3 * Copyright (c) 1991-2003, Larry Wall
5 * You may distribute under the terms of either the GNU General Public
6 * License or the Artistic License, as specified in the README file.
11 * "It all comes from here, the stench and the peril." --Frodo
15 * This file is the lexer for Perl. It's closely linked to the
18 * The main routine is yylex(), which returns the next token.
22 #define PERL_IN_TOKE_C
25 #define yychar PL_yychar
26 #define yylval PL_yylval
28 static char ident_too_long[] = "Identifier too long";
29 static char c_without_g[] = "Use of /c modifier is meaningless without /g";
30 static char c_in_subst[] = "Use of /c modifier is meaningless in s///";
32 static void restore_rsfp(pTHX_ void *f);
33 #ifndef PERL_NO_UTF16_FILTER
34 static I32 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen);
35 static I32 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen);
38 #define XFAKEBRACK 128
41 #ifdef USE_UTF8_SCRIPTS
42 # define UTF (!IN_BYTES)
44 # define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
47 /* In variables named $^X, these are the legal values for X.
48 * 1999-02-27 mjd-perl-patch@plover.com */
49 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
51 /* On MacOS, respect nonbreaking spaces */
52 #ifdef MACOS_TRADITIONAL
53 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
55 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
58 /* LEX_* are values for PL_lex_state, the state of the lexer.
59 * They are arranged oddly so that the guard on the switch statement
60 * can get by with a single comparison (if the compiler is smart enough).
63 /* #define LEX_NOTPARSING 11 is done in perl.h. */
66 #define LEX_INTERPNORMAL 9
67 #define LEX_INTERPCASEMOD 8
68 #define LEX_INTERPPUSH 7
69 #define LEX_INTERPSTART 6
70 #define LEX_INTERPEND 5
71 #define LEX_INTERPENDMAYBE 4
72 #define LEX_INTERPCONCAT 3
73 #define LEX_INTERPCONST 2
74 #define LEX_FORMLINE 1
75 #define LEX_KNOWNEXT 0
83 # define YYMAXLEVEL 100
85 YYSTYPE* yylval_pointer[YYMAXLEVEL];
86 int* yychar_pointer[YYMAXLEVEL];
90 # define yylval (*yylval_pointer[yyactlevel])
91 # define yychar (*yychar_pointer[yyactlevel])
92 # define PERL_YYLEX_PARAM yylval_pointer[yyactlevel],yychar_pointer[yyactlevel]
94 # define yylex() Perl_yylex_r(aTHX_ yylval_pointer[yyactlevel],yychar_pointer[yyactlevel])
99 /* CLINE is a macro that ensures PL_copline has a sane value */
104 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
107 * Convenience functions to return different tokens and prime the
108 * lexer for the next token. They all take an argument.
110 * TOKEN : generic token (used for '(', DOLSHARP, etc)
111 * OPERATOR : generic operator
112 * AOPERATOR : assignment operator
113 * PREBLOCK : beginning the block after an if, while, foreach, ...
114 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
115 * PREREF : *EXPR where EXPR is not a simple identifier
116 * TERM : expression term
117 * LOOPX : loop exiting command (goto, last, dump, etc)
118 * FTST : file test operator
119 * FUN0 : zero-argument function
120 * FUN1 : not used, except for not, which isn't a UNIOP
121 * BOop : bitwise or or xor
123 * SHop : shift operator
124 * PWop : power operator
125 * PMop : pattern-matching operator
126 * Aop : addition-level operator
127 * Mop : multiplication-level operator
128 * Eop : equality-testing operator
129 * Rop : relational operator <= != gt
131 * Also see LOP and lop() below.
134 /* Note that REPORT() and REPORT2() will be expressions that supply
135 * their own trailing comma, not suitable for statements as such. */
136 #ifdef DEBUGGING /* Serve -DT. */
137 # define REPORT(x,retval) tokereport(x,s,(int)retval),
138 # define REPORT2(x,retval) tokereport(x,s, yylval.ival),
140 # define REPORT(x,retval)
141 # define REPORT2(x,retval)
144 #define TOKEN(retval) return (REPORT2("token",retval) PL_bufptr = s,(int)retval)
145 #define OPERATOR(retval) return (REPORT2("operator",retval) PL_expect = XTERM, PL_bufptr = s,(int)retval)
146 #define AOPERATOR(retval) return ao((REPORT2("aop",retval) PL_expect = XTERM, PL_bufptr = s,(int)retval))
147 #define PREBLOCK(retval) return (REPORT2("preblock",retval) PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
148 #define PRETERMBLOCK(retval) return (REPORT2("pretermblock",retval) PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
149 #define PREREF(retval) return (REPORT2("preref",retval) PL_expect = XREF,PL_bufptr = s,(int)retval)
150 #define TERM(retval) return (CLINE, REPORT2("term",retval) PL_expect = XOPERATOR, PL_bufptr = s,(int)retval)
151 #define LOOPX(f) return(yylval.ival=f, REPORT("loopx",f) PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
152 #define FTST(f) return(yylval.ival=f, REPORT("ftst",f) PL_expect = XTERMORDORDOR,PL_bufptr = s,(int)UNIOP)
153 #define FUN0(f) return(yylval.ival = f, REPORT("fun0",f) PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
154 #define FUN1(f) return(yylval.ival = f, REPORT("fun1",f) PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
155 #define BOop(f) return ao((yylval.ival=f, REPORT("bitorop",f) PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
156 #define BAop(f) return ao((yylval.ival=f, REPORT("bitandop",f) PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
157 #define SHop(f) return ao((yylval.ival=f, REPORT("shiftop",f) PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
158 #define PWop(f) return ao((yylval.ival=f, REPORT("powop",f) PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
159 #define PMop(f) return(yylval.ival=f, REPORT("matchop",f) PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
160 #define Aop(f) return ao((yylval.ival=f, REPORT("add",f) PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
161 #define Mop(f) return ao((yylval.ival=f, REPORT("mul",f) PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
162 #define Eop(f) return(yylval.ival=f, REPORT("eq",f) PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
163 #define Rop(f) return(yylval.ival=f, REPORT("rel",f) PL_expect = XTERM,PL_bufptr = s,(int)RELOP)
165 /* This bit of chicanery makes a unary function followed by
166 * a parenthesis into a function with one argument, highest precedence.
167 * The UNIDOR macro is for unary functions that can be followed by the //
168 * operator (such as C<shift // 0>).
170 #define UNI2(f,x) return(yylval.ival = f, \
174 PL_last_uni = PL_oldbufptr, \
175 PL_last_lop_op = f, \
176 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
177 #define UNI(f) UNI2(f,XTERM)
178 #define UNIDOR(f) UNI2(f,XTERMORDORDOR)
180 #define UNIBRACK(f) return(yylval.ival = f, \
183 PL_last_uni = PL_oldbufptr, \
184 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
186 /* grandfather return to old style */
187 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
192 S_tokereport(pTHX_ char *thing, char* s, I32 rv)
195 SV* report = newSVpv(thing, 0);
196 Perl_sv_catpvf(aTHX_ report, ":line %d:%"IVdf":", CopLINE(PL_curcop),
199 if (s - PL_bufptr > 0)
200 sv_catpvn(report, PL_bufptr, s - PL_bufptr);
202 if (PL_oldbufptr && *PL_oldbufptr)
203 sv_catpv(report, PL_tokenbuf);
205 PerlIO_printf(Perl_debug_log, "### %s\n", SvPV_nolen(report));
214 * This subroutine detects &&=, ||=, and //= and turns an ANDAND, OROR or DORDOR
215 * into an OP_ANDASSIGN, OP_ORASSIGN, or OP_DORASSIGN
219 S_ao(pTHX_ int toketype)
221 if (*PL_bufptr == '=') {
223 if (toketype == ANDAND)
224 yylval.ival = OP_ANDASSIGN;
225 else if (toketype == OROR)
226 yylval.ival = OP_ORASSIGN;
227 else if (toketype == DORDOR)
228 yylval.ival = OP_DORASSIGN;
236 * When Perl expects an operator and finds something else, no_op
237 * prints the warning. It always prints "<something> found where
238 * operator expected. It prints "Missing semicolon on previous line?"
239 * if the surprise occurs at the start of the line. "do you need to
240 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
241 * where the compiler doesn't know if foo is a method call or a function.
242 * It prints "Missing operator before end of line" if there's nothing
243 * after the missing operator, or "... before <...>" if there is something
244 * after the missing operator.
248 S_no_op(pTHX_ char *what, char *s)
250 char *oldbp = PL_bufptr;
251 bool is_first = (PL_oldbufptr == PL_linestart);
257 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
259 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
260 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
262 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
263 if (t < PL_bufptr && isSPACE(*t))
264 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
265 t - PL_oldoldbufptr, PL_oldoldbufptr);
269 Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
276 * Complain about missing quote/regexp/heredoc terminator.
277 * If it's called with (char *)NULL then it cauterizes the line buffer.
278 * If we're in a delimited string and the delimiter is a control
279 * character, it's reformatted into a two-char sequence like ^C.
284 S_missingterm(pTHX_ char *s)
289 char *nl = strrchr(s,'\n');
295 iscntrl(PL_multi_close)
297 PL_multi_close < 32 || PL_multi_close == 127
301 tmpbuf[1] = toCTRL(PL_multi_close);
307 *tmpbuf = (char)PL_multi_close;
311 q = strchr(s,'"') ? '\'' : '"';
312 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
320 Perl_deprecate(pTHX_ char *s)
322 if (ckWARN(WARN_DEPRECATED))
323 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED), "Use of %s is deprecated", s);
327 Perl_deprecate_old(pTHX_ char *s)
329 /* This function should NOT be called for any new deprecated warnings */
330 /* Use Perl_deprecate instead */
332 /* It is here to maintain backward compatibility with the pre-5.8 */
333 /* warnings category hierarchy. The "deprecated" category used to */
334 /* live under the "syntax" category. It is now a top-level category */
335 /* in its own right. */
337 if (ckWARN2(WARN_DEPRECATED, WARN_SYNTAX))
338 Perl_warner(aTHX_ packWARN2(WARN_DEPRECATED, WARN_SYNTAX),
339 "Use of %s is deprecated", s);
344 * Deprecate a comma-less variable list.
350 deprecate_old("comma-less variable list");
354 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
355 * utf16-to-utf8-reversed.
358 #ifdef PERL_CR_FILTER
362 register char *s = SvPVX(sv);
363 register char *e = s + SvCUR(sv);
364 /* outer loop optimized to do nothing if there are no CR-LFs */
366 if (*s++ == '\r' && *s == '\n') {
367 /* hit a CR-LF, need to copy the rest */
368 register char *d = s - 1;
371 if (*s == '\r' && s[1] == '\n')
382 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
384 I32 count = FILTER_READ(idx+1, sv, maxlen);
385 if (count > 0 && !maxlen)
393 * Initialize variables. Uses the Perl save_stack to save its state (for
394 * recursive calls to the parser).
398 Perl_lex_start(pTHX_ SV *line)
403 SAVEI32(PL_lex_dojoin);
404 SAVEI32(PL_lex_brackets);
405 SAVEI32(PL_lex_casemods);
406 SAVEI32(PL_lex_starts);
407 SAVEI32(PL_lex_state);
408 SAVEVPTR(PL_lex_inpat);
409 SAVEI32(PL_lex_inwhat);
410 if (PL_lex_state == LEX_KNOWNEXT) {
411 I32 toke = PL_nexttoke;
412 while (--toke >= 0) {
413 SAVEI32(PL_nexttype[toke]);
414 SAVEVPTR(PL_nextval[toke]);
416 SAVEI32(PL_nexttoke);
418 SAVECOPLINE(PL_curcop);
421 SAVEPPTR(PL_oldbufptr);
422 SAVEPPTR(PL_oldoldbufptr);
423 SAVEPPTR(PL_last_lop);
424 SAVEPPTR(PL_last_uni);
425 SAVEPPTR(PL_linestart);
426 SAVESPTR(PL_linestr);
427 SAVEGENERICPV(PL_lex_brackstack);
428 SAVEGENERICPV(PL_lex_casestack);
429 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
430 SAVESPTR(PL_lex_stuff);
431 SAVEI32(PL_lex_defer);
432 SAVEI32(PL_sublex_info.sub_inwhat);
433 SAVESPTR(PL_lex_repl);
435 SAVEINT(PL_lex_expect);
437 PL_lex_state = LEX_NORMAL;
441 New(899, PL_lex_brackstack, 120, char);
442 New(899, PL_lex_casestack, 12, char);
444 *PL_lex_casestack = '\0';
447 PL_lex_stuff = Nullsv;
448 PL_lex_repl = Nullsv;
452 PL_sublex_info.sub_inwhat = 0;
454 if (SvREADONLY(PL_linestr))
455 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
456 s = SvPV(PL_linestr, len);
457 if (!len || s[len-1] != ';') {
458 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
459 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
460 sv_catpvn(PL_linestr, "\n;", 2);
462 SvTEMP_off(PL_linestr);
463 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
464 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
465 PL_last_lop = PL_last_uni = Nullch;
471 * Finalizer for lexing operations. Must be called when the parser is
472 * done with the lexer.
478 PL_doextract = FALSE;
483 * This subroutine has nothing to do with tilting, whether at windmills
484 * or pinball tables. Its name is short for "increment line". It
485 * increments the current line number in CopLINE(PL_curcop) and checks
486 * to see whether the line starts with a comment of the form
487 * # line 500 "foo.pm"
488 * If so, it sets the current line number and file to the values in the comment.
492 S_incline(pTHX_ char *s)
499 CopLINE_inc(PL_curcop);
502 while (SPACE_OR_TAB(*s)) s++;
503 if (strnEQ(s, "line", 4))
507 if (SPACE_OR_TAB(*s))
511 while (SPACE_OR_TAB(*s)) s++;
517 while (SPACE_OR_TAB(*s))
519 if (*s == '"' && (t = strchr(s+1, '"'))) {
524 for (t = s; !isSPACE(*t); t++) ;
527 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
529 if (*e != '\n' && *e != '\0')
530 return; /* false alarm */
535 CopFILE_free(PL_curcop);
536 CopFILE_set(PL_curcop, s);
539 CopLINE_set(PL_curcop, atoi(n)-1);
544 * Called to gobble the appropriate amount and type of whitespace.
545 * Skips comments as well.
549 S_skipspace(pTHX_ register char *s)
551 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
552 while (s < PL_bufend && SPACE_OR_TAB(*s))
558 SSize_t oldprevlen, oldoldprevlen;
559 SSize_t oldloplen = 0, oldunilen = 0;
560 while (s < PL_bufend && isSPACE(*s)) {
561 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
566 if (s < PL_bufend && *s == '#') {
567 while (s < PL_bufend && *s != '\n')
571 if (PL_in_eval && !PL_rsfp) {
578 /* only continue to recharge the buffer if we're at the end
579 * of the buffer, we're not reading from a source filter, and
580 * we're in normal lexing mode
582 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
583 PL_lex_state == LEX_FORMLINE)
586 /* try to recharge the buffer */
587 if ((s = filter_gets(PL_linestr, PL_rsfp,
588 (prevlen = SvCUR(PL_linestr)))) == Nullch)
590 /* end of file. Add on the -p or -n magic */
591 if (PL_minus_n || PL_minus_p) {
592 sv_setpv(PL_linestr,PL_minus_p ?
593 ";}continue{print or die qq(-p destination: $!\\n)" :
595 sv_catpv(PL_linestr,";}");
596 PL_minus_n = PL_minus_p = 0;
599 sv_setpv(PL_linestr,";");
601 /* reset variables for next time we lex */
602 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
604 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
605 PL_last_lop = PL_last_uni = Nullch;
607 /* Close the filehandle. Could be from -P preprocessor,
608 * STDIN, or a regular file. If we were reading code from
609 * STDIN (because the commandline held no -e or filename)
610 * then we don't close it, we reset it so the code can
611 * read from STDIN too.
614 if (PL_preprocess && !PL_in_eval)
615 (void)PerlProc_pclose(PL_rsfp);
616 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
617 PerlIO_clearerr(PL_rsfp);
619 (void)PerlIO_close(PL_rsfp);
624 /* not at end of file, so we only read another line */
625 /* make corresponding updates to old pointers, for yyerror() */
626 oldprevlen = PL_oldbufptr - PL_bufend;
627 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
629 oldunilen = PL_last_uni - PL_bufend;
631 oldloplen = PL_last_lop - PL_bufend;
632 PL_linestart = PL_bufptr = s + prevlen;
633 PL_bufend = s + SvCUR(PL_linestr);
635 PL_oldbufptr = s + oldprevlen;
636 PL_oldoldbufptr = s + oldoldprevlen;
638 PL_last_uni = s + oldunilen;
640 PL_last_lop = s + oldloplen;
643 /* debugger active and we're not compiling the debugger code,
644 * so store the line into the debugger's array of lines
646 if (PERLDB_LINE && PL_curstash != PL_debstash) {
647 SV *sv = NEWSV(85,0);
649 sv_upgrade(sv, SVt_PVMG);
650 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
653 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
660 * Check the unary operators to ensure there's no ambiguity in how they're
661 * used. An ambiguous piece of code would be:
663 * This doesn't mean rand() + 5. Because rand() is a unary operator,
664 * the +5 is its argument.
673 if (PL_oldoldbufptr != PL_last_uni)
675 while (isSPACE(*PL_last_uni))
677 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
678 if ((t = strchr(s, '(')) && t < PL_bufptr)
680 if (ckWARN_d(WARN_AMBIGUOUS)){
683 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
684 "Warning: Use of \"%s\" without parens is ambiguous",
691 * LOP : macro to build a list operator. Its behaviour has been replaced
692 * with a subroutine, S_lop() for which LOP is just another name.
695 #define LOP(f,x) return lop(f,x,s)
699 * Build a list operator (or something that might be one). The rules:
700 * - if we have a next token, then it's a list operator [why?]
701 * - if the next thing is an opening paren, then it's a function
702 * - else it's a list operator
706 S_lop(pTHX_ I32 f, int x, char *s)
713 PL_last_lop = PL_oldbufptr;
714 PL_last_lop_op = (OPCODE)f;
728 * When the lexer realizes it knows the next token (for instance,
729 * it is reordering tokens for the parser) then it can call S_force_next
730 * to know what token to return the next time the lexer is called. Caller
731 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
732 * handles the token correctly.
736 S_force_next(pTHX_ I32 type)
738 PL_nexttype[PL_nexttoke] = type;
740 if (PL_lex_state != LEX_KNOWNEXT) {
741 PL_lex_defer = PL_lex_state;
742 PL_lex_expect = PL_expect;
743 PL_lex_state = LEX_KNOWNEXT;
749 * When the lexer knows the next thing is a word (for instance, it has
750 * just seen -> and it knows that the next char is a word char, then
751 * it calls S_force_word to stick the next word into the PL_next lookahead.
754 * char *start : buffer position (must be within PL_linestr)
755 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
756 * int check_keyword : if true, Perl checks to make sure the word isn't
757 * a keyword (do this if the word is a label, e.g. goto FOO)
758 * int allow_pack : if true, : characters will also be allowed (require,
760 * int allow_initial_tick : used by the "sub" lexer only.
764 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
769 start = skipspace(start);
771 if (isIDFIRST_lazy_if(s,UTF) ||
772 (allow_pack && *s == ':') ||
773 (allow_initial_tick && *s == '\'') )
775 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
776 if (check_keyword && keyword(PL_tokenbuf, len))
778 if (token == METHOD) {
783 PL_expect = XOPERATOR;
786 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
787 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
795 * Called when the lexer wants $foo *foo &foo etc, but the program
796 * text only contains the "foo" portion. The first argument is a pointer
797 * to the "foo", and the second argument is the type symbol to prefix.
798 * Forces the next token to be a "WORD".
799 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
803 S_force_ident(pTHX_ register char *s, int kind)
806 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
807 PL_nextval[PL_nexttoke].opval = o;
810 o->op_private = OPpCONST_ENTERED;
811 /* XXX see note in pp_entereval() for why we forgo typo
812 warnings if the symbol must be introduced in an eval.
814 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
815 kind == '$' ? SVt_PV :
816 kind == '@' ? SVt_PVAV :
817 kind == '%' ? SVt_PVHV :
825 Perl_str_to_version(pTHX_ SV *sv)
830 char *start = SvPVx(sv,len);
831 bool utf = SvUTF8(sv) ? TRUE : FALSE;
832 char *end = start + len;
833 while (start < end) {
837 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
842 retval += ((NV)n)/nshift;
851 * Forces the next token to be a version number.
852 * If the next token appears to be an invalid version number, (e.g. "v2b"),
853 * and if "guessing" is TRUE, then no new token is created (and the caller
854 * must use an alternative parsing method).
858 S_force_version(pTHX_ char *s, int guessing)
860 OP *version = Nullop;
869 while (isDIGIT(*d) || *d == '_' || *d == '.')
871 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
873 s = scan_num(s, &yylval);
874 version = yylval.opval;
875 ver = cSVOPx(version)->op_sv;
876 if (SvPOK(ver) && !SvNIOK(ver)) {
877 (void)SvUPGRADE(ver, SVt_PVNV);
878 SvNVX(ver) = str_to_version(ver);
879 SvNOK_on(ver); /* hint that it is a version */
886 /* NOTE: The parser sees the package name and the VERSION swapped */
887 PL_nextval[PL_nexttoke].opval = version;
895 * Tokenize a quoted string passed in as an SV. It finds the next
896 * chunk, up to end of string or a backslash. It may make a new
897 * SV containing that chunk (if HINT_NEW_STRING is on). It also
902 S_tokeq(pTHX_ SV *sv)
913 s = SvPV_force(sv, len);
914 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
917 while (s < send && *s != '\\')
922 if ( PL_hints & HINT_NEW_STRING ) {
923 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
929 if (s + 1 < send && (s[1] == '\\'))
930 s++; /* all that, just for this */
935 SvCUR_set(sv, d - SvPVX(sv));
937 if ( PL_hints & HINT_NEW_STRING )
938 return new_constant(NULL, 0, "q", sv, pv, "q");
943 * Now come three functions related to double-quote context,
944 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
945 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
946 * interact with PL_lex_state, and create fake ( ... ) argument lists
947 * to handle functions and concatenation.
948 * They assume that whoever calls them will be setting up a fake
949 * join call, because each subthing puts a ',' after it. This lets
952 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
954 * (I'm not sure whether the spurious commas at the end of lcfirst's
955 * arguments and join's arguments are created or not).
960 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
962 * Pattern matching will set PL_lex_op to the pattern-matching op to
963 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
965 * OP_CONST and OP_READLINE are easy--just make the new op and return.
967 * Everything else becomes a FUNC.
969 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
970 * had an OP_CONST or OP_READLINE). This just sets us up for a
971 * call to S_sublex_push().
977 register I32 op_type = yylval.ival;
979 if (op_type == OP_NULL) {
980 yylval.opval = PL_lex_op;
984 if (op_type == OP_CONST || op_type == OP_READLINE) {
985 SV *sv = tokeq(PL_lex_stuff);
987 if (SvTYPE(sv) == SVt_PVIV) {
988 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
994 nsv = newSVpvn(p, len);
1000 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
1001 PL_lex_stuff = Nullsv;
1002 /* Allow <FH> // "foo" */
1003 if (op_type == OP_READLINE)
1004 PL_expect = XTERMORDORDOR;
1008 PL_sublex_info.super_state = PL_lex_state;
1009 PL_sublex_info.sub_inwhat = op_type;
1010 PL_sublex_info.sub_op = PL_lex_op;
1011 PL_lex_state = LEX_INTERPPUSH;
1015 yylval.opval = PL_lex_op;
1025 * Create a new scope to save the lexing state. The scope will be
1026 * ended in S_sublex_done. Returns a '(', starting the function arguments
1027 * to the uc, lc, etc. found before.
1028 * Sets PL_lex_state to LEX_INTERPCONCAT.
1036 PL_lex_state = PL_sublex_info.super_state;
1037 SAVEI32(PL_lex_dojoin);
1038 SAVEI32(PL_lex_brackets);
1039 SAVEI32(PL_lex_casemods);
1040 SAVEI32(PL_lex_starts);
1041 SAVEI32(PL_lex_state);
1042 SAVEVPTR(PL_lex_inpat);
1043 SAVEI32(PL_lex_inwhat);
1044 SAVECOPLINE(PL_curcop);
1045 SAVEPPTR(PL_bufptr);
1046 SAVEPPTR(PL_bufend);
1047 SAVEPPTR(PL_oldbufptr);
1048 SAVEPPTR(PL_oldoldbufptr);
1049 SAVEPPTR(PL_last_lop);
1050 SAVEPPTR(PL_last_uni);
1051 SAVEPPTR(PL_linestart);
1052 SAVESPTR(PL_linestr);
1053 SAVEGENERICPV(PL_lex_brackstack);
1054 SAVEGENERICPV(PL_lex_casestack);
1056 PL_linestr = PL_lex_stuff;
1057 PL_lex_stuff = Nullsv;
1059 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1060 = SvPVX(PL_linestr);
1061 PL_bufend += SvCUR(PL_linestr);
1062 PL_last_lop = PL_last_uni = Nullch;
1063 SAVEFREESV(PL_linestr);
1065 PL_lex_dojoin = FALSE;
1066 PL_lex_brackets = 0;
1067 New(899, PL_lex_brackstack, 120, char);
1068 New(899, PL_lex_casestack, 12, char);
1069 PL_lex_casemods = 0;
1070 *PL_lex_casestack = '\0';
1072 PL_lex_state = LEX_INTERPCONCAT;
1073 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
1075 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1076 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1077 PL_lex_inpat = PL_sublex_info.sub_op;
1079 PL_lex_inpat = Nullop;
1086 * Restores lexer state after a S_sublex_push.
1092 if (!PL_lex_starts++) {
1093 SV *sv = newSVpvn("",0);
1094 if (SvUTF8(PL_linestr))
1096 PL_expect = XOPERATOR;
1097 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1101 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1102 PL_lex_state = LEX_INTERPCASEMOD;
1106 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1107 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1108 PL_linestr = PL_lex_repl;
1110 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1111 PL_bufend += SvCUR(PL_linestr);
1112 PL_last_lop = PL_last_uni = Nullch;
1113 SAVEFREESV(PL_linestr);
1114 PL_lex_dojoin = FALSE;
1115 PL_lex_brackets = 0;
1116 PL_lex_casemods = 0;
1117 *PL_lex_casestack = '\0';
1119 if (SvEVALED(PL_lex_repl)) {
1120 PL_lex_state = LEX_INTERPNORMAL;
1122 /* we don't clear PL_lex_repl here, so that we can check later
1123 whether this is an evalled subst; that means we rely on the
1124 logic to ensure sublex_done() is called again only via the
1125 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1128 PL_lex_state = LEX_INTERPCONCAT;
1129 PL_lex_repl = Nullsv;
1135 PL_bufend = SvPVX(PL_linestr);
1136 PL_bufend += SvCUR(PL_linestr);
1137 PL_expect = XOPERATOR;
1138 PL_sublex_info.sub_inwhat = 0;
1146 Extracts a pattern, double-quoted string, or transliteration. This
1149 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1150 processing a pattern (PL_lex_inpat is true), a transliteration
1151 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1153 Returns a pointer to the character scanned up to. Iff this is
1154 advanced from the start pointer supplied (ie if anything was
1155 successfully parsed), will leave an OP for the substring scanned
1156 in yylval. Caller must intuit reason for not parsing further
1157 by looking at the next characters herself.
1161 double-quoted style: \r and \n
1162 regexp special ones: \D \s
1164 backrefs: \1 (deprecated in substitution replacements)
1165 case and quoting: \U \Q \E
1166 stops on @ and $, but not for $ as tail anchor
1168 In transliterations:
1169 characters are VERY literal, except for - not at the start or end
1170 of the string, which indicates a range. scan_const expands the
1171 range to the full set of intermediate characters.
1173 In double-quoted strings:
1175 double-quoted style: \r and \n
1177 backrefs: \1 (deprecated)
1178 case and quoting: \U \Q \E
1181 scan_const does *not* construct ops to handle interpolated strings.
1182 It stops processing as soon as it finds an embedded $ or @ variable
1183 and leaves it to the caller to work out what's going on.
1185 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @::foo.
1187 $ in pattern could be $foo or could be tail anchor. Assumption:
1188 it's a tail anchor if $ is the last thing in the string, or if it's
1189 followed by one of ")| \n\t"
1191 \1 (backreferences) are turned into $1
1193 The structure of the code is
1194 while (there's a character to process) {
1195 handle transliteration ranges
1196 skip regexp comments
1197 skip # initiated comments in //x patterns
1198 check for embedded @foo
1199 check for embedded scalars
1201 leave intact backslashes from leave (below)
1202 deprecate \1 in strings and sub replacements
1203 handle string-changing backslashes \l \U \Q \E, etc.
1204 switch (what was escaped) {
1205 handle - in a transliteration (becomes a literal -)
1206 handle \132 octal characters
1207 handle 0x15 hex characters
1208 handle \cV (control V)
1209 handle printf backslashes (\f, \r, \n, etc)
1211 } (end if backslash)
1212 } (end while character to read)
1217 S_scan_const(pTHX_ char *start)
1219 register char *send = PL_bufend; /* end of the constant */
1220 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1221 register char *s = start; /* start of the constant */
1222 register char *d = SvPVX(sv); /* destination for copies */
1223 bool dorange = FALSE; /* are we in a translit range? */
1224 bool didrange = FALSE; /* did we just finish a range? */
1225 I32 has_utf8 = FALSE; /* Output constant is UTF8 */
1226 I32 this_utf8 = UTF; /* The source string is assumed to be UTF8 */
1229 const char *leaveit = /* set of acceptably-backslashed characters */
1231 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1234 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1235 /* If we are doing a trans and we know we want UTF8 set expectation */
1236 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
1237 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1241 while (s < send || dorange) {
1242 /* get transliterations out of the way (they're most literal) */
1243 if (PL_lex_inwhat == OP_TRANS) {
1244 /* expand a range A-Z to the full set of characters. AIE! */
1246 I32 i; /* current expanded character */
1247 I32 min; /* first character in range */
1248 I32 max; /* last character in range */
1251 char *c = (char*)utf8_hop((U8*)d, -1);
1255 *c = (char)UTF_TO_NATIVE(0xff);
1256 /* mark the range as done, and continue */
1262 i = d - SvPVX(sv); /* remember current offset */
1263 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1264 d = SvPVX(sv) + i; /* refresh d after realloc */
1265 d -= 2; /* eat the first char and the - */
1267 min = (U8)*d; /* first char in range */
1268 max = (U8)d[1]; /* last char in range */
1272 "Invalid range \"%c-%c\" in transliteration operator",
1273 (char)min, (char)max);
1277 if ((isLOWER(min) && isLOWER(max)) ||
1278 (isUPPER(min) && isUPPER(max))) {
1280 for (i = min; i <= max; i++)
1282 *d++ = NATIVE_TO_NEED(has_utf8,i);
1284 for (i = min; i <= max; i++)
1286 *d++ = NATIVE_TO_NEED(has_utf8,i);
1291 for (i = min; i <= max; i++)
1294 /* mark the range as done, and continue */
1300 /* range begins (ignore - as first or last char) */
1301 else if (*s == '-' && s+1 < send && s != start) {
1303 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
1306 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
1318 /* if we get here, we're not doing a transliteration */
1320 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1321 except for the last char, which will be done separately. */
1322 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1324 while (s < send && *s != ')')
1325 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1327 else if (s[2] == '{' /* This should match regcomp.c */
1328 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1331 char *regparse = s + (s[2] == '{' ? 3 : 4);
1334 while (count && (c = *regparse)) {
1335 if (c == '\\' && regparse[1])
1343 if (*regparse != ')') {
1344 regparse--; /* Leave one char for continuation. */
1345 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1347 while (s < regparse)
1348 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1352 /* likewise skip #-initiated comments in //x patterns */
1353 else if (*s == '#' && PL_lex_inpat &&
1354 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1355 while (s+1 < send && *s != '\n')
1356 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1359 /* check for embedded arrays
1360 (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-)
1362 else if (*s == '@' && s[1]
1363 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1])))
1366 /* check for embedded scalars. only stop if we're sure it's a
1369 else if (*s == '$') {
1370 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1372 if (s + 1 < send && !strchr("()| \r\n\t", s[1]))
1373 break; /* in regexp, $ might be tail anchor */
1376 /* End of else if chain - OP_TRANS rejoin rest */
1379 if (*s == '\\' && s+1 < send) {
1382 /* some backslashes we leave behind */
1383 if (*leaveit && *s && strchr(leaveit, *s)) {
1384 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
1385 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1389 /* deprecate \1 in strings and substitution replacements */
1390 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1391 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1393 if (ckWARN(WARN_SYNTAX))
1394 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
1399 /* string-change backslash escapes */
1400 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1405 /* if we get here, it's either a quoted -, or a digit */
1408 /* quoted - in transliterations */
1410 if (PL_lex_inwhat == OP_TRANS) {
1417 if (ckWARN(WARN_MISC) &&
1420 Perl_warner(aTHX_ packWARN(WARN_MISC),
1421 "Unrecognized escape \\%c passed through",
1423 /* default action is to copy the quoted character */
1424 goto default_action;
1427 /* \132 indicates an octal constant */
1428 case '0': case '1': case '2': case '3':
1429 case '4': case '5': case '6': case '7':
1433 uv = grok_oct(s, &len, &flags, NULL);
1436 goto NUM_ESCAPE_INSERT;
1438 /* \x24 indicates a hex constant */
1442 char* e = strchr(s, '}');
1443 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1444 PERL_SCAN_DISALLOW_PREFIX;
1449 yyerror("Missing right brace on \\x{}");
1453 uv = grok_hex(s, &len, &flags, NULL);
1459 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
1460 uv = grok_hex(s, &len, &flags, NULL);
1466 /* Insert oct or hex escaped character.
1467 * There will always enough room in sv since such
1468 * escapes will be longer than any UTF-8 sequence
1469 * they can end up as. */
1471 /* We need to map to chars to ASCII before doing the tests
1474 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(uv))) {
1475 if (!has_utf8 && uv > 255) {
1476 /* Might need to recode whatever we have
1477 * accumulated so far if it contains any
1480 * (Can't we keep track of that and avoid
1481 * this rescan? --jhi)
1485 for (c = (U8 *) SvPVX(sv); c < (U8 *)d; c++) {
1486 if (!NATIVE_IS_INVARIANT(*c)) {
1491 STRLEN offset = d - SvPVX(sv);
1493 d = SvGROW(sv, SvLEN(sv) + hicount + 1) + offset;
1497 while (src >= (U8 *)SvPVX(sv)) {
1498 if (!NATIVE_IS_INVARIANT(*src)) {
1499 U8 ch = NATIVE_TO_ASCII(*src);
1500 *dst-- = (U8)UTF8_EIGHT_BIT_LO(ch);
1501 *dst-- = (U8)UTF8_EIGHT_BIT_HI(ch);
1511 if (has_utf8 || uv > 255) {
1512 d = (char*)uvchr_to_utf8((U8*)d, uv);
1514 if (PL_lex_inwhat == OP_TRANS &&
1515 PL_sublex_info.sub_op) {
1516 PL_sublex_info.sub_op->op_private |=
1517 (PL_lex_repl ? OPpTRANS_FROM_UTF
1530 /* \N{LATIN SMALL LETTER A} is a named character */
1534 char* e = strchr(s, '}');
1540 yyerror("Missing right brace on \\N{}");
1544 if (e > s + 2 && s[1] == 'U' && s[2] == '+') {
1546 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1547 PERL_SCAN_DISALLOW_PREFIX;
1550 uv = grok_hex(s, &len, &flags, NULL);
1552 goto NUM_ESCAPE_INSERT;
1554 res = newSVpvn(s + 1, e - s - 1);
1555 res = new_constant( Nullch, 0, "charnames",
1556 res, Nullsv, "\\N{...}" );
1558 sv_utf8_upgrade(res);
1559 str = SvPV(res,len);
1560 #ifdef EBCDIC_NEVER_MIND
1561 /* charnames uses pack U and that has been
1562 * recently changed to do the below uni->native
1563 * mapping, so this would be redundant (and wrong,
1564 * the code point would be doubly converted).
1565 * But leave this in just in case the pack U change
1566 * gets revoked, but the semantics is still
1567 * desireable for charnames. --jhi */
1569 UV uv = utf8_to_uvchr((U8*)str, 0);
1572 U8 tmpbuf[UTF8_MAXLEN+1], *d;
1574 d = uvchr_to_utf8(tmpbuf, UNI_TO_NATIVE(uv));
1575 sv_setpvn(res, (char *)tmpbuf, d - tmpbuf);
1576 str = SvPV(res, len);
1580 if (!has_utf8 && SvUTF8(res)) {
1581 char *ostart = SvPVX(sv);
1582 SvCUR_set(sv, d - ostart);
1585 sv_utf8_upgrade(sv);
1586 /* this just broke our allocation above... */
1587 SvGROW(sv, (STRLEN)(send - start));
1588 d = SvPVX(sv) + SvCUR(sv);
1591 if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */
1592 char *odest = SvPVX(sv);
1594 SvGROW(sv, (SvLEN(sv) + len - (e - s + 4)));
1595 d = SvPVX(sv) + (d - odest);
1597 Copy(str, d, len, char);
1604 yyerror("Missing braces on \\N{}");
1607 /* \c is a control character */
1616 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
1619 yyerror("Missing control char name in \\c");
1623 /* printf-style backslashes, formfeeds, newlines, etc */
1625 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
1628 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
1631 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
1634 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
1637 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
1640 *d++ = ASCII_TO_NEED(has_utf8,'\033');
1643 *d++ = ASCII_TO_NEED(has_utf8,'\007');
1649 } /* end if (backslash) */
1652 /* If we started with encoded form, or already know we want it
1653 and then encode the next character */
1654 if ((has_utf8 || this_utf8) && !NATIVE_IS_INVARIANT((U8)(*s))) {
1656 UV uv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
1657 STRLEN need = UNISKIP(NATIVE_TO_UNI(uv));
1660 /* encoded value larger than old, need extra space (NOTE: SvCUR() not set here) */
1661 STRLEN off = d - SvPVX(sv);
1662 d = SvGROW(sv, SvLEN(sv) + (need-len)) + off;
1664 d = (char*)uvchr_to_utf8((U8*)d, uv);
1668 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1670 } /* while loop to process each character */
1672 /* terminate the string and set up the sv */
1674 SvCUR_set(sv, d - SvPVX(sv));
1675 if (SvCUR(sv) >= SvLEN(sv))
1676 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
1679 if (PL_encoding && !has_utf8) {
1680 sv_recode_to_utf8(sv, PL_encoding);
1686 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1687 PL_sublex_info.sub_op->op_private |=
1688 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1692 /* shrink the sv if we allocated more than we used */
1693 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1694 SvLEN_set(sv, SvCUR(sv) + 1);
1695 Renew(SvPVX(sv), SvLEN(sv), char);
1698 /* return the substring (via yylval) only if we parsed anything */
1699 if (s > PL_bufptr) {
1700 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1701 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1703 ( PL_lex_inwhat == OP_TRANS
1705 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1708 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1715 * Returns TRUE if there's more to the expression (e.g., a subscript),
1718 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1720 * ->[ and ->{ return TRUE
1721 * { and [ outside a pattern are always subscripts, so return TRUE
1722 * if we're outside a pattern and it's not { or [, then return FALSE
1723 * if we're in a pattern and the first char is a {
1724 * {4,5} (any digits around the comma) returns FALSE
1725 * if we're in a pattern and the first char is a [
1727 * [SOMETHING] has a funky algorithm to decide whether it's a
1728 * character class or not. It has to deal with things like
1729 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1730 * anything else returns TRUE
1733 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1736 S_intuit_more(pTHX_ register char *s)
1738 if (PL_lex_brackets)
1740 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1742 if (*s != '{' && *s != '[')
1747 /* In a pattern, so maybe we have {n,m}. */
1764 /* On the other hand, maybe we have a character class */
1767 if (*s == ']' || *s == '^')
1770 /* this is terrifying, and it works */
1771 int weight = 2; /* let's weigh the evidence */
1773 unsigned char un_char = 255, last_un_char;
1774 char *send = strchr(s,']');
1775 char tmpbuf[sizeof PL_tokenbuf * 4];
1777 if (!send) /* has to be an expression */
1780 Zero(seen,256,char);
1783 else if (isDIGIT(*s)) {
1785 if (isDIGIT(s[1]) && s[2] == ']')
1791 for (; s < send; s++) {
1792 last_un_char = un_char;
1793 un_char = (unsigned char)*s;
1798 weight -= seen[un_char] * 10;
1799 if (isALNUM_lazy_if(s+1,UTF)) {
1800 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1801 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1806 else if (*s == '$' && s[1] &&
1807 strchr("[#!%*<>()-=",s[1])) {
1808 if (/*{*/ strchr("])} =",s[2]))
1817 if (strchr("wds]",s[1]))
1819 else if (seen['\''] || seen['"'])
1821 else if (strchr("rnftbxcav",s[1]))
1823 else if (isDIGIT(s[1])) {
1825 while (s[1] && isDIGIT(s[1]))
1835 if (strchr("aA01! ",last_un_char))
1837 if (strchr("zZ79~",s[1]))
1839 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1840 weight -= 5; /* cope with negative subscript */
1843 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1844 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1849 if (keyword(tmpbuf, d - tmpbuf))
1852 if (un_char == last_un_char + 1)
1854 weight -= seen[un_char];
1859 if (weight >= 0) /* probably a character class */
1869 * Does all the checking to disambiguate
1871 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1872 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1874 * First argument is the stuff after the first token, e.g. "bar".
1876 * Not a method if bar is a filehandle.
1877 * Not a method if foo is a subroutine prototyped to take a filehandle.
1878 * Not a method if it's really "Foo $bar"
1879 * Method if it's "foo $bar"
1880 * Not a method if it's really "print foo $bar"
1881 * Method if it's really "foo package::" (interpreted as package->foo)
1882 * Not a method if bar is known to be a subroutine ("sub bar; foo bar")
1883 * Not a method if bar is a filehandle or package, but is quoted with
1888 S_intuit_method(pTHX_ char *start, GV *gv)
1890 char *s = start + (*start == '$');
1891 char tmpbuf[sizeof PL_tokenbuf];
1899 if ((cv = GvCVu(gv))) {
1900 char *proto = SvPVX(cv);
1910 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1911 /* start is the beginning of the possible filehandle/object,
1912 * and s is the end of it
1913 * tmpbuf is a copy of it
1916 if (*start == '$') {
1917 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1922 return *s == '(' ? FUNCMETH : METHOD;
1924 if (!keyword(tmpbuf, len)) {
1925 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1930 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1931 if (indirgv && GvCVu(indirgv))
1933 /* filehandle or package name makes it a method */
1934 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1936 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1937 return 0; /* no assumptions -- "=>" quotes bearword */
1939 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1940 newSVpvn(tmpbuf,len));
1941 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1945 return *s == '(' ? FUNCMETH : METHOD;
1953 * Return a string of Perl code to load the debugger. If PERL5DB
1954 * is set, it will return the contents of that, otherwise a
1955 * compile-time require of perl5db.pl.
1962 char *pdb = PerlEnv_getenv("PERL5DB");
1966 SETERRNO(0,SS_NORMAL);
1967 return "BEGIN { require 'perl5db.pl' }";
1973 /* Encoded script support. filter_add() effectively inserts a
1974 * 'pre-processing' function into the current source input stream.
1975 * Note that the filter function only applies to the current source file
1976 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1978 * The datasv parameter (which may be NULL) can be used to pass
1979 * private data to this instance of the filter. The filter function
1980 * can recover the SV using the FILTER_DATA macro and use it to
1981 * store private buffers and state information.
1983 * The supplied datasv parameter is upgraded to a PVIO type
1984 * and the IoDIRP/IoANY field is used to store the function pointer,
1985 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1986 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1987 * private use must be set using malloc'd pointers.
1991 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1996 if (!PL_rsfp_filters)
1997 PL_rsfp_filters = newAV();
1999 datasv = NEWSV(255,0);
2000 if (!SvUPGRADE(datasv, SVt_PVIO))
2001 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
2002 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
2003 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
2004 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
2005 (void*)funcp, SvPV_nolen(datasv)));
2006 av_unshift(PL_rsfp_filters, 1);
2007 av_store(PL_rsfp_filters, 0, datasv) ;
2012 /* Delete most recently added instance of this filter function. */
2014 Perl_filter_del(pTHX_ filter_t funcp)
2017 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", (void*)funcp));
2018 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
2020 /* if filter is on top of stack (usual case) just pop it off */
2021 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
2022 if (IoANY(datasv) == (void *)funcp) {
2023 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
2024 IoANY(datasv) = (void *)NULL;
2025 sv_free(av_pop(PL_rsfp_filters));
2029 /* we need to search for the correct entry and clear it */
2030 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
2034 /* Invoke the n'th filter function for the current rsfp. */
2036 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
2039 /* 0 = read one text line */
2044 if (!PL_rsfp_filters)
2046 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
2047 /* Provide a default input filter to make life easy. */
2048 /* Note that we append to the line. This is handy. */
2049 DEBUG_P(PerlIO_printf(Perl_debug_log,
2050 "filter_read %d: from rsfp\n", idx));
2054 int old_len = SvCUR(buf_sv) ;
2056 /* ensure buf_sv is large enough */
2057 SvGROW(buf_sv, (STRLEN)(old_len + maxlen)) ;
2058 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
2059 if (PerlIO_error(PL_rsfp))
2060 return -1; /* error */
2062 return 0 ; /* end of file */
2064 SvCUR_set(buf_sv, old_len + len) ;
2067 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2068 if (PerlIO_error(PL_rsfp))
2069 return -1; /* error */
2071 return 0 ; /* end of file */
2074 return SvCUR(buf_sv);
2076 /* Skip this filter slot if filter has been deleted */
2077 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
2078 DEBUG_P(PerlIO_printf(Perl_debug_log,
2079 "filter_read %d: skipped (filter deleted)\n",
2081 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
2083 /* Get function pointer hidden within datasv */
2084 funcp = (filter_t)IoANY(datasv);
2085 DEBUG_P(PerlIO_printf(Perl_debug_log,
2086 "filter_read %d: via function %p (%s)\n",
2087 idx, (void*)funcp, SvPV_nolen(datasv)));
2088 /* Call function. The function is expected to */
2089 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2090 /* Return: <0:error, =0:eof, >0:not eof */
2091 return (*funcp)(aTHX_ idx, buf_sv, maxlen);
2095 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2097 #ifdef PERL_CR_FILTER
2098 if (!PL_rsfp_filters) {
2099 filter_add(S_cr_textfilter,NULL);
2102 if (PL_rsfp_filters) {
2105 SvCUR_set(sv, 0); /* start with empty line */
2106 if (FILTER_READ(0, sv, 0) > 0)
2107 return ( SvPVX(sv) ) ;
2112 return (sv_gets(sv, fp, append));
2116 S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2120 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2124 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2125 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2127 return GvHV(gv); /* Foo:: */
2130 /* use constant CLASS => 'MyClass' */
2131 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2133 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2134 pkgname = SvPV_nolen(sv);
2138 return gv_stashpv(pkgname, FALSE);
2142 static char* exp_name[] =
2143 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2144 "ATTRTERM", "TERMBLOCK", "TERMORDORDOR"
2151 Works out what to call the token just pulled out of the input
2152 stream. The yacc parser takes care of taking the ops we return and
2153 stitching them into a tree.
2159 if read an identifier
2160 if we're in a my declaration
2161 croak if they tried to say my($foo::bar)
2162 build the ops for a my() declaration
2163 if it's an access to a my() variable
2164 are we in a sort block?
2165 croak if my($a); $a <=> $b
2166 build ops for access to a my() variable
2167 if in a dq string, and they've said @foo and we can't find @foo
2169 build ops for a bareword
2170 if we already built the token before, use it.
2173 #ifdef USE_PURE_BISON
2175 Perl_yylex_r(pTHX_ YYSTYPE *lvalp, int *lcharp)
2180 yylval_pointer[yyactlevel] = lvalp;
2181 yychar_pointer[yyactlevel] = lcharp;
2182 if (yyactlevel >= YYMAXLEVEL)
2183 Perl_croak(aTHX_ "panic: YYMAXLEVEL");
2185 r = Perl_yylex(aTHX);
2195 #pragma segment Perl_yylex
2207 I32 orig_keyword = 0;
2209 /* check if there's an identifier for us to look at */
2210 if (PL_pending_ident)
2211 return S_pending_ident(aTHX);
2213 /* no identifier pending identification */
2215 switch (PL_lex_state) {
2217 case LEX_NORMAL: /* Some compilers will produce faster */
2218 case LEX_INTERPNORMAL: /* code if we comment these out. */
2222 /* when we've already built the next token, just pull it out of the queue */
2225 yylval = PL_nextval[PL_nexttoke];
2227 PL_lex_state = PL_lex_defer;
2228 PL_expect = PL_lex_expect;
2229 PL_lex_defer = LEX_NORMAL;
2231 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2232 "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr,
2233 (IV)PL_nexttype[PL_nexttoke]); });
2235 return(PL_nexttype[PL_nexttoke]);
2237 /* interpolated case modifiers like \L \U, including \Q and \E.
2238 when we get here, PL_bufptr is at the \
2240 case LEX_INTERPCASEMOD:
2242 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2243 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2245 /* handle \E or end of string */
2246 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2250 if (PL_lex_casemods) {
2251 oldmod = PL_lex_casestack[--PL_lex_casemods];
2252 PL_lex_casestack[PL_lex_casemods] = '\0';
2254 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2256 PL_lex_state = LEX_INTERPCONCAT;
2260 if (PL_bufptr != PL_bufend)
2262 PL_lex_state = LEX_INTERPCONCAT;
2266 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2267 "### Saw case modifier at '%s'\n", PL_bufptr); });
2269 if (s[1] == '\\' && s[2] == 'E') {
2271 PL_lex_state = LEX_INTERPCONCAT;
2275 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2276 tmp = *s, *s = s[2], s[2] = (char)tmp; /* misordered... */
2277 if (strchr("LU", *s) &&
2278 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U'))) {
2279 PL_lex_casestack[--PL_lex_casemods] = '\0';
2282 if (PL_lex_casemods > 10)
2283 Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2284 PL_lex_casestack[PL_lex_casemods++] = *s;
2285 PL_lex_casestack[PL_lex_casemods] = '\0';
2286 PL_lex_state = LEX_INTERPCONCAT;
2287 PL_nextval[PL_nexttoke].ival = 0;
2290 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2292 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2294 PL_nextval[PL_nexttoke].ival = OP_LC;
2296 PL_nextval[PL_nexttoke].ival = OP_UC;
2298 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2300 Perl_croak(aTHX_ "panic: yylex");
2304 if (PL_lex_starts) {
2313 case LEX_INTERPPUSH:
2314 return sublex_push();
2316 case LEX_INTERPSTART:
2317 if (PL_bufptr == PL_bufend)
2318 return sublex_done();
2319 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2320 "### Interpolated variable at '%s'\n", PL_bufptr); });
2322 PL_lex_dojoin = (*PL_bufptr == '@');
2323 PL_lex_state = LEX_INTERPNORMAL;
2324 if (PL_lex_dojoin) {
2325 PL_nextval[PL_nexttoke].ival = 0;
2327 force_ident("\"", '$');
2328 PL_nextval[PL_nexttoke].ival = 0;
2330 PL_nextval[PL_nexttoke].ival = 0;
2332 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2335 if (PL_lex_starts++) {
2341 case LEX_INTERPENDMAYBE:
2342 if (intuit_more(PL_bufptr)) {
2343 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2349 if (PL_lex_dojoin) {
2350 PL_lex_dojoin = FALSE;
2351 PL_lex_state = LEX_INTERPCONCAT;
2354 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2355 && SvEVALED(PL_lex_repl))
2357 if (PL_bufptr != PL_bufend)
2358 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2359 PL_lex_repl = Nullsv;
2362 case LEX_INTERPCONCAT:
2364 if (PL_lex_brackets)
2365 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2367 if (PL_bufptr == PL_bufend)
2368 return sublex_done();
2370 if (SvIVX(PL_linestr) == '\'') {
2371 SV *sv = newSVsv(PL_linestr);
2374 else if ( PL_hints & HINT_NEW_RE )
2375 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2376 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2380 s = scan_const(PL_bufptr);
2382 PL_lex_state = LEX_INTERPCASEMOD;
2384 PL_lex_state = LEX_INTERPSTART;
2387 if (s != PL_bufptr) {
2388 PL_nextval[PL_nexttoke] = yylval;
2391 if (PL_lex_starts++)
2401 PL_lex_state = LEX_NORMAL;
2402 s = scan_formline(PL_bufptr);
2403 if (!PL_lex_formbrack)
2409 PL_oldoldbufptr = PL_oldbufptr;
2412 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2413 exp_name[PL_expect], s);
2419 if (isIDFIRST_lazy_if(s,UTF))
2421 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2424 goto fake_eof; /* emulate EOF on ^D or ^Z */
2429 if (PL_lex_brackets)
2430 yyerror("Missing right curly or square bracket");
2431 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2432 "### Tokener got EOF\n");
2436 if (s++ < PL_bufend)
2437 goto retry; /* ignore stray nulls */
2440 if (!PL_in_eval && !PL_preambled) {
2441 PL_preambled = TRUE;
2442 sv_setpv(PL_linestr,incl_perldb());
2443 if (SvCUR(PL_linestr))
2444 sv_catpv(PL_linestr,";");
2446 while(AvFILLp(PL_preambleav) >= 0) {
2447 SV *tmpsv = av_shift(PL_preambleav);
2448 sv_catsv(PL_linestr, tmpsv);
2449 sv_catpv(PL_linestr, ";");
2452 sv_free((SV*)PL_preambleav);
2453 PL_preambleav = NULL;
2455 if (PL_minus_n || PL_minus_p) {
2456 sv_catpv(PL_linestr, "LINE: while (<>) {");
2458 sv_catpv(PL_linestr,"chomp;");
2461 if (strchr("/'\"", *PL_splitstr)
2462 && strchr(PL_splitstr + 1, *PL_splitstr))
2463 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
2466 s = "'~#\200\1'"; /* surely one char is unused...*/
2467 while (s[1] && strchr(PL_splitstr, *s)) s++;
2469 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s%c",
2470 "q" + (delim == '\''), delim);
2471 for (s = PL_splitstr; *s; s++) {
2473 sv_catpvn(PL_linestr, "\\", 1);
2474 sv_catpvn(PL_linestr, s, 1);
2476 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2480 sv_catpv(PL_linestr,"our @F=split(' ');");
2483 sv_catpv(PL_linestr, "\n");
2484 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2485 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2486 PL_last_lop = PL_last_uni = Nullch;
2487 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2488 SV *sv = NEWSV(85,0);
2490 sv_upgrade(sv, SVt_PVMG);
2491 sv_setsv(sv,PL_linestr);
2494 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2499 bof = PL_rsfp ? TRUE : FALSE;
2500 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2503 if (PL_preprocess && !PL_in_eval)
2504 (void)PerlProc_pclose(PL_rsfp);
2505 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2506 PerlIO_clearerr(PL_rsfp);
2508 (void)PerlIO_close(PL_rsfp);
2510 PL_doextract = FALSE;
2512 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2513 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2514 sv_catpv(PL_linestr,";}");
2515 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2516 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2517 PL_last_lop = PL_last_uni = Nullch;
2518 PL_minus_n = PL_minus_p = 0;
2521 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2522 PL_last_lop = PL_last_uni = Nullch;
2523 sv_setpv(PL_linestr,"");
2524 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2526 /* if it looks like the start of a BOM, check if it in fact is */
2527 else if (bof && (!*s || *(U8*)s == 0xEF || *(U8*)s >= 0xFE)) {
2528 #ifdef PERLIO_IS_STDIO
2529 # ifdef __GNU_LIBRARY__
2530 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
2531 # define FTELL_FOR_PIPE_IS_BROKEN
2535 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2536 # define FTELL_FOR_PIPE_IS_BROKEN
2541 #ifdef FTELL_FOR_PIPE_IS_BROKEN
2542 /* This loses the possibility to detect the bof
2543 * situation on perl -P when the libc5 is being used.
2544 * Workaround? Maybe attach some extra state to PL_rsfp?
2547 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2549 bof = PerlIO_tell(PL_rsfp) == (Off_t)SvCUR(PL_linestr);
2552 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2553 s = swallow_bom((U8*)s);
2557 /* Incest with pod. */
2558 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2559 sv_setpv(PL_linestr, "");
2560 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2561 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2562 PL_last_lop = PL_last_uni = Nullch;
2563 PL_doextract = FALSE;
2567 } while (PL_doextract);
2568 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2569 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2570 SV *sv = NEWSV(85,0);
2572 sv_upgrade(sv, SVt_PVMG);
2573 sv_setsv(sv,PL_linestr);
2576 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2578 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2579 PL_last_lop = PL_last_uni = Nullch;
2580 if (CopLINE(PL_curcop) == 1) {
2581 while (s < PL_bufend && isSPACE(*s))
2583 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2587 if (*s == '#' && *(s+1) == '!')
2589 #ifdef ALTERNATE_SHEBANG
2591 static char as[] = ALTERNATE_SHEBANG;
2592 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2593 d = s + (sizeof(as) - 1);
2595 #endif /* ALTERNATE_SHEBANG */
2604 while (*d && !isSPACE(*d))
2608 #ifdef ARG_ZERO_IS_SCRIPT
2609 if (ipathend > ipath) {
2611 * HP-UX (at least) sets argv[0] to the script name,
2612 * which makes $^X incorrect. And Digital UNIX and Linux,
2613 * at least, set argv[0] to the basename of the Perl
2614 * interpreter. So, having found "#!", we'll set it right.
2616 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV)); /* $^X */
2617 assert(SvPOK(x) || SvGMAGICAL(x));
2618 if (sv_eq(x, CopFILESV(PL_curcop))) {
2619 sv_setpvn(x, ipath, ipathend - ipath);
2625 char *bstart = SvPV(CopFILESV(PL_curcop),blen);
2626 char *lstart = SvPV(x,llen);
2628 bstart += blen - llen;
2629 if (strnEQ(bstart, lstart, llen) && bstart[-1] == '/') {
2630 sv_setpvn(x, ipath, ipathend - ipath);
2635 TAINT_NOT; /* $^X is always tainted, but that's OK */
2637 #endif /* ARG_ZERO_IS_SCRIPT */
2642 d = instr(s,"perl -");
2644 d = instr(s,"perl");
2646 /* avoid getting into infinite loops when shebang
2647 * line contains "Perl" rather than "perl" */
2649 for (d = ipathend-4; d >= ipath; --d) {
2650 if ((*d == 'p' || *d == 'P')
2651 && !ibcmp(d, "perl", 4))
2661 #ifdef ALTERNATE_SHEBANG
2663 * If the ALTERNATE_SHEBANG on this system starts with a
2664 * character that can be part of a Perl expression, then if
2665 * we see it but not "perl", we're probably looking at the
2666 * start of Perl code, not a request to hand off to some
2667 * other interpreter. Similarly, if "perl" is there, but
2668 * not in the first 'word' of the line, we assume the line
2669 * contains the start of the Perl program.
2671 if (d && *s != '#') {
2673 while (*c && !strchr("; \t\r\n\f\v#", *c))
2676 d = Nullch; /* "perl" not in first word; ignore */
2678 *s = '#'; /* Don't try to parse shebang line */
2680 #endif /* ALTERNATE_SHEBANG */
2681 #ifndef MACOS_TRADITIONAL
2686 !instr(s,"indir") &&
2687 instr(PL_origargv[0],"perl"))
2693 while (s < PL_bufend && isSPACE(*s))
2695 if (s < PL_bufend) {
2696 Newz(899,newargv,PL_origargc+3,char*);
2698 while (s < PL_bufend && !isSPACE(*s))
2701 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2704 newargv = PL_origargv;
2706 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
2707 Perl_croak(aTHX_ "Can't exec %s", ipath);
2711 U32 oldpdb = PL_perldb;
2712 bool oldn = PL_minus_n;
2713 bool oldp = PL_minus_p;
2715 while (*d && !isSPACE(*d)) d++;
2716 while (SPACE_OR_TAB(*d)) d++;
2719 bool switches_done = PL_doswitches;
2721 if (*d == 'M' || *d == 'm') {
2723 while (*d && !isSPACE(*d)) d++;
2724 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2727 d = moreswitches(d);
2729 if ((PERLDB_LINE && !oldpdb) ||
2730 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2731 /* if we have already added "LINE: while (<>) {",
2732 we must not do it again */
2734 sv_setpv(PL_linestr, "");
2735 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2736 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2737 PL_last_lop = PL_last_uni = Nullch;
2738 PL_preambled = FALSE;
2740 (void)gv_fetchfile(PL_origfilename);
2743 if (PL_doswitches && !switches_done) {
2744 int argc = PL_origargc;
2745 char **argv = PL_origargv;
2748 } while (argc && argv[0][0] == '-' && argv[0][1]);
2749 init_argv_symbols(argc,argv);
2755 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2757 PL_lex_state = LEX_FORMLINE;
2762 #ifdef PERL_STRICT_CR
2763 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2765 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2767 case ' ': case '\t': case '\f': case 013:
2768 #ifdef MACOS_TRADITIONAL
2775 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2776 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2777 /* handle eval qq[#line 1 "foo"\n ...] */
2778 CopLINE_dec(PL_curcop);
2782 while (s < d && *s != '\n')
2786 else if (s > d) /* Found by Ilya: feed random input to Perl. */
2787 Perl_croak(aTHX_ "panic: input overflow");
2789 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2791 PL_lex_state = LEX_FORMLINE;
2801 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2808 while (s < PL_bufend && SPACE_OR_TAB(*s))
2811 if (strnEQ(s,"=>",2)) {
2812 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2813 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2814 "### Saw unary minus before =>, forcing word '%s'\n", s);
2816 OPERATOR('-'); /* unary minus */
2818 PL_last_uni = PL_oldbufptr;
2820 case 'r': ftst = OP_FTEREAD; break;
2821 case 'w': ftst = OP_FTEWRITE; break;
2822 case 'x': ftst = OP_FTEEXEC; break;
2823 case 'o': ftst = OP_FTEOWNED; break;
2824 case 'R': ftst = OP_FTRREAD; break;
2825 case 'W': ftst = OP_FTRWRITE; break;
2826 case 'X': ftst = OP_FTREXEC; break;
2827 case 'O': ftst = OP_FTROWNED; break;
2828 case 'e': ftst = OP_FTIS; break;
2829 case 'z': ftst = OP_FTZERO; break;
2830 case 's': ftst = OP_FTSIZE; break;
2831 case 'f': ftst = OP_FTFILE; break;
2832 case 'd': ftst = OP_FTDIR; break;
2833 case 'l': ftst = OP_FTLINK; break;
2834 case 'p': ftst = OP_FTPIPE; break;
2835 case 'S': ftst = OP_FTSOCK; break;
2836 case 'u': ftst = OP_FTSUID; break;
2837 case 'g': ftst = OP_FTSGID; break;
2838 case 'k': ftst = OP_FTSVTX; break;
2839 case 'b': ftst = OP_FTBLK; break;
2840 case 'c': ftst = OP_FTCHR; break;
2841 case 't': ftst = OP_FTTTY; break;
2842 case 'T': ftst = OP_FTTEXT; break;
2843 case 'B': ftst = OP_FTBINARY; break;
2844 case 'M': case 'A': case 'C':
2845 gv_fetchpv("\024",TRUE, SVt_PV);
2847 case 'M': ftst = OP_FTMTIME; break;
2848 case 'A': ftst = OP_FTATIME; break;
2849 case 'C': ftst = OP_FTCTIME; break;
2857 PL_last_lop_op = (OPCODE)ftst;
2858 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2859 "### Saw file test %c\n", (int)ftst);
2864 /* Assume it was a minus followed by a one-letter named
2865 * subroutine call (or a -bareword), then. */
2866 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2867 "### %c looked like a file test but was not\n",
2876 if (PL_expect == XOPERATOR)
2881 else if (*s == '>') {
2884 if (isIDFIRST_lazy_if(s,UTF)) {
2885 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2893 if (PL_expect == XOPERATOR)
2896 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2898 OPERATOR('-'); /* unary minus */
2905 if (PL_expect == XOPERATOR)
2910 if (PL_expect == XOPERATOR)
2913 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2919 if (PL_expect != XOPERATOR) {
2920 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2921 PL_expect = XOPERATOR;
2922 force_ident(PL_tokenbuf, '*');
2935 if (PL_expect == XOPERATOR) {
2939 PL_tokenbuf[0] = '%';
2940 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2941 if (!PL_tokenbuf[1]) {
2943 yyerror("Final % should be \\% or %name");
2946 PL_pending_ident = '%';
2965 switch (PL_expect) {
2968 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2970 PL_bufptr = s; /* update in case we back off */
2976 PL_expect = XTERMBLOCK;
2980 while (isIDFIRST_lazy_if(s,UTF)) {
2981 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2982 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2983 if (tmp < 0) tmp = -tmp;
2999 d = scan_str(d,TRUE,TRUE);
3001 /* MUST advance bufptr here to avoid bogus
3002 "at end of line" context messages from yyerror().
3004 PL_bufptr = s + len;
3005 yyerror("Unterminated attribute parameter in attribute list");
3008 return 0; /* EOF indicator */
3012 SV *sv = newSVpvn(s, len);
3013 sv_catsv(sv, PL_lex_stuff);
3014 attrs = append_elem(OP_LIST, attrs,
3015 newSVOP(OP_CONST, 0, sv));
3016 SvREFCNT_dec(PL_lex_stuff);
3017 PL_lex_stuff = Nullsv;
3020 /* NOTE: any CV attrs applied here need to be part of
3021 the CVf_BUILTIN_ATTRS define in cv.h! */
3022 if (!PL_in_my && len == 6 && strnEQ(s, "lvalue", len))
3023 CvLVALUE_on(PL_compcv);
3024 else if (!PL_in_my && len == 6 && strnEQ(s, "locked", len))
3025 CvLOCKED_on(PL_compcv);
3026 else if (!PL_in_my && len == 6 && strnEQ(s, "method", len))
3027 CvMETHOD_on(PL_compcv);
3028 else if (!PL_in_my && len == 9 && strnEQ(s, "assertion", len))
3029 CvASSERTION_on(PL_compcv);
3031 else if (PL_in_my == KEY_our && len == 6 &&
3032 strnEQ(s, "unique", len))
3033 GvUNIQUE_on(cGVOPx_gv(yylval.opval));
3035 /* After we've set the flags, it could be argued that
3036 we don't need to do the attributes.pm-based setting
3037 process, and shouldn't bother appending recognized
3038 flags. To experiment with that, uncomment the
3039 following "else". (Note that's already been
3040 uncommented. That keeps the above-applied built-in
3041 attributes from being intercepted (and possibly
3042 rejected) by a package's attribute routines, but is
3043 justified by the performance win for the common case
3044 of applying only built-in attributes.) */
3046 attrs = append_elem(OP_LIST, attrs,
3047 newSVOP(OP_CONST, 0,
3051 if (*s == ':' && s[1] != ':')
3054 break; /* require real whitespace or :'s */
3056 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3057 if (*s != ';' && *s != '}' && *s != tmp && (tmp != '=' || *s != ')')) {
3058 char q = ((*s == '\'') ? '"' : '\'');
3059 /* If here for an expression, and parsed no attrs, back off. */
3060 if (tmp == '=' && !attrs) {
3064 /* MUST advance bufptr here to avoid bogus "at end of line"
3065 context messages from yyerror().
3069 yyerror("Unterminated attribute list");
3071 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
3079 PL_nextval[PL_nexttoke].opval = attrs;
3087 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3088 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3105 if (PL_lex_brackets <= 0)
3106 yyerror("Unmatched right square bracket");
3109 if (PL_lex_state == LEX_INTERPNORMAL) {
3110 if (PL_lex_brackets == 0) {
3111 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3112 PL_lex_state = LEX_INTERPEND;
3119 if (PL_lex_brackets > 100) {
3120 Renew(PL_lex_brackstack, PL_lex_brackets + 10, char);
3122 switch (PL_expect) {
3124 if (PL_lex_formbrack) {
3128 if (PL_oldoldbufptr == PL_last_lop)
3129 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3131 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3132 OPERATOR(HASHBRACK);
3134 while (s < PL_bufend && SPACE_OR_TAB(*s))
3137 PL_tokenbuf[0] = '\0';
3138 if (d < PL_bufend && *d == '-') {
3139 PL_tokenbuf[0] = '-';
3141 while (d < PL_bufend && SPACE_OR_TAB(*d))
3144 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3145 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3147 while (d < PL_bufend && SPACE_OR_TAB(*d))
3150 char minus = (PL_tokenbuf[0] == '-');
3151 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3159 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3164 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3169 if (PL_oldoldbufptr == PL_last_lop)
3170 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3172 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3175 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
3177 /* This hack is to get the ${} in the message. */
3179 yyerror("syntax error");
3182 OPERATOR(HASHBRACK);
3184 /* This hack serves to disambiguate a pair of curlies
3185 * as being a block or an anon hash. Normally, expectation
3186 * determines that, but in cases where we're not in a
3187 * position to expect anything in particular (like inside
3188 * eval"") we have to resolve the ambiguity. This code
3189 * covers the case where the first term in the curlies is a
3190 * quoted string. Most other cases need to be explicitly
3191 * disambiguated by prepending a `+' before the opening
3192 * curly in order to force resolution as an anon hash.
3194 * XXX should probably propagate the outer expectation
3195 * into eval"" to rely less on this hack, but that could
3196 * potentially break current behavior of eval"".
3200 if (*s == '\'' || *s == '"' || *s == '`') {
3201 /* common case: get past first string, handling escapes */
3202 for (t++; t < PL_bufend && *t != *s;)
3203 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3207 else if (*s == 'q') {
3210 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3214 char open, close, term;
3217 while (t < PL_bufend && isSPACE(*t))
3221 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3225 for (t++; t < PL_bufend; t++) {
3226 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3228 else if (*t == open)
3232 for (t++; t < PL_bufend; t++) {
3233 if (*t == '\\' && t+1 < PL_bufend)
3235 else if (*t == close && --brackets <= 0)
3237 else if (*t == open)
3243 else if (isALNUM_lazy_if(t,UTF)) {
3245 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3248 while (t < PL_bufend && isSPACE(*t))
3250 /* if comma follows first term, call it an anon hash */
3251 /* XXX it could be a comma expression with loop modifiers */
3252 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3253 || (*t == '=' && t[1] == '>')))
3254 OPERATOR(HASHBRACK);
3255 if (PL_expect == XREF)
3258 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3264 yylval.ival = CopLINE(PL_curcop);
3265 if (isSPACE(*s) || *s == '#')
3266 PL_copline = NOLINE; /* invalidate current command line number */
3271 if (PL_lex_brackets <= 0)
3272 yyerror("Unmatched right curly bracket");
3274 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3275 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3276 PL_lex_formbrack = 0;
3277 if (PL_lex_state == LEX_INTERPNORMAL) {
3278 if (PL_lex_brackets == 0) {
3279 if (PL_expect & XFAKEBRACK) {
3280 PL_expect &= XENUMMASK;
3281 PL_lex_state = LEX_INTERPEND;
3283 return yylex(); /* ignore fake brackets */
3285 if (*s == '-' && s[1] == '>')
3286 PL_lex_state = LEX_INTERPENDMAYBE;
3287 else if (*s != '[' && *s != '{')
3288 PL_lex_state = LEX_INTERPEND;
3291 if (PL_expect & XFAKEBRACK) {
3292 PL_expect &= XENUMMASK;
3294 return yylex(); /* ignore fake brackets */
3304 if (PL_expect == XOPERATOR) {
3305 if (ckWARN(WARN_SEMICOLON)
3306 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3308 CopLINE_dec(PL_curcop);
3309 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
3310 CopLINE_inc(PL_curcop);
3315 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3317 PL_expect = XOPERATOR;
3318 force_ident(PL_tokenbuf, '&');
3322 yylval.ival = (OPpENTERSUB_AMPER<<8);
3341 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3342 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Reversed %c= operator",(int)tmp);
3344 if (PL_expect == XSTATE && isALPHA(tmp) &&
3345 (s == PL_linestart+1 || s[-2] == '\n') )
3347 if (PL_in_eval && !PL_rsfp) {
3352 if (strnEQ(s,"=cut",4)) {
3366 PL_doextract = TRUE;
3369 if (PL_lex_brackets < PL_lex_formbrack) {
3371 #ifdef PERL_STRICT_CR
3372 for (t = s; SPACE_OR_TAB(*t); t++) ;
3374 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3376 if (*t == '\n' || *t == '#') {
3394 if (PL_expect != XOPERATOR) {
3395 if (s[1] != '<' && !strchr(s,'>'))
3398 s = scan_heredoc(s);
3400 s = scan_inputsymbol(s);
3401 TERM(sublex_start());
3406 SHop(OP_LEFT_SHIFT);
3420 SHop(OP_RIGHT_SHIFT);
3429 if (PL_expect == XOPERATOR) {
3430 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3433 return ','; /* grandfather non-comma-format format */
3437 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3438 PL_tokenbuf[0] = '@';
3439 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3440 sizeof PL_tokenbuf - 1, FALSE);
3441 if (PL_expect == XOPERATOR)
3442 no_op("Array length", s);
3443 if (!PL_tokenbuf[1])
3445 PL_expect = XOPERATOR;
3446 PL_pending_ident = '#';
3450 PL_tokenbuf[0] = '$';
3451 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3452 sizeof PL_tokenbuf - 1, FALSE);
3453 if (PL_expect == XOPERATOR)
3455 if (!PL_tokenbuf[1]) {
3457 yyerror("Final $ should be \\$ or $name");
3461 /* This kludge not intended to be bulletproof. */
3462 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3463 yylval.opval = newSVOP(OP_CONST, 0,
3464 newSViv(PL_compiling.cop_arybase));
3465 yylval.opval->op_private = OPpCONST_ARYBASE;
3471 if (PL_lex_state == LEX_NORMAL)
3474 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3477 PL_tokenbuf[0] = '@';
3478 if (ckWARN(WARN_SYNTAX)) {
3480 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3483 PL_bufptr = skipspace(PL_bufptr);
3484 while (t < PL_bufend && *t != ']')
3486 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3487 "Multidimensional syntax %.*s not supported",
3488 (t - PL_bufptr) + 1, PL_bufptr);
3492 else if (*s == '{') {
3493 PL_tokenbuf[0] = '%';
3494 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3495 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3497 char tmpbuf[sizeof PL_tokenbuf];
3499 for (t++; isSPACE(*t); t++) ;
3500 if (isIDFIRST_lazy_if(t,UTF)) {
3501 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3502 for (; isSPACE(*t); t++) ;
3503 if (*t == ';' && get_cv(tmpbuf, FALSE))
3504 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3505 "You need to quote \"%s\"", tmpbuf);
3511 PL_expect = XOPERATOR;
3512 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3513 bool islop = (PL_last_lop == PL_oldoldbufptr);
3514 if (!islop || PL_last_lop_op == OP_GREPSTART)
3515 PL_expect = XOPERATOR;
3516 else if (strchr("$@\"'`q", *s))
3517 PL_expect = XTERM; /* e.g. print $fh "foo" */
3518 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3519 PL_expect = XTERM; /* e.g. print $fh &sub */
3520 else if (isIDFIRST_lazy_if(s,UTF)) {
3521 char tmpbuf[sizeof PL_tokenbuf];
3522 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3523 if ((tmp = keyword(tmpbuf, len))) {
3524 /* binary operators exclude handle interpretations */
3536 PL_expect = XTERM; /* e.g. print $fh length() */
3541 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3542 if (gv && GvCVu(gv))
3543 PL_expect = XTERM; /* e.g. print $fh subr() */
3546 else if (isDIGIT(*s))
3547 PL_expect = XTERM; /* e.g. print $fh 3 */
3548 else if (*s == '.' && isDIGIT(s[1]))
3549 PL_expect = XTERM; /* e.g. print $fh .3 */
3550 else if (strchr("?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3551 PL_expect = XTERM; /* e.g. print $fh -1 */
3552 else if (*s == '/' && !isSPACE(s[1]) && s[1] != '=' && s[1] != '/')
3553 PL_expect = XTERM; /* e.g. print $fh /.../
3554 XXX except DORDOR operator */
3555 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3556 PL_expect = XTERM; /* print $fh <<"EOF" */
3558 PL_pending_ident = '$';
3562 if (PL_expect == XOPERATOR)
3564 PL_tokenbuf[0] = '@';
3565 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3566 if (!PL_tokenbuf[1]) {
3568 yyerror("Final @ should be \\@ or @name");
3571 if (PL_lex_state == LEX_NORMAL)
3573 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3575 PL_tokenbuf[0] = '%';
3577 /* Warn about @ where they meant $. */
3578 if (ckWARN(WARN_SYNTAX)) {
3579 if (*s == '[' || *s == '{') {
3581 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3583 if (*t == '}' || *t == ']') {
3585 PL_bufptr = skipspace(PL_bufptr);
3586 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3587 "Scalar value %.*s better written as $%.*s",
3588 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3593 PL_pending_ident = '@';
3596 case '/': /* may be division, defined-or, or pattern */
3597 if (PL_expect == XTERMORDORDOR && s[1] == '/') {
3601 case '?': /* may either be conditional or pattern */
3602 if(PL_expect == XOPERATOR) {
3610 /* A // operator. */
3620 /* Disable warning on "study /blah/" */
3621 if (PL_oldoldbufptr == PL_last_uni
3622 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3623 || memNE(PL_last_uni, "study", 5)
3624 || isALNUM_lazy_if(PL_last_uni+5,UTF)
3627 s = scan_pat(s,OP_MATCH);
3628 TERM(sublex_start());
3632 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3633 #ifdef PERL_STRICT_CR
3636 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3638 && (s == PL_linestart || s[-1] == '\n') )
3640 PL_lex_formbrack = 0;
3644 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3650 yylval.ival = OPf_SPECIAL;
3656 if (PL_expect != XOPERATOR)
3661 case '0': case '1': case '2': case '3': case '4':
3662 case '5': case '6': case '7': case '8': case '9':
3663 s = scan_num(s, &yylval);
3664 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3665 "### Saw number in '%s'\n", s);
3667 if (PL_expect == XOPERATOR)
3672 s = scan_str(s,FALSE,FALSE);
3673 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3674 "### Saw string before '%s'\n", s);
3676 if (PL_expect == XOPERATOR) {
3677 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3680 return ','; /* grandfather non-comma-format format */
3686 missingterm((char*)0);
3687 yylval.ival = OP_CONST;
3688 TERM(sublex_start());
3691 s = scan_str(s,FALSE,FALSE);
3692 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3693 "### Saw string before '%s'\n", s);
3695 if (PL_expect == XOPERATOR) {
3696 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3699 return ','; /* grandfather non-comma-format format */
3705 missingterm((char*)0);
3706 yylval.ival = OP_CONST;
3707 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3708 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
3709 yylval.ival = OP_STRINGIFY;
3713 TERM(sublex_start());
3716 s = scan_str(s,FALSE,FALSE);
3717 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3718 "### Saw backtick string before '%s'\n", s);
3720 if (PL_expect == XOPERATOR)
3721 no_op("Backticks",s);
3723 missingterm((char*)0);
3724 yylval.ival = OP_BACKTICK;
3726 TERM(sublex_start());
3730 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3731 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),"Can't use \\%c to mean $%c in expression",
3733 if (PL_expect == XOPERATOR)
3734 no_op("Backslash",s);
3738 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3742 while (isDIGIT(*start) || *start == '_')
3744 if (*start == '.' && isDIGIT(start[1])) {
3745 s = scan_num(s, &yylval);
3748 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3749 else if (!isALPHA(*start) && (PL_expect == XTERM
3750 || PL_expect == XREF || PL_expect == XSTATE
3751 || PL_expect == XTERMORDORDOR)) {
3755 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3758 s = scan_num(s, &yylval);
3765 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3805 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3807 /* Some keywords can be followed by any delimiter, including ':' */
3808 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3809 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3810 (PL_tokenbuf[0] == 'q' &&
3811 strchr("qwxr", PL_tokenbuf[1])))));
3813 /* x::* is just a word, unless x is "CORE" */
3814 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3818 while (d < PL_bufend && isSPACE(*d))
3819 d++; /* no comments skipped here, or s### is misparsed */
3821 /* Is this a label? */
3822 if (!tmp && PL_expect == XSTATE
3823 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3825 yylval.pval = savepv(PL_tokenbuf);
3830 /* Check for keywords */
3831 tmp = keyword(PL_tokenbuf, len);
3833 /* Is this a word before a => operator? */
3834 if (*d == '=' && d[1] == '>') {
3836 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3837 yylval.opval->op_private = OPpCONST_BARE;
3838 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
3839 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3843 if (tmp < 0) { /* second-class keyword? */
3844 GV *ogv = Nullgv; /* override (winner) */
3845 GV *hgv = Nullgv; /* hidden (loser) */
3846 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3848 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3851 if (GvIMPORTED_CV(gv))
3853 else if (! CvMETHOD(cv))
3857 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3858 (gv = *gvp) != (GV*)&PL_sv_undef &&
3859 GvCVu(gv) && GvIMPORTED_CV(gv))
3866 tmp = 0; /* overridden by import or by GLOBAL */
3869 && -tmp==KEY_lock /* XXX generalizable kludge */
3871 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3873 tmp = 0; /* any sub overrides "weak" keyword */
3875 else { /* no override */
3877 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
3878 Perl_warner(aTHX_ packWARN(WARN_MISC),
3879 "dump() better written as CORE::dump()");
3883 if (ckWARN(WARN_AMBIGUOUS) && hgv
3884 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3885 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
3886 "Ambiguous call resolved as CORE::%s(), %s",
3887 GvENAME(hgv), "qualify as such or use &");
3894 default: /* not a keyword */
3898 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3900 /* Get the rest if it looks like a package qualifier */
3902 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3904 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3907 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3908 *s == '\'' ? "'" : "::");
3913 if (PL_expect == XOPERATOR) {
3914 if (PL_bufptr == PL_linestart) {
3915 CopLINE_dec(PL_curcop);
3916 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
3917 CopLINE_inc(PL_curcop);
3920 no_op("Bareword",s);
3923 /* Look for a subroutine with this name in current package,
3924 unless name is "Foo::", in which case Foo is a bearword
3925 (and a package name). */
3928 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3930 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3931 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
3932 "Bareword \"%s\" refers to nonexistent package",
3935 PL_tokenbuf[len] = '\0';
3942 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3945 /* if we saw a global override before, get the right name */
3948 sv = newSVpvn("CORE::GLOBAL::",14);
3949 sv_catpv(sv,PL_tokenbuf);
3952 sv = newSVpv(PL_tokenbuf,0);
3954 /* Presume this is going to be a bareword of some sort. */
3957 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3958 yylval.opval->op_private = OPpCONST_BARE;
3959 /* UTF-8 package name? */
3960 if (UTF && !IN_BYTES &&
3961 is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
3964 /* And if "Foo::", then that's what it certainly is. */
3969 /* See if it's the indirect object for a list operator. */
3971 if (PL_oldoldbufptr &&
3972 PL_oldoldbufptr < PL_bufptr &&
3973 (PL_oldoldbufptr == PL_last_lop
3974 || PL_oldoldbufptr == PL_last_uni) &&
3975 /* NO SKIPSPACE BEFORE HERE! */
3976 (PL_expect == XREF ||
3977 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3979 bool immediate_paren = *s == '(';
3981 /* (Now we can afford to cross potential line boundary.) */
3984 /* Two barewords in a row may indicate method call. */
3986 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3989 /* If not a declared subroutine, it's an indirect object. */
3990 /* (But it's an indir obj regardless for sort.) */
3992 if ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
3993 ((!gv || !GvCVu(gv)) &&
3994 (PL_last_lop_op != OP_MAPSTART &&
3995 PL_last_lop_op != OP_GREPSTART))))
3997 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
4002 PL_expect = XOPERATOR;
4005 /* Is this a word before a => operator? */
4006 if (*s == '=' && s[1] == '>' && !pkgname) {
4008 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
4009 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
4010 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
4014 /* If followed by a paren, it's certainly a subroutine. */
4017 if (gv && GvCVu(gv)) {
4018 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
4019 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
4024 PL_nextval[PL_nexttoke].opval = yylval.opval;
4025 PL_expect = XOPERATOR;
4031 /* If followed by var or block, call it a method (unless sub) */
4033 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
4034 PL_last_lop = PL_oldbufptr;
4035 PL_last_lop_op = OP_METHOD;
4039 /* If followed by a bareword, see if it looks like indir obj. */
4042 && (isIDFIRST_lazy_if(s,UTF) || *s == '$')
4043 && (tmp = intuit_method(s,gv)))
4046 /* Not a method, so call it a subroutine (if defined) */
4048 if (gv && GvCVu(gv)) {
4050 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
4051 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4052 "Ambiguous use of -%s resolved as -&%s()",
4053 PL_tokenbuf, PL_tokenbuf);
4054 /* Check for a constant sub */
4056 if ((sv = cv_const_sv(cv))) {
4058 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
4059 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
4060 yylval.opval->op_private = 0;
4064 /* Resolve to GV now. */
4065 op_free(yylval.opval);
4066 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
4067 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
4068 PL_last_lop = PL_oldbufptr;
4069 PL_last_lop_op = OP_ENTERSUB;
4070 /* Is there a prototype? */
4073 char *proto = SvPV((SV*)cv, len);
4076 if (strEQ(proto, "$"))
4078 if (*proto == '&' && *s == '{') {
4079 sv_setpv(PL_subname, PL_curstash ?
4080 "__ANON__" : "__ANON__::__ANON__");
4084 PL_nextval[PL_nexttoke].opval = yylval.opval;
4090 /* Call it a bare word */
4092 if (PL_hints & HINT_STRICT_SUBS)
4093 yylval.opval->op_private |= OPpCONST_STRICT;
4096 if (ckWARN(WARN_RESERVED)) {
4097 if (lastchar != '-') {
4098 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4099 if (!*d && !gv_stashpv(PL_tokenbuf,FALSE))
4100 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
4107 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
4108 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4109 "Operator or semicolon missing before %c%s",
4110 lastchar, PL_tokenbuf);
4111 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4112 "Ambiguous use of %c resolved as operator %c",
4113 lastchar, lastchar);
4119 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4120 newSVpv(CopFILE(PL_curcop),0));
4124 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4125 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4128 case KEY___PACKAGE__:
4129 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4131 ? newSVsv(PL_curstname)
4140 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4141 char *pname = "main";
4142 if (PL_tokenbuf[2] == 'D')
4143 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
4144 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4147 GvIOp(gv) = newIO();
4148 IoIFP(GvIOp(gv)) = PL_rsfp;
4149 #if defined(HAS_FCNTL) && defined(F_SETFD)
4151 int fd = PerlIO_fileno(PL_rsfp);
4152 fcntl(fd,F_SETFD,fd >= 3);
4155 /* Mark this internal pseudo-handle as clean */
4156 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4158 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4159 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4160 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4162 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4163 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4164 /* if the script was opened in binmode, we need to revert
4165 * it to text mode for compatibility; but only iff it has CRs
4166 * XXX this is a questionable hack at best. */
4167 if (PL_bufend-PL_bufptr > 2
4168 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4171 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4172 loc = PerlIO_tell(PL_rsfp);
4173 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4176 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
4178 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4179 #endif /* NETWARE */
4180 #ifdef PERLIO_IS_STDIO /* really? */
4181 # if defined(__BORLANDC__)
4182 /* XXX see note in do_binmode() */
4183 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
4187 PerlIO_seek(PL_rsfp, loc, 0);
4191 #ifdef PERLIO_LAYERS
4194 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
4195 else if (PL_encoding) {
4202 XPUSHs(PL_encoding);
4204 call_method("name", G_SCALAR);
4208 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL,
4209 Perl_form(aTHX_ ":encoding(%"SVf")",
4227 if (PL_expect == XSTATE) {
4234 if (*s == ':' && s[1] == ':') {
4237 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4238 if (!(tmp = keyword(PL_tokenbuf, len)))
4239 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4253 LOP(OP_ACCEPT,XTERM);
4259 LOP(OP_ATAN2,XTERM);
4265 LOP(OP_BINMODE,XTERM);
4268 LOP(OP_BLESS,XTERM);
4277 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4294 if (!PL_cryptseen) {
4295 PL_cryptseen = TRUE;
4299 LOP(OP_CRYPT,XTERM);
4302 LOP(OP_CHMOD,XTERM);
4305 LOP(OP_CHOWN,XTERM);
4308 LOP(OP_CONNECT,XTERM);
4324 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4328 PL_hints |= HINT_BLOCK_SCOPE;
4338 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4339 LOP(OP_DBMOPEN,XTERM);
4345 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4352 yylval.ival = CopLINE(PL_curcop);
4366 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4367 UNIBRACK(OP_ENTEREVAL);
4385 case KEY_endhostent:
4391 case KEY_endservent:
4394 case KEY_endprotoent:
4405 yylval.ival = CopLINE(PL_curcop);
4407 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4409 if ((PL_bufend - p) >= 3 &&
4410 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4412 else if ((PL_bufend - p) >= 4 &&
4413 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4416 if (isIDFIRST_lazy_if(p,UTF)) {
4417 p = scan_ident(p, PL_bufend,
4418 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4422 Perl_croak(aTHX_ "Missing $ on loop variable");
4427 LOP(OP_FORMLINE,XTERM);
4433 LOP(OP_FCNTL,XTERM);
4439 LOP(OP_FLOCK,XTERM);
4448 LOP(OP_GREPSTART, XREF);
4451 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4466 case KEY_getpriority:
4467 LOP(OP_GETPRIORITY,XTERM);
4469 case KEY_getprotobyname:
4472 case KEY_getprotobynumber:
4473 LOP(OP_GPBYNUMBER,XTERM);
4475 case KEY_getprotoent:
4487 case KEY_getpeername:
4488 UNI(OP_GETPEERNAME);
4490 case KEY_gethostbyname:
4493 case KEY_gethostbyaddr:
4494 LOP(OP_GHBYADDR,XTERM);
4496 case KEY_gethostent:
4499 case KEY_getnetbyname:
4502 case KEY_getnetbyaddr:
4503 LOP(OP_GNBYADDR,XTERM);
4508 case KEY_getservbyname:
4509 LOP(OP_GSBYNAME,XTERM);
4511 case KEY_getservbyport:
4512 LOP(OP_GSBYPORT,XTERM);
4514 case KEY_getservent:
4517 case KEY_getsockname:
4518 UNI(OP_GETSOCKNAME);
4520 case KEY_getsockopt:
4521 LOP(OP_GSOCKOPT,XTERM);
4543 yylval.ival = CopLINE(PL_curcop);
4547 LOP(OP_INDEX,XTERM);
4553 LOP(OP_IOCTL,XTERM);
4565 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4597 LOP(OP_LISTEN,XTERM);
4606 s = scan_pat(s,OP_MATCH);
4607 TERM(sublex_start());
4610 LOP(OP_MAPSTART, XREF);
4613 LOP(OP_MKDIR,XTERM);
4616 LOP(OP_MSGCTL,XTERM);
4619 LOP(OP_MSGGET,XTERM);
4622 LOP(OP_MSGRCV,XTERM);
4625 LOP(OP_MSGSND,XTERM);
4631 if (isIDFIRST_lazy_if(s,UTF)) {
4632 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4633 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4635 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4636 if (!PL_in_my_stash) {
4639 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4647 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4654 if (PL_expect != XSTATE)
4655 yyerror("\"no\" not allowed in expression");
4656 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4657 s = force_version(s, FALSE);
4662 if (*s == '(' || (s = skipspace(s), *s == '('))
4669 if (isIDFIRST_lazy_if(s,UTF)) {
4671 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4673 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)
4675 && !(t[0] == '=' && t[1] == '>')
4677 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
4678 "Precedence problem: open %.*s should be open(%.*s)",
4679 d - s, s, d - s, s);
4685 yylval.ival = OP_OR;
4695 LOP(OP_OPEN_DIR,XTERM);
4698 checkcomma(s,PL_tokenbuf,"filehandle");
4702 checkcomma(s,PL_tokenbuf,"filehandle");
4721 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4725 LOP(OP_PIPE_OP,XTERM);
4728 s = scan_str(s,FALSE,FALSE);
4730 missingterm((char*)0);
4731 yylval.ival = OP_CONST;
4732 TERM(sublex_start());
4738 s = scan_str(s,FALSE,FALSE);
4740 missingterm((char*)0);
4742 if (SvCUR(PL_lex_stuff)) {
4745 d = SvPV_force(PL_lex_stuff, len);
4748 for (; isSPACE(*d) && len; --len, ++d) ;
4751 if (!warned && ckWARN(WARN_QW)) {
4752 for (; !isSPACE(*d) && len; --len, ++d) {
4754 Perl_warner(aTHX_ packWARN(WARN_QW),
4755 "Possible attempt to separate words with commas");
4758 else if (*d == '#') {
4759 Perl_warner(aTHX_ packWARN(WARN_QW),
4760 "Possible attempt to put comments in qw() list");
4766 for (; !isSPACE(*d) && len; --len, ++d) ;
4768 sv = newSVpvn(b, d-b);
4769 if (DO_UTF8(PL_lex_stuff))
4771 words = append_elem(OP_LIST, words,
4772 newSVOP(OP_CONST, 0, tokeq(sv)));
4776 PL_nextval[PL_nexttoke].opval = words;
4781 SvREFCNT_dec(PL_lex_stuff);
4782 PL_lex_stuff = Nullsv;
4788 s = scan_str(s,FALSE,FALSE);
4790 missingterm((char*)0);
4791 yylval.ival = OP_STRINGIFY;
4792 if (SvIVX(PL_lex_stuff) == '\'')
4793 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4794 TERM(sublex_start());
4797 s = scan_pat(s,OP_QR);
4798 TERM(sublex_start());
4801 s = scan_str(s,FALSE,FALSE);
4803 missingterm((char*)0);
4804 yylval.ival = OP_BACKTICK;
4806 TERM(sublex_start());
4814 s = force_version(s, FALSE);
4816 else if (*s != 'v' || !isDIGIT(s[1])
4817 || (s = force_version(s, TRUE), *s == 'v'))
4819 *PL_tokenbuf = '\0';
4820 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4821 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4822 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4824 yyerror("<> should be quotes");
4832 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4836 LOP(OP_RENAME,XTERM);
4845 LOP(OP_RINDEX,XTERM);
4855 UNIDOR(OP_READLINE);
4868 LOP(OP_REVERSE,XTERM);
4871 UNIDOR(OP_READLINK);
4879 TERM(sublex_start());
4881 TOKEN(1); /* force error */
4890 LOP(OP_SELECT,XTERM);
4896 LOP(OP_SEMCTL,XTERM);
4899 LOP(OP_SEMGET,XTERM);
4902 LOP(OP_SEMOP,XTERM);
4908 LOP(OP_SETPGRP,XTERM);
4910 case KEY_setpriority:
4911 LOP(OP_SETPRIORITY,XTERM);
4913 case KEY_sethostent:
4919 case KEY_setservent:
4922 case KEY_setprotoent:
4932 LOP(OP_SEEKDIR,XTERM);
4934 case KEY_setsockopt:
4935 LOP(OP_SSOCKOPT,XTERM);
4941 LOP(OP_SHMCTL,XTERM);
4944 LOP(OP_SHMGET,XTERM);
4947 LOP(OP_SHMREAD,XTERM);
4950 LOP(OP_SHMWRITE,XTERM);
4953 LOP(OP_SHUTDOWN,XTERM);
4962 LOP(OP_SOCKET,XTERM);
4964 case KEY_socketpair:
4965 LOP(OP_SOCKPAIR,XTERM);
4968 checkcomma(s,PL_tokenbuf,"subroutine name");
4970 if (*s == ';' || *s == ')') /* probably a close */
4971 Perl_croak(aTHX_ "sort is now a reserved word");
4973 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4977 LOP(OP_SPLIT,XTERM);
4980 LOP(OP_SPRINTF,XTERM);
4983 LOP(OP_SPLICE,XTERM);
4998 LOP(OP_SUBSTR,XTERM);
5004 char tmpbuf[sizeof PL_tokenbuf];
5005 SSize_t tboffset = 0;
5006 expectation attrful;
5007 bool have_name, have_proto, bad_proto;
5012 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
5013 (*s == ':' && s[1] == ':'))
5016 attrful = XATTRBLOCK;
5017 /* remember buffer pos'n for later force_word */
5018 tboffset = s - PL_oldbufptr;
5019 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
5020 if (strchr(tmpbuf, ':'))
5021 sv_setpv(PL_subname, tmpbuf);
5023 sv_setsv(PL_subname,PL_curstname);
5024 sv_catpvn(PL_subname,"::",2);
5025 sv_catpvn(PL_subname,tmpbuf,len);
5032 Perl_croak(aTHX_ "Missing name in \"my sub\"");
5033 PL_expect = XTERMBLOCK;
5034 attrful = XATTRTERM;
5035 sv_setpv(PL_subname,"?");
5039 if (key == KEY_format) {
5041 PL_lex_formbrack = PL_lex_brackets + 1;
5043 (void) force_word(PL_oldbufptr + tboffset, WORD,
5048 /* Look for a prototype */
5052 s = scan_str(s,FALSE,FALSE);
5054 Perl_croak(aTHX_ "Prototype not terminated");
5055 /* strip spaces and check for bad characters */
5056 d = SvPVX(PL_lex_stuff);
5059 for (p = d; *p; ++p) {
5062 if (!strchr("$@%*;[]&\\", *p))
5067 if (bad_proto && ckWARN(WARN_SYNTAX))
5068 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5069 "Illegal character in prototype for %"SVf" : %s",
5071 SvCUR(PL_lex_stuff) = tmp;
5079 if (*s == ':' && s[1] != ':')
5080 PL_expect = attrful;
5081 else if (!have_name && *s != '{' && key == KEY_sub)
5082 Perl_croak(aTHX_ "Illegal declaration of anonymous subroutine");
5085 PL_nextval[PL_nexttoke].opval =
5086 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
5087 PL_lex_stuff = Nullsv;
5091 sv_setpv(PL_subname,
5092 PL_curstash ? "__ANON__" : "__ANON__::__ANON__");
5095 (void) force_word(PL_oldbufptr + tboffset, WORD,
5104 LOP(OP_SYSTEM,XREF);
5107 LOP(OP_SYMLINK,XTERM);
5110 LOP(OP_SYSCALL,XTERM);
5113 LOP(OP_SYSOPEN,XTERM);
5116 LOP(OP_SYSSEEK,XTERM);
5119 LOP(OP_SYSREAD,XTERM);
5122 LOP(OP_SYSWRITE,XTERM);
5126 TERM(sublex_start());
5147 LOP(OP_TRUNCATE,XTERM);
5159 yylval.ival = CopLINE(PL_curcop);
5163 yylval.ival = CopLINE(PL_curcop);
5167 LOP(OP_UNLINK,XTERM);
5173 LOP(OP_UNPACK,XTERM);
5176 LOP(OP_UTIME,XTERM);
5182 LOP(OP_UNSHIFT,XTERM);
5185 if (PL_expect != XSTATE)
5186 yyerror("\"use\" not allowed in expression");
5188 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
5189 s = force_version(s, TRUE);
5190 if (*s == ';' || (s = skipspace(s), *s == ';')) {
5191 PL_nextval[PL_nexttoke].opval = Nullop;
5194 else if (*s == 'v') {
5195 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5196 s = force_version(s, FALSE);
5200 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5201 s = force_version(s, FALSE);
5213 yylval.ival = CopLINE(PL_curcop);
5217 PL_hints |= HINT_BLOCK_SCOPE;
5224 LOP(OP_WAITPID,XTERM);
5233 ctl_l[0] = toCTRL('L');
5235 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5238 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5243 if (PL_expect == XOPERATOR)
5249 yylval.ival = OP_XOR;
5254 TERM(sublex_start());
5259 #pragma segment Main
5263 S_pending_ident(pTHX)
5266 register I32 tmp = 0;
5267 /* pit holds the identifier we read and pending_ident is reset */
5268 char pit = PL_pending_ident;
5269 PL_pending_ident = 0;
5271 DEBUG_T({ PerlIO_printf(Perl_debug_log,
5272 "### Tokener saw identifier '%s'\n", PL_tokenbuf); });
5274 /* if we're in a my(), we can't allow dynamics here.
5275 $foo'bar has already been turned into $foo::bar, so
5276 just check for colons.
5278 if it's a legal name, the OP is a PADANY.
5281 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
5282 if (strchr(PL_tokenbuf,':'))
5283 yyerror(Perl_form(aTHX_ "No package name allowed for "
5284 "variable %s in \"our\"",
5286 tmp = allocmy(PL_tokenbuf);
5289 if (strchr(PL_tokenbuf,':'))
5290 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
5292 yylval.opval = newOP(OP_PADANY, 0);
5293 yylval.opval->op_targ = allocmy(PL_tokenbuf);
5299 build the ops for accesses to a my() variable.
5301 Deny my($a) or my($b) in a sort block, *if* $a or $b is
5302 then used in a comparison. This catches most, but not
5303 all cases. For instance, it catches
5304 sort { my($a); $a <=> $b }
5306 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
5307 (although why you'd do that is anyone's guess).
5310 if (!strchr(PL_tokenbuf,':')) {
5312 tmp = pad_findmy(PL_tokenbuf);
5313 if (tmp != NOT_IN_PAD) {
5314 /* might be an "our" variable" */
5315 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
5316 /* build ops for a bareword */
5317 SV *sym = newSVpv(HvNAME(PAD_COMPNAME_OURSTASH(tmp)), 0);
5318 sv_catpvn(sym, "::", 2);
5319 sv_catpv(sym, PL_tokenbuf+1);
5320 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
5321 yylval.opval->op_private = OPpCONST_ENTERED;
5322 gv_fetchpv(SvPVX(sym),
5324 ? (GV_ADDMULTI | GV_ADDINEVAL)
5327 ((PL_tokenbuf[0] == '$') ? SVt_PV
5328 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5333 /* if it's a sort block and they're naming $a or $b */
5334 if (PL_last_lop_op == OP_SORT &&
5335 PL_tokenbuf[0] == '$' &&
5336 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
5339 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
5340 d < PL_bufend && *d != '\n';
5343 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
5344 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
5350 yylval.opval = newOP(OP_PADANY, 0);
5351 yylval.opval->op_targ = tmp;
5357 Whine if they've said @foo in a doublequoted string,
5358 and @foo isn't a variable we can find in the symbol
5361 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
5362 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
5363 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
5364 && ckWARN(WARN_AMBIGUOUS))
5366 /* Downgraded from fatal to warning 20000522 mjd */
5367 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5368 "Possible unintended interpolation of %s in string",
5373 /* build ops for a bareword */
5374 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
5375 yylval.opval->op_private = OPpCONST_ENTERED;
5376 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
5377 ((PL_tokenbuf[0] == '$') ? SVt_PV
5378 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5384 Perl_keyword(pTHX_ register char *d, I32 len)
5389 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5390 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5391 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5392 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5393 if (strEQ(d,"__END__")) return KEY___END__;
5397 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5402 if (strEQ(d,"and")) return -KEY_and;
5403 if (strEQ(d,"abs")) return -KEY_abs;
5406 if (strEQ(d,"alarm")) return -KEY_alarm;
5407 if (strEQ(d,"atan2")) return -KEY_atan2;
5410 if (strEQ(d,"accept")) return -KEY_accept;
5415 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5418 if (strEQ(d,"bless")) return -KEY_bless;
5419 if (strEQ(d,"bind")) return -KEY_bind;
5420 if (strEQ(d,"binmode")) return -KEY_binmode;
5423 if (strEQ(d,"CORE")) return -KEY_CORE;
5424 if (strEQ(d,"CHECK")) return KEY_CHECK;
5429 if (strEQ(d,"cmp")) return -KEY_cmp;
5430 if (strEQ(d,"chr")) return -KEY_chr;
5431 if (strEQ(d,"cos")) return -KEY_cos;
5434 if (strEQ(d,"chop")) return -KEY_chop;
5437 if (strEQ(d,"close")) return -KEY_close;
5438 if (strEQ(d,"chdir")) return -KEY_chdir;
5439 if (strEQ(d,"chomp")) return -KEY_chomp;
5440 if (strEQ(d,"chmod")) return -KEY_chmod;
5441 if (strEQ(d,"chown")) return -KEY_chown;
5442 if (strEQ(d,"crypt")) return -KEY_crypt;
5445 if (strEQ(d,"chroot")) return -KEY_chroot;
5446 if (strEQ(d,"caller")) return -KEY_caller;
5449 if (strEQ(d,"connect")) return -KEY_connect;
5452 if (strEQ(d,"closedir")) return -KEY_closedir;
5453 if (strEQ(d,"continue")) return -KEY_continue;
5458 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5463 if (strEQ(d,"do")) return KEY_do;
5466 if (strEQ(d,"die")) return -KEY_die;
5469 if (strEQ(d,"dump")) return -KEY_dump;
5472 if (strEQ(d,"delete")) return KEY_delete;
5475 if (strEQ(d,"defined")) return KEY_defined;
5476 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5479 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5484 if (strEQ(d,"END")) return KEY_END;
5489 if (strEQ(d,"eq")) return -KEY_eq;
5492 if (strEQ(d,"eof")) return -KEY_eof;
5493 if (strEQ(d,"err")) return -KEY_err;
5494 if (strEQ(d,"exp")) return -KEY_exp;
5497 if (strEQ(d,"else")) return KEY_else;
5498 if (strEQ(d,"exit")) return -KEY_exit;
5499 if (strEQ(d,"eval")) return KEY_eval;
5500 if (strEQ(d,"exec")) return -KEY_exec;
5501 if (strEQ(d,"each")) return -KEY_each;
5504 if (strEQ(d,"elsif")) return KEY_elsif;
5507 if (strEQ(d,"exists")) return KEY_exists;
5508 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5511 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5512 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5515 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5518 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5519 if (strEQ(d,"endservent")) return -KEY_endservent;
5522 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5529 if (strEQ(d,"for")) return KEY_for;
5532 if (strEQ(d,"fork")) return -KEY_fork;
5535 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5536 if (strEQ(d,"flock")) return -KEY_flock;
5539 if (strEQ(d,"format")) return KEY_format;
5540 if (strEQ(d,"fileno")) return -KEY_fileno;
5543 if (strEQ(d,"foreach")) return KEY_foreach;
5546 if (strEQ(d,"formline")) return -KEY_formline;
5551 if (strnEQ(d,"get",3)) {
5556 if (strEQ(d,"ppid")) return -KEY_getppid;
5557 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5560 if (strEQ(d,"pwent")) return -KEY_getpwent;
5561 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5562 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5565 if (strEQ(d,"peername")) return -KEY_getpeername;
5566 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5567 if (strEQ(d,"priority")) return -KEY_getpriority;
5570 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5573 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5577 else if (*d == 'h') {
5578 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5579 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5580 if (strEQ(d,"hostent")) return -KEY_gethostent;
5582 else if (*d == 'n') {
5583 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5584 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5585 if (strEQ(d,"netent")) return -KEY_getnetent;
5587 else if (*d == 's') {
5588 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5589 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5590 if (strEQ(d,"servent")) return -KEY_getservent;
5591 if (strEQ(d,"sockname")) return -KEY_getsockname;
5592 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5594 else if (*d == 'g') {
5595 if (strEQ(d,"grent")) return -KEY_getgrent;
5596 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5597 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5599 else if (*d == 'l') {
5600 if (strEQ(d,"login")) return -KEY_getlogin;
5602 else if (strEQ(d,"c")) return -KEY_getc;
5607 if (strEQ(d,"gt")) return -KEY_gt;
5608 if (strEQ(d,"ge")) return -KEY_ge;
5611 if (strEQ(d,"grep")) return KEY_grep;
5612 if (strEQ(d,"goto")) return KEY_goto;
5613 if (strEQ(d,"glob")) return KEY_glob;
5616 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5621 if (strEQ(d,"hex")) return -KEY_hex;
5624 if (strEQ(d,"INIT")) return KEY_INIT;
5629 if (strEQ(d,"if")) return KEY_if;
5632 if (strEQ(d,"int")) return -KEY_int;
5635 if (strEQ(d,"index")) return -KEY_index;
5636 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5641 if (strEQ(d,"join")) return -KEY_join;
5645 if (strEQ(d,"keys")) return -KEY_keys;
5646 if (strEQ(d,"kill")) return -KEY_kill;
5652 if (strEQ(d,"lt")) return -KEY_lt;
5653 if (strEQ(d,"le")) return -KEY_le;
5654 if (strEQ(d,"lc")) return -KEY_lc;
5657 if (strEQ(d,"log")) return -KEY_log;
5660 if (strEQ(d,"last")) return KEY_last;
5661 if (strEQ(d,"link")) return -KEY_link;
5662 if (strEQ(d,"lock")) return -KEY_lock;
5665 if (strEQ(d,"local")) return KEY_local;
5666 if (strEQ(d,"lstat")) return -KEY_lstat;
5669 if (strEQ(d,"length")) return -KEY_length;
5670 if (strEQ(d,"listen")) return -KEY_listen;
5673 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5676 if (strEQ(d,"localtime")) return -KEY_localtime;
5682 case 1: return KEY_m;
5684 if (strEQ(d,"my")) return KEY_my;
5687 if (strEQ(d,"map")) return KEY_map;
5690 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5693 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5694 if (strEQ(d,"msgget")) return -KEY_msgget;
5695 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5696 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5701 if (strEQ(d,"next")) return KEY_next;
5702 if (strEQ(d,"ne")) return -KEY_ne;
5703 if (strEQ(d,"not")) return -KEY_not;
5704 if (strEQ(d,"no")) return KEY_no;
5709 if (strEQ(d,"or")) return -KEY_or;
5712 if (strEQ(d,"ord")) return -KEY_ord;
5713 if (strEQ(d,"oct")) return -KEY_oct;
5714 if (strEQ(d,"our")) return KEY_our;
5717 if (strEQ(d,"open")) return -KEY_open;
5720 if (strEQ(d,"opendir")) return -KEY_opendir;
5727 if (strEQ(d,"pop")) return -KEY_pop;
5728 if (strEQ(d,"pos")) return KEY_pos;
5731 if (strEQ(d,"push")) return -KEY_push;
5732 if (strEQ(d,"pack")) return -KEY_pack;
5733 if (strEQ(d,"pipe")) return -KEY_pipe;
5736 if (strEQ(d,"print")) return KEY_print;
5739 if (strEQ(d,"printf")) return KEY_printf;
5742 if (strEQ(d,"package")) return KEY_package;
5745 if (strEQ(d,"prototype")) return KEY_prototype;
5750 if (strEQ(d,"q")) return KEY_q;
5751 if (strEQ(d,"qr")) return KEY_qr;
5752 if (strEQ(d,"qq")) return KEY_qq;
5753 if (strEQ(d,"qw")) return KEY_qw;
5754 if (strEQ(d,"qx")) return KEY_qx;
5756 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5761 if (strEQ(d,"ref")) return -KEY_ref;
5764 if (strEQ(d,"read")) return -KEY_read;
5765 if (strEQ(d,"rand")) return -KEY_rand;
5766 if (strEQ(d,"recv")) return -KEY_recv;
5767 if (strEQ(d,"redo")) return KEY_redo;
5770 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5771 if (strEQ(d,"reset")) return -KEY_reset;
5774 if (strEQ(d,"return")) return KEY_return;
5775 if (strEQ(d,"rename")) return -KEY_rename;
5776 if (strEQ(d,"rindex")) return -KEY_rindex;
5779 if (strEQ(d,"require")) return KEY_require;
5780 if (strEQ(d,"reverse")) return -KEY_reverse;
5781 if (strEQ(d,"readdir")) return -KEY_readdir;
5784 if (strEQ(d,"readlink")) return -KEY_readlink;
5785 if (strEQ(d,"readline")) return -KEY_readline;
5786 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5789 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5795 case 0: return KEY_s;
5797 if (strEQ(d,"scalar")) return KEY_scalar;
5802 if (strEQ(d,"seek")) return -KEY_seek;
5803 if (strEQ(d,"send")) return -KEY_send;
5806 if (strEQ(d,"semop")) return -KEY_semop;
5809 if (strEQ(d,"select")) return -KEY_select;
5810 if (strEQ(d,"semctl")) return -KEY_semctl;
5811 if (strEQ(d,"semget")) return -KEY_semget;
5814 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5815 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5818 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5819 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5822 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5825 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5826 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5827 if (strEQ(d,"setservent")) return -KEY_setservent;
5830 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5831 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5838 if (strEQ(d,"shift")) return -KEY_shift;
5841 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5842 if (strEQ(d,"shmget")) return -KEY_shmget;
5845 if (strEQ(d,"shmread")) return -KEY_shmread;
5848 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5849 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5854 if (strEQ(d,"sin")) return -KEY_sin;
5857 if (strEQ(d,"sleep")) return -KEY_sleep;
5860 if (strEQ(d,"sort")) return KEY_sort;
5861 if (strEQ(d,"socket")) return -KEY_socket;
5862 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5865 if (strEQ(d,"split")) return KEY_split;
5866 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5867 if (strEQ(d,"splice")) return -KEY_splice;
5870 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5873 if (strEQ(d,"srand")) return -KEY_srand;
5876 if (strEQ(d,"stat")) return -KEY_stat;
5877 if (strEQ(d,"study")) return KEY_study;
5880 if (strEQ(d,"substr")) return -KEY_substr;
5881 if (strEQ(d,"sub")) return KEY_sub;
5886 if (strEQ(d,"system")) return -KEY_system;
5889 if (strEQ(d,"symlink")) return -KEY_symlink;
5890 if (strEQ(d,"syscall")) return -KEY_syscall;
5891 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5892 if (strEQ(d,"sysread")) return -KEY_sysread;
5893 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5896 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5905 if (strEQ(d,"tr")) return KEY_tr;
5908 if (strEQ(d,"tie")) return KEY_tie;
5911 if (strEQ(d,"tell")) return -KEY_tell;
5912 if (strEQ(d,"tied")) return KEY_tied;
5913 if (strEQ(d,"time")) return -KEY_time;
5916 if (strEQ(d,"times")) return -KEY_times;
5919 if (strEQ(d,"telldir")) return -KEY_telldir;
5922 if (strEQ(d,"truncate")) return -KEY_truncate;
5929 if (strEQ(d,"uc")) return -KEY_uc;
5932 if (strEQ(d,"use")) return KEY_use;
5935 if (strEQ(d,"undef")) return KEY_undef;
5936 if (strEQ(d,"until")) return KEY_until;
5937 if (strEQ(d,"untie")) return KEY_untie;
5938 if (strEQ(d,"utime")) return -KEY_utime;
5939 if (strEQ(d,"umask")) return -KEY_umask;
5942 if (strEQ(d,"unless")) return KEY_unless;
5943 if (strEQ(d,"unpack")) return -KEY_unpack;
5944 if (strEQ(d,"unlink")) return -KEY_unlink;
5947 if (strEQ(d,"unshift")) return -KEY_unshift;
5948 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5953 if (strEQ(d,"values")) return -KEY_values;
5954 if (strEQ(d,"vec")) return -KEY_vec;
5959 if (strEQ(d,"warn")) return -KEY_warn;
5960 if (strEQ(d,"wait")) return -KEY_wait;
5963 if (strEQ(d,"while")) return KEY_while;
5964 if (strEQ(d,"write")) return -KEY_write;
5967 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5970 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5975 if (len == 1) return -KEY_x;
5976 if (strEQ(d,"xor")) return -KEY_xor;
5979 if (len == 1) return KEY_y;
5988 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5992 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5993 if (ckWARN(WARN_SYNTAX)) {
5995 for (w = s+2; *w && level; w++) {
6002 for (; *w && isSPACE(*w); w++) ;
6003 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
6004 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
6005 "%s (...) interpreted as function",name);
6008 while (s < PL_bufend && isSPACE(*s))
6012 while (s < PL_bufend && isSPACE(*s))
6014 if (isIDFIRST_lazy_if(s,UTF)) {
6016 while (isALNUM_lazy_if(s,UTF))
6018 while (s < PL_bufend && isSPACE(*s))
6023 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
6027 Perl_croak(aTHX_ "No comma allowed after %s", what);
6032 /* Either returns sv, or mortalizes sv and returns a new SV*.
6033 Best used as sv=new_constant(..., sv, ...).
6034 If s, pv are NULL, calls subroutine with one argument,
6035 and type is used with error messages only. */
6038 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
6042 HV *table = GvHV(PL_hintgv); /* ^H */
6046 const char *why1, *why2, *why3;
6048 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
6051 why2 = strEQ(key,"charnames")
6052 ? "(possibly a missing \"use charnames ...\")"
6054 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
6055 (type ? type: "undef"), why2);
6057 /* This is convoluted and evil ("goto considered harmful")
6058 * but I do not understand the intricacies of all the different
6059 * failure modes of %^H in here. The goal here is to make
6060 * the most probable error message user-friendly. --jhi */
6065 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
6066 (type ? type: "undef"), why1, why2, why3);
6068 yyerror(SvPVX(msg));
6072 cvp = hv_fetch(table, key, strlen(key), FALSE);
6073 if (!cvp || !SvOK(*cvp)) {
6076 why3 = "} is not defined";
6079 sv_2mortal(sv); /* Parent created it permanently */
6082 pv = sv_2mortal(newSVpvn(s, len));
6084 typesv = sv_2mortal(newSVpv(type, 0));
6086 typesv = &PL_sv_undef;
6088 PUSHSTACKi(PERLSI_OVERLOAD);
6100 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
6104 /* Check the eval first */
6105 if (!PL_in_eval && SvTRUE(ERRSV)) {
6107 sv_catpv(ERRSV, "Propagated");
6108 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
6110 res = SvREFCNT_inc(sv);
6114 (void)SvREFCNT_inc(res);
6123 why1 = "Call to &{$^H{";
6125 why3 = "}} did not return a defined value";
6134 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
6136 register char *d = dest;
6137 register char *e = d + destlen - 3; /* two-character token, ending NUL */
6140 Perl_croak(aTHX_ ident_too_long);
6141 if (isALNUM(*s)) /* UTF handled below */
6143 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
6148 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
6152 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6153 char *t = s + UTF8SKIP(s);
6154 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6156 if (d + (t - s) > e)
6157 Perl_croak(aTHX_ ident_too_long);
6158 Copy(s, d, t - s, char);
6171 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
6181 e = d + destlen - 3; /* two-character token, ending NUL */
6183 while (isDIGIT(*s)) {
6185 Perl_croak(aTHX_ ident_too_long);
6192 Perl_croak(aTHX_ ident_too_long);
6193 if (isALNUM(*s)) /* UTF handled below */
6195 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
6200 else if (*s == ':' && s[1] == ':') {
6204 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6205 char *t = s + UTF8SKIP(s);
6206 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6208 if (d + (t - s) > e)
6209 Perl_croak(aTHX_ ident_too_long);
6210 Copy(s, d, t - s, char);
6221 if (PL_lex_state != LEX_NORMAL)
6222 PL_lex_state = LEX_INTERPENDMAYBE;
6225 if (*s == '$' && s[1] &&
6226 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
6239 if (*d == '^' && *s && isCONTROLVAR(*s)) {
6244 if (isSPACE(s[-1])) {
6247 if (!SPACE_OR_TAB(ch)) {
6253 if (isIDFIRST_lazy_if(d,UTF)) {
6257 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
6259 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
6262 Copy(s, d, e - s, char);
6267 while ((isALNUM(*s) || *s == ':') && d < e)
6270 Perl_croak(aTHX_ ident_too_long);
6273 while (s < send && SPACE_OR_TAB(*s)) s++;
6274 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
6275 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
6276 const char *brack = *s == '[' ? "[...]" : "{...}";
6277 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
6278 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
6279 funny, dest, brack, funny, dest, brack);
6282 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
6286 /* Handle extended ${^Foo} variables
6287 * 1999-02-27 mjd-perl-patch@plover.com */
6288 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
6292 while (isALNUM(*s) && d < e) {
6296 Perl_croak(aTHX_ ident_too_long);
6301 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
6302 PL_lex_state = LEX_INTERPEND;
6307 if (PL_lex_state == LEX_NORMAL) {
6308 if (ckWARN(WARN_AMBIGUOUS) &&
6309 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
6311 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
6312 "Ambiguous use of %c{%s} resolved to %c%s",
6313 funny, dest, funny, dest);
6318 s = bracket; /* let the parser handle it */
6322 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
6323 PL_lex_state = LEX_INTERPEND;
6328 Perl_pmflag(pTHX_ U32* pmfl, int ch)
6333 *pmfl |= PMf_GLOBAL;
6335 *pmfl |= PMf_CONTINUE;
6339 *pmfl |= PMf_MULTILINE;
6341 *pmfl |= PMf_SINGLELINE;
6343 *pmfl |= PMf_EXTENDED;
6347 S_scan_pat(pTHX_ char *start, I32 type)
6352 s = scan_str(start,FALSE,FALSE);
6354 Perl_croak(aTHX_ "Search pattern not terminated");
6356 pm = (PMOP*)newPMOP(type, 0);
6357 if (PL_multi_open == '?')
6358 pm->op_pmflags |= PMf_ONCE;
6360 while (*s && strchr("iomsx", *s))
6361 pmflag(&pm->op_pmflags,*s++);
6364 while (*s && strchr("iogcmsx", *s))
6365 pmflag(&pm->op_pmflags,*s++);
6367 /* issue a warning if /c is specified,but /g is not */
6368 if (ckWARN(WARN_REGEXP) &&
6369 (pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL))
6371 Perl_warner(aTHX_ packWARN(WARN_REGEXP), c_without_g);
6374 pm->op_pmpermflags = pm->op_pmflags;
6376 PL_lex_op = (OP*)pm;
6377 yylval.ival = OP_MATCH;
6382 S_scan_subst(pTHX_ char *start)
6389 yylval.ival = OP_NULL;
6391 s = scan_str(start,FALSE,FALSE);
6394 Perl_croak(aTHX_ "Substitution pattern not terminated");
6396 if (s[-1] == PL_multi_open)
6399 first_start = PL_multi_start;
6400 s = scan_str(s,FALSE,FALSE);
6403 SvREFCNT_dec(PL_lex_stuff);
6404 PL_lex_stuff = Nullsv;
6406 Perl_croak(aTHX_ "Substitution replacement not terminated");
6408 PL_multi_start = first_start; /* so whole substitution is taken together */
6410 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6416 else if (strchr("iogcmsx", *s))
6417 pmflag(&pm->op_pmflags,*s++);
6422 /* /c is not meaningful with s/// */
6423 if (ckWARN(WARN_REGEXP) && (pm->op_pmflags & PMf_CONTINUE))
6425 Perl_warner(aTHX_ packWARN(WARN_REGEXP), c_in_subst);
6430 PL_sublex_info.super_bufptr = s;
6431 PL_sublex_info.super_bufend = PL_bufend;
6433 pm->op_pmflags |= PMf_EVAL;
6434 repl = newSVpvn("",0);
6436 sv_catpv(repl, es ? "eval " : "do ");
6437 sv_catpvn(repl, "{ ", 2);
6438 sv_catsv(repl, PL_lex_repl);
6439 sv_catpvn(repl, " };", 2);
6441 SvREFCNT_dec(PL_lex_repl);
6445 pm->op_pmpermflags = pm->op_pmflags;
6446 PL_lex_op = (OP*)pm;
6447 yylval.ival = OP_SUBST;
6452 S_scan_trans(pTHX_ char *start)
6461 yylval.ival = OP_NULL;
6463 s = scan_str(start,FALSE,FALSE);
6465 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6466 if (s[-1] == PL_multi_open)
6469 s = scan_str(s,FALSE,FALSE);
6472 SvREFCNT_dec(PL_lex_stuff);
6473 PL_lex_stuff = Nullsv;
6475 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6478 complement = del = squash = 0;
6479 while (strchr("cds", *s)) {
6481 complement = OPpTRANS_COMPLEMENT;
6483 del = OPpTRANS_DELETE;
6485 squash = OPpTRANS_SQUASH;
6489 New(803, tbl, complement&&!del?258:256, short);
6490 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6491 o->op_private = del|squash|complement|
6492 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
6493 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
6496 yylval.ival = OP_TRANS;
6501 S_scan_heredoc(pTHX_ register char *s)
6504 I32 op_type = OP_SCALAR;
6511 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6515 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6518 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6519 if (*peek && strchr("`'\"",*peek)) {
6522 s = delimcpy(d, e, s, PL_bufend, term, &len);
6532 if (!isALNUM_lazy_if(s,UTF))
6533 deprecate_old("bare << to mean <<\"\"");
6534 for (; isALNUM_lazy_if(s,UTF); s++) {
6539 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6540 Perl_croak(aTHX_ "Delimiter for here document is too long");
6543 len = d - PL_tokenbuf;
6544 #ifndef PERL_STRICT_CR
6545 d = strchr(s, '\r');
6549 while (s < PL_bufend) {
6555 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6564 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6569 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6570 herewas = newSVpvn(s,PL_bufend-s);
6572 s--, herewas = newSVpvn(s,d-s);
6573 s += SvCUR(herewas);
6575 tmpstr = NEWSV(87,79);
6576 sv_upgrade(tmpstr, SVt_PVIV);
6581 else if (term == '`') {
6582 op_type = OP_BACKTICK;
6583 SvIVX(tmpstr) = '\\';
6587 PL_multi_start = CopLINE(PL_curcop);
6588 PL_multi_open = PL_multi_close = '<';
6589 term = *PL_tokenbuf;
6590 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6591 char *bufptr = PL_sublex_info.super_bufptr;
6592 char *bufend = PL_sublex_info.super_bufend;
6593 char *olds = s - SvCUR(herewas);
6594 s = strchr(bufptr, '\n');
6598 while (s < bufend &&
6599 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6601 CopLINE_inc(PL_curcop);
6604 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6605 missingterm(PL_tokenbuf);
6607 sv_setpvn(herewas,bufptr,d-bufptr+1);
6608 sv_setpvn(tmpstr,d+1,s-d);
6610 sv_catpvn(herewas,s,bufend-s);
6611 (void)strcpy(bufptr,SvPVX(herewas));
6618 while (s < PL_bufend &&
6619 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6621 CopLINE_inc(PL_curcop);
6623 if (s >= PL_bufend) {
6624 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6625 missingterm(PL_tokenbuf);
6627 sv_setpvn(tmpstr,d+1,s-d);
6629 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6631 sv_catpvn(herewas,s,PL_bufend-s);
6632 sv_setsv(PL_linestr,herewas);
6633 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6634 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6635 PL_last_lop = PL_last_uni = Nullch;
6638 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6639 while (s >= PL_bufend) { /* multiple line string? */
6641 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6642 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6643 missingterm(PL_tokenbuf);
6645 CopLINE_inc(PL_curcop);
6646 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6647 PL_last_lop = PL_last_uni = Nullch;
6648 #ifndef PERL_STRICT_CR
6649 if (PL_bufend - PL_linestart >= 2) {
6650 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6651 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6653 PL_bufend[-2] = '\n';
6655 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6657 else if (PL_bufend[-1] == '\r')
6658 PL_bufend[-1] = '\n';
6660 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6661 PL_bufend[-1] = '\n';
6663 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6664 SV *sv = NEWSV(88,0);
6666 sv_upgrade(sv, SVt_PVMG);
6667 sv_setsv(sv,PL_linestr);
6670 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6672 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6675 sv_catsv(PL_linestr,herewas);
6676 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6680 sv_catsv(tmpstr,PL_linestr);
6685 PL_multi_end = CopLINE(PL_curcop);
6686 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6687 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6688 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6690 SvREFCNT_dec(herewas);
6692 if (UTF && is_utf8_string((U8*)SvPVX(tmpstr), SvCUR(tmpstr)))
6694 else if (PL_encoding)
6695 sv_recode_to_utf8(tmpstr, PL_encoding);
6697 PL_lex_stuff = tmpstr;
6698 yylval.ival = op_type;
6703 takes: current position in input buffer
6704 returns: new position in input buffer
6705 side-effects: yylval and lex_op are set.
6710 <FH> read from filehandle
6711 <pkg::FH> read from package qualified filehandle
6712 <pkg'FH> read from package qualified filehandle
6713 <$fh> read from filehandle in $fh
6719 S_scan_inputsymbol(pTHX_ char *start)
6721 register char *s = start; /* current position in buffer */
6727 d = PL_tokenbuf; /* start of temp holding space */
6728 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6729 end = strchr(s, '\n');
6732 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6734 /* die if we didn't have space for the contents of the <>,
6735 or if it didn't end, or if we see a newline
6738 if (len >= sizeof PL_tokenbuf)
6739 Perl_croak(aTHX_ "Excessively long <> operator");
6741 Perl_croak(aTHX_ "Unterminated <> operator");
6746 Remember, only scalar variables are interpreted as filehandles by
6747 this code. Anything more complex (e.g., <$fh{$num}>) will be
6748 treated as a glob() call.
6749 This code makes use of the fact that except for the $ at the front,
6750 a scalar variable and a filehandle look the same.
6752 if (*d == '$' && d[1]) d++;
6754 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6755 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6758 /* If we've tried to read what we allow filehandles to look like, and
6759 there's still text left, then it must be a glob() and not a getline.
6760 Use scan_str to pull out the stuff between the <> and treat it
6761 as nothing more than a string.
6764 if (d - PL_tokenbuf != len) {
6765 yylval.ival = OP_GLOB;
6767 s = scan_str(start,FALSE,FALSE);
6769 Perl_croak(aTHX_ "Glob not terminated");
6773 bool readline_overriden = FALSE;
6774 GV *gv_readline = Nullgv;
6776 /* we're in a filehandle read situation */
6779 /* turn <> into <ARGV> */
6781 (void)strcpy(d,"ARGV");
6783 /* Check whether readline() is overriden */
6784 if (((gv_readline = gv_fetchpv("readline", FALSE, SVt_PVCV))
6785 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
6787 ((gvp = (GV**)hv_fetch(PL_globalstash, "readline", 8, FALSE))
6788 && (gv_readline = *gvp) != (GV*)&PL_sv_undef
6789 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
6790 readline_overriden = TRUE;
6792 /* if <$fh>, create the ops to turn the variable into a
6798 /* try to find it in the pad for this block, otherwise find
6799 add symbol table ops
6801 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6802 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
6803 SV *sym = sv_2mortal(
6804 newSVpv(HvNAME(PAD_COMPNAME_OURSTASH(tmp)),0));
6805 sv_catpvn(sym, "::", 2);
6811 OP *o = newOP(OP_PADSV, 0);
6813 PL_lex_op = readline_overriden
6814 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6815 append_elem(OP_LIST, o,
6816 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
6817 : (OP*)newUNOP(OP_READLINE, 0, o);
6826 ? (GV_ADDMULTI | GV_ADDINEVAL)
6829 PL_lex_op = readline_overriden
6830 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6831 append_elem(OP_LIST,
6832 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
6833 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
6834 : (OP*)newUNOP(OP_READLINE, 0,
6835 newUNOP(OP_RV2SV, 0,
6836 newGVOP(OP_GV, 0, gv)));
6838 if (!readline_overriden)
6839 PL_lex_op->op_flags |= OPf_SPECIAL;
6840 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6841 yylval.ival = OP_NULL;
6844 /* If it's none of the above, it must be a literal filehandle
6845 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6847 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6848 PL_lex_op = readline_overriden
6849 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6850 append_elem(OP_LIST,
6851 newGVOP(OP_GV, 0, gv),
6852 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
6853 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6854 yylval.ival = OP_NULL;
6863 takes: start position in buffer
6864 keep_quoted preserve \ on the embedded delimiter(s)
6865 keep_delims preserve the delimiters around the string
6866 returns: position to continue reading from buffer
6867 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6868 updates the read buffer.
6870 This subroutine pulls a string out of the input. It is called for:
6871 q single quotes q(literal text)
6872 ' single quotes 'literal text'
6873 qq double quotes qq(interpolate $here please)
6874 " double quotes "interpolate $here please"
6875 qx backticks qx(/bin/ls -l)
6876 ` backticks `/bin/ls -l`
6877 qw quote words @EXPORT_OK = qw( func() $spam )
6878 m// regexp match m/this/
6879 s/// regexp substitute s/this/that/
6880 tr/// string transliterate tr/this/that/
6881 y/// string transliterate y/this/that/
6882 ($*@) sub prototypes sub foo ($)
6883 (stuff) sub attr parameters sub foo : attr(stuff)
6884 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6886 In most of these cases (all but <>, patterns and transliterate)
6887 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6888 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6889 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6892 It skips whitespace before the string starts, and treats the first
6893 character as the delimiter. If the delimiter is one of ([{< then
6894 the corresponding "close" character )]}> is used as the closing
6895 delimiter. It allows quoting of delimiters, and if the string has
6896 balanced delimiters ([{<>}]) it allows nesting.
6898 On success, the SV with the resulting string is put into lex_stuff or,
6899 if that is already non-NULL, into lex_repl. The second case occurs only
6900 when parsing the RHS of the special constructs s/// and tr/// (y///).
6901 For convenience, the terminating delimiter character is stuffed into
6906 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6908 SV *sv; /* scalar value: string */
6909 char *tmps; /* temp string, used for delimiter matching */
6910 register char *s = start; /* current position in the buffer */
6911 register char term; /* terminating character */
6912 register char *to; /* current position in the sv's data */
6913 I32 brackets = 1; /* bracket nesting level */
6914 bool has_utf8 = FALSE; /* is there any utf8 content? */
6915 I32 termcode; /* terminating char. code */
6916 U8 termstr[UTF8_MAXLEN]; /* terminating string */
6917 STRLEN termlen; /* length of terminating string */
6918 char *last = NULL; /* last position for nesting bracket */
6920 /* skip space before the delimiter */
6924 /* mark where we are, in case we need to report errors */
6927 /* after skipping whitespace, the next character is the terminator */
6930 termcode = termstr[0] = term;
6934 termcode = utf8_to_uvchr((U8*)s, &termlen);
6935 Copy(s, termstr, termlen, U8);
6936 if (!UTF8_IS_INVARIANT(term))
6940 /* mark where we are */
6941 PL_multi_start = CopLINE(PL_curcop);
6942 PL_multi_open = term;
6944 /* find corresponding closing delimiter */
6945 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6946 termcode = termstr[0] = term = tmps[5];
6948 PL_multi_close = term;
6950 /* create a new SV to hold the contents. 87 is leak category, I'm
6951 assuming. 79 is the SV's initial length. What a random number. */
6953 sv_upgrade(sv, SVt_PVIV);
6954 SvIVX(sv) = termcode;
6955 (void)SvPOK_only(sv); /* validate pointer */
6957 /* move past delimiter and try to read a complete string */
6959 sv_catpvn(sv, s, termlen);
6962 if (PL_encoding && !UTF) {
6966 int offset = s - SvPVX(PL_linestr);
6967 bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
6968 &offset, (char*)termstr, termlen);
6969 char *ns = SvPVX(PL_linestr) + offset;
6970 char *svlast = SvEND(sv) - 1;
6972 for (; s < ns; s++) {
6973 if (*s == '\n' && !PL_rsfp)
6974 CopLINE_inc(PL_curcop);
6977 goto read_more_line;
6979 /* handle quoted delimiters */
6980 if (*(svlast-1) == '\\') {
6982 for (t = svlast-2; t >= SvPVX(sv) && *t == '\\';)
6984 if ((svlast-1 - t) % 2) {
6988 SvCUR_set(sv, SvCUR(sv) - 1);
6993 if (PL_multi_open == PL_multi_close) {
7000 for (w = t = last; t < svlast; w++, t++) {
7001 /* At here, all closes are "was quoted" one,
7002 so we don't check PL_multi_close. */
7004 if (!keep_quoted && *(t+1) == PL_multi_open)
7009 else if (*t == PL_multi_open)
7017 SvCUR_set(sv, w - SvPVX(sv));
7020 if (--brackets <= 0)
7026 SvCUR_set(sv, SvCUR(sv) - 1);
7032 /* extend sv if need be */
7033 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
7034 /* set 'to' to the next character in the sv's string */
7035 to = SvPVX(sv)+SvCUR(sv);
7037 /* if open delimiter is the close delimiter read unbridle */
7038 if (PL_multi_open == PL_multi_close) {
7039 for (; s < PL_bufend; s++,to++) {
7040 /* embedded newlines increment the current line number */
7041 if (*s == '\n' && !PL_rsfp)
7042 CopLINE_inc(PL_curcop);
7043 /* handle quoted delimiters */
7044 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
7045 if (!keep_quoted && s[1] == term)
7047 /* any other quotes are simply copied straight through */
7051 /* terminate when run out of buffer (the for() condition), or
7052 have found the terminator */
7053 else if (*s == term) {
7056 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
7059 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
7065 /* if the terminator isn't the same as the start character (e.g.,
7066 matched brackets), we have to allow more in the quoting, and
7067 be prepared for nested brackets.
7070 /* read until we run out of string, or we find the terminator */
7071 for (; s < PL_bufend; s++,to++) {
7072 /* embedded newlines increment the line count */
7073 if (*s == '\n' && !PL_rsfp)
7074 CopLINE_inc(PL_curcop);
7075 /* backslashes can escape the open or closing characters */
7076 if (*s == '\\' && s+1 < PL_bufend) {
7078 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
7083 /* allow nested opens and closes */
7084 else if (*s == PL_multi_close && --brackets <= 0)
7086 else if (*s == PL_multi_open)
7088 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
7093 /* terminate the copied string and update the sv's end-of-string */
7095 SvCUR_set(sv, to - SvPVX(sv));
7098 * this next chunk reads more into the buffer if we're not done yet
7102 break; /* handle case where we are done yet :-) */
7104 #ifndef PERL_STRICT_CR
7105 if (to - SvPVX(sv) >= 2) {
7106 if ((to[-2] == '\r' && to[-1] == '\n') ||
7107 (to[-2] == '\n' && to[-1] == '\r'))
7111 SvCUR_set(sv, to - SvPVX(sv));
7113 else if (to[-1] == '\r')
7116 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
7121 /* if we're out of file, or a read fails, bail and reset the current
7122 line marker so we can report where the unterminated string began
7125 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
7127 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
7130 /* we read a line, so increment our line counter */
7131 CopLINE_inc(PL_curcop);
7133 /* update debugger info */
7134 if (PERLDB_LINE && PL_curstash != PL_debstash) {
7135 SV *sv = NEWSV(88,0);
7137 sv_upgrade(sv, SVt_PVMG);
7138 sv_setsv(sv,PL_linestr);
7141 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
7144 /* having changed the buffer, we must update PL_bufend */
7145 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7146 PL_last_lop = PL_last_uni = Nullch;
7149 /* at this point, we have successfully read the delimited string */
7151 if (!PL_encoding || UTF) {
7153 sv_catpvn(sv, s, termlen);
7156 if (has_utf8 || PL_encoding)
7159 PL_multi_end = CopLINE(PL_curcop);
7161 /* if we allocated too much space, give some back */
7162 if (SvCUR(sv) + 5 < SvLEN(sv)) {
7163 SvLEN_set(sv, SvCUR(sv) + 1);
7164 Renew(SvPVX(sv), SvLEN(sv), char);
7167 /* decide whether this is the first or second quoted string we've read
7180 takes: pointer to position in buffer
7181 returns: pointer to new position in buffer
7182 side-effects: builds ops for the constant in yylval.op
7184 Read a number in any of the formats that Perl accepts:
7186 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
7187 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
7190 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
7192 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
7195 If it reads a number without a decimal point or an exponent, it will
7196 try converting the number to an integer and see if it can do so
7197 without loss of precision.
7201 Perl_scan_num(pTHX_ char *start, YYSTYPE* lvalp)
7203 register char *s = start; /* current position in buffer */
7204 register char *d; /* destination in temp buffer */
7205 register char *e; /* end of temp buffer */
7206 NV nv; /* number read, as a double */
7207 SV *sv = Nullsv; /* place to put the converted number */
7208 bool floatit; /* boolean: int or float? */
7209 char *lastub = 0; /* position of last underbar */
7210 static char number_too_long[] = "Number too long";
7212 /* We use the first character to decide what type of number this is */
7216 Perl_croak(aTHX_ "panic: scan_num");
7218 /* if it starts with a 0, it could be an octal number, a decimal in
7219 0.13 disguise, or a hexadecimal number, or a binary number. */
7223 u holds the "number so far"
7224 shift the power of 2 of the base
7225 (hex == 4, octal == 3, binary == 1)
7226 overflowed was the number more than we can hold?
7228 Shift is used when we add a digit. It also serves as an "are
7229 we in octal/hex/binary?" indicator to disallow hex characters
7235 bool overflowed = FALSE;
7236 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
7237 static char* bases[5] = { "", "binary", "", "octal",
7239 static char* Bases[5] = { "", "Binary", "", "Octal",
7241 static char *maxima[5] = { "",
7242 "0b11111111111111111111111111111111",
7246 char *base, *Base, *max;
7252 } else if (s[1] == 'b') {
7256 /* check for a decimal in disguise */
7257 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
7259 /* so it must be octal */
7266 if (ckWARN(WARN_SYNTAX))
7267 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7268 "Misplaced _ in number");
7272 base = bases[shift];
7273 Base = Bases[shift];
7274 max = maxima[shift];
7276 /* read the rest of the number */
7278 /* x is used in the overflow test,
7279 b is the digit we're adding on. */
7284 /* if we don't mention it, we're done */
7288 /* _ are ignored -- but warned about if consecutive */
7290 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7291 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7292 "Misplaced _ in number");
7296 /* 8 and 9 are not octal */
7299 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
7303 case '2': case '3': case '4':
7304 case '5': case '6': case '7':
7306 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
7310 b = *s++ & 15; /* ASCII digit -> value of digit */
7314 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
7315 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
7316 /* make sure they said 0x */
7321 /* Prepare to put the digit we have onto the end
7322 of the number so far. We check for overflows.
7327 x = u << shift; /* make room for the digit */
7329 if ((x >> shift) != u
7330 && !(PL_hints & HINT_NEW_BINARY)) {
7333 if (ckWARN_d(WARN_OVERFLOW))
7334 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
7335 "Integer overflow in %s number",
7338 u = x | b; /* add the digit to the end */
7341 n *= nvshift[shift];
7342 /* If an NV has not enough bits in its
7343 * mantissa to represent an UV this summing of
7344 * small low-order numbers is a waste of time
7345 * (because the NV cannot preserve the
7346 * low-order bits anyway): we could just
7347 * remember when did we overflow and in the
7348 * end just multiply n by the right
7356 /* if we get here, we had success: make a scalar value from
7361 /* final misplaced underbar check */
7363 if (ckWARN(WARN_SYNTAX))
7364 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
7369 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
7370 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
7371 "%s number > %s non-portable",
7377 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
7378 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
7379 "%s number > %s non-portable",
7384 if (PL_hints & HINT_NEW_BINARY)
7385 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
7390 handle decimal numbers.
7391 we're also sent here when we read a 0 as the first digit
7393 case '1': case '2': case '3': case '4': case '5':
7394 case '6': case '7': case '8': case '9': case '.':
7397 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
7400 /* read next group of digits and _ and copy into d */
7401 while (isDIGIT(*s) || *s == '_') {
7402 /* skip underscores, checking for misplaced ones
7406 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7407 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7408 "Misplaced _ in number");
7412 /* check for end of fixed-length buffer */
7414 Perl_croak(aTHX_ number_too_long);
7415 /* if we're ok, copy the character */
7420 /* final misplaced underbar check */
7421 if (lastub && s == lastub + 1) {
7422 if (ckWARN(WARN_SYNTAX))
7423 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
7426 /* read a decimal portion if there is one. avoid
7427 3..5 being interpreted as the number 3. followed
7430 if (*s == '.' && s[1] != '.') {
7435 if (ckWARN(WARN_SYNTAX))
7436 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7437 "Misplaced _ in number");
7441 /* copy, ignoring underbars, until we run out of digits.
7443 for (; isDIGIT(*s) || *s == '_'; s++) {
7444 /* fixed length buffer check */
7446 Perl_croak(aTHX_ number_too_long);
7448 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7449 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7450 "Misplaced _ in number");
7456 /* fractional part ending in underbar? */
7458 if (ckWARN(WARN_SYNTAX))
7459 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7460 "Misplaced _ in number");
7462 if (*s == '.' && isDIGIT(s[1])) {
7463 /* oops, it's really a v-string, but without the "v" */
7469 /* read exponent part, if present */
7470 if (*s && strchr("eE",*s) && strchr("+-0123456789_", s[1])) {
7474 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
7475 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
7477 /* stray preinitial _ */
7479 if (ckWARN(WARN_SYNTAX))
7480 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7481 "Misplaced _ in number");
7485 /* allow positive or negative exponent */
7486 if (*s == '+' || *s == '-')
7489 /* stray initial _ */
7491 if (ckWARN(WARN_SYNTAX))
7492 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7493 "Misplaced _ in number");
7497 /* read digits of exponent */
7498 while (isDIGIT(*s) || *s == '_') {
7501 Perl_croak(aTHX_ number_too_long);
7505 if (ckWARN(WARN_SYNTAX) &&
7506 ((lastub && s == lastub + 1) ||
7507 (!isDIGIT(s[1]) && s[1] != '_')))
7508 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7509 "Misplaced _ in number");
7516 /* make an sv from the string */
7520 We try to do an integer conversion first if no characters
7521 indicating "float" have been found.
7526 int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
7528 if (flags == IS_NUMBER_IN_UV) {
7530 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7533 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
7534 if (uv <= (UV) IV_MIN)
7535 sv_setiv(sv, -(IV)uv);
7542 /* terminate the string */
7544 nv = Atof(PL_tokenbuf);
7548 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7549 (PL_hints & HINT_NEW_INTEGER) )
7550 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7551 (floatit ? "float" : "integer"),
7555 /* if it starts with a v, it could be a v-string */
7558 sv = NEWSV(92,5); /* preallocate storage space */
7559 s = scan_vstring(s,sv);
7563 /* make the op for the constant and return */
7566 lvalp->opval = newSVOP(OP_CONST, 0, sv);
7568 lvalp->opval = Nullop;
7574 S_scan_formline(pTHX_ register char *s)
7578 SV *stuff = newSVpvn("",0);
7579 bool needargs = FALSE;
7582 if (*s == '.' || *s == /*{*/'}') {
7584 #ifdef PERL_STRICT_CR
7585 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7587 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7589 if (*t == '\n' || t == PL_bufend)
7592 if (PL_in_eval && !PL_rsfp) {
7593 eol = strchr(s,'\n');
7598 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7600 for (t = s; t < eol; t++) {
7601 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7603 goto enough; /* ~~ must be first line in formline */
7605 if (*t == '@' || *t == '^')
7609 sv_catpvn(stuff, s, eol-s);
7610 #ifndef PERL_STRICT_CR
7611 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7612 char *end = SvPVX(stuff) + SvCUR(stuff);
7624 s = filter_gets(PL_linestr, PL_rsfp, 0);
7625 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7626 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7627 PL_last_lop = PL_last_uni = Nullch;
7630 yyerror("Format not terminated");
7640 PL_lex_state = LEX_NORMAL;
7641 PL_nextval[PL_nexttoke].ival = 0;
7645 PL_lex_state = LEX_FORMLINE;
7647 if (UTF && is_utf8_string((U8*)SvPVX(stuff), SvCUR(stuff)))
7649 else if (PL_encoding)
7650 sv_recode_to_utf8(stuff, PL_encoding);
7652 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7654 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7658 SvREFCNT_dec(stuff);
7659 PL_lex_formbrack = 0;
7670 PL_cshlen = strlen(PL_cshname);
7675 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7677 I32 oldsavestack_ix = PL_savestack_ix;
7678 CV* outsidecv = PL_compcv;
7681 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7683 SAVEI32(PL_subline);
7684 save_item(PL_subname);
7685 SAVESPTR(PL_compcv);
7687 PL_compcv = (CV*)NEWSV(1104,0);
7688 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7689 CvFLAGS(PL_compcv) |= flags;
7691 PL_subline = CopLINE(PL_curcop);
7692 CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
7693 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7694 CvOUTSIDE_SEQ(PL_compcv) = PL_cop_seqmax;
7696 return oldsavestack_ix;
7700 #pragma segment Perl_yylex
7703 Perl_yywarn(pTHX_ char *s)
7705 PL_in_eval |= EVAL_WARNONLY;
7707 PL_in_eval &= ~EVAL_WARNONLY;
7712 Perl_yyerror(pTHX_ char *s)
7715 char *context = NULL;
7719 if (!yychar || (yychar == ';' && !PL_rsfp))
7721 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7722 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7725 The code below is removed for NetWare because it abends/crashes on NetWare
7726 when the script has error such as not having the closing quotes like:
7728 Checking of white spaces is anyway done in NetWare code.
7731 while (isSPACE(*PL_oldoldbufptr))
7734 context = PL_oldoldbufptr;
7735 contlen = PL_bufptr - PL_oldoldbufptr;
7737 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7738 PL_oldbufptr != PL_bufptr) {
7741 The code below is removed for NetWare because it abends/crashes on NetWare
7742 when the script has error such as not having the closing quotes like:
7744 Checking of white spaces is anyway done in NetWare code.
7747 while (isSPACE(*PL_oldbufptr))
7750 context = PL_oldbufptr;
7751 contlen = PL_bufptr - PL_oldbufptr;
7753 else if (yychar > 255)
7754 where = "next token ???";
7755 #ifdef USE_PURE_BISON
7756 /* GNU Bison sets the value -2 */
7757 else if (yychar == -2) {
7759 else if ((yychar & 127) == 127) {
7761 if (PL_lex_state == LEX_NORMAL ||
7762 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7763 where = "at end of line";
7764 else if (PL_lex_inpat)
7765 where = "within pattern";
7767 where = "within string";
7770 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7772 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7773 else if (isPRINT_LC(yychar))
7774 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7776 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7777 where = SvPVX(where_sv);
7779 msg = sv_2mortal(newSVpv(s, 0));
7780 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7781 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7783 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7785 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7786 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7787 Perl_sv_catpvf(aTHX_ msg,
7788 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7789 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7792 if (PL_in_eval & EVAL_WARNONLY)
7793 Perl_warn(aTHX_ "%"SVf, msg);
7796 if (PL_error_count >= 10) {
7797 if (PL_in_eval && SvCUR(ERRSV))
7798 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
7799 ERRSV, OutCopFILE(PL_curcop));
7801 Perl_croak(aTHX_ "%s has too many errors.\n",
7802 OutCopFILE(PL_curcop));
7805 PL_in_my_stash = Nullhv;
7809 #pragma segment Main
7813 S_swallow_bom(pTHX_ U8 *s)
7816 slen = SvCUR(PL_linestr);
7820 /* UTF-16 little-endian */
7821 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
7822 Perl_croak(aTHX_ "Unsupported script encoding");
7823 #ifndef PERL_NO_UTF16_FILTER
7824 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-LE script encoding\n"));
7826 if (PL_bufend > (char*)s) {
7830 filter_add(utf16rev_textfilter, NULL);
7831 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7832 PL_bufend = (char*)utf16_to_utf8_reversed(s, news,
7833 PL_bufend - (char*)s - 1,
7835 Copy(news, s, newlen, U8);
7836 SvCUR_set(PL_linestr, newlen);
7837 PL_bufend = SvPVX(PL_linestr) + newlen;
7838 news[newlen++] = '\0';
7842 Perl_croak(aTHX_ "Unsupported script encoding");
7847 if (s[1] == 0xFF) { /* UTF-16 big-endian */
7848 #ifndef PERL_NO_UTF16_FILTER
7849 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding\n"));
7851 if (PL_bufend > (char *)s) {
7855 filter_add(utf16_textfilter, NULL);
7856 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7857 PL_bufend = (char*)utf16_to_utf8(s, news,
7858 PL_bufend - (char*)s,
7860 Copy(news, s, newlen, U8);
7861 SvCUR_set(PL_linestr, newlen);
7862 PL_bufend = SvPVX(PL_linestr) + newlen;
7863 news[newlen++] = '\0';
7867 Perl_croak(aTHX_ "Unsupported script encoding");
7872 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
7873 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-8 script encoding\n"));
7878 if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */
7879 s[2] == 0xFE && s[3] == 0xFF)
7881 Perl_croak(aTHX_ "Unsupported script encoding");
7889 * Restore a source filter.
7893 restore_rsfp(pTHX_ void *f)
7895 PerlIO *fp = (PerlIO*)f;
7897 if (PL_rsfp == PerlIO_stdin())
7898 PerlIO_clearerr(PL_rsfp);
7899 else if (PL_rsfp && (PL_rsfp != fp))
7900 PerlIO_close(PL_rsfp);
7904 #ifndef PERL_NO_UTF16_FILTER
7906 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
7908 I32 count = FILTER_READ(idx+1, sv, maxlen);
7913 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7914 if (!*SvPV_nolen(sv))
7915 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7918 tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7919 sv_usepvn(sv, (char*)tmps, tend - tmps);
7925 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
7927 I32 count = FILTER_READ(idx+1, sv, maxlen);
7932 if (!*SvPV_nolen(sv))
7933 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7936 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7937 tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7938 sv_usepvn(sv, (char*)tmps, tend - tmps);