 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6850 append_elem(OP_LIST,
6851 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
6852 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
6853 : (OP*)newUNOP(OP_READLINE, 0,
6854 newUNOP(OP_RV2SV, 0,
6855 newGVOP(OP_GV, 0, gv)));
6857 if (!readline_overriden)
6858 PL_lex_op->op_flags |= OPf_SPECIAL;
6859 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6860 yylval.ival = OP_NULL;
6863 /* If it's none of the above, it must be a literal filehandle
6864 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6866 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6867 PL_lex_op = readline_overriden
6868 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6869 append_elem(OP_LIST,
6870 newGVOP(OP_GV, 0, gv),
6871 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
6872 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6873 yylval.ival = OP_NULL;
6882 takes: start position in buffer
6883 keep_quoted preserve \ on the embedded delimiter(s)
6884 keep_delims preserve the delimiters around the string
6885 returns: position to continue reading from buffer
6886 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6887 updates the read buffer.
6889 This subroutine pulls a string out of the input. It is called for:
6890 q single quotes q(literal text)
6891 ' single quotes 'literal text'
6892 qq double quotes qq(interpolate $here please)
6893 " double quotes "interpolate $here please"
6894 qx backticks qx(/bin/ls -l)
6895 ` backticks `/bin/ls -l`
6896 qw quote words @EXPORT_OK = qw( func() $spam )
6897 m// regexp match m/this/
6898 s/// regexp substitute s/this/that/
6899 tr/// string transliterate tr/this/that/
6900 y/// string transliterate y/this/that/
6901 ($*@) sub prototypes sub foo ($)
6902 (stuff) sub attr parameters sub foo : attr(stuff)
6903 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6905 In most of these cases (all but <>, patterns and transliterate)
6906 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6907 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6908 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6911 It skips whitespace before the string starts, and treats the first
6912 character as the delimiter. If the delimiter is one of ([{< then
6913 the corresponding "close" character )]}> is used as the closing
6914 delimiter. It allows quoting of delimiters, and if the string has
6915 balanced delimiters ([{<>}]) it allows nesting.
6917 On success, the SV with the resulting string is put into lex_stuff or,
6918 if that is already non-NULL, into lex_repl. The second case occurs only
6919 when parsing the RHS of the special constructs s/// and tr/// (y///).
6920 For convenience, the terminating delimiter character is stuffed into
6925 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6927 SV *sv; /* scalar value: string */
6928 char *tmps; /* temp string, used for delimiter matching */
6929 register char *s = start; /* current position in the buffer */
6930 register char term; /* terminating character */
6931 register char *to; /* current position in the sv's data */
6932 I32 brackets = 1; /* bracket nesting level */
6933 bool has_utf8 = FALSE; /* is there any utf8 content? */
6934 I32 termcode; /* terminating char. code */
6935 U8 termstr[UTF8_MAXLEN]; /* terminating string */
6936 STRLEN termlen; /* length of terminating string */
6937 char *last = NULL; /* last position for nesting bracket */
6939 /* skip space before the delimiter */
6943 /* mark where we are, in case we need to report errors */
6946 /* after skipping whitespace, the next character is the terminator */
6949 termcode = termstr[0] = term;
6953 termcode = utf8_to_uvchr((U8*)s, &termlen);
6954 Copy(s, termstr, termlen, U8);
6955 if (!UTF8_IS_INVARIANT(term))
6959 /* mark where we are */
6960 PL_multi_start = CopLINE(PL_curcop);
6961 PL_multi_open = term;
6963 /* find corresponding closing delimiter */
6964 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6965 termcode = termstr[0] = term = tmps[5];
6967 PL_multi_close = term;
6969 /* create a new SV to hold the contents. 87 is leak category, I'm
6970 assuming. 79 is the SV's initial length. What a random number. */
6972 sv_upgrade(sv, SVt_PVIV);
6973 SvIVX(sv) = termcode;
6974 (void)SvPOK_only(sv); /* validate pointer */
6976 /* move past delimiter and try to read a complete string */
6978 sv_catpvn(sv, s, termlen);
6981 if (PL_encoding && !UTF) {
6985 int offset = s - SvPVX(PL_linestr);
6986 bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
6987 &offset, (char*)termstr, termlen);
6988 char *ns = SvPVX(PL_linestr) + offset;
6989 char *svlast = SvEND(sv) - 1;
6991 for (; s < ns; s++) {
6992 if (*s == '\n' && !PL_rsfp)
6993 CopLINE_inc(PL_curcop);
6996 goto read_more_line;
6998 /* handle quoted delimiters */
6999 if (*(svlast-1) == '\\') {
7001 for (t = svlast-2; t >= SvPVX(sv) && *t == '\\';)
7003 if ((svlast-1 - t) % 2) {
7007 SvCUR_set(sv, SvCUR(sv) - 1);
7012 if (PL_multi_open == PL_multi_close) {
7019 for (w = t = last; t < svlast; w++, t++) {
7020 /* At here, all closes are "was quoted" one,
7021 so we don't check PL_multi_close. */
7023 if (!keep_quoted && *(t+1) == PL_multi_open)
7028 else if (*t == PL_multi_open)
7036 SvCUR_set(sv, w - SvPVX(sv));
7039 if (--brackets <= 0)
7045 SvCUR_set(sv, SvCUR(sv) - 1);
7051 /* extend sv if need be */
7052 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
7053 /* set 'to' to the next character in the sv's string */
7054 to = SvPVX(sv)+SvCUR(sv);
7056 /* if open delimiter is the close delimiter read unbridle */
7057 if (PL_multi_open == PL_multi_close) {
7058 for (; s < PL_bufend; s++,to++) {
7059 /* embedded newlines increment the current line number */
7060 if (*s == '\n' && !PL_rsfp)
7061 CopLINE_inc(PL_curcop);
7062 /* handle quoted delimiters */
7063 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
7064 if (!keep_quoted && s[1] == term)
7066 /* any other quotes are simply copied straight through */
7070 /* terminate when run out of buffer (the for() condition), or
7071 have found the terminator */
7072 else if (*s == term) {
7075 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
7078 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
7084 /* if the terminator isn't the same as the start character (e.g.,
7085 matched brackets), we have to allow more in the quoting, and
7086 be prepared for nested brackets.
7089 /* read until we run out of string, or we find the terminator */
7090 for (; s < PL_bufend; s++,to++) {
7091 /* embedded newlines increment the line count */
7092 if (*s == '\n' && !PL_rsfp)
7093 CopLINE_inc(PL_curcop);
7094 /* backslashes can escape the open or closing characters */
7095 if (*s == '\\' && s+1 < PL_bufend) {
7097 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
7102 /* allow nested opens and closes */
7103 else if (*s == PL_multi_close && --brackets <= 0)
7105 else if (*s == PL_multi_open)
7107 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
7112 /* terminate the copied string and update the sv's end-of-string */
7114 SvCUR_set(sv, to - SvPVX(sv));
7117 * this next chunk reads more into the buffer if we're not done yet
7121 break; /* handle case where we are done yet :-) */
7123 #ifndef PERL_STRICT_CR
7124 if (to - SvPVX(sv) >= 2) {
7125 if ((to[-2] == '\r' && to[-1] == '\n') ||
7126 (to[-2] == '\n' && to[-1] == '\r'))
7130 SvCUR_set(sv, to - SvPVX(sv));
7132 else if (to[-1] == '\r')
7135 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
7140 /* if we're out of file, or a read fails, bail and reset the current
7141 line marker so we can report where the unterminated string began
7144 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
7146 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
7149 /* we read a line, so increment our line counter */
7150 CopLINE_inc(PL_curcop);
7152 /* update debugger info */
7153 if (PERLDB_LINE && PL_curstash != PL_debstash) {
7154 SV *sv = NEWSV(88,0);
7156 sv_upgrade(sv, SVt_PVMG);
7157 sv_setsv(sv,PL_linestr);
7160 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
7163 /* having changed the buffer, we must update PL_bufend */
7164 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7165 PL_last_lop = PL_last_uni = Nullch;
7168 /* at this point, we have successfully read the delimited string */
7170 if (!PL_encoding || UTF) {
7172 sv_catpvn(sv, s, termlen);
7175 if (has_utf8 || PL_encoding)
7178 PL_multi_end = CopLINE(PL_curcop);
7180 /* if we allocated too much space, give some back */
7181 if (SvCUR(sv) + 5 < SvLEN(sv)) {
7182 SvLEN_set(sv, SvCUR(sv) + 1);
7183 Renew(SvPVX(sv), SvLEN(sv), char);
7186 /* decide whether this is the first or second quoted string we've read
7199 takes: pointer to position in buffer
7200 returns: pointer to new position in buffer
7201 side-effects: builds ops for the constant in yylval.op
7203 Read a number in any of the formats that Perl accepts:
7205 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
7206 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
7209 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
7211 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
7214 If it reads a number without a decimal point or an exponent, it will
7215 try converting the number to an integer and see if it can do so
7216 without loss of precision.
7220 Perl_scan_num(pTHX_ char *start, YYSTYPE* lvalp)
7222 register char *s = start; /* current position in buffer */
7223 register char *d; /* destination in temp buffer */
7224 register char *e; /* end of temp buffer */
7225 NV nv; /* number read, as a double */
7226 SV *sv = Nullsv; /* place to put the converted number */
7227 bool floatit; /* boolean: int or float? */
7228 char *lastub = 0; /* position of last underbar */
7229 static char number_too_long[] = "Number too long";
7231 /* We use the first character to decide what type of number this is */
7235 Perl_croak(aTHX_ "panic: scan_num");
7237 /* if it starts with a 0, it could be an octal number, a decimal in
7238 0.13 disguise, or a hexadecimal number, or a binary number. */
7242 u holds the "number so far"
7243 shift the power of 2 of the base
7244 (hex == 4, octal == 3, binary == 1)
7245 overflowed was the number more than we can hold?
7247 Shift is used when we add a digit. It also serves as an "are
7248 we in octal/hex/binary?" indicator to disallow hex characters
7254 bool overflowed = FALSE;
7255 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
7256 static char* bases[5] = { "", "binary", "", "octal",
7258 static char* Bases[5] = { "", "Binary", "", "Octal",
7260 static char *maxima[5] = { "",
7261 "0b11111111111111111111111111111111",
7265 char *base, *Base, *max;
7271 } else if (s[1] == 'b') {
7275 /* check for a decimal in disguise */
7276 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
7278 /* so it must be octal */
7285 if (ckWARN(WARN_SYNTAX))
7286 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7287 "Misplaced _ in number");
7291 base = bases[shift];
7292 Base = Bases[shift];
7293 max = maxima[shift];
7295 /* read the rest of the number */
7297 /* x is used in the overflow test,
7298 b is the digit we're adding on. */
7303 /* if we don't mention it, we're done */
7307 /* _ are ignored -- but warned about if consecutive */
7309 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7310 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7311 "Misplaced _ in number");
7315 /* 8 and 9 are not octal */
7318 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
7322 case '2': case '3': case '4':
7323 case '5': case '6': case '7':
7325 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
7329 b = *s++ & 15; /* ASCII digit -> value of digit */
7333 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
7334 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
7335 /* make sure they said 0x */
7340 /* Prepare to put the digit we have onto the end
7341 of the number so far. We check for overflows.
7346 x = u << shift; /* make room for the digit */
7348 if ((x >> shift) != u
7349 && !(PL_hints & HINT_NEW_BINARY)) {
7352 if (ckWARN_d(WARN_OVERFLOW))
7353 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
7354 "Integer overflow in %s number",
7357 u = x | b; /* add the digit to the end */
7360 n *= nvshift[shift];
7361 /* If an NV has not enough bits in its
7362 * mantissa to represent an UV this summing of
7363 * small low-order numbers is a waste of time
7364 * (because the NV cannot preserve the
7365 * low-order bits anyway): we could just
7366 * remember when did we overflow and in the
7367 * end just multiply n by the right
7375 /* if we get here, we had success: make a scalar value from
7380 /* final misplaced underbar check */
7382 if (ckWARN(WARN_SYNTAX))
7383 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
7388 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
7389 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
7390 "%s number > %s non-portable",
7396 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
7397 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
7398 "%s number > %s non-portable",
7403 if (PL_hints & HINT_NEW_BINARY)
7404 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
7409 handle decimal numbers.
7410 we're also sent here when we read a 0 as the first digit
7412 case '1': case '2': case '3': case '4': case '5':
7413 case '6': case '7': case '8': case '9': case '.':
7416 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
7419 /* read next group of digits and _ and copy into d */
7420 while (isDIGIT(*s) || *s == '_') {
7421 /* skip underscores, checking for misplaced ones
7425 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7426 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7427 "Misplaced _ in number");
7431 /* check for end of fixed-length buffer */
7433 Perl_croak(aTHX_ number_too_long);
7434 /* if we're ok, copy the character */
7439 /* final misplaced underbar check */
7440 if (lastub && s == lastub + 1) {
7441 if (ckWARN(WARN_SYNTAX))
7442 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
7445 /* read a decimal portion if there is one. avoid
7446 3..5 being interpreted as the number 3. followed
7449 if (*s == '.' && s[1] != '.') {
7454 if (ckWARN(WARN_SYNTAX))
7455 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7456 "Misplaced _ in number");
7460 /* copy, ignoring underbars, until we run out of digits.
7462 for (; isDIGIT(*s) || *s == '_'; s++) {
7463 /* fixed length buffer check */
7465 Perl_croak(aTHX_ number_too_long);
7467 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7468 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7469 "Misplaced _ in number");
7475 /* fractional part ending in underbar? */
7477 if (ckWARN(WARN_SYNTAX))
7478 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7479 "Misplaced _ in number");
7481 if (*s == '.' && isDIGIT(s[1])) {
7482 /* oops, it's really a v-string, but without the "v" */
7488 /* read exponent part, if present */
7489 if (*s && strchr("eE",*s) && strchr("+-0123456789_", s[1])) {
7493 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
7494 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
7496 /* stray preinitial _ */
7498 if (ckWARN(WARN_SYNTAX))
7499 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7500 "Misplaced _ in number");
7504 /* allow positive or negative exponent */
7505 if (*s == '+' || *s == '-')
7508 /* stray initial _ */
7510 if (ckWARN(WARN_SYNTAX))
7511 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7512 "Misplaced _ in number");
7516 /* read digits of exponent */
7517 while (isDIGIT(*s) || *s == '_') {
7520 Perl_croak(aTHX_ number_too_long);
7524 if (ckWARN(WARN_SYNTAX) &&
7525 ((lastub && s == lastub + 1) ||
7526 (!isDIGIT(s[1]) && s[1] != '_')))
7527 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7528 "Misplaced _ in number");
7535 /* make an sv from the string */
7539 We try to do an integer conversion first if no characters
7540 indicating "float" have been found.
7545 int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
7547 if (flags == IS_NUMBER_IN_UV) {
7549 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7552 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
7553 if (uv <= (UV) IV_MIN)
7554 sv_setiv(sv, -(IV)uv);
7561 /* terminate the string */
7563 nv = Atof(PL_tokenbuf);
7567 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7568 (PL_hints & HINT_NEW_INTEGER) )
7569 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7570 (floatit ? "float" : "integer"),
7574 /* if it starts with a v, it could be a v-string */
7577 sv = NEWSV(92,5); /* preallocate storage space */
7578 s = scan_vstring(s,sv);
7582 /* make the op for the constant and return */
7585 lvalp->opval = newSVOP(OP_CONST, 0, sv);
7587 lvalp->opval = Nullop;
7593 S_scan_formline(pTHX_ register char *s)
7597 SV *stuff = newSVpvn("",0);
7598 bool needargs = FALSE;
7601 if (*s == '.' || *s == /*{*/'}') {
7603 #ifdef PERL_STRICT_CR
7604 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7606 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7608 if (*t == '\n' || t == PL_bufend)
7611 if (PL_in_eval && !PL_rsfp) {
7612 eol = strchr(s,'\n');
7617 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7619 for (t = s; t < eol; t++) {
7620 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7622 goto enough; /* ~~ must be first line in formline */
7624 if (*t == '@' || *t == '^')
7628 sv_catpvn(stuff, s, eol-s);
7629 #ifndef PERL_STRICT_CR
7630 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7631 char *end = SvPVX(stuff) + SvCUR(stuff);
7643 s = filter_gets(PL_linestr, PL_rsfp, 0);
7644 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7645 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7646 PL_last_lop = PL_last_uni = Nullch;
7649 yyerror("Format not terminated");
7659 PL_lex_state = LEX_NORMAL;
7660 PL_nextval[PL_nexttoke].ival = 0;
7664 PL_lex_state = LEX_FORMLINE;
7666 if (UTF && is_utf8_string((U8*)SvPVX(stuff), SvCUR(stuff)))
7668 else if (PL_encoding)
7669 sv_recode_to_utf8(stuff, PL_encoding);
7671 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7673 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7677 SvREFCNT_dec(stuff);
7678 PL_lex_formbrack = 0;
7689 PL_cshlen = strlen(PL_cshname);
7694 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7696 I32 oldsavestack_ix = PL_savestack_ix;
7697 CV* outsidecv = PL_compcv;
7700 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7702 SAVEI32(PL_subline);
7703 save_item(PL_subname);
7704 SAVESPTR(PL_compcv);
7706 PL_compcv = (CV*)NEWSV(1104,0);
7707 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7708 CvFLAGS(PL_compcv) |= flags;
7710 PL_subline = CopLINE(PL_curcop);
7711 CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
7712 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7713 CvOUTSIDE_SEQ(PL_compcv) = PL_cop_seqmax;
7715 return oldsavestack_ix;
7719 #pragma segment Perl_yylex
7722 Perl_yywarn(pTHX_ char *s)
7724 PL_in_eval |= EVAL_WARNONLY;
7726 PL_in_eval &= ~EVAL_WARNONLY;
7731 Perl_yyerror(pTHX_ char *s)
7734 char *context = NULL;
7738 if (!yychar || (yychar == ';' && !PL_rsfp))
7740 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7741 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7744 The code below is removed for NetWare because it abends/crashes on NetWare
7745 when the script has error such as not having the closing quotes like:
7747 Checking of white spaces is anyway done in NetWare code.
7750 while (isSPACE(*PL_oldoldbufptr))
7753 context = PL_oldoldbufptr;
7754 contlen = PL_bufptr - PL_oldoldbufptr;
7756 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7757 PL_oldbufptr != PL_bufptr) {
7760 The code below is removed for NetWare because it abends/crashes on NetWare
7761 when the script has error such as not having the closing quotes like:
7763 Checking of white spaces is anyway done in NetWare code.
7766 while (isSPACE(*PL_oldbufptr))
7769 context = PL_oldbufptr;
7770 contlen = PL_bufptr - PL_oldbufptr;
7772 else if (yychar > 255)
7773 where = "next token ???";
7774 #ifdef USE_PURE_BISON
7775 /* GNU Bison sets the value -2 */
7776 else if (yychar == -2) {
7778 else if ((yychar & 127) == 127) {
7780 if (PL_lex_state == LEX_NORMAL ||
7781 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7782 where = "at end of line";
7783 else if (PL_lex_inpat)
7784 where = "within pattern";
7786 where = "within string";
7789 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7791 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7792 else if (isPRINT_LC(yychar))
7793 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7795 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7796 where = SvPVX(where_sv);
7798 msg = sv_2mortal(newSVpv(s, 0));
7799 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7800 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7802 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7804 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7805 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7806 Perl_sv_catpvf(aTHX_ msg,
7807 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7808 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7811 if (PL_in_eval & EVAL_WARNONLY)
7812 Perl_warn(aTHX_ "%"SVf, msg);
7815 if (PL_error_count >= 10) {
7816 if (PL_in_eval && SvCUR(ERRSV))
7817 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
7818 ERRSV, OutCopFILE(PL_curcop));
7820 Perl_croak(aTHX_ "%s has too many errors.\n",
7821 OutCopFILE(PL_curcop));
7824 PL_in_my_stash = Nullhv;
7828 #pragma segment Main
7832 S_swallow_bom(pTHX_ U8 *s)
7835 slen = SvCUR(PL_linestr);
7839 /* UTF-16 little-endian */
7840 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
7841 Perl_croak(aTHX_ "Unsupported script encoding");
7842 #ifndef PERL_NO_UTF16_FILTER
7843 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-LE script encoding\n"));
7845 if (PL_bufend > (char*)s) {
7849 filter_add(utf16rev_textfilter, NULL);
7850 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7851 PL_bufend = (char*)utf16_to_utf8_reversed(s, news,
7852 PL_bufend - (char*)s - 1,
7854 Copy(news, s, newlen, U8);
7855 SvCUR_set(PL_linestr, newlen);
7856 PL_bufend = SvPVX(PL_linestr) + newlen;
7857 news[newlen++] = '\0';
7861 Perl_croak(aTHX_ "Unsupported script encoding");
7866 if (s[1] == 0xFF) { /* UTF-16 big-endian */
7867 #ifndef PERL_NO_UTF16_FILTER
7868 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding\n"));
7870 if (PL_bufend > (char *)s) {
7874 filter_add(utf16_textfilter, NULL);
7875 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7876 PL_bufend = (char*)utf16_to_utf8(s, news,
7877 PL_bufend - (char*)s,
7879 Copy(news, s, newlen, U8);
7880 SvCUR_set(PL_linestr, newlen);
7881 PL_bufend = SvPVX(PL_linestr) + newlen;
7882 news[newlen++] = '\0';
7886 Perl_croak(aTHX_ "Unsupported script encoding");
7891 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
7892 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-8 script encoding\n"));
7897 if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */
7898 s[2] == 0xFE && s[3] == 0xFF)
7900 Perl_croak(aTHX_ "Unsupported script encoding");
7908 * Restore a source filter.
7912 restore_rsfp(pTHX_ void *f)
7914 PerlIO *fp = (PerlIO*)f;
7916 if (PL_rsfp == PerlIO_stdin())
7917 PerlIO_clearerr(PL_rsfp);
7918 else if (PL_rsfp && (PL_rsfp != fp))
7919 PerlIO_close(PL_rsfp);
7923 #ifndef PERL_NO_UTF16_FILTER
7925 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
7927 I32 count = FILTER_READ(idx+1, sv, maxlen);
7932 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7933 if (!*SvPV_nolen(sv))
7934 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7937 tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7938 sv_usepvn(sv, (char*)tmps, tend - tmps);
7944 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
7946 I32 count = FILTER_READ(idx+1, sv, maxlen);
7951 if (!*SvPV_nolen(sv))
7952 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7955 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7956 tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7957 sv_usepvn(sv, (char*)tmps, tend - tmps);
7964 Returns a pointer to the next character after the parsed
7965 vstring, as well as updating the passed in sv.
7967 Function must be called like
7970 s = scan_vstring(s,sv);
7972 The sv should already be large enough to store the vstring
7973 passed in, for performance reasons.
7978 Perl_scan_vstring(pTHX_ char *s, SV *sv)
7982 if (*pos == 'v') pos++; /* get past 'v' */
7983 while (pos < PL_bufend && (isDIGIT(*pos) || *pos == '_'))
7986 /* this may not be a v-string if followed by => */
7988 while (start < PL_bufend && isSPACE(*start))
7990 if ((PL_bufend - start) >= 2 && *start == '=' && start[1] == '>' ) {
7991 /* return string not v-string */
7992 sv_setpvn(sv,(char *)s,pos-s);
7997 if (!isALPHA(*pos)) {
7999 U8 tmpbuf[UTF8_MAXLEN+1];
8002 if (*s == 'v') s++; /* get past 'v' */
8004 sv_setpvn(sv, "", 0);
8009 /* this is atoi() that tolerates underscores */
8012 while (--end >= s) {
8017 rev += (*end - '0') * mult;
8019 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
8020 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
8021 "Integer overflow in decimal number");
8025 if (rev > 0x7FFFFFFF)
8026 Perl_croak(aTHX_ "In EBCDIC the v-string components cannot exceed 2147483647");
8028 /* Append native character for the rev point */
8029 tmpend = uvchr_to_utf8(tmpbuf, rev);
8030 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
8031 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev)))
8033 if (pos + 1 < PL_bufend && *pos == '.' && isDIGIT(pos[1]))
8039 while (pos < PL_bufend && (isDIGIT(*pos) || *pos == '_'))
8043 sv_magic(sv,NULL,PERL_MAGIC_vstring,(const char*)start, pos-start);